LIBRARY *y

MiChiQan State
University

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClRC/DateDuepSS-pJS

 

MICRO-SCALE IN VIVO IMAGING OF FREEZE INJURY IN TISSUES
By

Gloria D. Elliott

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Mechanical Engineering

2002

 

 

F“ ‘n. '. p

l\l.‘-_l‘v‘._\“

'1. H

*v - l0 ‘
I

.' .

3‘ -I

1 r»).
.\‘1"‘

‘1

I"... .
' “XI \

on
O

finh‘

an.
\L‘ .n N
K“. -
~. .
v. s
s
5
AN
‘w- \ .
‘k d‘i",«
'3.
l
.3,»
‘K.

ABSTRACT
MICRO-SCALE IN VIVO IMAGING OF FREEZING IN TISSUES
By

Gloria D. Elliott

Cryosurgery is the use of localized freezing to kill undesirable tissue. Although
rapid fieezing rates are known to create intracellular ice, a phenomenon that correlates
highly with cell death, the effects of slow freezing rates and high sub-zero temperatures
that characterize the edge of a cryo-lesion are not well understood. Understanding the
relationship between thermal history and physiological response in tissues relies on the
ability to generate quantitative measures of each.

A novel Green Fluorescent Protein (OF P) transfection based viability assay was
developed that allowed the continuous in vivo evaluation of injury in the cryo-treated
tissue. GFP produced by the transfected cell was found to leak from damaged cells but
was retained in living cells. Viability results with this new assay correlated highly with
other static indicators of death. A radial mass diffusion and dilution mathematical model
was found to adequately explain the transient changes in GP P fluorescence intensity
observed following death of cells in suspension and within tissues.

Obtaining accurate thermal measurements within small cryo-lesions is very
challenging. Typically only a few thermocouples can be used, and these are prone to fin
efl'ects and placement error. To create a cryo-lesion that was thermally well
characterized, a novel annular freeze apparatus was developed that minimized the spatial

temperature gradients. This device uniformly freezes tissue within a rat skin flap

 

lvmh”

(if; u -'

‘JO~ Hilli‘
. lbslfi‘“

A“. ' '3"
a.“ A~}\.-

Us...

"1‘5 .ir‘.
)U‘ [55}.
1‘?“ "I
an¢lh‘

.
1".” ".1.
‘05: 1,‘\

v‘.-
‘K‘T’M’

“5&4‘

0.

will ’3
'F‘w'o ,‘L

““"'- u.

" l-v
‘IH...~

chamber while allowing optical access to the tissue before, during, and afier treatment.
Meticulous thermocouple measurements were acquired in phantom tissues and in vivo,
and results were also compared to a numerical simulation to validate the uniformity of the
field.

These new technologies were brought together to investigate the effect of high
sub-zero temperatures and slow cooling rates on both the microvasculature and tumor cell
viability in a living rodent. The microvasculature was severely disrupted immediately
after freezing, evidenced by vessel dilation, hemorrhage, and stasis. The cells survived
the freezing, but died several hours after the treatment. These results provide the first
direct evidence that cells in vivo are not dying immediately after the initial freezing
insult, thus implicating the role of secondary effects and possibly solute injury in creating

the ultimate cryo-lesion.

To my nephews Cody and Jonathon, for being a
welcomed distraction when I needed to empty my head, and

a motivating influence when I needed to fill it

- ' an ‘
div- MIU

-'.' l
'4.

I I
. . ‘b‘ .
l“ i’kll Cl

.. a
‘ft'kw- \r_
“-":IA. A...

H.
-I ll"
I ‘k.

ACKNOWLEDGEMENTS

This work could not have been completed without the support and participation
of many outstanding individuals. First and foremost I would like to thank my supervisor,
John J. McGrath, for his patience and support throughout the duration of my program,
and also for setting the bar high and encouraging me to reach for it. I’d also like to thank
my guidance committee for their willingness to learn a new ‘language’ as engineering
collided with science in this multi-disciplinary dissertation. Their constructive criticism
and encouragement during difficult phases of the research helped me to stay the path.

A number of undergraduate students made many significant contributions to this
research effort. I’d like to thank Mandy Miller, Jill Brunner, Maureen Galloway, Sarah
Harbin, Suzy Smith, and Diane Doeing for the many long hours they spent collecting or
processing data, or in supporting the efforts that first got us to that point. I’d also like to
thank Damien Fron, Noelle Frerichs, and Cathy Peltz for their significant involvement in
this project. Through their various independent study projects, they were instrumental in
creating and developing the cryosurgical probe technology used in this work. Many
thanks are due my lab colleagues Al Aksan and Paul Hoke. The many conversations we
had about both personal and professional goals were both motivating and inspiring.

A major component of this work was made possible by the generous provision of
supplies, research equipment, and lab space by Dr. Elahe Crockett. I’ve appreciated this
opportunity to conduct a major proportion of my work in a surgical environment, but I’m
especially thankful for the friendship that developed out of this relationship. I’m also

grateful for the opportunity to work with Crystal Remelius. Her willingness to share her

 

:1}: will {ht '
l0?

3% l ‘ "is

w. u I;
anti mt;
Wl‘.

lo I

.1. “My?“
‘1‘“)- tut...

expertise and experience with cell culture and small animal surgical procedures provided
me with the skills and guidance I needed to complete this work.

For the many delightful meals and the wonderful company after long days in the
lab, I’d like to thank Allan and Colleen. We ate like kings and queens at your house and
created many memories that I will always remember as an integral part of my days at
MSU.

To my husband Rob, thank you for marrying me amidst the madness of writing

this dissertation and conducting those ‘last’ experiments. Your patience knows no end.

vi

 

:5: I'll

:3"
.. I k)!

l. l"

I. :0
H \
1.211
I.“
1.43

TABLE OF CONTENTS

 

 

 

 

LIST OF TABLES .......................................................................................................... x
LIST OF FIGURES ......................................................................................................... xi
1. Introduction .............................................................................................................. l
1.1 Motivation for the Research ............................................................................... l

1.2 Inna-vital Microscopy .......................................................................................... 4

1.3 Viability Assays .................................................................................................. 5

1.4 State of the Art in Cryosurgery Research ........................................................... 9
1.4.1 Historical Perspective ............................................................................... 9

1.4.2 Direct Cellular Injury ................................................................................ 12

1.4.3 Secondary Injury ....................................................................................... 15

2. Molecular and Physical Basis for Proposed GFP Viability Assay .......................... 19
2.1 GF P Transfection .............................................................................................. 19

2.2 Enhanced Green Fluorescent Protein ................................................................ 23

2.3 Photodynamics .................................................................................................. 26
2.4 GFP as the Basis for a Viability Assay ............................................................. 29

3. Significance of the Research .................................................................................. 31
4. Cell Line and Animal Model .................................................................................. 35
4.1 GFP-Transfected Rat 3230 Mammary Adenocarcinoma Cell Line .................. 35
4.1.1 Cell Culture _ - -- .............. 36

4.1.2 Doubling Time ........................................................................................ 36

4.1.3 Cell Size Distribution ............................................................................. 38

4.2 Fischer 344 Rat ................................................................................................ 39
4.3 Immune Response to GF P ............................................................................... 40

5. Characterization of Water Transport in GFP-R3230AC Cells .............................. 42
5.1 Overview .......................................................................................................... 42
5.2 Biophysical Model ........................................................................................... 46
5.3 Materials and Methods ..................................................................................... 51
5.3.1 Determination of Vb, Osmotically Inactive Volume ............................... 51

5.3.2 Determination of Lp and EH, ................................................................... 53

5.3.3 MathematicaTM Models ........................................................................... 59

5.3.3.1 Validation of DiffCham Results ................... 59

5.3.3.2 Simulation of Water Transport During Freezing ....................... 59

5.4 Results .............................................................................................................. 60
5.4.1 Boyle Vant’ Hoff Plot ................ .. ......... 60

5.4.2 Estimates of Lp and E”, .......................................................................... 61

vii

5.4
5.4
55 [)5

$660

6 \i‘iiil
61 A»:
6.1.

6l.

6.].
6.1.
6.2 (h:
6.2.
6.2..
6.2.

6-2 ‘

5.4.3 Diffcham Check ...................................................................................... 65

 

 

 

 

 

 

5.4.4 Water Transport During Freezing ........................................................... 66
5.5 Discussion ........................................................................................................ 68
5.6 Conclusions ...................................................................................................... 7O
. Validation of a GFP-Based Viability Assay at the Cellular Level .......................... 71
6.1 Assessment of Injury in R3230 AC cells ........................................................... 71
6.1.1 Comparison of GFP-based Imaging to Standard Viability
Assays Following Freezing ....................................................................... 71
6.1.2 Comparison of GFP-based Imaging to Standard Viability
Assays Following Solute Exposure .......................................................... 78
6.1.3 Clonogenic Assay ..................................................................................... 80
6.1.4 Photobleach and Photodarnage ................................................................. 82
6.2 Characterization of the GF P Assay .................................................................... 85
6.2.1 Intracellular GF P Half-Life ....................................................................... 86
6.2.2 Stability of GFP to Low Temperatures ..................................................... 91
6.2.3 Dilution-Diffusion Hypothesis ................................................................. 100
6.2.4 Mass Conservation ................................................................................... 107
6.2.5 Conclusions .............................................................................................. 111
. Validation of a GFP-based Viability Assay In Vivo ................................................ 113
7.1 R3230 AC Tumor Grown in the Dorsal Skin Flap Chamber: Implant
Procedure . - ................... 113
7.2 Experimental Design- - .................................................................. 117
7.3 Cryosurgical Probe Development ...................................................................... 120
7.4 Thermal History: Tissue Phantom Measurements to Characterize
Cryosurgery Treatment ...................................................................................... 121
7.5 Post-Cryosurgical Damage to GFP-Fluorescing Tumor Tissue ....................... 125
7.5.1 Intravital Assessment ................................................................................ 125
7.5.2 Histopathology Assessment ...................................................................... 128
7.5.3 Comparison of Results .............................................................................. 135
7.6 Dynamics of Fluorescence Intensity Change in vivo ......................................... 138
. Design and Characterization of a Novel Annular Freeze Apparatus for the
Dorsal Skin Flap Chamber .............................................................. 139
8.1 Motivation - ..................................... 139
8.2 Design Objectives .............................................................................................. 143
8.3 Construction of the Annular Freeze Apparatus ................................................ 145
8.4 Thermal Characterization of the Annular Freeze Apparatus ............................. 147
8.5 BioHeat Transfer Model Development ............................................. 154
8.5.1 Mathematical Models of Biological Tissue Freezing in the
Literature ................................................................................................. 154
8.5.2 Governing Equations ................................................................................ 155
8.5.3 Latent Heat Effects ................................................................................... 158
viii

85.5

9. Applica‘
Slim-Fr
9.111.323
9.2 Rel.

llbi.
9.3 Pill.
9.4159

l&\\

8.5.4 Benchmark Solutions ............................................................................... 158

8.5.4.1 Apparent Cp Method with F luentTM ............................................. 160
8.5.4.2 Neumann Solution: 2-Phase ........................................................ 162
8.5.4.3 Cho and Sutherland Solution: 3-Phase .......................................... 165
8.5.4.4 Comparison of Results .................................................................. 168
8.5.5 Thermal Response in DSFC Tissue: Apparenth Method ...................... 169

9. Application of the GFP Intra—vital Viability Assay: Injury Following
Slow-Freezing to High Sub-Zero Temperatures ................................................... 177
9.1 Motivation ........................................................................................................ 177

9.2 Relationship Between Micro-vascular Injury and Thermal

History In Vivo ................................................................................................ 181
9.3 Pathophysiology of Cell and Vascular Injury Following Freezing .................. 188
9.4 Hypothesis and Experimental Design .............................................................. 190
9.5 Materials and Methods ..................................................................................... 191
9.5.1 Chamber Implant and Tumor Propagation ............................................. 191
9.5.2 Development of Freezing Protocol ......................................................... 192
9.5.3 Measurement of Thermal History During Intravital Studies ................... 192
9.5.4 GFP-based Intravital Microscopy ........................................................... 194
9.5.5 Histology ................................................................................................. 195
9.6 Results .............................................................................................................. 196
9.6.1 Cryosurgerical Protocol ....................................................................... 196
9.6.2 Thermal History Measured During Intravital Studies ......................... 199
9.6.3 Post-Freezing Vascular Changes .......................................................... 201
9.6.4 Time-Course of Injury Based on GFP Intravital Viability Assay ........ 202
9.6.5 Histology .............................................................................................. 205
9.7 Discussion ......................................................................................................... 209
9.8 Summary and Conclusions ............................................................................... 211
10. Summary, Conclusions, and Future Work ............................................................. 213
10.1 Summary ......................................................................................................... 213
10.2 Conclusions .................................................................................................... 218
10.3 Future Directions ............................................................................................ 221
Appendix A: DIFFCHAM Program ............................................................................ 226
Appendix B: Vc_Freeze Program ................................................................................ 230
Appendix C: Mechanical Drawing of AF A ................................................................ 235
Appendix D: Thermal Characterization Data .............................................................. 237
Appendix E: Neumann Program ................................................................................. 239
Appendix F: Cho and Sutherland Program .................................................................. 244
Bibliography ................................................................................................................ 250

Table 5.1
c3115. rape;

Table 6.1

Whilhin l

n ,.
‘ t9. Q .~ 1
.01. gm

Table 6.2

Itbk 6.3
Table '.1

Table 8.1

LIST OF TABLES

Table 5.1 Lp Estimates. (T) and (S) referred to trypsinized cells and scraped

cells, respectively ............................................................................................................ 63
Table 6.1 GFP fluorescence intensity as a function of exposure time for a select
population of plated GFP-transfected R3230 AC cells. Data represent averages of

four replicates ................................................................................................................. 85
Table 6.2 Cell fluorescence intensity as a function of exposure time to
cycloheximide. AF U, arbitrary fluorescent units ........................................................... 90
Table 6.3 Literature diffusivity values applicable to GP P diffusion in viva .......... 107
Table 7.1 Experimental Design .............................................................................. 1 19
Table 8.1 Properties and Conditions Employed in the Benchmark Solution ......... 160

ligurr 1.1

9&4- 4'.- got";
....: ..>.\..1.I

Figure 2.1
antral of
ififdig‘d c
If)? VW

rill-1“
5

112m 2.2
{It}; Sim;

m m; it“:

Fm 23
For mt}? P a:

'. l...

\.
‘wbfil- l am
a" -.
“-kX‘CE 11"

6:14.16“..-

llI'ure 2.4

WWII
l5.151116 II.
it! 71121” l

lure 4.]
r121”: 42

Flinn 43
WIN-:1 (cl
{9.13.1 duh.
I‘D-ll”: "

3“»...lln “ .‘
L

l 13!": 5.]

“4331. Ir» .
Ci . .. ”Kn”
‘ ~ ’

.K "mi‘ray.
Iii" "

It,” 52
5&5»

‘ I a W" . l
“\nzdl
I

LIST OF FIGURES

Figure 1.1 Location and size of ice crystals at different freezing rates. Images in
this dissertation are presented in color ........................................................................... 15

Figure 2.1 Tertiary structure of a GFP molecule: a tightly-packed ‘B-can’

comprised of B-sheets (green) which enclose an alpha helix (red) with the

appendaged chromophore protected in the center of the B-barrel.

(From Yang et al., 1996) ............................................................................................... 21

Figure 2.2 The GF P chromophore, highlighted in green is comprised a
cyclic tripeptide (Ser-dehydroTyr-Thr). (From Cody et al., 1993;
www.biochemtech.uni-halle.de/PPSZ/projects/jonda/chromoph.htm) ........................... 22

Figure 2.3 Effect of pH on fluorescence of thFP and some of its mutants.

For thFP and EGFP, excitation was at 488 nm and emission was collected from
490-650 nm. Maximum signal strength was normalized to 100. EGFP fluorescence

can be seen to decrease in intensity at a higher pH than thFP. (From Patterson

et al., 1997) ..... . ....................................................................... 25

 

Figure 2.4 Fluorescence excitation spectra of EGF P as a function of pH. From
the top: 9.0, 8.0, 11.0, 7.0, 6.5, 5.5, and 5.0. Inset: fluorescence emission intensity
at 510 run (I) fluorescence excitation intensity at 490 nm (0) and absorption at

488 nm (6). (From Haupts er al., 1998) ....................................................................... 26
Figure 4.1 Growth curve for the wild-type R3230AC cell line .............................. 37
Figure 4.2 Growth curve for the transfected R3230AC cell line ............................ 37

Figure 4.3 Histogram and probability density function based on n=327

individual cell diameter measurements from a single sampling of the cell population
acquired during routine cell passage. The probability density function assumed the
population was normally distributed ............................................................................. 39

Figure 5.1 Schematic of a microscope diffusion chamber. The semi-permeable
dialysis membrane allows solute in the bulk flow region to passively diffuse across

the membrane while containing the cells in the sample region. (From Sherban,

1987) .............................................................................................................................. 51

Figure 5.2 Schematic of the temperature regulating channels and bulk flow
experimental solutions channels in the microscope diffusion chamber ........................ 55

xi

Figure 5.3

331mm: rou‘
gem lilt’. 0i
fluted mug

oooooooooo

Figun 5.6
rubric}. 53:2:

“,9 p\;l1-‘l

ms 5. 169:

and ‘
.M‘fi‘m
1341‘ 3?: Cl
.E‘Iiuq
wit ." 3K

.3
4. .35
.-.u:-
.A '
u . ‘
JIM
_~\: .3
t h.
-'1

Figure 5.3 Apparatus set-up for acquiring diffusion chamber data at temperatures
other than room temperature. Two refrigeration baths are used; the first to cool the
outer jacket of the diffusion chamber, and the second to cool the solutions that are
flushed through the optical region of the diffusion chamber ......................................... 56

Figure 5.4 Boyle Van’t Hoff Plot. Errors bars represent standard error of
the mean - - - ............................... 6O

 

Figure 5.5 An example of a good fit of experimental data to the biophysical
model ........................................................................................................................... 61

Figure 5.6 An example of a poor fit of experimental data to the biophysical
model. Samples with volume histories characterized by a rising volume at the end
were excluded from the data set ................................................................................... 62

Figure 5.7 Volume histories of cells prepared for analysis by trypsin exposure
(triangles) and by mechanical scraping (squares) following exposure to hypertonic
solution at 22 °C ............................................................................................................ 63

Figure 5.8 Lp as a function of temperature. E14, is estimated to be 4.2 kcal/mol.
Error bars are not shown on the plot but are on the order of the mean ......................... 64

Figure 5.9 Volume histories of trypsinized cells (22 °C) that were used to

estimate parameters in DIFFCHAM are plotted together with simulated results from

an independent program implementing the DIF F CHAM parameter estimates. The
results are consistent, indicating that the DIF F CHAM estimation program is

providing accurate estimates .......................................................................................... 65

Figure 5.10 Predicted temperature dependence of normalized water volume as a
function of cooling rate for R3230AC cells. Cooling rates are indicated on each
curve. The dashed line represents the equilibrium response .......................................... 67

Figure 5.11 Predicted cytoplasmic undercooling of R3230AC cells. Cooling rates
are indicated on each curve .............................................................................................. 67

Figure 6.1 (a) Recovery of cells frozen at 1 °C/min to various minimum
temperatures, as assessed by GFP fluorescence, GFP Fluorescence with EtBr

co-stain, and Trypan Blue Exclusion. (b) Correlation of recovery assessed by GP P
fluorescence with recovery assessed by EtBr. (0) Correlation of recovery assessed

by GFP fluorescence with recovery assessed by Trypan Blue. Data represent

average of three replicates +/- one standard error of the mean ...................................... 74

Figure 6.2 Fluorescent images of GFP-transfected R3230 AC cells after freezing
to various minimum temperatures and thawing in a 37°C water bath: (a) unfrozen
control (b) —10°C (0) —20°C (d) ~30°C ........................................................................ 75

xii

figurt 6.3 ‘
mart {Cll‘
nerd b} 1‘7}

ligun 6.4

4,. "'
6.611361: -.

figure 65

cells. act: held
Indians. 11
I In: ten gt"
gar: plane cur

:Icroscopx' .....

Figure 6.6

minis. Int.
irritated mt
:t'fait-d git 1:

\all' ._- l. .

“sun 6.7
23610 the p‘
r. 3-.»1Ij.\l Ir-
ltzurt 6.8
:L‘KEML‘d v.
11331 [1‘ I 5 I

“inn 6.9
3133!} I'Jl’x

Flint: 6.10
F31 c:;1}.l l]

“-r— .
\F‘tnt ‘

Hm. 6.11

I“ “
.1133”? 4‘ 1

\ l

Figure 6.3 Correlation of post-thaw recovery of GFP-transfected and wild-type
R3230AC cells frozen at 1 °C/min to various minimum temperatures. Recovery was
assessed by Trypan Blue exclusion ................................................................................ 76

Figure 6.4 Viability of GFP-transfected R3230AC cells following 30 5 exposure
to 1.6M and 2.7M NaCl ................................................................................................. 79

Figure 6.5 Images of cells in culture following different conditions. (a) control

cells were held at room temperature for 30 minutes then cultured under normal
conditions. (b) test cells were frozen in liquid nitrogen followed by passive thawing

at room temperature for three cycles prior to normal culture. The left image of each
pair: phase contrast microscopy, right image: epi-illumination fluorescence

microscopy ...................................................................................................................... 81

Figure 6.6 Plated R3230AC cells were incubated in cycloheximide, a protein
synthesis inhibitor and assayed at several time points in a 24-h period. Control cells
(incubated without inhibitor) appear normal whereas the cells in the cycloheximide-
incubated group were fewer in number and appeared slightly swollen with indistinct

cell boundaries (potentially due to leaky membranes) at the 24-h point ........................ 88

Figure 6.7 Viability assessed with ethidium bromide after different exposure
times to the protein synthesis inhibitor cycloheximide. Control cell were incubated
in D-MEM media with no cyclohexirnide present. (* t-test, p<0.0005) ....................... 89

Figure 6.8 Calibration curve obtained with InSpeck Green calibration beads
illuminated with blue light (510-560 BP filter), and detected with a Sensicam
camera, 0.1s exposure time ............................................................................................ 94

Figure 6.9 Calibration curve obtained with Sodium F luoroscein solutions in
capillary tubes, path length 0.1 mm, 0.1 8 exposure time ............................................. 95

Figure 6.10 Calibration curve obtained with rEGFP solutions in capillary tubes of
path length 0.02 m using a 0.4 3 exposure time. Error bars are smaller than the
data point markers ......................................................................................................... 97

Figure 6.11 Decrease in intensity measured following multiple freeze-thaw (FT)
cycles of 4x10'2 mM rEGF P solution. Control samples were maintained in a light
impermeable container at room temperature and sampled at the same time as the

treated solutions. Error bars represent the coefficient of variation based on the

standard error of n=9 bitmaps. *Significant difference against control (p<0.01,
one-sided t-test) .............................................................................................................. 99

Figure 6.12 Histogram of 12-bit gray scale intensity levels measured in a
population of R3230 AC cells ....................................................................................... 102

xiii

do" ‘

Firm 6.
haze-n:
'11: $5an
:73: law
'3: 3.3.“.
thurt ".
\L'l Dex

“Sure 7.

‘Cl 1.), ~.
‘ H\rl‘\| )1

figure '1;-
Sim ind

Ethan:

“are I.

s. . '
.m _
~'-...'th:1

L5 31'.“ $13}

“9111?"

if". ‘

5......4‘. \IL
.' u.
r\.jllng “
l a, .
“N: 11'“ .
- H.“

“We 7,1

figure “.1
if“ his.

a“
45.”
. .

n1"": 3.

Figure 6.13 Curve-fitting was used to estimate a diffusivity value on the basis of
molecular weight. Data for plot was obtained from Lehninger, 1976 .......................... 103

Figure 6.14 Schematic to illustrate the characteristic radius required for
determination of the time constant for diffusion of GFP ............................................... 105

Figure 6.15 Photomicrographs demonstrating conservation of GFP fluorescence
following lysis of GFP-R3230AC cells: (a) before freezing (b) post-lysis .................... 109

Figure 6.16 Change in GFP fluorescent intensity of background media following
freeze-thaw lysis of suspended cells. Arbitrary fluorescent units (AF U) represent

the measured gray scale intensity normalized to the initial data point. All y-axis

error bars represent the standard error of the mean. The x-axis error bars represent

the uncertainty in measuring volume fraction ................................................................ l l 1

Figure 7.1 Sketch of Implanted Dorsal Skin Flap Chamber, Courtesy of Dr.
Mark Dewhirst, Duke University ................................................................................... l 14

Figure 7.2 Recently implanted dorsal skin flap chamber in an anesthatized
Fischer 344 rat - - - - ........ 116

Figure 7.3 Series of images depicting growth of a solid tumor from a tumor
slurry in the dorsal skin flap chamber: fresh tumor preparation (a), the tumor at day
3 (b), and day 6 (c) ................................................. 116

 

Figure 7.4 H & E section of the tissue contained in the dorsal skin flap
chamber at day 6 afier implant of chamber and tumor slurry. The vessel pictured
is the same vessel seen in the Figure 7.3 series .............................................................. 1 17

Figure 7.5 Illustration of the two different biopsy techniques used in the first

animal study. (a) Square tissue section was first removed from the skin flap. Tissue
sections were cut (b) in-the plane of the tissue (x-y) and also (c) cross-sections
wereobtainedatx=0andx=2.5mm ....... 119

Figure 7.6 Depiction of tissue that was overgrown at the time of treatment ........... 120

Figure 7.7 Cryosurgical probe system. (a) Linkam LNP9311 Cryopump with
liquid nitrogen dewar and probe. (b) Close-up of 6 mm diameter cryosurgical probe

tip ................ - - -_ .............................................................. 121

 

Figure 7.8 Phantom set-up for acquiring thermal data ............................................. 122

Figure 7.9 Location of thermocouples and cryosurgical probe during the
phantom measurements .................................................................................................. 123

xiv

tic pix-III

112m 7.
511.13 or. (
reset: inc

- o
3515 38m:

Figurt 7.?
firmer
I? mi 61

4.)” p. 1"
'Je Mikjds

52m 7.1

M" o—H

‘ A .i
"H“ . 6M L~

a D...“
m. “H,“

Cc.
'he. I"v~
IA AL“

I

Figure 7.10 Characteristic freezing curves measured at specific locations in
the phantom tissue system .............................................................................................. 124

Figure 7.11 Normalized gray scale intensity of GFP fluorescing tumor as a

function of time for cryosurgically treated animals (CSl-CSS). The rectangular

region indicates the average upper and lower autofluorescence levels of untreated

and treated normal tissue, respectively ........................................................................... 127

Figure 7.12 Image of the glass window and e-clip (a) and tiled composite

fluorescent images of the entire window chamber viewing field: before cryosurgery

(b) and 6 hours after the last fieeze-thaw cycle (c). The cryo-lesion is clearly

demarcated in (c) ............................................................................................................ 127

Figure 7.13 (a) GFP-fluorescing tumor before treatment. (b) GFP-fluorescing
tumor after treatment. (c) histpathology micrograph from the center of the tissue in
(b). Damage is complete, with only cell debris and necrotic cells (arrows) present ..... 129

Figure 7.14 (a) GFP-fluorescing tumor before treatment. (b) GFP-fluorescing

tumor afier treatment. (0) Histpathology micrograph from the center of the tissue in

(b). Damage is partial with some necrotic cells (arrows) present amidst a group of
normal tumor cells .......................................................................................................... 130

Figure 7.15 (a) GFP-fluorescing tumor before treatment. (b) GFP-fluorescing
tumor after treatment. (c) histpathology micrograph from sample at the center of
the tissue in (b). No damage is observed ...................................................................... 131

Figure 7.16 H&E micrographs of cryo-treated tumor tissue: The tissue

surrounding the lesion (a) is undamaged. At the margin of the lesion (b) necrotic

cells are evident (arrows). In the central region (c) nuclear and cytoplasmic

remnants are evident--.- - - ..... - .................................................... 133

 

Figure 7.17 H&E micrographs of acute inflammatory cells (neutrophils) located
near the edge of the cryo-lesion (arrows) 134

 

Figure 7.18 H&E micrograph of neutrophils phagocytosing red blood cells
(arrows) .......................................................................................................................... 134

Figure 7.19 Spatial comparison of GP P intensity changes (approximate lesion
diameter ~7.5 mm based on magnitude of intensity change) and histopathology

results (approximate lesion diameter 6.5 mm based on region of hemorrhage) for

one case of tumor cryosurgery treatment ....................................................................... 137

XV

Figurc 8.?
1116 stain}:
putts. Ag
DSFC. Ib

mil-"rant
‘ b

Figure 8.1
1.1311. ll]

‘ T! ‘.T
1.91.41.

5

its:
and rant;

figure 8.3

Figure 8.4

. 1.
In“ \hH-JL .\

\
0.1
m ‘
Li‘s. -l|4\‘\u:

Figure 8.5
res-Ia w;
and l. the i

01 the wild

FlSure 8.6

Figure 8.1 Annular Freeze Apparatus (AF A). (a) Cold N2 vapor passes through

the stainless steel freezing apparatus by means of hypodermic needle inlet and outlet
parts. A groove on the surface of the annuli provides the point of attachment to the

DSF C. (b) The symmetrical annular probes are attached to the exterior surface of the
protruding annuli of the DSFC ....................................................................................... 147

Figure 8.2 Characteristic data obtained with the Annular Freezing Apparatus

(AP A). Three thermocouples were attached to each side of the DSF C (per the

insert diagram) prior to assembly. The AF A was attached and cooling was initiated

and maintained for 20 minutes ....................................................................................... 149

Figure 8.3 Demonstration of thermal control possible with the AF A system ......... 150

Figure 8.4 A DSFC glass window implemented with machined grooves of 400
um diameter. The red and blue lines overlaid on the image illustrate how
thermocouples can be mounted on this window ............................................................ 152

Figure 8.5 Schematic illustrating solidification of a homogeneous semi-infinite
material with a phase change region. Ts represents the surface temperature imposed

and T0 the initial temperature. The mushy region (m) is identified by the positions

of the solidus and liquidus phase change boundaries .................................................... 156

Figure 8.6 Functional relationship of apparent Cp with temperature ....................... 161
Figure 8.7 Schematic illustrating solidification of a homogeneous semi-infinite
material initially at T0 with a single-phase change temperature, Ti, and a temperature
boundary condition T=Ts at x=0. The position of the phase change boundary is

given by X(t) .................................................................................................................. 163

Figure 8.8 Predicted thermal profiles using three different solutions for
solidification in a rectilinear semi-infinite geometry ..................................................... 169

Figure 8.9 Schematic illustrating the geometry used to represent a ‘/4 section of
tissue in the DSFC. The point of contact of the DSFC annulus, indicated with the

dark dotted line, was modeled as a constant heat flux boundary condition ................... 170
Figure 8.10 Mesh used to generate solidification solution ........................................ 171

Figure 8.11 Temperature profiles generated for phantom tissue (properties of

water used) in the tissue geometry depicted in Figure 8.9 ............................................. 172

Figure 8.12 Temperature distribution in DSFC tissue (properties of water used)

as simulated at 240 s ....................................................................................................... 175
xvi

Figurr 8.
{63218111

tr" '1'
55...:t. an

finn8
DSFC I1.

ligurr 9.
Emmi
ward

31333.3(

.'

5
.'._ ‘
mc Jakflk“
WNW

‘ 1
N u.
6;!
\q\ ar‘
r .I“
‘ .
W ‘ I.‘
Win! 1r

1.:‘3- .»
3 Of \4

Figure 8.13 Temperature difference between the top surface of the chamber
geometry (glass and annulus regions) and the tissue midline as a function of

temperature ..................................................................................................................... 1 76

Figure 8.14 Temperature difference between the top and bottom surfaces of the
DSFC glass window as a function of time and radial position ...................................... 176

Figure 9.1 Conceptualization of a cryosurgically treated subcutaneous lesion.
Regions close to the probe will experience rapid cooling and reach temperatures

close to the temperature of the probe; tissue further away from the probe will

experience slow cooling rates and will have temperatures closer to the phase

change temperature of water ........................................................................................... 178

Figure 9.2 Relationship between freezing rate and survival for various cell types.
From Mazur, 1970 ......................................................................................................... 182

Figure 9.3 Illustration of ischemic injury caused by damage to blood vessels ........ 183

Figure 9.4 (a) Photomicrograph demonstrating the location and positioning of

a single thermocouple for in viva temperature sensing. (b) Schematic of

thermocouple positioning for the purpose of minimizing fin effects of in viva
temperature measurements .............................................................................................. 194

Figure 9.5 Characteristic freezing curves measured in viva at specific locations....197

Figure 9.6 Experimental measurements for the in viva case presented in Figure
9.5 are plotted against predicted measurements (properties of water used) for
two locations on the surface of the tissue ....................................................................... 199

Figure 9.7 Characteristic thermal history measured using a T-type thermocouple
in contact with the tissue at the center of the annulus in the DSFC during slow
freezing ........................................................................................................................... 200

Figure 9.8 Individual vessel measurements from treated animals, normalized

to the starting diameter of each vessel, are compared to a baseline of pooled sham
animal data For the sham animals the error bars represent the standard deviation of
three animals. Remaining measurements represent the average and standard

deviation of five repeated measurements of the same vessel ........................................ 202

Figure 9.9 Micrograph series of GFP intensity change over time in a centrally
located DSFC tumor treated with slow freezing to —I 8 °C ........................................... 204

Figure 9.10 Normalized gray scale intensity plots of GFP fluorescence intensity
loss of cells following slow freezing to —1 8 °C ............................................................. 205

xvii

figure 9.l l
ram. sh.
Trim ctr

'r
l‘i films.

11M lli‘It'i‘il‘

figure 9.12
I) 48 ’C. 1;
sale; earrou
semis. 1b
5:33:23) her:

Figure 9.11 Micrographs of sham-treated tissue fiom the dorsal skin flap

chamber, showing no evidence of necrosis, and blood vessels of normal patency.

(a) Transverse (*) and longitudinal (**) sections of blood vessels can be seen.

(b) Fibroblast nuclei (arrows) are normal in appearance ............................................... 207

Figure 9.12 Micrographs of dorsal skin flap chamber tissue frozen at ~ 3 °C/min

to —l 8 °C. (a) Congestion and engorgement of blood vessels (*) and condensed

nuclei (arrows) were observed, indicating both microvascular injury and tissue

necrosis. (b) Extravasated red blood cells can be seen to dominate this

severely hemorrhaged section ........................................................................................ 208

xviii

3W0. .‘

Ve. ' ‘9-
"H‘i \
y

I)... .
is :‘KIV: ..I
“It 4
< l“‘u

'1
did“ -....'
‘§.\‘ If)?
~u

‘r’

P

I»

“r.".
\

“k mate’-
. H

Chapter 1

Introduction

1.1 Motivation for the Research

Low temperature injury to bio-materials has wide-reaching significance in arenas
as vast as maintaining ecological diversity to preventing death by cancer. Although
injuries of this nature have been studied for some decades, there remains considerable
ambiguity as to the mechanisms by which these injuries occur (Rubinsky, 1991). As the
temperature of a biological material is lowered, the effect can be preserving or
deleterious. These conflicting mam'festations are the product of an intricate interaction
between thermal, chemical, electrical, and mechanical forces (McGrath, 1993).
Depending on the degree of cooling, the final end temperature reached, and rate at which
the material is cooled, stored, and warmed, different injury or preservation mechanisms
dominate.

The primary goal in a cryosurgical process is to use localized freezing to achieve
maximal tumor destruction, without injury to healthy tissue. The current state of the art

limits the optimization of existing cryosurgical devices and processes as well as the

1.1.2.1:

0151". at

"-3..

{Mil-u
‘ O
a -'0_ PK

Lip...

HIT. L11

1“"
I r?

L“J

development of new devices and procedures that could extend the use of cryosurgery to
other applications. The limiting factor is that the mechanisms by which the freezing and
thawing kill the cancerous tissue are not well understood (Bischof et al., 1993, 1997). A
fundamental understanding of the response of tissue to freezing and thawing is lacking
whether the tissue is normal or neoplastic. To develop a comprehensive understanding of
the damage inflicted by cryosurgery, it is necessary to understand the direct cellular
injury caused by the decreased temperature and the formation of ice, and also the
contribution of secondary effects such as cell-mediated inflammatory response, host
immune response, and vascular damage to the overall injury. The mechanisms of injury
are not well understood primarily because the tools to easily delineate direct kill and
ischemia and inflammatory effects do not exist. As well, understanding the time course
of damage is difficult and requires the sacrifice of many animals. This lack of
understanding means that sub-optimal results are obtained from procedures designed to
destroy tumors by freezing (cryosurgery) or, conversely, the development of protocols
designed to preserve tissues and organs for storage. Development of a non-toxic viability
assay that can provide information about the vital state of a tissue over time, without
sacrifice of the animal, can significantly improve experimental efficiency as well as
provide continuous viability information that can begin to uncouple direct injury from
secondary effects.

Understanding the thermal history that gives rise to the freeze injury is also a
fundamental aspect of what will hereafier be generally referred to as ‘the cryosurgery
problem’. Quantifying the thermal history eliciting the particular tissue response is an

important and non-trivial task. For example, Bischof et al., 1997, predicted that the

e‘Ui“
vi:
ELM“!

l

LY
It
I
., ’
J

t.
Bu:_‘.,l‘)‘

,bP'
Ll..\re

§

.M .
1!“ , 1

n .1-.-

"x-n.

LP; V
“Va-421K)!
i

.‘b ’
"“‘el.".“

'!.-l |)'
“-1. \ .
‘ h. \

Teller»

.u
3-.
.‘_
ll ‘ "v-
"p .6
're
it. 9'»
If".
."
‘1" I...
\‘\
9.
3““; .
P.
in.”
.'.
q
5' — T.

cooling rates in biological tissue treated with a 3-mm liquid nitrogen-cooled probe would
range from >1000 °C/min to 5 °C/min, with temperatures ranging from —196° C at the
surface of the cryosurgical probe to ~0 0C at the edge of the ice ball. These extremes of
temperature and fieezing rates are manifested in an ice ball of order 1 cm. To understand
the spatial variation in cell death that occurs in cryosurgically-treated tissue, the thermal
history must be understood throughout the cryo-lesion. Although discrete thermocouples
can be used to record local temperatures, the accuracy of the measurements are somewhat
questionable due to inherent difficulties in acquiring measurements in steep spatial
gradients of temperature that are rapidly changing in time. Thermal field information can
only be obtained by interpolating between these discrete measurements and/or theoretical
modeling. Not only is this temperature field estimation challenging due to tissue
inhomogeneities and a lack of reliable thermal property data, but the warming effect of
circulating blood, as well as metabolic heat generation must be considered when freezing
a biomaterial in situ. This results in complicated governing equations, limiting most
mathematical models of biological freezing to numerical solutions. Optimizing the
measurement of temperature in biological materials and/or developing suitable
mathematical models are thus important complementary aspects of creating technologies
to improve the understanding of thermal history-viability functional relationships.

The cryosurgery problem presented here represents a sub-class of bio-heat
problems in which the primary objectives are to quantify a transient temperature field in a
biological material and understand the associated biophysical response. Thus, developing
technologies to address challenges of this nature in the cryosurgery problem will have

positive implications for other bio-heat transfer problems as well.

1

O
«a,
VAL.“

Ci.

1.2 Infra-vital Microscopy

Just as the march of engineering technology of recent decades has been towards
smaller, better, and faster, so has medical treatment and sensing progressed from gross
anatomical assessment to the scale of sub-cellular structures. Disease is treated at the
genetic level and molecular imaging is providing us with unprecedented information
about the molecules that comprise our tissues and organs. The bulk of biomedical
research addresses in some way the impaired transport of molecules or loss of cellular
control. Consequently, engineering tools developed to study disease are geared towards
micro-scale cellular sensing, imaging, or manipulation.

Intra-vital preparations are becoming exceedingly popular as a tool to understand
cellular injury, dysfunction, and repair. The complex interplay of inflammation and direct
cell injury following ischemia-reperfusion has been studied using intravital microscopy in
rodent models for direct visualization of cellular and tissue response by way of neutrophil
infiltration and vessel leakage of fluorescently-labeled macromolecules (Schuder et al.,
1999). Trans-illuminated tissue model systems such as the dorsal skin flap chamber
(DSFC), hamster cheek pouch, and rabbit ear preparations have been used effectively for
many types of studies. Such systems provide direct, in vivo optical access to tissue and
vasculature. These intra-vital approaches have been utilized to understand the
extravasation of chemotherapy drugs (Gaber et al., 1996), micro-vascular permeability
(Yuan et al., 1993), angiogenesis (Li et al., 2000), and vascular injury following burn
injury (Aggarwal et al., 1989), laser treatment (Gourgouliatos et al., 1991), cryosurgery
(Hoffman and Bischof, 2001ab), and radiation (Wu et al., 1994). However, the single

feature missing from all of these studies was the ability to monitor cell death dynamically

' l
.19me
bllni‘l

III .4.

1" c.

“1...:

‘V -o
\t \ 7“”
~ 3‘51.

\ i

m...
1‘; _‘\‘:.~
in “I

n“

‘V‘L, I

u. ; ~11
anl

' a

l H.“ o
0“: I‘

‘gll

\\

in the same animal. Independent of the observation period for the phenomena under
investigation, cell death could only be established at a single end point in a given animal,
either by delivery of a vital dye and subsequent termination, or by termination and biopsy
for standard pathological analysis.

Because of the preponderance of intra-vital preparations and techniques currently
being utilized to study a range of tissue injury phenomena, and also because of the
immediate impact that a new intra-vital continuous viability assay could have in these

systems, the in viva model used in this work will implement the dorsal skin flap chamber.

1.3 Viability Assays

Evaluation of the viability of individual cells, as well as cells within tissues and
organs after freezing is of fundamental importance for cryopreservation and cryosurgery
research. Although cryopreservation seeks to sustain cell and tissue viability and
conversely, cryosurgery seeks to devitalize neoplastic tissue, both rely on some
measurable firnction of viability to determine protocol success. Measuring viability is a
difficult task whether single cells or tissues are under consideration, partly because
viability is often poorly defined and can be associated with many different aspects of the
cell or tissue (Pegg, 1989). These aspects include morphology, cell attachment, motility,
chemotaxis, fertilization capacity, membrane integrity, enzyme activity, protein
synthesis, metabolism, reproduction, etc. Although there is no universally agreed upon
definition for viability that suits all contexts, and perhaps none possible, the convention
adopted by Pegg (1989) is embraced in this work. That is, viability can be considered

“the ability of a heated sample to exhibit a specific function or ftmctions, expressed as a

MW”
0 U . er ‘ .‘
1,]:an “n
.‘ i '|
m me CC“.
:31 mcmi‘i
11 it ‘ J '
lamlicu nu;
mr'k’ {hill P

lnt

m: arr
8:111] tissui
Imam
’hr Celia};

‘ l

u? .ntior

:2.“
“‘5
31.731015 3‘
imitation
Isicnsd in
i
.S a dth
4. .
was. kirk
It.
i. - - ~
115.". CD111!
lb

Yam

5 l ('7’.
3.113;}

proportion of the same function exhibited by the same sample before treatment, or an
identical fresh untreated sample”. There exist a number of assays that evaluate functions
of the cell, such as metabolic activity (e.g., MTT, Alarnar Blue), and others that indicate
cell membrane integrity (e.g., Ethidium Bromide, Trypan Blue). However, there are a
limited number of assays available for determining tissue viability, especially in situ, and
none that provide continuous dynamic measurements.

In the case of single cells it is relatively easy to add substrates such as fluorescent
dyes and these ofien correlate well with other accepted measures of viability. In contrast,
targeting dyes and substrates for tissue viability applications is difficult. A major problem
is that application of the dye or substrate is often diffusion-limited even for seemingly
small tissue sizes. The current gold standard for evaluating tissue injury is
histopathology. This biopsy and stain approach provides very detailed information about
the cellular characteristics of the injury (vacuolization, membrane disruption, nuclear
coagulation, vascular congestion, etc.) but the information is static, and represents the
injurious state of the tissue only at the time of biopsy. Moreover, there is a significant
incubation time following injury before the damage progresses to an extent that it can be
detected in histopathology specimens using standard light microscopy evaluation. Injury
is a dynamic phenomenon occurring immediately after the insult, and progressing for
days, with tissue remodeling continuing for weeks and months.

To understand the dynamics of this tissue injury, as well as when other
interventions can accelerate or ameliorate its progression, the sacrifice of many animals is
necessary. Because of variation from animal to animal, the number of animals included

per group needs to be large to ensure statistical significance. To provide comprehensive

utilimit:
tit pheru

perhaps -

information about injury progression requires experimental groups that span the range of
the phenomenon in small increments of time. This can translate into hundreds, and
perhaps even thousands of animals to effectively understand the development of the
injury. The consequence is either interpolation between discrete static measures or else
exorbitant research costs. The latter case is especially so with the current trends toward
utilization of ‘knock-out’ or transgenic animal models, with a cost per animal as much as
5-50 times that of the wild-type. The histological processing of tissues may also
introduce artifacts and it is often difficult to correlate the processed tissue in space and
time with other phenomena of interest that were applied to or occurred within the tissue.

Intra-vital dyes used in conjunction with microscopy, without fixation, are now
the state of the art for dynamic viability imaging (Schuder et al., 1999; Wanner et al.,
1999). However, the techniques for dye perfusion and observation are invasive and the
dyes can be toxic, consequently requiring termination of the host before recovery from
anesthesia. This restricts the duration of dynamic evaluation to experiments that can be
completed within a single anesthesia period. These in situ assays do not permit dynamic
and long-term evaluation of viability.

To date, most of the studies attempting to address the morphological, biophysical,
and thermal response of mammalian tissues to freezing have been done using tissue slices
or cell monolayers and suspensions. It is well known however, that models based on
parameters derived from these in vitro studies under-predict the extent of damage that
occurs in an actual cryosurgery (Bischof, 1997). It has been suggested that this
differential is due to a vascular mechanism of injury, yet the nature of the injury has not

been well-characterized due to a lack of adequate in viva assays. Conventional

:7.
LL.

116C ;

’Q14.;
‘fH-nj

kii‘ih r
' \
run

I

.6; | .
l .4 ‘
M‘.'k‘

fluorophores and chromophores are inserted into cells using microporation techniques or
via passive transport by incubation of cells in culture containing the probe. Although,
tissues can be suffused with assay solution, often the result is toxic to the tissue, and there
is difficulty with close-packed cells (Muldrew, 1994). Moreover it becomes difficult to
distinguish tumor and healthy tissue margins. A different class of fluorescent
biomarkers, called Green Fluorescent Proteins (GFP) may be able to furnish this very
necessary tissue assay.

The assay proposed in this work does not require staining or fixation of the tissue
to indicate viability, but relies on the expression and retention of an intracellular protein,
GFP, produced by a transfected cell. Because the fluorescent compound is produced
intracellularly, there is no requirement for electroporation techniques or diffusion
(passive or active) to insert the probe into the cell. Moreover, no exogenous substrate is
required to produce the fluorescence. The cells can be assayed without altering the
system in any way, other than illuminating with light of the proper excitation wavelength.
This allows continuous, real-time, non-destructive evaluation of GP P fluorescence when
optical access is possible. These features may provide the basis for a powerful tissue
viability assay as well as an assay for isolated cells. Because the introduction of dyes is
not required, limitations associated with diffusion of dyes are superceded. Moreover
biopsy is not necessary to assess viability. The GF P fluorescence may be assessed in
real-time, in a living host, without sacrificing the animal, allowing the study of host-
mediated effects on the protocol of interest. This would also permit chronic and long-

term studies on the same animal or system.

be rel-cur
treated
nesting

pitiit‘lil.

1.4 S

.....

":w...
Kd\._

Li!
I p.

(1‘

Although GFP has been shown to be non-toxic to cells, toxicity to an animal may
be relevant in a system where GFP can be released from cells, as in the case of the
proposed system. The fact that GF P mice and GFP bunnies have been produced by
inserting GFP into embryonic cells suggests that organisms can be very tolerant of this

protein. This aspect of toxicity will be discussed in further detail in Chapter 4.

1.4 State of the Art in Cryosurgical Research

Because the proposed research has been motivated by a class of bio-heat transfer
problems that are large in scope, to provide comprehensive validation for the new intra—
vital imaging techniques and yet ensuring a tractable problem, the cryosurgery problem
was chosen. The injury phenomena associated with cryosurgery are complex and occur
on several time scales and there are ambiguities about the nature of injury in certain
regions of the cryo-treated tissue. These features will serve to illustrate the power of the
technology developed in this work. It is believed that a GFP-based intra-vital viability
assay can provide new information about the cryosurgery problem that has not previously
been possible. Because the GF P intra-vital technology will be developed and
characterized in the context of low temperature injury with Specific attention to the
thermal conditions of relevance to cryosurgery, a review of the state-of- the-art in

cryosurgery research is warranted.

1.4.1 Historical Perspective
The use of low temperatures to treat pathological conditions dates back to the

1850’s when iced saline solutions (-18 to -20 °C) were used to irrigate cancers of the

breast and
mm of
liqueur-1 '

m“ [it\\

I
. .
-hl. TC"

'
.‘N'f‘i l":
I

breast and cervix, primarily for palliative purposes (Gage, 1992; Bird, 1949). With the
advent of liquefied gases came the ability to apply much colder temperatures (e. g.
liquefied N2, -196 °C). Thus, freezing at faster rates and to much lower temperatures was
now possible in the tissue. As a direct result of this innovation, successful cryosurgeries
of topical cancers, primarily dermal and opthamalgic, were reported (Gage, 1992, 1998;
White, 1899). In the mid-80’s breast cancer cryolumpectomies were performed (Rand et
al., 1985, 1987) wherein the tumorous mass was first fiozen through an open wound and
then removed as in a normal resection procedure. Later, with advances in imaging
technology (Hong et al., 1994; Gilbert et al., 1997), cryosurgery moved deeper inside the
body, in the treatment of visceral disease. The ability to monitor location of the freezing
probe and the position of the advancing ice front permitted much greater control and
resulted in much less invasive procedures.

Cryosurgery has been successfully performed on a diversity of organs including
the brain, liver, bronchus, prostate, pancreas, and oral and nasal cavities (Gage, 1992;
Baust et al., 1997). Moreover, liver and prostate cryosurgeries are becoming routine in
clinical settings (Baust et al., 1997). The ability to perform cryosurgeries through stab
incisions has allowed the cryosurgical treatment of awkwardly seated tumors and lesions
not possible to treat using an open wound conventional approach. Because the
undesirable tissue is not removed from the body, post- surgical results are cosmetically
pleasing with less tissue loss than surgical excision. In addition to being a primary care
alternative, cryosurgery has also been shown to be beneficial in the management of
recurrent carcinomas following standard excisions. Cure of metastases has also been

reported in rare cases.

10

Although cryosurgery has been shown to be an effective means of treating various
cancers (Gage, 1992; Baust et al., 1997), there are still many ambiguities about the nature
of the freeze injury that occurs and the best approach to maximizing the damage accruing
in a single freeze-thaw cycle (Gage and Baust, 1998). To improve the state-of-the art in
cryosurgical research, there is a need for an improved understanding of the response of
tumor tissue to freezing and thawing in viva. Studies on biopsied tissues have
contributed significantly to our understanding of the biophysical response of healthy and
tumor tissue to freezing (Rubinsky and Ikeda, 1985; Bischof et al, 1993). However, in
vitro cell and tissue studies cannot account for cell-mediated immunological responses or
the presence of circulating blood. Both of these can significantly impact cryosurgical
outcome (Gage and Baust, 1998). Additionally, information from post-operative tissue
biopsy is constrained to the time it was obtained and therefore offers only a static
interpretation of tissue response to cryosurgery. However, evidence suggests that cell
death following cryosurgery is not a discrete phenomenon, but that ischemic necrosis,
occurring on a longer time scale than direct injury, contributes to the overall lesion size
(Hoffman et al., 1999, 2001b). This not only complicates the assessment of cryosurgical
success, but also makes it difficult for the surgeon to delineate suitable margins for
debridement of dead tissue following frostbite injury (look at al., 1998). Understanding
the dynamics of tissue degeneration following freezing injury thus has important clinical
ramifications.

It is known that in tissues such as the liver (Bischof et al., 1993) and prostate
(Bischof et al., 1997) regions near the cryosurgical probe experience fast freezing rates

and are associated with intracellular ice formation. Peripheral regions of the tumor

11

experience slower rates and tissue death in this region is typically related to solution
effect or solutes injury. A vascular mechanism of injury has also been suggested as
accounting for some of the necrosis at the edge of the cryo-lesion (Bischof et al., 1997).
The precise conditions that give rise to these types of damage are not well-understood
(Rubinsky and Onik, 1991; McGrath, 1993; Bischof et al., 1997) and they vary from
tissue to tissue (Shepherd and Dawber, 1984). More complete knowledge of thermal
damage in any system affords a greater sense of control and can result in more effective

tumor de-vitalization and less damage to healthy tissues.

1.4.2 Direct Cellular Injury

Considerable knowledge has been obtained from freezing studies on isolated cells
(Mazur, 1977, 1984; McGrath et al., 1975; Smith et a1. 1999). At fast rates of freezing,
heat transfer dominates. Water is trapped within cells and thermodynamic equilibrium is
attained by intracellular ice formation (HF), which is generally lethal (Mazur, 1984).
This type of damage is considered to be primarily due to mechanical forces on
membranes and organelles. At slow rates of freezing mass transfer dominates, the cell
becomes severely dehydrated and the cells are exposed to high solute concentrations for
extended periods at relatively high subzero temperatures. This can cause both chemical
and mechanical membrane stress (McGrath, 1993). Mazur (1984) has referred to this
phenomenon as the two-factor hypothesis; viability as a function of cooling rate is
described as an inverted-U-shape, representing a fairly universal trend to low viability at

the extremes of fieezing rates.

12

lo
the forma‘
and P‘gg.
M01116 Cl
cell mcml
the cell 10
31? €1.00:
llLL'TIt’It‘Lls

85.”: in

“(K -
1.4.: 11\ i
I

8" .efiv I
ylkrudel£

For slow cooling rates, the immediate effect of temperatures lower than 0 °C is
the formation of ice crystals in the extracellular space and microvasculature (Rubinsky
and Pegg, 1988). As liquid water is converted to ice, solutes in the extracellular fluid
become concentrated, causing a decrease in chemical potential of water external to the
cell membrane. As the concentration of solutes steadily increases, water begins to leave
the cell to osmotically equilibrate this imbalance. During this slow freezing process cells
are exposed to progressively damaging hyperosmotic conditions. There have been
numerous observations made and mechanisms proposed to explain the nature of solutes
effect injury. Reduction of the cell volume past a critical level during slow
cooling/dehydration has been thought to be associated with a significant drop in cell
viability (Meryman, 1970). It has also been proposed that damage to or alteration of the
plasma-lemma membrane may be related to solutes injury (Gordon-Kamm and
Steponkus, 1984). Steponkus et al. (1983) indicated that an irreversible loss of
membrane material due to excessive shrinkage could result in irreversible and destructive
changes to the cell. Additionally precipitation of buffering salts can cause damaging
changes in pH (McGrath, 1993).

At fast freezing rates, the rate at which water can leave the cell cannot adequately
keep pace with the departure from osmotic equilibrium and intracellular freezing occurs
to offset the imbalance in chemical potential across the membrane (Toscano et al. , 1975).
Experimentally, this formation of intracellular ice has been described as a sudden
darkening of the cell, or “flashing”. The precise nature of how IIF relates to injury is
somewhat controversial. There is disagreement as to whether the force generated by

rapid water flux in and of itself causes lethal membrane damage and then this

13

"8
Li

H161.

permeabilization of the membrane allows intracellular ice to subsequently form
(Muldrew et al., 1990, 1994), or if the intracellular ice is indeed precipitating the cell
damage (Mazur, 1984). This can be generalized as a cause versus consequence debate.
Steponkus et al. (1983) have provided substantial evidence for the membrane rupture
hypothesis, suggesting a critical temperature at which the membrane fails. Mazur (1966)
has theorized that at a certain critical temperature the radius of ice crystals will match the
radius of aqueous pores in the intact membrane and intracellular ice will be subsequently
seeded by extra-cellular ice. During warming, re-crystallization or fusion of small non-
lethal ice crystals to form large crystals can be quite destructive (Mazur, 1977). The
more slowly these are warmed, the greater the likelihood that this type of re-
crystallization will occur. Although recognized as being an important component of the
injury, few studies investigate the drawing process with the level of detail given to the
fi'eezing rates and minimum temperatures. Toner et al. (1990) proposed a nucleation
theory of [IF that suggests that at certain critical temperatures the plasma membrane can
change in such a way that it becomes an effective ice nucleator. An alternative theory by
Muldrew et al. (1994) suggests that membrane rupture may occur due to the osmotically
driven water efflux, consequently allowing extracellular ice to propagate into the cell.
Despite the divergence of Opinion on how ice manifests intracellularly, it is generally
agreed that HF does correlate strongly with cell injury (Toner, 1993). Figure 1.1
summarizes the physical nature of the damage that is found at the different freezing rates.
Although the aforementioned theories offer plausible explanations to describe the
two classes of injury there is still much to be learned about the mechanisms that precede

or supersede cell damage and about the primary sites of injury.

14

rlzurt

3N1;

 

Slow Freeze {1‘}
fl . ’ “it? {I} Solute Injury

{32? £3,525?

 

Fast Freeze

if: 3:} Slow Thaw
<1“? :2?

Intracellular Ice

 

 

 

Figure 1.1 Location and size of ice crystals at different freezing rates. Images in this
dissertation are presented in color.
1.4.3 Secondary Injury

Frostbite research has contributed significantly to the understanding of how the
micro-vasculature responds to freezing (Sjostrom er al., 1964; Reite, 1965; Weatherly-
White et al., 1964, I969; Snider et al., 1975; Waris and Kyosala, 1982; Marzella et al.,
1989; Manson et al., 1991; Bourne et al., 1986). A common problem with
comprehensively interpreting the results of these studies is the absence of thermal control
and monitoring of the freezing process during the experiments. Oftentimes the
temperature of one discrete point in the tissue was monitored and data then generalized to
characterize the entirety of the frostbite injury. A number of the experimental techniques
relied on immersion of extremities in liquids of varying sub-zero temperatures to create
injury (e.g. Sjostrom et al., 1964, Weatherly-White er al., 1964, 1969; Manson et al.,

1991), with temperature histories in the actual tissue not documented.

15

 

 

 

,.
it:

I
‘,.._

uni.

-L.
‘CI

..
1w

uilu

Oil

6.

he.)
fix».

In addition to the preponderance of data available from frostbite studies,
investigations into the effects of lower temperatures and faster freezing rates
characteristic of cryosurgery procedures have also been performed. Zacarian er al.
(1970) and Rabb et al. (1974) have utilized liquid-nitrogen cooled copper probes to
directly examine the effects of cryogenic temperatures on microcirculation in the hamster
cheek pouch. More recent efforts by Schuder et al. (1999) and Hoffman et al. (1999,
2001b) have begun to explore micro-vascular changes in response to cryosurgically
relevant freezing in rat models.

These different experimental approaches to understanding the effects of freezing
on the micro-vasculature has resulted in the emergence of a general pattern of vascular
freezing injury. As the tissue is cooled, the initial therrnoregulatory physiologic response
is increased blood flow to the tissue (Weatherley-White, 1969), termed the “rewarming”
response. With sustained cooling the biological system begins to lose its protective
function, and blood flow to the cooled site is reduced. When the phase change
temperature is reached, the tissue begins to freeze first in the extra-cellular space
(Rubinsky er al., 1987, 1988, 1990). Circulatory stasis occurs in the affected tissue.
Vascular distention has been observed both in frozen section (Rubinsky et al., 1990,
1997) and intra-vitally post-thaw (Schuder et al., 1999). During thawing, circulation is
temporarily restored to the previously frozen tissue however progressive failure of the
micro-circulation is observed. Edema develops (Weatherley-White et al., 1964; Reite,
1965; Marzella et al., 1989; Zook et al., 1998; Manson et al., 1991; Zook et al., 1998),
vascular permeability is increased (Waris and Kyosala, 1982), impaired platelet function

and platelet aggregation leads to thrombosis and vascular occlusion (Weatherley-White et

16

11.5.:
limit

. . .
,
u.\,\

A
4“

hug

3‘94

.3; i_’i

al., 1964; Reite, 1965; Zacarian et al, 1970; Bourne et al., 1986; Marzella et al., 1989;
Zook et al., 1998), and endothelial cells die (Waris and Kyosala, 1982; Marzella et al.,
1989; Manson et al., 1991). With failure of circulation, nourishment of tissues is lost and
tissue death is certain. Adverse microvascular changes have been observed to occur on
several time scales. Changes have been observed as early as immediately after freezing
(endothelial cell damage) by Manson et al. (1991) or 3-min after thawing (small vessel
occlusions) by Bellman et al. (1956) to as long as 24 hours after freezing (necrosis of
endothelial cells) by Waris and Kyosola (1982).

Recent efforts to provide more control over freezing protocols applied as well as
better accuracy in the registration of temperatures that are reached in the tissues, have led
to the formulation of freezing rate and temperature sensitive models of vascular
distention occurring during freezing (Rubinsky et al., 1988, 1989; Bischof et al. 1993).
However, many questions remain as to how thermo-physical effects such as vascular
distention and the physiological effects of cell-mediated changes in vascular permeability
and platelet function inter-relate to give the final microvascular lesion.

The acute inflammatory response, characterized by neutrophil adhesion to the
endothelium and subsequent transmigration into the tissue following thawing are reported
to occur at different time points following cryosurgery. The magnitude and localization
of this inflammatory response can affect the ultimate size of the lesion, though this
contribution has not been well studied. Recently some investigators are beginning to use
infra-vital models to study the magnitude of this contribution however (Shuder et al.,

1999).

17

35a

Hoffman and Bischof (2001ab) have used the transparent skin flap model to
assess vascular integrity following freezing in a Copenhagen rat / AT-l tumor model.
Their results indicate that at the edge of the cryo—lesion where high sub-zero temperatures
are predicted, vascular damage correlates with histological assessment of cellular damage
3 and 7 days following treatment. The temperatures recorded at the edge of the lesion
were shown to be considerable higher than those necessary to kill cells in vitro,
suggesting that the injury in this region was associated with the secondary effects of
vascular injury as opposed to the immediate effect of freezing on the cells. This study
begins to decouple the effects of direct cellular injury and the secondary ischemia effects
of freeze injury. Development of a continuous tissue viability assessment technique
would enhance the utility of this experimental system and could provide the first direct
evidence that cells at the edge of a cryo-lesion die not as a result of the initial freezing

insult, but instead by the secondary effects of ischemia.

18

2.1 Cfl

(IIL‘C
gate rexpon
Subsequent},
.ii'tlluugl‘. (1
al.. 1994; 2

current lbc

i0“ Chimnrl
’ltan er a

anplbm

I ’ ~“ 5

‘9“ fix: ,1. a
Tm;

{Drier

Chapter 2

Molecular and Physical Basis for the
Proposed GFP Technique

2.1 GFP Transfection

Green Fluorescent Protein is native to the jellyfish species Aequarea Victoria. The
gene responsible for the production of GFP was first cloned by Prasher et al. (1992), and
subsequently expressed in both prokaryotic and eukaryotic cells by Chalfie et al. (1994).
Although GFP was first implemented as a reporter molecule for gene transfer (Chalfie et
al., 1994; Zhang et al., 1996), this molecule’s unique fluorescent properties underlie its
current use in a diversity of applications such as protein localization (Cubitt et al., 1995),
ion channel expression (Marshall et al, 1995), cardiac development and pathophysiology
(Xian et al., 1999), erythroid differentiation (Chida et al., 1999), measurement of
cytoplasm viscosity (Swarninathan et al., 1997), tumor metastasis (Chishima et al.,
1997a—e), and angiogenesis (Li et al., 2000).

Transfected cell lines are produced by introducing a GFP-encoding segment of
foreign DNA into the cell line of interest. This study utilizes the GFP-transfected Rat

3230 adenocarcinoma (R3230AC) tumor cell line. A GFP-transfected cell manufactures

19

the GFP protein using its inherent protein synthesis mechanism and native amino acids.
Furthermore, the vector can remain stably expressed in successive generations, which
makes it especially attractive for studying tumor metastasis. Chishima et al. (1997a-e)
and Yang et al. (1998b) have effectively utilized transduction of a GFP vector to
characterize successive stages of metastatic colony growth of human lung
adenocarcinoma, and the metastatic potential of hamster ovarian tumors. More recently
this group has been developing a GFP-based mouse model of metastatic human prostate
cancer (Yang et al., 1999). Higashijima et al. (2000) have used GFP transfection to
study cranial motor neurons. Others have used GFP transfection to monitor herpes virus
transmission (Brewis et al., 2000). This protein has recently been used to study gap
junctions (F alk and Lauf, 2001). Despite their relative novelty, the use of GFP proteins is
wide and varied.

The water-soluble cylindrical GFP monomer, comprised of 238 amino acids
(Prasher et. al., 1992), is 3 nm in diameter and 4 nm long (Yang et al., 1996). The
reported molecular weight of the GFP ranges from 27,000 (Shimura, 1979) to 30,000
(Prendergast and Mann, 1978). GFP and its mutants can be fused with other proteins
both at the N- and C- terminus without loss of biological activity of either protein and
without loss of fluorescent properties (Cubitt er al., 1995; Marshall at al., 1995; Yokoe
and Meyer, 1996; Rizzuto et al., 1995; Windoffer and Leube, 1999). However in some
instances the biological function can be attenuated to some extent (Yokoe and Meyer,
1995), therefore preliminary screening for functionality as well as fluorescence is

important.

20

€0an
and s?

lye-val

IRLIEI

Mud

F‘QUrr

i rib-‘\l~
ll‘

Yang et al. (1996) describe the tertiary structure of GFP as cylindrical, and
comprised of a tightly woven, ll-stranded [3- sheet, with an or-helix inside the cylinder
and short helical segments on each end. This structure is frequently referred to as a [3-
barrel or B-can. The fluorophore portion of the protein is encapsulated within the barrel,
utilizing the at end-segments as scaffolding (Yang et al., 1996). This structure is pictured
in Figure 2.1. Although truncations of the protein can be tolerated, they are minimal, and
residues 7-229 must be maintained (Li et al., 1997) to ensure proper formation of the B-
barrel. The protective barrel ensures the proper environment for stable fluorescence by

excluding oxygen and adverse solvents.

 

Figure 2.1 Tertiary structure of a GFP molecule: a tightly-packed ‘B-can’ comprised
of B-sheets (green) which enclose an alpha helix (red) with the appendaged chromophore
protected in the center of the B-barrel. (From Yang et al. , 1996)

21

Although the nascent protein is itself not fluorescent, the formation of the
chromophore that imparts the fluorescence occurs autocatalytically (Chalfie et al., 1994;
Heirn et al., 1994; Kolb et al., 1996), requiring only molecular oxygen to complete this
post-translational modification. The tripeptide chromophore depicted in Figure 2.2, p-
hydrux,‘ , "" ' " "" , is generated by post-translational cyclization and
oxidation of a Ser-dehydroTyr-Gly sequence at residues 65-67 (Cody et al., 1993). This
chromophore is responsible for the fluorescent absorption bands at 280, 395, and 480 nm,
as well as the emission maximum at 508-509 nm (Morise et al., 1974) in wild-type GFP.

The chromophore is only fluorescent when embedded in the GFP protein in its fully

folded state.

 

Figure 2.2 The GFP chromophore, highlighted in green is comprised a cyclic
tripeptide (Ser-dehydroTyr-Thr). (From Cody et al., 1993; www.biochemtech.uni-
halle.de/PPS2/r j ‘ ’j.....‘...’ L 1* km)

 

22

'H

be.

CN‘A

4.;

~ Q .

.
ul.._g

53-,-

”Va.
Hl -.»'

0:"-

Mg.

F‘Q
&| [’
'-\"i?..~
e..._
'5,

2.2 Enhanced Green Fluorescent Protein

Wild-type GFP does not have high levels of expression in higher eukaryotes such
as human and primate cells (Kain, 1999). It also has a relatively low level of
fluorescence when compared to some of the variants of GFP that have been recently been
constructed. These mutants are now preferred for many applications. There are
numerous variations or mutants of the wild-type GF P (e.g. Heim et al., 1994). The GFP
mutant used in this work is the Enhanced GFP (EGFP), which is a red-shifted variant of
wild type GFP that has been engineered for brighter fluorescence. ‘Red-Shifted’ refers to
the direction of the change in spectral properties that occur as a result of the point
mutations, with the excitation peak shifted towards the red end of the photodynamic
spectrum. The mutant GFPmutl, described by Cormack et al. (1996), contains two point
mutations in the chromophore. The serine residue at position 65 is replaced with
threonine and at position 64 the phenylalanine residue is converted to leucine. The
resulting mutant has a single excitation peak at 488 nm (Cormack et al., 1996) as
opposed to the two peaks of wild type, and is well suited for use with standard
fluoroscein isothiocyanate (FITC) filter cubes. The emission Spectrum is basically
unaltered from wild-type GFP. Enhanced GFP is a specialized form of GFPmutl. They
have exactly the same amino acid sequences, but the EGF P variant has also been
optimized for improved expression in mammalian cells by the replacement of some of the
native codons (i.e. originating in the jellyfish) with codons that are highly expressed in
human proteins (Zhang et al., 1996; Haas et al., 1996; Clonetech Protocol # PT2040-1).
A related mutant worth mentioning is the S65T mutant that includes the Ser to Thr

substitution at position 65 but without the position 64 mutation. The Phe-64 to Leu

23

mutation imparts a thermo-tolerance not present in the thFP. Patterson el al. (1997)
have shown that on the basis of protein folding studies, the EGFP fluorescence intensity
expressed at 28 °C and 37 °C are equivalent. In the case of thFP the exhibited
fluorescence of transfected cells has been shown to diminish at higher culture
temperatures (Lim et al., 1995).

The spectral resolution achieved by such alterations in the excitation spectra
makes the red-shifted mutants more readily utilized in conjunction with other probes.
Crossover-free dual labeling experiments with EGF P and DsRed, a novel red fluorescent
protein, in E. Cali were recently performed by Jakobs et al. (2000). Triple combinations
of red, green, and blue proteins have also been reported (Beavis and Kalejta, 1999).

The mutant EGFP retains the same tertiary structure as the wild-type protein (a [3-
barrel configuration), but is 35X brighter than wild- type GFP when excited with blue
light (Cormack et al. , I996). The extinction coefficient (Em) of EGFP is ~55,000 cm'lM'l
compared to 7000 cm'lM'l for wild-type GFP (thFP). The mutant EGFP also has
improved solubility and reduced photo-bleaching rates when compared to wild type GFP
(Jain, I999). The pH stability for EGFP is also different. Although thF P fluorescence
has been reported to be stable between pH 6 to 10 (Ward, 1981), EGFP fluorescence
begins to decrease below pH 7 (Patterson et al., 1997). This sensitivity to pH is shown in
Figure 2.3. Haupts et al. (1998) demonstrated that the drop in fluorescence at lower pH
is evident in the excitation, emission, and absorption spectra, as shown in Figure 2.4. The
authors suggest that a single protonation step (protonation of the hydroxyl group of Tyr-
66) is responsible for conversion of the protein into a non-fluorescent state. The pKa of

this reaction is given as 5.8. Wachsmuth et al. (2000) have also examined this pH

24

:‘Iji’lltl
611 z

A
raids

phenomemon using Fluorescence Correlation Spectroscopy (F CS) and give the value as
6.0 i 0.4. Fluorescence of EGFP was found to be independent of the ionic strength

(Haupts et al., 1998).

 

Normalized Fluorescence

 

 

Figure 2.3 Effect of pH on fluorescence of thFP and some of its mutants. For
thFP and EGF P, excitation was at 488 nm and emission was collected from 490-650
nm. Maximum signal strength was normalized to 100. EGFP fluorescence can be seen
to decrease in intensity at a higher pH than thFP. From Patterson et al., 1997.

25

 

1°
0

  
 

 

 

intensity'10’ [cps]
;

,4 s 6 ’1 6170710
pH

1"
O

.0
or
q -

i

 

 

 

9
o

400L420 140460730300
wavelength [nm]

Figure 2.4 Fluorescence excitation spectra of EGFP as a function of pH. From the
top: 9.0, 8.0, 11.0, 7.0, 6.5, 5.5, and 5.0. Inset: fluorescence emission intensity at 510
nm (I) fluorescence excitation intensity at 490nm (e) and absorption at 488nm (O).
From Haupts et al., 1998.

Other mutants have been optimized for localization within particular organelles
and cell compartments (De Giorgi et al., 1999), as well as for different spectral ranges

(Cormack et al., 1996; Heim et al., 1994), which indicates many possibilities for

specifically targeting damage following insults of the nature explored in this work.

2.3 Photodynamics

Swaminathan et a1. (1997) have used fluorescence recovery after photo-bleaching
(FRAP) and anisotropy decay techniques to study the translational and rotational
diffusion of green fluorescent protein in solution and biological cells. The GFP mutant
used in their study was S65T. The FRAP technique involves irradiation of a small area in
a larger sample of aqueous or in situ fluorophore, using laser excitation of suitable

wavelength, power, and duration to induce photo-bleaching. The laser intensity is then

26

 

r2220
during
proteii
130755
area

fimfa‘ir

hieachi

'3‘ )o-I‘
"'LANE
. ‘7. ‘T '

114m
;-0."'ur
“"45111

1
:2'1 ‘lfi-l‘i

A}. flat

attenuated significantly to allow subsequent monitoring of the fluorescent intensity
during the recovery period without causing further photo-bleaching. Since the GFP-S65T
protein was found to irreversibly photo-bleach in aqueous solution, recovery of
fluorescence was assumed to be dependent only on translational diffusion into the probed
area. This recovery was also found to be independent of concentration of the
fluorophore. Based on the half-life of the fluorescence recovery following photo-
bleaching, the diffusion coefficient of GFP-S65T in aqueous solution was determined to
be 86.7 umz/s.

The Swaminathan et al. (1997) study also demonstrated that the absence of
reversible photo—bleaching was evident only in dilute solutions. When the viscosity was
increased with glycerol, a rapid recovery process was observed. However, the significant
difference in time constants for the rapid recovery process (reversible and irreversible
photo-bleaching) allowed the determination of diffusion coefficients in viscous media.
As expected, the same time-resolved phenomena were observed in the viscous cytoplasm
of the cell. Based on tug of the reversible recovery process, the relative viscosity
difference between cytoplasm of CHO cells and aqueous solution was determined to be
3.2. Because the diffusivity was found to vary directly with viscosity, this suggests that
the primary barrier to GFP diffusion is interaction with macromolecular solutes.
Recovery curves were not affected by the concentration of GFP-S65T, suggesting that a
concentration-dependent monomer-to-dimer transition characteristic of thF P (Chalfie,
1994) does not occur in this mutant.

As mentioned previously, Wachsmuth et al. (2000) have used FCS to study the

spatial diffusion of EGF P in cellular environments. Because the time resolution of video

27

05.

(’9’ ‘

in9

or:

microscopy does not permit observation of the motion of a single molecule, they use
fluorescence intensity fluctuations in laser-irradiated femtoliter volumes of fluorochrome
solutions, which contain just a few fluorophores, to determine hydrodynamic and
photophysical properties of the molecules. These fluctuations in intensity are caused by
Brownian motion of the molecules, as the fluorophores move into and out of the
observation window. Based on the lifetime of fluorescence decay it was possible to
compute diffusion coefficients. Similar to the results of Swaminathan et al. (1997) the
diffusion coefficient of EGFP in aqueous solutions was determined to be 87 umZ/s. In
cells, the autocorrelation function (ACF), which characterizes the fluctuations in
fluorescent intensity, generally lent itself to a single freely diffusing species model in the
cytoplasm, and either a two-species model (model 1) or obstructed diffusion model
(model 2) in the nucleoplasm. However, there were variations that suggested the
organelles in the cell or some other cytoplasmic structure caused some hindered diffusion
in the cytoplasm as well. On average, using the fast component of the two-species model
or the obstructed diffusion model gave very similar results for diffusion of EGFP in two
types of cells (less than 5% variation). Based on the reported ratios of diffirsion
coefficients, the diffusivity of EGFP in AT-l cells was given as 15.8 (model 2) or 16.4
umz/s (model 1), and in COS-7 cells it is 19.3 umz/s (model 2) or 18.5 umz/s (model 1).
1n the same study, based on the amplitude of the ACE and the size of the detection
volume, the Wachsmuth group (2000) found the average GF P protein concentrations in

the AT-l cell to vary between 10 and 150 nM.

28

CO")

C1114

2.4 GFP as the basis for a Viability Assay

Wild-type GFP, as well as EGFP, is known to be distributed throughout the
cytoplasm and nucleus when expressed in mammalian cells (Cubitt er al., 1995). With
targeted mutations to the vector, GFP can also be tagged specifically to other organelles
(Rizzuto et al., 1995, 1996). Because GFP accumulates in the cytoplasm of normally
metabolizing transfected cells, it seems likely that damage to the plasma membrane could
result in leakage of cell contents, including GFP. In this work, a GFP viability assay,
based on a loss of GFP fluorescence in damaged cells, is compared to the viability results
determined by a number of classical indicators of viability, including Trypan Blue
exclusion, Ethidium Bromide exclusion, and standard histopathology. Cells and tissues
were stressed by treatment with a range of freezing protocols and ice-free hyper-osmotic
solutions with concentrations Similar to those that might be experienced during freezing.
These conditions have implications for both cryosurgery and cryopreservation.

A cell viability assay based on a transfection process has also been explored by
Coombe et al. (1998) and Baumstark-Khan er al. (1999). The assay developed by
Coombe is based on the expression of luciferase that has been transfected into cells. This
assay requires addition of a substrate however, and as currently implemented, has little
utility as an in situ viability assay for tissues because it has the same limitations of
delivery as any other viability probe. A similar GFP-based assay, requiring no
exogenous substrate, was developed for biotechnology applications by Hunt at al. (1999)
to quantitatively assess colony grth in bioreactors. Their studies indicated that GFP
fluorescence increased with cell concentration to a maximum that was dependent on the

original plating density. The high correlation of GFP fluorescence to cell number

29

«Gum

WORM:

the sum

bunliH
Trixie.

urns

Futile.
5.510 '
will It
Cllt'dpg
Kim‘s
Wild)
1035 0
Sink-Ad

hot, 3‘

(R2=0.997) supports the use of GFP fluorescence as an indicator of cell viability as
proposed in this work. The assay utilized by Baumstark-Khan et al. (1999) was based on
the same concept but implemented differently. The colony forming ability of wild-type
and GFP-transfected CHO cells was measured after irradiation. They found that colonies
surviving radiation expressed GFP and that the surviving fractions were comparable for
transfected and wild-type cells. None of these transfection-based assays were explored as
the basis for a tissue viability assay.

As the tertiary structure suggests, GFP is physico-chemically very stable.
Purified samples of GP P have been shown to be highly fluorescent at pHs ranging from
5.5 to 12, and thermally stable to 65 °C (Bokman et al., 1981). Denaturation of the GF P
will result in a loss of fluorescence, primarily because the chromophore is no longer
encapsulated and protected (Cody er al., 1993; Prasher er al., 1992). Fluorescence is
recovered upon renaturation however (Bokman et al., 1981). The successive freeze-thaw
method typically used to isolate proteins has also been employed to isolate GF P without
loss of fluorescence (e.g. Gonzalez and Ward, 2000), and purified GFP is shipped and
stored in the frozen state (e.g. Clonetech, Palo Alto, CA), suggesting cold-denaturation is

not a problem when using GFPs at low temperatures.

30

liti. " '

t 11\.\’.It
Cl
animal ratltt
reduces the :
that can by 1:

di

Chapter 3

Significance of the Research

Successful development of a GP P fluorescence-based viability method capable of
producing instantaneous in viva viability data within tissues would create future
opportunities for many researchers. The GFP method has numerous advantages
compared to traditional viability assessment methods. These advantages include:

a) Instantaneous, quantitative, non-invasive viability determination within
living tissues. This would eliminate the need for labor-intensive, time-consuming
histological processing of tissues, and avoid the possibility of fixation artifacts.

b) Continuous monitoring of spatio-temporal viability relationships within
living tissue.

c) Ability to monitor complex time-course experiments within a single
animal rather than having to use many animals sacrificed at sequential times. This
reduces the number of animals required for experimentation and eliminates ambiguities
that can be introduced by animal-to-animal variations.

d) Possibilities to combine optically-based GFP tissue cell viability method

31

"4
‘1
Jun

spec

INS-

. 9"

”1 iii}

if ‘R"

TH .
.‘irr

0.»!
ECL'i‘h.
.j‘;.

with other optical methods (fluorescence-based or otherwise) for simultaneous
measurements of physio-chemical, thermal, morphological, physiological, inflammatory
and immunological responses to freezing/thawing or other processes by using multi-
spectral probes, infrared thermography, and other existing technologies.

e) Improved knowledge of spatial relationships that are difficult to track after
histological processing (e.g. accurate positioning of tissue and/or vascular damage
relative to location of frozen regions).

There are some limitations to the use of GFP for viability assessment. The stable
transfection of cells with GFP is not a trivial task, and reselection may be necessary
periodically to ensure that a portion of the colony has not lost the GFP vector. The use of
a GFP viability assessment technique in a quantitative way will also be limited to
situations in which optical access with exciting wavelength light is possible.

The use of GFP-transfection techniques can provide a non-toxic way of
introducing fluorescent markers into living tissue and thereby offers the potential to
examine cryosurgery phenomena in viva, in real time. This approach can provide more
realistic information about the effects of low temperature on tumor tissue as it relates to
cryosurgery because the studies are being conducted in a living and normally functioning
system. Prior to this study, GFPs had not yet been evaluated as a marker/viability
indicator in the low temperature regime. If this application is successful, GFPs can
provide a means of directly observing freeze injury phenomena, dynamically, in a living
system. Not only is there potential to develop an in viva viability marker, but this

technique simultaneously provides a means of studying the effects of freeze-induced

32

\axt‘ulrl

lolltmir

freezing
correlazt
utilized
during c
constitut
{urchin

lite tin:

mute C

vasculature damage on tumor devitalization as well as the risk of tumor metastasis
following a cryosurgical procedure. To date, this has not been possible in a living host.

The current work seeks to provide the first in viva evidence that following
freezing, fluorescent intensity changes of tumors grown from GFP-transfected cells
correlate with tissue damage as assessed by histological methods. The freezing protocols
utilized were chosen to span the range of damage (both type and severity) experienced
during cryosurgery. The experimental assay proposed to evaluate viability relies on the
constitutive expression and retention of stable GFP. The GFP molecule therefore
functions primarily as an indicator of membrane integrity and normal anabolic activity.
The first part of this study looks at the effectiveness of the GFP assay in individual cells,
and the second part of the work investigates the validity and limitations of the assay in
viva. This approach thus answers questions pertaining to its utility both in vitro and in
viva. Additionally it may highlight potential differences in how the assay should be
implemented for maximum reliability in isolated cells and cells within tissues.

The GFP method described here provides an exciting new possibility to monitor
events relevant to improving our understanding of the mechanisms of cryosurgical
damage at the tissue level. In addition to immediate applications, it should be possible to
develop different GFP mutants with different wavelengths which may make it possible to
simultaneously label and observe multiple cell types such as endothelial cells and cells of
the immune system via transfection into embryonic cells such as has been done in the
case of the GFP mouse (Okabe, 1997) and the GFP bunny (Kac, 2000). If such cell—
specific GFP mutants become available, it may be possible to observe the absorption of

antigenic material into the lymphatic channels and lymph nodes, as well as observe cell-

33

IIICC;

1ch

£11.11:

andt

1t mu

11551.1;

l’:“()o

mediated immune response to the cryosurgical assault, thus providing additional tools for
understanding cryo-immunologic phenomena. Although the thermal conditions used to
challenge the cells in this work corresponds to the injury experienced during cryosurgery,
and to some extent during cryopreservation, the assay is not limited to these applications.
It may also prove to be a valid viability assay for a wide range of phenomena, including
tissue hypoxia, drug-induced tissue death, and hyperthermia, and do so dynamically, in

viva, without sacrifice of the host animal.

34

Chapter 4

Cell Line and Animal Model

4.1 GFP-Transfected Rat 3230 Mammary Adenocarcinoma Cell Line

A GFP-transfected rat mammary adenocarcinoma (R3230AC) cell line was
provided as a gift from Drs. Chuan-Yuan Li and Mark Dewhirst, Dept. of Radiation
Oncology, Duke University Medical Center. This cell line can be utilized to grow
transplantable fluorescing mammary adenocarcinomas in the Fischer 344 rat. The cells
had been transfected with a pEGFP-NI plasmid using DMRIE liposomes and G418
selection (Li et al., 2000). The R3230AC cell line is a subline of a R323OAB cell line
originally derived from a spontaneous mammary adenocarcinoma in the Fischer rat and
characterized by Hilf er al. (1964). Like its parent line, the R3230AC line is hormone-
sensitive, responding to both estrogen and androgen. This cell line has extensive
secretory activity, and can be induced to lactate by estrogen. This R3230AC cell line
does not grow as fast as the parent R3230AB, and has biochemical and morphological

characteristics similar to normal breast tissue (Hilf et al., 1965).

35

 

4.1.1

sup?”
1111113 1
fungi}

IHCUO‘

412 1

he ufld
phhdat
atrmu
hrnuh.
wash

r ,-
:OICdthl

lhuau

4.1.1 Cell Culture

Cells were cultured in T—flasks in Dulbecco’s Modified Eagle Medium (D-MEM)
supplemented with 10% fetal bovine serum, 4% penicillin streptomycin solution (10,000
units/ml penicillin G sodium and 10,000 pig/ml Streptomycin in 0.85% saline) and 1%
fungizone (250 ug/ml) [all products from Gibco BRL, Grand Island, NY], with
incubation at 37°C and 5% C02 atmosphere. Cells were harvested by exposure to 0.25%
trypsin solution (Gibco BRL, Grand Island, NY) for 2 minutes at room temperature,
subsequently removing the trypsin solution by pipette, and then incubating the flask for 3
minutes at 37°C to release the cells from the flask surface. Cells were then re-suspended
in supplemented D-MEM and diluted 1:10 for normal cell passage. To cryo-preserve the
cells a solution of 10% DMSO, 30% fetal bovine serum, and 60% media was used. Cells
were prepared to 1x106 cells/ml and then immediately placed inside a cardboard storage

box in a —80 °C freezer.

4.1.2 Doubling Time

To understand the growth characteristics of the cells, the doubling time for both
the wild-type and transfected cell lines was determined. Five flasks per cell line were
plated at a density of 0.5-1 x 10 4 cells/ml then incubated at 37 °C. Beginning 24 hours
after plating, the cells were released from the flask using trypsin and counted at regular
intervals, using a hemacytometer and Trypan Blue. Figures 2.1 and 2.2 give the growth
curves for the wild-type and transfected R3230AC cells, respectively. The doubling time
for each line was calculated from the exponential fit of the growth data. The wild-type

line was determined to have a doubling time of 16 hours whereas the doubling time for

36

the trans

'6' ..
‘ l..CIt’Ilt

cells tha

ditiding

ligure 4.1

Cn"n-n.--.‘-_ A‘-

Figure 4.2

the transfected line was determined to be 20 hours. There is no apparent reason for the
difference in growth rates. However one might speculate that in the process of selecting
cells that were brightly fluorescing during the transfection process, a more rapidly

dividing colony was isolated from the parent line.

 

8.00E+05 t
7.00E+05 ‘
6.00E+05
5.00E+05 t - - - ~ -
4.00E+05
3.00E+05 ~ - , 7*
2.00E+05 ..

Concentration (# ceiislml)

1 .00E+05

 

 

 

 

 

1 0.00900 . . , _ ,
1 0 20 4o 60 so 100 120 140 160 t
1 Time (hrs) '

Figure 4.1 Growth curve for the wild- -type R3230AC cell line.

 

§ 2.00905
O

0 1.00905 ._
0.00900 -

1 , .

t O 20 40 60 80 100 120 140 160 t
l 1
1

 

 

 

 

 

Time (hrs)

Figure 4.2 Growth curve for the transfected R3230AC cell line.

37

4.1.3 1

mar iii;
ice. an 1
lmmcdiu
cclls stili
Niger 5
50611111111
cells had
diameter it
02mm ml
the“ trim
4.3. m.
Ca*"Cltlatctl i
also Pltlttcd
COIandc. “ll
microns. dtk

dc‘l‘llsfllltln “i1

4.1.3 Cell Size Distribution

Because the Size of a cell figures significantly into how fast a cell can transport
water (discussed further in Chapter 5), and consequently its tendency to form intracellular
ice, an understanding of the average cell diameter and its distribution was desired.
Immediately following a routine cell passage, a culture flask containing GFP-R3230AC
cells still in suspension was placed on the stage of a Nikon Diaphot inverted microscope
(Mager Scientific, Sterling Heights, MI). Using a SPOT digital camera (Mager
Scientific, Sterling Heights, MI) seven images were acquired of the suspension before the
cells had time to attach to the flask. Images of these cells were then measured for
diameter in both the horizontal and vertical axial direction of the image, using the SPOT
camera software and the appropriate stage micrometer-based calibration. An average was
then taken of the two values. A histogram created from the results is shown in Figure
4.3. The total number of cells measured was 327. A probability distribution was
calculated based on the assumption that the population was normally distributed and was
also plotted in Figure 4.3. As can be seen from these results, the sample does not
coincide with a perfect normal distribution. The mean diameter of the population, 14
microns, does not correspond with the mode of the population, 15 microns. The standard

deviation was calculated to be 2 microns.

38

0.30

 

0.25 ..
0.20 4

j,

ity Function

0.15

 

0.10 ~~
0.05 -.

 

 

  

Probability Dene

 

0.00: .
910111213141516171819 20 2122

Cell Diameter (microns)

__ L, _ -__.__._.__.__._______ _.___-_.__ L _.. _ _L.i

Figure 4.3 Histogram and probability density function based on n=327 individual cell
diameter measurements from a single sampling of the cell population acquired during
routine cell passage. The probability density function assumed the population was
normally distributed.

4.2 Fischer 344 Female Rat

Female Fischer 344 rats, originating from Harlan (Indianapolis, IN) and Charles
River Laboratories (Raleigh, NC) were used in this study. Female Fischer rats are an
inbred species with a high incidence of naturally occurring mammary tumors (41%) that
are also receptive to growth of transplantable R3230 mammary carcinomas (Altman and
Katz, 1979). Because R3230 cells were to be used in this study, this species was chosen

as the animal model for experiments.

39

4.3 Immune Response to EGFP

One of the applications motivating the in viva assay proposed in this work is the
study of the secondary effects of freeze injury (ischemia and inflammation).
Consequently it is important to understand how a possible immune response to GFP
released from cells could complicate the tissue response.

The recent interest in GFP co-expression as a means of optimizing therapeutic
gene delivery and expression in humans (Levenson et al., 1998, Aran et al., 1998) has
prompted the investigation of the immune response to GFP. Stripecke et al. (1999) have
studied T cell immune response against EGFP+ cells in murine (mouse) models. AS
indicated by Stripecke, other non-GF P transgene constructs have been shown to evoke T-
cell mediated toxicities in patients. As an example, in clinical trials with HIV positive
patients, Riddell et al. (1996) found that five out of six of their patients developed T-
lymphocyte responses specific for a novel protein produced by the intentionally
transduced genes. Novel protein by-products can present themselves at the T-receptor
sites, thus potentially stimulating the immune system to produce antibodies to this
antigen, as well as elicit an immediate strike on the cell containing the protein, by way of
the killer T-cells.

Stripecke et al. were able to establish that when EGFP-expressing leukemia cells
(EGFP+ BM185 pre-B) were grown in syngeneic BALB/c mice there was a marked
decrease in the kinetics of leukemia growth over the wild-type cells. Only one of twelve
mice developed a tumor, and this lymphoma was considerably smaller than normal, and
over time regressed in size. Several finther experiments supported the hypothesis that the

effect was due to a T-cell response. The same EGFP+ line was injected in

40

immunodeficient Nu/Nu mice, which by definition lack T cells. Normal lymphomas
developed, indicating that T-cells were likely the cause of decreased tumor growth in the
immunocompetent BALB/c mice. Cytotoxic T lymphocyte (CTL) lytic activity
experiments were also supportive of this hypothesis.

Although these experiments were performed in mouse models, it is interesting
that a significant T-cell mediated immune response is not detected in the RC3230 AC cell
line/Fischer Rat syngeneic system. The Fisher rat is immunocompetent, yet readily
grows EGFP+ adenocarcinomas. It is possible that there are enough differences in the
immune systems of the rat and mouse that the observations of Stripecke in the BALB/c
mouse model will not be representative of the expected response in the rat. More likely
the differences are at the cellular level. Stripecke did evaluate tumor growth kinetics
using a different EGFP+ tumor cell line, the T cell lymphoma EL-4, and found generality
of results. However these cell lines are both of lymphoma origin and certainly do not
represent a good cross-section of existing tumor cell lines.

In light of these results however, if GFP-transfected cells were to be used in
studies related to immune response where low level T-cell effects could be significant, a

closer investigation along the lines of Stripecke’s study is warranted.

41

Chapter 5

Characterization of Water Transport
in GFP-R3230AC Cells

5.1 Overview

If the goal of an in viva model is to determine the role that secondary injuries such
as the inflammatory response and vascular injury play in the determination of the
ultimate cryo-lesion size, then to begin with, the basic freeze injury response in the
isolated cell system must be well characterized. Limitations in existing technologies
require the extrapolation from isolated cell and tissue slice behavior to expected in viva
response. The rates and end temperatures that cause both intracellular ice and damaging
dehydration must be understood in order to separate direct cell injury from host-mediated
responses or secondary ischemic effects that arise from progressive failure of the micro-
circulation. The manner in which a cell responds to freezing protocols is directly related
to the cell’s ability to transport water. Presently data pertaining to the water permeability
of R3230AC cells exists neither above nor below the freezing temperature of water.
Comprehensive viability information at a variety of freezing rates spanning the range of

the different types of freeze injury is also lacking. The goal of this research component is

42

to delineate these freezing regimes for the R3230 AC cell line using biophysical
modeling. Determining membrane transport properties allows the utilization of
predictive models to understand when cells can be expected to dehydrate in response to
extra-cellular ice formation, or when HF would be more probable.

The focus of this component of the work is to develop an understanding of the
thermal histories that result in the two distinct classes of direct freeze injury (solute injury
and HF). Generally, the goal is to understand what constitutes ‘slow freezing’ and ‘fast
freezing’ for these cells. This knowledge will help in the interpretation and application of
GP P fluorescence signal intensity changes following injury by these two modes, and will
also assist in the uncoupling of direct and indirect damage in later in viva studies.

There are two key parameters that are recognized as characterizing the response
of cells to freezing, namely the hydraulic conductivity of the plasma membrane, Lp, and
the activation energy of the hydraulic conductivity, E14, (Schwartz and Diller, 1983). The
surface area to volume ratio (S/A) also plays an important role in governing how a cell
responds to freezing (Toner, 1990). Although these parameters are of critical importance
for thermodynamic modeling of the response of cells to freezing, most of the data
available are for properties measured above the freezing temperature of water. Once Lpg
and E14, have been determined for a particular cell type, thermodynamic models of
volume change during freezing can be used to predict intracellular concentration changes
during fieezing for those cells.

The development of successful cryo-preservation and cryosurgery protocols can be
accelerated by knowledge of the cell parameters that govern the transport of water during

freezing. Using models of water transport that have been developed in the past few

43

decades (e.g. Mazur, 1963; Levin, 1976), the amount of water inside a cell at a given
temperature during freezing can be predicted. Knowledge of these parameters can
provide considerable insight into the probability as well as type of injury that would be
expected for given thermal processing conditions.

When the temperature of a cell suspension is lowered, freezing preferentially
begins in the extra-cellular space. As liquid water is converted to solid water in the extra-
cellular solution, a difference in water activity is produced across the plasma membrane.
The water activity of the extracellular solution is lowered in comparison to the
intracellular water activity. If the freezing rate is slow enough, this thermodynamic
imbalance results in an efflux of water from the cell. The rate at which the water can
leave the cell is governed by the parameters Lp and ELp. When no further water can be
removed from the cell, the remaining volume is referred to as Vb, the osmotically inactive
volume. This is an important experimentally determined parameter that is utilized in
most thermodynamic models of freezing (e.g. Mazur, 1963; Schwartz and Diller, 1983;
Levin, 1976).

If the cooling rate is rapid, water cannot leave the cell fast enough to satisfy
equilibrium, the consequence of which is the formation of intracellular ice to reduce the
water activity inside of the cell. The probability for IIF thus increases with increasing
fi'eezing rate. However, ‘fast’ is relative to the particular cell type, and is intricately tied
to the water permeability of the cell. For example, the cooling rate yielding 50% HF in
red blood cells is over 3 orders of magnitude greater than for mouse and hamster oocytes
(Toner, 1993). Red blood cells are known to have unusually high water permeabilities

and high surface area-to-volume ratios.

44

 

Although the membrane permeability of a cell plays a very important role in
determining the fate of water inside the cell during freezing, most measurements of this
value have been obtained at temperatures above zero. There are a few cryomicroscopy-
based approaches for determining permeability values at sub-zero temperatures that have
been described in the literature. In 1979, Levin er al. determined sub-zero permeability
values for yeast, as did Schwartz and Diller in 1983. Levin’s estimation of sub-zero
permeability parameters was based on the experimental data of Ushiyama and Cravalho
(1973), wherein cells were suspended in distilled water and frozen at various freezing

rates. Lp and EM, were then estimated by fitting the volume change data as a function of
temperature to a thermodynamic volume regulation equation using multiple non-linear
regression fitting schemes. Papanek (1978) measured the permeability of the human
erythrocyte down to —10 °C using a stopped flow-light scattering approach. Schwartz and
Diller (1983) used a similar approach to Levin (1979) but instead super-cooled the cell
suspensions to temperatures between —2 and —1 0 °C, and then nucleated ice formation. In
thi s manner the sudden change of extra-cellular concentration induced by the freezing
efl‘ected a mass transfer response by the cell. A novel approach by Bischof and
col leagues uses differential scanning calorimetry to estimate the membrane transport
properties during freezing (Devireddy et al., 1998).
In this work measurements of LI, were undertaken in the non-frozen state. It is
r ec()gnized that the extrapolation of membrane permeability values determined at supra-
fi‘eezing temperatures may not be representative of permeability values at sub-zero
temperatures. However, at present there is no permeability information available for

R3 23OAC cells at any temperature. Therefore, the approach of determining Lpg and E”,

45

 

at temperatures in the range of 0 °C to 22 °C using diffusion chamber methods will be
taken as a first step to understanding the response of cells to hypertonic stress and

consequently freezing conditions.

5.2 Biophysical Model
Considering the cell as an open thermodynamic system, with the cell membrane
represented as a semi-permeable barrier (permeable to water but not to salt) across which
we describe water transport, the equation describing the volume flux of water, JV (um/s),

is given as:

4ch
V Ac dt

 

= _Lp(c::ll '03:) [5‘1]

where Ac is the surface area of the cell (umz), V6 is the volume of the cell (um3), t is time
(s), Lp is the hydraulic permeability (um/atm—min), cexsan is the extracellular salt
concentration (Osmoles) and cinsan is the intracellular salt concentration (Osmoles). An
increase in the extracellular salt concentration will cause water to leave the cell thereby
increasing the intracellular salt concentration. This change in intracellular salt

concentration can be related to the changing volume of the cell according to:

in “:1:
e58] = in [5.2]
t Vc -Vb _nsaltUsalt

 

46

where Vb is the osmotically inactive volume (um3), ninsalt is the number of moles of
intracellular salt, and 058“ is the molar volume of salt (um3/mol).

When the change in extracellular concentration is also accompanied by a
temperature change (such as can occur in the case of freezing) the temperature
dependence of the water activity needs to be considered in the derivation of water
transport equations. The first equation to quantify the relationship between temperature
and the amount of water leaving a cell during freezing was developed by Mazur (1963).
This mathematical relationship has provided the basis for almost all subsequent

thermodynamic models of cellular dehydration related to freezing (Mansoori, 1975;
Silvares et al., 1975; Levin et al., 1976; Shabana and McGrath, 1988; Toner et al., 1990).
There are a number of assumptions used to develop this model including: (1) absence of
hydrostatic pressure gradients across the membrane, (2) negligible temperature gradients
across the membrane, (3) concentration gradients existing only across the membrane, (4)
i nfinite extracellular space, (5) constant surface area, Ac, (6) spherical cellular space, (7)
ideal and dilute intracellular solution, and (8) a constant latent heat of fusion (McGrath,
1 985). These assumptions together with basic thermodynamic laws permit an expression

0 f water flux across the cell membrane as:

 

 

—L RT in
“W = " APE”) [5.3]

where aw is the activity of water, nw is the number of moles of water, T is absolute

temperature (K), R is the universal gas constant (cm3atm/mol-K), and ow is the molar

47

volume of water (um3/mol). The intracellular water activity, if modeled as an ideal

solution, can be given by the mole fraction of available water, Xw:

a: =x‘; =___n_..__ [5.4]
nw +(vsnsah)

where nw is the number of moles of water and V5 is the dissociation constant of the solute.
In this work the solute is sodium chloride, which is assumed to undergo complete ionic
dissociation to yield a dissociation constant of 2. Using the molar volume of water

Equation 5.4 can be rewritten as:

. . V
a: 2X: = w [5.5]
Vw +(uwv n. )

5 salt

 

The water content of a cell is composed of both free and bound water, as well as
salt, proteins, and other components that make up cell membranes and organelles. It is

convenient to express this relationship as:

Vc = vw + Vb [5'6]

where VW represents the volume of free water, or water available for transport across the
cell membrane, and Vb is the volume of the remaining components of the cell once all

poSSible free and unbound water has lefi the cell. This allows equation 5.5 to be

3' fortnulated as:

48

in _ (Vc —Vb)
a“ ‘ (v. -v.)+u.(v.n.) ”'71

 

The extracellular water activity can be related to the extracellular concentration and

temperature, assuming constant pressure, using a Gibbs-Helmholtz relationship:

AH,

6 ..
Edna... )P = —RT, [5.8]

Assuming constant AHf, this can be integrated to yield an expression for extracellular

water activity:

 

as: = AH‘[—1——l] [5.9]

where AHf (J/mol) is the latent heat of fusion for water, and TR is the reference
temperature of 273.15 °K, the freezing point of pure water.

At constant cooling rates, B, the temperature is related to time according to:

B = 111 [5.10]
dt

49

Combining Equations 5.3, 5.7, 5.9, and 5.10 results in the following water transport

equation, expressed as a function of temperature:

 

 

 

 

 

an: _LpART ln (VG—Vb) _AH,. _'__l [511]
dT uiB (Vc — Vb)+ VSans R TR T .
If Vb is independent of time and temperature then:
ch : dVw =Uw dnw [5.12]
dt dt (it
and equation [5.1 1] can be rewritten as:
L ART _

fizz—v In (v, Vb) _AHf _‘___1. [5.13]
dT uwB (vc — vb)+ vsuwns R TR T

The temperature dependence of Lp can be given by an Arrhenius mathematical

description (Levin et al. , 1976):

E 1 1
L =1. ex i —-— [514}
P P8 p[ R [TR TJ]

where Lpg is the membrane permeability at the reference temperature, TR, and E”, is the

"ctivation energy of Lp.

50

5.3 Materials and Methods
5.3.1 Determination of Vb, Osmotically Inactive Volume

The osmotically inactive volume can be determined by a number of different
experimental approaches. A microscope diffusion chamber, which utilizes a change of
extracellular solute concentration to effect a change in cell volume, was utilized in this
work. The model of diffusion chamber utilized is depicted in Figure 5.1. The
microdiffusion chamber allows cells to be directly observed with a microscope as they

respond to changing osmotic conditions produced by an external flow. This apparatus

was described in detail by McGrath (1985). The cells are separated from the bulk flow

region by a semi-permeable dialysis membrane. As the bulk flow concentration is

changed, the change in the local environment of the cell (sample region, Figure 5.1) can
be quantified using a simple mass transfer analysis across this membrane.

Cells were cultured according to the methodology described in Chapter 4. Cells
were detached from the flask with trypsin when they had grown to nearly a monolayer
and were then concentrated by centrifugation. The supernatant was decanted and the

cells were then re-suspended in ~1 ml of un-supplemented DMEM media. The cells
were placed on the thin plastic window of the diffusion chamber and then covered with a
semi-permeable dialysis membrane. The assembly was then inverted and inserted into
the customized stage component. The dialysis membrane mechanically separates the
eel ls fiom the bulk fluid flow, but allows the hypertonic solution to diffuse across the
membrane in a quantitative rate-limited manner. A Sony charge-coupled device (CCD)
Video camera was mounted on the camera port of a Zeiss Universal microscope and the

video signal merged with the output of a video timer, the final image being recorded on a

51

 

 

Sony SLV-420 videocassette recorder (VCR). A 25x microscope objective was used

together with a 1.25x microscope tube head magnifying lens. The cells were then

exposed to and equilibrated in progressively hypertonic solutions (308-1200 mOsm) and

the change in cell dimensions monitored microscopically and videotaped for offline

analysis. Offline images were obtained using a video capture board and then processed

with Image Pro® PLUS sofiware (Media Cybernetics, Silver Spring, MD). Following

the 1200 mOsm equilibration step, the cells were returned to isotonic concentration.

Only cells that maintained a spherical shape throughout this process and subsequently

returned to the starting volume at the end were selected for further analysis.

Spacer Ring

”\O

 

 

 

Microscope
Objective

 

 

 

 

 

 

 

 

 

 

 

 

 

 

' —\ /—— Cover Glass
Tap Fitting
\\i\ \ Dialysis
Membrane
‘— Sample
III/IIIIIIZD Region I
1*
:> Bulk Flow Diffusion C
llflllllll‘ln (Iii/L] Chamber
k Plastic Body
\ /
\ /

NOT TO SCALE

Light Source

figure 5.1 Schematic of a microscope diffusion chamber. The semi-permeable
lEi-llt’Sis membrane allows solute in the bulk flow region to passively diffuse across the
membrane while containing the cells in the sample region. (From Sherban, 1987).

52

Cell volumes were determined from 2-D image area measurements by manually
tracing the outline of the cell three consecutive times and then computing the average of
these measurements. With the assumption that the cells were spherical, these
circumference measurements were then converted to volume measurements using the

following relationship:
V = —— [5.15]

where V is volume, and C is circumference. The computed volumes were normalized to
the starting cell volume and then plotted as a function of inverse osmolarity and a linear
regression analysis of the data was performed. The linear equation was then solved for
infinite osmolarity (x-intercept = l/Osm = O) to obtain the osmotically inactive volume

Vb.

, v
_Y_=M...[ milky—b— [5.16]

5-3-2 Determination of LI, and EL,

The microdiffusion chamber used in this work has channels incorporated in the
aSSembly to allow for a temperature regulating fluid to be circulated around the exterior

o f the sample region. This concept is depicted in Figure 5.2. In the current study the Lp

53

value was desired at temperatures ranging from ambient to near the freezing temperature
of water. Cooled ethylene glycol refi'igerant was circulated through the diffusion
chamber channels using a circulating programmable refrigeration bath (RTE-140). The
bath temperature was maintained at a slighter cooler temperature than the desired
operating temperature of the microdiffusion chamber to compensate for heat losses in the
TygonTM tubing used to deliver the refi'igerant. The hypertonic solutions were
maintained at the desired operating temperature in a reservoir containing ethylene glycol
refrigerant. The temperature of this fluid was regulated using a second Neslab circulating
refrigeration bath, also maintained several degrees below the desired operating
temperature of the microdiffusion chamber. This set-up is depicted in Figure 5.3. The
refrigeration baths were mechanically isolated from the microscope to prevent
undesirable vibrations in the microdiffusion chamber. The gravimetric flow of fluids to
the chamber was controlled by a set of solenoid valves that allowed rapid switching from
one solution to another. A bleed line was placed between the solution bottles and the
solenoid valves to allow the release of standing solution in the hose before switching

from isotonic to hypertonic flow.

54

 

<w_._.__-._s__ LL _._ s

Temperature Regulating Fluid

 

. :1!
, z 1111'-
W W, i
l 7 z i
_
z , z
I w z
, o z
i H
. k , 7 ,
m , . .
L V
) , 7
1
_ 7 7
_ , v
. 1‘ ii iim
_ rift-
_ A: 1.11-11; 1.11
if} 1.5311: s .
Ilia lirltlrlllllllll. z

Schematic of the temperature regulating channels and bulk flow

Figure 5.2

eXperimental solutions channels in the microscope diffusion chamber.

55

r’t‘s 4 4
_ Video Dyerlay. - 4‘ Monitor ‘.

 

 

 

 

 

 

1"“...1 r :
i V__ _ L.._.__A_.fl._- DJ
L ’ng M‘______ _i— Hm , ,_i A A 7 , J
l
l
l ' """""""""""""""""" ‘: ‘ ........... 1
. _::- ,é:_ as“”* "*-*7
1 ‘ngfi 1 incl“-.. i, i [Y , 4 1
[A _ ‘ * _‘_ * fl i Eqiuilibrativon Bath for l 1 4
3 WV 7 TSolutions ’ * "4 '= 4 1
if e 7.- v74} l__v ,l 1 4
l 4 ‘
Video Camera i 4 ‘
.- a s 7 4 4
l
,l - z z I, l 3
l l i I
_--fl__.___ L _ .QTLJ ___ 4!-.-n
1(é ., ~ , FT, E l Diffusion 1 l
4 W 7 | '75) ‘ Chamber ‘ E
consul“ A —_ ‘C‘ms‘ant i
Temperature if“ ? T ’ “" j ”if“ m lTemperature 1
Bath # l __-4 L .- a; __ “*4; r vBath # 2 4
11 l _,,,_lf.n.,.l 4
‘ : Microscope I l

 

 

 

      

/
Ele ctron ic /’
Solenoid Valves

To Discharge
Beaker

Figure 5.3 Apparatus set-up for acquiring diffusion chamber data at temperatures
other than room temperature. Two refi'igeration baths are used; the first to cool the outer
jacket of the diffusion chamber, and the second to cool the solutions that are flushed
through the optical region of the diffusion chamber.

56

To determine Ea, the activation energy of the membrane permeability, Lp, was
determined at 22, 9.5, and 1 °C. The 22 °C data was acquired with the device and fluids
at ambient temperature with only air in the temperature regulating channels. In the case
of the 9.5 and 1 °C data, the temperature was recorded at the inlet and at the outlet of the
microdiffusion chamber and the average temperature chosen to be representative of the
thermal environment of the cells in the sample chamber. The temperature difference
between the inlet and the outlet was on average 1.5 °C.

Cell suspensions were prepared in the same manner as in the previous section.
The cells were placed in the diffusion chamber and first equilibrated in isotonic saline.
Next, the cells were exposed to a 900 mOsm hypertonic solution by switching the flow
from the isotonic line to the hypertonic line using the solenoid valve switch. The
corresponding osmotic response of the cell over time was recorded. The data were
quantified by post-processing with Image Pro® Plus software as previously described,
except a time course of volume measurements was obtained instead of equilibrium
volumes. Based on a measured flow rate through the chamber an adjustment was made
in the data to account for the lag time for the solution to traverse the short length of
tubing from the solenoid valve to the chamber.

The volume history in the form of radius as a function of time was entered into
DIFFCHAM, a parameter estimation program that implements the Kedem-Katchalsky
(K-K) model of passive coupled transport across the plasma membrane (Kedem and

Katchalsky, 1958; McGrath, 1992). In the absence of a permeating cryoprotectant the K-
K model reduces to equation 5.1. The DIFFCHAM program has a built-in routine that

compensates for the lag in solute diffusion across the dialysis membrane. The Box-

57

Kanemasu method of parameter estimation was utilized together with the Gauss method
of linearization and ordinary least squares. The Lp coefficient was determined by

minimization of the sum of squares:

sir-met

where Y. is the volume history datum at time i, and m the corresponding predicted cell

volume at that time, which is based on the current estimate of parameters, given by the

vector B. To minimize the sum of squares, the following conditions must be

simultaneously satisfied for p, the number of parameters being estimated:

3%— = O [5.18]
68k
where k=l to p.

Values of 1.1, obtained at these different temperatures were then plotted as a
function of inverse temperature, and ELp was determined from a linear fit of this data
according to equation 5.13. The slope of this line provided the activation energy.

To ensure that the use of trypsin was not affecting the measured permeability
Values, a series of measurements was made using cells that had been mechanically

removed from the flask. When the cells had grown to nearly a monolayer, the medium

Was removed and the flask surface was gently scraped with a sterile disposable cell

58

scraper (Costar, Cambridge, MA). The detached cells were then collected by adding
culture media and the cells were transferred by pipette to a centrifugation tube. The rest
of the procedure was identical to the trypsin protocol. Because some mechanical damage
of cells was expected, cells were evaluated for the presence of OF P fluorescence prior to
analysis. A GFP positive cell was considered to have an intact and normal cell
membrane as will be validated in Chapter 6. Only GFP positive cells were used for

experimentation.

5.3.3 MathematicaTM Programs
5.3.3.1 Validation of Diffcham Results

Because the results of the Lp measurements were unexpected, a check of
DIFFCHAM results was performed. Assuming that the sample area chemical profile
predicted by DIFFCHAM was correct, an exponential fit to this extracellular
concentration history was obtained and implemented in a MathematicaTM program,
Difi‘cham _Check. This simple Mathematica program models cell volume response when
provided with concentration histories and parameters. This program is given in

Appendix A.

5.3.3.2 Simulation of Water Transport During Freezing
With knowledge of the values L1) and E. for the GFP-R3230AC cell line, it was
possible to predict cell volume change as a fimction of cooling rate according to
Equations 5.13-5.14. A MathematicaTM program was written to simulate the cell volume

change at a variety of cooling rates. This program, Vc _Freeze, is given in Appendix B.

59

5.4 Results

5.4.1 Boyle Van’t Hoff Plot

A Boyle Van’t Hoff plot was constructed from the equilibrium cell volumes at
concentrations ranging from 308 to 1200 mOsm, as shown in Figure 5.4. Each data point
represents the average of 13 cells and 3 separate experiments. The y-intercept yields an
osmotically inactive volume of 32.7% of the isotonic volume. The correlation was high

(R2=0.9832) supporting the validity of results.

 

  
 

 

 

 

 

1Inorm Osm

T 1
1 1.2 1
‘ 1
| 1
l 1 l
I 1
. c 1
g 0.8 « 1
g 1

“u: 0.6 - y = 0.3982x + 0.327 5

E R = 0.9832 1

1

73 0.4 « ‘
> 1

1 0.2 i
1 1
1 1
1 0 * A 1
1 0 0.2 0.4 0.6 0.8 1 1.2 1
j 1
1 1
1 1

Figure 5.4 Boyle Van’t Hoff plot. Errors bars represent standard error of the mean.

60

5.4.2 Estimates of L9 and E14,

The cell volume curves obtained for the purposes of estimating Lp did not all fit
the simplified K-K model implemented in DIFFCHAM. Data were removed from the
final data set on the basis of: (1) blatant deviatory behavior from the expected volume
curve, and (2) inability of the program to fit the data to the model. Deviatory behavior
was characterized by what appeared to be overshoot and re-swelling behavior near the
equilibrium volume. An example of what is considered a good fit to the data is shown in
Figure 5.5. In contrast, a poor fit of data classified as ‘deviatory’ and not included in
average Lp results is shown in Figure 5.6. Approximately 50% of the data sets had to be

removed due to deviatory behavior or inability of the program to fit the data.

Normalized Volume History
Sample Data

*Expoata
-81m0au

 

 

 

 

 

K: 107.003 rel-11.760733 Waco)
Vb: 20.000 “-1 350421 (percent)
Vo: 10000.12! +4.0.oooooo (un')

figure 5.5 An example of a good fit of experimental data to the biophysical model.

61

Normalized Volume History
Sample Data

Normalhed Volume

 

 

K: 1oooo.ooo +1.1 27344539441 (my
Vb: 30.323 dam (percent)
Vo: 190nm +I-0.000000 (mm)

Figure 5.6 An example of a poor fit of experimental data to the biophysical model.
Samples with volume histories characterized by a rising volume at the end were excluded
from the data set.

To understand if the use of trypsin could change the basic osmotic response of
cells, the averaged volume histories and Lp estimates were compared to those of cells that
had been prepared for analysis by mechanical scraping instead of trypsin exposure. As
can be seen in Figure 5.7, at 22 °C the cell volume histories were similar. Statistically

identical estimates of LD were obtained for these cells as well, as can be seen in Table 5.1.

62

{ ........ , ......... ...................... ................ t

1.2 F

l
| _‘-_,,______._.,_,
l
l

- l a Scraped, n=3 , -
lflIEXQsjnl3§$ Eli-'9 7i

 

2‘ £

.0
on
1
1

Normalized Volume

.0
03

{d

l

I 1

l

l l

t ,

l 1

i. ;
I
'-
J

 

 

 

 

0.4 .....................................
0.2 7777777777777777777777777777777777777777777
|
l o f
1

o 10 20 3o 40 50 60 7o 80 90 l
l

Figure 5.7 Volume histories of cells prepared for analysis by trypsin exposure
(triangles) and by mechanical scraping (squares) following exposure to hypertonic
solution at 22 °C.

Table 5.1 Lp Estimates (um/s). (T) and (S) referred to trypsinized cells and scraped

 

 

 

cells, respectively.
Temperature (°C) n Average Lp SD SEM
1 (T) 6 164 283 115
9.5 (T) 10 197 158 50
22 (T) 9 302 274 91
22 (S) 3 305 215 124

 

63

The Lp values determined for each temperature are shown in Table 5.1 and are
also plotted as a function of inverse temperature in Figure 5.8. A linear regression was
used to determine the activation energy from this plot. An ELp value of 4.2 kcal/mol was
obtained. It is important to note that the temperature dependence cannot be considered
statistically significant due to the large standard deviations of Lp values. This value of
EM, is consistent with the presence of aquaporins in the cell membrane, as will be

discussed shortly.

 

 

N
l

y _= -21fi21.5x#+ 9.7771 g

F k2 é 0.9751

 

Ln Lp (um/s)
3

A

”WW” EéEUVCFailTrfioil F F *1 if f * ST ’7! 7”“, 7W“ i

.0
01
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
L

 

 

0 e. . - .
0.00335 0.0034 0.00345 0.0035 0.00355 0.0036 0.00365 0.0037

1IT ‘

 

,.__—. #A4 —fi4___.7 ___ -__ ____._7 tin—_—

Figure 5.8 Lp as a fimction of temperature. E”, is estimated to be 4.2 kcal/mol. Error
bars are not shown on the plot but are on the order of the mean.

5.4.3 DIFFCHAM Check

The average volume history of trypsinized cell data (22 °C) used to estimate
parameters in DIFFCHAM was plotted together with the simulated results of
Diffcham_Check. The predicted local extracellular concentration history and the
estimated parameters generated in DIFFCHAM were used in Diffcham_Check. The
predicted results from Diffcham_Check were consistent with the original data used to
estimate the parameters. Sample results are shown in Figure 5.9. This agreement
suggests that DIF F CHAM was providing acceptable estimates of the parameters. It does
however assume that the local concentration history in the sample chamber provided by

the DIF F CHAM program is accurate.

 

 

1",
0.9 77"”.7..7 ******************
0.8777 e Ii”! ‘ rrrrrrrrrrrrrrrr
0.7 7 77777 men’rwfi’"! III.
0.6 iiiiiiiiiii 77 777777 ’
0.51 777777777 77* *7 777777777777777
0.477777777777 7'
0.3 77e’77’rrririrr ,,,{

0.2 777777777777777 t . :7
0_1,,,,,,,,,,,_ Mil-Experimental).

Normalized Volume

 

 

 

0 5 10 1 5 20 25 30
Time(s)

 

 

 

Figure 5.9 Volume histories of trypsinized cells (22 °C) that were used to estimate
parameters in DIFFCHAM are plotted together with simulated results from an
independent program implementing the DIFFCHAM parameter estimates. The results
are consistent, indicating that the DIFFCHAM estimation program is providing accurate
estimates.

65

 

th

m;

5.4.4 Water Transport During Freezing

The predicted normalized water volume was plotted as a function of temperature
and cooling rate in Figure 5.10. As can be seen in the figure, even at cooling rates as
high as 100 °C/min there is very little deviation from the equilibrium curve, suggesting
that intracellular ice formation would be unlikely at any rate equal to or less than this.
This is further illustrated in Figure 5.11. Using the volume-concentration relationship in
the Boyle van’t Hoff plot shown in Figure 5.4, the static intracellular concentration can
be estimated from the predicted volume curves in Figure 5.10. This osmolarity
information, together with the phase change relationship Tr = -1.858 x Osm (for NaCl),
allows the equilibrium phase change temperature to be estimated from the predicted cell
volume. This can then be used to compute the amount of under-cooling, or deviation
from equilibrium, at a given temperature and freezing rate. The under-cooling represents
the difference between the actual temperature and the equilibrium freezing temperature
for a given cytoplasmic concentration. The probability of HF will increase as the amount
of under—cooling increases. The results in Figure 5.11 suggest that cooling rates of at
least 500 °C/min are necessary to achieve more than a few degrees of undercooling and
thereby increase the likelihood of intracellular ice formation. This is similar to the
freezing rates predicted by Toner et al. (1992) for hepatocytes, also reported to have a

high L, value (1%: 1.5 x10'12 m3/Ns).

66

 

    
 

1mmcmm

(15 -

Normalized Volume

 

(l4

 

 

 

l

l

1

l 0.3 L

. l
l 273 271 269 267 265 263 i
L i
1

Temperature

 

l ,,,,,,,,,,,,,, _m _ _ i v 7 v 77,, , if , ,7 W i , W 7

Figure 5.10 Predicted temperature dependence of normalized water volume as a
function of cooling rate for R3230AC cells. Cooling rates are indicated on each curve.
The dashed line represents the equilibrium response.

 

 

 

 

 

l 6 .
l l
l 1000 Clmin ;
l 5 l
l- i
‘9, l
O 4 l
._._=. 500 l
o
3 3 r
h l
a 2 a
5 l
l
100 l
i 1 t l
l I
l l
l O .
{ 1 -1 -3 -5 -7 -9 i
l Temperature (C) i
L—__flflz_n._ufl_nu_ , _ ________ #J

 

Figure 5.11 Predicted cytoplasmic undercooling of R3230AC cells. Cooling rates are
indicated on each curve.

67

5.5 Discussion

Approximately 50% of the volume histories curves obtained did not fit the K-K
model of water transport. The non-K-K behavior could indicate that cell membrane
damage may be present, however the basic appearance of the cell and its ability to
respond osmotically when returned to isotonicity suggested otherwise. It would be
desirable to explore the behavior of these deviatory cells to understand if the observations
are artifactual or if they are representative of phenomenological behavior of a portion of
the population. If the latter case proves to be true, it is important to understand this
behavior in the context of cryo-sensitivity.

The volume response of scraped cells was seen to be more erratic than in the
case of trypsinized cells. This may be a phenomenon related to the method of cell
preparation or it may be accounted for by the different cell numbers included in the
averages, with the trypsinized cell response more effectively ‘smoothed’ by the larger
data set.

The water permeability value estimated for R32230 AC cells is quite high when
compared to many of the published [1, values for other cell types. Other cell types
characterized as having high water permeabilities include red blood cells (Terwillinger
and Soloman, 1981), tumor ascites cells (Hempling, 1977), human myelogenous
leukemia K562 cells (Martinez de Maranon et al., 1997), and Saccharomyces cerevisiae
yeast (Martinez de Maranon et al., 1997). Many of these high permeabilities have been
associated with the presence of aquaporin water channels (Martinez de Maranon et al.,

1997)

68

Aquaporins are a class of trans-membrane water selective transporter proteins
that facilitate the rapid transport of water both across cells and across vascular structures
by causing the formation of channels in the cell membrane (Agre et al., 1995).
Aquaporins are widely expressed in epithelia and endothelia that have a fluid-
transporting physiologic role (Chou et al., 1999). The presence of aquaporin in cell
membranes tends to correlate with high cell membrane or transepithelial water
permeability values. For example, Chou et al. (1998) demonstrated a fourfold reduction
in trans—epithelial water permeability in the inner medullary collecting duct of aquaporin-
4 knockout mice. However, there are many tissues that have not yet been examined for
the presence or localization of water channels, and pathological tissues have been studied
even less. The immunoblot and immunofluorescence study by Endo et al. (1999) was the
first to examine Aquaporin-l (AQPl) expression in different types of pathologically
altered tissues. They found that AQPl was not expressed in the cell membranes of
normal epithelial cells but was expressed in the cell membranes of tumor cells to varying
degrees. A dependence of AQP] expression on location of the tumor in the animal
(cranial or subcutaneous) was found for two out of three of the glioblastomas that were
investigated. The protein AQPl was found to be expressed in the mammary carcinomas
they studied, which included the R3230AC cells used in this work. The presence of
aquaporins explains the high permeability values measured for R3230AC cells as well as
the magnitude of the activation energy. Cells with water channels are known to have low
activation energies, due to the insensitivity of water channel transport to temperature

(Kleinhans, 1998).

69

The bulk of studies to characterize the water permeability of biological cells has
been performed for a limited range above 0 °C. However when using this information for
freezing studies it is necessary to extrapolate these data to sub-zero temperatures. There
are many potential dangers in doing so. The activation energy for Lp above and below
zero can be markedly different. For example, changes in water conductivity due to phase
changes in membranes at sub-zero temperatures or due to increasing solute
concentrations, desorption of membrane (Pegg, 1988), or formation of external ice may
result in substantial polarization of the membrane proteins and lipids (Toner et al., 1990).
The complexity introduced by the behavior of aquaporin water channels as the phase
change temperature is traversed is also not well-studied, further complicating the
extrapolation of data from above zero to sub-zero. Kleinhans (1998) has indicated that
the temperature dependence of Lp for aquaporin-containing cells would be expected to be
non-linear due to the different temperature dependences of membrane diffusion and water

channel diffusion that give rise to the overall effect.

5.6 Conclusions

The high membrane water permeabilities measured for the R3230AC cell line
suggest that cooling rates on the order of 500 °C/min will be necessary to induce the
formation of intracellular ice in these cells. The conclusions drawn are meant to be a
general guideline for future experimental design, not a definitive indicator of when
particular modes of injury are expected. Further cryomicroscopy studies are necessary to
determine phenomenological parameters that can be used in the HF probability models of

Toner (1990).

70

Chapter 6

Validation of a GFP-Based Viability
Assay at the Cellular Level

6.1 Assessment of Injury in R3230 AC Cells

6.1.1 Comparison of GFP-based Imaging to Standard Viability Assays Following
Freezing

Cells transfected with GFP and cells without the GF P vector were prepared to a
concentration of 1 x 105 cells/ml in D-MEM culture media. Aliquots of 220 p1 were
added to 5 x 50 mm plastic vials (Corning, NY), and placed in a programmable
refrigeration bath (Neslab RTE-140) at -5°C for 2 minutes. Ice formation was initiated
by touching the solution with a cold probe, and a 1-min period allowed for the latent heat
of fusion to dissipate. A convenient range of freezing damage was produced in both of
these cell lines by freezing from —5°C to minimum temperatures between -10°C and —
30°C at 1°C/min. When the bath temperature reached the desired minimum temperature,
samples were held in the bath for 10 seconds. The samples were then rapidly thawed by

immersion and agitation in a 37°C water bath. After thawing, 20 pl of the cell

suspension were added to 20 pl of Trypan Blue solution (0.1% in PBS) and a portion of

71

nu
po
six
rcr.
om
Brt
add

108

COL”

Com

Hum
fillm
lite 1'

0133,.

new,

the solution was then placed in a hemacytometer. Viability was assessed by counting the
number of cells excluding Trypan Blue approximately ten minutes afier thaw. Data
points represent the average of 3 replicates, each replicate representing the average of
sixteen 1 x 10'4 ml volume hemacytometer well counts (~80 cells per replicate). The
remaining 200 pl of cell suspension (for transfected R3230 AC cells only) were added to
one well of a 96 well tissue-culture microplate and either 20 pl of 1.25 mM Ethidium
Bromide in D-MEM media (EtBr assay) or 20 pl of DMEM media (GFP assay) was
added. Following a 10-15 minute period of incubation, the microplates were transferred
to a Nikon Diaphot inverted fluorescent microscope equipped with a 100 W mercury arc
lamp. The wells were illuminated with blue light (450-490 nm) by inserting a FITC filter
cube (Chroma Technology Corp., Brattleboro, VT) into the light path. Seven color
digital images were captured from approximately the same regions within each well using
a SPOT thermoelectric cooled color CCD camera (Diagnostic Instruments, Inc., Sterling
Heights, MI). It took approximately 10 minutes to locate each assay region, focus the
frame of interest, and obtain a 3-channel RGB image of each test area. The average cell
count per control well image was 133 (absolute range: 86-253 cells) therefore each
control replicate represented the average of approximately 1000 cells. Images were later
analyzed using Image Pro software (Media Cybernetics, Silver Spring, MD). The total
number of fluorescent green cells was counted in the GFP assay, and the total number of
fluorescent green, EtBr negative (non-red) cells was counted in the EtBr assay. Although
the number of EtBr stained cells was not explicitly counted, control wells were generally
observed to have less than 3% EtBr positive cells and this percentage increased with

decreasing minimum temperature. A color-cube-based counting method, which

72

identifies all cells within the color range selected by the user was utilized. In this case the
cells with the brightest and weakest GFP fluorescence were used to set the color limits.
In all assays, cell counts were normalized to non-frozen control counts. Each assay point
represents the average of 3 replicates.

The recovery of cells frozen at 1°C/minute to various end temperatures, as
assessed by the three assays, is shown in Figure 6.1. A freeze damage curve was
obtained, with recovery decreasing as the minimum temperature decreased. For all
minimum temperatures there was no statistical difference between the three methods used
to assess recovery (Students t-test, p=0.05). It was assumed that in all cases the sample
parameter was normally distributed. A linear regression analysis of GFP viability as a
function of ethidium bromide viability yielded a correlation coefficient of 0.95
(y=1.1271x). A similar correlation coefficient of 0.97 was obtained for a linear
regression analysis of GP P viability as a function of Trypan Blue viability (y=l .0628x).

Figure 6.2 illustrates the change in GFP fluorescence intensity of cells after
freezing to different end temperatures. The total number of cells in each template was
approximately equal. The GFP intensity in the damaged cells is so low that the cells are
undetectable without counterstaining or use of phase-contrast microscopy.

The response of non-transfected R3230AC cells, as assessed by Trypan Blue
exclusion (n=3), was comparable to that of transfected cells, as show in Figure 6.3. A
correlation of recovery of transfected cells as a function of recovery of wild-type cells
yielded a high correlation coefficient (R= 0.98) and a nearly perfectly linear relationship
(slope= 0.95). This suggests that, as expected, the presence of GFP does not alter the

basic freeze-thaw response.

73

 

 

 

 

 

 

! 09 l . l 1 0.8 4. +f—‘I—l i i

' A “'T " 1&06 ' . I

0.8 i ‘ l . GFP : i ““504 .,/7/ 1 1

8 l I 9 , [ lEtBr ‘ l 0.2 i ‘

‘- 071 l 4!": __ _______, l

a) ' ' ATrypan I l 0 T T 3 ‘ l

> . L _ _ _ }

8 0.6-1 1 l o 0.2 0.4 0.6 0.8 1 ‘

”05' . i ’—--.7-.._E.tBi-.__- J

S 0.4 l 1 h

z ‘ l
O 0.3 1 ‘

ls . I w - . - - - , _ -

“1.0.2; . l 1 1r““’*“ _g, ,7 i

l f l 0.8 , .'

l 0.1 . ! n. / l

0.6 ,,

l U. ,/’ I

1 o +— ; ~ ». - g -, »— . _- | 00,4 {1 . l

‘ o -5 -1o -15 -20 -25 -30 I 0.2 lg... J t

I l O . -‘r ‘——r- *' 7r— — r l

L W ,,,, 7* __End Temperature (C) _ - if 1 o 0.2 0.4 0.6 0.8 1 l

i _ _ __T_'YB?£§_'“° _ _ _é

a c

Figure 6.1 (a) Recovery of cells frozen at 1 °C/min to various minimum
temperatures, as assessed by GFP fluorescence, OF P Fluorescence with EtBr co-stain,
and Trypan Blue Exclusion. (b) Correlation of recovery assessed by GFP fluorescence
with recovery assessed by EtBr. (c) Correlation of recovery assessed by GFP
fluorescence with recovery assessed by Trypan Blue. Data represent average of three
replicates +/- one standard error of the mean.

74

 

tan
‘b).

a b
c (I

Figure 6.2 Fluorescent images of GFP-transfected R3230 AC cells after freezing to
various minimum temperatures and thawing in a 37°C water bath: (a) unfrozen control
(b) —10°C (c) —20°C (d) —30°C.

75

 

1.4

 

1.2
y = 0.9511x “

14 W=0%&

    

0.8

 

0.6 1

Transfected

0.4 .
0.2 ~

 

 

0 - - a ,
0.00 0.20 0.40 0.60 0.80 1.00

\MMWm

 

 

 

 

Figure 6.3 Correlation of post-thaw recovery of GFP-transfected and wild-type
R3230AC cells frozen at 1 °C/min to various minimum temperatures. Recovery was
assessed by Trypan Blue exclusion.

Good agreement between GFP and conventional membrane integrity indicators
was obtained as is evident from the correlations (R = 0.95 EtBr; R = 0.97, Trypan Blue)
for fi'eeze injury at slow fi'eezing rates. Although the slope of the correlation curves was
greater than 1 in both cases, the deviation from the ideal fit is thought to be a dynamic
effect, resulting from a slight variation in the time for GFP to diffuse into and ethidium
bromide out of the cells. The water-soluble cylindrical GFP monomer is 3 nm in diameter
and 4 nm long (Yang et al. , 1996). Calculations based on the proposed structure of wild-
type GFP yield a molecular weight of 26,862 g/mol (Inouye et al., 1994). Because GFP
is a much larger molecule than ethidium bromide (MW: 394.3 g/mol) or Trypan Blue
(MW: 960.8 g/mol) it can be expected to diffuse more slowly out of the cell. This
combination of diffusion dynamics can give rise to small differences in recovery at a

given assay time. Recovery measured with the GFP assay would be expected to be

76

higher than recovery with EtBr or Trypan Blue assay at a specific point in time, giving a
slope greater than 1 in the regression analysis, as observed.

Image-Pro analysis was performed in a way such that the threshold for distinction
of GFP-positive cells was chosen as 1 color index (of 10 possible) above background.
This was a very conservative measure and included as viable, cells with fluorescence
diminished by as much as 90%. A less stringent threshold (eg. 50% of average starting
fluorescence) would have resulted in unifome lower values of recovery. Alternatively,
the threshold value could be specifically determined by fitting the GFP recovery data to
the recovery data of another viability indicator, such as ethidium bromide or Trypan Blue
as used in this study. From a statistical standpoint however, these assays were found to
yield equivalent results (Students t-test, p=0.05) and therefore the threshold value of 1
color index above background was maintained throughout all experiments.

Some semi-quantitative observations were made over the course of these
experiments that also support the use of GP P as a viability indicator. No intracellular
GFP fluorescence was observed following rapid fi'eezing by plunge freezing
(~500°C/min) in liquid nitrogen (-196°C) and slow thawing (~10°C/min). The EtBr
assay also revealed 0% viability for these conditions, suggesting that the GFP assay will
be a valid viability assay at rapid freezing rates as well. Overgrowth of cells or lack of
fresh media can give rise to nutritional stresses that eventually lead to cell death. When
these circumstances were allowed to occur, the GFP fluorescence was found to decay in
time as long as these conditions were allowed to persist. At various stages ethidium
bromide was added to the flasks and the relative fraction of EtBr(+) cells was

qualitatively comparable to the disappearance of GFP fluorescence.

77

6.1.2 Comparison of GFP-based Imaging to Standard Viability Assays Following
Solute Injury

24-well flat bottom tissue-culture-treated polystyrene cell wells (Corning, NY)
were plated at a density of 1 x 104 cells/mL and incubated at 37°C for a minimum of 24
hours. For the GFP assay the old media was removed and 1 mL of fresh media added. In
the ethidium bromide assay, 1 mL of media with ethidium bromide (2.5 pM) was added.
This solution was prepared fresh daily. In both assays a region of non-confluent colonies
was selected and an image acquired with a Spot digital color camera. The media was
then removed and 1.6M or 2.7M NaCl solution added to the well for a total exposure time
of 30 seconds. The NaCl solution was then removed and fresh media (GFP assay) or
media with ethidium bromide (EtBr Assay) added to the well. Images were acquired at
15, 30, 60, and 90 minutes following return to isotonicity. All assays were performed
between 24 and 60 hours afier plating, and all solutions were at room temperature.
Images were later analyzed using Image Pro Software. In the GF P assay the total number
of green fluorescent cells was counted. In the ethidium bromide assay the number of
green fluorescent, ethidium bromide negative (non-red) cells were counted. Cell counts
were normalized to the starting cell counts to determine the fiaction recovered in plated
colonies.

Figure 6.4 gives the viability response following exposure of cells to two different
concentrations of NaCl solution, as assayed with GF P fluorescence and with ethidium
bromide. There are no statistically significant differences between the two assays in
time, or with solute strength. The standard deviation of the data for the 2.7M
experimental series is substantially higher than the 1.6M series. This is thought to be

attributed to the sensitivity of the experiment to human error involved in delivering and

78

removing the hypertonic solution. As more damage is expected at higher concentrations,
error in the time of exposure (estimated at +/- 6%) will result in larger absolute standard
deviations under these conditions. Mechanical delivery/removal of solutions would be
necessary to improve upon the accuracy of these results. Because the assays tracked
reasonably well despite these limitations, an extension of the 2.7 M experiments was not

pursued.

 

 

0.70 I in.

,— .fi. ‘ I i 7 i .7 *7 ,_. 1 _ .

I i g A I
! 090 2-..-2-_e~-._ .__., 2,- _ _.._,#_ ___ _ .*—— I
I 0.80 .0- i «— — ~~--— ; 7—i—7 }1-5M I
I
‘I

Fraction Recovered
p o o
T I

I
I

I
l I

I
I

I
I I
I
I

I
I I
I I
I .

I
I

I
I I
I I
I I
I
I
I
I
I
I
I

.0
TI

I

I

I

I

I

I

I

I

. E

I

I

I

I

I

I

 

 

I 0.20 I____ .... -. .. _ . _ _ _ _.._. _ —— — — -—» —
I

 

0.10~e—-~——#-~—~fi ~ 7 7 a _ W 3 3 _-_--..__._ ,1 ,

 

 

0.00

 

I
I
I

0 20 40 60 80 100
. Time Poet Exposure (min)

L---.___..__.,,_..__ _. - _ _4 _,._ "i _ ____ _. i ___k _EJ

 

Figure 6.4 Viability of GFP-transfected R3230AC cells following 30 3 exposure to
1.6M and 2.7M NaCl.

79

6.1.3 Clonogenic Assay

Practical experience with OF P transfected cells in culture has revealed that plated
cells that have been stressed by overgrth and lack of fresh media will eventually lose
their fluorescence. Cells that reach this stage have not survived subsequent passage
and/or supplementation with fresh media. To understand if this observation was also
consistent with freezing-induced loss of GFP fluorescence, cells that were predicted to be
dead on the basis of OF P assay were subsequently cultured to test for positive cell
growth.

One ml of cell suspension (~ 1 x 105 cells) was placed in six 1.5 ml eppendorf
tubes. Three tubes were held at room temperature for 30 minutes and three were
immersed in liquid nitrogen for 2.5 minutes followed by passive thawing at room
temperature. Three cycles of freezing and thawing were performed on the contents of
these vials. The contents of each vial were then added to a separate T-50 culture flask
and 9 ml of D-MEM added. Prior to culture, images were acquired of all flasks using
both phase contrast and epi-illumination fluorescence microscopy. Although cells could
be identified in the frozen-thawed samples using phase contrast microscopy, they did not
have the appearance of control cells, as can be seen in Figure 6.5. Furthermore,
fluorescence microscopy revealed dim fluorescence levels in only a few cells in the entire
flask, suggesting the population was irreversibly damaged. All cells were then cultured
under normal conditions for 3 days (normal passage schedule). On day 3 flasks were
removed from the incubator and imaged in the same manner as day 0 cells. The flasks
containing the frozen-thawed cells had no identifiable green fluorescing cells and phase

contrast microscopy did not reveal any cell growth. Control flasks had reached

80

confluency. Flasks containing the frozen-thawed cells were cultured for three more days

with no growth observed.

Day 0 Day 3

 

(b) Frozen —Thawed Cells

Figure 6.5 Images of cells in culture following different conditions. (a) control cells
were held at room temperature for 30 minutes then cultured under normal conditions. (b)
test cells were frozen in liquid nitrogen followed by passive thawing at room temperature
for three cycles prior to normal culture. The left image of each pair: phase contrast
microscopy, right image: epi-illumination fluorescence microscopy.

This experiment demonstrated that for the freezing conditions examined, when
GFP fluorescence was lost from cells, the cells were irreversibly damaged. To date, it
has not been possible to sub-lethally damage a cell, with loss of GFP. Although the

results support the hypothesis that GFP fluorescence loss is indicative of cell death, it

81

does not definitively answer the question of whether it is possible that GFP loss can occur
without cell death. Such an experimental condition is difficult to contrive using freezing
as the modality of injury. However, because of the lack of any results to suggest
otherwise, it is reasonable to assume this is not a high probability occurrence in the GFP

assay.

6.1.4 Photobleaching and Photodamage

Although GFP is known to be resistant to photo-bleaching (Cubitt et al., 1995)
this effect needs to be evaluated in the context of each system of study. Additionally,
because one of the goals of developing this assay is to provide a means of continuously
monitoring cells and tissue, the potentially damaging effects of photo-absorption warrant
consideration. For example, Miller and Silverston (1979) were able to kill neurons
within a few minutes by pre-loading them with Lucifer Yellow CH and irradiating with
blue light from a fluorescent microscope. Although bright GF P fluorescence is desirable
for most applications, the capacity to absorb exciting energy can cause damage to system
of study if the exposure time is excessive. The limiting total exposure time before the
irreversible photo damage occurs is therefore a parameter of interest.

To assess the effects of continuous exposure on the intensity of GFP fluorescence,
as well as on the viability of the cells, 24-well flat bottom tissue-culture-treated
polystyrene plates were plated at a density of 1 X 104 cells/ml. After 24 hours of
incubation at 37°C, a population of plated cells was continuously exposed to blue light
with the shutter fully open, and digital images were taken of this same population at

regular intervals over a time span of 40 minutes. An equivalent assay was performed,

82

with 2.5 pM ethidium bromide present in the media. Four replicates of the EtBr assay
were performed. Images were later analyzed using Image Pro Software. In the GFP
assay the total number of green fluorescent cells was counted. In the EtBr assay the
number of green fluorescent, ethidium bromide negative (non-red) cells were counted.
Cell counts were normalized to the starting cell counts to determine the fraction
recovered in plated colonies.

Because GFP is thought to leak out of cells with a damaged membrane, to assess
photo-bleaching it is necessary to evaluate fluorescent intensity in cells that are not
membrane-compromised. The ethidium bromide assay images that were captured at 40
minutes were examined to isolate specific cells that remained viable throughout the test
period. In each test series 3-5 fluorescing cells were selected in the 40 minute images
and identified and marked in all of the preceding images using Image Pro software. All
images were then converted to gray-scale and the gray-scale intensities of fluorescence of
these specific cells were determined as a function of time.

At full illumination (shutter completely open) with blue light, the critical limit
before detectable damage accrued was 15 minutes (3% of cells ethidium bromide
positive). Significant damage (>10% of cell ethidium bromide positive) was observed
beyond 30 minutes. Controls, which were not illuminated, but were incubated at room
temperature, did not show any evidence of damage at the end of the 40-minute period,
and also displayed bright fluorescence.

Normalized gray scale intensity measurements are shown in Table 6.1 as a
function of time for cells that remained viable afier 40 minutes of exposure. Data beyond

30 minutes was not reported due to the possibility of photodamage artifacts in the

83

intensity data. A steady decrease in fluorescent intensity over time can be observed for
these cells. At 20 minutes the fluorescent intensity has decreased by 15%. This is
comparable to the approximate 20% decrease in fluorescent emission at 510 nm
measured with a monochromator by Cubitt et al. (1995) when the S65T mutant of OF P
was irradiated at 280 nm with light from a xenon lamp.

Although it can be difficult to separate photobleaching effects from leakage due to
photodamage, some degree of distinction was achieved by measuring the decrease in
fluorescence intensity for cells that remained viable after 40 minutes of exposure. A
steady but slight decrease in fluorescent intensity was observed for these cells over the
time interval assayed, indicating that the fluorescence is quite stable for reasonably long
exposure times (1-5 minutes). It is possible that this decrease in intensity is due to slight
leakage of cytoplasmic GFP caused by damage to the membrane that has not fully
manifested itself yet. Given the relative size of the viability markers however, red
fluorescence from ethidium bromide would likely be detected before significant loss of
fluorescence due to diffusion of the much larger OF P molecule. To avoid photo-induced
damage, cells assayed in this work were exposed to blue light illumination for the

minimum possible times.

84

Table 6.1 GFP fluorescence intensity as a function of exposure time for a select
population of plated GFP-transfected R3230 AC cells. Data represent averages of four
replicates.

 

 

Normalized

Time Fluorescent
(min) Intensity SEM
0 1.000 0.000
1 0.980 0.019
2 0.987 0.007
3 0.988 0.009
4 0.986 0.015
5 0.983 0.017
6 0.970 0.017
7 0.968 0.015
8 0.964 0.024
9 0.970 0.026
10 0.964 0.015
15 0.928 0.020
20 0.848 0.018
30 0.783 0.012

 

6.2 Characterization of the GFP Assay

The loss of GFP fluorescence following exposure of cells to subzero temperatures
was rapid in cell suspensions, with dramatic decreases in intensity observed directly after
freezing and thawing. It was not immediately obvious whether or not this loss of
fluorescence was due to destruction of the chromophore and/or denaturation of the
protein, or if it is was simply a leakage and dilution phenomenon. The results of section
6.1 suggest that the GFP may be functioning as a membrane integrity indicator because
of the close agreement of damage predictions as assessed by other viability dyes that fall
into this category. However, a more systematic approach to understanding the source of

GFP fluorescence loss after injury was sought.

85

Three testable hypotheses were identified that could possibly explain the source
of GFP fluorescence loss following exposure of cells to lethal subzero temperatures.
These were:

(1) shutdown of protein synthesis
(2) protein denaturation
(3) leakage of OF P from the damaged membrane
Each of these hypotheses was evaluated in an attempt to develop a mechanistic

understanding of the basis of the GFP assay.

6.2.1 Intracellular GF P Half-life

Proteins are not immortal. Each protein has a characteristic rate of proteolytic
degradation into constituent amino acids for a given set of physiological circumstances
(Creighton, 1996). This turnover rate can vary widely for different proteins. For
example, omithine decarboxylase in the rat liver has a half-life of 11-min. By contrast,
actin in rabbit skeletal muscle has a half-life of 50-days (Kay, 1978). Because the GFP
molecule is extremely robust to thermal denaturation, temperature extremes, and pH
extremes (see Chapter 4), it seems likely that the GFP molecule would have a relatively
long turnover rate. Determination of the OF P half-life in R3230 AC cells is important in
terms of establishing whether protein synthesis shutdown can contribute to fluorescence
loss immediately following fieezing, but also in determining the limits and maximizing
the utility of a GFP viability assay.

To determine the half-life of GFP in R3230 AC cells, cells were incubated in

cycloheximide, a protein synthesis inhibitor. When using a protein synthesis inhibitor to

86

determine the rate of OF P turnover, it is essential to differentiate cytoplasmic GFP loss as
a consequence of cell death and possible leakage, with true turnover of GFP resulting
from degradation or metabolism of the protein. To do this it was important to establish
the integrity of the cell membrane with an indicator other than GFP fluorescence
intensity. Ethidium bromide was used to identify a membrane breach.

Twenty-four well plates were plated at a density of 1 x 104 cells/ml two days prior
to use and incubated at 37°C. At the time of experiment the cells were washed twice with
D-MEM media then 900 pl of fresh media was added. A 100-pl aliquot of
cycloheximide solution (10 mg/ml) was added to each test well. The control wells were
supplemented with 100 pl of media. Plates were then incubated at 37°C for 0, 2, 6, l6,
and 24 hours. Prior to imaging the phenol red containing media was removed and
replaced with colorless media. Images were acquired using the Spot digital camera, and
a 20x objective. Seven images per well were obtained. Characteristic results are shown
in Figure 6.6. It can be seen that the control cells (without inhibitor) are growing
normally, whereas the cycloheximide-incubated cells do not double as rapidly, and at 16
hours are showing signs of cell death. The OF P fluorescence appeared more diffuse and

the location of the boundaries of the cells were more ambiguous.

87

   
  

Control

C cloheximide

0 Hours 6 Hours 16 Hours 24 Hours
Figure 6.6 Plated R3230AC cells were incubated in cycloheximide, a protein
synthesis inhibitor and assayed at several time points in a 24-h period. Control cells
(incubated without inhibitor) appear normal whereas the cells in the cycloheximide-

incubated group were fewer in number and appeared slightly swollen with indistinct cell
boundaries (potentially due to leaky membranes) at the 24-h point.

Images were converted to 16 bit gray scale using Image Pro software. Intensity
thresholding tools were utilized to create a photo-mask that fixed all background pixels at
zero intensity. Histogram data of the entire field were then obtained to determine the
cumulative fluorescent intensity of the cell population. This value was then divided by
the area of the un-masked area (the cells), to yield a value of gray scale intensity per sq.
pixel for the population. This procedure was reversed to determine a value for the
background intensity.

A complementary set of experiments was performed to establish the effect of long
cycloheximide incubation times on the viability of the cell. The inhibition of protein
synthesis by cycloheximide will certainly affect the vitality of the cell at some stage, as
proteins critical for the integrity and normal functioning of the cell will no longer be

manufactured. Cells were plated as described above, with identical experimental design.

After incubation at 6, 16, and 24 hours, viability was assessed using ethidium bromide to
detect membrane—compromised cells. Seven images were obtained per well yielding
approximately 700 total cells per data point.

The results of the viability assessment are given in Figure 6.7. A dramatic
decrease in viability is seen between the 16 and 24-hour data points. This viability
assessment suggests that analysis of cells for evidence of protein turnover beyond 16
hours will give ambiguous results. Using the methodology described here, after 16 hours
of incubation, it would not be possible to discern whether GFP intensity loss in cells was

due to leakage fiom damaged cells or from OF P turnover.

 

 

 

 

 

Incubation Time (hours)

I .0. * I
‘5’ 60 ~ I
.2
> 40- I ”HUMV
I°\° g 707 control I , I
I 20 I cyclohex I '
I 0 7””"T“ . 1 I
I 0 5 10 15 20 25 I
I
I

 

Figure 6.7 Viability assessed with ethidium bromide after different exposure times to
the protein synthesis inhibitor cycloheximide. Control cell were incubated in DMEM
media with no cycloheximide present. (* t-test, p<0.0005).

89

The data are plotted in Table 6.2. There was no statistically significant difference
between control wells and the cycloheximide-incubated wells at all time points prior to
16 hours, however a small but significant decrease was found at 16 hours (student t-test,
p<0.01). This suggests that the half-life of GFP exceeds 16 hours. Because cells begin to
die after 16 hours of incubation in cycloheximide, persistence of OF P beyond this time is
difficult to measure. These results are qualitatively comparable to the half-life of 20
hours reported by Li et al., 1998. Their study did not address the possibility of cell death
giving rise to decreased GFP signals, therefore it is possible that GFP does persist much
longer than can be effectively measured using protein synthesis inhibitors.

These results establish that the loss of fluorescence directly following freezing
cannot be accounted for by a failure of the protein synthesis mechanisms. Even if new
protein is no longer manufactured because of damage to the cell components involved in

protein manufacture, the GFP produced prior to injury will continue to fluoresce.

Table 6.2 Cell fluorescence intensity as a function of exposure time to
cycloheximide. AF U, arbitrary fluorescent units.

Fluorescence Intensity (AFU)

 

Control Experiment

 

Time(h) Ave SD Ave SD

 

 

0 3.2 0.4 2.9 0.4
2 3.3 0.4 3.2 0.3
6 3.2 0.3 3.2 0.2
16 3.6* 0.1 3.4* 0.1
(t-test, p<0.001)

90

6.2.2 Stability of CF P to Low Temperatures

There is considerable information in the literature regarding the stability of OF P
at high temperatures, at high and low pH, and in the presence of denaturants (Bokman et
al., 1981). To the author’s best knowledge there exists no information on the stability of
OF P at low temperatures. Unfortunately there is no structural feature of proteins that
definitively indicates a propensity for stability at high or low temperatures, as there are
many molecular features contributing to stabilization of molecules at physiologic
extremes. Some of these factors include the number of disulfide bonds, the number of
salt bridges, the preponderance of ion-pair networks, the inaccessibility of hydrophobic
groups, and compactness of the molecule (Scandurra et al., 2000). Proteins that are
stable at extremes of temperature and other adverse environmental conditions (termed
extremophilic proteins) each contain different manifestations of these characteristics.
Thus a detailed examination of the various levels of structure of the EGF P molecule may
serve to explain observed stability phenomena, but it does not provide predictive power
for a range of conditions, such as expected low temperature behavior.

Generally protein firnctionality or activity at low temperature is achieved at the
expense of stability at high temperature, because the molecular flexibility required for
catalytic activity (the raison d’etre for many proteins) at low temperatures is in opposition
to the rigid nature of proteins that are stable at high temperatures. Conversely, the
rigidity associated with high temperature proteins can result in a non-functional protein at
low temperatures. From an evolutionary standpoint however, GFP does not have a
catalytic role per se. It serves as an energy transfer molecule, producing a shift in light

emission from blue to green when combined with luciferin and luciferase (Ward and

91

Corrnier, 1979). Rigidity that might be associated with the molecule’s reported high
temperature stability may not necessarily result in a stability or fluorescence functionality
trade-off at the low end of the temperature spectrum.

Despite the complexity of structure-stability relationships that exist in proteins,
there is one structural feature of GFP worth highlighting that may suggest expected low
temperature behavior. Because the chromophore is encapsulated within the OF P ‘B-can’
(see Chapter 2), in general the fluorescence fimctionality will likely be protected until
such time that the protein molecule either becomes completely denatured by normal
thermodynamic means or the tertiary structure becomes significantly altered by
mechanical forces that might be associated with ice crystal formation.

Nardid et al. (1997) studied the effect of freezing and thawing on the fluorescent
proteins P450 and spectrin using fluorescent spectroscopy. They specifically examined
the structural state of the molecules as measured by changes in the location and
magnitude of spectroscopic peaks. Their results revealed a small shift towards longer
wavelength of the emission peak maximum following liquid nitrogen immersion and
subsequent thawing of these proteins in buffer solution. The authors suggested that
freezing and thawing induced macromolecule surface loosening and consequently a
conformational change that exposed the protein chromophores to the solution molecules.
These changes were irreversible. Although studies of this nature have not been
performed for a wide variety of fluorescent proteins, the fact that both spectrin and P450
had similar structural changes resulting in mildly altered fluorescent spectra parameters
suggests that this phenomenon might be a minor factor in decreased fluorescent intensity

measured in EGFP-transfected cells following freezing.

92

All experiments performed up to this point suggest that the absence or dramatic
decrease of green fluorescence from a transfected cell that was fluorescent prior to the
event, means the cell is dead. However, it is critical to understand if, following freezing
conditions that are non-lethal to GFP-transfected cells, fluorescence can be diminished by
denaturation or degradation of the protein to an extent that a live cell can be deemed non-
viable by the GFP assay. The high correlations of the GFP assay with other viability
assays together with the results of Clonogenic assays suggest that this would be highly
improbable. However, this hypothesis was tested directly with pure EGFP in solution.

Before using the expensive recombinant protein, troubleshooting of the
experimental system was performed using InSpeck Green (505/515 nm) calibration beads
of fixed intensity [Molecular Probes, Eugene, OR]. A small amount of each calibrated
bead solution (~ 5 pL sample) was placed on a microscope slide and allowed to dry
before mounting and sealing with a cover slip. The mounted samples were then placed on
the stage of a Nikon Eclipse Fluorescent microscope [Mager Scientific, Detroit, MI]
equipped with a 460-500 nm excitation filter, a 505 nm dichroic mirror, and a 510-560
band pass filter. Images were taken using a 12-bit Cooke SensiCam [Cooke Corp.,
Auburn Hills, MI] with a 0.1 5 exposure time. Using Image Pro software [Media
Cybernetics, Silver Spring, MD] to obtain measurements of gray scale intensity of the
central portion of the beads, a calibration curve was constructed and is shown in Figure
6.8. Each data point represents the average of 5 calibrated beads. The correlation

obtained was consistent with the manufacturer’s specifications.

93

 

0.9 ._.______E__,,,,M,---,-,- A A”- 4“ — Wk 5 A ”u-” —-— -——--

' — _T‘l

Measured Normalized lntenelty
.0
Ul

.0
a)

.o .o
09 \l
I I
I I
I I
I I
I
I
I
I
I
I
I
I
I
I
I

,1: 0.0097x + 0013,? , - ,

.0
a

 

 

 

 

 

 

 

" I272; 5.5955 ‘
0.3 — __ ____
0.2
0.1 ......................... .-
0 . . . . .
40 50 60 70 80 90 100

I
I
I
I Normalized Intensity (Manufacturer's Specs)

Figure 6.8 Calibration curve obtained with InSpeck Green calibration beads
illuminated with blue light (510-560 BP filter), and detected with a Sensicam camera,
0.15 exposure time.

The stability of the Opti-Quip Kobosha light source was tested by examining one
set of beads over an extended time period to identify drift or fluctuations in the intensity
measured. Images were obtained every 30 seconds for a 3 minute period for the 30%
intensity beads. Small bitmaps (4x4) of the central portion of 5 beads were obtained and
analyzed for gray scale intensity. The average deviation of measured intensity was 1.1%
(one standard deviation from the average) during this time period, with a maximum
deviation of 1.8%, suggesting that the light source was relatively stable over short periods
of time.

Because the rEGFP measurements were going to be performed on liquid solution

samples contained in capillary tubes, sodium fluorescein solutions were implemented to
understand any complexities that might be introduced by this approach. Serial dilutions

of sodium fluorescein were made up in distilled water, to bracket the concentration range

94

0.02 to 0.1 mg/mL. Samples were drawn up in 0.1 mm path length capillary tubes (In
Vitro Diagnostics, NJ), sealed on both ends, and placed on a glass microslide. Upon
illumination with blue light (ex. 460-500 nm; em. 510-560 nm), images were captured
with the Cooke SensiCam with a 0.1 5 exposure time. A 250 pixel x 250 pixel bitmap of
the capillary tube was analyzed with Image Pro for average gray scale intensity. The
calibration curve is plotted in Figure 6.9. As evidenced by the value of the coefficient of
variation (0.98), the calibration curve was quite linear, which suggested that the
experimental system was adequate for attempting to construct a similar curve for rEGFP

in solution.

 

 

 

 

 

E e
g 0.8I ,

3 I
E I
2 |
30.67 - -- - ,-..-_-._.-

a," y =10.156x + 0.0102 I
g . R2=0.9863 I
1,0.4I«unff...i __E -g__g_,_g_ I
O I
.u

E"

c’02 ———— -_ -p-___3 - _f_ _ __ i
z
I I
I I
I 0 r ' I
I 0 0.02 0.04 0.06 0.08 0.1 I
I _i

Concentration (mglml)

Figure 6.9 Calibration curve obtained with Sodium Fluoroscein solutions in capillary
tubes, path length 0.1 mm, 0.1 3 exposure time.

95

A calibration curve of fluorescence as a function of concentration of pure OF P
was constructed to understand the levels of EGFP that can be detected with the
equipment currently available, and to serve as a standard for estimation of EGFP
concentrations in the population. In the case of liquids, the fluorescent signal comes from
both in-focus and out-of-focus regions of the sample, with light absorption and scattering
also influencing the overall fluorescence measured (Yuan et al., 1993). To avoid possible
artifacts from light attenuation differences associated with scattering and reflectance, 20
pm capillaries were chosen for determination of calibration curves. These capillaries are
within a few micrometers of the diameter of a suspended cell. The intensity from the
center of cells in suspension would be expected to represent an integrated intensity over
the nominal 14- micron diameter. In this manner the calibration curve can be used to
directly estimate the concentration of EGF P in cells based on gray scale levels.

Calibration curves were constructed using recombinant EGFP (rEGFP) [Clontech
Laboratories Inc., Palo Alto, CA]. Capillary tubes very close to the diameter of the cells
were used to contain the samples of rREGP for determination of sample intensity. Serial
dilutions of rEGFP were made up in 10 mM TRIS-HCI buffer (pH 8.0) to bracket the
concentration range 0.0025 to 1 mg/mL. Similar to the procedure for preparing sodium
fluorescein solutions, small samples of protein solution were drawn up into 0.02 mm
capillary tubes [Vitro Dynamics Inc., Rockaway, NJ] and placed on a microscope slide.
Upon illumination with blue light, images were captured with the Cooke Sensi-cam
camera with a 0.4 3 exposure time and analyzed with Image Pro. Three images of each
sample were taken at approximately the same depth in the capillary with a 0.4 3 exposure

time. Three images of each capillary were taken and test area bitmaps of 249 pixels x

96

251 pixels were acquired of the imaged capillaries. Gray scale intensity results of the
three areas were averaged to obtain each data point. Data to construct the calibration
curve was also acquired after the solutions were held at 5 hours at room temperature in
light-imperrneable containers. This curve deviated less than 3% from the original curve,
hence the two sets of data were pooled to obtain the calibration curve plotted in Figure
6.10. Data represents the average of six bitmap images from two separate capillary tubes.
Background corrections based on blank capillary intensity levels were applied to the data

in Figure 6.10

 

___ _.A' __7 ..- .____-__ __—..

R2 = 0.9999

_l
L
I

D
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

§
I
I
I
I
I
I
I
I
I
I
I
I

 

Raw Gray Scale lntenelty
I
I
I
I
I
I
I
I
I
I
I

§
I
I
I
I
I
I
I
I

 

 

O

 

0 0.5 1 1.5 2 2.5 3 3.5 4
Concentration (mgImL) '100

,__L____z___.__ _ L , i _ ,____ I

Figure 6.10 Calibration curve obtained with rEGFP solutions in capillary tubes of path
length 0.02 m using a 0.4 3 exposure time. Error bars are smaller than the data point
markers.

97

Patterson (1997) has indicated a cytoplasmic concentration threshold of ~200 nM
for EGFP (approximately 10,000 GFP molecules per cell) versus ~1 pM for wild type
(Niswender et al., 1995) is necessary to detect a fluorescent signal that is twice the
autofluorescence level. These studies were performed in HeLa cells. The detection
threshold determined in these experiments with capillary tubes was similar to the results
of Patterson, with a threshold of 220 nM to detect a signal that is twice the auto-
fluorescence level of capillaries filled with buffer solution.

To test the stability of the EGF P protein to extreme low temperatures a dilute
sample of rEGFP was treated with what considered in the cryosurgery literature to be a
highly damaging freezing protocol. A 0.5 mL Eppendorf tube containing a 50-pl sample
of 4 x 10’2 mM rEGFP was plunged into liquid nitrogen and held for 2.5 minutes,
followed by uncontrolled warming at ambient room conditions. Ten minutes after the
sample had melted it was plunged into liquid nitrogen and the procedure repeated for a
total of three freeze-thaw cycles. Control samples were maintained in a light
impermeable container at room temperature and sampled at the same time as the treated
solutions. For intensity analysis, samples were drawn up into small capillaries of path
length 0.05 m (V itro Dynamics Inc., Rockaway, NJ) and placed on the stage of a Nikon
Eclipse Fluorescent microscope [Mager Scientific, Detroit, MI] equipped with a 460-500
nm excitation filter, a 505 nm dichroic mirror, and a 510-560 band pass filter. Three
images of each sample were taken at approximately the same depth in the capillary using
a 12-bit Cooke SensiCam [Cooke Corp., Auburn Hills, MI] with a 0.4 s exposure time.
The test was performed in triplicate with a fresh working solution of rEGFP. Test area

bitmaps of 250 pixels x 250 pixels were acquired of the imaged capillaries using Image

98

Pro software [Media Cybernetics, Silver Spring, MD] and the gay scale intensity results
of the three areas were averaged to obtain each data point. The resultant decrease in
intensity is presented in Figure 6.11. Error bars represent the coefficient of variation

based on one standard deviation of nine bitmaps.

 

 

I 1.270 I I 7 7 77,17,771: 7 '7' 771,717 :7:;7:,7A7:7:77777‘
7. Control I Experimental .
E 100
G
.5
£0
20
8
to
a. 0.60 I
6
'0
3 040
ii
E
z °‘I20
0.00 I ,
post FT1 post FT2 post FT3 A

Figure 62.1] Decrease 1n intensity measured following multiple freeze-thaw (FT) cycles
of 4x10'2 mM rEGFP solution. Control samples were maintained 1n a light impermeable
container at room temperature and sampled at the same time as the treated solutions.
Error bars represent the coefficient of variation based on the standard error of n=9
bitmaps. *Sigiificant difference against control (p<0.01, one-sided t-test).

Results from direct freezing of rEGFP solutions suggest that damage or
denaturation of the protein would be expected to contribute very minimally to a

decreased fluorescence sigral in cells following freezing. Based on a one-sided t-test of

treated and control samples, a significant decrease (p<0.01) was not observed until the

99

ques
1100

the]

6.2.

18m

YES;

am}

end of three fi‘eeze-thaw cycles. The level of decrease measured even after three cycles
of aggessive freezing and thawing (~15%) was not of a magnitude that would result in
identification of a cell as non-viable per the GFP assay. This result suggests that there is
likely a more pronounced mechanism responsible for the GFP intensity change in cells
following freezing. Because membrane breach is widely accepted as a consequence of
cell injury following freezing, it is logical to assume that GF P leakage would occur. The
question now remains as to whether or not leakage and subsequent dilution of the
fluorescence signal can take place on a spatial and temporal scale that is consistent with

the loss of signal observed experimentally.

6.2.3 Diffusion-Dilution Hypothesis

With the knowledge that denaturation and degadation of EGF P at low
temperatures was not likely to contribute significantly to the overall cell intensity
response following freezing, the question of EGFP leakage was addressed. To the
author’s best knowledge, there is no direct evidence in the literature that indicates the size
of lesion that might be induced in a cell membrane following freezing. Because of this
lack of information, for the time being it is assumed that damaged cells will have a
membrane breach of adequate size such that diffusion of EGF P into the aqueous media is
not inhibited. Thus, the hypothesis is that EGFP can diffuse through a damaged
membrane immediately upon thawing, on a time and distance scale that would result in
dilution of the EGF P concentration to an extent that it falls below a certain threshold
shortly after cell injury. This concept will subsequently be referred to as the dilution-

diffusion hypothesis.

100

A series of experiments was performed next to establish the concentration of
EGFP in EGFP-transfected R3230AC cells. Concentrated solutions of EGFP-transfected
R3230 cells were drawn up into capillaries and images acquired under identical
conditions as the calibration curve. A total of nine images were acquired of three
separately drawn capillaries. The fluorescent intensity of the cells was determined by
averaging the pixels in an 11x11 bitmap positioned at the center of the cell. This bitmap
size was chosen because it covered approximately 30% of the cell area and was relatively
uniform in intensity. In this manner 345 cells were analyzed and a histogram of the
results are plotted in Figure 6.12. There appears to be two distinct peaks in the
histogam. The bulk of the population (90%) is distributed in the 850- 3650 range and is
positively skewed, but the remaining population appears to be centered near 4050.
Because the population goes off-range at this point it is difficult to describe its
distribution. All cells that were off-scale were lumped together at the theoretical 4150
level, as only 4096 levels of gay scale are possible with a 12-bit camera. It is likely that
cells in this region of the distribution represent the fraction of the population undergoing
mitosis, with an effective doubling of the cytoplasmic contents, including EGFP. Using
the mean gay scale intensity level of the first segment of the distribution, 2075, one
would expect the mitotic sub-fraction to have a mean gay scale intensity level of 4150.
This is consistent with the observed histogram. The median and mode of the main
population (850-3650) are 2027 and 1550, respectively. The mean of the entire
population was 2295. Based on the mode of the main population shown, this direct gay

scale intensity level, when corrected for backgound media fluorescence, would yield a

101

cellular EGF P concentration of 0.044 mg/mL according to the calibration curve in Figure

6.10.

Frequency
3

 

  

 

 

 

 

 

:III II. III

I I I . I l l I I I I I I l I

_—

 

77-777 A? gagsm _
9:0 =

Q Q
\Q'x'x‘f) 5’69 \Q‘bb 0°" 063$ ‘1

I
I
I
I
o (133% o o o ‘
66° $639,? (0% Me? 9665 (5,33 59" I
12-Bit Gray Scale Intensity:b Levels I

Figure 6.12 Histogarn of 12-bit gay scale intensity levels measured in a population of
R3230 AC cells.

A simple mathematical model of EGFP diffusion was desired to probe the
dilution-diffusion hypothesis. Although an understanding of nominal EGFP
concentrations in cells was now possible with the available experimental data, a value of
EGF P diffusivity was necessary for the calculations.

To obtain an estimate of the mass diffusivity coefficient for EGFP, known values
of diffusivity for other proteins were plotted as a function of molecular weight of the

molecule. As can be seen from Figure 6.13, the value of diffusivity for proteins tends to

102

Fig
mI

correlate highly with molecular weight. Based on a molecular weight of 27,000 g/mol,
the diffusivity of EGF P would be estimated as 9.24 x 10'7 cmZ/s in aqueous solution if the
linear best-fit line in Figure 6.11 were used. As discussed in Chapter 2, the diffusivity
coefficient measured for the GFP mutant 8651‘, a protein very similar to EGFP, was 8.67
x 10'7 cmz/s. These values are very similar. The average of these two values, 9.0 x 10'7

cmz/s, will be used in subsequent calculations.

 
    

. ,

I
I 1.00E-06 I *
I , y = -5.88E-07x + 3.53E-06 I I

8.00E-07 I 2
R = 9.51E-01

D (cm‘le)
O)
O
O
rn
c':
\I

Figure 6.13 Curve-fitting was used to estimate a difl°usivity value on the basis of
molecular weight. Data for plot was obtained from Lehninger, 1976.

A 1-D radial diffusion model was used to calculate the expected time constant for

diffusion of OF P into the extracellular space:
EzD i22( 2E] [6.1]
at r 6r Or

103

In this equation C is the concentration of GFP, r is radius, and t, time. Laws of scaling
analysis were used to simplify the derivatives and achieve an order-of-magnitude
estimate of the time constant for radial diffusion. The derivatives in Equation 6.1 were

simplified as follows:

 

2
9— ~ 01—21 ’ C [6.2]
t r r r
that leads to a characteristic diffusion time of:
2
r
t ~ — 6.3
D l l

The characteristic radius used in this calculation can be understood from the
diagram in Figure 6.14. Using the mode of the population described by the histogram in
Figure 6.12, and the rEGFP calibration curve given in Figure 6.10 to determine the
concentration, a 1:20 dilution factor would be necessary to result in a decrease of the
fluorescence of an average cell to background intensity levels (i.e. that of culture media
in capillaries). The radius of a sphere that would result in this decrease in intensity was
computed based on the change in volume necessary to achieve the 1:20 dilution. The
characteristic radius to be used in Equation 6.3 is not the radius of this sphere, but the
difference between the starting radius and the final radius necessary to achieve this
dilution. This is the distance that the outermost molecule would have to travel to reach

the outermost edge of this sphere. This characteristic radius was determined to be 12 um.

104

for

fol]

Fig
0ft

UI‘Id
CXlr
We .
a] 50

hi nc

Using this value together with a diffusivity value of 90 umZ/s yields a characteristic time
for diffusion of 1.6 s. This is consistent with the observation that GFP fluorescence

following cell injury is of order seconds to minutes in cell suspensions.

 

Figure 6.14 Schematic to illustrate the characteristic radius required for determination
of the time constant for diffusion of GFP.

In an attempt to extrapolate to expected in vivo behavior it is important to
understand that cells in a tumor are close-packed with only a finite amount of
extracellular space into which the GFP could diffuse. Using a cell volume fraction of 0.5,
we could thus expect a dilution factor of approximately 1:2. The diffusivity value would
also be expected to be higher because the transport in vivo would be expected to be more

hindered than in a pure aqueous environment. In choosing an appropriate estimate of

105

\\

[C

If

10

dr

GFP diffusion in viva, several models were examined which are summarized in Table
6.3. It is noteworthy that the values listed are within a narrow range. The Nugent and
Jain (1984) value was chosen as the situation most closely resembling the steady state in
viva environment. Their value represents the diffusion coefficient of a 27. kD molecule in
the interstitium of intact tissue. The situation following freezing in viva would be
expected to be more hindered however, as cell debris and extravasated red blood cells
would serve to slow the movement of the OF P molecule. Of the values considered, the
AT-l nucleoplasm model provided the most hindered environment. A value intermediate
to these, 1.9 x 10'7 cm2/s, was therefore chosen to determine the expected time constant
for GFP diffusion in viva, following freezing. Because the expected dilution factor in
viva is expected to be only 2-fold, the in viva characteristic radius was calculated to be 2
pm. The time for diffusion to occur to an extent that the signal drops 50% is calculated
to be 0.2 s. This indicates that if freezing in viva directly damages cells, the characteristic

drop in fluorescence (~50% decrease) would be very rapid.

106

Table 6.3 Literature diffusivity values applicable to GFP diffusion in viva.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7
D*12x10 Molecule Matrix Source

(cm /s)

2.23 27kD (1) Carcinoma Nugent and Jain (1984)
2.71 EGFP Cells Swaminathan, (1997)
1.64 EGFP AT-l cytoplasm Wachsmuth et al. (2000)
1.85 EGF P COS-7 cytoplasm Wachsmuth et al. (2000)
1.58 EGFP AT-l nucleoplasm Wachsmuth et al. {2000)
1.93 EGFP COS-7 nucleoplasm Wachsmuth et al. (2000)
1.97 IgG Q) Human Adenocarcinoma Netti et al. (2000)

1.89 IgG Mouse Carcinoma Netti et al. (2000)

 

(1) D calculated on the basis of molecular weight of GP P
(2) IgG has a molecular weight of ~1 SOkD

6.2.4 Mass Conservation

Based on the evidence presented so far, the diffusion-dilution hypothesis best
explains the loss of GFP fluorescence following freezing. To further test the validity of
these results, a mass conservation type of experiment was performed. Cells were grown
to confluence in 50 ml Falcon flasks (Becton Dickinson, Lincoln Park, NJ) and 800 ml
Corning flasks (Corning, NY) and following detachment from the flask with trypsin, cells
were concentrated by centrifugation and then re-suspended in 500 u] of D-MEM media.
Cells were prepared to varying volume fractions. The volume fraction of cells in solution
or ‘cytocrit’ was measured using an International Equipment Company Model MB
micro-capillary centrifuge (International Equipment Company, Needharn Hts, MA) and a
Lancer Spiracrit micro-hematocrit capillary tube reader (St. Louis, MO). Following
volume fraction determination, samples for imaging were drawn up in three 20-um
microcapillary tubes and placed on a microslide on the stage of the Nikon microscope.

Samples were illuminated with blue light and images were acquired in three areas of each

107

VC

Ir:

(:1I

IILK

bitr.

6.15
bitrr
3V8}:
him;
is SIN.

USIng

“here I

the Cell

capillary. Aliquots of 30-50 pl of cell suspension were then transferred to 0.5 ml micro-
eppendorf tubes and the samples immersed in liquid nitrogen for 2.5 minutes followed by
passive thawing at room temperature. Freezing and thawing was repeated for an
additional cycle. The samples were briefly mixed on a Fisherbrand Vortex Genie 2
(Fischer Scientific, Pittsburgh, PA) and then three fresh capillaries drawn from the
solution. Again, 3 images were acquired of each capillary. In this manner, for each
volume fraction studied, 9 images were obtained for analysis both before and after the
freezing and thawing protocol.

Images were analyzed using Image Pro Plus. Using a rectangular area of interest
(AOI) of 49x50 pixels (~50 umz), 20 separate bitmaps were acquired of the background
fluorescence in the cell suspension. This AOI size was specifically chosen to allow
bitmaps to be collected between suspended cells. This concept is illustrated in Figure
6.15. In the case of frozen-thawed cells, no intact cells were ever visible and therefore
bitmaps were collected in a systematic 5 x 4 array covering the field of interest. An
average ‘before’ and ‘afier’ intensity was calculated from the appropriate set of 180
bitmaps. The difference in intensity was plotted as a function of cell volume fraction and
is shown in Figure 6.16. Also plotted was a predicted intensity curve that was calculated

using the following equation:

Ilys¢d = VCcllsICcns + VMedia 1 Media [6-4]

where Ilysed is the GF P fluorescent intensity predicted for the solutions following lysis of

the cells, 11mm;a is the mean starting background intensity as measured in the current

108

experiments, ICells is the mean cell fluorescent intensity computed from the entire
population represented in the Figure 6.12 histogram, and VCells and VMedIa are the
volume fraction of cells and media, respectively. The error in [mm was predicted on the
basis of one standard error of the mean of both ICells and IMedia. Error was calculated

according to:

slam 2 SIM:- 2
SIM E I 1M I—‘ + I— [6.5]
Cells Media

 

Background
~275 AFU

 

 

 

 

Background
~450 AF U

 

 

 

 

(b)

Figure 6.15 Photomicrographs demonstrating conservation of GFP fluorescence
following lysis of GFP-R3230AC cells: (a) before freezing, (b) post-lysis.

109

As can be seen in Figure 6.16, the data are in reasonable agreement with the
predicted curve, with a correlation coefficient of 0.95, suggesting that the estimate of the
mean fluorescent intensity in previous experiments is in good agreement with current
data. The average percent difference between calculated and experimental data was
determined to be 12.4 %. The slope of the line originating from a linear fit of the
experimental data in Figure 6.16 revealed that within the magnitude of the experimental
error, a doubling of the volume fraction yields a 2-fold increase in fluorescence intensity.
This result provides direct evidence for the validity of the diffusion-dilution hypothesis
by establishing that fluorescence lost from the cells following freezing is quantitatively
accounted for by the magnitude of the increase in fluorescence intensity in the extra-

cellular space.

110

 

Fig

I‘ICI

rep
in l

6.2.

of ;
follr
mo].
COId
GFP
Iong

Chara

 

    

 

 

TTTTTTTT ’ 1' W” " "” ’ "I +Predicted

 

3 ' a, ,, ,,,,,,._
2 , , , /$¢4 ,,,,, , , , H I 9 Experimental ‘
1 , ‘ I
0

0 0.05 0.1 0.15 0.2
Volume Fraction of Cells

Figure 6.16 Change in GFP fluorescent intensity of background media following
freeze-thaw lysis of suspended cells. Arbitrary fluorescent units (AFU) represent the
measured gray scale intensity normalized to the initial data point. All y—axis error bars
represent the standard error of the mean. The x—axis error bars represent the uncertainty
in measuring volume fraction.

6.2.5 Conclusions

The previous results provide support for the hypothesis that diffusion and dilution
of the GFP molecule is the basis for the GFP fluorescent intensity loss observed
following freeze-thaw injury. Experiments using pure rEGFP demonstrated that the
molecule was robust to multiple cycles of freezing to extreme temperatures and therefore
cold denaturation of protein was not a relevant aspect of this application of GFP. The
GFP was shown to persist in cells that are not L r ' ‘, at least for as

long as 16 hours and probably for much longer. Order-of-magnitude estimates on the

characteristic time for GFP to diffuse to an extent that significant changes in fluorescence

111

are observed were consistent with experimental observation. Furthermore, mass
conservation experiments performed on the basis of quantification of GFP intensity
further indicated that the GFP molecule was not being destroyed but was transported into
the extra-cellular space. All experimental results and theoretical predictions were
consistent with GFP leakage following freeze-induced injury.

If cells become irreversibly damaged but not immediately membrane-
compromised, the GFP will likely persist until the necrotic pathway reaches the stage of
membrane degradation. This suggests that although the OF P assay has two hypothesized
features: (1) GFP will leak from membrane-compromised cells (2) GFP will not be
manufactured in cells that are metabolically damaged, the latter feature seems to be
difficult to practically capitalize on due to the long (>16 hours) turnover time of OF P in
the cell. Necrosis will probably progress to the extent that membrane damage manifests

prior to this latter mode becoming a rational basis for deeming a cell non-viable.

112

flI
11>
1h:
do
Ob
the

in}.

MI
Sub
and

hEm

Chapter 7

Validation of a GFP-based Viability
Assay in viva

7.1 R3230 AC Tumors Grown in the Dorsal Skin Flap Chamber:
Implant Procedure

The current work seeks to establish that following freezing, changes in the
fluorescent intensity of tumors grown from GFP-transfected cells will correlate with
tissue damage as assessed by other independent measures of viability, thereby evaluating
the use of a transfection-based assay for in vivo purposes. To achieve these ends the
dorsal skin flap chamber (DSF C) was utilized to grow and treat microscopically
observable tumors. The DSFC has been recently implemented by several researchers for
the study of freezing injury in rodents (Hoffman and Bischof, 1999, 2001ab). Following
injury the tissue was evaluated using GFP intravital fluorescence, and histopathology.

Female Fischer 344 Rats weighing between 150 and 180 grams were implanted
with dorsal skin flap chambers. Because this chamber permits chronic optical access to
subcutaneous or other implanted tissues, it is very desirable for studying both frostbite
and cryosurgery-induced tissue injury phenomema. Tissue integrity and microscale

hemodynamics can be studied without sacrifice of the animal. This procedure has been

113

 

I)

h}

mi

described by Papenfuss (1979) and few deviations from the methodology were made.
The primary modification was that one-sided preparations were utilized instead of
removal of the dermis on both sides of the skin flap. The traditional preparation involves
removal of the epidermal layer of opposing skin flaps, leaving a thin layer of vascularized
dermal tissue. Because this study focused on fluorescent epi-illumination of tumor
preparations, thin sections for trans-illumination were not necessary, and the less
traumatic one-sided preparations were found to be most suitable. All of the experimental
procedures were carried out under aseptic conditions using sterile surgical instruments

and reagents. A sketch depicting an implanted skin flap chamber is shown in Figure 7.1.

 

Figure 7.1 Sketch of Implanted Dorsal Skin Flap Chamber, Courtesy of Dr. Mark
Dewhirst, Duke University.

Animals were injected subcutaneously (so) with 0.3 mg/kg of buprenorphine
hydrochloride analgesic (Reckitt & Colman Pharmaceuticals, Inc., Richmond, VA). Ten

minutes later animals were anesthetized with sodium pentobarbitol (35 mg/kg

114

the
dis
DIG

EX;

intraperitoneal (ip), diluted 1:5 in 0.9% NaCl). The dorsal side fur was clipped and
depilatory used to remove the fine hair. The dorsal skin flap was then sterilized and
aseptically sutured to a stainless steel frame to allow through—the-skin visualization of
vasculature. Upon selection of a suitable vascular bed, the skin was dissected away on
one side of the dorsal flap (~1 cm diameter) using a scalpel and micro scissors, exposing
the microvasculature and subcutaneous tissue. The hypodermal tissue was trimmed until
the microvasculature on the opposing side was neatly exposed.

Holes were made at three sites surrounding the dissected region and the two
halves of the chamber assembly inserted and mounted together. A glass window was
then placed over the exposed subcutaneous tissue and it was sealed in place with an E-
ring. Three sutures were added to stabilize the chamber. At the time of chamber
implant, using a sterile syringe, the tissue was bathed with a small bolus (~200 uL) of
GFP-transfected R3230 AC tumor cells in DMEM media (1 x 105 cells/ml) before sealing
the preparation with a glass window. Prior to implantation Trypan Blue was used to
discriminate live and dead cells, and cells were counted with a hemacytometer and
prepared to the desired concentration in fresh supplemented D-MEM. In sham
experiments sterile saline was used instead of a cell suspension. An implanted chamber
(animal 1 hour post-operative and still anesthetized) is shown in Figure 7.2.

Animals were placed on an insulating pad and 4.5 mL of Ringer’s lactate was
administered every 2 hours until animals were ambulatory. They were then moved to a
room maintained at 91°F and ambient humidity and housed there for the remainder of the
tumor incubation period. Animals were monitored for tumor growth activity until the

tumor tissue dominated the viewing area, creating a relatively uniform spatial

115

(J) ”1'1

‘si

ti

F5g
tht
60

fluorescence signal. This was achieved in a 6 to 9 day time period. The tissue was then

treated with cryosurgery.

 

Figure 7.2 Recently implanted dorsal skin flap chamber in an anesthatized Fischer
344 rat.

The growth of implanted tumor cells was monitored in the first few animals that
were studied to understand the appropriate incubation time to achieve a substantial tumor
size. Shown in Figure 7.3 is a series of images that reflect a typical growth pattern.
Although quantification was not attempted, in areas of adequate blood supply, growth in

vivo tended to parallel grth in culture, with a doubling time on the order of 20 hours.

a b c

Figure 7.3 Series of images depicting growth of a solid tumor from a tumor slurry in
the dorsal skin flap chamber: fresh tumor preparation (a), the tumor at day 3 (b), and day
6 (c).

 

116

II

F I

et

In

“t

The tissue shown in Figure 7.3 was biopsied at day 6, fixed in 10% formalin,
embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E). The
same region of the tissue identified in Figure 7.3 was also located in the fixed specimen

and is shown in Figure 7.4.

 

Figure 7.4 H & E section of the tissue contained in the dorsal skin flap chamber at
day 6 afier implant of chamber and tumor slurry. The vessel pictured is the same vessel
seen in the Figure 7.3 series.

7.2 Experimental Design

Four groups were evaluated in this study. To determine baseline levels of
fluorescence in the tissue, independent of any fluorescent tumor growth, chambers were
implanted and were analyzed for fluorescent intensity over a 6-hour period (Sham, n=3).
Following the observation period, tissues were biopsied and fixed. To determine the
effect of treatment on this background level of fluorescence, several animals without
tumors were treated with cryosurgery (Sham Treated, n=3). Tumors were raised in
another group of animals (Control, n=5) that served as controls. The tissue in n=2 of the
animals in this group were biopsied in the plane of the chamber, whereas the other n=3

were biopsied in such a way to allow observation through the thickness of the chamber

117

tie

tissue (cross-section). See Figure 7.5 for an illustration of the biopsy procedure. For
animals in this category, the tumor tissue was not treated but exposed to the same period
of fluorescent light illumination (intensity, magnitude, and frequency), and then at 6
hours the animal was euthanized and the tissue within the chamber biopsied, fixed in
formalin, and stained for histological analysis. The final group received both a tumor
preparation and a cryosurgery treatment (Cryosurgery, n=9). Fluorescent images of the
tumors were acquired at the center of the viewing area before, during, and for 6 hours
following the freezing procedure. In some instances images were acquired of the entire
field to allow visualization of lesion diameters. Following the 6-hour observation period
the animal was immediately euthanized and the tissue within the chamber was fixed in
formalin and stained for histological analysis. The experimental design is summarized in

Table 7.1.

118

Ilsca

Figure 7.5 Illustration of the two different biopsy techniques used in the first animal
study. (a) Square tissue section was first removed from the skin flap. Tissue sections were
cut (b) in-the plane of the tissue (x-y) and also (c) cross-sections were obtained at x = 0
and x = 2.5 m.

Table 7.1 Experimental Design

 

Group Acronym #
Animals

 

Control:
tumor, 6 hr C n=5
observation
Sham:

no tumor, 6hr S- =3
observation
Sham Treated:
no tumor, S+ =3
cryosurgery,

6 hr observation
Cryosurgery:
tumor, CS n=
cryosurgery,

6 hr observation

 

 

 

 

 

 

 

 

119

Although not part of the original experimental design, the absence of a second
glass window on the non-dissected side occasionally resulted in considerable non-tumor
cell proliferation. This resulted in tissues two to three times the intended thickness, as
illustrated in Figure 7.6. All tissues, independent of thickness, were treated with an

identical freezing protocol.

 

 

Glass window

Figure 7.6 Depiction of tissue that was overgrown at the time of treatment.

7.3 Cryosurgical Probe Development
As part of a MSU indepedendent study, Fron (2000) created a simple cryosurgical
probe that was manufactured to achieve a range of cryosurgically relevant end

temperatures (lower than -20 °C) and fieezing rates (faster than ~50 °C/min) (Gage and

120

IJJ

Fl
nit

Baust, 1998) in the tumor tissue within the dorsal skin flap chamber. The body of the
probe was manufactured from 0.5” OD copper tubing. The 12 mm length probe tip was
made of 6 mm diameter brass stock. Extending from the top end of the probe were short
lengths of 1/8” diameter brass tubing, which provided connections for the liquid nitrogen
supply and return. Mixed phase nitrogen was circulated through the cryosurgical probe
using a variable flow rate Linkam LNP9311 Cryopump. The probe was insulated with
3/8” thick foam tubing. The evaporation of liquid nitrogen in the body of the probe
provided a minimum tip temperature of —l92 °C. Different tip temperatures and cooling
rates could be achieved by adjusting the flow rate on the Cryopump. The cryosurgery

system is depicted in Figure 7.7.

   

a b

Figure 7.7 Cryosurgical probe system. (a) Linkam LNP9311 Cryopump with liquid
nitrogen dewar and probe. (b) Close-up of 6 mm diameter cryosurgical probe tip.

121

‘

r1
53

1'1

F.‘g

7.4 Thermal History: Tissue Phantom Measurements to Characterize
Cryosurgery Treatment

To achieve the most accurate record of thermal history without invasively
instrumenting the tissue within an implanted window chamber, a phantom system was
used to optimize the cryosurgery protocol. A stage was built to facilitate instrumentation
of the interior ring of the window chamber with Omega 0.001” and 0.005” wire
thermocouples (see Figure 7.8) and subsequent measurement of freezing histories.
Biopsied rat skin samples from the Fischer 344 rat were utilized to achieve estimates of
the fieezing rates and end temperatures that would be expected in viva. It was assumed
that the effects of microvascular blood flow that were not accounted for in this phantom
system would not contribute significantly to the thermal resistance of the tissue for the in
viva case. The interior ring and glass window of the DSFC was instrumented with

thermocouples as shown in Figures 7.8 and 7.9.

 

Figure 7.8 Phantom set-up for acquiring thermal data.

122

ll

 

 

r=4J mm

Figure 7.9 Location of thermocouples and cryosurgical probe during the phantom
measurements.

The stage was draped with thick absorbent paper towels in the same manner as the
animal would be draped during the procedure. The flow rate of the cryopump as well as
the duration of freezing and thawing were optimized to create freezing rates with a
magnitude of -50 °C/min or greater and end temperatures lower than -20°C, without
extending the ice ball to the edge of the metal chamber. After a series of trials a protocol
involving three twenty-second freezing cycles was selected, while allowing 2 minutes
between freezing for the tissue to thaw completely. This protocol did not create sub-zero

temperatures on the surface of the metal chamber, yet created a well-circumscribed lesion

123

It

It

[0

dr

Flgl
phat

in the viewable area (~6-8 mm diameter). This optimized freezing protocol was applied
in viva to tissue within the window chamber.

Characteristic freezing curves obtained for phantom skin samples are shown in
Figure 7.10. The most variation amongst replicates occurred at the minimum
temperature. Therefore the standard error of the mean is reported at these points.
Measurements are for three replicates. In all cases the center thermocouple registered a
minimum temperature that was below —20 °C for each freezing cycle. The average
freezing rate was calculated to be of order 100 °C/min. The 2-min thaw time was chosen
to provide enough time for the tissue to completely thaw and return to above 5 °C.
Thermal measurements taken on the surface of the metal chamber indicated that the
draping used to protect the animal from cold air shedding from the probe was adequate,
as sub—zero temperatures were not recorded in the metal insert at any point during the

simulated procedure (data not shown).

 

 

 

 

 

30 . .______. __ __.__.. ._ _- _.
2 I —e-Center ~n— R=2.5 mm —-—R=4.0mm y
33. n. /’/ n,._.,./
2 I” ‘ /’" '
a _ n I __ _
2
8.
.5, I
P ._ d _
40 I — — — — — ——
-50
0 100 200 300 400
Time(s)

Figure 7.10 Characteristic freezing curves measured at specific locations in the
phantom tissue system.

124

8V:

at]

C01

Cr}.

7.5 Past-Cryosurgical Damage to GFP Fluorescing Tumor Tissue

7.5.1 Intravital GF P Intensity Assessment

The animal was anesthetized with sodium pentobarbital (35 mg/kg) and placed on
its side on a modified microscope stage. Images were acquired of a centrally located test
section (directly beneath the cryosurgical probe) before, during, and after freezing using a
SPOT thermoelectric cooled color CCD camera and a Nikon Diaphot inverted fluorescent
microscope equipped with a 100 W mercury arc lamp (both from Diagnostic Instruments,
Inc. (Sterling Heights, MI). The tissue was illuminated with blue light (450-490 nm) by
inserting a FITC filter cube (Chroma Technology Corp., Brattleboro, VT) into the light
path. Three-channel RGB fluorescent images of the tumor were acquired at the center of
the viewing area before, during, and for 6 hours following freezing (exposure times, Red:
1.533 sec, Green:1.538 sec, Blue: 2.307 sec). These test area images were then evaluated
for changes in fluorescent intensity. The test area images were converted to gray scale
and analyzed using Image Pro software (Media Cybernetics, Silver Spring, MD). The
average fluorescent intensity/ sq. pixel of tissue was measured in each of the test sections
at key time points. The area analyzed was 775 microns x 612 microns. Representative
composite images spanning the viewable area (~1.0 cm diameter) were also taken before
cryosurgery and at 2.5 and 6 hours following the procedure for some of the animals.

The gray scale intensities per square pixel were measured for intravital
microscopy images. Intensities were normalized to the starting local intensity measured
for each animal. The normalized results for animals that experienced lethal fi'eezing on
the basis of histopathology are shown in Figure 7.11. The results of five identical

cryosugeries for five different animals with tissue of comparable thickness are shown

125

(CSl-CSS). These results reveal a large reduction in fluorescent intensity immediately
following the final fi'eeze-thaw cycle, with little subsequent change for the duration of the
6-hour observation period. Data from treated (8+) and untreated (S-) sham animals
(animals received no tumor implant but were incubated for 6-9 days) were also plotted in
this figure. These data were normalized to the average starting intensity of tumor-bearing
treatment and control animals. At the 6—hour point all of the cryosurgery data from
cryosurgically-treated tumor-bearing animals fell within the limits of these average
autofluorescence intensity levels. The difference between these intensities was observed
to be directly related to the amount of hemorrhage and blood pooling in the observation
area, which served to moderately attenuate the fluorescence levels. For treated animals,
the x-axis represents the time following the last of three fieeze-thaw cycles. The
untreated control animals received an equivalent time series of blue light illumination
(~60 minutes total) but did not undergo cryosurgery. Some variation was noted in the
intensity of the control tissue but the deviation was small. The average variation was 3%
from the average starting intensity of 14 animals. This indicated that the GFP expressing
tumor tissue was robust to photo-bleaching effects and that the excitation source was
relatively stable.

Composite tiled images of the viewing area are shown in Figure 7.12 for intravital
images acquired before cryosurgery and 6 hours following treatment for the same animal.
Note the retaining ring at the edge of the images in Figure 7.12. This ring has an ID of
10.6 mm. The cryolesion in Figure 7.12 (b), judged by fluorescence decay, was
measured to be ~7.5 mm in diameter. The 6-hour images were similar to the 2.5-hour

images.

126

+035 i—s-

 

 

 

 

 

 

I
I .
I 1.00 *********** ' I
I 30.90 I
.2 0.80
I 8 0.70
5 0.60 I
g 0.50, I
-- 0.40
E l. J. J.
E W \ I
o 0.20 _\ /L\,é I
0.10 \d—ae/ I
0.00 . I
1 10 100 1000]I

 

n_n_ ,. _Tl'flfl'fl-

 

Figure 7.11. Normalized gray scale intensity of GFP fluorescing tumor as a function of
time for cryosurgically treated animals (CSI-CSS). The rectangular region indicates the
average upper and lower autofluorescence levels of untreated and treated normal tissue,
respectively.

 

Figure 7.12 Image of the glass window and e-clip (a) and tiled composite fluorescent
images of the entire window chamber viewing field: before cryosurgery (b) and 6 hours
after the last freeze-thaw cycle (c). The cryo-lesion is clearly demarcated in (c).

127

7.5.2 Histopathology Assessment

Following the cryosurgical protocol the animal was immediately euthanized and
the tissue within the chamber was fixed in 10 % formalin and embedded in paraffin. To
establish the relationship between the presence or absence of GFP fluorescence and cell
devitalization in viva, S-micron thick tissue sections were obtained either in the plane of
the skin flap or in cross-section, and stained with hematoxylin and eosin.

Before the fixed specimens were dissected for paraffm embedding, the
impressions that the window chamber left in the tissue were used as markers to estimate
the amount of tissue shrinkage that occurred during the fixation process. Shrinkage
varied across the specimen. The inner tissue with the dermis dissected away had an
average shrinkage of 2.5%, the outer non—dissected tissue had a value of 4.5 %.

As previously indicated, some tissues experienced overgrowth. They were treated
with the same protocol but as expected, little damage occurred. In five of the nine
animals treated the tissues were of normal thickness. In these cases the GFP intensity
was seen to drop by 50% or more within 10 minutes after treatment. Histopathology
confirmed the necrosis or ablation of this tissue. In the remaining cases, cell death either
did not occur or was partial. In each of these cases histopathology correlated with

intravital observations. Characteristic results are shown in figures 7.13 through 7.15.

128

500 rim rS—O'O :m

It w? it ’

(C)

 

Figure 7.13 (a) GFP-fluorescing tumor before treatment. (b) GFP-fluorescing tumor
after treatment. (c) histpathology micrograph from the center of the tissue in (b). Damage
is complete, with only cell debris and necrotic cells (arrows) present.

129

 

(C)

Figure 7.14 (a) GFP-fluorescing tumor before treatment. (b) GFP-fluorescing tumor
afler treatment. (c) Histpathology micrograph from the center of the tissue in (b).
Damage is partial with some necrotic cells (arrows) present amidst a group of normal
tumor cells.

130

 

Figure 7.15 (a) GFP-fluorescing tumor before treatment. (b) GFP-fluorescing tumor
afier treatment. (c) histpathology micrograph from sample at the center of the tissue in
(b). No damage is observed.

131

The remaining observations will focus on the tissues in which distinct cryo-
lesions were formed. In the center of the targeted tissue there was near complete ablation
of the tumor tissue with only nuclear and cytoplasmic remnants of tumor cells scattered
within a markedly edematous and congested connective tissue matrix. Viable tumor
tissue containing a high mitotic index circumscribed the outer edge of the targeted area.
Between the central area that was nearly devoid of tumor cells and the outer edge of the
remaining tumor tissue, there was a distinct zone of degeneration and necrosis of
individual tumor cells, vascular congestion and hemorrhage, and a conspicuous influx of
neutrophils with lesser numbers of mononuclear cells (i.e. acute inflammation).

Characteristic results are shown in Figures 7.16-7.18.

132

Fig
I981).

I am-

 

(0)

Figure 7.16 H&E micrographs of cryo-treated tumor tissue: The tissue surrounding the
lesion (a) is undamaged. At the margin of the lesion (b) necrotic cells are evident
(arrows). In the central region (0) nuclear and cytoplasmic remnants are evident.

133

 

Figure 7.17 H&E micrographs of acute inflammatory cells (neutrophils) located near
the edge of the cryo-lesion (arrows).

 

Figure 7.18 H &E micrograph of neutrophils phagocytosing red blood cells (arrows).

134

C011

prcc
non
caus
IUmI
leak;
GFP
0113.»,
this I

lobc

Derick
lie

EZII

fluore:

7.5.3 Comparison of Results

The freezing protocol applied to the tissue was expected to create a region of
complete tissue damage in the center of the tissue. In the tissues that were considered
normal thickness, H&E analysis confirmed that significant tissue damage was present in
the central region of the tissue, which suggests that the protocol was successful in
creating a well-circumscribed cryo-lesion in a subset of the tissues studied. In the central
region of the tissue, GFP fluorescence intensity decreased dramatically following the
final freeze-thaw cycle, consistent with the proposed basis for the GF P assay: leakage of
GFP from cells with damaged membranes. The freezing rates and end temperatures
predicted at the center of the tissue were of order -100°C/min and —30°C, respectively for
normal thickness tissues. Multiple applications of this thermal history were likely to have
caused some intracellular ice formation and other damaging events in the cells within the
tumor. Membrane damage on the thaw cycle was thus expected, giving rise to GP P
leakage into the extracellular fluid. This would cause a dilution of GF P and a decrease in
GP P fluorescence signal, as observed. Although a dramatic decrease in signal was
observed immediately following the cryosurgery treatment for all four cases included in
this group, the absolute signal observed varied between animals. This was established
to be due to the fluorescent dampening effects of blood pooling in the observation area.

In some of the cryosurgery cases the tissue had ‘overgrown’ during the incubation
period, and had pushed out the back of the chamber. The general appearance of this
tissue was shown conceptually in Figure 7.6. These tissues were treated with the same
freezing protocol as the ‘normal thickness’ tissue. The pattern of disappearance of GFP

fluorescence was not the same. Obviously with a thicker tissue section, the penetration of

135

the

“a

C0]

the

(in

6th

qu

I10]

the freezing front was not as deep, and the tissue thus was likely to have experienced
benign thermal histories. What is important is that the intravital images qualitatively
agreed with the histopathology results. In specimens where the tissue presented as viable
in intravital GFP microscopy, the tissue resembled that of control tissues, with no damage
observed. As can be seen from the characteristic results shown in figures 7.13-7.15, the
gradiation in damage observed over the set of GF P intravital images can also be seen in
histopathology in the same region of the tissue.

The spatially distributed percent change of GFP fluorescence after cryosurgery
was evaluated for the 2.5 and 6-hour composite images for a number of cases, and
compared to histopathology. For a sample case shown in Figure 7.19, it can be seen that
the lesion diameter predicted by GFP was larger than that predicted by histopathology
(image dimensions corrected for shrinkage). It seems likely that the cells will exhibit
evidence of damage by GFP assay more quickly than histopathology can reveal at the
same time point. It is interesting to note that in both cases, the assays reveal an area of
tissue with a gradiation of damage, suggesting that the GFP assay might be used to
quantify partial damage in whole tissues as well as complete damage, on the basis of

normalized intensity change.

136

m
I

I

I

I
8I
I

I

I

I

I

I

I

I

I

I

I

I

I
n__I

 

O

Intensrty
Change (%)

 

     

L.
o
o

 

Figure 7.17 Spatial comparison of GFP intensity changes (approximate lesion
diameter ~7.5 mm based on magnitude of intensity change) and histopathology results
(approximate lesion diameter 6.5 mm based on region of hemorrhage) for one case of
tumor cryosurgery treatment.

137

ti

0.

al

Ih

CC

dc

dc

7.6 Dynamics of Fluorescence Intensity Change in viva

How would the dynamics of GFP intensity change be expected to change in viva?
Instead of having an aqueous pool of media or buffer to diffuse into, as in the case of
many of the in vitro experiments performed to date, the EGFP molecules must diffuse
through a normal or disrupted tissue matrix upon leaking from a damaged cell. In the
case of in vitro studies the volume fraction of cells was such that there was an ample sink
for the EGFP molecules. It is necessary to understand the capacity for dilution in viva.
The volume fraction of the interstitium in tumors was determined to be in the range of
0.35 — 0.55 by Jain et al. in 1987. The values reported for the vascular component by
Jain et al. in 1988, were 0.01-0.2. This results in a cellular fraction of 0.25-0.64. Thus if
all cells in the center treatment region were to be completely lysed following three fieeze-
thaw cycles, the dilution factor would be somewhere between 1.3 and 4. This is
consistent with in viva observations, as GF P fluorescence intensities have been seen to
decrease up to ~60% following lethal freezing. As demonstrated in Chapter 6, this could
be achieved on the order of seconds, consistent with some of the observations in viva that

demonstrate a loss of fluorescence directly following the thaw.

138

Chapter 8

Design and Characterization of an
Annular Freezing Apparatus for the

Dorsal Skin Flap Chamber

8.1 Motivation

The time course of freeze injury in viva, as well as the magnitude of inflammatory
and ischemia contributions to overall injury are not well understood. Traditional in viva
approaches for studying cryosurgery frequently involve inserting a cryosurgical probe
into a lesion, evoking several freeze-thaw cycles, and then biopsying the treated tissue at
discrete time points after the cryosurgery to obtain histopathology specimens. Species
such as sheep (e.g., Rabin et al., 1998) have frequently been used as animal models for
cryosurgery studies because their size allows the use of traditionally sized cryosurgical
probes in large tumor masses. In these types of animal models, tumors can be grown that
are analogous to human tumors, both in size and character. However in terms of
evaluating the temporal nature of tissue injury, use of large animals to achieve
statistically significant data at a large number of time points can be prohibitively
expensive. Scaling down the clinical situation to smaller animals such as rodent models

makes the evaluation of injury dynamics more accessible, however new challenges are

139

introduced. Large animal studies allow the placement of discrete thermocouples, or
permit simultaneous imaging of the procedure (e.g. ultrasound, MRI) so that phase
change boundaries (the ice front) can be tracked. In contrast, small animals create
difficulties for monitoring thermal histories during cryosurgeries, and when treating large
tumors the risk of hypothermic shock is high. The use of small optical chambers, while
providing excellent detail regarding the microvascular injury, does not facilitate thermal
history measurement. Numerical simulations of the therrnophysical response are thus
useful for generating thermal field information in these systems.

Thermocouples have been used to measure temperatures within the dorsal skin
flap chamber, both for cryosurgery studies (Hoffman et al., 1999) and high temperature
studies such as hypertherrnia (Gross et al., 1982) and burn injury (Diller and Hayes,
1991). Thermal measurement within the DSFC and other optical chambers is non-trivial
because of the small scale of the system. Since large temperature gradients are typical in
scaled-down cryosurgery model, small thermocouples are required to achieve a localized
temperature measurement. Thermocouples with bead diameters smaller than the tissue
thickness (0.2 to 2.0 mm) are difficult to handle and they are fragile. Moreover,
accurately locating the measurement junction once it has been inserted into the tissue can
be problematic. Although some have used ‘needle’ sensors, thermocouples embedded in
the shafts of needles, these constructions are known to introduce significant ‘fin effects’
error into the measured temperature value (Rabin, 1998). Thermal contact resistance is
also an issue when thermocouples are used to measure surface temperatures. Generally
the number of thermocouples used is limited and interpolation, extrapolation, and

symmetry arguments are invoked to infer the desired thermal field information. For the

140

SI“

the
his
im I

On

tiss
ex;
the

Ho

C(Ir
cor
hi 5
Re;

ah

Pat

gal

specific case of using a dorsal skin flap chamber (DSFC) for studying cryosurgery, the
Bischof group has used thermocouples as a means of quantifying the thermal field so that
the physiological and biophysical responses can be correlated with defined thermal
histories (Hoffman et al., 1999, 2001ab). This approach has already contributed to an
improved understanding of the importance of the vasculature in tissue freezing response.
On the other hand, fin effect errors in these thermocouple measurements are estimated to
be of order 10°C, which limits the utility of such an approach. Moreover, instrumenting
tissue within the DSFC with thermocouples is an invasive process. The author
experienced inadvertent hemorrhage of blood vessels during instrumentation of some of
the chambers used in this study, and some minor problems were also noted in the
Hoffman and Bischof study (2001b).

In this work, the focus of the first animal series (Chapter 7) was to determine the
correlation of the GFP assay with histopathology at different extremes of damage (i.e.
complete, partial, none) to validate its use in viva. Knowledge of the precise thermal
history was not important as long as it was within a cryosurgically relevant regime.
Repeated cycles of freezing, following the protocol optimized in the phantom system, had
a high probability of creating damage in the area directly beneath the cryosurgical probe
in thin tissue sections. This was observed in all of the animals included in the normal
tissue thickness group. Although observations at the edge of the cryolesion were made
during the first experimental animal series, the focus of the study was the utilization of
GP P as a cellular viability assay, and not correlating thermal history with the
pathophysiology of injury at the edge of the cryo-lesion. Nevertheless, it was possible to

gain valuable insight into the challenges inherent in achieving viability-thermal history

141

correlations. Using this same approach for inducing freeze injury would result in
difficulties in reproducibly localizing the cryo-lesion edge for intra-vital experiments, as
the diameter of the lesion ranged by as much as 10 mm for the animals treated (i.e. no
lesion to complete ablation). Although much of this variability was due to the tissue
thickness, even if the tissue thickness was constrained, tracking the interface remains
challenging. Localizing a region of tissue experiencing temperatures within the narrow
range considered to characterize the edge of the cryo-lesion in such a scaled-down
system, would equate to trying to identify the temperature gradients and viability
response within a region of sub-mm dimensions. Certainly from the standpoint of
thermocouple measurements this is nearly impossible.

Another one of the challenges in using the dorsal skin flap chamber to study
cryosurgically relevant low temperature processes is the difficulty in achieving a
controlled and well-characterized cryolesion within the ~1 cm viewing area. The
structural complexity of the microscale system limits the size of the ice-ball that can be
created within the chamber before the boundary conditions become extremely
complicated and extrapolation from thermocouple measurements questionable. These
complexities also present challenges in implementing numerical models to predict
thermal histories in the treated tissue. Development of technologies that would allow
better control and resolution of temperature gradients and temperatures in the DSFC
would thus be advantageous from many standpoints.

To produce a means of studying vascular injury, ischemia, and inflammation at
temperatures and freezing rates consistent with the edge of a standard cryo-lesion, a non-

conventional approach was used. Instead of creating a spherical tumor, a disk tumor, a

142

ch

5}

1h.

10

IE

fr

characteristic of tumors grown in a dorsal skin flap chamber, is chosen as the model
system. The cryosurgical probe was not inserted into the center of the lesion or placed on
the surface of the tumor as performed in the first animal series, but instead a novel probe
was developed to create a lesion by applying a uniform and controllable thermal history.
Although devices of this nature have been created to study freezing of tissues in vitro
(e.g. Rubinsky, 1995), in viva devices are less common.

The device in this study was annular in design, thereby permitting optical access
to the tissue during treatment. By creating a device that could provide uniform and
reproducible high sub-zero temperatures and slow freezing histories in viva, the
possibility of freezing tissue within the chamber without eliciting significant direct cell
injury was feasible. Cooling rates and temperatures could be applied that would not
result in significant direct cell injury. This provided a measure of ‘uncoupling’ not
possible with the more conventional scaled-down cryosurgery approach. In this latter
case, both fast and slow freezing rates are experienced in the same system, and therefore
spatial resolution of the high sub-zero temperature range is necessary. A novel freezing
device as described would allow a direct means of determining both the extent and
dynamic nature of the ischemic injury that occurred following freezing. The remainder
of this chapter details the construction, characterization, and numerical modeling of this

novel Annular Freeze Apparatus (AF A).

8.2 Design Objectives

A freezing device was desired that attempted to maximize the utility of the DSF C

for studying low temperature injury. Fast freezing injury, such as occurs at the center of

143

 

a typical cryolesion, is well-characterized but there are many ambiguities associated with
injury at the edge of the cryolesion where slow freezing rates and high sub-zero
temperatures dominate. Injury in this region tends to be associated with loss of vascular
integrity, ischemia, and inflammation, but this complex pathophysiology is not well-
characterized. Even less is known about the range of thermal conditions that elicits the
response. For this reason the design goal was to create an apparatus that could
reproducibly achieve slow freezing rates to minimum temperatures in the -5 to —60 °C
range, in a large portion of tissue in the DSFC. By creating a thermal history in the tissue
that was the same on both sides of the chamber, temperature measurements could be
made on one side of the chamber while microscopic observations (brightfield and epi-
illumination) were made on the opposing side. Symmetry of thermal response was also
desired to simplify the modeling of tissue freezing within the DSFC. If symmetry
arguments can be evoked, the size of the mesh necessary to solve the heat transfer
problem can be reduced by a factor of 2. This results in a decrease in computational
time.
The design objectives for the AF A were:
1. Controllable and reproducible slow freezing (1-10 °C/min) of tissue in the
pftrggllyéiccessible region of the DSFC to high sub-zero temperatures (-5

2. Symmetry of device and/or symmetry of temperature response at the
interior surface of the DSF C annulus profile.

3. Spatial uniformity of temperature over the annulus.

4. Minimization of frozen tissue in the region surrounding the DSFC.

144

(I:

II

Iis

8.3 Construction of the Annular Freeze Apparatus

Several prototypes were investigated as part of an MSU independent study
(Frerichs, 2001), with designs falling into two distinct classes: those that directly
implemented the annulus of the DSFC as a cooling chamber and those that used an
external device. The latter category was preferred because it involved the construction of
a single device as opposed to the retrofit of individual chambers to suit the design
requirements. A final design was selected from this category.

The final freezing apparatus design was comprised of two stainless steel
symmetrical rings that have an internal 2.3 mm channel for circulation of cold nitrogen
vapor. The mechanical drawing for this device is given in Appendix D. A Linkam LN2
Cryopump was used to pull cold nitrogen vapor through the apparatus, via inlet and outlet
ports manufactured from hypodermic needles. Stainless steel parts were assembled and
attached using Omegabond 400 high-temperature air-set cement (Omega Engineering,
Stamford, CT). Although the cement was effective at low temperatures, condensation
and associated moisture adversely affected the bonds; hence the cemented regions were
sealed with acrylo adhesive prior to repetitive use. An external groove on the surface of
these rings provided the point of attachment for interfacing with the DSFC. When
implemented in viva the symmetrical annular probes are attached to the exterior surface
of the protruding annulus of the DSFC. Use of the AFA together with the DSFC is
pictured in Figure 8.1. The extraction of heat from the tissue is achieved via the DSFC
annulus, therefore the optical port is not obstructed during the freezing procedure.
Cooling begins in the tissue at the point where the annulus of the DSFC contacts the

tissue, followed quickly by extraction of energy from the center of the chamber. In this

145

situation the optical access area of the DSFC can be fully utilized during the freezing
protocol, thereby permitting the use of various imaging methods (e.g., fluorescence,
infrared) throughout the freezing procedure to identify ice front position or thermal
gradients. This apparatus was designed to have flexibility of imaging modes, i.e.
microscope-mounted or otherwise. Care was taken to ensure that the AF A did not
directly contact the metal surrounding the annulus, by providing a clearance between the
AF A and the metal region surrounding the DSFC annulus. This design aspect minimizes

adverse heat conduction from regions of the chamber that contact the tissue at other sites.

146

 

Figu
Stain]

annul

8.4

5381c
expe:
regen
m0

appn

diam

Surfa

 

(a) (b)

Figure 8.1 Annular Freeze Apparatus (AFA). (a) Cold N2 vapor passes through the
stainless steel freezing apparatus by means of hypodermic needle inlet and outlet ports.
A groove on the surface of the annuli provides the point of attachment to the DSFC. (b)
The symmetrical annular probes are attached to the exterior surface of the protruding
annuli of the DSFC.

8.4 Thermal Characterization of the Annular Freeze Apparatus

To assess a number of features related to the thermal behavior of the AF A-DSF C
system, processed meat was used to simulate tissue. In all of the characterization
experiments, thick slices of chicken bologna (~2 mm thickness) were u'immed to
resemble the hemispherical shape of a dorsal skin flap and then sandwiched between the
two aluminum flames that comprise the DSFC. This caused the tissue to compress by
approximately a factor of two.

In the first series of characterization experiments, thermocouples with bead
diameters of ~ 0.01” diameter were attached to the center of the annulus on the interior

surface of the dorsal skin flap chamber using acrylo adhesive at three equidistant

147

locations, as shown in the inset picture of Figure 8.2. Thermocouples were placed
identically on the opposing side of the chamber for a total of 6 mounted thermocouples.
To maximize the thermal contact between the freezing apparatus and the annular portion
of the DSFC, Omegathenn 201 thermal paste (Omega Engineering, Inc., Stamford, CT)
was applied to the internal groove of the freezing device, the surface that contacts the
DSFC. This provided approximately 10 °C degrees of further cooling capacity by
reducing the contact resistance between the chamber and the freezing device. The AFA
was attached to the DSF C and cooling initiated by increasing the power of the Linkam
Cryopump to its maximum setting. Temperature was recorded in l 3 intervals using a
DAS-TC/B data acquisition board (Keithley Instruments, Cleveland, OH). Cooling was
maintained for 20 minutes and then the Cryopump was turned off. The apparatus was not
removed from chamber during the passive thaw period. This protocol was repeated three
times with a fresh sample placed in the chamber after each experiment.

A characteristic thermal history is given in Figure 8.2. The similarity of
temperature measurements at each of the three locations on the surface of a single
annulus indicated that the temperature distribution over the surface was uniform. The
average spatial variation (measured as differences between thermocouples at a single
time) when comparing thermocouples attached on a single side of the DSFC was 0.68 i
0.44 °C (average 3: standard deviation). Of the 3 tests performed, the maximum
temperature difference recorded amongst all 6 thermocouples (excluding the sharp
discontinuity region around the latent heat release) was 4.0 °C. To test for differences
between temperatures recorded on opposing halves of the DSF C, the thermocouple

readings on each side were averaged and compared. The temperatures measured on

148

opposing sides were in very good agreement. When the three thermocouples
measurements per side were averaged and the difference compared, the average
difference over all three experiments was 1.8 i 1.0 °C (average i standard deviation).
The most significant differences were recorded at the lowest minimum temperatures
reached. This indicated that the freezing apparatus was in good contact with the
chamber throughout the protocol and that the cooling power delivered to each side of the
device was effectively the same. This was the desired result. The minimum temperature
reached was -83.4 i: 2.96 °C (average i standard deviation) after 20 minutes of thawing.

A detailed summary of the data is given in Appendix E.

 

 

 

 

 

 

 

Figure 8.2 Characteristic data obtained with the Annular Freezing Apparatus (AF A).
Three thermocouples were attached to each side of the DSFC (per the insert diagram)
prior to assembly. The AF A was attached and cooling was initiated and maintained for 20
minutes.

149

As shown in Figure 8.3, hold periods were possible with the AFA by manually
adjusting the pumping rate once the desired temperature was achieved. Cooling was
initiated in the AF A attached to a dorsal skin flap chamber containing processed meat.
Temperature was recorded in 1 5 intervals. When the sample reached — 44.5 °C the pump
speed was decreased until temperature stabilized. The pump speed was then adjusted as

needed to maintain the temperature during the hold period.

 

Temp (deg C)

 

 

 

 

Time (min)

Figure 8.3 Demonstration of thermal control possible with the AF A system.

Once acceptable uniformity of cooling over the surface of the DSF C annulus and
symmetry of response between both halves of the DSFC was established, the next goal
was to determine the magnitude of the temperature gradient within the viewing region of

the DSFC. A grid window was used to position a central thermocouple. This grid is

150

pictured in Figure 8.4. Thermocouples were mounted on the annular surface as
previously described together with a centrally located thermocouple on the grid window.
The window was placed groove—side down, with the thermocouple in contact with the
tissue. Cooling was initiated and the sample cooled until the central thermocouple
reached a temperature of —20.0 °C. This temperature was held for several minutes and
then cooling discontinued. This procedure was repeated three times. During cooling, the
maximum variation between the temperature of the annulus and the temperature recorded
at the central thermocouple was 2.2 °C (range: 1.7-2.6 °C). If the temperature gradient is
approximated as a linear function of distance from the center, these results suggest that
the temperature gradient is of order 0.5 °C/mm. The actual nature of the temperature

distribution will be discussed further in Chapter 9.

151

 

 

Figure 8.4 A DSFC glass window implemented with machined grooves of 400 um
diameter. The red and blue lines overlaid on the image illustrate how thermocouples can
be mounted on this window.

In the Hoffman and Bischof study (2001a) involving freezing in the DSFC, large
thermocouples were implemented for measurements. These were mounted with the
thermocouple bead on the surface of the tissue, with the remaining lead wires at ambient
conditions. Conduction error was considered to be significant and so they performed an
experiment to quantify this contribution. For a thermocouple of the size used in this work
(0.22 mm), the conduction error reported at —10 °C was 3 °C. In the present study the
temperature gradient in the environment of the thermocouple wire is an order of
magnitude smaller than in the Hoffman study, much more favorable for accurate

temperature measurements. Since the temperature gradient is the driving force for

152

conduction errors, the error due to conduction can be approximated to be an order of
magnitude less. This would yield conduction errors of approximately 0.3 °C. All
thermocouple measurements would be expected to be affected similarly in the
configuration described, hence fin effects should not be significantly altering the
character of the temperature gradient.

Temperatures and freezing rates characteristic of the edge of a cryolesion were
achieved using the novel AF A together with the DSFC. The initial design objective was
met and freezing rates between 1 and 10 °C/min were achieved. Rates faster than 5
°C/min require the probe to first be cooled and then attached to the surface of the DSFC.
At the fastest pumping rates the freezing apparatus is capable of achieving a temperature
of —130 °C on the surface of the AFA. When the apparatus was connected to the DSFC,
it was possible to achieve —1 10 °C on the inner annulus of the DSFC. Temperature
gradients within the optical port of the DSFC were determined to be small (<1 °C/mm)
thereby facilitating a more accurate experimental determination of localized thermal
history. The symmetrical nature of the apparatus, and the reproducibility and uniformity
of fi'eezing rates achieved on the inner ring of the DSFC, resulted in a much more
tractable computational problem compared to a conventional miniaturized ice-ball
approach, which can lead to significant improvements in numerical efficiency of

modeling efforts.

153

8.5 Bioheat Transfer Model Development

In previous experiments the AFA was characterized by obtaining thermocouple-
based temperature measurements on the surface of the AFA, on the surface of the DSFC
inner annulus (i.e. the side that contacts the tissue), as well as in phantom material and
living tissue within the DSFC. This experimental information can be used to develop a
numerical heat transfer model of freezing in the DSFC geometry that can provide field
information not possible using discrete thermocouples. The mathematical model can
subsequently be used to help design better in viva experiments. With knowledge of the
expected temperature gradients, thermocouples can be localized to areas that support
effective extrapolation or provide representative temperature measurements. A numerical
model can also be helpful for understanding how cryosurgical probes and window

chambers of various designs influence the tissue thermal response.

8.5.1 Mathematical Models of Biological Tissue Freezing in the Literature
Mathematical models exist that predict the thermal field in tissue surrounding a
cryosurgical probe (Cooper and Trezek, 1970). Numerical methods for solving
multidimensional phase change problems of inanimate materials (Rabin and Karin, 1993)
as well as biological materials with temperature-dependent thermophysical properties,
metabolic heat generation and blood perfirsion have been published (Rabin and Shitzer,
1998). Budman et al. (1995) developed a one-dimensional integral solution for the
inverse Stefan problem for a non-ideal binary solution with a phase change occurring
over a range of temperatures. Rabin and Shitzer (1995) published an exact analytical

solution to the one-dimensional inverse Stefan problem in tissues with a non-ideal phase

154

change over a range of temperatures. The same authors also produced a one-dimensional
combined integral solution in the frozen region and a numerical enthalpy-based integral
solution in the unfrozen region for the inverse Stefan problem for freezing/thawing in
biological tissues (Rabin and Shitzer, 1997). Non-ideal phase change over a range of
temperatures was included, as were the effects of metabolic heat generation and blood
perfusion in the unfrozen region. The author has also recently published a morphology-
specific solidification model of mammary tissue freezing based on the Neumann exact
solution for l-D fi'eezing (Elliott and McGrath, 1999). A numerical phase change
model, specific to the DSF C geometry, was published by Hoffman and Bischof (2001a).
These authors used a 2-D transient numerical solution incorporating blood flow for a
cylindrical geometry. While providing a useful conceptual framework, this body of
published work cannot be applied in any detailed fashion to the specific case of freezing
tissue within the DSFC as imposed by the AFA. Consequently a geometry specific
numerical solution was desired that could adequately represent the current situation.
Before proceeding to the geometry of the chamber, exact solutions from the literature
provided a means of benchmarking to verify that the numerical models were adequately

predictive.

8.5.2 Governing Equations

Prediction of thermal fields that are generated in biological tissues during
cryosurgery generally rely on solutions of conduction heat transfer governing equations
that include the effects of latent heat evolution that occur during phase change. The

initial condition is typically one of uniform temperature, To, ofien ambient temperature

155

(20-25°C) or physiologic temperature (37°C). In practice the material is subjected to a
low temperature probe located centrally or distally, which is generally modeled as either
a temperature or heat flux boundary condition. These can be steady or transient boundary
conditions. Beginning from ambient or physiologic conditions, energy in the form of
sensible heat is thus removed from the tissue/material until the solidification phase
change temperature (Tf) is reached. As the material begins to solidify, latent heat is
released in the process.

The geometry of a typical l-D problem is shown in Figure 8.5.

 

To

/

S m l

 

 

—>x

 

Figure 8.5 Schematic illustrating solidification of a homogeneous semi-infinite
material with a phase change region. TS represents the surface temperature imposed and
To the initial temperature. The mushy region (m) is identified by the positions of the
solidus and liquidus phase change boundaries.

The governing equation for heat transfer in the solid region is given by the simple

heat conduction equation:

156

 

 

ps 5 a, ’ ax 8 6x 0<x<X., [8.1]

where p, , ks, and cs, are the density (kg/m3), thermal conductivity (W/mK), and specific
heat (J/kgK) of the solid, respectively. These properties may vary spatially and can be
temperature-dependent. The position of the solidus interface is given by X,. The
governing equation in the liquid region will contain a buoyancy term that represents
convection due to a change in density as the material changes from solid to liquid. This
term can be neglected if the change in density is not significant. The energy equation in

the liquid region, including this term, is:

6T] 6T] 0 I 5T1]
plCIBT-plcluzza- kl};- XI< X<°°’ [8'2]
where the subscript 1 denotes the liquid values of the previously defined properties, and u
is the bulk velocity of the liquid region where X. defines the location of the liquidus
interface.

Although pure materials undergo phase change at a single temperature, many
materials solidify over a temperature range. Thus as the material changes phase three
distinct regions are observed: a solid region, a liquid region, and a region that contains
both the solid and liquid phase. This concept was illustrated in Figure 8.1. A liquidus
phase change temperature, Ta, and a solidus phase change temperature, Tfs, are defined to
describe the thermal limits of the phase change process. This region is commonly termed

the ‘mushy zone’. The governing equation for the mushy region is given by:

157

 

6x2 + k dt =0r 6t x,<x<x. [331

where subscript m refers to the mushy zone properties, fS is the fraction of solid in the
mushy zone, and L is the latent heat of fusion (kl/kg). The thermal diffusivity, denoted by
or, represents the thermal conductivity divided by the product of density and specific heat.
The various solutions available for solidification problems can generally be classified in
terms of how the mushy region is modeled. Specifically they vary in how the

contribution from the latent heat term is handled.

8.5.3 Latent Heat Effects

Biological solutions and tissues contain water, salts, proteins, sugars and many
other bio-molecules. In such complex mixtures, freezing does not occur at a single
temperature but rather spans a range of temperatures, and the evolution of latent heat has
a functional relationship with temperature in this freezing range (V oller and Prakash,

1987)

8.5.4 Benchmark Solutions

Analytical solutions do not exist for complex geometries such as those of the
DSFC. Consequently a numerical solution was desired to provide thermal field
information for freezing in DSF C tissue. To generate thermal field information of tissue
frozen in the DSFC, a computational fluid dynamics (CFD) software program, FluentTM

5.0 was utilized. Fluent utilizes the finite volume method (FVM) of discretization based

158

on a fixed grid to arrive at a numerical solution to the governing equations of the field of
interest. Fixed grid methods are routinely used for numerical solutions of phase change
phenomena (V oller et al., 1990). The alternative is to use a front-tracking method that
employs a deforming grid scheme. The front-tracking approach works well for
isothermal phase change of a pure material. However, for a multi-component
solidification problem involving a mushy region, tracking an interface becomes
computationally challenging (V oller et al., 1990). Other commercial packages can be
used or a specific numerical analysis code can be written that adequately generates
solutions to solidification heat transfer problems. FluentTM was chosen on the basis of its
availability, and the author’s familiarity with this software package.

Before beginning to develop complex geometries for FluentTM analysis it was
necessary to ensure that the numerical solver was producing accurate results. The
disadvantage of numerical models is that they frequently produce ‘reasonable’ results
regardless of how accurately the problem is specified. Consequently it is the task of the
user to validate their accuracy by whatever means possible. A benchmark case was
solved using three different l-D solutions, an apparent Cp method in Fluent, the
simplified 2-phase Neumann solution (Lunardini, 1981) and the 3-phase Cho and
Sutherland solution (Cho and Sutherland, 1969). The properties used in the solution are

given in Table 8.1.

159

Table 8.1 Properties and Conditions Employed in the Benchmark Solution

 

 

 

 

 

Initial Temperature (K) 310
Upper Limit of Phase Transition (K) 272
Lower Limit of Phase Transition (K) 265
Surface Temperature (K) 77

 

Thermal Conductivity — frozen (W/mK) 1.88
Thermal Conductivity — unfrozen (W/mK) 0.61
Thermal Conductivity - mush LW/mK) 1.245

 

 

 

 

 

 

 

 

 

 

 

 

Specific Heat — frozen (kJ/kgK) 2.04
Specific Heat — unfrozen (kJ/kgK) 4.18
Specific Heat — mush (kJ/kgK) 3.1 1
Density (kg/my) 1000
Latent heat (kl/kg) 334
Time(s) 22

 

8.5.4.1 Apparent Cp Method with FluentTM
The apparent Cp method involves the representation of the latent heat effect as a

component of the specific heat, CL. That is:

CL = —— [8.4]

or some other function of temperature, where AT is the temperature range over which the
latent heat is released. The latent heat effect can then be incorporated into an ‘apparent’
Cp that will have a functional relationship with temperature that varies from phase to

phase. The apparent specific heat, C,, is thus given as:

160

 

I Cs (solid)
Ca =4 C“ :C' +CL (mush) [8.5]
CI (liquid)

 

Consequently the governing equation in the mushy region takes the form of that in the
solid and liquid region equations. The latent heat release was modeled as a piecewise
linear function between —8 and —1 °C with the peak CL(T) value occurring at —3 °C.
This functional relationship is plotted in Figure 8.6. This relationship is a geometrically
simple representation that closely mirrors the more complex pattern of latent heat
evolution. It was used by Rabin and Shitzer (1997) with good results. Bonacini et al.
(1974) demonstrated that the shape of the assumed curve does not matter as long as the
area beneath the specific heat curve accurately reflects the magnitude of the latent heat

effect.

 

90

Apparent Cp (kJ/kgK)

 

 

 

 

 

 

I-10-9-a-7Is-54-3-2-10
Templdegcl

Figure 8.6 Functional relationship of apparent Cp with temperature.

161

The geometry representing the physical situation was first constructed in
GambitTM, a preprocessor for CFD analysis. A 0.25 cm x 2.0 cm geometry was used. A
mesh resolution of 10 microns was chosen with quadrilateral volumes. This yielded a
5000 element mesh. Boundaries were assigned using the pre-processer, with one surface
at T5 and the remaining boundaries assigned as adiabatic. The solution was then
generated using a second order upwind scheme with a convergence criterion of 1x10'7

and a time step of Is.

8.5.4.2 Neumann Solution: 2-Phase

The Neumann solution, an exact solution for 2-phase solidification in a semi-
infinite tissue slab, was also used to solve the benchmark phase change problem
(Lunardini, 1984). In the Neumann solution the mushy zone is reduced to a single phase-
change interface. An illustration of this is given in Figure 8.7. The Neumann
formulation is characterized by a distinct moving phase change boundary at which the
heat in the solid and liquid phases must balance.

In the solid and liquid region the governing equations are given by Equations 8.1
and 8.2, respectively, however convection transport due to change in density is neglected.
In the following derivation the numerical subscripts are assigned as follows: (1) liquid (2)

solid. The energy balance at the phase change interface is given as:

3x ax - dt [8.6]

162

 

  

T. a X(t)

 

 

 

 

 

Figure 8.7 Schematic illustrating solidification of a homogeneous semi-infinite
material initially at T0 with a single-phase change temperature, Ty, and a temperature
boundary condition T=TS at x=0. The position of the phase change boundary is given by
X(t).

At the interface of the solid and liquid region the temperature is given by the

phase change temperature of the material:

T1(X,I) = T2(X9t) = Tf [8'7]

The governing partial differential equations can be transformed into an ordinary

differential equation using the following similarity transformation:

 

1 [8.8]

163

Direct integration of the resulting ordinary differential equation for either region, yields

the solution:

T, = C [(5213 [8.9]
The error function is defined as:
erfn=_2_ I’e-fi’dfi [8.10]
72

which, when implemented with the boundary condition at x = 0 allows us to express the

temperature distributions in each region as:

-T +———(Tr —T’)erf x

 

 

 

_ 3.11

l s 611(3) 2 alt [ l
(T —T,) x

T =T — ° rf 8.12

2 ° erfc(k,/0t:)[e c:2. alt] I I

The constant 7» can be determined from:

164

634'2 _kanaIz (To-'Tfk-m'212 = L’A/IT
erfOt) (r, —r,)erfc>.,/a,, CI(T. —T.)

 

 

[8.13]

where 0.122(11/(12, and k21=k2/k.. This constant can then be implemented in the

expression:

X = 21M [8.14]

to determine the location of the phase-change front with time.

The previously defined thermal properties and conditions were implemented in
Equation 8.13. Using MathematicaTM software to determine the root of this equation (see
Appendix E for program), the location of the solidification interface with time was
determined from equation 8.14. The position of the interface delineates the appropriate
spatial range for each of the temperature equations, thereby allowing the continuous

temperature distribution to be plotted.

8.5.4.3 Cho and Sutherland Solution: 3-Phase

Cho and Sutherland (1969) developed an exact solution to the Neumann problem
for the case of a latent heat being evolved over a phase change region. This solution is
reviewed in Lunardini (1981). In the following solution the numerical subscripts refer to
each phase as follows: (1) solid, (2) mush, (3) liquid. At the solid-mush and mush-liquid
interfaces, the temperatures are given by the solidus and liquidus phase change

temperatures of the material:

165

Tl(x59t) = T2(X89t) : Tfs

T2(Xl,t) = T3(Xl,t) 2 T0

The energy balance at the solidus front is given by:

5T2

 

k'gzk’ ax

Similarly, the energy balance at the liquidus front is given by:

x
—+p,dets atx =xs

[8.15]

[8.16]

[8.17]

[8.18]

In the same manner as the Neumann solution presented previously, the energy balance at

each phase change interface provides constants that are used to determine the position of

each interface. The resultant equations that must be simultaneously solved for A. and n

are:

 

 

 

 

 

2
expO. )+ fikflfl _
erfOt) 2C (T —T )(n—k)
2 fs 5
1f 4.2
Tfl —TfS +— exp ——
k2] c2 0L21 = lb/ffl-f)
a2] '1 c1(Tfs —Ts)
T -—T erf —erf
fl fs (1 a
21 21

166

[8.19]

and

 

 

 

 

 

_nz _nZ
[Tn —Tfs + ——)exp( I 1f (1,, exp[ J
2 an __ “(121 ___ (T —T )k “31 [820]
[erf n —erf 7t] 2( ) erf—a“
an 0'21 3'

Once these roots have been determined, they can be implemented to determine the

position of each interface according to:

X5 = 2}»,Iorlt [8.21]
XI = 2n,/a,t [8.22]

as well as the temperature profiles in each region. The final expressions for the

temperature distribution in each region are given by:

 

erf[2 x I
_ O. t
T‘ T5 = ' [8.23]

Trs — T erf)»

S

 

 

 

S

— -— +
Tf _T c2(Tf's _Tan-A) (Tfs _Ts)

S S

x A.
X If erf - ——
If —}t _ _ I—

 

 

 

[8.24]
n A

2a2t azt

erf

 

 

167

and

 

erfc
2 at2 t
T, 2T —(r —T,) [8.25]

II

 

erf

The MathematicaTM program written to solve for the roots and to generate this solution is

given in Appendix F.

8.5.4.4 Comparison of Results

The results of all three solutions are plotted together in Figure 8.8 for a solution at
t = 223. The solutions of Neumann and those of Cho and Sutherland are in very close
agreement. In the phase change region the two solutions are expected to vary, as the
latent heat is being released over a range of temperatures in the latter solution and at a
single temperature in the former. This is indeed the case, however the deviation is very
small, with a percent difference of < 2.5 % in the freezing range. The apparent Cp
solution using FluentTM is expected to more closely resemble that of Cho and Sutherland.
The results of the FluentTM solution are very similar to the 3-phase exact solution,
however the region of latent heat release does not track exactly. The maximum percent
difference between temperatures for these two solutions is 3.3 % and this occurs in the
phase change region. The average difference over the temperature range in the plot is
1.4%. These results suggest that the apparent Cp method with FluentTM is providing

acceptable estimates of temperature.

168

350

 

 

300 -

 

250 a eeeeeeeeeeeee , <75!" IIIIIIIIIIIIIII l

 

 

 

 

;
1
l
1
1
:
l
l
I
t
l
g Position(m)
l

a. l

E ggggggggggggggggggggggg t
g, .

'- l

i - Neumann : l

> l

50 _4 iiiiiiiiiiiiiiiiii J - Fluent-App Cp L l

%; PDOLSWCPBGJ i

0 ~ , T 3

0 0.002 0.004 0.006 0.008 0.01 1

g

l

|

Figure 8.8 Predicted thermal profiles using three different solutions for solidification
in a rectilinear semi-infinite geometry.

8.5.5 Thermal Response in DSFC Tissue: Apparent Cp Method

The apparent Cp method for a 1-D solidification problem (solved for a 2-D
geometry) appeared to give reasonable estimates of the thermal field when compared to
1-D exact solutions. This was taken as adequate verification that the numerical model
and apparent Cp approach was valid. It was assumed that using this same solution
method (F luentTM numerical solution with apparent Cp) would produce accurate results
for a 2—D solidification problem in a more complex geometry.

A simplified 2-D model of the DSFC and tissue was created using Gambit. Based
on the characterization work of previous sections, it was decided that the thermal
response of the DSFC was adequately symmetrical to model the tissue according to the

schematic in Figure 8.9. The tissue thickness varied in experiments. For the purposes of

169

understanding the temperature gradient through the thickness of the tissue as well as the

temperature distribution over the viewing field, a representative tissue thickness of 1 mm

was chosen. The thickness of the glass window in the DSF C was 0.9 mm. The mesh is

pictured in Figure 8.10.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Constant
L<__ Heat Flux
is” g m
, , Drawing Not to Scale
tissue
glass 3.”
tissue

 

 

'6— 0.0076 In

 

 

 

.<__ 0.0049m ——>[

 

_\L

Figure 8.9 Schematic illustrating the geometry used to represent a ‘A section of tissue
in the DSF C. The point of contact of the DSFC annulus, indicated with the dark dotted
line, was modeled as a constant heat flux boundary condition.

170

 

 

 

 

re:

Pr:

 

Figure 8.10 Mesh used to generate solidification solution.

Because the amount of tissue surrounding the DSFC varies depending on the
location, selecting the appropriate length of the tissue required some approximation. The
side of the annulus closest to the body of the animal might be modeled as a semi-infinite,
whereas because of the paucity of tissue on the uppermost side, this area can be
considered of finite length. A tissue length that was twice the diameter of the skin flap
tissue was chosen to represent the extent of tissue surrounding the chamber in an attempt
to create a boundary condition that represented an average of the two conditions
described.

The glass and tissue were meshed separately using a SO-micron triangular mesh.
This yielded a 5855-node mesh, with 6918 elements comprising the tissue, and 3930
elements comprising the window. The metal annulus was not represented by elements in
the mesh but instead was modeled as a constant heat flux boundary condition. This was a
reasonable approximation of the thermal situation and it was simpler to implement than a

prescribed cooling rate. All other boundaries not represented by a symmetry condition

171

were considered adiabatic. The heat flux boundary condition was adjusted in the model
until a cooling rate on the order of that seen on the surface of the annulus during
characterization measurements (~5 °C/min) was achieved. This resulted in a heat flux
condition of -1600 W. A second order upwind scheme was used to generate an energy
solution. The convergence criterion was set at 1x10'7 and time steps of 1 s were used.
Properties of water as used in the benchmark solution were utilized. The starting
temperature implemented was 295 K. Solutions were generated at periodic intervals and

are plotted in Figure 8.11.

 

i 300 4 |<-- Glass -—>|<- Ann ->|
l

 

 

 

 

 

 

Position (m)

l l
l ’ l
. l
l l l
l l
A 290
2 280 1 [I I!
3 l l ‘
E 270 l --1205 3 t
260 i l ' l
l |,
: g l 4805 fu
. l- 250 ‘ "‘ i l
i l 1 i
f 240 1 —600s ‘ 1
. I ,
1
g 230 I , l __. _ _ __J:
3 0 0.005 0.01 0.015 1
'l r
t. _. .'

Figure 8.11 Temperature profiles generated for phantom tissue (properties of water
used) in the tissue geometry depicted in Figure 8.9.

172

As expected, the temperature can be seen to be minimal directly beneath the
surface of the annulus, which is centered at x = 0.0625 m (see schematic in Figure 8.9 for
orientation). As cooling progresses the temperature gradient within the chamber is seen
to decrease. The rate of cooling in the tissue decreases when the phase change
temperature region is reached. Latent heat release in this region serves to slow down the
advance of the freezing front. By the time the chamber tissue reached —18 °C, the
temperature gradient was very flat, with a spatial variation of < 0.5 °C/mm. Even at the
beginning of the freezing process the temperature difference between the center and the
edge of the chamber was seen to be small (~5 °C after 30 s).

In the simulation the tissue surrounding the optical area of the chamber appears to
cool significantly as well. This has also been observed in the course of in vivo
experiments, as will be discussed in Chapter 9. Future probe or protocol designs could
benefit from modifications that minimize the cooling in the region outside the chamber.

The temperature distribution through the thickness of the tissue was studied to
understand if significant temperature gradients existed in the axial direction. It is
generally assumed that the temperature gradient is negligible through the thickness of
thermally treated intra-vital preparations of the type used in this work. Using the
numerical simulation, this assumption was explored for the present case involving a
relatively thick preparation (1 mm). For illustration purposes the predicted thermal field
at 2405 is displayed on the grid geometry in Figure 8.12. Evident fiom this plot is the
nature of the temperature gradient through the thickness of the tissue. The temperature
difference between the top surface of the tissue and the midline of the tissue (i.e. the

bottom symmetry plane in the mesh geometry) was calculated for various times and the

173

results are plotted in Figure 8.13. Temperatures were seen to be effectively constant in
the axial direction, suggesting that measurements taken on the surface of the tissue would
be representative of those throughout the thickness. The largest temperature difference
was observed around the edge of the window, suggesting this would not be a good
location for a thermocouple if a representative tissue temperature was desired.

The temperature difference across the thickness of the glass was also examined.
The difference in temperature between the top surface and the bottom surface of the glass
window was determined and is plotted in Figure 8.14. Again the temperature can be seen
to be effectively uniform through the thickness of the glass, with the temperature
differing by less than 0.2 °C. The area near the edge of the DSFC annulus has a higher
variation in temperature through the thickness of the glass, as was observed for the tissue.
Still this variation is less than 1 °C.

Two key observations come from these simulations: (1) the temperature
throughout the thickness of the tissue and glass appears to be uniform, and (2) the
temperature gradient is less than 1 °C/mm in the window chamber. These are important
observations that can be used in the design of experiments for future studies. The
prediction of uniform temperature through the thickness of both the glass and tissue is an
interesting result relevant to the potential use of optical techniques that can measure
surface temperatures in these systems. This observation was consistent at all time points.
These results suggest that optical methods for measuring surface temperature would be
highly successful in this system, as the measured surface temperature would be

representative of the tissue temperature, and extrapolation or interpolation would not be

necessary.

174

T: 2816+02
2. 796+02
2. 786+02
2. 7760-02
2. 7564-02

3
f 2. 749+02

 

 

 

 

'- 2,726+02
27164-02
2.708t-02

_ 2. 68et-02

 

 

Ej 26794-02

 

Contours or Static Temperature (k) (rtme=2.4000e+02) Sep 23, 2001
FLUENT 5,5 (20, segregated. lam. unsteady)

 

 

 

Figure 8.12 Temperature distribution in DSFC tissue (properties of water used) as
simulated at 240 s.

175

 

 

 

 

 

 

 

1 """ 1
08 |<———GIau———->|<—Annulus—>| 1 . 305 l
A . . l‘ ’
1&06 iiiiiiiiiiiiii :_,2,,,z,,‘
0
l 304 ,,i,,,,,,s,_,{',h2,2,,,1”205
2
g 0.2 ‘
° 0240s
2 0
3 l 6 11‘
E-OZ , . , ' H l
E I I l .3605
o-O.4< 777777777777 .222i2iq 1
I- . I
_6_,,,,,,,,-2,22,2,, 7 7~ ,
4) i -600s
-0.8 L1.)
Position(mm) }

 

Figure 8.13 Temperature difference between the top surface of the chamber geometry
(glass and annulus regions) and the tissue midline as a function of temperature.

 

 

Temperature (K) A

 

 

 

 

 

 

0 0.001 0.002 0.003 0.004 0.005
Position (m)

 

 

Figure 8.14 Temperature difference between the top and bottom surfaces of the DSFC
glass window as a function of time and radial position.

176

Chapter 9

Application of the GFP Intra-vital Viability Assay:
Injury Following Slow Freezing to
High Sub-Zero Temperatures

9.1 Motivation

The deleterious effects of freezing on biological tissue can be sub-divided into
two main categories, direct cell injury, and secondary injury associated with micro-
vasculature damage and host response. Although the nature of direct cell injury has been
well-studied (Meryman, 1970; McGrath et al., 1975; Mazur, 1977; Bischof et al., 1997)
comprehensive information on the contribution of secondary injury to the ultimate size of
the cryosurgically induced lesion is not available (Gage and Baust, 1998).

Because of the high correlation between intracellular ice formation and cell death,
cryosurgery instrumentation is generally designed to maximize this mode of direct cell
injury. Cryosurgical probes are created with increasingly colder operating temperatures
and increased cooling power (Baust et al., 1997). However, when applying rapid
freezing rates and low temperatures to tissues using a conventional cryosurgical probe,
basic heat transfer principles dictate slower freezing rates and high sub-zero temperatures

in distal areas of the tissue as a natural consequence of the treatment. Treated tissue thus

177

experiences different thermal histories at different spatial locations and this thermal
history will depend on the thermal properties of the tissue, as well as the location, shape,
and cooling power of the cryosurgical probe. This general concept is illustrated in Figure
9.1 for a cylindrical probe. Although this clinical approach yields acceptable levels of
certainty about cell death at the center of the lesion where the probability of intracellular
ice formation is high (Whittaker, l974ab), in peripheral areas the injury is complex and
not very predictable. Consequently clinical protocol involves extending the ice ball into
the healthy tissue circumscribing the tumor to ensure that critical isotherms (discussed

shortly) reach the edge of the diseased tissue.

 

Cryosurgery Treatment

 

Sub-Cutaneous Lesion

 

 

Zone 1: Rapid Freezing Rates,
Low Sub-Zero Temperatures

Zone 2: Slow Freezing Rates,
High Sub-Zero Temperatures

 

 

 

 

Figure 9.1 Conceptualization of a cryosurgically treated subcutaneous lesion.
Regions close to the probe will experience rapid cooling and reach temperatures close to
the temperature of the probe; tissue further away fiom the probe will experience slow
cooling rates and will have temperatures closer to the phase change temperature of water.

178

 

 

R
th.
17?,
not

Clin

Many early in vitro studies suggested that —20 °C was the necessary critical
temperature to achieve certain cell death (Mazur, 1963; Meryman, 1956; Cooper, 1964),
hence many clinical protocols were designed to ensure that the entirety of the lesion
experienced temperatures at least as low as —20 °C. More recent in vitro studies suggest
that this critical temperature is as low as — 40 °C (Tatsutani et al., 1996; Roberts et al.,
1997) or — 50 °C (Jacob et al., 1985), and in their review on mechanisms of
cryosurgically relevant tissue injury, Gage and Baust (1998) recommend a critical
temperature of —50 °C to be certain of achieving a good clinical outcome when treating
neoplastic tissue. To achieve this temperature at the edge of the lesion requires the
fi'eezing of a large margin of healthy tissue.

Although the nature of injury at the edge of a cryo-lesion is not well understood, it
is recognized that micro-vascular damage occurring at slow freezing rates and high sub-
zero temperatures may contribute to tumor de-vitalization, and therefore can figure
significantly into the size of the ultimate lesion that develops. Consequently, estimation
of cryosurgical success using a lethal temperature prescribed on the basis of intracellular
ice formation in vitro can under-predict the amount of damage that will occur in viva.
Recent attempts to understand the lethal temperature for cell injury in viva suggest that
the lethal temperature in viva is considerably higher than prescribed on the basis of in
vitro tests (Hoffman, 1999, 2001ab; Schuder, 2000). The results of such studies should
not necessarily result in the prescription of a higher sub-zero temperature criterion for
clinical usage, but rather serve to provide a more comprehensive understanding of how
all of the various cellular responses inter-relate to create the final lesion. In this manner

rationally based adjuvant therapies or variations of the traditional algorithm can be

179

identified and implemented to possibly improve clinical outcome and prevent secondary
recurrence. Understanding the relationship between thermal history and tissue injury in
viva, in lieu of in vitro extrapolation, would thus help advance the state of the art in
cryosurgery research.

Tumor cells are generally more resistant to intracellular ice formation (IIF), as
evidenced by both direct experimental observation and a trend towards higher membrane
water permeability values (McGrath, 1988). Cells with a greater ability to dehydrate in
the presence of extra-cellular ice are less likely to super-cool and the formation of
intracellular ice is thus less probable. In light of this tendency, the secondary role that
vascular injury plays in destroying the treated neoplastic tissue is important to understand
so that in the development of new therapies appropriate end-points for evaluation of
therapeutic success are chosen. Also, if microvascular damage is a significant mode of
injury, adjuvent therapies such as the use of anti-angiogenic drugs may be desirable to
improve clinical success. Combined chemical and cryosurgery therapies such as this are
currently under investigation (Baust et al., 2000; Clarke et al., 2001). It is especially
important to understand the balance of primary and secondary contributions to tissue
destruction where treatment of aggressive and rapidly proliferating cancers is concerned.
If the time course of injury following cryosurgery is such that the destruction of cells by
ischemia is outstripped by the rapid growth and spread of these cells into a well-
vascularized region, then cryosurgery without other interventions may not be the best

treatment option.

180

9.2 Relationship between micro-vascular injury and thermal history
in viva

It is well known that the thermal processing conditions causing cell death vary
enormously from cell-type to cell-type. The relationship between freezing rate and
survival for various cells has been comprehensively studied. Figure 9.2 illustrates this
relationship for a number of different cell types. The variation evident in this figure
demonstrates the difficulty in generalizing the relationship between freeze injury and
thermal history in just a direct cell injury sense. Cell injury is not just a firnction of
freezing rate. It is also a fimction of the minimum temperature, the thaw rate, and the
hold time at a given temperature, and these variables are often intricately interrelated.
For example, a recent study indicates that the minimum lethal temperature for human
primary prostatic adenocarcinoma cells depends on freezing rate (Tatsutani et al., 1999),
suggesting there isn’t a single critical isotherm associated with cell death. In addition to
these thermal variables there are certain host features that become important in viva. The
local tissue environment can impact how well a cell can recover from an applied freezing
stress (Jacob et al., 1985). Immune response can result in cell de-vitalization above and
beyond that inflicted directly by freezing. The local blood vessel architecture can also
figure significantly into the problem by creating heat sinks that make it difficult to cool
certain regions of tissue, leading to thermal histories that are difficult to quantify and
relate to clinical outcome. Understanding the importance of thermal history variables, in
viva, where the effects of cell-communication, cell immune response, and micro-vascular
contributions are in play, is an important goal if the state of the art in cryosurgery is to be

advanced. However, uncoupling direct cellular injury in viva fiom the secondary cell

181

injury brought about by a myriad of host factors can be a daunting task. The technologies

available to researchers have precluded such a comprehensive investigation to date.

SURVIVAL 00
§

1?

20-1

tO-t

 

 

 

O! ‘0 0.000
COOLING RATE ('C/mtnl

Figure 9.2 Relationship between freezing rate and survival for various cell types.
From Mazur, 1970.

It is well-established that malignant tumor tissue requires a significant blood
vessel supply to meet the high metabolic demands of continuously dividing cells (e.g. Li
et al., 2000). A number of interventive therapies exploit this fact. For example, anti-
angiogenic drugs are specifically designed to inhibit the growth of new vessels
(angiogenesis) thereby preventing the tumor fi'om uninhibited growth. When cells aren’t
adequately nourished they will die from ischemic injury, a lack of adequate nutrition.
This concept is illustrated in Figure 9.3. Ischemic injury is hypothesized to contribute in
part to the overall damage that accrues during cryosurgery, as impaired blood flow has

been demonstrated to be a consequence of the freezing process. For example, Brown et

182

al. (1993; 1994) have demonstrated reduced and complete shutdown of perfusion in cryo-
treated liver, and cremaster muscle, respectively, following cryosurgery. Also, there
exists a preponderance of evidence in the frostbite literature that implicates the role of
micro-vascular damage in freezing induced injury. Although frostbite injury is generally
associated with slow freezing to sub-zero temperatures that are considerably higher than
typically desired for a cryosurgery treatment, fi'ostbite injury does mirror the character of
freezing at the margin of a cryo-lesion. Despite the cornucopia of studies detailing the
existence of micro-vascular damage and reduced or absent flow for many hours following
cryosurgical intervention, none of these studies provide direct evidence that necrosis is
caused by ischemia and not other injuries such as slow freezing ‘solutes injury’ damage,
or an increased cryo-sensitivity of cells growing in an in viva environment. Consequently

the balance of direct cellular injury versus ischemic necrosis is not well understood.

 

Blood Vessel
Damage

___—1

    

 

 

 

Figure 9.3 Illustration of ischemic injury caused by damage to blood vessels.

183

Most attempts to study the relationship between thermal history and vascular
injury that leads to tissue necrosis have the general weakness of inaccurate temperature
sensing. Although there have been several studies done in the last 2 years that attempt
to quantitatively correlate thermal history and micro-vascular injury (Hoffman et al.,
1999, 2000ab; Schuder et al., 2000), the experimental design and temperature sensing
technologies utilized resulted in an ambiguous determination of lethal temperatures for
freezing induced micro-vascular injury.

In the study by Hoffman et al. (1999, 2001ab), a dorsal skin flap chamber (DSF C)
system was utilized with an AT-l tumor in the Copenhagen rat. Thermocouples
embedded in the tumor tissue, using a polycarbonate window as a guide, were used to
measure the temperature history at several radial locations during fieezing with a small
diameter cylindrical cryoprobe. Because the thermocouples were mounted perpendicular
to the tissue surface, fin effects were significant hence a theoretical correction based on a
steady state conduction fin model was applied to the experimental measurements. These
data were compared with two simple analytical heat transfer models to interpolate the
temperature field between the discrete thermocouple measurement sites. This thermal
information was linked to in viva measurements of vascular disruption determined by
macromolecule leakage and to histological assessment of the tissue after sacrificing the
animal. It was found that the region of vascular stasis (determined by macromolecule
leakage) correlated with the region of tissue necrosis (determined by histology) for both
normal tissue and tumor tissue. Correlation of lesion size with thermal modeling suggests
that the minimum temperature required for tissue necrosis is in the range of -10°C to -

13°C which is a much higher range than the temperatures near -40°C required to kill

184

single cells of this cell line in vitro. The results suggested that vascular injury may be the
dominant mechanism for determining the size of the cryolesion due to freezing in viva.
However, difficulties in making the thermocouple measurements as well as modeling
assumptions and simplifications in the thermal models combined to produce an estimated
temperature uncertainty of at least 7°C. Also, no direct in viva evidence was given to
support the hypothesis that cells in the lesion periphery did not die from direct cell injury.
The steep gradients within the small working area of the DSFC make it very difficult to
localize the lesion periphery and even more difficult to obtain accurate temperature
measurements. With these limitations it can be inferred that the minimum temperature
required for AT-l tumor tissue necrosis in viva was in the range of —5 °C to —20 °C. The
cooling rate experienced by the tissue was not explored.

Despite the fact that the specific goal of the Schuder study (Schuder et al., 2000)
was to determine a lethal temperature for vascular injury in a rat hepatic model, the
attempts to measure temperature were coarse. Although they used a plastic guide for
thermocouple placement normal to the surface, no adjustment was made for fin effects.
Details of the thermocouple size and construction were also not provided; hence other
errors may be present in the data The high treatment temperatures they predict as
characterizing the edge of the ultimate lesion (~8°C), were likely caused by conduction
along the thermocouple wires that caused a warming of the thermocouple junction.
Although the intra-vital aspect of their work was sound, the correlation with temperature
can be considered qualitative at best.

In an earlier study Rabb (1974) studied the effects of fieezing and thawing on the

microcirculation in a hamster cheek pouch using a cold block freezing device with

185

fieezing rates controlled by regulating the flow of cold helium. In this model the cheek
pouch was everted and pinned over the cold block. A 0.05” thermocouple was placed
between the pouch and block and this was used as the basis for thermal control. No
experimental or theoretical validation was presented that indicated this temperature was
representative of the temperature history throughout the thickness of the sample, nor was
there any mention of the thermal contact resistance that would be inherent in such a set-
up. In light of this, one would expect that the temperatures recorded were probably
colder than those actually experienced in the bulk of the tissue. However, the protocol
used involved freezing at 1 °C/min therefore at this slow rate, barring significant contact
issues, it is not unreasonable to assume that the thermal history in the tissue was
reasonably close to that reported. Although this study demonstrated that micro-vascular
changes are extensive and would likely lead to cell necrosis, at —30 °C it is not
uncommon for cells to already be damaged from the direct effects of fi'eezing. No
evidence was presented that demonstrated cells were not already dead from direct cell
injury (solute injury or otherwise).

The study by Le Febvre (1971) also lacks detail about the thermal history actually
experienced by the tissue undergoing cryosurgery. Using the hamster cheek pouch
method, different depths of cooling were achieved in the tissue via the use of a stainless
steel trough circumscribing the cheek pouch preparation. A solution consisting of 40%
propylene glycol in Ringer’s solution that had been cooled in an alcohol-dry ice bath was
used to charge the trough and elicit temperature changes in the tissue. Because a
combined forced-air vacuum system was used to change the freezing solution from 37 °C

Ringer’s solution to the cooled propylene glycol solution, the trough temperature likely

186

responded quickly and might be modeled as a step change in temperature. However, the
thermal diffusivity of the tissue is an order of magnitude less than that of the stainless
steel and thus the tissue will experience a transient change in temperature. The freezing
rate in the tissue was not reported, nor was the spatial distribution of temperature in the
optically studied section. However, the strength of this paper lies in the correlation
between duration of freeze at given trough temperature and tissue necrosis as determined
over a 30-day period following cryosurgical treatment. What can be concluded from the
reported results is that —15 °C is the critical ‘trough’ temperature below which a
significant degree of necrosis can be expected with a minimum time of exposure.
Complete necrosis was reported for any exposure to —18 °C trough temperatures and
below. If these trough temperatures are accurate, the critical ‘tissue’ temperatures are
expected to be somewhat warmer. At such high sub-zero temperatures, generally
speaking, cell viability in vitro is quite high (e.g. Elliott et al. 2000), and thus the necrosis
observed can be supposed to be originating primarily from the observed microcirculatory
stasis and associated ischemia.

The Zacarian et al. hamster cheek pouch study (1970) was similar to that of
Hoffman et al. (1999) in the sense that a ‘traditional’ cryosurgery was performed. In this
case however a single thermocouple was used to record the temperature at various
locations distal to the cryosurgical probe. The thermocouple was located in the plane of
the tissue, beneath the surface of the cheek pouch. Temperatures reached in the tissue
were likely colder than actually recorded by the temperature sensor, as the cryosurgical
probe was applied on the opposing side of the tissue. The authors noted that ‘a single

freeze-thaw experience between the temperature range of -20 to -30 °C to the vascular

187

network of the cheek pouch affords as much lethality as would be expected from extreme
temperatures, suggesting the dominant role that micro-vascular effects can play in tissue
necrosis. Temperatures in the —20 to —30 °C range can also elicit direct cell injury,
making it difficult to discern the actual contribution of micro-vascular injury to overall
tissue necrosis.

Based on these previous studies that have examined cellular and micro-vascular
injury following freezing at high sub-zero temperatures, it appears that the temperature
regime least likely to cause direct cell injury, but still having a high likelihood of causing
stasis and significant micro-vascular injury is the —10 to —20 °C regime. Studies in this
temperature regime may be able to effectively uncouple direct cell injury from secondary
effects and allow the investigation of both the extent and the time course of cellular
injury following freeze-induced ischemia. At these temperatures, some cell death may
occur, but the bulk of cells in the tumor mass should remain viable immediately afier the

initial freeze-thaw cycle.

9.3 Patho—physiology of Cellular and Vascular Injury Following
Freezing

The general pattern of vascular injury following freeze injury was delineated in
the first chapter of this dissertation, as gleaned from both the frostbite and the
cryosurgery literature. As a tissue is cooled, the initial thermoregulatory physiologic
response is increased blood flow to the tissue (Weatherley-White, 1969). Continued
cooling produces vasoconstriction and a decrease in blood flow. When the phase change
temperature is reached, freezing is expected to begin to freeze first in the extra-cellular

space (Rubinsky et al., 1987, 1988, 1990). Depending on the rate of cooling and the

188

temperature reached, ice may also form intracellularly. During thawing, circulation is
temporarily restored to the previously frozen tissue. However, progressive failure of the
micro-circulation is observed. Vascular distention has been observed both in frozen
section (Rubinsky et al., 1990, 1997) and intravitally (Schuder et al., 1999). Edema
develops (Weatherley-White etal., 1964; Reite, 1965; Marzella et al., 1989; Zook et al.,
1998; Manson et al., 1991; Zook et al., 1998), vascular permeability is increased (Waris
and Kyosala, 1982), impaired platelet function and platelet aggregation leads to
thrombosis and vascular occlusion (Weatherley-White etal., 1964; Reite, 1965; Zacarian
et al, 1970; Bourne et al., 1986; Marzella et al., 1989; Zook et al., 1998), and endothelial
cells die (Waris and Kyosala, 1982; Marzella et al., 1989; Manson et al., 1991). The
increased vascular permeability appears to be a result of damaged endothelial cells and
defects in the endothelial cell junctions. This endothelial damage also results in platelet
aggregation and micro-thrombus formation. This, in turn, causes blood flow stagnation
(stasis) that ultimately deprives cells of oxygen, nutrients and a sink for waste products.
The cells die of ischemia. The time scale on which damaging micro-vascular changes
occur has been observed as early as immediately after freezing (endothelial cell damage)
by Manson et al. (1991) or 3-min afier thawing (small vessel occlusions) by Bellman et
al. (1956) to as long as 24 hours after freezing (necrosis of endothelial cells) by Waris
and Kyosola (1982).

While this general pattern is understood, the specific thermal history, or the
minimum thermal insult to elicit this general response has not been well-studied. Also
the bulk of studies have demonstrated that micro-vascular injury exists, but few have

provided any direct evidence that micro-vascular injury is responsible for tumor cell

189

death. The lack of efficient methods to study the time course of cellular injury has
precluded researchers from directly establishing the relative contributions of direct
cellular injury and the secondary effects of vascular injury and ischemia to overall lesion
size. Also because of the lack of correlation with thermal data, cryosurgical technology
has not advanced to the point where the secondary aspects of cryosurgery injury can be

predicted and controlled.

9.4 Hypothesis and Experimental Design

There were several objectives in this experiment. The first was to determine the
extent of microvascular damage that can be achieved by slow cooling to —18 °C in viva.
The second objective was to provide the first direct in viva evidence that cells in tissue
frozen slowly to high sub-zero temperatures are not immediately injured by the freezing
insult. If the microvascular damage from this protocol was determined to be substantial,
a final objective was to determine the time course of cellular injury that might occur due
to the possible secondary effects of ischemia.

Dorsal skin flap chambers bearing tumor slurries as described in Chapter 7 were
implanted in eight animals. Five chambers were treated with a freezing protocol using
the annular freeze apparatus described in Chapter 8, and the remaining three animals
were used for sham experiments, in which identical procedures were used except the

annular freeze apparatus was not cooled.

190

9.5 Materials and Methods
9.5.1 Chamber Implant and Tumor Propagation

Similar to the methodology described in Chapter 7, female Fischer 344 Rats
weighing between 150 and 180 grams were implanted with dorsal skin flap chambers,
according to the procedure described by Papenfuss (1979). In the chapter 7 animal
series the thickness of the tissue was unrestricted because the non-dissected side was not
sealed in any matter. In the current animal series the traditional double-sided preparation
was constructed, with epidermis being removed fiom both sides of the skin flap. This
technique involves dissection of opposing skin flaps, leaving a thin layer of vascularized
dermal tissue. At the time of chamber implant a small bolus (~200 uL) of GFP-
transfected R3230 AC tumor cells was injected into the chamber before sealing the
preparation with a glass window. The tumor cell suspensions were prepared by counting
cells with a hemacytometer, using Trypan Blue to discriminate live and dead cells, and if
necessary concentrating by centrifugation to achieve cell counts of 1-2 x 105 cells/mL in
D-MEM. In previous work, this protocol has resulted in high tumor take rates and
predictable tumor grth (Elliott et al., 2000).

Animals with dorsal skin flap chambers were maintained in a 90 °F room at
ambient humidity. The warm room was necessary to maintain circulation in the skin flap
chamber. The integrity of tissue in the chamber was macroscopically monitored until
tissue could be seen to entirely fill the viewing area. This generally required a minimum

of 10 days. The tissue was then treated with freezing.

191

 

 

It

Ex.

001

Ce”

9.5.2 Development of Freezing Protocol

To optimize the operating conditions of the annular freezing apparatus for intra-
vital studies, three animals were implanted with dorsal skin flap chambers containing
tumor slurries and incubated for 20 days. These animals were used to understand the
temperature distribution that could be expected in viva. Animals were anesthetized with
sodium pentobarbital (35 mg/kg) and placed on their side on a stack of towels.
Polyethylene bags containing 37°C water were used to prevent hypothermic shock in the
rat, and these were changed frequently. The window on one side of the chamber was
removed and replaced with the thermocouple grid window described in Chapter 8.
Calibrated T-type thermocouples constructed from 0.005” wires had been previously
localized at radial positions of 2, 3, and 4 mm (+/- 200 microns). Use of this grid
minimized the problem of thermocouple positioning error, as well as positioned the
thermocouples in the most favorable way for minimizing fin effects error (i.e. in the
direction of minimal temperature gradient). After threading the thermocouple wires
through the center of the annular fi'eeze apparatus (AF A), this fieezing device was
attached to both sides of the dorsal skin flap chamber. Cooling was initiated and

temperature measurements recorded with a DAS-TC/B board.

9.5.3 Measurement of Thermal History During Intravital Studies

For intra-vital studies, the animals were identically prepared and kept warm,
except the thermocouple grid window was not implemented. The glass window on the
one side of the chamber was removed and replaced with a glass window containing a

central l-mm hole. A T-type thermocouple was inserted into the hole and sealed laterally

192

against the surface of the window using a thin layer of transparent nail enamel. The e-clip
used to keep the window fixed in place in the DSF C facilitated positioning the
thermocouple in the manner depicted in Figure 9.4. The lead wires were then threaded
through the annular freezing device and attached to the appropriate T-type connecter.
Measurements were recorded manually at 30-s intervals using a hand-held Omega model
CL24 calibrator-thermometer (Omega Engineering Inc., Stamford, CT). Cooling was
initiated and maintained until an end temperature of —l 8 °C was reached at the central
thermocouple. Cooling was then discontinued and the tissue was allowed to passively

thaw with the annular freeze apparatus remaining in place.

193

 

   

T-type
/ Thermocouple

Aluminum

   

b

Figure 9.4 (a) Photomicrograph demonstrating the location and positioning of a
single thermocouple for in viva temperature sensing. (b) Schematic of thermocouple
positioning for the purpose of minimizing fin effects of in viva temperature
measurements.

9.5.4 GFP-based Intravital Microscopy

A freezing protocol, designed to achieve —18 “C in tissue at the center of the
DSFC with a minimal temperature gradient across the entire viewing surface, was applied
to five animals. Where possible a selected area in the central region of the tissue was
monitored for both GFP intensity and vascular injury by alternating between fluorescent

and transmitted light. These circumstances occurred in only two out of five animals.

194

Some preparations did not adequately grow into tumors. Others were not of suitable
clarity for vascular monitoring. The maximum information possible was obtained from
each preparation. The group of three sham animals were prepared identically to the
experimental group animals but cooling was not initiated in the AFA. None of these
animals grew tumors, but all provided good microvascular information.

Cell viability was monitored using the GFP intra-vital technique before, during,
and after the cryosurgery procedure for pre-determined observation periods to determine
if cell death is immediate and associated with direct cell injury fiom the fieezing insult,
or delayed and consistent with ischemic injury. Images were acquired before, during, and
after freezing using a SPOT thermoelectric cooled color CCD camera and a Nikon
Diaphot inverted fluorescent microscope equipped with a 100 W mercury arc lamp (both
from Diagnostic Instruments, Inc. (Sterling Heights, MI). The tissue was illuminated
with blue light by inserting a FITC filter cube (ex. 450-490 nm, em. 510-540 nm)
(Chroma Technology Corp., Brattleboro, VT) into the light path.

The blood vessels were also monitored for evidence of stasis before, during, and
afler the cryosurgery procedure, using only brightfield illumination. lrnages were
acquired at predetermined times over a lO-hour period to allow for offline analysis of
blood vessel dimensions. These images were later analyzed for evidence of vascular

distention.

9.5.5 Histology
At the end of the lO-hour observation period the animal was immediately

euthanized by cardiac puncture and injection of an overdose of sodium pentobarbitol.

195

 

 

rn

\Pi

1(

the
and
11111:
5qu
Bee;

aHal]

 

The tissue within the chamber was fixed in 10 % formalin for a minimum of 20 hours and
then embedded in paraffin. Tissue sections of S-micron thickness were obtained and

stained with hematoxylin and eosin.

9.6 Results

9.6.1 Cryosurgical Protocol

Using the grid window for thermocouple measurements, it was found that similar
to results with phantom tissue, the DSFC tissue in living animals could be reproducibly
cooled to high sub-zero temperatures with a minimal temperature gradient across the
field. As can be see in Figure 9.5, thermocouples positioned at the center and at a 2 mm
radial position, record nearly identical profiles. This was true in all three animals that
were studied. The thermocouple positioned at 4-mm recorded temperatures that were
seen to deviate from the central thermocouple by up to 2 °C (excluding the region of
latent heat release). Unfortunately the 4 mm thermocouple was damaged in two of the
three experiments. Repeated attempts were made to maintain three working
thermocouples in the configuration described. This task proved to be extremely difficult,
and therefore measurements were focused on the central region of the tissue where
intravital imaging was performed. The single result from the 4 mm thermocouple does
suggest however that the temperature gradient within the chamber is less than 1 °C/mm.
Because of the small field examined during intravital studies (<1 mm), the area of

analysis can safely assumed to be isothermal.

196

 

TemEFaEeTC)

 

 

 

 

0 2 4 6 8 10 12
Time (min)

Figure 9.5 Characteristic freezing curves measured in viva at specific locations.

For comparison purposes the previous results were plotted together with the
predicted temperature at x = 0 and x = 0.004 m originating from the numerical model
developed in Chapter 8. These combined results are given in Figure 9.6. In the
experimental case neither the thickness of the sample nor the thermal properties were
known precisely. However, since the water content of biological tissue is generally quite
high, the simulation should be reasonably representative of the experimental situation,
certainly in an order-of-magnitude sense. When comparing results it could be seen that
the simulation predicted that the temperature difference between the center and edge
thermocouples would be minimal at later times (< 2 °C), and this was indeed observed to
be the case in experiments. There is a considerable deviation when comparing these

results in the region of phase change. Whereas the model predicts a smooth evolution of

197

latent heat over a 7 °C freezing range, results measured in samples have been seen to
have considerable variability in terms of the onset of freezing. In the case of the
experiment shown in Figure 9.6, freezing did not begin until —8 °C. At this temperature
all of the latent heat in the model had been released already. It should also be
emphasized that the constant heat flux condition chosen to represent cooling at the
surface of the annulus is an approximation. Because the AF A system is itself at ambient
conditions at the beginning of the procedure, the initial cooling load delivered to the
system first cools the hosing and attachments of the AF A before beginning to extract
significant heat from the annular components of the device. This results in a heat flux
condition that will increase as the system approaches it steady operating condition.
Results of the simulations are thus expected to be less representative of the experimental
situation at early times. In future models, precise cooling histories could be
implemented in the model, via the use of user defined-firnctions (separately compiled
programs). Also, for purposes of comparisons, the use of materials with well-defined
thermal properties would be more suitable. Despite these limitations, the results shown

in Figure 9.6 demonstrate reasonable agreement at the later times.

198

 

 

 

 

 

300i'a. --
I
A 2 2 o
g 280 ; ........ r ............ .
2 : Q t A
3 ....................... O a,
E 260 '
a
E 240 eeeeeeeeeeeeeeee , r
220 1 ex=4 mm Predicted Ix=0 mm Predicted 1. 3;,
1 ax=4 mm Measured ex=0 mm Measured1
200 " *"“‘r;r*r“****fiA‘¥
0 200 400
Time(s)

 

 

 

Figure 9.6 Experimental measurements for the in viva case presented in Figure 9.5

are plotted against predicted measurements (properties of water used) for two locations
on the surface of the tissue.

9.6.2 Thermal History Measured During Intravital Studies.

Characteristic results of temperature measured at the center of the tissue during
the freezing protocol are shown in Figure 9.7. This general pattern of freezing was
observed in each animal. In Figure 9.7 it is evident fiom the temperature data that latent
heat was evolved at ~ -8 °C and that thawing occurred around 0 °C. In all cases it was
possible to identify the latent heat release region based on the sudden spike in the
temperature data.

The onset of freezing occurred within 8.5 to 10.5 minutes afier

cooling was initiated, which corresponded to an onset temperature range of —1.0 to —9.7

°C. The starting temperatures of all the skin flaps were between 25 °C and 28 °C. The

199

nominal fieezing rate was determined to be ~ 4 °C/min. The nominal thaw rate was also

~ 4 °C/min.

 

Temp (C)

 

 

 

 

" Time (min) __

Figure 9.7 Characteristic thermal history measured using a T-type thermocouple in
contact with the tissue at the center of the annulus in the DSFC during slow freezing.

200

9.6.3 Post-Freezing Vascular Changes

Images that were acquired during the experimental observation period were
analyzed for blood vessel diameter using Image Pro Plus soflware. Two to four vessels
were identified in each series of images. At every time point, five measurements of
diameter were made of each vessel at the same location for both sham animals and
treated animals. The vessel diameters of sham animals were observed to undergo very
little deviation over time and so for presentation purposes in Figure 9.8, the results of
sham animals were averaged. The vessels in the treated animals were all examined
individually to define the extent of distention and the nature of the variation amongst
vessels. Experimental samples were not pooled because the combination of unequal
sample numbers per time point and the large variation between vessels obscured the
observed response. The results displayed in Figure 9.8 demonstrate that vessel distention
following freezing is a frequently observed phenomenon, with 7 out of the 8 vessels
demonstrating vessel dilation. The vessel-to-vessel variation is very high with dilation
observed to span the range of approximately 1.1 to 2.5. In some cases the amount of
vessel distention over time does not appear to be steady, and appears to increase with
time post-thaw. In these vessels the time at which maximal dilation is seen to occur (5-
10 minutes post-thaw), coincides with the onset of secondary stasis in these vessels (4.5—
6 minutes post-thaw). In other cases, by the time vessels could be resolved in images (~
3 minutes after thaw) they already appeared distended. This tended to be true only in
cases where the vessel distention never exceeded 1.5. Some extravasation of red blood
cells was observed to occur immediately post-thaw, but severe hemorrhage did not occur

until 30 to 40 minutes post-thaw.

201

 

 

 

 

 

1 1
. 1*‘——‘ _ ___ _._-. _ _ _ — -_-_.-_- 1
1 1 o 9_6_A_a I 9_6_A_b 6 14_A_a 1 1
1 3 1 x 6_14_A_b x 9_5 A_a o 9_5 A_d 1 1
1
1 g l + Sham Average - 9_5_A_b - 9_5_A_c ‘ 1
1 1.__.---*.__~_‘—d*Iw<~II'—i—"*‘*'"i‘*"'T’TTT
-2.5 -
1 >1! 1 -F 1
1 h a =- 1
1%5 2 3 '
l
1
13 (g g I g ‘ E 1
_ 3 : .
1 § 31.5 .: x 1 1
= 7". "‘ I z" I
> E g I I! . I 1
o 1 #i‘fiiiii i g I 1
l 2 1
1 l
1 0.5 1
1 l
, l
1 o . . 1 1
1 0 5 10 15 20 25 30 1
1 Time Post-Thaw (min) 1

 

Figure 9.8 Individual vessel measurements from treated animals, normalized to the
starting diameter of each vessel, are compared to a baseline of pooled sham animal data
For the sham animals the error bars represent the standard deviation of three animals.
Remaining measurements represent the average and standard deviation of five repeated
measurements of the same vessel.

9.6.4 Time Course of Injury Based on GFP Intra-vital Viability Assay

Only two animals in the treated group were found to have suitable tumor growth
for analysis. As can be seen in the image series in Figure 9.9, fluorescence loss from
GFP-transfected tumor cells was gradual, with a minimal loss minutes after freezing, and
progressive changes occurring over several hours. The change was seen to be most

dramatic in the first three hour time period. By two hours post thaw, the skin flaps were

202

observed to be extremely swollen and in some cases the tissue was beginning to blacken,
consistent with the gross character of frostbite injury and appearing consistent with the in
viva observations. To quantify the time course of GFP intensity change in the two treated
animals that grew tumors, gray scale intensity values normalized to the starting intensity
values were acquired of the central region of the tumor tissue in each image. These
results are plotted in Figure 9.10. The GF P signal reduction for both animals is very
similar, with very little change occurring after 4 hours. The tumor appears to have a
considerable amount of viable cells immediately after the thaw, but these cells slowly die

over the subsequent three hours.

203

Start 5-min.
10-min. 30-min.

 

10 hr.

 

2 hr.

Figure 9.9 Micrograph series of GF P intensity change over time in a centrally located
DSFC tumor treated with slow freezing to —1 8 °C.

204

..L 1
N 1
l

 

 

 

 

l

1 .! __....- l
g 106_14_Al 1
c 1 .................... l I
1l8_15B 1

20.8 I V, , ‘
1nI 1
113 e.- .
lgoea—‘i _________________________ 1 1
1:- e l
10 O 1
30.4 - ' -2, .7 7777777777777 1
1.! . . . - . 1
'5 1
[E02in?_*17-7__,fi-i-i-~-L-H 1
o l
12 1
3 ° . 1 . - '1
1 0 100 200 300 400 500 600 700 1
l

Time(min) 1

I 1

Figure 9.10 Normalized gray scale intensity plots of GP P fluorescence intensity loss of
cells following slow freezing to —1 8 °C.

9.6.5 Histology

As can be seen in Figure 9.11, histological sections from sham-treated tissue
demonstrated vessels of normal patency. Fibroblast cells in these specimens had the
expected spindle morphology and flattened nuclei. In contrast, histological sections from
treated tissue were seen to have dilated and engorged vessels and significant hemorrhage.
Figure 9.12(A) illustrates the extent of red blood cell extravasation that was typically
observed. The engorged vessel seen in Figure 9.12(A) appears to have been

mechanically damaged, perhaps fiom dilation beyond physically tolerable limits. The

205

nuclei were seen to be condensed in treated tissue, consistent with tissue necrosis. These
histology images were consistent with both gross observation and intra-vital
measurements and observations, as both vessel distention and severe hemorrhage was

observed, the latter being resolved at both microscopic and macroscopic levels.

206

 

(b)

Figure 9.11 Micrographs of sham-treated tissue from the dorsal skin flap chamber,
showing no evidence of necrosis, and blood vessels of normal patency. (a) Transverse
(*) and longitudinal (“) sections of blood vessels can be seen. (b) Fibroblast nuclei
(arrows) are normal in appearance.

207

100 pm
as A “

 

Figure 9.12 Micrographs of dorsal skin flap chamber tissue frozen at ~ 3 °C/min to —
18 °C. (a) Congestion and engorgement of blood vessels (‘) and condensed nuclei
(arrows) were observed, indicating both microvascular injury and tissue necrosis. (b)
Extravasated red blood cells can be seen to dominate this severely hemorrhaged section.

208

9.7 Discussion

The results of this study suggest that temperatures as high as -18 °C can cause
severe disruption of microvasculature integrity and functionality, which may be effecting
cell changes that manifest on a time scale consistent with this injury, i.e. some hours
delayed. Vessel dilation was observed to occur post-thaw to varying extents, as
evidenced by both offline analysis of intravital images and histopathology specimens.

The intravital microvascular observations of this study are very similar to those
reported by Rabb et al. (1974) in their studies in the hamster cheek model. They
observed constriction of venules and arterioles and sluggish blood flow as the tissue was
cooled, with freezing taking place from —3 to —9 °C. Upon thaw, circulation commenced
from 0 to —3 °C. Consistent with our observations, they also reported vasodilatation
beginning shortly after thaw and continuing throughout the observation period.
Hemostasis was observed to begin at 1 minute afier thaw with flow stoppage in small
vessels by 20 minutes attributed to platelet occlusions. In the current study the earliest
episode of secondary stasis that occurred could be seen 4.5 minutes following thaw.
Although they reported return of blood flow after 30 minutes, this was not observed in
our experiments, as vessels remained static throughout the ten-hour observation period.

Shuder et al. (1999), in their investigation of the response of hepatic
microcirculation to cryosurgery, found complete stasis at the center of the lesion directly
following freeze injury, whereas the sinusoids at the edge of the lesion were
heterogenously perfused. This region experienced progressive failure as parenchymal
cell aggregates lodged in the sinusoidal lurnens and caused terminal blockages.

Microvascular perfusion failure was associated with Ito cell disintegration and possible

209

loss of Ito—cell regulated vasomotor control. Consistent with the results of the current
study, marked dilation of sinusoids (15.6 vs. 8.8 microns) was also observed following
freeze-thaw.

This increase in sinusoidal diameter was also observed by Rubinsky (1987) in
studies on freeze-substituted liver sections that had been previously directionally
solidified at various fi'eezing rates. Rubinsky subsequently modeled this response as a
biophysical response (1990). The general basis for his theory is that during slow freezing
ice crystals would form first in the microvasculature, resulting in an exclusion of solutes,
thereby establishing a concentration gradient that would result in water transport into the
blood vessels from the extra-cellular space. This model therefore predicts an increase in
the diameter of the sinusoids as a direct consequence of this water transport, consistent
with his experimental observations in frozen section.

The Rubinsky theory does not address the expected thawing response, but it is
possible that water will leave the vascular when the system melts and returns to ambient
temperatures. Whether the vessels are dilated or not when the system thaws is not clear
from the studies done on liver tissue, therefore it is difficult to extrapolate to the current
case. It is worth noting that liver tissue has the unique feature of being a relatively
homogenous tissue compared to many other tissue types. Hepatic tissue is uniformly
perfused with vessels of very similar diameters. In contrast the tissue studied in this work
has vessels of many sizes coursing through the tissue and are very irregularly spaced.
The fact that the window chamber is implanted in a wound results in a dynamically
changing vascular bed. The relevance of this environment to a ‘normal’ tumor

vasculature needs exploration from the standpoint of mass transfer, but also begs the

210

question of how other tissues compare to the hepatic blood vessel architecture, and
whether the Rubinsky model can be considered valid for other tissue types. In any event,
if the freezing events have created mechanical damage to the microvasculature, the stress
on the vessels appears to be further augmented by dilation post-thaw. The likely
mechanism for the post-thaw swelling is the plugging of vessels and subsequent
mounting of hydrostatic pressure gradients as blood fi'om the uninjured microvasculature

tries to make its way into this region.

9.8 Summary and Conclusions

This pilot study provided the first *direct* in viva evidence that cells frozen to —
18°C at a slow freezing rate are not being immediately damaged by the freezing protocol,
but that the damage manifests at a later time. Only two animals were able to grow tumors
adequately for GFP intensity changes to be studied together with vascular changes,
therefore, statistical conclusions cannot be drawn about the GF P signal from this small
subset. The information provided from these pilot studies, together with experience
gleaned from the animal series described in Chapter 7, can however be used to develop
strategies for future experimental design. The GF P assay provided a measure of injury
dynamics in the living system that prior to now has not been possible. This represents a
significant advance in uncoupling the effects of direct and indirect cellular injury
following freezing. Based on these results, cell injury associated with high sub-zero
temperatures and slow freezing rates can be seen to occur on a time scale of several
hours. This is not inconsistent with the predictions of Hoffman and others that

microvascular injury is indirectly bringing about the death of cells that would otherwise

211

survive the initial freezing insult. However, the timing of viability loss does not preclude
the effects of solute injury, which have been observed in this study to progress over a
period of hours following hypertonic exposure. The possibility of osmotic and thermal
shock induction of apoptosis is also not ruled out. The pilot results obtained do suggest
however that the relevant time scale to explore these slow freezing primary and
secondary injury mechanisms is on the order of hours and not days. This information
alone can directly improve the experimental efficiency of firture studies, independent of
the particular hypotheses of injury that might be explored.

Although —18°C was the temperature used in this study, it is possible that higher
sub-zero temperatures may bring about as much lethality.

The vascular injury model was shown to be highly reproducible with all treated
tissues demonstrating signs of severe vascular hemorrhage. Tissues with vessels of
adequate resolution for monitoring blood flow were seen to undergo secondary stasis in a
very narrow time range post-thaw, specifically, 4.5 to 8 minutes. Marked vessel dilation
(~2-fold) was observed in many of the vessels, though this was not true for all vessels.
This variation may be due to the different vessel sizes analyzed as well as variation in the
distance between vessels. These results did not support nor refute the hypothesis of
Rubinsky that distention is occurring during freezing, as it was not possible to measure
dilation during the frozen period as the ice obscures the sample. Some vessels appeared
to be consistent with the model of water transport into vessels during freezing as they
appeared dilated directly after thawing, yet others appeared to dilate over time and
appeared to coincide with the onset of secondary stasis. It is likely that both mechanisms

are contributing to the overall microvascular injury that occurs.

212

Chapter 10

Summary, Conclusions and
Future Work

10.1 Summary

At the onset of this work, it was hypothesized that GF P transfection of tumor cells
prior to introduction into animal models could provide the basis for a tissue viability
assay in the mature tumor. It was speculated that the presence of GFP within transfected
tumor cells would indicate that two features of cell viability exist: a metabolically
functioning system capable of producing the GF P and an intact cell membrane. Because
of the unique tertiary structure of the GFP molecule, we postulated that the molecule
itself would remain stable with respect to freezing and that the attenuation of intracellular
GFP fluorescence would occur when protein synthesis was disrupted or when membrane
damage caused leakage of cytoplasmic contents, including the GF P molecules. Freezing
is known to affect the cell in such a way that the membrane is either directly impaired, or
the cell is damaged such that necrosis and/or apoptosis will prevail, inducing a cascade

that inevitably leads to both metabolic failure and eventual membrane degradation. The

213

 

 

 

results obtained following freezing of both isolated cells and tumors grown in the rodent
dorsal skin flap chamber confirmed these hypotheses.

It was established that GFP fluorescence is lost from isolated transfected
R3230AC cells following exposure to cryosurgically relevant freezing and thawing rates.
The presence of GFP fluorescence in these tumor cells correlated very well (0.9] < R2 <
0.94) with several standard cell survival assays such as ethidium bromide and Trypan
Blue. This level of correlation compared very favorably with correlations made between
non-GFP-based viability indicators (Yang et al., 1998), suggesting that the GFP-based
approach is comparable to established standard methods. GFP fluorescence was also lost
following exposure to hypertonic stresses comparable to those experienced during these
freeze/thaw conditions. The correlation between GFP fluorescence as a measure of
viability and the other common viability indicators used was very high not only for static,
end point measures of viability but throughout the dynamics of the apparent loss of
viability.

The GFP-expressing rat tumor cells were used to produce fluorescent tumors in
viva in a rodent-implanted dorsal skin flap chamber. At the time of chamber implant, the
tissue inside the chamber was bathed with a solution of tumor cells in media. The animals
were placed in a warm room (91°F) for 6-11 days to grow large tumors (~ 10 mm
diameter) in the skin flap window chamber. The tumors were then treated using three
cycles of freezing and thawing. GFP intensity fields were measured before, during, and
after the cryosurgery protocol. The tissue was also biopsied for hematoylin and eosin
(H&E) staining. Histopathology evaluation presented evidence for direct cell injury at

locations near the cryoprobe where the tissue is expected to experience the most rapid

214

freezing rates. In these regions the histology revealed nearly complete ablation of the
tumor tissue, with only nuclear and cytoplasmic remnants scattered within a congested
connective tissue matrix. The GFP fluorescence signal was also diminished to
background levels in these same central regions. The loss of fluorescence occurred
within seconds to minutes. Thus, cells that presented as definitively dead in the
histopathology also presented as dead in the intra-vital GF P microscopy. Likewise,
outside of the region of treatment, cells unaffected by therapy as demonstrated by
unaltered levels of GFP fluorescence presented as viable and highly mitotic in
histopathology. The histopathology revealed a mixed population of healthy cells and
cells in various stages of necrosis (evidenced by nuclear pyknosis, karyorrhexis, and
cytoplasmic eosinophilia) in a band between the definitively dead tissue and the healthy
tissue in the margin. A region of similar ‘mixed’ character (average GFP signal
diminished to levels above background) was identified during intravital GFP microscopy.

It was hypothesized that the GFP assay would have characteristics that likened it
to two classes of existing viability assays: membrane integrity assays and protein
synthesis assays. The detection of intracellular GF P directly following insult would
correspond to retention of membrane integrity. Although membrane integrity is generally
accepted as a reasonable indicator that the cell will survive, it is a necessary but not
sufficient condition. Because the transfected cells continuously produce GF P, the
maintenance of GFP synthesis following insult was hypothesized to be a second indicator
of cell survival. To implement this feature accurately it was first necessary to understand
the half-life of the GF P inside the cell. Experiments with cycloheximide, a protein

synthesis inhibitor, were undertaken to measure the half-life of GFP in R3230AC cells.

215

 

The half-life was not definitely established, as cells began to die and leak their contents
before GF P turnover was observed. However it was established that the half-life was
greater than 16 hours.

Experiments were performed to establish the stability of pure recombinant GF P to
freezing. Successive rapid plunge freezing of aqueous solutions of rEGFP in liquid
nitrogen and slow thawing at ambient conditions revealed a minimal loss of fluorescence
intensity after three cycles, suggesting that cold denaturation of the GF P molecule was
not occurring to an appreciable extent under these conditions. An analysis of
extracellular green fluorescence intensity before and after cell lysis revealed an increase
in background fluorescence intensity that was directly proportional to the volume fiaction
of GFP-transfected cells in the starting cell suspension prior to lysis. Consequently a
conservation of fluorescence was established, indicating that intracellular GFP was not
being destroyed during this process.

A number of supporting studies were also completed that provided the foundation
for the work previously described. Linking the thermal history of the tissue during
cryosurgery to the biophysical and physiological responses is understood to be critical to
a better understanding of tissue freezing and thawing. The biophysical characteristics of
the GFP-R3230AC line were determined for individual cells to predict biophysical
behavior. These results indicated that the cell membrane permeability for cultured cells
was quite high (Lp = 302 urn/s, 22 °C) suggesting that rapid fi'eezing rates will be
necessary to induce intracellular ice in these cells. The magnitude of the activation
energy (Ea = 4.2 kcal/mol) together with the high Lp values, were consistent with the

presence of aquaporins.

216

An annular cryo-probe system was developed that allowed the application of slow
freezing rates to high sub-zero temperatures. The initial design objective was met and
freezing rates between 1 and 10 °C/min were achieved. At the fastest pumping rates the
freezing apparatus was capable of achieving a surface temperatures of —130 °C on the
surface of the annular probe, and when in contact with the dorsal skin flap chamber, ——
110 °C on the inner surface of the chamber. Temperature gradients within the optical
port of the DSFC were determined to be small thereby facilitating a more accurate
experimental determination of localized thermal history. A numerical model of freezing
in the dorsal skin flap chamber was developed and compared to discrete thermocouple
measurements in phantom systems. Because of the symmetrical nature of the apparatus
and the fact that reproducible and uniform freezing rate results were measured on the
inner ring of the DSFC, the temperature profiles could be computed for a ‘A section of the
DSF C. These results were representative of the temperature profiles in the firll geometry
but were computed with less computational time than a full grid would have required.
The similarity of experimental and computational results suggested that the numerical
model was suitably accurate to use for future experimental design.

The concluding work in this study was the application of the new GF P viability
assay together with the new freezing device to evaluate the hypothesis that cells in the
periphery of cryo-lesions (i.e. high sub-zero freezing rates) do not die fi'om the immediate
effect of freezing but are the result of delayed or secondary effects (solute damage or
ischemia/inflammation). High sub-zero temperatures and slow freezing rates were
chosen because they were not expected to induce significant intracellular ice formation.

In the eight animals studied the microvascular injury induced was significant. Vessel

217

dilation, stasis, edema, and hemorrhage characterized the vascular lesion. In the two
cases where GF P tumors were grown and treated using this same protocol, GFP
fluorescence intensity in the cells did not decrease immediately following thawing, but
instead the fluorescence was gradually lost over several hours. These preliminary results

indicated that the time scale for this cellular injury was ~2 hours.

10.2 Conclusions

The principal conclusion based on the results of in vitro studies with GFP-
transfected R3230AC cells was that GFP fluorescence could be used as a viability
indicator to monitor the dynamic loss of viability in a population of cells with a
predictive ability equal to that of conventional assays. This work was the first application
of a GFP-based viability assay to cryobiological problems (Elliott et al., 2000b).

Studies on the application of the GFP viability assay to tissues also established a
good correlation between the loss of fluorescence intensity fiom tumors grown from
GFP-transfected cells and tissue damage as assessed by histological methods. To the
author’s best knowledge this work represented the first time the validity of this assay was
established in tissues, and as well the first utilization of the GF P viability protocol for in
viva assessment of tissue damage following injury (Elliott et al., 2000a).

In terms of the nature of the viability assay, i.e. whether it functioned as a
membrane integrity assay, as a metabolic indicator, or as a combination of both, it was
established that its primary utility was as a membrane integrity assay. Because the half-
life of GFP inside the R3230AC cells was found to exceed 16 hours, this suggested that if
freezing (or some other stress) impaired protein synthesis without damaging the

membrane, the GFP signal would likely not decay significantly until necrosis progressed

218

to the stage where cell leakage occurred. The continuous nature of the assay would thus
be in the tracking of membrane degradation over time. These results suggested that in
instances where membrane damage was not expected after an injurious stress, but instead
some other component of the cell stressed or injured (eg. nuclear damage), the
appropriate time to assay for lethality of injury would be greater than 16 hours after the
insult. More specifically it would be on the time scale of necrosis and/or apoptosis in that
system.

It was concluded on the basis of pure recombinant GF P protein stability studies
together with a conservation of fluorescence signal analysis that the mechanism
responsible for the attenuation of the GF P signal following cell injury was diffusion and
dilution. A basic l-D model of GFP diffusion in aqueous solutions and in hindered
cellular environments yielded time constants that agreed in an order-of-magnitude sense
with characteristic attenuation times generally observed in experiments. All the
experiments and analyses performed were consistent with a leakage, diffusion, and
dilution model of GFP fluorescence attenuation following injury. Cold denaturation of
the protein was ruled out as a significant mechanism of fluorescence attenuation under
the circumstances studied.

Although the conclusions presented so far represent the most significant
contributions of this work, several other pieces of work were completed in support of
both the short-term and long-term goals of this project. The therrnophysical
characterization work that was performed as part of this comprehensive study allowed the
conclusion that fast fieezing rates (order 100-1000°C/min) were necessary to induce

definitive IIF injury in cultured cells. However, the extrapolation of in vitro behavior to

219

in viva behavior is not straightforward. It is possible that the high permeability
associated with cultured R3230AC cells will not translate into identical freezing behavior
in viva. The interesting result that came out of this study was that the cultured cells
appear to possess AQP channels for transporting water, and it is therefore likely that the
in viva cells possess the same channels. If the water channel transport is the dominant
means of moving water across the cell, it can be hypothesized that the in viva tumor cells
will have high water permeabilities and consequently also require very rapid freezing
rates to induce significant IIF within cells in the tissue. Differences between cultured
cells and in viva cells that might result in differences in passive diffusion of water across
the cell membrane may not be relevant for water transport when AQPs are present. This
hypothesis that in viva cells will parallel in vitro cells in their freezing response remains
to be tested however.

The AF A device that was developed was able to achieve freezing rates and
temperatures in the range desired, without sacrifice of optical access in the DSFC.
Moreover a significant portion of the tissue could be treated uniformly, thereby
minimizing the challenges associated with measuring temperature in small systems
amidst large temperature gradients. This development represents a contribution to the
state-of-the-art in small-scale cryosurgical probes, as well as extends the utility of DSFC
preparations. There were limitations with this device however, the most significant being
the magnitude of the injury it induced. By virtue of the device design, the tissue
surrounding the experimental region was also damaged to the same extent as the tissue in

the experimental window. In instances where the volume of tissue damaged can come to

220

bear on the variables being studied, this AFA leaves little room for control, and
consequently the AF A would not be reccomended for such studies.

The new GF P technology that was developed in this work was used together with
the novel AFA to perform a pilot study on a relevant problem in the cryosurgery field,
namely freeze injury at high sub-zero temperatures. At temperatures of —18.0 °C
significant vascular injury was observed, supporting the hypothesis that vascular injury
can contribute significantly to the overall lesion that forms following a cryosurgery,
consequently killing cells that otherwise would be expected to remain viable. Although
only 2 of the 8 animals that were kept in the study group grew tumors of suitable
fluorescence for transient injury observations, the GF P attenuation response post-injury
was very similar in the two cases. Complete attenuation was seen to occur by ~2 hours.
This was a somewhat unexpected result, and suggests that although vascular injury may
be playing a role, it is possible that the cells may by dying as the result of a more direct
mechanism. Despite the lack of statistical significance due to the small group size (n=2)
these preliminary results are valuable with respect to future experimental design.
Experiments to delineate the nature of this high sub-zero injury should not focus solely
on vascular injury as the dominant mechanism, without first ruling out such possible
alternatives as delayed primary injury (e. g. osmotic shock induced apoptosis or

accumulative solute injury).

10.3 Future Work

While GFP technology has revolutionized biological research in many application

areas, we believe that we are the first to use it as a tissue viability assay and the first to

221

apply it to cryobiology problems. The results presented here suggest that the GFP-based
viability assay is not limited to isolated cells but can be used as the basis of a tissue
viability assay as well, and moreover can be used in viva. This tool should allow the
dynamics of cellular injury to be quantitatively evaluated continuously and dynamically,
in viva, without sacrifice of the host animal, wherever optical access is possible. This
opens the door for obtaining unprecedented information in time and space about
cryosurgical tissue injury. The application of GFP technology to a number of
fundamental and applied research areas in the field of cryobiology appears to have
excellent potential. Moreover, this assay has possible utility for assessing viability
following a diversity of tissue injury modalities. The next steps of this research would
involve exploring the application of this assay to other modes of injury. Because other
applications may involve injury mechanisms and tissue changes that could impair GFP
diffusion in viva, it would be prudent to explore GFP diffusion dynamics in the context of
several different types of injury.

Both modeling and experimental evidence suggested differences in the threshold
levels to be used for the in vitro and in viva experimental cases. The signal-to-noise ratio
in the in viva case was poor (2:1) due to the background green autofluorescence of
biological tissue. In future work it would be advisable to explore different colors of
fluorescent protein to improve the signal-to-noise ratio in viva.

Long-term follow up studies in viva would be advisable to evaluate re-growth
potential after cells are deemed non-viable by the GF P assay. All cell culture experience
to date has implicated no potential for re-growth once GFP has been lost as a result of

injury, however this hypothesis remains to be tested in any long-term study in an animal

222

model. Because of the extent of the edema seen in the skin flaps following fieezing it
would be advisable to implement an intra-vital tissue model that is not constrained inside
a device for such studies. The liver model of Shuder et al. (2000) is one such example,
however to the author’s knowledge this system has not yet been suitably developed for
long-term or chronic studies. Using an in viva model that is not as prone to hemorrhage
during instrumentation with thermocouples would also be desirable. A preparation that
does not require removal of a window could improve the experimental efficiency of the
DSFC system as utilized in this work. Establishing the limitations of the GFP technique
for trans-epithelial imaging without a dorsal skin flap chamber could open up potential
new intravital applications of the GF P assay.

Perhaps the single greatest advance in this area would be the development of non-
contact methods for measuring thermal field information in viva, in conjunction with the
intravital information. The AF A developed in the present work creates a uniform
temperature field during slow freezing of tissue within the DSFC. This allows
thermocouples to be used to quantify local temperature and freezing rates with a low
sensitivity to placement error. Also, the optical access provided by the annular shape of
the device allows thermocouples to be placed in a manner that minimizes fin effect error.
If this device is implemented for rapid freezing, the temperature gradients are not
expected to be as uniform, and therefore accurate temperature sensing becomes more
challenging. In experimental situations where the environment is not amenable to
accurate thermocouple measurements, other methods of temperature sensing are

necessary. Avoiding the use of thermocouples also minimizes the risk of microvascular

223

hemorrhage. Novel methods of measuring temperature that require less manipulation of
the DSFC or the underlying tissue would be advantageous.

Initial feasibility tests performed by other researchers in the Low Temperature
Biology Lab, suggest that infiared thermography and F ERIT technology can be applied
successfully to measure the temperature field on the surface of the window of the dorsal
skin flap chamber or on the rat skin tissue itself if the window is removed. It has been
shown that these thermography methods can be performed during freezing and thawing
of tissue and tumors within the dorsal skin flap chamber. The application of
thermography has several advantages compared to the use of thermocouples. The method
is non-contacting, the method is not subject to the fin—effect artifacts that arise from the
use of thermocouples and the data are in the form of a complete field rather than discrete
points. The continued development and application of these technologies would help
improve the ability the correlate thermal history information in time and space with
viability information, could improve the overall efficiency of the experimental systems

when using the DSF C to study thermal injury.

224

APPENDICES

225

APPENDIX A

DIFFCHAM Program

226

(*DiffCham_Check*)
(***********************#************************************)
(* This program checks the results of Diffcham *)
(************************************************************)
Areacellinit = .

Concsaltext = .

Concsaltextinit = .

Concsaltinit = .

Lp = .

meod = .

Nsaltinit = .

R = .

ri = .

t = .

T = .

Vb = .

vsalt = .

Vbcell = .

Vbcelli = .

Vwinit = .
(********************************************$***************)

(*Parameters that generally remain constant“)

R=8.31 (*J/K/mol“) (*Universial Gas Constant*);

(*Properties requiring definition per cell type“)

Lp=l 3.2 (*um/min/atm“) (*Membrane Permeability Constant*);
Vb=0.33(*dimensionless) (*Osmotically-Inactive Volume F raction*);
T=295 (*K*) (“Temperature");

ri=9.05 (*um") (*Initial Cell Radius“);

(*Experimental conditions that must be specified*)

(*Define initial extracellular salt concentration“)
(*Note: Conversion of Osm to moles/um3 with 10'15 factor“);

Concsaltexti = 0.308 * 10'l5 (*Salt conc, moles/pm“);
vsalt = 17.7 * 10"2(*NaC1*)(*um3/mole*);

(***************************************************************)

227

(******************************************************************)

(* Adjustment for mass transfer through the dialysis membrane *)
(******************************************************************)
Clear [Concsaltext];

Concsaltext = (0.308 + 0.592*(1-Exp[-0.038*t]))*10‘15;

3
(*Salt conc, moles/um *)
(*******************************************************************)

Plot [Concsaltext, {t,0,180}, PlotRange ——> {0.0,1 * 10"5},
AxesLabel —* {“Time (Sec)”, “Ext Concn (moles/um3)”}]

E‘xt Concn (moles /tm3)
lxlO-151

8x10‘1fl
6x10'161 _
4xlO—l61/

~16.

2x10 1

1
IILMLMH n+_ LIL LMII.IMI.‘ Tune (88:)

25 50 75 100 125‘ 150 i75

(*Necessary Conversions“)
meod=R*T*Lp*(10'8)*(1/60)*(10'5)(* um‘Vmol/sec”);

("‘10'8 for m3 to um3; 1/60 for nrin to see; 10'5 for atrn to (N/m2)*);
Vcelli=(4/3)"'3.14"‘ri3 (*assuming spherical, um“);
Areacellinit-——4"'3.14"‘ri2 (*umz“);

Vbcell=Vcelli*Vb (*um *);

(*Define me0d*A factor; (m4/mol/sec)*um2*) ;

(*Calc initial volume of cell water, urn3; Assume Vcell=Vw+Vsalt+Vb*)
Vwinit=(Vcelli-Vbcell)/( 1 +(vsalt*Concsaltexti));

(*Calc initial number of moles of intracellular salt*);
Nsaltinit=Concsaltexti*Vwinit;

(*Define instantaneous intracellular salt cone (mol/um3)*)
(*Where y[t] below is Vcell“)
Concsaltint=(Nsaltinit)/(y[t]-Vbcell-Nsaltinit*vsalt);

Clear[y, sol]

sol=NDSolve[{y’[t] = = -meod*Areacellinit*((Concsaltext)-Concsaltinit),
3'10] = = Vcelli}. YItl, {L0,180}l

(********************************************************************)

228

 

Plot[Evaluate[y[t]Ncelli)/.sol], {t,0,60}, PlotRange -+ {0.0, 1.0},
AxesLabel -—>{“Time (Sec)”, “Normalized Volume”}]

Nomalized Volume
1 .

0.81
0.6- ' I
0.41
0.2

. ,.,.- ~ **‘*"TJ'ITE(SeC)
10 20 30 4O 50 60

229

APPENDIX B

Vc__Freeze Program

 

230

(*Vc_Freeze*)

(************************************************************)

(* This program simulates the biophysical response *)

(* of cells to freezing. *)
(************************************************************)

ELp=.
ELp l =.
Lp=.
Lpg=.
meod=.
B=.

R=.

T=.

TR=.
ri=.

H=.

HI =.
A=.

V=.

Vi=.

Vb=.
Vbcell=.
vs=.

vw=.
concexini=.
concex=.
nsalt=.

(************************************************************)

(* Parameters that generally remain constant *)
(************************************************************)

R=8.314 (*J/K-mol*) (*Universial Gas Constant*);
vw=18.0*1012(*p.m3/mol*);
H1=333 (*latent heat of fusion; kJ/kg“);

(********t***************************************************)

231

(************************************************************)

(* Reference Conditions *)
(************************************************************)

Lpg=7.466 (*um/min-atm“);
TR=273.15 (*Temperature in K*);

(**********************#*************************************)

(* Cell Parameters *)
(************************************************************)

ri=9.0 (*um“);
ELpl=4.2 (*kcal/mol*);
Vb=0.33 (*osmotically-inactive volume fraction*);

(************************************************************)

(* Experimental conditions that must be satisfied *)
(*********************************#**************************)

concexini=0. 154* 10'15 ;
(*initial extracellular salt concentration; um3/mole*)

(************************************************************)

(* Conversions to ensure consistent units *)
(*******#****************************************************)

ELp=ELp1*4.184*103 (*J/mol*);
H=H1*18.015 (*J/mol*);

(************************************************************)

(* Calculations *)
(********************************#***************************)

A=4*3.14*ri2 mm“);
Vi=(4/3)*3.14*ri3;

Vbcell=Vb*Vi;
nsalt=concexini*(Vi-Vbcell);
LP=LPg*EXP[(ELp/R2*(( 1[TR)-(1f1"))];
meod=Lp*R*T* 10l (* um4/mol-atm“);

(************************************************************)

232

 

(************************************************************)

(* Plots of Calculations *)
(************************************************************)

Plot[Lp, {T, 274, 253}, PlotRange —-> {0.0, 8.0},
AxesLabel —> {“Temp(K)”, “Lp (pm/min-atm)”}];

um
min xatm

 

.uv—v—v'

>-v.,_..v..r~
.

V .vv

H4 k) U) Q: U1 05 \J(D

260 265 270 erp

(***************************#********************************)

Clear [B];
B=lOO (*Cooling rate in °C/min*);

(************************************************************)

(* Solution *)
(*****#**************************************************#***)

sol=NDSolve[{V’[T] = = ((meod*A)/(vw*B))*
((Log[(V[T]-Vbcell)/((V[T]-Vbcell)+(2*vw*nsalt))]) -—
(H/R)*((l/TR)-(1/T))), V[272.4] = = Vi}, V[T],
{T, 272.4, 253}]

(************************************************************)

(* Results *)
(************************************************************)

233

Plot[Evaluate [((V[T])/Vi)/.sol], {T, 272, 253},
PlotRange —+ {0, 1.1}, AxesLabel —>

{“Temp(K)”, “Normalized Volume”}]

Table[(V[T]/Vi)/.sol, {T, 253, 272}];

TableForm[%]

Normalized

OOOOOOOOOOOOOOOOOOOO

0.

.345868
.346861
.347964
.349196
.350581
.352151
.353943
.356009
.358418
.361261
.364669
.36883

.374027
.380705
.389618
.402163
.421334
.462304
.699509
.956958

l

Volume

A 3 25725 23610; 2621326526715— 2710”

234

Temp

(K)

APPENDIX C

Mechanical Drawing of AFA

235

 

 

 

 

 

 

 

 

236

APPENDIX D

Thermal Characterization Data

237

Characterization of AFA using processed meat (n=3)

 

 

3 Ioria5.lo .
e loria6.lo .

3 loria7.lo

 
 
   

 

 

Lowest Minimum Temperature

 

W
3 ioria6.lo-

8

 

 

 

 

   

I

(o

 
 

 

Average Spatial Variation

 

 

loria5. lo
loriaGJ
loria7.l
VE 0.680161
D 0.439 . .
EM mm
Symmetry

  

three T/Cs . - r side averaced, then differences summarized)

AVE=average difference between sides
MAX=maximum difference between sides

 

A=TC1-TCZ D=TC4-TC5
B=TC2-TC3 E=TC4~TCG
C=TC1-TC3 F=TC5~TC6

238

APPENDIX E

Neumann Program

239

(*NeumannVl .nb*)
(t***********************************************************)
(* This program finds the solution to the Neumann 2-phase *)

(* solidification problem, with constant properties *)
(************************************************************)

Clear[rhol, rh02, k1, k2, a1, a2, c1, c2, T1, T2, Tf, To, Ts, l, L, lam, X]

(************************************************************)

(* properties; l=solid region, 2=liquid region“)

(*water*)

rhol=1000; (*kg/m3“)
rh02=1000; (*kg/m3*)
k1=1.88; (*W/mK*)
k2=0.61; (*W/mK*)
cl=2.04 103; (*ngK“)
c2=4.18 103; (*J/kgK“)
1=333.4 103; (*J/m *)

(************************************************************)

(*boundary, initial, and phase change conditions; l=solid region,
2=liquid region*)

Tf=269 (*K*);
To=310 (*K*);
Ts=77(*K*);

(************************************************************)

a] =kl/(rhol cl);
a2=k2/(rh02 02);
a12=al/a2;
k21=k2/k1;

(************************************************************)

FindRoot[(1/(Exp[lam2 Erf[lam]))— k2]*a120'5*(To-Tf))/((Tf-Ts)
*Exp[a12*(lam )] Erfc[a12°‘ *lam]) = =
1* lam*(Pi°'5)/(cl*(Tf-Ts)), {lam, 0.01, 0.01, 100}]

(************************************#***********************)

9.21569x10'7
{lam—+0.545159}

240

(************************************************************)

(*
(*
(*
(*

With knowledge of the root the position of the interface at
any given time can be computed.

Note: al needs to be re-written from the previous sub-routine

*)
*)
*)
*)

(************************************************************)

Table[(2*0.545159*((t*9.21569 1050-5)), {t, 0,22, 1}]

(************************************************************)

CDCD<D CDCDCD CDCDCD CDCD<D CDCD<D C>CDCD C>CD<D C>CD

.00104669
.00148024
.00181292
.00209338
.00234046
.00256385
.00276928
.00296048
.00314006
.00330992
.00347147
.00362583
.00377389
.00391635
.0040538

.00418675
.0043156

.00444072
.00456241
.00468093
.00479653
.0049094

241

(******************************************************************)

(* For a given time/interface position the temperature distribution *)
(* can be determined. *)
(******************************************************************)

Clear [rho], rh02, k1, k2, a1, a2, cl, c2, T1, T2, Tf, To, Ts, l, L, lam, X];

rhol=1000; (*kg/m3“)
rh02=1000; (*kg/m3“)
k1=1.88; (*W/mK*)
k2=0.61; (*W/mK*)
c1=2.04 103; (*J/kgK*)
c2=4.18 103; (*J/kgK“)
1=333.4103;(*J/m *)

 

Tf=269 (*K*);
To=310 (*K*);
Ts=77(*K*);

a1=k1/(rhol cl);
a2=k2/(rh02 c2);
a12=a1/a2;
k21=k2/k1;

Tl= Ts + (((Tf-Ts)*(ErfIx/(2*(al *22)°'5}))/(Erfl0.545159]));
T2=To — (((To-Tf)*(Erfc[x/(2*(a2*22)°' ])/(Erfc[0.545159*(a120'5)])));

plotl=Plot[T1 , {x, 0, 0.0049094”

plot2=Plot[T2, {x, 0.004904, 0.01}]
Show [plot 1, plot2]

.A. ._,. -_. 7 __ - ___.__.,74 . . ,h—g, -

16.0014 2 ”01.002“ 0.003 0.004 16.305

242

- _ -310 1
3001
2901
2801

A_- ._#

1 T 4. 4s- ..

300 1
250
200 1
150 1

1001 i if.
.1 0.002

4;. g . _Vfif ,

1 6.0041 H 0.0061 ilvo.608 ‘0151

243

APPENDIX F

Cho and Sutherland Program

244

(*Neumann 3 phase.v1"‘)
(************************************************************)

(* This program provides the exact solution to the Neumann *)
(* problem with a melting temperature range, *)
(* infinite slab geometry *)

(************************************************************)

ClearAll[a1, a2, a3, a12, a2], a13, a23, a31, c1, c2, c3, eta, f, kl, k2, k21,
k3], k32, l, lam, rho, t, T1, T2, T3, T0, Tfl, Tfs, x]

(#*****************************#*****************************)

(* Material properties *)
(* l=solid phase; 2=mush; 3=liquid phase *)

(************************************************************)

(* Define properties *)
(************************************************************)
k1=1.88 (*W/mK*);

k2=1.245 (*W/mK*);

k3=0.61 (*W/mK*);

rho=1000 (*kg/m3*);

c1=2040 (*J/kgK*);

c2=31 10 (*J/kgK*);

c3=4180 (*J/kg/K*);

Tfs=265 (*K*);

Tfl=272 (*K*);

To=310 (*K*);

Ts=77 (*K*);

f—‘l (“dimensionless") (* =1 pure water =eutectic concn for mixtures“);
1=334000(*J/kg*);

(************************************************************)

(* Conversions *)
(************************************************************)
a1=k1/(rho*c1);

a2=k2/(rho*c2);

a3=k3/(rho"‘c3);

alZ=al/a2;

a21=a2/a1;

a23=a2/a3;

a32=a3/a2;

a31=a3/a1;

k12=k1/k2;

k21=k2/k1;

k23=k2/k3;

k32=k3/k2;

(************************************************************)

245

 

 

(************************************************************)

(* Solution *)
(************************************************************)

I 1

1 lf 1

1 k21*(Tfl-Tfs+ )e a21 l \fn(1—f)llan

1 *‘** 'A *"* ‘”” *‘ii""**’*—*"|== "* ”“ *‘c’ r

, 1 i c1(Tfs-Ts)

i‘JaZI*(Tfl3-TS)*in[ ($3 l—Erf[shm1]11

i l 1621 11
1 eta2 1

lf ~- ~ 1 , , I, z , _eta2

1 [Hi-JMS+_V 1e am- 1 l“/fl*dfl.*l* f 1Au_(fl3-Tflj}GZN/d8 e 33.}
1Erf[€ta1—Erf Jim]! 112*c2*(eta—lan) 1 - Eirfc[—_e.ta.-] '
,_ ‘
1 \/a21 \/a21 , \/a31

{eta, 0.45}, {lam 0.4},
MaxIteraticns 4800,th 40.2]

(*NOTE: Do not set eta=larn. It will generate an infinity term")
{eta —+ 0.55469, 1am 90.531195}

eta=0.55469;
lam=0.531195;

246

 

(********************** TEST ROOTS *************************)

1121—19—13]? 1. / x/nkzlfi 1
KBrat-111111) ‘12c2irfs—Ts) (eta —lam)1

( ( _1an121 1

k21*(Tf1—Tfs+lf)e 12121“: 1
C2 .

i 1 3 x/Jru-fnlam
_ i/ ’11:- l(Tf—T)I
,1/a21“Tris—'1‘s)11:1.:;rf[9'CC‘1‘]—Iarf[lam]?-l c S S
1 1 16121 151211;
( _et:a2 1
(Tfl -Tfs+lf)e a21 . 1 f
amass c2_1 1—((e\/7ra21)*
iErf[ eta1—Erf1 13’“ : 2
'1 \/a21 \/a21 1
eta2

(To-Tfl) 1132 V32? e— a31
Erfc[ eta -]
\/a31

((11. f/ (c2 1. (eta —lam))))) —

(************************************************************)

(* Roots solved in the previous equation need to be defined below *)
(4‘***********18**************III********************************)

eta=0.55469
lam=0.53195
(IIIIt*******************#**************************************)
(* The next segment solves for the temperature distribution at a *)
(* pre-determined time *)

(***********************************************#************)

t=22; (*sec*)

(************************************************************)

247

((Erf1 ’8- 1 1
1 2*\/al*t 1
T1211“ 7 *7 1*(TfS—TS)1+TS;
11. Erf[lam] ‘
11 1 1
T2:
11 I1 1
1 j r 1 1
. 1 l*f*' X : -lan.
111 _I 12*\/a1*t 1 1
1 _

e a -- 7, +
1C2*(Tfs-Ts)*(eta—lan)1

1 1

 

1
1 (1 111
1 I X _ £1,311 1
1/(Tfl—Tfs+(l*f111 11m11215/a2t1] m1 a211 ..11
11,- .-____.. _C2__4,,1_,_,_ __*___ ___ ___1 11
1'1 Tf -TS 1 1 1
11 1S 1 11 Wiesel-minim]: 1%!
1 1 1 .1121 1:121) 111
1
1
1
1
1
*(Tfs-Ts)11+‘I‘s;
1
1
1
l I 11
1 1EIJCfC[---M 11
1 1 2*\/a3*t11
T3=TO-‘(TO—Tfl)* .. E4”-
1 1
11cm its]! 1
1 1 \/a31} 1

248

(************************************************************)

(* Plotting: need to solve for the location of the mushy region *)
(* limits *)
(************************************************************)

XS=2*lan*\/Wiéflil;gi

x1 =2*eta*x/a{1;t;

X3

X1

0.00478365

0.00499523
(*******tit***********titt****tit******#****#****************)
(* Using these limits the results can then be appropriately *)
(* plotted *)

(********************************************************#***)

plot1=Plot[T1, {x, 0, 0.00478365}]
plot2=Plot[T2, {x, 0.00478366, 0.00499523}]
plot3=Plot[T3, {x, 0.00499524, 0.01}]
Show[plot1, plot2, plot3]

249

BIBLIOGRAPHY

250

Aggarwal, S.J., Shah, S.J., Diller, K.R., and Baxter, C.R. Fluorescent digital microscopy
of interstitial molecular diffusion in burn injury. Comput. Biol. Med. 19(4), 245-261
(1989).

Agre, P., Brown, D., and Nielson, S. Aquaporin water channels: unanswered questions
and unresolved controversies. Curr. Opin. Cell. Biol. 7, 472-483 (1995).

Altman N.H., and Goodman D.G. In “The Laboratory Rat” (H.J. Baker, J .R. Lindsey, and
SH. Weisbroth, Eds), pp. 361-362. Academic Press, Orlando, 1979.

Aran, J .M., Gottesman, M.M., and Pastan, 1. Construction and characterization of
bicistronic retroviral vectors encoding the multi-drug transporter and b-galactosidase or
green fluorescent protein. Cancer. Gene Ther. 5, 195-206 (1998).

Baumstark-Khan, C., Palm, M., Wehner, J., Okabe, M., Ikawa, M., and Homeck, G.
Green fluorescent protein (GF P) as a marker for cell viability afier UV irradiation.
Journal of Fluorescence 9(1), 37-43 (1999).

Baust, J., Gage, A.A., Ma, H., and Zhang, C.M. Minimally invasive cryosurgery-
technological advances. Cryobiology 34, 373-384 (1997).

Baust, J.G., Clarke, D., Chatterjee, S., Baust, J.M., Friedland, D.M., Cohen, J.K., and
VanBuskirk, R. Cryosurgical modeling: sequenced cryo-chemotherapy increases human
prostate cancer cell death. Cryobiology 41, 358-359 (2000).

Beavis, A.J., and Kalejta, R.F. Simultaneous analysis of the cyan, yellow, and green
fluorescent proteins by flow cytometry using single-laser excitation at 458 nm.
Cytometry 37, 68-73 (1999).

Bellman, S., and Adams-Ray, J. Vascular reactions afier experimental cold injury: a
microangiographic study on rabbit ears. Angiology 7, 329-367 (1956).

Bischof, J., Christov, K., and Rubinsky, B. A morphological study of cooling rate

response in normal and neoplastic human liver tissue: cryosurgical implications.
Cryobiology 30, 482-492 (1993).

Bischof, J., Smith, D., Pazhayannur, P.V., Manivel C., and Hulbert, J. Cryosurgery of
Dunning AT-l rat prostate tumor: thermal, biophysical, and viability responses at the
cellular and tissue level. Cryobiology 34, 42-69 (1997).

Bird, HM. James Amott, M.D. (Aberdeen) 1797-1833: A pioneer in refrigeration
analgesia. Anesthesia 4, 10-17 (1949).

Bokman, SH, and Ward, W.W. Renaturation of Aequorea Green fluorescent protein.
Biochem. Biophys. Res. Commun. 101, 1372-1380 (1981).

251

 

 

Bonacina, C., Comini, G., Fasano, A., and Primicerio, M. On the estimation of
thermophysical properties in nonlinear heat conduction problems. Int. J. of Heat & Mass
Transfer 17, 861-867 (1974).

Brewis, N., Phelan, A., Webb, 1., Drew, J., Elliott, G., and O’Hare, P. Evaluation of
VP22 spread in tissue culture. Journal of Virology 74(2), 1051-6 (2000).

Brown, N.J., Bayoo, P., and Reed M.W.R. Effect of cryosurgery on liver blood flow. Br.
J. Cancer 68, 10-12 (1993).

Brown, N.J., Pollack, K.J., Bayoo, P., and Reed M.W.R. The effect of cryotherapy on
the cremaster muscle microcirculation in vivo. Br. J. Cancer 68, 10-12 (1994).

Bourne, M.H., Piepkorn, M.W., Clayton, F., and Leonard, L.G. Analysis of
microvascular changes in frostbite injury. J. of Surgical Research 40, 26-35 (1986).

Budman, H., Shitzer, A., and Dayan, J. Analysis of the inverse problem of freezing and
thawing of a binary solution during cryosurgical processes. JBME 117, 193-202 (1995).

Chalfie, M., Tu, Y., Euskirchen, G., Ward, W., and Prasher, D.C. Green fluorescent
protein as a marker for gene expression. Science 263, 802-805 (1994).

Chida, D., Miura, O., Yoshimura, A., and Miyajima, A. Role of cytokine signaling
molecules in erythroid differentiation of mouse fetal liver hematopoietic cells: functional
analysis of signaling molecules by retrovirus-mediated expression. Blood 93, 1567-1578
(1999)

Chishima, T., Miyagi, Y., Li, L., Tan, Y., Baranov, E., Yang, M., Shimada, H., Moossa,
A.R., and Hoffman, R.M. Use of histoculture and green fluorescence protein to visualize
tumor cell host interaction. In vitro Cell. Dev. Biol. 33, 745-747 (1997a).

Chishima, T., Miyagi, Y., Wang, X., Baranov, E., Tan, Y., Shimada, H., Moossa, A.R.,
and Hoffman, R.M. Metastatic patterns of lung cancer visualized live and in process by
green fluorescence protein expression. Clin. Exp. Metastasis 15, 547-552 (1997b).

Chishima, T., Miyagi, Y., Wang, X., Tan, Y., Shimada, H., Moossa, A.R., and Hoffman,
R.M. Visualization of the metastatic process by green fluorescent protein expression.
Anticancer Research 17, 2377-2384(19970).

Chishima, T., Miyagi, Y., Wang, X., Yamaoka, H., Shimada, H., and Moossa, A.R.
Cancer invasion and micrometastasis visualized in live tissue by green fluorescent protein
expression. Cancer Res. 57, 2042-2047 (1997d).

Chishima, T., Yang, M., Miyagi, Y., Li, L., Tan, Y., Baranov, E., Shimada, H., Moossa,

A.R., Penman, S., and Hoffman, R.M. Governing step of metastasis visualized in vitro.
Proc. Natl. Acad. Sci. USA 94, 11573-11576 (l997e).

252

 

Cho, SH. and Sutherland, J.E. Heat conduction problems with melting or freezing.
Journal of Heat Transfer, 91, 421-426 (1969).

Chou, C.L., Ma, T., Yang, B., Knepper, M.A., and Verkman, A.S. Fourfold reduction of
water permeability in inner medullary collecting duct of aquaporin-4 knockout mice. Am.
J. Physiol. 274(2), C549—554 (1998).

Chou, C.L., Knepper, M.A., van Hock, A.N., Brown, D., Yang, H., Ma, T., and Verkman,
A.S. Reduced water permeability and altered ultrastructure in thin descending limb of
Henle in aquaporin-1 null mice. J. Clinical Investigation 103 (4), 491-496 (1999).

Clarke, D.M., Baust, J .M., VanBuskirk, R, and Baust, J .G. Cryo-chemo combination
therapy: an adjunctive model for the treatment of prostate cancer. Cryobiology 42, 274-
285(2001)

Clontech Protocol # PT2040-1, Version #PR67195

Cody, C.W., Prasher, D.C., Westler, W.M., Prendergast, F.G., and Ward, W.W.
Chemical structure of the hexapeptide chromophore of the Aequorea green fluorescent
protein. Biochemistry 32, 1212-1218 (1993).

Coombe, D.R., Nakhoul, A.M., Stevenson, S.M., Peroni, SE, and Sanderson, C.J.
Expressed luciferase viability assay (ELVA) for the measurement of cell growth and
viability. J. of Immun. Methods 215, 145-150 (1998).

Cooper, [S Cryobiology as viewed by the surgeon. Cryobiology 1, 44-54 (1964).

C00per, T.E., and Trezek, G.J. Analytical prediction of the temperature field emanating
from a cryogenic surgical cannula. Cryobiology 7, 79-93 (1970).

Cormack, B.P., Valdivia, RH, and Falkow, S. FACS-optimized mutants of the Green
Fluorescent Protein (GF P). Gene 173, 33-38 (1996).

Creighton, T.E. “Proteins, Structures, and Molecular Properties, second edition” W.H.
Freeman and Company, New York, 1996.

Cubitt, A.B., Heim, R., Adams, S.R., Boyd, A.E., Gross, L.A., and Tsien, R.Y.
Understanding, improving and using Green Fluorescent Proteins. Trends in Biochemical
Science 20, 448-455 (1995).

De Giorgi, F., Ahmed, Z., Bastianutto, C., Brini, M., Jouaville, L.S., Marsault, R.,
Murgia, M., Pinto, P., Pozzan, T., and Rizzuto, R. Targetting GFP to organelles. In
“Methods in Cell Biology: Green Fluorescent Proteins” (K.F. Sullivan, and S.A. Kay,
Eds.), Vol. 58, pp. 75-85. Academic Press, San Diego, 1999.

253

Devireddy, R.V., Raha, D., and Bischof, J.C. Measurement of water transport during

freezing in cell suspensions using a differential scanning calorimeter. Cryobiology 36,
124-155 (1998).

Diller, K.R., and Hayes, L.J. Analysis of tissue injury by burning: comparison of in situ
and skin flap models. Int. J. Heat Mass Transfer 34, 1393-1406 (1991).

Elliott, G.D., and J .J . McGrath. Freezing response of mammary tissue: a mathematical
study. Advances in Heat and Mass Transfer in Biotechnology. HTD-Vol. 363/BED-Vol.
44, 59-64 (1999).

Elliott, G.D., Fron, D., Seip, C., McGrath, J .J ., and Crockett-Torabi, E. Development of
a novel in vivo assay of tissue viability: a cryosurgery application. Advances in Heat and
Mass Transfer in Biotechnology, HTD-Vol. 368, BED-Vol.47, 149-155 (2000a).

Elliott, G., McGrath, J ., and Crockett-Torabi, E. Green Fluorescent Protein: A novel
viability assay for cryobiological applications Cryobiology 40, 360-369 (2000b).

Endo, M., Jain, R.K., Witwer, B., and Brown, D. Water channel (Aquaporin 1)

expression and distribution in mammary carcinomas and glioblastomas. Microvascular
Research 58, 89-98 (1999).

Falk, M.M., and Lauf, U. High resolution, fluorescence deconvolution microscopy and
tagging with the autofluorescent tracers, CF P, GF P, and YFP to study structural

composition of gap junctions in living cells. Microscopy Research and Technique 52(3),
251-262 (2001).

Frerichs, N. Independent Study, Department of Mechanical Engineering, Michigan State
University, 2001.

Fron, D. Independent Study, Department of Mechanical Engineering, Michigan State
University, 2000.

Gaber, M.H., Wu, N.Z., Hong, K., Huang, S.K., Dewhirst, M.W., and Papahadjopoulos,
D. Thermosensitive liposomes: extravasation and release of tumor contents in tumor
micro-vascular networks. Int. J. Radiat. Oncol. Biol. Phys. 36(5), 1177-1187 (1996).

Gage, A.A. Cryosurgery in the treatment of cancer. Surg. Gynecol. Obstet. 174, 73-92
(1992)

Gage, A.A. History of cryosurgery. Seminars in Surgical Oncology 14, 99-109 (1998).

Gage, A. and Baust, J. Mechanisms of tissue injury in cryosurgery. Cryobiology 37,
171-186 (1998).

254

Gilbert J.C., Rubinsky B., Wong S.T.S., Brennan K.M., Pease GR, and Leung P.P.
Temperature determination in the frozen region during cryosurgery of rabbit liver using
MR image analysis. Magnetic Resonance Imaging 15, 657-667 (1997).

Gonzalez, D.G., and Ward, W.W. Large-scale purification of recombinant green
fluorescent protein from Escherichia coli. In “Methods in Enzymology:
Bioluminescence and Chemiluminescence Vol. 305(C)” (M.M. Ziegler and TD.
Baldwin, Eds.) pp. 212-223, Academic Press, San Diego, 2000.

Gordon-Karnm W.J., Steponkus P.L. Lamellar-to-hexagonaln phase transitions in the

plasma membrane of isolated protoplasts following freeze-induced dehydration. Proc.
Natl. Acad. Sci, USA 81, 6373-6377 (1984).

Gourgouliatos, Z.F., Welch, A.J., and Diller, K.R. Microscopic instrumentation and
analysis of laser-tissue interaction in a skin flap model. JBME 113, 301-307 (1991).

Gross, J.F., Roemer, R., Dewhirst, M., and Meyer, T. A uniform thermal field in a
hyperthennia chamber for microvascualr studies. Int. J. Heat Mass Transfer 25 1313-
1320 (1982).

Haupts, U., Maiti, S., Schwille, P., and Webb, W.W. Dynamics of fluorescence
fluctuations in green fluorescent protein observed by fluorescence correlation
spectroscopy. Proc. Natl. Acad Sci. USA 95, 13573-13578 (1998).

Haas, J., Park, EC, and Seed, B. Codon usage limitation in the expression of HIV-1
envelope glycoprotein. Curr. Biol. 6(3) 315-324 (1996).

Heim, R., Prasher, DC, and Tsien, R.Y. Wavelength mutations and post-translational
auto-oxidation of Green Fluorescent Protein. Proc. Natl. Acad. Sci. USA 91, 12501-
12504 (1994).

Hempling, H.G., Thomson, S., and Dupre, A.M. Osmotic properties of human
lymphocytes. J. Cell Physiol. 93 293-302 (1977).

Higashijima, S., Hotta, Y., and Okamoto, H. Visualization of cranial motor neurons in
live transgenic zebrafish expressing green fluorescent protein under the control of the
islet-1 promoter/enhancer. J. Neuroscience 20(1), 206-218 (2000).

Hilf, R., Freeman, J.J., Michel, L, and and Borman, A. Characterization of a
transplantable lactating mammary tumor: endocrinological, morphological, and
biochemical aspects. Cancer Res. 24, 812-824 (1964).

Hilf, R., Michel, 1., Bell, C.F., and Borman, A. Biochemical and morphologic properties
of a new lactating mammary tumor line in the rat. Cancer Res. 25, 286-299 (1965).

255

Hoffman, N., Swanlund, D. and Bischof, J. Observation of vascular injury after freezing:
investigating the response of normal skin and subcutaneous AT-l tumor tissue to
cryosurgery in the dorsal skin flap chamber. Advances in Heat and Mass Transfer in
Biotechnology, HTD-Vol. 363/BED-Vol. 44, 39-42 (1999).

Hoffman, N., and Bischof, J. Cryosurgery of normal and tumor tissue in the dorsal skin
flap chamber: part I-therrnal response. J. of Biomechanical Engineering 123, 301-309
(2001)

Hoffman, N., and Bischof, J. Cryosurgery of normal and tumor tissue in the dorsal skin

flap chamber: part II-injury response. J. of Biomechanical Engineering 123, 310-316
(2001)

Hong J ., and Rubinsky B. Patterns of ice formation in normal and malignant breast tissue.
Cryobiology 31, 109-120 (1994).

Hong J ., Wong S., Pease G., and Rubinsky B. MR imaging assisted temperature
calculations during cryosurgery. Magn. Reson. Imaging 12(7), 1021-1031 (1994).

Hunt, L., Jordan, M., De Jesus, M., and Wurm, F.M. GFP-expressing mammalian cells

for fast, sensitive, non-invasive cell growth assessment in a kinetic mode. Biotechnol. &
Bioeng. 65(2), 201-205 (1999).

Jacob, G., Kurzer, M.N., and Fuller, B.J. An assessment of tumor cell viability after in
vitro freezing. Cryobiology 22, 417-426 (1985).

Jain, R.K. Transport of molecules in the tumor interstitium: a review. Cancer Res. 47
3039-51 (1987).

Jain, R.K., Stock, R.J., Chary, S.R., and Reuter, M. Convection and diffusion
measurements using fluorescence recovery afier photobleaching and video analysis: in
vitro calibration and assessment. Microvascular Research 39, 77-93 (1990).

Jain, R.K., Weissbrod, J.M., and Wei, J. Mass transport in tumors: characterization and
applications to chemotherapy. Advances in Cancer Research 33, 251-311 (1998).

Jain, R.K. Transport of molecules, particles, and cells in solid tumors. Ann. Rev.
Biomed. Eng. 1, 241-263 (1999).

Jakobs, S., Subramaniam, V., Schonle, A., Jovin, T.M., and Hell, S.W. EGFP and DsRed
expressing cultures of Escherichia coli imaged by confocal, two-photon and fluorescence
lifetime microscopy. F EBS Letters 479, 131-135 (2000).

Kac, E. GFP Bunny. In “Telepresence, Biotelematics, and Transgenic Art” (P.T.
Dobrila, Ed.) Kibla, Maribor, Slovenia, 2000.

256

Kain, S.R. Enhanced variants of the Green Fluorescent Protein for greater sensitivity,
different colors, and detection of apoptosis. In “Fluorescent and Luminescent Probes, 2"d
Edition”, pp. 284-292, Academic Press, 1999.

Kay, J. Intracellular protein degradation. Biochem. Soc. Trans. 6, 789-797 (1978).

Kedem, O. and Katcalsky, A. Thermodynamic analysis of the permeability of biological
membranes to non-electrolytes. Biochimica et Biophysica ACTA 27, 229-246, (1958).

Kleinhans, F.W. Membrane permeability modeling: Kedem-Katchalsky vs. a two-
parameter formalism. Cryobiology 37(4), 271-89 (1998).

Kolb, V.A., Makeyev, E.V., Ward. W.W., and Spirin, A.S. Synthesis and maturation of
green fluorescent protein in a cell-free translation system. Biotechnology Letters 18,
1447-1452 (1996).

LeFebvre, J .H., and F olke, L.E. Effects of subzero temperatures on the microcirculation
in the oral mucous membrane. Microvascular Research 10, 360-372 (1975).

Lehninger, A. “Biochemistry, The Molecular Basis of Cell Structure and Function,
Second Edition”. Worth Publishers, Inc., New York (1976).

Levin, R.L., Cravalho, E.G., and Huggins, C.E. A membrane model describing the effect
of temperature on the water conductivity of erythrocyte membranes at subzero
temperatures. Cryobiology 13, 415—429 (1976).

Levin, R.L. Water permeability of yeast cells at subzero temperatures. Journal of
Membrane Biology 46, 191-214 (1979).

Levonson, V.V., Transue, E.D.E., and Roninson, I.B. Internal ribosomal entry site -

containing retroviral vectors with green fluorescent protein and drug resistance markers.
Hum. Gen. Ther. 9, 1233-1235 (1998).

Li, X., Zhang, G., Ngo, N., Zhao, X., Kain, S.R., and Huang, C.C. Deletions of the
Aequorea victoria Green Fluorescent Protein define the minimal domain required for
fluorescence. J. Biol. Chem. 272, 28545-28549 (1997).

Li, X., Zhao, X., Fang, Y., Jiang, X., Duong, T., Fan, C., Huang, C-C., Kain, S.R.
Generation of destabilized Green Fluorescent Protein as a transcription reporter. J. Biol.
Chem. 273(52), 34970-34975 (1998).

Li, C-Y., Shan, S., Huang, Q., Braun, R., Lanzen, J., Hu, K., Lin, P. and Dewhirst, M.

Initial stages of tumor cell-induced angiogenesis: evaluation via skin window chambers
in rodent models. Journal of the National Cancer Institute 92(2), 143-7 (2000).

257

 

Lim, C.R., Kimata, Y., Oka, M., Nomaguchi, K., and Kohno, K. Thermosensitivity of
green fluorescent protein fluorescence utilized to reveal nuclear-like compartments in a
mutant nucleoporin NSPl. J. Biochem. 118, 13-17 (1995).

Lunardini, V. Heat Transfer in Cold Climates. Van Nostrand Reinhold Company, New
York, (1981).

Lunardini, V. Heat Transfer with Freezing and Thawing. Elsevier, Amsterdam, (1991).

Manson, P.N., Jesudass, R., Marzella, L., Bulkley, G.B., Irn, M.J., and Narayan, K.K.
Evidence for an early free-radical mediated reperfusion injury in frostbite. Free Radical
Biology & Medicine 10, 7-11 (1991).

Mansoori, G.A. Kinetics of water loss fiom cells at subzero centigrade temperature.
Cryobiology 12, 34-45 (1975).

Marshall, J ., Molloy, R., Moss, G.W.J., Howe, J .R., and Hughes, T.E. The jellyfish green
fluorescent protein: a new tool for studying ion channel expression and function. Neuron
14, 211-215 (1995).

Martinez de Maranon, 1., Gervais, P., and Molin, P. Determination of cells’ water
membrane permeability: unexpected high osmotic permeability of Saccharomyces
cerevisiae. Biotechnology and Bioengineering 56(1), 62-70 (1997).

Marzella, L., Jesudass, R.R., Manson, P.N., Myers, R.A., and Bulkley, G.B. Plast.
Reconstr. Surg. 83 (1), 67-76 (1989).

Mazur, P. Kinetics of water loss from cells at subzero temperatures and the likelihood of
intracellular freezing. Journal of General Physiology 14, 347-369 (1963).

Mazur, P. Theoretical and experimental effects of cooling and warming velocity on the
survival of frozen and thawed cells. Cryobiology 2(4), 181-92 (1966).

Mazur, P. The role of intracellular freezing in the death of cells cooled at supraoptimal
rates. Cryobiology 14, 251-272 (1977).

Mazur P. Freezing of living cells: mechanisms and implications. Amer. J. Physiol. 247,
C125-C142 (1984).

McGrath, J.J., Cravalho, E.G., and Huggins, C.E. An experimental comparison of
intracellular ice formation and freeze-thaw survival of Hela S-3 cells. Cryobiology 12,
540-550 (1975).

McGrath, J .J. Preservation of biological material by freezing and thawing. In “Heat

Transfer in Medicine and Biology” (A Shitzer, and RC. Eberhart, Eds.) Plenum, New
York, 1985.

258

 

 

McGrath, J .J . Membrane transport properties. In “Low Temperature Biotechnology:
Emerging Applications and Engineering Contributions” (J .J . McGrath and KR. Diller,
PAS.) pp. 271-330. ASME Press, New York, 1988.

McGrath, J .J . Low temperature injury processes. In “Advances in Bioheat and Mass
Transfer: Microscale Analysis of Thermal Injury Processes, Instrumentation, Modeling,
and Clinical Applications” ASME-HTD 268, 125-132 (1993).

Meryman, H.T. Mechanics of freezing in living cells and tissues. Science 124, 515-521
(1956)

Meryman H.T. The exceeding of a minimum tolerable cell volume in hypertonic
suspension as a cause of freezing injury. In “The Frozen Cell” (G.E. Wolstenholme, M.
0' Conner, Eds.) pp. 51-67. J. & A. Churchill, London, 1970.

Miller, J.P., and Selverston, A.I. Rapid killing of single neurons by irradiation of
intracellularly injected dye. Science 206, 702-704 (1979).

Morise, H., Shirnomura, 0., Johnson, F.H., Winant, J. Intermolecular energy transfer in
the bioluminescent system. Biochemistry 13(12), 2656-2662 (1974).

Muldrew, K., and McGann, L.E. Mechanisms of intracellular ice formation. Biophys J.
57, 525-532 (1990).

Muldrew, K., Hurtig, M., Novak, K., Schachar, N., and McGann, L. Localization of
freezing injury in articular cartilage. Cryobiology 31, 31-38 (1994).

Nardid,O., Dyubko, T, and Repina, S. A comparative study of the effect of freeze-
thawing on peripheral and integral membrane proteins. Cryobiology 34, 107-113 (1997).

Netti, P.A., Berk, D.A., Swartz, M.A., Grodzinsky, A.J., and Jain, R.K. Role of
extracellular matrix assembly in interstitial transport in solid tumors. Cancer Research
60, 2497-2503 (2000).

Niswender, K.D., Blackman, S.D., Rhode, L., Magnuson, M.A., and Piston, D.W.
Quantitative imaging of green fluorescent protein in cultured cells: comparison of

microscopic techniques, use in fusion proteins, and detection limits. J. Microbiol. 180,
109-116 (1995).

Nugent, L.J., and Jain, R.K. Pore and fiber-matrix models for diffusive transport in
normal and neoplastic tissues. Microvascular Research 28, 270-274 (1984).

Okabe, M., Ikawa, M., Kominami, K., Nakanishi, T., and Nishimune, Y. ‘Green mice’ as
a source of ubiquitous green cells. F EBS Lett 407, 313-319 (1997).

259

Papanek, T.H. in The Water Permeability of the Human Erythrocyte in the Temperature
Range +25 C to -10 C, PhD dissertation, Massachusetts Institute of Technology (1978).

Papenfuss, H.D., Gross, J.F., Intaglietta, M., and Treese, F.A. A transparent access
chamber for the rat dorsal skin fold. Microvascular Research 18(3), 311-318 (1979).

Patterson, G.H., Knobel, S.M, Sharif, W.D., Kain, S.R., and Piston, D.W. Use of the
green fluorescent protein and its mutants in quantitative fluorescence microscopy.
Biophysical Journal 73, 2782-2790 (1997).

Pegg, D.E. Viability assays for preserved cells, tissues, and organs. Cryobiology 26,
212-231 (1989).

Prasher, D.C., Eckenrode, V.K., Ward, W.W., Prendergast, F.G., and Corrnier, M.J.
Primary structure of the Aequorea victoria Green Fluorescent Protein. Gene 111, 229-
233 (1992).

Prendergast, F .G., and Mann, K.G. Chemical and physical properties of Aequorin and
the Green Fluorescent Protein isolated from Aequoreaforskalea. Biochemistry 17, 3448-
3453 (1978).

Rabb, J.M., Renaud, M.L., Brandt, P.A., and Witt, C.W. Effect of freezing on the
microcirculation capillary endothelium of the hamster cheek pouch. Cryobiology 11(6),
508—18 (1974).

Rabin, Y., and Korin, E. An efficient numerical solution for the multi-dirnensional

solidification (or melting) problem using a microcomputer. Int. J. Heat and Mass
Transfer 36(3), 673-683 (1993).

Rabin, Y., and Shitzer, A. Exact solution to the one-dimensional inverse Stefan problem
in non-ideal biological tissues. Journal of Heat Transfer 117, 425-431 (1995).

Rabin, Y., and Shitzer, A. Combined solution of the inverse Stefan problem for
successive freezing/thawing in non-ideal biological tissues. JBME 119, 146-152 (1997).

Rabin, Y., Julian, T.B., Olson, P., Taylor, M.J., and Wolmark, N. Evaluation of post
cryosurgery injury in a sheep breast model using the vital stain 2,3,5-triphenytetrazolium
chloride. Cryo-Letters 19, 255-262 (1998).

Rabin, Y., and Shitzer, A. Numerical solution of the multi-dimensional freezing problem
during cryosurgery. JBME 120, 32-37 (1998).

Rand R.W., Rand R.P., Eggerding F.A., Field M., DenBesten L., King W., and Camici S.

Cryolumpectomy for breast cancer: an experimental study. Cryobiology 22, 307-318
(1985)

260

Rand RW, Rand RP, Eggerding F, DenBesten L, King W. Cryolumpectomy for
carcinoma of the breast. Surg. Gynecol. Obstet. 165, 392-396 (1987).

Reite, O.B. Functional qualities of small blood vessels in tissue injured by freezing and
thawing. Acta. Physiol. Scand. (Suppl) 63, 111-120 (1965).

Riddell, S.R., Elliott, M., Lewinsohn, D.A., Gilbert, M.J., Wilson, L., Manley, S.A.,
Lupton, S.D., Overell, R.W., Reynolds, T.C., Corey, L., and Greenberg, P.D. T-cell
mediated rejection of gene-modified HIV -specifrc cytotoxic T lymphocytes in HIV-
infected patients. Nature Medicine 2(2), 216-223 (1996).

Rizzuto, R., Brini, M., Pizzo, P. Murgia, M., and Pozzan, T. Chimeric green fluorescent
protein as a tool for visualizing subcellular organelles in living cells. Current Biology 5
(6), 635-642 (1995).

Rizzuto, R., Brini, M., DeGiorgi, F., Rossi, R., Heim, R., Tsien, R.Y., and Pozzan, T.
Double labelling of subcellular structures with organelle-targeted GFP mutants in vivo.
Current Biology 6 (2), 183-8 (1996).

Roberts, K.P., Smith, D., Ozturk, H., Kazem, A., Pazhayannur, P., Hulbert, J.C., and
Bischof, J .C. Biochemical alterations and tissue viability in AT-l tumor tissue afier in
vitro cryo—ablation. Cryo-letters 18, 241-250 (1997).

Rubinsky B., and Ikeda M. A cryomicroscope using directional solidification for
controlled freezing of a biological material. Cryobiology 22, 55-68 (1985).

Rubinsky, B., Lee, C.Y., Bastacky, J., and Hayes, T.L. The mechanism of freezing in
biological tissue: the liver. Cryo-letters, 8, 370-381 (1987).

Rubinsky, B., and Pegg, D. A mathematical model for the freezing process in biological
tissue. Proceedings of the Royal Society (1988).

Rubinsky, B. The energy equation for freezing of biological tissue. Journal of Heat
Transfer 111, 988-997 (1989).

Rubinsky, B., Lee, C.Y., Bastacky, J., and Onik, G. The process of freezing and the
mechanism of damage during hepatic cryosurgery. Cryobiology 27, 85-97 (1990).

Rubinsky B., and Onik G. Cryosurgery: advances in the application of cold to medicine.
Int. J. of Refiigeration 14, 190-199 (1991).

Scandurra, R., Consalvi, V., Chiaraluce, R., Politi, L., and Engel, P. Protein stability in
extremophilic archaea. Frontiers in Bioscience 5, d787-795 (2000).

Schuder, G., Vollrnar, B., Richter, S., Pistorus, G., Fehringer M., Feifel, G., and Menger,

M.D. Epi-Illumination fluorescent light microscopy for the in vivo study of rat hepatic
microvascular response to cryotherrnia. Hepatology 29 (3), 801-808 (1999).

261

Schuder, G., Pistorius, G., Fehringer, M., Feifel, G., Menger, M.D., and Vollmar, B.
Complete shutdown of microvascular perfusion upon hepatic cryotherrnia is critically
dependent on local tissue temperature. Br. J. Cancer 82(4), 794-799 (2000).

Schwartz, G.J., and Diller, K.R. Osmotic Response of Individual Cells During Freezing.
II. Membrane Permeability Analysis. Cryobiology 20 (1983).

Schwartz, G.J., and Diller, K.R. Analysis of the water permeability of human
granulocytes at subzero temperatures in the presence of ice. JBME 105, 360-366 (1983).

Shabana, M., and McGrath, J .J . Cryomicroscope investigation and thermodynamic
modeling of the freezing of unfertilized hamster ova. Cryobiology 25, 338-354 (1988).

Shepherd J .P., and Dawber R.P.R. Wound healing and scarring after cryosurgery.
Cryobiology 21, 157-169 (1984).

Sherban, K.D. Application and analysis of a microscope diffusion chamber. MS.
Thesis, Mechanical Engineering Department, Michigan State University, 1987.

Shimura, 0. Structure of the chromophore of Aequorea Green Fluorescent Protein.
F EBS Letters 104(2), 220-222 (1979).

Silvares, O.M., Cravalho, E.G., Toscano, W.M., and Huggins, CE. The thermodynamics
of water transport from biological cells during freezing. ASME Journal of Heat Transfer
97, 582-588 (1975).

Sjostrom, B., Weatherly-White, R.C.A., and Paton, B.C. Experimental studies in cold
injury I. The individual response to a standard cold environment. JSR IV (1), 12-16
(1964).

Smith, D.J., Fahssi, W.M., Swanlund, DJ, and Bischof, J.C. A parametric study of
freezing injury in AT-l rat prostrate tumor cells. Cryobiology 39, 13—28 (1999).

Snider, R.L., and Porter, J .M. Treatment of experimental frostbite with intra-arterial
sympathetic blocking drugs. Surgery 77(4), 557-561 (1975).

Steponkus P.L., Dowgert M.F., and Gordon-Karnm W.J. Destabilization of the plasma
membrane of isolated plant protoplasts during a fieeze-thaw cycle: the influence of cold
acclimation. Cryobiology 20, 448-465 (1983).

Stripecke, R., del CarmenVillacres, M., Skelton, D.C., Stake, N., Halene, S., and Kohn,

D.B. Immune response to green fluorescent protein: implications for gene therapy. Gene
Therapy 6, 1305-1312 (1999).

262

Swaminathan, R., Bicknese, S., Periasamy, N., and Verkman, A.S. Cytoplasmic
viscosity near the cell plasma membrane: translational diffusion of a small fluorescent

solute measured by total internal reflection-fluorescence photobleaching recovery.
Biophys. J. 71, 1140-1151 (1996).

Swaminathan, R., Hoang, OR, and Verkman, A.S. Photobleaching recovery and
anisotropy decay of green fluorescent protein GFP-S65T in solution and cells:
cytoplasmic viscosity probed by green fluorescent protein translational and rotational
diffusion. Biophys. J. 72, 1900-1907 (1997).

Tatsutani, K., Rubinsky, B., Onik, G., and Dahiya, R. Effect of thermal variables on
frozen human primary prostatic adenocarcinoma cells. Urology 48(3), 441-447 (1996).

Terwillinger, T.C., and Solomon, A.K. Osmotic water permeability of human red cells.
J. Gen. Physiol. 77, 549-570 (1981).

Toscano, W.M., Cravalho, E.G., Silvares, O.M., and Huggins, CE. The thermodynamics
of intracellular ice nucleation in the freezing of erythrocytes. ASME Journal of Heat
Transfer 97, 326-332 (1975).

Toner, M., and Cravalho, E. Thermodynamics and kinetics of intracellular ice formation
during freezing of biological cells. J. Appl. Phys. 67 (3), 1582-1593 (1990).

Toner, M., Tompkins, R.G., Cravalho, E.G., and Yarmush, M.L. Transport phenomena
during freezing of isolated hepatocytes. Amer. Inst. of Chem. Eng. J. 38, 1512-1522
(1992)

Toner, M. Nucleation of ice inside biological cells. In “Advances in Low-Temperature
Biology, Vol. 2” pp. 1-51, JAI Press Ltd, 1993.

Tsien, R., Pozzan, T., and Rink, TJ. J. Cell. Biol. 94, 325-334 (1982).

Ushiyama, M., Cravalho, E.G., Diller, K.R., Huggins, C.E., Intracellular water content of
Saccharomyces cerevisae during freezing at constant cooling rates. Cryobiology 10, 517
(1973)

Voller, V.R., and Prakash, C. A fixed grid numerical modeling methodology for
convection-diffusion mushy region phase-change problems. Int. J. Heat Mass Transfer
30(8), 1709-1719 (1987).

Voller, V.R., Brent, A.D., and Prakash, C. Modeling the mushy region in a binary alloy.
Appl. Math. Modeling, 14, 320-326 (1990).

Voller, V.R. An overview of numerical methods for solving phase change problems. In
“Advances in Numerical Heat Transfer”, 1995.

263

Wachsmuth, M., Waldeck, W., and Langowski, J. Anomalous diffusion of fluorescent
probes inside living cell nuclei investigated by spatially-resolved fluorescence correlation
spectroscopy. J. Mol. Biol. 298, 677-689 (2000).

Wanner, G.A., Mica, L.L., Wanner-Schmid, E., Kolb, Stefan A., Hentze, H.; Trentz, 0.,
and Wolfgang, E. Inhibition of caspase activity prevents CD95-mediated hepatic

microvascular perfusion failure and restores Kupffer cell clearance capacity. FASEB J.
13, 1239-1248 (1999).

Ward, W.W., and Cormier, M.J. An energy transfer protein in coelenterate
bioluminescence. J. Biol. Chem. 274(3), 781-788 (1979).

Ward, W.W. Properties of the coelenterate green-fluorescent protein. In
“Bioluminescence and Chemiluminescence” (M.A. DeLuca and W.D. McElroy, Eds.) pp.
235-242. Academic Press, New York, 1981.

Waris, T., and Kyosala, K. Cold injury of the rat skin. Scand. J. Plast. Reconstr. Surg.
16, 1-9 (1982).

Weatherly-White, R.C.A., Sjostrom, B., and Paton, B.C. Experimental studies in cold
injury: the pathogenesis of frostbite. J. Surg. Res. 4(1), 17-22 (1964).

Weatherly-White, R.C.A., Knize, D.M., Geisterfer, DJ, and Paton, B.C. Experimental
studies in cold injmy. V. circulatory hemodynamics. Surgery 66 (1), 208-214 (1969).

White, A.C. Liquid air: its application in medicine and surgery. Med. Rec. 56, 109-112
(1899)

Whittaker, D.W. Ice crystals formed in tissues during cryosurgery. 1. Light microscopy.
Cryobiology 11, 192-201 (1974).

Whittaker, D.W. Ice crystals formed in tissues during cryosurgery. 11. Electron
microscopy. Cryobiology 11, 202-217 (1974).

Windoffer, R., Leube, R.E. Detection of cytokeratin dynamics by time-lapse
fluorescence microscopy in living cells. J. Cell Sci. 112, 4521-4534 (1999).

Woods, B.J., Liu, J., Derrow, C., Smith, F.O., Williams, D.A., and Critser, J.K.
Osmometric and permeability characteristics of human placental /umbilical cord blood

CD34+ cells and their applications to cryopreservation. J. of Hematotherapy and Stem
Cell Research 9, 161-173 (2000).

Wu, N., Ross, B., Gulledge, C., Klitzrnan, B., and Dewhirst, M. Differences in

leucocyte-endotheliurn interactions between normal and adenocarcinoma-bearing tissues
in response to radiation. Br. J. Cancer Research 69, 883-889 (1994).

264

Xian, M., Honbo, N., Zhang, J., Liew, C-C., Karliner, J.S., and Lau, Y-F.C. The green
fluorescent protein is an efficient biological marker for cardiac myocytes, J. Mol. Cell.
Cardiol. 31, 2155-2165 (1999).

Yang, F., Moss, LG, and Phillips, G.N., Jr. The molecular structure of Green
Fluorescent Protein. Nature Biotechnology 14, 1246-1251 (1996).

Yang, H., Acker, J ., Chen, A., and McGann, L. In situ assessment of cell viability. Cell
Transplantation 7(5), 443-451 (l998a).

Yang, M., Hasegawa, S., Jiang, P., Wang, X., Tan, Y., Chishima, T., Shimada, H.,
Moossa, A.R., and Hoffman, R.M. Widespread skeletal metastatic potential of human

lung cancer revealed by Green Fluorescent Protein expression. Cancer Res. 58, 4217-
4221 (1998b).

Yang, M., Jiang, P., Sun, F., Hasegawa, S., Baranov, E., Chishima, T., Shimada, H.,
Moossa, A.R., and Hoffman, R.M. A fluorescent orthotopic bone metastasis model of
human prostate cancer. Cancer Res. 59 (4), 781-6 (1999).

Yokoe, H., and Meyer, T. Spatial dynamics of GFP-tagged proteins investigated by local
fluorescence enhancement. Nature Biotechnology 14, 1252-1256 (1996).

Yuan, F ., Leunig, M., Berk, D. A. and Jain, R. K. Microvascular permeability of
albumin, vascular surface area, and vascular volume measured in human adenocarcinoma
LSI74T using dorsal chamber in SCH) mice. Microvascular Research 45, 269-289
(1993)

Zacarian, S.A., Stone, D., and Clater, M. Effects of cryogenic temperatures on
microcirculation in the golden hamster cheek pouch. Cryobiology 7, 27 (1970).

Zhang, G., Gurtu, V., and Kain, S. An enhanced green fluorescent protein allows
sensitive detection of gene transfer in mammalian cells, Biochemical and Biophysical
Research Communications 227, 707-711 (1996).

Zhang, Z., Rydel, R.E., Drzewieke, G.J., Fuson, K., Wright, S., Wogulis, M., Audia, J .E.,
May, RC, and Hyslop, P.A. Amyloid B-mediated oxidative and metabolic stress in rat

cortrial neurons: no direct evidence for a role of H202 generation. J. Neurochemistry 67,
1595-1606 (1996).

Zook, N., Hussmann, J., Brown, R., Russell, R., Kucan, J., Roth, A., and Suchy, H.

Microcirculatory studies in frostbite injury. Annals of Plastic Surgery 40 (3), 246-255
(1998)

265