Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

ABCB5 is a limbal stem cell gene required for corneal development and repair

Abstract

Corneal epithelial homeostasis and regeneration are sustained by limbal stem cells (LSCs)1,2,3, and LSC deficiency is a major cause of blindness worldwide4. Transplantation is often the only therapeutic option available to patients with LSC deficiency. However, while transplant success depends foremost on LSC frequency within grafts5, a gene allowing for prospective LSC enrichment has not been identified so far5. Here we show that ATP-binding cassette, sub-family B, member 5 (ABCB5)6,7 marks LSCs and is required for LSC maintenance, corneal development and repair. Furthermore, we demonstrate that prospectively isolated human or murine ABCB5-positive LSCs possess the exclusive capacity to fully restore the cornea upon grafting to LSC-deficient mice in xenogeneic or syngeneic transplantation models. ABCB5 is preferentially expressed on label-retaining LSCs2 in mice and p63α-positive LSCs8 in humans. Consistent with these findings, ABCB5-positive LSC frequency is reduced in LSC-deficient patients. Abcb5 loss of function in Abcb5 knockout mice causes depletion of quiescent LSCs due to enhanced proliferation and apoptosis, and results in defective corneal differentiation and wound healing. Our results from gene knockout studies, LSC tracing and transplantation models, as well as phenotypic and functional analyses of human biopsy specimens, provide converging lines of evidence that ABCB5 identifies mammalian LSCs. Identification and prospective isolation of molecularly defined LSCs with essential functions in corneal development and repair has important implications for the treatment of corneal disease, particularly corneal blindness due to LSC deficiency.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Figure 1: ABCB5 marks LSCs.
Figure 2: ABCB5 regulates corneal development and repair.
Figure 3: Regenerative role of ABCB5+ LSCs.

Accession codes

Primary accessions

GenBank/EMBL/DDBJ

Data deposits

The murine Abcb5 messenger RNA sequence has been deposited in GenBank under accession number JQ655148.

References

  1. 1

    Davanger, M. & Evensen, A. Role of the pericorneal papillary structure in renewal of corneal epithelium. Nature 229, 560–561 (1971)

    CAS  Article  ADS  Google Scholar 

  2. 2

    Cotsarelis, G., Cheng, S. Z., Dong, G., Sun, T. T. & Lavker, R. M. Existence of slow-cycling limbal epithelial basal cells that can be preferentially stimulated to proliferate: implications on epithelial stem cells. Cell 57, 201–209 (1989)

    CAS  Article  Google Scholar 

  3. 3

    Majo, F., Rochat, A., Nicolas, M., Jaoude, G. A. & Barrandon, Y. Oligopotent stem cells are distributed throughout the mammalian ocular surface. Nature 456, 250–254 (2008)

    CAS  Article  ADS  Google Scholar 

  4. 4

    Dua, H. S., Joseph, A., Shanmuganathan, V. A. & Jones, R. E. Stem cell differentiation and the effects of deficiency. Eye (Lond.) 17, 877–885 (2003)

    CAS  Article  Google Scholar 

  5. 5

    Rama, P. et al. Limbal stem-cell therapy and long-term corneal regeneration. N. Engl. J. Med. 363, 147–155 (2010)

    CAS  Article  Google Scholar 

  6. 6

    Frank, N. Y. et al. Regulation of progenitor cell fusion by ABCB5 P-glycoprotein, a novel human ATP-binding cassette transporter. J. Biol. Chem. 278, 47156–47165 (2003)

    CAS  Article  Google Scholar 

  7. 7

    Schatton, T. et al. Identification of cells initiating human melanomas. Nature 451, 345–349 (2008)

    CAS  Article  ADS  Google Scholar 

  8. 8

    Pellegrini, G. et al. p63 identifies keratinocyte stem cells. Proc. Natl Acad. Sci. USA 98, 3156–3161 (2001)

    CAS  Article  ADS  Google Scholar 

  9. 9

    Luo, Y. et al. Side population cells from human melanoma tumors reveal diverse mechanisms for chemoresistance. J. Invest. Dermatol. 132, 2440–2450 (2012)

    CAS  Article  Google Scholar 

  10. 10

    Wilson, B. J. et al. ABCB5 identifies a therapy-refractory tumor cell population in colorectal cancer patients. Cancer Res. 71, 5307–5316 (2011)

    CAS  Article  Google Scholar 

  11. 11

    Frank, N. Y. et al. ABCB5-mediated doxorubicin transport and chemoresistance in human malignant melanoma. Cancer Res. 65, 4320–4333 (2005)

    CAS  Article  Google Scholar 

  12. 12

    Lin, J. Y. et al. Genetically determined ABCB5 functionality correlates with pigmentation phenotype and melanoma risk. Biochem. Biophys. Res. Commun. 436, 536–542 (2013)

    CAS  Article  Google Scholar 

  13. 13

    Notara, M. & Daniels, J. T. Biological principals and clinical potentials of limbal epithelial stem cells. Cell Tissue Res. 331, 135–143 (2008)

    Article  Google Scholar 

  14. 14

    Rocco, A. et al. MDR1-P-glycoprotein behaves as an oncofetal protein that promotes cell survival in gastric cancer cells. Lab. Invest. 92, 1407–1418 (2012)

    CAS  Article  Google Scholar 

  15. 15

    Cheung, S. T., Cheung, P. F., Cheng, C. K., Wong, N. C. & Fan, S. T. Granulin-epithelin precursor and ATP-dependent binding cassette (ABC)B5 regulate liver cancer cell chemoresistance. Gastroenterology 140, 344–355 (2011)

    CAS  Article  Google Scholar 

  16. 16

    Liu, P., Jenkins, N. A. & Copeland, N. G. A highly efficient recombineering-based method for generating conditional knockout mutations. Genome Res. 13, 476–484 (2003)

    CAS  Article  Google Scholar 

  17. 17

    Rodríguez, C. I. et al. High-efficiency deleter mice show that FLPe is an alternative to Cre-loxP . Nature Genet. 25, 139–140 (2000)

    Article  Google Scholar 

  18. 18

    Hutcheson, D. A. & Kardon, G. Genetic manipulations reveal dynamic cell and gene functions: Cre-ating a new view of myogenesis. Cell Cycle 8, 3675–3678 (2009)

    CAS  Article  Google Scholar 

  19. 19

    Lakso, M. et al. Efficient in vivo manipulation of mouse genomic sequences at the zygote stage. Proc. Natl Acad. Sci. USA 93, 5860–5865 (1996)

    CAS  Article  ADS  Google Scholar 

  20. 20

    Frank, N. Y. et al. VEGFR-1 expressed by malignant melanoma-initiating cells is required for tumor growth. Cancer Res. 71, 1474–1485 (2011)

    CAS  Article  Google Scholar 

  21. 21

    Pellegrini, G. et al. Location and clonal analysis of stem cells and their differentiated progeny in the human ocular surface. J. Cell Biol. 145, 769–782 (1999)

    CAS  Article  Google Scholar 

  22. 22

    Meyer-Blazejewska, E. A. et al. Preservation of the limbal stem cell phenotype by appropriate culture techniques. Invest. Ophthalmol. Vis. Sci. 51, 765–774 (2010)

    Article  Google Scholar 

  23. 23

    Krulova, M. et al. A rapid separation of two distinct populations of mouse corneal epithelial cells with limbal stem cell characteristics by centrifugation on percoll gradient. Invest. Ophthalmol. Vis. Sci. 49, 3903–3908 (2008)

    Article  Google Scholar 

  24. 24

    Lathrop, K. L., Gupta, D., Kagemann, L., Schuman, J. S. & Sundarraj, N. Optical coherence tomography as a rapid, accurate, noncontact method of visualizing the palisades of Vogt. Invest. Ophthalmol. Vis. Sci. 53, 1381–1387 (2012)

    Article  Google Scholar 

  25. 25

    Kao, W. W. et al. Keratin 12-deficient mice have fragile corneal epithelia. Invest. Ophthalmol. Vis. Sci. 37, 2572–2584 (1996)

    CAS  PubMed  Google Scholar 

  26. 26

    Pal-Ghosh, S., Tadvalkar, G., Jurjus, R. A., Zieske, J. D. & Stepp, M. A. BALB/c and C57BL6 mouse strains vary in their ability to heal corneal epithelial debridement wounds. Exp. Eye Res. 87, 478–486 (2008)

    CAS  Article  Google Scholar 

  27. 27

    Hutcheon, A., Sippel, K. C. & Zieske, J. D. Examination of the restoration of epithelial barrier function following superficial keratectomy. Exp. Eye Res. 84, 32–38 (2007)

    CAS  Article  Google Scholar 

  28. 28

    Klocke, J. et al. Spontaneous bacterial keratitis in CD36 knockout mice. Invest. Ophthalmol. Vis. Sci. 52, 256–263 (2011)

    CAS  Article  Google Scholar 

  29. 29

    Pellegrini, G. et al. The control of epidermal stem cells (holoclones) in the treatment of massive full-thickness burns with autologous keratinocytes cultured on fibrin. Transplantation 68, 868–879 (1999)

    CAS  Article  Google Scholar 

  30. 30

    Meyer-Blazejewska, E. A. et al. From hair to cornea: toward the therapeutic use of hair follicle-derived stem cells in the treatment of limbal stem cell deficiency. Stem Cells 29, 57–66 (2011)

    CAS  Article  Google Scholar 

  31. 31

    Veilleux, I., Spencer, J. A., Biss, D. P., Côté, D. & Lin, C. P. In vivo cell tracking with video rate multimodality laser scanning microscopy. IEEE J. Sel. Top. Quant. Electron. 14, 10–18 (2008)

    CAS  Article  ADS  Google Scholar 

  32. 32

    Wang, K. et al. Three-color femtosecond source for simultaneous excitation of three fluorescent proteins in two-photon fluorescence microscopy. Biomed. Opt. Express 3, 1972–1977 (2012)

    CAS  Article  ADS  Google Scholar 

Download references

Acknowledgements

We thank R. Maas for critical reading of the manuscript. We also thank P. Mallen for assistance with graphic illustrations; the Heartland Lions Eye Bank for providing clinical specimens; G. Berg for assistance with mouse colony maintenance; D. Dombkowski, F. Preffer and R. Huang for their assistance with cell sorting and viability studies. This work was supported by National Institutes of Health (NIH)/National Institutes of Neurological Disorders and Stroke grant K08NS051349, VA BLR&D 1I01BX000516 and VA RR&D 1I01RX000989 Merit Review Awards, and a Harvard Stem Cell Institute grant to N.Y.F., NIH/National Cancer Institute grants R01CA113796, R01CA158467 and R01CA138231 to M.H.F., Department of Defense grant PR0332453 to B.R.K., NIH R01EY018624 and P30EY014801 grants to V.L.P., NIH grant R01EY021768 to W.W.K., NIH New Innovator Award DP2OD007483 and a Corley Research Foundation grant to B.A.T., NIH R01CA138231 to G.F.M., Western Pennsylvania Medical Eye Bank Core Grant for Vision Research (EY08098) to K.L.L., and NIH grants R01 EB017274, U01HL100402 and P41EB015903 to C.P.L. F.C.G. is a Howard Hughes Medical Institute Fellow of the Life Sciences Research Foundation.

Author information

Affiliations

Authors

Contributions

N.Y.F., M.H.F. and B.R.K. designed the study. P.E.K., B.R.K., N.Y.F., M.H.F., B.J.W., K.R.S., Q.G., J.M., S.P.M., M.S.G., W.J.B.V., Q.Z., K.L.L., C.P.L., C.A. and L.J.M. performed experiments. V.L.P., F.C.-G. and B.A.T. provided reagents and specimens. W.W.Y.K. and M.K.C. provided technical assistance. P.E.K., B.J.W., Q.G., J.M., S.P.M., M.S.G., W.J.B.V., J.D.Z., G.F.M., B.R.K., M.H.F. and N.Y.F. analysed the data. N.Y.F., M.H.F. and B.R.K. wrote the manuscript. M.H.F. and N.Y.F. are co-senior investigators.

Corresponding authors

Correspondence to Markus H. Frank or Natasha Y. Frank.

Ethics declarations

Competing interests

M.H.F. is co-inventor of the ABCB5-related US patent 6,846,883 (Gene encoding a multidrug resistance human P-glycoprotein homologue on chromosome 7p15-21 and uses thereof) assigned to Brigham and Women’s Hospital, Boston, Massachusetts, and licensed to Ticeba GmbH (Heidelberg, Germany) and Rheacell GmbH & Co. KG (Heidelberg, Germany). M.H.F. serves as scientific advisor to Ticeba GmbH and Rheacell GmbH & Co. KG.

Extended data figures and tables

Extended Data Figure 1 BrdU label-retaining cells and optical coherence tomography identification of the palisades of Vogt.

a, Schematic summary of the experimental design for BrdU pulse-chase experiments. b, Representative flow cytometric analyses depicting specific staining of BrdU label-retaining cells in limbal epithelial cells of wild-type mice that did not receive BrdU (left two panels) or wild-type mice that received BrdU followed by an 8-week chase (right two panels). Limbal epithelial cells were recovered and stained with either anti-BrdU antibody or with an isotype control antibody. The percentages of BrdU-positive cells within the gate are indicated on each plot. ce, Schematic illustration of the optical coherence tomography (OCT) imaging algorithm used for the human limbus. fh, Cross-sectional images of human cornea depicting a sagittal view (f), a coronal C-mode image reconstructed to reveal the palisades (green arrow) and the rete pegs (g), and an axial view of the corneolimbal junction showing the conjunctival stroma beneath the limbal epithelium (green arrow identifies the basal epithelial layer) (h). i, j, Schematic representation of the limbus (i) and the anterior limbus (j), illustrating the orientation of OCT images used to identify and dissect palisade-rich regions within the limbus of corneal rims (approximately 1 cm2 tissue blocks). These smaller sections were then stained with anti-ABCB5 monoclonal antibody and analysed by confocal microscopy, as is shown in Fig. 1e of the main manuscript. Supplementary Video 3 consists of sequential confocal images depicting the location of ABCB5+ cells within palisades.

Extended Data Figure 2 Limbal biopsies from normal donors or patients with LSCD.

ac, Limbal biopsies were obtained from normal donors or patients with LSCD. a, Typical findings are shown for a patient (patient 1) with a chemical burn before receiving a penetrating keratoplasty plus kerato-limbal allograft from a cadaveric donor eye (donor 1). Serial cross-sections of the biopsies were stained with either H&E, isotype control monoclonal antibody, or ABCB5 moncolonal antibody. ABCB5 staining in the limbal epithelium of donor 1 reveals nests of ABCB5-positive cells, whereas ABCB5-positivity is reduced in the limbal epithelium of patient 1. Photographs of immunofluorescent staining are montages of sequential photos at ×20 magnification. In these studies, equal-sized biopsies were recovered from a portion of the patient and donor limbus, frozen, and sectioned to produce eight sequential sections. All epithelial cells were counted in each section. A total of 2,031 and 2,051 epithelial cells were counted in patient 1 and donor 1, respectively. b, Limbal biopsies were obtained from a patient (patient 2) with an autoimmune corneal melt, peripheral ulcerative keratitis, and partial limbal stem cell deficiency before receiving a kerato-limbal autograft from the patient’s normal contralateral eye (donor 2). Serial sections of the biopsies were stained with either H&E, isotype control monoclonal antibody, or ABCB5 monoclonal antibody. ABCB5 positivity was present in the basal layer of the limbal epithelium of donor 2, while a dramatically reduced epithelial layer and no ABCB5 staining was observed in the limbus of patient 2. Photographs of immunofluorescent staining are montages of sequential photos at ×20 magnification. In these studies, equal-sized biopsies were recovered from a portion of the patient and donor limbus, frozen, and sectioned to produce eight sequential sections. All epithelial cells were counted in each section. A total of 563 and 2,662 epithelial cells were counted in patient 2 and donor 2, respectively. Patient 2 had a reduced number of epithelial cells due to the extensive damage from chronic autoimmunity. c, LSCD patient information. *Donor 1: cadaveric donor; **Donor 2: autologous transplant from contralateral eye. KLAL, kerato-limbal allograft (limbal tissue was harvested from donor eye); KLAU, kerato-limbal autograft (part of limbal tissue was resected from uninjured contralateral eye); OD, right eye; PKP, penetrating keratoplasty; PUK, peripheral ulcerative keratitis.

Extended Data Figure 3 Phenotypic evaluation of Abcb5 knockout versus wild-type mice.

a, H&E staining of the normal murine eye depicts location of the limbus and central cornea (left panel). Representative immunofluorescence staining of the wild-type (WT) murine eye illustrates the presence of an Abcb5+ cell population (green) in the limbus but not the central cornea (middle panel). Abcb5 immunofluorescence staining of the knockout (KO) mouse demonstrates loss of Abcb5 expression in the limbus (right panel). b, In situ hybridization with murine antisense or sense Abcb5 mRNA probes spanning 144 bp of murine Abcb5 cDNA encoding exon 10 of the murine Abcb5 gene (GenBank accession number JQ655148) reveals loss of Abcb5 mRNA expression in Abcb5 knockout mice. c, Western blot analyses of murine protein lysates with rabbit polyclonal antibody directed against the Abcb5 N terminus (Abgent) reveal loss of a 80 kDa protein band of the predicted size in Abcb5 knockout mice. d, H&E cross-section of a wild-type eye (left) and an Abcb5 knockout eye (right) depicts a normal shape of the Abcb5 knockout eye and the presence of all major structures including cornea, conjunctiva, iris, lens and retina. e, H&E staining of methacrylate-embedded Abcb5 wild-type (left) and Abcb5 knockout (right) age-matched adult corneas. f, Inflammatory cell marker immunostaining and respective isotype control immunostaining in positive control tissues (left columns) and inflammatory cell marker immunostaining of ABCB5 wild-type and ABCB5 knockout corneas (right columns). Ly-6G, neutrophil marker; F4/80, macrophage marker; CD3, T-cell marker; B220, B-cell marker. Positive control tissues: Staphylococcus aureus-infected murine cornea (Ly-6G); murine spleen (F4/80 and B220); murine lymph node (CD3). g, Representative flow cytometry analyses of either the central corneal (left) or the limbal (right) epithelium of wild-type and Abcb5 knockout mice. Forward scatter (FSC) and side scatter (SSC) indicate cellular size and granularity, respectively. The central corneal epithelium of Abcb5 knockout mice showed reduced numbers of epithelial cells compared to wild-type epithelium (left panels), caused by a reduction in larger cells (right gates), but not smaller cells (left gates). There was no reduction in the numbers of limbal epithelial cells (right panels). Representative results of samples pooled from four eyes are shown (n = 3 experiments). h, Representative flow cytometry analyses of epithelial cells harvested from either the limbus or the central cornea of wild-type and Abcb5 knockout mice. Recovered cells were stained with isotype control antibody, anti-Pax6 antibody, or anti-Krt12 antibody. There was a reduced frequency of Pax6+ cells in the limbus of Abcb5 knockout mice and a reduced frequency of Pax6+ and Krt12+ epithelial cells in the central cornea of Abcb5 knockout mice. Red gates identify Pax6+ or Krt12+ cells compared to isotype control staining. Representative plots of n = 3 experiments for each marker are shown. Magnification in a, b, e, f: ×20; d: ×4.

Extended Data Figure 4 Functional evaluation of Abcb5 knockout versus wild-type mice.

a, Increased corneal fragility in Abcb5 knockout mice. H&E-stained sections of wild-type (WT) and Abcb5 knockout (KO) corneas collected immediately after brushing with a wet Microsponge were examined for the presence or absence of epithelial defects. Only 33% of wild-type animal-derived cornea sections exhibited small epithelial defects (<25% of epithelium), whereas 100% of Abcb5 knockout cornea sections exhibited significant epithelial injury (n = 3 mice per group, 25 sections per mouse, Fisher’s exact test: P < 0.001). be, Wound healing following corneal epithelial debridement of wild-type and Abcb5 knockout mice. b, The area to be debrided was marked with a 2 mm trephine and the epithelium was removed with a small scalpel. c, DAPI-stained cross-section of the cornea immediately following central epithelial debridement depicting the wound margins and exposed central corneal stroma. Image is a montage of sequential photos at ×10 magnification. d, Corneal epithelial wound closure was monitored at 1, 24, and 48 h post-debridement via fluorescein staining. e, Wound closure rates were not significantly different between wild-type and Abcb5 knockout mice (summary of n = 2 replicate experiments, n = 4 mice per group, unpaired t-test, P = not significant). f, Reduced re-epithelialization of wounded corneas in Abcb5 knockout mice. Representative DAPI-stained composite corneal cross sections of wild-type (left) and Abcb5 knockout (right) mice 48 h after a corneal epithelial debridement wound, demonstrating a reduced number of epithelial cells in Abcb5 knockout mice. The white dashed line demarcates the epithelium from stroma; the white box indicates area shown at ×20 magnification (montage pictures are at ×10 magnification); white lines demarcate the area in which epithelial cells were counted. Epithelial cells were counted within the standardized area in at least three consecutive composite cross sections in three replicate mice per group in two separate experiments (aggregate data shown in Fig. 2i). g, Increased apoptosis in wounded corneas of Abcb5 knockout mice. Representative TUNEL-stained composite corneal cross-sections of wild-type (left) and Abcb5 knockout (right) mice 48 h after a corneal epithelial debridement wound, demonstrating increased numbers of apoptotic cells in Abcb5 knockout mice. Areas defined by the white box are shown at ×20 magnification (montage pictures at ×10 magnification). The number of TUNEL-positive epithelial cells was counted, and the data from two replicate experiments in n = 2 mice are summarized in Fig. 2k. Error bars indicate s.e.m. NS, not significant.

Extended Data Figure 5 Phenotypic characterization of wild-type versus Abcb5 knockout retina.

Analysis of H&E-stained sections from 9-month-old wild-type mice (left) or Abcb5 knockout mice (right) revealed changes in the retina and retinal pigment epithelial cells (RPEs). a, Compared to wild-type mice (left), RPEs in Abcb5 knockout mice (right) were enlarged and distended, possibly due to the presence of vacuoles. b, Compared to wild-type mice (left), areas of abnormal RPE in Abcb5 knockout eyes (right) coincided with changes in the overlaying photoreceptors and the outer nuclear layer. There was a thinning and attenuation of photoreceptor outer segments along with a disruption of inner segments, which was associated with a loss of cells in the outer nuclear layer. Magnification: ×20.

Extended Data Figure 6 ABCB5 regulates LSC quiescence and functions as an anti-apoptotic molecule.

a, Flow cytometric analysis showing depletion of BrdU label-retaining cells in Abcb5 knockout (KO) versus Abcb5 wild-type (WT) limbal epithelial cells after an 8-week chase. Analysis was performed in n = 6 Abcb5 wild-type mice and in n = 12 Abcb5 knockout mice. The experiment was performed three times. Data were analysed using the unpaired t-test. Data are shown as mean ± s.e.m., P < 0.05. b, Equivalent BrdU labelling in Abcb5 wild-type and Abcb5 knockout mice after a 24 h chase is shown (mean ± s.e.m.). The experiment was performed using n = 4 mice per group and was performed twice. Data were analysed using the unpaired t-test, P = not significant. c, Schematic illustrating identification of the limbal epithelium within corneal whole mounts via identification of posteriorly localized limbal vessels within the underlying stroma. Far right, confocal Z-stack images displaying limbal vessels alone (top, white arrows) and limbal vessels (white arrows) with overlying limbal epithelium (bottom) (see also Supplementary Video 2). d, Sequential immunofluorescence histological examination of limbal epithelium in corneal whole mounts (localized as illustrated in c and Supplementary Video 2), showing equivalent BrdU incorporation after a 24 h chase in Abcb5 knockout and wild-type mice (column 2), but progression to selective loss of BrdU label-retaining cells in Abcb5 knockout mice at the 8-week chase time point (far right column). Far left column, negative BrdU immunostaining result (negative control) of BrdU-untreated wild-type and Abcb5 knockout mouse limbal epithelium. e, Immune fluorescence analysis of Ki67 expression in Abcb5 wild-type and Abcb5 knockout mouse limbus and cornea. Bar graphs on the right illustrate the percentage of Ki67+ cells in Abcb5 wild-type versus Abcb5 knockout mice in the limbus and cornea. The percentages of Ki67+ cells in Abcb5 wild-type versus Abcb5 knockout mice in the limbus and cornea were determined using n = 4 mice per group. The experiment was performed twice. Within a standardized area, all corneal epithelial cells were counted in at least three consecutive cross-sections. Data were analysed using the unpaired t-test. Data are shown as mean ± s.e.m. f, Flow cytometric analysis of ABCB5 expression by p63α-rich human limbal epithelial cells. g, Cell viability measured in relative luciferase units (RLU) following 48 h of ABCB5 monoclonal antibody or isotype control monoclonal antibody treatment (n = 5 experimental replicas, mean ± s.e.m.). Data were analysed using the unpaired t-test, P < 0.001. h, i, Differential expression of apoptosis pathway-associated proteins detected by Proteome Profiler Apoptosis Array (ARY009, R&D Systems) following ABCB5 monoclonal antibody or isotype control monoclonal antibody treatment, analysed using ImageJ software. Error bars indicate s.e.m. *P < 0.05, ***P < 0.001. NS, not significant. Magnification in ce: ×20.

Extended Data Figure 7 Limbal stem cell transplantation protocol and cell sorting for purification of ABCB5+ and ABCB5 limbal epithelial cells.

a, Recovery and separation of ABCB5+ and ABCB5 limbal epithelial cells from donor corneas followed by preparation of fibrin gels containing donor cells. b, Induction of limbal stem cell deficiency in recipient mice and transplantation of donor grafts. c, e, Representative flow cytometry analyses showing sorting gates and viability of murine (c) and human donor limbal epithelial cells (e). d, f, Post-sort analysis depicting the purity and viability of ABCB5+-enriched and ABCB5-enriched subpopulations of limbal epithelial cells isolated from mice (d) and human donors (f). Viability is shown as the percentage of cells excluding DAPI. g, KRT12 expression in ABCB5+ and ABCB5 limbal cell populations. h, Number and viability of donor cells used for transplantation.

Extended Data Figure 8 Restoration of LSCD by donor murine Abcb5+ or human ABCB5+ cell transplants.

Representative H&E composite corneal cross-sections of recipient C57BL/6J mice 5 weeks after receiving an induced LSC deficiency followed by engraftment of donor fibrin gel transplants containing the following syngeneic murine limbal epithelial cell subpopulations: (1) no cells (negative control), (2) Abcb5+ cells, (3) Abcb5 cells, or (4) unsegregated cells. A normal untreated cornea (no LSCD) served as a positive control. The positive control displays the typical stratified corneal epithelium and iridocorneal angle. Mice receiving transplants with no cells displayed the typical conjunctivalization that occurs following a LSCD, that is, unstratified conjunctival epithelium covers the cornea with extensive inflammation, neovascularization, and stromal oedema. Synechia (where the iris adheres to the cornea) is typical of intense anterior segment inflammation. In contrast, mice that received transplants of Abcb5+ cells, but not Abcb5 cells, displayed a restored stratified corneal epithelium with no evidence of inflammation, neovascularization, stromal oedema, or synechia. Mice that received transplants of unsegregated limbal epithelial cells displayed areas of stromal oedema with unstratified epithelium, while other parts of the cornea contained normal stratified epithelial cells. b, Restoration of Krt12 expression by donor murine Abcb5+ cell transplants. Representative immunofluorescent Krt12 staining (green) of recipient C57BL/6J mice 5 weeks after an LSCD induction followed by transplantation of donor fibrin gel grafts containing grafts as in a. Normal untreated murine cornea (no LSCD), shown here as a positive control, displays high intensity of Krt12 staining. Mice that received grafts containing no cells displayed no Krt12 expression. In contrast, mice transplanted with Abcb5+ cells, exhibited significantly enhanced Krt12 expression in comparison to the mice transplanted with unsegregated limbal epithelial cells. No Krt12 expression was detected in the mice transplanted with Abcb5 cells. The white box depicts the area shown at ×40 magnification. Montage images are shown at ×10 magnification. c, Restoration of LSCD by donor human ABCB5+ limbal cell transplants. Representative H&E composite corneal cross-sections of recipient immunodeficient NSG mice 5 weeks after LSCD induction followed by transplantation of donor fibrin gel grafts containing the following human limbal epithelial cell subpopulations: (1) no cells (negative control), (2) ABCB5+ cells, (3) ABCB5 cells, and (4) unsegregated cells. The positive control (normal untreated NSG cornea (no LSCD)) displays the typical stratified corneal epithelium and iridocorneal angle. Mice that received transplants with no cells displayed evidence of conjunctivalization that occurs following a LSC deficiency; that is, unstratified conjunctival epithelium covers the cornea with extensive neovascularization and synechia (anterior segment inflammation is muted in NSG mice due to their immunodeficiency). In contrast, mice that received transplants containing ABCB5+ cells displayed areas of restored stratified epithelium, whereas recipients receiving ABCB5 cell grafts did not.

Extended Data Figure 9 Long-term corneal restoration by donor human ABCB5+ cell transplants 13 months after transplantation.

a, Representative H&E composite corneal cross-sections of recipient immunodeficient NSG mice 13 months after LSCD induction followed by transplantation of donor fibrin gel grafts containing: (1) no cells (negative control), (2) ABCB5+ cells, (3) ABCB5 cells, and (4) unsegregated cells. A normal untreated NSG cornea (no LSCD) served as a positive control. The positive control displays the typical stratified corneal epithelium. Mice that received transplants with no cells displayed evidence of conjunctivalization that occurs following a LSC deficiency; that is, unstratified conjunctival epithelium covers the cornea with extensive neovascularization and synechia (anterior segment inflammation is muted in NSG mice due to their immunodeficiency). In contrast, mice that received transplants containing ABCB5+ cells displayed restored stratified epithelium, whereas recipients receiving ABCB5 cell grafts did not. b, Representative immunofluorescent KRT12 staining (green) of corneas derived from NSG mice 13 months after LSCD induction followed by transplantation of donor fibrin gel grafts containing the following human limbal epithelial cell subpopulations: (1) no cells (negative control), (2) ABCB5+ cells, (3) ABCB5 cells, or (4) unsegregated cells. Normal untreated murine cornea (no LSCD), shown here as a positive control, displayed a high intensity of KRT12 staining. As expected, recipients of grafts containing no cells displayed no KRT12 expression. Mice transplanted with ABCB5+ cells exhibited significant KRT12 expression, also enhanced compared to mice transplanted with unsegregated limbal epithelial cells. No KRT12 expression was detected in mice transplanted with ABCB5 cells. The white arrow depicts the area shown at ×40 magnification. Montage images are shown at ×10 magnification. c, Representative immunofluorescent KRT12 staining (red) of human and mouse cornea confirms specific antibody reactivity with human KRT12 (top left), and no cross-reactivity with murine Krt12 (bottom left). Isotype control antibody staining is shown for the respective tissues in the right panels. Nuclei are stained with DAPI (blue). Bar graph (bottom) demonstrates aggregate antibody staining data of either human cornea (pixel intensity 142.3 ± 2.4 pixels µm−2, mean ± s.e.m.) or mouse cornea (pixel intensity 1.3 ± 0.7 pixels µm−2, mean ± s.e.m.). Aggregate human KRT12 antibody staining data of either human or mouse cornea was derived from the analyses on n = 2 corneas per group. Within a standardized area in (b), all corneal epithelial cells were counted in at least n = 3 consecutive cross-sections. Data were analysed using the unpaired t-test. Error bars show s.e.m. ***P < 0.001. NS, not significant.

Extended Data Figure 10 Corneal epithelial restoration in 13-month-old human transplants.

ac, Corneal epithelial restoration in 13-month-old human transplants examined by reflectance confocal microscopy. a, Schematic illustration of the cornea. b, c, Confocal microscopic reflectance was used to image 400 × 400 µm areas in the central cornea in the normal eye, as well as control and treatment groups. Representative en face images of normal corneal epithelial layers depicting superficial squamous and basal cuboidal epithelial cells, and the corresponding cross-sectional image depicting the epithelial layer thickness, are shown in the extreme left panels of b and c, respectively. Additional panels in b and c from left to right show, show representative en face (b) and cross-sectional (c) images of recipient NSG mice 13 months after LSCD induction followed by transplantation of donor fibrin gel grafts containing the following human limbal epithelial cell subpopulations: (1) no cells (negative control), (2) ABCB5 cells, (3) unsegregated cells, or (4) ABCB5+ cells. Normal untreated cornea (no LSCD), shown here as a positive control, displayed a typical stratified epithelium of normal thickness. Mice that received grafts containing no cells displayed no stratified epithelium and a significantly reduced epithelial layer. Mice transplanted with ABCB5 cells displayed a thin unstratified epithelium that was not significantly different from the negative control group. Mice transplanted with unsegregated limbal cells displayed a mixture of stratified and unstratified epithelium that was significantly thinner compared to normal corneas. In contrast, only mice transplanted with ABCB5+ cells displayed a normal stratified epithelium with superficial squamous and basal cuboidal epithelial cells, and a thickness not only significantly greater than in alternative treatment or untreated control groups, but comparable to normal healthy cornea (no significant difference), as determined by measurements of cross-sectional image data. Epithelial layer thickness measurements were performed in all groups using ImageJ software and cross-sectional reflectance confocal microscopy imaging (4 mice per group, 10 measurements per cornea) through the central region of the cornea. The measurements were performed on mice from two independent experiments. Data were analysed using the one-way ANOVA and Bonferroni multiple comparisons tests. Aggregate results are illustrated in c, bottom panel bar graph (mean thickness in micrometres (µm) ± s.e.m.). df, Corneal stromal architecture of 13-month-old human transplants examined by reflectance confocal and second harmonic generation microscopy. d, Schematic illustration of the cornea. e, f, Reflectance confocal and second harmonic generation microscopy was used to image 400 × 400 µm areas in the central stroma of the normal eye, as well as of control and treatment groups. Representative en face images of normal cornea (e, extreme left panels) show normal stromal keratocytes as determined by confocal reflectance (top) and stromal architecture as determined by second harmonic generation of collagen fibrils (magenta images, bottom). The corresponding cross-sectional image depicting the stroma layer thickness is shown in the extreme left of panel f. Additional panels in e and f from left to right show representative en face and cross-sectional images of recipient NSG mice 13 months after LSCD induction followed by transplantation of donor fibrin gel grafts containing the following human limbal epithelial cell subpopulations: (1) no cells (negative control), (2) ABCB5 cells, (3) unsegregated cells, or (4) ABCB5+ cells. Normal untreated cornea (no LSCD), shown here as a positive control, displayed typical stromal keratocytes and a normal collagen fibril pattern, with normal stromal thickness determined in cross-sectional images. Mice that received grafts containing no cells displayed a high level of reflectance due to inflammation (also compare H&E staining in Extended Data Fig. 9a) and stromal oedema as shown by increased stromal thickness. In addition, an abnormal collagen fibril pattern was observed, possibly due to deposition of new collagen by infiltrating inflammatory cells. Mice transplanted with ABCB5 cells displayed a high level of reflectance, an abnormal collagen fibril pattern, and stromal oedema that was not significantly different from the negative control group. Mice transplanted with unsegregated limbal cells also displayed increased reflectance, an abnormal collagen fibril pattern, and stromal oedema. In contrast, only mice transplanted with ABCB5+ cells displayed a normal pattern of stromal keratocytes and collagen fibrils, and a stromal thickness comparable to normal healthy cornea (no significant difference) and indicative of absent oedema, as determined by measurements of cross-sectional image data. Stromal thickness measurements were performed in all groups using ImageJ software and cross-sectional second harmonic microscopic images of collagen fibrils (4 mice per group, 5 measurements per stroma) through the central region of the stroma. The measurements were performed on mice from two independent experiments. Data were analysed using the one-way ANOVA and Bonferroni multiple comparisons tests. Aggregate results are illustrated in f, bottom panel bar graph (mean thickness in µm ± s.e.m.). Error bars show s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. NS, not significant. Magnification in af: ×60.

Supplementary information

Limbal epithelial cell Abcb5/BrdU co-expression.

Immunofluorescent staining of limbal epithelial cells for Abcb5 (red) and BrdU (green) co-expression counterstained with DAPI (blue), analyzed by confocal microscopy in the limbal epithelium of mouse corneal whole mounts. This supplementary video consists of sequential confocal images depicting a cluster of ABCB5(+)/BrdU(+) label-retaining cells surrounded by double-negative ABCB5(-)/BrdU(-) cells and relatively rare, single-positive ABCB5(+)/BrdU(-) cells. Magnification: 60x. (AVI 13246 kb)

Identification of the limbal epithelium.

Identification of the limbal epithelium within corneal whole mounts via identification of posteriorly localized limbal vessels within the underlying stroma. Cell nuclei are counterstained with DAPI (blue) and sequential confocal images are displayed beginning from the apical corneal epithelial layer and moving down into the stroma of the cornea and limbus. Appearance of the limbal vessels in the stroma is used to identify the overlaying limbal epithelial layer. This technique was used for immunostaining analyses. Magnification: 20x. (MOV 2108 kb)

ABCB5(+) cells within palisades of Vogt.

Sections through the palisades of Vogt in the anterior limbus were stained with anti-ABCB5 mAb (red) with nuclear counterstaining (blue), and analyzed by confocal microscopy. This supplementary video consists of sequential confocal images depicting the location of ABCB5(+) cells within palisades. Magnification: 20x. (AVI 321 kb)

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ksander, B., Kolovou, P., Wilson, B. et al. ABCB5 is a limbal stem cell gene required for corneal development and repair. Nature 511, 353–357 (2014). https://doi.org/10.1038/nature13426

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing