Clusters of Corneal Epithelial Cells Reside Ectopically in
Human Conjunctival Epithelium
Satoshi Kawasaki, Hidetoshi Tanioka, Kenta Yamasaki, Norihiko Yokoi, Aoi Komuro,
and Shigeru Kinoshita
PURPOSE. The ocular surface is covered by two biologically
distinct epithelia: corneal and conjunctival. The expression of
keratin12 (K12) is currently considered a hallmark of cornea-
type differentiation. In the current study, the biological fea-
tures of K12-positive cells in human bulbar conjunctival epi-
thelium were examined.
METHODS. Human conjunctival tissues were subjected to inves-
tigate the K12-positive cells in conjunctiva by immunostaining,
in situ hybridization, Western blot analysis, reverse transcrip-
tase–polymerase chain reaction (RT-PCR), and fluorescence-
activated cell sorting (FACS). Gene expression profiling of
these cells was performed with introduced amplified-fragment
length polymorphism (iAFLP). To determine the presence of
stem- or progenitor cells, immunostaining and colony-forming
assays were performed.
RESULTS. Western blot analysis, RT-PCR revealed that K12 was
expressed in conjunctival epithelium. Immunostaining analysis
showed that K12-positive cells reside mainly in clusters in
conjunctival epithelium. FACS analysis showed that 0.2% to
1.7% of conjunctival epithelial cells collected from the inferior
bulbar conjunctiva were K12 positive. iAFLP analysis revealed
that the gene expression patterns of these cells were highly
similar to that of corneal epithelial cells. p63 and ABCG2 were
expressed beneath the K12-positive cells. Some colony-form-
ing cells expressed K12.
CONCLUSIONS. The K12-positive cells appear to be ectopically
residing, self-maintaining corneal epithelial cells in the con-
junctival epithelium. (Invest Ophthalmol Vis Sci. 2006;47:
Corneal epithelial cells are continuously supplied from the
limbus where their stem cells reside.1–5Conjunctival epithelial
stem cells were found primarily at the conjunctival fornix by
colony-forming assay in humans5and by in vivo label-retaining
experiments in rabbits6and mice.7,8More recently, time-lapse
studies in green fluorescent protein (GFP) mice conjunctiva
disclosed the uniform distribution of stem cells in the bulbar
he ocular surface is covered by two different types of
epithelia: the conjunctival and the corneal epithelium.
conjunctiva.9In addition, the mucocutaneous junction con-
junctiva has been shown to contain stem cells that migrate
toward the fornix.10,11Thus, the site where conjunctival epi-
thelial stem cells reside remains controversial.
Besides their anatomic segregation, the two types of ocular
epithelium possess unique tissue- and cytological properties.
For example, conjunctival epithelium does, while corneal ep-
ithelium does not, contain mucin-secreting goblet cells.12Our
previous gene expression analysis13,14disclosed that many
genes are differentially expressed by these epithelia. Wei et
al.15reported that rabbit conjunctival and corneal epithelial
cells belong to two separate lineages. Based on these observa-
tions, corneal and conjunctival epithelial cells appear to be
The current dogma is that the expression of K3/12 is
thought to be a hallmark of epithelia with cornea-type differ-
entiation1,15–20and to be indispensable for corneal epithelial
homeostasis.21However, K3 is also expressed in other epithe-
lia, including snout,16gingiva, and tongue22and palpebral
conjunctiva.6Also, it has been reported that bovine bulbar
conjunctival epithelial cells expressed trace amounts of K3 and
were induced to express K3/12 by inoculation onto corneal
basement membrane.23Similar findings were made on cul-
tured rabbit conjunctival epithelial cells.6These reports
strongly suggest that the actual expression patterns of K3/12 in
ocular surface epithelium are not as clear cut as the current
We investigated expression of K12 in human conjunctival
epithelium. We found that K12-positive cells are present in this
tissue primarily as clusters and appear to possess cellular fea-
tures highly similar to corneal epithelial cells. Furthermore,
they seem to have their own stem- or progenitor cells. Based
on our observations, we postulated that the K12-positive cells
in conjunctival epithelium are ectopically residing corneal ep-
This study was approved by the Committee for Ethical Issues on
Human Research of Kyoto Prefectural University of Medicine and was
performed in accordance with the tenets of the Declaration of Hel-
Normal conjunctival tissues were obtained from otherwise healthy
eyes at cataract or conjunctivochalasis surgery. The resected normal
conjunctivae from the patients with cataract (n ? 10; three men and
seven women; mean age, 69.0 ? 13.1 years) were 3 ? 3 mm2in size
and located at inferior bulbar conjunctiva 5 mm distant from the
surgical limbus. The resected conjunctivae from the patients with
conjunctivochalasis (n ? 10; three men and seven women; mean age,
71.6 ? 8.7 years) were 3 to 6 mm (vertical) ? 15 mm (horizontal) and
located at the inferior bulbar conjunctiva, at least 2 mm distant from
the limbus. Prior informed consent was obtained from all subjects after
a detailed explanation of the procedures. Cadaveric corneas were
obtained from the Northwest Lions’ EyeBank (Seattle, WA). Permission
to use the donated corneas for research was obtained from all donor
From the Department of Ophthalmology, Kyoto Prefectural Uni-
versity of Medicine, Kyoto, Japan.
Supported by Grants 15791001 and 16390502 from the Japanese
Ministry of Education, Science, Culture and Sports; a grant from the
Japanese Ministry of Health, Labor and Welfare; and research funds
from the Kyoto Foundation for the Promotion of Medical Science.
Submitted for publication August 17, 2005; revised November 26,
2005; accepted February 15, 2006.
Disclosure: S. Kawasaki, None; H. Tanioka, None; K. Ya-
masaki, None; N. Yokoi, None; A. Komuro, None; S. Kinoshita,
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be marked “advertise-
ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Corresponding author: Satoshi Kawasaki, Department of Ophthal-
mology, Kyoto Prefectural University of Medicine, 465 Kajii-cho, Hi-
rokoji-agaru, Kawaramachi-dori, Kamigyo-ku, Kyoto, 602-0841, Japan;
Investigative Ophthalmology & Visual Science, April 2006, Vol. 47, No. 4
Copyright © Association for Research in Vision and Ophthalmology
Corneal and conjunctival tissues were embedded in OCT compound
(Tissue-Tek; Sakura Finetechnical Co. Ltd., Tokyo, Japan) and snap
frozen with liquid nitrogen. Sections were placed on glass slides for
immunostaining and in situ hybridization or on slides (Penfoil; Leica
Microsystems, Co., Ltd., Wetzlar, Germany) for laser microdissection.
Sections or cells were dried and fixed at 4°C with Zamboni’s fixative.
Then they were incubated in blocking solution (1% BSA in 0.01 M PBS),
incubated in the primary antibodies (Table 1), washed with 0.01 M
PBS, incubated again with the corresponding fluorescence-labeled sec-
ondary antibody, immersed in mounting medium, and covered with
Western Blot Analysis
Conjunctival tissue was stretched on a filter paper epithelial side up.
With a spatula, the epithelium was mechanically scraped from the
tissue with special care taken to avoid breaking the underlying stroma.
The collected epithelium was lysed in a buffer containing 25 mM
Tris-HCl (pH 7.4), 0.6 M KCl, 1% Triton X-100, and a protease inhibitor
cocktail (Complete Mini; Roche Diagnostics, Penzberg, Germany). Af-
ter centrifugation, pellets were solubilized in sample buffer containing
25 mM Tris-HCl (pH 6.8), 2% SDS, 5% ?-mercaptoethanol, 10% glycerol,
and 0.005% bromphenol blue (BPB). The samples were electrophoresed,
transferred to a polyvinylidene difluoride (PVDF) membrane (HybondP;
GE Healthcare, Piscataway, NJ) and immunostained with anti-K12 anti-
body (1:1000, sc-17098; Santa Cruz Biotechnology, Santa Cruz, CA).
Fluorescence-Activated Cell Sorting
The conjunctival epithelium was separated from underlying stroma
with 1.2 U/mL dispase24and further disintegrated with 0.05% trypsin/
EDTA. After fixation with 4% paraformaldehyde, the cells were incu-
bated in a blocking buffer containing 0.1 M PBS and 0.5% BSA. After
incubation in a permeabilization buffer (0.5% saponin in the blocking
buffer), the cells were immunostained with anti-K12 antibody (1;100,
sc-17098) or normal goat IgG. After washing in 0.1 M PBS, the cells
were stained with Alexa488 anti-goat IgG (Invitrogen, Carlsbad, CA)
and analyzed by fluorescence-activated cell sorting (FACS; FACSCali-
ber, BD Biosciences, San Jose, CA).
Fluorescence In Situ Hybridization
A specific region for K12 mRNA (nucleotide positions 1337-1792 in
NM000223) was amplified and cloned into a T-overhang vector
(pGEM-T Easy Vector; Promega, Madison, WI). After confirmation by
sequencing, the plasmid was digested with restriction enzyme and
used to prepare a sense or antisense digoxigenin (DIG)-labeled RNA
Frozen conjunctival sections (10 ?m) were fixed with 4% parafor-
maldehyde and incubated in a 0.2-?g/mL proteinase-K solution. Then,
they were acetylated with 0.25% acetic anhydride in 0.1 M triethanol-
amine (TEA; pH 8.0), washed with PBS, and dehydrated with a graded
series of ethanol. After 30-minute air drying, they were incubated in
hybridization buffer containing 50% formamide, 0.3 M NaCl, 20 mM
Tris-HCl (pH 7.5), 5 mM EDTA, 10% dextran sulfate, 1? Denhardt
solution (Wako Pure Chemical Industries, Ltd., Osaka, Japan), 500
ng/mL salmon sperm DNA (Invitrogen), 0.5 mg/mL yeast tRNA (Roche
Diagnostics), and 10 mM dithiothreitol [DTT] plus 10 ng/mL of the
sense or antisense probe. After an 18-hour incubation at 60°C, the
sections were washed twice at 52°C for 30 minutes in 0.5? SSC and
50% formamide and then washed twice for 15 minutes at room tem-
perature in 0.2? SSC. After a 30-minute incubation in a blocking
buffer, the sections were incubated in horseradish peroxidase (HRP)-
labeled anti-DIG antibody (1:100; Roche Diagnostics) solution. After
signal intensification by tyramide signal amplification (Biotin-TSA kit;
Perkin-Elmer Life Sciences Inc., Boston, MA), the sections were incu-
bated with Alexa488-labeled streptavidin (Invitrogen).
Reverse Transcriptase–Polymerase Chain
Reaction and Real-Time PCR
RNAs were extracted from corneal or conjunctival epithelium, reverse-
transcribed, amplified with primer pairs against the genes (Table 2),
and electrophoresed in 2% agarose gels. Southern blot analysis was
performed to validate the results.
Real-time PCR was performed to quantitate the relative gene ex-
pression of K12 and ribosomal RNA (for normalization) using a se-
quence-detection system (Prism 7000 Sequence Detection System;
Applied Biosystems, Ltd. [ABI], Tokyo, Japan). (Sequences for these
primers and internal probes were not disclosed.)
For introduced amplified fragment length polymorphism (iAFLP) anal-
ysis, K12-positive and K12-negative cells were individually harvested
from three individual conjunctivas (Fig. 1) by using a laser-microdis-
section device (AS LMD; Leica Microsystems).
Gene Expression Analysis by iAFLP
Corneal or conjunctival epithelial cells were mechanically peeled from
five corneas or five conjunctivas. RNAs were extracted from these cells
or from six microdissected samples (TRIzol reagent; Invitrogen).
Comprehensive gene expression profiles were examined with the
iAFLP method of Kawamoto et al.25slightly modified. Briefly, double-
stranded cDNA was synthesized with a pUC119-based vector primer,
as described previously,26and digested with MboI for subsequent
adaptor ligation. Small aliquots (approximately one-sixth) of all di-
gested cDNAs were pooled to obtain a reference sample to connect the
data among the different sample sets. Each of the cDNA samples,
including the reference sample, was ligated with an individual length
polymorphic adaptor (TTnew33-TTnew45 adaptors for individual sam-
ples and TTnew48 adaptor for the reference sample). Five different
cDNA samples and the reference sample were pooled to make four
sample sets in total. After PCR amplification with AntVpPst and
T7revBam primers, the four sample sets were digested with BamHI,
ligated with the T7-3000 adaptor, and amplified by PCR with a fluo-
rescent-labeled MA20 primer and a gene-specific primer. Gene-specific
primers (288 genes) were designed to analyze genes that were domi-
nantly and/or specifically expressed in corneal epithelial cells (for gene
selection, we referred to the Bodymap database; http://bodymap.
ims.u-tokyo.ac.jp/). Each amplified product was electrophoresed on a
fluorescence autosequencer (ABI3100 DNA analyzer; ABI) and the
results were analyzed on computer (Genescan and Genotyper soft-
ware; ABI). The resultant gene expression data were further analyzed
with Cluster and Treeview software.27All oligomers and adaptors
except for gene-specific primers used in iAFLP analysis are listed in
TABLE 1. List of Antibodies
Animal Source Dilution
Mono: monoclonal antibody, Poly: polyclonal antibody; PROGEN,
PROGEN Biotechnik GmbH, Heidelberg, Germany; Novocastra, Novo-
castra Laboratories Ltd, Newcastle, UK; Santa Cruz, Santa Cruz Biotech-
nology Inc., Santa Cruz, CA; NC, not commercially available; KAMIYA,
Kamiya Biomedical Company, Seattle, WA.
1360Kawasaki et al.
IOVS, April 2006, Vol. 47, No. 4
Virtual Northern Blot
Full-length cDNAs were amplified by a cDNA synthesis kit (Super Smart
PCR; BD-Clontech, Mountain View, CA). The cDNAs were electropho-
resed, transferred to a nylon membrane, and hybridized with biotin-
A colony-forming assay was performed as described previously.5
Briefly, conjunctival epithelial cells were enzymatically dissociated and
seeded on a feeder layer of MMC-treated 3T3 cells. After 4 to 5 days, the
cells were fixed with Zamboni’s fixative and subjected to immunostain-
Conjunctival epithelial cells were organotypically cultured on an hu-
man amniotic membrane, according to a previously described meth-
od.28,29After 7 days of culture at the air–liquid interface, the cultured
conjunctival epithelial sheet was embedded in OCT compound, cryo-
sectioned, and subjected to immunostaining.
All fluorescent images were acquired with a confocal laser (TCS SP2
AOBS; Leica Microsystems) or a fluorescent microscope (Olympus
Corp., Tokyo, Japan). All chemiluminescent images were acquired in
an intelligent dark box (VersaDoc 5000; Bio-Rad Laboratories, Inc.,
K12-Positive Cells in Conjunctival Epithelium
Immunostaining analysis using five donor tissues extending
from the peripheral cornea to the bulbar conjunctiva revealed
TABLE 2. Oligomers
For the 3? or 5? modification, B means biotinylation, P means phosphorylation, NH2 means amino-
linker, and F means fluorescent dye (6-Fam). Length polymorphic adaptors were made by pooling an
equimolar amount of NH1400 oligomer and one of TTnew (TTnew33-TTnew48) oligomers. The T7_3000
adaptor was made by pooling an equimolar amount of T7_3000 and NH14_rev oligomers. All oligonucle-
otides were synthesized by Promega.
siding in conjunctival epithelium. Conjunctival K12-positive or -nega-
tive cells were selectively collected by laser microdissection (B) by
inspecting contiguous K12-immunostained sections (A).
Microdissection of K12-positive and K12-negative cells re-
IOVS, April 2006, Vol. 47, No. 4
Corneal Epithelial Cell Clusters in Conjunctival Epithelium1361
immunostained with anti-K12 antibody (green) and counterstained with propidium iodide (red). Distance from the end of the Bowman’s membrane
(open arrowhead) to the most distal K12-positive cell cluster (filled arrowhead) was shown at the right of each sample. (B) Expression of K12 in
conjunctival epithelium. Conjunctival tissues were immunostained with anti-K12 antibody (green) and counterstained with propidium iodide (red). Nine
samples (Ba–i) exhibit K12-positive cell cluster(s) while 1 sample (Bj) does not. (C) Expression of the four abundant keratins in conjunctival epithelium.
Conjunctival tissues were double-immunostained against K3 (Ca) and K12 (Cb), K4 (Cd), and K12 (Ce), or K13 (Cg) and K12 (Ch). Note that K3 and
K12 colocalize (Cc), whereas expression of K4 and K12 (Cf) and that of K13 and K12 (Ci) are almost mutually exclusive. (D) K12 expression using two
different antibodies. Conjunctival tissue was immunostained against K12 (green) using either goat polyclonal antibody (Da) or rabbit polyclonal antibody (Db).
Note that these two different K12 antibodies produced consistent immunostaining results. (E) Expression of K12 mRNA in conjunctival epithelium. The
photographs were taken at low (Ea–c) and high (Ed–f) magnification. Images of conjunctival tissues processed by immunostaining (Ea, Ed) or fluorescent in
situ hybridization (Eb, Ee; antisense probe, Ec, Ef; sense probe) demonstrate the consistent expression pattern between K12 protein (green) and its
Tissue localization of K12-positive cells in conjunctival epithelium. (A) Expression of K12 in corneoscleral tissues. Corneoscleral tissues were
1362 Kawasaki et al.
IOVS, April 2006, Vol. 47, No. 4
the presence of K12-positive cell clusters in conjunctiva at a
site far from the limbus (Fig. 2A). On average, the farthest
K12-positive cell cluster in each sample was located at a dis-
tance of approximately 3.7 ? 2.3 mm from the end of Bow-
man’s membrane. Among these, the farthest was 7.4 mm from
the end of Bowman’s membrane, a region that can be consid-
ered to be the bulbar conjunctival or the conjunctival fornix. In
the limbal area, the expression pattern of K12 was almost the
same as that in cornea, except that intermediate to superficial
layers tended to retard K12 expression. We examined the
existence of such cell clusters in conjunctival epithelia of 10
different subjects. All conjunctivae, except that of one subject
(Fig. 2Bj), exhibited K12-positive cells and clusters in the
conjunctival epithelium (Fig. 2B). Double-immunostaining
analysis against K4 and K12 revealed that these keratins were
expressed in a mutually exclusive manner (Figs. 2Cd–f), sug-
gesting that these differentially stained cells have properties
different from each other. We further examined the expression
of K3 and K13, known to form a heterodimer with K12 and K4,
respectively. As expected, the expression patterns for K12 and
K3 were very similar (Figs. 2Ca–c). In contrast, K12 and K13
(Figs. 2Cg–i) presented an image that was similar to the images
produced by double-immunostaining against K4 and K12.
The antibody we used (sc-17098) was raised against the
N-terminal partial peptide sequence. To examine the possibil-
ity that this antibody reacted with other molecules, immuno-
staining was again performed with a different K12 antibody.17
The hypothetical cross-reactivity was almost completely abol-
ished, as this antibody yielded an expression pattern very
similar to that obtained with the other antibody (Fig. 2D).
Tissue localization of K12 mRNA, analyzed by in situ hy-
bridization against sections contiguous with immunostained
sections, demonstrated a tissue distribution pattern that was
highly consistent with that of K12 protein (Fig. 2E). This
finding strongly supported our immunostaining results. West-
ern blot analysis demonstrated that conjunctival epithelial ex-
tracts produced a faint but specific band for K12 (Fig. 3A).
RT-PCR analysis revealed that K12 mRNA was expressed in
both corneal and conjunctival epithelium (Fig. 3B). However,
real-time PCR analysis disclosed that the expression level of
K12 mRNA in conjunctival epithelium was no more than 1% of
that in corneal epithelium (Fig. 3B). FACS analysis also revealed
that 0.2% to 1.7% of the conjunctival epithelial cells collected
from the inferior bulbar conjunctiva was K12 positive (Fig.
3C). Taken together, despite individual variations, our findings
suggest that as many as 1% of conjunctival epithelial cells seem
to be K12 positive.
As some K12-positive cell clusters were located at quite a
distance from the limbus, we hypothesized that these clusters
are physiologically independent and spatially segregated from
this region. To rule out the possibility that they are simply part
of an extended limbal epithelium, we first looked for goblet
cells, thought to be present only in conjunctival epithelium, in
the vicinity of the K12-positive cell clusters. Double-immuno-
staining analysis against K12 and Mus5AC clearly demonstrated
some MUC5AC-positive goblet cells very close to the K12-
positive cell clusters (Fig. 4A). Next, we carefully inspected the
tissue localization of the K12-positive cells in a series of con-
tiguous sections. We found that some K12-positive clusters
existed as solitary islands in conjunctival epithelium (Fig. 4B).
denotes negative control. Protein samples prepared from the insoluble fraction of conjunctival (lanes 1 to 5, 11) or corneal (lane 6 to 10, 12)
epithelial lysates were electrophoresed, transferred, and immunostained against K12 (lanes 1 to 10) or normal goat IgG (lanes 11, 12). Note that
the loaded protein amount of corneal samples was reduced to 1:100 of that of conjunctiva to avoid signal quenching. (B) Expression of the keratin
genes detected by RT-PCR and real-time PCR. (Ba) The expression of K3, -4, -12, and -13 was analyzed by RT-PCR and validated by Southern blot
analysis. Expression of K3 was not detected in conjunctival epithelium with a normal three-temperature thermal setting but was detected under
a touchdown thermal condition (*). (Bb) Quantitative expression of K12 in corneal (Cr1, Cr2, Cr3) and conjunctival (Cj1, Cj2, Cj3) epithelia by
real-time PCR analysis. The expression level of K12 in conjunctival epithelium was no more than 1% of that in corneal epithelium. (Bc) The kinetics
of K12 gene amplification were monitored by real-time PCR. (Open arrowhead) Three corneal samples; (filled arrowhead) Cj2; (filled arrow) Cj3;
(open arrow) Cj1. Red horizontal line: the threshold line. (C) FACS analysis of K12-positive cells in conjunctival epithelium. Conjunctival (Ca–d)
and corneal (Ce) epithelial cells were dispersed by enzymatic dissociation, fixed, and immunostained against K12. In the conjunctiva, K12-positive
cells (purple, M2 region) comprised approximately 0.2% to 1.7% of total analyzed cells. The positive–negative cutoff line was defined according
to the signal distribution of the isotype-negative control (red line).
Expression of K12 in conjunctival epithelium. (A) Expression of K12 protein in conjunctival epithelium (Western blot analysis). NC
IOVS, April 2006, Vol. 47, No. 4
Corneal Epithelial Cell Clusters in Conjunctival Epithelium1363
Similarity of K12-Positive Cells in Conjunctival
Epithelium to Corneal Epithelial Cells
As gene expression patterns vary significantly with tissue or
cell type, we performed gene expression analysis using iAFLP
to assign the K12-positive cells in the conjunctiva to the proper
type of ocular surface epithelium. Of the 288 genes we exam-
ined, 185 could be analyzed; others could not, possibly due to
improper primer sequences. Cluster analysis of the iAFLP data
clearly demonstrated that the gene expression profiles of the
K12-positive cells in the conjunctiva were similar to those of
corneal epithelial cells (Figs. 5A, 5B). Among the genes exam-
ined here, 25 genes exhibited apparently different expression
patterns (Fig. 5B) between conjunctival and corneal epithe-
lium. Especially, TKT30and ALDH331are known as dominant
proteins in the cornea. Of note, TGFBI (keratoepithelin), a
gene involved in hereditary corneal dystrophies,32was highly
expressed both in corneal epithelial cells and conjunctival
K12-positive cells. These data were further validated by RT-
PCR (Fig. 5C) and virtual Northern blot analysis (Fig. 5D). The
results strongly suggest that the K12-positive cells in the con-
junctiva possess properties identical or very similar to those of
corneal epithelial cells.
Stem Cells Associated with the K12-Positive Cells
in the Conjunctiva
If the K12-positive cells in the conjunctiva are not derived from
the limbus, where do these cells come from? We hypothesized
that their stem cells reside just beneath them. Therefore, we
looked for the expression of stem/progenitor cell markers
around K12-positive cell clusters in the conjunctival epithe-
lium. Among several putative markers for limbal basal stem
cells,33we examined K12,34p63,35,36and ABCG2.37Some
basal cells under the K12-positive cell clusters did not express
K12 (Fig. 6A), implying that they were the stem cells of the
overlying K12-positive cell cluster. Also, some basal-to-supra-
basal cells beneath the K12-positive cell cluster expressed p63
(Fig. 6Aa) and ABCG2 (Fig. 6Ab), implying that these cells are
stem/progenitor cells of the overlying K12-positive cell cluster.
To test this hypothesis further, we investigated the colony-
forming activity of these cells. As a result, some of the colonies
expressed K12 (Fig. 6B), indicating that such K12 positive
colonies are stem/progenitor cells of the K12-positive cells
residing in conjunctiva. Then, we tested whether the K12-
positive cells in the conjunctiva can be maintained after orga-
notypic culture. We identified K12-positive cells in cultivated
conjunctival epithelial sheets (Fig. 7).
Immunostaining analysis clearly demonstrated the existence of
K12-positive cells in human conjunctival epithelium. The re-
sults of Western blot analysis, in situ hybridization, FACS, and
RT-PCR analyses further supported this observation. Moreover,
gene expression analysis by iAFLP strongly suggests that the
K12-positive cells in the conjunctiva possess properties highly
similar to those of corneal epithelial cells. In addition, the
K12-positive cells in the conjunctiva appeared to be main-
tained by their own stem or progenitor cells. Based on these
results, we postulate that these cells are ectopically residing
corneal epithelial cells self-maintained in the conjunctiva.
The most important issue in this study appears to be
whether the K12-positive cells in conjunctiva are linked to
corneal epithelial stem cells residing in the limbal basal layer.
Our data strongly suggest that these cells are self-maintained in
conjunctiva and are independent of the limbal basal stem cells.
However, the possibility that limbal epithelium extends to
such a distant region cannot be completely ruled out. Some
radially sectioned corneoscleral tissues (Fig. 2A) demonstrate
contiguous K12-positive cells from the limbus, implying this
possibility. Investigation of whole ocular surface epithelium
from cadaveric donors would shed light on this question.
Data derived from animal experiments led to the classic
concept of conjunctival epithelial transdifferentiation—that is,
conjunctival epithelial cells can become corneal epithelial cells
under certain conditions,38–41thereby making it possible for
eyes with total limbal failure to recover completely and exhibit
limbal epithelium. (A) Expression of K12 and Muc5AC in conjunctival
epithelium. Conjunctival tissue was immunostained against K12 (green)
and Muc5AC (red, arrowhead) to demonstrate the spatial proximity of
the conjunctival K12-positive cell clusters and goblet cells. (B) Isolated
K12-positive cell cluster in conjunctival epithelium. The photographs
show a series of contiguous sections to demonstrate that a conjunctival
K12-positive cell cluster exists as a solitary island. Note that photographs
2 and 29 represent both edges of the K12-positive cluster.
Segregation of conjunctival K12-positive cell clusters from
1364 Kawasaki et al.
IOVS, April 2006, Vol. 47, No. 4
by hierarchical clustering. The 16 samples comprised corneal epithelial cells from five subjects (Corn-1–Corn-5), conjunctival epithelial cells from
five different subjects (Conj-1–Conj-5), and laser microdissected K12-positive (K12(?)1–K12(?)3) and K12-negative (K12(?)1–K12(?)3) cells
from three different conjunctivae. Each row represents an individual gene and each column an individual sample. The data were log-transformed
(base 2) and centered in row-direction by subtracting the median observed value (log space). The data are depicted according to the color scale
(log space) shown at top left. Gray data indicate that the electrophoresis data for the row were under the cutoff. (A) Whole image of
two-dimensional hierarchical clustering of 185 genes across 16 samples. The horizontal hierarchical trees show the degree of similarity in the gene
expression pattern among the 16 samples. Note that the 16 samples are clearly divided into two groups (red and blue trees). The area demarcated
in yellow includes genes with expression that was significantly different in these two groups. (B) Differentially expressed genes between corneal
and conjunctival epithelium . The color-coded matrix is a zoomed image of the area demarcated in yellow in (A). At the right, some well-known
genes are represented by their symbols: GJA1, gap junction protein; ?1 (connexin43); KRT3, keratin3; KRT12, keratin12; TKT, transketolase;
CTSL2, cathepsinL2; TGFBI, beta IgH3; ALDH3, aldehyde dehydrogenase 3. Note that microdissected K12-positive samples manifest gene
expression patterns highly similar to those of corneal epithelial cells. (C) Validation of the iAFLP results by RT-PCR. Lanes 1, 2: corneal epithelium,
lane 3, 4: conjunctival epithelium. All genes, except for the ?-actin gene, demonstrate dominant expression in corneal epithelium. All amplicons
were confirmed by sequencing analysis. (D) Validation of the iAFLP results by virtual Northern blot. Lane 1, 2: corneal epithelium, lane 3, 4:
conjunctival epithelium. Equal amount of amplified cDNAs were electrophoresed and hybridized. Each arrowhead indicates a band of authentic
full-length cDNA of each gene.
Gene expression profiling of the K12-positive cells in conjunctiva. Gene expression data on 185 genes from 16 samples were analyzed
IOVS, April 2006, Vol. 47, No. 4
Corneal Epithelial Cell Clusters in Conjunctival Epithelium1365
a transparent cornea. Although clinical studies provided evi-
dence in support of the transdifferentiation hypothesis42in
humans, other animal experiments and data based on biochem-
ical studies appeared to render this concept invalid,43–47be-
cause neither regenerating conjunctival epithelium covering
the cornea nor organotypically cultured conjunctival epithe-
lium exhibited cornea-specific phenotypes; rather, the pheno-
type was that of conjunctival epithelium. Based on our results,
we postulate that during the epithelial regenerating process in
patients with compromised limbal stem cells, ectopically resid-
ing corneal epithelial cells in the conjunctival epithelium mi-
grate, cover the denuded cornea, and exhibit bona fide corneal
epithelial properties. In this sense, the transdifferentiation con-
cept would be incorrect from a cytological but correct from a
In summary, ours is the first study to demonstrate clearly the
existence of K12-positive cells in in vivo human conjunctival
epithelium. We identified their cellular features by comprehen-
sive gene-expression analysis and found them to be similar to the
features of corneal epithelial cells. Moreover, K12-positive cells
appear to have their own stem or progenitor cells. We submit the
hypothesis that these cells are ectopically residing corneal epithe-
lial cells and that they are self-maintained, even in conjunctival
epithelium. Our preliminary findings that K12-positive cells exist
in organotypically cultured conjunctival epithelium suggest that
these cells can be maintained during the culture process. Sorting
of these cells by FACS may allow us to generate cultured corneal
epithelial sheets from conjunctiva. Studies are under way in our
laboratory to investigate the potential usefulness of conjunctival
epithelium to reconstruct the corneal surface in patients with
limbal stem cell deficiency.
The authors thank Michelle A. Kurpakus for providing the rabbit
polyclonal K12 antibody; the staff at the Northwest Lion’s EyeBank
Foundation, especially Monty Montoya, Bernie Iliakis, Doug Marcoux,
Malcom An, and Jeremy Shuman for helping to obtain fresh human
corneal tissues; and Yoshihide Nakai for helping to obtain fresh human
Conjunctival tissue was double-immunostained against K12 (red) and p63 (Aa, green) or K12 (red) and ABCG2 (Ab, green) and then counterstained
with 4?,6?-diamino-2-phenylindole (DAPI; blue). Note that basal cells beneath the conjunctival K12-positive cell cluster are devoid of K12. (B)
Expression of K12 by colony-forming cells of conjunctival epithelium. Conjunctival epithelial cells were dispersed by enzymatic digestion and
seeded on MMC-treated 3T3 cells. After colonies became obvious, the cells were immunostained against K12 (green). (Ba) Some colonies expressed
K12 (arrow), whereas others did not (arrowhead). (Bb) Zoomed image of the K12-positive colony identified by the arrowhead in (Ba). (Bc) The
K12-negative colony identified by the arrow in (Ba).
Presence of stem cells for conjunctival K12-positive cells. (A) Expression of stem cell markers in conjunctival K12-positive cell clusters.
epithelium. Organotypically cultured conjunctival epithelium was im-
munostained against K12 (green) and then counterstained with 4?,6?-
diamino-2-phenylindole (DAPI; blue).
Expression of K12 in organotypically cultured conjunctival
1366 Kawasaki et al.
IOVS, April 2006, Vol. 47, No. 4
1. Schermer A, Galvin S, Sun TT. Differentiation-related expression of a
major 64K corneal keratin in vivo and in culture suggests limbal
location of corneal epithelial stem cells. J Cell Biol. 1986;103:49–62.
2. Ebato B, Friend J, Thoft RA. Comparison of limbal and peripheral
human corneal epithelium in tissue culture. Invest Ophthalmol Vis
3. Cotsarelis G, Cheng SZ, Dong G, Sun TT, Lavker RM. Existence of
slow-cycling limbal epithelial basal cells that can be preferentially
stimulated to proliferate: implications on epithelial stem cells. Cell.
4. Kruse FE, Tseng SC. A tumor promoter-resistant subpopulation of
progenitor cells is larger in limbal epithelium than in corneal
epithelium. Invest Ophthalmol Vis Sci. 1993;34:2501–2511.
5. Pellegrini G, Golisano O, Paterna P, et al. Location and clonal
analysis of stem cells and their differentiated progeny in the human
ocular surface. J Cell Biol. 1999;145:769–782.
6. Wei ZG, Wu RL, Lavker RM, Sun TT. In vitro growth and differentia-
tion of rabbit bulbar, fornix, and palpebral conjunctival epithelia:
implications on conjunctival epithelial transdifferentiation and stem
cells. Invest Ophthalmol Vis Sci. 1993;34:1814–1828.
7. Lavker RM, Wei ZG, Sun TT. Phorbol ester preferentially stimulates
mouse fornical conjunctival and limbal epithelial cells to prolifer-
ate in vivo. Invest Ophthalmol Vis Sci. 1998;39:301–307.
8. Wei ZG, Cotsarelis G, Sun TT, Lavker RM. Label-retaining cells are
preferentially located in fornical epithelium: implications on con-
junctival epithelial homeostasis. Invest Ophthalmol Vis Sci. 1995;
9. Nagasaki T, Zhao J. Uniform distribution of epithelial stem cells in the
bulbar conjunctiva. Invest Ophthalmol Vis Sci. 2005;46:126–132.
10. Wirtschafter JD, McLoon LK, Ketcham JM, Weinstock RJ, Cheung JC.
Palpebral conjunctival transient amplifying cells originate at the mu-
cocutaneous junction and their progeny migrate toward the fornix.
Trans Am Ophthalmol Soc. 1997;95:417–429; discussion 429–432.
11. Wirtschafter JD, Ketcham JM, Weinstock RJ, Tabesh T, McLoon
LK. Mucocutaneous junction as the major source of replacement
palpebral conjunctival epithelial cells. Invest Ophthalmol Vis Sci.
12. Inatomi T, Spurr-Michaud S, Tisdale AS, et al. Expression of secre-
tory mucin genes by human conjunctival epithelia. Invest Oph-
thalmol Vis Sci. 1996;37:1684–1692.
13. Nishida K, Adachi W, Shimizu-Matsumoto A, et al. A gene expres-
sion profile of human corneal epithelium and the isolation of
human keratin 12 cDNA. Invest Ophthalmol Vis Sci. 1996;37:
14. Dota A, Nishida K, Adachi W, et al. An expression profile of active
genes in human conjunctival epithelium. Exp Eye Res. 2001;72:
15. Wei ZG, Sun TT, Lavker RM. Rabbit conjunctival and corneal
epithelial cells belong to two separate lineages. Invest Ophthalmol
Vis Sci. 1996;37:523–533.
16. Cooper D, Sun TT. Monoclonal antibody analysis of bovine epithe-
lial keratins: specific pairs as defined by coexpression. J Biol
17. Kurpakus MA, Stock EL, Jones JC. Expression of the 55-kD/64-kD
corneal keratins in ocular surface epithelium. Invest Ophthalmol
Vis Sci. 1990;31:448–456.
18. Liu CY, Zhu G, Westerhausen-Larson A, et al. Cornea-specific
expression of K12 keratin during mouse development. Curr Eye
19. Liu CY, Zhu G, Converse R, et al. Characterization and chromo-
somal localization of the cornea-specific murine keratin gene
Krt1.12. J Biol Chem. 1994;269:24627–24636.
20. Wu RL, Zhu G, Galvin S, et al. Lineage-specific and differentiation-
dependent expression of K12 keratin in rabbit corneal/limbal
epithelial cells: cDNA cloning and northern blot analysis. Differ-
21. Kao WW, Liu CY, Converse RL, et al. Keratin 12-deficient mice
have fragile corneal epithelia. Invest Ophthalmol Vis Sci. 1996;37:
22. Collin C, Ouhayoun JP, Grund C, Franke WW. Suprabasal marker
proteins distinguishing keratinizing squamous epithelia: cytokera-
tin 2 polypeptides of oral masticatory epithelium and epidermis
are different. Differentiation. 1992;51:137–148.
23. Kurpakus MA, Stock EL, Jones JC. The role of the basement
membrane in differential expression of keratin proteins in epithe-
lial cells. Dev Biol. 1992;150:243–255.
24. Geggel HS, Gipson IK. Removal of viable sheets of conjunctival
epithelium with dispase II. Invest Ophthalmol Vis Sci. 1985;26:
25. Kawamoto S, Ohnishi T, Kita H, Chisaka O, Okubo K. Expression
profiling by iAFLP: a PCR-based method for genome-wide gene
expression profiling. Genome Res. 1999;9:1305–1312.
26. Okubo K, Hori N, Matoba R, et al. Large scale cDNA sequencing for
analysis of quantitative and qualitative aspects of gene expression.
Nat Genet. 1992;2:173–179.
27. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and
display of genome-wide expression patterns. Proc Natl Acad Sci
28. Koizumi N, Cooper LJ, Fullwood NJ, et al. An evaluation of culti-
vated corneal limbal epithelial cells, using cell-suspension culture.
Invest Ophthalmol Vis Sci. 2002;43:2114–2121.
29. Ang LP, Tan DT, Beuerman RW, Lavker RM. Development of a
conjunctival epithelial equivalent with improved proliferative
properties using a multistep serum-free culture system. Invest
Ophthalmol Vis Sci. 2004;45:1789–1795.
30. Sax CM, Salamon C, Kays WT, et al. Transketolase is a major protein
in the mouse cornea. J Biol Chem. 1996;271:33568–33574.
31. Kays WT, Piatigorsky J. Aldehyde dehydrogenase class 3
expression: identification of a cornea-preferred gene promoter in
transgenic mice. Proc Natl Acad Sci USA. 1997;94:13594–13599.
32. Munier FL, Korvatska E, Djemai A, et al. Kerato-epithelin mutations in
four 5q31-linked corneal dystrophies. Nat Genet. 1997;15:247–251.
33. Chen Z, de Paiva CS, Luo L, et al. Characterization of putative stem
cell phenotype in human limbal epithelia. Stem Cells. 2004;22:
34. Kurpakus MA, Maniaci MT, Esco M. Expression of keratins K12, K4
and K14 during development of ocular surface epithelium. Curr
Eye Res. 1994;13:805–814.
35. Pellegrini G, Dellambra E, Golisano O, et al. p63 identifies keratin-
ocyte stem cells. Proc Natl Acad Sci USA. 2001;98:3156–3161.
36. Yang A, Schweitzer R, Sun D, et al. p63 is essential for regenerative
proliferation in limb, craniofacial and epithelial development. Na-
37. Watanabe K, Nishida K, Yamato M, et al. Human limbal epithelium
contains side population cells expressing the ATP-binding cassette
transporter ABCG2. FEBS Lett. 2004;565:6–10.
38. Kinoshita S, Friend J, Thoft RA. Biphasic cell proliferation in
transdifferentiation of conjunctival to corneal epithelium in rab-
bits. Invest Ophthalmol Vis Sci. 1983;24:1008–1014.
39. Shapiro MS, Friend J, Thoft RA. Corneal re-epithelialization from
the conjunctiva. Invest Ophthalmol Vis Sci. 1981;21:135–142.
40. Thoft RA, Friend J. Biochemical transformation of regenerating ocular
surface epithelium. Invest Ophthalmol Vis Sci. 1977;16:14–20.
41. Friedenwald JS. Growth pressure and metaplasia of conjunctival
and corneal epithelium. Doc Ophthalmol. 1951;5–6:184–192.
42. Kinoshita S, Ohashi Y, Watanabe K, Manabe Y. New surgical
treatment for secondary ocular surface disorders. Jpn J Clin Oph-
43. Tsai RJ, Sun TT, Tseng SC. Comparison of limbal and conjunctival
autograft transplantation in corneal surface reconstruction in rab-
bits. Ophthalmology. 1990;97:446–455.
44. Kruse FE, Chen JJ, Tsai RJ, Tseng SC. Conjunctival transdifferen-
tiation is due to the incomplete removal of limbal basal epithelium.
Invest Ophthalmol Vis Sci. 1990;31:1903–1913.
45. Huang AJ, Tseng SC. Corneal epithelial wound healing in the
absence of limbal epithelium. Invest Ophthalmol Vis Sci. 1991;32:
46. Chen WY, Mui MM, Kao WW, Liu CY, Tseng SC. Conjunctival
epithelial cells do not transdifferentiate in organotypic cultures:
expression of K12 keratin is restricted to corneal epithelium. Curr
Eye Res. 1994;13:765–778.
47. Moyer PD, Kaufman AH, Zhang Z, et al. Conjunctival epithelial
cells can resurface denuded cornea, but do not transdifferentiate
to express cornea-specific keratin 12 following removal of limbal
epithelium in mouse. Differentiation. 1996;60:31–38.
IOVS, April 2006, Vol. 47, No. 4
Corneal Epithelial Cell Clusters in Conjunctival Epithelium1367