Identification of BST2 as a conjunctival epithelial stem/progenitor cell marker

Summary The conjunctival epithelium consists of conjunctival epithelial cells and goblet cells derived from conjunctival epithelial stem/progenitor cells. However, the source of these cells is not well known because no specific markers for conjunctival epithelial stem/progenitor cells have been discovered. Therefore, to identify conjunctival epithelial stem/progenitor cell markers, we performed single-cell RNA sequencing of a conjunctival epithelial cell population derived from human-induced pluripotent stem cells (hiPSCs). The following conjunctival epithelial markers were identified: BST2, SLC2A3, AGR2, TMEM54, OLR1, and TRIM29. Notably, BST2 was strongly positive in the basal conjunctival epithelium, which is thought to be rich in stem/progenitor cells. Moreover, BST2 was able to sort conjunctival epithelial stem/progenitor cells from hiPSC-derived ocular surface epithelial cell populations. BST2-positive cells were highly proliferative and capable of successfully generating conjunctival epithelial sheets containing goblet cells. In conclusion, BST2 has been identified as a specific marker of conjunctival epithelial stem/progenitor cells.

Single-cell RNA-seq of hiPSC-derived conjunctival epithelial cells revealed BST2 BST2 strongly expressed in the basal conjunctival epithelium rich in stem cells BST2 can be used to sort conjunctival epithelial stem cells BST2-positive cells generated conjunctival epithelial sheet containing goblet cells

INTRODUCTION
The ocular surface comprised the conjunctival epithelium, corneal epithelium, lacrimal glands, meibomian glands, and tear film, all of which play essential roles in maintaining good vision. 1 The conjunctiva, which covers the sclera and the posterior surface of the eyelids, facilitates smooth eye movement and protects the ocular surface by regulating the immune response. 2 The conjunctival epithelium also contains goblet cells that produce mucin, an essential component of the tear film. 3 Despite substantial advances in corneal epithelium research, much remains unknown about the conjunctival epithelium. There is abundant evidence, for example, that corneal epithelial stem/progenitor cells exist at the limbus, with several limbal epithelial stem/progenitor cell markers previously reported. 4 From a clinical perspective, limbal stem cell insufficiency can be treated using a variety of regenerative therapies, including cultivated corneal epithelial sheets, oral mucosal epithelial sheets, and human-induced pluripotent stem cell (hiPSC)-derived corneal epithelial sheets. 5,6 However, the source of conjunctival epithelial stem/progenitor cells remains controversial and no specific markers for conjunctival epithelial stem/progenitor cells have yet been identified. 7 Moreover, although conjunctival epithelial stem/progenitor cells are believed to be abundant in the conjunctival fornix, a full appreciation of their localization and the molecular mechanisms that affect their behavior remain unclear. 8 It is also more difficult to obtain tissue from the conjunctival epithelium compared to the corneo-scleral limbal epithelium, which hinders research on the conjunctival epithelium, and relatively few regenerative therapies exist for the conjunctival epithelium because conjunctival epithelial sheets cannot currently be fabricated with high purity. 7 Recently, we generated hiPSC-derived self-formed ectodermal autonomous multi-zone (SEAM) organoids that mimic whole-eye development. 9,10 In each of four concentric cellular zones, cells had characteristics of cells that comprise different ocular component tissues such as the lens, neuro-retina, ocular surface epithelium, and lacrimal gland. 11 We also succeeded in isolating and differentiating conjunctival epithelial stem/ progenitor cells and corneal epithelial stem/progenitor cells, respectively, via different differentiation culture methods. Conjunctival epithelial stem/progenitor cells, for example, preferentially emerged in epidermal growth factor (EGF)-treated SEAMs, whereas corneal epithelial stem/progenitor cells predominated in keratinocyte growth factor (KGF)-treated SEAMs. 12 There are no known conjunctival epithelial  iScience Article stem/progenitor cell markers, especially cell surface markers, capable of separating conjunctival epithelial stem/progenitor cells from corneal epithelial stem/progenitor cells, and hiPSC-derived conjunctival epithelial sheets made using conventional SEAM cultivation methods occasionally contain some corneal epithelial cells, with a lack of goblet cells. Identifying a new conjunctival epithelial stem/progenitor cell marker would allow cell sorting and enable the generation of high-purity conjunctival epithelial sheets to contribute to the development of regenerative medicine applications and drug discovery as applied to the conjunctival epithelium.
Currently, single-cell RNA sequencing (scRNA-seq) has completely changed our understanding of diseases and associated biological processes. This technology has led to the discovery of new cell types, unique cell states, and novel tissue-specific genes, including stem/progenitor cell markers. 13 In a recent study, scRNAseq was used to identify limbal/corneal epithelial stem/progenitor cell markers in the human limbus. 14 However, few studies have employed scRNA-seq to research conjunctival epithelial stem/progenitor cell markers owing to the difficulty of obtaining human conjunctiva, especially the conjunctival fornix, which are predicted to be rich in conjunctival epithelial stem/progenitor cells. To overcome this problem, we performed scRNA-seq on hiPSC/SEAM-derived conjunctival epithelial stem/progenitor cell populations to identify new conjunctival epithelial stem/progenitor cell markers, especially cell surface markers, that could be used to isolate conjunctival epithelial stem/progenitor cells from corneal epithelial stem/progenitor cells, similarly analyzed by scRNA-seq. A comparison of the conjunctival and corneal epithelial stem/progenitor scRNA-seq data obtained from hiPSC/SEAM-derived cell populations identifies several novel conjunctival epithelial cell markers, including one stem/progenitor cell marker, BST2.

RESULTS
Single-cell RNA sequencing of ocular surface epithelial cells, conjunctival epithelial cells, and corneal epithelial cells derived from hiPSCs hiPSCs were cultivated for four weeks, during which time they spontaneously formed typical SEAMs with four identifiable concentric zones. Each SEAM exhibited four concentric cellular zones: cells in the innermost zone-1 were similar to the neuroectoderm; cells in zone-2, more peripherally, most closely resembled those of the optic cup and neural crest; cells in zone-3, more peripherally again, contained cells of the ocular surface ectoderm and lens; and cells in the outermost zone-4 had characteristics of the surface ectoderm, including skin. Cells in SEAM zones-3 and -4 at four weeks' cultivation were identified as a group of ocular surface epithelial cells (OEC) containing stem/progenitor cells. At this juncture, zone-1 and -2 cells were removed from the SEAMs by manual pipetting and EGF or KGF was added to the culture medium to induce the emergence of OEC of a conjunctival or corneal epithelial cell lineage, respectively (Figures 1A and S1A). After 12 weeks of cultivation, EGF-treated SEAMs (minus zones 1 and 2) were subjected to fluorescent-activated cell sorting (FACS) as described previously, 12 with the CD200-/CD104+/SSEA4weakly+ cell population containing stem/progenitor cells designated as conjunctival epithelial cells (CjEC) ( Figures 1A and S1B). KGF-treated SEAMs at 12 weeks were also subjected to FACS as described previously, 9,15 with the CD200-/CD104+/SSEA4+ cell population containing stem/progenitor cells designated as corneal epithelial cells (CEC) ( Figures 1A and S1C).
scRNA-seq analysis was performed and t-distributed stochastic neighbor embedding data for CjEC, CEC, and OEC were obtained ( Figures 1B, S1D, and S1G). We first considered CjEC, and Figure 1C shows feature plots of K13 (a mature conjunctival epithelial cell marker), K4 (a mature conjunctival epithelial cell marker), TP63 (an epithelial stem cell marker), and PAX6 (an ocular surface epithelial cell marker). Clusters of CjEC were then classified, with clusters 2, 3, 5, and 7 identified as immature conjunctival epithelial cells and clusters 6 and 8 as mature conjunctival epithelial cells. RNA velocity, a high-dimensional vector that can predict the future state of individual cells on a temporal scale, 16 was also used to investigate the developmental lineages of the immature conjunctival epithelial cell population. This determined that CjEC cluster 2 comprised the conjunctival epithelial stem/progenitor cell population ( Figure 1D). A further investigation iScience Article of the highly expressed markers in CjEC cluster 2 that were simultaneously absent or expressed at low levels in other clusters identified BST2, SLC2A3, AGR2, TMEM54, OLR1, and TRIM29 ( Figure 1E). Accordingly, these are proposed as novel conjunctival epithelial cell markers, all of which are cell surface markers, which could be useful for FACS analysis and cell sorting.
We similarly analyzed the feature plots of K12 (a mature corneal epithelial cell marker), K3 (a mature corneal epithelial cell marker), TP63, and PAX6 expression in CEC ( Figure S1E), and classified clusters 0, 1, 2, 9, and 10 as immature corneal epithelial cells and clusters 3, 4, 5, 6, 7, and 8 as mature corneal epithelial cells. An analysis of the RNA velocity further determined that CEC cluster 0 comprised the corneal epithelial stem/ progenitor cell population ( Figure S1F). An analysis of the feature plots of TP63 and PAX6 expression in OEC determined that cluster 1 comprised the ocular surface epithelial stem/progenitor cell population ( Figure S1H).

Immunolocalization of novel conjunctival epithelial cell markers and stem/progenitor cell marker
Tissue sections of human conjunctiva, limbus, and cornea were immunostained with antibodies to each of the proposed conjunctival epithelial cell markers, BST2, SLC2A3, AGR2, TMEM54, OLR1, and TRIM29, and all were found to strongly stain the human bulbar conjunctival epithelium ( Figure 2A). Of these, BST2, AGR2, and TMEM54 were not identified in the limbal or corneal epithelium. SLC2A3 was identified in the conjunctival and limbal epithelium but not in the corneal epithelium, whereas OLR1 and TRIM29 stained the conjunctival, limbal, and corneal epithelium. Of the novel conjunctival epithelial markers, BST2 was notable for its positive labeling of the basal conjunctival epithelium. As we could not obtain human fornix conjunctiva for this study, we analyzed tissue from a cynomolgus monkey and found that basal regions of both the bulbar and fornix conjunctival epithelia were positive for BST2, especially the basal region of the fornix conjunctiva on the bulbar side ( Figure 2B). All other markers also labeled monkey fornix conjunctiva ( Figure S2). A gene expression analysis confirmed that BST2 was upregulated in the human bulbar conjunctival epithelium, compared to the limbal or corneal epithelia ( Figure 2C). Of the novel conjunctival epithelial markers, BST2, AGR2, and TMEM54 are conjunctival epithelium-specific. Among them, only BST2 exclusively immunostained the basal layer of the conjunctival epithelium, suggesting it to be the most representative marker of conjunctival epithelial stem/progenitor cells 7 (Figure 2A). AGR2 did not stain the basal layer, while TMEM54 stained not only the basal layer but also cells in the outer conjunctival layers, so both differed from the typical immunostaining patterns of conjunctival epithelial stem/progenitor cells. Accordingly, we excluded AGR2 and TMEM54 from our ongoing studies, and shifted our focus to BST2.
CFA disclosed that P1 and P2 cells had a high proliferative potential, whereas P3 and P4 cells had no proliferative potential. Cells from P1, moreover, were larger and formed more colonies than cells from P2 (Figure 3C), indicating that the colony-forming efficiency (CFE) of P1 was significantly higher than that of P2 ( Figure 3D). These findings indicated that P1 or P2 cells (i.e. CD104+ cells) contain conjunctival epithelial stem/progenitor cells, whereas P3 and P4 cells (i.e. CD104-) do not. Overall, P1 cells (i.e. those that are CD104+/BST2+) had the highest proliferative potential among P1-P4 cells. Although this result was obtained using the YZWJs524 hiPSC line, we have also confirmed, using line 201B7, that P1 cells had the highest proliferative potential among P1-P4 cells ( Figures S4A-S4C). iScience Article cell markers) were elevated in P2. We speculated that P1 contained conjunctival epithelial stem/progenitor cells, and that P3 contained differentiated conjunctival epithelial cells because MUC5AC (a goblet cell marker) and MUC4 (a membrane mucin marker) were elevated in P3 ( Figure 3E).
Fabrication of epithelial sheets from hiPSC-derived BST2+ cells P1 and P2 fractions, with their high CFE, were cultivated in maturation medium and compared by en face observation and gene expression analysis ( Figure 4A). After four weeks, homogeneous epithelial cell sheets had formed ( Figure 4B), with those derived from P1 revealing the presence of MUC5AC (a goblet cell marker) and K13 (a conjunctival epithelial cell marker) ( Figure 4C). Immunostaining of the epithelial sheet derived from P2, on the other hand, revealed the presence of K12 (a corneal epithelial cell marker) ( Figure 4C). Gene expression analysis of P1 and P2 before maturation (i.e. FACS group) indicated that the expression levels of conjunctival differentiation markers such as K13/K4, mucin markers such as MU-C5AC/MUC4, and corneal differentiation markers such as K12/K3 were not elevated ( Figure 4D). After maturation (i.e. Sheet group), however, conjunctival differentiation markers (K13 and K4 as conjunctival epithelial cell markers), MUC5AC (a goblet cell marker), K7 (a unicellular gland marker), and MUC4 (a membrane mucin marker) were highly expressed in the sheet derived from P1. Corneal differentiation markers (K12 and K3 as corneal epithelial cell markers) were highly expressed in the sheets derived from P2. Ocular surface differentiation markers (PAX6 and p63 as ocular surface epithelial cell markers) were expressed in sheets from both P1 and P2 ( Figure 4E). Although this result was obtained from using the YZWJs524 hiPSC line, we have also confirmed, using line 201B7, that the expression of K13 (a conjunctival epithelial cell marker) and MUC5AC (a goblet cell marker) was detected in the sheets derived from P1 ( Figures S4D-S4F).

Immunostaining of epithelial sheets fabricated from hiPSC-derived BST2+ cells
The immunostaining characteristics of P1-derived and P2-derived epithelial sheets were compared (Figure 5A). This showed that the goblet cell marker MUC5AC, the conjunctival epithelial cell markers K13 and K4, the membrane-bound mucin MUC4, and the unicellular gland marker K7, all of which are characteristic of the conjunctival epithelium, were present in the P1-derived epithelial sheets. In contrast, positive staining for the corneal epithelial cell markers K12 and K3 was seen in P2-derived epithelial sheets. Both P1and P2-derived sheets were immunopositive for the ocular surface epithelial cell markers PAX6 and p63 ( Figure 5A). PAS-positive cells were also identified in the P1-derived epithelial sheets, indicating the presence of mucin, with no PAS-positive cells observed in the P2-derived epithelial sheets ( Figure 5B).

DISCUSSION
Numerous markers for corneal epithelial cells exist; however, conjunctival epithelial cell markers have seldom been reported and no cell surface marker has been discovered that can distinguish conjunctival epithelial stem/progenitor cells from corneal epithelial stem/progenitor cells, probably because the respective cell types are similar in nature. Now, however, we report six novel putative conjunctival epithelial cell markers-BST2, SLC2A3, AGR2, TMEM54, OLR1, and TRIM29-for example, one of which, AGR2, is known to be associated with intestinal and pulmonary goblet cells 17,18 where it helps regulate goblet cell differentiation and protein glycosylation. 19 AGR2 has not been reported previously in the conjunctival epithelium, a tissue that also contains mucin-producing goblet cells. Although we did not perform further experiments with these conjunctival epithelial cell markers with the exception of BST2, we believe that all these markers will help advance research on the human conjunctiva.
Further analysis ascertained that cultured conjunctival epithelial sheets containing goblet cells could be fabricated from conjunctival epithelial stem/progenitor cells that had been obtained from hiPSC-derived SEAM eye-like organoids and sorted by FACS using CD104, CD200, and one of the new conjunctival epithelial cell markers, BST2. BST2, also referred to as HM1.24, tetherin, or CD317, is a lipid raft-associated type II transmembrane glycoprotein that localizes on the cell surface. 20 It is expressed to varying degrees in iScience Article all organs examined so far and has been detected in many specialized cell types, including pneumocytes, hepatocytes, the ducts of major salivary glands, kidney, pancreas, various epithelia, Paneth cells, Leydig cells, monocytes, bone marrow stromal cells, plasma cells, and the vascular endothelium. 21 BST2 is reported to be highly expressed in various tumors and is involved in the proliferative potential of cancers such as esophageal, gastric, colorectal, and breast cancer. 22,23 Specifically, BST2 expression was upregulated in these tumor tissues compared to adjacent non-tumorous tissues. Here, we performed CFA and confirmed the higher proliferative potential of hiPSC/SEAM-derived BST2+ cells compared to BST2À cells, implying that BST2+ cells are present in the stem/progenitor cell population of the conjunctival epithelium. BST2 was strongly positive on basal regions of the fornix conjunctiva; this agrees with reports of iScience Article conjunctival epithelial stem/progenitor cells that were abundant in the fornix conjunctiva. 7,8 We also discovered that BST2 is more abundant on the bulbar side of the fornix conjunctiva than on the palpebral side of the fornix conjunctiva. We therefore hypothesize that conjunctival epithelial stem/progenitor cells are also more abundant in the bulbar side of the fornix conjunctiva.
BST2 is also an innate immune protein that responds to enveloped viruses in mammals 24 and plays an important role in the defense of the host cell by preventing the release and spread of various enveloped viruses from the infected cells. We speculated that the difference in BST2 distribution between the corneal epithelium and conjunctival epithelium might be because of the differences in the immune responses of the respective epithelia. For example, the conjunctiva contains organized lymphoid tissue comprising adjacent lymphocytes, subepithelial lymphoid follicles, intraepithelial lymphatics, and blood vessels. This conjunctiva-associated lymphoid tissue, which does not exist in the cornea, forms an immunological interface between the ocular surface and external environment, 25 and protects against infection by enveloped viruses such as adenovirus. 26 While most cases of conjunctivitis caused by an adenovirus are self-limiting, corneal inflammation caused by adenoviruses can continue or recur for a long term after infection, leading to reduced vision. 27 Although the immune response of the ocular surface is not completely understood, differences in the respective responses of the cornea and conjunctiva to viral infection could be potentially related to the different expression of BST2 in the epithelia of the two tissues.
Previously, we succeeded in isolating conjunctival epithelial cells, including stem/progenitor cells, from hiPSC-derived SEAMs by conducting FACS with specific cell surface markers, SSEA4, CD104, and CD200. 12 However, no available cell surface marker could accurately separate conjunctival epithelial stem/progenitor cells from corneal epithelial stem/progenitor cells. In the previous report, we used SSEA4, which was weakly expressed in the native conjunctival epithelium and strongly expressed in the native corneal epithelium, as a negative marker. This approach 12 meant that some corneal epithelial stem/progenitor cells would invariably be mixed in with the conjunctival epithelial stem/progenitor cells.
The new approach using BST2 is able to accurately separate conjunctival epithelial stem/progenitor cells from corneal epithelial stem/progenitor cells to enable the generation of a hiPSC/SEAM-derived conjunctival epithelial sheet with high purity. This represents an important advance for our fundamental understanding of conjunctival epithelial cell biology and has applications for regenerative medicine-based treatments of ocular surface diseases, as well as drug discovery research targeting the conjunctival epithelium.

Limitations of the study
The present study demonstrates that BST2-positive cells give rise to both conjunctival epithelial cells and goblet cells. Conjunctival epithelial stem/progenitor cells are generally believed to possess the capability of multipotent differentiation into both cell types; 7 however, the results we provide here do not provide direct evidence to definitively confirm the multipotency of BST2-positive cells.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:   iScience Article and ''FindClusters'' functions for use in t-distributed stochastic neighbor embedding (t-SNE). Differentially expressed genes were identified for each cluster using the Wilcoxon rank sum test implemented in Seurat. The RNA velocity analyses were performed based on Cellranger bam files. The spliced and unspliced UMIs for each gene in each cell were calculated using Velocyte (10x Genomics).

QUANTIFICATION AND STATISTICAL ANALYSIS
All data are presented as the mean G the standard deviation; n represents the number of biological replicates. The Wilcoxon rank-sum test (two-sided) was performed for two-group comparisons. Statistical significance was set to p < 0.05. All statistical analyses were performed using JMP Pro software (v17.0).