Regulation of Cell Motility on Polymer Substrates via ...

TISSUE ENGINEERING
Volume 8, Number 2, 2002
© Mary Ann Liebert, Inc.

Regulation of Cell Motility on Polymer Substrates via
“Dynamic,” Cell Internalizable, Ligand Microinterfaces

JANE S. TJIA, Ph.D.,1 and PRABHAS V. MOGHE, Ph.D.1,2

ABSTRACT

In vivo, motile epidermal tissues frequently encounter ligand microinterfaces that are dy-
namic, owing to rapid cell-mediated substrate phagocytosis. In this study, we have exam-
ined cell motility phenomena in response to the adhesion ligand, collagen, which was pre-
sented on cell-internalizable 400-nm colloidal gold microcarriers. Normal human
keratinocytes were seeded onto collagen-adsorbed poly(lactide-co-glycolide) (PLGA) films
that were predeposited with varying densities of the ligand-associated microcarriers (LAMs),
such that the overall ligand density and the ligand loading per microcarrier were invariant.
Cells seeded on LAMs exhibited rapid and distinct cytoskeletal redistribution resulting in
numerous filopodial extensions, indicative of activation of cell motility processes. We report
that the population-averaged cell migration rate of cells, m, was increased from 2 mm2/min
on collagen substrates lacking the microcarriers, to ,50 mm2/min on collagen presenting
LAMs. An increase in LAM density to 1.2 LAMs per mm2 was found to maximize m, as well
as the directional persistence and speed of cell migration, whereas very high LAM densities
saturated cell internalization and diminished migration rates. The cell–LAM interactions es-
sential to enhanced migration were ligand-activated, as m values were reduced 400% on in-
ternalizable microcarriers lacking ligands; moreover, cooperative ligand elicitation was pos-
sible, for example, when 10 mg/mL soluble fibronectin was introduced as a costimulant of
keratinocytes, yielding the highest reported m values (80 mm2/min) on collagen LAM-PLGA
substrates. Notably, cell migration rates were severely repressed when cell internalization
processes were challenged through covalent conjugation of ligand carriers to the substrate,
indicating that signals from cell–LAM binding alone were inadequate for elevated levels of
cell migration. Further analysis indicated that the presence of LAMs did not alter the pro-
tease-resistant adhesivity between the cell and the underlying substrate, suggesting that the
activation governing ligand interactions likely arose at the cell–LAM interface rather than
the cell–substrate interface. This study highlights the novel use of secondary ligand-pre-
senting microscale/nanoscale depots at polymer substrates to elicit, via dynamic cell inter-
nalization processes, significantly enhanced levels of cell migration over traditional inter-
faces with ligand-bearing substrates.

1The Department of Chemical and Biochemical Engineering and 2Department of Biomedical Engineering, Rutgers
University, Piscataway, New Jersey.

247

TJIA AND MOGHE

INTRODUCTION

DYNAMIC LIGAND INTERFACES that target ligand-carrying microvehicles directly to a cell population of in-
terest can result in increased sampling efficiency and bioavailability of the delivered molecule. Ligand-
bound microparticle based systems have been widely utilized for the intracellular delivery of drugs, genes,
and peptides, as well as for the rapid and precise detection of protein antigens and molecules.1 In addition
to cell targeting, ligand interfaces also play an important role in a variety of in vivo cellular activities, in-
cluding differentiation, proliferation, and migration, which are spatiotemporally regulated by cell receptor
interactions with the local ligand field.

Cell migration is central to a number of physiologic phenomena, and there has been much interest in de-
termining ligand properties that govern cell migration processes, particularly on ligand-borne biomaterials
that are widely used in the field of tissue engineering.2,3 Systematic variations in cell migratory responses
to ligands have been widely investigated by the modulation of the ligand substrate density, receptor-ligand
affinity, and the number of integrin receptors,4,5 and have been examined by altering the availability of in-
tegrin cell surface receptors via antibody blocking.6 More recently, variations in the two-dimensional (2-
D) spatial organization of substrate ligands have been shown to alter the rate of the establishment of ligand-
receptor contracts and to serve as a basis to selectively activate intracellular signaling pathways controlling
cell migration.7 Cell migratory behavior has also been shown to be enhanced in three-dimensional (3-D) li-
gand clustered systems, wherein ligand properties have been varied in conjunction with the microgeome-
try of the substrate.8 Since receptor clustering is known to be necessary for full migratory activation,9 3-D
microconfiguration of ligand interfaces may be a significant parameter to activate receptor-regulated cell
migration. However, significantly elevated levels of local receptor-ligand binding can also suppress cell
motility in these systems, because of increased cell adhesion to the ligand immobilized substrate, a limita-
tion that may be potentially overcome using a dynamic type of a ligand microinterface.

We have previously proposed the use of a secondary microinterface between ligands and substrate-de-
posited microcarriers, which can activate receptor-based cell signals transiently, and which is dynamic in
the sense that cells internalize the ligand.10 Similar transient interactions may exist during the unique phago-
kinetic migration of skin keratinocytes during wound healing, as cells can rapidly migrate on matrices with
high ligand concentration internalizing the matrix along their migratory path.11 Our studies have specifi-
cally focused on the migration behavior of keratinocytes, on poly(lactide-co-glyolide) (PLGA) substrates,
in the presence of ligand-bound colloidal gold microcarriers.10 The ligand-adsorbed microcarriers (LAMs)
exhibited dynamic interactions with migratory keratinocytes and were rapidly cleared upon binding to the
cells through phagocytic internalization as the cells migrated.12

In this study, we have examined a model system of cell phagokinetic migration on ligand immobilized
substrates, following the systematic incorporation of a second, dynamic ligand-presenting microinterface.
The microinterface was established through varying densities of collagen-bound 400-nm scale gold micro-
carriers deposited on the underlying PLGA substrate (Fig. 1). Our studies show that an active, dynamic in-
teraction with the LAMs can significantly enhance the rate of cell migration elicited by the adhesion ligand.
The enhancement is directly governed by the extent of cell–LAM interactions and is dependent on cell-me-
diated LAM internalization.

MATERIALS AND METHODS

Cell culture

Human keratinocytes from neonatal foreskin were isolated by enzyme digestion13 and cultured in serum-
free keratinocyte growth medium (KGM; Clonetics, San Diego, CA) containing 0.1 ng/mL epidermal growth
factor (EGF), 5 mg/mL insulin, 0.5 mg/mL hydrocortisone, 50 mg/mL gentamicin, 50 ng/mL amphotericin-
B, 0.15 mM calcium, and 30 mg/mL bovine pituitary extract (BPE). Each isolation was screened for con-
taminating fibroblasts using indirect immunofluorescence staining for keratin 14 and vimentin (Sigma, St.
Louis, MO). In all experiments, cells were used at passage 2 or 3. To fully define the culture media for ex-

248

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES

FIG. 1. Schematic illustration of ligand-bound microcarrier coated substrates. Collagen ligands were adsorbed onto
PLGA-coated coverslips in the presence or absence of microcarriers such that the overall surface concentration of the
ligand was constant. Soluble FN was added as an exogenous stimulant in selected conditions and is not diagrammed
here.

periments, media was switched to KGM without BPE and EGF at 6 h prior to the initiation of the experi-
ment.
Substrate preparatio n

Glass coverslips (25-mm-diameter) were cleaned using soap and sonication, and stored in 100% ethanol
until use. Thin polymer films were obtained by spincoating a 1% w/v solution of 50:50 poly-D,L(lactic
acid–glycolic acid) (PLGA; Medisorb, Cincinnati, OH) in chloroform onto cleaned coverslips. PLGA films
were then coated with type I rat tail collagen (Collaborative, Bedford, MA) at a concentration of 5 mg/cm2
by slowly evaporating 1 mL of a 24.2 mg/mL collagen solution in 0.2 N acetic acid at room temperature.
The substrates were then placed in six-well plates and held in place with a small amount of silicone glue.

Prepared substrates were coated with collagen-adsorbed particulate gold salts in order to stimulate phago-
cytosis. Gold salts were prepared as previously described.14 Briefly, 11 mL of distilled water was combined
with 6 mL of 36.5 mM Na2CO3 and 1.8 mL of 14.5 mM AuCl4H (J.T. Baker, Phillipsburg, NJ). The so-
lution was heated just to boiling, and 1.8 mL of 0.1% formaldehyde was added. In order to prevent any
substrate denaturation, the solution was then cooled to below 70°C. The cooled solution was layered over
the prepared substrates (2 mL) and incubated at room temperature for 45 min. The addition of the gold so-
lution at these temperatures results in desorption of a significant portion of the collagen. The collagen then
binds to the gold particles through noncovalent interactions. In some experiments, substrates were coated
with a solution made without AuCl4H prior to the addition of gold particle solution to obtain final sub-
strates with nonligand adsorbed microparticles. We have previously shown that substrates prepared in this
manner contain a final collagen concentration of 612 ng/cm2.10

Experiments were also performed with substrates on which the gold particles were covalently attached.
Since we have previously found the extent of migration on collagen-adsorbed PLGA is equivalent to that
observed on collagen-adsorbed glass,10 the gold particles were attached directly to glass coverslips. Glass
coverslips were covalently modified using chemistry similar to that previously used to micropattern sur-
faces in order to examine cell migratory and adhesion phenomena.15–17 Surface hydroxyl groups on cleaned
glass coverslips were first activated by immersion in 20 mL of a solution containing nine parts sulfuric acid
(Fisher, Pittsburgh, PA) and one part 30% hydrogen peroxide in water (Sigma, St. Louis, MO). After 15

249

TJIA AND MOGHE

min, the coverslips were extensively rinsed with water and then air-dried. The activated coverslips were
then modified with an amine functionalized silane. Briefly, activated coverslips were immersed in a 1%
solution of N-(2-aminoethyl)-3-aminopropyltrimethoxysilane (Aldrich, Milwaukee, WI)/ethanol/water in
a 1:95:4 v/v/v ratio. After 2 min of sonication, the coverslips were thoroughly rinsed with ethanol and
then dried at 105°C for 30 min.15 After collagen adsorption onto these activated coverslips, addition of
the gold particle solution resulted in substrates containing covalently attached ligand-adsorbed mi-
croparticles. 18

In order to decrease the surface particle concentration, the coating solution was diluted. To increase
the surface concentration, the total amount of gold solution layered over the substrates was increased.
Previous studies have shown that the average diameter of gold particles formed in this manner is ap-
proximately 400 nm.19 In order to verify the final surface particle concentration, transmitted microscopy
images of the substrate surface were obtained at 1003 magnification and analyzed using image analy-
sis. In experiments where no gold particulates were used, substrates were coated with a solution made
without AuCl4H.

Cytoskeletal morphology

Indirect immunofluorescence was used to visualize the actin cytoskeleon. Seeded keratinocytes were al-
lowed to attached to substrates for 2 h and were then fixed with a solution of 3% formaldehyde in PBS.
After permeabilization with 0.1% Triton-X in PBS for 4 min, cells were incubated for 45 min at room tem-
perature with a dilution of fluorescein phalloidin (Molecular Probes, Eugene, OR). Cells were then visual-
ized using fluorescence microscopy on a Zeiss Axiovert microscope (Carl Zeiss, Inc., Thornwood, NY).

Time-lapse microscopy and image analysis of cell migration

Keratinocytes were plated onto prepared substrates in six-well plates in KGM media without BPE at a
density of 2000 cells/cm2 and allowed to attach for 2 h at 37°C. Thirty ng/mL of EGF (Sigma) was si-
multaneously added with the cells. In some experiments, soluble fibronectin stimulation was incorporated
by adding 10 mg/mL fibronectin (Sigma) to the media. After the attachment period, the six-well plate was
transferred to the motorized, heated stage of a Zeiss Axiovert laser scanning microscope (Carl Zeiss, Inc.).
The microscope stage was kept at 37°C using a specially designed, temperature-controlled, humidified,
Plexiglas incubation chamber. All cell tracking was performed under brightfield at 103 magnification with
1.53 zoom. Using LSM 410 software (Carl Zeiss, Inc.), a macro was written to store and track three rep-
resentative fields of view from each coverslip in the 6 well plate. Scanned images were obtained every 3
min up to a total time of 5 h. Images were then processed using Image-Pro software (Media Cybernetics,
Silver Spring, MD) to analyze the paths of the individual migrating cells. Between 5 and 12 cells were an-
alyzed in each field of view. Only cells whose paths did not intersect those of other cells over the migra-
tion period were selected for analysis. Selected cells were highlighted and their centroids at each time point
recorded. In addition, the area cleared by the cells at each time point was also determined.

Determination of random motility coefficient, m

The random cell motility coefficient was determined using the formulation originally proposed by Dunn
and later by Othmer et al.20,21 Briefly, the random motility of individual cells was characterized by re-
gression of the mean-squared displacement behavior of the cell population as a function of time:

kd2(t)l 5 4m[t 2 P(1 2 e2t/P)] (1)

where kd2(t)l is the mean-squared displacement, P is the cell directional persistence time, and t is the time
interval of the experiment. Using the (x,y) coordinates of the cell centroids, squared displacements were
calculated to be:

d2(t) 5 (xt1Dt 2 xt)2 1 (yt1Dt 2 yt)2 (2)

250

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES
In order to take advantage of all positional data available, overlapping intervals were used. Mean squared
displacements were calculated at given time intervals and fit to Equation 1 using the Levenberg-Marquardt
nonlinear least squares regression algorithm to obtain estimates for m and P.
Cell adhesion strength

We found that traditional mechanical assays used to measure cell adhesion strength, such as the cen-
trifugation assay5 or the shear flow assay,22 were unable to adequately determine the cell adhesion strength
in our system. Thus, we used an enzymatic assay that measures the trypsin sensitivity of adhered cells to
determine the effect of LAMs on cell–substrate adhesion strength. This assay has previously been shown
to accurately determine the cell–substrate adhesion strength of cells adhered to synthetic polymer sub-
strates.23,24

Keratinocytes were seeded onto prepared substrates and allowed to adhere for 30 min. Adhered cells were
then treated with 0.008% trypsin-EDTA (Clonetics) for 5 min to 1 h at 37°C with gentle shaking. Cells de-
tached during treatment were counted using a hemocytometer. Aliquots were taken a 5, 10, 20, 40, and 60
min, and initial volumes were restored by the addition of fresh trypsin-EDTA solution. Nondetached cells
at the end of enzymatic treatment were released by 0.25% trypsin-EDTA within 15 min. The area beneath
a graph of the percentage of cells released as a function of time was taken to be inversely proportional to
the strength of cell adhesion to the substrate.

FIG. 2. Ligand microcarrier activation of cytoskeletal stress fiber organization. Visualization of F-actin stress fibers
in the absence of microcarriers (A), in the presence of substrate-adsorbed ligand microcarriers at 1.21 LAMs/mm2 (B),
in the presence of ligand microcarriers covalently attached to the substrate (C), and in the presence of adsorbed mi-
crocarriers without ligand (D). No fibronectin was incorporated in any of these conditions.

251

TJIA AND MOGHE

FIG. 3. Enhanced keratinocyte migration via induction of ligand-mediated phagocytosis. Keratinocytes were seeded
onto prepared substrates in the absence (A–C) or presence (D–F) of ligand microcarriers and tracked using time-lapse
microscopy. Ligand microcarrier density in D–F is 1.21 LAMs/mm2. Representative images are shown at 0 h (A,D),
2.5 h (B,E), and 5 h (C,F).

252

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES
RESULTS

Effects of LAMs on cell morphology
The actin cytoskeleton of migrating cells was visualized in the presence and absence of ligand-adsorbed

microcarriers (Fig. 2). In the absence of microcarriers, distinct stress fibers were seen throughout the cell.
However, in the presence of LAMs, cells were found to exhibit a significantly greater amount of filopodial
extensions with fewer distinct stress fibers evident throughout the cell. In the absence of ligand on the mi-
crocarriers, as well as when LAMs were covalently conjugated to the substrate, cells still exhibited distinct
filopodia, but the extensions were shorter and fewer in number.

FIG. 4. Effect of ligand microcarrier concentration on the mean-squared displacement. Temporal change of the mean
squared displacement of migrating keratinocytes as a function of the initial LAM substrate density in the presence of
basal activation (A) and fibronectin stimulation (B). For each condition, 40–50 cells were analyzed. Bar 5 SE.

253

TJIA AND MOGHE

FIG. 5. Effect of ligand microcarrier concentration on the cell random motility coefficient. Estimates of the random
motility coefficient were obtained as a function of the initial ligand microcarrier density. For each carrier concentra-
tion, 40–50 cells were analyzed. Bar 5 SE.

Cell migration behavior on LAM substrates

Time-lapse microscopy was used to track migrating cells under various conditions. Representative im-
ages of cells in the absence and presence of LAMs are shown in Figure 3. In order to quantify migration,
the random motility coefficient was calculated using nonlinear least-squares regression. A comparison of
cell population averaged mean squared displacement values with respect to time revealed a biphasic re-
sponse to surface LAM density (Fig. 4), both in the presence and absence of fibronectin stimulation. Val-
ues of m were found to confirm the biphasic trend as a function of ligand-bound gold microcarrier surface
concentration with an maximum value of m of 49.69 6 12.25 mm2/min found to occur at a particle con-
centration of 1.216 LAMs/mm2 (Fig. 5). In the absence of particles, the value of m was found to be 1.75 6
0.27 mm2/min, and at 3.15 particles/mm2, m was found to be 4.87 6 0.74 mm2/min. Stimulation with fi-

TABLE 1. EFFECT OF LAM DENSITY ON CELL MOTILITY PARAMETERS, CELL SPEED, AND PERSISTENCE
BEHAVIOR IN THE PRESENCE AND ABSENCE OF FN STIMULATION

0 mg/mL FN stimulation 10 mg/mL FN stimulation

LAM density Cell speed Persistence time Cell speed Persistence time
(mm/min) (min) (mm/min) (min)

0 0.78 6 0.12 22.52 6 5.15 0.83 6 0.09 20.55 6 0.07

0.69 0.93 6 0.23 37.70 6 7.45* 1.39 6 0.23* 40.38 6 7.99*

1.21 0.98 6 0.10 104.23 6 10.79* 1.42 6 0.34* 54.77 6 7.90*

3.15 0.60 6 0.24 25.30 6 1.64 0.69 6 0.02 22.57 6 3.12

Estimates for root mean squared cell speed and directional persistence time were regressed from the mean squared dis-
placement behavior versus time.

*A statistically significant difference, p , 0.05, between the values for a given condition and respective control (zero
LAM density). For each LAM density, 40–50 cells were analyzed.

254

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES

FIG. 6. Roles of ligand specificity and microcarrier mobility in cell motility. Estimates of the random motility coef-
ficient for migrating cells were obtained in the absence or presence of collagen ligand on the carriers and when inter-
nalization was inhibited by covalent attachment of the ligand adsorbed microcarriers to the substrate. In addition, ex-
periments were also performed in the presence of stimulation with 10 mg/mL fibronectin. All data were obtained at a
carrier density of 1.21 microcarriers/mm2. For each condition, 40–50 cells were analyzed. Bar 5 SE.

bronectin caused a significant increase in the maximum value of m to 79.46 6 9.67 mm2/min (Fig. 5). How-
ever, the particle concentration at which this maximum occurred remained at 1.21 particles/mm2. Persis-
tence times and cell speeds were also calculated (Table 1). In the absence of fibronectin, cell speeds were
found to be relatively unchanged with an average value of 0.831 6 0.07 mm/min. Persistence times were
found to exhibit a biphasic trend similar to that observed with the random motility coefficient, with the
greatest persistence time of 104.23 6 10.79 min occurring at a LAM density of 1.21 LAMs/mm2. Follow-
ing 10 mg/mL fibronectin stimulation, however, persistence times decreased to a maximum value of
54.766 6 7.90 min. Moreover, modest biphasic trends were observed for both cell speed and persistence
time with respect to LAM density.

In the absence of collagen ligand on the microparticles, cells were relatively non-motile with a m value
of 12.51 6 1.99 mm2/min at a LAM concentration of 1.216 particles/mm2 (Fig. 6). The presence of colla-
gen ligand on the particles significantly increased m to 49.69 6 12.25 mm2/min. Prevention of cellular in-
ternalization of the LAMs via covalent attachment to the underlying substrate lowered values of m to 5.01 6
0.92 mm2/min. The presence of soluble fibronectin was found to markedly enhance m values on collagen
LAM substrates. This effect required internalizable, collagen adsorbed microcarriers, since fibronectin alone,
or on collagen-deficient microcarriers (data not shown) did not elicit equivalent levels of m.

Using studies of LAM coverage dynamics over time, the amount of cleared area per cell was plotted ver-
sus time (Fig. 7). In general, clearance kinetics was uniformly maintained at low and intermediate LAM den-
sities; however, at 3.15 LAMs/mm2, the clearance profiles approach a plateau by 5 h. In general, the net LAM
clearance behavior qualitatively correlates with that of the rate of cell migration, with the greatest amount of
clearance observed with cells migrating on the optimal LAM density (1.216 LAMs/mm2; Fig. 7). Higher and
lower LAM densities resulted in lower levels of clearance. For the condition of highest motility, the time de-

255

TJIA AND MOGHE

FIG. 7. Effect of phagocytosis on cleared area. Using time-lapse microscopy and image analysis, the cumulative
cleared area was experimentally determined at various carrier concentrations under conditions of basal activation (A)
and fibronectin stimulation (B).
rivative of clearance area was greater in the presence of fibronectin (Fig. 7B), indicating that fibronectin cos-
timulation could consistently enhance clearance kinetics prior to saturation effects due to internalized LAMs.

Effect of LAMs on cell adhesion strength
The relative biological adhesion strength of cells adhered to substrates with and without microcarriers

was assessed using an enzymatic assay. After 60 min of enzymatic treatment, approximately 100% of ad-
256

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES

FIG. 8. Effect of ligand-adsorbed microcarriers on cell-substrate adhesion. The relative adhesion strength of adhered
keratinocytes was determined by enzymatic assay. The cumulative percentage of cells released after various times of
enzymatic treatment was determined via hemocytometer cell counts.

hered cells had become detached. No significant differences were observed between substrates containing
microcarriers and those without carriers (Fig. 8).

DISCUSSION
Engineering ligand-specific cell motility processes is key to the control of tissue interactions with im-
planted biomaterials25,26 and during cell transplantation for tissue engineering applications.27,28 In this study,
we have quantitatively examined the behavior of keratinocyte migration in the presence of a secondary, dy-
namic ligand interface. We report that the induction of ligand-mediated phagocytic processes via LAMs
can significantly enhance cell migratory responsiveness to the basal substrate ligand.
Our studies show that keratinocyte binding to ligand carriers alone is not sufficient to activate signifi-
cant migratory activity; LAM internalization appears to be an essential coactivator for enhanced migration.
In fact, covalent immobilization of LAMs to the underlying substrate resulted in severely repressed migra-
tion. Previous investigations have presented evidence suggesting that phagocytic internalization events may
contribute to motility dynamics by providing a means for membrane and integrin receptor recycling.29,30
In addition, the phagocytic internalization process can effect cytoskeletal activation pathways that play a
role in cell migration.31–33 It is possible, therefore, that the LAM internalization process contributes to cell
migratory activation through temporally accumulated signaling intermediates that comediate migration as
well as cytoskeletal processes, and like conventional phagocytic processes, effects faster receptor recycling.
Cell migration on ligands immobilized to substrates is induced via the activation of intracellular signal
transduction pathways arising from receptor-ligand binding.34,35 By using ligand-presenting microcarriers
or nanocarriers as secondary ligand depots in our system, we expect that the migratory activation may be
promoted by altering the rate or degree of ligand exposure. The progressive increase in intracellular acti-
vation due to LAMs is indicated by the differences in membrane spreading activity.36 A significant increase
in the number of long filopodial extensions was observed in cells exposed to internalizable LAMs com-
pared to those exposed to ligand-deficient microcarriers; very few filopodia were seen at all on purely li-

257

TJIA AND MOGHE

FIG. 9. Parameters hypothesized to govern rate of cell–LAM dynamics underlying migratory activation. Alteration
of the ligand density on the LAM microcarriers could lead to variations in the number of bound ligands, possible al-
tering the rate of signal transduction. In addition, variation in the substrate LAM density, in addition to altering the
number of bound ligands, could also effect the number of LAMs which are internalized, and is a possible means of
controlling the balance between the rates of ligand binding and internalization.

gand immobilized substrates (no LAMs) or with carriers that were noninternalizable. Thus, the relatively
mobile nature of the LAMs is critical for enhanced migration, and the increase in migratory activity ob-
served is not simply due to the microtopography imparted to the substrate by the LAMs.

In our studies, phagokinetic cell migration processes were quantitatively characterized in terms of two
components: the rate of net mean squared displacement (msd) of cells and the population-averaged clear-
ance, which quantifies progressive decrease in the degree of LAM coverage. By comparing these two in-
dices, one can infer cell motility characteristics as a function of LAM density and the presence of a co-
stimulant. For example, at 1.21 LAMs/mm2, the clearance values as well as msd values were observed to
increase monotonically over time, indicating a uniform rate of “new encounters” between cells and LAMs.
On the other hand, for 3.15 LAMs/mm2, the clearance profile levels off, (with very low values of msd), in-
dicating that internalized LAM concentration reaches saturation, leading to impaired cell migration. The
fact that the combination of increasing msd but a saturation in clearance was not observed in any of our
conditions, indicates that new LAM interactions are critical for increased migration in our system. It is also
possible that the rapid cell-LAM binding events cannot sufficiently “turn on” signal transduction processes
sustaining migration—continuous LAM stimulation may be necessary to sustain enhanced motility. Rea-
sons for this phenomenon are still unclear, although we believe that the relative time constants for cell motil-
ity and cell phagocytosis may underlie the rate-limiting signaling processes arising from cell–LAM inter-
actions, which will be the subject of a subsequent theoretical study.12

Regardless of the stimulatory conditions used, LAM density was observed to govern cell motility in a
distinctly biphasic behavior. It is notable that the accompanying changes in the cell persistence behavior
were also distinctly biphasic (Table 1), particularly in the absence of FN stimulation. We believe that cell
binding to LAMs strengthens the directional persistence behavior, and that LAM density can intensify this
further, within limits. Why 1.21 LAMs/mm2 promote persistence is unclear, although it is likely that at this

258

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES

density, LAM clearance at the leading edge of cells can be “spatially autocatalytic” for further LAM bind-
ing and clearance, thus eliciting tracks with highly correlated nature. Cell motility has been similarly shown
to correlate with persistence in systems where contact guidance due to ligand concentration predisposes per-
sistence behavior.37 At very high LAM density, in our system, intracellular LAM saturation downregulates
LAM binding and migratory activation, further decreasing persistence behavior. Interestingly, the presence
of soluble FN stimulation lowers persistence (indicating that cell binding has less regioselectivity), while
enhancing cell speed (presumably via upregulated activation pathways following LAM binding10), as well
as enhancing the cell speed sensitivity to LAM density. The latter effect indicates that the time constant for
binding and migration may be lowered in the presence of fibronectin. Overall, we believe that LAM den-
sity controls the initial rate of cell–LAM interactions, and that there exists a threshold range of cell-LAM
interactions that activates migration. (Extremely high levels of cell–LAM interactions lead to local clear-
ance, but cause only insignificant increases in msd.) Our data suggests that once migration is induced, the
rate of cell–LAM interactions may be dependent upon both the LAM density, and the rate of cell migra-
tion. There is, therefore, an optimal LAM density that is able to stimulate migration at a rate that may be
conducive to maintaining an optimal rate of cell–LAM interactions. At higher LAM densities, the rate of
initial cell–LAM interactions is high, initially activating migration, but the rate of migration combined with
the high LAM density quickly leads to saturation of cell–LAM binding sites, which results in a decrease
in the rate of new cell–LAM interactions, resulting in reduced migration. At far lower LAM densities, on
the other hand, the rate of cell–LAM interactions is not rapid enough to activate migration.

Previous studies using substrate adsorbed cell adhesion proteins have shown that the rate of cell migra-
tion can be ultimately limited by the short-term adhesion strength of the cell to the substrate, which in turn
is governed by the initial ligand–receptor interactions.5,38 In contrast, our studies highlight a configuration
wherein, regardless of the presence of microcarriers, the short-term values of keratinocyte adhesion (ex-
pressed in units of protease-resistant adhesion), remained uniformly low and invariant even under condi-
tions of significant changes in the levels of cell motility (Fig. 7). We expect that, in our system with LAMs,
the substrate–cell adhesion strength is primarily governed by the number of ligand–receptor interactions at
the substrate, not by the receptor–ligand interactions at the secondary LAM interface, which is relatively
mobile in nature. Since the overall (LAM/substrate) collagen ligand density used in our studies (0.6 mg/cm2)
is significantly lower that previously reported to be optimal for keratinocyte migration on collagen (3.7–4.4
mg/cm2),39,40 in all of our studies herein, cell–substrate adhesion strength is low and, in the absence of
LAMs, keratinocyte motility is limited by weak traction forces on the underlying substrate. However, the
presence of LAMs induces phagocytosis, during which temporally coordinated forces within the cell cy-
tostructure and cell spreading can result in the generation of a significant cell contractile force.41,42 We be-
lieve that the presentation of collagen ligand on the internalizable microcarriers is necessary in our system
to cause increased activation of the signal transduction pathways, resulting in the enhancement of a con-
tractile force that is able to overcome the low adhesion strength to the substrate. We have preliminary data
suggesting, for example, that the net intracellular tyrosine kinase activity of motile cells34 depends on the
rate of LAM internalization,12 which is cooperatively upregulated by collagen–LAM and soluble fi-
bronectin.10 For intermediate levels of LAM density, the LAM binding dynamics (sensitively governed by
the collagen concentration) may also influence the tyrosine kinase specific activation. The balance between
the relative rates of cell–LAM binding versus internalization and their variation via alteration of the sur-
face LAM density and microcarrier ligand concentrations, may, therefore, be parameters to control migra-
tory activation on these substrates (Fig. 9). Understanding how cell motility processes are regulated at the
level of cell–LAM interactions will have to await further mechanistic studies.12

In conclusion, we have shown that the controlled induction of phagocytic events through substrate-asso-
ciated ligand-microcarriers can be used to increase cell motility responsiveness to a substrate ligand. We
propose that there exists a parametric regimen wherein presentation of ligand on dynamic, mobile micro-
carrier substrates may result in enhanced activation of migration. This approach can sustain greatly increased
migration for a given ligand field due to the “decoupling” of biophysical effects of the primary ligand (the
substrate adhesion strength) and activation induced through a secondary ligand, and through further acti-
vation due to microcarrier internalization. The results from our studies could have direct implications for
the design of biomimetic strategies toward directed epithelialization of skin grafts as well as the emerging

259

TJIA AND MOGHE

fields of tissue-targeted cell delivery43 and functional noninvasive bioimaging, wherein the ligand specific
uptake of nano- or microparticulate matter may be used for the imaging and detection of alterations in tis-
sue molecular properties. Applications also exist in the fields of functional proteomics,44 rapid biosensors,45
and models of phagokinetic cell migration during wound healing.

ACKNOWLEDGMENTS

This study was supported by Johnson & Johnson Discovery Award, ConvaTec Young Professor Award,
Hoechst Celanese Innovative Research Award, and a National Science Foundation CAREER Award to P.
Moghe. We would like to thank Howard Salis, a NJ Center for Biomaterials Summer Fellow, and Patrick
Hossler for their help with image analysis. J. Tjia was partially supported by the Rutgers–UMDNJ National
Institutes of Health Biotechnology Training Grant, and the Rutgers University Louis Bevier Fellowship.

REFERENCES

1. Weissleder, R., Moore, A., Mahmood, U., et al. In vivo magnetic resonance imaging of transgene expression. Nat.
Med. 6, 351, 2000.

2. Dee, K.C., Anderson, T.T., and Bizios, R. Osteoblast population migration characteristics on substrates modified
with immobilized adhesive peptides. Biomaterials 20, 221, 1999.

3. Langer, R., and Vacanti, J.P. Tissue engineering. Science 260, 920, 1993.
4. Maheshwari, G., Wells, A.G., Griffith, L.G., et al. Biophysical integration of effects of epidermal growth factor

and fibronectin on fibroblast migration. Biophys. J. 76, 2814, 1999.
5. Palecek, S.P., Loftus, J.C., Ginsberg, M.H., et al. Integrin-ligand binding properties govern cell migration speed

through cell-substratum adhesiveness. Nature 385, 537, 1997.
6. Duband, J.-L., Dufour, S., Yamada, S.S., et al. Neural crest cell locomotion induced by antibodies to beta-1 inte-

grins. J. Cell Biol. 122, 729, 1991.
7. Maheshwari, G., Brown, G., Lauffenburger, D.A., et al. Cell adhesion and motility depend on nanoscale RGD clus-

tering. J. Cell Sci. 113, 1677, 2000.
8. Ranucci, C.S., and Moghe, P.V. Substrate typography can enhance cell adhesive and migratory responsiveness to

matrix ligand density. J. Biomed. Mater. Res. 54, 140, 2001.
9. Dedhar, S., and Hannigan, G.E. Integrin cytoplasmic interactions and bidirectional transmembrane signaling. Curr.

Opin. Cell Biol. 8, 657, 1996.
10. Tjia, J.S., Aneskievich, B.J., and Moghe, P.V. Substrate-adsorbed collagen and cell-secreted fibronectin concert-

edly induce cell migration on poly(lactide-glycolide) substrates. Biomaterials 20, 2223, 1999.
11. Clark, R.A.F., Ed. The Moelcular and Cellular Biology of Wound Repair, 2nd ed. New York: Plenum Press, 1996.
12. Tjia, J.S., and Moghe, P.V. Cell-internalized ligand microinterfaces on biomaterials: Design of regulatory deter-

minants of cell migration. In: Dillow, A., and Lowman, T. (eds.). Biomimetic Materials and Design: Biointerfacial
Strategies, Tissue Engineering and Targeted Drug Delivery. New York: Dekker, 2002, pp. 335–373.
13. Stenn, K.S., Link, R. Moellmann, G., et al. Dispase, a neutral protease from Bacillus polymyxa, is a powerful fi-
bronectinase and type IV collagenase. J. Invest. Dermatol. 93, 287, 1989.
14. Albrecht-Buehler, G. The phagokinetic tracks of 3T3 cells. Cell 11, 395, 1977.
15. Saneinejad, S., and Shoichet, M.S. Patterned glass surfaces direct cell adhesion and process outgrowth of primary
neurons of the central nervous system. J. Biomed. Mater. Res. 42, 13, 1998.
16. DiMilla, P.A., Folkers, J.P., Biebuyck, H.A., et al. Wetting and protein adsorption of self-assembled monolayers
of alkanethiolates supported on transparent films of gold. J. Am. Chem. Soc. 116, 2225, 1994.
17. Matsuda, T., and Sugawara, T. Control of cell adhesion, migration and orientation on photochemically micropat-
terned surfaces. J. Biomed. Mater. Res. 32, 165, 1996.
18. Yi, X., Huang-Xian, J., and Hong-Yuan, C. Direct electrochemistry of horseradish peroxidase immobilized on a
colloid/cysteamine-modified gold electrode. Anal. Biochem. 278, 22, 2000.
19. Albrecht-Buehler, G., and Yarnell, M.M. A quantitation of movement of marker particles in the plasma membrane
of 3T3 mouse fibroblasts. Experimental methods. Exp. Cell Res. 78, 67, 1973.
20. Dunn, G.A. Characterizing a kinesis response: time averaged measures of cell speed and directional persistence.
Agents Actions Suppl. 12, 14, 1983.
21. Othmer, H.G., Dunbar, S.R., and Alt, W. Models of dispersal in biological systems. J. Math. Biol. 26, 263, 1988.

260

CELL MOTILITY AT DYNAMIC LIGAND MICROINTERFACES

22. van Kooten, T.G., Schakenraad, J.M., van der Mei, H.C., et al. Influence of substratum wettability on the strength
of adhesion of human fibroblasts. Biomaterials 13, 897, 1992.

23. Lampin, M., Warocquier-Clerout, R., Legris, C., et al. Correlation between substratum roughness and wettability,
cell adhesion, and cell migration. J. Biomed. Mater. Res. 36, 99, 1997.

24. Duval, J.L., Letort, M., and Sigot-Luizard, M.F. Comparative assessment of cell/substratum static adhesion using
an in vitro organ culture method and computerized analysis system. Biomaterials 9, 155, 1988.

25. Cook, A.D., Hrkack, J.S., Gao, N.N., et al. Characterization and development of RGD-peptide-modified poly(lac-
tic acid-co-lysine) as an interactive, resorbable biomaterial. J. Biomed. Mater. Res. 35, 513, 1997.

26. Webb, K., Huang, S.-C., and Caldwell, K.D. Low protein binding surface coating for covalent immobilization of
extracellular matrix proteins on hydrophobic materials. Trans 24th Annu. Mtg. Soc. Biomater. 21, 14, 1998.

27. Beumer, G.J., Van Blitterswijk, C.A., Bakker, D., et al. A new biodegradable matrix as part of a cell seeded skin
substitute for the treatment of deep dermal skin defects: a physico-chemical characterization. Clin. Mater. 14, 21,
1993.

28. Boyce, S.T. Cultured skin substitutes: a review. Tissue Eng. 2, 255, 1996.
29. Bretscher, M.S. Endocytosis: relation to capping and cell locomotion. Science 224, 681, 1984.
30. Bretscher, M.S. Moving membrane up to the front of migrating cells. Cell 85, 465, 1996.
31. Kwiatkowska, K., and Sobota, A. Signaling pathways in phagocytosis. BioEssays 21, 422, 1999.
32. Coutant, K.D., Corvaia, N., and Ryder, N.S. Bradykinin induces actin reorganization and enhances cell motility in

HaCat keratinocytes. Biochem. Biophys. Res. Commun. 237, 257, 1997.
33. Ng, T., Shima, D., Squire, A., et al. PKCa regulates b1 integrin-dependent cell motility through association and

control of integrin traffic. EMBO J. 18, 3909, 1999.
34. Huttenlocker, A., Ginsberg, M.H., and Horwitz, A.F. Modulation of cell migration by integrin-mediated cytoskeletal

linkages and ligand-binding affinity. J. Cell. Biol. 134, 1551, 1996.
35. Miyamoto, S., Teramoto, H., Gutkind, J., et al. Integrins can collaborate with growth factors for phosphorylation

of receptor tyrosine kinases and MAP kinases activation: roles of integrin aggregation and occupance of receptors.
J. Cell Biol. 135, 1633, 1996.
36. Kawahara, E., Tokuda, R., and Nakanishi, I. Migratory phenotypes of HSC-3 squamous carcinoma cell line in-
duced by EGF and PMA: relevance to migration of loosening of adhesion and vinculin-associated focal contacts
with prominent filopodia. Cell Biol. Int. 23, 163, 1999.
37. Burgess, B.T., Myles, J.L., and Dickinson, R.B. Quantitative analysis of adhesion-mediated cell migration in three-
dimensional gels of RGD-grafted collagen. Ann. Biomed. Eng. 28, 110, 2000.
38. DiMilla, P.A., Barbee, K., and Lauffenburger, D.A. Mathematical model for the effects of adhesion and mechan-
ics on cell migration speed. Biophys. J. 60, 15, 1991.
39. Woodley, D.T., Bachmann, P.M., and O’Keefe, E.J. Laminin inhibits human keratinocyte migration. J. Cell. Phys-
iol. 136, 140, 1988.
40. Sarret, Y., Woodley, D.T., Grigsby, K., et al. Human keratinocyte locomotion: the effect of selected cytokines. J.
Invest. Dermatol. 98, 12, 1992.
41. Evans, E., Leung, A., and Zhelev, D. Synchrony of cell spreading and contraction force as phagocytes engulf large
pathogens. J. Cell Biol. 122, 1295, 1993.
42. Lowry, M.B., Duchemin, A.-M., Robinson, J.M. et al. Functional separation of psudopod extension and particle
internalization during Fcg receptor-mediated phagocytosis. J. Exp. Med. 187, 161, 1998.
43. Langer, R. Drug delivery and targeting. Nature 392, 5, 1998.
44. Soskic, V., Gorlach, M., Poznanovic, S., et al. Functional proteomics analysis of signal transduction pathways of
the platelet-derived growth factor beta receptor. Biochemistry 38, 1757, 1999.
45. Guetens, G., Van Cauwenberghe, K., DeBoeck, G. et al. Nanotechnology in bio/clinical analysis. J. Chromatogr.
B. Biomed. Sci. Appl. 739, 139, 2000.

Address reprint requests to:
Prabhas V. Moghe, Ph.D.

Department of Chemical and Biochemical Engineering
Rutgers University
98 Brett Rd.

Piscataway, NJ 08854

E-mail: [email protected]

261