Author contributions: K.H.-T.: provision of study material or patients, collection and/or assembly of data; T.N.: conception and design, collection and/or assembly of data, data analysis and interpretation, writing manuscript; N.O.: collection and/or assembly of data, S. KAWASAKI: collection and/or assembly of data; E.P.K.: data analysis and interpretation; Y.B.: financial support; N.K.: data analysis and interpretation; financial support; S. KINOSHITA: collection and/or assembly of data. K.H.T. and T.N. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
first published online in STEM CELLSEXPRESS April 3, 2013.
Leucine-rich repeat-containing G protein-coupled receptor 5 (LGR5), a target of Wnt signaling, is reportedly a marker of intestine, stomach, and hair follicle stem cells in mice. To gain a novel insight into the role of LGR5 in human corneal tissue, we performed gain- and loss-of-function studies. The findings of this study show for the first time that LGR5 is uniquely expressed in the peripheral region of human corneal endothelial cells (CECs) and that LGR5(+) cells have some stem/progenitor cell characteristics, and that in human corneal endothelium, LGR5 is the target molecule and negative feedback regulator of the Hedgehog (HH) signaling pathway. Interestingly, the findings of this study show that persistent LGR5 expression maintained endothelial cell phenotypes and inhibited mesenchymal transformation (MT) through the Wnt pathway. Moreover, R-spondin-1, an LGR5 ligand, dramatically accelerated CEC proliferation and also inhibited MT through the Wnt pathway. These findings provide new insights into the underlying homeostatic regulation of human corneal endothelial stem/progenitor cells by LGR5 through the HH and Wnt pathways. STEM Cells2013;31:1396–1407
In most vertebrates, including humans and other primates, the majority of external information is gained through eyesight, and the cornea is a very important avascular tissue related to the maintenance of this vision system. The cornea consists of a stratified surface epithelial cell layer, a thick collagenous stroma, and an inner single-cell-layered endothelium. Through the combination of these three cell layers, corneal tissue is kept optically clear, and ocular homeostasis and integrity are maintained. According to the World Health Organization, an estimated 25-million people worldwide are affected by cornea-related blindness . Therefore, it is important to understand the underlying mechanisms by which corneal integrity is maintained.
From the medical standpoint, corneal endothelial cells (CECs) represent the most important component of the cornea, as they are crucial for maintaining corneal integrity . CECs, which are derived from the neural crest, play an essential role in the maintenance of corneal transparency through their barrier and pump functions. Although human CECs are mitotically inactive and are arrested at the G1 phase of the cell cycle in vivo , they retain the capacity to proliferate in vitro . However, a recent study has shown that to date, culturing human CECs for a long period of time is extremely difficult . In view of these findings, it is now understood that the molecular mechanism, including the stem cell biology of corneal endothelial behavior, is an important research subject to explore to better understand the role and function of the cornea, as well as to elucidate the most effective means by which to reconstruct damaged corneal tissue.
It is well known that stem cells facilitate the maintenance of self-renewing tissues and organs [6–8]. With regard to corneal tissue, various studies indicate that corneal epithelial stem cells reside in the basal layer of the peripheral cornea in the limbal zone [9–11]. In contrast, even though it has been reported that CECs from the peripheral area of the cornea retain higher replication ability , the corneal endothelial stem cells have yet to be specifically identified and their exact locations are also not fully understood owing to the lack of unique markers and the absence of stem cell assay [13–15].
Recently, genetic mouse models have allowed for the visualization, isolation, and genetic marking of leucine-rich repeat G protein-coupled receptor 5 (LGR5)-positive cells and have provided evidence that there are stem cells in the stomach, small intestine, colon, and hair follicles of those mice [16–18]. LGR5 reportedly is expressed downstream of Hedgehog (HH) signaling in basal cell carcinoma, and LGR5high cells in hair follicles reportedly show active HH signaling [16, 19]. To gain more insights on the mechanism of corneal stem cells, we performed Affymetrix Microarray (Affymetrix, Inc., Santa Clara, CA) analyses using holoclone-type human corneal keratinocytes, and LGR5 was identified as a potential marker for human corneal keratinocyte stem/progenitor cells (data not shown). These findings have led us to an interesting hypothesis that a common stem cell marker exists between developmentally distinct tissues, yet to date, there have been no reports regarding the role and function of LGR5 in CECs.
In this study, we show for the first time that LGR5 is uniquely expressed in the peripheral region of human CECs and that LGR5+ cells have some stem/progenitor cell characteristics. In addition, the findings of this study show that LGR5 is a key molecule for maintaining the integrity of CECs and is mainly regulated by HH and Wnt signaling. Moreover, R-spondin-1 (RSPO1), an LGR5 ligand, was found to dramatically influence the maintenance of CECs. Thus, our data provide new insights into the underlying homeostatic regulation of corneal endothelial stem/progenitor cells by LGR5.
MATERIALS AND METHODS
All human donor cornea tissues were obtained from SightLife (Seattle, WA) eye bank, and all corneas were stored at 4°C in storage medium (Optisol; Bausch&Lomb, Rochester, NY, http://www.bausch.com). A total of 80 donor corneas were used for all experiments (donor age: 61.8 ± 8.6 years (mean ± SD); mean time to preservation: 7.6 ± 5.6 hours; mean endothelial cell density: 2,757 ± 221 mm2; mean storage time: 6.0 ± 0.9 days). All experiments were performed in accordance with the tenets set forth in the Declaration of Helsinki. Eight corneas obtained from cynomolgus monkeys (donor age: 7.1 ± 4.5 years (mean ± SD); estimated equivalent human age: 15–42 years) housed at NISSEI BILIS Co., Ltd., Koka, Japan and Eve Bioscience, Co., Ltd., Japan, respectively, were used for this study. For other research purposes, the monkeys were given an overdose of sodium pentobarbital for euthanization intravenously according to the approval by the Laboratory Animal Use and Ethics Committee of the Shiga Laboratory, NISSEI BILIS Co., Ltd. and the institutional animal care and use committee of Eve Bioscience, Co., Ltd., respectively. The corneas of the cynomolgus monkeys were harvested after confirmation of cardiopulmonary arrest by veterinarians, and were then provided for our research. All corneas were stored at 4°C in Optisol storage medium for less than 24 hours before the experiment. All animals were housed and treated in accordance with the The Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.
The human and monkey CECs were cultured using the method of our previously reported system [2, 20–22]. Briefly, the Descemet's membrane including CECs was stripped and digested with 2 mg/ml collagenase A (Roche Applied Science, Penzberg, Germany, http://www.roche-applied-science.com) at 37°C. After incubation for 3 hours, the CECs (individual cells and cell aggregates) obtained from individual corneas were resuspended in culture medium containing OptiMEM-I (Invitrogen), 5% fetal bovine serum (FBS), 50 μg/ml gentamicin, and 10 μM Y-27632 (Calbiochem, LA Jolla, CA) and then plated in one well of a 12-well plate coated with FNC Coating Mix (Athena Environmental Sciences, Inc., Baltimore, MD, http://www.athenaes.com). The CECs were cultured in a humidified atmosphere at 37°C in 5% CO2. The culture medium was changed every 2 days. When cells reached subconfluence, they were rinsed in Ca2+ and Mg2+-free phosphate-buffered saline (PBS), trypsinized with TrypLE Select (Life Technologies) for 10 minutes at 37°C, and passaged at ratios of 1:2–4.
Immunohistochemical studies followed our previously described method [23, 24]. Briefly, 8-μm-thin sections and whole-mount sections prepared by peeling the Descemet's membrane from cornea tissues were placed on silane-coated slides, air dried, and fixed in 100% acetone at 4°C for 15 minutes. After washing in PBS containing 0.15% TRITON X-100 surfactant (The Dow Chemical Company, Midland, MI, http://www.dow.com) at room temperature (RT, 24°C) for 15 minutes, sections were incubated with 1% bovine serum albumin (Sigma-Aldrich) at RT for 30 minutes to block nonspecific binding. Sections were then incubated with primary antibody at RT for 1 hour and washed three times in PBS containing 0.15% TRITON X-100 for 15 minutes. Control incubations were conducted with the appropriate normal mouse and rabbit IgG at the same concentration as the primary antibody, and the primary antibody for the respective specimen was omitted. The sections were then incubated with the appropriate secondary antibodies at RT for 1 hour. After being washed three times with PBS, the sections were then coverslipped using glycerol-containing propidium iodide (PI) (Nacalai Tesque, Inc., Kyoto, Japan, https://www.nacalai.co.jp), and examined under a confocal microscope (FluoView; Olympus Corporation, Tokyo, Japan, http://www.olympus.co.jp).
Real-Time Polymerase Chain Reaction
Real-time polymerase chain reaction (PCR) was performed following our previously described method . To prepare the samples, we first separated the central cornea from the peripheral cornea using an 8-mm trephine. We then stripped the Descemet's membrane including CECs using micro forceps under a microscope to separate the corneal epithelium, stroma, and endothelium in the central and peripheral cornea, respectively. We then separated the corneal epithelial cells from the corneal stroma using dispase treatment (37°C for 1 hour). All samples were homogenized in lysis buffer (Buffer RLT; QIAGEN, Inc., Valencia, CA http://www.qiagen.com) and total RNA was eluted by use of the RNeasy Mini Kit (QIAGEN) according to the manufacturer's instructions. The relative abundance of transcripts was detected by use of SYBR Green PCR Master Mix (Applied Biosystems, Inc., Foster City, CA http://www.appliedbiosystems.com) according to the manufacturer's instructions. The primers that were used are shown in Table 1.
Table 1. Sequences for PCR and shRNA
For the cell sorting of LGR5high cells, monkey CECs prepared as described above were passaged in 1:2 dilutions and cultured to subconfluence. The CECs were dissociated to single cells by use of TrypLE Select. We then performed the following two experiments. First, the CECs were fixed in 70% (wt/vol) ethanol at 4°C for 2 hours, washed with PBS, and incubated at RT for 15 minutes with 1% FBS. The CECs were then incubated with 1:100-diluted anti-rabbit LGR5 and 1:100-diluted anti-mouse Ki67, washed, and incubated with 1:1500-diluted Alexa Fluor 488-conjugated goat anti- rabbit IgG (Life Technologies) and 1:1000-diluted Cy3 anti-mouse IgG (Jackson Immunoresearch Laboratories). Flow cytometric analyses were then performed with FACS Aria II (BD Biosciences).
Second, the CECs were washed with PBS, and then incubated at RT for 15 minutes with 1% FBS. They were then incubated with 1:100-diluted anti-rabbit LGR5 at RT for 20 minutes, washed, and incubated with 1:1500-diluted Alexa Fluor 488-conjugated goat anti-rabbit IgG (Life Technologies). LGR5high and LGR5low cells were isolated by use of fluorescence activated cell sorting (FACS) Aria II, and the resultant cells were then cultured on an eight-well chamber slide with poly-L-lysine (Sigma-Aldrich). After 5 days of culture, those cells were immunostained by anti-mouse Ki67 as described above, and the Ki67high cells in each group were then counted (n = 4).
Measurement of Cell Area
Each isolated cell fraction was centrifuged and resuspended in culture medium. Cells (approximately 100 cells/ml) were placed in a six-well plate and photographed under an inverted microscope. Cell areas were measured randomly (200 cells/fraction) using Scion Image software and statistically analyzed .
Short hairpin RNA (shRNA) was purchased from Sigma-Aldrich. The LGR5 shRNA targeted sequences and the non-target (NT) shRNA sequences are shown in Table 1. The lentivirus plasmid DNA was transfected to the HEK293T cells along plasmid packaging plasmid mixture (MISSION Lentiviral Packaging Mix; Sigma-Aldrich) using a commercially available transfection reagent (FuGENE HD; Roche Diagnostics Corporation, Indianapolis, IN, http://www.roche-diagnostics.com). After 18 hours, the media was aspirated off and replaced with complete medium. The quantity of lentiviral particles was assessed by HIV-1 p24 Antigen ELISA (ZeptoMetrix Corp., Buffalo, NY, http://www.zeptometrix.com) according to the manufacturer's instructions.
Construction of Lentivirus Plasmid Vector for Gene Expression
For the construction of the lentivirus plasmid vector that expresses the introduced gene, LGR5, a commercially available lentiviral vector (pLenti6.3_V5-TOPO; Life Technologies) was used. cDNAs were amplified with a primer pair (Forward Primer: CTACTTCGGGCACCA TGGACACCT, Reverse Primer: CACATATTAATTAGAGACATGGGA) encompassing an entire coding sequence of LGR5, gel-purified, and then ligated into the lentivirus plasmid vector.
The expression lentivirus Production and Infection were in a modified version of our protocol used for the shRNA . Briefly, the lentivirus plasmid DNA was transfected to the HEK293T cells along with the plasmid packaging plasmid mixture ViraPower Lentiviral Packaging Mix (Life Technologies) which contains pLP1, pLP2, and pLP/VSVG plasmids and FuGENE HD as the transfection reagent. After 18 hours, the media was aspirated off and replaced with complete medium and the quantity of lentiviral particles was then assessed.
The culture supernatant containing the infection-competent virus particle was harvested to human CECs at 5,000 cells/well in a six-well plate with FNC Coating Mix for 24 hours (Multiplicity of infection (MOI) = 1) using the culture medium described above. The supernatant was applied onto cultivated CECs in the presence of 4 μg/ml polybrene. As puromycin-resistant colonies (shRNA experiment) and blasticidin-resistant colonies (overexpression model) were collected, cells were cultured in the presence of 0.4 μg/ml of puromycin and 2 μg/ml of blasticidin, with the media being changed every 2 days.
The cultivated human CECs were washed with PBS and then lysed with lysis buffer containing PBS, 1% TRITON X-100, 0.5 M EDTA, Phosphatase Inhibitor Cocktail two (Sigma-Aldrich), and Protease Inhibitor Cocktail (Roche Diagnostics). Detection of activated β-catenin (nonmembrane bound) was performed according to the previously reported protocol . Briefly, cell lysates treated with Con A Sepharose 4B (GE Healthcare) were incubated at 4°C for 1 hour. After centrifugation at 4°C for 10 minutes, the supernatants were transferred to new tubes and Con A Sepharose was added to each tube and incubated at 4°C for 1 hour. Finally, after a brief centrifugation, the supernatants were transferred to new tubes and their protein concentration was determined.
The proteins were then separated by SDS polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. The membranes were then blocked with 1% ECL Advance Blocking Reagent (GE Healthcare) in Tris Buffered Saline with Tween 20 (TBS-T) buffer and were incubated with primary antibody at 4°C overnight. After being washed three times in TBS-T buffer, the polyvinylidene fluoride (PVDF) membranes were incubated with appropriate HRP-conjugated anti-rabbit or mouse IgG secondary antibody at RT for 1 hour. The membranes were exposed by use of the ECL Advance Western Blotting Detection Kit (GE Healthcare), and then examined by use of the LAS-3000 (FujiFilm Corporation, Tokyo, Japan, http://www.fujifilm.com) imaging system.
Unique Expression Pattern of LGR5 in Human Donor CECs
The expression pattern of LGR5 in human CECs was investigated by indirect immunofluorescence. On examination of the CECs of those tissues, intensive LGR5 expression was observed, especially in the peripheral area. However, LGR5 was only minimally expressed in the corneal epithelium and stroma (Fig. 1A). Real-time PCR showed that compared with stroma and epithelium, mean LGR5 messenger RNA (mRNA) expression was significantly upregulated in the CECs (*p < .05, n = 4, mean age: 60 years) (Fig. 1B). Thus, among the corneal tissues, the expression of LGR5 was found to be most prominent in the CECs.
Next, we examined the location pattern of LGR5 using whole-mount immunofluorescence (n = 3, mean age: 64 years). The expression of LGR5 was observed in the peripheral-region CECs, yet its level gradually decreased in CECs located towards the central region (Fig. 1C, 1D). Real-time PCR clearly showed that the expression of LGR5 in the peripheral regions was upregulated in comparison with the central region (8-mm diameter) (*p < .05, n = 3, mean age: 70 years) (Fig. 1E). These findings indicate that in corneal tissue, LGR5 is uniquely expressed in the peripheral CECs.
Downregulation of LGR5 in In Vitro Culture Conditions
It is well known that the proliferative potential of CECs varies among species . To date, it is extremely difficult to consistently culture human CECs which retain a healthy morphology and high cell density. In contrast, we previously reported that under the proper in vitro conditions, monkey and rabbit CECs can proliferate reasonably well [2, 20–22]. Thus, to gain an insight into the molecular mechanism that underlies the varying proliferative potentials of CECs, we examined the expression of LGR5 in vitro.
Phase contrast microscopy photographs of human peripheral donor CECs revealed that they exhibited a confluent monolayer of smaller-size homogeneously hexagonal cells (Fig. 2A). In contrast, cultured CECs (P0, P1) were found to be enlarged and not homogeneously hexagonal (Fig. 2A). Immunostaining showed that LGR5 was well-expressed in the peripheral donor CECs (Fig. 2A). Worthy of note, the expression of LGR5 was only minimally observed in the cultured CECs in vitro (P0, P1) (Fig. 2A). Real-time PCR showed that the mean LGR5 mRNA expression was significantly downregulated in in vitro CECs as compared to that in peripheral donor CECs (*p < .05) (Fig. 2B).
Phase contrast photographs of monkey CECs showed that both the peripheral donor and the in vitro (P0, P1) cells exhibited a confluent monolayer of smaller-size homogeneously hexagonal cells (Fig. 2C). The expression pattern of LGR5 in the monkey CECs closely mimicked that of in the human donor CECs (data not shown). Immunostaining of those cells showed that LGR5 is moderately expressed both in the donor cells and in vitro cells (Fig. 2C), even though the mean LGR5 mRNA expression in vitro gradually decreased through the cell passages (*p < .05) (Fig. 2D). In view of these findings using human and monkey cells, it is likely that LGR5 may play a role in maintaining the cell integrity of CECs.
LGR5+ CECs Were Small and Exhibited Higher Proliferative Potential
To examine the characteristics of the LGR5+ and LGR5− cell fractions, the subsets were isolated by flow cytometry. To validate the cell sorting procedure, immunofluorescence for LGR5 confirmed its expression at the protein level in the purified fraction (Fig. 3A).
As the highest clonogenicity is reportedly found in the smallest keratinocytes , the cell size in each isolated fraction was measured by use of Scion Image software. Viewed under an inverted microscope, the LGR5+ cells were found to be clearly smaller than the LGR5− cells (Fig. 3B), and the average size of the LGR5+ cells was significantly smaller than that of the LGR5− cells (184.6 ± 45.8 μm2 vs. 326.78 ± 78.8 μm2, respectively, n = 35, **p < .01).
Next, to assess the cell-cycling status of each isolated cell fraction, FACS was used for double-staining with LGR5 and Ki67. FACS analysis showed that the LGR5high/Ki67high cell fraction was 3.4%, whereas the LGR5high/Ki67low cell fraction was 3.8% (Fig. 3C). Most interestingly, all LGR5low cell fractions showed the Ki67 low level (92.8%). To further examine the proliferative capacity of each isolated cell fraction in detail, isolated cell fractions were cultivated on cell chamber slides. Five days later in culture, the percentage of Ki67-labeled cells in the LGR5+ and LGR5− cells was 14.2 ± 3.87% and 0.58 ± 0.5%, respectively, rendering the difference in the Ki67-labeling index statistically significant (*p < .05) (Fig. 3D), suggesting that without the LGR5 expression, CECs do not have proliferative ability.
Active HH Signaling Induced LGR5 Expression
HH signaling reportedly plays a key role in various kinds of biological processes, such as cell differentiation, proliferation, and growth [16, 29, 30]. To define the properties of LGR5 in CECs at the molecular level, we first examined the expression of HH signaling-related molecules in human donor CECs. Of interest, the levels of SHH, Gli1, and Gli2 mRNA were found to be elevated in CECs located in the peripheral-region as compared to those in the central region (Fig. 4A). On the other hand, the expression level of smoothened (Smo) and protein patched homolog one (Ptch1) receptor molecules in the HH pathway was similar (Fig. 4A). Thus, HH signaling was clearly activated in the peripheral-region CECs, suggesting that a regional variation of HH signaling activity does exist.
To determine whether the expression of LGR5 in the CECs was regulated by the HH signaling pathway, the peripheral donor CECs (outside the 8-mm central cornea area) were incubated in culture medium (Dulbecco's modified Eagle's medium + 5% FBS) and stimulated using recombinant SHH (an HH ligand, 100 ng/ml), purmorphamine (an HH agonist, 2 μM) , and cyclopamine (an HH antagonist, 2 μM)  for 24 hours at 37°C in 5% CO2. As expected, expression of LGR5 in the peripheral-region CECs, yet not in the central-region CECs, was found to be upregulated by SHH and purmorphamine stimulation, whereas LGR5 expression was reduced by cyclopamine stimulation at both the mRNA and protein levels (Fig. 4B, 4C). The expression patterns of Gli1 and Gli2 were similar to that of LGR5, but HH activation did not dramatically have an influence on Ptch1, the HH receptor (Fig. 4B).
Next, immunohistochemistry for Ki67 was performed to elucidate whether or not the HH pathway induced donor CEC proliferation. As human CECs are mitotically inactive and show weak-to-no proliferative capacity in vivo , an elevated expression of Ki67 was not observed in all experimental groups, suggesting that stimulation of the HH pathway alone is not sufficient to induce donor CEC proliferation (Fig. 4C). However, CECs reportedly retain the capacity to proliferate in vitro , so we investigated whether the HH pathway induced CEC proliferation in vitro. The expression of Ki67 was found to be upregulated in response to SHH- and purmorphamine-stimulation, however, it was not upregulated in response to cyclopamine (Fig. 4D). These findings indicate that in the in vitro situation, the HH pathway is able to induce CEC proliferation. We posit that CECs treated with cyclopamine were unable to maintain their normal hexagonal morphology (Fig. 4D). Furthermore, Real-time PCR showed that the expression of LGR5 in the cultured CECs with SHH stimulation was elevated as compared to those without SHH stimulation. Immunohistochemistry showed that after SHH stimulation, the expression of LGR5 in the cultured CECs was elevated in some of the cells, yet not in all of the cells (supplemental online Fig. 1). In view of these findings, we discovered for the first time that LGR5 is the target molecule of HH signaling in CECs and that CEC maintenance is partially regulated by the HH pathway.
Downregulation of LGR5 Decreased the Proliferation of CECs
The direct effect of LGR5 on the CECs was elucidated by the knockdown of LGR5 by shRNA. For this experiment, primate cultivated CECs were used, due to the fact that cultured human CECs rarely express LGR5 (Fig. 2A, 2B). Nine sets of shRNA were designed, and the efficacy of their knockdown potential was then examined. Of those, shRNA-589 was found to be the most effective for knocking down the LGR5 mRNA expression (approximately 60% knockdown) (Fig. 5A). Real-time PCR for Ptch1, Gli1, and Gli2 showed that no significant differences were found between the short hairpin LGR5 (shLGR5) group and the control (Fig. 5A). To demonstrate the effect of knocking down the LGR5 gene on CEC proliferation, immunocytochemistry for Ki67 was performed. Compared with the control, cell morphology of the shLGR5-treated cells was not dramatically changed, however, the number of Ki67+ cells in the shLGR5-treated cells was greatly reduced (Fig. 5B). These findings indicated that downregulation of LGR5 did not have an effect on the HH pathway, but did decrease CEC proliferation in vitro.
Persistent LGR5 Expression Inhibited MT Through the Wnt Pathway
To investigate the direct effects of persistent LGR5 expression on CECs, we attempted to overexpress LGR5 using lentivirus containing CMV-LGR5- mRFP. In this experiment, human cultivated CECs (fourth passage, 62-year-old donor) were used, as they rarely express LGR5 (Fig. 2A). Real-time PCR showed that the expression of LGR5 in LGR5-transfected cells (6 days after transfection) was about 60 times higher than that in NT vector-transfected cells (p < .01) (Fig. 6B). Immunofluorescence was used to confirm that the expression of LGR5 in the LGR5-transfected cells was elevated in comparison with that in the NT cells (Fig. 6A). Of great interest, the relative mRNA levels of the HH signaling molecules in LGR5-transfected cells were downregulated as compared to those in the NT cells (Fig. 6B), indicating that LGR5 operates as a negative feedback regulator of the HH pathway.
Human CECs are reportedly vulnerable to morphological fibroblastic change under normal culture conditions . To better demonstrate the effect of persistent LGR5 expression, we used fourth-passaged cultivated CECs. After lentivirus transfection, some of the NT cells still exhibited an enlarged and elongated shape (fibroblastic change) (Fig. 6A). Of great interest, the LGR5-transfected cells gradually changed their morphology and were shown to be compact, smaller-size, homogeneously hexagonal cells, resuming the normal physiological morphology (Fig. 6A). Cell density of the LGR5-transfected cells was found to be greatly elevated compared with that of the NT cells (Fig. 6C). To examine the function of cultivated CECs transfected with the NT and LGR5 vector, immunohistochemistry was performed for Na+/K+ ATPase and ZO1. The expression of these two functional proteins was found to be much greater in the LGR5-transfected cells than in the NT cells, even though these expression patterns were not typical in comparison with those in in vivo CECs (Fig. 6A). In view of these findings, it is likely that LGR5 may be the key molecule for maintaining normal CEC phenotypes.
Transformation of endothelial cells to fibroblastic cells is known as endothelial–mesenchymal transformation (MT) . The interesting findings observed in the LGR5-transfected cells led us to further study whether or not the persistent expression of LGR5 was able to block the MT process. The expression level of epithelial-MT (EMT)-related molecules (Snail, Slug, Twist, and Collagen1)  was examined using real-time PCR. Of great importance, the relative mRNA level of all EMT markers except Slug were lower in the LGR5-transfected cells than in the NT cells (Fig. 6D), suggesting that persistent LGR5 expression blocked the MT process. We further examined which pathway regulates the endothelial-MT observed in CECs. Recent studies suggest that the Wnt/β-catenin signaling pathway plays an important role in EMT . Therefore, the expression level of Wnt/β-catenin-related molecules was examined using western blot analysis. Worthy of note, the protein level of cytosolic (non membrane bound) β-catenin and phosphorylated low-density-lipoprotein receptor-related protein 6 (p-LRP6) was greatly decreased in the LGR5-transfected cells (Fig. 6F). We found that the expression of β-catenin shifted from the cell membrane to the cytoplasm and nucleus, which is well observed in the typical EMT process, in most of nontarget transfected CECs. In contrast, we could observe the expression of β-catenin in cell membrane of LGR5-transfected CECs (supplemental online Fig. 2). These findings indicated that persistent LGR5 expression inhibited the corneal endothelial-MT through the Wnt/β-catenin pathway.
RSPO1-Accelerated CEC Proliferation and Inhibited MT Through the Wnt Pathway
Previously, LGR5 was thought to be an orphan receptor of the G protein-coupled receptor superfamily, and its ligand was unknown. However, several recent reports demonstrated that RSPOs function as ligands of LGR5 to regulate Wnt/β-catenin signaling [35–37]. Interestingly, we discovered that RSPO1, 2, 3, and 4 mRNA were expressed in the corneal epithelium, stroma, and endothelium, and that RSPO1, 2, and 3 mRNA were only expressed in the peripheral-region CECs (supplemental online Fig. 3A). To determine the function of RSPOs on CEC differentiation, we cultured the primary human CECs with or without human recombinant RSPOs. Worthy of note, 7 days after culture, only cultivated human CECs treated with RSPO1 [50 ng/ml] showed the compact, smaller-size, homogeneously hexagonal cells, whereas other RSPOs did not have an obvious effect on CEC differentiation in vitro (supplemental online Fig. 3B). To determine the function of RSPOs on donor CEC proliferation, we performed immunohistochemistry for Ki67. Most surprisingly and very interestingly, human donor CECs incubated with RSPO1 (50 ng/ml) for 48 hours at 37°C showed a dramatically increased level of Ki67+ cell ratios as compared to other RSPOs (supplemental online Fig. 3C). In view of these findings, we think that among the RSPOs family, RSPO1 in particular may play an important role in the maintenance of CECs.
Finally, to further determine the effect of RSPO1 on CECs, we maintained the secondary culture of human CECs with or without RSPO1. Through culturing the CECs in both conditions, we clearly observed that the cultured cells with RSPO1 maintained their hexagonal morphology, whereas some of the cultured cells without RSPO1 still showed fibroblastic phenotypes (Fig. 6E). Moreover, the cell density of RSPO1-treated cells was elevated in comparison with that of the nontreated cells (Fig. 6E). To demonstrate which pathway regulates this type of corneal endothelial MT, we examined the expression level of Wnt/β-catenin-related molecules using western blot analysis. We performed the experiments twice, and the results were nearly identical; the protein level of cytosolic β-catenin and p-LRP6 in the LGR5-transfected cells treated with RSPO1 was decreased in comparison with that in the NT cells (Fig. 6F). Moreover, the protein levels of the RSPO1-treated NT and LGR5-transfected cells were more decreased as compared to the cells not treated with RSPO1 (Fig. 6F). These results suggested that the stimulation of cells overexpressing LGR5 with RSPO1 accelerates pLRP degradation and β-catenin turnover.
Cornea tissue is extremely important, as most mammals acquire the majority of their external information through it. Recently, particular attention has been focused on CECs due to the fact that the corneal transplantation procedure is currently undergoing a paradigm shift from keratoplasty to endothelial keratoplasty. Therefore, both scientifically and clinically, to establish the next generation of novel therapy for treating cornea-related blindness worldwide, it is extremely important to understand the molecular mechanism of corneal endothelial stem/progenitor cells. However, very little is presently known about those molecular mechanisms.
It has been reported that the characteristics and proliferative potential of CECs are different between those located at the central region of the cornea and those located at the peripheral region of the cornea [38, 39], and a study has shown that the cornea has a higher density of endothelial cells in the peripheral region than in the central region . Moreover, CECs from the peripheral region reportedly retain higher replication ability than those from the central region , and peripheral-region CECs contain more precursors and have a stronger self-renewal capacity than CECs in the central region . He et al. recently identified a novel anatomic organization in the peripheral region of human corneal endothelium, suggesting a continuous slow centripetal migration of CECs from specific niches . Thus, it is most likely that human corneal endothelial stem/progenitor cells are mainly distributed in the peripheral region. In fact, no stem/progenitor cell marker for CECs has thus-far been elucidated, and the results of this study demonstrate for the first time that CECs exhibit regional diversity with respect to LGR5 expression. In view of these findings and the unique expression pattern of LGR5 in CECs, LGR5 might represent a first marker for corneal endothelial stem-cell-containing populations.
It has been reported that cell size may distinguish keratinocyte stem cells from transient amplifying cells or differentiated cells . In the epidermis, the response to phorbol esters of the smallest keratinocytes is different from that of other cells. Those keratinocytes also exhibited the highest clonogenicity. Even though CECs are different from ectoderm-derived keratinocytes, the average diameter of the LGR5+ cells in this study was in fact smaller than that of the LGR5− cells. Based on these findings, and on the findings of the above-cited previous report regarding the size of peripheral CECs, it is possible that cell size might be a potential indicator of corneal endothelial stem/progenitor cells.
We found that LGR5 is a key molecule for maintaining the integrity of CECs and regulating normal cell phenotypes in vitro. We also found that isolated cells fractionated based on the intensity of their LGR5 expression could produce different cell populations with different properties. Only cells in the LGR5+ population exhibited exceptionally high proliferative potential, features associated with stem/progenitor cell populations. Based on these findings, the unique expression pattern and necessity in the in vitro condition, there is possibly a link between LGR5 and the function of corneal endothelial stem/progenitor cells.
Previous studies have indicated that high concentration of SHH caused a marked increase in retinal progenitor cell proliferation and a general increase in the accumulation of differentiated cells . The findings of this present manuscript show that in the in vitro situation, the HH pathway is able to induce CEC proliferation, consistent with the findings of previous reports. HH is a family of secreted molecules that serve as morphogens during multiple aspects of development in a wide range of tissue types. HH is involved in the left-right asymmetry decision and anterior–posterior axis decision in limb pattern determination by regulating cell proliferation and survival. In CECs, there is regional variation of HH signal activity, and based on our findings, HH signaling might possibly control corneal endothelial morphogenesis.
RSPOs are a family of four cysteine-rich secreted proteins that were isolated as strong potentiators of Wnt/β-catenin signaling. A vast amount of information regarding the cell biological functions of RSPOs has emerged over the last several years, especially with respect to their role as ligands of the orphan receptors LGR 4/5/6. These updated and important findings led us to further study whether RSPOs may have an effect on the function of human CECs. As human CECs are mitotically inactive and are essentially nonregenerative in vivo, corneal endothelial loss due to disease or trauma is followed by a compensatory enlargement of the remaining endothelial cells. To the best of our knowledge, there are no reports regarding a useful inductive reagent or molecule to increase the level of human CEC proliferation and CEC density, although we previously developed the CECs culture protocol using Y-27632 [21, 22]. We examined the expression of RSPO1 in CECs and found that its protein level is quite low (data no shown), suggesting that external RSPO1, rather than internal RSPO1, plays a critical role in maintaining the CEC function. Moreover, although there was no expression of LGR5 in the cultured CECs, RSPO1 did have some effect on the condition of CECs in vitro. We do not precisely know the reason why, but from our results, we presume that the effect of RSPO1 on CECs might be of both an LGR5-dependent and -independent manner. The findings of this study show for the first time that CECs incubated with RSPO1 exhibited a dramatically increased level of cell proliferation and cell density, suggesting that it might represent a first candidate molecule for reconstructing the damaged cornea through topical application or for use as a culture reagent.
Several studies suggest that the Wnt/β-catenin pathway plays an important role in EMT and that activation of Wnt/β-catenin-dependent signaling modulates the expression of EMT-related genes . However, previous reports have indicated that RSPOs potentiate Wnt/β-catenin signaling by actually functioning as a ligand of LGR5 [35–37]. The exact mechanism involved in this activation is still unclear and there are several conflicting findings as to whether LGR5 is a positive or negative regulator of the Wnt pathway [42–44]. One possible explanation is that the molecular mechanism depends on the tissues, organs, and the species of animal. The cornea is a unique avascular tissue, and its health is maintained by tears and aqueous humor. In contrast, the health of most other organs is maintained by vascular support, suggesting that the characteristics and mechanism of corneal cells are fundamentally different from the epithelial cells of other tissues. Thus, based on the findings of this study, RSPO1 dramatically accelerates CEC proliferation and inhibits corneal endothelial MT through the Wnt pathway.
In conclusion, the findings of this study are the first to demonstrate the function of LGR5 in human CECs (supplemental online Fig. 4). LGR5 has proven to be a powerful tool in identifying a multitude of stem/progenitor cell populations. Through the regulation of LGR5 through the HH and Wnt pathways, CEC integrity was well structured and maintained. In addition, the LGR5 ligand RSPO1 may exploit the novel substantial protocol to provide the efficient expansion of CECs, suggesting that RSPO1-based three dimensional culture and medical treatments hold promise for regenerative therapy, not only for the treatment of corneal dysfunctions, but also for a variety of severe general diseases.
The authors wish to thank Yuiko Hata, Rie Yasuda, Kenta Yamazaki, Yuji Sakamoto, and Shoki Okura for assisting with the experimental procedures, and John Bush for reviewing the manuscript. Drs. Noriko Koizumi and Shigeru Kinoshita had applied for the patent regarding the use of ROCK inhibitor on corneal endothelial research (WO/2009/028631). This study was supported in part by Grants-in-Aid for scientific research from the Highway Program for realization of regenerative medicine, the JST-ETH Strategic Japanese-Swiss Cooperative Program and OptiStem.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.