NANOENGINEERED TISSUE SCAFFOLDS FOR REGENERATIVE MEDICINE IN
NEURAL CELL SYSTEMS
By
Volkan Mujdat Tiryaki

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Electrical Engineering—Doctor of Philosophy
2013

ABSTRACT
NANOENGINEERED TISSUE SCAFFOLDS FOR REGENERATIVE MEDICINE IN
NEURAL CELL SYSTEMS
By
Volkan Mujdat Tiryaki
Central nervous system (CNS) injuries present one of the most challenging problems.
Regeneration in the mammal CNS is often limited because the injured axons cannot regenerate
beyond the lesion. Implantation of a scaffolding material is one of the possible approaches to this
problem. Recent implantations by our collaborative research group using electrospun polyamide
nanofibrillar scaffolds have shown promising results in vitro and in vivo. The physical properties
of the tissue scaffolds have been neglected for many years, and it has only recently been
recognized that significant aspects include nanophysical properties such as nanopatterning,
surface roughness, local elasticity, surface polarity, surface charge, and growth factor
presentation as well as the better-known biochemical cues.
The properties of: surface polarity, surface roughness, local elasticity and local work of
adhesion were investigated in this thesis. The physical and nanophysical properties of the cell
culture environments were evaluated using contact angle and atomic force microscopy (AFM)
measurements. A new capability, scanning probe recognition microscopy (SPRM), was also used
to characterize the surface roughness of nanofibrillar scaffolds. The corresponding
morphological and protein expression responses of rat model cerebral cortical astrocytes to the
polyamide nanofibrillar scaffolds versus comparative culture surfaces were investigated by AFM
and immunocytochemistry. Astrocyte morphological responses were imaged using AFM and
phalloidin staining for F-actin. Activation of the corresponding Rho GTPase regulators was
investigated using immunolabeling with Cdc42, Rac1, and RhoA. The results supported the

hypothesis that the extracellular environment can trigger preferential activation of members of
the Rho GTPase family, with demonstrable morphological consequences for cerebral cortical
astrocytes.
Astrocytes have a special role in the formation of the glial scar in response to traumatic
injury. The glial scar biomechanically and biochemically blocks axon regeneration, resulting in
paralysis. Astrocytes involved in glial scar formation become reactive, with development of
specific morphologies and inhibitory protein expressions. Dibutyryl cyclic adenosine
monophosphate (dBcAMP) was used to induce astrocyte reactivity. The directive importance of
nanophysical properties for the morphological and protein expression responses of dBcAMPstimulated cerebral cortical astrocytes was investigated by immunocytochemistry, Western
blotting, and AFM. Nanofibrillar scaffold properties were shown to reduce immunoreactivity
responses, while PLL Aclar properties were shown to induce responses reminiscent of glial scar
formation. Comparison of the responses for dBcAMP-treated reactive-like and untreated
astrocytes indicated that the most influential directive nanophysical cues may differ in woundhealing versus untreated situations.
Finally, a new cell shape index (CSI) analysis system was developed using volumetric AFM
height images of cells cultured on different substrates. The new CSI revealed quantitative cell
spreading information not included in the conventional CSI. The system includes a floating
feature selection algorithm for cell segmentation that uses a total of 28 different textural features
derived from two models: the gray level co-occurance matrix and local statistics texture features.
The quantitative morphometry of untreated and dBcAMP-treated cerebral cortical astrocytes was
investigated using the new and conventional CSI, and the results showed that quantitative
astrocyte spreading and stellation behavior was induced by variations in nanophysical properties.

Copyright by
VOLKAN MUJDAT TIRYAKI
2013

Dedicated to my family.

v

ACKNOWLEDGEMENTS

First, I would like to thank my advisor, Professor Virginia M. Ayres, for her generous support
and guidance throughout my PhD studies at Michigan State University. Her broad vision has
always inspired me. It has been a great privilege having her as my advisor during these five
years. Her enthusiasm and wise criticism greatly influenced my professional development.
I would like to thank Professors Hayder Radha, Martin Crimp, Alicia Pastor, and Yiying
Tong for serving on my Academic Guidance Committee. Their technical feedback greatly
expanded the scope of my research.
I would like to thank Professor David I. Shreiber for his technical discussions and essential
contributions to our research. All primary cell cultures involving vertebrate animals in this
dissertation are prepared at Rutgers, The State University of New Jersey. The Animal Welfare #
is A3262-01. Fixed cell cultures are sent to Michigan State University. I thank Dr. Ijaz Ahmed
for teaching me primary neural cell preparation and analysis techniques, and also his hard work
in preparing the cell cultures, Western blotting, and immunofluorescence samples. I am also
grateful to Professor Sally Meiners for introducing our research group the new polyamide
nanofibrillar scaffolds, and her collaboration in our publications. I thank Professor Harini
Sundaraghavan for providing tissue scaffolds including polycaprolactone (PCL), methacrylated
hyaluronic acid (MeHA), and A6 gels. I also thank Corning Life Sciences for providing us
polyamide nanofibrillar scaffolds.
I thank Dr. Melinda Frame for invaluable discussions about colocalization and quantitative
immunolabeling analysis using confocal laser scanning microscopy. I thank Dr. Baokang Bi for
teaching me use of SEM and Bruker Dimension 3100 AFM at the MSU Keck Microfabrication
vi

Center, Lars Haubold (Fraunhofer, USA) for teaching me contact angle measurements, and
Katerina Moloni and Sharath Nair (nPoint Inc.) for their assistance in closed-loop multimode
AFM investigation. I am grateful to Abigail Tirrell for her help in imaging the AFM probe with
FESEM and gold coating of various tissue scaffolds, and Kan Xie for his help in SEM imaging
of polyamide nanofibers, PCL, MeHA, and A6 gels.
I also thank the Office of the Turkish Educational Attaché in New York for their support
during my PhD studies. I thank the National Science Foundation of the United State of America
(USA) for their support during my PhD studies through grant PHY-097776, awarded through the
Physics of Living Systems Program, Dr. Kraston Blagoev, Program Director. The Dissertation
Completion Fellowship provided by Michigan State University’s Graduate School was a great
honor and instrumental in helping me complete my work.
Finally, I would like to thank my father and my mother for understanding my wish to come
to the USA for my PhD studies, which opened a new era in my professional career. They always
loved and supported me. I dedicate my dissertation to them.

vii

TABLE OF CONTENTS

LIST OF TABLES

xii

LIST OF FIGURES

xiii

Chapter 1 Introduction
1.1 Motivation
1.2 Overview of tissue engineering for neural cell systems
1.3 Nanoscale imaging of tissue scaffolds
1.4 Summary of the following chapters

1
1
3
4
5

Chapter 2 Central nervous system
2.1 Cellular components of the CNS
2.2 CNS extracellular matrix
2.3 Glial scar
2.4 History of neural tissue engineering applications

6
6
8
9
10

Chapter 3 Scanning probe microscopy
3.1 Atomic Force Microscopy
3.1.1 Contact mode AFM
3.1.2 Tapping Mode AFM
3.2 Applications of AFM
3.3 SPRM and its applications

13
13
14
16
16
19

Chapter 4 Confocal Laser Scanning Microscopy
4.1 The principle of CLSM
4.2 Applications of CLSM

21
21
22

Chapter 5 Nanophysical Properties of Tissue Scaffolds
5.1 Surface Roughness
5.2 Local elasticity
5.3 Contact angle
5.4 Work of adhesion
5.5 Growth factor coverage
5.6 Effect of aging on nanofibrillar scaffolds
5.6.1 Experimental Procedures
5.6.1.1 Preparation of scaffold samples
5.6.1.2 Preparation of cell cultures
5.6.1.3 Analytical techniques
5.6.2 Experimental Results
5.6.2.1 TEM and SAED investigation of new and aged nanofibers
5.6.2.2 Raman spectroscopy investigation
5.6.2.3 AFM investigation of cerebellar granular neuron responses
5.6.3 Discussion

24
24
25
29
31
31
32
33
33
34
34
36
36
36
39
40

viii

Chapter 6 The Effect of Nanophysical Properties on Untreated and dBcAMP-treated
Astrocyte Differentiation
41
6.1 Nanofibrillar scaffolds induce preferential activation of Rho GTPases in cerebral
cortical astrocytes
41
6.1.1 Material and methods
44
6.1.1.1 Nanofibrillar Scaffolds
44
6.1.1.2 Planar Culture Surfaces
44
6.1.1.3 Primary Astrocyte Culture
45
6.1.1.4 Contact Angle Measurements
46
6.1.1.5 Optical Microscopy
46
6.1.1.6 Atomic force microscopy
46
6.1.1.7 Antibody and F-actin staining of astrocytes
47
6.1.1.8 Image Capture Conditions for Immunocytochemistry Structural Investigation
48
6.1.1.9 Image Capture Conditions for Protein Expression Estimate
48
6.1.1.10 Analysis of Cytoskeletal Colocalization
49
6.1.2 Results
50
6.1.2.1 Contact Angle Measurement Results of Culture Surfaces
50
6.1.2.2 AFM Surface Roughness Results
51
6.1.2.3 Optical Microscopy Survey of Astrocyte Morphology and Cell Density
54
6.1.2.4 Astrocyte Responses to Poly-L-lysine-functionalized Glass Surfaces
55
6.1.2.5 Astrocyte Responses to Nanofibrillar Scaffolds
58
6.1.2.6 Astrocyte Responses to Unfunctionalized Aclar Planar Surfaces
61
6.1.2.7 Astrocyte Responses to PLL functionalized Aclar Planar Surfaces
62
6.1.2.8 AFM-based Assessment of Astrocyte Stellation Response
63
6.1.2.9 Immunocytochemistry
64
6.1.2.10 Immunocytochemistry Structural Investigation
65
6.1.2.11 Protein Expression Estimate
67
6.1.2.12 Colocalization Analysis
69
6.1.3 Discussion
71
6.1.4 Conclusion
74
6.2 Differentiation of reactive-like astrocytes cultured on nanofibrillar and comparative
culture surfaces
75
6.2.1 Materials and Methods
78
6.2.1.1 Nanofibrillar Scaffolds
78
6.2.1.2 Planar Culture Surfaces
78
6.2.1.3 Primary dBcAMP-treated astrocyte cultures
79
6.2.1.4 AFM
80
6.2.1.5 Astrocyte Cell Shape Index
80
6.2.1.6 Astrocyte Soma Height Measurements
81
6.2.1.7 Astrocyte Process Measurements and Number of Astrocyte Processes
81
6.2.1.8 AFM Force Curves of Culture Surfaces
82
6.2.1.9 Immunolabeling for Cdc42, Rac1, RhoA, and F-actin
82
6.2.1.10 Image Capture Conditions for Cdc42, Rac1 and RhoA expression estimate 82
6.2.1.11 Immunolabeling for GFAP and Tubulin
83
6.2.1.12 Quantitative GFAP and Tubulin expression estimate
84
ix

6.2.2 Results
85
6.2.2.1 Local Elasticity Investigation of Culture Environments and Parameter Space
Analysis
85
6.2.2.2 Surface Energy Investigation and Parameter Space Analysis
87
6.2.2.3 Astrocyte Responses: Morphology Investigation
91
6.2.2.3.1 Cell density
91
6.2.2.3.2 Astrocyte morphology investigation by AFM
92
6.2.2.3.3 Astrocyte quantitative morphology assessment and comparison
94
6.2.2.4 Astrocyte Responses: Protein Expressions
97
6.2.2.4.1 GFAP and Tubulin Staining Results
97
6.2.2.4.2 GFAP and Tubulin Quantification Results
99
6.2.2.4.3 Cdc42, Rac1, RhoA, and F-actin Immunolabeling Results
101
6.2.2.4.4 Cdc42, Rac1, and RhoA Quantification Results
103
6.2.3 Discussion
104
6.2.4 Conclusions
108
Chapter 7 AFM Feature Definition for Neural Cells on Nanofibrillar Tissue Scaffolds
110
7.1 Introduction
110
7.2 Material and Methods
112
7.2.1 Neural Cell Culture
112
7.2.2 Nanofibrillar Culture Surface
112
7.2.3 Epi-Fluorescence Microscopy
113
7.2.4 Atomic Force Microscopy
113
7.2.5 Image Processing Methods
113
7.2.6 Gaussian High Pass Filter Implementation
114
7.2.7 High Boost Filtering Implementation
115
7.2.8 Histogram Equalization for Contrast Enhancement
115
7.3 Results
116
7.3.1 Feature Definition Problem
116
7.3.2 Systematic Approach to Problem Diagnosis
117
7.3.3 Filter Design Based on Problem Diagnosis
120
7.3.4 Biomedical Interpretations Based on Fluorescence Microscopy, AFM and GHPF
AFM Height Images
122
7.4 Discussion
124
Chapter 8 A new AFM Based Cell Shape Index for Quantitative Cell Spreading and Stellation
Investigation
127
8.1 Introduction
127
8.2 Experimental Procedures
129
8.2.1 Preparation of samples
129
8.2.2 Analytical techniques
131
8.2.2.1 Astrocyte Morphology Investigation by Conventional CSI
133
8.2.2.2 Astrocyte Morphology Investigation by New AFM-based Cell Shape Index 134
8.3 Experimental Results
135
8.3.1 Initial Segmentation Step
135
8.3.2 Astrocyte Morphology Investigation by Conventional CSI
138
x

8.3.3 Astrocyte Morphology Investigation by New AFM-based Cell Shape Index
8.4 Discussion

139
140

Chapter 9 Conclusions and future work

141

BIBLIOGRAPHY

145

xi

LIST OF TABLES

Table 6.1: Quantitative astrocyte stellation response results. [12*].

64

Table 6.2: Reactive-like astrocyte cell density measurement results (mean±SEM). [117*]. 91
Table 8.1: The textural features used to segment the cell surface from the cell substrate.
(STD: standard deviation; GHPF: Gaussian high pass filter; GLCM: gray level co-ocurance
matrix; IMC1: information measure of correlation 1; IMC2: information measure of
correlation 2)
133

xii

LIST OF FIGURES

Figure 2.1: The blood brain barrier is formed by endothelial cells, basement membrane, and
astrocyte endfoot. [22]. For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this dissertation.
8
Figure 2.2: Epi-fluorescent images of ECM proteins and astrocytes in blood brain barrier.
Immunolabeling was performed with antibodies against glial fibrillary acidic protein
(GFAP) in order to dye astrocytes green, and either fibronectin or laminin with red (a, b).
An astrocyte endfoot is shown by a single arrow, and an astrocyte cell body is shown by a
double arrow in (b). Scale bar, 35 μm. [15].
9
Figure 2.3: Comparison of regeneration in (a) PNS and (b) CNS. [29].

11

Figure 3.1: Schematic of contact mode AFM. [41].

15

Figure 3.2: ((a) and (b)) The effect of AFM probe cone angle on the tip dilation artifact in
AFM. (c) Dashed lines show the measured AFM height data and the solid lines show the
real surface data. [41].
18
Figure 4.1: Schematic of confocal microscopy. The detector aperture obstructs the light that
is not coming from the focal point, as shown by the dotted line. [53].
22
Figure 5.1: Elasticities of cellular microenvironments. The glass and plastic tissue cultures
are stiffer than the stiffest bone tissue. [61].
26
Figure 5.2: AFM force versus z-piezo displacement (F-z) curves including retraction and
extension parts. (a) Typical F-z curve of a soft material, (b) typical F-z curve of a material
with large adhesion component, and (c) F-z curve of PLL Aclar substrate.
28
Figure 5.3: Schematic of a contact angle measurement. The contact angle θ_C depends on
interfacial energies between solid and liquid (γ_SL), solid and vapor (γ_SG), and liquid and
gas (γ_LG).
30
Figure 5.4: The relationship between surface roughness, surface energy, and their influence
on cell adhesion. [66].
30

xiii

Figure 5.5: Tapping mode AFM images of unmodified polyamide nanofibers (a and c) and
nanofibers covalently cross-linked with FGF-2 (b and d). Images in (a) and (b) are height
and (c) and (d) are amplitude images. The difference in surface roughness of FGF-2
modified nanofibers is apparent in amplitude images. Images on same row have same scale
bar. [10].
32
Figure 5.6: TEM, Raman spectroscopy, and SAED results of aged ((a), (c), and (e)) and new
((b), (d), and (f)) polyamide nanofibers. No structural differences were evident in TEM
images of (a) aged and (b) new nanofibers. Arrows in the aged nanofibers (a) show dark
contrast inclusions. Scale bars, 100 nm. Raman spectroscopy, given in (c) and (d), showed
FWHM increases on the polyamide peaks and development of a broad disordered carbon
peak for the > 3 yr samples (c). AU: arbitrary units. Both SAED in (e) and (f) showed
demonstrated evidence of crystallinity within the nanofibers that were investigated.
Directionality in SAED of new nanofiber (f) further indicates potential crystal texture.
[67*].
38
Figure 5.7: A composite AFM image (a) shows the burrowing response of a cerebellar
granular neuron (CGN) on aged (> 3 yr old) nanofibrillar scaffolds. The neurite length is
approximately 0.5 mm. The close-up of the dashed box in (a) as shown in (b) shows that the
neurons with associated neurite outgrowth are underneath the nanofibers. Arrows show cell
debris or some CGN which lacked neurite. (c) Neurons cultured on new nanofibers clustered
on the top and lacked neurite outgrowth. Scale bars, 20 µm. [67*].
39
Figure 6.1: Contact angle measurements of culture surfaces. Aclar was significantly less
hydrophilic than nanofibrillar scaffolds, PLL glass and PLL Aclar. No significant
differences were identified from pairwise comparisons among the PLL Aclar, PLL glass,
and nanofibrillar scaffolds. [12*].
51
Figure 6.2: AFM surface roughness investigation of culture surfaces. High-resolution AFM
deflection images of (a) nanofibrillar scaffold, (b) PLL glass, (c) Aclar, and (d) PLL Aclar
by an A scanner. Scale bars, 250 nm. RMS surface roughness measurement results (e) of
nanofibrillar scaffolds, along nanofibers, PLL glass, Aclar, and PLL Aclar. The mean RMS
surface roughness measurements showed that nanofibrillar scaffolds had the highest surface
roughness but surface roughness along individual nanofibers was comparable to the planar
surfaces (inset (e)). Error bars show the standard deviation. [12*].
53
Figure 6.3: DIC images of astrocytes cultured on (a) PLL glass, (b) nanofibrillar scaffolds,
(c) Aclar, and (d) PLL Aclar. Scale bar, 100 µm. [12*].
55
Figure 6.4: AFM astrocyte morphology investigation on PLL glass. (a) AFM high pass
filtered composite height image of astrocytes by J scanner show lamellipodia formation
(scale bar, 20 µm). (b–d) Close-ups by E scanner of arrow regions a–d shown in (a): stress
xiv

fibers (b); filopodia anchors (c); and cell–cell interaction (d) (scale bars, 2 µm). (e and f)
Cross-section profiles of cells e and f in (a). Axis units are in µm with astrocyte soma extent
identified by red arrow. [12*].
57
Figure 6.5: AFM astrocyte morphology investigation on nanofibrillar scaffolds. (a) AFM
GHPF composite height image of astrocytes cultured on nanofibrillar surfaces by J scanner.
Scale bar, 20 µm. (b–d) Cross section profiles of cells b–d in (a). Axis units are in µm with
astrocyte soma extent identified by red arrow. [12*].
59
Figure 6.6: AFM high pass filtered composite height images of astrocytes cultured on
nanofibrillar scaffolds. (a) Dashed box identifies the intercellular connection. (b) Close-up
of the intercellular connection by E scanner. Dashed box identifies a contacted area. (c)
Close-up of a contacted area by E scanner. The scale bars show 10, 2, and 0.5 µm in (a), (b),
and (c), respectively. [12*].
60
Figuer 6.7: (a) AFM GHPF composite height image of astrocytes cultured on Aclar surfaces
by J scanner. The cross sections shown by b–d in (a) were given in corresponding section
profiles. Scale bar shows 20 µm and axis units are in µm. [12*].
62
Figure 6.8: AFM investigation of astrocytes cultured on PLL Aclar surfaces. (a) Composite
AFM deflection image of astrocytes by J scanner. Scale bar, 20 µm. (b) AFM deflection
image of an astrocyte by J scanner. Scale bar, 10 µm. Dotted arrow identifies cross section.
Dashed box identifies a cell edge. (c) Corresponding cell edge showing nanoscale filopodia
by E scanner. Scale bar, 2 µm. (d) Corresponding cross section profile. Axis units are in µm
with astrocyte soma extent identified by red arrow. [12*].
63
Figure 6.9: Immunocytochemistry structural investigation results. Left to right: fluorescence
images of astrocytes cultured on PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar
substrates. Immunolabeling and corresponding phalloidin are represented in red and green,
respectively. Astrocytes were labeled with (a) anti-Cdc42, (b) anti-Rac1, and (c) anti-RhoA.
Scale bars, 10 µm. [12*].
66
Figure 6.10: The white, gray, and black bars show the Cdc42, Rac1, and RhoA intensities,
respectively. The data is based on maximum intensity level images. Bars show the mean,
and error bars show standard error of n = 20 cells. [12*].
68
Figure 6.11: Overlaid fluorescence images of astrocytes cultured on (a) nanofibrillar
scaffolds and (c) PLL glass labeled with phalloidin (F-actin) and anti-RhoA. The
fluorescence was represented in green for F-actin and red for RhoA in (a) and (c). The
scatter plots of F-actin and RhoA for nanofibrillar scaffolds (b) and PLL glass (d) showed
partial colocalization with a mean PC of 0.66 and 0.58 respectively. Scale bar in (a) and (c)
show 10 µm and 20 µm, respectively. [12*].
70
xv

Figure 6.12: Elasticity of culture surfaces by AFM using DMT model for Young’s modulus.
(a) Young’s modulus box plot: Solid lines show the median, and the box edges show the
25th and 75th percentiles. Histogram distribution plots of Young’s modulus for (b) PLL
glass, (c) PLL Aclar, (d) Aclar and (e) nanofibrillar scaffolds. Dashed lines show the mode
values in (b), (c) and (d) in descending order. Unit x axis for (b), (c), (d), and (e) is 100
MPa. PLL G, PLL A, ACL, and NFS denote, PLL glass, PLL Aclar, Aclar and nanofibrillar
scaffolds respectively. [117*].
86
Figure 6.13: Work of adhesion characterization of culture surfaces. (a) Work of adhesion
box plot. Solid lines show the median, and the box edges show the 25th and 75th
percentiles. Histogram distribution plots of adhesion of (b) PLL glass, (c) Aclar, (d)
nanofibrillar scaffolds and (e) PLL Aclar. Dashed lines show the mode values in (b), (c) and
(d) in descending order. PLL G, ACL, NFS and PLL A, denote PLL glass, Aclar,
nanofibrillar scaffolds and PLL Aclar, respectively. [117*].
89
Figure 6.14: Two-dimensional parameter space representation of work of adhesion and
surface roughness of culture surfaces. (b) Two-dimensional parameter space representation
of surface polarity and surface roughness of culture surfaces. PLL G, NFS, ACL and PLL A
denote PLL glass, nanofibrillar scaffolds, Aclar and PLL Aclar, respectively. [117*].
90
Figure 6.15: Morphology investigation of reactive-like astrocytes with AFM height GHPF
images. (a) Astrocytes cultured on PLL glass, with close-up from dashed box region inset.
The close up image shows the nanoscale process formation. (b) Astrocytes cultured on
nanofibrillar scaffolds. (c) Representative morphologies for astrocytes cultured on Aclar. (d)
Astrocytes cultured on PLL Aclar shows glial-scar like chain formation. PLL G, NFS, ACL,
and PLL A, and denote PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar,
respectively. Scale bars, 20 µm. [117*].
93
Figure 6.16: Quantitative morphology assessment of reactive-like and quiescent astrocytes
cultured on PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar. Gray bars: reactivelike astrocytes, White bars: quiescent astrocytes. (a) Average length of astrocyte processes.
(b) Average number of processes. (c) Average maximum soma height. (d) Average CSI
results. Error bars show standard error of n=20 cells. PLL G, NFS, ACL, and PLL A, and
denote PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar, respectively. [117*].
97
Figure 6.17: Wide area GFAP staining images of reactive-like cerebral cortical astrocytes in
vitro. Maximum intensity projection images of reactive-like cerebral cortical astrocytes
cultured on (a) PLL glass, (b) nanofibrillar scaffolds, (c) Aclar and (d) PLL Aclar. Scale bar,
100 µm. [117*].
98
Figure 6.18: Close-up GFAP, tubulin, and DAPI staining images of reactive-like cerebral
cortical astrocytes. Maximum intensity projection fluorescence images of astrocytes
xvi

cultured on PLL glass, nanofibrillar scaffolds, Aclar and PLL Aclar. First, second and third
rows show the GFAP, tubulin, and overlaid images of GFAP, tubulin, and nuclei. GFAP,
nuclei, and tubulin, were represented in green, blue, and red, respectively. Scale bar, 50 µm.
[117*].
99
Figure 6.19: Quantitative total GFAP and tubulin expression estimation results for reactivelike astrocytes. Culture surfaces modulate the total GFAP (a) and tubulin (b) expression in
reactive-like astrocytes. Error bars show standard error of n=50 cells. PLL G, NFS, ACL,
and PLL A, and denote PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar,
respectively. [117*].
101
Figure 6.20: Cdc42, Rac1, RhoA, and F-actin staining images of reactive-like astrocytes.
Left to right column: fluorescence images of astrocytes cultured on PLL glass, nanofibrillar
scaffolds, Aclar, and PLL Aclar. Immunolabeling and corresponding F-actin were
represented in red, and green, respectively. Astrocytes were labeled with (a) anti-Cdc42, (b)
anti-Rac1, and (c) anti-RhoA. Scale bar, 100 µm. [117*].
102
Figure 6.21: Quantitative Cdc42, Rac1 and RhoA expression estimation results for (a)
reactive-like and (b) quiescent astrocytes. The white, gray and dark gray bars show the
average Cdc42, Rac1 and RhoA intensities, respectively. Error bars show standard error of
n=20 cells. PLL G, NFS, ACL, and PLL A, and denote PLL glass, nanofibrillar scaffolds,
Aclar, and PLL Aclar, respectively. [117*].
104
Figure 7.1: Feature definition problems of astrocyte neural cells on nanofibrillar surfaces.
(a) The structures marked by arrows in the AFM height image could be either nanofibers or
processes. (b) A higher resolution close-up tapping mode phase image of the region shown
by dashed box in (a). Its edge features are still unclear. [83*].
117
Figure 7.2: (a) Section measurement of nanofibrillar surface along dashed line, (b)
corresponding surface profile, and (c) DFT of nanofibrillar surface profile. (d) Section
measurement of cell surface along dashed line (e) corresponding surface profile, and (f)
DFT of cell surface profile. [83*].
119
Figure 7.3: GHPF results (a) filtered nanofibrillar surface profile, (b) DFT spectrum of the
nanofibrillar surface profile, (c) filtered astrocyte surface profile, and (d) DFT spectrum of
the astrocyte surface profile. [83*]
120
Figure 7.4: (a) Perspective plot and (b) radial cross section of the GHPF transfer function of
order 1 and normalized cutoff frequency 0.5. The zero frequency sample is at the center in
(a). [83*]
121

xvii

Figure 7.5: (a) Epi-fluorescence microscopy image of “isolated” astrocytes at 24 h. (b)
GHPF AFM image revealed significant leading edge formation. (c) Unfiltered AFM image
did not show full extent of leading edge formation. (d) Inclusion of 15% of original image in
high-boost filtering resulted in loss of feature definition to a level comparable to unfiltered
AFM image. [83*].
123
Figure 7.6: In the composite AFM height image (a), cellular processes and cell edges are
indistinguishable from the nanofibrillar background. (b) In the GHPF result, the cellular
edges and processes are clearly distinguished from nanofibrillar scaffold. Scale bars, 20 μm.
[83*].
124
Figure 8.1: The segmentation procedure results of AFM height images of an astrocyte
cultured on cell substrates. AFM height images of the astrocytes (left column), the
segmentation results (middle column), and individual cell area outlines (right column).
Scale bar, 10 μm.
137
Figure 8.2: The conventional CSI results of untreated and reactive-like astrocytes cultured
on Aclar, PLL glass, nanofibrillar scaffolds and PLL Aclar. Error bars show standard error.
138
Figure 8.3: New CSI results of untreated and reactive-like astrocytes cultured on Aclar, PLL
glass, nanofibrillar scaffolds, and PLL Aclar. Error bars show standard error.
139
Figure 9.1: A. Confocal GFAP staining image of an intercellular contact (arrow) between
untreated astrocytes cultured on nanofibrillar scaffolds. An AFM image of a contacting
structure is shown in the inset. B. In the higher resolution confocal image, the contact points
between the two cells (arrows) are not clearly resolved. C. In the STED image, the
continuity of the connections to the process is clearly resolved (right arrow). A potential
interaction between cells is shown by left arrow. The GFAP polymers were measured to be
about 90 nm using the STED image and Huygens software (asterisk). Images, D. I. Shreiber
and V. M. Ayres laboratories. Scale bars show 10, 2, 2.5, and 2.5 μm in A, A inset, B, and
C, respectively.
143

xviii

Chapter 1

Introduction
1.1 Motivation
Tissue engineering, or regenerative medicine, is a multidisciplinary field that merges expertise
from engineering, physics, and biology. Tissue engineering is based on experimental
observations that, when injury sites are supplied with artificial scaffolds, the cell populations can
regenerate and re-establish original functions within the surrounding tissue. Tissue engineering
applications depend on the location of the organ or injury. Fundamental research is needed to
design tissue scaffolds with appropriate mechanical, topographical, and biochemical properties.
Central nervous system (CNS) injuries present one of the most challenging problems.
Regeneration in the mammal CNS is often limited because the injured axons cannot regenerate
past the lesion. Implantation of a scaffolding material is one of the possible approaches to this
problem, and recent experiments have shown promising results. However, more research is still
needed before CNS tissue scaffolds can be designed and implemented successfully.
Characterizing physical properties of tissue scaffolds and determining the corresponding cell
behavior are required in order to improve the current tissue scaffold applications. The physical
properties of the tissue scaffolds have been neglected for many years, and it has only recently
been recognized that significant aspects include nanophysical properties such as nanopatterning
[1, 2], surface roughness [3, 4, 5], local elasticity [6], surface energy [7, 8], surface charge [9],
and growth factor presentation [10]. In this dissertation, the rat neural cell system was used to

1

model human CNS injuries. The advantages of using rats over other organisms as a model of
human disease have been reviewed by Iannaccone and Jacob [11].
The characterization of tissue scaffold physical properties must be at the same nanoscale
level as the cell-matrix interactions. Atomic force microscopy (AFM), a milestone invention in
nanotechnology, has been used to investigate the surface roughness, elasticity, nanofibrillar
architecture, and even growth factor presentation of tissue scaffolds [10]. Therefore, AFM has
great potential for determining physical properties of tissue scaffolds. Conventional AFM was
designed for planar substrates and therefore cannot provide automated property measurements
along individual cylindrical nanofibers. Scanning probe recognition microscopy (SPRM), the
integrated version of AFM with image processing, pattern recognition, and computer vision
techniques, was developed in the Electronic and Biological Nanostructures Laboratory at
Michigan State University. It can auto-track individual nanofibers, and therefore has the
capability of collecting data from an individual nanofiber, and was recently used to characterize
the physical properties of nanofibrillar scaffolds with higher detail than a conventional AFM [12,
13, 14, 15]. SPRM is expected to be used for new investigations of tissue scaffolds.
AFM is a surface technique that provides three-dimensional topography of samples. To
identify cellular mechanisms, internal protein expressions must also be analyzed through the use
of immunostaining. Immunostaining is a general term in biochemistry that applies to any use of
an antibody-based method to detect a specific protein in a sample. Immunostaining covers a
broad range of techniques used in histology, cell biology, and molecular biology that utilize
antibody-based staining methods. Two of those techniques are immunocytochemistry and
Western blotting, which are used in this dissertation. Both techniques are used to investigate cell
behavior in response primarily to nanophysical cell environments.
2

1.2

Overview of tissue engineering for neural cell systems

Tissue engineering uses cells and/or scaffolding materials that will facilitate regeneration around
a lesion, and improve or replace the biological function of an organ. This is a difficult task
because the material has to provide appropriate signals to the tissue over time. To be successful,
the scaffolding material has to mimic all aspects of the natural environment of cells. It should
provide appropriate mechanical support to the cells in the lesion area, and should be degraded
during the healing process. The type of tissue dictates the appropriate methods of tissue
engineering.
Mammals have a nervous system that is divided into the peripheral nervous system (PNS)
and the CNS. The CNS consists of the brain, spinal cord, optic, olfactory, and auditory systems,
and the PNS includes all other neural cells outside the CNS. The regeneration capability differs
between the two systems. The neurons in the PNS are able to regenerate to some extent; but the
neurons in the CNS often cannot regenerate and CNS injury usually results in paralysis. Neural
tissue engineering seeks the optimum engineering and life science approach that will provide the
appropriate environment for the neurons to grow past the lesion and re-gain their original
function.
Current neural tissue engineering applications have been tested on animals, and promising
results have been obtained [1]. The regeneration capability of rodents is known to be better than
that of humans. Until the techniques can be perfected for rodents, clinical applications are not yet
warranted.

3

1.3

Nanoscale imaging of tissue scaffolds

Optical microscopy has been the most commonly used technique in the biomedical sciences;
however, its resolution is limited by the optical diffraction limit, which is on the order of half the
wavelength of light. Assuming a light source with a wavelength of 500 nm, the optical
diffraction limit, also known as Abbe diffraction limit, would be approximately 250 nm. This
resolution is not enough to visualize the nanostructures of cells such as receptors, protein
assemblies, channels, and nanoscale membrane protrusions [16]. The Abbe diffraction limit was
one of the reasons why researchers sought a higher-resolution instrument. AFM, invented in
1986, is able to provide atomic resolution of samples under certain conditions. Since its
invention, AFM has been successfully used for biomedical investigations including tissue
engineering [10,14], drug delivery [17], protein folding [18], and clinical medicine [19].
Other techniques that provide nanoscale images are transmission electron microscopy (TEM)
and scanning electron microscopy (SEM), which were invented in 1931 and 1938, respectively.
Both techniques have been used to observe biological nanostructures. AFM has several
advantages over TEM. TEM provides a two-dimensional projection of a three-dimensional
volume of the sample while AFM provides a three-dimensional surface profile of the sample.
TEM sample preparation involves expertise in fixation and sectioning, and the sample must be
observed in vacuum. AFM sample preparation is easier than that of TEM as thin sectioning is not
required and standard culture preparations can be used. Capturing an AFM image can be done in
ambient air, and even in a liquid environment, which makes investigation of living cells possible.
SEM and TEM techniques do not allow live cell imaging because they require a vacuum. A
modified version of SEM, called the environmental SEM (ESEM), operates at atmospheric
pressure and can be used to investigate living cells. Image capture time by AFM is longer than
4

by TEM, but the scan speed of AFM in the newest commercially available instruments is
increased. AFM can be used not only for imaging at the nanoscale but also for characterizing the
mechanical properties and surface charges of regenerative tissue scaffolds.

1.4

Summary of the following chapters

Details about the CNS, neural tissue engineering applications, and CNS injury are discussed in
Chapter 2. The new SPM family member, SPRM, its principle, and its recent applications are
presented in Chapter 3. Chapter 4 addresses confocal laser scanning microscope (CLSM), which
provides three-dimensional localization of proteins and other cellular structures. The protein
expression estimation technique using CLSM is included in this chapter. In Chapter 5,
characterization of nanophysical properties of tissue scaffolds are described. The consequences
of the nanophysical environment for the neural cell system are presented in Chapter 6. In Chapter
7, a new cell morphology evaluation technique is proposed. The advantages of the technique are
compared to the conventional method. Finally, the summary of the dissertation and
recommended future work are given in Chapter 8.
Figures that use images that originate from the candidate’s thesis work are identified by an
asterisk in the figure caption.

5

Chapter 2

Central nervous system
The nervous system is responsible for signal transmission throughout the animal body. The
system is divided into the CNS and the PNS. The CNS consists of the brain, spinal cord, optic,
olfactory, and auditory systems, and the PNS includes all other neural cells outside the CNS. The
CNS transmits and interprets signals and also excites the PNS, whereas PNS innervates the
muscles, and transmits and receives signals from the spinal column.
This dissertation will focus on neural tissue engineering applications for the rat CNS model.
The differences between rat and human brain have not yet been fully addressed. The human
brain is the largest among mammals relative to body size. Another difference is that the cerebral
cortex of rodents is smoother than that of humans. Nonetheless, the advantages of using rats over
other organisms as a model of human disease have been documented [11] and use of a rat model
system is an accepted neural cell system investigative procedure.

2.1 Cellular components of the CNS
Neurons and glial cells are the two classes of cells in the CNS. Neurons are the main signaling
cells, and glial cells are the supporting cells in the CNS. There are three different types of
neurons in terms of function: sensory neurons respond to touch, sound, light and other stimuli
affecting cells of the sensory organs that then send signals to the spinal cord and brain. Motor
neurons receive signals from the brain and spinal cord, cause muscle contractions, and affect
6

glands. Interneurons connect neurons to other neurons within the same region of the brain or
spinal cord. Neurons cannot undergo mitosis, whereas glial cells are able to divide. Although
neurons cannot divide, they can extend or sprout axons under certain conditions.
Glia are not directly involved in information processing, but have vital roles such as
supporting and insulating neurons, removing debris after injury, directing outgrowth of axons,
and releasing growth factors. Glial cells are grouped as microglia and macroglia. Microglial cells
are phagocytes that become activated after injury, infection, or disease. Macroglial cells are
classified into oligodendrocytes, pericytes, and astrocytes. Oligodendrocytes insulate axons by
providing a myelin sheath to enable efficient signal transmission in the CNS. Pericytes are
connective tissue cells that occur around small blood vessels.
The importance of astrocyte functions is apparent from the ratio of the number of astrocytes
to neurons, which is 1:6 in worms, 1:3 in the rodent cortex, and 1.4:1 in the human cortex. This
indicates that the role of astrocytes increases as the complexity of the tissue increases [20]. The
majority of the glial cells are astrocytes, which have a star-shaped cell body (Figure 2.1).
Astrocytes are the key support cells for CNS neurons and have many roles; their functions
include transferring nutrients from capillaries to neurons, removing wastes from neurons,
contributing to the blood brain barrier, maintaining the potassium ion concentration in the
extracellular space between neurons, and controlling the blood flow. Astrocytes are a very
heterogeneous population of cells with specific morphologies and properties [21]. They are
considered to be the cellular bridge between the capillary basement membrane and the neurons
[22].

7

Figure 2.1: The blood brain barrier is formed by endothelial cells, basement membrane, and
astrocyte endfoot. [22]. For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this dissertation.

The cerebral cortex contains two types of matter: gray and white. White matter consists of
myelinated and unmyelinated axons, whereas grey matter consists of the dendrites, cell bodies,
neuroglia, unmyelinated axons, and axon terminals. Astrocytes are classified into fibrous and
protoplasmic on the basis of cell morphology and anatomy. The astrocytes in gray matter are
protoplasmic, whereas those in white matter are fibrous. Human astrocytes are known to be
larger, structurally more complex, and more diverse than rodent astrocytes [23].

2.2 CNS extracellular matrix
The extracellular matrix (ECM) is the part of animal tissues that usually provides structural
support and regulates intercellular communication. Animal cells, except blood cells, are
8

anchorage-dependent, and they adhere to a solid tissue composed of an ECM and resident cells
or an adjacent cell to survive. The solid tissue ECM may have physical properties such as soft or
stiff, thick or thin, and smooth or rough.
Cells adhere to other cells and the ECM through cell adhesion molecules. The ECM contains
many types of structural and directive proteins such as collagens (type I, II, III), multiadhesive
matrix proteins (laminin, fibronectin, entactin) and proteoglycans. The CNS ECM includes
fibronectin, laminin, tenascin C, and proteoglycans. Epi-fluorescent images of ECM proteins in a
matrix surrounding a capillary in a rat cerebrum are shown in Figure 2.2.

(a)

(b)

Figure 2.2: Epi-fluorescent images of ECM proteins and astrocytes in blood brain barrier.
Immunolabeling was performed with antibodies against glial fibrillary acidic protein (GFAP) in
order to dye astrocytes green, and either fibronectin (a) or laminin (b) with red. An astrocyte
endfoot is shown by a single arrow, and an astrocyte cell body is shown by a double arrow in
(b). Scale bar, 35 μm. [15].

2.3 Glial scar
Recent research indicates that the regenerative capability depends on the glial cells. After the
CNS injuries in mammals, glial scar formation creates a barrier to axon regeneration. Glial scars
consist mainly of reactive astrocytes and proteoglycans. Glial cells are believed to have a key

9

role in determining CNS regenerative capability. The response of astrocytes to injuries is called
reactive astrogliosis. During astrogliosis, astrocytes enlarge, become stellate, proliferate, and upregulate GFAP expression. Astrocytes are present in all CNS tissues without overlapping
domains. However, in severe reactive astrogliosis astrocytes form overlapping processes.
Astrocytes express four classes of proteoglycans: CSPG, heparan sulfate proteoglycans (HSPG),
and keratan sulfate proteoglycans (KSPG) are inhibitory to neurite outgrowth [24]; whereas
dermatan sulfate proteoglycans (DSPG) promotes axon regeneration in hippocampal neurons
[25]. FGF2, a growth factor in the CNS, is known to be up regulated after injury, and promotes
the proliferation and stellation of astrocytes in culture [26].
The reactive astrogliosis and glial scar formation has some beneficial functions such as
protection of neural cells, restriction of the spread of inflammation and infection, and promoting
tissue repair [27]. However, reactive astrogliosis has potential detrimental effects such as
inhibition of axon regeneration [27]. Consequently, developing a scaffolding environment that
would preserve the beneficial and attenuate the potential detrimental aspects of reactive
astrogliosis would be highly useful in the CNS regenerative medicine [27].

2.4 History of neural tissue engineering applications
Currently, there is no available treatment to restore the nerve function in the CNS (Figure 2.3).
Research in the CNS regenerative medicine increased after the 1980s. The developments were
complicated due to the controversial results from several studies. Embryonic spinal cord grafts
and peripheral nerve tissue grafts were able to support regeneration in the CNS; however, the
nerve fibers often did not grow back across the peripheral-central transition zone [28]. Therefore,
10

efforts are focused on designing an environment that is permissive to axonal regeneration.
Recent research shows that the permissive environment has to contain not only biomolecular
signals, but also physical guidance cues [29, 30].

(a)

Nerve cell
body
Injury >>>

Muscle
fibers
Proliferating
Schwann cell

Macrophage

Monocyte

(b) Nerve cell
body
Injury >>>

Postsynaptic
cell

Astrocyte

Oligodendrocyte

Figure 2.3: Comparison of regeneration in (a) PNS and (b) CNS. [29].

Physical guidance of axons is considered to be an important part of nerve repair. In the
nineteenth century, autologous nerve grafts [31], metal tube [32], bone [33], and fat sheaths [34]

11

were used to physically guide the regeneration of the PNS. Autologous nerve grafts were
accepted as the gold standard for treatment of PNS injuries in the late 1980s [35,36]. The other
material types that are currently under research are nonautologus tissue, natural-based materials,
and synthetic materials. Among those materials, synthetics are attractive because their physical
and chemical properties such as surface roughness, surface energy, degradation rate, elasticity,
and growth factor presentation can be optimized for specific applications. The challenges of the
synthetic materials are the different inflammatory response of a body and compatibility with cell
adhesion and tissue repair. Ideally, a synthetic material that would induce axonal regeneration
should be biodegradable, semipermeable, and non-cytotoxic.
Recent research has identified an implantable scaffold composed of randomly oriented
electrospun polyamide nanofibers that appears to have unusual wound healing properties for the
CNS and that acts via modulation of the astrocyte response [12]. The polyamide nanofibers
promote neuronal attachment and neurite generation [1], and produce no cytotoxicity in vitro [1]
and in vivo [37]. They are also biodegradable, but not rapidly [37], which is an advantage
because the materials has to remain until full recovery. Furthermore, the polyamide nanofibrillar
scaffolds are flexible and therefore ideal for neural tissue engineering applications. Finally and
more importantly, polyamide nanofibrillar scaffolds have shown promising results for the repair
of spinal cord injury [38]. The present dissertation further investigates both the nanophysical
properties of the polyamide nanofibrillar scaffolds and the neural cell responses to these
scaffolds.

12

Chapter 3

Scanning probe microscopy
3.1 Atomic Force Microscopy

Scanning probe microscopy began with the invention of the scanning tunneling microscope
(STM) in 1982 [39]. The invention of this instrument opened a new era of surface science. The
STM tunneling current between the probe and the sample is kept constant while the probe scans
over the surface of the sample. The vertical movement of the z piezo is recorded at each x,y
coordinate, which yields an image of the sample topography. One of the limitations of this
instrument is that the sample has to be electrically conductive. Although images of lowconductivity samples with STM have been obtained [40], it is not easy to establish a tunneling
current with low-conductivity samples such as biological samples.
AFM, or scanning force microscopy, is another type of scanning probe microscope. AFM
was developed within four years of the invention of STM. An AFM system consists of scanning,
probe, sensor, vibration isolation, controller, and computer systems. The principle of AFM is to
scan the surface of a sample with an AFM probe that is attached to a cantilever. The cantilever is
flexible and has a specific spring constant. The probe is kept atomically close to the surface. The
tip scans the specimen line by line by an accurate piezo-scanner. The top of the cantilever has a
polished region, and the vertical movement of the tip is measured by transmitting a laser beam to
the polished surface, and then detecting the reflected laser light from the cantilever.

13

In this thesis, the SPM technique that was primarily used was AFM. Contact mode AFM,
tapping mode AFM, and SPRM were used to image the cell cultures and to investigate the
nanophysical properties of culture surfaces.

3.1.1 Contact mode AFM
Contact mode was the first developed imaging mode of AFM, and it can be used to capture
images of non-conductive samples. In this mode, a probe attached to the end of a cantilever is
scanned over the sample surface while a split photodiode detector measures the magnitude of
cantilever deflection from the reflected laser beam. The photodiode measures the intensity of
laser light on top minus the intensity of laser light on the bottom divided by the total incident
light. The feedback loop maintains constant deflection via the input from the photodiode detector
[41]. A contact mode schematic representation of AFM is shown in Figure 3.1.

14

Feedback loop maintains
constant tip-sample interaction
A+B-C-D
A+B+C+D

Detector
Electronics

Controller
Electronics

Laser

Cantilever & Tip
Split Photodiode
Detector

Cantilever
substrate
Sample

Y
X

Piezoelectric
scanner

Z

Figure 3.1: Schematic of contact mode AFM. [41].

The force that the tip applies to the sample can be calculated using Hooke’s law in contact mode
AFM:
(3.1)
where F, k, and x are the force, spring constant and cantilever deflection, respectively. The
constant deflection means a constant force is applied to the sample. AFM height data is
constructed by recording the voltage applied to the z piezo at the same time the probe’s (x,y)
position. The deflection data in contact mode AFM is obtained by recording the cantilever
deflection that occurs prior to re-establishing the constant force.

15

Contact mode AFM investigation can be performed in ambient air and liquid environment.
However, lateral and normal forces may reduce spatial resolution and may damage biological
samples.

3.1.2 Tapping Mode AFM
In tapping mode AFM [41], the cantilever is oscillated approximately at its resonance frequency
and an amplitude range of typically from 20 nm to 100 nm. The tip lightly “taps” the specimen,
and the tip contacts the surface at its minimum point of the cycle. The tip sample distance affects
both the amplitude and the frequency of the tip. The change in frequency gives information about
the sample. The feedback loop keeps the amplitude constant based on the four cell photodiode
RMS amplitude measurement. At each lateral (x,y) point the vertical z point is recorded that
maintains the setpoint amplitude. The z point recordings generate the topography of the sample.
A phase image is obtained when the frequency change or the phase shift of the tip is recorded at
each (x,y) point. The phase shift is very sensitive to the material and topographic properties.
In tapping mode AFM lower forces are applied to sample as compared to the contact mode,
which means less damage to soft tissues. The scan speed is slightly slower than contact mode
AFM.

3.2 Applications of AFM
AFM is an instrument that can reach macromolecular resolution with relatively short sample
preparation time and over a wide temperature range. High-resolution imaging capability led
AFM to have many applications for biomedical sciences. AFM is a marker-free imaging tool for
obtaining sample topography. The tapping mode, pico-newton imaging system of AFM, was
16

developed after the nanonewton forces of contact mode was shown to damage the soft biological
tissues in some cases. The successful operation of AFM in liquid environment allowed
biomedical scientists to perform experiments in nearly native environment conditions [42,43].
There are some disadvantages of AFM that are also important to realize. AFM requires
complementary investigations such as CLSM to obtain sub-surface information. The scanning
speed of AFM is slow compared to optical and fluorescence microscopy and SEM. A typical
scan in AFM requires several minutes, which may be longer depending on the resolution and
scan rate. The relatively slow rate of AFM may also reflect thermal drift of the sample [44].
Finally the lateral resolution is limited by the shape of probe. AFM image is affected by the
radius of the probe curvature and tip sidewall angles. The effect of AFM probe shape on the
AFM height image is illustrated in Figure 3.2.

17

(a)

(b)

(c)
Figure 3.2: ((a) and (b)) The effect of AFM probe cone angle on the tip dilation artifact in
AFM. (c) Dashed lines show the measured AFM height data and the solid lines show the real
surface data. [41].

AFM is not only an instrument to visualize samples with macromolecular resolution. AFM
force curves can be used to investigate material properties such as elasticity, Hamaker constant,
adhesion, and surface charge densities, which was elegantly reviewed [45]. Therefore, AFM is
an instrument that is likely to contribute to the biomedical sciences.
Numerous publications have been done with the contribution of AFM in biomedical sciences.
The surface biology of DNA by AFM was investigated by Hansma [46]. AFM was used to
18

investigate the morphology of quiescent astrocytes cultured on different cell substrates [12]. It
was used to image the molecular structure of aggrecan [47] and laminin on a planar surface [48].
AFM was also used to determine the elasticity of living tissue [49]. These applications and others
demonstrate the capabilities of the instrument in the biomedical sciences. The increased scan
speed of AFM in novel commercial instruments indicates the contributions of this instrument
would likely to continue in the future.

3.3 SPRM and its applications
Scanning probe recognition microscopy (SPRM) is a new scanning probe microscopy capability
developed recently in an NSF GOALI collaborative partnership with Veeco Instruments (now
Bruker). In SPRM, the scanning probe microscope system itself is given the power to auto-track
specific regions of interest through feature recognition coupled with adaptive scan plan
generation and implementation. 'Feature' is not just topography, e.g. nanoscale elasticity was
investigated as a physical property feature. SPRM is an approach that works directly with the
interaction sensing capability of a scanning probe microscope, which inherently has atomic to
nanometer scale resolution. The human operator interaction is now focused to the decisionmaking level rather than the execution level, and the research emphasis evolves from sample
imaging to feature/property investigation.
For AFM based surface roughness measurements, if the region of interest is not rectangular,
such as nanofibers, the surface roughness measurements need to be repeated many times. SPRM
allows adaptively following along individual nanofibers within a nanofibrillar tissue scaffold.
The comparison of SPRM auto-track imaging versus traditional AFM imaging was demonstrated

19

and histograms of various scaffold surface roughness were analyzed by Fan et al. [50]. This work
has demonstrated the advantages of SPRM for nanoscale tissue scaffold investigations.
SPRM was recently used [51] for nanoscale elasticity measurements to collect local force
curves at nanofiber median points with the tip normal to the nanofiber cylinder axis.
Measurements from several nanofibers were compiled into a statistical representation of the
nanofibrillar matrix. The measurements were related to a Young’s modulus using a Hertz model.
SPRM was used to investigate the surface roughness along individual nanofibers and given
in Chapter 6 of this thesis.

20

Chapter 4

Confocal Laser Scanning Microscopy
Confocal laser scanning microscopy (CLSM) is a technique that captures high-resolution infocus images from selected depths of a sample. Images are acquired at each point on the focal
plane and reconstructed by a computer, which yields three-dimensional images of samples. The
image quality is enhanced compared to optical microscopy because images are acquired from
multiple depths and are not superimposed. The principle of the CLSM was patented by Marvin
Minsky in 1957 [52], but the development of this instrument was not achieved until the late
1980s, when laser scanning systems became available.

4.1 The principle of CLSM
In CLSM, the laser beam passes through an aperture and then focuses on the small focal volume
of the sample. Scattered, reflected, or fluorescent laser light from the illuminated spot is
collected by an objective lens. A beam splitter reflects some of the collected light into the
detection apparatus, which is usually a photomultiplier tube. The light intensity is then recorded
by a computer. The detector aperture passes only the light that is coming from the focal point and
prevents passing of the out-of-focus light (Figure 4.1). This yields sharper images than those of
conventional fluorescence microscopy and enables capturing images of focal planes at various
depths within the sample. Series of these images are called z-stacks [53].

21

Beam splitter

Light source
Aperture

Detector
aperture

Focal plane

Light detector
Figure 4.1: Schematic of confocal microscopy. The detector aperture obstructs the light that is
not coming from the focal point, as shown by the dotted line. [53].

The detected light captured from the focal point within the sample represents one pixel in the
final image. The laser beam is scanned over the horizontal plane of the sample by using servocontrolled oscillating mirrors. The focal plane is changed by raising or lowering the microscope
stage, and a computer assembles the multiple successive focal plane images into a threedimensional image of the sample. The lateral and axial resolution of CLSM can be as low as 66
nm and 175 nm, respectively [54].

4.2 Applications of CLSM
CLSM is currently used in many biological science disciplines, quantum optics, and nano-crystal
imaging. In this research, CLSM was used to estimate the total expression of the cytoskeletal
protein GFAP in astrocytes. In the healthy CNS, the level of GFAP expression in astrocytes may
22

not be detectable [27]. However, after injury to the CNS, the astrocytes are activated, or become
reactive, and the level of GFAP is known to be up-regulated. Therefore, to determine the level of
reactivity of astrocytes it is necessary to estimate the total GFAP expression within them. This
requires a microscopy technique that is able to illuminate entire cytoplasm and detect protein
levels in astrocytes individually. This can be achieved by CLSM. In this research, CLSM was
used to estimate the total GFAP expression of astrocytes cultured on standard tissue culture
surfaces and polyamide nanofibrillar scaffolds. Tubulin expression was also investigated. Cell
nuclei were also stained to determine cell density.

23

Chapter 5

Nanophysical Properties of Tissue Scaffolds

When injury sites are supplied with appropriate biomaterials that physically and biochemically
mimic the ECM, cells can regenerate and re-establish functional connections with surrounding
tissue. However, studies indicate that tissue regeneration requires a very high level mimicry of
nature [55].
The tissue scaffolding material under investigation in this research is a synthetic non-woven
electrospun polyamide nanofibrillar matrix that has demonstrated promise for the repair of the
injured spinal cord both in vivo and in vitro [37, 38]. A fundamental set of nanoscale properties
of nanofibrillar scaffolds serve as the primary cues for the re-establishment of cell systems. The
nanoscale cues for tissue scaffolds [56] were previously defined as surface roughness, surface
energy, elasticity, and growth factor coverage, which are discussed below.

5.1 Surface Roughness
Surface roughness of the tissue scaffolds can influence cell adhesion and fates [57]. Human
mesenchymal cells are more sensitive to exogenous signals when grown on surfaces with
periodic surface roughness [58]. Lipski et al. have demonstrated that glass substrates coated with
50- to 300-nm diameter silica nanoparticles influenced cell morphology and metabolism in
bovine aortic endothelial cells (BAECs) and mouse calvarial preosteoblasts (MC3T3-E1) [8].
24

These studies show that it is necessary to consider the surface roughness of the synthetic tissue
scaffolds. The surface roughness is the root mean square of the height values in a specific region
and is given by:

∑

√

(5.1)

where N is the total number of pixels in the region, Zi is the height value of the pixel, and Zave is
the average height value for AFM height data.
For AFM-based surface roughness measurements, if the region of interest is not rectangle,
such as nanofibers, the measurements need to be repeated many times. The comparison of SPRM
auto-track imaging versus traditional AFM imaging was demonstrated and histograms of various
scaffold surface roughness values were analyzed by Fan et al. [50]. This work demonstrated the
advantages of SPRM for surface roughness investigations of nanofibrillar scaffolds along
individual nanofibers.
The surface roughness investigations of polyamide nanofibrillar scaffolds and comparative
culture surfaces are presented in Chapter 6 of this thesis in combination with the astrocyte
morphological and protein expression responses to this property.

5.2 Local elasticity
Elasticity is the physical property of the material that returns into its original shape after the
deforming stress is removed. Stress is the force per unit area exerted on an object, and strain is

25

the relative deformation or change in shape and size of the object. If there is a linear relationship
between stress and strain, the material is in the linear elastic regime, e.g., a metal spring.
Stiffness is a measure of the resistance offered by an elastic object to deformation, which is:

(5.2)

where k is stiffness, F is the force applied to the object, and δ is the displacement produced by
the force.
The elasticity of the cellular environment regulates cell adhesion, cell spreading, and cell
differentiation [6]. Pelham and Wang [59] examined the responses of 3T3 cells to mechanical
properties of a collagen-coated polyacrylamide substrate, and found that cells on flexible
substrates had less spreading behavior and higher lamellipodial and motility activity than cells on
rigid surfaces. A recent study showed that the stiffness of the scaffold material influenced the
tissue bridging and axonal ingrowth for spinal cord injury [60]. These studies indicate that
elasticity is a significant physical parameter for a tissue scaffold design.
The elasticity range of the cellular environment varies with tissue type [61]. The range of
elasticity of cellular microenvironments is shown in Figure 5.1.
brain

fat

muscle

bone
Glass and
plastic tissue culture

[kPa] 0.1

1

10

100

Figure 5.1: Elasticities of cellular microenvironments. The glass and plastic tissue cultures are
stiffer than the stiffest bone tissue. [61].

26

The overall elasticity of the nanofibrillar scaffold that is sensed by the cell is dependent on
the elasticity of individual nanofibers and the mesh density. AFM can be used to measure the
elasticity of individual cells or culture surfaces by acquiring force curves on the surface [49].
To determine the elasticity of a material with AFM, force versus z-piezo displacement curves
from a tissue culture are collected using the force mode analysis in Nanoscope Software version
5.30r3.sr3. The deflection sensitivity is determined on a silicon wafer, which is known to be
stiffer than the planar culture surface [62]. The mean value of approximately ten measurements
from the wafer is determined. The force versus z-piezo displacement data are then converted to
force versus distance data, where distance is the gap between the atom at the far end of the AFM
probe and the sample. Using the cantilever stiffness of the AFM provided by the manufacturer
(0.58 N/m), the elasticity of tissue cultures was calculated from the retraction curve according to
the Derjaguin, Muller, and Toporov (DMT) model [63].

(5.3)

(

)

(5.4)

(5.5)

(

27

)

(5.6)

where δ is the indentation depth, a is the contact radius, R is the tip radius, Fn is the applied
force, Fad is the adhesion force, K is the total elastic modulus of the tip-substrate, W132 is the
work of adhesion required to separate the AFM tip (subscript 2) from the cell substrate (subscript
1) in air (subscript 3) [64], νtip is the Poisson’s ratio of the AFM tip, νsample is the Poisson’s ratio

for the sample, Etip is the Young’s modulus of the AFM tip, and Esample is the Young’s modulus
of the sample. The calculations were implemented in MATLAB. Examples of AFM force curves
that are relevant to these investigations are shown in Figure 5.2. These are: typical force curves
from a “soft” material (Figure 5.2 (a). and from a “harder” material with a large adhesive
component Figure 5.2 (b). All force curves investigated in this thesis resembled the second type,
with an actual example given in Figure 5.2 (c). This motivated the use of the DMT model for
extraction of the Young’s modulus.

0

z-piezo displacement
(nm)

(c)

0

z-piezo displacement
(nm)

Force (nN)

(b)

Force (nN)

Force (nN)

(a)

0

z-piezo displacement
(nm)

Figure 5.2: AFM force versus z-piezo displacement (F-z) curves including retraction and extension
parts. (a) Typical F-z curve of a soft material, (b) typical F-z curve of a material with large
adhesion component, and (c) F-z curve of PLL Aclar substrate.

28

To measure the elasticity of individual nanofibers, it is necessary to auto-track the
nanofibers for statistically reliable measurements. SPRM can directly interact with an individual
nanofiber without using a zoom box. The elasticity of individual tubules within a tissue scaffold
was investigated by SPRM [65]. In another work, the elasticity histograms of a series of
electrospun carbon nanofiber tissue scaffolds were performed with SPRM by Fan et al. [14].
The local elasticity of polyamide nanofibrillar scaffolds and comparative culture surfaces are
investigated and are presented in Chapter 6 of this thesis in combination with the astrocyte
morphological and protein expression responses to this property.

5.3 Contact angle
The surface energy is the free energy change

when the surface area of a medium is increased

by unit area. The contact angle of a liquid droplet can be used to determine the surface energy of
a material. The equilibrium contact angle

is determined from Young’s equation:

(5.7)

where

is the solid-vapor interfacial energy,

is the liquid-vapor interfacial energy, and

is the solid-liquid interfacial energy,

is the equilibrium contact angle. A schematic of a

liquid drop on a solid surface is given in Figure 5.3.

29

 LG
C

 SG

 SL

Figure 5.3: Schematic of a contact angle measurement. The contact angle
on interfacial energies between solid and liquid (
liquid and gas (

), solid and vapor (

depends
), and

).

Hallab et al. suggested that for 3T3 fibroblast cells the surface energy may be a more
important determinant of cell adhesion and proliferation [66], and may be more useful than
surface roughness for influencing cell adhesion and cell colonization on engineered tissue
scaffoldings (Figure 5.4).

Influence on cellular
adhesion strength

Surface energy

Surface roughness
Increasing roughness
Increasing surface energy
Figure 5.4: The relationship between surface roughness, surface energy, and their influence
on cell adhesion. [66].

The contact angle measurements of polyamide nanofibrillar scaffolds and comparative
culture surfaces are presented in Chapter 6 of this thesis in combination with the astrocyte
morphological and protein expression responses to this property.
30

5.4 Work of adhesion
The surface energy of a material can also be calculated by measuring the work of adhesion using
an AFM. Work of adhesion is the work needed to separate the AFM tip from the culture surface
in ambient air. The adhesive component of the AFM force curves can be extracted and used to
investigate the work of adhesion as previously given in Equation 5.5.
The work of adhesion of polyamide nanofibrillar scaffolds and comparative culture surfaces
are presented in Chapter 6 of this thesis in combination with the astrocyte morphological and
protein expression responses to this property.

5.5 Growth factor coverage
Growth factors are compounds that stimulate nearby target cells to grow and reproduce. Cells
can receive and respond to the growth factors produced by other cells in their surroundings.
Growth factors act as signaling molecules between cells. Cytokines and hormones are examples
of growth factors that bind to specific receptors on the surface of target cells.
Transforming growth factor-β, glial cell line-derived neurotrophic factor, leukaemia
inhibitory factor, and basic fibroblast growth factor are examples of growth factors that are
involved in the blood brain barrier. FGF-2 is produced by astrocytes and induces specific
functions in endothelial cells and astrocytes. It has proliferation and stellation effects on
astrocytes. Delgado-Rivera et al. have demonstrated that FGF-2-modified nanofibrillar scaffolds
are more permissive to neurite outgrowth than unmodified nanofibers and standard tissue culture
surfaces [10]. This study showed the potential of both nanofibrillar scaffolds and the growth
factors for neural tissue engineering.
31

Growth factor coverage characterization of tissue scaffolds can be obtained with AFM.
Tapping mode AFM images of unmodified and covalently cross-linked FGF-2 polyamide
nanofibrillar scaffolds are shown in Figure 5.5. The close-up images in (c) and (d) revealed the
increased surface roughness of the FGF-2 cross-linked nanofibers. The increased surface
roughness may be due to changes in both topography and surface chemistry of the modified
nanofibrillar scaffolds [10].
(a)

(b)

(c)

(d)

Figure 5.5: Tapping mode AFM images of unmodified polyamide nanofibers (a
and c) and nanofibers covalently cross-linked with FGF-2 (b and d). Images in (a)
and (b) are height and (c) and (d) are amplitude images. The difference in surface
roughness of FGF-2 modified nanofibers is apparent in amplitude images. Images
on same row have same scale bar. [10].

5.6 Effect of aging on nanofibrillar scaffolds
The physical properties of tissue scaffolds can change over time. This issue of tissue scaffold
shelf life, with possible changes in directive signals to cells, was studied by Ayres et al. [67]. The
32

aging behavior of polyamide material was tested by Roggendorf [68], who concluded that one
year after implantation, amorphous polyamide-6 sheets showed increased brittleness, crazing,
and crystallization of the material. The effect of aging on biomaterial properties is under
investigation by several groups [69]. The questions how scaffolds change over time, and how
these changes may induce the cellular responses, are crucial for tissue scaffold design.
The nanoscale physical properties of newly electrospun polyamide nanofibrillar matrices < 1
year old versus those that were > 3 years old were investigated with TEM, contact angle
measurements, and Raman spectroscopy. Significant differences in hydrophobicity and
chemistry were observed. The cell responses of cerebellar granular neurons (CGNs) to the
changed nanophysical properties of the new versus aged tissue scaffolds were also investigated
using AFM. CGNs were selected because this is a comparatively homogeneous CNS cell
population. In a healthy situation, CGNs respond by well-fasciculated (branched) neurite
(dendrites and axon) outgrowth. CGNs can also display a guidance response to nano-patterning
in their environment, including nanofibers [37]. CGN neuron outgrowth on the new versus aged
tissue scaffolds were shown to differ. A new result was that a strong CGN drive toward
obtaining a 3-dimensional environment, achievable with the aged nanofibrillar scaffolds, was
observed.

5.6.1 Experimental Procedures
5.6.1.1 Preparation of scaffold samples
The polyamide nanofibrillar scaffolds electrospun on Aclar substrates were purchased from
Donaldson Co., Inc. (Minneapolis, MN). The polyamide nanofibers were electrospun on Aclar
33

substrates for use as cell culture surfaces. Astrocyte cell cultures were prepared as previously
reported [10, 37]. The old and new nanofibrillar scaffolds were kept in dark storage at room
temperature for about 3 years and 1 year, respectively, except for experimental usage.

5.6.1.2 Preparation of cell cultures
Purity-enhanced cultures of primary CGNs were prepared from postnatal day 8 (P8) Sprague
2

Dawley rats and grown to confluence in 75 cm tissue culture flasks. The culture medium was
composed of Dulbecco’s Modified Eagle’s Medium (DMEM, Invitrogen, Carlsbad, CA) + 10%
fetal bovine serum (Invitrogen), 1% penicillin, and streptomycin, (Sigma-Aldrich, St. Louis,
MO) to which 25 mM KCl was added. CGNs were then subcultured in 0.5 mL medium at a
density of 30,000 cells/well directly on Aclar coverslips coated with nanofibers in 24 well plates.
Cells were fixed with paraformaldehyde at 24 h.

5.6.1.3 Analytical techniques
AFM height images of the neuron cell cultures were captured with a Nanoscope IIIA (Bruker
AXS Inc, Madison WI, formerly Veeco Metrology). TEM investigations of old and new
nanofibers were performed using both a JEOL 100CX TEM operated at 100 kV and a JEOL
FS2200 TEM operated at 200 kV (Japan Electron Optics Laboratories, Tokyo, Japan). The JEOL
FS2200 TEM was used to obtain the selected area electron diffraction (SAED) of nanofibers
with the camera length set to 60 cm. TEM sample preparation was as follows. For the plan-view
TEM, the nanofibers were gently detached from the nanofibrillar surface with tweezers and
suspended in 100% ethanol. The dispersion was then ultrasonicated for 30 sec (Branson
34

Ultrasonic Corporation, Danbury, CT). A droplet was placed on a lacey film coated TEM grid by
pipette and dried in ambient air. For the cross-section TEM, nanofibrillar matrix samples
electrospun onto 200 μm plastic coverslips were embedded in Poly/Bed 812 epoxy resin
(Polysciences) and polymerized for 24h at 60 °C. Thin sections (70 nm thickness) of embedded
nanofibers, were obtained using a MTX ultramicrotome (RMC, Boeckeler Instruments, Tucson,
AZ) and using glass knives. Glass knives were prepared with a Glass Knife Maker (RMC,
Boeckeler Instruments, Tucson, AZ). Thin sections were placed on 200 mesh copper grids and
positively stained with 2% uranyl acetate in 50% ethanol and lead citrate (Reynold’s
formulation). Positively stained samples resulted in high contrast TEM images [70].
Contact angle measurements of culture surfaces were acquired using VCA Optima contact
angle analysis equipment (AST Products Inc., Billerica, MA). One sample from each of five
randomly selected areas per culture was analyzed.
Raman spectroscopy investigation was performed with a Kaiser Optical Systems HoloProbe
Micro-Raman Spectrograph (Kaiser Optical Systems Inc., Ann Arbor, MI) coupled to an
Olympus BX-60 optical microscope (Olympus, Center Valley, PA). For surface enhanced
Raman spectroscopy (SERS), glass coverslips were coated with approximately 20-nm-thick gold
in an Emscope Sputter Coater model SC 500 (Ashford, Kent, England) purged with argon gas.
Old and new nanofibers were deposited by direct scraping with a clean razor blade to avoid
chemical contamination.

35

5.6.2 Experimental Results
5.6.2.1 TEM and SAED investigation of new and aged nanofibers
TEM investigation of individual nanofibers did not show obvious structural differences between
the new and aged nanofibers, e.g., the hollow, bamboo, and solid core structures described [14].
A light contrast interior with darker contrast walls suggestive of a hollow core structure was
observed for both, as shown in the representative examples of Figure 5.6 (a) and (b). A similar
conclusion of structural homogeneity was drawn from AFM investigation of new and aged
nanofibrillar matrix surfaces. Dark contrast inclusions (arrows, Fig. 5.6 (a) ) were also noted for
the aged nanofibers investigated in this study. SAED results indicated that many nanofibers both
new and aged were substantially crystalline and there were differences in the crystallinity
between the new and the aged nanofibers (Figure 5.6 (e) and (f)).
The contact angles of the new and aged nanofibrillar surfaces were measured on five
randomly selected regions. The new and aged nanofibers had 52.3 ± 2.2 ° and 63.3 ± 2.5 ° (mean
± standard deviation) contact angles, respectively. These results indicated that the polyamide
nanofibrillar scaffolds became less hydrophilic with age.

5.6.2.2 Raman spectroscopy investigation
SERS Raman investigation of seven regions of each sample showed a typical polyamide
spectrum with an Amide I: C=O stretch at about 1300 cm
at about 1630 cm

-1

-1

and an Amide III: C=O/C=N stretch

for both new and aged samples. A cis N-H peak at about 1400 cm

-1

, which

has previously been reported for polyamide nanofibers, was also observed for both samples.

36

However, full width at half maximum (FWHM) increases on the polyamide peaks and the
development of a broad disordered carbon peak [71] for the aged sample clearly indicated that
changes in chemical bonding had occurred over time. It was noted that the Amide I C=O stretch
appeared to be stronger in spectra from the aged nanofibrillar matrices. Representative spectra
from both new and aged nanofibrillar matrices are shown in Figure 5.6 (c) and (d).

37

Old (> 3 yr)

O
C

H
N

0
Overtones
1300

(b)

New

(c)
Intensity (AU)

Intensity (AU)

Old

Old

New (< 1 yr)

(a)

0

(d)

Cis
Amide III N-H
C=O/C=N 1440
1300 1630 Amide I
C=O

2600 cm-1

1300
New

(e)

2600 cm-1
(f)

Figure 5.6: TEM, Raman spectroscopy, and SAED results of aged ((a), (c), and (e)) and new
((b), (d), and (f)) polyamide nanofibers. No structural differences were evident in TEM
images of (a) aged and (b) new nanofibers. Arrows in the aged nanofibers (a) show dark
contrast inclusions. Scale bars, 100 nm. Raman spectroscopy, given in (c) and (d), showed
FWHM increases on the polyamide peaks and development of a broad disordered carbon
peak for the > 3 yr samples (c). AU: arbitrary units. Both SAED in (e) and (f) showed
demonstrated evidence of crystallinity within the nanofibers that were investigated.
Directionality in SAED of new nanofiber (f) further indicates potential crystal texture.
[67*].

38

5.6.2.3 AFM investigation of cerebellar granular neuron responses
High pass filtered AFM height images of CGNs are shown in Figure 5.7. The most frequently
observed result for CGNs cultured on the aged (> 3 yr old) scaffolds was that the neurons
burrowed beneath the scaffold (Figure 5.7 (a) and (b)). CGNs on aged scaffolds extended
neurites as shown in Figure 5.7 (a). In contrast, CGNs cultured on the newly electrospun (< 1 yr
old) scaffolds were clustered and lacked neurite outgrowth (Figure 5.7 (c)).

(a)

(b)

(c)

Figure 5.7: A composite AFM image (a) shows the burrowing response of a cerebellar granular
neuron (CGN) on aged (> 3 yr old) nanofibrillar scaffolds. The neurite length is approximately
0.5 mm. The close-up of the dashed box in (a) as shown in (b) shows that the neurons with
associated neurite outgrowth are underneath the nanofibers (arrows). Arrow heads show cell
debris or some CGN which lacked neurite. (c) Neurons cultured on new nanofibers clustered on
the top and lacked neurite outgrowth. Scale bars, 20 µm. [67*].

39

5.6.3 Discussion
This study demonstrated that the age of tissue scaffolds affects CGN migration and neurite
outgrowth. Careful investigation of the nanofibrillar layers on the Aclar surfaces, after the
burrowing response was observed, showed no obvious breaks. The neurons were therefore
strongly driven to reach a three-dimensional environment, either by penetrating through the
nanofiber meshes or by infiltrating between the nanofiber and the Aclar layers at their edges and
then migrating considerable distances (0.5-1.0 mm) to reach some of the regions investigated by
AFM. Contact angle measurements showed that there was a measurable difference in the surface
polarity of new and aged scaffolds. Raman spectroscopy results confirmed the degradation of the
basic polyamide chemistry over time. The dark contrast inclusions identified by arrows in Figure
5.5 (a) were noted in all of the aged samples. Similar inclusions were not observed in any of the
new samples.
The experimental results suggest that local crystallization may be present both in the aged
and new nanofibers (Figure 5.5). Facilitation of neuron growth cone development into axons on
hydrogel matrices with compliances comparable to those of brain tissue (100-400 Pa, considered
soft) has been reported [72]. It was further reported [72] that neuron single cultures grew only on
soft hydrogels, which suggests a strong response to the mechanical environment. In the present
study, healthy neuron morphologies with neurite outgrowth were observed in association with
the aged scaffolds while non-process-bearing morphologies were observed on the new scaffolds.
The properties of the aged nanofibrillar scaffolds evoked a response in neurons, which burrowed
into a more three-dimensional environment and showed neurite outgrowth.

40

Chapter 6

The Effect of Nanophysical Properties on Untreated and dBcAMPtreated Astrocyte Differentiation
In this chapter, the effect of tissue scaffold nanophysical properties on the untreated and
dBcAMP-treated astrocytes was investigated. The purpose of the untreated and dBcAMP treated
astrocytes were to obtain astrocytes like quiescent and reactive astrocytes in vivo. The promising
results of polyamide nanofibrillar scaffolds motivated the investigation of reactive-like astrocyte
differentiation in response to the external physical cues provided by the nanofibrillar scaffolds
and additional comparative culture surfaces. The four substrates were used in the present study:
(1) the electrospun polyamide nanofibrillar scaffolds and three comparative culture surfaces, (2)
poly-L-lysine (PLL) functionalized glass, (3) PLL functionalized Aclar, and (4) unfunctionalized
Aclar. These scaffold environments themselves present an array of potentially directive physical
properties.

6.1 Nanofibrillar scaffolds induce preferential activation of Rho GTPases in
cerebral cortical astrocytes

Astrocytes cultured on electrospun polyamide nanofibers whose nanophysical properties may
reproduce key aspects of native extracellular matrices have demonstrated promising results in
both in vitro and in vivo situations. In vitro, astrocytes cultured on nanofibrillar scaffolds
assumed morphologies that appeared to recapitulate those observed in native tissues.
41

Furthermore, autocrine up-regulation of FGF-2 was demonstrated by Western blot, along with
increased neurite outgrowth by co-cultured neurons [10]. In vivo, the same scaffolds introduced
into spinal cord wound sites promoted accelerated hindlimb recovery measured by standardized
observational scoring (Beattie, Basso, Breshnahan – BBB score) with aligned and fasciculated
axon development and re-vascularization throughout wound sites [37, 38]. Furthermore, low
levels of astrocytic scarring were observed at three and five weeks after injury in comparison to
injury-only controls, with reduced levels of glial fibrillary acidic protein, a marker for reactive
astrocytes, and chondroitin sulfate proteoglycans, a group of molecules expressed by both glial
and fibroblastic components of a glial scar and believed to be inhibitory to axonal regrowth [73,
74]. The in vitro and in vivo results suggest that nanofibrillar scaffolds could induce preferential
astrocyte differentiation leading to minimized glial scar formation. Attenuation of scarring may
be due to the nanofibrillar nanostructure that closely resembles features of naturally occurring
extracellular matrix/basement membrane scaffolds and may therefore cause the nanofibers to
look like ‘self’, or to the nanofibrillar chemistry, or to both. It is therefore important to
investigate the physical properties of the nanofibrillar environment and to elucidate their
directive effects on neural cell responses.
The present research explores the hypothesis that external physical cues of the nanofibrillar
scaffolds can trigger the initiation of specific signaling cascades with morphological
consequences. The external physical cues of surface polarity and surface roughness were
investigated in this study. Previous investigations suggested that surface polarity might be an
influential parameter of neural cell responses [75]. Because the macroscopic surface roughness
introduced by the nanofibrillar scaffolds differed significantly from previous investigations [76,
77], this parameter was also considered. Additionally, new nanoscopic investigations of surface
42

roughness were performed. Contact angle measurements were used to assess surface polarity and
AFM was used to investigate both macroscopic and nanoscopic surface roughness.
In this research, the morphological response of cerebral cortical astrocytes to nanofibrillar
scaffolds was investigated at high-resolution using AFM. The three dimensional capability of
AFM was also used to characterize cell spreading. An initial study of the corresponding
activation of GTPase upstream regulators was performed using immunocytochemistry. For each
immunolabel, phalloidin staining of F-actin was also performed to allow assessment of
colocalization with the cytoskeleton. The immunocytochemistry study focused on investigation
of the main GTPase regulators for the observed morphological responses: filopodia,
lamellipodia, and stress fiber formation, and stellation. Immunolabeling with Cdc42, Rac1, and
RhoA was performed as these are upstream regulators of filopodia, lamellipodia, and stress fiber
formation, respectively [78, 79]. Inhibition of the RhoA effector kinase (ROCK) through
depression of RhoA is one pathway that is associated with the cessation of stress fiber formation
and the onset of stellation [80].
As only external cues were used to trigger cell responses, astrocyte responses to nanofibrillar
surfaces were studied in comparison with their responses to three additional culture surfaces:
poly-L-lysine-functionalized planar glass (PLL glass), unfunctionalized planar Aclar (Aclar), and
PLL-functionalized planar Aclar (PLL Aclar). PLL glass is a standard astrocyte culture surface,
and astrocyte responses to it are well characterized [1]. The polyamide nanofibrillar scaffolds
were electrospun on Aclar substrates; therefore, astrocyte responses to Aclar surfaces were
investigated to distinguish responses to the nanofibrillar scaffolds from possible responses to the
underlying Aclar substrate. Astrocyte responses to PLL Aclar surfaces were studied to clarify the
role of the underlying substrate versus surface functionalization.
43

The results of the morphological investigations and immunocytochemistry experiments
consistently indicated that preferential activations of Rho GTPases were induced by the
variations in the physical environments. These were directive for specific cell morphologies and
responses.

6.1.1 Material and methods
6.1.1.1 Nanofibrillar Scaffolds
Polyamide (proprietary blend) nanofibers (median diameter ~180 nm) were electrospun from a
blend of two polymers [(C28O4N4H47)n and (C27O4.4N4H50)n] onto Aclar coverslips by
Donaldson Co., Inc. (Minneapolis, MN) [81]. The resulting non-woven polymeric nanofibrillar
scaffold was approximately 2 µm thick when measured on edge by optical microscopy. The
matrix comprised approximately ten layers of randomly oriented nanofibers and appeared to be
2

uniformly dense, with ~25 nanofiber segments per 5x5 µm scan area per surface layer when
measured by AFM [10] and no direct openings to the coverslip surface. Astrocytes could attach
to the nanofibrillar surfaces but not migrate into them. The nanofibrillar scaffolds in the present
study were not chemically modified.

6.1.1.2 Planar Culture Surfaces
Glass coverslips (12 mm, No. 1 coverglass, Fisher Scientific, Pittsburgh, PA) and Aclar
coverslips (Ted Pella, Redding, CA) were used as underlying planar substrates. Glass or Aclar
coverslips were placed in a 24-well tissue culture plate (one coverslip/well) and covered with 1

44

ml of PLL solution (50 μg PLL/ml in dH2O) overnight. The coverslips were then rinsed with
dH2O, sterilized with UV light, and used for the cultures. One set of Aclar coverslips was left
unfunctionalized.

6.1.1.3 Primary Astrocyte Culture
Purity-enhanced cultures of primary astrocytes were established as previously reported [10, 37,
38]. All procedures were approved by the Rutgers Animal Care and Facilities Committee
(IACUC Protocol #02-004). This includes a number of steps taken to assure the purity of the
primary astrocyte cultures. Briefly, rat cerebral cortical astrocytes were prepared from postnatal
2

day P2 Sprague Dawley rats and grown to confluence in astrocyte medium in 75 cm tissue
culture flasks. This enhances astrocyte purity because, after birth, the neurons in the cerebral
cortex are post mitotic and do not grow in tissue cultures (tissue for culturing cortical neurons
tissue is obtained from embryos). Astrocyte medium was composed of Dulbecco’s Modified
Eagle’s Medium (DMEM, Invitrogen, Carlsbad, CA) + 10% fetal bovine serum (Invitrogen), 1%
penicillin and streptomycin (Sigma-Aldrich, St. Louis, MO), which does not support growth of
primary, dissociated neurons in culture. Cultured astrocytes were then harvested with 0.5%
Trypsin/EDTA (Sigma-Aldrich) and re- seeded at a density of 30,000 cells/well directly on 12
mm Aclar or PLL Aclar coverslips, PLL glass coverslips, or on Aclar coverslips coated with
nanofibers in 24 well trays in astrocyte medium (0.5 ml), and then stained for F-actin, Cdc42,
Rac1 and RhoA. The sub-culturing processes further purified the primary cultures. Parallel
cultures were immunostained with GFAP, a marker for astrocytes, and more than 95% were
found to be GFAP positive.
45

6.1.1.4 Contact Angle Measurements
Contact angle measurements of culture surfaces were acquired using VCA Optima contact angle
analysis equipment (AST Products Inc., Billerica, MA). Two samples from each set of culture
surfaces were analyzed by five randomly selected measurements from each sample. Variations in
contact angle data among the culture surfaces were analyzed using ANOVA followed by
pairwise post hoc comparisons with Tukey’s test [82]. Significance levels were set at P < 0.05.

6.1.1.5 Optical Microscopy
Differential Interference Contrast (DIC) images (8 per substrate) using an Olympus Software
FV10-ASW Version 03.01.02.02 (Olympus, Center Valley, PA) were captured to ensure that
representative specimens were considered for AFM investigation. The 30,000 cell plating density
versus the cell adhesion density for the individual substrates was also investigated.

6.1.1.6 Atomic force microscopy
AFM images of individual astrocytes at 24 h were acquired using a Veeco Instruments
Nanoscope IIIA (Bruker AXS Inc, Madison WI, formerly Veeco Metrology) operated in ambient
air. Three scanners were used as noted in the text: a J scanner with 125x125x5.548 μm x-y-z
scan range, an E scanner with 13.5x13.5x3.08 μm x-y-z scan range and an A scanner with
1x1x0.66 μm x-y-z scan range. Both contact mode, using silicon nitride tips with a nominal tip
radius of 20 nm and cantilever spring constant k = 0.58 N/m, and tapping mode, using silicon
tips with a nominal tip radius of 10 nm and drive frequency of ~320 kHz, were used. The
closed-loop Multimode scanner (iCXY100Z15-MM, nPoint Inc., Middleton, WI) with
46

100x100x15 μm x-y-extended z scan range was used in order to investigate astrocyte
morphology on Aclar surfaces.
As the heights of the cellular processes investigated on the nanofibrillar scaffolds were on the
same order as the background nanofibers, it was not possible to clearly distinguish them in
conventional height, amplitude, or phase images. Frequency domain Gaussian high pass filtering
(GHPF) was therefore used to segment the cellular extensions from the nanofibrillar background
in height images [83]. AFM images of at least 30 astrocytes for each culture surface were
evaluated to ensure that the results were representative. Astrocytes with having stellate
morphologies were counted to obtain the percentage stellation for each cell culture used in this
work using the sum criteria as previously reported [84]. Astrocytes having at least one process
longer than the width of the cell soma were considered as stellate.
AFM surface roughness analysis of the scaffolds was performed by Nanoscope Software
2

2

version 5.30r3.sr3. Square 1x1 μm and 10x10 μm regions were randomly selected, and five
measurements were taken from each sample. The flatten command was used to remove the bow
of the AFM height images. Surface roughness data along individual nanofibers were collected
using SPRM [14, 51]. Variations in RMS surface roughness data among the culture surfaces
were analyzed using ANOVA followed by pairwise post hoc comparisons with Tukey’s test [82].
Significance levels were set at P < 0.05.

6.1.1.7 Antibody and F-actin staining of astrocytes
The astrocytes cultured on coverslips were fixed with 4% paraformaldehyde for 10 minutes,
permeabilized with 0.5% Triton X-100 for 5 minutes, and blocked with 10% normal goat serum
47

for 30 minutes. After removing the normal goat serum, the cells were double stained with one of
the primary antibodies mouse anti-Cdc-42 (1:200), mouse anti-RhoA (1:200), or mouse antiRac1 (1:25) (all from Cytoskeleton, Denver, CO)), and phalloidin (Invitrogen, Carlsbad, CA).
The incubation in the primary antibodies was done overnight at room temperature in a humidity
chamber. Coverslips were then washed with phosphate buffered saline (PBS) and stained with
goat anti-mouse Alexa 568 IgG secondary antibody (ex 579 nm, em 603 nm, Invitrogen) at a
1:500 dilution for 1 hour. Following the antibody staining the cells were rinsed, stained with
Phalloidin-488 (ex 498 nm, em 520 nm, Invitrogen) at a 1:100 dilution for 1 hour, mounted on
microscopic slides with GelMount (Biomeda, Foster City, CA), and observed under an Olympus
IX81 inverted microscope (Olympus, Center Valley, PA).

6.1.1.8 Image Capture Conditions for Immunocytochemistry Structural Investigation
A first set of six representative projection images for each substrate and immunolabeling
condition: Cdc42, Rac1, RhoA, each with corresponding phalloidin staining, were captured using
fluorescence microscopy, at scales and pixel resolutions that gave optimal representation of
structural features (72 cells total).

6.1.1.9 Image Capture Conditions for Protein Expression Estimate
A second set of maximum intensity level grayscale images were captured specifically for protein
estimation. These had identical scale bars and pixel resolutions as required for accurate
comparative analysis of activation intensity patterns. Five images (via 20X objective) were
captured for each of the three antibodies on each of the four culture surfaces. At this
48

magnification, each image typically continued several whole cells and a total of 240 cells were
investigated. Images of the astrocyte cultures were captured using Metamorph software
(Molecular Devices Sunnyvale, CA). For each condition, a minimum of five images was
captured from different regions of three coverslips to ensure that the results were representative
for the cultures as a whole. The amount of detected protein was quantified by selecting a
constant rectangular region and measuring the total fluorescence intensity. The mean background
intensities of the fluorescence images were calculated and then subtracted from the total
fluorescence intensity value. ImageJ software version 1.44p was used for the quantitative
analysis of light intensity. Intensity data were analyzed using ANOVA followed by pairwise post
hoc comparisons with Tukey’s test. Significance levels were set at P < 0.05.

6.1.1.10 Analysis of Cytoskeletal Colocalization
Colocalization analysis of RhoA and F-actin was performed on the first set of images using
Olympus Software FV10-ASW Version 03.01.02.02 (Olympus, Center Valley, PA). A Pearson
Coefficient (PC) measure of the colocalization of two proteins was calculated as [54]

∑(
√∑ (

̅̅̅̅̅) (
̅̅̅̅̅) (

̅̅̅̅̅)

(6.1)

̅̅̅̅̅)

for six examples from each of the three combinations. F1,i is average brightness of wavelength

λ1 at i-th pixel, F2,i is average brightness of wavelength λ2 at i-th pixel,

49

̅̅̅̅

∑

is average brightness of wavelength λ1 and

̅̅̅̅̅

∑

is average brightness of

wavelength λ2.
Prior to the calculation of each PC, the background was removed by setting the threshold
level manually subsequent to cell segmentation. Individual values were used to calculate a mean
value PC ± standard deviation. All results met two separate reliability criteria: a mean PC value
> 0.5 and a standard deviation of less than 25%.

6.1.2 Results
6.1.2.1 Contact Angle Measurement Results of Culture Surfaces
The hydrophobicity of tissue cultures has been suggested as a possible material property for
reduced astrocyte growth [76]. In this work, the hydrophobicity of the PLL glass, nanofibrillar
scaffolds, Aclar, and PLL Aclar surfaces were characterized by contact angle analysis (Figure 6.
1).

50

Contact Angle (degrees)

90
80
70
60
50
40
30
20
10
0

*

Aclar

Nanofibrillar
scaffolds

PLL glass

PLL Aclar

Figure 6.1: Contact angle measurements of culture surfaces. Aclar was significantly less
hydrophilic than nanofibrillar scaffolds, PLL glass and PLL Aclar. No significant differences
were identified from pairwise comparisons among the PLL Aclar, PLL glass, and
nanofibrillar scaffolds. [12*].
The contact angle was significantly different among the different surfaces (ANOVA,
P=0.0002). Post hoc pairwise comparisons with Tukey's test demonstrated that the contact angle
on Aclar surfaces was significantly different than each of the other surfaces (max P=0.017).
Aclar surfaces were significantly less hydrophilic than nanofibrillar scaffolds, PLL glass, and
PLL Aclar surfaces.
6.1.2.2 AFM Surface Roughness Results
Surface roughness measured with AFM showed that two scales of surface roughness were
present on the nanofibrillar surfaces: a large overall RMS roughness and a much smaller RMS
51

roughness along the nanofibers (Figure 6.2). The overall nanofiber roughness was more than an
order of magnitude greater than the roughness of any of the other surfaces (ANOVA, followed
by post hoc pairwise comparisons with Tukey's test, all P<0.0001). A separate analysis followed
by pairwise comparisons of the surface roughness among PLL glass, PLL Aclar, and Aclar
surfaces indicated that each was different than the other (max P=0.0111).
SPRM [14, 51] was used to investigate the surface roughness along individual nanofibers.
SPRM AFM data was obtained by raster-scan auto-tracking along individual nanofibers. The
RMS surface roughness was calculated on each pixel based on a local neighborhood region, after
curvature removal by Kasa circle fit [85]. The local neighborhood region was a rectangular box
around each pixel whose box size was set to be close to the nanofiber diameter. Multiple sets of
overlapping surface roughness information were generated, with the provision that any box that
extended outside the nanofiber boundaries was automatically truncated. Based on a sampling of
over 1000 data points from multiple nanofibers, the mean value of the RMS surface roughness
along the individual nanofibers was midway between the values for PLL glass and Aclar. The
AFM surface roughness results therefore indicated that the nanofibrillar surfaces had
significantly higher RMS surface roughness at the macroscopic level but lower RMS surface
roughness at the cell receptor level with values that were comparable to those of PLL glass.
Representative high-resolution A-scanner AFM height images of the four surfaces are shown in
Figure 6.2 (a-d) and the mean values for RMS surface roughness ± standard deviation are shown
in Figure 6.2 (bottom).

52

(a)

(b)

(c)

(d)

(e)
100

RMS surface roughness (nm)

90
80
70
60
50
40
30

12
10
8
6
4
2
0

Along
Glass Aclar
Along PLL
PLL Glass
Aclar
Nanofibers
Nanofibers

PLL
PLLAclar
Aclar

20
10
0

Nanofibrillar Along
Along Glass+PLL
PLL Glass
Nanofibrillar
Scaffolds
Scaffolds Nanofibers
Nanofibers

Aclar
Aclar

PLL Aclar
Aclar+PLL

Figure 6.2: AFM surface roughness investigation of culture surfaces. High-resolution AFM
deflection images of (a) nanofibrillar scaffold, (b) PLL glass, (c) Aclar, and (d) PLL Aclar
by an A scanner. Scale bars, 250 nm. RMS surface roughness measurement results (e) of
nanofibrillar scaffolds, along nanofibers, PLL glass, Aclar, and PLL Aclar. The mean RMS
surface roughness measurements showed that nanofibrillar scaffolds had the highest surface
roughness but surface roughness along individual nanofibers was comparable to the planar
surfaces (inset (e)). Error bars show the standard deviation. [12*].

53

6.1.2.3 Optical Microscopy Survey of Astrocyte Morphology and Cell Density
Differential Interference Contrast (DIC) images (8 per substrate) were captured to ensure that
representative specimens were considered for AFM investigation. The results shown in Figure
6.3 indicated that relatively homogeneous populations of astrocytes with different morphologies
were present on each substrate. The 30,000 cell plating density versus the cell adhesion density
2

for the individual substrates was also investigated. The mean value results in cells per mm were:
PLL glass: 86; nanofibrillar scaffolds: 56; Aclar: 90; and PLL Aclar: 91. It is noted that DIC and
brightfield images were unable to resolve the cell morphologies of astrocytes on nanofibrillar
scaffolds (Figure 6.3 (b)). This is because the cellular edges and processes are approximately the
same order in height as the background nanofibers, ~100 to 200 nm. As discussed in Chapter 7,
this issue was resolved for corresponding AFM images. The adhesion density for astrocytes on
nanofibrillar scaffolds was therefore estimated from GHPF AFM images.

54

(a)

(b)

(c)

(d)

Figure 6.3: DIC images of astrocytes cultured on (a) PLL glass, (b) nanofibrillar scaffolds,
(c) Aclar, and (d) PLL Aclar. Scale bar, 100 µm. [12*].

6.1.2.4 Astrocyte Responses to Poly-L-lysine-functionalized Glass Surfaces
Different stages of attachment and growth were observed simultaneously at 24 h, as shown in the
representative seven-member group shown in Figure 6.4 (a). Most cells had lamellipodial
vertices, suggesting activation of Rac [86]. Fan-shaped astrocytes were also observed; this is a
morphological indicator of a migratory phenotype via lamellipodial leading edges [87]. Cell
55

clusters were observed, such as shown in the upper left. Potentially apoptotic cells were
occasionally observed, as shown in the lower right. Close-ups of areas marked by arrows in
Figure 6.4 (a) show details of: (b) a lamellipodial migration leading edge, with stress fibers
evident; (c) filopodial anchor spikes; and (d) interacting lamellipodia between two cells.
AFM profile measurements of cell soma typically showed well-delineated nuclear and
cytoplasmic regions. Representative section measurements across cells (along the dotted lines of
Figure 6.4 (a)) are shown in Figure 6.4 (e-f). Nuclear regions in these desiccated samples had
maximum heights of 0.5-1.0 μm.

56

b

(a)

c

(b)

e
(c)

d

(d)
f

3
2
1
0

3
2
1
0

(e)
0

20

40 60 80

(f)
0

20

40 60 80

Figure 6.4: AFM astrocyte morphology investigation on PLL glass. (a) AFM high
pass filtered composite height image of astrocytes by J scanner show lamellipodia
formation (scale bar, 20 µm). (b–d) Close-ups by E scanner of arrow regions a–d
shown in (a): stress fibers (b); filopodia anchors (c); and cell–cell interaction (d)
(scale bars, 2 µm). (e and f) Cross-section profiles of cells e and f in (a). Axis units
are in µm with astrocyte soma extent identified by red arrow. [12*].

57

6.1.2.5 Astrocyte Responses to Nanofibrillar Scaffolds
Astrocyte responses to the nanofibrillar scaffolds were different from those observed for PLL
glass. Astrocytes cultured on nanofibrillar surfaces typically had stellate morphologies, as shown
in the composite AFM images of Figures 6.5 (a) and 6.6 (a). Also, for most cells at 24 h, one
vertex had developed into a process that traversed most or all of the distance to a neighboring
cell as shown in Figure 6.5 (a). AFM section measurements of cell somata showed welldelineated nuclear regions about 1-2 µm in height with elevated perinuclear regions [84].
Representative profiles taken along the dotted lines in Figure 6.5 (a) are shown in Figure 6.5 (bd).

58

3

(a)

(b)

2
1
0

b

0

20

40

60

3

80

(c)

2
1

c

0
0

20

40

60

3

(d)

2

d

80

1
0
0

20

40

60

80

Figure 6.5: AFM astrocyte morphology investigation on nanofibrillar scaffolds. (a)
AFM GHPF composite height image of astrocytes cultured on nanofibrillar surfaces
by J scanner. Scale bar, 20 µm. (b–d) Cross section profiles of cells b–d in (a). Axis
units are in µm with astrocyte soma extent identified by red arrow. [12*].

Process-mediated cell-cell interactions appeared to result from contacts between neighboring
cells. Multiple examples of process growth were studied with the following observations.
Processes typically tapered to ~1 µm wide and ~100 nm tall at their narrowest points and then
developed broadened tips (~8-15 μm wide) within about 5 μm of a neighboring cell, as shown by
the interacting cell pair in Figure 6.6. Ribbon-like elongations from the broadened tips formed
the actual physical contact to the neighboring cell. In the example shown in Figure 6.6, the
ribbon-like structures are visible on the top surface of a neighboring cell (boxed region Figure
6.6 (a) and its close-up (b)); in other cell-cell interactions, they extended across an open gap to
59

contact a cell at their tips. Interdigitating, 10 nm-diameter villar-like structures were typically
observed near the contacted region as shown in Figure 6.6 (c), which is a close-up image of the
boxed region in Figure 6.6 (b).

(a)

(b)

(c)

Figure 6.6: AFM high pass filtered composite height images of astrocytes cultured on
nanofibrillar scaffolds. (a) Dashed box identifies the intercellular connection. (b) Close-up of
the intercellular connection by E scanner. Dashed box identifies a contacted area. (c) Close-up
of a contacted area by E scanner. The scale bars show 10, 2, and 0.5 µm in (a), (b), and (c),
respectively. [12*].

60

6.1.2.6 Astrocyte Responses to Unfunctionalized Aclar Planar Surfaces
Astrocyte responses to the Aclar surfaces were different from those observed for both PLL glass
and nanofibrillar scaffolds. Extensive networks of astrocyte processes about 100-150 nm in
diameter were observed between astrocytes, as shown in the composite AFM image of Figure 6.7
(a). The three dimensional AFM investigation demonstrated that the cell somata were not well
spread, as shown in the representative section measurements of Figure 6.7 (b-d), taken along the
dotted lines in Figure 6.7 (a). Maximum somata heights frequently exceeded 3 μm, requiring use
of an extended z-range scanner.

61

3

(a)

(b)

2
1
0
0

20

40

60

3

(c)

2

b

80

1
0
0

20

40

60

3

c

d

80

(d)

2
1
0
0

20

40

60

80

Figure 6.7: (a) AFM GHPF composite height image of astrocytes cultured on Aclar surfaces
by J scanner. The cross sections shown by b–d in (a) were given in corresponding section
profiles. Scale bar shows 20 µm and axis units are in µm. [12*].

6.1.2.7 Astrocyte Responses to PLL functionalized Aclar Planar Surfaces
PLL functionalization of Aclar surfaces induced astrocyte behavior similar to that observed for
cells cultured on PLL glass. Astrocytes were once again observed in clusters as shown in Figure
6.8 (a) and were well spread. The typical heights of the well-defined nuclear regions were also
similar, 0.5-1.0 µm, as shown in Figure 6.8 (b-c). The chief difference between the astrocyte
responses to PLL Aclar versus PLL glass was the increased cell edge interaction with the
substrate. High resolution AFM investigation of cell edges revealed branched structures that
62

appeared to infiltrate along the valleys created by the surface roughness, as shown in Figure 6.8
(d).

(a)

(b)

(c)

3

(d)

2
1
0
0

20

40

60

80

Figure 6.8: AFM investigation of astrocytes cultured on PLL Aclar surfaces. (a) Composite
AFM deflection image of astrocytes by J scanner. Scale bar, 20 µm. (b) AFM deflection image
of an astrocyte by J scanner. Scale bar, 10 µm. Dotted arrow identifies cross section. Dashed
box identifies a cell edge. (c) Corresponding cell edge showing nanoscale filopodia by E
scanner. Scale bar, 2 µm. (d) Corresponding cross section profile. Axis units are in µm with
astrocyte soma extent identified by red arrow. [12*].

6.1.2.8 AFM-based Assessment of Astrocyte Stellation Response
A stellation response, defined as in Cox et al. as the development of a process longer than the
cell body, was assessed based on AFM images of 35-45 cells per each culture surface. Use of the
63

GHPF as described in Tiryaki et al. enabled accurate assessment of process lengths on
nanofibrillar surfaces [83]. The stellation responses based on GHPF AFM height images or on
AFM deflection images are shown in Table 6.1. The highest percentage of stellate cells was
found on the nanofibrillar scaffolds.
Table 6.1: Quantitative astrocyte stellation response results. [12*].

Cell substrate

# Cells

# Stellate cells

% Stellate

PLL glass

45

25

56

Nanofibrillar
scaffolds

44

31

70

Aclar

35

20

57

PLL Aclar

36

15

42

6.1.2.9 Immunocytochemistry
Rac1, Cdc42, and RhoA are Rho GTPase family members that are involved in actin and
microtubule dynamics [88, 89]. Cdc42 produces filopodial extensions [78], Rac1 induces
lamellipodia formation and membrane ruffling [86], and RhoA causes actin bundles and focal
adhesions in non-neuronal cells [90]. Kalman et al. have shown that Rac1 and RhoA GTPases
induce growth of hippocampal astrocyte processes [91]. The four key morphologies observed in
varying proportions on the different surfaces using AFM suggested preferential Rac1, Cdc42 and
RhoA involvement. The expression of these GTPase proteins, as well as F-actin organization,
was investigated with immunocytochemistry and phalloidin staining, respectively.

64

6.1.2.10 Immunocytochemistry Structural Investigation
An immunocytochemistry structural investigation was performed to investigate whether protein
expression was generally consistent with the observed AFM morphology results. Six
representative images for each substrate and immunolabeling condition were captured and
studied, with typical results shown in Figure 6.9. From analysis of the phalloidin results,
astrocytes on PLL glass displayed prominent stress fibers. Astrocytes on nanofibers displayed a
stellate response with obviously less stress fiber formation. Astrocytes on PLL Aclar consistently
displayed extensive filopodia formation. Astrocytes on Aclar appeared to display less filopodia
formation; however, the full extent of the branched networks shown in Figure 6.9 consistently
required AFM for observation. The phalloidin results of the immunocytochemistry structural
investigation therefore generally supported the results of the AFM morphology study.
Analysis of the immunolabeling results suggested up-regulation of RhoA for astrocytes on
PLL glass versus down-regulation of RhoA for astrocytes on nanofibers. It also indicated
possible differences in the sequestration of the protein expression, i.e. whether the greatest
intensity was observed near the cell periphery or the cell interior. A semi-quantitative protein
expression estimate and a co-localization analysis were therefore performed to address these two
points.

65

Aclar

PLL Aclar

Phalloidin

Cdc42

PLL glass

Nanofibrillar
scaffolds

Phalloidin

Rac1

(a)

Phalloidin

RhoA

(b)

(c)
Figure 6.9: Immunocytochemistry structural investigation results. Left to right: fluorescence
images of astrocytes cultured on PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar
substrates. Immunolabeling and corresponding phalloidin are represented in red and green,
respectively. Astrocytes were labeled with (a) anti-Cdc42, (b) anti-Rac1, and (c) anti-RhoA.
Scale bars, 10 µm. [12*]. Fluorescent images by I. Ahmed, Shreiber laboratory with analysis in
this thesis.
66

6.1.2.11 Protein Expression Estimate
A protein expression estimate based on maximum intensity level images was performed. A
second set of maximum intensity level grayscale images was captured specifically for
quantification analysis and estimation. These had identical scale bars and pixel resolutions and
were therefore accurate for the comparative analysis of activation intensity patterns. Twenty
whole cell images (via 20X objective) for each of the three antibodies co-stained with phalloidin
on each of the four culture surfaces were investigated using maximum intensity levels (240 cells
total). The ImageJ software [92] free-hand tool was used to individually outline the area of each
cell. ImageJ software was further used to calculate a mean intensity defined as the total intensity
divided by the selected area. The results were compiled into an average mean intensity with
standard error for each antibody type on each culture surface. The average mean intensities of the
maximum intensity levels were used to investigate the activation intensity patterns for each
surface.
The activation intensity patterns for Cdc42, Rac1, and RhoA are shown in Figure 6.10. For
each surface, significant differences were observed in the intensity of immunostaining among
Cdc42, RhoA, and Rac1 (ANOVA, max P = 0.023), which regulate actin dynamics and cell
adhesion. For astrocytes cultured on PLL glass, post hoc analysis with Tukey's test revealed that
the intensity of RhoA labeling was significantly greater than Cdc42 (P = 0.0005) and Rac1 (P =
0.0289). Labeling of Cdc42 and Rac1 were not statistically different (P = 0.366). For astrocytes
cultured on nanofibers, the intensity of Cdc42 immunolabeling was significantly greater than
RhoA (P = 0.0154). No significant differences were detected between Cdc42 and Rac1 (P
=0.663) or Rac1 and RhoA (P = 0.120).

67

For astrocytes cultured on Aclar, Cdc42 immunolabeling was significantly greater than
Rac1 (P=0.041). No differences was observed in labeling of Cdc42 and RhoA (P=0.995). Large
differences were observed between Rac1 and RhoA expression on Aclar, but the P-value
following Tukey's test (P=0.051) did not reach the significance threshold of P<0.05. For cells
cultured on PLL Aclar, the intensity of each immunolabel was significantly different from the

Mean intensity

other two (max P= 0.001).

PLL glass

Nanofibrillar
scaffolds

Aclar

PLL Aclar

Figure 6.10: The white, gray, and black bars show the Cdc42, Rac1, and RhoA intensities,
respectively. The data is based on maximum intensity level images. Bars show the mean,
and error bars show standard error of n = 20 cells. [12*].

The results for astrocytes on PLL glass shows RhoA > Rac1 > Cdc42. This result is
consistent with the AFM morphology and immunocytochemistry structural investigations, and
indicated that RhoA activation had occurred by 24 h. The results for astrocytes on nanofibrillar
scaffolds were reversed: Cdc42 > Rac1 > RhoA. The depression of RhoA relative to Rac1 is
68

consistent with the stellate formation of the astrocytes, as RhoA negatively regulates stellation
[93]. The activation intensity results for astrocytes on Aclar and PLL Aclar are similar to each
other but different from the PLL glass and nanofiber patterns, with Rac1 was depressed relative
to both Cdc42 and RhoA: Cdc42 > Rac1 < RhoA.

6.1.2.12 Colocalization Analysis
The possible colocalization of actin with RhoA was investigated, using six 100X magnification
fluorescence microscopy images of leading edges. Nanofiber auto fluorescence apparent in the
Cdc42 channel was minimized prior to PC calculation as discussed in the Materials and Methods
section.
Investigation of the colocalization of RhoA with F-actin yielded reliable results for astrocytes
cultured on the nanofibrillar and PLL glass surfaces, with PCs equal to 0.656 ± 0.153 and 0.584
± 0.096, respectively. While both exhibited partial colocalization, its character appeared to be
different.

Scatter plots obtained from astrocytes on the nanofibrillar surfaces consistently

showed a weak tendency toward negative colocalization, which was supported by the appearance
of the corresponding colocalization images. Scatter plots obtained from astrocytes on the PLL
glass surfaces consistently showed partial colocalization but tended to be noisy. The slopes of
the scatter plots were also different, astrocytes on the nanofibrillar surface has shown higher
intensity of RhoA whereas the PLL glass surfaces has indicated higher intensity of F-actin
(Figure 6.11).

69

(b) 250

Intensity

200

Red

(a)

150
100

50
0

Green
50 100 150 200 250

Intensity

(d)

250

Intensity

200

Red

(c)

150
100
50

Green

0 50 100 150 200 250

Intensity

Figure 6.11: Overlaid fluorescence images of astrocytes cultured on (a) nanofibrillar scaffolds
and (c) PLL glass labeled with phalloidin (F-actin) and anti-RhoA. The fluorescence was
represented in green for F-actin and red for RhoA in (a) and (c). The scatter plots of F-actin and
RhoA for nanofibrillar scaffolds (b) and PLL glass (d) showed partial colocalization with a
mean PC of 0.66 and 0.58 respectively. Scale bar in (a) and (c) show 10 µm and 20 µm,
respectively. [12*]. Fluorescent images by I. Ahmed, Shreiber laboratory with colocalization
analysis in this thesis.

70

6.1.3 Discussion
Investigation of the morphological responses of cerebral cortical astrocytes to the four different
culture surfaces demonstrated a series of responses, which involved combinations of filopodia,
lamellipodia, and stress fiber formation, and stellation. The three-dimensional nature of AFM
imaging demonstrated that differences in cell spreading were involved. Preferential activation of
Rho GTPase upstream regulators known to be associated with the morphological responses was
demonstrated using immunocytochemistry. The observed responses were induced by the four
different culture surfaces without the introduction of any chemical agents. The results therefore
support the hypothesis that the extracellular environment can trigger preferential activation of
members of the Rho GTPase family with demonstrable morphological consequences for cerebral
cortical astrocytes.
The maximum intensity level images of immunolabeling results of Cdc42, Rac1, and RhoA
indicated that preferential activation of Cdc42 compared with RhoA occurred for the astrocytes
on the nanofibrillar surfaces, while the opposite, preferential activation of RhoA compared with
Cdc42, occurred for the astrocytes on the PLL glass. Partial colocalization of RhoA and F-actin
was also demonstrated by astrocytes on the PLL glass. The standard interpretation of the
immunolabeling results for PLL glass would be that at 24 h, the astrocytes on the PLL glass
surfaces had formed contractile stress fibers [94]. The morphological response observed by the
AFM and phalloidin imaging showed prominent stress fiber formation, consistent with the
standard interpretation. In contrast, astrocytes on the nanofibrillar surfaces showed a different
morphological response. The astrocytes on nanofibrillar scaffolds displayed the highest
percentage of stellate response, with long processes instead of the lamellipodia/filopodia
observed on the PLL functionalized surfaces. This is consistent with previously reported results
71

[1, 10, 37, 38, 72]. There was also a difference in the typical cell body shape, with vertices
instead of a rounded, lamellipodial morphology. Using GHPF AFM, it was also shown that by 24
h direct cell-cell interaction by long processes with specialized contacting structures had
occurred. The immunocytochemistry results indicated that RhoA was depressed relative to
Cdc42. This would be consistent with stellation induced by inhibition of the RhoA effector
kinase ROCK through depression of RhoA [80]. It has been reported that stellation can proceed
via either protein kinase C (PKC) or cyclic AMP dependent protein kinase A pathways [84].
Additional investigations suggest that RhoA is an effector of the PKC signaling pathway [93]. A
possible hypothesis consistent with the present results is that stellation was associated with
inhibition of Rho kinase [80] achieved through depression of RhoA levels, potentially with
alteration to the PKC cycle [93].
As no chemical agents were used, the differences in the responses were attributed to
differences in the physical environments. The contact angle measurements demonstrated that the
nanofibrillar scaffolds had surface polarities that were about 10% more hydrophobic than those
of the PLL glass and PLL Aclar surfaces. This enables some of the known properties of PLL
functionalization to be extrapolated to the unfunctionalized nanofibrillar scaffolds.

PLL

functionalization is known to add positive charge to culture surfaces. The net charge behavior of
cell membranes is dominated by negatively charged sialic-acid-containing glycoproteins [95,
96]. Therefore, the expected interaction between the astrocytes and the PLL functionalized
surfaces would be to maximize the Coulomb interaction, resulting in an increase of the cellsubstrate interaction area and spreading of the cell body; and this was the effect that was
observed. Astrocytes on the nanofibrillar surfaces displayed less cell spreading, while astrocytes
on the Aclar surfaces, about 30% more hydrophobic than the PLL surfaces, were characterized
72

by a marked lack of cell spreading. These results and others [97] suggest that astrocytes may be
fairly sensitive to the charge characteristics of their environments, which is a hypothesis that can
be tested in future work.
The surface roughness variations considered in the present work were those that existed
“naturally” in the four culture environments. These measurements demonstrated that two scales
of surface roughness were present on the nanofibrillar surfaces: a large-scale RMS roughness
between nanofibers and a small-scale RMS roughness along the nanofibers. The RMS roughness
along the individual nanofibers was close in value to that of the PLL glass surfaces. Therefore,
one distinguishing physical property for the nanofibrillar surfaces was the large-scale component
of their surface roughness. This suggests that multiple receptor types may be involved in the
astrocyte response to nanofibrillar scaffolds.
Astrocyte responses to the PLL glass and PLL Aclar surfaces appeared to reflect the
differences in their surface roughness. The AFM surface roughness measurements demonstrated
that significant differences in RMS surface roughness existed between the PLL glass and PLL
Aclar surfaces although similarity in their surface polarity, confirmed by the contact angle
measurements, was achieved through PLL functionalization. The AFM results indicated that
there was a stronger stimulation of filopodia formation on the PLL Aclar surfaces although the
cell clustering and cell spreading behavior was similar for the astrocytes on the PLL glass and
PLL Aclar surfaces. The AFM images further demonstrated that the astrocyte filopodia appeared
to infiltrate the valleys on PLL Aclar (Figure 6.8 (c)). The immunocytochemistry results
indicated that Cdc42 activation was significantly greater than Rac1 activation for the PLL Aclar
surfaces, which is consistent with a filopodia response. The Cdc42 activation was significantly
less than Rac1 activation for the PLL glass surfaces. These results suggest a possible link
73

between the nanoscale surface roughness and activation of Cdc42 with consequent filopodia
formation as an exploratory response.
Surface roughness may be a meaningful parameter for certain response aspects rather than a
dominant factor for an overall response in astrocytes. As the latter may be an aggregate of all
meaningful parameters, not just one, the nanoscale influence of surface roughness should
continue to be investigated.

6.1.4 Conclusion
Morphological differences among astrocytes cultured on PLL glass, nanofibrillar scaffolds,
Aclar, and PLL Aclar surfaces were demonstrated with AFM and phalloidin staining of F-actin.
The differences and similarities of the morphological responses were consistent with differences
and similarities of the surface polarity and surface roughness of the four surfaces investigated in
this work. An initial quantitative immunolabeling study further identified significant differences
in the activation of Rho GTPases. Astrocytes cultured on nanofibrillar scaffolds showed a
response that included stellation, cell-cell interactions by stellate processes and evidence of
depression of RhoA. The astrocyte response to the nanofibrillar scaffolds was unique among four
surfaces studied. The results support the hypothesis that the extracellular environment can trigger
preferential activation of members of the Rho GTPase family with demonstrable morphological
consequences for cerebral cortical astrocytes, and suggest that the large-scale component of the
nanofibrillar scaffold surface roughness (a form of nano-patterning) is a possible environmental
trigger.

74

6.2 Differentiation of reactive-like astrocytes cultured on nanofibrillar and
comparative culture surfaces
Repair of traumatic injury to the CNS remains a challenging problem. One of the hurdles that
must be overcome is the glial scar, which is established after traumatic injury in mammalian
systems and creates a barrier to regeneration of axons [98, 99]. A glial scar consists mainly of
reactive astrocytes and proteoglycans. Reactive astrocytes at a wound site form chain-like
clusters, and also undergo a morphological change, extending interwoven processes [100].
Furthermore, they express tenascins [101], and inhibitory chondroitin (C)- and keratan (K)sulfate

proteoglycans

[102].

Uninterrupted

expression

of

myelinating

proteins

by

oligodendrocytes adds to the blockage. The result is a permanent scar that biomechanically and
biochemically blocks axonal elongation and reconnection [30].
The effect of a physical environment on cell adhesion and differentiation has recently been
recognized, especially in neural cells. Collectively, studies of cell responses to nano-patterning
[12, 103, 104, 105, 106, 107], elasticity [72, 108, 109, 110, 111], surface roughness [5, 77], and
surface polarity [75] suggest that controlling aspects of the physical environment holds potential
for inducing preferential differentiation of reactive astrocytes into non-inhibitory pathways.
During the last decade, promising results have been obtained using nanofibrillar scaffolds [10,
112, 113] and hydrogels [114] for regenerative repair of traumatic CNS injuries. The scaffold
environments themselves present an array of potentially directive physical properties.
The promising results of polyamide nanofibrillar scaffolds motivated the investigation of
reactive-like astrocyte differentiation in response to the external physical cues provided by the
nanofibrillar scaffolds and additional comparative culture surfaces. Cerebral cortical astrocytes, a
relatively uniform population, were treated with dibutyryladenosine cyclic monophosphate
75

(dBcAMP), and used to investigate the responses of reactive-like astrocytes to external physical
cues. Astrocytes treated with dBcAMP develop morphologies that resemble the reactive
astrocytes [115]. The same four substrates used in the recent investigation of untreated cerebral
cortical astrocytes responses to surface polarity and surface roughness variations [12] were used
in the present study: (1) the electrospun polyamide nanofibrillar scaffolds and three comparative
culture surfaces, (2) poly-L-lysine (PLL) functionalized glass, (3) PLL functionalized Aclar, and
(4) unfunctionalized Aclar. This enabled investigation of the differences between the untreated
and dBcAMP-treated astrocyte responses to the same culture environments and properties. In the
present study, the local elasticity and surface energies of the substrates were investigated and the
results were added to the previous investigations of surface polarity and surface roughness. The
physical property trends were then compared with the morphological and protein expression
responses.
Morphological differentiation of dBcAMP-treated astrocytes was investigated using AFM
and immunocytochemistry. GFAP is an intermediate filament protein thought to be involved in
the structure and function of the astrocyte cytoskeleton although its function is not well
understood. Up-regulation of GFAP is typically considered as an indication of increased
reactivity. Quantification of GFAP expression was performed using the confocal laser scanning
microscopy (CLSM) z-series. Astrocyte stellation is a tubulin-dependent rearrangement of
cytoskeleton [84]. Tubulin expression was also quantified using confocal z-series.
Immunolabeling of dBcAMP-treated astrocytes with the Rho GTPase upstream regulators for
filopodia, lamellipodia, and stress fiber formation, namely Cdc42, Rac1, and RhoA [78], was
also performed, enabling comparisons of these cytoskeletal proteins with the previously reported

76

results for untreated astrocytes [12]. The protein expressions were quantified using maximum
intensity level images acquired by fluorescence microscopy.
The study was organized as follows. Investigations of the physical properties of each culture
environment, assessed in terms of local elasticity, work of adhesion, RMS surface roughness,
and surface polarity, are presented first. The results indicated that the culture surfaces, including
the nanofibrillar scaffolds, occupied distinct ranges of physical property parameter space.
Investigations of the astrocyte responses, assessed in terms of morphology and protein
expression, are presented next. The morphological response and Rho GTPase expression for the
dBcAMP-treated astrocytes are also compared with those of untreated astrocytes [12].
Quantitative analyses for morphological and protein expression estimates are given. Results from
investigations of two proteins associated with astrocyte reactivity, GFAP and tubulin, are then
presented. The GFAP and tubulin investigations were performed exclusively for the dBcAMPtreated astrocytes.
The results indicate that the nanophysical cues of the extracellular environment can modulate
the reactive response of dBcAMP-stimulated cerebral cortical astrocytes, and that different cues
could have more directive importance for astrocytes in scarring mode versus in untreated mode.
DBcAMP-treated astrocytes cultured on the nanofibrillar scaffolds showed a unique Rho GTPase
and morphological non-response. Cdc42, Rac1, and RhoA expressions and consequent
morphologies were almost unchanged from their untreated values while the Rho GTPase
expressions and morphologies of the three comparative culture surfaces were greatly changed.
Furthermore, dBcAMP-treated astrocytes cultured on the PLL Aclar surfaces exhibited responses
that were reminiscent of glial scar formation. The comparison of dBcAMP-treated and untreated
astrocyte responses suggested that different nanophysical cues may have more directive
77

importance in a wound-healing situation. The elasticity property was indicated as potentially
more directive for dBcAMP-treated astrocytes than for untreated astrocytes [116].

6.2.1 Materials and Methods
6.2.1.1 Nanofibrillar Scaffolds
Polyamide nanofibers (median diameter ∼180 nm) were electrospun from a blend of two
polymers [(C28O4N4H47)n and (C27O4.4N4H50)n] onto Aclar coverslips by Donaldson Co, Inc
(Minneapolis, MN). The resulting non-woven nanofibrillar scaffold was approximately 2 µm
thick when measured by optical microscopy [12].

6.2.1.2 Planar Culture Surfaces
Glass coverslips (12 mm, No 1 coverglass; Thermo Fisher Scientific, Waltham, MA) and Aclar
coverslips (Ted Pella, Inc., Redding, CA) were used as underlying substrates. The substrates
were placed in a 24-well tissue culture plate and covered with 1 mL of PLL solution (50 µg
PLL/mL in dH2O) overnight. The coverslips were then rinsed with dH2O, sterilized with
ultraviolet light, and used for the cultures. One set of Aclar coverslips was left unfunctionalized
[12].

78

6.2.1.3 Primary dBcAMP-treated astrocyte cultures
The cerebral hemispheres of new born Sprague Dawley rats (postnatal day 1 or 2) were isolated
aseptically. The cerebral cortices were dissected out, freed of meninges, and collected in HBSS.
They were minced with sterile scissors and digested in 0.1% trypsin and 0.02% DNase for 20
min at 37°C. The softened tissue clumps were then triturated by passing several times through a
fine bore glass pipette to obtain a cell suspension. The cell suspension was washed twice with
culture media (DMEM+10% FBS) and filtered through a 40 μm nylon mesh. For culturing, the
2

cell suspension was placed in 75 cm flasks (one brain/flask in 10 mL growth medium) and
incubated at 37ºC in a humidified CO2 incubator. After 3 days of incubation the growth media
was removed, cell debris was washed off, and fresh medium was added. The medium was
changed every 3-4 days. After reaching confluency (about 7 days), the cultures were shaken to
remove macrophages and other loosely adherent cells. To obtain reactive-like astrocytes,
0.25mM dBcAMP was added to the culture medium and the serum concentration was reduced to
1%. The cultures in dBcAMP containing medium were incubated for 7-8 more days with a media
change every 3-4 days. The morphology of the cells was observed on alternate days under a
phase contrast microscope. In the control cultures, the cells were fed with DMEM + 1% FBS
(without dBcAMP). All procedures were approved by the Rutgers Animal Care and Facilities
Committee (IACUC Protocol #02-004).
After completing dBcAMP treatment, reactive-like astrocytes were then harvested with
0.25% Trypsin/EDTA (Sigma-Aldrich) and re-seeded at a density of 30,000 cells per well
directly on 12 mm Aclar or PLL Aclar coverslips, PLL glass coverslips, or on Aclar coverslips
coated with nanofibers in 24-well plates in astrocyte medium containing dBcAMP (0.5 mL),

79

fixed with paraformaldehyde and then stained for F-actin, Cdc42, Rac1, and RhoA. Parallel
cultures were immunostained with GFAP, an identification marker for astrocytes, and more than
95% were found to be GFAP-positive. Another set of parallel cultures were immunostained for
GFAP/β-tubulin.

6.2.1.4 AFM
AFM images of astrocytes were captured using a Veeco Instruments Nanoscope IIIA (Bruker
AXS Inc, Madison, WI; formerly Veeco Metrology) operated in ambient air. A J scanner with
125 µm × 125 µm × 5.548 µm x-y-z scan range and an E scanner with 13.5 µm × 13.5 µm × 3.08
µm x-y-z scan range were used. The AFM was operated in contact mode silicon nitride tips with
a nominal tip radius of 25 nm and cantilever spring constant k = 0.58 N/m (Nanoprobe SPM tips,
Bruker AXS Inc, Madison, WI; formerly Veeco Metrology). Frequency domain Gaussian high
pass filtering (GHPF) was used to segment the astrocytes from the substrate backgrounds in
AFM height images [83]. AFM images of at least 20 astrocytes for each culture surface were
evaluated to ensure that the results were representative.

6.2.1.5 Astrocyte Cell Shape Index
Cell soma investigations by AFM indicated that the reactive-like astrocytes had comparable cell
spreading. Therefore, the appropriate cell shape index (CSI) is defined as the ratio of the
perimeter squared to the two-dimensional cell area [117]:

(6.2)
80

where P is the cell perimeter and A is the cell area. This definition describes stellation as a
departure from CSI = 1 for a circular cell. The cell perimeter and area were calculated following
manual segmentation of cells from GHPF AFM height images. The procedure was implemented
in MATLAB version 7.7.0 (R2008b) using Image Processing Toolbox (The MathWorks).

6.2.1.6 Astrocyte Soma Height Measurements
Untreated and dBcAMP-treated astrocyte maximum soma height values were measured using the
section analysis of Nanoscope Software version 5.30r3.sr3 and NanoScope Analysis 1.10
(Bruker AXS Inc, Madison, WI; formerly Veeco Metrology). Twenty cells from each set of
culture surfaces were analyzed by section profile through the nucleus region. Variations in
maximum soma height data among the culture surfaces were analyzed using analysis of variance
(ANOVA) followed by pairwise post hoc comparisons with Tukey’s test [82]. Significance
levels were set at P < 0.05.

6.2.1.7 Astrocyte Process Measurements and Number of Astrocyte Processes
Untreated and dBcAMP-treated astrocyte cell process lengths were measured based on AFM
height and GHPF AFM height images. Twenty cells from each set of culture surfaces were
analyzed by measuring the length of processes manually, and the number of processes of each
astrocyte was counted. An extension longer than the diameter of a cell soma was considered to
be a process. Variations in astrocyte process length and number of process data among the
culture surfaces were analyzed using ANOVA followed by pairwise post hoc comparisons with
Tukey’s test [82]. Significance levels were set at P < 0.05.
81

6.2.1.8 AFM Force Curves of Culture Surfaces
The local elasticity of the polyamide nanofibrillar scaffolds and comparative culture surfaces
were investigated by AFM force curves as described in Chapter 5.2 of this thesis.

6.2.1.9 Immunolabeling for Cdc42, Rac1, RhoA, and F-actin
The astrocytes cultured on coverslips were fixed with 4% paraformaldehyde for 10 minutes,
permeabilized with 0.5% Triton X-100 for 5 minutes, and blocked with 10% normal goat serum
for 30 minutes. After removing the normal goat serum, the cells were double stained with one of
the primary antibodies mouse anti-Cdc-42 (1:200), mouse anti-RhoA (1:200), or mouse antiRac1 (1:25) (all from Cytoskeleton, Denver, CO), and phalloidin (Life Technologies, Grand
Island, NY). The incubation in the primary antibodies was done overnight at room temperature in
a humidity chamber. Coverslips were then washed with phosphate buffered saline (PBS) and
stained with goat anti-mouse Alexa 568 IgG secondary antibody (ex 579 nm, em 603 nm; Life
Technologies) at a 1:500 dilution for 1 hour. Following the antibody staining, the cells were
rinsed, stained with Phalloidin-488 (ex 498 nm, em 520 nm; Life Technologies) at a 1:100
dilution for 1 hour, mounted on microscopic slides with GelMount (Biomeda, Foster City, CA),
and observed under an Olympus IX81 inverted microscope (Olympus, Tokyo, Japan).

6.2.1.10 Image Capture Conditions for Cdc42, Rac1, and RhoA expression estimate
Maximum intensity level images were captured specifically for protein expression estimation
using an Olympus IX81 inverted microscope (Olympus). The images had identical scale bars and
pixel resolutions as a requirement for an accurate comparative analysis of activation intensity
82

patterns. Eight images (via 20× objective) were captured for each of the three antibodies on each
of the four culture surfaces to ensure that the results were representative for the cultures as a
whole. At this magnification, each image typically had several whole cells, and a total of 240
cells were investigated. Images of the astrocyte cultures were captured using Metamorph
software (Molecular Devices LLC, Sunnyvale, CA). The amount of detected protein was
quantified by segmenting each cell manually and measuring the average fluorescence intensity.
The average background intensities of the fluorescence images were calculated, and subtracted
from the average fluorescence intensity value. ImageJ software version 1.46r (National Institutes
of Health, Bethesda, MD) was used for the quantitative analysis of light intensity. Intensity data
were then analyzed using ANOVA followed by pairwise post-hoc comparisons with Tukey’s
test. Significance levels were set at P < 0.05.
6.2.1.11 Immunolabeling for GFAP and Tubulin
The astrocytes cultured on coverslips were fixed with 4% paraformaldehyde for 10 minutes,
permeabilized with 0.5% Triton X-100 in PBS for 5 minutes, and blocked with 10% normal goat
serum for 30 minutes. After removing the normal goat serum, the cells were double stained with
one of the primary antibodies and phalloidin. Rabbit anti-GFAP (Dako, Carpinteria, CA) was
diluted 1:500 while mouse anti-β-tubulin (Sigma-Aldrich, St. Louis, MO) was diluted 1:200. The
incubation in the primary antibodies was done overnight at room temperature in a humidity
chamber. Coverslips were then washed three times with PBS and stained appropriate secondary
anti-body either goat anti-rabbit Alexa 568 IgG (Life Technologies) or goat anti-mouse Alexa
568 IgG (Life Technologies). Both secondary anti-bodies were diluted 1:500 and incubation was
for 1 hour at room temperature. Secondary anti-body staining coverslips were rinsed three times
with PBS. Following PBS rinse, the coverslips were stained with Alexa Fluor® 488 Phalloidin
83

(Life Technologies) diluted 1:100 for 1 hour, rinsed again with PBS as above, and stained with
DAPI (Life Technologies) for 5 minutes. After completing staining, coverslips were mounted on
microscopic slides with GelMount (Biomeda), and observed under an Olympus FluoView 1000
Laser Scanning Confocal Microscope system attached to an Olympus IX81 automated inverted
microscope platform.
6.2.1.12 Quantitative GFAP and Tubulin expression estimate
Confocal z-series images were captured using an Olympus FluoView 1000 Laser Scanning
Confocal Microscope system attached to an Olympus IX81 automated inverted microscope
platform equipped with a 40× oil immersion objective (N.A.=1.3). The increments between the
z-series were set to 1.13 µm in order to avoid oversampling of fluorophores. The z-series were
acquired under identical imaging conditions, including the same high voltage (HV) and offset
settings, same objective, and same resolution (1024×1024 pixels). The HV and offset levels were
set to a level so that the pixels were not saturated. This is crucial to ensure the accuracy of the
relative intensity comparisons. Images were analyzed using MATLAB version 7.7.0 (R2008b)
Image Processing Toolbox (The MathWorks) and ImageJ version 1.46r (National Institutes of
Health). The maximum intensity projection images were obtained from the z-series, the
boundaries of the cells were determined manually, and then the total intensity and the number
pixels were calculated by using the z-series and cell boundaries. The average background
intensity values were calculated for each z-series because the autofluorescence was different on
different culture surfaces. The average background intensity was then multiplied by the number
of pixels of z-series, and subtracted from the total intensity. Finally, the total amount of GFAP or
tubulin expression / cell was determined. At least 50 cells were analyzed for each culture surface.
Variations in GFAP and tubulin expression estimation data among the culture surfaces were
84

analyzed using ANOVA followed by pairwise post hoc comparisons with Tukey’s test [33].
Significance levels were set at P < 0.05.

6.2.2 Results

6.2.2.1 Local Elasticity Investigation of Culture Environments and Parameter Space
Analysis
Reported investigations [72, 108, 109] have identified local elasticity as potentially directive.
Local elasticity along individual nanofibers was characterized in the present work. AFM force
curve measurements along individual nanofibers were taken and interpreted using the DMT
elasticity interpretative model for the Young’s modulus that included adhesion forces. Analysis
of the Young’s modulus results using both median-value box plots and mode-value distribution
plots are shown in Figure 6.12.
Mode value analysis indicated that the Young’s modulus comparison for the culture surfaces
was

(6.3)

at room temperature and for interaction with an AFM tip. Both median and mode-value analyses
indicated that the PLL glass substrate presented the most uniform elastic environment. Median
versus mode value analysis suggested that PLL functionalization could result in a less uniform
elastic environment for PLL Aclar, indicated by the emergence of tail values in the distribution.

85

The median value/mode value analysis further indicated that substrate elasticity dominated over
the chemical functionalization effect since unfunctionalized and PLL functionalized Aclar had
similar elasticity variances (shown as percentile boxes and lines) and distributions while PLL
functionalized glass and PLL functionalized Aclar had different results for both.

Figure 6.12: Elasticity of culture surfaces by AFM using DMT model for Young’s modulus.
(a) Young’s modulus box plot: Solid lines show the median, and the box edges show the 25th
and 75th percentiles. Histogram distribution plots of Young’s modulus for (b) PLL glass, (c)
PLL Aclar, (d) Aclar and (e) nanofibrillar scaffolds. Dashed lines show the mode values in
(b), (c) and (d) in descending order. Unit x axis for (b), (c), (d), and (e) is 100 MPa. PLL G,
PLL A, ACL, and NFS denote, PLL glass, PLL Aclar, Aclar and nanofibrillar scaffolds
respectively. [117*].

86

The nanofiber median value Young’s modulus was higher than that of PLL glass. However,
mode value analysis showed that the median value resulted from its wide range of Young’s
modulus values. For the nanofibers, therefore, a median value analysis may not accurately
represent the nanofiber elastic property. Specific investigation indicated that the higher elasticity
values did not correspond to nanofiber crossings. However, the wide range of elasticity values
may be due to the presence of hard and soft components in individual nanofibers, indicated by
previously reported transmission electron microscopy analysis [70]. The wide range of elasticity
values along individual nanofibers was accurately reflected in the percentile box and lines,
compared with those for PLL glass, PLL Aclar, and Aclar.

6.2.2.2 Surface Energy Investigation and Parameter Space Analysis
Use of the DMT interpretive model further enabled investigation of the surface energies of the
cell substrates by measuring the work of adhesion, W132, between the tip and each culture
surface [64]. W132 is the work that is needed to separate AFM tip from the culture surface in
ambient air. The adhesive component of the AFM force curves was extracted, and used to
investigate the work of adhesion.
Analysis of the work of adhesion using both median-value box plots and mode-value
distribution plots, shown in Figure 6.13, revealed several features. First, despite the similar
functionalization of PLL glass and PLL Aclar, the work of adhesion median value was higher for
PLL Aclar (Figure 6.13 (a)). This resulted from the different distributions: the PLL glass
distribution was single mode (Figure 6.13 (b)) with a narrow distribution and consistent median
2

and mode values at about 0.16 J/m The Aclar distribution (Figure 6.13 (c)) was also single
87

2

mode with consistent median and mode values at about 0.13 J/m . The widest range of work of
adhesion values was observed for the nanofibrillar scaffolds (Figure 6.13 (d)). The majority were
low work of adhesion values, which was reflected in the 0.119 J/m

2

median value and

2

indications of a mode value at about 0.1 J/m (Figure 6.13 (a) and (d)). The PLL Aclar
distribution also displayed a range of work of adhesion values (Figure 6.13 (e)). Comparison of
Aclar and PLL Aclar (Figure 6.13 (c) and (e)) indicated that this range resulted from the PLL
functionalization. This suggests that the combination of chemical functionalization with surface
roughness can result in higher work of adhesion, possibly through PLL conformation changes.

88

Figure 6.13: Work of adhesion characterization of culture surfaces. (a) Work of adhesion box
plot. Solid lines show the median, and the box edges show the 25th and 75th percentiles.
Histogram distribution plots of adhesion of (b) PLL glass, (c) Aclar, (d) nanofibrillar scaffolds
and (e) PLL Aclar. Dashed lines show the mode values in (b), (c) and (d) in descending order.
PLL G, ACL, NFS and PLL A, denote PLL glass, Aclar, nanofibrillar scaffolds and PLL
Aclar, respectively. [117*].

The data ranges for the work of adhesion and RMS surface roughness [12] were analyzed as
a two-dimensional parameter space as shown in Figure 6.14 (a). The results indicated that the
culture surfaces, including the nanofibrillar scaffolds, occupied distinct ranges of physical
property parameter space. Comparison of the work of adhesion and RMS surface roughness
89

(Figure 6.14 (a)) versus surface polarity [12] and surface roughness data ranges (Figure 6.14 (b))
suggested that the more local work of adhesion environment could present variations that would
not be evident in the more macroscopic environment assessed by contact angle measurements.

Figure 6.14: Two-dimensional parameter space representation of work of adhesion and surface
roughness of culture surfaces. (b) Two-dimensional parameter space representation of surface
polarity and surface roughness of culture surfaces. PLL G, NFS, ACL and PLL A denote PLL
glass, nanofibrillar scaffolds, Aclar and PLL Aclar, respectively. [117*].
90

6.2.2.3 Astrocyte Responses: Morphology Investigation
6.2.2.3.1 Cell density
Cell density on different cell substrates was measured by counting the nuclei from at least 22
DAPI-stained images per culture surface. Initially, cell plating density on all surfaces was 66
2

2

cells/mm . The dBcAMP-treated astrocyte densities for each substrate in cells/mm are given in
Table 6.2.
Table 6.2: Reactive-like astrocyte cell density measurement results (mean±SEM). [117*].

Cell density
[#cells/mm2]

Aclar

PLL glass

PLL Aclar

Nanofibrillar
scaffolds

117 7

189 24

202 17

206 14

Cell density on Aclar was low compared to other substrates, indicating that the cell adhesion
and/or proliferation on Aclar surfaces were negatively influenced by the properties of this
substrate. The low cell density on Aclar may be due to its lower hydrophilicity resulting in lower
cell adhesion [12]. The higher cell densities on the PLL functionalized surfaces were possibly
because of increased cell adhesion by the positive charges introduced to the substrate by
functionalization [118]. The highest cell density was observed on the nanofibrillar scaffolds,
which indicated that these surfaces were able to successfully modulate cell adhesion and/or
proliferation without requiring PLL functionalization.

91

6.2.2.3.2 Astrocyte morphology investigation by AFM
DBcAMP-treated astrocyte morphologies and astrocyte-astrocyte interactions by AFM provided
details of the different morphological responses to the culture surfaces investigated in this work.
The dBcAMP-treated astrocytes cultured on PLL glass surfaces developed astrocyte-astrocyte
interactions via long processes that suggested contact spacing [119] (Figure 6.15 (a)). Extensive
nanoscale process formation was also observed (inset). The nanoscale processes were about ~1050 nm in diameter and were observed around both processes and somata.
The dBcAMP-treated astrocytes cultured on nanofibrillar scaffolds developed long processes
in a stellate pattern (Figure 6.15 (b)). Astrocyte-astrocyte interactions suggestive of contact
spacing were observed. Extensive cell clustering was not observed.
The dBcAMP-treated astrocytes cultured on Aclar surfaces developed 2-3 directional
processes (Figure 6.15 (c)). Astrocyte-astrocyte interactions were few in number; when present,
they were suggestive of contact spacing. As astrocyte-astrocyte interactions were not
representative for this culture, AFM images of two individual cells that displayed commonly
observed morphologies are shown in Figure 6.15 (c).
The dBcAMP-treated astrocytes cultured on PLL Aclar surfaces also developed long
processes, but astrocyte-astrocyte interactions consisted of intertwined processes rather than
contact spacing. This resulted in chain-like cell clusters reminiscent of glial scar formation
(Figure 6.15 (d)).

92

(a)

(b)

PLL G

NFS

(c)

(d)

ACL
PLL A
Figure 6.15: Morphology investigation of reactive-like astrocytes with AFM height GHPF
images. (a) Astrocytes cultured on PLL glass, with close-up from dashed box region inset.
The close up image shows the nanoscale process formation. (b) Astrocytes cultured on
nanofibrillar scaffolds. (c) Representative morphologies for astrocytes cultured on Aclar. (d)
Astrocytes cultured on PLL Aclar shows glial-scar like chain formation. PLL G, NFS, ACL,
and PLL A, and denote PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar, respectively.
Scale bars, 20 µm. [117*].

93

6.2.2.3.3 Astrocyte quantitative morphology assessment and comparison
The morphological responses for the dBcAMP-treated astrocytes were quantified as process
length, the number of processes, and maximum soma heights using measurements taken from the
AFM height images.

A cell shape index was also calculated from perimeter and area

measurements taken from the AFM height images.

These measures were investigated for

statistically significant differences using Tukey’s test pairwise comparisons with P < 0.05.
Pairwise comparisons were made for the dBcAMP-treated astrocytes, untreated astrocytes, and
dBcAMP-treated versus untreated astrocytes cultured on the same four culture surfaces. A total
of 20 astrocytes per each culture were selected randomly from both dBcAMP-treated and
untreated astrocyte samples. The results, shown in Figure 6.16, are displayed as mean-value bar
graphs with standard error of the mean.
Analysis of process lengths is shown in Figure 6.16(a). For dBcAMP-treated astrocytes
alone, post hoc analysis with Tukey’s test revealed that the process lengths of astrocytes cultured
on PLL glass were significantly longer than the ones on nanofibrillar scaffolds (P < 0.0001) and
Aclar (P = 0.0246). The astrocyte process lengths on PLL Aclar were significantly longer (P =
0.0088) than the ones on nanofibrillar scaffolds. No significant differences were detected
between PLL glass versus PLL Aclar (P = 0.512), PLL Aclar versus Aclar (P = 0.4293), and
Aclar versus nanofibrillar scaffolds (P = 0.4058).
For untreated astrocytes alone, post hoc analysis with Tukey’s test revealed that the process
lengths of astrocytes cultured on PLL glass were significantly longer (P = 0.0166) than the ones
on Aclar. Other pairwise comparisons, PLL glass versus PLL Aclar (P = 0.292), PLL glass
versus nanofibrillar scaffolds (P = 0.3556), nanofibrillar scaffolds versus Aclar (P = 0.5694),

94

nanofibrillar scaffolds versus PLL Aclar (P = 0.9996), and PLL Aclar versus Aclar (P = 0.6254)
were not significant.
For dBcAMP-treated versus untreated astrocytes, the process length statistical analysis
showed that dBcAMP-treated astrocyte process lengths were significantly longer (max P =
0.00369) than untreated ones except for the astrocytes on nanofibrillar scaffolds (P = 0.28561).
As shown in Figure 6.16 (a), the greatest process length increases were observed for the
astrocytes cultured on the PLL functionalized surfaces, while the smallest increase was observed
for astrocytes cultured on the nanofibrillar scaffolds.
Analysis of the number of processes is shown in Figure 6.16 (b). No significant differences
in the number of processes were observed for dBcAMP-treated astrocytes alone, untreated
astrocytes alone, or for comparisons of dBcAMP-treated versus untreated astrocytes.
Analysis of astrocyte maximum soma height investigation, shown in Figure 6.16 (c), was
performed to investigate the cell spreading behavior. No significant difference in the maximum
soma height of reactive-like astrocytes alone was observed (min P = 0.3477), which justified use
of the two-dimensional CSI [120] analysis shown in Figure 6.16 (d). For untreated astrocytes
alone, Tukey’s all pairs comparison of untreated astrocytes showed that the maximum soma
height of astrocytes cultured on Aclar was significantly higher than the ones of PLL Aclar
(P<0.0001) and PLL glass (P < 0.0001). The maximum soma height of astrocytes on
nanofibrillar scaffolds was significantly higher than the ones of PLL Aclar (P < 0.0001) and PLL
glass (P < 0.0001). For reactive-like versus untreated astrocytes, the pairwise comparisons were
all significant (max P = 0.00652), and indicated contraction for reactive-like astrocytes cultured

95

on Aclar and nanofibrillar scaffolds and swelling for the ones cultured on PLL Aclar and PLL
glass.
Analysis of the two-dimensional CSI is shown in Figure 6.16 (d). A two-dimensional CSI
uses a ratio of area to perimeter to describe stellation as a departure from CSI = 1 for a circular
cell [40, 41]. The pairwise comparisons of reactive-like astrocytes alone showed the CSI of
reactive-like astrocytes cultured on nanofibrillar scaffolds was significantly less than ones of
PLL Aclar (P = 0.0055) and PLL glass (P < 0.0001). For untreated astrocytes alone, the
statistical results of CSI analysis showed that no significant difference exists (min P = 0.0611).
For reactive-like versus untreated astrocytes, the CSI increased on all surfaces (max P = 0.033)
except Aclar (P = 0.6), which showed a decrease. The most significant CSI change was observed
for reactive-like versus untreated astrocytes cultured on PLL Aclar (P = 0.0037).

96

Figure 6.16: Quantitative morphology assessment of reactive-like and quiescent astrocytes
cultured on PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar. Gray bars: reactive-like
astrocytes, White bars: quiescent astrocytes. (a) Average length of astrocyte processes. (b)
Average number of processes. (c) Average maximum soma height. (d) Average CSI results.
Error bars show standard error of n=20 cells. PLL G, NFS, ACL, and PLL A, and denote PLL
glass, nanofibrillar scaffolds, Aclar, and PLL Aclar, respectively. [117*].

6.2.2.4 Astrocyte Responses: Protein Expressions

6.2.2.4.1 GFAP and Tubulin Staining Results
Wide area GFAP staining images of the reactive-like astrocytes are shown in Figure 6.17. The
observed astrocyte-astrocyte interactions differed significantly as a function of culture surfaces
and cell density. Reactive-like astrocytes cultured on PLL glass surfaces (Figure 6.17 (a)),
nanofibrillar scaffolds (Figure 6.17 (b)), and Aclar surfaces (Figure 6.17 (c)) were typically not
clustered, while the reactive-like astrocytes on the PLL Aclar surfaces (Figure 6.17 (d))

97

displayed pronounced astrocyte-astrocyte interactions that appeared to recapitulate the chain-like
clustering of glial scar formation.

Figure 6.17: Wide area GFAP staining images of reactive-like cerebral cortical astrocytes in
vitro. Maximum intensity projection images of reactive-like cerebral cortical astrocytes
cultured on (a) PLL glass, (b) nanofibrillar scaffolds, (c) Aclar and (d) PLL Aclar. Scale bar,
100 µm. [117*].

Morphological and biochemical differentiation of reactive-like astrocytes was investigated by
GFAP and tubulin staining. Representative GFAP, tubulin, and overlay images with nuclei

98

staining images of reactive-like astrocytes are shown in Figure 6.18. Overlay images of GFAP
and tubulin consistently indicated an interaction between GFAP and tubulin.

NFS

ACL

PLL A

GFAP Tubulin Nuclei

Tubulin

GFAP

PLL G

Figure 6.18: Close-up GFAP, tubulin, and DAPI staining images of reactive-like cerebral
cortical astrocytes. Maximum intensity projection fluorescence images of astrocytes cultured on
PLL glass, nanofibrillar scaffolds, Aclar and PLL Aclar. First, second and third rows show the
GFAP, tubulin, and overlaid images of GFAP, tubulin, and nuclei. GFAP, nuclei, and tubulin,
were represented in green, blue, and red, respectively. Scale bar, 50 µm. [117*].

6.2.2.4.2 GFAP and Tubulin Quantification Results
The total GFAP and tubulin expression of reactive-like astrocytes was investigated by confocal
microscopy z-series. Both total GFAP and tubulin expression were modulated by the culture

99

surfaces. The results, shown in Figure 6.19, are displayed as mean-value bar graphs with
standard error of the mean.
The astrocytes cultured on PLL Aclar had highest expressions of both GFAP (Figure 6.19
(a)) and tubulin (Figure 6.19 (b)). Reactive-like astrocytes cultured on PLL glass displayed
comparatively low expression of GFAP but the second highest expression of tubulin. The
reactive-like astrocytes cultured on nanofibrillar scaffolds had comparatively higher expression
of GFAP but a lower expression of tubulin. Reactive-like astrocytes cultured on Aclar expressed
the least amounts of tubulin and the second least amounts of GFAP.
For each surface, significant differences were observed in the total intensity of GFAP among
astrocytes cultured on nanofibrillar scaffolds, Aclar, PLL Aclar, and PLL glass. Tukey’s all pairs
comparison results revealed that the total GFAP intensity of astrocytes cultured on PLL Aclar
was significantly higher than PLL Glass (P < 0.0001) and Aclar (P = 0.0386). Tukey’s all pairs
comparison also showed that GFAP intensity on nanofibrillar scaffolds was significantly higher
(P = 0.007) than PLL glass, whereas other pairwise comparisons were not significant.
Significant differences were observed in the intensity of tubulin among astrocytes cultured on
the cell substrates. Tukey’s all pairs comparison results showed that all pairwise comparisons
were significant (max P = 0.0349) except nanofibrillar scaffolds versus Aclar.

100

Figure 6.19: Quantitative total GFAP and tubulin expression estimation results for reactivelike astrocytes. Culture surfaces modulate the total GFAP (a) and tubulin (b) expression in
reactive-like astrocytes. Error bars show standard error of n=50 cells. PLL G, NFS, ACL,
and PLL A, and denote PLL glass, nanofibrillar scaffolds, Aclar, and PLL Aclar,
respectively. [117*].

6.2.2.4.3 Cdc42, Rac1, RhoA, and F-actin Immunolabeling Results
The reactive-like astrocytes were immunolabeled by antibodies against Cdc42, Rac1, and RhoA,
members of the Rho GTPase family, which are known to be upstream regulators of filopodia,
lamellipodia, and stress fiber formation, respectively [78]. Furthermore, inhibition of RhoA
effector Kinase (ROCK) through depression of RhoA is associated with the cessation of stress
fiber formation and the onset of stellation [93]. Representative results are shown in Figure 6.20
for Cdc42, Rac1, RhoA, with parallel phalloidin staining for F-actin. F-actin staining images
showed lack of stress fibers in reactive-like astrocytes for all culture surfaces. Chain-like cell
clustering on PLL Aclar was observed from F-actin staining images, as were low cell densities
and highly directional cell morphologies on Aclar. The F-actin staining results were therefore
consistent with the GFAP staining results. The RhoA intensity on all culture surfaces was
dimmer than other protein intensities (Figure 6.20 (c) versus (a) and (b)), which would be
consistent with depression of RhoA.
101

NFS

ACL

PLL A

F-actin

RhoA

F-actin

Rac1

F-actin

Cdc42

PLL G

Figure 6.20: Cdc42, Rac1, RhoA, and F-actin staining images of reactive-like astrocytes. Left
to right column: fluorescence images of astrocytes cultured on PLL glass, nanofibrillar
scaffolds, Aclar, and PLL Aclar. Immunolabeling and corresponding F-actin were represented
in red, and green, respectively. Astrocytes were labeled with (a) anti-Cdc42, (b) anti-Rac1, and
(c) anti-RhoA. Scale bar, 100 µm. [117*]. Fluorescent images by I. Ahmed, Shreiber
laboratory with analysis in this thesis.

102

6.2.2.4.4 Cdc42, Rac1, and RhoA Quantification Results
Rho GTPase expressions were investigated for statistically significant differences using Tukey’s
test to perform pairwise comparisons with P < 0.05. A total of 20 astrocytes per each culture
were selected randomly from both reactive-like and untreated astrocyte samples. The results,
shown in Figure 15, are displayed as mean-value bar graphs with standard error of the mean.
For the reactive-like astrocytes alone (Figure 6.21 (a)), RhoA was depressed relative to Rac1
for all culture surfaces, with the greatest differences observed for PLL glass (min P = 0.0008)
and PLL Aclar (min P = 0.0002). This result is consistent with reactive process development
through depression of RhoA in dBcAMP-treated astrocytes. The RhoA expression of reactivelike astrocytes was also significantly depressed relative to Cdc42 on PLL Aclar (P = 0.0175) and
PLL glass (P < 0.0001) (Figure 6.21 (a)).
For untreated astrocytes alone (Figure 6.21 (b)), RhoA was significantly depressed relative to
Rac1 only for astrocytes cultured on the nanofibrillar scaffolds (P = 0.0025). RhoA was also
significantly depressed relative to Cdc42 only for astrocytes cultured on the nanofibrillar
scaffolds (P = 0.015). Both findings were consistent with the previously reported results [12].
For reactive-like versus untreated astrocytes, the Rho GTPase expressions for reactive-like
astrocytes were shown to be modulated by the different culture surfaces (Figure 6.21 (a)) in ways
that differed from the modulation observed for untreated astrocytes (Figure 6.21 (b)).
Comparison of Figure 6.21 (a) and (b) showed that the dBcAMP treatment of astrocytes reversed
the relative expression of Cdc42, Rac1, and RhoA on all culture surfaces, except the nanofibrillar
scaffolds.

103

Figure 6.21: Quantitative Cdc42, Rac1 and RhoA expression estimation results for (a)
reactive-like and (b) quiescent astrocytes. The white, gray and dark gray bars show the
average Cdc42, Rac1 and RhoA intensities, respectively. Error bars show standard error of
n=20 cells. PLL G, NFS, ACL, and PLL A, and denote PLL glass, nanofibrillar scaffolds,
Aclar, and PLL Aclar, respectively. [117*].

6.2.3 Discussion
In the present work, the effects of culture surface physical properties on the morphological and
biochemical differentiation of reactive-like rat cerebral cortical astrocytes were investigated in
vitro. Additionally, the responses of the reactive-like astrocytes were compared with the
responses of untreated astrocytes. Differences in morphology, astrocyte-astrocyte interactions,
and immunocytochemistry of astrocytes cultured on four culture surface surfaces: nanofibrillar
104

scaffolds, and comparative PLL glass, PLL Aclar, and Aclar culture environments were
investigated. Four physical properties of the culture surfaces were considered as potentially
directive: elasticity, work of adhesion, surface polarity, and surface roughness. The results
demonstrated that the morphological and biochemical differentiation of reactive-like astrocytes
were modulated by the physical properties of the culture surfaces. Comparison of reactive-like
versus untreated responses further demonstrated that the responses were differently modulated
by the same culture surface properties.
The present study indicated that nanofibrillar scaffolds reduced the immunoreactivity
responses of the reactive-like astrocytes. This was consistently indicated by quantitative
morphology assessments, quantitative Rho GTPase expression estimation, and quantitative
GFAP and tubulin expression estimation. The quantitative morphology assessments showed that
process lengths of reactive-like astrocytes cultured on nanofibrillar scaffolds were significantly
shorter than those for reactive-like astrocytes on PLL Aclar and PLL glass (Figure 6.16(a)).
Reactive-like and untreated pairwise comparisons also showed that the reactive-like astrocytes
developed significantly longer processes, with the exception of the ones cultured on the
nanofibrillar scaffolds. Furthermore, the pairwise comparisons of the CSI of reactive-like
astrocytes showed that the CSI of astrocytes cultured on the nanofibrillar scaffolds was
significantly less than the ones of PLL glass and PLL Aclar (Figure 6.16 (d)). Therefore, results
from the quantitative morphology assessment indicated that the reactive-like astrocyte response
to nanofibrillar scaffolds was similar to the untreated response.
The present study is one of few investigations to combine immunoreactivity proteins with
Rho GTPase investigations [121], and the first to compare Rho GTPase investigations of
dBcAMP-stimulated and untreated cerebral cortical astrocytes. The dBcAMP-stimulated
105

astrocyte response to the nanofibrillar scaffolds was unique in that the relative Rho GTPase
expressions were unchanged within experimental error while Rho GTPase expressions to the
three other surfaces were significantly altered from their untreated expressions.
The dBcAMP-stimulated astrocyte responses to all surfaces showed morphological evidence
of stellation with depression of RhoA relative to Rac1, which agrees with the previously reported
work [93, 122]. In addition, the lack of stress fiber formation in reactive-like astrocytes, as
shown in the Figure 6.20 F-actin staining images, is consistent with the RhoA depression. The
quantitative Rho GTPase expression estimation results showed increased expression of Cdc42
for reactive-like astrocytes on PLL glass. The AFM morphology images showed that the
reactive-like astrocytes on PLL glass formed nanoscale processes, but actin staining images
suggest that those tiny processes are either not filopodia or were below the level of experimental
resolution by confocal microscopy.
The immunoreactivity proteins investigation enabled comparison of the amounts of GFAP
and tubulin expression among the four culture surfaces. Tubulin and GFAP expressions were
both comparatively high for the astrocytes on PLL Aclar surfaces. Comparatively high GFAP
expression was observed for astrocytes on the nanofibrillar scaffolds; however, tubulin
expression was comparatively low and all other Rho GTPase and morphological responses
indicated minimal change from untreated behavior. The overlay of GFAP and tubulin staining
images, shown in Figure 6.18, suggested interaction between these proteins. This is consistent
with previously reported results [122].
In sharp contrast with the reduction in immunoreactivity responses displayed by reactive-like
astrocytes on the nanofibrillar scaffolds, the responses to PLL Aclar exhibited features

106

reminiscent of glial scar formation. Chain-like clustering of reactive-like astrocytes with
interwoven processes was evident in immuno-staining and AFM images. The immunoreactivity
investigation demonstrated the largest comparative expressions of both GFAP and tubulin.
Furthermore, the most significant depression of RhoA relative to Rac1 was observed for the
reactive-like astrocytes on PLL Aclar. The chain-like clustering of reactive-like astrocytes
appeared to be induced by the local elastic property of the PLL Aclar surface in combination
with the surface polarity property. The directive nature of the elasticity property was indicated by
the chain-like versus non-chain-like clustering observed for reactive-like astrocytes on PLL
Aclar (soft) versus PLL glass (hard). The directive nature of the surface polarity property was
indicated by the chain-like clustering versus the unusual morphology and minimal cell-cell
interactions observed for reactive-like astrocytes on PLL Aclar (soft elasticity and high
hydrophilicity) versus Aclar (soft elasticity but low hydrophilicity).
It is also interesting to note that the responses of reactive-like astrocytes on PLL Aclar versus
PLL glass surfaces differed from the responses of untreated astrocytes on the same culture
surfaces. The untreated astrocytes on PLL Aclar surfaces developed extensive filopodia
formation that infiltrated along valleys created by the nanoscale surface roughness of these
surfaces [12]. The filopodia formation response was not observed on the nanoscale-smooth PLL
glass surfaces [12]. For the reactive-like astrocytes, although the surface roughness differences
were the same, a glial scar type response was observed instead of filopodia formation. The
investigations summarized in Figure 6.12 demonstrated that elasticity differences were also
present between the PLL glass and PLL Aclar culture surfaces. Previously reported results [123,
124, 125] have indicated increased sensitivity to local elasticity as a response to wound-healing

107

situations. Studies of hydrogel-assisted wound healing [125] further suggest that the combination
of local elasticity and surface charge may be directive.
The results support the hypothesis that the nanophysical cues of the cell culture surfaces can
modulate the morphological and biochemical differentiation of dBcAMP-stimulated cerebral
cortical astrocytes. The present study indicated that the properties of the nanofibrillar scaffolds
reduced the immunoreactivity responses of the reactive-like astrocytes, while the properties of
the PLL Aclar surfaces induced responses reminiscent of glial scar formation. The comparison of
reactive-like versus untreated astrocyte responses to elasticity versus surface roughness
differences suggested that different nanophysical cues may have more directive importance in a
wound-healing situation versus a maintenance situation.
The property investigations and corresponding astrocyte in vitro behavior in the present study
showed that a combination of tissue scaffold physical properties can modulate cellular
mechanisms of reactive-like astrocytes. This suggests that scaffolds with appropriate physical
properties could alter the immunoreactivity of astrocytes, which may promote axonal
regeneration around a lesion site. The polyamide nanofibrillar scaffolds appear to have the
optimum physical properties among the culture surfaces considered in this work. Therefore, it
remains as a promising candidate for future in vitro and in vivo investigations.

6.2.4 Conclusions
The morphological and biochemical differentiation of reactive-like dBcAMP-treated astrocytes
cultured on electrospun polyamide nanofibrillar scaffolds, PLL glass, Aclar, and PLL Aclar was
investigated by immunocytochemistry, fluorescence, and atomic force microscopy. Four physical
108

properties of the culture surfaces were considered as potentially directive: elasticity, work of
adhesion, surface polarity, and surface roughness. The physical properties of the culture
environment were shown to trigger biochemical and morphological consequences for reactivelike astrocytes. The properties of the nanofibrillar scaffolds reduced immunoreactivity responses
while the properties of the PLL Aclar surfaces induced responses reminiscent of glial scar
formation. The comparison of reactive-like and untreated astrocyte responses suggested that
different nanophysical cues may have more directive importance in a wound-healing situation
than in a maintenance situation. Local elasticity and surface polarity were indicated as potentially
directive for reactive-like cerebral cortical astrocytes.

109

Chapter 7

AFM Feature Definition for Neural Cells on Nanofibrillar Tissue
Scaffolds

7.1 Introduction
The use of AFM in biomedical science has grown rapidly, with recent exciting applications in
diverse fields including regenerative medicine (tissue engineering) [14], drug delivery [17],
protein folding [126] and clinical medicine [127]. Even so, AFM remains an under-utilized
technique within the biomedical research community and more importantly an under-developed
enabler of significant new nanoscale biomedical discoveries due to a general problem with
inconsistent feature definition. When a feature definition problem is encountered, the standard
approach is to use instrument-supplied hardware or software capabilities to resolve it. Deflection
imaging (contact mode) or phase imaging (tapping mode) can improve feature definition when
changes in cantilever deflection or RMS cantilever oscillation are greatest at boundaries.
Alternatively, image processing can be used to extract information that actually exists in an AFM
image but is inaccessible prior to processing. Useful filters are a standard component of
commercial image processing packages for AFMs, and low pass filtering for noise reduction is a
known and popular approach. However for either hardware or software approaches to be
successful, it is key to diagnose and accurately identify the nature of the feature definition
problem involved. As will be shown, low pass filtering can be the wrong approach to improve
feature definition. There has been comparatively little systematic methodology developed for

110

image diagnosis, other than user experience. This can make the AFM learning curve a lengthy
one for new biomedical researchers, especially for certain classes of biomedical problems that
have feature definition issues.
In the present work, a severe problem with feature definition of astrocyte neural cells on a
promising prosthetic nanofibrillar scaffold for brain and spinal cord injury repair [37, 38] is first
diagnosed and then resolved. Recent studies indicate that cells grown on nanofibrillar surfaces
that approximate their native extracellular matrix environments behave differently, and in
seemingly more biomimetic ways [10, 72]. Many details of the cell-cell and cell-scaffold
interactions that may induce the biomimetic response are not presently well known. This is
therefore a research area in which the nanoscale resolution capability of AFM could offer
significant biomedical insights.

The difficulty with AFM investigation is that cells on

nanofibrillar surfaces interact with these surfaces via nanoscale edges and processes that are not
distinguishable from the nanofibrillar background by height, deflection or phase imaging. This is
because the cellular edges and processes are approximately the same order in height as the
background nanofibers, ~100 to 200 nm. This problem can be resolved by filtering and present a
novel diagnostic approach based on standard AFM section measurements to enable
knowledgeable filter selection and design. The requirements for successful image processing
were identified as a combination of low frequency component suppression with dynamic range
enhancement. This design was implemented to filter the harmonic components present in the
images in such a way that the cellular edges and processes became distinguishable from the
nanofibrillar backgrounds. The new information revealed in the filtered AFM images would
change the biomedical interpretations drawn about the cell-cell interactions present, especially
when compared with a more typical analysis of fluorescent microscopy images [83].

111

7.2 Material and Methods
7.2.1 Neural Cell Culture
Rat cerebral cortical astrocytes were prepared from postnatal day one (P1) Sprague Dawley rats
2

and grown to confluence in astrocyte medium in 75 cm tissue culture flasks as previously
described [10]. The astrocyte culture medium was comprised of Dulbecco’s Modified Eagle’s
Medium (DMEM, Invitrogen, Carlsbad, CA) + 10% calf serum (Invitrogen). After reaching
confluence (~8-10 days), flasks were shaken overnight on a rotary shaker at 37°C to remove any
loosely adherent oligodendrocytes, neurons, or macrophages. Astrocytes were then subcultured
in astrocyte medium (0.5 ml) at a density of 50,000 cells/well onto 12 mm coverslips coated with
nanofibers in 24 well trays. The astrocytes were maintained for 24 hours (h). The astrocytes
were then fixed with paraformaldehyde (4%) and stained with phalloidin [113] for fluorescent
microscopy investigation. Phalloidin staining did not affect the AFM imaging.

7.2.2 Nanofibrillar Culture Surface
Randomly oriented polyamide nanofibers (median diameter ~180 nm) were electrospun from a
blend [81] of two polymers [(C28O4N4H47)n and (C27O4.4N4H50)n] onto plastic ACLAR
coverslips (Ted Pella, Reading, CA) by Donaldson Co., Inc. (Minneapolis, MN). Cross-linking
of nanofibers was done in the presence of an acid catalyst. The resulting nonwoven polymeric
nanofibrillar matrix was approximately 2.0 μm thick when measured on edge by optical
microscopy, with no direct openings to the coverslip surface [1, 128].

112

7.2.3 Epi-Fluorescence Microscopy
Epi-fluorescence microscopy images of astrocyte neural cell cultures at 24 h were captured using
a Zeiss Axioplan microscope (Carl Zeiss Microimaging GmbH, Germany). Fluorescence optical
microscopy is the most widely utilized technique for cell culture analysis.

7.2.4 Atomic Force Microscopy
AFM images of astrocyte neural cell cultures at 24 h were captured using a Nanoscope IIIA
(Bruker AXS Inc, Madison WI, formerly Veeco Metrology) operated in ambient air. Wide-area
images showing cell groups were acquired using a J scanner with a maximum scan range of 125
2

x 125 μm x-y range and

2.774 μm z range with the AFM was operated in contact mode, using

silicon nitride tips with a nominal tip radius of 20 nm and a cantilever spring constant k = 0.58
N/m. Close-up images of cell-scaffold interactions were acquired using an E scanner with a
2

maximum scan range of 13.5 x 13.5 μm x-y range and

1.54 μm z range with the AFM was

operated in tapping mode, using etched silicon tips with a nominal tip radius of 10 nm and a
drive frequency of ~320 kHz.

7.2.5 Image Processing Methods
Digital image processing techniques were implemented with MATLAB version 7.6.0 (R2008a).
Digital images were exported as ASCII files from the Nanoscope Software version 5.30r3.sr3 by
converting the units to nm. Four different types of two-dimensional Finite Impulse Response
(FIR) digital filters were evaluated in this work: frequency domain Gaussian and Butterworth
high pass filters, spatial domain high pass filters and high boost filters. The Gaussian and

113

Butterworth high pass filters were implemented over frequency domain with normalized cutoff
frequencies (ω/π) from 0 to 1 and with integer degrees from 1 to 5. As a final step, histogram
equalization was applied for contrast enhancement (MATLAB®, The MathWorks, Natick, MA).

7.2.6 Gaussian High Pass Filter Implementation
Image enhancement in the frequency domain is based on the computation of the two-dimensional
discrete Fourier transform (DFT) of the input image, followed by multiplication of the result by a
filter transfer function. The final output is obtained by taking the inverse two-dimensional DFT
of the product. The fast Fourier transform (FFT), the computationally efficient algorithm [129]
for computing DFT, was performed for all of the DFT computations throughout this work.

The two-dimensional DFT of an M x N pixel image was calculated as:

∑

∑

[

(

)]

(7.1)

where x and y are the spatial variables, f(x,y) is the raw image, u and v are the frequency domain
variables, and F is the two-dimensional DFT of the M x N pixel image.
The Gaussian high pass filter (GHPF) transfer function was implemented as:

[

114

]

(7.2)

where D0 is the cutoff frequency and D(u,v) is the distance from (u,v) to the origin. To apply the
GHPF to the image, F(u,v) and H(u,v) were multiplied by array multiplication.
G(u,v)=F(u,v)H(u,v) and the inverse two dimensional DFT defined as:

∑

∑

[

(

)]

(7.3)

was applied, and the final GHPF result was obtained.

7.2.7 High Boost Filtering Implementation
High boost, or high-frequency-emphasis, filtering is based on adding a specified percent of the
original image to the high pass filtered image [130]. This addition restores the low frequency
components that were lost in the high pass filtering operation, so the resulting image may look
more like the original image. This was attractive for re-inclusion of cell surface features. A highboost filter was implemented as:

(7.4)
where Y is the filter output, A is the amplification factor, O is the original image and H is the
high pass filtered image.

7.2.8 Histogram Equalization for Contrast Enhancement
A histogram equalization was performed in order to enhance the contrast of the filtered images.
The image contrasts were enhanced by transforming the values in the filtered image, so that the
115

histogram of the output image had a roughly equal number of pixels mapped to each of its 256
levels. The histogram equalization operation converted the low-contrast and dark images to
relatively higher contrast and brighter images.

7.3 Results
In this section, the feature definition problem found in distinguishing thin neural cell processes
and edges from the tissue culture surfaces that have nanoscale features is identified. The
quantitative problem diagnosis procedure is presented next. The optimal filter design based on
the problem diagnosis is then given, and finally the biomedical interpretations drawn from
analyzing images with missing versus complete information are discussed.

7.3.1 Feature Definition Problem
A composite AFM height image of a three-astrocyte group on a nanofibrillar surface is shown in
Figure 7.1 (a). The feature definition is poor for both the cell edges and the cellular processes
(extensions with which a cell explores its environment). The structures marked by arrows in the
AFM height image of Figure 7.1 (a) could be either nanofibers or cellular processes. Close-ups
of potentially important cell-scaffold interactions (dashed box in Figure 7.1 (a)) were
investigated by AFM tapping mode phase images. As shown in Figure 7.1 (b) this did not
improve the feature definition. The problem is that the cellular edges and processes are
approximately the same order in height as the background nanofibers, ~100 to 200 nm.

116

Figure 7.1: Feature definition problems of astrocyte neural cells on nanofibrillar surfaces. (a)
The structures marked by arrows in the AFM height image could be either nanofibers or
processes. (b) A higher resolution close-up tapping mode phase image of the region shown by
dashed box in (a). Its edge features are still unclear. [83*].

7.3.2 Systematic Approach to Problem Diagnosis
A diagnostic approach based on standard AFM section measurements was developed. Individual
section measurements of the nanofibrillar surface and an astrocyte cell body are shown in Figure
7.2 (a)/(b) and (d)/(e). A one-dimensional DFT was then applied to the section measurement
data of Figure 7.2 (b) and (e). This converted the section measurements into the frequency
domain where the harmonic components could be studied and analyzed. The magnitudes of the
DFT spectra were then calculated, and the logarithmic DFT spectra were plotted. The
logarithmic DFT spectra shown in Figure 7.2 (c) and (f) can be used to identify the key

117

differences between the cell and nanofibrillar surfaces, which can then be used to create an
optimal filter design. In the present case, the differences were:
1. The one dimensional DFT spectra in Figure 7.2 (c) and (f) demonstrated that the
amplitude of the zero frequency sample of the cell profile was approximately 5 fold
higher than the amplitude of the zero frequency of the nanofibrillar profile. This
suggested that attenuation of the zero frequency sample and the other low frequency
components with a high pass filter would result in deemphasizing the astrocyte surface
relative to the nanofibers so that the astrocyte surface could be distinguished from the
nanofibrillar background. This is known as dynamic range enhancement.
2. The nanofibrillar surface had sharper features, meaning more power in the higher
frequency harmonics than the astrocyte surface. The high frequency harmonic region
above cut-off frequency 0.5 (red dotted lines) had 52% more power than the same
frequency region for the astrocyte DFT spectrum.
3. The low frequency harmonics of astrocytes and nanofibers were overlapping in the DFT
spectra (Figure 7.2 (c) and (f)). Therefore, total elimination of cell surface information
while retaining the nanofibrillar surface information is not possible by filtering
techniques. However, because the high frequency components of nanofibrillar surface
had more power than the high frequency components of astrocyte surfaces, as shown in
Figure 7.2 (c) and (f), amplification of the high frequency components would exaggerate
the difference between nanofibers and astrocytes. This is achieved by using a high pass,
not a low pass, filter. The results of a preliminary test using a GHPF, shown in space and
k-space domains in Figure 7.3, confirmed this approach. When the GHPF surface profiles
shown in Figure 7.3 (a) and 7.3 (c) are compared, it can be seen that the nanofibrillar

118

surface has higher edge features than the astrocytes. Figure 7.3 (b) and 7.3 (d) shows that
the high frequency harmonics of nanofibrillar substrate has more power than the high
frequency harmonics of astrocyte surface. These quantitative differences are the basis for
the subsequent successful filter design.

DFT of Surface Profile

Surface Profile

(µm)

5
4
3
2
1
0

3
2
1
0
0

10

20

(a)
(µm)

30
(μm)

40

50 1

(b)

(c)

Surface Profile

DFT of Surface Profile

3
2
1
0
0

(d)

10

0.5
0
0.5
1
Frequency Index (ω/π)

20

30
(μm)

(e)

40

50

5
4
3
2
1
0

1

0.5
0
0.5
1
Frequency Index (ω/π)

(f)

Figure 7.2: (a) Section measurement of nanofibrillar surface along dashed line, (b)
corresponding surface profile, and (c) DFT of nanofibrillar surface profile. (d) Section
measurement of cell surface along dashed line (e) corresponding surface profile, and (f) DFT
of cell surface profile. [83*].

119

GHPF Nanofibrillar Surface Profile
(a)
(b)3
3

3
2
2
1
1
0
0

2
1

0

0

10

20

30
(μm)

40

50

GHPF Astrocyte Surface Profile

(c)

DFT of Surface Profile

1

3

2

2

1

1

0

0

0

10

20

30
(μm)

40

50

1

DFT of Surface Profile

(d)

3

0.5
0
0.5
Frequency Index (ω/π)

1

0.5
0
0.5
Frequency Index (ω/π)

1

Figure 7.3: GHPF results (a) filtered nanofibrillar surface profile, (b) DFT spectrum of the
nanofibrillar surface profile, (c) filtered astrocyte surface profile, and (d) DFT spectrum of the
astrocyte surface profile. [83*].

7.3.3 Filter Design Based on Problem Diagnosis
The Gaussian and Butterworth frequency domain high pass filters were implemented by
changing the order and the cutoff frequency of the filter. Optimum results were obtained with a
GHPF transfer function of order 1 and normalized cutoff frequency 0.5. The perspective plot of
the GHPF transfer function of order 1 and normalized cutoff frequency 0.5 is shown in Figure
7.4 (a) and the radial cross section of the transfer function is shown in Figure 7.4 (b). Figure 7.4
120

(a) shows that the Gaussian frequency domain high pass filter is positional invariant or isotropic.
Application of the Butterworth filter yielded noisier results, therefore identifying the optimized
GHPF as preferable.

(a)

(b)
1

1
0.75
0.5

0.5

0

0.25
400

400

200
0

200

0

0.25

0.5

0.75

1

(ω/π)

Figure 7.4: (a) Perspective plot and (b) radial cross section of the GHPF transfer function of order
1 and normalized cutoff frequency 0.5. The zero frequency sample is at the center in (a). [83*].
Spatial domain filters were also investigated, both as an alternative to frequency domain filter
and for implementation of high boost filtering, a variant of spatial filtering, for return of cell
features to filtered images. Spatial domain filters require the specification of a mask. In the
present work, 3 x 3, 5 x 5 and 7 x 7 spatial high pass masks were investigated, and 3 x 3 mask
size was identified as optimal. High boost filtering was therefore performed using a 3 x 3 spatial
high pass mask. Amplification factors of 1.05, 1.10, 1.15 and 1.20 were implemented, and 1.15
was optimum. However, the high boost filtering technique was not successful because return of
the %15 of the original image also compressed the dynamic range to the point where cellular
edge and process feature definitions were unacceptably reduced. Furthermore, analysis of spatial
121

domain transfer functions revealed positional anisotropy that would introduce distortions into
filtered images of randomly oriented nanofibers. These investigations enabled the systematic
selection of the GHPF as best for the investigations.

7.3.4 Biomedical Interpretations Based on Fluorescence Microscopy, AFM and
GHPF AFM Height Images
Cell culture protocols include inspection for cell morphology at regular intervals, e.g., at 24 h
and 48 h, typically performed using optical microscopy. In a coordinated epi-fluorescence
microscopy/AFM investigation [131], a series of images from different regions of the astrocyteon-scaffold cell cultures were acquired using epi-fluorescence microscopy. The epi-fluorescence
microscopy images (Figure 7.5 (a)) indicated that the astrocytes were stellate but largely isolated
from each other at 24 h. AFM height image of astrocytes cultured on nanofibrillar surfaces,
shown in Figure 7.5 (c), would not contradict the conclusion that the cells were stellate and
largely isolated at 24 h. However, the clear feature definition of cellular edges and processes
obtained by application of the GHPF filter, shown in Figure 7.5 (b), demonstrated that significant
leading edge formation beyond the previously assumed process ending had occurred at 24 h. The
high boost filtering result shown in Figure 7.5 (d) demonstrates that loss of dynamic range
enhancement by inclusion of as little as 15% of the original image resulted in loss of feature
definition to a level comparable to that of an unfiltered AFM image.

122

Figure 7.5: (a) Epi-fluorescence microscopy image of “isolated” astrocytes at 24 h. (b) GHPF
AFM image revealed significant leading edge formation. (c) Unfiltered AFM image did not
show full extent of leading edge formation. (d) Inclusion of 15% of original image in highboost filtering resulted in loss of feature definition to a level comparable to unfiltered AFM
image. [83*].
The unfiltered height image of Figure 7.1 (a) is reproduced in Figure 7.6 (a) and compared
with its GHPF equivalent in Figure 7.6 (b). In addition to the long process described in Figure
7.5, the filtered image revealed a second complete cellular bridge that extended between the
“isolated” cells. Direct contact via cellular bridges has implications for inter-cellular information
exchange [132] and, using the methods developed in this paper, this is the first time to report this
possibility for neural cells on nanofibrillar surfaces.

123

(a)

(b)

Figure 7.6: In the composite AFM height image (a), cellular processes and cell edges are
indistinguishable from the nanofibrillar background. (b) In the GHPF result, the cellular edges
and processes are clearly distinguished from nanofibrillar scaffold. Scale bars, 20 μm. [83*].

7.4 Discussion
In this work, DFTs of standard AFM section measurements of astrocytes versus nanofibrillar
surfaces were used as the basis for the diagnosis of the feature definition problem. These showed
that the high frequency components of nanofibrillar surface had more power than the high
frequency components of astrocyte surfaces and identified the use of a frequency domain high
pass filter as the potential solution to the feature definition problem. Such a filter has two
advantages. The high frequency components of the nanofibrillar surface are amplified, making
them easier to distinguish from the cellular features. Furthermore, suppression of the low
frequency components yields a dynamic range enhancement of the image. The diagnostic
conclusions were therefore that the combination of dynamic range enhancement with low
124

frequency component suppression that could be achieved through frequency domain high pass
filtering would be likely to enhance feature definition of cellular edges and process relative to
nanofibrillar tissue scaffolds. A GHPF filter of appropriate degree and cutoff frequency was
designed and used to selectively emphasize the high frequency harmonics that belonged to
nanofibers of the tissue culture scaffold, enabling the nanoscale cellular edges and processes to
be distinguished from nanofibrillar surfaces.

The high boost filtering technique, although

attractive for inclusion of cell surface features, was not successful in the problem because the
return of as little as 15% of the original image also compressed the dynamic range with
unacceptable loss of cellular feature definition relative to the nanofibrillar surface. The
information needed for clear feature definition was contained in the AFM images all along. The
problem was not a damaged or dirty tip, or poor tracking.
The GHPF filter of the present study was successfully applied to more than 250 contact and
tapping mode AFM height images of cerebral cortical astrocytes cultured on poly-L-lysine
functionalized glass, nanofibrillar scaffolds and fibroblast growth factor-2 derivatized
nanofibers; and also cerebellar granular neurons cultured on nanofibrillar scaffolds and poly-Llysine functionalized glass. While this suggests that the GHPF filter with selected degree and
cutoff frequencies has general utility in cell culture investigations, the major emphasis of this
study is on the simple diagnostic approach that identified it. The diagnosis and custom filter
design approach presented here expands user options beyond the selection supplied with the
instrument or towards more knowledgeable use of instrument supplied filters. It also enables the
user to connect with the growing body of work in image enhancement methods for AFM images
for biological samples [133], for non-biological samples [134], and in general [135, 136, 137].

125

The electrospun polyamide nanofibrillar matrices of the present study have shown promising
results for spinal cord repair [38]. There are many other types of nanofibrillar prosthetics
currently under investigation for different injury repair situations, including self-assembling
peptide nanofibers [138], porous hydrogels [139], self-assembling nanofibrous gels [140] and
ECM allografts [141]. Many potentially significant details of the cell-cell and cell-scaffold
interactions are not presently known and the nanoscale resolution of AFM could offer significant
biomedical insights. However, AFM investigation of cells on any of these surfaces would
encounter the same problem described in the present work: that the dynamic range of the
captured images is insufficient to allow thin ~20-100 nm layers of cytoplasm to be distinguished
from ~100-200 nm nano-featured surfaces.
GHPF AFM height images revealed that these nanofibrillar surfaces enable the development
of previously unknown astrocyte cell-cell interactions by 24 h, which provides new information
for ongoing investigations of why cells cultured on nanofibrillar surfaces seem more biomimetic.
The same techniques were successfully applied for cerebellar granular neurons cultured on
nanofibrillar surfaces, and should be useful in similar investigations, e.g., cardiomyocytes on
nanofibrillar surfaces. The methods developed here can therefore extend the usefulness of AFM
nanoscale imaging in regenerative medicine.

126

Chapter 8

A new AFM Based Cell Shape Index for Quantitative Cell Spreading
and Stellation Investigation

8.1 Introduction
A cell shape index (CSI) is a measure of the morphological differentiation of a cell that is used to
provide a quantitative comparison between cells. In vivo, cells have different morphologies
depending on their type and environment, e.g., astrocytes have a stellate morphology in the
central nervous system (CNS) to allow neuron and capillary interactions [142] while endothelial
cells in heart arteries have an elongated morphology with actin and microtubule fibers aligned
parallel to the direction of blood flow [143]. In vitro, cells also have different morphologies
depending on their type and environment, and recent studies have used quantitative cell shape
investigations to explore the potentially significant directive impact of environments for healthy
or pathological cellular outcomes [12, 116, 144, 145, 146].
A conventional CSI is the ratio of perimeter squared to the cell projection area [142]:

2

CSI = P / (4 π A)

(8.1)

where P is cell perimeter, and A is the cell projection area. This equation describes stellation as a
cell’s departure from a circular cross section since:
2

2

(2 π r) / 4 π (πr ) = 1

127

(8.2)

The conventional CSI has been previously used to provide two-dimensional shape
information of astrocytes [142] and vascular endothelial cells [143]. However, this definition for
CSI, which was created for use with two-dimensional optical microscopy images, does not
include three-dimensional effects, e.g., cell-spreading or hypotonic swelling.
New research directions are actively being pursued to enhance the usefulness of cell shape
analysis in biomedical research. Two important directions are research to incorporate threedimensional information and research to segment cells for CSI analysis. Addressing the former,
Canham and Burton have proposed a “sphericity index” for the study of human red blood cells
[147]. Chvátal et al. has recently developed a three-dimensional cell morphometry definition for
use with z-series confocal microscopy images [148]. Tiryaki et al. has recently developed a new
CSI that incorporates volumetric information acquired from high-resolution AFM height images
of cells on culture surfaces [120]. Addressing the segmentation issue, CSI analysis is a revealing
but currently under-utilized approach because CSI calculation requires a clearly defined cell
perimeter. Automatic extraction of cell boundary information using, e.g., NIH Image J 1.46r is
limited to isolated cells with perimeters that display sharp contrast. In recent work by Pincus and
Theriot [149], a mask and template matching approach was innovatively applied to confluent
cells in culture to create an accurate numerical two-dimensional representation for individual
cells with extractable boundaries.
In the present work, the new CSI definition based on analysis of high-resolution AFM
images is used [120] and additionally incorporate a new cell segmentation algorithm that uses a
new multi-dimensional texture analysis to segment cells from substrate regions. The cell area and
cell contours are found automatically without user bias. Segmentation of cells from substrate
regions is important for CSI analysis applied to cells cultured on the new, biomimetic substrates

128

as nanofibrillar [10] or hydrogel scaffolds, which have with significant surface features. AFM
images retain volumetric information such as cell spreading for cells cultured on surfaces,
making the new CSI segmentation and analysis system appropriate for investigations of
substrate-adherent cell cultures and surface seeding of three-dimensional matrix cultures.
The new AFM-based CSI segmentation and analysis system is applied to analysis of the
responses of untreated and dBcAMP-treated reactive-like astrocytes to the nanophysical cues
provided by four culture environments including polyamide nanofibrillar scaffolds. Astrocyte
responses to polyamide nanofibrillar scaffolds appear to favorably modulate the glial scar
response that blocks axon regeneration in CNS traumatic injury [37, 38]. AFM studies [12] have
identified that responses include cell spreading as well as process formation, and that the cell
spreading can vary depending on the surface polarity of the cell substrate. In the present work,
the new CSI segmentation and analysis system is used to perform the first quantitative cell
spreading and stellation response investigation for astrocytes in response to (1) changes in
nanophysical environment cues, and (2) untreated versus reactive-like states of astrocytes.

8.2 Experimental Procedures
8.2.1 Preparation of samples
Four culture surfaces were investigated in this work: PLL glass, PLL Aclar, Aclar and polyamide
nanofibrillar scaffolds. Glass coverslips (12 mm, No. 1 coverglass, Fisher Scientific, Pittsburgh,
PA) and Aclar coverslips (12 mm, Ted Pella, Redding, CA) were used as underlying surfaces for
the PLL functionalization. Glass or Aclar coverslips were placed in a 24-well tissue culture plate
(one coverslip/well) and covered with 1 mL of poly-L-lysine (PLL) solution (50 μg PLL/mL in
129

dH2O) overnight. The coverslips used for the cultures were then rinsed with dH2O and sterilized
with 254 nm UV light using a Spectronics Spectrolinker XL-1500 (Spectroline Corporation,
Westbury, NY). The polyamide nanofibrillar scaffolds electrospun on Aclar substrates were
obtained from Donaldson Co., Inc. (Minneapolis, MN) and Corning Life Sciences (Lowell, MA).
The fiber diameter for the nanofibrillar scaffolds has a range from 100 to 300 nm.
Primary untreated and reactive-like astrocyte cultures were prepared as previously reported
[115, 150]. The astrocytes cultured on coverslips were fixed in 4% paraformaldehyde for 10
minutes, rinsed with distilled water and air-dried. All cells in this study were fixed at 24 h, at
which time cells were either isolated or in small clusters.
Astrocyte responses to the nanofibrillar surfaces were studied in comparison with their
responses to three additional culture surfaces: poly-L-lysine-functionalized planar glass (PLL
glass), unfunctionalized planar Aclar (Aclar), and PLL-functionalized planar Aclar (PLL Aclar).
PLL glass is a standard astrocyte culture surface, and astrocyte responses to it are well
characterized, making it useful for identifying differences in astrocyte responses to other
surfaces. The polyamide nanofibrillar scaffolds were electrospun on Aclar substrates; therefore,
astrocyte responses to Aclar surfaces were investigated to distinguish responses to the
nanofibrillar scaffolds from possible responses to the underlying Aclar substrate. Astrocyte
responses to PLL Aclar surfaces were studied to clarify the role of the underlying substrate
versus surface functionalization. The four culture surfaces were previously characterized [12,
116] in terms of four of the nanophysical cues that they presented: local surface roughness, local
elasticity, local work of adhesion and global surface polarity.

130

8.2.2 Analytical techniques
The AFM investigations were performed using a Nanoscope IIIa (Bruker AXS Inc, Madison WI,
formerly Veeco Metrology) operated in contact mode and in ambient air. A J scanner with 125
μm x 125 μm x 5.548 μm x-y-z scan range, and Bruker DNP silicon nitride probes with a 35º ±
2º cone angle and a nominal 20 nm tip radius of curvature were used for the AFM investigations.
For each culture surface, a minimum of 50 images were captured from different regions of at
least three different cell substrates. The AFM height images are 512x512 and 256x256 pixels as
identified in the text with 16 bit gray level depth. In this work, the field of view of the images
was 100 μm x 100 μm. Cell segmentation for both the conventional and new CSI calculations
was implemented with MATLAB version 7.7.0 (R2012b) (The MathWorks, Natick, MA) using
the neural network and image processing toolboxes.
Cell segmentation was performed automatically by texture segmentation. The AFM height
data was exported as an ASCII file, and the textural features were extracted from the height and
the Gaussian High-Pass Filtered height image. For comparative culture surfaces, Aclar, PLL
Aclar and PLL glass, the cell and substrate surface training data was determined manually. The
cell boundaries on nanofibrillar scaffolds are not easily distinguishable. Therefore, for astrocytes
on nanofibrillar scaffolds the cell surface training data features were collected from AFM cell
culture images, and scaffold surface data features were collected from bare nanofibrillar scaffold
AFM images to ensure correct training data. A total of 28 types of textural features were
extracted from AFM height images, and are listed in Table 8.1. The textural features were
extracted from each pixel of the image, and the moving window size was set to 5x5 for local
statistics textural features. The texture measure A and GHPF textural features were calculated as
described in Ref [151] and Ref [83]. Sequential forward selection (SFS) and sequential floating
131

forward selection (SFFS) feature selection algorithms [152] were used to select features and
reduce the computational cost for segmentation. The texture features were then used to classify
each pixel of the AFM image either to cell or cell substrate region via a two-hidden-layer
feedforward neural network with resilient backpropagation algorithm [153]. The number of
training iteration was set to 30 times the number of input features to prevent the overtraining of
the classifier. The number of hidden nodes was approximately 10% of the number of training
patterns. For target values, each digit was encoded with +1 and -1 values to improve
performance of the classifier. After segmentation, the morphological close operation was used to
remove the noise. The small regions in the segmented image which are less than 1% of the whole
image were eliminated. The segmented image was used to calculate the conventional and new
CSI. In this work, the cell segmentation method was not intended for finding the boundaries of
cells in clusters. The boundaries of cells were outlined manually when the cells were clustered.
The segmentation performance on PLL Aclar surfaces was low. When 7x7 pixels moving
window was used, the accuracy was 85%. Therefore, the segmentation was done manually when
needed.
GLCM is a statistical method of extracting texture features from images. It is a tabulation of
how often different combinations of pixel brightness values occur in an image [154]. The texture
idea came from Haralick’s paper [155]. In this work, a subset of 28 textural features derived from
two models: the gray level co-occurance matrix and local statistics texture features were used to
classify cell region versus cell substrate region. The 28 textural features are listed in Table 8.1.

132

Table 8.1: The textural features used to segment the cell surface from the cell substrate. (STD:
standard deviation; GHPF: Gaussian high pass filter; GLCM: gray level co-ocurance matrix;
IMC1: information measure of correlation 1; IMC2: information measure of correlation 2.

#

textural feature

model

#

1

mean

local statistics

15 inverse difference moment GLCM

2

STD

local statistics

16 sum average

GLCM

3

mean by STD

local statistics

17 sum variance

GLCM

4

GHPF

local statistics

18 sum of variances

GLCM

5

texture measure A

local statistics

19 sum entropy

GLCM

6

entropy

local statistics

20 entropy

GLCM

7

skewness

local statistics

21 difference variance

GLCM

8

kurtosis

local statistics

22 difference entropy

GLCM

9

power spectrum

local statistics

23 IMC1

GLCM

10 local binary partition

local statistics

24 IMC2

GLCM

11

GLCM

25 dissimilarity

GLCM

12 correlation

GLCM

26 inertia

GLCM

13 energy

GLCM

27 cluster shade

GLCM

14 homogeneity

GLCM

28 cluster prominence

GLCM

contrast

textural feature

model

8.2.2.1 Astrocyte Morphology Investigation by Conventional CSI
After segmentation, the conventional CSI definition given in equation 1 was used to
investigate the quantitative morphometry of the cerebral cortical astrocytes cultured on the four
different culture surfaces that presented different nanophysical cues [120]. The conventional CSI
measures were investigated for statistically significant differences using Tukey’s test pairwise
comparisons with P < 0.05.

133

8.2.2.2 Astrocyte Morphology Investigation by New AFM-based Cell Shape Index
The new CSI that includes volumetric information extracted from AFM images of cells on
surfaces is defined as:
3

2

New CSI= (1/18 π) x (SAcell) / (Vcell)

(8.3)

where SAcell is the cell surface area, and Vcell is the cell volume. This assumes that the cell
is resting on, not penetrating, a surface. For a hemispherical “cell” on a surface, the new CSI is:
2 3

3 2

(1/18 π) x (0.5[4πr ]) / (0.5[4/3 πr ]) = 1

(8.4)

Therefore, the new CSI increases as the cell morphology become more stellate and/or the cell
spreading increases.
After cell segmentation from the cell substrate, the surface area of each cell was calculated
by splitting the AFM topography faces into triangles. The area of a triangle in three-dimensional
space was computed using the cross product given by:
SAtriangle = 0.5 |(v2 - v1) x (v3 - v1)|

(8.5)

where SAtriangle is the area of a triangle on the cell surface, and the coordinates of the vertices
are given by vi=(xi, yi, zi). The surface area of each triangle was then computed over the
segmented cell area, and cell surface area was obtained. The volume of each astrocyte was
calculated by assuming each pixel and its z coordinate as a square prism [156]. The volume
under each pixel was computed by multiplying the unit pixel area by the height of that pixel.
This was repeated for all the pixels of the segmented cell region. The base level for all culture
134

surfaces was assumed to be a planar surface. The base level for planar culture surfaces was
determined taking the average of the user-defined culture surface region. The base level for the
nanofibrillar scaffolds was determined similarly except when the average value of the culture
surface is not lower than the minimum height point on the cell region, the base level was
assumed to be 10 nm (the thickness of cell membrane) lower than the minimum height point on
the cell region. The new CSI was computed as given in equation 8.3. Departure from unity
reflects the average departure from a three-dimensional hemispherical volume by both stellation
and cell spreading.

8.3 Experimental Results
8.3.1 Initial Segmentation Step
CSI analysis requires mathematical segmentation of cell from the cell substrate. The difficulty of
segmentation varies depending on the cell and the substrate. In this work, the cell segmentation
on nanofibrillar scaffolds and PLL Aclar surfaces were more challenging than the PLL glass and
Aclar surfaces. Cells on nanofibrillar scaffolds interact with these surfaces via nanoscale edges
and processes that are not distinguishable from the nanofibrillar background by AFM height,
deflection, or phase imaging. This is because the cellular edges and processes are approximately
the same order in height as the background nanofibers, ~100 to 200 nm. This difficulty was
overcome by using GHPF as described in detail [83].
The textural features of astrocytes and PLL Aclar surface were similar which reduced the
segmentation performance. The surface roughness of PLL Aclar surfaces was higher than other
cell substrates when the surface roughness values were measured from individual nanofibers [3].
135

The surface roughness of the cell region has a lot of variance because of the many different types
of protein assemblies, glycoproteins, cell protrusions, and ion channels on astrocyte membrane.
Therefore, in terms of surface roughness, the cell and the PLL Aclar surfaces were closer than
other cell substrate surfaces. This explains the reason of the difficulty in segmenting the cell
region from the PLL Aclar surfaces.
In previous work, a diagnostic approach based on analysis of discrete Fourier transform of
standard AFM section measurements was developed and used to identify quantitative differences
in the frequency components of nanoscale cellular edges and processes of neural cells cultured
on nanofibrillar scaffolds. Accurate identification of differences formed the basis for the
subsequent successful Gaussian high-pass filter (GHPF) design that enabled the edges and
processes to be clearly distinguished in AFM images. In the present work, the GHPF was applied
to AFM height images and followed by histogram equalization. The final cell boundaries were
determined by manual segmentation. The results of the segmentation procedure steps for a
representative astrocyte AFM height image were shown in Figure 8.1 (a)-(c).

136

Cell segmentation
result

Cell outline and
noise removal

PLL Aclar

PLL glass

Nanofibrillar
scaffolds

Aclar

AFM height image
of astrocytes

Figure 8.1: The segmentation procedure results of AFM height images of an astrocyte
cultured on cell substrates. AFM height images of the astrocytes (left column), the
segmentation results (middle column), and individual cell area outlines (right column).
Scale bar, 10 μm.
137

8.3.2Astrocyte Morphology Investigation by Conventional CSI
After segmentation, the conventional CSI definition given in equation 8.1 was used to investigate
the quantitative morphometry of the cerebral cortical astrocytes cultured on the four different
culture surfaces that presented different nanophysical cues. At 24 h, astrocytes on all substrates
exhibited variation in their morphologies. The conventional CSI results, shown in Figure 8.2,
indicated that untreated astrocytes cultured on Aclar surfaces are more stellate than the ones on
other surfaces. The astrocytes on PLL glass and nanofibrillar scaffolds became more stellate
when they were treated by dBcAMP, whereas other astrocytes did not.

15
13

2D CSI

11
9
7

untreated

5

reactive-like

3

1
Aclar

PLL glass Nanofibrillar PLL Aclar
scaffolds

Figure 8.2: The conventional CSI results of untreated and reactive-like astrocytes cultured on
Aclar, PLL glass, nanofibrillar scaffolds and PLL Aclar. Error bars show standard error.

138

8.3.3 Astrocyte Morphology Investigation by New AFM-based Cell Shape Index
The new CSI results shown in Figure 8.3 indicated that untreated and reactive-like astrocytes
cultured on the nanofibrillar scaffolds had the lowest CSI values. This implies that, with the
same stellation counted in the new CSI, there is also dominant low spreading behavior. The other
astrocyte group with low CSI values was on Aclar. AFM and phalloidin staining results [12]
confirmed that astrocytes cultured on Aclar surfaces were significantly more hemisphere-like
than the ones on other substrates. The new CSI results further indicated that the astrocytes
cultured on the two PLL functionalized surfaces were similar in terms of combined cell
spreading and stellation possibly due to PLL functionalization. The new CSI of astrocytes on
PLL glass reduced when the astrocytes became reactive-like, whereas the ones of PLL Aclar
were increased. This result may be due to the different elastic property of cell substrate.

8000
7000

3D CSI

6000
5000

4000

untreated
reactive-like

3000
2000

1000
0

Aclar

PLL glass Nanofibrillar PLL Aclar
scaffolds

Figure 8.3: New CSI results of untreated and reactive-like astrocytes cultured on Aclar,
PLL glass, nanofibrillar scaffolds, and PLL Aclar. Error bars show standard error.

139

8.4 Discussion
The field of measuring cell morphology changes is known as quantitative morphometry. CSI is a
dimensionless two-dimensional cell shape analysis method [142]. A CSI that incorporates
volume information acquired from high-resolution AFM height images of cells on culture
surfaces was recently reported [120]. In this work, a new automatic cell segmentation method
that can be successfully applied to cells on rough non-planar surfaces is presented. The new CSI
is also dimensionless but includes quantitative information about the cell spreading and
stellation.
Cell-matrix adhesion is mediated by biomolecular interactions between adhesion receptors on
the plasma membrane and cell substrate. Cell spreading is an F-actin and myosin dependent
process, and it is important for cell growth and cell motility. To date, cell spreading has been
investigated by using confocal microscopy [157], and recently with AFM [158]. One of the
advantages of AFM over CLSM is that AFM resolution with the J scanner used in this work, in z
direction is less than 1 nm, whereas it is around 500 nm for CLSM. Thus, cell volume and cell
surface area calculations with AFM are more accurate than CLSM. As a result, this study
presents a very accurate quantitative evaluation of cell spreading.
The PLL functionalized surfaces are commonly used as a cell substrate for neural cells to
increase cell adhesion. The astrocytes cultured on Aclar adopted a hemisphere-like shape, but the
ones on PLL Aclar and PLL glass were well spread. The new CSI is a new quantitative cell
morphology evaluation method that makes use of the volumetric information provided by highresolution AFM investigation, expected to useful in biomedical science.

140

Chapter 9
Conclusions and future work

There is currently no treatment option for human CNS injury. However, CNS axons have been
proven to grow in permissible environments; therefore, the primary problem to CNS injury is
finding the chemical and physical conditions that will promote axon regeneration [159]. This
dissertation demonstrates studies to elucidate the effects of nanophysical properties of a
promising CNS scaffold on the modulation of the cellular and molecular components of glial
scar.
In this research, dBcAMP-treatment was performed to obtain astrocytes that are similar to
reactive astrocytes in the glial scar, and named as reactive-like astrocytes. Nanophysical cues of
tissue scaffolds have been shown to affect morphology of untreated and dBcAMP-treated
astrocytes, and the astrocyte cytoskeletal protein expressions including actin, GFAP, and tubulin
(Chapter 6). Nanophysical cues that were investigated in this thesis are surface polarity, work of
adhesion, surface roughness, and elasticity. The expression of F-actin, GFAP, and tubulin were
quantified by CLSM z-series, and western blot. Western blot is used in biomedical science to
detect specific proteins in the given sample but does not show the distribution of the protein in
the cell cytoplasm. CLSM presents three-dimensional localization of protein which can also be
used for quantification of specific proteins.
In conclusion, polyamide nanofibrillar scaffolds presented a mid-range of multiple cues to
cells, while the comparative substrates presented cues that were extreme in one or more physical
parameters. One major finding from this research was identification of the nanofibrillar scaffold
141

cues that corresponded to unchanged morphologies and cytoskeletal protein expressions even for
cells with deliberately induced reactivity. Another important major finding was identification of
a different set of nanophysical cues that induced morphological and protein expressions that
recapitulated glial scar formation. A further major finding was that all of the following showed
statistical differences as a function of the nanophysical cues presented by the culture
environments:

cytoskeletal proteins, reactivity proteins, and morphological expressions

including process length, soma height, and cell shape index.
The special contribution to image processing for AFM investigations was presented in
Chapter 7. GHPF is a useful feature for segmenting neural cells from nanofibrillar scaffolds. The
comparison of GHPF and other local statistics and GLCM textural features on the recognition
performance is given in Chapter 8. A new CSI definition which includes stellation and spreading
behavior of cells is also explained in Chapter 8.
The investigation of neural cell responses with super-resolution microscopy, a form of high
resolution light microscopy, continues. A preliminary study on the quiescent astrocyte-astrocyte
interactions on the nanofibrillar scaffolds demonstrated contact spacing at 24 h, with interactions
by stellate processes. While multiple examples of contacting structures that appeared to be in
various stages of development were discovered using AFM, a protein composition investigation
was below the resolution limit of the CLSM. A new capability, super-resolution microscopy, is
starting to enable protein investigations below the optical diffraction limit using strategic
combinations of emission, depletion and pixel-by-pixel analysis. The Figure 9.1 images, which
are immunostained for GFAP, were acquired using on a gated stimulated emission depletion
(GSTED) super-resolution microscopy system. An apparent contact (arrow) is located in the
large field of view confocal image shown in Figure 9.1 A. The GFAP fibril thickness was
142

measured to be about 90 nm using the GSTED image and Huygens software (asterisk), and this
is well below the level of resolution of confocal microscopy. The new information provided by
the increased resolution indicates that function of the contacting process may be to increase the
interaction area. Furthermore, approximate location of the contacting cell GFAP with that of the
contacted cell is observed (left arrow), suggesting possible interaction(s) that can be explored.

Figure 9.1. A. Confocal GFAP staining image of an intercellular contact (arrow) between
untreated astrocytes cultured on nanofibrillar scaffolds. An AFM image of a contacting
structure is shown in the inset. B. In the higher resolution confocal image, the contact points
between the two cells (arrows) are not clearly resolved. C. In the STED image, the
continuity of the connections to the process is clearly resolved (right arrow). A potential
interaction between cells is shown by left arrow. The GFAP polymers were measured to be
about 90 nm using the STED image and Huygens software (asterisk). Images, D. I. Shreiber
and V. M. Ayres laboratories. Scale bars show 10, 2, 2.5 and 2.5 μm in A, A inset, B, and C,
respectively.

143

Finally, there is a need for investigating the changes in the physical properties of tissue
scaffolds over time. An ideal scaffold should have growth inhibitory cues for axonal regeneration
at the acute phase of the recovery, but a growth promoting behavior in the chronic phase [160].
The degradation behavior of polyamide nanofibrillar scaffolds, the change in the physical
properties of scaffolds over time, and the influence of physical property changes on the neural
cell response remain as a future work.

144

BIBLIOGRAPHY

145

BIBLIOGRAPHY
[1] Ahmed I, Liu HY, Mamiya PC, Ponery AS, Babu AN, Weik T, Schindler M, Meiners S.
Three-dimensional nanofibrillar surfaces covalently modified with tenascin-C-derived peptides
enhance neuronal growth in vitro. J Biomed Mater Res A 76(4), 851–860 (2006).
[2] Nur-E-Kamal A, Ahmed I, Kamal J, Schindler M, Meiners S. Three-dimensional
nanofibrillar surfaces induce activation of Rac. Biochem Biophys Res Commun 331(2), 428–434
(2005).
[3] Tejeda-Montes E, Smith KH, Poch M, López-Bosque MJ, Martín L, Alonso M, Engel E,
Mata A. Engineering membrane scaffolds with both physical and biomolecular signaling. Acta
biomater 8(3), 998–1009 (2012).
[4] Ereifej ES, Matthew HW, Newaz G, Mukhopadhyay A, Auner G, Salakhutdinov I,
Vandevord PJ. Nanopatterning effects on astrocyte reactivity. J Biomed Mater Res A 101(6) 1–
15 (2012).
[5] Hallab NJ, Bundy KJ, O’Connor K, Moses RL, Jacobs JJ. Evaluation of metallic and
polymeric biomaterial surface energy and surface roughness characteristics for directed cell
adhesion, Tissue Eng 7(1), 55–71 (2001).
[6] Discher DE, Janmey P, Wang Y. Tissue Cells Feel and Respond to the Stiffness of Their
Substrate. Science, 310(5751) 1139–1143 (2005).
[7] Lampin M, Warocquier-Clerout R, Legris C, Degrange M, Sigot-Luizard MF. Correlation
between substratum roughness and wettability, cell adhesion, and cell migration. J Biomed Mater
Res 36(1) 99–108 (1997).
[8] Lipski AM, Pino CJ, Haselton FR, Chen IW, Shastri VP. The effect of silica nanoparticlemodified surfaces on cell morphology, cytoskeletal organization and function. Biomaterials,
29(28) 3836–3846 (2008).
[9] Janmey PA, van Oosten AS. Extremely charged and incredibly soft: physical characterization
of the pericellular matrix. Biophys J 104(5) 961–963 (2013).
[10] Delgado-Rivera R, Harris SL, Ahmed I, Babu AN, Patel R, Ayres VM, Flowers DA,
Meiners S. Matrix Bio 28 137-147 (2009).
[11] Iannaccone PM, Jacob HJ. Rats! Dis Model Mech 2(5–6), 206–210 (2009).
[12] Tiryaki VM, Ayres VM, Khan AA, Ahmed I, Shreiber DI, Meiners S. Nanofibrillar
scaffolds induce preferential activation of Rho GTPases in cerebral cortical astrocytes. Int J
Nanomedicine, 7, 3891-3905, (2012).
146

[13] Chen Q, Fan Y, Ayres VM, Udpa L, Schindler MS, Rice AF. Scanning probe recognition
microscopy investigation of the elastic properties of tissue scaffolding. In Materials Research
Society, Boston, MA (2005).
[14] Fan Y, Chen Q, Ayres VM, Baczewski AD, Udpa L, Kumar S. Scanning probe recognition
microscopy investigation of tissue scaffold properties. Int J Nanomed 2(4), 651–661 (2007).
[15]Delgado Rivera R, Ayres VM, Chen Q, Fan Y, Flowers DA, Meiners S, Ahmed I. Scanning
probe recognition microscopy investigation of neural cell prosthetic properties. Int J
Nanomanufacturing 6, 279–290 (2010).
[16] Wang X, Bukoreshtliev NV, Gerdes HH. Developing neurons form transient nanotubes
facilitating electrical coupling and calcium signaling with distant astrocytes. PloS one 7(10),
e47429 (2012).
[17] Sitterberg J, Ozcetin A, Ehrhardt C, Bakowsky U. Utilising atomic force microscopy for the
characterisation of nanoscale drug delivery systems. Eur J Pharm Biopharm 74(1), 2–13 (2010).
[18] Wang CC, Tsong TY, Hsu YH, Marszalek PE. Inhibitor binding increases the mechanical
stability of staphylococcal nuclease. Biophys J 100(4) 1094–1099 (2011).
[19] Kreplak L, Wang H, Aebi U. Atomic force microscopy of mammalian urothelial surface. J
Mol Biol 374(2) 365–373 (2008).
[20] Nedergaard M, Ransom B, Goldman SA. New roles for astrocytes: redefining the functional
architecture of the brain. Trends Neurosci 26(10) 523–530 (2003)
[21] Matyash V, Kettenmann H. Heterogeneity in astrocyte morphology and physiology. Brain
Res Rev, 63(1-2), 2–10 (2010).
[22] Abbott NJ, Rönnbäck L, Hansson E. Astrocyte-endothelial interactions at the blood-brain
barrier. Nat Rev Neurosci 7(1), 41–53 (2006).
[23] Oberheim NA, Takano T, Han X, He W, Lin JH, Wang F, Xu Q, Wyatt JD, Pilcher W,
Ojemann JG, Ransom BR, Goldman SA, Nedergaard M Uniquely hominid features of adult
human astrocytes. J Neurosci 29 3276–3287 (2009)
[24] McKeon RJ, Jurynec MJ, Buck CR. The chondroitin sulfate proteoglycans neurocan and
phosphacan are expressed by reactive astrocytes in the chronic CNS glial scar. J Neurosci
19(24), 10778–10788 (1999).
[25] Hikino M, Mikami T, Faissner A, Vilela-Silva AC, Pavão MS, Sugahara K. Oversulfated
dermatan sulfate exhibits neurite outgrowth-promoting activity toward embryonic mouse
hippocampal neurons: implications of dermatan sulfatein neuritogenesis in the brain. J Biol
Chem 278, 43744–43754 (2003).
147

[26] Raisman, G. Myelin inhibitors: does NO mean GO? Nat Rev Neurosci 5, 157–161 (2004).
[27] Sofroniew MV, Vinters HV. Astrocytes: biology and pathology. Acta neuropathol 119(1),
7–35 (2010).
[28] Carlstedt T. Nerve fibre regeneration across the peripheral-central transitional zone. J Anat
190 51–56 (1997).
[29] Schmidt CE, Leach JB. Neural tissue engineering: strategies for repair and regeneration.
Annu Rev Biomed Eng 5, 293–347 (2003).
[30] Zhou K, Nisbet D, Thouas G, Bernard C, Forsythe J. Bio-nanotechnology Approaches to
Neural Tissue Engineering. Tissue engineering, InTech North America. 459–485 (2010).
[31] Albert E. Einige operationen an nerven. Wien Med 26 1285 (1885).
[32] Gluck T. Ueber Neuroplastik auf dem Wege der Transplantation. Arch Klin Chir 25 606–
616 (1880).
[33] Payr E. Beitr¨age zur Technik der Blutgefass und Nervennaht nebst Mit-teilungen ¨ uber die
Verwendung eines resorbibaren Metalles in der Chirurgie. Arch Klin Chir 62, 67 (1900).
[34] Kirk EG, Lewis D. Fascial tubulization in the repair of nerve defects. JAMA 65:486–92
(1915).
[35] Lundborg G. Nerve injury and repair. New York: Longman Group UK (1988).
[36] Mackinnon SE, Dellon AL. Surgery of the peripheral nerve. New York: Thieme Med. Publ.
(1988).
[37] Meiners S, Ahmed I, Ponery AS, Amor N, Harris SL, Ayres VM, Fan Y, Chen Q, DelgadoRivera R, Babu AN. Engineering electrospun nanofibers for spinal cord repair: A discussion.
Polymer Int 56, 1340-1348 (2007).
[38] Meiners S, Harris SL, Delgado-Rivera R, Ahmed I, Babu AN, Patel RP, Crockett DP. A
nanofibrillar prosthetic modified with fibroblast growth factor-2 for spinal cord repair In; Chang
WN. editor. Nanofibers: Fabrication, Performance, and Applications. Hauppage, NY: Nova
Science Publishers, Inc. 327–343 (2009).
[39] Binnig A, Rohrer G. Scanning tunneling microscopy. Helvetica Physica Acta, 55, 726–735
(1982).
[40] Guckenberger R, Heim M, Cevc G, Knapp HF, Wiegräbe W, Hillebrand A. Scanning
tunneling microscopy of insulators and biological specimens based on lateral conductivity of
ultrathin water films. Science 266(5190), 1538–1540 (1994).
148

[41] A practical guide to scanning probe microscopy, Veeco Instruments Inc, (2005).
[42] Parpura V, Haydon PG, Henderson E. Three-dimensional imaging of living neurons and glia
with the atomic force microscope. J Cell Sci 104, 427–432 (1993).
[43] Schaus SS, Henderson ER. Cell viability and probe-cell membrane interactions of XR1 glial
cells imaged by atomic force microscopy. Biophys J 73(3), 1205–1214 (1997).
[44] Lapshin RV. Feature-oriented scanning methodology for probe microscopy and
nanotechnology. Nanotechnology 15(9), 1135–1151 (2004).
[45] Butt HJ, Cappella B, Kappl M. Force measurements with the atomic force microscope:
Technique, interpretation and applications. Surface Science Reports 59(1–6), 1–152 (2005).
[46] Hansma HG. Surface biology of DNA by atomic force microscopy. Annu Rev Phys Chem
52, 71–92 (2001).
[47] Ng L, Grodzinsky AJ, Patwari P, Sandy J, Plaas A, Ortiz C. Individual cartilage aggrecan
macromolecules and their constituent glycosaminoglycans visualized via atomic force
microscopy. J Struct Biol 143(3), 242–257 (2003).
[48] Chen CH, Clegg DO, Hansma HG. Structures and dynamic motion of laminin-1 as observed
by atomic force microscopy. Biochemistry 37(22), 8262–8267 (1998).
[49] Aifantis KE, Shrivastava S, Pelidou SH. Capturing the elasticity and morphology of live
fibroblast cell cultures during degradation with atomic force microscopy. J Microsc 249(1), 62–
68 (2013).
[50] Fan Y, Chen Q, Ayres VM, Baczewski AD, Udpa L, Kumar S. Scanning probe recognition
microscopy investigation of tissue scaffold properties. Int J Nanomed 2(4), 651–61 (2007).
[51] Ayres VM, Chen Q, Fan Y, Flowers DA, Meiners SA, Ahmed I, Delgado-Rivera R.
Scanning probe recognition microscopy investigation of neural cell prosthetic properties. Int J
Nanomanufacturing 6, 279–290 (2010).
[52] Minsky MM. Microscopy apparatus, US 3013467 (1961)
[53] Pawley JB. Handbook of Biological Confocal Microscopy, Berlin: Springer. (2006).
[54] Bolte S, Cordelières FP. A guided tour into subcellular colocalization analysis in light
microscopy. J Microsc 224, 213–232 (2006).
[55] Shastri VP, Lendlein A. Engineering materials for regenerative medicine, MRS Bulletin
35(8), 571-577 (2010).
149

[56] Ayres VM, Chen Q, Fan Y, Flowers DA, Meiners SA, Ahmed I, Delgado-Rivera R. Neural
cell prosthetic investigation by scanning probe recognition microscopy. 6th International
Symposium on Nanomanufacturing, Athens, Greece, 5.B.2.(2008).
[57] Zhao BH, Lee IS, Han IH, Park JC, Chung SM. Effects of surface morphology on human
osteosarcoma cell response. Current Applied Physics 7S1, e6-e10 (2007).
[58] Lipski AM, Jaquiery C, Choi H, Eberli D, Stevens M, Martin I, Chen IW, Shastri VP.
Nanoscale engineering of biomaterial surfaces. Adv Mater 19, 553-557 (2007).
[59] Pelham RJ, Wang YL. Cell locomotion and focal adhesions are regulated by substrate
flexibility. Proc Natl Acad Sci USA 94, 13661–13665 (1997).
[60] Kubinová Š, Horák D, Hejčl A, Plichta Z, Kotek J, Proks V, Forostyak S, Syková E.
SIKVAV-modified highly superporous PHEMA scaffolds with oriented pores for spinal cord
injury repair. J Tissue Eng Regen Med doi: 10.1002/term.1694 (2013).
[61] Thomas WE, Discher DE, Shastri VP. Mechanical regulation of cells by materials and
tissues, MRS Bulletin 35(8), 578-583 (2010).
[62] Khan A, Philip J, Hess P. Young's modulus of silicon nitride used in scanning force
microscope cantilevers. J Appl Phys 95(4), 1667-1672 (2004).
[63] Derjaguin BV, Muller VM, Toporov YP. Effect of Contact Deformations on the Adhesion
of Particles. Prog Surf Sci 45(1-4), 131-143 (1994).
[64] Jogikalmath G, Stuart JK, Pungor A, Hlady V. Adhesion mapping of chemically modified
and poly(ethylene oxide)-grafted glass surfaces. Colloid Surface A 154(1-2), 53-64 (1999).
[65] Chen Q, Fan Y, Ayres VM, Udpa L, Schindler MS, Rice AF. Scanning probe recognition
microscopy investigation of the elastic properties of tissue scaffolding. MRS Warrendale, PA
(2005).
[66] Hallab NJ, Bundy KJ, O’Connor K, Moses RL, Jacobs JJ. Evaluation of metallic and
polymeric biomaterial surface energy and surface roughness characteristics for directed cell
adhesion. Tissue Eng 7(1), 55–71 (2001).
[67] Ayres VM, Xie K, Tiryaki VM, Ahmed I, Shreiber DI. Investigation of Nanophysical
Properties of Aging Polyamide Nanofibrillar Tissue Scaffolds by TEM, SAED, Contact Angle
and Raman Spectroscopies. In MRS Boston, MA (2011).
[68] Roggendorf E. The biostability of silicone rubbers, a polyamide and a polyester. J Biomed
Mater Res 10, 123 (1976).

150

[69] Hatamleh MM, Polyzois GL, Silikas N, Watts DC. Effect of extraoral aging conditions on
mechanical properties of maxillofacial silicone elastomer. J Prosthodont 20, 439-446 (2011).
[70] Tiryaki VM, Khan AA, Pastor A, Alduhaileb RA, Delgado-Rivera R, Ahmed I, Meiners S,
Ayres VM. Quantitative investigations of nanoscale elasticity of nanofibrillar matrices. In MRS
Boston, MA, (2010).
[71] Ferrari AC, Robertson J. Resonant Raman spectroscopy of disordered, amorphous, and
diamondlike carbon. Phys Rev B 64, 075414 (2001).
[72] Georges PC, Miller WJ, Meaney DF, Sawyer ES, Janmey PA. Matrices with compliance
comparable to that of brain tissue select neuronal over glial growth in mixed cortical cultures.
Biophys J 90, 3012-3018 (2006).
[73] McKeon RJ, Schreiber RC, Rudge JS, Silver J. Reduction of neurite outgrowth in a model
of glial scarring following CNS injury is correlated with the expression of inhibitory molecules
on reactive astrocytes. J Neurosci 11, 3398–3411 (1991).
[74] Tang X, Davies JE, Davies SJA. Changes in distribution, cell associations, and protein
expression levels of NG2, neurocan, phosphacan, brevican, versican V2, and tenascin-C during
acute to chronic maturation of spinal cord scar tissue. J Neurosci Res 71, 427–444 (2003).
[75] Biran R, Noble MD, Tresco PA. Characterization of cortical astrocytes on materials of
differing surface chemistry. J Biomed Mater Res 46, 150–159 (1999).
[76] Patel KR, Tang H, Grever WE, Simon Ng KY, Xiang J, Keep RF, Cao T, McAllister JP
2nd. Evaluation of polymer and self-assembled monolayer-coated silicone surfaces to reduce
neural cell growth. Biomaterials 27:1519–1526, (2006).
[77] Tonazzini I, Bystrenova E, Chelli B, Greco P, Stoliar P, Calò A, Lazar A, Borgatti F,
D'Angelo P, Martini C, Biscarini F.. Multiscale morphology of organic semiconductor thin films
controls the adhesion and viability of human neural cells. Biophys J 98, 2804–2812 (2010).
[78] Nobes CD, Hall A. Rho, Rac, and Cdc42 GTPases regulate the assembly of multimolecular
focal complexes associated with actin stress fibers, lamellipodia, and filopodia. Cell 81(1), 53–62
(1995).
[79] Biyasheva A, Svitkina T, Kunda P, Baum B, Borisy G. Cascade pathway of filopodia
formation downstream of SCAR. J Cell Sci 117, 837–848 (2004).
[80] Racchetti G, D’Alessandro R, Meldolesi J. Astrocyte stellation, a process dependent on
Rac1 is sustained by the regulated exocytosis of enlargeosomes. Glia 60(3), 465–475 (2011).

151

[81] Chung HY, Hal JRB, Gogins MA, Douglas G. Crofoot, Thomas M. Weik. Donaldson
Company, Inc. Polymer, polymer microfiber, polymer nanofiber and applications including filter
structures. US 6743273 B2, (2004).
[82] Spatz C. Basic Statistics: Tales of Distributions. 10th ed. Belmont: Wadsworth Cengage
Learning (2011).
[83] Tiryaki VM, Khan AA, Ayres VM. AFM feature definition for neural cells on nanofibrillar
tissue scaffolds. Scanning 34, 316 (2012).
[84] Davis-Cox MI, Turner JN, Szarowski D, Shain W. Phorbol ester-stimulated stellation in
primary cultures of astrocytes from different brain regions. Microsc Res Tech 29, 319–327
(1994).
[85] Kasa I. A curve fitting procedure and its error analysis. IEEE Trans Inst Meas 25 8–14,
(1976).
[86] Ridley AJ, Paterson HF, Johnston CL, Diekmann D, Hall A. The small GTP-binding protein
Rac regulates growth factor-induced membrane ruffling. Cell 70:401–410 (1992).
[87] Jones KJ, Korb E, Kundel MA, Kochanek AR, Kabraji S, McEvoy M, Shin CY, Wells DG.
CPEB1 regulates beta-catenin mRNA translation and cell migration in astrocytes. Glia 56(13),
1401–1413 (2008).
[88] Etienne-Manneville S, Hall A. Rho GTPases in cell biology. Nature 420, 629–635 (2002).
[89] Bishop AL, Hall A. Rho GTPases and their effector proteins. Biochem J 348:241–255
(2000).
[90] Ridley AJ, Hall A. The small GTP-binding protein Rho regulates the assembly of focal
adhesions and actin stress fibers in response to growth factors. Cell 70, 389–399 (1992).
[91] Kalman D, Gomperts SN, Hardy S, Kitamura M, Bishop JM. Ras family GTPases control
growth of astrocyte processes. Mol Biol Cell 10, 1665–1683, (1999).
[92] Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image
analysis. Nature Methods 9, 671–675 (2012).
[93] Abe K, Misawa M. Astrocyte stellation induced by Rho kinase inhibitors in culture. Brain
Res Dev Brain Res 143, 99–104 (2003).
[94] Hall A. Rho GTPases and the actin cytoskeleton. Science 279, 509–514 (1998).
[95] Cook GMW. Glycoproteins in membranes. Biol Rev 43, 363–391 (1968).

152

[96] Finkelstein EI, Chao PG, Hung CT, Bulinski JC. Electric field-induced polarization of
charged cell surface proteins does not determine the direction of galvanotaxis, cell motility and
the cytoskeleton. Cell Motil Cytoskeleton 64, 833–846 (2007).
[97] Seil JT, Webster, TJ. Electrically active nanomaterials as improved neural tissue
regeneration scaffolds. Wiley Interdiscip Rev Nanomed Nanobiotechnol 2, 635–647 (2010).
[98] Sivron T, Schwartz M. Glial-cell types, lineages, and response to injury in rat and fish implications for regeneration. Glia 13(3), 157-165 (1995).
[99] Ridet JL, Malhotra SK, Privat A, Gage FH. Reactive astrocytes: cellular and molecular cues
to biological function. Trends Neurosci 20(12), 570-577 (1997).
[100] Wilhelmsson U, Bushongt EA, Price DL, Smarr BL, Phung V, Terada M, Ellisman MH,
Pekny M. Redefining the concept of reactive astrocytes as cells that remain within their unique
domains upon reaction to injury. P Natl Acad Sci USA 103(46), 17513-17518 (2006).
[101] Wiese S, Karus M, Faissner A. Astrocytes as a source for extracellular matrix molecules
and cytokines. Front Pharmacol 3, 120 (2012).
[102] Silver J, Miller JH. Regeneration beyond the glial scar. Nat Rev Neurosci 5(2), 146-156
(2004).
[103] Vignaud T, Galland R, Tseng QZ, Blanchoin L, Colombelli J, Thery M. Reprogramming
cell shape with laser nano-patterning. J Cell Sci 125(9), 2134-2140 (2012).
[104] Koegler P, Clayton A, Thissen H, Santos GNC, Kingshott P. The influence of
nanostructured materials on biointerfacial interactions. Adv Drug Deliver Rev 64(15), 1820-1839
(2012).
[105] Yim EKF, Pang SW, Leong KW. Synthetic nanostructures inducing differentiation of
human mesenchymal stem cells into neuronal lineage. Exp Cell Res 313(9), 1820-1829 (2007).
[106] Wood JA, Ly I, Borjesson DL, Nealey PF, Russell P, Murphy CJ. The modulation of
canine mesenchymal stem cells by nano-topographic cues. Exp Cell Res 318(19), 2438-2445
(2012).
[107] Ahmed I, Ponery AS, Nur-E-Kamal A, Kamal J, Meshel AS, Sheetz MP, Schindler M,
Meiners S. Morphology, cytoskeletal organization, and myosin dynamics of mouse embryonic
fibroblasts cultured on nanofibrillar surfaces. Mol Cell Biochem 301(1-2), 241-249 (2007).
[108] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage
specification. Cell 126(4), 677-689 (2006).

153

[109] Nemir S, West JL. Synthetic materials in the study of cell response to substrate rigidity.
Ann Biomed Eng 38(1), 2-20 (2010).
[110] Leipzig ND, Shoichet MS. The effect of substrate stiffness on adult neural stem cell
behavior. Biomaterials 30(36), 6867-6878 (2009).
[111] Banerjee A, Arha M, Choudhary S, Ashton RS, Bhatia SR, Schaffer DV, Kane RS. The
influence of hydrogel modulus on the proliferation and differentiation of encapsulated neural
stem cells. Biomaterials 30(27), 4695-4699 (2009).
[112] Liu T, Houle JD, Xu JY, Chan BP, Chew SY. Nanofibrous collagen nerve conduits for
spinal cord repair. Tissue Eng Pt A 18(9-10), 1057-1066 (2012).
[113] Schindler M, Ahmed I, Kamal J, Nur-E-Kamal A, Grafe TH, Young Chung H, Meiners S.
A synthetic nanofibrillar matrix promotes in vivo-like organization and morphogenesis for cells
in culture. Biomaterials 26(28), 5624-5631 (2005).
[114] Jiang FX, Yurke B, Firestein BL, Langrana NA. Neurite outgrowth on a DNA crosslinked
hydrogel with tunable stiffnesses. Ann Biomed Eng 36(9), 1565-1579 (2008).
[115] Fedoroff S, Mcauley WaJ, Houle JD, Devon RM. Astrocyte Cell Lineage .5. Similarity of
Astrocytes That Form in the Presence of Dbcamp in Cultures to Reactive Astrocytes In vivo. J
Neurosci Res 12(1), 15-27 (1984).
[116] Tiryaki, VM, Ayres, VM, Ahmed, I, Shreiber, DI. Differentiation of reactive-like
astrocytes cultured on nanofibrillar and comparative culture surfaces. In press, Nanomedicine
(2013).
[117] East E, Golding JP, Phillips JB. A versatile 3D culture model facilitates monitoring of
astrocytes undergoing reactive gliosis. J Tissue Eng Regen M 3(8), 634-646 (2009).
[118] Rao S, Winter J. Adhesion molecule-modified biomaterials for neural tissue engineering.
Front Neuroeng 2, 1-14 (2009).
[119] Dreher Z, Tout S, Stone J. Interactions of living astrocytes in vitro - evidence of the
development of contact spacing. Glia 11(1), 57-63 (1994).
[120] Tiryaki VM, Adia-Nimuva U, Hartz SA, Xie K, Ayres VM, Ahmed I, Shreiber DI. New
atomic force microscopy based astrocyte cell shape index. In MRS. UU: Scanning Probe
Microscopy - Frontiers in Nanotechnology (2013).
[121] Yoshimura S, Sakai H, Nakashima S, Nozawa Y, Shinoda J, Sakai N, Yamada H.
Differential expression of Rho family GTP-binding proteins and protein kinase C isozymes
during C6 glial cell differentiation. Mol Brain Res 45(1), 90-98 (1997).

154

[122] Charrierebertrand C, Leprince G, Bouchier C, Fages C, Tardy M, Nunez J. Expression of
tubulin, GFAP and of their encoding messenger-RNAs during the proliferation and
differentiation of cultured astrocytes. Neurochem Int 15(2), 215-222 (1989).
[123] Georges P, McCormick M, Flanagan L, Ju Y, Sawyer E, Janmey P. Tuning the elasticity of
biopolymer gels for optimal wound healing. In: MRS. Boston, MA (2005).
[124] Taber LA. Towards a unified theory for morphomechanics. Philos T R Soc A 367(1902),
3555-3583 (2009).
[125] Zuidema JM, Pap MM, Jaroch DB, Morrison FA, Gilbert RJ. Fabrication and
characterization of tunable polysaccharide hydrogel blends for neural repair. Acta Biomater 7(4),
1634-1643 (2011).
[126] Wang CC, Tsong TY, HsuYH, Marszalek PE. Inhibitor binding increases the mechanical
stability of staphylococcal nuclease. Biophys J 100, 1094–1099 (2011).
[127] Kreplak L, Wang H, Aebi U, Kong X. Atomic force microscopy of mammalian urothelial
surface. J Mol Biol 374, 365–373 (2007).
[128] Grafe T, Graham K. Polymeric nanofibers and nanofiber webs: a new class of nonwovens.
Proceedings of the International Nonwovens Technical Conference, Atlanta, GA, 24–26 (2002).
[129] Frigo M, Johnson SG. FFTW: an adaptive software architecture for the FFT. Proceedings
of the International Conference on Acoustics, Speech, and Signal Processing 3:1381–1384
(1998).
[130] Gonzalez RC, Woods RE. Digital Image Processing 3rd ed. Prentice Hall, Upper Saddle
River, NJ (2008).
[131] Tiryaki VM, Ayres VM, Khan AA, Flowers DA, Ahmed I, Flowers, DA, Ahmed, I,
Delgado-Rivera, R and Meiners, S. Investigation of nanofibrillar influence on cell-cell
interactions of astrocytes by atomic force microscopies. In MRS, Boston, MA (2011).
[132] Gerdes HH, Carvalho RN. Intercellular transfer mediated by tunneling nanotubes. Curr
Opin Cell Biol 20, 474–475 (2008).
[133] Kienberger F, Pastushenko VP, Kada G, Chtcheglova L, Riethmueller C, Rankl C, Ebner
A, Hinterdorfer P. Improving the contrast of topographical AFM images by a simple averaging
filter. Ultramicroscopy 106, 822–828 (2006).
[134] Liu Q, Wang H, Liu J, Huang H. AFM image processing for estimating the number and
volume of nanoparticles on a rough surface. Surf Interface Anal 43:1354–1359 (2010).

155

[135] Osher SJ, Rudin LI. Feature-oriented image enhancement using shock filters. SIAM J
Numer Anal 27, 919-940 (1990).
[136] Alvarez L, Mazorra L. Signal and image restoration using shock filters and anisotropic
diffusion. SIAM J Numer Anal 31, 590–605 (1994).
[137] Gilboa G, Sochen N, Zeevi YY. Regularized shock filters and complex diffusion.
Proceedings of the European Conference on Computer Vision’02 399–313 (2002).
[138] Zou Z, Zheng Q, Wu Y, Guo X, Yang S, Li J, Pan H. Bio-compatibility and bioactivity of
designer self-assembling nanofiber scaffold containing FGL motif for rat dorsal root ganglion
neurons. J Biomed Mater Res A 95, 1125–1131 (2010).
[139] Hwang CM, Sant S, Masaeli M, Kachouie NN, Zamanian B, Lee SH, Khademhosseini A.
Fabrication of three-dimensional porous cell-laden hydrogel for tissue engineering.
Biofabrication 2, 1–12 (2010).
[140] Zhou M, Smith AM, Das AK, Hodson NW, Collins RF, Ulijn RV, Gough JE. Selfassembled peptide-based hydrogels as scaffolds for anchorage-dependent cells. Biomaterials 30,
2523–2530 (2009).
[141] Lisy M, Pennecke J, Brockbank KG, Fritze O, Schleicher M, Schenke-Layland K, Kaulitz
R, Riemann I, Weber CN, Braun J, Mueller KE, Fend F, Scheunert T, Gruber AD, Albes JM,
Huber AJ, Stock UA. The performance of ice-free cryopreserved heart valve allografts in an
orthotopic pulmonary sheep model. Biomaterials 20, 5306–5311 (2010).
[142] Matsutani S, Yamamoto N. Neuronal regulation of astrocyte morphology in vitro is
mediated by GABAergic signaling. Glia 20, 1-9, (1997).
[143] Vartanian KB, Kirkpatrick SJ, Hanson SR, Hinds MT. Endothelial cell cytoskeletal
alignment independent of fluid shear stress on micropatterned surfaces. Biochem Biophys Res
Commun 371, 787-792 (2008).
[144] Dike LE, Chen CS, Mrksich M, Tien J, Whitesides GM, Ingber DE. Geometric control of
switching between growth, apoptosis, and differentiation during angiogenesis using
micropatterned substrates. In Vitro Cell Dev Biol Anim 35, 441-448, (1999).
[145] Chen CS, Mrksich M, Huang S, Whitesides GM, Ingber DE. Micropatterned surfaces for
control of cell shape, position, and function. Biotechnol Prog 14, 356-363 (1998).
[146] Chen CS, Mrksich M, Huang S, Whitesides GM, Ingber DE. Geometric control of cell life
and death. Science 276(5317), 1425-1428, (1997).
[147] Canham P B, Burton AC. Distribution of size and shape in populations of normal human
red cells. Circ Res 22(3), 405–422 (1968).
156

[148] Chvátal A, Andrová M, Kirchhoff F. Three-dimensional confocal morphometry - a new
approach for studying dynamic changes in cell morphology in brain slices. J Anat 210, 671-683
(2007).
[149] Pincus Z, Theriot A. Comparison of quantitative methods for cell-shape analysis. J
Microsc 227, 140-156 (2007).
[150] Petroski RE, Grierson JP, Choikwon S, Geller HM. Basic fibroblast growth-factor
regulates the ability of astrocytes to support hypothalamic neuronal survival in vitro. Dev Biol
147(1), 1-13 (1991).
[151] Bolle R, Connell J, Haas N, Mohan R, Taubin G. Veggievision: a produce recognition
system. Proceedings of the 3rd IEEE Workshop on Applications of Computer Vision, Sarasoto,
FL, 244–251 (1996).
[152] Pudil P, Novovicova J, Kittler J. Floating Search Methods in Feature Selection. Pattern
Recognition Letters, 15, 119-125 (1994).
[153] Riedmiller M, Braun H. A direct adaptive method for faster backpropagation learning: the
RPROP algorithm. Proceedings of the IEEE International Conference on Neural Networks, 586591 (1993).
[154] Hall-Beyer M. http://www.fp.ucalgary.ca/mhallbey/ Version: 2.10 February (2007).
[155] Haralick RM, Shanmugam K, Dinstein I. Textural features for image classification. IEEE
Transactions on Systems, Man and Cybernetics 3(6), 610-620 (1973).
[156] Korchev YE, Gorelik J, Lab MJ, Sviderskaya EV, Johnston CL, Coombes CR, Vodyanoy
I, Edwards CRW. Cell volume measurement using scanning ion conductance microscopy.
Biophys J 78, 451-457, (2000).
[157] Fardin MA, Rossier OM, Rangamani P, Avigan PD, Gauthier NC, Vonnegut W, Mathur
A, Hone J, Iyengarband R, Sheetz MP. Cell spreading as a hydrodynamic process. Soft Matter 6,
4788-4799, (2010).
[158] Shao W, Jin H, Huang J, Qiu B, Xia R, Deng Z, Cai J, Chen Y. AFM investigation on OxLDL-induced changes in cell spreading and cell-surface adhesion property of endothelial cells
Scanning, 35(2), 119-126 (2012).
[159] Recknor JB, Mallapragada SK. Nerve regeneration: tissue engineering strategies, The
Biomedical Engineering Handbook: Tissue Engineering and Artificial Organs, Taylor & Francis:
New York (2006).
[160] Rolls A, Shechter R,Schwartz M. The bright side of the glial scar in CNS repair. Nat Rev
Neurosci 10(3), 235–241 (2009).
157