Identification of human fibroblast cell lines as a feeder layer for human corneal epithelial regeneration.

Abstract

There is a great interest in using epithelium generated in vitro for tissue bioengineering. Mouse 3T3 fibroblasts have been used as a feeder layer to cultivate human epithelia including corneal epithelial cells for more than 3 decades. To avoid the use of xeno-components, we evaluated human fibroblasts as an alternative feeder supporting human corneal epithelial regeneration. Five human fibroblast cell lines were used for evaluation with mouse 3T3 fibroblasts as a control. Human epithelial cells isolated from fresh corneal limbal tissue were seeded on these feeders. Colony forming efficiency (CFE) and cell growth capacity were evaluated on days 5-14. The phenotype of the regenerated epithelia was evaluated by morphology and immunostaining with epithelial markers. cDNA microarray was used to analyze the gene expression profile of the supportive human fibroblasts. Among 5 strains of human fibroblasts evaluated, two newborn foreskin fibroblast cell lines, Hs68 and CCD1112Sk, were identified to strongly support human corneal epithelial growth. Tested for 10 passages, these fibroblasts continually showed a comparative efficiency to the 3T3 feeder layer for CFE and growth capacity of human corneal epithelial cells. Limbal epithelial cells seeded at 1 × 10(4) in a 35-mm dish (9.6 cm(2)) grew to confluence (about 1.87-2.41 × 10(6) cells) in 12-14 days, representing 187-241 fold expansion with over 7-8 doublings on these human feeders. The regenerated epithelia expressed K3, K12, connexin 43, p63, EGFR and integrin β1, resembling the phenotype of human corneal epithelium. DNA microarray revealed 3 up-regulated and 10 down-regulated genes, which may be involved in the functions of human fibroblast feeders. These findings demonstrate that commercial human fibroblast cell lines support human corneal epithelial regeneration, and have potential use in tissue bioengineering for corneal reconstruction.

Full text

Page 1

Identification of Human Fibroblast Cell Lines as a Feeder
Layer for Human Corneal Epithelial Regeneration
Rong Lu1,2, Fang Bian2,3, Jing Lin2,4, Zhitao Su2, Yangluowa Qu2, Stephen C. Pflugfelder2, De-Quan Li2*
1Zhongshan Ophthalmic Center, State Key Laboratory of Ophthalmology, Sun Yat-Sen University, Guangzhou, China, 2Ocular Surface Center, Cullen Eye Institute,
Department of Ophthalmology, Baylor College of Medicine, Houston, Texas, United States of America, 3Department of Ophthalmology, Union Hospital of Tongji Medical
College, Huazhong Science and Technology University, Wuhan, China, 4Department of Ophthalmology, The Affiliated Hospital of Qingdao University Medical College,
Qingdao, China
Abstract
There is a great interest in using epithelium generated in vitro for tissue bioengineering. Mouse 3T3 fibroblasts have been
used as a feeder layer to cultivate human epithelia including corneal epithelial cells for more than 3 decades. To avoid the
use of xeno-components, we evaluated human fibroblasts as an alternative feeder supporting human corneal epithelial
regeneration. Five human fibroblast cell lines were used for evaluation with mouse 3T3 fibroblasts as a control. Human
epithelial cells isolated from fresh corneal limbal tissue were seeded on these feeders. Colony forming efficiency (CFE) and
cell growth capacity were evaluated on days 5–14. The phenotype of the regenerated epithelia was evaluated by
morphology and immunostaining with epithelial markers. cDNA microarray was used to analyze the gene expression profile
of the supportive human fibroblasts. Among 5 strains of human fibroblasts evaluated, two newborn foreskin fibroblast cell
lines, Hs68 and CCD1112Sk, were identified to strongly support human corneal epithelial growth. Tested for 10 passages,
these fibroblasts continually showed a comparative efficiency to the 3T3 feeder layer for CFE and growth capacity of human
corneal epithelial cells. Limbal epithelial cells seeded at 16104in a 35-mm dish (9.6 cm2) grew to confluence (about 1.87–
2.416106cells) in 12–14 days, representing 187–241 fold expansion with over 7–8 doublings on these human feeders. The
regenerated epithelia expressed K3, K12, connexin 43, p63, EGFR and integrin b1, resembling the phenotype of human
corneal epithelium. DNA microarray revealed 3 up-regulated and 10 down-regulated genes, which may be involved in the
functions of human fibroblast feeders. These findings demonstrate that commercial human fibroblast cell lines support
human corneal epithelial regeneration, and have potential use in tissue bioengineering for corneal reconstruction.
Citation: Lu R, Bian F, Lin J, Su Z, Qu Y, et al. (2012) Identification of Human Fibroblast Cell Lines as a Feeder Layer for Human Corneal Epithelial
Regeneration. PLoS ONE 7(6): e38825. doi:10.1371/journal.pone.0038825
Editor: Irina Kerkis, Instituto Butantan, Brazil
Received February 15, 2012; Accepted May 11, 2012; Published June 18, 2012
Copyright: ? 2012 Lu et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was supported by Department of Defense CDMRP PRMRP FY06 PR064719 (DQL), Lions Foundation for Sight, Fundamental Research Funds
of State Key Laboratory of Ophthalmology (China), Research to Prevent Blindness, Oshman Foundation, and William Stamps Farish Fund. The funders had no role
in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: dequanl@bcm.tmc.edu
Introduction
Ocular surface diseases with corneal epithelial stem cell
deficiency are sight threatening and often cause blindness [1].
There is great interest in using corneal epithelium generated in
vitro to treat these conditions and restore vision [2–7]. This
requires development of culture technique that mimics the in vivo
environment. Mesenchymal-epithelial interactions play an essen-
tial role in organogenesis and tissue regeneration at both
embryonic [8,9] and adult stages [10–12], and have been applied
to in vitro cell culture systems for growing epithelial cells that were
difficult to be cultivated.
The single-cell clonal growth was first achieved for keratinocytes
by co-culturing with mouse 3T3 fibroblasts as a feeder layer
[13,14]. This 3T3 fibroblast co-culture system has been proven to
be the most powerful system for human epithelial cultivation
[7,15–17], including human ocular surface epithelia [18–21], over
the past 3 decades. Based on our observation, the proliferative rate
and regenerative capacity of human corneal epithelial cells on the
3T3 fibroblast feeder layer were 10–50 times higher than that on
explant cultures [22]. As few as 200–1000 limbal epithelial cells,
which can be obtained from as small as 0.5 mm2of limbal biopsy,
could regenerate a whole human corneal epithelium in 2 weeks
with this co-culture system. Human corneal epithelium regener-
ated on 3T3 fibroblast feeder layer preserved more stem/
progenitor cells than explant cultures [20,22,23].
However, human epithelia generated on mouse 3T3 fibro-
blasts may bear a risk of transmitting pathogens and as a
consequence their use is not permitted by US FDA for
cultivating corneal epithelia for human transplantation. To
develop this powerful co-culture system for clinical use without
xeno-components, mouse 3T3 fibroblasts must be replaced by
human fibroblasts. Encouraged by reports [24–27] that human
fibroblasts could replace mouse fibroblasts as feeder layers to
support ex vivo expansion of human embryonic stem cells, we
hypothesize that human fibroblasts can be used as a feeder layer
to promote corneal epithelial regeneration for human trans-
plantation. The present study was attempted to identify stable
human fibroblast cell lines that could serve as an alternative
feeder layer to mouse 3T3 cells.
PLoS ONE | www.plosone.org1 June 2012 | Volume 7 | Issue 6 | e38825

Page 2

Results
Identification of Human Fibroblast Cell Lines as a Feeder
Layer Supporting Ex-vivo Expansion of Human Corneal
Epithelial Cells
To identify whether human fibroblasts can serve as a feed layer
to support ex vivo expansion of human corneal epithelial cells, five
commercially available human fibroblast cell lines were tested with
mouse 3T3 fibroblasts as controls. To serve as a feeder for
epithelial regeneration, mouse 3T3 fibroblasts were lethally treated
with 5 mg/ml mitomycin C for 2 hours, then trypsinized and
plated as single cells at 26104/cm2in a 35-mm dish or 6-well
plates. We have observed that higher doses and longer period of
mitomycin C treatment are required for human fibroblasts than
mouse 3T3 cells. The treatment with 5 mg/ml mitomycin C for 16
hours was tested to be an optimal condition for these human
fibroblasts serving as a feeder layer.
Human corneal epithelial cells were isolated from donor corneal
limbus, and seeded at 16103cells/cm2on these human fibroblast
feeder layers with mouse 3T3 cells as controls [22]. We have
observed that human corneal epithelial cells formed fewer colonies
and grew slowly on the feeders prepared from human fetus skin
fibroblasts Detroit 551 and two newborn foreskin fibroblasts (Hs27
and BJ), when compared with that on a 3T3 feeder layer.
Interestingly, two newborn foreskin fibroblasts, Hs68 and
CCD1112Sk, displayed a comparative capacity to mouse 3T3
fibroblast feeder layer in supporting human corneal epithelial
growth. Tested for 10 passages at 16–25 and 6–15 respectively,
Hs68 and CCD1112Sk fibroblasts showed strong support for
colony forming efficiency (CFE) and growth capacity of human
corneal epithelial cells. As shown in Fig 1A, the CFE was
1.4360.13% at day 4 and 4.9160.47% at day 8 on Hs68
fibroblasts, and 1.5260.17% at day 4 and 5.1660.47% at day 8
on CCD1112Sk fibroblasts, similar to the CFE on mouse 3T3
feeder layer, 1.5660.07% at day 4 and 5.4160.57% at day 8,
respectively. The corneal epithelial cells grew rapidly from clonal
colonies to confluencein
1.8722.416106cells covering 9.6 cm2area) in 12–14 days on
Hs68, CCD1112Sk or 3T3 feeder layers (Fig 1B), representing
187–241 fold expansion with over 7–8 doublings on human
feeders, similar to 3T3 feeder layer.
35-mm dishes(about
The Phenotype of Human Corneal Epithelium
Regenerated on Human Fibroblast Feeder Layer
The phenotype of human corneal epithelium regenerated on
Hs68 and CCD1112Sk feeder layers were further characterized by
evaluating their morphology and epithelial cell markers. The
corneal epithelium generated on these human feeders contained
small compact cells and expressed epithelial markers (Fig 2).
Immunofluorescent staining revealed that corneal epithelial
specific marker cytokeratin (K) 3 was immunolocated in cytoplasm
of about 45% cells with relatively large size, which also expressed a
differentiation marker gap junction protein connexin 43 (Cx43) at
cell membrane. Interestingly, three epithelial progenitor cell
markers, nuclear transcription factor p63, integrin b1 and
epidermal growth factor receptor (EGFR), immunolocated in the
nucleus, cytoplasm or membrane, respectively, were strongly
expressed by 22–30% cells with relatively small size. The
percentages of positive cells for these markers, characteristics of
corneal epithelial progenitor cells, in human corneal epithelia
regenerated on human fibroblasts Hs68 and CCD1112Sk were
comparable to those on mouse 3T3 feeder layer, as shown in
Table 1. To confirm these markers at transcription levels, we
further compared their mRNA expression by these corneal
epithelial cells generated on human fibroblasts (Hs68 and
CCD1112Sk) and 3T3 cells. As shown in Fig 3 evaluated by
RT-PCR, the cells grown on these two human feeders expressed
typical corneal epithelial markers, including progenitor cell
markers, DNp63 and integrin b1, and differentiation markers,
Cx43, K3 and K12, which was similar to the expression profile of
the cells grown on 3T3 mouse feeder. These results indicated that
the epithelium regenerated on human feeder layer resembled the
phenotype of human corneal epithelium and contained undiffer-
entiated progenitor cells [22], which are important for tissue
engineering in regenerative medicine.
Identification of Potential Stroma-derived Factors from
the Supportive Human Fibroblasts that Promoted
Corneal Epithelial Regeneration
Mouse 3T3 fibroblast feeder layer has been widely used for
long-term growth and serial propagation of many types of
epithelial cells since 1975 [13], however, the exact identity of this
fibroblast-derived supporting activity remains to be elucidated.
While human foreskin fibroblast cell lines have been identified to
be capable of supporting the corneal epithelial regeneration, we
further attempted to identify stroma-derived factors from these
supportive fibroblasts. We performed cDNA microarray using
Agilent GeneChip to characterize the expression profile and
molecular changes of supportive human fibroblasts (Hs68) before
and after mitomycin C treatment.
Among 22575 transcripts on a chip, 3400 transcripts, account-
ing for 15.1% of expressed genes, were regulated after treatment
with mitomycin C (5 mg/ml for 16 hours). Among 3,400 changed
genes, 202 genes were up-regulated more than 2 fold with p value
,0.05; while 331 genes were down-regulated more than 50% with
p value ,0.05. The best 3 up-regulated genes were identified:
CDKN1A (cyclin-dependent kinase inhibitor 1A, p21, Cip1),
GDF15 (growth differentiation factor 15), and FDXR (ferredoxin
reductase). The best 10 down-regulated genes were also identified:
HIST1H4C (histone cluster 1, H4c), TNFRSF11B (tumor necrosis
factor receptor superfamily, member 11b), ITGB1 (integrin, beta
1), COL6A3 (collagen type VI, alpha 3), TWIST2 (twist homolog
2), SMARCD3 (SWI/SNF related, matrix associated, actin
dependent regulator of chromatin, subfamily d, member 3), FN1
(fibronectin 1), THBS2 (thrombospondin 2), PTX3 (pentraxin 3,
long), and CDK2 (cyclin-dependent kinase 2). Functional pathway
analysis revealed that there were 7 genes down-regulated (.2-fold)
in 20 genes of the cell proliferation regulation group (data not
shown). These genes and related signaling pathways might be
potential targets, and further investigations are important for
validation and functional identification of stroma-derived genes
that promote proliferation and regeneration of human corneal
epithelial cells.
Discussion
Mouse 3T3 fibroblasts have been used as a feeder layer for more
than three decades [13] and proven to be the most powerful
system among all epithelial culture methods. To avoid xeno-
components per clinical use, the mouse 3T3 fibroblasts need to be
replaced by human feeders. In the present study, we have
evaluated five commercial human fibroblast cell lines, and
successfully identified two cell lines of human newborn foreskin
fibroblasts, Hs68 and CCD1112Sk, which showed comparative
efficiency to a 3T3 feeder layer in supporting human corneal
epithelial cell growth and regeneration.
Based on evaluation of epithelial clonal colony forming
efficiency and epithelial regeneration capacity, human newborn
Human Feeder for Corneal Epithelial Regeneration
PLoS ONE | www.plosone.org2June 2012 | Volume 7 | Issue 6 | e38825

Page 3

Human Feeder for Corneal Epithelial Regeneration
PLoS ONE | www.plosone.org3 June 2012 | Volume 7 | Issue 6 | e38825

Page 4

foreskin fibroblasts, Hs68 and CCD112Sk, showed strong
supportive capacity, similar to the mouse 3T3 feeder layer
(Fig 1). The seeded epithelial cells (16104cells) grew rapidly from
colonies to confluence reaching about 1.8722.416106cells in 12–
14 days, which represented 187–241 fold expansion with over 7–8
doublings. Cell morphology and immunofluorescent staining
further revealed that the regenerated epithelium on these human
feeders resembled the phenotype of human corneal epithelium
containing undifferentiated progenitor cells. As shown in Fig 2 and
3, the regenerated epithelium expressed three progenitor cell
markers, p63, integrin b1 and EGFR, in addition to corneal
epithelial specific and differentiated markers K3, K12 and Cx43.
The cells expressing the progenitor cell markers reached 22–30%,
consistent to our previous report [22].
Signals transmitted from mesenchyme to epithelia or vice versa
constitute the basis of reciprocal mesenchymal-epithelial interac-
tions, which are thought to play an essential role in organogenesis
and tissue regeneration. In adult tissues, mesenchymal-epithelial
cell communications maintain tissue homeostasis and play a key
role in epithelial repair [10,28,29]. Mesenchymal-epithelial
interactions have been successfully applied to in vitro cell culture
systems for growing epithelial cells that were difficult to cultivate
and serially propagate without the presence of fibroblasts or their
products. In the presence of a fibroblast feeder layer, epithelial
cells expand rapidly and maintain their progenitor phenotype,
while in the absence of fibroblasts, epithelial cells undergo terminal
differentiation. After identification of human fibroblast cell lines,
Hs68 and CCD112Sk, supporting corneal epithelial regeneration,
we performed cDNA microarray to search stroma-derived factors
from these supportive fibroblasts. Interestingly, three up-regulated
genes and 10 down-regulated genes have been identified to be
involved in the functions of fibroblast feeders. Functional pathway
analysis using GenMAPP 2.0 and MAPPFinder further revealed
that 7 genes were down-regulated in 20 genes of cell proliferation
regulation group. These genes and related pathways are potential
targets that are useful for further validation and functional
identification of the stroma-derived genes that promote prolifer-
ation of limbal stem cells.
In conclusion, we have identified two commercially available
cell lines of human newborn foreskin fibroblasts, Hs68 and
CCD112Sk, which support human corneal epithelial regeneration.
The cultivated corneal epithelium on human fibroblast feeder
layers resembles the phenotypes of human corneal epithelium
containing progenitor cells. The development of human fibroblast
feeders will represent a major advance in the field of corneal
epithelial regeneration. Human feeders have potential in use for
tissue bioengineering, not only for corneal epithelium, but also
possibly for other epithelial tissues. In particular, this new
technique would greatly promote corneal tissue bioengineering
in regenerative medicine to improve the treatment for patients
with blinding corneal disease.
Materials and Methods
Preparation of Fibroblasts as a Feeder Layer
Mouse 3T3 embryo fibroblasts (CCL-92) and five human
fibroblast cell lines were purchased from ATCC (Manassas, VA),
including fetus skin fibroblasts (Detroit 551, CCL-110), and 4
newborn foreskin fibroblasts, Hs27 (CRL-1634), Hs68 (CRL-
1635), CCD1112Sk (CRL-2429) and BJ (CRL-2522). All these
fibroblasts were cultured in DMEM containing 10% FBS, and
subcultured with 1:4 split for several passages when they reached
80–90% confluence The confluent fibroblasts were lethally treated
with mitomycin C at 5 mg/ml for different time periods (2 hours
for 3T3 cells, and 16 hours for human fibroblasts), and then
trypsinized and plated at 26104cells/cm2in 35-mm dishes or 6-
well culture plates as feeder layers used for corneal epithelial cell
culture.
Isolation of Limbal Epithelial Single Cells for Ex vivo
Expansion in Co-cultures
Cadaveric human corneoscleral tissues were obtained from the
Lions Eye Bank of Texas (LEBT, Houston, TX). Single epithelial
cells were isolated from corneal limbal tissues as previously
described [22,30,31]. In brief, the limbal rims were incubated with
5 mg/ml of dispase II (Roche, Indianapolis, IN) at 37uC for 2
Figure 1. Human limbal epithelial cells (HLE) grown on feeder layers of Hs68, CCD1112Sk or 3T3 fibroblasts after seeded at
16103cells/cm2in 6-well plate. A. Colony forming efficiency (CFE) of HLE on three fibroblasts in days 4–8. B. Phase images showing a comparative
efficiency of Hs68, CCD1112Sk and 3T3 fibroblasts as a feeder layer in supporting clonogenicity and growth capacity of HLE on days 1–14.
doi:10.1371/journal.pone.0038825.g001
Figure 2. Representative immuno-fluorescent staining profiles for corneal epithelial phenotype. Corneal epithelial markers, including
the differentiation markers, keratin 3 (K3), and connexin 43 (Cx43), as well as progenitor markers, EGFR, nuclear p63 and integrin b1, were expressed
by HLE regenerated on human feeder of Hs68 fibroblasts; Hoechst 33342 was used for nuclear counterstaining.
doi:10.1371/journal.pone.0038825.g002
Human Feeder for Corneal Epithelial Regeneration
PLoS ONE | www.plosone.org4 June 2012 | Volume 7 | Issue 6 | e38825

Page 5

hours. The detached limbal epithelial sheets were collected and
incubated with 0.05% trypsin/0.03% EDTA at 37uC for 5–10
minutes to isolate single cells. Single cells of human limbal
epithelium were then seeded at 16103cells/cm2on different
fibroblast feeders in a supplemented hormonal epidermal medium
(SHEM) containing 5% FBS to evaluate the colony forming
efficiency, growth capacity and phenotypes of cultivated corneal
epithelium [22].
Assays for Clonogenicity and Proliferative Potential
Limbal epithelial single cells were seeded, in triplicate, at
16103cells/cm2on a feeder layer (human or mouse fibroblast) in
6-well culture plates [13,22,30]. The colony forming efficiency is
calculated as a percentage of the number of colonies at days 4–8
generated by the number of epithelial cells plated in a well. The
clonal growth capacity was evaluated on days 12–14 when the
cultures were stained with 1% crystal violet.
Immunofluorescent and Immunohistochemical Staining
Cell cultures were fixed in cold methanol or 2–4% paraformal-
dehyde and permeabilized with 0.2% Triton X-100 in PBS for 10
minutes each. Immunofluorescent staining were performed with
our previous methods [22,28,32,33] using primary antibodies
against p63 [34], integrin b1, EGFR, K14, K3, connexin 43
(Calbiochem, Labvision, ICN Pharmaceuticals, Santa Cruz
Biotechnology, Chemicon International, or Invitrogen, respective-
ly), with Alexa Fluor 488 conjugated secondary antibodies
(Invitrogen), and counter-staining with Hoechst 33342 (Sigma).
Specimens without primary antibody or with IgG isotype were
used as negative controls. The staining were evaluated and
photographed with a Nikon Eclipse microscope.
Total RNA Extraction and Reverse Transcription
Polymerase Chain Reaction (RT-PCR)
Total RNA was isolated from cells using a Qiagen RNeasyH
Mini kit according to manufacturer’s protocol. The RNA samples
were stored at –80uC before use. The RNA quality and
concentration were measured by Agilent 2100 Bioanalyzer and
NanoDropH ND-1000 Spectrophotometer. The mRNA expres-
sion of different molecular markers was analyzed by RT-PCR as
previously described [33,35]. The specific primer pairs for each
gene were designed from published human gene sequences
(Table 2).
Agilent Human cDNA Microarray Analysis
Agilent Human cDNA Microarray (Agilent Technologies, Santa
Clara CA) assays were performed by the Microarray Core Facility
at Baylor College of Medicine. Total RNA (15 mg) of human
fibroblast cell line Hs68 (CRL-1635) without or with mitomycin C
treatment was converted into cDNA by reverse transcription and
labeled with green Cye3- (for control) or red cye5- (for treated)
dUTP. An array chip was hybridized with labeled cDNAs
overnight, scanned with an Axon 4200, and analyzed using the
program Acuity from Axon Technologies. Data is viewed as a
normalized ratio of Cye5/Cye3. The ratio was indicatives of no
changed (Cye5/Cye3=1), decreased (Cye5/Cye3,1) or in-
creased (Cye5/Cye3.1) levels of gene expression relative to the
control sample. Two- or more fold (Log Ratio of Cye5/Cye3.1
Table 1. Properties of human corneal epithelia co-cultured
on human (Hs58 and CCD1112Sk) and mouse (3T3)
fibroblasts.
Fibroblast Feeder
Human Corneal Epithelium*
Hs68CCD1112Sk 3T3
Proliferation
CFE % on Day 41.4360.131.5260.171.5660.07
CFE % on Day 84.9160.47 5.1660.47 5.4160.57
Cells at confluence (x106) 2.360.18 2.160.362.460.21
Confluence at days12–1412–1412–14
Doubling time (hours)42.843.942.4
Epithelial Markers
K3+ Cells%
K14+ Cells%
Cx43+ Cells%
EGFR+ Cells%
Nuclear p63+ Cells%
Integrin b1+ Cells%
44.269.345.3611.2 42.3610.4
50.7613.352.27615.1 48.7612.3
55.5614.857.4613.559.3615.2
27.164.426.366.724.365.9
24.665.822.365.423.964.7
28.165.729.966.826.365.4
*Initiated from 16104limbal epithelial cells/35 mm dish (9.6 cm2area).
doi:10.1371/journal.pone.0038825.t001
Figure 3. Gene expression profiles for corneal epithelial
markers by HLE grown on Hs68, CCD1112Sk (CCD) and 3T3
fibroblasts. Representative RT-PCR images show the mRNA expression
of DNp63 (440 bp), integrin b1 (104 bp), Cx43 (154 bp), K3 (145 bp) and
K12 (150 bp) with GAPDH (498 bp) as internal control. A 100 bp DNA
ladder is shown at the left lane.
doi:10.1371/journal.pone.0038825.g003
Human Feeder for Corneal Epithelial Regeneration
PLoS ONE | www.plosone.org5 June 2012 | Volume 7 | Issue 6 | e38825