| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Center for Engineering in Medicine and Surgical Services, Massachusetts General Hospital, Harvard Medical School and Shriners Burns Hospital, Boston, Massachusetts 02114, USA
1Correspondence: Shriners Hospital for Children, 51 Blossom St., Boston, MA 02114, USA. E-mail: jmorgan{at}sbi.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: tissue engineering • basal lamina • dermal analog • polyimide chip
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. These engineered tissue
analogs, often composed of cultured cells, biomaterials or composites
combining cells and biomaterials, have achieved some clinical success
as substitutes for skin and cartilage (2
3
4
5)
. Future
developments in the design of tissue substitutes will be aided by an
improved understanding of the role of 3-dimensional scaffolds in the
regulation of cell function, maintenance of tissue architecture, and
the regeneration of important tissue features. Significant effort has
been devoted to producing biocompatible scaffolds with defined pore
sizes that help ensure proper cell–cell contacts, cell–matrix
interactions, and preserve cellular function. For example, in collagen
sponges the average pore diameter (skin: 20 to 125 µm, nerve: 5 to 10
µm) and orientation (skin: random, nerve: axial) are critical for
optimal analog activity (6
, 7)
. The pore size (
500
µm) of cylindrical poly(L-lactic acid) (PLLA) devices also enhances
the rate of fibrovascular tissue growth into porous 3-dimensional
scaffolds (8)
. Likewise, 12 µm diameter polyglycolic
acid (PGA) fibers processed into 96–97% porous, nonwoven meshes
facilitate proliferation of chondrocytes, production of sulfated
glycosaminoglycans and type II collagen, as well as the formation of
neocartilage with specific, reproducible 3-dimensional shapes and
histological features that closely resemble cartilage (9
, 10)
. Although sponges, foams, and matrices with defined
porosities are useful for the fabrication of relatively large and
highly porous 3-dimensional analogs of the extracellular matrix, their
use is limited for the creation of a basal lamina, a thin membranous
layer of connective tissue found in many organs and tissues.
The basal lamina performs several important functions in a variety of
tissues. In the glomerulus of the kidney, the basal lamina separates
endothelial cells from podocytes and acts as a selectively permeable
barrier to plasma molecules (reviewed in ref 11
). In the
skin, the basal lamina functions as a barrier that separates cells
(epidermal keratinocytes) from underlying connective tissue (dermis).
Although it prevents fibroblasts in the dermis from contacting
epidermal keratinocytes, it does not prevent the movement of immune
cells in and out of the epidermis, nor does it prevent the innervation
of the epidermis. During wound healing as well as during normal
development, the basal lamina acts as a guide and template that helps
control cell migration and differentiation (12)
. As with
most tissues, the basal lamina in the skin is not a simple flat plane
of connective tissue, rather it conforms to a series of ridges and
invaginations known as rete ridges and papillary projections. In
addition to effects on the biomechanical properties of the skin, the
pattern and depths of these ridges are thought to have a role in the
proliferation and differentiation of epidermal keratinocytes. The
fabrication of a basal lamina with controlled dimensions would help
elucidate the influence of topography on cell function and should have
applications in tissue engineering of skin substitutes as well as other
basal lamina-containing tissues.
In this study, we used a microfabrication approach to produce an analog of a basal lamina with precisely engineered topographic features. After designing a test pattern with features comparable in length scale to the topographic features of native basal lamina structures, a master chip was created by laser machining the geometric specifications into a polyimide chip. Using negative replicates of the chip, we produced thin, biocompatible membranes with topographic features that reproduced the dimensions of the master chip. The utility of the microfabricated membrane was demonstrated by laminating it to the surface of a collagen sponge to form a composite analog of the dermis. When epidermal keratinocytes were cultured on the surface of this composite, the cells formed a continuous, translucent, and hydrophobic epidermis with gross topographic features that conformed to the surface of the microfabricated membrane.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Production of a basal lamina analog
Two types of materials were used with the negative replicates to
make membranes: gelatin or a white coprecipitate containing type I
collagen (5 mg/ml) and glycosaminoglycan (GAG, 0.18 mg/ml). A 1% (w/v)
gelatin solution was prepared by stirring 1.0 g of gelatin (Sigma
Chemical, St. Louis, Mo.) into 100 ml of a 0.05 M acetic acid solution
that was warmed to 65°C until the gelatin was completely dissolved.
The coprecipitate was prepared according to published protocols
(6)
. Briefly, 3.6 g of lyophilized bovine collagen
(Medicol F, Integra Medicus, West Chester, Pa.) was dispersed in 600 ml
of a 0.05 M acetic acid solution by blending at 20,000 rpm for 90 min
at 4°C in a refrigerated homogenizer. The coprecipitate was formed by
adding 120 ml of a 0.11% w/v solution of shark cartilage chondroitin
6-sulfate (Sigma Chemical) to the blending collagen dispersion, then
blending the collagen-GAG copolymer for an additional 90 min. The
collagen-GAG dispersion was degassed under vacuum to remove trapped
air, then stored at 4°C.
To create membranes, a small volume of either gelatin or collagen-GAG dispersion (220–330 µl/cm2) was poured onto the PDMS-negative replicate, where it conformed to the surface. The dispersion was air-dried overnight at room temperature in a laminar flow hood and the resulting dried collagen membrane was gently peeled from the negative replicate. Finally, dried membranes were covalently cross-linked by thermal dehydration at 105°C in a vacuum of 100 mTorr for 24 h.
Production of a dermal analog
Composite dermal analogs were prepared by laminating a
microfabricated membrane to the surface of a collagen sponge. The
collagen sponge was produced by methods similar to those previously
described by Yannas et al. (13)
and Boyce et al.
(14)
. Briefly, 10 ml of collagen-GAG dispersion was poured
into an aluminum pan with a surface area of 38.5 cm2
(Fisher Scientific, Springfield, N.J.), and a microfabricated membrane
was gently floated on the surface of the dispersion. The dispersion was
rapidly frozen at -80°C, placed on a shelf in a freeze dryer
initially set at -45°C, then lyophilized overnight (Virtis Genesis,
Virtis, Gardner, N.Y.) at a vacuum of 100 mTorr. After lyophilization,
the composites were covalently cross-linked by thermal dehydration at
105°C in a vacuum of 100 mTorr for 24 h, rehydrated in a 0.05 M
acetic acid solution for 24 h, cross-linked in a 0.25%
glutaraldehyde solution for 24 h, then washed exhaustively with
sterilized water, PBS, and keratinocyte seeding medium (described
below).
Keratinocyte culture
Normal human keratinocytes derived from neonatal foreskins were
cultured by the method of Rheinwald and Green (15)
.
Keratinocytes were cocultivated with 3T3-J2 mouse fibroblasts
(originally provided by H. Green, Harvard Medical School, Boston,
Mass.) that had been pretreated with 15 µg/ml mitomycin C (Boehringer
Mannheim Co., Indianapolis, Ind.). Culture medium was changed every
3–4 days with a 3:1 mixture of Dulbecco’s modified Eagle’s medium
(DMEM) (high glucose) (Life Technologies, Inc.-BRL, Gaithersburg, Md.)
and Ham’s F-12 medium (Life Technologies, Inc.-BRL) with 10% fetal
bovine serum (FBS; JRH Bioscience, Lenexa, Kans.). Supplements were
added as described previously (16)
. Cells were subcultured
by first removing the feeder layer cells with a brief EDTA wash, 5 mM
in phosphate-buffered saline (PBS), and then treating the keratinocytes
with trypsin-EDTA.
Production of skin equivalents
Keratinocytes were seeded onto the dermal analogs using methods
similar to those previously described (17)
with media
changes as described by Ponec et al. (18)
. Dermal analogs
were placed into 35 mm tissue culture dishes, microfabricated membrane
side up, and cells in keratinocyte seeding medium (described below)
were seeded onto the surface (5x105
cells/cm2). After ~2 h, the cell-seeded dermal
analogs were submerged in keratinocyte seeding medium for 24 h.
Keratinocyte seeding medium was a 3:1 mixture of DMEM (high glucose)
(Life Technologies, Inc.-BRL) and Ham’s F-12 medium (Life
Technologies, Inc.-BRL) supplemented with 1% FBS (JRH Bioscience),
10-10 M cholera toxin (Vibrio Cholerae, Type
Inaba 569 B; Calbiochem, La Jolla, Calif.), 0.2 µg/ml hydrocortisone
(Calbiochem), 5 µg/ml insulin (Novo Nordisk, Princeton, N.J.), 50
µg/ml ascorbic acid (Sigma Chemical), and 100 IU/ml and 100 µg/ml
penicillin-streptomycin (Boehringer Mannheim Co.). After 24 h,
keratinocyte seeding medium was removed and the skin equivalents were
submerged for an additional 48 h in a keratinocyte priming medium.
Keratinocyte priming medium was composed of keratinocyte seeding medium
supplemented with 24 µM bovine serum albumin, 1.0 mM L-serine, 10
µM L-carnitine, and a mixture of fatty acids including 25 µM oleic
acid, 15 µM linoleic acid, 7 µM arachidonic acid, and 25 µM
palmitic acid, all from Sigma Chemical (19)
. After 48 h in priming medium, skin equivalents were placed on stainless steel
screens, raised to the air-liquid interface, and cultured for 7 days
with an air-liquid interface medium composed of serum-free keratinocyte
priming medium supplemented with 1.0 ng/ml epidermal growth factor
(Collaborative Biomedical Products, Bedford, Mass.).
Histological and quantitative morphometric analyses
For histological analysis, skin equivalents were fixed in a 3%
glutaraldehyde/4% paraformaldehyde solution, dehydrated with
increasing concentrations of ethanol, infiltrated first at -80°C and
then at 4°C with glycolmethacrylate (JB-4, Polysciences, Inc.,
Warrington, Pa.), and finally embedded in glycolmethacrylate. Sections
of skin equivalents, 5 µm in thickness, were collected by cutting
samples in a plane perpendicular to the surface of the microfabricated
membrane. Sections were mounted on glass slides with Tissue-Tack
Adhesive (Polysciences, Inc.), stained with Gill’s hematoxylin and
ethanolic eosin solutions, then viewed with a Nikon Eclipse 800
microscope. The dimensions of the channels in a collagen membrane were
measured in triplicate with a Nikon Diaphot 300 microscope coupled to
MetaMorph Imaging Software (West Chester, Pa.). The thickness of the
stratified epidermal layer in each channel and between channels was
measured vertically from the lower surface of the basal layer to the
outer edge of the granular layer. Conversely, invaginations of the
epidermal layer in each channel were measured vertically the granular
layer at the bottom of each channel up to the plane that corresponded
to the granular layers of the adjacent interchannel regions.
For scanning electron microscopy, skin equivalents were fixed in a 3% glutaraldehyde/4% paraformaldehyde solution, postfixed with a 1% osmium tetroxide solution, en bloc stained with a 2% uranyl acetate solution, dehydrated with increasing concentrations of ethanol, then critical point dried with liquid carbon dioxide under pressure. Samples were sputter coated with a thin layer of gold-palladium and viewed with an Amray 1000 scanning electron microscope.
The dimensions of the channels in the laser machined master chip were analyzed with a Zygo New View 200 Scanning White Light Interference Microscope (Zygo Corp., Middlefield, Conn.) fitted with a 20x Mirau objective lens. The microscope produced 3-dimensional images of each channel and 2-dimensional surface profiles in planes perpendicular to the surface of the master chip. The shoulder-to-shoulder widths and maximum depths of each channel were measured with the caliper tools in the microscope software.
The dimensions of the channels on a PDMS-negative replicate were analyzed with a Nikon Diaphot 300 microscope coupled with MetaMorph Imaging Software. Sections of a negative replicate were collected by cutting samples in a plane perpendicular to the microfabricated surface with a razor blade. Sections were placed on glass slides, viewed with a low power objective lens, and digitally imaged with the microscope software. Shoulder-to-shoulder widths and maximum depths of each channel were measured in triplicate with the tools provided in the software package.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Our strategy was to create a single master chip that could be used as
a template to produce numerous negative replicates that could then be
used to produce membranes with topographic features corresponding to
the design in the master chip. The master chip was produced by laser
machining our specified test pattern into the surface of a single
polyimide chip (Fig. 1)
.
|
To test the fidelity with which laser machining satisfied the
specifications of our design, we measured the dimensions of the
channels in the master chip by Scanning White Light Interference (SWLI)
microscopy. SWLI microscopy provided noncontact quantitative 2- and
3-dimensional topographic data with submicron resolution for each of
the channels, although its ability to image vertical surfaces such as
the side walls of some channels was limited. SWLI images of the master
chip showed that the narrow channels (specified widths: 40 or 80 µm)
had smooth side walls and flat bottom surfaces close to the original
design. Conversely, the bottom surfaces of the wider channels
(specified widths: 120, 160, or 200 µm) tapered to points or had
irregular surfaces. In addition, the widest channels (specified width:
200 µm) had notched shoulders in the side walls (Fig. 2
).Since the shapes of the channels were not rectangular, we simplified
and standardized our quantitative analyses by measuring the maximum
shoulder-to-shoulder width and maximum depth of each channel (Fig. 2)
.
Although the widths in the master chip deviated from the original
design, they covered a range of widths from 114 to 310 µm, with five
distinct channel widths (Table 1
). For example, channels whose width was specified to be 40 µm had an
average maximum width of 117 ± 3 µm for all depths specified.
|
|
The maximum channel depths, with some exceptions, were closer to the
design specifications (Table 1)
and covered a range of depths from 42
to 290 µm. For example, channels whose depth was specified to be 40
µm had an average maximum depth of 62 ± 21 µm for all widths
specified. Some depths, such as those channels whose width was
specified to be 120 µm, were consistently deeper than specified.
PDMS faithfully replicates the polyimide chip
To make negative replicates of the master chip, we poured liquid
PDMS over the polyimide chip and polymerized the PDMS by heat curing
(Fig. 1)
. To determine whether PDMS had faithfully replicated the
features of the master chip, we sectioned the PDMS and measured the
maximum shoulder-to-shoulder width and maximum depth of each of the
protruding PDMS ridges (Fig. 2)
. When compared to the values from the
master chip, there was no significant difference over the entire range
of widths and depths (Table 1)
. Thus, PDMS had faithfully replicated
the polyimide chip and had even replicated many of the small (micron)
irregular features found in the bottom of some of the channels.
Thin membranes with complex topographies can be laminated to a
porous collagen sponge
To determine the dimensions over which we could microfabricate
topographic features in a thin, biocompatible membrane, we poured a
small volume of gelatin or collagen-GAG coprecipitate onto the negative
replicates. The solutions were air-dried; membranes with features
similar to the master chip could be gently peeled off the negative
replicates (Fig. 1)
.
To form a composite dermal analog, these membranes were floated on the
surface of a small volume of collagen-GAG coprecipitate, which was
subsequently frozen and lyophilized to form a collagen-GAG sponge
(Fig. 3
). Scanning electron micrographs of a cross section of these composites
(Fig. 3)
showed that the surface of a highly porous collagen/GAG sponge
had been laminated with a thin membrane containing a series of
channels.
|
To determine whether the membranes had reproduced the surface of the
negative replicate, we rehydrated a composite and analyzed the
dimensions of the channels in the membrane. After rehydration, we
observed that the thin membrane became supple and flexible. The dermal
equivalent was subsequently fixed, embedded in glycolmethacrylate, and
sections were cut perpendicular to the surface. Depending on the volume
of collagen/GAG or gelatin used, the microfabricated membranes ranged
in thickness from 21.7 ± 4.1 to 30.5 ± 7.1 µm, with 25
channels that were supported by a porous collagen-GAG sponge. Distinct
widths and depths were produced in the membrane (Fig. 3)
as well as
many of the small irregular surface features that were observed in the
negative replicate (Figs. 2
and 3)
. In contrast to the polyimide chip
and the PDMS, the flexibility of these membranes, particularly the
deepest channels, precluded the accurate measurements of the width and
depth of the channels. Nevertheless, we estimated the fidelity of these
membranes by measuring the perimeter of each channel and comparing it
to the PDMS negative replicate. There was excellent correlation between
the perimeters of the membrane and its corresponding PDMS replicate
(Fig. 4
).
|
An epidermis with complex topography can be formed on the surface
of a microfabricated membrane
To investigate the response of cells to the topographic features
of these membranes, cultured keratinocytes were seeded on the surface
of the composite dermal analogs. The resulting skin equivalents were
cultured at the air-liquid interface to induce differentiation and
cornification of the epidermal layer. After 10 days, a continuous,
translucent, and hydrophobic epidermis that conformed to the
microfabricated surface of the skin equivalent was evident by gross
inspection. Histological analyses showed that the surface of the skin
substitute contained a series of channels of varied depths and widths,
as well as spaces between channels, covered with a stratified epidermis
(Fig. 5
).
|
Stratification of the deeper channels was enhanced compared to the
shallower channels or the flat interchannel regions. To quantitate this
response, we measured the thickness of the epidermal layer in each
channel. Epidermal thickness of the flat interchannel regions was 37.1
µm and thickness increased as the depth of each channel was increased
(Fig. 6
). The best fit for this relationship was for those channels whose
widths were between 250 and 350 µm.
|
It was also apparent that the channels induced the formation of
invaginations of the epidermal surface. To determine the influence of
channel geometry on these gross skin surface features, we also measured
the distance of the epidermal invagination for each channel. Infolds of
the epidermis occurred when the depth of the channels were greater than
~25 µm, and these invaginations increased in distance as channel
depth increased (Fig. 6
). The best fit was also for those
channels whose widths were in the range of 250 to 350 µm.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Native basal laminae exhibit a range of thickness (50–90 nm in skin to
240–340 nm in kidney glomerulus) and contain a variety of collagenous
and noncollagenous extracellular matrix molecules, as well as sulfated
proteoglycans. Although the exact composition varies between tissues,
all basement membrane structures are believed to contain type IV
collagen, laminin, fibronectin, and heparin sulfate proteoglycan
(reviewed in ref 20
). The function of basal laminae can
also vary, but generally laminae act as adhesive membranes that
separate cell types or anchor epithelial cell layers to mesenchymal
tissue. In addition to maintaining complex spatial organization of
tissues (skin, intestinal mucosa, mammary ducts, and salivary glands),
they can also act a selective barrier to cells and macromolecules.
Our study demonstrates for the first time that microfabrication approaches can be used to produce analogs of the basal lamina and that these first analogs exhibit some, but not all, properties of native basal laminae. Principally, these analogs served as membranes that supported the attachment and growth of keratinocytes in complex topographies and prevented keratinocytes from passing through, while allowing for the diffusion of nutrients. Compared to a native basal lamina (50–340 nm), these analogs were relatively thick, in the range of 21 µm; however, future efforts to optimize the production of thinner membranes should be able to reduce this thickness considerably.
Another difference from a native basal lamina is composition. Although the analogs in this study were produced with type I collagen or gelatin, our preliminary data shows that similar membranes can be produced with material containing type IV collagen, laminin, fibronectin, and proteoglycans. Thus, microfabricated analogs with different macromolecular compositions should be useful for investigating the role of various macromolecules in basal lamina structure/function and how these structure/functions are influenced by topography.
The microfabrication technology we chose for this study was based on laser machining a polyimide chip. This approach, although extremely useful for generating a single test pattern with a wide range of widths and depths, has some drawbacks. Our measurements of the chip and the negative replicates show that the fidelity with which the laser was able to machine the specified design was not as accurate as might be desired for some applications. Other microfabrication technologies, such as photolithography, can be used to generate patterns that more accurately reflect the intended design. Moreover, this technology can be used to produce topographies far more complex than the parallel channels in our master chip. For example, photolithography could easily be used to create a pattern of hexagonally packed arrays of pegs that mimic the dermal papillae in native skin.
Stratification of the epidermis was significantly enhanced in the deep
channels, as is the case for rete ridges of normal skin. This finding
suggests that the topographical microenvironment influences
keratinocyte differentiation/proliferation by affecting the spatial
arrangement of cell–matrix contacts and/or cell–cell contacts. Note
that in human palmar epidermis, the highest percentage (80%) of the
proliferating basal and suprabasal cells is in the deeper rete ridges
(21)
. It has also been shown that expression of
2ß1 integrin, a marker
of epidermal stem cells, also varies with topography. High expression
is found in patches of basal cells located on the tips of the dermal
papillae (foreskin, scalp) or at the bottom of the deep rete ridges
(palm) (22
, 23)
. A microfabricated membrane may be a
useful tool to examine the relationship between the dimensions of
topographic features and the proliferation and differentiation of basal
keratinocytes in the epidermis.
Microfabricated membranes may be useful for investigating the critical
topographical features of skin that contribute to its mechanical
stability. In areas of the skin exposed to excessive friction (plantar
and palmar surfaces), the dermal papillae and epidermal ridges are
longer and more numerous, suggesting that the enhancement of the
interface between the epidermis and dermis helps provide additional
mechanical stability (24)
. By testing the properties of
skin equivalents that have rete ridges and dermal papillae of various
geometries, it should be possible to establish a direct and
quantitative link between topography and mechanical properties.
Moreover, these studies can be extended to the molecular level by
examining the distribution and numbers of known adhesive proteins such
as integrins and hemidesmosomes (22
, 25)
.
Previous studies have shown that tissue-engineered skin equivalents
also have clinical application for the treatment of patients suffering
from severe burns and chronic ulcers (reviewed in ref 26
, 27
). Boyce et al. produced a composite skin substitute by
laminating the surface of a flat collagen sponge originally developed
by Yannas et al. (13
, 28)
. Fibroblasts were seeded into
the porous collagen sponge and a flat epidermal layer was cultured on
the laminated surface (14
, 29)
. In clinical studies, these
composite skin substitutes have had some success for the treatment of
full-thickness burns (2)
. Another skin equivalent has been
produced by culturing a flat epidermal layer on the surface of a
fibroblast populated collagen gel (30)
. These skin
equivalents have recently been approved for the treatment of venous leg
ulcers (4)
. Recently, novel skin equivalents with
capillary-like networks have also been produced (31)
.
Microfabricated membranes may be useful for improving the performance of these skin substitutes. Each of these skin equivalents has a flat interface between the epidermis and the dermal component. A microfabricated membrane that created a complex interdigitating interface between these two layers might be expected to improve the resistance of these skin substitutes to failure due to shear forces. Such a complex interface might also facilitate improved mass transport of nutrients and growth factors to the epidermis and aid in graft take. Moreover, since microfabrication can be used to create complex patterns, it should be possible to mimic the fine lines and pore structures of native skin to create a more ‘natural’ and perhaps cosmetically acceptable skin substitute.
Since basal lamina are an integral part of many other tissues besides
skin, microfabricated membranes may have other applications in tissue
engineering and developmental biology. For example, it may be possible
to enhance mass transport in tissue-engineered small intestine by using
a microfabricated membrane that mimics the topography of the intestinal
mucosa. During development, it is well known that basal laminae perform
critical roles in embryologic tissue compartmentalization and organized
cellular migration leading to morphogenesis and organogenesis (reviewed
in ref 20
, 32
). With the recent isolation of embryonic
stem cell lines derived from human blastocysts (33)
,
microfabricated analogs of the basal lamina may be useful for exploring
the roles of cell and matrix interactions on tissue morphogenesis and
mammalian development.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J.-M. Lim, S. Byun, S. Chung, T. H. Park, J.-M. Seo, C.-K. Joo, H. Chung, and D.-i. Cho Retinal Pigment Epithelial Cell Behavior is Modulated by Alterations in Focal Cell-Substrate Contacts Invest. Ophthalmol. Vis. Sci., November 1, 2004; 45(11): 4210 - 4216. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. N. Yousaf, B. T. Houseman, and M. Mrksich Using electroactive substrates to pattern the attachment of two different cell populations PNAS, May 22, 2001; 98(11): 5992 - 5996. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
