Low fucosylation defines the glycocalyx of progenitor cells and melanocytes in the human limbal stem cell niche

Abstract

It is widely recognized that the glycocalyx has significant implications in regulating the self-renewal and differentiation of adult stem cells; however, its composition remains poorly understood. Here, we show that the fucose-binding Aleuria aurantia lectin (AAL) binds differentially to basal cells in the stratified epithelium of the human limbus, hair follicle epithelium, and meibomian gland duct. Using fluorescence-activated cell sorting in combination with single-cell transcriptomics, we find that most epithelial progenitor cells and melanocytes in the limbus display low AAL staining (AAL^low) on their cell surface, an attribute that is gradually lost in epithelial cells as they differentiate into mature corneal cells. AAL^low epithelial cells were enriched in putative limbal stem cell markers and displayed high clonogenic capacity. Further analyses revealed that AAL^low epithelial cells had reduced expression of GDP-mannose-4,6-dehydratase, an enzyme catalyzing the first and limiting step in the de novo biosynthesis of GDP-fucose, and that inhibition of fucosylation using a small-molecule fucose analog stimulated the proliferative potential of limbal epithelial cells ex vivo. These results provide crucial insights into the distinctive composition of the glycocalyx in adult stem cells and underscore the significance of fucose modulation in the therapeutic regeneration of the human limbal stem cell niche.

Keywords: GDP-mannose-4,6-dehydratase; fucosylation; glycocalyx; human limbus; stem cell niche.

Figure 1

AAL binds differentially to basal cells in human stratified epithelia (A) Schematic representation of adult stem cells located in the basal epithelial layer of the human limbus. (B) Simplified diagram showing the post-translational modification of proteins by fucosyltransferases (FUT1-11, POFUT1-2) and the interactions between AAL and fucose residues. (C) Representative image showing AAL histochemistry (green) and DAPI (blue) staining in the human cornea (top) of five human donors. Bottom images show magnified images from the same corneal section. Arrows indicate pockets of cells with negligible AAL staining. AAL binding was inhibited in the presence of competing L-fucose. Scale bar, 100 μm. (D) Quantification of AAL staining across the basal layer of the human limbal and central corneal epithelia. Two independent sections were analyzed from each donor. The box-and-whisker plots show the mean fluorescence intensity, MFI (n = 5 human donors; ^∗∗p < 0.01, paired t test). (E) Immunohistochemical analysis of AAL (green) co-stained with K15 or ABCG2 (red) in the human limbus. Scale bars, 100 μm. (F) Immunohistochemical analysis of AAL (green) co-stained with K15 (red) in human hair follicle and meibomian gland specimens. Scale bars, 100 μm.

Figure 2

Fucosylation defines different cell subpopulations in the human limbus (A) Representative FACS histogram showing AAL binding to cell suspensions from the limbus and central cornea. (B) Box-and-whisker plots showing the percentage of cells displaying low AAL staining (AAL^low) in the limbus and central cornea (n = 3 human donors; ^∗∗∗p < 0.001, unpaired t test). (C) Representative fluorescence-activated cell sorting of limbal cells. All events are plotted in the graph. The red circles define two populations of limbal cells with low and high levels of AAL binding (AAL^low and AAL^high) collected for scRNA-seq experiments. Propidium iodide (PI) was used to exclude dead cells. (D) UMAP visualization of AAL^low and AAL^high cell fractions pooled from limbal tissue of three human donors. The dot plot depicts expression levels of canonical marker genes together with the percentage of cells expressing the marker. (E) Cell types identified in the human limbus with UMAP projections of scRNA-seq data. LPC, limbal progenitor cell; B, basal; SB, suprabasal; Conj, conjunctival. (F) Proportions of cell types identified in the AAL^low and AAL^high cell fractions.

Figure 3

Melanocytes and limbal progenitor epithelial cells display low surface fucosylation (A) Top: representative micrographs of the human limbus labeled with antibodies to MelanA or isotype control (green). Nuclear DNA was stained with DAPI (blue). Bottom: phase-contrast images of melanocytes and epithelial cells cultured from human limbus. Images to the right show AAL histochemistry (green) and DAPI (blue) staining of non-permeabilized melanocytes and limbal epithelial cells. AAL binding was inhibited in the presence of competing L-fucose (Fuc). Scale bars, 100 μm. (B) Normalized abundance of epithelial cells expressing limbal progenitor cell markers in AAL^low and AAL^high cell fractions. Percentages are shown at the top of the scatterplot. Each dot in the scatterplot represents an individual cell. KRT12 is a marker of cells undergoing corneal epithelial differentiation. (C) Representative phase-contrast images showing colony-forming ability of AAL^low and AAL^high cell subpopulations. Unstained cells were processed as control. The box-and-whisker plots show the colony-forming efficiency (CFE) of AAL^low and AAL^high cells (n = 5–6 human donors; ^∗∗p < 0.01, unpaired t test). (D) Cumulative population doublings of AAL^low and AAL^high cell fractions obtained from 4 human donors.

Figure 4

The GDP-fucose de novo pathway is hampered in limbal progenitor epithelial cells (A) Left: schematic overview of the biosynthesis of fucosylated proteins through de novo and salvage pathways. Relevant enzymes are labeled in red. Right, top: bubble heatmap showing fucosylation pathway genes in the AAL^low and AAL^high cell fractions. The size and color of the dot indicate the fraction of expressing cells and averaged scaled expression level for each gene, respectively. Right, bottom: bar graphs depicting genes whose expression is significantly altered between the AAL^low and AAL^high cell fractions (n = 3 human donors; bar graphs indicate mean ± SEM; ^∗p < 0.05, ^∗∗∗∗p < 0.0001, Mann-Whitney test). (B) Representative image of the human cornea co-stained with antibodies to GMDS (green) and K15 (red). Bottom images show magnified images from the same corneal section. Nuclear DNA was stained with DAPI (blue). Dashed lines denote the epithelial-stromal junction. Scale bar, 100 μm. (C) Immunoblot analysis of human limbal epithelial cells grown for 3 and 10 days. The box-and-whisker plots show the densitometry analysis of the protein band intensity, expressed as the ratio of the target protein/housekeeping protein (n = 4–5 human donors; ^∗p < 0.05, ^∗∗p < 0.01, Mann-Whitney test). IVL, involucrin. (D) Representative lectin blot of human limbal epithelial cells treated with 2F-peracetyl-fucose (2FF) for 14 days. DMSO was used as vehicle control. Right: box-and-whisker plots of qPCR data showing the relative expression of C/EBPdelta and IVL after 7 and 14 days, respectively (n = 3–6 human donors; ^∗p < 0.05, paired t test). (E) Cumulative population doublings of human limbal epithelial cells treated with 2FF or DMSO (n = 4 human donors). Bottom: box-and-whisker plots showing the cumulative population doublings on senescence (n = 4 human donors; ^∗p < 0.05, unpaired t test).