SILTY LBESIRARI

Illllllllllllllllllllllllllllllllllllll

"3 1 3 1293 02050 96

 

an 7,? k
K‘ J “UV

This is to certify that the

dissertation entitled
A NOVEL TOTAL INTERNAL REFLECTION FLUORESCENCE (TIRF)
TECHNIQUE FOR INVESTIGATING MACROMOLECULAR ADSORPTION
AND INTERACTIONS AT THE LIQUID-LIQUID INTERFACE.

presented by

Arivalagan Gajraj

has been accepted towards fulfillment
of the requirements for

Ph.D. Chemical Engineering

 

degree in

 

 

Majoflessor 7 V

Date

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11m macros-0mm.“

 

A NOVEL TOTAL INTERNAL REFLECTION FLUORESCENCE (TIRF)
TECHNIQUE FOR INVESTIGATING MACROMOLECULAR ADSORPTION AND
INTERACTIONS AT THE LIQUID-LIQUID INTERFACE.

By

Arivalagan Gajraj

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

1999

ABSTRACT
A NOVEL TOTAL INTERNAL REFLECTION FLUORESCENCE (TIRF)
TECHNIQUE FOR INVESTIGATING MACROMOLECULAR ADSORPTION AND
INTERACTIONS AT THE LIQUID-LIQUID INTERFACE.
By

Arivalagan Gajraj

This dissertation presents a total internal reflection fluorescence microscopy (TIRFM) tool
that allows for direct, quantitative and in situ measurement of molecular-level phenomena
at the liquid-liquid interface. The apparatus uses a novel experimental cell designed to

create a stable and easily maintained liquid-liquid interface and is completely non-invasive.

Protein adsorption at the liquid-liquid interface plays a significant role in a wide range of
biological, pharmaceutical, cosmetic, and food related processes. In the first part of this
study, the apparatus was used to measure the equilibrium adsorption isotherms of two
proteins (BSA and lysozyme), labeled with fluorescein isothiocyanate (FITC) at the oil-
water interface. To evaluate the integrity of the interface, the measured profiles were
compared to adsorption data collected independently using interfacial tensiometry.
Agreement between these sets of data indicated that, in spite of the thinness of the
immiscible liquid layers, the interface behaves energetically like a normal liquid-liquid

interface.

An important aspect of the study of protein dynamics is the attachment of an extrinsic

fluorescent label to enable tracking of proteins by optical techniques. Consequently, the

effect of labeling on protein dynamics is of considerable scientific interest. Unfortunately,
this has not been adequately addressed in the literature to date. In the second part of this
work, the effect of F ITC labeling on BSA was investigated. Ion-exchange
chromatography (IEC) was used to exploit the differences in electronegativity between
proteins labeled with various quantities of FITC, to obtain samples with difi‘erent
monodisperse labeling ratios. TIRFM was then combined with fluorescence
photobleaching recovery (FPR) to measure the effect of labeling on adsorption kinetics
and protein difi‘usion. The results showed that the attachment of an extrinsic label does
have a pronounced effect on both adsorption and diffusion of proteins near the interface.
For instance, we estimated that the diffusion coefficient of a BSA molecule conjugated
with 2 FITC molecules is 40% greater than that of BSA to which only a single label has

been attached.

The last part of this work concerned the development of a calibration protocol to enable
quantification of interfacial protein concentrations, based solely on the intensity of
fluorescence emission. Photobleaching was used to separate the bulk and surface
contributions to the fluorescence signal by exploiting differences in mobilities between free
and adsorbed proteins. The procedure is instantaneous and non-invasive, and does not
require the use of external standards or additional equipment. The surface coverages
measured were lower than the corresponding values reported in the literature for the solid-
liquid interface. However, this was attributed to the increased mobility afforded to

proteins at a liquid-liquid interface.

Dedicated to my mother and my father.

iv

ACIG‘IOWLEDGEMENTS

Firstly, I would like to thank my beloved advisor, Dr. Bob Ofoli, for his guidance, support,
encouragement, and friendship. More than anything, he exposed me to the true
philOSOphy of research and taught me, by example, to be a scientist.

I would also like to thank my committee members Dr. Lira, Dr. Ng and Dr. Scranton for
taking the time out of their busy schedules to serve on my committee.

Thanks are also due to Sharon, Ilona and Mary who patiently processed every small
materials request and put up with me when I fretted over delays.

Big thanks to all my fiiends and labmates: Parijat, Shawn, Julie, Sumei, Brian, Jamaica,
Avani, Ajeet, Bill, Kiran, ...for the good times and the times wasted, the Tetris
competitions and the philosophical discussions.

Special thanks to the food science crew, who shared with me their delicious food (Oh
yes!) and their fiiendship and support: Dr. Ng, Vince, Monica, Alfred, Ryu, Pascal, Ling
Thanks to Nicole for her fiiendship and for her work on the calibration protocols, Rob, for
his tensiometry work and enlightening technical discussions and Margarida, for her life
saving help with the SAS programming and so much more.

Finally, a shout out to my boys in the Mau Mau. .. I’d have probably gone insane if not for

the music: Jahyouth, Mo, Ossama, Raja, Richard, Tarig, Temesgen

TABLE OF CONTENTS

TABLE OF CONTENTS ............................................................................................... vi
LIST OF FIGURES ........................................................................................................ ix
LIST OF TABLES ........................................................................................................ xii
NOMENCLATURE ..................................................................................................... xiii
1. INTRODUCTION ...................................................................................................... 1
1.1 Review of studies at the liquid-liquid interface ....................................................... 3
1.2 Challenges and difficulties presented by the liquid-liquid interface ......................... 5
1.2.1 Interfacial stability ........................................................................................ 5
1.2.2 Reproducibility ............................................................................................. 5
1.2.3 Difficulties in calibration ............................................................................... 6

1.3 Focus of this dissertation ....................................................................................... 6

2. A TIRF TECHNIQUE FOR INVESTIGATIN G MACROMOLECULAR

ADSORPTION AND INTERACTIONS AT THE LIQUID-LIQUID

INTERFACE. ............................................................................................................. 9
2.1 Introduction and background ................................................................................ 9
2.1.1 Rationale for using TIRFM for the liquid-liquid interface ............................ 10
2.2 Objectives ........................................................................................................... 13
2.3 Theory ................................................................................................................ 13
2.3.1 The Evanescent wave ................................................................................. 13
2.4 Description of apparatus ..................................................................................... 16
2.4.1 Developmental challenges ........................................................................... 16
2.4.2 TIRFM set-up ............................................................................................ 17
2.4.3 The sample cell ........................................................................................... 21
2.4.4 Materials .................................................................................................... 22
2.5 Experimental Procedures .................................................................................... 24
2.5.1 Coating of microscope slide with oil ........................................................... 24
2.5.2 Protein labeling ........................................................................................... 24
2.5.3 Beam alignment and microscope focusing protocols ................................... 25
2.5.4 Introduction of proteins into the flow cell ................................................... 26
2.5.5 Experimental precautions ............................................................................ 27
2.6 Problems and opportunities for the liquid-liquid interface .................................... 27
2.7 Results and discussion ......................................................................................... 28
2.7.1 Timescale of equilibrium adsorption ............................................................ 28
2.7.2 Assessment of interfacial integrity ............................................................... 30
2.8 Conclusions ........................................................................................................ 33

. EFFECT OF EXTRIN SIC FLUORESCENT LABELS ON DIFFUSION AND
ADSORPTION KINETICS OF BSA AT THE LIQUID-LIQUID
INTERFACE. ........................................................................................................... 3 5

3.1 Introduction and background .............................................................................. 35

3.2 Objectives ........................................................................................................... 38
3.3 Theory ................................................................................................................ 39
3.3.1 Chemistry of the protein-labeling reaction ................................................... 39
3.3.2 Diffusion of BSA-FITC .............................................................................. 40
3.3.3 Reaction kinetics ofBSA-FITC at the interface .......................................... 42
3.3.4 Concentration quenching: definition and concerns ....................................... 44
3.4 Materials and methods ........................................................................................ 45
3.4.1 Flowchart of strategy .................................................................................. 45
3.4.2 Total internal reflection fluorescence (TIRF) ............................................... 48
3.4.3 Fluorescence photobleaching recovery (FPR) ............................................. 48
3.4.4 Protein labeling reaction ............................................................................. 49
3.4.5 Ion exchange chromatography .................................................................... 50
3.4.6 Determination of starting conditions: pH and molarity ................................ 51
3.4.7 Elution of the protein .................................................................................. 52
3.4.8 Bufi‘er exchange ......................................................................................... 53
3.4.9 TIRFM experiments ................................................................................... 53
3.5 Data analyses ...................................................................................................... 54
3.5.1 Processes contributing to fluorescence recovery. ........................................ 54
3.5.2 The difiirsion-limited regime ....................................................................... 58
3.5.3 Reaction-limited regime .............................................................................. 60
3.6 Results and discussion ......................................................................................... 61
3.6.1 The diflirsion limited regime ....................................................................... 61
3.6.2 The reaction-limited regime ........................................................................ 67
3.7 Conclusions ........................................................................................................ 77

. A CALIBRATION PROTOCOL FOR QUANTITATION OF INTERFACIAL
CONCENTRATIONS OF MACROMOLECULES AT THE LIQUID-LIQUID

INTERFACE. ........................................................................................................... 80
4.1 Introduction ........................................................................................................ 80
4.2 Theory ................................................................................................................ 82
4.2.1 The evanescent wave .................................................................................. 82
4.2.2 Interfacial selectivity of TIRF ..................................................................... 84
4.2.3 The calibration protocol .............................................................................. 87
4.2.4 Relation between evanescent intensity (lo) and monitoring beam intensity 92
4.3 Materials and methods ........................................................................................ 95
4.3.1 TIRFM calibration experiments .................................................................. 95
4.3.2 Calculation of C3 by absorption spectroscopy ............................................. 96
4.3.3 Prevention of protein adsorption on the bottom slide .................................. 97
4.4 Data analyses ...................................................................................................... 98
4.5 Results .............................................................................................................. 105
4.6 Scope and application of the calibration protocol .............................................. 107
4.7 Discussion ........................................................................................................ 109
4.7.1 The detection system ................................................................................ 109
4.7.2 Heating effects of the laser beam .............................................................. 110
4.7.3 The quantum yield problem ...................................................................... 112

4.7.4 Does adsorption change the bulk concentration? ....................................... 114

4.7.5 Elimination of PMT light hysteresis: the shutter system ............................. 115
4.8 Conclusions ...................................................................................................... 117
5. SUMMARY AND CONCLUSIONS ...................................................................... 119
6. RECOMMENDATIONS FOR FUTURE WORK ................................................... 123
7. BIBLIOGRAPHY ................................................................................................... 128
8. APPENDICES ........................................................................................................ 143
8.1 Appendix A: Development of mathematical model for TIRF/PPR
experiments in the diffusion-limited regime ........................................................ 143
8.2 Appendix B: FITC ............................................................................................ 147
8.2.1 Structure .................................................................................................. 147
8.2.2 Prototropic forms ..................................................................................... 148
8.2.3 Absorption and emission spectra ............................................................... 150
8.2.4 Appendix C: SAS programs for data analysis ............................................ 152
8.3 Appendix D: Absorbance calibration experiments .............................................. 156
8.3.1 BSA at pH = 7.4, PBS buffer .................................................................... 156
8.3.2 FITC at pH = 5.1, acetate bufl‘er ............................................................... 157
8.3.3 FITC in pH = 7.4, PBS bufi‘er ................................................................... 158
8.4 Appendix B: Data collection program ............................................................... 159

viii

LIST OF FIGURES

Figure 2.1. Total internal reflection at the oil-water interface. The resulting
evanescent wave decays exponentially into the rarer aqueous phase. .............................. 15

Figure 2.2. Layout of the experimental setup for TIRFM experiments. The
double-flat system is used to first separate the incident beam into monitoring and
photobleaching streams, and then recombine the two beams. This scheme provides
control over beam intensities while ensuring superposition at the interface. .................... 20

Figure 2.3. The experimental cell, showing the location of the interface. The
thickness of the oil-water layer is less than 1.0 mm, which helps to ensure
interfacial stability. ......................................................................................................... 23

Figure 2.4. Typical TIRFM protein adsorption experiment showing fluorescence
intensity versus time during the equilibrium adsorption of BSA at the oil-water
interface. From the data, the time required to reach adsorption equilibrium was
estimated at about 20 minutes. ....................................................................................... 29

Figure 2.5. Equilibrium adsorption profiles for lysozyme and BSA conjugated with
the FITC label. Plot a) for BSA is at a much higher concentration, and portrays a
solution that has approached and exceeded interfacial saturation. The plot shows a
clear break in profile at the crnc. Plot b) for lysozyme shows a gradual approach
to interfacial saturation, with no indication of a cmc. ...................................................... 32

Figure 3.1. Flowchart of the experimental strategy used to determine the efi'ect of
extrinsic labeling on protein dynamics at and near the interface ....................................... 47

Figure 3.2. Dependence of a) r (characteristic time of difliision) and h) f (mobile
fraction) on labeling ratio in the difliision-limited regime for BSA at the oil-water
interface. The bulk protein concentration was 0.01 mM. ............ . .................................... 65

Figure 3 .3. Dependence of a) r0 (immobile fraction), b) r, (slowly reversible
fraction) and c) r; (rapidly reversible fi'action) on labeling ratio in the reaction-
limited regime for BSA at the oil-water interface. The bulk protein concentration
was 0.025 mM. .............................................................................................................. 68

Figure 3.4. Dependence of a) k4] (rate of slow desorption), b) kdz (rate of rapid
desorption) and c) kpb (rate of photobleaching) on labeling ratio in the reaction-
limited regime for BSA at the oil-water interface. The bulk protein concentration
was 0.025 mM. .............................................................................................................. 69

ix

Figure 3.5. Check for concentration quenching effects in the equilibrium
adsorption of L = 2, 0.025 mM BSA at the oil-water interface. ................................... 75

Figure 3.6. a. Normal (quenching free) fluorescence recovery of L = 2, 0.025 mM
BSA after preparative chromatography. b. Fluorescence recovery of L ~ 2.5.
0.025 mM BSA showing the efl‘ects of concentration quenching .................................... 76

Figure 4.1. Evanescent wave intensity as a function of depth of penetration into
the rarer medium. The penetration depth, (1,, is 75 nm in our setup, for an
interface between oil (n1 = 1.52), and water (n; = 1.33) for an incident light of
wavelength A = 488 nm, at an angle of incidence 9 = 70°. .............................................. 85

Figure 4.2. Evanescent intensities LS and I-p for perpendicular and polarized
incident light (A = 488 nm) at the oil-water interface (2 = 0) as a function of angle
of incidence. The critical angle, 9c, in our experiments is 61° for oil (111 = 1.52) and
water (n; = 1.33). .......................................................................................................... 94

Figure 4.3. PMT signal outputs showing typical levels of dark current,
background noise and shot noise proportional to fluorescence signals. The PMT is
operated at 800 V and —22°C ....................................................................................... 100

Figure 4.4. RMS shot noise as a function of the magnitude of the fluorescence
signal. .......................................................................................................................... 103

Figure 4. 5. A typical fluorescence recovery curve for TIRF/FPR experiments. F r
is the equilibrium fluorescence emission prior to photobleaching, and F M, is the
fluorescence emission immediately following a 15 mW, 100 ms photobleaching

pulse. ........................................................................................................................... 104
Figure 4.6. Interfacial concentration of BSA, at the oil-water interface, as a
function of bulk concentration estimated by the calibration protocol ............................. 107
Figure 4.7. Beam path of Axiovert 135 microscope Showing placement of the
PMT shutter (81) ......................................................................................................... 116
Figure 8.1. Structure of fluorescein isothiocyanate (5-isomer) ...................................... 147

Figure 8.2. Prototropic forms of fluorescein at different pH with fluorescence
efficiencies. .................................................................................................................. 149

Figure 8.3. Absorption and emission spectra (excitation at 488 nm) for FITC in
pH = 7.4 PBS bufi‘er, showing spectral overlap and Stokes’ shift ................................. 151

Figure 8.4. Absorbance versus concentration of BSA in pH 7.4 PBS bufl'er,
measured with a diode-array spectrophotometer (8452A Hewlett Packard) .................. 15 6

Figure 8.5. Absorbance versus concentration of FIT C in pH 5.1 acetate bufl‘er,
measured with a diode-array spectrophotometer (8452A Hewlett Packard) .................. 157

Figure 8.6. Absorbance versus concentration of FITC in pH 7.4 PBS buffer,
measured with a diode-array spectrophotometer (8452A Hewlett Packard) .................. 158

LIST OF TABLES

Table 1. Summary of results for BSA-FITC in the diffusion-limited regime. The
bulk protein concentration was 0.01 mM ....................................................................... 66

Table 2. Summary of results for BSA-FITC in the reaction-limited regime. The
bulk protein concentration was 0.025 mM ...................................................................... 72

xii

w(r)

WOO)

17(1)

NOMENCLATURE

incident angle of laser light

critical angle for nl-nz interface

wavelength of incident laser light in free space
spectroscopic absorbance

protein concentration before photobleaching
concentration of diffusing species

two-dimensional interfacial concentration of proteins
bulk concentration of proteins

characteristic difl‘usion time

dimensionless function of the evanescent intensity profile
protein diffusion coefficient

penetration depth of the evanescent wave

depth of focus of microscope objective

perpendicularly polarized component of transmitted electric field
amplitude of E;

mobile fi'action of adsorbed proteins

fluorescence signal before photobleaching

asymptote of the fluorescence recovery curve

fluorescence signal as a fianction of time

xiii

Fr
Fads

Fbulk

1(2)

1:10)

MO)

1641
kd2

km

kr

n1

722

*Q

SE“?

total fluorescence seen by the PMT
interfacial fluorescence

bulk fluorescence

evanescent wave intensity at a depth of z
incident light intensity at a depth of z = 0
p-polarized incident light intensity
s-polarized incident light intensity
bleaching parameter

adsorption rate constant

slow desorption rate constant

rapid desorption rate constant
photobleaching rate constant

rate constant for inter-fluorophore quenching energy transfer
depth of aqueous phase in experimental cell
molar labeling ratio of proteins

refractive index of the dense media
refractive index of rare media

relative refi'active index

numerical aperture of microscope objective
overall system constant

quantum yield of fluorophores

quantum yield of bulk fluorophores

quantum yield of adsorbed fluorophores

xiv

q; overall instrument constant

q, relative quantum yield

r0 immobile fraction of adsorbed proteins
r, slowly desorbing fraction of proteins
r2 rapidly desorbing fi'action of proteins
r radial position vector

R inter-fluorophore separation

R0 Forster distance

3 thickness of adsorbed protein layer at the interface
1 time

T duration of photobleaching pulse

w half width at e'2 height of the Gaussian intensity profile of the laser beam
2 depth coordinate into the rare media
Abbreviations

BSA bovine serum albumin

FITC fluorescein isothiocyanate

FPR fluorescence photobleaching recovery
IEC ion-exchange chromatography

PBS phosphate buffered saline buffer

PEO polyethylene oxide

PMT photomultiplier tube

TIRF total internal reflection fluorescence

1. INTRODUCTION

Proteins are amphiphilic polymers that play crucial roles in all biological processes. The
word protein is derived fiom the Greek word “proteios” which means “of the first rank.”
The term was coined by Berzelius in 1838 to emphasize the importance of this class of
molecules (1). The ability of proteins to influence a variety of interfacial phenomena stems
fi'om an amphiphilic nature that makes them highly surface active. Proteins are known to
adsorb fi'om solution onto almost any surface. In fact, early studies of protein adsorption
at the solid-liquid interface were aimed at the development of non-thrombogenic materials
for use in implanted cardiovascular prosthetic devices (2). Furthermore, practical
problems such as contact lens fouling, plaque formation on teeth and denitrifices, foaming
of protein solutions, and fouling of equipment used in the food processing industry are a

direct consequence of the relatively high surface activity of proteins.

Due to the importance of proteins in many practical situations, there is a great interest in
the study of their adsorption at a variety of interfaces. A large number of studies have
been conducted on equilibrium adsorption as well as interactions of various
macromolecules at the fluid-solid interface. A small sampling of these studies include
interfacial adsorption (3, 4, 5, 6, 7, 8, 9, 10), equilibrium conformation (11, 12, 13),
binding equilibrium (13, 14, 15, 16, 17, 18), and binding kinetics (19, 20, 21, 22, 23, 24).

Other studies include adsorption of proteins onto solid surfaces (25) and polymer films

(26, 27, 2), binding of antibody to an immobilized antigen (28), characterization of cell-

substrate interactions (29, 30, 31), and surface diffusion of adsorbed species (32, 33).

Proteins at the liquid-liquid interface play a significant role in a wide range of biological,
pharmaceutical, cosmetic, and food-related processes. In molecular biology,
transmembrane proteins (adsorbed at the lipid bilayer) act as channels or pumps and thus
play a vital role in energy transduction, ion exchange, and intercellular communication.
Just as important, a wide variety of physiological processes are the result of ligand
interactions with transmembrane protein receptors. Receptor-driven cell behavior is now
known to be extremely important; for example growth, secretion, contraction, motility,
and adhesion are all receptor-mediated functions (34). Also, cytoskeletal components (e. g.
spectrin in red blood cells) form attachments to cell membranes to define cell motility,

shape and mechanoelastic properties (1).

In biomedical applications, the behavior of drugs at biological membranes is an important
factor in determining pharmacological activity (35). In such cases, the oil-water interface
can be used as a first approximation to study drug delivery in the human body. The water-
hydrocarbon solvent interface is the simplest representation of a biological membrane
surface that separates the aqueous cytosol or external medium from the apolar central
region of a bilayer membrane (3 6). In fact, oil-water interfaces have been used to model
not only drug delivery (37, 35, 38) but also biological systems such as skin, nerve and

biomembranes related to taste and olfaction (3 9).

In areas related to food and agricultural science, protein adsorption plays an important
role in the formation and stability of dispersed systems (foams, emulsions, etc). Protein
adsorption afi‘ects the processing, rheology, texture, appearance and shelf life of food

emulsions like mayonnaise and butter and foams like meringue, whipped cream and bread.

There is now also speculation that oil water interfaces might have played an important role
in the origin of life. Dearner and Volkov (40) have theorized that the most basic form of
life could have utilized energy trapped at the hydrocarbon/water interface to control
photochemical reactions, thus giving rise to the beginning of life as defined by biological

science.

1.1 Review of studies at the liquid-liquid interface

Various studies have been performed at the liquid-liquid interface using optical and other
techniques. The optical techniques used include resonance Raman spectroscopy for a
surface-active dye excited by total internal reflection of incident light (41, 42, 43), total
internal reflection fluorescence combined with polarization measurements of bis-AN S at
the decalin-water interface (3 6), total internal reflection sum-frequency vibrational
spectroscopy to study the conformational order of sodium dodecyl sulfate (SDS) adsorbed
at the D20-CC14 interface (44), attenuated tOtal internal reflection spectroscopy of the
kinetics of metal ion extraction at the decalin-water interface (45), optical second
harmonic generation used to determine orientation of molecular monolayers at the decane-
water and carbon tetrachloride-water interfaces (46), and surface light scattering to

determine the dynamic surface pressure of a spread monolayer (47).

Other techniques that measure macroscopic properties and derive molecular parameters
include measurement of interfacial tension (47, 48, 49), measurement of the interfacial
concentration of proteins tagged with 1“C by scintillation counting (5 0), measurement of
potential drop to characterize ion and electron transfer in phospholipids at the interface
between two immiscible electrolyte solutions (ITIES) (51), ion transfer voltametry (35),

and measurement of interfacial viscosity and shear (52, 53 48).

Markin and Volkov (54) derived a modified Frumpkin adsorption isotherm in a theoretical
study and Leaver et al. (55) used trypsin-catalyzed hydrolysis to study variations in the
binding of B-casein to the oil-water interface. Ando et al. (56) have cultured human
endothelial cells at the liquid-liquid interface pretreated with proteins to study cell-matrix
interactions. Also the partitioning of proteins between two liquid phases has been used to

model the transfer of membrane proteins into membranes (5 7).

More recently, Malmsten (58) has combined ellipsometry with TIRF to study adsorption
processes in parenteral drug delivery at the liquid-liquid interface. Aoyama et al. (59) have
exploited surface mobility at the liquid-liquid interface to develop a new method for two-
dimensional crystallization of soluble proteins. Shiba et al. (60) have studied the growth
and morphology of anchorage-dependent animal cells. By cultivating the cells at a liquid-
liquid interface, they were able to prevent detrimental proteolysis effects usually seen at

the solid-liquid interface.

1.2 Challenges and difficulties presented by the liquid-liquid interface

A survey of the literature reveals relatively few optically based techniques for the liquid-
liquid interface. This is because of the various difficulties presented by the inherent nature
of this interface. Some of the main difliculties and our approaches to dealing with them are

discussed below.

1. 2. 1 Interfacial stability

The interface between two fluids is by nature unstable and susceptible to deformation. For
example, room and building vibrations can destabilize the oil-water interface; even
convectional currents can induce flow causing the interface to fluctuate. To counteract this
tendency, our interface is made up of very thin layers of oil and water (approximately an
80 um layer of oil on top of a 900 um layer of water). Thus, fi'om a global or
macrOSCOpic point of view, the system functions like a solid-liquid interface. Therefore it
does not suffer from the typical vibration-induced instability problems encountered at
liquid-liquid interfaces of normal depth. However, in relation to the largest linear
dimension of a protein (approximately 15 nm for BSA) an oil layer of 80 um is suficiently
thick, and should firnction as a liquid-liquid interface of normal depth. To provide firrther

stability, our entire experimental apparatus is housed on a vibration isolation table.

1.2.2 Reproducibility

Forming a reproducible oil-water interface is nearly impossible as minute changes in shape,
size and location are unavoidable. This means that experimental runs made on different

days or even at different locations on the same interface may not be comparable.

Therefore, we have been constrained to design our experiments to stand on their own.
Our calibration protocol, described in chapter 4 of this dissertation, has been developed to

circumvent this problem.

1. 2.3 Difficulties in calibration

The common techniques used for quantifying protein adsorption at the solid-liquid
interface will not work at the liquid-liquid interface. This is because
1. the exact thickness of the liquid-liquid interface is non-uniform and unknown;
2. the liquid-liquid interface cannot be isolated fi'om the experimental cell; and
3. there is no reliable way to drain species in the aqueous bulk without
incorporating species at the interface.
A calibration protocol that enables one to circumvent these liabilities is discussed in

further detail in chapter 4 of this dissertation.

1.3 Focus of this dissertation

The primary goal of this dissertation was to design a quantitative tool to study protein
adsorption at the liquid-liquid interface to enable non-intrusive, in Situ investigation of the
same. A major portion of the work that went into this project involved the development,
design and assembly of the instrumentation of the technique. The basic experimental tool
was adapted from various TIRF studies at the solid-liquid interface. However, there were
several differences between the two protocols. For example, the nature of the oil-water
interface precludes the use of a standard microscope, and the use of an inverted

microscope necessitated a total rethinking of the experimental cell and beam optics in

relation to the position of the microscope objective. Several generations of equipment
were developed, modified and improved. The key concerns addressed were enhancements
of the signal-to-noise ratios of the PMT and improvement of the surface selectivity of the
TIRF technique. To illustrate we were able to reduce the noise levels from approximately

40% to 0.5%.

This dissertation was divided into three studies with auxiliary information provided in the
appendices. The first study introduces the TIRF technique and provides a description of
the development, design criteria, construction, and trouble shooting of the experimental
set-up. The study also presents details of the experimental procedure and preliminary
results of equilibrium adsorption of BSA and lysozyme. The qualitative agreement of the
adsorption curves with the theory, independent interfacial tensiometry experiments and
previous results in the literature confirms the integrity of our liquid-liquid interface, and

validates the technique.

An important aspect of modern protein chemistry is the attachment of an extrinsic label to
proteins. The second study investigates the effect of a label on the adsorption kinetics and
difliision of proteins. Several researchers have discussed the possibility that extrinsic
fluorophores afi‘ect protein behavior; however, there are no definitive conclusions reported
in the literature. This study uses ion-exchange chromatography to obtain protein samples
with difl‘erent monodisperse labeling ratios. Subsequent TIRF experiments on these
samples revealed that the label does have a pronounced effect on both diffusion and

adsorption kinetics.

The third study presents a calibration protocol for the quantitation of interfacially
adsorbed proteins in our system. The method is self-contained, in that no other measuring
systems besides the TIRF apparatus is required. The protocol uses an absolute criterion
for focusing the microscope that automatically accounts for unevenness of the oil-water
interface. Thus, protein concentrations estimated at difl‘erent locations on the interface are

comparable.

33's
m

cm

2. A TIRF TECHNIQUE FOR INVESTIGATING
MACROMOLECULAR ADSORPTION AND
INTERACTIONS AT THE LIQUID-LIQUID

INTERFACE.

2.1 Introduction and background

The liquid-liquid interface is of considerable importance in biological systems, particularly
in biomedical engineering, pharmacology and food processing. The hydrocarbon/water
interface provides an excellent first approximation to cellular interfaces where many
important biological fimctions and/or processes occur. These include receptor-ligand
interactions, energy transduction, electron transfer, enzyme coupling, and triggering of
cells by hormones and neurotransmitters (1). In biomedical applications, the behavior of
drugs at biological membranes is an important factor in determining pharmacological
activity (3 5). In such cases, the oil-water interface can be used to characterize and/or
understand drug delivery issues, since it can be used to approximate membrane structure
and/or activity. In effect, while biomembranes are structurally complex, immiscible oil-

water interfaces may be used as simplified models in drug delivery experiments.

The liquid-liquid interface provides an excellent model for interactions of macromolecules
at relatively mobile interfaces (61). It also affords enormous flexibility with respect to
system complexity. For example, by adding surfactants and ions to a simple oil-water
interface and/or controlling the pH, the interface can be made as complex as desirable,

enabling one to mimic many biological systems. In particular, the hydrocarbon-water

interface is a simple representation of a biological membrane surface that separates the
aqueous cytosol or external medium from the apolar central region of a bilayer membrane

(36).

The liquid-liquid interface is also crucial to food systems and food processing. For
example, the distribution and/or interactions of proteins and surfactants at this interface
greatly influence the properties of food emulsions and microemulsions. The stability of
emulsions such as mayonnaise depends to a large extent on the adsorption and/or
interactions of proteins and/or surfactants at the oil-water interface. Protein-stabilized
emulsions also influence such properties as shelf life and foam stability in many food
systems. The food industry (as well as the pharmaceutical industry) also routinely uses
proteins to lower the interfacial tension between two immiscible liquid phases, to facilitate

the dispersion of one phase in the other (48).

In spite of its considerable importance in biological systems, this interface has, until
recently, received relatively little attention. This is primarily due to the lack of appropriate
quantitative techniques that may be used to probe the interface at the molecular level. One
of our goals in developing the technique described in this chapter was to help mitigate this

shortcoming in experimental tools.

2.1.1 Rationale for using T [RPM for the liquid-liquid interface

Total internal reflection fluorescence microscopy (TIRFM) is an invaluable in-situ, non-

invasive and non-destructive tool for studying interfacial phenomena at the molecular

10

 

US

1‘8

re.

level. The ability to directly measure molecular-level phenomena represents an
improvement over techniques that measure macroscopic properties and derive micrOSCOpic
parameters fiom the data. TIRFM provides the ability to selectively excite species within
tens of nanometers from a given interface. It can be used to quantify the concentration of
species as a fiInction of distance fi'om a given interface, allowing one to directly determine
adsorption and desorption kinetic parameters (62). One of the earliest studies using TIRF
for fluid-fluid interfacial adsorption was done by Tweet et al. (63) when they used the
technique to measure the emission spectrum and quenching behavior of chlorophyll a
monolayers at the air-water interface. By using TIRF, the weak fluorescence generated by
the dilute monolayer was detected while light from direct and scattered mercury arc

excitation was excluded.

There are a few published applications of TIRF at the liquid-liquid interface. Among
these, Morrison and Weber (3 6) studied the adsorption of 4,4’-bis-l-phenylamino—8-
naphthalenesulfonate (bis-AN S) at the decalin-water interface; and Takenaka and
Nakanaga (42), Nakanaga and Takenaka (41) and Takenaka (43) have presented results of
a series of experiments on adsorption of anionic dyes to the water-chloroform interface,
using Raman scattering to infer the orientation of dye molecules at the interface. More
recently, TIRFS has been used by Wirth and Burbage (64) to study acridine orange
reorientation at a water/(octadecylsilyl)silica interface, Besso et al. (65) have studied the
microenvironment of 8-anilino- 1 -naphthalenesulphonate at the heptane/water interface,

Watarai and Saitoh (66) have studied the ion-association adsorption of water-soluble

11

porphyrins at the liquid-liquid interface, and Tupy et al. (67) have studied dynamic

adsorption at the oil-water interface.

Several standard (non-TIRF) techniques have also been used to study adsorption at the
liquid-liquid interface (68, 69, 70). These techniques include monitoring changes in
interfacial tension (48, 71, 72, 73); measuring the interfacial concentration of 14C labeled
species by scintillation counting (50); measuring rheological properties such as interfacial
viscosity and/or shear (48, 71, 74); monitoring film viscosities and viscous traction (48,
52); evaluating surface voltage potential differences (particularly in the case of spread

monolayers); measuring surface difliision; and characterizing electrokinetic phenomena.

Among other optical studies, Grubb et al. (46) used second harmonic generation to
determine the orientation of molecular monolayers of surfactants at various liquid-liquid
interfaces, Conboy et al. (44) used sum-fiequency vibrational spectroscopy (in TIR
geometry) to investigate the conformation and order of sodium dodecyl sulfate adsorbed
at the water-carbon tetrachloride interface, and Sauer et al. (47) used quasi-elastic light
scattering to determine the dynamic surface pressure of dipalmitoylphosphatidylcholine at

the heptane-water interface.

12

2.2 Objectives

Our goal in this study was to develop a total internal reflection fluorescence microscopy
(TIRFM) technique to enable molecular-level investigations of adsorption and interactions
of amphiphilic macromolecules at the liquid-liquid interface. In this chapter, we present
the details of the experimental set-up and the nature of the interface, and briefly outline the
results of experiments on the equilibrium adsorption of a hard protein (lysozyme) and a

soft protein (BSA) at the oil-water interface.

2.3 Theory

The theory of total internal reflection has been well treated elsewhere (62, 75, 25, 30). As
such, only a brief discussion will be given here. A more detailed description of

development of the evanescent wave equation can be found in Chapter 4.

2.3.1 The Evanescent wave
Total internal reflection occurs when a light beam, emanating from an optically dense

medium (72,) reaches the interface between that medium and an optically rarer medium

(n2) at an angle which exceeds a critical angle (9,) defined by

9, s sin"[£2—) n, >n, (1)

"I

where nl and n2 are the reflective indices of the dense and rare media, respectively.

13

 

in:

When the beam undergoes total internal reflection, it induces an electromagnetic field in
the rarer medium beyond the interface (Figure 2.1). The intensity of this induced
electromagnetic field (or evanescent wave) decays exponentially with perpendicular
distance 2 from the interface. As a result, the light impinging on a body in the evanescent
wave also decays exponentially with distance {Tom the interface according to the

relationship
1(2) = Ioe""’r (2)
where 1(2) is the intensity at any depth 2 and I, is the incident light intensity at the

interface.

The penetration depth, d, , of the evanescent wave is given by

’1
d = 0 (3)
p 471721 ‘/sin2 6—11212

 

where 110 is the wavelength of the incident radiation in vacuum, 6 is the angle of

incidence, and n21 2 n2 /n1 is the relative refractive index.

14

 

Laser beam

 

 

Evanescent Wave

 

 

 

Figure 2.1. Total internal reflection at the oil-water interface. The resulting evanescent
wave decays exponentially into the rarer aqueous phase.

15

IE

2.4

2.4

POIIII
Calibr.

direct]

There are two optical techniques that exploit the properties of the evanescent wave to
measure the sub-micron distances between macromolecules and interfaces. These are total
internal reflection microscopy (TIRM) and total internal reflection fluorescence (TIRF).
TIRM has been used to measure the energies of interaction between a single particle and a

flat surface. By measuring 10 and 1(2), the separation distances 2 between the particle and

the surface can be calculated from Eq. (2) and converted to a potential energy of
interaction (76). TIRF also exploits the evanescent wave at the interface between two
materials with different refractive indices, except that fluorescence emission intensities
rather than light scattering intensities are measured. Because TIRFM can accommodate
much smaller particles for which light scattering intensities are generally not measurable, it

is an important technique for studying protein adsorption.

2.4 Description of apparatus

2. 4. 1 Developmental challenges

There were several challenges in developing and assembling this apparatus. The first
concerned the adaptation of the standard solid-liquid interface TIRF apparatus to the
liquid-liquid interface. In particular, several precautions and mechanisms had to be put in
place to ensure interfacial stability. The second challenge was to ascertain that the steps
taken to help ensure interfacial stability had not modified the interfacial energetics to the
point that one could not guarantee the integrity of the interface. The third was to devise a
calibration scheme to enable the calculation of interfacial fluorophore concentrations

directly from fluorescence intensity data.

16

0P
Un
per
mi
achj
Unit

deSCl

2. 4.2 T IRFM set-up

The experimental setup is shown in Figure 2.2. In summary, the apparatus uses an inverted
microscope (Zeiss Axiovert 135 M, Carl Zeiss, Inc., Thornwood, NY), a 5 W continuous
wave argon ion laser (Lexel Lasers Model 95, Fremont, CA), a side-on photomultiplier
tube (Hamamatsu R463 2, Bridgewater, NJ) jacketed in a thermoelectrically cooled
housing (TE177TSRF, Products for Research, Danvers, MA), a CCD camera (MTI,
VE1000, Zeiss), a computerized data acquisition system (Viewdac, Keithley/Asyst,
Rochester, NY), a modular automation controller with an R8232 interface (MAC 2000,
Ludl Electronic Products, Hawthorne, NY), and a double-syringe pump system (Model
551382, Harvard Apparatus, South Natick, MA) used to simultaneously infiise and
withdraw samples from the flow cell. The entire assembly of components is mounted on a
vibration isolation table (RS 4000, Newport Corporation, Irvine, CA) to ensure interfacial

stability.

We use the 488 nm line of an argon ion laser housed in a light-tight box. The laser light is
first attenuated using neutral density filters (03FNQ, Melles Griot, Boulder, CO). An
optical chopper (SR540, Stanford Research Systems, Sunnyvale, CA) is used to prevent
unintended photobleaching of fluorophores during experiments conducted over long
periods of time. Our work requires a rapid and reproducible method for instantaneously
switching from bleaching laser beam intensities to monitoring beam intensities. This is
achieved by a system of two identical optical flats, a programmable shutter (D122,
Uniblitz, Vincent Associates, Rochester, NY) and a set of neutral density filters, as

described below.

17

bit
in

ad

fir

ranc
500i“
Cont

ceH,i

miCl'oS

The front of the first flat is coated with magnesium fluoride and the back with aluminum,
to discourage and encourage reflection, respectively. The monitoring beam is formed by
the weak reflection off the front surface while the photobleaching beam reflects ofl‘ the
aluminized back surface. The second flat is aligned to exactly recombine the two beams to
ensure that the monitoring beam is accurately superimposed on the photobleaching beam
at the interface. The shutter, placed between the two flats, blocks the photobleaching
beam only, but allows for photobleaching with a sub-millisecond laser flash upon
activation. A typical bleach pulse duration of 100 msecs is used in our experiments. This
was found to be sufficient to entirely bleach the interfacial fluorophores and yet not cause
any undesirable heating effects (77). A second shutter placed inside the microscope cavity
blocks ofl‘ the PMT for the duration of the photobleaching pulse. The driver/controllers
for the two shutters are wired such that they can be triggered by a single footswitch. By
adjusting the delay and exposure times, the two shutters can be exactly synchronized.
This is important to prevent damage to the PMT during photobleaching and also aids in
the mitigation of light hysteresis of the PMT (see chapter 4 for firrther discussion). The
ratio of photobleaching to monitoring beam intensities can be varied from 100021 up to
5000:1, and is typically 4000:1 in our experiments. Beyond the flats, the laser beam is
conveyed by a system of mirrors, and finally focused onto the interface in the experimental

cell, using a plano-convex lens (f = 200mm, Oriel Corporation, Stratford, CT).

The sample cell (described more fully below) is accessed through a Zeiss Achroplan

microsc0pe objective (LD 440651, 32X, 0.4 or 440091, 40X, 0.75, Carl Zeiss Inc.,

18

Thornwood, NY) focused at the interface. The fluorescence excited by the incident laser
light is captured by the objective and can either be viewed by a CCD camera or measured
by the PMT. An interference filter (5 OOEFLP-EM, Omega Optical, Brattleboro, VT)
placed inside the microscope ensures that only the wavelength range of the emitted

fluorescence is transmitted.

l9

 

 

  
    

'11

neutral density 11hr
opdal 1hr

minor

11mm:

Almldnnm coating

  
 
 
 
  
  
  

mm); bum
photobleaching been

‘85 “33'“5

‘1‘ «(5‘ x“

inverted microscope

 

 

Figure 2.2. Layout of the experimental setup for TIRFM experiments. The double-flat
system is used to first separate the incident beam into monitoring and photobleaching
streams, and then recombine the two beams. This scheme provides control over beam
intensities while ensuring superposition at the interface.

20

2. 4.3 The sample cell

The flow cell that houses the oil-water interface is of crucial importance in this
experimental set-up. Two microscope slides are separated by a 1-mm thin aluminum
spacer, with an oval cut into it to form the experimental sample cell. The underside of the
top microscope slide is coated with a thin layer of oil. This layer, approximately 80 pm
thick, forms the oil part of the interface. Two holes drilled in the bottom slide enable
infusion and withdrawal of water and/or protein solutions from the apparatus. The bottom
slide is ground to a thickness of 0.75 mm to allow for the use of the appropriate
microscope objective. The top slide is optically coupled to a 70° dovetail prism and the
entire assembly is firmly screwed onto an anodized aluminum shell that sits on top of an
inverted microscope stage (Figure 2.3). The microscope stage can be moved precisely
and repeatably by a computer, enabling difi‘erent areas of the interface to be studied in a

series of runs.

The double-syringe pump is used to simultaneously infuse and withdraw protein solution
into and out of the experimental cell in a controlled manner. The setup can be used for
flow-through experiments, but has so far been mostly used in the stopped-flow mode.
Prior to each experiment, the chamber is filled with phosphate bufl‘ered saline (PBS)
bufl‘er, which is replaced by a protein solution of the appropriate concentration for each

experiment.

A critical and novel aspect of this setup is that the thickness of the entire oil-water

assembly is approximately 1.0 mm. The primary advantage of this construction is that,

21

fiom a global or macroscopic point of view, the system firnctions like a solid-liquid
interface. Therefore, it does not suffer from the typical vibration-induced instability
problems encountered at liquid-liquid interfaces of normal depth. Thus, interfacial stability
is essentially guaranteed. However, from a microscopic or molecular point of view, it
behaves as a true liquid-liquid interface since, for a protein molecule with a linear
dimension of no more than 20 run, an 80 um oil layer is sufficiently thick and should not

affect its dynamics. This was the first goal in our apparatus development.

The second goal was to put the thin oil layer in contact with a protein solution
approximately 1000 um thick. This water/solution layer is thin enough to give us an
equilibrium adsorption time on the order of minutes rather than hours. In addition, the
apparatus makes it possible to use hydrodynamic shear to sweep the interface clean and
create a new interface within a reasonable amount of time although, in practice, this has

not been very easy to accomplish.

2. 4.4 Materials

The proteins used for the study presented in this chapter were essentially fatty acid-free
BSA (A-7511, Sigma Chemical Co, St. Louis, MO) and lysozyme (L-2879, Sigma
Chemical Co, St. Louis, MO ). Protein concentrations in all our regular experiments were
kept below the critical micelle concentration, therefore; only single protein interactions are
normally characterized with this equipment. In this chapter, however, we present standard
protein adsorption data, therefore concentrations range from infinite dilution to above the

critical micelle concentration.

22

 

WMoIMMrm

 

 

Oil- Water Interface

 

 

 

Ahmirrimn Spacer

 

O-rings Bottom Glass Side

 

Teflon Tubing
To Syr‘nge Pump

 

 

 

(o 0 } O-rina

 

 

 

 

 

 

 

Figure 2.3. The experimental cell, showing the location of the interface. The thickness of
the oil-water layer is less than 1.0 mm, which helps to ensure interfacial stability.

23

2.5 Experimental Procedures

2. 5.1 Coating of microscope slide with oil

The lyophilic layer at the interface is formed with Cargille Type A immersion oil (Fisher
Scientific, Pittsburgh, PA). The microscope slide is cleaned by the procedure outlined by
Cheng et al. (78). Before coating with oil, the slide is first heated for 5 minutes under a
heat lamp. Then a drop of oil, of a pre-determined volume, is placed at the top of the slide.
A clean cylindrical glass rod is smoothly dragged and gently rolled once over the slide,
thereby sweeping the surface of the slide and evenly coating it with oil. The slide is then
optically coupled to the prism by applying an appropriate refractive index matching fluid
(Cargille Type A immersion oil, Fisher Scientific, Pittsburgh, PA) to the upper surface in
contact with the prism. The slide and the prism are then placed on top of a flow cell
(initially filled with double de—ionized water or protein solution) to form the oil-water
interface. We estimated that the oil layer interacting with the water to form the oil-water
interface is approximately 80 um thick. This was done by carefully comparing the weights
of microscope slides, before and after coating with oil, to deduce the weight of the oil
layer. The thickness of the oil layer was then deduced from the surface area of the slide

and the density of the oil.

2. 5.2 Protein labeling

Proteins were labeled with fluorescein-S-isothiocyanate (F ITC, F—1907, Molecular Probes,
Eugene, OR) in our laboratory, using the techniques outlined by Edidin (79). The
unreacted labels were removed by a two-stage dialysis process performed over a period of

24 hours. Absorbance spectroscopy measurements with a diode array spectrophotometer

24

(8452 A, Hewlett Packard, Brielle, NJ) at 280 nm and 500 nm were used to determine the
protein and label concentrations, respectively. The chemistry of the labeling reaction is

discussed in chapter 3 of this dissertation.

2. 5.3 Beam alignment and microscope focusing protocols

The proper alignment of the TIRF system is primarily dependent on achieving normal
incidence of the laser beam on the TIR prism, while simultaneously ensuring that the point
of TIR is exactly above the microscope objective. This is not a trivial matter, because the
optical alignments required to focus the microscope tend to destroy the normal incidence
of the beam and vice versa. However, repeated iterations of the two procedures, along
with horizontal translation of the beam steering device, will eventually yield the desired

result.

Normal incidence of the laser beam was achieved by tracking the reflection of the beam
from the prism surface. The optical components of the system were adjusted till the back
reflection of the laser beam re-entered the laser. This indicates that the beam has achieved
normal incidence at the prism. Each optical component, starting from the mirror closest to
the prism, was adjusted in turn to ensure that the reflection fiom the next optic exactly
retraces the incident beam path. The optical component furthest upstream from the laser
was later slightly displaced, to prevent the back reflection from re—entering the laser during

the course of the experiments.

25

The microscope was focussed as follows: a coarse alignment was first performed with the
naked eye to roughly place the point of TIR over the microscope objective. Then,
looking through the PMT cavity on the microscope (the PMT was removed for this
procedure), the top mirror of the beam steering device was adjusted to bring the
evanescent wave under the microscope objective. The microscope was then focussed till
the evanescent wave fills the objective viewing area. This was performed after removing

the interference filter fiom the microscope at a laser beam intensity of about 0.1 mW.

The PMT was then replaced and minute adjustments were performed to the microscope

and the beam steering device to maximize the PMT signal.

2. 5.4 Introduction of proteins into the flow cell

The experimental cell was first flushed with pure PBS buffer at a pH of 7 .4, for a period of
25 minutes. This served to remove any air bubbles in the tubing and the experimental cell.
At this time, the PMT shutter was opened and the background noise present in the system
was estimated. This level, usually about 100 arbitrary intensity units, was later subtracted
from all PMT readings. The proteins were then injected into the cell using a reciprocating
syringe pump that simultaneously withdraws PBS buffer solution. To prevent artifacts
due to mechanical shearing of the interface, care was taken to ensure that the flowrate
does not exceed ~0.1 mL/min. The solution was allowed to flow into the sample cell until
the PBS bufl‘er initially occupying the flow cell had been completely replaced by the
protein solution. At this time, the syringe pump was turned off, and a stable interface was

re-established.

26

2. 5. 5 Experimental precautions

The first concern was with unintended photobleaching of labeled adsorbates.
Photobleaching is the random, dynamic process in which fluorophore molecules undergo
photo-induced chemical destruction and thus lose their ability to fluoresce. This was
always a strong possibility in TIRF experiments when adsorption was slow to reach
equilibrium and one continuously monitors the process. We followed the procedure of
Lok et al. (27, 2) and used a beam chopper during long experiments. The photobleaching
beam was always blocked during periods of fluorescence measurement through the use of
the triggered shutter system described earlier. The trigger that activates the
photobleaching beam also simultaneously closes an LS6Z2 shutter (Vincent Associates,
Rochester, NY) mounted inside the microscope. This serves to protect the PMT from the
high power photobleaching flash and prevent PMT hysteresis, which distorts the initial

fluorescence recovery curve.

A second experimental necessity was the prevention of photochemically induced cross-
linking of proteins at the interface. When this occurs, it was accompanied by a slow
continual illumination-dependent increase in the fluorescence signal (62). To prevent this,

the samples were deoxygenated before introduction into the sample chamber.

2.6 Problems and opportunities for the liquid-liquid interface

Flow cells are routinely used in studies at the solid-liquid interface without the risk of

destroying the interface. \Vrth the liquid-liquid interface, however, there is a strong

27

possibility of sweeping away molecules adsorbed at the interface during continuous flow
experiments, as a result of hydrodynamic shear. While this is a potential problem, the fact
that the interface can be swept clean is also an advantage of this setup. The ability to use
hydrodynamic shear to clear the interface of adsorbates enables us to greatly minimize the
time required to create a new interface. By using this clean-in-place strategy, a given
quartz slide can be left in place and used to run several experiments without having to
change the slide each time a clean interface is required. Our experiments were usually
conducted in stopped-flow mode, but we have also determined the maximum flow rate at
which flow-through experiments can be conducted without inducing unacceptable levels of

hydrodynamic shear at the interface.

2.7 Results and discussion

2. 7.1 Timescale of equilibrium adsorption

Figure 2.4 shows the results of a TIRFM experiment to measure the equilibrium
adsorption of BSA at the oil-water interface. The bulk protein concentration was 0.01
mM with a labeling ratio of one FITC per BSA. The syringe pump was operated at an
injection rate of 0.05 ml/min. To reduce the chances of photobleaching by the monitoring
beam, the optical chopper was completely masked with aluminum foil, except for a 1-cm
wide slit. This reduced the duration of the monitoring beam incidence on the interface by

approximately 95%.

28

1200

 

 

 

 

 

S
1000-
:5
>
3:
m 800*
C
o
E
g 600‘
C
8 400‘
m
2
O
2 200‘
u.
o l r I l
O 5 1O 15 2O 25
Time(min)

Figure 2.4. Typical TIRFM protein adsorption experiment showing fluorescence intensity
versus time during the equilibrium adsorption of BSA at the oil-water interface. From the
data, the time required to reach adsorption equilibrium was estimated at about 20 minutes.

29

At t = 5 min, the flow of protein solution was started. As the protein adsorbed, the
fluorescence signal increased steadily and reached a steady-state value at interfacial
saturation. This was observed to occur at t ~ 20 minutes. This confirms the fact that the

thinness of the oil-water assembly allows for faster equilibrium adsorption times.

Note that the signal to noise ratio (~10%) is larger than what is typically seen in other
TIRFM experiments presented in this dissertation. This is due to the presence of the
optical chopper. Since PMT data collection is not synchronized with beam chopping
frequency, a part of the collection cycle frequently occurs with the incident beam partially
blocked. This leads to a random artificial lowering of the PMT signal, which results in an

increased signal to noise ratio.

2. 7.2 Assessment of interfacial integrity

A key concern of this study was whether a thin oil-water interface, configured to a
dimension of no more than 80 pm of oil on top of approximately 1000 um of water,
would behave like a “true” oil-water interface. As already discussed, the thinness of the
interface was necessary to ensure interfacial stability, and prevent the interface from
shitting or rolling in the presence of vibration. To assess the interfacial integrity of our
apparatus, we measured the emitted fluorescence and, from the signal, computed the
relative interfacial concentration of adsorbed species. The adsorption isotherms were then
compared to isotherms obtained from interfacial tension measurements at a "normal" oil-
water interface, using the Kriiss 12K Interfacial tensiometer (Kri‘rss, USA). The results

for BSA and lysozyme are discussed below.

30

2.7.2.1 Equilibrium adsorption of BSA at the oil-water interface

Most amphiphilic or surface-active solutes have the property of forming colloid-size
clusters called micelles when their concentration in an aqueous solution is at or above the
critical rrricelle concentration (cmc). A graph of most physical properties that are firnctions
of size or number density of solute particles display a break in the neighborhood of the
cmc. We postulated that an important test of the integrity of the oil-water interface in our
apparatus is whether the apparatus can be used to estimate the cmc. To do this
assessment, we measured the interfacial fluorescence intensities as a function of the

concentration of FITC-conjugated BSA molecules in solution.

The results are shown in Figure 2.5a. As expected, the fluorescence intensity rises with
concentration until the interface approaches saturation. At that point, the fluorescence
intensity begins to level off. However, there is a clear break in slope between the region
of sharply rising fluorescent intensities and the region where fluorescent intensities appear
to be stabilizing with respect to concentration. This break occurs at a BSA concentration
of 0.1 mM, which is approximately the cmc of this protein, as determined by interfacial
tension measurements. These data provide strong evidence that we are indeed measuring
phenomena at a thermodynamically true oil-water interface, and that the thinness of the

oil-water layer has not affected protein dynamics.

31

 

 

 

 

 

 

 

 

 

 

   

 

 

a 1400
A . .. - 1200
g 0 ‘ 1 1000
§ . 800
8 o
rn — 600
Q o
3 ‘ a 400
“- o

. o - 200
. O
I I r 0
-20 -15 -10 -5 0
In (concentration of BSA-F ITC (mM))
b.
12 T
10 Ar #3
s 8 “
s 6 _
§
8 4
In
S 2
2
1.1..
0 i i i I I I
O.E+00 1.E-05 2.E-05 3.E-05 4.E-05 5.E-05 6.E—05
Lysozyme-FITC concentration (M)

 

 

 

Figure 2.5. Equilibrium adsorption profiles for lysozyme and BSA conjugated with the
FITC label. Plot a) for BSA is at a much higher concentration, and portrays a solution
that has approached and exceeded interfacial saturation. The plot shows a clear break in
profile at the cmc. Plot b) for lysozyme shows a gradual approach to interfacial
saturation, with no indication of a cmc.

32

2.7.2.2 Equilibrium adsorption of lysozyme at the oil-water interface

The adsorption isotherm for lysozyme is plotted in Figure 2.5b. It shows a region where
the fluorescence intensity initially rises linearly with concentration, as would be expected.
As the concentration increases fiirther and the interface becomes saturated with proteins,
the fluorescence intensity levels off. This is a classical adsorption isotherm. We should
note that there is no point of inflection or change of slope to indicate the formation of
micelles in this system. Therefore, the profile is as expected, since lysozyme is a

predominantly hydrophilic protein that does not form micelles in aqueous media (80).

2.8 Conclusions

We have developed a total internal reflection fluorescence microscopy (TIRFM) technique
for quantitative molecular-level investigations of macromolecular adsorption and
interactions at the oil-water interface. Due to the novel construction of the oil-water
interface in our sample cell, we have circumvented the typical vibration-induced instability

problems encountered with optical techniques at liquid-liquid interfaces of normal depth.

An important goal during our development was to put a thin oil layer in contact with a
protein solution that is no more than 1000 um deep. The thinness of the oil-water
assembly makes it possible for us to achieve a diffusion time constant of no more than 3
minutes (see Table 1, chapter 3), and an equilibrium adsorption time on the order of
minutes rather than hours. It also provides the possibility of using hydrodynamic shear to

create a new interface.

33

In spite of the thinness of the resulting interface, it behaves as a true liquid-liquid interface,
as evidenced by the classical adsorption data we obtained for lysozyme and BSA To
evaluate the interfacial integrity of our sample cell, we used the apparatus to estimate the
critical micelle concentration (cmc) of BSA A clear break in the adsorption isotherm,
indicative of the cmc, occurred at approximately 0.1 mM, which is a typical value for this
protein. On the other hand, the adsorption isotherm measured for lysozyme displayed a
smooth curve without any breaks to indicate the formation of micelles, which is to be

expected for a predominantly hydrophilic protein such as lysozyme.

The system is very flexible with respect to sample complexity. For example, by adding
surfactant molecules and ions to this simple oil-water interface and/or controlling the pH,
one can create more biologically relevant systems and interfaces. System flexibility also
allows the study of macromolecular adsorption at difi‘erent liquid-liquid interfaces, as long

as both liquids are optically transparent.

34

3. EFFECT OF EXTRIN SIC FLUORESCENT LABELS ON
DIFFUSION AND ADSORPTION KINETICS OF BSA AT
THE LIQUID-LIQUID INTERFACE.

3.1 Introduction and background

An important and widely-used technique in interfacial macromolecular adsorption
experiments is the covalent attachment of an extrinsic fluorescent label. The purpose of
labeling a macromolecule is to provide it with a "marker" that makes it readily trackable by
optical means. Ideally, the marker emits a measurable fluorescence upon excitation at an
appropriate wavelength. This fluorescence is an extremely sensitive optical technique for
the detection of biological materials. This sensitivity, combined with the ease of use and
the potential for multiparameter analysis, has resulted in widespread use of fluorescence in

biology and medicine (81).

Extrinsic labels have been used to study a variety of macromolecular characteristics and
behaviors. These include adsorbate enzyme activity (82), receptor-ligand binding (11, 12)
and binding of proteins to specific membrane targets (18). They have also been used to
model planar membranes (15, 20) and study conformational changes of proteins upon
adsorption (83, 33, 84), as well as kinetics of protein adsorption (4, 12). Labels have also
been used to study various aspects of protein adsorption at the solid-liquid (27, 32, 9, 78,
85) and liquid-liquid interfaces (86, 67, 87), and at biological membranes (88, 89, 90).
Labeled proteins have also been very useful in such techniques as radioirnmunoassays,

enzyme immunoassay, fluorescent antibody staining and nucleic acid probes (91).

35

Ideally, labeling a macromolecule should not compromise the target properties or
characteristics of interest in the native state. However, several researchers have discussed
the possibility that extrinsic fluorophores do affect the biological phenomena being
studied. For example, Axelrod et al. (62) and Burghardt and Axelrod (4) stated that, in
principle, such labels probably affect the adsorption of proteins. Hoyland (92) has also
reviewed the potential artifacts resulting fi'om using fluorescence probes in membranes.
Zimmerman et al. (14) showed that there were no difl‘erences in thermal denaturation (as
measured by static light scattering) or enzymatic activity between labeled and unlabeled
lysozyme. They concluded, however, that the labeled lysozyme, while being a functional
protein, could have other altered properties that might be reflected in their adsorption

behavior.

As early as 1978, Brynda et al. (93) evaluated alternatives to radioactive-labeling of
proteins and concluded that BSA exhibited changes in its native electrophoretic properties
when labeled with DAN SI, fluorescamine or F IT C. Sorption properties were also
observed to be affected by conjugation with fluorescamine. However, no attempt was
made to quantify the effects of the label. Crandall et al. (94) used electrophoretic mobility
and ion-exchange chromatography, together with measurements of the degree of
conjugation and sulfliydryl content, to study the effect of FITC, fluorescamine and
radioiodine labels on BSA. The authors provided evidence of instability of protein
conjugates and labeling-induced changes in their chromatographic and electrophoretic
properties. They concluded that fluorescein conjugates should not be used for adsorption

studies. In contrast, Robeson and Tilton (95) showed that FITC-Rnase A did not

36

preferentially adsorb in comparison to the native protein. However, due to possible
complications arising from concentration quenching, the labeling ratios used were very
low (41.02). Therefore, this study may not provide conclusive evidence of the effect of

labeling.

Lok et al. (27) used BSA doubly labeled with both tritium [3H] and FITC to develop a
calibration technique for adsorption of macromolecules at the solid-liquid interface. They
investigated the influence of the label on protein adsorption by using mixtures of
[3H]F ITC—B SA and unlabeled protein. They showed that mixtures containing 50 and
25% labeled proteins gave adsorption results with fluorescence intensities and tritium
counts that were both 50 and 25%, respectively, of adsorption experiments with 100%
labeled proteins. This study, performed in 1983, has since been widely cited to support
the claim that there is no preferential adsorption of labeled protein over unlabeled protein
or vice versa. However, we believe that the conclusions inferred in this study are unique
and applicable only to the dual-labeled protein system used. We speculate that attaching a
positively charged tritium molecule to the protein-label conjugate effectively nullifies the
increase in electronegativity of the protein due to the addition of the fluorescence label.
Hence, the surface hydrophobicity of the lysine group (to which both [3H] and FITC are
attached) remains unaltered, and therefore the sorption properties of the dual-labeled

protein are unchanged in comparison to the native species.

Burghardt and Axelrod (4) measured the differences in equilibrium adsorption of

unlabeled BSA and rhodamine-labeled BSA over various labeling ratios. Labeling ratios

37

were varied by either changing the molar rhodamine to BSA ratio or diluting labeled
samples with unlabeled proteins. They concluded that the measured differences fall within
the intrinsic error of the measuring system and could, therefore, not be totally attributed to
labeling difl'erences. However, they could not completely rule out the possibility that
rhodamine-B SA sorption behavior was different from that of unlabeled BSA Sacco et al.
(96) showed that benzenehexacarboxylate (BHC) dextran-labeled haemoglobin exhibited a

decreased oxygen afinity, rendering it suitable for use as a blood substitute.

With this uncertainty on the efl‘ect of labeling on protein dynamics, several studies have
completely avoided using extrinsic labels by studying the intrinsic fluorescence of
tryptophan or tyrosine residues in the protein (7, 9, 10, 8). Clearly, understanding the
efl‘ect of fluorescence labeling on protein dynamics is important, and there is a need for
further investigation into its effect on protein adsorption. However, a comprehensive
search of the literature did not yield any studies that have specifically attempted to
quantify this effect. This study aims to address this gap, by quantifying the changes in
diffusive properties and adsorption kinetics that result from the conjugation of fluorescein-

5-isothiocyanate (FITC) to bovine serum albumin (BSA).

3.2 Objectives

Our goal in this study was to quantify the effect of labeling BSA with FITC on the
adsorption and diffusion of BSA. We have analyzed three different levels of BSA-FIT C
conjugations: BSA-IFITC (with exactly one conjugated FITC per BSA molecule), BSA-
2FITC (with exactly two FITC per BSA molecule) and BSAm—FITC (with an average of

38

approximately 1.7 FITC per BSA molecule). We studied the three protein samples under

two different concentration regimes:

1. The diffusion-limited (low concentration) regime, where we determined the mobile
fraction and the apparent diffusion coefficient; and

2. The reaction-limited (high concentration) regime, where we detemrined the slow and
rapid desorption rates and the fiaction of slowly reversible and rapidly reversible

proteins.

All studies were conducted at the oil-water interface.

3.3 Theory

3. 3.1 Chemistry of the protein-labeling reaction

Fluorescein derivatives are the most common reagents for covalent labeling of proteins,
because fluorescein has a relatively high absorptivity and an excellent quantum yield
(~09). Additionally, it has an absorption maxima of about 490-494 nm that matches the
spectral wavelength of an Arg-Ion laser (488 nm), and produces a fairly stable protein

conjugate. The general labeling reaction is shown below for BSA.

R‘—N=C=S + Rz—NH2 —> R’-MT—CS—NHR2
isothiocyanate primary amine thiourea conjugate [4]
(FITC) (BSA) (BSA — F ITC)

In most cases, fluorescent labeling utilizes the primary amino group on the target

molecule. In the case of a polypeptide like BSA, the primary amino acid is usually in the

39

lysine group. The isothiocyanate group on the fluorophore provides eficient amino group
labeling (81). A higher pH favors the -NH; group, which reacts well with the

isothiocyanate over the unreactive —NH3+.

3.3.2 Diffusion of BSA-FIT C

Fick’s second law governs the lateral diffusion of a single non-interacting species of

protein in the absence of convection.
L2: ’) = DV2C(r, t) [5]

where C(r, t) is the concentration of the diffusing species, D is the diffusion coemcient, t is
the time of diffusion and r is the radial position of the species at the interface.
The concentration profile immediately after photobleaching is given by

C (r,0) = Ce‘“ m [6]

where C is the concentration before photobleaching, and K is the bleaching parameter,
which is a function of the rate constant for photobleaching and the duration and intensity

of the photobleaching pulse (29).

The fluorescence recovery profile following photobleaching is given by (29)

F(t) = (31)} I(r)C(r, t) dzr [7]

40

where F (t) is the fluorescence at any time after the photobleaching flash (which occurs at t
= 0 by definition). The quantum yield, q, is a product of the eficiencies of excitation light
absorption and fluorescence emission and detection, I(r) is the evanescent wave intensity
profile at the interface, and A is the attenuation factor of the beam during the period of

recovery.

The task now is to determine the firnctional form of C(r, t) so that Eqn. [5] can be
integrated. This problem has been resolved by Axelrod et al. (29), using the solution of
the Fourier transform of Eqn. [5] (subject to the initial condition in Eqn. [6]) to yield a
functional form of Eqn. [6]. Note that the initial condition depends on the bleaching
parameter, K, which is a measure of the degree of bleaching in the system. For a Gaussian
laser beam profile and for small values of the bleaching parameter K, the resulting equation

is

K
2[1+ 1]
r

where F is the pre-bleach fluorescence, and T is the characteristic diflirsion time for spot

F(t) = F 1- [91

 

photobleaching with a Gaussian laser beam. The detailed steps required to arrive at Eqn.

[8], along with associated assumptions, are discussed in Appendix A.

41

3. 3.3 Reaction kinetics of BSA-FIT C at the interface

Given molecules of bulk concentration A(r, z, t) fieely difl‘using in solution and reacting
with binding sites at an interface with a two-dimensional surface concentration B(r, t) to
form fluorescent complexes of surface concentration C (r, t), we can write the reaction

equilibrium as

A(.-,z,t)+19(r,t)”“"E : C(r,t) [10]

where k, and k, are the adsorption and desorption rate constants, respectively. The

vector r is the position at the interface fi'om the center of the observation area, 2 is the
perpendicular distance from the interface observation area to a point in the bulk solution,

and t is time.

The equilibrium of the system is described by

5
AB d I 1

3:- law-

where Z , B , and C" are the equilibrium concentrations of the various species.
Thompson et al. (6) derived a mathematical model for use with TIRF/PPR experiments
using the spot photobleaching technique. A brief summary of the model, as well as the

final equations, is presented below.

The model is based on the monotonically decreasing function G (t), which is defined as

42

G (t) = F—F(t) [12]
where F (t) is the fluorescence at any time t and F is the pre-bleach fluorescence.

Combining Eqn. [12] with a form of Eqn. [42] (see Appendix A) that relates fluorescence

intensity to interfacial concentration gives
G(t)=qJ'I(r)C,,(r,t)d2r [131

where C,» = C - C(r, t) and q is the quantum eficiency discussed earlier.

The differential equations for A(r, z, t) and C(r, t) were solved by linear transformation
theory. The initial conditions were derived from the intensity profile of the monitoring
laser beam. The boundary conditions were based on the fact that only surface bound
molecules within a finite surface area are photobleached. Away from the bleached region

A(r, z, t) and C(r, t) do not depart from their equilibrium values.

Upon integration and application of these boundary conditions, and using a simplification
valid for the reaction limited regime where the desorption rate is much smaller than the

bulk normal diffusion rate, Eqn. [13] can be written as follows (6)

G(t) = G(O)e 4‘“ [14]

In our analysis, we have incorporated a modification, as recommended by Burghardt and
Axelrod (4), to account for two difi‘erent classes of adsorbing proteins. We have also

included an additional modification, based on earlier work in our laboratory by Shulda et

43

al. (87), to account for the unintended photobleaching of fluorophores by the monitoring

beam. These modifications are discussed later in this paper.

3. 3. 4 Concentration quenching: definition and concerns

Concentration quenching is a manifestation of the Forster non-radiative energy transfer
phenomenon (97). This type of nonradiational deactivation of an excited fluorophore is
called long-range quenching or Forster quenching. It is generally considered to be due to
dipole-dipole coupling between a donor (the excited fluorophore) and an acceptor (the
quencher) where the donor and the acceptor are different species. However, many
fluorophores satisfy the requirement of self-energy transfer, namely that there is an overlap
between the excitation and emission spectra. An inspection of the spectral profiles of

fluorescein (appendix B) reveals a large overlap in the emission and excitation spectra.

Concentration quenching results in a drop in fluorescence emission upon increasing the
fluorophore concentration. Understandably, this is to be avoided. The easiest way to do
this is to keep the labeling ratio low, thus reducing the probability of two fluorophores
coming into the close proximity required for quenching to occur. The reaction constant

for energy transfer, k1, is given by (98)

k. = 6(3) [151

where R is the inter-fluorophore separation, r0 is the luminescence lifetime of the donor
fluorophore, and R0 is the Forster distance. R, depends on the spectral overlap, the

reflective index of the medium and the relative orientations of the donor and acceptor

fluorophores. Quenching may occur whenever fluorophores are within the Forster

distance.

Previous research using conjugated BSA-FITC (33) has shown that there are no
concentration quenching effects at a labeling ratio of 1.7 . However, no preparative
chromatography was performed on the protein conjugates used in that study. Therefore,
the reported labeling ratio of 1.7 is actually an average molar ratio of bound FITC to BSA
In reality, the sample could potentially contain BSA with higher degrees of labeling, for
instance BSA-3FITC or BSA-4FIT C. It is our hypothesis that the concentration
quenching artifacts suffered at average labeling ratios greater than 1.7 are due to BSA
molecules in the sample that have higher labeling ratios. We speculate that, after
preparative chromatography, a sample with a homogenous labeling ratio of 2 will not be

subject to concentration quenching. This proposition was validated, as shown later in this

chapter.

3.4 Materials and methods

3. 4.1 Flowchart of strategy
The flowchart of the step by step procedure used to determine the effect of the extrinsic

label is shown in Figure 3.1. The individual steps are discussed in detail later in this
section.

The protein labeling reaction was carried out in 0.1 M, pH 9.2, borax bufl‘er. An alkaline
pH was required to ensure maximum conjugation of FITC to BSA. It was then necessary

to separate out fractions of BSA-1FITC and BSA-2FITC from the polydispersed labeled

45

protein sample. This was performed using an ion-exchange chromatography column that
was operated under isocratic elution conditions. Optimum operating pH for an anion
exchange column is within 1 pH unit of the isoelectric point of the protein conjugate being
fractionated. For BSA-FITC, the isoelectric point was measured at a pH of 4.6, hence the
column was pre-equilibrated with a 0.01 M, pH 5.1 acetate buffer. To maintain isocratic
elution conditions, it was necessary to reduce the pH of the protein solution to 5.1. This
was accomplished in a two-stage stirred dialysis process. In the first stage the 0.1 M, pH
9.2, borax bufl‘er was completely replaced by a 0.1 M, pH 5.1, acetate buffer. In the
second stage the concentrated acetate bufi‘er was exchanged with the 0.01 M, pH 5.1

acetate buffer, which was also used to pre-equilibrate the column.

The protein sample was then loaded onto the column and the elutant collected. The elutant
from the ion-exchange column was analyzed using absorption versus concentration
profiles developed for BSA and FITC in pH 5.1 acetate bufi‘er, using a spectrophotometer.
The absorbance values at 490 nm and 280 nm are used to determine the concentration of
protein and the labeling ratio. However, at pH 5.1, the FITC label exists as a monoanion
(pk. = 6.3) which is four times less fluorescent than the dianion. The higher quantum yield
of the fluorophore dianion makes it the preferred ionic state in subsequent TIRF/PPR
experiments. The elutant was therefore dialyzed against a 0.03 M, pH = 7.4 PBS solution
in order to completely replace the acetate buffer. While this process of dialysis does not
affect the label ratio of the protein sample, it could alter the protein concentration.
Therefore, the protein concentration was then measured again, using an absorbance versus

protein concentration calibration curve developed in pH 7.4, PBS buffer.

46

 

       

     

 
   

. are
«was:

   

 

 

 

Figure 3.1. Flowchart of the experimental strategy used to detemrine the effect of

extrinsic labeling on protein dynamics at and near the interface.

47

TIRF/FPR experiments were then performed to obtain the fluorescence recovery data
after spot photobleaching of the interface. The data were then analyzed and fitted to the
theoretical models, using non-linear regression algorithms, and the diflirsion and

adsorption parameters were extracted for protein samples with different labeling ratios.

3.4.2 Total internal reflection fluorescence (T IRE)
The total internal reflection technique used in this study is described in detail in Chapter 2

of this dissertation. The same equipment was used with minor experimental changes that

are discussed later in this chapter.

3.4.3 Fluorescence photobleaching recovery (FPR)
After equilibrium adsorption of labeled proteins to the interface, a small region of the

interface was exposed to a brief, intense flash of laser light in the total internal reflection
geometry. Under ideal conditions, this results in the photobleaching of all the labels within
the two-dimensional surface in the field of the evanescent wave. Photobleaching is the
dynamic process in which fluorophore molecules undergo photo-induced chemical
destruction and thus lose their ability to fluoresce. Several techniques based on this
photobleaching phenomenon have been developed and successfully applied. The most
widely used technique is fluorescence photobleaching recovery (FPR), first developed by

Peters et al. (99).

Afier the photobleaching flash, the fluorescence intensity recovers towards the pre-bleach

level by exchange of bleached interfacial species with unbleached fluorescent species in the

43

bulk. In FPR, this recovery is observed using a much-attenuated monitoring beam.
Transport coemcients can be determined by measuring the rate of fluorescence recovery.
Depending on the experimental conditions used, the recovery profile can be used to
determine bulk diffusion and/or lateral diffusion of proteins, as well as the kinetics of
adsorption and/or desorption. In this study, we have used FPR to determine bulk

difi‘usion parameters and adsorption kinetics.

3. 4. 4 Protein labeling reaction
Bovine serum albumin (BSA), essentially fatty acid free, with greater than 97% albumin

(A-751 1) was obtained from Sigma Chemical Co. (St. Louis, MO) and used without
further preparation. The proteins were labeled with fluorescein-S-isothiocyanate (FITC, F -
1907, Molecular Probes, Eugene, OR) in our laboratory. The reaction was done in 0.10
M, pH 9.2 borax buffer for up to 6 hours at room temperature in the dark (27). Two
different molar ratios of FITC to BSA were used to obtain the BSA-1FITC and BSA-
2FITC samples. For BSA-1FITC, a molar ratio of 1.3:1 was used; for BSA-ZFITC, a
molar ratio of 3:1 was used. The average labeling ratio in the former sample was

approximately 1 and in the latter was approximately 2.

The unreacted labels were removed by a two-stage dialysis process performed over a
period of 48 hours. This length of time was found to be suficient to completely remove
any unreacted FITC. The protein solution (pH 9.2) was dialyzed with continuous stirring
against a pH 5.1 acetate buffer, using a molecularporous regenerated cellulose dialysis

membrane (Spectra/Por 1, molecular weight cutofi‘ = 6,000, The Spectrum Companies,

49

Gardena, CA). Absorbance spectroscopy measurements with a diode array
spectrophotometer (8452 A, Hewlett Packard, Brielle, NJ) at 280 nm and 500 nm were
used to determine the protein and label concentrations, respectively. This is discussed in
more detail in Chapter 4 and the absorption calibration curves for BSA and FITC are

presented in appendix D.

3. 4. 5 Ion exchange chromatography
Ion exchange is probably the most fi'equently used chromatographic technique for the

separation and purification of proteins. We selected this method because of its simplicity
and high resolving power. Separation by ion exchange chromatography depends on the
reversible adsorption of charged solute molecules to immobilized ion exchange groups of
opposite charge. In this study, we exploited the anticipated difl‘erences in
electronegativity created by the sequential addition of an extrinsic fluorophore to obtain

separate fi'actions of unlabeled BSA, BSA-1FITC and BSA-ZFITC.

Ion exchange chromatography was performed using DEAE sepharose CL-6B gel (17-
0710-01, Pharmacia Biotech AB, Uppsala, Sweden). DEAE sepharose CL-6B, a
macroporous bead-formed (mean particle size of 90 um diameter) ion exchanger, is
derived fiom the attachment of DEAE groups to a cross-linked agarose gel. The gel was
packed into a SR 25/45 column fitted with SRA 25 flow adapters (19-0880-01 and 19-
0867-01, Amersham Pharmacia Biotech, NJ). The DEAE column is flexible and can be
run as an anion exchange or a cation exchange, depending on the working pH. In this

study, the ion exchange chromatography was performed under isocratic conditions.

50

3.4. 6 Determination of starting conditions: pH and molarity
The working buffer pH should be chosen such that substances to be bound to the ion

exchange gel are charged. The isoelectric point (pl) of BSA, determined from several
sources in the literature, was approximately 4.5. This means that, above a pH of 4.5, BSA
is negatively charged; below this pH, BSA is positively charged. For an anion exchange,
the starting pH should be at least 1 pH unit above the pI of the protein sample. Similarly,
the pH should be at least one pH unit below the pI, for a cation exchange. We chose an
anion exchange to protect the BSA-FTIC conjugation bond which might be endangered at

a low, acidic pH.

To determine the pI for labeled BSA, the approach outlined in Amersham (100) was used.
1.5 ml of the gel was added to a series often 15 ml test tubes. The gel in each tube was
equilibrated to a difl‘erent pH by washing 10 times with 10 ml of 0.5 M of the selected
starting acetate bufi‘er (101). A pH range of 5 to 9 was chosen, with intervals of 0.5 pH
units. The gel in each tube was then equilibrated to a lower ionic strength of 0.01 M, by
washing 5 times with 10 ml of the buffer at the same pH but lower ionic strength. A
known amount of sample was added to each tube and the contents were shaken for 10
minutes. After the gel settled, the supernatant was visually inspected and then assayed
using absorption spectroscopy. A pH of 5.1 was selected. This would allow the protein to
be bound to the gel, and still be sufficiently close to the isoelectric (or release) point. Ifthe
pH were too high (for example, more than 1.5 pH units away fi'om the isoelectric point),
the protein would be firmly bound to the gel, resulting in a difficult elution and

necessitating the use of high salt concentrations.

51

A similar experiment was conducted to determine the molarity of the elution bufl‘er. After
equilibrium with bufl‘er at the selected beginning pH, the gel samples were further
equilibrated with solutions of varying ionic strength of the salt (N aCl) at a constant pH. A
known amount of protein sample was added and the tubes were stirred for 10 minutes.
The supernatant was then assayed to determine the highest ionic strength at which binding

occurs and the lowest ionic strength at which elution occurs.

3. 4. 7 Elution of the protein

Elution of the protein was performed by increasing the ionic strength or lowering the pH
of the elution buffer toward the isoelectric point. We used SmM, lOmM, 15mM and 20
mM NaCl solutions to perform a stepwise elution procedure. Since the buffer used in the
column is also an anionic buffer, care must be taken to ensure that equilibrium is attained
in all stages of loading and elution. The column was connected to a double-Syringe pump
system (Model 551382, Harvard Apparatus, South Natick, MA) which served to
introduce protein solutions and elution bufl‘er via a T-valve. Once the elution was started,
2 ml samples of elutant were collected and analyzed for concentration and labeling ratio by
absorption spectroscopy. Samples that fell within an acceptable range of labeling ratio and
protein concentrations were pooled together to make up the final protein solutions of
BSA-1FITC and BSA-2FITC. For example, samples with labeling ratios ranging from

0.98 to 1.02 were pooled together to make up BSA-1FITC.

52

3. 4. 8 Buffer exchange

The elutant from the chromatography column was at pH 5.1. At this pH, however, the
FITC label exists as a monoanion (pk. = 6.3) which is four times less fluorescent than the
dianion (102) (please see appendix B for firrther details). It was thus necessary to perform
a buffer exchange to replace the acetate bufi‘er (pH = 5.1) with PBS buffer (pH = 7 .4).
This was performed by a two-stage, 36-hour, continuously stirred dialysis operation using
a molecularporous regenerated cellulose dialysis membrane (Spectra/Por 1, molecular
weight cutofl‘ = 6,000, The Spectrum Companies, Gardena, CA). The completion of the
buffer transfer was checked with a pH meter. Also, a spectroscopic check of the dialysis
bath revealed no trace of FITC, indicating that the protein conjugation bond was not

affected by the reduction in pH and subsequent ion-exchange chromatographic process.

3. 4.9 TIRFM experiments

Earlier studies of the timescale of the adsorption of BSA-FITC (see chapter 2 of this
dissertation) showed that BSA-FITC took minutes to equilibrate at the oil-water interface
in our experimental cell. Typically the protein sample was introduced into the
experimental cell and, alter equilibration, the monitoring beam of 5 11W was switched on.
Approximately 10 secs later, the shutter system was triggered. This caused a ~100 msec
photobleaching flash of 15mW at the interface. All power readings were taken with a
Newport detector (818-SL, Newport Corporation, Irvine, CA) and measured with a
Newport power meter (1815-c, Newport Corporation, Irvine, CA). The shutter system
closes and protects the PMT for the duration of the bleach. Immediately after the bleach,

the PMT was opened to record the fluorescence recovery. The microscope stage was then

53

slowly moved via the modular automation control from the computer, to allow another
location on the interface to be photobleached and studied. After three difl‘erent locations
were studied, the cell was flushed for 20 minutes with protein solution at the rate of
0.1ml/min. Since the volume of the experimental cell was 960 uL, this completely replaces

the bulk solution.

A detailed description of the computer software used to collect data, the RS-232 interface
with the MAC 2000 (modular automation controller), the sample rate and averaging
techniques of data collection, as well as the photomultiplier tube (PMT) settings can be

found in appendix E.

3.5 Data analyses
The data were analyzed with the PROC NLIN (N onLINear regression) program in SAS.

Two iterative algorithms, the modified Gauss-Newton method and the Marquardt method,
were used. Both methods regress the residuals onto the partial derivatives of the model
with respect to the parameters until the estimates converge. The complete SAS code

written for the two algorithms is given in appendix C.

3. 5. 1 Processes contributing to fluorescence recovery.

Following photobleaching, four processes contribute to the redistribution of bleached and

unbleached fluorophores and combine to affect the speed of recovery and the shape of the

recovery curve (14):

54

1. Difiirsion-controlled motion of free bulk molecules within the ‘cylinder’ of the
evanescent wave;

2. bulk diflusion-controlled refilling of the depletion layer of fluorescent molecules built
up during the bleaching pulse;

3. exchange of surface-bound bleached molecules with the unbleached bulk molecules, as
governed by the adsorption and desorption rate constants; and

4. lateral diffusion of surface-bound molecules into the monitoring beam.

In addition to these, one might theoretically consider rotational diffusion of the
fluorophores, orientation and conformational changes of the irreversibly adsorbed proteins
and possibly spontaneous recovery of photobleaching (dark recovery). However, these

processes do not affect measurements in our system, as discussed below.

Previous studies (103, 104) have shown that for FPR experiments involving one non-
interacting protein species, rotational and translational contributions to the fluorescence
recovery are naturally separated, because they occur on different time scales. However, it
is possible, for a sufliciently short photobleaching pulse, that only those fluorophores with
transition moments parallel to the electric field of the excitation beam will be bleached.
Then a fluorescence recovery would be observed as the rotational diffusion of unbleached
fluorophores cause their transition moments to align with the electric field vector of the
monitoring beam and emit fluorescence. This recovery is independent of the translational
motion of the protein and would compromise the validity of our theory. However,

rotational diffusion as measured by time-resolved phosphorescence anisotropy decay

55

(Edidin 1987) takes place over very short timescales (IO-100 11 seconds). Typical bleach
pulse durations in our experiments are 100 ms; therefore, we assume that fluorophores are

bleached isotropically.

This ensures that rotational diffusion effects do not compromise the qualitative nature of
the crucial early stage of our fluorescence recovery curves. Additionally, steady-state TIR
fluorescence polarization (3 6) and polarized fluorescence recovery after photobleaching
(PFRAP) studies (105) have shown that the rotational diffusion of interfacially adsorbed
molecules is restricted. Since our apparatus is designed to be extremely interfacially
sensitive, we believe that this renders the effects of rotational difl‘usion negligible in our

studies.

Orientational and conformational changes in an adsorbed protein layer are due to changes
in the tertiary structure of the protein, which typically occur over a period of several
hours. This has been seen in interfacial tension experiments conducted in our laboratory
(106) by a slow gradual decrease of interfacial tension over a period of several hours,
following a quick initial drop. However, the entire timescale of our adsorption
experiments in this study is only 60 seconds. Hence the effects of orientational and

conformational changes in the adsorbed BSA layer can be ignored.

Stout and Axlerod (107) first identified possible complications arising fi'om dark recovery
or reversible photobleaching. However, the study also showed that, for a non-

deoxygenated, air-equilibrated system, photobleaching is totally irreversible. For this

56

study, we did not deoxygenate our protein samples. It could potentially be argued that this
could result in artifacts caused by photochemical crosslinking of proteins at the interface.
However, this is a slow phenomenon that is seen after equilibrium adsorption of the
protein. Over the timescale of our experiments, reversible photobleaching is a more

potentially significant factor.

Processes (2) and (4) are essentially absent in our experiments. The size of the
photobleaching beam in relation to the monitoring beam (3: 1) prevents any contributions
from lateral difl‘usion over the time scale of our fluorescence recovery curve. Jauhari
(108) calculated an approximate time of 900 seconds for the lateral diffusion of
unbleached proteins into the monitoring beam. Similarly, due to the relatively large bulk
diffusion coefficient of BSA (calculated in an earlier experiment to be 10’7 cmz/s), the bulk
exchange takes place too quickly (2 ms) to be captured during the timescale of our
experiments. Thus, the fluorescence recovery depends mainly on the bulk difl‘usion of
desorbed molecules away from the interface, and the kinetics of the exchange between

surface-adsorbed and bulk free proteins.

At lower concentrations when bulk diffusion is slow, reversibly adsorbed bleached
molecules diffuse very slowly from the vicinity of the interface. Sometimes they even
readsorb onto the interface, thereby slowing the recovery. In this case, the rate of
recovery, governed by the slowest step, depends on the bulk diffusion coefficient and not
on the adsorption reaction kinetics. This is the bulk-diffusion regime, and for BSA in our

experimental set-up, it has been shown to exist at concentrations less than 0.015mM (1

57

mg/ml) (87). At higher protein concentrations, the bleached solute rapidly disperses into
the bulk after desorption, and the probability of readsorption is very low. In this reaction-
limited regime, the region close to the interface has a high concentration of unbleached
protein molecules and serves as an infinite reservoir to exchange with desorbing proteins.

Here, the desorption rate constant, kD, controls the recovery of fluorescence.

Experiments were conducted in both diffusion-limited and reaction-limited regimes. They

provide information about the efi‘ect of conjugation on diffusion and adsorption kinetics,

respectively.

3. 5.2 The diffusion-limited regime
The equation for analysis of bulk diffusion was modified to account for possible
photobleaching of fluorescent moieties by the monitoring beam (87). The resulting

equation is

K
2[1+ 1]
T

where F is the pre-bleach fluorescence and r is the characteristic diflirsion time for spot

F(t) = F.e("‘”")l l—

 

[16]

photobleaching with a Gaussian laser beam:

r=-—- [17]

58

Here, w is the half—width at e'2 of the height of the Gaussian intensity profile of the laser
beam. Since this is difiicult to estimate in practice, and the non-ideality of the beam leads

to firrther complications, absolute values of D are not reported.

Monitoring beam photobleaching is essentially negligible in our current set-up, which is an
improved version of the one used by Shukla (87) for diflirsion-limited experiments.

Further evidence of this may be discerned from the low values obtained for kpg.

The mobile fraction, 1: defined as the fiaction of proteins reversibly adsorbed, is given by

_Fo-F(°°)
Fo-i:

f [18]

where F (00) is the asymptote of the fluorescence recovery curve and F0 is the post-bleach

fluorescence.

It must be noted that the term ‘reversibly adsorbed’ is always subject to both the timescale
and environment of the experiment. In this case, the proteins are reversibly adsorbed over

the timescale of our experiments (60 seconds) at pH 7 .4.

The prebleach value, F , is determined directly from the data by inspection. The
parameters, r, K and kpB, are obtained by curve fitting, after initial guesses are provided to
the non-linear regression program. The iterative process fits the recovery data to Eqn. [16]

and returns the fitted parameters.

59

3. 5.3 Reaction-limited regime

A visual inspection of the fluorescence recovery curves of BSA-FITC in the reaction-
limited regime reveals three qualitative features: a fast initial recovery, followed by a
slower recovery manifested by a less steep upward slope, and a prolonged period of
fluorescence recovery towards the prebleach level (Figure 4.5). To characterize the
recovery curve quantitatively, Burghardt and Axelrod (4) approximated the recovery as a

sum of exponentials.

kdl

g(t) : ro + rle- ' + r224” [19]

In an attempt to ascribe a physical significance to the fitted parameters, they classified the
adsorbing proteins into two classes, a slowly desorbing fraction (r, at a rate kg), and a

more rapidly desorbing fraction (r; at a rate kdz).

Later studies of proteins at various interfaces have confirmed that fluorescence recovery
curves in the reaction-limited regime are best described by two exponential desorption
terms (21, 17). Further, an additional term to account for unintended photobleaching of

fluorophores by the monitoring beam has been suggested by Shukla (87)

The final modeling equation, using G (t) = F — F (t) is then
F(-) —- Fa) = [F(-) — F(0)e"‘»'](ro +r'le"""‘ +r2e"‘“') [20]

The mobile fraction, f, is given by

f=l—ro=r3+r2 [21]

60

The prebleach value is determined directly from the data. F (0), r0, r1, r2, lid], [‘42, and k“,

are obtained by non-linear curve-fitting, after initial guesses are provided to the PROC

NLIN program.

3.6 Results and discussion

3. 6. 1 The diffusion limited regime

Table 1 gives final values for the time constant (I), the mobile fraction (f), the
photobleaching rate constant (19%) and the bleaching parameter (K) for labeled BSA at a
concentration of 0.01 mM (0.7 mg/ml). We obtained the values by non-linear regression
(Gauss-Newton method) of the data to Eqn.[l6] using the Statistical Analysis Software
(SAS). We report the average and standard deviation of 10 runs performed in the dark at

room temperature (20 °C) for protein samples with L = 1, L = 2 and L ~ 1.7.

Figure 3.2a shows the dependence of r, the characteristic diffusion time, on labeling ratio
measured in the diffusion-limited regime. The characteristic diffusion time for spot
photobleaching with a Gaussian beam, r, for BSA-1FITC was nearly 60 seconds longer
than for BSA-2FITC. Based on Eqn.[l6], which was used to calculate the diffusion
coefficient, this translates into an apparent diffusion coefficient for BSA-2FITC that is
41% greater than that calculated for BSA-1FITC. The primary driving force for the
diffusion of a protein in an aqueous solution toward an interface is the hydrophobicity of
the molecule. Upon conjugation with BSA, the FITC molecule attacks the highly polar

(and hence hydrophilic) side chain of the lysine group. This results in a loss of lysine

61

groups and the acquisition of carboxyl groups, thus increasing the overall negative charge
of the protein. The characteristic diffusion time obtained for BSA-mFIT C, with an
average labeling ratio of 1.7, was approximately between the value for BSA-1FITC and

BSA-2FITC.

It has been previously reported (91) that the formation of an e—fluorescein thiocarbamyl-
lysyl bond lowers the isoelectric point of the protein, rendering it more electronegative at a
given pH. The net effect of the loss of the hydrophilic lysine groups and hence the increase
in the electronegativity of the protein is an increase in the hydrophobic nature of the
protein surface. There is very good correlation between the Nakai surface hydrophobicity
(a measure of protein hydrophobicity introduced by Nakai et al. (109)) and the surface
activity of a protein (53). We believe that the increased hydrophobicity of BSA-2FIT C
leads to greater surface activity, which results in a higher apparent diffusion coeflicient.
Note that the diffusion coefficient measured here is an apparent diffusion coeflicient which
reflects a combination of bulk diffusion and lateral mobility of proteins associated with
(but not adsorbed at) the interface. The increased surface hydrophobicity results in a .
greater driving force for proteins seeking to remove themselves from the aqueous bulk
environment (by adsorption at the interface), thus resulting in a faster diffusion rate. It
may be argued that an increase in surface activity should result in slower exchange
kinetics, due to larger residence times for proteins at the interface. This should slow the
rate of recovery of fluorescence and reduce the apparent diffusion coemcient. However,

in the diffirsion-lirrrited regime, the rate-limiting step is the diffusive process, and thus

62

increased surface hydrophobicity is manifested in this step, resulting in a higher apparent

diffusion coemcient.

The fraction of mobile proteins f, does not show a dependence on labeling ratio within
experimental error (Figure 3 .2b). During equilibrium adsorption at lower concentrations,
there is sufiicient time for the protein molecules to unfold and fully exploit the
hydrophobic residues buried within the globular structure of the native protein. This is
because, at lower concentrations, during the finite time required for unfolding of a protein,
it is not affected by steric hindrances from the adsorption of another protein molecule. In
this situation, the increased surface hydrophobicity caused (by the presence of extrinsic
fluorophores has a negligible effect on the total surface activity that governs the degree of
irreversibility of protein adsorption. Therefore, the measured mobile fi'actions 1, do not

show a dependence on the labeling ratio.

In our apparatus, it is difficult to precisely determine w, the half-width at e'2 of the height
of the Gaussian intensity profile of the monitoring beam. This makes it difficult to
determine the absolute diffusion coefficient; hence we do not report absolute values in this
study. However, using rough estimates of w based on Jauhari (108), we calculated
diflirsion coefficients comparable to values reported in the literature (on the order of 10"
cm2/s) for BSA-FITC under similar conditions. The apparent diffusion coemcients
calculated in this study should not be confiised with bulk diffusion coeficients. Physically,

the apparent diffusion coefficient reported here can be thought of as a parameter that

63

characterizes a combination of bulk diflirsion and lateral mobility of proteins near or

loosely associated with the interface.

To ensure that the two sets of experiments were comparable, we examined the bleaching
parameter K, expressed as shown in Eqn. [22].

K = al (0)T [22]
Physically, K denotes the amount of bleaching induced in time T, and al(0) is the rate
constant of the simple irreversible first order photobleaching reaction at the interface. K
was constant throughout all our experiments, indicating that the bleaching conditions, as

well as the beam profile, were identical for both sets of experiments.

64

 

300

250 i

200 -

150 -

100 -

r (characteristic diffusion time) s

 

 

 

5O
1 2

Labeling ratio (FITC:BSA)

 

b- 0.8
0.7 i
0.6 <
0.5 l
0.4 ~

0.3 « {/i

0.2 ~

 

HH

f (mobile fraction)

0.1 .

 

 

 

0.0
1 2

Labeling ratio (FITC:BSA)

Figure 3.2. Dependence of a) r (characteristic time of diffiision) and b) f (mobile fraction)
on labeling ratio in the difl‘usion—limited regime for BSA at the oil-water interface. The
bulk protein concentration was 0.01 mM.

65

 

Table 1. Summary of results for BSA-FITC in the difl‘usion—limited regime. The bulk
protein concentration was 0.01 mM

 

 

 

 

 

Protein Conc. = 0.01 mM L = l L ~ 1.7 L = 2

r(s) 204.8 :t 39.4 156.1 1 20.3 145.0 i 28.1
Time constant
f 0.30 i 0.03 0.30 :l: 0.04 0.36 :l: 0.06
Mobile fiaction

kpa (s") 0.0013 i 0.0002 0.0015 :1: 0.0002 0.0013 :1: 0.0003
Rate of photobleaching
K 0.69 i 0.08 0.65 i 0.08 0.70 i 0.14
Bleaching parameter

 

 

 

 

 

66

3. 6.2 The reaction-limited regime

Table 2 shows a summary of the results of experiments conducted in the reaction-limited
regime, at a protein concentration of 0.025 mM (1.7 mg/ml). The fluorescence recovery
data were non-linearly regressed (Marquardt method) to fit Eqn. [20] using SAS. Again,
we report the average and standard deviation of 10 runs performed in the dark at room

temperature (20 °C), for L = 1, L = 2 and L ~ 1.7.

Three parallel non-interacting classes of proteins account for the shape of the
fluorescence recovery curves (4): an irreversible protein fraction (r0) called the immobile
fiaction, a slowly reversible fiaction r; with a slow desorption rate of k.” (3") and a rapidly
reversible fraction (r;) with a rapid desorption rate of kg; (3"). Physically, the simplest

model that appears to be consistent with the data is as follows.

The three classes of adsorbed proteins may be thought of as representing different bound
layers. The irreversible bound layer is the layer closest to the oil-water interface and is in
intimate contact with the oil. After this layer come the reversible classes with the slowly
reversible class closest to the adsorbed layer. The rapidly desorbing class constitutes
proteins that are loosely attached to the adsorbed layers in a surface-associated state. A
more realistic model might include several classes of binding proteins that interact with
each other. However, adding adjustable parameters to the non-linear regression algorithms
increases the dependencies between/of the fitted parameters to unacceptable levels. This
decreases the reliability of the regression, and makes it increasingly difficult to attribute

physical significance to the experimental results.

67

 

0.8
0.7 <
0.6 i
0.5 r
0.4 <
0.3 .
0.2 1
0.1 4
0.0

r0 (immobile friction)

 

 

 

1 2
Lebelhg ratio (FITC:BSA)

0.8
0.7 <
0.6 .
0.5 1
0.4 .
0.3 .
0.2 4
0.1 .
0.0

 

r, (slowly reversible fraction)

 

 

 

1 2
Labeling tltlo (F ITCIBSA)

0.4

 

0.3 .

0.2 .

(11.

r,(repidly reversible fraction)

 

 

0.0

 

1 2
Labeling ratio (FITC:BSA)

Figure 3 .3. Dependence of a) r0 (immobile fraction), b) r; (Slowly reversible fraction) and
c) r; (rapidly reversible fraction) on labeling ratio in the reaction-limited regime for BSA at
the oil-water interface. The bulk protein concentration was 0.025 mM.

68

 

O
3

o
S

 

0.01 <

kd, (slow desorption rate) 3"

 

 

 

 

 

 

 

 

 

 

 

 

0.00
1 2
b_ Labeling ratio (FITC:BSA)
0.8
’3’ 0.7
8 0.6 <
i 0.5+
3 04 1
3 .
a 0.3 .
g 0.2
,9 0.1«
0.0
1 2
Labeling ratio (FITC:BSA)
C.
'7 0015 <
0
3
E
,5 0.010 « t
E
i
g 0.005 A
8.
“ 0.000
1 2
Labeling ratio (FITC:BSA)

Figure 3.4. Dependence of a) kd, (rate of slow desorption), b) k4; (rate of rapid
desorption) and c) kph (rate of photobleaching) on labeling ratio in the reaction-limited
regime for BSA at the oil-water interface. The bulk protein concentration was 0.025 mM.

69

3.6.2.1 The immobile fraction r0

Figure 3.3a illustrates the dependence of the immobile fi'action on the labeling ratio. The
immobile fraction increases nearly linearly with labeling ratio. For BSA-2FITC, the
fraction of proteins classified as irreversible is 0.61 i 0.06 which is almost double the
value computed for BSA-1FITC, 0.30 i 0.07. This can be attributed to the increased
hydrophobic nature of BSA-ZFITC, which increases r0, the fraction of irreversibly
adsorbed proteins in two ways. First, since it is more surface-active, BSA-2FITC is more
irreversibly adsorbed at the interface. Secondly, the higher diflirsion coefficient of BSA-
2FITC enables it to saturate the interface at a faster rate. Since the time allowed for
equilibrium adsorption is identical in all experiments, this provides a longer period of time
for the adsorbed protein layer to ‘settle’ into the interface, thus increasing the fraction of

irreversibly adsorbed proteins.

3.6.2.2 The reversible fractions, r, and r,

Figure 3 .3b shows the values of n, the slowly reversible protein fi'action plotted against
the labeling ratio. The graph is fairly linear, with a negative slope. For BSA-1FITC, r, =
0.52 :t 0.06 which is greater than the corresponding value of 0.32 i 0.06 reported for
BSA-ZFITC. This might seem to be contrary to the hypothesis expounded earlier, as it
seems to indicate that BSA-1FITC is more firmly adsorbed. However, if one examines the
rates of desorption, plotted against labeling ratio in Figure 3.4a, one sees that k,” for BSA-
lFITC is 0.022 :1: 0.011 s ’1 which is much faster than the corresponding rate of 0.001 i

0.0005 5'1 reported for BSA-2FITC. In other words, even though there is a larger fraction

70

of BSA-1FITC in the slowly desorbing class, the rate of desorption is much faster than

that of the corresponding BSA-2FITC class.

To get a clearer picture of the efl‘ect of the label, we can consider the ratio r1: r;, a
measure of the distribution of the protein population amongst the rapidly and slowly
desorbing classes. For BSA-1FITC this ratio is approximately 3 and for BSA-2FITC it is
approximately 4. We can thus conclude that a larger percentage of BSA-ZFITC is in the
slowly desorbing class of proteins and, in addition, desorbs at a slower rate. This is in

agreement with the increased hydrophobicity of BSA-2FIT C discussed earlier.

Figure 3.4b shows the dependence of lid; on labeling ratio. We observed that BSA-2FITC
desorbs with a faster kdz (the rapid desorption rate constant) than BSA-1FITC, 0.48 :I:
0.21 compared to 0.23 i 0.11. This is presumably due to the fact that, since this class of
proteins is loosely associated with the interface, lateral diflirsion plays a more important
role in the desorption process. The higher rate constant of BSA-2FITC then simply
reflects the higher diffusion coefficient, which we measured in the difl’usion-limited regime

(discussed earlier).

The dependence of the photobleaching rate constant kpb is shown in Figure 3.4c. The
salient feature of the graph is that the rate constant of photobleaching is much larger at a
labeling ratio of 1. The random statistical probability of photobleaching a fluorophore has
been estimated between 1:1000 and 1:5000 (110). In other words, a fluorophore will go

through the excitation-emission cycle between 1000 and 5000 times before being

71

Table 2. Summary of results for BSA-FITC in the reaction-limited regime. The bulk
protein concentration was 0.025 mM

 

 

 

 

 

 

 

 

 

 

 

Protein Cone. = 0.025 mM L = l L ~ 1.7 L = 2

ro 0.30 i 0.07 0.48 :l: 0.11 0.61 :t 0.06
Immobile fraction

n 0.52 i 0.06 0.43 :I: 0.095 0.32 i 0.06
Slowly reversible fraction

r2 0.18 i 0.10 0.096 1: 0.024 0.08 i 0.02
Rapidly reversible fi'action

in (s") 0.022 :t 0.011 0.0048 1: 0.0012 0.001 i 0.0005
Rate of slow desorption

kdz (s") 0.23 i 0.11 0.30 :i: 0.08 0.48 :i: 0.21
Rate of rapid desorption

kpa (8") 0.01 i 0.005 0.001 :I: 0.0003 0.001 :I: 0.0005
Rate of photobleaching

 

72

 

photochemically destroyed. At a lower labeling ratio of 1, there are fewer fluorophores at
the interface that are undergoing the excitation-emission cycle for the same laser intensity.
This results in a higher rate of destruction of these fluorophores since they go through the

excitation-emission cycles more often.

Concentration quenching can be detected by observing the qualitative shape of the
fluorescence signal during the adsorption process. Concentration quenching is present
when the fluorescence signal passes through a maximum and then decreases, as the
concentration of labeled proteins at the interface increases, as shown in the theoretical
curve in Figure 3 .5. This overshoot in the fluorescence is due to the competition between
an increase in the number of fluorophores at the interface with adsorption and the decrease
in the quantum yield of these fluorophores as the average interfluorophore separation falls
within the concentration quenching range (within the Forster distance). Larger labeling

ratios therefore produce more dramatic signal maxima that occur at earlier times (33).

We conducted checks for concentration quenching in a sample of 0.025 mM protein with
a monodisperse labeling ratio of 2. The data, shown in Figure 3.5, clearly do not portray a
maximum in the TIRF signal. This proves our earlier hypothesis that preparative
chromatography eliminates concentration quenching by removing the higher labeled
proteins from the sample. Concentration quenching can also be manifested in the early
phase of fluorescence photobleaching recovery. Again, the effect of concentration

quenching is to produce an overshoot in the recovery.

73

Figure 3.7a shows a typical recovery curve for 0.025 mM protein concentration at a
labeling ratio of 2. One can see a smooth increase in post-bleach fluorescence, typical of a
system free from the efl‘ects of concentration quenching. Figure 3 .7 b, shows a recovery
curve for 0.025 mM protein concentration with an average labeling ratio of 2.5. This
sample was prepared without preparative chromatography. The plot shows the distortion

of the initial stages of the recovery curve, due to the effects of concentration quenching.

74

1200

 

 

 

Fluorescence (AU)

 

 

 

 

 

we I I I I I l I
0 100 200 300 400 500 600 700
Time (sec)
Actual recovery curve, L I 2, C I 0.025mM
- - Hypothetical recovery curve shoMmflects of concentration quenching

 

 

Figure 3.5. Check for concentration quenching effects in the equilibrium adsorption of
L = 2, 0.025 mM BSA at the oil-water interface.

75

 

 

 

 

 

 

Whoa htensly

Mm]

 

 

 

 

 

9 9.5 10 10.5 11 11.5 12
tin-Isl

 

 

 

 

 

 

 

 

 

 

 

 

 

7m 1 T I T I
12 13 14 15 16 17 1a
tin-Isl

 

 

Figure 3.6 a. Normal (quenching free) fluorescence recovery of L = 2, 0.025 mM
BSA after preparative chromatography. b. Fluorescence recovery of L ~ 2.5. 0.025 mM
BSA showing the effects of concentration quenching

76

3.7 Conclusions

Preparative (ion exchange) chromatography was used to separate BSA conjugated with
FITC into two samples with labeling ratios (L = number of FITC molecules attached to a
single BSA molecule) of l and 2. Experiments were conducted by total internal reflection
fluorescence microscopy (TIRFM) along with fluorescence photobleaching recovery
(FPR), using proteins labeled with l, 2 and approximately 1.7 FITC molecules. The
diffusion and adsorption kinetics of BSA-FITC were investigated under two different

concentration regimes at each labeling ratio.

In the diffusion-limited regime (protein concentration of 0.01 mM or 0.7 mg/ml), BSA-
2FITC has a difl‘usion coeflicient that is 40% larger than BSA-1FITC. We attributed this
to an increase in surface hydrophobicity caused by the presence of the additional
fluorophore molecule. If we assume a simple linear relationship for the efl‘ect of the
fluorophore on the surface hydrophobicity of the macromolecule, this would imply that
previous studies of this nature have overestimated the lateral mobility of labeled proteins

by up to 40%.

In the reaction-limited regime (protein concentration of 0.025 mM or 1.7 mg/ml), the
additional label led to an increase in surface hydrophobicity and hence a more irreversible

adsorption of BSA-2FITC. The result was that BSA-2FITC has:

1. An immobile fraction of 0.6 i 0.06 which is larger than the r0 for BSA-1FITC,

which was 0.30 i 0.07

77

2. A smaller rate constant k.“ of 0.001 t 0.0005 s‘1 for the slowly desorbing
proteins than for BSA-1FITC for which we measured a k.“ of 0.022 i 0.0011
s"; and

3. a higher rzzr, ratio (the ratio of the slowly desorbing protein class to the rapidly

desorbing protein class) of 4:1 compared to a similar ratio of 2.9:] for BSA-

1FITC.

We have shown that the attachment of an extrinsic label does have a pronounced effect on
the adsorption properties and the diffusion of proteins near the interface. Intuitively, one
would expect that the addition of a planar fluorescein label to a globular protein (BSA)
would increase the characteristic timescale of diffusion, due to viscous drag. This is most
likely true for pure bulk diffusion of BSA in solution. However, for diffusion measured
near the interface in these studies, our experiments show that the addition of the label

significantly decreased the timescale of the diffusion process.

Experiments performed in the reaction limited concentration regime conclusively prove
that the adsorption kinetics of proteins are affected by the presence of the fluorophore
label. Due to an increased surface hydrophobicity, the protein-label conjugate is more
surface-active and tends to adsorb more irreversibly than the native protein. This leads to
a slower desorption rate for labeled proteins and hence efi‘ectively a preferential

adsorption of labeled proteins over unlabeled proteins.

78

Based on the results presented in this study, we recommend that the covalent labeling of
proteins to quantitatively investigate interfacial phenomena be undertaken with caution.
Preparative chromatography can be used to separate proteins into fractions with difi’erent
monodisperse labeling ratios. Experiments can be performed on protein samples with
difl‘erent labeling ratios to quantify the effect of the label unique to the characteristic
property being investigated. The results can then be carefirlly extrapolated to determine

the properties of the native unlabeled protein.

79

4. A CALIBRATION PROTOCOL FOR QUANTITATION
OF INTERFACIAL CONCENTRATIONS OF
MACROMOLECULES AT THE LIQUID-LIQUID
INTERFACE.

4.1 Introduction

We have successfully used total internal reflection fluorescence microscopy to study
protein adsorption at the liquid-liquid interface (86, 111, 87). So far in these studies, it
has not been necessary to precisely calculate the interfacial concentrations of proteins.
However, in studying many other adsorption phenomena, it is important to know the
concentrations of macromolecules at the interface. In these types of situations, it would
be particularly beneficial to be able to directly calculate the concentration of adsorbed

interfacial species from fluorescence emission intensity data.

In general, devising a reliable calibration technique to enable quantification of interfacial
amphiphilic concentrations on the basis of fluorescence emission intensity is a very difficult
problem at the liquid-liquid interface, at least in comparison to the solid-liquid interface.
Techniques that have been routinely used to determine interfacial concentrations at the
solid-liquid interface have been made possible by the fact that a) protein adsorption is
essentially irreversible at this interface, b) the interface is very stable, and c) the interface
can be easily disassembled to enable adsorption on the solid surface to be quantified.
Common techniques include assessment of depletion of adsorbates in the bulk phase (4),

use of non-adsorbing species (such as polystyrene spheres) as standards, use of

80

scintillation counting in combination with TIRF (27, 2), and combination of reflectometry

with TIRF (33).

Unfortunately, none of these techniques will work at the liquid-liquid interface, because of
two primary limitations. Even if adsorption at this interface were irreversible, there is no
reliable technique that can be used to drain the bulk solution from the sample cell without
incorporating adsorbed species as well as some of the oil layer into the aqueous efiluent.
Quantitation methods that combine reflectometry or scintillation counting with TIRF rely
on the immobilization of proteins on one of the surfaces that make up the interface. For
example, adsorbed species at the solid-liquid interface can be isolated on the solid surface.
The solid can then be separated from the sample cell and the quantity of adsorbed species
determined reasonably accurately by any one of several methods. Such an operation is
obviously impossible at the liquid-liquid interface. Therefore, a very different approach is

required for this interface.

In this study, we propose a calibration technique that uses fluorescence photobleaching
recovery to determine the interfacial concentration of adsorbed species in situ, without
destabilizing the oil-water interface. In addition, no external measuring devices are

required in this protocol.

81

4.2 Theory

4. 2. 1 The evanescent wave

A general overview of the theory of total internal reflection is described in Chapter 2 of
this dissertation. Here, we present a detailed analysis of the development of the equation
characterizing the evanescent wave. This exponentially decaying wave is the partial
emergence of the electromagnetic field into the rarer medium at the point of total internal
reflection. The success of TIRF as an interfacial tool is due to the unique properties of the

evanescent wave.

A thorough mathematical development of Maxwell’ 3 wave equations under conditions of
total internal reflection has been presented by Reichert et al. (112). They showed that the
perpendicularly polarized component of the transmitted electric field could be expressed
by the following equation

2 2 2 i
E; = jig—0‘s exp —j—;E[fl]sin 61 —A—m{[fi-) sin2 19, 4] [23]

2 2 2 "2

where E0". is the amplitude of the electric field at the interface in the optically rarer

medium 2, n, and n; are the refractive indices of media 1 and 2, respectively, 2.2 is the
wavelength of incident light in medium 2, and 6, is the incident angle of the excitation
beam measured normal to the interface. The variables, x, y, and z, are three mutually
perpendicular coordinates, with the interface lying in the x-y plane. The superscript, 1,

indicates that the equation describes the transmitted electric field.

82

For an incident beam of finite width, Eqn. [23] represents the surface wave that
propagates along the interface as demonstrated by Goos and Hanschen (please see Herrick
(75)). The first term in the equation accounts for the longitudinal displacement (in the y
direction) between the incident and reflected beams at the interface, and is called the Goos
Hanschen shift. The second term accounts for the penetration of the surface wave into the
rarer medium (in the z direction); this is a standing wave of exponentially decaying

intensity, and is referred to as the evanescent wave.

We can relate the second term to the penetration depth, dp, of the evanescent wave as

follows:

1
2 5
-—1— : fig-"4] sin 2 9, - I] [24]
dp 22 712

and write Eqn. [23] as

"'r —t' 2
Es = Es CXP[— 27] [25]
P

where E; = 91,13; S exp[— jZ—E(§‘—) sin 0,]
’12 ”2

The effective depth of penetration, dp, is defined for mathematical convenience to be the
distance 2 required for the electric field to decay to e'1 of its original value. Ifwe note that
71.1, the wavelength of the excitation wave in the medium 1, can be given by 2., = Mm,
where 20 is the free space wavelength of the incident beam, then Eqn. [24] reduces to the
expression for the penetration depth presented in Chapter 2 of this thesis:

83

d ’10 [26]

p 471711 (lsin2 19—n2,2

where n2, = 722 /n1 is the relative refractive index.

Given randomly oriented dipoles on the fluorophores, the probability of absorption of
energy from the evanescent wave is proportional to the intensity; as such, I s IE‘.|2 (62).

Thus, Eqn. [25] reduces to the more familiar expression for the evanescent wave:

1(2) = 1,5“! [27]

It is important to note that 10 in the above equation refers to the incident beam intensity in
the optically rarer medium. Experimentally, it is easier to measure the intensity of the
incident beam in medium 1, the optically denser medium. However, 10, for incident angles
within 10° of the critical angle, can be up to five times larger than the intensity of the
incident wave. This is of particular concern since underestimating 10 could potentially
cause unintended photobleaching of fluorophores by the monitoring beam since, in this
situation, the monitoring beam would be more intense than would be anticipated. The
calculation of lo and its relation to the monitoring beam intensity will be discussed later in

this chapter.

4. 2.2 Interfacial selectivity of T IRF

The major challenge in studying the interfacial adsorption of any macromolecule is the

need to exclude the properties of the bulk molecules fi'om those at the interface. In TIRF,

84

 

120.0

 

100.0 1

80.0 ~-

Evanescent wave intensity (AU)
8

20.0 ~~

0.0 i i i i :

0 50 99 100 150 200 250
Depth into the rarer medium (nm)

 

 

 

 

 

 

 

Figure 4.1. Evanescent wave intensity as a function of depth of penetration into the rarer
medium. The penetration depth, 11,, is 75 nm in our setup, for an interface between oil (n;
= 1.52), and water (n; = 1.33) for an incident light of wavelength A = 488 nm, at an angle
of incidence 0 = 70°.

85

the exponential nature and shallow penetration depth ((1,) of the evanescent wave help to
effectively accomplish this. The (1,, for most TIR systems ranges from 75 to 150 nm. This

allows preferential and selective illumination of a very thin layer closest to the interface.

Figure 4.1 shows a plot of the evanescent wave intensity with distance into the rarer
medium for our system (clp = 75 nm). From the graph, it can be seen that the evanescent
wave intensity drops to about 25% of its initial value within the first 100 nm of the

interface.

In addition, microscope optics allow reasonably precise focussing at the interface, further
contributing to better interfacial selectivity by rejecting out of focus fluorescence from the
bulk. A microscope objective with a high numerical aperture gives a small depth of focus
a} according to the relation (Technical support, Melles Griot, personal communication):

2
d ~ “ 28
’ 4.1VA2 [ ]

 

where 21.... is the wavelength of the light and NA is the numerical aperture of the objective.

The depth of focus refers to a region where all objects are simultaneously in focus,
determined by just meeting the Rayleigh criteria of resolution (which is +/- M4). It
should be noted that, in our system, the light collected by the microscope objective is the
emitted fluorescence for which 21 is 519 nm. With NA = 0.75 and 21.... = 519 nm, Eqn. [28]

gives a depth of focus of approximately 200 nm. Therefore, carefirl focusing of the

86

microscope leads to the rejection of bulk fluorescence emissions more than 200 nm away

from the interface.

Despite these precautions, there is still a sizable contribution of fluorescence from
fluorophores in the bulk in most practical situations. However, by reducing the incident
light intensities in our experiments to 1-5 uW and using an extremely sensitive detection
system, we are able to minimize these unwanted contributions. For an accurate, absolute
quantitation of interfacial proteins, it is necessary to completely account for emissions
from the bulk and subtract this contribution from the total fluorescence signal. This is
accomplished by using a photobleaching pulse to eliminate fluorescence emission fiom
proteins adsorbed at the interface and thus enable the calculation of the bulk fluorescence.

The details and theory are presented below.

4. 2.3 The calibration protocol

The calibration protocol presented here was adapted from Zimmerman et al. (14), who
applied the technique to measure the adsorption of BSA at the quartz/water interface.
The method is self-contained, in that it does not require any other external measuring
techniques such as ellipsometry, y-detection of radiolabeled proteins or spectrofluorimetry.
The method is based on the fact that the total fluorescence detected by the PMT is a

combination of contributions fi'om the bulk and the interface (29):

F. =F... +F... =4, [I(z)C(z)q,(z)dz r291

87

where q, is the overall instrument constant which accounts for the collection efliciency of
the optical system as well as the fi'action of labeled proteins, and q; is the quantum yield of

the fluorophore.

Incorporating the expression for the evanescent wave fi'om Eqn. [27] gives
' is.)
F. =F... +F... =21. I e C(z)q,(z)dz [301
0

where C(z) is the concentration of proteins, and is a function of the z direction

perpendicular to the interface.

Note that the upper limit of the integration is I, the depth of the aqueous phase in our
experimental cell. We assume a simple model of protein adsorption, whereby there is a
layer of adsorbed protein molecules at the interface of thickness 3 and two-dimensional
concentration CS, in equilibrium with a bulk concentration of C3. We can now split up the
definite integral into two terms, the first representing F a , the surface fluorescence, and

the second F M, the bulk fluorescence:
' {is}
FT:Fadr+Fbulk=q110 CquS+CquBJe , dz [31]
S

where (1,3 and qu are quantum efficiencies of the fluorophore in the adsorbed layer and in

the bulk, respectively.

88

In this analysis, we have assumed that (1,3 is equal to qfa (this and other assumptions are
discussed in the next section). We can therefore combine the fluorophore quantum
efficiency with q], the overall instrument constant to get q, the overall system constant.
Note that we have assumed here that total internal reflection occurs on a plane parallel to
the interface and at a depth of z = s. This is a valid assumption for a dense adsorbed layer
of proteins, since the index of refraction of proteins is approximately 1.5 (113), which is

approximately the same as the reflective index of the oil phase in our system.

Based on the shape and dimensions of a BSA molecule, we can make a rough estimate of
the magnitude of s in our system. BSA has been well described hydrodynamically as a
prolate ellipsoid. Wright and Thompson (114) have used transient electric birefiingence to
calculate dimensions of 14.1 nm for the major axis and 4.1 nm for the minor axis. For a
multilayer adsorption model, with at least three different classes of proteins (discussed in
more detail in Chapter 3), this means that s lies somewhere in the range of 10 nm to 50
nm. Therefore, s is clearly much smaller than I (s< <1), the depth of the aqueous phase
(which is approximately 0.9 mm). Therefore, we can simplify the expression for F M, the

bulk fluorescence as follows:

 

1 _<_-_> -(H) ’ -Q;_)
11,, =qC, [ e ‘9 dzqu,[-d,e "’1 ] =qC,[—d,e ”i +d,]~qc,d, [32]

Incorporating Eqn. [32], we can rewrite Eqn. [31] as
Fr :Fad:+Fbulk:qIO(CS+Cde) [33]

where the factor qu is called the instrument constant.

89

Since the bulk concentration is known, we need only to measure its fi'actional contribution
to the total fluorescence to calculate both the unknown instrument constant and the
interfacial concentration. In fact, according to Axelrod et al. (62), one can easily deduce
this fractional contribution by two simple techniques. The first is to abolish the
fluorescence fi'om the interface with a strong photobleaching pulse of incident light, so the
bulk fluorescence emission can be measured. The second is to replace the bulk solution
with a fluorescence-free rinse, so the measured fluorescence will only come fi'om the
interface. Of course, this second technique is only appropriate if adsorption is irreversible

at the interface of interest.

Previous work in our laboratory (87) has demonstrated that protein adsorption at the oil-
water interface is not entirely irreversible. Therefore, the introduction of a fluorescent-
free rinse (by either an equivalent solution of unlabelled proteins or a buffer solution free
of proteins) into the flow cell will affect the determination of the concentration of
adsorbed proteins. In the case of introducing an unlabelled protein rinse, there is the
possibility of exchange of the reversibly adsorbed labeled proteins with bulk unlabeled
proteins, which will reduce the fluorescence intensity detected at the interface. The result
is that the measured concentration will be less than the actual prevailing surface
concentration. On the other hand, the introduction of a buffer solution into the flow cell
will, at the least, change the microenvironment of the fluorescent tag on the protein. This
will induce changes in the quantum yield of the fluorophore (2) that could potentially lead

to erroneous estimates of the interfacial concentration.

90

Fortunately, photobleaching interfacially adsorbed species allows us to circumvent these
potential problems. Using controlled photobleaching experiments, the difl‘erence in
mobilities between surface-adsorbed and free bulk proteins can be exploited to separate
the surface and the bulk components of the fluorescence signal. The proteins in the bulk
are mobile with a diffusion coeflicient on the order of 10'7 cmz/s (115). The time taken for
exchange of unbleached fluorophores with bleached fluorophores in the bulk can be

calculated by the following simple equation:

1 = — [34]

where x is the distance traveled to efi‘ect the exchange, and D is the bulk diffusion

coefficient.

For an evanescent wave penetration depth of 75 run, this corresponds to an average
exchange time on the order of milliseconds. This is less than the time resolution of our
experiments (10-12 ms) and results in an apparent unbleachability of the bulk
fluorophores. In other words, even though the photobleaching beam bleaches proteins in
the bulk, the fast diffusion coeflicient causes the recovery to occur before the detection
system can record it. Experimentally then, F M, can be determined by measuring the
fluorescence intensity after complete bleaching of the adsorbate layer and can be expressed
as

Fbulk : quCde [35]

If we make the reasonable assumption that the concentration of proteins in the bulk is not

affected by adsorption at the interface (the so-called “infinite well” assumption), then C3 is

91

essentially the protein concentration injected into, the experimental cell and is thus known.
Combining Eqn. [33] and Eqn. [35], we can express the concentration of interface-

associated proteins, Cs , as

C, = CBd, ({1— — 1] [36]
bulk

Therefore, Cs can be calculated by measuring F r and F M, for a given bulk concentration

C3.

4. 2. 4 Relation between evanescent intensity (1.) and monitoring beam intensity

The light intensity at the interface (10 in Eqn. [42]) is a function of the angle of incidence
and the polarization of the incident laser beam. The incident light can either be p-polarized
(transverse magnetic field) with the electric field vector parallel to the plane of incidence,
or s-polarized (transverse electric field) with the electric field vector perpendicular to the
plane of incidence. The intensity of the evanescent wave at z = 0, for unit incident

intensities, are given by Harrick (75):

 

(4 cos2 6)(sin2 6?)
I 0 = 37
’( ) n4 cos2 9+sin2151—n2 [ l
2
1,(o)=(4—C95—,‘31 r381

l—n

where 0 is the angle of incidence and n = 1121 2 712 /nl is the relative refiactive index.

For incident angles less than the critical angle (9c = 61° for our system), the intensity of

the transmitted electromagnetic waves can be calculated from the Fresnel equations:

92

(2n cos6) J [39]

n2 cos6+~fn2 - sin2 6

I,(0)=E2,(0)=[

 

I.(0)=E%<0) =( (““6" j 1401

0056+ an - sin2 6

The transmitted light intensities in our system for incidence angles ranging fi'om subcritical
to supercritical are plotted in Figure 4.2 using Eqns. [37]-[40]. Note that the evanescent
intensities near the interface and within 10° of the critical angle are up to five times larger

than the intensities of the incident plane waves.

The incident angle of the laser beam in our system was fixed at 70° by the geometry of the
TIR prism. For p-polarized incident light, this gives a relative intensity of 2.5. In other
words, the evanescent intensity at z = 0 is 2. 5 times greater than the intensity of light
striking the oil-water interface. This is approximately equivalent to the monitoring beam
intensity. However, the monitoring beam intensity is measured outside the experimental
cell, i.e. before the laser beam is incident on the prism. To relate the intensity of the
monitoring beam to the evanescent wave intensity, it is necessary to estimate losses due to

reflection at the prism surface and absorption by the prism and the slide.

In a separate experiment, we used a Newport detector (818-SL, Newport Corporation,
Irvine, CA) and a Newport power meter (1815-c, Newport Corporation, Irvine, CA) to
measure laser light intensities before and after passing through the prism and slide. An

average transmission efficiency of 0.85 was measured which, when coupled with the

93

 

 

5 - ------ l-e polarized
l-p polarized

 

 

 

 

relative intensity
0’

 

 

 

 

 

angle of incidence (degrees)

Figure 4.2. Evanescent intensities LS and I-p for perpendicular and polarized incident
light (A = 488 nm) at the oil-water interface (2 = 0) as a firnction of angle of incidence.
The critical angle, 6c, in our experiments is 61° for oil (71, = 1.52) and water (72; = 1.33).

94

relative intensity factor (2. 5), resulted in an evanescent wave intensity that was a factor of
2.1 times greater than the monitoring beam intensity. Thus, in our system, typical
monitoring beam intensities of 1-5 pW resulted in evanescent wave intensities of 2-10

11W.

4.3 Materials and methods

The experimental apparatus used is discussed in detail in Chapter 2 of this dissertation.
Protein solutions of concentrations ranging fi'om 10" mM to 0.3 mM were injected into
the experimental cell. An FITC:BSA molar labeling ratio of 1:1 was used for all
experiments. Absorption spectroscopy was used to measure protein concentrations and

labeling ratios, as discussed later in this chapter.

4.3.1 T IRFM calibration erqreriments

A description of the experimental used in general TIRFM experiments can be found in
Chapter 2. Here we present the procedures specific to experiments conducted to develop

the calibration protocol.

All calibration experiments were performed in the dark at room temperature, with protein
concentrations ranging from 10’3 mM to 0.15 mM at a monodisperse labeling ratio of 1.
The experimental cell was first flushed with pure PBS buffer at a pH of 7.4, for a period of
25 minutes. This served to remove the air bubbles in the tubing and the experimental cell.

The proteins were then injected into the cell using a reciprocating syringe pump that

95

simultaneously withdrew PBS bufl‘er solution. All flowrates in and out of the cell were
held constant at 0.5 ml/min. At this flowrate, it took approximately 20 minutes to
completely replace the bulk solution in the experimental cell, assuming that the flow
regime in the cell could be modeled as plug-flow. However, to ensure equilibrium, the
proteins were injected into the cell for 45 minutes, after which the syringe pump was
turned off and another 20 minutes was allowed to ensure equilibrium adsorption.

The interface was then bleached for 150 ms with a 15 mW laser beam and the subsequent
drop in fluorescence was recorded by the PMT. The microscope stage, on which the
experimental cell rests, was then slowly moved to a new position. The exact destination,
coordinates and speed of motion were controlled from the computer via the RS-232
interface. By doing this, a new location on the interface was brought into focus to be
bleached by the photobleaching beam. This procedure was repeated for 10 different
locations on the interface. The average FT to PM ratio was used to estimate the surface

concentration.

4. 3.2 Calculation of C3 by absorption spectroscopy

A diode array spectrophotometer (8452 A, Hewlett Packard) was used to determine the
bulk concentration of proteins in the aqueous phase. The absorbance of the protein sample
was measured at 280 nm and compared with previously developed calibration curves of
absorbance versus unlabelled protein concentration. These calculations were complicated
by the fact that FITC absorbs at both 280 nm and 490 nm. Thus, the absorbance of the

labeled protein sample at 280 nm has contributions from FITC as well as BSA. To obtain

96

an accurate value for the absorbance of BSA alone, it was necessary to subtract the
absorbance of FITC at 280 nm from the total absorbance at 280 nm. To this end, a
calibration chart of absorbance at 490 nm versus concentration of pure F IT C was
prepared. From this, we could calculate the concentration of FITC in the protein sample.
Another absorbance calibration chart for pure FITC at 280 nm was then used to calculate
the absorbance of FITC at 280 nm. This was then subtracted from the total absorbance of
the protein sample at 280 nm to obtain the actual absorbance attributable to protein alone.
The protein concentration corresponding to this absorbance could then be calculated from
the unlabeled protein calibration chart. The calibration graphs along with sample spectra

for BSA and FITC are given in Appendix B and D.

4. 3.3 Prevention of protein adsorption on the bottom slide

In the calibration protocol discussed earlier, we made the assumption that the bulk
concentration of proteins is not affected by adsorption (this assumption is discussed in
detail in the Discussion section). In order to ensure that this assumption is valid at even
the lowest concentrations of proteins (10’8 M), it is necessary to prevent adsorption of
proteins from the bulk on to the waterwhite glass slide that serves as the bottom of the

flowcell.

Coating of a glass surface with polyethylene glycol, also called polyethylene oxide (PEO),
has been shown to effectively prevent adsorption of BSA (78). PEO-based protein

resistant surfaces firnction primarily by a steric exclusion mechanism involving very high

97

 

surface mobility and surface dynamics of the PEO chains. Andrade et al. (116) have
shown that, for a PEO-coated surface, protein resistance depends on several factors,
including protein size, surface density and chain length of PEO, as well as the degree of
polymerization. PEO polymers with the longest chain lengths at the optimum surface
density were shown to give maximum resistance to protein adsorption. To this end, we
used PEO with a molecular weight (M.) of approximately 8x106 (NO. 37,283-8, Aldrich
Chemical Company, Milwaukee, WI). Deionized water was used to make a 600-ppm PEO
solution. A bottom glass slide, which had previously been cleaned using the procedure
described by Cheng et al. (78), was immersed in the PEO solution for two hours to form a
coating on the slide. Before the slide was used, it was rinsed with water to remove excess

PEO.

4.4 Data analyses

The magnitude and quality of the output signal fiom the PMT depends on several factors
besides the actual fluorescence emission. These factors include dark current, background
light and flicker noise. In order to reliably interpret PMT data, it is necessary to examine
and account for these factors. Typical PMT outputs showing levels of dark current,
background light and concentration-dependent flicker noise are superimposed on a single

diagram in Figure 4.3

Dark current is the small amount of current that flows in a photomultiplier tube even when
operated in a completely dark environment. It is primarily due to thermionic emissions
from the photocathode and dynodes, leakage current (between the anode and the other

98

electrodes in the tube head) and ionization of residual gases in the tube. The level of dark
current in our system depends on the high voltage applied to the PMT and the set point of
the PMT cooling system. We measured a dark current of 50 counts at a high voltage of
800 V at a temperature of —22°C. These settings were maintained throughout all our

calibration experiments, and the dark current value was subtracted fi'om all PMT readings.

Even with the utmost precautions, stray light from various sources can enter the PMT
tube. For example, fluorescence from the immersion oil used to optically couple the prism
and the top microscope slide, scattered evanescent energy that bypasses the interference
filter or reflected laser light that enters the PMT tube from outside the microscope light
path. Together, these constitute the background noise in the PMT signal and can be
measured in the system at the pure buffer/oil interface. In, Figure 4.3 the fluorescence
signal jumps from the dark current level to the background noise level upon opening the
PMT shutter. Note that this is prior to the injection of proteins into the system.
Background light levels of 40 counts were typical in our calibration experiments, and were

subtracted from all PMT readings.

99

 

i

 

 

 

2‘ Fluorescence Signal 2
V 120°“ W —"7°+’- ‘6
2:
'Z’
0) 1000/5 x2
E 600/ Fluorescence Signali /
8 WWW — 471 *" 1°
g 400 .
(I)
9 .
O 200 - Background horse
5 W _ 96 +/- 5

0 Dark current

0 500 1000 1500 2000 2500 3000
time

Figure 4.3. PMT signal outputs showing typical levels of dark current, background noise
and shot noise proportional to fluorescence signals. The PMT is operated at 800 V and -
22°C.

100

 

 

The output from the PMT has an AC component noise that can be categorized by origin
as follows:

1. Shot noise fi'om signal light (flicker and quantum noise);

2. Shot noise fi'om background light; and

3. Shot noise fi'om dark current

The first kind of noise can come either fiom the inherent randomness of the fluorescence
excitation process (quantum noise) or from non-ideal experimental conditions (flicker
noise). For example, scattering of light by components of the experimental cell (prisnr,
microscope slide, oil etc) can excite fluorophores in the bulk, causing flicker noise. The
magnitude of the shot noise in the signal is usually expressed as the root mean square
value of the signal. The shot noise calculated for several different levels of PMT signals is
plotted in Figure 4.4. The rms value of the shot noise varies linearly with the signal and is

approximately 0.5 % of the magnitude of the signal.

In our calibration experiments, the PMT records data at a rate of 82 readings per second.
Each reading is the average of 256 samples of the detector by the photometer device. The

process of sampling the data 256 times takes approximately 4-6 msecs.

A typical fluorescence recovery curve is shown in Figure 4.5. After equilibrium
adsorption, the PMT was switched on at t = 0 seconds. For the first ten seconds of the
experiment, we measured the total fluorescence from the interface and the bulk. After ten

seconds, the shutter system was triggered causing the interface to be bleached while

101

simultaneously shuttering the PMT. This is manifested as a brief drop of approximately
100 ms in the fluorescence signal. The fluorescence signal immediately after
photobleaching is the bulk fluorescence (F bulk). A simple arithmetic average of the first 10

seconds of the fluorescence data gives F T, the total fluorescence.

Estimating the bulk fluorescence is however a more complicated task. Taking a simple
time average, as in the case of F r is unacceptable since the fluorescence signal increases
with time during the recovery process. Additionally, the first data point after opening the
shutter is not a valid reading since it can be artificially low. This is because the opening of
the PMT shutter is not always precisely synchronized with the start of the sampling
process. Hence, for a part of the sampling duration of 4-6 msecs, the PMT shutter might
have been closed, and thus some of the 256 samples of the detector that make up the first
reading could be only a measure of the dark current levels. Therefore the second data
point after photobleaching is used as the bulk fluorescence signal. The error in estimating
the bulk fluorescence signal in this fashion is approximately 0.5 %, which is the rrns value

of the shot noise of the signal as discussed earlier.

102

&
0|

 

‘5
O

 

 

we»
00’!

 

N
0|

 

 

 

N
O

 

RMS shot noise

.a
U"

 

 

.a
0

0|

 

 

 

 

o I I I I I I

0 1 000 2000 3000 4000 5000 6000 7000

Fluorescence Signal

Figure 4.4. RMS shot noise as a function of the magnitude of the fluorescence signal.

103

 

 

 

Normalized thoreecenee Inbnalty

 

 

 

 

0 10 20 30 40 50 60
Time (s)

Figure 4.5. A typical fluorescence recovery curve for TIRF/FPR experiments. F r is the
equilibrium fluorescence emission prior to photobleaching, and F M, is the fluorescence
emission immediately following a 15 mW, 100 ms photobleaching pulse.

104

4.5 Results

The concentration of adsorbed BSA calculated by the calibration protocol at different bulk
protein concentrations is shown in Figure 4.6. The interfacial concentration increases
linearly with bulk concentration up to 0.025 mM, after which the slope of the curve levels
off as the proteins begin to saturate the interface. The interface is saturated with BSA at a
surface coverage of approximately 0.07 mg/m2 at a bulk concentration of approximately
0.1 mM. Saturation at this bulk concentration is expected, since the critical micelle
concentration (cmc) of BSA, estimated in chapter 2 of this dissertation, is approximately

0.1mM.

The adsorbed quantities of BSA reported here are lower than the saturation adsorption
concentrations typically reported for the solid-liquid interface. Previous estimates of the
surface coverage of BSA using TIRF on various polymeric surfaces ranged from 0. 1-1
mg/mz. However, these studies were performed at a bulk concentration above the cmc of
BSA (7 8). On the other hand, lower estimates of surface saturation have been reported at
the oil-water interface. For example, Beaglehole et al. (117) used ellipsometry to estimate
the adsorption of BSA at the water-oleyl alcohol interface. They measured an adsorption

of 0.45 mg/m2 at a concentration of 0.05 mM.

Two factors can potentially contribute to the low estimates of interfacial protein
concentration reported in this work. First, it is possible that the high intensity laser beam
flash is not completely bleaching the interfacially adsorbed proteins. In this case, the post-

photobleach fluorescence will have a contribution fi'om the interface, leading to an

105

overestimation of the bulk emission at the expense of the interfacial emission. This will
result in an erroneously low calculation of the interfacial protein concentration. However,
previous experiments in our laboratory have shown that increasing the duration of the
bleach pulse above 150 ms has a negligible effect on the post-bleach fluorescence.
Secondly, as discussed earlier, the assumption that the quantum yields of fluorophores do
not change on adsorption can lead to an artificially low estimate of interfacial protein
concentration. Since, upon conjugation to BSA, the quantum yield of a fluorophore drops
by about 60% (92), it is unlikely that adsorption will cause a firrther reduction of similar
magnitude. However, even correcting for a 60% reduction in the quantum yield of a
fluorophore on adsorption will only increase the interfacial protein concentration by a
factor of 2.5. This, for the improbable and extreme case, is not large enough to account
for difl’erences between the surface coverages reported for the solid-liquid interface, and

those measured at our liquid-liquid interface.

We believe that the smaller interfacial concentrations measured in our system are due to
the fluid nature of the liquid-liquid interface. According to Dickinson ‘the greater mobility
at a fluid interface (e. g. liquid-oil) allows greater opportunity for protein unfolding than at
a solid surface’ (53). Given the dilute concentrations used in this study, one can postulate
that proteins arriving first at the fresh interface have the time and space to unfold and
denature, occupying a greater surface area per molecule and preventing fiuther
adsorption. This leads to a lower surface coverage at the liquid-liquid interface.
Additionally, at the higher concentrations typically used in the solid-liquid experiments

cited earlier, the finite time required for unfolding is very limited. Therefore a molecule is

106

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

........................................................................

 

 

Interfacial concentration Cs (m glm2)

 

 

I l

0.05 0.10 0.15 0.20 0.25 0.30 0.35
Bulk concentration Cb (m M)

 

Figure 4.6. Interfacial concentration of BSA, at the oil-water interface, as a filnction of
bulk concentration estimated by the calibration protocol.

rapidly surrounded by others, producing a close-packed state in which any configurational

readjustment is severely restricted. The rigid nature of the solid-liquid interface further

107

restricts the relaxation of the protein and thus encourages closer packing of the unfolded

globular proteins, leading to higher surface coverages.

4.6 Scope and application of the calibration protocol

The interfacial protein concentration estimated by the calibration protocol is dependent on
the accuracy with which the microscope is focussed at the interface. Improper focusing
could lead to a large bulk contribution to the total fluorescence signal and hence an
underestimation of the interfacial concentration. This is because, even though the
evanescent wave is naturally only formed at the interface, the absolute distance between
the microscope objective and the origin of the evanescent wave changes due to bumps and
ridges at the interface. For a fixed beam alignment and objective position, the accuracy of
the focus will then vary with these surface irregularities at the liquid-liquid interface. This
variance makes it impossible to compare data from different interfaces or even difl’erent
locations on the same interface. It is therefore necessary to refocus the microscope at
every new location on the interface to maximize the PMT signal, thereby ensuring an
optimum focus. After the microscope stage is moved to examine a new location, minute
changes in the focus of the microscope can be performed to maximize the ‘live’ PMT
signal (read directly off the MAC terminal). Now, even though the absolute PMT signal
may change at different locations on the interface, the ratio of total to bulk fluorescence is
constant. Thus a ratio of F p’F bun; obtained by bleaching one location on the interface will
yield the surface concentration (C,) at any other point on the interface. In our calibration
experiments, at a given interface the standard deviation of the F p’F by”, ratio at various
locations was less than 3%.

108

In theory, it is possible that the F 1'/F m ratio at a given bulk concentration will be the same
at different oil-water interfaces. By meticulously maintaining the experimental conditions
observed in the calibration experiments, this was shown to be true within an experimental
error of 5%. The variances between two oil-water interfaces are no larger than the
variances between difl'erent locations on the same interface. However, precisely
duplicating the laser beam intensity, the PMT voltage and operating conditions, the beam
alignment, and the quality of the top slide and its optical coupling to the prism is not easy

in practice.

4.7 Discussion

Calibration of the TIRF apparatus to enable direct estimation of interfacial concentration
of proteins from fluorescence emission data is a difficult endeavor. Several experimental
precautions needed to be rigorously observed to ensure that the integrity of the TIRF

instrumentation is not compromised. These are discussed below.

4. 7. 1 The detection system

A photomultiplier tube (PMT) is at the heart of our current TIRF system. The
fluorescence emission from the interfacial species is focussed on to the photocathode of
the PMT, causing the ejection of photoelectrons. A system of dynodes (the electron
multiplier) uses high voltages to multiply and amplify the photoelectrons eventually

reaching the anode, after which they are converted into an intensity signal. With current

109

amplifications as large as 108, it is important to ensure that the PMT electronics do not
drift. Drift usually occurs only during initial operation, as the signal tends to stabilize as
operating time elapses. We used a combination of aging and warm-up techniques (118) to
effectively eliminate drifting. Before an experiment, the PMT was run at the desired
operating voltage and exposed to light levels anticipated during the experiment for a
period ranging from 45 minutes to an hour. This was judged to be a sufficient enough
duration to stabilize the signal. Additionally, the laser was also run for a period of one
hour at the desired intensity prior to initiating experiments, as recommended by the

manufacturer.

An additional source of concern was the thermionic emission of photoelectrons. Due to
the extremely low work firnction of the multi alkali (N a-K-Sb-Cs) photocathode material,
thermally induced ejection of photoelectrons can occur even at room temperature. To
prevent this, we jacketed the PMT in a thennoelectrically cooled system, maintaining a

temperature of -25°C throughout the course of data collection.

4. 7.2 Heating effects of the laser beam

A source of concern in fluorescence photobleaching experiments is the possibility of
temperature variations due to the intense bleach pulse and the subsequent attenuated
monitoring laser beam. Potentially, this could cause several unwanted results, including

the following:

110

l. Distortion of the fluorescence recovery curves by convectional currents in the
experimental cell, thus interfering with the calculation of protein diflilsion;

2. Photodestruction of the protein molecule or alteration of its native properties
upon bleaching; and

3. Apparent photochemically induced cross-linking of proteins at the interface by

the monitoring beam.

Axelrod (77) has shown that, for a system with similar laser intensities, the local
temperature rise fiom the bleach pulse is approximately 0.03 °C, and thus essentially
negligible. Also, since proteins predominantly absorb in the ultraviolet region of the
spectrum, the laser beam (which is at a wavelength of 488 nm, the excitation wavelength

of the fluorophore) is not expected to have any efl‘ect on protein structure or function.

The apparent photochemically-induced crosslinking of proteins adsorbed at the interface is
typically manifested by a very slow illumination-dependent increase in the observed
fluorescence. This is difiicult to observe with our equipment, since bleaching of
fluorophores by the monitoring beam could possibly mask this efl‘ect. For experiments on
a short timescale (< 90 seconds), this effect is negligible and can be ignored. For longer
experiments, we use a spinning disc beam chopper that only allows periodic sampling of

the interface by the monitoring beam, so it is not always incident on the interface.

111

4. 7.3 The quantum yield problem

In the calibration protocol discussed earlier, we assumed that there was no difference in
the quantum yield of a fluorophore whether it is in the adsorbed layer (qys) or in the bulk
(qu). This enabled us to directly equate fluorescence fi'om the bulk and the surface,
eliminating the need to explicitly calculate either the quantum yield or the instrument

constant. We now discuss what we believe to be the validity of this assumption.

Fluorescence is a radiational transition between electronic states of the same multiplicity.
After an electron is excited to the first electronic state, it can return to the ground state by
two possible routes: emission or radiationless decay (including intersystem crossing). The
fraction of excited photons that decay through emission is called the quantum yield. Thus,
the quantum yield can also be defined as the ratio of the number of photons emitted to the
number absorbed. Environmental factors such as temperature, pH, and the presence of
other species can affect the quantum yield of any given molecule (98). These factors can
affect the rate constant of the luminescence and radiationless deactivation processes or the

nature of the lowest-lying excitation state.

Since fluorophores attached to surface-adsorbed proteins and bulk free proteins are
exposed to different microenvironments, we expect differences in their quantum yields. In
our experiments, we have eliminated changes in quantum yield due to fluorophore-

fluorophore interactions at the interface, which is caused by the close proximity of

112

adsorbed proteins. This is the phenomenon of concentration quenching, and was avoided

by using low labeling ratios and preparative chromatography, as discussed in chapter 3.

Additionally, the efl’ective quantum yield, a factor included in the overall instrument
constant, can difl’er between bulk and surface-associated fluorophores, due to the variance
of the collection emciency of the microscope objective with distance fiom the interface.
In general, the dipole emission of the fluorophore has angular characteristics that depend
on the distance from the surface of TIR. However, for an isotropic angular orientation of
transition moments excited by p-polarized light and given a high detection aperture, the
integral quantum yield was shown to be independent of the distance from the interface,

with a maximum difference of 5% (62).

However, the presence of quenchers, such as molecular oxygen in the bulk, and the efi’ects
of surface properties like charge and hydrophobicity may have an efi‘ect on the quantum
yield. With our experimental set-up, it is not possible to determine the importance of
these effects, or quantify them, and so we have neglected them in our analysis. At the
present time, the approach we have taken is widely used in the literature, although firrther
investigation into the quantum yield problem is clearly warranted. Lok et al. (2) reported
that exposing adsorbed BSA to a PBS solution caused an increase in extrinsic
fluorescence over that seen when unlabeled protein (in PBS) contacted the adsorbed

protein, but no attempt was made to quantify this efi‘ect.

113

It is possible, however, to qualitatively state that the quantum yield of the fluorophore at
the surface is less than that in the bulk. Physicochemical processes occurring as a result of
protein adsorption probably tend to decrease the quantum yield of an extrinsic label such
as fluorescein because its quantum yield is initially high. Further conformational changes
of the protein at the interface cannot increase the fluorescence quantum yield (119).
Therefore, the result of equating (1,3 to qbs is to potentially overestimate the quantum yield
of the fluorophores in the surface adsorbed layer. This causes the calibration protocol to

underestimate the concentration of interfacially adsorbed protein species.

4. 7.4 Does adsorption change the bulk concentration?

In our calibration protocol, we have assumed that the bulk concentration of proteins in the
aqueous phase of the experimental cell is unaltered by adsorption at the interface. This
enabled us to determine the bulk concentration by simply measuring the protein
concentration injected into the experimental cell. Since the experimental cell is flushed
with approximately three times its volume of protein solution before the cell is allowed to
achieve equilibrium, this would appear to be a valid assumption. Even at the lowest bulk
protein concentration (104 M), we calculated an aqueous phase bulk protein content that
was at least two orders of magnitude greater than the total surface excess of protein at the
interface. The total surface excess was estimated by multiplying the total interfacial area

of our cell with the unit surface excess measured by the calibration protocol.

114

As an additional check, we used absorption spectroscopy to determine the protein
concentrations entering and leaving the experimental cell. These were sampled fiom the
infusion and withdrawal syringes of the syringe pump respectively, and were equal within
experimental error. Hence, even with the low volume of our bulk aqueous phase (~ 1 ml),
we are confident that the bulk concentration remains unchanged following interfacial

adsorption.

4. 7. 5 Elimination of PMT light hysteresis: the shutter system

When the incident light or the supply voltage is changed in a step function, a
photomultiplier tube may not produce an output comparable with the same step firnction.
This phenomenon is known as PMT hysteresis and can be classified into either light
hysteresis or voltage hysteresis. Since we maintain a stabilized, constant high voltage
supply to the PMT, voltage hysteresis is not a problem in our experiments. However, at
the end of the bleach pulse, there is a drastic step change in the incident intensity as the
light level drops from the photobleaching level (20 mW) to the monitoring beam level (1 —
5 uW). At constant voltage, such a step change across nearly the entire measuring range
of the PMT will result in a temporary variation in the anode output. This variation is
highly undesirable, since it compromises the shape of the initial recovery curve and the

accuracy of the bulk fluorescence measurement.

115

 

 

 

 

 

lncldent laser light

I

8 Observation beam path
P Photo beam path

K Condenser

0 Objective

S1 PMT shutter

Figure 4.7. Beam path of Axiovert 135 microscope showing placement of the PMT
shutter (S 1).

116

To prevent hysteresis, we shuttered the PMT for the duration of the photobleaching pulse
using a fast-acting, electro—programmable shutter (LS6ZZ, Vincent Associates, Rochester,
NY). The microscope was modified to allow the shutter to be placed in the path of the
observation beam inside the microscope cavity as depicted in Figure 4.7. The PMT
shutter (St) has to be perfectly synchronized with the shutter blocking the photobleaching
beam (S2), such that when S; is open S1 is closed, and when S2 is closed, S; is open.
However, the exact synchronization of the shutter system is complicated by the fact that
the two shutters have difl‘erent response and transfer times on opening and closing. Each
shutter is driven by a model T-132 shutter driver (U niblitz, Vincent Associates, Rochester,
NY) that allows the control of delay and exposure times. A remote footswitch (44-610D,
Radio Shack, Fort Worth, TX) was used to simultaneously trigger the two shutter drivers.
Separate experiments were performed to determine the optimum settings of the shutter

drivers to enable synchronization of the shutters.

4.8 Conclusions

We have developed a calibration protocol for quantitation of interfacial protein species
based on the intensity of fluorescence emission. The method exploits the fact that the total
fluorescence detected by the PMT is a combination of contributions from the bulk and the
interface. We used photobleaching techniques to eliminate the interfacial fluorescence, to
enable determination of the fluorescence emission from a known bulk concentration.
From this data, we deduced the instrument constant, which subsequently allows the

calculation of the interfacial protein concentrations from fluorescence emission data.

117

The technique is instantaneous and self-contained, in that it does not require the use of
external standards or additional equipment. Also, since the protocol automatically
accounts for irregularities in the oil-water interface, the interfacial protein concentration in
a given situation can be determined by bleaching any random location at the interface.
This is highly desirable as it allows non-invasive sampling of the interface. In other words,
by bleaching in a comer of the interface, the adsorbed protein concentration can be
determined without compromising the nature of the macromolecular phenomena being

investigated at any other location on the interface.

We have measured a saturation surface coverage for BSA at the oil-water interface of
approximately 0.07 mg/mz, at a bulk concentration of 0.1mM. This is in agreement with
previously reported values of 0.1 mM for the critical micelle concentration (cmc) of BSA.
However, the values we calculated for the interfacial concentration of BSA are lower than
corresponding values reported for the solid-liquid interface. While this could be partially
attributable to experimental error and simplifications inherent in the theory of the
calibration protocol, we believe that it primarily reflects the increased mobility afi‘orded to

proteins at a liquid-liquid interface in comparison to those at the solid-liquid interface.

A source of concern in the development of the theory for the calibration protocol is the
assumption that the quantum yield of fluorophores does not change on adsorption. We
have discussed the validity of this assumption and its effect on various aspects of the

measuring system.

118

5. SUMMARY AND CONCLUSIONS

Protein adsorption dynamics at liquid-liquid (and liquid-like) interfaces play an important
role in several biological, pharmaceutical and food-related processes. For example, the
action of hormones in the body is mediated by the binding of a ligand protein to a receptor
protein adsorbed at the cell membrane. Also, the pharmacological activity of drugs is
governed by their behavior at bio-membranes, which can be modeled to a first
approximation by liquid-liquid interfaces. In food systems, protein adsorption at the oil-
water interface plays an important role in the stability of food colloids such as mayonnaise,
margarine, ice cream, etc., thus affecting their appearance and shelf life, as well as storage
and processing conditions. Thus, understanding and accurately characterizing the
adsorption and dynamics of amphiphiles at the liquid-liquid interface is of both scientific

and technological interest.

The overall goal of this dissertation was to investigate, develop and verify a novel
quantitative and in-situ technique for investigating macromolecular adsorption and
interactions at the liquid-liquid interface. The interfacial selectivity of the total internal
reflection geometry was exploited to enable extraction of molecular level information
about the dynamics of protein adsorption at the oil-water interface. An important
component of the apparatus is the development of a novel oil-water interface, which is
constructed fi'om a thin film of oil coated on a microscope slide in contact with water

approximately 1 mm deep. This innovation has circumvented the typical vibration-induced

119

instability problems encountered with optical techniques at liquid-liquid interfaces of

normal depth.

As a means of verifying the accuracy of the apparatus, the equilibrium adsorption of two
proteins, bovine serum albumin (BSA) and lysozyme, was investigated. The two proteins
were labeled with fluoresceine isothiocyanate (FIT C), and were selected for evaluating the
equipment because they are of different sizes, and are contrasts in nature. BSA is a large
globular protein that tends to unfold upon adsorption, whereas lysozyme is a smaller,
compact enzyme that undergoes very little structural change upon adsorption. The
adsorption profiles obtained correlated quite well with independently developed estimates
using the traditional method of interfacial tensiometry, and accurately reflected the nature
of the two proteins. In particular, the equilibrium adsorption profile of BSA showed a
pronounced break in the slope at a bulk aqueous concentration of approximately 0.1 mM,
indicating the onset of micellization. In contrast, lysozyme, which is a predominantly
hydrophilic protein and is thus not expected to form micelles, did not portray any evidence

of aggregate formation.

The effect of attaching an extrinsic fluorophore on the adsorption and difiirsion of proteins
was also investigated in this work. Ion-exchange chromatography was used to separate
proteins into samples with different monodisperse labeling ratios. This process helped to
ensure that we do not simply use a sample of an appropriate average labeling ratio which,
in reality, may contain many species with very high degrees of fluorophore conjugation.

Therefore, concentration quenching effects, which may be normally present in samples at

120

an equivalent average labeling ratio, were avoided. Near interface diflirsion coeficients
measured for proteins conjugated with two FITC molecules were 40% larger than the
coemcients for proteins labeled with a single F ITC molecule. This difference was
attributed to an increase in surface hydrophobicity caused by the presence of the additional
fluorophore. These results indicate that previous studies of this nature most likely

overestimated the lateral mobility of proteins by as much as 40%.

Fluorescence photobleaching recovery (FPR) was combined with TIRF to develop a
calibration protocol for the quantitation of proteins adsorbed at the interface. The method
is instantaneous, self-contained and non-invasive. Bulk BSA concentrations required to
produce interfacial saturation were measured, and were in very good agreement with
information available in the technical literature. However, surface coverages of BSA
measured at the oil-water interface were lower than those typically reported for the solid-
liquid interface. This difi’erence was attributed to greater fluidity at the oil-water interface,
which allowed proteins to unfold to a greater extent. This, in turn, increased the surface

area per protein molecule and thus decreased the surface coverage.

As a result of the work presented here, a new tool is now available for investigating
molecular-level protein dynamics at the oil-water interface. For situations where it is
necessary to calculate the concentration of interfacial moieties, a calibration procedure has
been developed and tested to enable absolute quantitation of adsorbed species. The effect
of attaching an extrinsic label to a protein has been of long-standing interest, since it is

generally believed that labeling affects the dynamics of the protein relative to its native

121

behavior. This work has clearly demonstrated that this is indeed the case, and that
accurate determination of native protein characteristics must account for the efl‘ect of any
extrinsic labels. It is expected that this apparatus will make it possible to study several
important phenomena relating to the dynamics of macromolecules at liquid-liquid

interfaces, including the study of receptor-ligand interactions at the oil-water interface.

122

6. RECONIMENDATIONS FOR FUTURE WORK

The work presented in this dissertation has resulted in the development of a total internal
reflection fluorescence microscopy (TIRFM) tool to study macromolecular interactions at
the liquid-liquid interface. So far, the study has verified the integrity of the novel thin
liquid-liquid interface, and produced a calibration protocol for quantitation of the
concentration of interfacially adsorbed species. The stage is thus set for investigation of a

wide variety of more complex and application-oriented interfacial phenomena.

Receptor-ligand interactions are of great importance in understanding cell behavior in
terms of molecular properties, which is a goal of molecular cell biologists and
bioengineers. Most of the efforts of the pharmaceutical industry are aimed at formulating
drugs that mimic, replace or interfere with natural compounds which regulate cell
firnction. Thus, understanding cell behavior at the molecular level is vital for work of this
nature. Additionally, the initial step in viral penetration of cells is binding to cell receptors.
Hence, this research would also provide a valuable insight into viral targeting. The
TIRFM tool developed in this work can be used to study various aspects of receptor-
ligand interactions, including, determination of the number of ligands bound by a given
receptor and the molecular mechanism and/or kinetics of receptor-ligand complexation.
Initial adsorption of the receptor onto the oil-water interface can be followed by
introduction of the appropriate ligand into the system. An experimental protocol

developed to differentiate between free and bound ligands, can then be used to study

123

receptor-ligand binding. For instance, l-anilino-8-naphthalene sulfonic acid (ANS), which
is practically nonfluorescent in aqueous solution, shows markedly increased fluorescence
with a blue shift in emission wavelength upon binding to BSA Thus, this property can be
exploited to monitor the binding of ANS too BSA Additionally, varying the solvent
conditions of the aqueous phase can provide valuable information on the dependence of

biological reactions on pH, ionic strength, temperature, etc.

In food colloids, as well as in in viva biological systems, there is rarely a lone protein
species involved in the adsorption process at the interface. Typically, the adsorbed layer in
a food product is a mixture of several difl‘erent kinds of proteins, surfactants, as well as
small molecules and ions. The food technologist/scientist would ideally like to be able to
predict the behavior of a mixture of proteins over a range of processing and storing
conditions (53). Similarly, in biological systems, ligands can be multivalent, crosslinked or
coupled to other molecules. Thus, any studies beyond the most fundamental would need
to consider the competitive adsorption and interaction between difl’erent adsorbing
species. Up to the present time, most competitive adsorption studies of proteins at liquid-
liquid interfaces have been performed by interfacial tensiometry. These studies lack
molecular detail and are limited in scope. Our TIRF technique is admirably suited to
examining the molecular aspects of competitive adsorption. Different proteins can be
introduced sequentially or simultaneously to compare surface efficiency, surface activity
and replacement kinetics. Also, the addition of ions and small molecule surfactants to the
aqueous phase of the interface can be used to improve the membrane mimetic chemistry of

the interface. Additionally, excitation with a tunable laser will allow the use of several

124

extrinsic fluorophores so that each protein species can be monitored at a difl‘erent

wavelength.

An interesting idea is the use of a two-laser excitation system to study simultaneous
competitive adsorption. An Arg-ion laser can be used to monitor the FITC labeled protein
at 488 nm, and a tunable laser can monitor another protein-label conjugate at a difi’erent
wavelength. A suitable emission filter can be chosen to block both the laser wavelengths
while allowing the fluorescence emissions to reach the PMT. The contributions to the

PMT signals from the two different fluorophores can be distinguished by:

1. laser excitation on different timescales, i.e. one laser in a continuous mode
while the other is either optically chopped or in a pulsed mode; and/or
2. differences in laser monitoring intensities coupled with inherent difi’erences in

quantum yields of the two fluorophores.

The TIRF apparatus can also be adapted to study the behavior of proteins at the interface
between natural hydrocarbons and water. The refi’active index of natural hydrocarbons is
typically lower than the oil currently used in our system. Therefore, this will necessitate a
change in the angle of incidence of the laser beam and possibly the use of a prism of lower
refractive index. For example, the interface between crown glass (n = 1.46) and water has
a critical angle of 68°. However, one should proceed with caution since it has been

reported that certain protein solutions have refractive indices quite difl‘erent from pure

125

water. In fact, the development of an auxiliary technique to measure the reflective indices

of protein solutions would be very helpful.

Estimation of the intensity profile and exact dimensions of the photobleaching and
monitoring beam would be useful in the determination of absolute difiirsion coefficients
and development of more accurate models of interfacial phenomena. Conventional
methods of translation against a knife edge or use of a light sensitive diode are tedious and
subject to experimental error. An immobilized fluorophore system (FIT C in gelatin),
developed by Robeson et a1 (33) to measure the amplitude of interference fiinges, can be

used to elegantly accomplish this as described below.

FITC is reacted with the gelatin and the resulting solution is spin cast onto the top
microscope slide prior to gelation. This three-dimensional system is free of concentration
quenching artifacts (Forster 1959), and using a gel with a high water content preserves the
conditions necessary for total internal reflection at the substrate/water interface. The
immobilized fluorophore system is photobleached either with a flash of the photobleaching
beam or for a longer duration by the monitoring beam. Since the fluorophores are
immobile, there is no recovery of fluorescence over time. Therefore, the degree of
bleaching and the size of the bleached spot will accurately reflect the beam properties.
Then, the beam geometry and size can be deduced fi'om the PMT signal obtained by
subsequent translation of the monitoring beam across the interface. In our system, the

monitoring beam can be translated across the interface by moving the microscope stage.

126

The velocity and distance of the microscope stage can be accurately controlled from the

computer via an RS-232 interface.

The second study in this dissertation investigated the effect of an extrinsic fluorophore on
the filnctionality of proteins. This research compared the behavior of protein samples with
monodisperse labeling ratios of 1 and 2. In a related experiment, we also showed that the
efi’ects of concentration quenching were not manifested in a sample with a monodisperse
labeling ratio of 2. It would be of interest to study samples with progressively higher yet
monodisperse labeling ratios to determine when the efl'ects of concentration quenching are
first manifested. Subsequent investigation of quenching in protein samples with average
labeling ratios equivalent to the monodisperse samples which are fi'ee of concentration
quenching can be performed. These experiments would provide information on the

efficacy of monodisperse labeling as a technique to avoid concentration quenching.

127

[1]

[2]

[3]

[4]

[5]

[6]

[7]

7. BIBLIOGRAPHY

Stryer, L., "Protein Structure and Function, "Protein Structure and Function, W.
H. Freeman and Company, New York, 1995.

Lok, B. K.; Cheng, Y.-L.; Robertson, C. R. , "Protein adsorption on crosslinked
polydirnethylsiloxane using total internal reflection fluorescence, ”J. Colloid
Interface Sci. 1983b, 91, 104-116.

Axelrod, D. , "Cell-substrate contacts illuminated by total internal reflection
fluorescence, "J. Cell Biology 1981, 89, 141-145.

Burghardt, T. P.; Axelrod, D. , "Total internal reflection/fluorescence
photobleaching recovery study of serum albumin adsorption dynamics, "Biophys.
J. 1981, 33, 455-468.

Burghardt, T. P.; Axelrod, 1D. , "Total internal reflection fluorescence study of
energy transfer in surface-bound and dissolved bovine serum albumin,
"Biochemistry 1983, 22, 979-985.

Thompson, N. L.; Burghardt, T. P.; Axelrod, D. , "Measuring surface dynamics of
biomolecules by total internal reflection fluorescence with photobleaching recovery
or correlation spectroscopy, "Biophys. J. 1981, 33, 43 5-454.

Iwamoto, G. K.; Winterton, L. C.; Soker, R. S.; VanWagenen, R A; Andrade, J.
D.; Mosher, D. F. , "Fibronectin adsorption detected by interfacial fluorescence,

"J. Colloid Interface Sci. 1985, 106, 459-463.

128

[3]

[9]

[10]

[11]

[12]

[13]

[14]

Lowe, R; Hlady, V.; Andrade, J. D.; VanWagenen, R A , "Human haptoglobin
adsorption by a total internal reflection fluorescence method Biomaterials, " 7:
1986, 41-44, 1986.

Horsely, D.; Herron, J.; V., H; Andrade, J. D., "Human and hen lysozyme
adsorption: a comparitive study using total internal reflection fluorescence
spectroscopy and molecular graphics, " In Proteins at Interfaces: Physicochemical
and Biochemical Studies, Brash, J. L., Horbett, T. A Eds, 1987.

Hlady, V.; Rickel, J .; Andrade, J. D. , "Fluorescence of adsorbed protein layers H
Adsorption of human lipoproteins studies by total internal reflection intrinsic
fluorescence, "Colloids Surf. 1988, 34, 171-183.

Hlady, V.; Andrade, J. D. , "Fluorescence emission fi'om adsorbed bovine albumin
and alburnin-bound l-anilinonaphthalene-8-sulfonate studied by TIRF, " Colloids
Surf 1988, 32, 359-368.

Hlady, V.; Andrade, J. D. , "A TIRF titration study of l-anilinonaphthalene-S-
sulfonate binding to silica-adsorbed bovine serum albumin, "Colloids Surf 1989,
42, 85-96.

Schmidt, C. F .; Zimmerman, R M.; Gaub, H. E. , "Multilayer adsorption of
lysozyme on a hydrophobic substrate, "Biophst. 1990, 57, 577-588.

Zimmerman, R. M.; Schmidt, C. F.; Gaub, H. E. , "Absolute quantities and
equilibrium kinetics of macromolecular adsorption measured by fluorescence
photobleaching in total internal reflection, "J. Colloid Interface Sci. 1990, 139,

268-280.

129

[15]

[16]

[17]

[13]

[19]

[20]

[21]

Pisarchick, M. L.; Thompson, N. L. , "Binding of a monoclonal antibody and its
Fab fragment to supported phospholipid monolayers measured by total internal
reflection microscopy, "Biophys. J. 1990, 58, 123 5-1249.

Pisarchick, M. L.; Thompson, N. L. , "Surface binding kinetics of a monoclonal
Fab fragment on supported phospholipid monolayers measured by total internal
reflection/fluorescence photobleaching recovery, ”Biophys. J. 1991, 59, 350a.
Pisarchick, M. L.; Gesty, D.; Thompson, N. L. , "Binding kinetics of an anti-
dinitrophenyl monoclonal Fab on supported phospholipid monolayers measured by
total internal reflection with fluorescence photobleaching recovery, "Biophys. J.
1992, 63, 215-223.

Kalb, E.; Engel, J.; Tamm, L. K. , "Binding proteins to specific target sites in
membranes measured by total internal reflection fluorescence microscopy,
"Biochemistry. 1990, 29, 1607-1613.

Hellen, E. H.; Axelrod, D. J. , "Kinetics of epidermal grth factor/receptor
binding on cells measured by total internal reflection/fluorescence recovery afier
photobleaching, " J. Fluor. 1991, 1, 113-128.

Pearce, K. H.; Hiskey, R G.; Thompson, N. L. , "Binding kinetics of fluorescently
labeled bovine prothrombin fragment 1 at planar model membranes measured by
total internal reflection fluorescence microscopy, "Biophys. J. 1991, 59, 622a.
Pearce, K. H.; Hiskey, R. 6.; Thompson, N. L. , "Surface binding kinetics of
prothrombin fragment 1 on planar membranes measured by total internal reflection

fluorescence microscopy, "Biochemistry 1992, 31, 5983-5995.

130

[22]

[23]

[24]

[25]

[26]

[27]

[23]

[29]

[30]

Poglitsch, C. L.; Sumner, M. T.; Thompson, N. L. , "Binding of IgG to MoFc*RH
purified and reconstituted into supported planar membranes as measured by total
internal reflection fluorescence microscopy, "Biochemistry 1991, 30, 6662-6671.
Shibata, C. T.; Lenhofl‘, A. M. , "TIRF of salt and surface effects on protein
adsorption, I Equilibrium, "J. Colloid and Interface Sci. 1992, 148, 469-484.
Shibata, C. T.; Lenhofl‘, A. M. , "TIRF of salt and surface efl‘ects on protein
adsorption, 11 Kinetics, "J. Colloid cmd Interface Sci. 1992, 148, 485-507.
Harrick, N. J .; Loeb, G. I. , ”Internal-reflection techniques in fluorescence
spectroscopy, "Modern Fluorescence Spectroscopy 1973, 1, 211-225.

Watkins, R W.; Robertson, C. R. , "A total internal-reflection technique for the
examination of protein adsorption, ”J. Biomed Mater. Res. 1977, 11, 915-938.
Lok, B. K.; Cheng, Y. -L.; Robertson, C. R. , "Total internal reflection
fluorescence: A technique for examining interactions of macromolecules with solid
surfaces, "J. Colloid Interface Sci. 1983a, 91, 87-103.

Kronick, M. N.; Little, W. A. , ”A new immunoassay based on fluorescence
excitation by internal reflection spectroscopy, "J. Immunol. Methods 1975, 8, 23 5-
240.

Axelrod, D.; Koppel, D. E.; Schlessinger, J.; Elson, B; Webb, W. W. , "Mobility
measurement by analysis of fluorescence photobleaching recovery kinetics,
"Biophy. J. 1976, 16, 1055-1069.

Axelrod, D.; Thompson, N. L.; Burghardt, T. P. , "Total internal reflection

fluorescent microscopy, "J. Microscopy 1983, 129 (Pt. 1), 19-28.

131

[31]

[32]

[33]

[34]

[3 5]

[36]

[37]

[33]

Weis, R. M.; Balakrishnan, K.; A, S. B.; McConnel, H. M. , "Stimulation of
fluorescence in a small contact region between rat basophfl leukemia cell and
planar lipid membrane targets by coherent evanescent radiation, "J. Biol. Chem
1982, 257, 6440-6445.

Tilton, R D.; Gast, A. P.; Robertson, C. R. , "Surface diffusion of interacting
proteins: Effect of concentration on the lateral mobility of adsorbed bovine serum
albumin, "Biophys. J. 1990, 58, 1321-1326.

Robeson, J. L., "Application of TIRF to probe surface difl‘usion and orientation of
adsorbed proteins, " Ph.D. dissertation. Carnegie Mellon University, 1995,
Laufl‘enburger, D. A; Linderman, J. J ., "Receptors: Models for binding, tramcking
and signaling, "Receptors: Models for binding, trafficking and signaling. , Oxford
University Press, New York, 1996.

Arai, K.; Kusu, F.; Takamura, K., "Electrochemical behaviour of drugs at the
oil/water interfaces, " In Liquid-liquid interfaces: Iheory and Methods, Volkov,
A. G., Deamer, D. W. Eds; CRC Press, 1996, p. 375-400.

Morrison, L. E.; Weber, G. , "Biological membrane modeling with a liquid/liquid
interface, "Biophys. J. 1987, 52, 367-379.

Arai, K.; Ohsawa, M.; Kusu, F.; Takamura, K. , "Drug ion transfer across an oil-
water interface and pharmacological activity, "Bioelectrochern. Bioenerg. 1993,
31, 65.

Provost, C.; Herbots, H.; Kinget, R , "Transparent oil water gels. study of some
physicochemical and biopharmaceutical characteristics. Part IV: the in vitro release

of hydrophilic and lipophilic drugs,, "Acta. Pharma. techno]. 1989, 35, 143.

132

[39]

[40]

[41]

[42]

[43]

[44]

[45]

Yoshikawa, K.; Shoji, M.; Nakata, S.; Maeda, S.; Kawakarni, H. , "An excitable
liquid membrane possibly mimicking the sensation of tase., "Langmuir 1988, 4,
759.

Deamer, D. W.; Volkov, A G., ”Oil/water interfaces and the origin of life, " In
Liquid-liquid interfaces: Theory cmd Methods, Volkov, A. G., Deamer, D. W.
Eds; CRC Press, 1996.

Nakanaga, T.; Takenaka, T. , "Resonance Raman spectra of monolayers of a
surface-active dye adsorbed at the oil-water interface, "J. Phys. Chem. 197 7, 81,
645-649.

Takenaka, T.; Nakanaga, T. , "Resonance Raman spectra of monolayers adsorbed
at the interface between carbon tetrachloride and an aqueous solution of a
surfactant and a dye, "J. Phys. Chem. 1976, 80, 475-480.

Takenaka, T. , "Efl‘ect of electrolyte on the molecular orientation in monolayers
adsorbed at the liquid-liquid interface: Studies by resonance Raman spectra,
"Chem. Phys. Lett. 1978, 55, 515-518.

Conboy, J. C.; Messmer, M. G; Richmond, G. L. , "Investigation of surfactant
conformation and order at the liquid-liquid interface by total internal reflection
sum-frequency Vibrational Spectroscopy, "J. Phys. Chem. 1996, 100, 7 617-7622.
Perera, J. M.; McCulloch, J. K.; Murray, B. S.; Grieser, F.; Stevens, G. W. , "An
attenuated total internal reflection spectrometer study of the kinetics of metal ion

extraction at the decane aqueous solution interface, "Langmuir 1992, 8, 336.

133

[46]

[47]

[43]

[49]

[50]

[51]

[52]

[53]

[54]

Grubb, S. G.; Kim, M. W.; Rasing, T.; Shen, Y. R. , "Orientation of molecular
monolayers at the liquid-liquid interface as studied by optical second harmonic
generation, "Langmuir 1988, 4, 452-454.

Sauer, B. B.; Chen, Y.; Zografi, G.; Yu, H. , "Static and dynamic interfacial
tensions of a phospholipid monolayer at the oil/water interface:
Dipalmitoylphosphatidylcholine at heptane/water, "Langmuir 1986, 2, 683-685.
Dickinson, E.; Euston, S. R; Woskett, C. M. , "Competitive adsorption of food
macromolecules and surfactants at the oil-water interface, " Progr. Colloid Polym.
Sci. 1990, 82, 65-75.

Plucinski, P.; Nitsch, W. , "Kinetics of interfacial phenylalanine solubilization in a
liquid-liquid microemulsion system, "J. Phys. Chem. 1993, 97, 8983-8988.
Graham, D. E.; Chatergoon, L.; Phillips, M. C. , "A technique for measuring
interfacial concentrations of surfactants at the oil-water interface, "J. Phys. E:
Scientific Instruments 1975, 8, 696-699.

Kakiuchi, T., "Equilibirum electric potential between two immiscible electrolyte
solutions, " In Liquid-liquid interfaces: Iheory and methods, Volkov, G. A.,
Deamer, D. W. Eds; CRC Press: New York, 1996.

Dickinson, E.; Stainsby, G.,"Colloids in Foods, "Colloids in Foods, Applied
Science Publishers, Barking, UK, 1982.

Dickinson, E.,"Food Colloids, "Food Colloids, 1982.

Markin, V. S.; Volkov, A. G., "Adsorption isotherms and the structure of oil-
water interface, " In Liquid-liquid interfaces: Theory and methods, Deamer, D.

W., Volkov, A. G. Eds; CRC Press: New York, 1996.

134

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

Leaver, J.; Dalgleish, D. G. , "Variations in the binding of b—casein to oil-water
interfaces detected by trypsin-catalyzed hydrolysis, "J. Coll. Interface. Sci. 1992,
149, 49-55.

Ando, J .; Albeda, S. M.; Levine, E. M. , "Culture of human adult endothelial cells
on liquid-liquid interfaces- A new approach to the study of cell-matrix interactions,
"In Vitro Cell. Dev. Bio 1991, 27, 525-532.

Murray, B. 8.; Nelson, P. V. , "A novel langmuir trough for equilibrium and
dynamic measurements oil air-water and oil-water monolayers, ”Langmuir 1996,
12, 5973-5976.

Malmsten, M. , "Ellipsometry and TIRF studies of adsorption processes in
parenteral drug delivery, "Interface Sci. 1997, 5, 159-167.

Aoyama, K.; Ogawa, K.; Kimura, Y.; Fujiyoshi, Y. , "A method for 2D
crystallization of soluble-proteins at liquid-liquid interface, " Ultramicroscopy
1995, 57, 345-354.

Shiba, Y.; Ohshima, T.; Sato, M. , "Growth and morphology of anchorage-
dependent animal cells in a liquid-liquid interface system, "Biotech Bioeng. 1998,
57, 583-5 89.

Andrade, J. D.; Smith, L. M.; Gregonis, D. E., "The Contact Angle and Interface
Energetics Surface and Interfacial Aspects of Biomedical Polymers Vol 1:, " In
Surface Chemistry and Physics, Andrade, J. D. Ed; Plenum Press: NY, 1985.
Axelrod, D.; Hellen, E. H.; Fulbright, R. M., "Biochemical Applications, " In
Topics in Fluorescence Spectroscopy, Vol. 3, Lakowicz, J. R. Ed., 1992, p. 289-

343.

135

[63]

[64]

[65]

[66]

[67]

[63]

[69]

[70]

[71]

Tweet, A. G.; Gaines, G. L.; Bellamy, W. D. , "Fluorescence of chlorophyll * in
monolayers, "J. Chem. Phys. 1964, 40, 2596-2600.

Wrrth, M. J .; Burbage, J. D. , ”Adsorbate reorientation at a
water/(octadecylsilyl)silica interface, "Anal. Chem. 1991, 63, 1311.

Bessho, K.; Uchida, T.; Yamauchi, A.; T., S.; Teramae, N. , "Microenvironments
of 8-anilino- l -naphthalenesulfonate at the heptane-water interface: Time-resolved
total internal reflection fluorescence spectroscopy, " Chem. Phys. Lett. 1997, 264,
381.

Watarai, H.; Saitoh, Y. , "Total internal reflection fluorescence measurements of
ion-association adsorption of water-soluble porphyrins at liquid/liquid interface,
"Chem. Lett. 1995, 283, 4.

Tupy, M. J .; Blanch, W. H.; Radke, J. C. , "Total internal reflection fluorescence
spectrometer to study dynammic adsorption phenomena at liquid/liquid interfaces,
"Ind Eng. Chem. Res. 1998, 37, 3159-3168.

Cheesman, D. F.; Davies, J. T. , "Physicochemical and biological aspects of
proteins at interfaces, "Adv. Protein Chem. 1954, 9, 439-501.

Aveyard, R; Vincent, B. , "Liquid-liquid interfaces: In isolation and in interaction,
"Prog. Surface Sci. 1977, 8, 59-102.

MacRitchie, F. , "Proteins at interfaces, "Adv. Protein Chem. 1978, 32, 283-326.
Davies, J. T.; Mayers, G. R. , " Studies on the interfacial viscosities of monolayers,

" Trans. Faraday Soc. 1960, 59, 691-696.

136

[72]

[73]

[74]

[75]

[76]

[77]

[73]

[79]

[30]

[31]

Fisher, L. R; Mitchell, E. B, "Effect of adsorbed proteins on interactions between
emulsion droplets, " In Food Colloids, Bee, R. D., Richmond, P., Mingins, J.
Eds, 1989, p. 123-137.

Yonese, M.; Hioki, Y.; Yano, M.; Kishimoto, H. , "Isooctane/water interfacial
adsorption of chondroitin sulfates having hydrophobic counterions, “J. Colloid
Interface Sci. 1992, 149, 314-321.

Castle, J .; Dickinson, E.; Murray, B. S.; Stainsby, G. , "Mixed-protein films
adsorbed at the oil-water interface, "ACS Symposium Series, 1987, 343, 118-134.
Harrick, N. 1.," Internal Reflection Spectroscopy, " Internal Reflection
Spectroscopy, Interscience Publishers, New York, NY, 1967.

Prieve, D. C.; F rej, N. A. , " Total internal reflection microscopy: A quantitative
tool for the measurement of colloidal forces, "Langmuir 1990, 6, 396-403.
Axelrod, D. , "Cell surface heating during fluorescence photobleaching recovery
experiments, "Biophys. J. 1977, 18, 129-131.

Cheng, Y. L.; Darst, S. A; Robertson, C. R , "Bovine serum albumin adsorption
and desorption rates on solid surfaces with varying surface properties, "J. Colloid
Interface Sci. 1987, 118, 212.

Edidin, M., "Fluorescent labeling of cell surfaces, " In Methods in Cell Biology,
Wang, Y.-L., Taylor, D. L. Eds, 1989, p. 87-102.

Canfield, E. R. , "The amino acid sequence of egg white lysozyme, "J. Biol. Chem.
1963, 238, 2698-2707.

Waggoner, A , " Covalent labeling of proteins and nucleic asids with fluorophores,

"Methods in Enzymology 1995, 246, 363-3 73.

137

[32]

[33]

[34]

[35]

[36]

[87]

[33]

Golander, C.-G.; Hlady, V.; Caldwell, K.; Andrade, J. D. , "Absorption of human
lysozyme and adsorbate enzyme activity as quantified by means of total internal
reflection fluorescence, 1251 labelling and ESCA, "Colloids Surf 1990, 50, 113-
130.

Robeson, J. L.; Tilton, R. D. , "Spontaneous reconfiguration of adsorbed lysozyme
layers observed by total internal reflection fluorescence with a pH=sensitive
fluorophore, ”Langmuir 1996, 12, 6104-6113.

Britt, D. W.; Buijs, J .; Hlady, V. , "Tobacco mosaic virus adsorption on self-
assembled and Langrnuir-Blodgett monolayers studied by TIRF and SFM, "Ihin
Solid Films 1998, 327, 824-824.

Tilton, R D.; Robertson, C. R; Gast, A P. , "Lateral difl‘usion of bovine serum
albumin adsorbed at the solid-liquid interface, "J. Coll. Interfacial Sci. 1990, 137,
192-203.

Gajraj, A; Ofoli, R , "A quantitative technique for investigating macromolecular
adsorption and interactions at the liquid-liquid interface, "submitted to Langmuir
1999.

Shulda, B. A, "Kinetics of desorption of BSA at the oil-water interface using total
internal reflection fluorescence photobleaching recovery, " Ph.D. dissertation.
Michigan State University, 1999,

Johns, L. M.; Shelden, E. A; Levitan, E. S.; Holz, R. W.; Axelrod, D. , "Motion
of chromafiin granules near the plasma membrane as visualized with total internal

reflection fluorescence microscopy, "Biophys. J. 1999, 76, A70-A70.

138

[89]

[90]

[91]

[92]

[93]

[94]

[95]

[96]

[97]

[93]

Sund, S. E.; Swanson, J .; Dzakpasu, R; Axelrod, D. , ”Cell membrane curvature
visualized by polarized total internal reflection fluorescence, ”Biophys. J. 1999, 76,
A7 5-A75.

Fulbright, M. R; Axelrod, D. , "Dynamics of nonspecific adsorption of insan to
erythrocyte membranes, "J. Fluorescence 1993, 3, 1-16.

Anderson, A. J. , "Covalent labeling of a protein, ”J. Chem. Edu. 1991, 68, 521-
522.

Hoyland, J ., "Fluorescent probes in practice - potential artifacts, " In Fluorescent
and luminescent probes for biological activity, Mason, W. T. Ed.; Academic
Press: Cambridge, UK, 1993.

Brynda, E.; Drobnik, J.; Vacik, J .; Kalal, J. , "Protein sorption on polymer surfaces
measured by fluorescence labels, "J. Biomed Mat. Res. 1978, 12, 55-65.

Crandall, R. E.; Janatova, J.; Andrade, J. D. , "The effects of radioiodination and
fluorescent labelling on albumin, "Preparative Biochemistry 1981, 11, 111-138.
Robeson, J. L.; Tilton, R. D. , "Effect of concentration quenching of fluorescence
recovery after photobleaching measurements, "Biophys. J. 1995, 68, 2145-2155.
Sacco, D.; Dellacherie, E.; Prochayret, F. , "Effect of covalent labeling of dextran-
benzenehexacarbosylate on haemoglobin, "J. Protein Chem. 1994, 13, 1-8.

Stryer, L. , "Ann. Rev. Biochem. 1978, 47.

Ingle, J. D.; Crouch, S. R,"Spectrochemical Analysis, "Spectrochemical Analysis,

Prentice Hall, NJ, 1988.

139

[99]

[100]

[101]

[102]

[103]

[104]

[105]

[106]

[107]

Peters, R; Peters, J .; Tews, K. H.; Bahr, W. , "Microfluorimetric study of
translational diffusion of proteins in erythrocyte-membrane, "Biochim. Biophys.
Acta. 1974, 367, 282-294.

Amersham, P.-B.,"Ion Exhange Chromatography: Principles and Methods, "Ion
Exhange Chromatography: Principles and Methods, Amersham Pharmacia
Biotech, Upsala, Sweden, 1996.

Lange, N. A,"Lange's Handbook of Chemistry, "Lange's Hancfiook of Chemistry;
vol. 13th edition, McGraw-Hill, 1985.

Ygueribide, J. E.; Talavera, E.; Alvarez, J. M.; Quintero, B. , "Photochem.
Photobiol. 1994, 60, 435.

Wegener, W. A; Rigler, R. , "Separation of translational and rotational
contributions in solution studies using fluorescence photobleaching recovery,
"Biophys. J. 1984, 46, 787-793.

Wegener, W. A. , ”Fluorescence recovery spectroscopy as a probe of slow
rotational motions, "Biophys. J. 1984, 46, 795-803.

Axelrod, D.; Rubinstein, J. L.; Yuan, Y. , "Fast rotational difl‘usion measured by
fluorescence recovery after photobleaching, ”A bstr. Pap. Amer. Chem. Soc. 1992,
203.

Ofoli, R Y.; Patel, A. , "to be submitted to Langmuir 1999.

Stout, A. L.; Axelrod, D. , "Spontaneous recovery of fluorescence by
photobleached surface-adsorbed proteins, "Photochem. photobiol. 1995, 62, 23 9-

244.

140

[108]

[109]

[110]

[111]

[112]

[113]

[114]

[115]

[116]

J auhari, P., "A study of the difiiision of bovine serum albumin at the oil-water
interface using total internal reflection fluorescence and fluorescence
photobleaching recovery, " Ph.D. dissertation. Michigan State University, 1997,
Nakai, S.; Lichan, E.; Hayakawa, S. , ”Contribution of protein hydrophobicity to
its firnctionality, "Nahrung-Food 1986, 30, 327-336.

Thomas, J.; Webb, W. W., "Fluorescence photobleaching recovery: A probe of
membrane dynamics, " In Non-invasive techniques in cell biology, Wiley-Liss Inc,
1990.

Ofoli, R Y.; Gajraj, A; Jauhari, P. , "Bulk and lateral difl‘usion of bovine serum
albumin at the oil-water interface, "Colloids and Surfaces (to be submitted) 1999.
Reichert, M.; Rockhold, S.; Van Wagenen, R. A; Andrade, J. D.,"Interfacial
aspects of biomedical polymers, "Interfacial aspects of biomedical polymers; vol.
2, Plenum Press, New York, 1984.

Norde, W. , ”Adv. Colloid. Interface Sci., "Adv. Colloid Interface Sci. 1986, 25,
267.

Wright, A. K.; Thompson, M. R. , "Hydrodynamic structure of bovine serum
albumin determined by transient electric birefi'ingence, "Biophys. J. 1975, 15, 137-
141.

Cha, W. N.; Beissinger, R. L. , "Macromolecular mass transport to a surface:
Effects of shear rate, pH and ionic strength, "J Coll. Interface. Sci 1996, 177,
666-674.

Andrade, J. D.; Hlady, V.; J eon, S. I. , "Poly(ethylene oxide) and protein resistance

- principles, problems, and possibilities, ”Advan. Chem. Ser. 1996, 248, 51-59.

141

[117] Beaglehole, D.; Lawson, F .; Harper, G. , "Tryptophan, tryptophan-leucine, and
BSA adsorption at an oil-water interface, "J Colloid Interf Sci. 1997, 192, 266-
268.

[118] Hamatsu,"Photomultiplier tyubes: Principle to application, "Photomultiplier
tyubes: Principle to application, Hamamatsu Photonics, 1994.

[119] Rockhold, S. A; Quinn, R. D.; VanWagenen, R A; Andrade, J. D.; Reichert, M. ,
"Total intemal-reflectionfluorescence (TIRF) as a quantitative probe of protein

adsorption, ”J. Electroanal Chem. 1983, 150, 261-275.

142

8. APPENDICES

8.1 Appendix A: Development of mathematical model for TIRF/FPR
experiments in the diffusion-limited regime.

The following steps present the details of the development of a model of fluorescence
recovery for pure-two-dimensional difiirsion monitored by a laser beam with a Gaussian
intensity profile. For simplicity, we approximate the photobleaching of fluorescent species
as a simple, irreversible first order reaction with a rate constant proportional to the

intensity profile.

Then the concentration of unbleached fluorophore, C (r, t), at a point r on the interface

and at time t is given by

dC (r, t) =

dt —a I (r) C (r, t) [41]

where I(r) is the bleaching intensity profile and a is a proportionality constant. For a
bleaching pulse that lasts a time interval T that is small compared with the characteristic

time for transport, the post-photobleaching fluorophore concentration at t = 0 is given by

C(r,0) = C0 e‘l‘m "’1 [42]

where Co is the uniform pre-bleach fluorophore concentration.

The amount of bleaching in time T is expressed by k, where

143

k = a1 (O)T

For a Gaussian intensity profile, I(r) is given by

2P 2
I(r) = ”2 exp[— %—) [431

 

where w is the half-width at e" height, and P is the total laser power.

The differential equation for lateral transport of a single non-interacting species of protein

in the absence of convection is given by Fick’s law
L; t) = DV2C(r, t) [441

where D is the two dimensional difl‘usion coeflicient. The boundary condition is

C(ao, t) = C0, and the initial condition is given by Eqn. [42]

The fluorescence F (t) observed at time t 2 0 is given by
_ q 2
F(t) .. (2)] I(r)C(r, t) d r [45]

where C(r, t) is the solution to Eqn. [44], subject to the boundary condition and k
dependant initial condition described earlier. q is the product of all the quantum
efficiencies of the fluorophore emission, absorption and detection, and A is the attenuation

factor of the beam during observation of the recovery.

The pre—bleach fluorescence F can be written as

144

20$:

 

F = A [46]
The Fourier transform of [44] yields

d—figt’iL-[p’bmm [471
where p2 = u: + pyz. The solution of Eqn. [47] is

5(#.1)=5(#,0)eXP[-#201] [471
Since,

(701,0) = (27:)‘1 [dzre‘”"C(r, 0) [48]
with C (r, 0) given by initial conditions and

C (r, t) = (2n)-1Ie"’”C(p, t)d2p [49]
We obtain fiom Eqn. [45]

F( 1):“?- )] d2#-¢(#)-e“’”"’" [501

It A

where 450:) is a function which depends on the laser intensity profile but not on the
mechanism of transport. On expanding ¢(p) and performing the integrations explicitly for

a general term in the series we obtain a series solution applicable for all values of k and t.

145

F( t) = Fi[%)-:Il+n[l+?]]-

For k values smaller than 1 this reduces to the simple form

k

2M

 

F(t)= ’ 1—

146

[51]

[52]

8.2 Appendix B: FITC
8. 2.1 Structure

Fluorescein isothiocyanate (FITC) is a derivative of fluorescein, an arnine-reactive
fluorescent molecule that is widely used in covalent labeling of proteins. The FITC
molecule shown in Figure 8.1 is identical to the fluorescein molecule except for the
presence of the isothiocyanate group at the 5- or the 6- position of the “bottom” ring of
the fluorescein structure. The 5-isomer is the more common form because it is easier to

isolate and is used in all experiments.

 

Figure 8.1. Structure of fluorescein isothiocyanate (S-isomer)

147

8. 2.2 Prototropic forms

The fluorescence efiiciency of FITC is closely dependent on the ionic state of the
molecule, which in turn is affected by environmental factors. Figure 8.2 shows the various
prototropic states of fluorescein at different pH values. An excellent source of
information on the different states of protonation of fluorescein can be found in
Ygueribide et al. (102). At low pH, the cationic form of fluorescein predominates and the
molecule is slightly fluorescent. In the range 2.2 <pH< 4.4 the neutral prototropic form
dominates and the molecule is non-fluorescent. At pH greater than 4.4 but less than 6, the
slightly fluorescent monoanion is present and this converts to the dianion at pH greater
than 6. The dianion shows a drastic, four-fold increase in fluorescence efficiency over the
anion. All our experiments were run at a pH of 7 .4, to ensure maximum fluorescence

efficiency of FITC.

148

OH I O i OH OH O O O 0
<-—>
00H
0 °°“ 0

Cation 4) = 0.39 Neutral 4’ = 0
pK2=4.4
0' O O I O I OH

H -

- COO
a 0
. . Anion

Dtanlon¢ = 0.93 4> — 0.26

Figure 8.2. Prototropic forms of fluorescein at different pH with fluorescence efficiencies.

149

8. 2.3 Absorption and emission spectra

The absorption and emission spectra for pure FITC in a pH 7 .4 PBS bufl‘er are shown in
Figure 8.3 below. Fluorescein displays a Stokes’ shift of about 20-25 run between
emission and excitation maxima. This is due to vibrational decay of excited molecules.
The dotted straight line at 500 nm denotes the 500nm EFLP (emission fiequency long
pass) filter. This filter was placed after the microscope objective in the light path (fi'om
the interface to the PMT), and blocks all light of wavelength below 500 nm. This
excludes excitation laser light, thus ensuring that the light impinging on the PMT is only

fiom the emission of fluorescence.

The region of overlap between the emission and excitation spectra can also be seen in the
figure. In this region, it is possible for a fluorophore to undergo a self-quenching process,

in which emission energy is lost in a radiationless transfer.

150

 

pa
«in

p
q

Fluormence Intensity (AU)
9 p r:- p r:

P
d
l

 

I

' ---A

T

r

“II

I

.1
an

A.

\/\\ 500+nm|ongpassflter

r 8915di m
/

 

V
r

m-flflflflfifl

 

T U- H
on an “DUI!

j t T l

webngth (nm)

Figure 8.3. Absorption and emission spectra (excitation at 488 nm) for FITC in pH = 7 .4
PBS bufl‘er, showing spectral overlap and Stokes’ shifi

151

8.3 Appendix C: SAS programs for data analysis

Mam-limited regime

The Statistical Analysis Software (SAS) code used to perform non-linear regression of
fluorescence recovery data in the difiirsive regime model is given below. The maximum
number of iterations allowed was 250, after which the data was abandoned as non-
convergent. On an average, each regression program took 30 seconds to return the fitted
parameters, depending on the size of the matrix of initial values provided. Comments

describing the code are provided in italics immediately after the relevant line.

libname client ‘c:\data';
filename adsasS dde "excellnew81r1009c3 1800004";
import of data fiom excel spreadsheet
data clientadsa58;
infile adsas8;
input 11 time;
proc nlin method=gauss;
identification of non-linear regression algorithm
parameters
p=0.0004
k=0.8
tau=1000
f=7 86;
declaration of initial parameters

bounds k>0;
bounds tau>0;

152

run;

declaration of constraints
temp=timeltau;
pb = eXP(-p*time);
model fi= pb*(f+f‘(-k*(2+2*temp)**(-1)+k**2/2*(3 +4*temp)**(-1)-
k"3/6*(4+6*temp)**(-1)));
description ofdrfitsion-limited model
der.k=t‘pb"(—(2+2*temp)"(-1)+k*(3+4*temp)"(-1)-k"2/2*(4+6*temp)**(-1));
pcatial derivative with respect to bleaching parameter
der.tau=f'pb*(-k*time/2*(tau+time)**(-2)+2*k"2*time*(3 *tau+4*time)**(-2)
-k"”"3*time*(4*tau+6*time)**(-2));
partial derivative with respect to characteristic time of diffusion
der.p=-tirne*pb*(f+i"'(-k*(2+2*temp)**(-1)+k**2/2*(3+4"'temp)**(-l)-
k"3/6"'(4+6*temp)**(-1)));
partial derivative with respect to photobleaching rate constant

153

Reaction-limited re gime
The Statistical Analysis Software (SAS) code used to perform non-linear regression of

fluorescence recovery data in the reaction regime model is given below. The maximum
number of iterations allowed was 250, after which the data was abandoned as non-
convergent. On an average, each regression program took 40 seconds to return the fitted
parameters, depending on the size of the matrix of initial values provided. Comments

describing the code are provided in italics immediately after the relevant line.

libname client 'c:\data';
filename brl dde "excellbrl !r1029c3:r8001c4";
import of data from excel spreadsheet
data clientbrl;
infile brl;
input 11 time;
proc nlin method=marquardt maxiter = 250 g4singular;
identification of non-linear regression algorithm
parameters
f0=400
kpb=0.001
r0=0.64
r1=0.05
k1=0.5
k2=O. O 1
f=596;
declaration of initial parameters
bounds fD>0;
bounds O<r0<1;

bounds O<rl<1;

154

run;

bounds kl>0;

bounds k2>0;

bounds kpb>0;

declaration of constraints

temp=exp(-time"‘kpb);

temp 1=exp(-time*kl);

temp2=exp(-time*k2);

model fi=f-(f-f0*temp)“(r0+r1*templ+(1-r0-r1)*temp2);
description of reaction-limited model

der.fl)=temp*(r0+rl ‘temp 1+(1 -rO-r1)*temp2);

partial derivative with respect to post-bleach fluorescence
der.kpb=-f0*time*temp“‘(r0+rl *temp1+(l-rO-r1)*temp2);
partial derivative with respect to bleaching rate constant
der.rO=-(f-fl)"'temp)*(l-temp2);
der.r1=-(f-fl)*temp)*(templ-temp2);

partial derivatives with respect to protein fractions
der.k1=(f-fD*temp)*(time*rl *templ);
der.k2=(f-f0*temp)*(time*(l-r0-rl)*temp2);

partial derivatives with respect to desorption rate constants

155

8.4 Appendix D: Absorbance calibration experiments

8.4.1 BSA atpH 7.4, PBS buffer

 

Absorbance calibration of BSA

1.2

A = 42.5oac
R’ = 0.9974

0.8

0.6

Absorbance Q 278 nm

0.4

02

 

0 0.005 0.01 0.015 0.02 0.025 0.03

Concentration mM

 

 

 

Figure 8.4. Absorbance versus concentration of BSA in pH 7.4 PBS bufl‘er, measured
with a diode-array spectrophotometer (8452A Hewlett Packard)

156

8.4.2 FIT C at pH 5. 1, acetate buffer

Absorbance calibration of FITC 9 pH 5.1

1.4 ..-

  

9
at

Absorbance
.0
O

0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055
Concentration (mM)

Figure 8.5. Absorbance versus concentration of FITC in pH 5.1 acetate buffer, measured
with a diode-array spectrophotometer (8452A Hewlett Packard)

157

8.4.3 FITC 1'an 7.4, PBS buffer

Absorbance calibration of FITC 9 pH 7.4

Absorbance

2 .....
1-8 R2 = 0.9994 ' ._
1:5 R2: 0.9971

 

o 0.01 0.01 0.02 0.02 0.03 0.03 0.04 0.04 0.05 0.05 0.06 0.06 0.07 0.07
Concentration mM

Figure 8.6. Absorbance versus concentration of FIT C in pH 7.4 PBS buffer, measured
with a diode-array spectrophotometer (8452A Hewlett Packard)

158

8.5 Appendix E: Data collection program

The Viewdac program code for reading data fi'om the PMT head memory is listed below.
An explanation of the code is presented in italics.
The PMT head is interfaced to an IBM 386 computer using an RS-232 interface with

RTS/CTS handshaking protocols. The baud rate of data transfer is 9600 bytes/ sec.

Start P
Begins data collection with the desired parameters

Set CW (control word)

Sets PMT gain levels, enables sync of mode. A gain of 10 was typically used

Set HV (high voltage)

Sets PMT dynode voltage. A voltage of 800 Vwas typical.

Set AVER (average)

Sets the number of times the PMT head arhotometer device) samples the detector in the
calculation of a PMT reading. T )pical values range fiom 25 6 to 64. Time for one
reading for an AVE = 25 6 was 4-6 ysecs.

Set INTER (time interval)

Sets the time interval between PMT readings, i. e. the time the PMT head rests between
sampling cycles of the detector. T )pical values ranged from 1 ms, for photobleaching
experiments to 150 ms for experiments monitoring the timescale of equilibrium
adsorption.

Set COLQTY (collection quantity)

Sets the number of readings to be taken by the PMT. T )picaly values ranged from 8000
to 20000.

159

"Illllllllllllllllillllli