R-Spondin1 expands Paneth cells and prevents dysbiosis induced by graft-versus-host disease

Trillions of microbes inhabit mammals, with the largest population in the intestinal tract, and play vital roles in a host’s metabolism and immunity (Qin et al., 2010). Composition of intestinal microbiota is mainly maintained by luminal secretion of antimicrobial peptides (AMPs) from Paneth cells. α-Defensins are the most bactericidal AMPs released from Paneth cells (Ayabe et al., 2000; Salzman et al., 2010). Emerging evidence demonstrates that Paneth cell functions are impaired in various inflammatory and metabolic disorders, resulting in unfavorably altered intestinal microbiota (dysbiosis; Salzman and Bevins, 2013). Dysbiosis, however, exacerbates the underlying diseases, thus creating a vicious cycle between the host and microbiota.

Graft-versus-host disease (GVHD) is an alloreactive, donor T cell–mediated inflammatory disease that occurs after allogeneic hematopoietic stem cell transplantation (SCT), involving the skin, liver, and gastrointestinal tract (Ferrara et al., 2009). We and others have shown that GVHD leads to a loss of Paneth cells and mediates intestinal dysbiosis (Eriguchi et al., 2012; Jenq et al., 2012). The dysbiosis that occurs in MHC-mismatched mouse models of GVHD is remarkable and thus represents a feasible tool to test novel strategies to modulate dysbiosis (Eriguchi et al., 2012).

Current strategies to restore the gut ecosystem are bacteriotherapy, using diet, prebiotics/probiotics, and fecal microbiota transplantation; however, no physiological approach has been developed so far. Here, we demonstrate a novel approach to restore intestinal microbial ecology and prevent dysbiosis by Wnt agonist R-Spondin1 (R-Spo1; Kim et al., 2005; Takashima et al., 2011) or recombinant α-defensin (Tomisawa et al., 2015) in mice. The Wnt agonist R-Spo1, which binds to leucine-rich repeat–containing G protein–coupled receptor (Lgr) 5, is one of the essential factors to build intestinal villus-crypt units from a single Lgr5⁺ intestinal stem cell (ISC; Sato et al., 2009; de Lau et al., 2011; Farin et al., 2016). We found that R-Spo1 stimulates ISCs to differentiate to Paneth cells and enhanced luminal secretion of α-defensins. In addition, administration of R-Spo1 or the recombinant mouse α-defensin cryptdin-4 (Crp4) prevents GVHD-mediated dysbiosis after SCT. Such approaches represent a physiological approach at modifying the gut ecosystem to restore intestinal homeostasis and host–microbiota cross talk toward therapeutic benefits. Because dysbiosis has a role in the pathogenesis of many diseases, such approaches have broad potential in individuals at risk or with various diseases.

R-Spo1 enhances the proliferation of cycling ISCs via the Wnt/β-catenin signaling pathway and generates crypt-villus organoids from ISCs in vitro (Sato et al., 2009). We previously showed that administration of R-Spo1 stimulated proliferation of ISCs and induced crypt cell hyperplasia in vivo (Kim et al., 2005; Takashima et al., 2011). However, the effects of R-Spo1 on Paneth cell proliferation and function remain to be determined. Here, we first addressed whether R-Spo1 could increase the number of Paneth cells in vivo. R-Spo1 was i.v. injected to B6D2F1 mice at a dose of 200 µg for 6 d. The number of Paneth cells morphologically identified as cells containing eosinophilic granules in H&E staining was significantly increased in all sites of the small intestine, including duodenum, jejunum, and ileum of R-Spo1–treated mice (Fig. 1, A and B). R-Spo1 significantly elongated crypt depth (Fig. 1 C). Although Kim et al. (2005) showed that daily injection of R-Spo1 at a dose of 100 µg for 3 d did not increase Paneth cell numbers, differences in dose and duration of the R-Spo1 used may explain the discrepancy in the results between studies. Immunofluorescence studies demonstrated that Paneth cells generated by R-Spo1 coexpress lysozyme, Crp1, a subtype of α-defensins, and matrix metalloproteinase-7 (MMP-7), which converts pro–α-defensins into active form (Fig. 1, D and E). These results indicate that they are functionally mature Paneth cells (Wilson et al., 1999). Similar results were obtained in BALB/c mice, ruling out the strain-specific effects of R-Spo1 on Paneth cell expansion (Fig. S1, A–C). There were some MMP-7⁺ Crp1⁻ cells in R-Spo1–treated mice (Fig. 1 E and Fig. S1 C). Although characteristics of these cells remain to be elucidated, Wnt activation may lead to precocious differentiation of progenitors into Paneth cells (Tian et al., 2015).

R-Spo1 also increased the number of enhanced GFP (EGFP)⁺ ISCs in B6-Lgr5-EGFP-creER mice, in which Lgr5⁺ ISCs were marked, as we previously reported (Fig. S1, D and E; Takashima et al., 2011). All intestinal epithelial cells, including Paneth cells, are progenies of ISCs (Barker et al., 2007). We therefore addressed whether R-Spo1 stimulated ISCs to differentiate to de novo Paneth cells or expanded mature Paneth cells in fate-mapping reporter mice, in which Lgr5⁺ ISCs and their progenies are visually identified as RFP⁺ cells. Administration of R-Spo1 for 3 d increased RFP⁺ Paneth cells, indicating that R-Spo1 stimulated differentiation of ISCs toward Paneth cells (Fig. 1, F and G).

We next addressed whether other intestinal cell types were also expanded by R-Spo1. Alcian blue staining of the small intestine demonstrated that R-Spo1 increased goblet cells, but the expansion of goblet cells occurred to a lesser extent compared with that of Paneth cells (Fig. 1, B and H). Immunofluorescence staining of chromogranin A showed R-Spo1 did not significantly increase enteroendocrine cells (EECs) in the small intestine (Fig. 1, I and J). The mechanisms by which R-Spo1 preferentially expands Paneth cells compared with other cell types in the small intestine remains to be elucidated. One possible explanation could be that Paneth cell differentiation is more strictly dependent on Wnt/β-catenin signaling than other cell types (Yin et al., 2014). Both Wnt activation and Notch suppression in ISCs promote ISCs to differentiate to Paneth cells (VanDussen et al., 2012; Tian et al., 2015). We next studied expression of delta-like (Dll) 1 and Dll4, canonical Notch ligands that suppress progenitor differentiation to Paneth cells. R-Spo1 significantly decreased the expression of Dll4 in the small intestine, and there was a trend toward reduced expression of Dll1 in R-Spo1–treated animals compared with controls (P = 0.06), suggesting that R-Spo1 activated Wnt signaling but suppressed Notch signaling (Fig. 1, K and L). Although the molecular mechanisms by which R-Spo1 suppresses Notch ligands remain to be elucidated, Wnt and Notch signaling in ISCs could be reciprocally controlled, as Notch inhibition enhances Wnt activation (Tian et al., 2015).

We then evaluated whether administration of R-Spo1 could lead to increased luminal secretion of α-defensins, the most potent AMP class from Paneth cells (Ayabe et al., 2000; Salzman et al., 2010). Fecal samples were collected after administration of R-Spo1. An ELISA (Nakamura et al., 2013) demonstrated that R-Spo1 significantly increased fecal levels of α-defensins Crp1 and Crp4 (Fig. 1, M and N; and Fig. S1, F and G). We addressed whether increased production of α-defensins could be caused by an increase in Paneth cell production on a per-cell basis. Quantitative PCR of Paneth cells purified from the isolated crypts showed expression levels of Defa4 and Mmp7 were not increased, indicating that increased production of α-defensins was merely a reflection of an increased number of Paneth cells by R-Spo1 (Fig. 1, O and P).

These data showed novel in vivo effects of R-Spo1: it promptly stimulates ISCs to differentiate to functionally maturated Paneth cells and increases intraluminal levels of α-defensins. We then addressed whether increased secretion of α-defensins into the lumen could alter the composition of intestinal microbiota of healthy naive mice using bacterial 16S ribosomal RNA (rRNA) sequences of the fecal pellets. Permutational analysis of variance (PERMANOVA) of the genus composition did not detect a significant alteration of intestinal microbial composition after R-Spo1 treatment (P = 0.62), although there was a minor alteration of some genera (Fig. S1 H). These data suggested that transient up-regulation of α-defensins has minimal bactericidal activity against certain commensals living together with a host in a symbiotic relationship, as previously reported (Ouellette et al., 1994; Ayabe et al., 2000; Masuda et al., 2011). However, the effects of chronic administration of R-Spo1 on microbial composition remain to be elucidated because genetic overexpression of human defensin 5 in mice alters gut microbiota composition (Salzman et al., 2010).

Historically, an impact of the intestinal microbiota on disease development was first described in mouse models of GVHD using germ-free mice or gut-decontaminating antibiotics in the early 1970s (Jones et al., 1971). GVHD is an alloreactive donor T cell–mediated inflammatory disease involving the gut, skin, and liver after allogeneic SCT (Teshima et al., 2002). We and others have reported that GVHD mediates Paneth cell injury and subsequent intestinal dysbiosis in association with reduced levels of α-defensins (Eriguchi et al., 2012; Jenq et al., 2012, 2015; Levine et al., 2013; Holler et al., 2014). Because α-defensins have minimal bactericidal effects against certain commensal bacteria and predominantly target pathogenic bacteria, loss of α-defensins may be causally related to the reduction of commensal bacteria and the compensatory outgrowth of pathogenic bacteria. Alternately, AMP loss may alter the ecological interaction between bacterial species or host immune reaction against commensal bacteria or reduce bacterial niche for commensal bacteria (Faust and Raes, 2012). To evaluate dysbiosis associated with the loss of Paneth cells and the reduced secretion of α-defensin, lethally irradiated B6D2F1 or B6D2F1-Lgr5-EGFP-creER (H-2^b/d) mice were i.v. injected with 5 × 10⁶ bone marrow (BM) cells + 5 × 10⁶ splenocytes from MHC-mismatched B6 (H-2^b) donors on day 0. Immunofluorescence analysis of the small intestine on day 7 showed that α-defensin–expressing Paneth cells were severely reduced in allogeneic animals compared with syngeneic controls (Fig. 2, A–C; and Fig. S2 A). In parallel with Paneth cell loss, fecal levels of α-defensin Crp4 were significantly lower in allogeneic animals on day 7 than those in syngeneic controls (Fig. 2 D), although bacterial load in fecal pellets was not changed between syngeneic and allogeneic mice (Fig. 2 E). Bacterial 16S rRNA sequences of the fecal pellets showed significant dysbiosis 7 d after allogeneic SCT, whereas there was no alternation of bacterial composition in syngeneic controls compared with those in naive mice (Fig. 2 F and Fig. S2, B–D). A principal component analysis and PERMANOVA demonstrated a significant difference in the bacterial composition between allogeneic and syngeneic animals (Fig. 2, G and H). There were significant correlations between GVHD severity and the proportion of specific microbes on day 7 after allogeneic SCT. Abundances of Proteobacteria at the phylum level, Enterobacteriales at the order level, and Escherichia and Bacteroides at the genus level were negatively correlated with the body weight of mice, whereas those of the phylum Firmicutes, the order Clostridiales, and the genus Lactobacillus were positively correlated with the body weight of mice (Fig. 2, I–K; and Fig. S2, E–H). These data confirmed that GVHD-mediated reduction of AMPs resulted in the loss of symbiotic bacteria that have immunoregulatory or homeostatic roles in the intestine, as well as the expansion of pathogenic proinflammatory bacteria (Gerbitz et al., 2004; Heimesaat et al., 2010; Eriguchi et al., 2012; Jenq et al., 2012). We hypothesize that R-Spo1 could inhibit the expansion of pathogenic noncommensals with the protection of symbiotic beneficial bacteria by expanding Paneth cells and increasing α-defensins after SCT.

To test whether R-Spo1 could protect Paneth cells after SCT, recipient mice were i.v. injected with R-Spo1 at a dose of 100 µg from day −3 to −1 and day 1 to 3 after SCT, and Paneth cells were enumerated on day 7 and 28 after SCT. As expected, R-Spo1 protected Paneth cells against GVHD-mediated damages, and significantly more Paneth cells persisted on days 7 and 28 after SCT in R-Spo1–treated allogeneic recipients compared with allogeneic controls (Fig. 2, A–C; and Fig. S2 A). We also confirmed that R-Spo1 was protective against ISC damages (Fig. S3 A) and ameliorated GVHD pathology such as severe blunting of villi and inflammatory infiltration (Fig. 2 B). Furthermore, R-Spo1 treatment significantly restored fecal levels of mouse α-defensin Crp4 as high as the levels of syngeneic mice (Fig. 2 D). To evaluate whether R-Spo1 reduces intestinal bacterial load, fecal bacterial loads in R-Spo1–treated mice were measured with quantitative PCR targeting 16S rRNA and compared with those in the mice treated with a combination of four antibiotics, including ampicillin, streptomycin, vancomycin, and metronidazole from day −7 of transplantation. In sharp contrast with the dramatically reduced bacterial load in mice treated with antibiotics, R-Spo1 did not affect bacterial load (Fig. 2 E). 16S rRNA sequences of the fecal pellets showed that microbiome compositions of R-Spo1–treated mice significantly differed from those of allogeneic controls but were rather similar to that of syngeneic controls (Fig. 2, F–H; and Fig. S2, B–D). R-Spo1 restored intestinal microbiome diversity to the levels of that in syngeneic controls, as determined by a Simpson index (Fig. 2 L; Simpson, 1949). It also inhibited GVHD-associated outgrowth of Escherichia and Bacteroides and reduction of Barnesiella and Lactobacillus (Fig. 2 M and Table 1). Together with the data showing R-Spo1 had minimal effects on commensals in naive mice, these results represent in vivo confirmation of a selective bactericidal activity of α-defensins increased by R-Spo1 against noncommensal bacteria, which we and others have shown in vitro (Ouellette et al., 1994; Masuda et al., 2011). In association with prevention of dysbiosis, brief administration of R-Spo1 significantly suppressed donor T cell infiltration into the small intestine (Fig. S3 B). These data indicate that R-Spo1 prevents GVHD-associated dysbiosis and dampens allogeneic immune responses after allogeneic SCT.

α-Defensins are the most potent AMPs, which account for 70% of the Paneth cell-derived AMP activity (Ayabe et al., 2000; Salzman et al., 2010). Finally, we addressed whether direct administration of α-defensin could modify the intestinal microbiota. We generated Crp4, the most bactericidal mouse α-defensin (Tomisawa et al., 2015). Crp4 was orally administered at a dose of 125 µg twice daily from day 3 to 7 after SCT and resulted in increased fecal levels of Crp4 measured by ELISA, which detects only conformationally native form, not changed or reduced form, indicating that Crp4 was not degenerated before reaching the intestines (Fig. 3 A; Nakamura et al., 2013). Administration of Crp4 significantly ameliorated GVHD-mediated dysbiosis (Fig. 3 B) and restored microbiome diversity (Fig. 3 C) without altering the bacterial load on day 7 (Fig. 3 D). There was a significant difference in the bacterial composition between Crp4-treated animals and allogeneic controls on day 7 (Fig. 3, E and F). Crp4 significantly inhibited the GVHD-associated outgrowth of Bacteroides at genus level (44.9 ± 4.5% vs. 28.0 ± 3.8%; P = 0.038), and there were trends of reduced Enterobacteriales (11.3 ± 3.2% vs. 3.78 ± 3.3%; P = 0.11) and increased Clostridiales (17.6 ± 3.0% vs. 32.6 ± 6.2%; P = 0.067) at order levels in Crp4-treated animals compared with allogeneic controls (Fig. 3, G and H). Oral administration of Crp4 also significantly suppressed donor T cell infiltration into the small intestine, liver, and spleen (Fig. S3, C–E) and significantly mitigated weight loss and clinical GVHD scores on day 21 after SCT (Fig. 3, I and J). The flora shift toward the widespread prevalence of Bacteroides and Escherichia in GVHD (Fig. 2 F and Fig. 3 B), as has been shown (Eriguchi et al., 2012), increased the translocation of pathogen-associated molecular patterns, such as lipopolysaccharides into systemic circulation, and augmented donor T cell activation (Cooke et al., 2001). Crp4 suppressed the outgrowth of such pathogens (Fig. 3 B), as has been shown (Masuda et al., 2011), and thereby could suppress systemic donor T cell expansion. However, dysbiosis recurred after discontinuation of Crp4, and effects of Crp4 against GVHD were abrogated on day 35 (Fig. 3, B, C, E, F, I, and J). Although Crp4-treated mice showed a trend toward better survival early after SCT, the difference was eventually not significant (Fig. 3 K), suggesting that the brief administration of a single Crp subset may not be impactful enough to improve survival.

Although both R-Spo1 and Crp4 prevented dysbiosis, the effects of R-Spo1 were more potent than those of Crp4. We previously reported that short-term administration of R-Spo1 significantly prolonged survival after SCT, but the effects of short-term Crp4 on GVHD were modest. This is probably because R-Spo1 potentially increases AMPs from Paneth cells other than Crp4, such as Crp1 and lysozyme (Fig. 1, E, F, and N), and possibly other AMPs from other cells such as regenerating islet-derived protein 3 (Reg3) from enterocytes and exerts protective effects against ISCs and epithelial injury (Takashima et al., 2011).

Epithelial regeneration is critical for barrier maintenance for host defense and favorable immune responses (Takashima et al., 2011; Lindemans et al., 2015). α-Defensins lack protective effects on epithelium; intestinal epithelial injury impairs the beneficial effects of commensals on the host immune system (Nakahashi-Oda et al., 2016). IL-22 also promotes ISC-mediated intestinal epithelial regeneration (Lindemans et al., 2015); however, differentiation of ISCs toward Paneth cells requires Wnt/β-catenin signaling (Yin et al., 2014). Thus, R-Spo1 could more potently stimulate Paneth cell proliferation and secretion of α-defensins than IL-22. However, IL-22 increases enterocyte production of another AMP, Reg3 (Lindemans et al., 2015). A combined use of R-Spo1 and IL-22 may be an attractive strategy.

There is a cross talk between a host and microbiota. Normal microbiota is important for the host’s health and a balanced immune system. Emerging data suggest that changes in the microbiota play a crucial role in the pathogenesis of both intestinal and nonintestinal disorders (Kamada et al., 2013; Mathewson et al., 2016; Teshima et al., 2016). The direct administration of a single AMP could be a novel approach to restore the gut ecosystem in dysbiosis. However, administration of R-Spo1 may modify the gut ecosystem more potently by increasing multiple AMPs from Paneth cells and possibly by other intestinal epithelial cells. Thus, the administration of R-Spo1 represents a novel physiological approach to restore the gut microbial ecosystem in order to ameliorate disease activity not only in GVHD, but also in various diseases in association with intestinal dysbiosis.

Female C57BL/6 (B6: H-2^b, CD45.2⁺), C57BL/6-Ly5a (B6-CD45.1: H-2^b, CD45.1⁺), B6D2F1 (H-2^b/d, CD45.2⁺), and DBA/2 (H-2^d, CD45.2⁺) were purchased from CLEA Japan. Female BALB/c (H-2^d, CD45.2⁺) mice were purchased from Charles River Japan. B6-Lgr5-EGFP-creER (Lgr5-EGFP-IRES-creER^T2) and B6-R26^Tomato [B6.Cg-Gt(ROSA)26Sor^{tm14(CAG-tdTOMATO)Hze/J}] mice were purchased from Jackson Laboratory. Lgr5-EGFP-creER×R26^Tomato mice were generated by crossing B6-R26^Tomato female mice with B6-Lgr5-EGFP-creER male mice. B6D2F1-Lgr5-EGFP-creER mice were generated by crossing B6-Lgr5-EGFP-creER male mice with DBA/2 female mice. All animal experiments were performed under the auspices of the Institutional Animal Care and Research Advisory Committee (approval number 12–0106). Experiments in this manuscript were performed in a nonblinded fashion. All mice for microbial analysis or the recipients of SCT were purchased from the same vendor and cohoused until they were used in the experiments.

Recombinant human R-Spo1 was generated as previously reported (Kim et al., 2005). Recombinant Crp4 was produced and purified as previously described (Tomisawa et al., 2015). Streptomycin and metronidazole purchased from Sigma-Aldrich and ampicillin and vancomycin purchased from Wako were given at a concentration of 1 mg/ml in drinking water.

Mice were transplanted as previously described (Takashima et al., 2011). In brief, after lethal x-ray total body irradiation (11–13 Gy) delivered in two doses at 4-h intervals, mice were i.v. injected with 5 × 10⁶ BM cells with 5 × 10⁶ splenocytes. Female mice at 8–12 wk old were allocated randomly for each experimental group, ensuring the mean body weight in each group was similar. Mice were maintained in specific pathogen-free conditions and received normal chow and autoclaved acidic water, pH 2.5, after SCT. Survival after SCT was monitored daily, and the degree of clinical GVHD was assessed weekly by using a scoring system (Cooke et al., 1996). Sample size for survival experiments was aimed for 10 per group for 75% power to detect 60% difference in survival probabilities from the threshold with a one-sided type I error of 0.05.

For pathological analysis, samples of the small intestine were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained with H&E or Alcian blue. Immunofluorescence analysis was performed using primary antibodies (Abs), including rabbit antilysozyme (A0099; Dako), rat anti-Crp1 (77-R63), rabbit anti–MMP-7 (D4H5; Cell Signaling Technology), chicken anti-EGFP (ab13970; Abcam), rabbit anti-RFP (ab34771; Abcam), and rabbit anti–chromogranin A (ab85554; Abcam), and visualized with Alexa Fluor 488–, 555–, and 647–conjugated secondary Abs. Pictures of tissue sections were taken at room temperature using a digital camera (DP20; Olympus) mounted on a microscope (BX50; Olympus), fluorescence microscope (BZ-X700; Keyence), and confocal laser microscope (FV-1000D; Olympus).

Lgr5-EGFP-creER×R26^Tomato mice were i.p. injected with 40 mg/kg tamoxifen (Sigma-Aldrich) daily for 3 d to label ISCs, followed by i.v. injection of R-Spo1 of PBS daily for 3 d.

Monoclonal Abs conjugated with fluorescein isothiocyanate, phycoerythrin, phycoerythrin-Cy7, peridinin–chlorophyll protein complexes, allophycocyanin, and allophycocyanin-Cy7 were purchased from BD PharMingen, eBioscience, and Biolegend. In flow cytometric analysis, at least 300,000 live samples were analyzed using FACSCantoII (BD Biosciences) and FlowJo software (Tree Star).

The small intestine was isolated and opened with scissors along intestinal length. Samples were then incubated on a shaker in complete medium (CM; 2% FCS in PBS) in the presence of 1 mM DL-dithiothreitol (Sigma-Aldrich) at 37°C for 20 min and subsequently incubated with 1.3 mM EDTA (Nippon Gene) in CM at 37°C for 40 min. They were rinsed twice in CM and digested with 0.3 mg/ml of type IV collagenase (Sigma-Aldrich) at 37°C for 45 min, homogenized, filtered, and washed.

Fecal samples were prepared as previously described (Nakamura et al., 2013; Eriguchi et al., 2015). In brief, samples were air dried, powdered using a bead beater–type homogenizer (Beads Crusher µT-12; TAITEC). Fecal extract was collected after blending with PBS using a vortex mixer for 1 h at 4°C and centrifugation at 20,000 g for 20 min, and levels of Crp1 and Crp4 were measured by sandwich ELISA as previously described (Nakamura et al., 2013; Eriguchi et al., 2015).

For crypt isolation, mouse small intestine was flushed with cold PBS and cut open lengthwise in 10-cm-long pieces. The villi were scraped off using a scalpel blade, and remaining tissues were washed with cold PBS. After incubation with 30 mM EDTA in HBSS for 10 min at room temperature, the tissue fragments were shaken vigorously in fresh HBSS to exfoliate crypts. The dissociated crypts were further digested with shaking at 180 rpm in HBSS supplemented with 200 U/ml collagenase (Sigma-Aldrich), 10 µM Y-27632 (Sigma-Aldrich), and 1 mM N-acetylcysteine (Sigma-Aldrich) for 5 min at 37°C on a horizontal shaker (TAITEC). Crypt cells were then briefly treated with 50 µg/µl DNase I (Roche), filtered with a 40-µm cell strainer (BD Falcon), and subjected to Paneth cell purification.

Crypt cells were incubated with 10 µM Zinpyr-1 (Santa Cruz Biotechnology) for 10 min at 37°C to stain secretory granules in Paneth cells and filtered with Cell Strainer Snap Cap with 35-µm nylon mesh (BD Falcon). Zinpyr-1⁺SSC^low cells were sorted as Paneth cell–rich fraction using a cell sorter (JSAN; Bay Bioscience), and then Paneth cells identified as Zynpyr-1⁺ granular cells using a confocal microscope (A1; Nikon) were aspirated one by one using a 50-µm glass micropipette (1-GT50S-5; Nepa Gene) with micromanipulators (MN-4 and MMO-202ND; Narishige) and an electronic pipette (PicoPipet; Nepa Gene).

50 Paneth cells were lysed in 4 µl lysis buffer of SingleShot cell lysis kit (Bio-Rad Laboratories) and subjected to reverse transcription using an iScript Advanced cDNA synthesis kit for quantitative real-time PCR (Bio-Rad Laboratories) and a thermal cycler (Veriti; Thermo Fisher Scientific). For preparation of cDNA from the small intestine, total RNA from frozen tissues was extracted using ISOGEN II (Nippon Gene). cDNA was synthesized using ReverTra Ace qPCR RT Master Mix with gDNA Remover (FSQ-301; Toyobo).

Quantitative real-time PCR was performed on the ABI StepOnePlus system using TaqMan Fast Advanced Master Mix and the primers and fluorescent TaqMan probe sets specific for mouse Defa4 (Mm00651736_g1), Mmp7 (Mm00487724_m1), Dll1 (Mm01279269_m1), and Dll4 (Mm00444619_m1; all from Applied Biosystems). The reactions were performed in a 96-well plate in triplicate. The 18S rRNA primer probe set was 5′-GCTCTTTCTCGATTCCGTGGG-3′ for the forward primer, 5′-ATGCCAGAGTCTCGTTCGTTATC-3′ for the reverse primer, and FAM-CTCCACCAACTAAGAACGGCCATGCACC-TAMRA for the probe (Sigma-Aldrich), and the set was separately amplified in the same plate as an internal control for variation in the amount of cDNA in PCR.

Fecal bacterial load was measured as previously described (Liang et al., 2015).

Fresh fecal pellets were collected from each mouse. Total DNA was extracted using a PowerFecal DNA isolation kit (MO BIO Laboratories). The V3–V4 variable region of the 16S rRNA gene was amplified from fecal DNA extracts using the 16S metagenomic sequencing library protocol provided by Illumina. Samples were sequenced on the MiSeq sequencing platform according to standard Illumina sequencing protocols. Sequenced read data were deposited to the DNA Databank of Japan Sequence Read Archive with DRA accession nos. DRA005119 and DRA005031.

We discarded the reads that (a) contained ambiguous nucleotides and (b) were mapped to the PhiX genome sequence by a Bowtie 2 (version 2.2.3) search with default parameters (Langmead and Salzberg, 2012). Each forward and reverse read for the paired-end library was then merged by a USEARCH (version 7.0.1090) with a –fastq_truncqual 7 parameter (Edgar, 2010). Both the forward and reverse primer sequences were removed by a TagCleaner search with four mismatches allowed (Schmieder et al., 2010). We obtained the high-quality reads after removal of the reads that (a) contained <350 or >650 nt and (b) were associated with a mean Phred-like quality score of <25 as calculated by the Illumina MiSeq sequencer. Sequence clustering of the high-quality reads was conducted by using the UCLUST (version 7.0.1090) with identity >97% (Edgar, 2010) and query and reference coverage >80%. Chimeric operational taxonomic units were detected and removed if the operational taxonomic units were assigned to the chimera in both of the following two methods: (1) a UCHIME (version 7.0.1090) reference mode search against the reference gold database (http://drive5.com/uchime/gold.fa) and (2) a UCHIME de novo mode search (Edgar et al., 2011). Taxonomic assignment of the high-quality reads was performed by an RDP MultiClassifier (version 1.1) search with a bootstrap value >0.5 (Wang et al., 2007). Compositional differences of genera among mice with different treatments were visualized by a principal component analysis and statistically analyzed by a PERMANOVA with Bray–Curtis dissimilarity index in the vegan library of the R software. For multiple testing corrections, a false discover rate estimation was used in Fig. 3 F, whereas more conservative multiple testing corrections with Bonferroni correction was performed to reduce type I errors with the larger sample sizes in Fig. 2 H. Diversity of the microbial community was calculated using the Simpson’s index with genus level (Simpson, 1949).

Data were checked for normal distribution and similar variance between groups. Student’s t tests were used to compare data between two groups when the data follow a normal distribution. Mann–Whitney U tests were used to compare data between two groups when the data did not follow a normal distribution. The Kaplan–Meier product limit method was used to obtain survival probability, and the log-rank test was applied to compare survival curves. Analyses were performed using JMP Pro (version 11.0.0; SAS) and Prism (version 7.0; GraphPad Software). P < 0.05 was considered statistically significant.

Fig. S1 shows that R-Spo1 increases Lgr5⁺ ISCs and Paneth cells in vivo. Fig. S2 that shows R-Spo1 protects Paneth cells against GVHD and ameliorates intestinal dysbiosis. Fig. S3 shows effects of R-Spo1 on ISCs, Paneth cells, and donor T cells and effects of Crp4 on donor T cells after SCT.

This study was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology/Japan Society for the Promotion of Science KAKENHI (25293217 and 221S0002 to T. Teshima, 26461438 to D. Hashimoto, 26462831 to K. Nakamura, and 26440072 to T. Aizawa), Health and Labor Science Research grants from the Ministry of Health, Labour and Welfare (to T. Teshima), the Promotion and Standardization of the Tenure-Track System (to D. Hashimoto), the Astellas Foundation for Research on Metabolic Disorders (to D. Hashimoto), the Takeda Science Foundation (to D. Hashimoto), SENSHIN Medical Research Foundation (to D. Hashimoto), Global Station for Soft Matter, a project of the Global Institution for Collaborative Research and Education at Hokkaido University (to T. Aizawa), and the Center of Innovation Program from the Japan Science and Technology Agency (to T. Teshima, K. Nakamura, and T. Ayabe.).

The authors declare competing financial interests: T. Teshima is a recipient of a research grant from Kyowa Hakko Kirin.

Author contributions: E. Hayase, D. Hashimoto, and T. Teshima developed the conceptual framework of the study, designed the experiments, conducted studies, analyzed data, and wrote the paper. C. Noizat, R. Ogasawara, S. Takahashi, H. Ohigashi, Y. Yokoi, R. Sugimoto, K. Ebata, T. Yamakawa, S. Matsuoka, E. Yokoyama, T. Ara, and K. Nakamura conducted experiments. Y. Ogura, T. Hayashi, H. Mori, and K. Kurokawa conducted 16S rRNA gene sequencing and data analysis. K. Tomizuka produced R-Spo1. R. Hiramine and T. Aizawa produced Crp4. T. Kondo and T. Ayabe supervised experiments.