Intact function of Lgr5 receptor-expressing intestinal stem cells in the absence of Paneth cells

Lgr5⁺ Cells Localize in Intestinal Intervillus Regions Before Birth and Exhibit Stem-Cell Properties in the Absence of Paneth Cells.

Because Lgr5 marks a stem-cell population in intestinal (5) and distal stomach (10) epithelia and in hair follicles (11), Lgr5⁺ cell properties have been characterized in detail in normal adult tissues (22), but their emergence has not been examined in detail during intestine development. We examined the distribution of Lgr5⁺ intestinal cells in fetal and newborn mice, using native GFP expression to monitor Lgr5⁺ cells in Lgr5^{GFP-IRES-CreER} mice, which express GFP in Lgr5⁺ CBCs in a mosaic fashion (5). At embryonic day 15 (E15), when the pseudostratified intestinal epithelium has formed the first villi lined with columnar cells, no GFP expression was evident in the duodenum, the proximal segment of the intestine where Lgr5⁺ cells are most abundant in adults (Fig. S1 A and B). On postnatal day 1 (P1), Lgr5⁺ cells already were sharply localized to the intervillus zone of proliferating cells (Fig. 1A), suggesting that developing progenitors acquire Lgr5 expression and not that ubiquitous Lgr5 expression becomes restricted. Lgr5⁺ cells remained confined to the intervillus regions and emerging crypts at postnatal days 5 (P5), 10 (P10), 15 (P15), and 21 (P21) (Fig. 1 B–D and Fig. S1C), as they are in adults. Although GFP⁺ intervillus cells were interspersed with GFP⁻ cells at P1 (Fig. S1D), this intermingling was rarely the case at P5 or P10, when the deep intervillus space of GFP⁺ crypts contained only GFP⁺ cells (Fig. 1 B and C); this distribution is distinct from the intermingled pattern of Lgr5⁺ and Paneth cells at P21 (Fig. 1D) and thereafter. Consistent with previous reports that recognizable Paneth cells appear gradually after birth (23), immature cryptdin-expressing Paneth cells were infrequent at P5 or P10 (Fig. 1E) and were observed to mingle among GFP⁺ cells only at P15 and P21 (Fig. 1 F and G); only 20% of P5 intervillus regions expressed the Paneth-cell marker Crs4c1. Lysozyme staining revealed more mature Paneth cells at P15 and fuller maturation by P21 (Fig. S1E). Cells at the intervillus base uniformly exhibited ultrastructural characteristics attributed to CBCs (5): a broad base, narrow apical cytoplasm, wedge-shaped basal nucleus, numerous supranuclear mitochondria, and absence of secretory granules (Fig. 1H and Fig. S2A). Only 10 of 100 intervillus regions showed one or more cells with immature Paneth-like granules (Fig. S2B).

To test whether early Lgr5⁺ intervillus cells manifest stem-cell properties in the absence of Paneth cells, we crossed Lgr5^{GFP-IRES-CreER} and Rosa26^YFP reporter mice (24) and then administered single doses of tamoxifen (0.05 mg/g body weight) to Lgr5^{GFP-IRES-CreER};Rosa26^YFP pups at P5 and 10. This treatment resulted in YFP expression in villus cells within 5 d (Fig. S1F) and uniform YFP expression in villi 3 wk later (Fig. 1 I and J). Taken together, these data demonstrate that Lgr5⁺ cells cluster in intervillus regions without intervening cells and behave like self-renewing progenitors early in gut maturation; thus, Paneth cells are dispensable in the development of functioning Lgr5⁺ stem cells.

Lgr5⁺ CBCs Occupy the Paneth Cell Zone in the Absence of Paneth Cells.

A recent study reported on the requirement for Paneth cell-mediated Wnt signaling in Lgr5⁺ cell function in epithelial organoid cultures and genetically engineered mice (19). In these mouse models, the decline in stem-cell numbers generally followed that of Paneth cell numbers, but no model eliminated Paneth cells completely (19). We therefore reexamined the requirement for Paneth cells in a mouse model with complete and unambiguous loss of Paneth cells in affected crypts. The intestine- and neural-restricted transcription factor Atoh1 is necessary to produce intestinal secretory cells, including Paneth cells (21, 25). We crossed Lgr5^{GFP-IRES-CreER} mice to Atoh1^flx/flx mice, which carry a conditional floxed null allele, and studied the compound mutants 3 wk after administration of tamoxifen. Because Paneth cells have a long life (26), and a few cells persisted after 3 wk (Fig. S3A), we also examined mice 3 mo after tamoxifen injection. Atoh1^+/+ mice showed the expected alternating pattern of Lgr5⁺ (GFP⁺) CBCs and Crs4c1-expressing Paneth cells (Fig. 2 A–C). No Paneth cells were present after 3 mo in Atoh1-deleted crypts, in contrast to adjacent crypts that lacked Lgr5-GFP expression and had thereby avoided Cre-mediated Atoh1 recombination (Fig. 2 D–F). These data confirm efficient Atoh1 and Paneth cell loss in numerous intestinal crypts. Lgr5⁺ CBCs not only were intact but occupied the whole base in such Paneth cell-depleted crypts, as revealed by uniform GFP staining (Fig. 2 D–F).

To exclude an artifact that might follow from mosaic Cre expression in Lgr5^Cre mice, we also crossed Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx mice to the Villin^CreER strain (20); in the resulting progeny, Villin^CreER drives Atoh1 deletion in all crypts, and Lgr5^GFP provides a means to visualize Lgr5⁺ CBCs in a significant fraction of those crypts. Within 2 wk of tamoxifen injection, Villin^Cre-mediated Atoh1 deletion removed Paneth cells throughout the intestine, consistent with a requirement for Atoh1 in Paneth cell survival and representing total, sustained Paneth cell loss of unprecedented scope. Fewer than 1 in 500 crypts escaped total Paneth cell depletion; the crypts that avoided total depletion of Paneth cells contained residual long-lived cells from a previous renewal cycle before Cre-mediated Atoh1 deletion (Fig. S3B). In crypts that entirely lacked Paneth cell markers, we again observed uniform GFP staining, demonstrating that Lgr5⁺ CBCs were intact and lacked intermingled cells (Fig. 2 G–I). In situ hybridization analysis for the independent CBC marker Olfm4 (27) after Villin-Cre–mediated Atoh1 and Paneth cell loss revealed intact Olfm4⁺ stem cells throughout crypt bases (Fig. 2 J and K), indicating that these cells are bona fide CBCs and are not residual GFP⁺ descendants. Moreover, compared with control mice, GFP⁺ cells were modestly increased in number in the absence of Paneth cells (Fig. S3C). To see if CBCs may be affected more adversely by reduced Paneth cell numbers than by their total absence, we modulated each animal model, administering a single 0.1-mg dose of tamoxifen to Villin^CreER;Lgr5^GFP;Atoh1^flx/flx mice or examining Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx mice 2 wk after tamoxifen treatment. These approaches allowed us to examine individual crypts containing small numbers of residual Paneth cells; neither approach to subtotal Paneth cell depletion decreased GFP⁺ CBCs (Fig. S3C).

Proliferation of Lgr5⁺ CBCs Is Increased and Stem-Cell Activity Is Preserved in the Absence of Paneth Cells.

Two weeks after tamoxifen administration, the proliferative marker Ki67 revealed that all crypt cells in Villin^CreER;Atoh1^flx/flx mice were replicating actively, compared with 82% of cells in wild-type crypts (Fig. 3 A and B). To look specifically at proliferation of Lgr5⁺ CBCs, we introduced the S-phase marker BrdU into Villin^CreER;Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx mice 3 wk after tamoxifen injection, confirming that Paneth cells were eliminated by this time (Fig. S2C). Sequential analysis of native GFP/Lgr5 expression and BrdU immunostaining revealed a 14% absolute increase in the fraction of BrdU⁺ cells among Lgr5⁺ CBCs compared with control Lgr5^{GFP-IRES-CreER} intestines (Fig. 3 C–G). To test if this increased proliferation of Lgr5⁺ CBCs accelerates epithelial turnover, we traced BrdU after a single pulse. Progression of BrdU-stained cells along the crypt–villus axis at 1 h, 1 d, and 2 d was similar in mutant and wild-type mice (Fig. S4A). In agreement with enhanced cell replication, however, we did find 13% and 15% increases in crypt height in the duodenum and ileum, respectively, in Atoh1-depleted intestines (Fig. S4B).

To test long-term stem-cell activity of these proliferating Lgr5⁺ CBCs, we examined Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx;Rosa26^YFP mice 3 mo after tamoxifen administration. These mice revealed Lgr5⁺ CBC-derived mature cells in entire villi in the complete and persistent absence of Paneth cells (Fig. 4 A and B). Although villus cells derived from Lgr5⁺ CBCs lacked secretory markers, as expected because of the Atoh1^−/− genetic background (25), these data demonstrated the cardinal stem-cell property of self-renewal over a period of at least 3 mo. To exclude again the possibility of artifacts resulting from mosaic Cre expression in Lgr5^CreER mice, we asked the same question when Paneth cells were eliminated throughout the intestine in Villin^CreER mice. Three months after pulse tamoxifen exposure, Villin^CreER;Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx;Rosa26^YFP mice also showed YFP staining throughout intestinal villi (Fig. 4C). The duration of these experiments rules out short-term ROSA26 recombination in short-lived mature cells as the basis for mature YFP⁺ villus cells. Because Villin-Cre marks cells throughout crypts, including Lgr5⁺ CBCs, this experiment alone does not distinguish between stem-cell activity of Lgr5⁺ and other putative stem-cell populations. It nevertheless reveals intact long-term self-renewal in the gut epithelium in the absence of Paneth cells and, together with data in Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx;Rosa26^YFP mice (Fig. 4 A and B), attributes this Paneth cell-independent stem-cell activity to Lgr5⁺ CBCs.

Cell Replication-Associated Transcriptional Targets of Wnt Signaling Are Intact in the Absence of Paneth Cells.

To decipher possible mechanisms for Paneth cell regulation of Lgr5⁺ CBCs, Sato et al. (19) compared gene expression in isolated CBCs and Paneth cells. Coupled with organoid cultures in vitro, their data implied that Paneth cells may elaborate Wnt ligands to activate Wnt signaling in receptor-expressing Lgr5⁺ CBCs. To examine Wnt signaling, we sorted GFP⁺ cells from Paneth cell-depleted Villin^CreER;Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx and control Villin^CreER;Lgr5^{GFP-IRES-CreER};Atoh1^+/flx crypts 14 d after tamoxifen exposure. We used the purified Lgr5⁺ CBCs from crypts with and without Paneth cells (Fig. 5A) for quantitative RT-PCR (qRT-PCR) analysis of Wnt target genes. Consistent with enhanced proliferation of Lg5⁺ CBCs in Paneth cell-depleted crypts, the levels of Wnt target transcripts, including genes associated with cell proliferation such as CD44, Myc, and Ccnd1, were unaffected or even increased compared with wild-type cells (Fig. 5B). Immunohistochemistry further showed expansion of the CD44 expression domain into the crypt base (Fig. 5C), which is fully occupied by Lgr5⁺ CBCs, and verified the modestly reduced Sox9 expression evident by qRT-PCR (Fig. 5D). Thus, robust crypt cell proliferation in the absence of Paneth cells (Fig. 3) reflects a substantially preserved segment of the Wnt transcriptional response, implying that another source of Wnt signaling must remain active in vivo. To ask if Lgr5⁺ CBCs themselves might elaborate Paneth-expressed Wnt ligands, thus representing such a source, we tested Wnt3 and Wnt11 mRNA levels in sorted Lgr5⁺ cells by qRT-PCR. We found no Wnt11 expression, irrespective of Paneth cell exposure, and low Wnt3 levels were reduced further in Lgr5⁺ cells from Paneth cell-depleted crypts (Fig. S4C). Unless CBCs express another Wnt ligand, these data suggest in vivo Wnt sources other than Lgr5⁺ or Paneth cells.

Experimental Animals.

Villin^CreER transgenic mice were generously provided by S. Robine (Institut Curie, Paris, France) (20). Lgr5^GFP-CreER, Math^flx, and Rosa26^YFP mice were purchased from Jackson Laboratories. Animals were housed under specific pathogen-free conditions and handled in accordance with protocols approved by the Animal Care and Use Committee of the Dana-Farber Cancer Institute.

Drug Administration.

Tamoxifen (Invitrogen) was dissolved in 10% (vol/vol) ethanol and sunflower oil. Adult mice were injected i.p. with 1 mg for 5 consecutive d. Pups under the age of 21 d received a single i.p. dose of 0.05 mg/g body weight. Adult mice were killed 1 h, 1 d, or 2 d after i.p. administration of BrdU (50 mg/kg; BD Biosciences).

Immunohistochemistry.

Tissues were fixed overnight in 4% paraformaldehyde at 4 ºC and were washed three times in PBS. For cryo-embedding, fixed tissues were incubated overnight in 30% sucrose in PBS at 4 ºC and then were embedded in Optimal Cutting Temperature compound (OCT; Sakura); sections were cut at 10-μm thickness. For paraffin-embedding, fixed tissues were dehydrated, embedded in paraffin, and sectioned at 5-μm thickness. Antigens were retrieved in 10 mM Na citrate buffer (pH 6) followed by blocking of endogenous peroxidase activity in methanol and 3% H₂O₂. After blocking with 10% FBS, samples were incubated overnight at 4 ºC with one of the following antibodies: Crs4c1 [1:1,000; gift of Andre Ouellette (University of Southern California, Los Angeles, CA)], lysozyme (1:200; Invitrogen), Ki67 (clone MM1; 1:1,000; Vector Laboratories), BrdU (1:300; AbD Serotec), CD44 (1:500; eBioscience), or Sox9 (1:300; Millipore). Immunohistochemistry samples were washed and treated with biotin-conjugated anti-mouse or anti-rabbit IgG (1:300; Vector Laboratories). Color reactions were developed with Vectastain avidin-biotin complex (ABC kit; VectorLaboratories) and diaminobenzidine substrate (Sigma-Aldrich).

In Situ Hybridization.

After overnight fixation in 4% paraformaldehyde at 4 ºC, intestines were embedded in OCT compound (Sakura). Cut 10-μm sections were treated with10 μg/mL proteinase K (Roche) and hybridized overnight at 64 ºC with digoxigenin-labeled antisense. The riboprobe for Olfm4 was generated from IMAGE clone IRCKp5014N1115Q (Source BioScience imaGenes). Slides were washed in 2× SSC, incubated with alkaline phosphatase-conjugated digoxigenin Ab (1:2,000; Roche) and, to develop the stain, treated with nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Roche).

Crypt Isolation, Lgr5⁺ CBC Sorting, and qRT-PCR Analysis.

Intestines of control Lgr5^{GFP-IRES-CreER} and Villin^CreER;Lgr5^{GFP-IRES-CreER};Atoh1^flx/flx mice were dissected and washed in cold PBS. Villi were removed by scraping with glass microslides (Surgipath). To isolate crypt epithelium, samples were transferred to 5 mM EDTA in PBS (pH 8), followed by three 1-min shakings by hand, a 15-min incubation at 4 ºC, and passage through 70-μm filters (BD Falcon) to collect the flowthrough. After treatment of crypts for 45 min in 3.3% TrypLE (Invitrogen) to disaggregate cells, Lgr5⁺ CBCs were sorted by flow cytometry. RNA was isolated using TRIzol reagent (Invitrogen) and reverse transcribed using SuperScript enzyme (Invitrogen). cDNA was assessed using SYBR green master mix (Applied Biosystems). Threshold cycle (C_t) values for the test transcripts first were normalized with respect to Gapdh and are expressed as ratios of transcript levels in mutant cells to transcript levels in wild-type cells. Means and SDs for each group (n = 2) were calculated using Microsoft Excel.