Wnt6 plays a complex role in maintaining human limbal stem/progenitor cells

The corneal epithelium is consistently regenerated by limbal stem/progenitor cells (LSCs), a very small population of adult stem cells residing in the limbus. Several Wnt ligands, including Wnt6, are preferentially expressed in the limbus. To investigate the role of Wnt6 in regulating proliferation and maintenance of human LSCs in an in vitro LSC expansion setting, we generated NIH-3T3 feeder cells to overexpress different levels of Wnt6. Characterization of LSCs cultured on Wnt6 expressing 3T3 cells showed that high level of Wnt6 increased proliferation of LSCs. Medium and high levels of Wnt6 also increased the percentage of small cells (diameter ≤ 12 µm), a feature of the stem cell population. Additionally, the percentage of cells expressing the differentiation marker K12 was significantly reduced in the presence of medium and high Wnt6 levels. Although Wnt6 is mostly known as a canonical Wnt ligand, our data showed that canonical and non-canonical Wnt signaling pathways were activated in the Wnt6-supplemented LSC cultures, a finding suggesting that interrelationships between both pathways are required for LSC regulation.

. Favorable long-term outcome has been reported 13 , which indicates the ability of the NIH-3T3 feeder cells to serve as a surrogate niche. However, the success rate has not improved over the last 2 decades. Further improvement of the culture condition by supplementing additional niche factors could improve the clinical outcome. We previously reported that one of the critical factors of the LSC niche both in vivo 20,21 and in vitro 22,23 , is Wnt signaling. Four Wnt ligands, Wnt2, 6, 11 and 16b had a higher expression in the limbus compared to the cornea 20,21 . Wnt ligands are small lipid-modified proteins that act over a relatively short distance and serve as niche factors 24 . Wnt signaling is required for the survival of LSCs 22 . However, characterization of the individual Wnt functions has not been conducted. Elucidation of the molecular events involved in the maintenance of the LSC is necessary for the fine manipulation of the biological factors involved in their ex vivo expansion.
Wnt6, a Wnt family member responsible for the maintenance of epithelia during vertebrate embryogenesis 25,26 , is upregulated and preferentially expressed at the basal limbal epithelial layer 20 . Frizzled 7 (Fzd7) which is present in the basal layer of limbal epithelium is a receptor of Wnt6 23,25 . Therefore, Wnt6 possibly signals via Fzd7 and may be an important player in the LSC niche. NIH-3T3 cells do not express Wnt6. Consequently, if Wnt6 does play a vital role in the maintenance of the LSC niche, NIH-3T3 feeder cells may not provide an optimal niche for LSCs. In the currently study, the role of Wnt6 in the regulation of LSCs was investigated by supplementing Wnt6 in the LSC culture via ectopic expression of Wnt6 in the NIH-3T3 feeder cells.

Results
Wnt6 mRNA is expressed in the human limbus and in Wnt6-3T3. Quantitative (q) RT-PCR showed that the expression of Wnt6 mRNA was 5.17-fold greater in the limbus than in the central cornea (Fig. 1A). The level of endogenous Wnt6 mRNA expression in the NIH-3T3 feeder cells was minimal or absent. To supplement Wnt6 in LSC cultures, the NIH-3T3 feeder cells were transduced with a modified lentiviral vector containing the full Wnt6 coding sequence (Wnt6-3T3). The vector, pRRL.CAG.Wnt6.IRES.GFP, induced the expression of Wnt6 under a CAG promoter, and the expression of green fluorescent protein (GFP) under an internal ribosome entry site (IRES) element (Fig. 1B). Wnt6-3T3 cells were sorted on the basis of GFP fluorescence intensity (low, medium, and high). NIH-3T3 cells that expressed GFP only, at low-, medium-, and high-levels, were sorted and used as controls (GFP-3T3). Correlation between GFP intensity and Wnt6 mRNA levels were verified by qRT-PCR (Fig. 1C). In human limbus, mRNA level of Wnt6 expression was comparable to the level in low-Wnt6-3T3 cells (Fig. 1D). In situ hybridization using the RNAscope V2 Fluorescent assay (ACD, Newark, California) con-   20 , as Wnt6 mRNA was highly expressed, preferentially in the basal limbal epithelial cell layer ( Fig. 2 and Supplemental Fig. 1).

Effects of Wnt6 on stem cell properties of human LSCs in vitro.
To determine the effects of Wnt6 on human LSCs, freshly isolated limbal epithelial cells were seeded on Wnt6-3T3 feeder cells that expressed low-, medium-, and high-levels of Wnt6 and their respective low-, medium-, and high-GFP-3T3 controls. The cell morphology was similar in the controls and Wnt6 supplemented cultures (Fig. 3A), indicating that Wnt6 expression had no immediate cytotoxic effect on LSC growth. Analysis of the colony-forming efficiency (CFE), a classic functional analysis of the clonogenic capacity of the progenitor cell population, showed that the CFE was significantly increased in high-Wnt6-3T3 cells (13.22%) as compared to the control high-GFP-3T3 cells (12.19%, 1.03% increase; P < 0.05) and the uninfected NIH-3T3 cells (11.26%, 1.96% increase; P < 0.05) ( Fig. 3B-D). Stem cells must conserve resources in times of quiescence and replication, which usually results in smaller cell size from reduced cytoplasmic/nuclear ratios 27,28 . This trend is observed in the limbus with stem/progenitor cells being about 12 µm or smaller 29,30 . Quantification of cultured cells revealed that the percentage of small cells (≤ 12 µm) significantly increased for cells cultured under medium (2.01%, P < 0.05) and high (2.64%, P < 0.05) levels of Wnt6 (Fig. 3E). We also examined expression of p63α bright , the currently accepted LSC marker 31 , however, expression was reduced in comparison to uninfected controls (Fig. 3F). To demonstrate that cultured LSCs maintained undifferentiated status, we analyzed populations for cytokeratin (K)14, a marker for undifferentiated cells 32 . Comparison between all treatment and control groups showed comparable percentages of cells positive for K14 (Fig. 3G). The presence and maintenance of LSCs is further demonstrated by the significant decrease detected in K12 + cells (Fig. 3H), the differentiation marker 33 . The expression of K12 + in the cultured cells was significantly reduced in both medium-(0.25%, P < 0.05) and high-(0.31%, P < 0.05) Wnt6-3T3 cultures, demonstrating the maintenance of a progenitor state. Although the expression of K12 + decreased by a small percentage it is large considering that the maximum percentage of differentiating cells was less than 1% of the entire cultured population. Altogether, these results demonstrate that Wnt6 supports a more a suitable growth conditions for LSC expansion, consistently with findings from our previous studies 20,22 .

Wnt6 activation of canonical signaling in LSCs.
To determine which signaling pathways are activated in LSCs upon Wnt6 exposure, freshly isolated single LSCs were cultured on control 3T3 feeder cells and treated with Wnt6-conditioned media (CM). Wnt signaling may be dosage-dependent, as demonstrated by morphogen gradients, where high concentrations may predominantly initiate the canonical pathway and low concentrations may initiate non-canonical signals, or vice versa [34][35][36] . The canonical pathway activation is demonstrated by increasing levels of active β-catenin 37 , whereas increases in RhoA or CamKII 38 levels can represent activation of the non-canonical PCP and Ca 2+ pathways, respectively.
Analysis of β-catenin levels showed various effects, from no response to 2.3-fold increases after Wnt6 exposure. However, the trends between the donors were similar. Increases were detected immediately after treatment with low-Wnt6 CM, whereas use of med-and high-Wnt6 CM resulted in delayed β-catenin responses (Fig. 4A). We further investigated β-catenin activity, which can be measured by phosphorylation levels. Phosphorylation of β-catenin at tyrosine-142 has been found to release β-catenin from the membrane and a subsequent increase transcriptionally active β-catenin in the nucleus [39][40][41] . Low-, medium-, and high-Wnt6 CM increased Wnt6 regulates the non-canonical pathway. Just as β-catenin is associated with the membrane and is associated with the canonical pathway, RhoA is present at adherent junctions in cellular membranes and is associated with the non-canonical Wnt/planar cell polarity (PCP) pathway 42,43 . Figure 4C shows that the total RhoA levels generally decreased over time with all treatments of Wnt6 CM. However, analysis of phosphorylated RhoA at the serine 188 site (phospho-RhoA S188 ), which negatively regulates RhoA 38 , showed an immediate decrease in phospho-RhoA S188 levels upon treatment with medium-and high-Wnt6 CM by 54.18% and 76.70%, respectively (Fig. 4D). This suggests that RhoA activity was upregulated upon Wnt6 exposure. We also examined the expression of CamKII protein, which mediates the non-canonical Wnt/Ca 2+ pathway 44,45 . Upon Wnt6 exposure, CamKII expression increased immediately by 2.13-fold. Maximum levels were reached 30 or 60 min after treatment with Wnt6 CM in a dose dependent manner: a 3.37-fold increase was seen with low-Wnt6 CM treatment; a 3.88-fold increase, with medium-Wnt6 CM treatment; and a 5.33-fold increase, with high-Wnt6 CM treatment (Fig. 4E). Evaluation of phosphorylated CamKII levels, the activated form of the protein 45 , revealed the opposite trend in which low-Wnt6 CM treatment resulted in high phosphorylation levels, whereas medium-Wnt6 CM and high-Wnt6 CM resulted in 2.86-and 2.21-fold decreases, respectively www.nature.com/scientificreports/ (Fig. 4F). The differences detected immediately and 30 min after treatment suggested that divergent pathways may exist upon Wnt6 receptor binding.

Discussion
Both Wnt6 and its receptor Fzd7 are present in the limbus 25 , suggesting a potential role in the regulation of LSCs in their niche. However, the 3T3 feeder cells used to mimic a niche environment in LSC expansion system do not express Wnt6. To explore the specific role and mechanisms of action of Wnt6 in LSCs, we developed a system in which Wnt6 was supplemented by the Wnt6-expressing 3T3 feeder layer. Low Wnt6-expressing 3T3 cells supplement Wnt6 at a similar level as in the limbal tissue, whereas high Wnt6-expressing 3T3 cells generated significantly higher level of Wnt6 in the culture. We demonstrated that medium to high levels of Wnt6 are required to increase LSC proliferation while maintaining their undifferentiated phenotype, as confirmed by the increased percentage of small cells, the stable percentage of K14 + cells 46 , and the decreased percentage of K12 + cells 33 . The percentage of p63α bright cells was decreased in both the GFP-3T3 controls and Wnt6-3T3 cultures. It is possible that the GFP or the transduction protocol might affect the expression level of p63α in the LSCs, rather than affected by Wnt6 alone.  www.nature.com/scientificreports/ Wnt6 is expressed in the limbus in a gradient fashion as shown in Fig. 2. The basal limbal epithelial layer has the highest Wnt6 level compared to the wing and superficial layers. The critical roles of Wnt gradients for distinct intracellular responses have been reported for other cell types. For example, this gradient pattern in the limbus is similar to that reported in the intestine stem cell niche. High levels of Wnt ligands are retrieved at the bottom of the intestinal crypts, near the intestine stem cells, and function to maintain the stem cell characteristics and to promote their proliferation through Wnt/β-catenin activation 47,48 .
To further understand how Wnt6 could mechanistically promote stem/progenitor cells, we treated cultures with 3 levels of Wnt6 for short periods of time. The 3 concentrations of Wnt6 allowed us to examine the differential effect Wnt dosages can have on the downstream effectors of both canonical and non-canonical pathways, as demonstrated in other systems 34 . Interestingly, Wnt6 appears to activate both Wnt/β-catenin canonical and noncanonical pathways in LSCs. The level of the active form of phosphorylated β-catenin (Y142) increased immediately upon Wnt6 exposure at a dose dependent manner, confirming activation of Wnt/β-catenin canonical pathway in the cultivated LSCs followed by a trend of decline in the level of active phosphorylated β-catenin afterwards, whereas the opposite trend of continuous increase over time was observed in the control. Wnt6 also activates the non-canonical Wnt/Ca 2+ at low Wnt6 level and Wnt/PCP signaling at a dose dependent manner. Given that the changes in the levels of proteins participating in the non-canonical pathways occurred as early as 15 min after the addition of Wnt6 when the β-catenin canonical pathway activation started to decline, the activation of the non-canonical pathway might be also direct. Nalesso et al. reported that low concentrations of Wnt3a in chondrocytes trigger the non-canonical Wnt/Ca 2+ pathway and thus promote chondrocyte differentiation, whereas high concentrations of Wnt3a activate Wnt/β-catenin signaling and thus, promote proliferation of de-differentiated chondrocytes. It is possible that a single Wnt could simultaneously activate different pathways and that Wnt gradients are critical for fine stem cell regulation 49,50 . Like Wnt3a, here we demonstrated that Wnt6 can directly activate the Wnt/β-catenin canonical and noncanonical pathway effectors in LSCs. The inverse relationship of the protein levels of participants in the canonical and canonical pathways during the first hour of exposure to Wnt6 suggests that the Wnt/β-catenin canonical and noncanonical pathways activation was also time-modulated, raising the hypothesis that Wnt6 levels are finely controlled in the LSC niche, subsequently the timed activation of Wnt canonical and noncanonical pathways.
As suggested by Van Amerongen and Nusse, Wnt activation may function in integrated signaling pathways, instead of in independent ones 51 . The limbal stiffness has been shown to regulate LSCs proliferation and phenotype through interactions with several signaling pathways both in vitro and in vivo 52,53 . Gouveia et al. reported that inactivation of yes associated protein (YAP) at the basal layer, due to the softer properties of the limbus and the interactions with adhesion molecules, maintained high level of expression of ΔNp63, β-catenin, and ABCG2 53 . Subsequently, bone morphogenetic protein 4 was inhibited, preventing cell differentiation. Conversely, in the supra-basal and superficial layers, the influence of the limbal stiffness and adhesion molecules is decreased, leading to a progressive differentiation of the suprabasal cells, possibly via YAP activation. Similarly, we showed that Wnt6 is expressed in a gradient manner among the limbal layers. Overall, these results suggest that cross talks between Wnt/β-catenin signaling, and YAP-dependent mechanotransduction pathway likely work in an integrated fashion in the limbus to regulate LSC phenotype and proliferation. Moreover, crosstalk between canonical and non-canonical Wnt pathways has been reported in many cell types 49,54-56 ; a growing body of literature suggests that interrelationships between these pathways are required for fine tuning of Wnt signaling 57 . To further confirm the roles of Wnt6 in the human LSC niche, exploration of downstream genes activated by the canonical and non-canonical pathways is necessary in the future studies.
The findings from our study are important in the characterization of the functions of Wnt6 in the LSC niche but also from a pharmacologic point of view. Wnt is an attractive therapeutic target for stem cell therapy, but its regulation of the multiple processes and their complexity have long slowed the task of identifying a suitable activator. Understanding how downstream biological events are individually regulated by Wnt in the LSC niche could lead to a higher degree of target specificity. Broad Wnt small molecules are able to modulate LSCs differentiation 22,58 . The availability of molecules and drugs that can selectively interfere with CamKII, RhoA, or the canonical Wnt/β-catenin pathway offers the possibility of targeting individual outcomes of Wnt stimulation, while avoiding issues related to broad Wnt activation.
In conclusion, Wnt6 is an important factor that can promote the proliferation of LSCs and maintain their phenotype in the LSC niche. Wnt6 gradients have distinct intracellular effects, modulating both canonical and non-canonical pathways; these effects suggest that modulation of Wnt6 levels could optimize cultivation of LSCs.

Materials and methods
Generation of NIH-3T3 cells overexpressing Wnt6. To generate NIH-3T3 feeder layers that overexpressed Wnt6 (Wnt6-3T3) and control NIH-3T3 feeder layers that overexpressed GFP only (GFP-3T3), we obtained the Wnt6 coding sequence from Dr. Eric Wexler (University of California, Los Angeles) 59  Colony-forming efficiency and cell proliferation rate. CFE was measured at the end of culture by dividing the number of colonies by the number of seeded LECs. The LSC colonies were fixed with 4% paraformaldehyde (Sigma-Aldrich) for 10 min and stained with 0.5% rhodamine B (Sigma-Aldrich) for 15 min at room temperature. The proliferation rate was calculated by dividing the number of harvested cells by the number of seeded LECs, as previously described 58 .
Cell cultures and colony forming efficiency imaging. Images of cell cultures and CFE were taken with the digital inverted Keyence BZ-X710 fluorescence microscope (Keyence, Osaka, Japan) and the image capture system BZ-X viewer. Cell size and quantitation of expression of markers such as p63α (p63α bright cells) was performed by using the BZ-X analyzer software (version 1.3.0.3) with the macro hybrid cell-count function. By specifying the mask area, the software tracked information about multiple parameters such as the intensity of fluorescence signals in different channels, cell counts, and target area measurements 65,66 . For the cell size counting, phase contrast images of single LSCs placed in a hematocytometer were used.

Conditioned media and LSC treatment.
Conditioned media (CM) was obtained from 1 × 10 4 /cm 2 Wnt6-transduced NIH-3T3 cultured in DMEM supplemented with 10% BCS (Invitrogen) and 1% penicillin/ streptomycin (Invitrogen) for 4 days. Following collection, the CM underwent centrifugation at 1000×g for 10 min and was concentrated by using 30-kDa Amicon Ultra Centrifugal Filters (Sigma-Aldrich). Single-cell cultures of LSCs were washed once with 1 × phosphate-buffered saline (PBS) and incubated with 200 µL of low-, medium-, and high-Wnt6-CM for specific time intervals. The LSCs were then washed twice with 1 × PBS and harvested for analysis.
Western blot. Cells were lysed with ice cold RIPA buffer (ThermoFisher Scientific, Carlsbad, CA), and proteins in the lysate were denatured in Sample Loading Buffer (Li-cor, Lincoln, Nebraska) with 10% β-mercaptoethanol (Sigma Aldrich) for 5 min at 95 °C. Samples were loaded onto NuPAGE™ 4-12% Bis-Tris Protein Gels (ThermoFisher Scientific) and transferred onto Immobilon-FL PVDF membranes (Millipore, St. Louis, MO). Membranes were incubated for 1 h at room temperature in Odyssey blocking buffer and then overnight with the following primary antibodies: rabbit phospho-β-catenin (Abcam, ab27798; dilution ratio, 1:500); rabbit phospho-CamKII (Abcam, ab32678; dilution ratio, 1:1000), rabbit phospho-RhoA (Abcam, ab41435; dilution ratio, 1:1000); mouse active β-catenin (Millipore, 05-665; dilution ratio, 1:1000); mouse CamKII (Santa Cruz Biotechnologies, Dallas, TX, sc-5306; dilution ratio, 1:200); mouse RhoA (Abcam, ab54835; dilution ratio, 1:1000); or mouse GAPDH (Millipore, MAB374; dilution ratio, 1:500). After overnight incubation, the membranes were washed 3 times for 10 min with TBS-Tween (0.1%) (ThermoFisher Scientific). Membranes were incubated at room temperature for 1 h with the following secondary antibodies: IRDye 680LT Donkey antimouse IgG (Li-cor, 1:10,000) and IRDye 800CW Donkey anti-rabbit IgG (Li-cor, 1:10,000). The original output of the quantification of the fluorescent intensity from the bands of the Western blot raw images is provided in Supplemental Table 1. RNA extraction and qRT-PCR. Levels of Wnt6 mRNA were compared between human corneas, NIH-3T3 feeder cells, and low-, medium-, high-Wnt6-3T3 cells. Death-to-preservation time was < 6 h for all tissues used for RNA extraction. RNA extraction was performed as previously described 20,67 . Briefly, the cornea and limbal epithelia along with the immediate adjacent stroma were dissected from donor tissues and stored at − 80 °C until RNA extraction was performed. Homogenization of tissue samples was achieved by using a serrated homogenizer (Omni International, Marietta, GA). Total RNA was extracted (Qiagen RNeasy Mini Kit; Qiagen). The quantity and quality of total RNA were assessed by a spectrophotometer (NanoDrop 1000 Immunocytochemistry and quantitative analysis. Cultured colonies were incubated for 2 h in SHEM5 growth medium supplemented with 2.4 U/mL Dispase II (Roche) at 37 °C. Single cells were obtained with 0.25% trypsin and 1 mM EDTA (Gibco) for 7 min at 37 °C. Harvested cells underwent cytospinning onto slides by a cytocentrifuge (Cytofuge; ThermoFisher Scientific), and these slides were stored at − 80 °C until further use. Slides were fixed with 4% paraformaldehyde at room temperature for 15 min, washed 3 times with PBS, blocked and permeabilized with 1X PBS containing 1% bovine serum albumin (BSA) and 0.5% Triton X-100 (Sigma-Aldrich) for 30 min at room temperature. Slides were incubated with primary antibodies in 1X PBS containing 1% BSA and 0.1% Triton X-100 overnight at 4 °C in a moist chamber. Primary antibodies used were the following: anti-K12 (Santa Cruz Biotechnologies, sc-25722; dilution ratio, 1:100); anti-K14 (ThermoFisher Scientific, MS-115-R7); and anti-p63α (Cell Signaling Technologies, 4892; dilution ratio, 1:100). After incubation, cells were washed with 1X PBS, incubated with the following secondary antibodies: goat anti-mouse IgG conjugated to AlexaFluor 488 (ThermoFisher Scientific, A11029, 1:500) or goat anti-rabbit IgG conjugated to AlexaFluor 546 (ThermoFisher Scientific, A11035, 1:500) at 20 °C for 1 h. Nuclei were labeled with Hoechst 33342 (Invitrogen, 33342, 4 µg/mL) at 20 °C for 15 min and mounted in Fluoromount medium (Sigma-Aldrich). Images were taken with a Keyence BZ-X710 inverted microscope (Osaka, Japan), and characteristics such as fluorescence intensity, number of cells, and size were evaluated with the BZ-X analyzer software (version 1.3.0.3) as described above. mRNA in situ hybridization using RNAscope assays. The RNAscope V2 Fluorescent Assay (ACD Biosystems, Newark, CA) was performed according to the ACD Biosystems protocol for fresh-frozen tissue. Corneal sections from 3 donors were hybridized with the Wnt6 mRNA probe. Assays were performed in 2 independent experiments for each donor. To confirm the mRNA integrity of the tissues, positive control probes targeting human housekeeping genes Polr2a, PPIB, and HPRT were visualized (Supplemental Fig. 1). The probes were amplified according to the manufacturer's instructions and labeled with the fluorophore Opal 520 nm (Akoya Biosciences, FP1487001KT, 1:250), Opal 570 nm (Akoya Biosciences, FP1488001KT, 1:1500), and Opal 690 nm (Akoya Biosciences, SKU FP1497001KT, 1:1500). The Olympus FluoView FV1000 was used to visualize the fluorescence in situ hybridization signals. Quantitation of the dots among the limbal layers was performed with Imaris software and the surface and dots functions (V. 9.7.0, Imaris, Oxford Instruments, Oxon, UK). Statistical analysis. The Student's t test was used to analyze data from at least 4 independent experiments in which 4 different donor tissues were used for cell culture. The mean ± standard error of the mean (SEM) are reported in each bar graph. Statistical significance was established with P values < 0.05.

Data availability
All data and materials are available upon reasonable request.