Epithelial phosphatidylinositol-3-kinase signaling is required for β-catenin activation and host defense against Citrobacter rodentium infection

Abstract

Citrobacter rodentium infection of mice induces cell-mediated immune responses associated with crypt hyperplasia and epithelial β-catenin signaling. Recent data suggest that phosphatidylinositol-3-kinase (PI3K)/Akt signaling cooperates with Wnt to activate β-catenin in intestinal stem and progenitor cells through phosphorylation at Ser552 (P-β-catenin(552)). Our aim was to determine whether epithelial PI3K/Akt activation is required for β-catenin signaling and host defense against C. rodentium. C57BL/6 mice were infected with C. rodentium and treated with dimethyl sulfoxide (DMSO) (vehicle control) or with the PI3K inhibitor LY294002 or wortmannin. The effects of infection on PI3K activation and β-catenin signaling were analyzed by immunohistochemistry. The effects of PI3K inhibition on host defense were analyzed by the quantification of splenic and colon bacterial clearance, and adaptive immune responses were measured by real-time PCR (RT-PCR) and enzyme-linked immunosorbent assay (ELISA). Increased numbers of P-β-catenin(552)-stained epithelial cells were found throughout expanded crypts in C. rodentium colitis. We show that the inhibition of PI3K signaling attenuates epithelial Akt activation, the Ser552 phosphorylation and activation of β-catenin, and epithelial cell proliferative responses during C. rodentium infection. PI3K inhibition impairs bacterial clearance despite having no impact on mucosal cytokine (gamma interferon [IFN-γ], tumor necrosis factor [TNF], interleukin-17 [IL-17], and IL-1β) or chemokine (CXCL1, CXCL5, CXCL9, and CXCL10) induction. The results suggest that the host defense against C. rodentium requires epithelial PI3K activation to induce Akt-mediated β-catenin signaling and the clearance of C. rodentium independent of adaptive immune responses.

Fig. 1.

C. rodentium induces epithelial hyperplasia and PI3K signaling. Colon tissue samples from controls (day 0) or mice inoculated with C. rodentium are shown. (A) From left to right are representative photomicrographs of tissues from infected mice at the specified time points stained with H&E and for BrdU, P-Akt, P-β-catenin552, and alcian blue. Arrows indicate positively stained cells within crypt bases, and arrowheads indicate staining in the mid-crypt to upper crypt regions. (B to E) Infection resulted in crypt lengthening (B) as proliferation (BrdU incorporation) increased 6-fold (C) as well as increased PI3K signaling determined by the enumeration of P-Akt-positive epithelial cells/crypt (D) and P-β-catenin⁵⁵²-positive IECs/crypt (E). (F) Enumeration of alcian blue-positive cells/crypt demonstrates the progressive loss of goblet cells coinciding with increased PI3K signaling to day 14. Goblet cells repopulate crypts as the PI3K signal returns to baseline. Values are means ± SEM. Magnification, ×200 with ×400 inserts. Scale bar, 100 μm. *, P ≤ 0.02; #, P ≤ 0.01 (versus day 0).

Fig. 2.

PI3K inhibition impairs C. rodentium-induced Akt and β-catenin signaling and attenuates epithelial cell proliferation. The distal colon was removed from C57BL/6 mice 14 days after oral infection with C. rodentium (C. rod) treated with either DMSO (vehicle control) (middle) or LY294002 (LY) (right). Left panels represent uninfected mice treated with LY294002. (A) Colonic tissue stained with H&E and for BrdU, P-Akt, P-β-catenin⁵⁵², and alcian blue. Arrows indicate positively stained cells within crypt bases, and arrowheads indicate staining in the mid-crypt to upper crypt regions. (B to F) LY294002 treatment alone had no effect relative to baseline but impaired C. rodentium-induced crypt lengthening, PI3K signaling, proliferation, and goblet cell depletion. Values are means ± SEM. Magnification, ×200 with ×400 inserts. Scale bar, 100 μm. *, P < 0.01.

Fig. 3.

PI3K inhibition does not affect epithelial cell apoptosis. (A) Distal colon sections from mice infected 14 days earlier with C. rodentium and treated with either DMSO or LY294002 were stained for apoptotic cells by the TUNEL technique. (B) Enumeration of TUNEL-positive epithelial cells per ×400 field. Values are means ± SEM. HPF, high-power field.

Fig. 4.

Inhibition of PI3K signaling reduces β-catenin signaling in C. rodentium colitis. (A and B) Colon sections from uninfected (day 0) and C. rodentium-infected mice treated with DMSO or LY294002 (day 14) were processed and stained to assess c-Myc (green) expression (A) and cyclin D1 (brown) expression (B). (C) Enumeration of c-Myc- and cyclin D1-positive epithelial cells per crypt. Values are means ± SEM. Magnification, ×200 with ×400 inserts.

Fig. 5.

Inhibition of PI3K signaling increases and prolongs weight loss. C. rodentium infection caused transient weight loss at day 4 with recovery by day 7 and subsequent weight gain. The blockade of PI3K signaling worsened weight loss that did not recover until day 19 and was significantly reduced until day 28. *, P < 0.01 versus C. rodentium (n = 4 to 8 mice for each group).

Fig. 6.

PI3K signaling is required for clearance of C. rodentium. In vivo colonization and clearance dynamics were monitored from day 7 after inoculation with C. rodentium to day 28. (A) Mice receiving LY294002 showed prolonged colonization within the colon and shedding of bacteria in the stool, demonstrating that the PI3K inhibition reduces the ability to clear C. rodentium infection. (B) PI3K inhibition resulted in persistence of systemic infection within the spleen. Arrows indicate time points for LY294002 injection. *, P < 0.05 versus C. rodentium (n = 3 to 4 mice for each group, representative of 3 separate experiments).

Fig. 7.

Interruption of PI3K signaling disrupts the epithelial barrier. (A, top) Two separate representative distal colons stained with H&E demonstrating the expected crypt hyperplasia with relatively intact epithelial lining 14 days after infection with C. rodentium. (Bottom) Two separate representative distal colons in mice infected with C. rodentium and treated with LY294002. PI3K inhibition of host epithelial cell responses resulted in an increased disruption of the epithelial cell barrier and frank ulceration. Arrows indicate ulcer edges. Magnification, ×100. (B) Histological scores for ulceration, inflammatory cell infiltration, and edema were assessed in colonic tissue samples from infected mice treated with DMSO or LY294002.

Fig. 8.

PI3K is dispensable for adaptive and innate immune responses to C. rodentium. (A) Time course of mRNA (adaptive immunity) isolated from the distal colon after C. rodentium infection alone or with LY294002 treatment. (B) Distal colon segments were removed 14 days after infection, with and without LY294002 treatment; washed; and cultured overnight for direct cytokine analysis by ELISA corresponding to Fig. 8A mRNA. (C) Time course of mRNA (innate immunity) isolated from the distal colon after C. rodentium infection alone or with LY294002 treatment (n = 3 to 4 mice per group).

Fig. 9.

PI3K inhibition does not reduce IL-22 production. PI3K inhibition does not affect either mRNA (A) or colon explant culture (B) ELISA measurements of IL-22 production 4 and 7 days after infection with C. rodentium.

Fig. 10.

PI3K inhibition with wortmannin (WM) impairs the host defense against C. rodentium. The distal colon was removed from C57BL/6 mice 14 or 18 days after oral infection with C. rodentium treated with either DMSO (vehicle control) or 0.1 mg/kg wortmannin (see Materials and Methods). (A) Wortmannin treatment alone had no effect relative to baseline but impaired C. rodentium-induced crypt lengthening. (B) C. rodentium infection caused transient weight loss with recovery by day 7 and subsequent weight gain. The blockade of PI3K signaling worsened weight loss that did not begin to recover until day 16. (C) Mice receiving wortmannin showed prolonged colonization within the colon and shedding of bacteria in the stool as well as a persistence of systemic infection within the spleen, demonstrating that PI3K inhibition reduces the ability to clear C. rodentium infection. (D) mRNA levels at days 14 and 18. Values are means ± SEM. *, P < 0.05 versus C. rodentium plus DMSO; #, P < 0.01 versus day 14 (n = 5 to 8 mice per group).