Time-resolved fate mapping identifies the intestinal upper crypt zone as an origin of Lgr5+ crypt base columnar cells

Abstract

In the prevailing model, Lgr5+ cells are the only intestinal stem cells (ISCs) that sustain homeostatic epithelial regeneration by upward migration of progeny through elusive upper crypt transit-amplifying (TA) intermediates. Here, we identify a proliferative upper crypt population marked by Fgfbp1, in the location of putative TA cells, that is transcriptionally distinct from Lgr5+ cells. Using a kinetic reporter for time-resolved fate mapping and Fgfbp1-CreER^T2 lineage tracing, we establish that Fgfbp1+ cells are multi-potent and give rise to Lgr5+ cells, consistent with their ISC function. Fgfbp1+ cells also sustain epithelial regeneration following Lgr5+ cell depletion. We demonstrate that FGFBP1, produced by the upper crypt cells, is an essential factor for crypt proliferation and epithelial homeostasis. Our findings support a model in which tissue regeneration originates from upper crypt Fgfbp1+ cells that generate progeny propagating bi-directionally along the crypt-villus axis and serve as a source of Lgr5+ cells in the crypt base.

Keywords: Fgfbp1; Lgr5; crypt base columnar cells; intestinal epithelium; intestinal stem cells; intestinal upper crypt; lineage hierarchy; time-resolved fate mapping; tissue regeneration; transit-amplifying cells.

Figure 1.. Single-cell analysis identifies Fgfbp1 as a marker of a proliferative upper crypt zone population that is transcriptionally and spatially distinct from Lgr5+ CBC cells.

(A) Left: co-embedded uniform manifold approximation and projection (UMAP) of 13,095 single cells consisting of 11,218 FACS-sorted Lgr5–GFP+ (Lgr5-GFP^High and Lgr5-GFP^Low) and 1,877 Lgr5–GFP- cells (GEO: GSE92865) following Wnt/R-spondin modulation, colored by cluster. Right: UMAP highlighting in red either the Lgr5–GFP- control cells or Lgr5–GFP+ cells. (B) DotPlot visualization of selected differentially expressed genes of the annotated clusters in (A), highlighting the putative transit-amplifying (TA) cluster. (C) Density plots showing expression of Lgr5, Ascl2, TA cell marker Fgfbp1, and the predicted cell cycle distribution of cells. (D) Distribution of cell types across individual treatment conditions (highlighted in red). (E) Crypt localization of Fgfbp1 and Lgr5 expression by in situ hybridization (ISH) combined with Ki67 immunostaining in WT murine intestine (low magnification view). Scale bar, 250 μm. (F) Top: spatial segregation of Fgfbp1 and Lgr5 transcripts along the WT murine intestine. Bottom: higher magnification views of boxed regions (TA, TA zone; CBC, CBC zone). Scale bars, 50 μm. (G) Spatially segregated expression of Fgfbp1 and Lgr5-GFP confirmed by Fgfbp1 ISH combined with Lgr5-GFP and Ki67 immunostaining in Lgr5-DTR-GFP intestine. Scale bar, 50 μm. (H) Positional distribution of crypt cells with detectable Fgfbp1 expression along the intestine length demonstrates an upper crypt expression domain, as determined by ISH. The center-most nucleus at the very crypt base was assigned as position “0”. Data represented as mean with SD, n =3 mice. (I) Quantitative multiplex mRNA single-molecule fluorescence ISH (smFISH) analysis of jejunal crypts highlights two distinct domains, an upper crypt (UC) Fgfbp1+ zone and an Lgr5+ CBC zone. Data represent mean mRNA integrated density with SEM for Lgr5 and Fgfbp1 transcripts acquired from >50 individual crypts, n = 2 mice. (J) Projection of Yan et al. single-cell clusters dataset (GEO: GSE92865) onto UMAP coordinates of Haber et al. (GEO: GSE92332). (K) RNA-velocity analysis of Haber et al. (GEO: GSE92332). Velocity field arrows inferred using scVelo stochastic model and projected onto the UMAP space indicate predicted lineage trajectories.

Figure 2.. Fgfbp1-TimeR allele reveals that the Fgfbp1+ upper crypt population exhibits multi-potency and gives rise to the Lgr5+ CBC cells during homeostasis.

(A) Targeting strategy for the dual-fluorescence Fgfbp1-TimeR knock-in allele. ** indicates destabilized dsRed; SP, signal peptide. (B) Spatial restriction of labile dsRed signal to the upper crypt zone and relatively broad distribution of long-lived mTagBFP2 along the crypt-villus axis in Fgfbp1-TimeR intestine. Jejunum, scale bars, 50 μm. (C) dsRed versus mTagBFP2 distribution along the Fgfbp1-TimeR intestine. Left: swiss-roll of small intestine, low magnification. Scale bars, 1 mm. Right: higher magnification views of crypts from the proximal (a) and distal (b) intestine, scale bars, 50 μm. (D) Left: co-localization of mTagBFP2 with enterocyte (FABP1), goblet (MUC2), enteroendocrine (CHGA), and Paneth cell (LYZ) markers in Fgfbp1-TimeR intestine. Jejunum, scale bar, 25 μm. Top right, representative examples for LYZ+/mTagBFP2+ (solid lines, white arrowhead) and LYZ+/mTagBFP2- (dashed lines) Paneth cells. Jejunum, scale bar, 25 μm. Bottom right: quantification of mTagBFP2+ signal among LYZ+ Paneth cells. Data represented as mean with SD, n = 3 mice. (E) Overlapping expression of mTagBFP2 and Lgr5-GFP, but not dsRed and Lgr5-GFP, is observed in Fgfbp1-TimeR; Lgr5-DTR-GFP intestinal crypts. Jejunum, scale bar, 50 μm. (F) Extensive co-expression of mTagBFP2 and Lgr5-GFP is observed throughout the length of the Fgfbp1-TimeR; Lgr5-DTR-GFP intestine. mTagBFP2 is pseudo-colored in red to visualize overlap with Lgr5-GFP (yellow). Scale bars, 25 μm. (G) Left: flow cytometry analysis of EPCAM+/Lgr5-GFP+ cells from Fgfbp1-TimeR; Lgr5-DTR-GFP jejunum. Right: mTagBFP2+ signal is observed in both Lgr5-GFP^High and Lgr5-GFP^Low epithelial cells. (H) Quantitation of mTagBFP2 and dsRed distribution among Lgr5-GFP+ cells in Fgfbp1-TimeR; Lgr5-DTR-GFP intestine by IF. Extensive overlap between Lgr5-GFP/mTagBFP2 versus minimal overlap between Lgr5-GFP/dsRed are observed. The Lgr5-GFP/dsRed overlap is predominantly restricted to the +4/+5 cell positions. Data represented as mean with S.D., n=3 mice.

Figure 3.. Fgfbp1-CreER^T2 lineage tracing reveals rapid regeneration of the Lgr5+ CBC compartment from the upper crypt during homeostasis.

(A) Targeting strategy for the Fgfbp1-CreER^T2 knockin allele. SP, signal peptide. (B) Labeling of tdTomato+ cells in the upper crypt (UC), but not the crypt base (CBC), 18 h post tamoxifen administration in Fgfbp1-CreER^T2; Rosa26-tdTomato mice. Jejunum, scale bars, 50 μm. (C) Fgfbp1-CreER^T2; Rosa26-tdTomato lineage tracing time-course demonstrates rapid reconstitution of villus cells from the Fgfbp1+ upper crypt zone. Jejunum, scale bars, 50 μm. (D) Distribution of tdTomato+ cells along the crypt-villus axis over time in Fgfbp1-CreER^T2; Rosa26-tdTomato mice post tamofixen administration. Data from >50 crypts/region in n = 3 mice. (E) Fgfbp1-CreER^T2; Rosa26-tdTomato lineage tracing shows rapid Lgr5+ CBC cell replacement by the Fgfbp1+ upper crypt zone over the course of 14 days. Infiltrating tdTomato+ crypt base cells (white arrowheads) are observed by day 4, followed by an explosion of labeling events in CBC positions between days 7 and 14 post tamoxifen. Scale bars, 50 μm. (F) Lgr5+ CBC cells are found within tdTomato+ traces as demonstrated by co-immunostaining of tdTomato with Lgr5-GFP (top), OLFM4 (center), and LYZ (bottom) in the Fgfbp1-CreER^T2; Rosa-tdTomato; Lgr5-DTR-GFP jejunum, day 7 post tamoxifen. Scale bars, 50 μm. (G) Both absorptive (FABP1⁺) and secretory (CHGA⁺, MUC2⁺, and LYZ⁺) cells are found by co-immunostaining within Fgfbp1-CreER^T2; Rosa26-tdTomato lineage traces. Day 21 post tamoxifen, jejunum, scale bars, 20 μm. (H) Both absorptive (ACE2⁺) and secretory (MUC2+) cells localize within a Confetti-RFP⁺ (Cf-RFP) clonal trace in Fgfbp1-CreER^T2; Rosa26-Confetti jejunum, consistent with multi-potency at the clonal level. Jejunum, day 21 post tamoxifen. Dashed lines outline Cf-RFP⁺/ACE2⁺ enterocyte, * indicates Cf-RFP⁺/MUC2⁺ goblet cell. Scale bars, 25 μm. (I) Extensive tdTomato+ lineage tracing is observed throughout the Fgfbp1-CreER^T2; Rosa26-tdTomato intestine at day 4 post tamoxifen. Boxes note regions shown at higher magnification. Scale bar for low and high magnification = 2.5 mm and 50 μm, respectively. (J) Persistence of extensive tdTomato+ lineage traces in Fgfbp1-CreER^T2; Rosa26-tdTomato intestine at 13 months post tamoxifen confirms long-term self-renewal of the Fgfbp1+ cell population. Boxes note regions shown at higher magnification. Scale bars for low and high magnification, 2.5 mm and 50 μm, respectively.

Figure 4.. Fgfbp1+ cells persist to repopulate crypt bases upon targeted ablation of Lgr5+ CBC cells and exhibit differential responses to Rspo modulation.

(A) Top: experimental schema for DT-mediated ablation of Lgr5-GFP+ cells in Fgfbp1-TimeR; Lgr5-DTR-GFP mice. Bottom: flow cytometry confirms complete Lgr5-GFP+ cell ablation post DT administration. (B) dsRed+ Fgfbp1+ cells persist unperturbed despite complete DT-mediated ablation of Lgr5+ CBC cells. The Fgfbp1 progeny (mTagBFP2+/dsRed-) repopulate crypt bases upon CBC cell ablation. Duodenum, scale bars, 50 μm. (C) Fgfbp1+ cells support regeneration throughout the crypt-villus axis in the absence of Lgr5+ CBC cells following DT-mediated ablation, as evidenced by extensive mTagBFP2 signal throughout the Fgfbp1-TimeR; Lgr5-GFP-DTR tissue. Duodenum, harvested at 72h per experimental schema in (A). Scale bars for low and high magnification views, 250 μm and 50 μm, respectively. (D) H&E and Lgr5 ISH validate effects of systemic R-spondin modulation via adenoviral-mediated hepatic transduction and systemic overexpression of Lgr5-ECD (Ad Lgr5-ECD) for loss of function and Rspo1 (Ad Rspo1) for gain of function in Fgfbp1-TimeR intestine, day 4 post treatment. Paneth cell and Lgr5+ cell loss are observed upon Lgr5-ECD treatment, whereas crypt enlargement and Lgr5+ cell expansion are seen upon Rspo1 augmentation. Duodenum, scale bars, 50 μm. (E) dsRed+ Fgfbp1+ cells persist and expand down into the crypt bases upon Rspo inhibition and subsequent loss of Lgr5+ CBC cells by Ad Lgr5-ECD. dsRed+ Fgfbp1+ cells expand to occupy the enlarged crypts upon Rspo augmentation by Ad Rspo1. Co-immunostaining of Ki67, dsRed and mTagBFP2 in Fgfbp1-TimeR duodenum, day 4 post adenovirus, scale bars, 50 μm. (F) Positional distribution of dsRed+ cells along the crypt-villus axis upon Rspo signaling modulation, highlighting changes in crypt occupancy of Fgfbp1+ cells. Duodenum, day 4 post adenovirus. Data represented as mean with SD, n = 3 mice. (G) Fgfbp1+ cell progeny (mTagBFP2+) repopulate the entire epithelium under conditions of Rspo modulation, including Rspo inhibition and Lgr5+ CBC cell depletion by Lgr5-ECD, as shown by mTagBFP2 immunostaining. Fgfbp1-TimeR duodenum, day 4 post adenovirus, scale bars, 50 μm. (H) Rspo1 overexpression induces expansion of both Fgfbp1+ and Lgr5+ cells, and Fgfbp1+ cell progeny reconstitute the tissue as shown by co-immunostaining of Lgr5-GFP, dsRed and mTagBFP2 in Fgfbp1-TimeR; Lgr5-DTR-GFP mouse duodenum, day 4 post treatment. Scale bars, low magnification, 250 μm and high magnification, 100 μm. (I) Lineage tracing of Lgr5+ versus Fgfbp1+ cells upon Rspo1 overexpression reveals striking functional differences. Top: Lgr5+ CBC cells and all their progeny remain confined to the bases of the expanded crypts in Lgr5-GFP-IRES-CreER^T2; Rosa26-tdTomato intestine. Bottom: Fgfbp1+ cells give rise to the expanded villi and to the expanded crypts, including the expanded Lgr5+ CBC compartment in Fgfbp1-CreER^T2; Rosa26-tdTomato; Lgr5-GFP-DTR intestine. Lgr5-GFP and tdTomato co-immunostaining of duodenum day 4 post Ad Rspo1 and tamoxifen, scale bars, 50 μm.

Figure 5.. Fgfbp1 is essential for intestinal epithelial homeostasis.

(A) Experimental schema for Vil-CreER^T2-driven Fgfbp1 conditional knockout (cKO) in the adult murine intestinal epithelium. (B) Vil-CreER^T2-driven Fgfbp1 cKO results in grossly shortened intestine, Scale bar, 5 cm. (C) Confirmation of knock-out by anti-FLAG immunostaining of control (Fgfbp1^f/f) and Fgfbp1 cKO (Vil-CreER^T2; Fgfbp1^f/f) jejunum. Scale bars, 50 μm. (D) Disrupted epithelial homeostasis and histological distortion is observed by H&E staining in Fgfbp1 cKO jejunum, Scale bars, 50 μm. (E) Architectural disruption of crypts and villi in Fgfbp1 cKO intestine. Villi are shortened and filled with cleaved caspase 3+ (CC3+) apoptotic cells (top). Crypt loss is observed, and remnant crypts are filled with nests of LYZ+ Paneth cells without intervening CBC cells (bottom). CC3 and LYZ immunostaining of control (Fgfbp1^f/f) versus Fgfbp1 cKO (Vil-CreER^T2; Fgfbp1^f/f) jejunum, scale bars, 50 μm. (F) Organoid culture shows impaired epithelial organoid generation in Fgfbp1 cKO versus control Fgfbp1^f/f epithelium. Day 4 of culture, jejunum, scale bars, 1 mm. (G) Histological analysis reveals marked crypt loss by H&E staining in Fgfbp1 cKO versus control intestine. Cross-sectional view, jejunum, scale bars, 50 μm. (H) Crypt loss in Fgfbp1 cKO intestine occurs with dampened proliferation and complete loss of Lgr5+ CBC cells. Multiplex ISH for Lgr5 and Fgfbp1 and Ki67 immunostaining of control and Fgfbp1 cKO jejunum. Scale bars, 50 μm.

Figure 6.. Model of the upper crypt zone as the source of intestinal epithelial regeneration.

(A) Fgfbp1+ cells in the upper crypt are multi-potent ISCs that give rise to all the mature lineages, including the Lgr5+ CBC cells, along the length of the intestine during homeostasis. (B) Loss of Lgr5+ cells via selective ablation or pharmacological Rspo inhibition results in reconstitution of crypt bases by a pre-existing pool of Fgfbp1+ ISCs. (C) The Rspo signaling axis controls the transition between Fgfbp1+ upper crypt cells and Lgr5+ CBC cells. Fgfbp1+ ISCs are Rspo-responsive and expand upon Rspo stimulation, but unlike Lgr5+ CBC cells they do not require Rspo signaling (via LGR5) for self-renewal. In the absence of Rspo, Lgr5+ CBC cells are depleted without regeneration from Fgfbp1+ cells. (D) FGFBP1 is essential for intestinal epithelial homeostasis. Fgfbp1 loss disrupts intestinal tissue architecture and results in crypt dropout, impaired proliferation, and complete Lgr5+ CBC cell loss, consistent with impaired regeneration.