Mitogenic Influence of Human R-Spondin1 on the Intestinal Epithelium

See allHide authors and affiliations

Science  19 Aug 2005:
Vol. 309, Issue 5738, pp. 1256-1259
DOI: 10.1126/science.1112521


Several described growth factors influence the proliferation and regeneration of the intestinal epithelium. Using a transgenic mouse model, we identified a human gene, R-spondin1, with potent and specific proliferative effects on intestinal crypt cells. Human R-spondin1 (hRSpo1) is a thrombospondin domain-containing protein expressed in enteroendocrine cells as well as in epithelial cells in various tissues. Upon injection into mice, the protein induced rapid onset of crypt cell proliferation involving β-catenin stabilization, possibly by a process that is distinct from the canonical Wnt-mediated signaling pathway. The protein also displayed efficacy in a model of chemotherapy-induced intestinal mucositis and may have therapeutic application in gastrointestinal diseases.

The intestinal epithelium undergoes rapid and continuous self-renewal along the crypt-villus axis, and the β-catenin/T cell factor (TCF) signal transduction pathway plays a pivotal role in the proliferation, differentiation, and oncogenesis of the intestine (13). The Wnt proteins provide a prototype for the ligand-mediated activation of this signaling pathway, and their activities are considered central to the maintenance of the undifferentiated state of intestinal crypt progenitor cells (46). The R-spondin protein family includes four human paralogs (R-spondin1–4), each of which contains a leading signal peptide, two cystein-rich, furin-like domains, and one thrombospondin type 1 domain (7, 8). Studies of the Xenopus R-Spondin ortholog have shown that this protein class acts as secreted activators of Wnt/β-catenin signaling and Wnt-dependent myogenesis in Xenopus (7).

We have undertaken an investigation of secreted protein biology by using a transgenic mouse knock-in (KI) approach that allows B cell restricted–transgene expression under control of an immunoglobulin κ promoter (9) (figs. S1 and S2). Among the proteins tested, we observed that expression of a cDNA encoding human R-spondin1 (hRSpo1) in KI chimeras led to a dramatic abdominal distention by 8 weeks of age. Examination of hRSpo1-KI chimeras at necropsy revealed a substantial increase in the diameter, length, and weight of the small intestine (Table 1, Fig. 1A, and fig. S3). We also observed an increase in the length of the colon of hRSpo1-KI chimeras (Table 1). Histological analysis and immunohistochemistry of the small intestine of hRSpo1-KI chimeras revealed a marked, diffuse thickening of the mucosa, crypt epithelial hyperplasia, and a greatly expanded zone of proliferating cells as evidenced by the expression of a proliferation marker, Ki-67 (Table 1 and fig. S3).

Fig. 1.

Expression of hRSpo1 stimulates growth of the intestinal epithelium. (A) Gross abdominal anatomy from 8-week-old control (left) and hRSpo1-KI (right) chimeras. Scale bar, 1 cm. (B) Recombinant hRSpo1 protein increases the proliferation of intestinal crypt epithelial cells. BALB/c mice were injected daily intravenously (iv) with hRSpo1 (100 μg per injection) or the same volume of phosphate-buffered saline (PBS) as a vehicle control for 3 days. Bromodeoxyuridine (BrdU) immunohistochemistry assays on mid-jejunum and transverse colon sections are shown. (C) Crypt proliferative index. BrdU immunohistochemistry assays were performed on mid-jejunal sections after a single injection of hRSpo1 (100 μg per mouse). A total of 40 crypts from two mice were analyzed for BrdU incorporation and are represented as the mean ± SD. *, P < 0.01, analysis of variance (ANOVA). (D) hRSpo1 treatment does not change the number of goblet cells in small intestine as detected by Alcian blue staining. Animals were injected daily with hRSpo1 (100 μg) or with PBS as a control for 3 or 7 days, and mid-jejunal sections were analyzed (n = 3). Error bars, mean ± SD. (E) Immunohistochemistry of mid-jejunal sections using antibody to lysozyme. Animals were treated with hRSpo1 for 3 days. Paneth cells are confined to the bottom of crypts, and hRSpo1 treatment did not alter their numbers.

Table 1.

Phenotypic data of hRSpo1-KI chimeras and recombinant protein–injected mice. All data are presented as the mean ± SD. Recombinant protein injection: BALB/c mice received a daily injection of 100 μg of hRSpo1 protein for 3 days (n = 3).

KI Recombinant protein injection
Control hRSpo 1 Control hRSpo 1
Small intestine
    Diameter (mid jejunum, mm)View inline 2.1 ± 0.2 4.0 ± 0.8View inline 2.3 ± 0.2 3.64 ± 0.1View inline
    Wet weight (KI: including cecum, g)View inline 3.8 ± 0.4 8.5 ± 2.6View inline 0.9 ± 0.04 1.34 ± 0.1View inline
    Crypt number/circumference (KI: 4 weeks of age)View inline 144 ± 9.3 246 ± 33.9View inline 125 ± 7.4 154 ± 6.0View inline
    Ki67-positive cell number/crypt unitView inline 11.5 ± 3.3 60.3 ± 21.1View inline ND ND
Colon length (mm)View inline 99 ± 9 125 ± 18View inline 67 ± 3 85 ± 5View inline
  • View inline n = 4 to 6 animals per group, 7 to 8 weeks of age.

  • View inline For each animal, two mid-ileal circumferences were analyzed (n = 2 animals per group).

  • View inline§ Total 30 longitudinally, well-oriented crypt units were counted for mid-ileal sections from hRSpo 1-KI and control animals (n = 2 animals per group, 4 weeks of age).

  • View inline* P < 0.05;

  • View inline** P < 0.01 (control versus hRSpo 1). ND, not determined.

  • To further investigate the activity of hRSpo1, we expressed recombinant hRSpo1 protein (10) and injected the purified protein into normal mice. Histological evaluation of gastrointestinal sections after 3 days of treatment with hRSpo1 showed significant proliferation of the intestinal crypt epithelial cells in the small intestine and colon, consistent with the results obtained in the hRSpo1-KI chimeras (Table 1 and Fig. 1B). A more detailed examination of the crypts in the small intestine after a single bolus injection of hRSpo1 showed increased proliferation within 3 hours and a peak proliferative effect at approximately 24 hours after injection (Fig. 1C). Immunohistochemistical analysis of the small intestine demonstrated no significant changes in the number of goblet cells and Paneth cells in the small intestine of hRSpo1-treated mice (Fig. 1, D and E), indicating that the transient exposure to hRSpo1 does not affect the maturation and migration of differentiated cells along the crypt-villus axis.

    Wnt-mediated activation of epithelial cells results in the stabilization of cytosolic β-catenin, which subsequently translocates to the nucleus and transactivates TCF target genes (11, 12). Isolated colonic crypt cells from mice treated with hRSpo1 displayed stabilization of cytosolic β-catenin (Fig. 2A), and this correlated with an increase in the nuclear localization of β-catenin in the crypts of hRSpo1-KI chimeras (Fig. 2B), indicating functional activation of β-catenin. This phenotype is reminiscent of the increased crypt proliferation and nuclear accumulation of β-catenin by the conditional deletion of adenomatosis polyposis coli (APC) in mice (13). However, crypt proliferation induced by administration of hRSpo1 protein was completely resolved within 4 days after withdrawal of hRSpo1 (fig. S4), indicating that the action of hRSpo1 protein is transient and reversible, in contrast to the effect of APC deletion. Consistent with the activation of β-catenin by hRSpo1, the expression of β-catenin target genes, including Axin2, EphB3, CD44, and Pla2g2a (13), was increased in hRSpo1-KI chimeras (Fig. 2C) and corresponded to the elevated levels of hRSpo1 mRNA.

    Fig. 2.

    Modulation of β-catenin signaling by hRSpo1. (A) hRSpo1 causes an increase in cytosolic β-catenin in isolated colonic crypts. Colonic crypts were isolated from mice injected with hRSpo1 (100 μg iv) or PBS control 6 hours after injection, and cytosolic fractions were analyzed by Western blotting with antibody to β-catenin. (B) hRSpo1 increases nuclear β-catenin in KI mice. β-catenin immunostaining in representative crypts from control (left) and hRSpo1-KI (right) chimeras (4 weeks of age) are shown. Cells with strong nuclear β-catenin staining can be observed in the bottom of the crypts from control chimeras. In hRSpo1-KI chimeras, cells with nuclear β-catenin staining are also found in the upper crypt region corresponding to the extended morphology of the crypt compartment. Magnification, 400x. (C) Increased expression of β-catenin target genes in small intestine of hRSpo1-KI chimeras correlates with hRSpo1 expression. Reverse transcription polymerase chain reaction was performed to evaluate time-dependent intestinal expression of transcripts as described (10). C, control chimeras; R, hRSpo1-KI chimeras. (D) Western analysis of cytosolic β-catenin. HEK 293 cells were treated with recombinant hRSpo1 protein at indicated concentrations for 3 hours, and cytosolic fractions were analyzed by Western blot using antibodies to β-catenin or β-Actin (loading control). NTC, not treated control. (E) Time course of β-catenin stabilization. HEK293 cells were treated with 40 ng/ml of hRSpo1, and cytosolic fractions were analyzed. (F) hRSpo1 enhances Wnt3A-induced β-catenin stabilization in HEK293 cells. The HEK293 cells were incubated with various concentrations of Wnt3A in the presence or absence of hRSpo1(10 ng/ml). (G) Dkk-1 effects on β-catenin stabilization; HEK293 cells were preincubated with 125 ng/ml Dkk-1 for 2 hours, followed by treatment with hRSpo1 or Wnt3A at indicated concentrations. hRSpo1-induced β-catenin stabilization is only slightly inhibited by Dkk-1 protein, whereas Wnt3A-induced β-catenin stabilization is completely inhibited. (H) β-catenin stabilization in mouse L cells. Cells were incubated with hRSpo1 or Wnt3A at indicated concentrations for 3 hours, and cytosolic fractions were resolved in SDS–polyacrylamide gel electrophoresis. Whereas Wnt3A causes a dose-dependent increase of cytosolic β-catenin, hRSpo1 fails to stabilize β-catenin in these cells.

    To further evaluate the molecular mechanism of RSpo1-induced β-catenin activation, we treated various cell lines with recombinant protein. hRSpo1 induced the stabilization of β-catenin in human embryonic kidney (HEK) 293 cells in a dose-dependent manner, with detectable activity at 10 ng/ml (∼300 pM) (Fig. 2D). Similar effects were seen upon treatment with Wnt-3A (Fig. 2F). Peak stabilization of β-catenin occurred ∼6 hours after application of hRSpo1 (Fig. 2E), which correlates with the hRSpo1-induced increase in crypt proliferation (Fig. 1C). Simultaneous treatment of HEK 293 cells with hRSpo1 and Wnt3A protein indicated an additive, or mildly synergistic, effect (Fig. 2F). Interestingly, we found that Dickkopf-1 (Dkk-1), a potent Wnt pathway antagonist that interacts with the Wnt coreceptor LRP5/6 (14), only partially inhibited hRSpo1-induced β-catenin stabilization, whereas it completely inhibited Wnt3A-mediated β-catenin stabilization in these cells (Fig. 2G). Furthermore, in the mouse fibroblast L cell line, Wnt3A induced β-catenin stabilization, whereas hRSpo1 had no effect (Fig. 2H). These results suggest that hRSpo1 is capable of stabilizing β-catenin by a pathway that does not entirely overlap with the canonical Wnt/β-catenin pathway.

    Mucosal damage (mucositis) is frequently encountered as a result of anticancer therapies, and mitogenic agents, including fibroblast growth factor (FGF) family members and glucagon-like peptide-2 (GLP2), have shown efficacy in disease model studies and in clinical applications (1518). When tested in mice, FGF-7 and GLP2 also stimulated intestinal crypt proliferation, and isolated colonic crypts from FGF-7- or GLP2-treated mice displayed increased levels of β-catenin (fig. S5, A and B), which suggests that β-catenin is a common mediator of crypt proliferation in vivo. We tested hRSpo1 in a mouse xenograft model that included treatment with the chemotherapeutic agent 5-fluorouracil (5-FU) (19, 20) and evaluated both the extent of mucositis and the effects of hRSpo1 on tumor growth in vivo. The murine colon carcinoma cell line CT26 was inoculated into mice, and tumors were allowed to develop (20). hRSpo1 treatment substantially reduced 5-FU-induced diarrhea and weight loss (Fig. 3A), whereas no effect on tumor volume was observed (Fig. 3B). The CT26 line showed an increase in β-catenin levels in response to Wnt3A treatment, indicating that these cells do not have a constitutively activated phenotype, whereas hRspo1 treatment was minimally effective in activating β-catenin (fig. S6). hRSpo1 appeared to exhibit its protective effect by preserving the architecture of both the small intestine and the colon in the presence of 5-FU (Fig. 3C and fig. S7), as well as preventing 5-FU–induced loss of tongue epithelium proliferation (Fig. 3D).

    Fig. 3.

    hRSpo1 reduces 5-FU–induced mucositis in mice without compromising chemotherapy effectiveness. CT26 murine carcinoma cells (1 × 106 cells in Hanks' balanced salt solution buffer) were injected subcutaneously into the left flank of each mouse on Day-5, and solid tumors were allowed to form. Starting on Day 1, 5-FU was injected daily (30 mg per kg of body weight) for 5 days, followed by a 3-day recovery period and sacrifice (Day 8). hRSpo1 was delivered 50 μg/day iv, Day 0 through Day 8. (A) Diarrhea score and animal body weight loss were measured on Day 8 of the study (n = 3). (B) Tumor volume and extirpated tumor weight were measured on Day 8. Error bars, mean ± SD. *, P < 0.05 (ANOVA). (C) Hematoxylin/eosin stained sections (Day 8) include small intestine (SI) and colon (Co). (D) hRSpo1 protects mouse tongue from 5-FU–induced damage. BrdU immunohistochemistry was performed on tongue sections to evaluate proliferation of basal epithelial cells. hRSpo1 prevents 5-FU–induced loss of proliferation in the tongue epithelium. BrdU index was measured in the ventral tongue of mice from each treatment (n = 3). Error bars, mean ± SD. *, P < 0.05 (ANOVA, 5-FU/saline versus vehicle/saline); #, P < 0.05 (ANOVA, 5-FU/saline versus 5-FU/hRSpo1).

    Although the intrinsic function of hRSpo1 in the development and maintenance of normal intestinal epithelium is not yet defined, the expression of hRSpo1 in human intestinal enteroendocrine cells (fig. S8) is consistent with the proposed role of the protein as a crypt cell mitogen and is reminiscent of the expression of the intestinotrophic factor GLP2 (2123). Expression was also observed in the epithelium of kidney (renal tubules) and prostate (seminal vesicles), whereas in the adrenal gland and pancreas, expression appeared to be confined to neuroendocrine-type cells (fig. S8).

    Ascribing biological function for orphan secreted proteins remains a major challenge in the postgenomic era, and the application of an unbiased in vivo screen, as represented by our transgenic mouse system (9), can identify unexpected activities for orphan ligands. Systemic administration of hRSpo1 potently affects proliferation of the intestinal epithelium through activation of β-catenin, and the activation of this pathway by hRSpo1 in vitro would indicate that this effect may result directly from receptor-mediated binding. An epistatic analysis of R-spondin2 has suggested that this protein acts upstream of dishevelled, most likely at the receptor level, which suggests that the protein works within the context of the canonical Wnt/frizzled pathway (7). In contrast, our experiments show that induced accumulation of β-catenin by hRspo1 is relatively insensitive to inhibition by Dkk-1, and we have identified a cell line that responds to Wnt3A but not to hRSpo1. These results suggest that hRSpo1-mediated signaling is not completely dependent on the canonical Wnt/frizzled pathway, although it is possible that hRSpo1 may require a distinct frizzled receptor complex. Clearly it will be very important to identify receptors for R-spondins to fully understand the biology of this important class of activating ligands.

    Supporting Online Material

    Materials and Methods

    Figs. S1 to S8

    References and Notes

    View Abstract

    Navigate This Article