Disruption of Pten Speeds Onset and Increases Severity of Spontaneous Colitis in Il10^−/− Mice

Abstract

BACKGROUND & AIMS

Early-onset ulcerative colitis, severe colonic inflammation that develops in infants and young children, can be caused by alterations in interleukin 10 (IL10) signaling, although other factors are involved in its pathogenesis. We investigated whether loss of phosphatase and tensin homologue (PTEN), which regulates many important cell functions including cell proliferation, cell survival, and toll-like receptor (TLR) signaling pathways, contributes to development of colitis in Il10^−/− mice.

METHODS

We generated Il10^−/− mice (in C57BL/6 and C3H/HeJBir background strains) with disruption of Pten in the intestinal epithelium (Ints^{ΔPten/ΔPten}; Il10^−/− mice), and Ints^ΔCont;Il10^−/− (control) mice. Colon tissues were collected and histologic, transmission electron microscopy, and gene expression analyses were performed. Fecal microbiota samples were analyzed by sequencing of 16S ribosomal RNA genes. We disrupted Tlr4 in Ints^{ΔPten/ΔPten};Il10^−/− mice. Lipopolysaccharide signaling via TLR4 was blocked by feeding mice polymyxin B.

RESULTS

Il10^−/− mice developed colitis when they were 6–7 months old, whereas Ints^{ΔPten/ΔPten};Il10^−/− mice developed severe colitis and colon tumors by the time they were 36 days old. Within 3 months of birth, 80% of Ints^{ΔPten/ΔPten};Il10^−/− mice developed severe colitis and colonic malignancy, whereas none of Ints^{Δ Cont};Il10^−/− mice had these phenotypes. Ints^{ΔPten/ΔPten};Il10^−/− mice had alterations in fecal microbiota compared with controls, such as increased proportions of Bacteroides species, which are Gram negative. Disruption of Tlr4 or feeding Ints^{ΔPten/ΔPten};Il10^−/− mice polymyxin B delayed the development of colitis and reduced disease severity.

CONCLUSIONS

Disruption of Pten in the intestinal epithelium of Il10^−/− mice speeds the onset and increases the severity of colitis. Fecal microbiota from Ints^{ΔPten/ΔPten};Il10^−/− mice have increased proportions of Bacteroides species. Development of colitis is delayed and reduced by blocking TLR4 signaling. Ints^{ΔPten/ΔPten};Il10^−/− mice might be studied as a model for early-onset ulcerative colitis and used to identify new therapeutic targets.

INTRODUCTION

Inflammatory bowel disease (IBD) is a chronic inflammatory condition occurring in the gastrointestinal tract and classified into Crohn’s disease (CD), ulcerative colitis (UC), and indeterminate colitis^{1, 2}. Although IBD is able to occur at any age, intriguingly, a subset of the patients develops severe colonic inflammation at infants and young children, termed early-onset ulcerative colitis³, accompanied with an increased risk for primary sclerosing cholangitis⁴. In addition to the involvement in both CD and UC, defective mutation at the gene encoding interleukin10 (Il10) or IL10 receptor (Il10r) was identified in the patients of early-onset ulcerative colitis^{1, 5, 6}, indicating that impaired IL10 signaling should be involved in colitis development at early ages.

Similarly, Il10^−/− mice in specific pathogen free (SPF) or conventional housing environment are predisposed to developing spontaneous colitis⁷. Il10^−/− mice, however, do not develop colitis under germ-free conditions, indicating that the flare-up of colitis in these mice is driven by antigens in gut microflora⁸. Thus, enteric microbes influence the development of colitis. The timing of developing colitis, however, depends on each housing condition and/or genetic background. Indeed, an elegant study demonstrated that both the severity of colitis and the frequency of colitis development in Il10^−/− mice are greatly influenced by the genetic background⁹, indicating that another genetic factor would participate in eliciting colitis in an Il10 deficient environment.

Pten is a phosphoinositide-3-phosphatase converting phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3] to PI(4,5)P2. With an antagonistic effect against the phosphatidylinositol 3-kinase signaling regulating cell proliferation and apoptosis, Pten plays an important role in suppressing tumorigenesis in multiple organs. In keeping with this notion, conventional Pten^+/− mice exhibited hyperplastic-dysplastic changes in the prostate and skin¹⁰. In contrast to the embryonic lethality of conventional Pten knockout mice¹⁰, global Pten knockout induced by interferon-α/β-inducible Cre-expression system using Pten-floxed (Pten^loxP/loxP) mice resulted in polyposis in the intestine¹¹. On the other hand, Langlois et al¹² demonstrated that intestinal epithelial cell specific-Pten knockout mice using Villin promoter-driven Cre expression system were healthy and normal without any gross abnormalities such as tumorigenesis. Indeed, we confirmed that those intestinal epithelial cell specific Pten knockout mice (Ints^{ΔPten/ΔPten}) generated in our laboratory did not exhibit any gross abnormalities in the intestine (Supplementary Figure 1 and Supplementary Figure 2). Accordingly, the impact of Pten in the intestine still remains to be elucidated.

Although Pten plays a critical role in regulating cell proliferation, growth, and apoptosis¹³ which are the propensity of intestinal epithelial cells, we recently demonstrated that Pten could be an important factor regulating TLR5-midiated signaling pathways, indicating the involvement of Pten in host-microbial interaction in the gut¹⁴. With this in mind, we hypothesized that Pten would be involved in developing and perpetuating colitis on which enteric microflora has profound impacts. To test this hypothesis, we investigated whether Pten in the intestinal epithelial cells could affect the development of colitis in an Il10 deficient environment.

MATERIALS AND METHODS

Animals

Pten floxed (Pten^loxP/loxP)¹⁵, Villin promoter-driven Cre expression mouse (Vil^Cre/+)¹⁶, and Il10^−/− (C3H/HeJBir or C57BL/6 background) mouse⁷ were obtained from the Jackson Laboratory (Bar Harbor, ME). As Langlois et al¹² did, intestinal epithelial cell specific Pten knockout mouse (Ints^{ΔPten/ΔPten}) was generated by crossing Pten^loxP/loxP mouse with Vil^Cre/+ mouse. Then, Ints^{ΔPten/ΔPten} mouse was crossed with Il10^−/− mouse to generate Ints^{ΔPten/ΔPten}; Il10^−/− mouse in which Il10 was conventionally deleted and simultaneously, Pten was deleted in the intestinal epithelial cells. Ints^{ΔPten/ΔPten};Il10^−/− mouse was backcrossed into C57BL/6 or C3H/HeJBir background for at least 6 generations prior to performing the experiments. Mice were bred and maintained under standard SPF conditions with normal drinking water ad libitum at the animal facility of the Division of Laboratory Animal Medicine, University of California, Los Angeles, under the approval by the Institutional Animal Care and Use Committee of the UCLA. All animal experiments were approved by the Animal Research Committee of UCLA.

All other protocols for materials and experiments are detailed in Supplementary Materials and Methods.

RESULTS

Intestinal epithelial Pten deletion in Il10^−/− mice elicits spontaneous colitis at early ages

We generated Ints^{ΔPten/ΔPten};Il10^−/− (C57BL/6 background) mice having both global Il10 gene knockout and intestinal epithelial cell specific Pten deletion. Ints^{ΔPten/ΔPten};Il10^−/− mice grew smaller than age- and gender-matched littermate control Ints^Cont;Il10^−/− (Figure 1A). Subsequently, we identified that Ints^{ΔPten/ΔPten};Il10^−/− mice had severe inflammation occurring along the colon and cecum characterized with pale and enlarged colon and reduced colonic contents, while Ints^Cont;Il10^−/− mice had normal intestine (Figure 1B). More than 60% of Ints^ΔPten/Δ;Il10^Pten−/− mice failed to survive 3 months after their birth due to the complication of severe colitis, whereas Ints^Cont;Il10^−/− mice remained healthy during the experimental period (Figure 1C). It is of interest to note that Ints^{ΔPten/ΔPten};Il10^−/− mice died as early as 36 days after their birth due to severe colitis. Notably, we confirmed that Il10^−/− mice on C57BL/6 or C3H/HeJBir background took at least 6~7 months after their birth to show signs of severe colitis in our animal facility, as evaluated by mortality (Figure 1D). Considering both the onset timing of colitis in Il10^−/− mice and the baseline life span for laboratory mice which is approximately more than 2 years, spontaneous colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice occurred at very early ages. With this in mind, the onset of colitis in young Ints^{ΔPten/ΔPten};Il10^−/− mice seems to be similar to the development of early-onset ulcerative colitis in humans with which loss-of-function mutations at Il10 or Il10r gene are associated.

Figure 1.

Development of spontaneous colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice at early ages. (A) Gender (female)- and age-matched (3 months old) Ints^{ΔPten/ΔPten};Il10^−/− and littermate Ints^Cont;Il10^−/− mice on C57BL/6 background were presented. (B) Gross appearance of the colon and cecum from these mice was compared. (C) Survival was monitored until the mouse reached to the age of 3 months. Ints^{ΔPten/ΔPten};Il10^−/− (n=22), Ints^Cont;Il10^−/− (n=15). (P < 0.0001). (D) Survival of Il10^−/− mice on C3H/HeJBir (HeJBir) (n=9) and C57BL/6 (BL/6) (n=19) background were monitored. P<0.0001. (E) Body weight change was monitored every other day, starting at the age of 32 days. Results are means ± SD. Ints^{ΔPten/ΔPten};Il10^−/− (n=8), Ints^Cont;Il10^−/− (n=7). *P<0.05, **P<0.01, ***P<0.001, ^#P<0.0001. (F) Full-length of the colon was prepared in a ‘swiss-roll’ method and subjected to H&E staining. Low and high magnifications of representative photograph were presented at the upper and lower position, respectively (scale bars, 100 μm). The yellow inset areas were enlarged in the lower panel. Note, gross image (scale bar, 1 mm) of the colon swiss-roll of Ints^{ΔPten/ΔPten};Il10^−/− mouse was positioned at ‘F-i’. The red-inset area of ‘F-i’ was enlarged at ‘F-ii’. Data (C,D,E) were analyzed with the results accumulated by 6 independent experiments.

In keeping with the incidence of spontaneous colitis, Ints^{ΔPten/ΔPten};Il10^−/− mice were characterized with slow and limited weight-gaining compared to the gradual and consistent pattern in Ints^Cont;Il10^−/− mice (Figure 1E). While Ints^Cont;Il10^−/− mice exhibited normal colonic mucosa, Ints^{ΔPten/ΔPten};Il10^−/− mice revealed dramatic changes in colonic histology caused by inflammation (Figure 1F). Together, these data demonstrate that Ints^{ΔPten/ΔPten};Il10^−/− mice develop spontaneous colitis which can be incurred at very early ages.

Colonic malignancies are found in Ints^{ΔPten/ΔPten}; Il10^−/− mice

Ints^Cont;Il10^−/− mice at 3 months old age had healthy colon (Figure 2A). In contrast, age-matched Ints^{ΔPten/ΔPten};Il10^−/− mice diagnosed with colitis were characterized with various pathological and histological abnormalities in the colon. The colon of Ints^{ΔPten/ΔPten};Il10^−/− mice were featured with hyperplastic mucosa with diffused leukocyte infiltrations causing colon obstruction (Figure 2B-i). Surprisingly, sessile adenocarcinoma attached directly by its base without a stalk was also found along with the penetration of cancerous cells into the submucosa (Figure 2B-ii). Moreover, in Ints^{ΔPten/ΔPten};Il10^−/− mice, a large tumor mass was developed in the colonic epithelium and grown up to the extent of causing bowel obstruction (Figure 2B-iii). These observations suggested that the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice is not only highly susceptible to early-onset spontaneous colitis, but also prone to colonic malignancy. Accordingly, microscopic examination demonstrated that the colonic mucosa of Ints^{ΔPten/ΔPten};Il10^−/− mice were characterized with prominent abscesses, massive neutrophil infiltration, increased epithelial erosion, marked architectural distortion of crypts with epithelial cell atypia, and enhanced necrosis, while these clinical parameters were almost negligible in the colon of Ints^Cont;Il10^−/− mice (Figure 2C). Due to the inflammation and malignancy, colonic mucosal thickness of Ints^{ΔPten/ΔPten};Il10^−/− mice was excessively increased, compared to that of Ints^Cont;Il10^−/− mice (Figure 2D). With the histological analysis, we determined that more than 80% of Ints^{ΔPten/ΔPten};Il10^−/− mice had a diagnosis of colitis and malignancies in the colon by 3 months after their birth, whereas none of the age-matched Ints^Cont;Il10^−/− mice showed such phenotypes (Figure 2E).

Figure 2.

Development of colonic malignancy in Ints^{ΔPten/ΔPten};Il10^−/− mice. (A) Microscopic examination revealed normal colonic mucosa in Ints^Cont;Il10^−/− mice. (B) Ints^{ΔPten/ΔPten};Il10^−/− mouse colon had diffused hyperplasia with marked inflammation (B-i), sessile adenocarcinoma (B-ii), and large tumor mass in the colon with typical granuloma (B-iii). Area of red arrow in the left panel was magnified in the right. In ‘B-iii’, area of yellow arrow was enlarged in the lower. Scale bars, 100 μm. (C) Histological parameters were quantified with H&E sections (n=10/group). (D) The thickness of colonic mucosa was measured using Axio Imager Z1 microscope (Carl Zeiss). (n=20–27/group). (E) Percentage of the mice diagnosed with colitis and colonic malignancy until the mice became 3 months old. Ints^{ΔPten/ΔPten};Il10^−/− (n=22), Ints^Cont;Il10^−/− (n=15). ^# P<0.0001 (Mann-Whitney U test). Representative images are presented.

Electron microscopy demonstrates microvilli degeneration and epithelial lesions in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice

We examined the colonic epithelium from Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice under transmission electron microscopy. In the colonic epithelium of Ints^Cont;Il10^−/− mice, columnar epithelial cells were held well one another along the lateral surface by cell-to-cell junctions and identified with well-developed microvilli and intact intracellular organelles including mitochondria, vacuoles, and lysosomes (Figure 3A–C). In contrast, in the colonic epithelium of Ints^{ΔPten/ΔPten};Il10^−/− mice, epithelial microvilli were effaced, cell-to-cell junctions were disrupted, and the cytoplasm became electron lucent due to necrosis (Figure 3D). Well differentiated adenocarcinoma cells harboring atypical nuclei with excessive mitosis and chromosomal irregularities were observed (Figure 3E). We also observed many apoptotic cells displaying distinctive features of apoptosis such as cell shrinkage, apoptotic body, condensation and fragmentation of chromatin, and cellular disintegration (Figure 3F). Electron micrographs demonstrated that Ints^{ΔPten/ΔPten};Il10^−/− mice had morphological alterations, tissue degeneration, and epithelial cell differentiation occurred in the colon, reflecting the pathophysiology of colonic inflammation and malignancy.

Figure 3.

Electron micrographs of colonic epithelium. (A to C) Ints^Cont;Il10^−/− mice showed normal columnar epithelium in which microvilli (MV), cell-to-cell adhesion, and intracellular organelles such as mitochondria (M), lysosome (L), vacuole (V), goblet cells possessing mucin droplets (Md), and solid nucleus (N) and nucleolus (Nc) are intact. (D to F) The colonic epithelium of Ints^{ΔPten/ΔPten};Il10^−/− mouse exhibited effaced microvilli, necrosis, damaged cell-to-cell adhesion (D), adenocarcinoma cells with atypical and irregular nuclei (open arrowhead) (E), and apoptosis (closed arrowhead) (F). Inset areas in ‘A-i’ and ‘D-i’ were enlarged in ‘A-ii’ and ‘D-ii’, respectively. Representative images are presented.

The development of inflammation in Ints^{ΔPten/ΔPten};Il10^−/− mice is limited to the colon

In this study, Pten deletion in the intestinal epithelium of Ints^{ΔPten/ΔPten};Il10^−/− mice was accomplished by Villin promoter-driven Cre expression. Villin promoter activity is specific to both the small intestine and the colon. Thus, we next investigated whether Ints^{ΔPten/ΔPten};Il10^−/− mice would also develop inflammation in the small intestine. The small intestine of Ints^{ΔPten/ΔPten};Il10^−/− mice, however, did not show any signs of abnormalities. Just like the normal epithelial mucosa from Ints^Cont;Il10^−/− mice, the small intestine of Ints^{ΔPten/ΔPten};Il10^−/− mice exhibited normal epithelial mucosa (Supplementary Figure 3A and B). Of importance is the fact that both Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice did not have any abnormalities in the liver, kidney, and spleen (Supplementary Figure 3C), indicating that the incidence of inflammation and tumorigenesis in Ints^{ΔPten/ΔPten};Il10^−/− mice is specific to the colon.

Blood serum chemistry and hematologic analysis identified that albumin, total protein, alkaline phosphatase (ALP), and glucose levels were substantially decreased in the blood serum of Ints^{ΔPten/ΔPten};Il10^−/− mice, compared to the levels in Ints^Cont;Il10^−/− mice (Figure 4A). The level of ALT and AST signifying liver-related diseases and the total amount of fatty substances (triglyceride and cholesterol) indicative of metabolic defects were similar between Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice (Supplementary Figure 4A). Malnutrition is related to lower ALP level, while elevated ALP is associated with bile duct obstruction or liver diseases including hepatitis and liver cancer. Having found that the liver, kidney, and spleen were normal in Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice, however, these data excluded a possibility of inflammation or tumorigenesis in the liver. Thus, reduced levels of ALP, albumin, total protein, and glucose in Ints^{ΔPten/ΔPten};Il10^−/− mice are likely due to the complication of colitis and colon malignancy and consequent malnutrition.

Figure 4.

Serum chemistry and hematologic analysis using blood samples. (A) Serum chemistry data with significant difference were presented. Ints^{ΔPten/ΔPten};Il10^−/− (n=13); Ints^Cont;Il10^−/− (n=14). (B) Complete blood count was performed for hematologic analysis. Ints^{ΔPten/ΔPten};Il10^−/− (n=12), Ints^Cont;Il10^−/− (n=7–9). *P<0.05, **P<0.01, ***P<0.001 (Mann-Whitney U test). Complementary results of this analysis were presented in Supplementary Figure 4. HB, hemoglobin; HCT, hematocrit; NE, neutrophil; RBC, red blood cell.

In addition, neutrophil level in Ints^{ΔPten/ΔPten};Il10^−/− mice (35.15 %) was substantially increased compared to Ints^Cont;Il10^−/− mice (17.98 %) (Figure 4B), which was in line with the diagnosis of severe inflammation in the colon. Conversely, the level of lymphocytes in Ints^{ΔPten/ΔPten};Il10^−/− mice (56.57 %) was slightly lower than Ints^Cont;Il10^−/− mice (74.73 %), while the levels of monocytes, eosinophils, and platelets were not altered in these mice (Supplementary Figure 4B). Since the neutrophil count rose in Ints^{ΔPten/ΔPten};Il10^−/− mice, its lymphocyte count appeared to be lowered proportionally. In contrast, there were no differences in the level of red blood cell, hematocrit, and hemoglobin counts between these animal groups, indicating that the mice did not suffer from anemia or internal bleeding. Taken together, these results support that the diagnosis of severe inflammation and malignancy in Ints^{ΔPten/ΔPten};Il10^−/− mice are limited to the colon.

The genes regulating inflammation and tumorigenesis are differentially expressed in the colon of Ints^{ΔPten/ΔPten}; Il10^−/− mice

We next analyzed the inflammation- or tumorigenesis-associated gene expression in the colon tissues using inflammation (Supplementary Figure 5A) and cancer (Supplementary Figure 5B) pathway-focused PCR array assay (SABiosciences), followed by individual quantitative real time PCR (qPCR). We found that various pro-inflammatory genes including Cxcl5, Cxcr2, Nos2, Ifnγ, Il-1β, Tnfα, Il6, and Il17a were markedly up-regulated, and simultaneously, potent anti-inflammatory genes such as Nr3c1 and Il5 were substantially down-regulated in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice, compared to the levels in Ints^Cont;Il10^−/− mouse colon (Supplementary Figure 5 and Supplementary Figure 6).

Similarly, Hmox1 and Ccl2 promoting tumorigenesis were significantly up-regulated in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice, compared to Ints^Cont;Il10^−/− mice. In parallel, Il18 and Sirt1 capable of inducing potent anti-tumor activity were significantly down-regulated in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice relative to Ints^Cont;Il10^−/− mice. It is worth noting that Wee1 expression was reduced in Ints^{ΔPten/ΔPten};Il10^−/− mouse colon, compared to Ints^Cont;Il10^−/− mouse colon. Since Wee1 was under-expressed in human colon cancer patients’ tissues and several colon cancer cell lines¹⁷, we believe that reduced Wee1 expression in Ints^{ΔPten/ΔPten};Il10^−/− mouse colon is indicative of the malignancy. In accordance with the induction of colitis and colonic malignancy, together, our data show that the colon tissues of Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice have the opposing expression patterns of pro- vs. anti-inflammatory or tumor promoting- vs. tumor suppressing genes.

Anti-bacterial responses are altered in the colon of Ints^{ΔPten/ΔPten}; Il10^−/− mice

Considerable evidence suggested that impaired Pten alters TLR-induced signaling pathways^{14, 18} which play a key role in activating the innate immunity including the expression of antimicrobial peptides. Having this in mind, we tested whether anti-bacterial responses would be altered in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice using an anti-bacterial response-focused PCR array (SABiosciences), followed by individual qPCR. Various genes associated with anti-bacterial responses were differentially expressed in the colon tissues (Figure 5A). Among them, Tlr5, a specific receptor recognizing bacterial flagellin was prominently down-regulated in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints;Il10^Cont−/− mice (Figure 5B). Moreover, the levels of Tlr1 and TLR-associated signaling molecules, Irak1 and Tirap/Mal, were substantially lower in Ints^{ΔPten/ΔPten};Il10^−/− colon than Ints^Cont;Il10^−/− colon, while Tlr4 expression was similar in these colon tissues. Notably, lactotransferrin (Ltf), a potent anti-bacterial protein for a variety of different microorganisms, was dramatically up-regulated in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mice. Together, these data indicate that anti-bacterial responses are altered in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice.

Figure 5.

Altered anti-bacterial responses in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mouse relative to Ints^Cont;Il10^−/− mouse. (A) The expression of anti-bacterial responses-focused genes was analyzed using mouse colon tissues. Gene expression profiles with significant difference were visualized in the heat map. (B) Representative gene expression having significant difference was separately determined by qPCR. Error bars indicate ± SEM. n=6/group. *P<0.05, **P<0.01 (Mann-Whitney U test). Data are the representative of three independent experiments.

Fecal microbiota of Ints^{ΔPten/ΔPten};Il10^−/− mice is distinguished from that of Ints^Cont;Il10^−/− mice

In Ints^{ΔPten/ΔPten};Il10^−/− mice, spontaneous inflammation was developed only in the colon (Figure 2 and Supplementary Figure 3). Moreover, in early-onset ulcerative colitis patients having a defective IL10 signaling, inflammation develops primarily in the colon. Given that the colon is the primary reservoir of gut microbiota and anti-bacterial responses are altered in the colon of Ints^{ΔPten/ΔPten};Il10^−/− mice (Figure 5), we speculated that the development of spontaneous colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice would be associated with altered microbiota in the colon. Furthermore, emerging evidence suggests that the pathogenesis of human IBD^{19, 20} and colon tumorigenesis²¹ are profoundly influenced by altered gut microbiota. Those considerations prompted us to examine whether Ints^{ΔPten/ΔPten};Il10^−/− mice predisposed to developing early-onset colitis would have altered colonic microbiota, as compared to Ints^Cont;Il10^−/− mice. To test this, fecal samples were harvested from age (6 weeks old)- and gender matched Ints ^{ΔPten/ΔPten};Il10^−/− and littermate Ints^Cont;Il10^−/− mice. 16S ribosomal RNA (rRNA) gene in the fecal samples was sequenced. 100% of analyzed sequences belonged to the kingdom bacteria, and assigned to 7 phyla including Bacteroidetes, Firmicutes, and Proteobacteria encompassing the majority of sequences (> 95%) (Supplementary Figure 7A and B).

From dual hierarchal dendrogram (Figure 6A) and principal coordinates analysis (PCoA) plot (Figure 6B) based upon the predominant genera, we identified that Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice formed its own strong single cluster. From the PCoA plot, we noted that the fecal microbiota from Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice were easily distinguished within the primary 3 axis which together accounted for 92.3% of the sample variability. It is notable that one of Ints^{ΔPten/ΔPten};Il10^−/− samples harvested from the mouse exhibiting no signs of colitis did cluster away from the rest of its group and instead, close to Ints^Cont;Il10^−/− group. Nevertheless, these data demonstrate that the fecal microbiota communities are indeed different in Ints^{ΔPten/ΔPten};Il10^−/− and Ints^Cont;Il10^−/− mice.

Figure 6.

Changed fecal microflora in Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mice. (A) Based upon the predominant genera using Ward’s minimum variance clustering and Manhattan distances, a dual hierarchal dendrogram was constructed. Samples were clustered along the x-axis. The heatmap describes the relative percentage in each sample of the associated genera with a legend provided in the upper left of the figure. Predominant genera were clustered along the Y-axis. (B) PCoA plot of weighted unifrac metrics was generated. Samples were separated along principal coordinate (PC) 3 (92.3% of the sample variability). Each dot represents individual microbiota samples from the mouse. (C to E) The abundance of major bacteria at the genus (C), family (D), and species (E) level were analyzed. Horizontal bar indicates a mean value. *P<0.05; **P<0.01. (Mann-Whitney U test). (n=6/group).

The abundance of Bacteroides is increased in Ints^ΔPten/Δ;Il10^−/−Pten mice

At the genus level, we identified that Bacteroides (Mean: 22.75 % in Ints^Cont;Il10^−/− vs. 60.31 % in Ints^{ΔPten/ΔPten};Il10^−/−) was the most abundant, followed by Prevotella (16.36 % vs. 3.92 %) in the mouse feces (Figure 6C). It was of importance that the abundance of Bacteroides was dramatically increased in Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mice. In contrast, the abundance of Prevotella, Tanerella, Roseburia, Butyrivibrio, and Adlercreutzia were reduced in Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mice. Similarly, the family level of Bacteroidaceae was markedly increased in the feces of Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mice (21.21 vs. 59.91%) (Figure 6D). The level of Prevotellaceae was substantially reduced in Ints^{ΔPten/ΔPten};Il10^−/− mice relative to Ints^Cont;Il10^−/− mice (16.36 vs. 3.92%). Among the genus of Bacteroides which were dramatically increased in the fecal sample of Ints^{ΔPten/ΔPten};Il10^−/− mice, Bacteroides acidifaciens species surged in the samples (5.50 vs. 46.97 %), while other Bacteroides species were reduced in the fecal sample of Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mice (Figure 6E). Taken together, these results suggest that compared to healthy Ints^Cont;Il10^−/− mice, the family of Bacteroidaceae, the genus of Bacteroides, and the species of Bacteroides acidifaciens are greatly increased in the feces of Ints^{ΔPten/ΔPten};Il10^−/− mice which are prone to spontaneous colitis at early ages.

Defective TLR4 inhibits colitis development in Ints^{ΔPten/ΔPten};Il10^−/− mice

In the fecal samples of Ints^{ΔPten/ΔPten};Il10^−/− mice, Bacteroides were by far the biggest genus (60.31%), followed by Prevotella (3.92%) and Tannerella (2.10%). Simultaneously, the abundance of Bacteroides was greatly increased in the fecal samples of Ints^{ΔPten/ΔPten};Il10^−/− mice, compared to Ints^Cont;Il10^−/− mice. Given that Bacteroides are classified as Gram-negative bacteria, we hypothesized that lipopolysaccharide (LPS) from these bacteria would be associated with the pathogenesis of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice. To test this, we generated Ints^{ΔPten/ΔPten};Il10^−/− mice on C3H/HeJBir background which has a missense mutation (P712H) at Tlr4 gene, rendering the mouse non-responsive to LPS from Gram-negative bacteria^{22, 23}. Surprisingly, we did not observe gross abnormalities in the appearance of age (3 months old)- and gender-matched Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) and littermate control Ints^Cont;Il10^−/− (HeJBir) mice (Figure 7A). These mice showed no difference in mortality over 3 months after their birth, as virtually most of Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) (93.3%) and Ints^Cont;Il10^−/− (HeJBir) (100%) mice survived during this experimental period (Figure 7B). In line with these results, there was no difference in the weight gaining until the age of 80 days (Figure 7C). However, the weight gaining pattern of Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) mice was a little slower than that of control mice as they aged. Indeed, from the age of 82 days, Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) mice began to exhibit significantly reduced body weight compared to Ints^Cont;Il10^−/− (HeJBir) mice. Importantly, there was no difference in hematologic data of these mice (Supplementary Figure 8A). Instead, the levels of albumin, total protein, and ALP were reduced in Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) mice compared to Ints^Cont;Il10^−/− (HeJBir) mice (Supplementary Figure 8B). Therefore, these data indicated that enhanced susceptibility of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice could still remain on the C3H/HeJBir background. But the development of colitis was substantially delayed, and the severity of the colitis was greatly ameliorated on the C3H/HeJBir background having non-functional TLR4.

Figure 7.

Reduced colitis development in Ints^{ΔPten/ΔPten};Il10^−/− mice by blocking LPS-TLR4 engagement. (A) Gender (female)- and age (3 months)-matched Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) and littermate Ints^Cont;Il10^−/− (HeJBir) mice were presented. (B) No difference in survival rate was observed over 3 months after their birth. Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir), n=15; Ints^Cont;Il10^−/− (HeJBir), n=22. (C) Body weight change was monitored, starting at the age of 32 days. Results are means ± SD. *P<0.05, **P<0.01. Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) (n=10), Ints^Cont;Il10^−/− (HeJBir) (n=32). Data (B, C) were analyzed with the results accumulated by 6 independent experiments. (D) Representative images of colon H&E sections were presented. Inset areas at low magnification photographs were enlarged at the lower (scale bars, 100 μm). Histological parameters were quantified with H&E sections. Ints^{ΔPten/ΔPten};Il10^+/− (n=7), Ints^Cont;Il10^−/− (n=30), Ints^{ΔPten/ΔPten};Il10^−/− (n=20). (E and F) Ints^{ΔPten/ΔPten};Il10^−/− mice on C57BL/6 background at the age of 32 days started to be under polymyxin B (100 ppm)-mixed or control diet until the age of 3 months. Data were analyzed with the results accumulated by 7 independent experiments. Polymyxin B-treated (n=18), control group (n=13), P=0.0143 (E). H&E sections were prepared from the colon of the mouse survived until the age of 3 months. Representative micrographs were presented. Inset areas inside the upper panel were magnified at the lower. Scale bars, 100 μm. Histological parameters were evaluated. Polymyxin B-treated (n=14), control (n=22). ^# P≤0.0001 (Mann-Whitney U test) (F).

Il10^−/− mice on C3H/HeJBir background are more vulnerable to spontaneous colitis than C57BL/6 background Il10^−/− mouse. In agreement with this, none of Ints^Cont;Il10^−/− on C57BL/6 background exhibited any signs of colitis over 3 months after their birth (Figure 1 and Figure 2). However, Ints^Cont;Il10^−/− (HeJBir) mice at the same age revealed a patchy mild to moderate inflammation in the colon, in contrast to normal colonic mucosa of age-matched littermate Ints^{ΔPten/ΔPten};Il10^+/− (HeJBir) mice (Figure 7D). By histologic evaluation, we confirmed that Ints^Cont;Il10^−/− (HeJBir) and Ints^{ΔPten/ΔPten};Il10^−/−(HeJBir) mice had similar degree of mild colitis. Since we suggest that the severity of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice is dramatically exacerbated on C57BL/6 background, our data from C3H/HeJBir background mice suggest that the absence of functional TLR4 is associated with less recognizable severity of colitis in Ints^{ΔPten/ΔPten};Il10^−/− (HeJBir) mice.

LPS scavenger, polymyxin B, ameliorates colitis in Ints^{ΔPten/ΔPten}; Il10^−/− mice

We next tested whether blocking LPS-TLR4 engagement would reduce the pathogenesis of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice on C57BL/6 background. Polymyxin B detoxifies the endotoxicity of LPS by specifically interacting with LPS and disrupting the supramolecular structure of LPS²⁴. Thus, Ints^{ΔPten/ΔPten};Il10^−/− mice at 32 days old were fed with polymyxin B (100 ppm)- or vehicle mixed chow for 2 months. Ints^{ΔPten/ΔPten};Il10^−/− mice with polymyxin B-mixed diet revealed substantially improved survival rate compared to the mice with control diet (Figure 7E). At the age of 3 months old, mice fed with control diet had severely inflamed colon. In contrast, age matched mice treated with polymyxin B preserved the integrity of colonic mucosa (Figure 7F). Compared to enhanced degree of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice under control diet, the histological parameters of colitis were greatly reduced in polymyxin B-treated Ints^{ΔPten/ΔPten};Il10^−/− mice. Taken together, these data indicate that inhibiting endogenous LPS-TLR4 engagement not only delays the onset of colitis, but also ameliorates the severity of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice.

DISCUSSION

Emerging evidence indicated that Pten could be an important signaling molecule regulating the inflammation and innate immunity in response to commensal and pathogenic microbes^{14, 18}. Commensal microbes mainly reside in the luminal side of the intestine. Therefore, it is of interest to investigate the impact of Pten in gut homeostasis controlled by the coherence of pro- and anti-inflammatory responses upon enteric microbiota. Our study demonstrated that loss of Pten in the intestinal epithelium induced severe colitis and colon malignancy at very early ages in Il10^−/− mice, indicating that together with defective IL10 signaling, impaired Pten could be a polygenic factor to colitis occurring at very early ages.

In addition to colitis, Ints^{ΔPten/ΔPten};Il10^−/− mice had tumor development in the colon. Inflamed colon of Ints^{ΔPten/ΔPten};Il10^−/− mice had dysplastic lesions that can be localized, flat, or polypoid characteristics of colitis-associated colon cancer (CAC). As chronic UC precedes CAC development in humans, we believe that colon malignancy in Ints^{ΔPten/ΔPten};Il10^−/− mice could result from severe colitis developed at early ages.

Our data showed that only Bacteroides were greatly increased in the fecal sample of Ints^{ΔPten/ΔPten};Il10^−/− mice compared to Ints^Cont;Il10^−/− mouse samples, implying that Bacteroides could be responsible for eliciting spontaneous colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice. This notion is supported by the following evidence: The abundance of Bacteroides is much higher in the colon of UC patients than healthy control subjects²⁵. Oral gavage of intestinal Bacteroides induces colitis in IBD-susceptible mice²⁶. Helicobacter species was known to affect the pathogenesis of colitis in Il10^−/− mice^{27, 28}. However, the presence of Helicobacter was not detected from the fecal samples of the mice (Supplementary Table 1 and Supplementary Table 2). Accordingly, it is reasonable to believe that increased Bacteroides should be associated with inducing colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice.

Gram-negative bacterial LPS is one of the most potent microbial factors in terms of proinflammatory properties. Despite the protective effect from LPS in a mouse epithelial injury model²⁹, numerous studies indicated that LPS in the intestine promoted the intestinal inflammation^{30, 31}. Gram-negative bacteria, moreover, are remarkably enhanced in the intestine of IBD patients^32–34 and accumulated at high concentrations in an inflamed lesion of IBD patients³⁵. These studies implied that LPS might be the microbial antigen eliciting intestinal inflammatory diseases. In agreement with this notion, our data showed that genetic or pharmacologic intervention to LPS-TLR4 engagement greatly delayed the onset of colitis and markedly reduced the severity of colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice (Figure 7). Given that some antibiotic regimens improve clinical outcomes in IBD patients³⁶, polymyxin B-driven effects in Ints^{ΔPten/ΔPten};Il10^−/− mice might be interpreted as a general effect of an antibiotic treatment. But, loss of the LPS specific receptor Tlr4 in Ints^{ΔPten/ΔPten};Il10^−/− mice dramatically inhibited the incidence of colitis. These data, therefore, substantiate that endogenous LPS in the colon plays a critical role in developing spontaneous colitis in Ints^{ΔPten/ΔPten};Il10^−/− mice.

In summary, intestinal epithelial Pten deletion in an Il10 deficient environment triggers spontaneous colitis and colon malignancy at very early ages by inducing gut dysbiosis with increased Gram-negative Bacteroides, in which endogenous LPS from altered gut microflora plays a critical role in colitis development. Our study, therefore, suggests that in addition to defective IL10 signaling, impaired Pten in the intestine may participate in the pathogenesis of early-onset ulcerative colitis, as a polygenic component.

Abbreviations

Ccl2

Chemokine (C-C motif) ligand 2

Cxcl5

chemokine (C-X-C motif) ligand 5

Cxcr2

chemokine (C-X-C motif) receptor 2

Hmox1

heme oxygenase (decycling) 1

Il 5

interleukin 5

Il 18

interleukin 18

Irak 1

interleukin 1 receptor-associated kinase 1

Nos2

nitric oxide synthase 2

Nr3c1

nuclear receptor subfamily 3, group C, member 1

Sirt1

sirtuin 1

Tirap/Mal

toll-interleukin 1 receptor (TIR) domain containing adaptor protein/ Myeloid differentiation factor 88 (MyD88)-adapter-like