Regulation of Spontaneous Intestinal Tumorigenesis Through the Adaptor Protein MyD88

See allHide authors and affiliations

Science  06 Jul 2007:
Vol. 317, Issue 5834, pp. 124-127
DOI: 10.1126/science.1140488


Inflammation is increasingly recognized as an important component of tumorigenesis, although the mechanisms and pathways involved are not well understood. Tumor development is regulated by products of several modifier genes, but instructions for their tumor-specific expression are currently unknown. We show that the signaling through the adaptor protein MyD88 has a critical role in spontaneous tumor development in mice with heterozygous mutation in the adenomatous polyposis coli (APC) gene. We found that MyD88-dependent signaling controls the expression of several key modifier genes of intestinal tumorigenesis and has a critical role in both spontaneous and carcinogen-induced tumor development. This study thus reveals the important role of an innate immune signaling pathway in intestinal tumorigenesis.

Inflammatory responses contribute to carcinogenesis through multiple mechanisms (13). Activation of the transcription factor nuclear factor κB(NF-κB), a key mediator of inflammation, has a critical role in the regulation of tumor development resulting from chronic inflammation or exogenous mutagens (4, 5). NF-κBis activated by multiple stimuli (6), and it is currently unknown which pathway is critically involved in cancer-associated inflammation and the tissue repair response (7). The role of inflammatory and tissue repair responses in spontaneous carcinogenesis, independent of chronic inflammation or administration of exogenous carcinogens, has not yet been characterized. However, signaling through Toll-like receptors (TLRs) of the innate immune system to MyD88 (a signaling adaptor of TLRs) has a critical role in the control of tissue renewal responses (811).

A link between intestinal tissue renewal and tumorigenesis was established when the genetic basis of familial associated polyposis (FAP) was mapped to the APC gene (12). Germline and sporadic mutations in APC occur in >85% of FAP and >80% of sporadic colorectal tumors (12). A mouse model of spontaneous intestinal tumorigenesis was discovered in a forward genetic screen (13). These mice, designated ApcMin/+, have a mutation in the APC gene and develop 60 to 80 intestinal adenomas, mostly at the distal small intestine (14). Given the role of the MyD88 signaling in intestinal tissue renewal and repair, we investigated the role of this pathway in spontaneous intestinal carcinogenesis in ApcMin/+ mice.

To examine a potential role of the MyD88 signaling pathway in spontaneous intestinal tumorigenesis, we generated ApcMin/+ mice on MyD88-sufficient and -deficient backgrounds and analyzed sex- and age-matched cohorts. On average, ApcMin/+ mice die within 6 months of age from complications of intestinal tumors (13). In contrast, mortality of ApcMin/+ MyD88–/– mice was dramatically reduced as compared with ApcMin/+ littermate controls (Fig. 1A). We investigated the red blood cell (RBC) status of these mice, a marker of intestinal tumorigenesis, and found that the anemia observed in ApcMin/+ mice was significantly ameliorated in ApcMin/+ MyD88–/– mice (Fig. 1B). MyD88-dependent signaling therefore contributes substantially to the severe mortality and morbidity caused by inactivation of APC.

Fig. 1.

Effects of MyD88-deficiency on mortality and blood loss in ApcMin/+ mice. (A) ApcMin/+ MyD88+/+ or MyD88+/– (ApcMin/+; N = 32) and ApcMin/+ MyD88–/– (ApcMin/+ MyD88–/–; N = 29) mice were passively followed for long-term survival. Difference in survival was determined significant by the Mantel-Haenszel/Log-rank test. (B) At time of sacrifice, hematocrit of peripheral blood of age- and sex-matched mice (20 to 22 weeks old) was determined. Statistical analysis was performed using the Mann-Whitney U test.

The number of visible polyps (≥0.5 mm in diameter) was next quantified by stereoscopic microscopy. The number of macroadenomas in the small intestines or colon of ApcMin/+ MyD88–/– mice was reduced compared with that in ApcMin/+ controls (Fig. 2A and fig. S1). Loss of MyD88 decreased the number of small intestinal tumors in all regions of the small intestine (Fig. 2B), most notably in the distal small intestine where the majority of tumor formation in ApcMin/+ mice occurs (13) (Fig. 2B).

Fig. 2.

Characterization of tumors in ApcMin/+ mice in the presence or absence of MyD88. (A) (Left) The number of visible polyps in the small intestine (≥0.5 mm in diameter) was quantified by stereoscopic microscopy in age (20 to 33 weeks old)- and sex-matched ApcMin/+ and ApcMin/+ MyD88–/– mice. (Right) The number of visible polyps in the large intestine in age (20 to 28 weeks old)- and sex-matched ApcMin/+ and ApcMin/+ MyD88–/– mice. (B) (Left) Frequency of the polyps in ApcMin/+ and ApcMin/+ MyD88–/– mice stratified by small intestinal region. (Right) Size range of tumors from age- and sex-matched ApcMin/+ and ApcMin/+ MyD88–/– mice. Error bars, ±SEM. **, P < 0.01; ***, P < 0.001 (compared with ApcMin/+); Mann-Whitney U test.

The number of polyps in ApcMin/+ MyD88–/– mice was reduced, and the polyps that were present, both at the proximal and distal small intestine, were smaller in size than those present in age-matched ApcMin/+ mice (Fig. 2B and fig. S1). This finding suggested that the MyD88-dependent signaling pathway regulated intestinal tumor growth in ApcMin/+ mice. Because the size of tumors can be influenced by the balance of cell proliferation and death within the tumor mass, we investigated the proliferative and apoptotic rates within size-matched tumors from ApcMin/+ and ApcMin/+ MyD88–/– mice (15). Polyps from both genotypes showed similar proportions of proliferating cells (Fig. 3A). However, higher numbers of apoptotic cells were found in tumors from ApcMin/+ MyD88–/– mice than in samples from ApcMin/+ mice (Fig. 3B and fig. S2A). Thus, MyD88 may contribute to tumor progression through regulation of apoptosis.

Fig. 3.

MyD88-dependent and -independent epithelial homeostasis in intestinal tumors and formation of early neoplastic lesions. (A) 5-bromodeoxyuridine (BrDU), (B) (left panel) diamidino-phenylindole (DAPI), and (right panel) terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) stain of intestinal tissue from WT and MyD88–/– mice and size-matched tumors from ApcMin/+ and ApcMin/+ MyD88–/– mice. (C) Representative histologic microphotos of distal small intestine from littermate mice at 4.5 weeks of age. Haemotoxylin and eosin stain; images are at 200x magnification. (D) Representative β-catenin staining of small intestine of age (20 weeks)- and sex-matched mice. Images are at 400x magnification. Arrowheads indicate nuclear localization of β-catenin at cells present at the crypt base in WT and MyD88–/– mice. Increased cytoplasmic and nuclear β-catenin staining can be seen in numerous cells in tumors of ApcMin/+ and ApcMin/+ MyD88–/– mice.

Intestinal tumorigenesis is a stepwise process (12). At the first step, referred to as initiation, genetic changes in either oncogenes or tumor suppressors lead to the transformation of normal epithelium to dysplastic cells, resulting in the formation of microadenomas. Tumor progression results in the expansion of microadenomas and an increase in tumor size (4, 12). The decrease in polyp frequency and size observed in ApcMin/+ MyD88–/– compared with those in ApcMin/+ mice indicated that MyD88 regulates initiation or the progression from microadenomas to macroadenomas. The cancer initiation step in both FAP patients and ApcMin/+ mice involves the loss of heterozygosity of the APC gene and results in the formation of microadenomas (12). To investigate whether MyD88 affected early events in neoplasia, we analyzed ApcMin/+ and ApcMin/+ MyD88–/– mice for the presence of microadenomas. Four- to five-week-old ApcMin/+ mice, an age at which the majority of microadenoma formation occurs (13), were observed to have similar frequencies [ApcMin/+: 2.06 ± 1.63 (SEM); ApcMin/+ MyD88–/–: 2.19 ± 0.95; microadenomas per Swiss roll section of entire small intestine; N = 4 mice per genotype] and morphology of microadenomas regardless of the presence or absence of MyD88 (Fig. 3B). Thus, MyD88 appears not to influence the formation of early neoplastic lesions in intestinal tumorigenesis but rather contributes to tumor growth and progression. In addition, macroadenomas of both ApcMin/+ and ApcMin/+ MyD88–/– mice showed an equivalent increase in cytoplasmic and nuclear accumulation of β-catenin (Fig. 3D) and frequencies of cells with nuclear β-catenin (fig. S2B), suggesting that MyD88-dependent signaling does not affect β-catenin nuclear translocation in dysplastic epithelium.

We further investigated the role of MyD88 in intestinal tumorigenesis by comparing gene expression in size-matched polyps from ApcMin/+ and ApcMin/+ MyD88–/– mice. This analysis revealed two classes of genes that are overexpressed in tumors compared with normal tissue. One class of genes was MyD88-independent in that their expression was similarly upregulated in ApcMin/+ and ApcMin/+ MyD88–/– tumors (fig. S3). This class of genes included, for example, c-Myc, which is known to be a direct target of β-catenin (16). These tumor-upregulated, but MyD88-independent, genes serve as internal controls for tumor tissue sampling.

However, in these same size-matched tumors, we found many genes that showed increased expression in a tumor-specific and MyD88-dependent manner (Fig. 4). These included genes encoding positive regulators of intestinal tumorigenesis, such as cyclooxygenase-2 (COX-2) (17, 18), matrix metalloprotease (MMP) 7 (19), and cytosolic phospholipase A2 (cPLA2) (20, 21) (Fig. 4). MyD88-dependent signaling was also required for the tumor-specific expression of genes encoding keratinocyte growth factor 2 (KGF2) [also called fibroblast growth factor 10 (FGF10)], CD44, MMP10, insulin-like growth factor binding protein (IGFBP) 5, insulin-like growth factor–1 (IGF1), regenerating gene product IIIβ (RegIIIβ), and various cytokines and chemokines such as keratinocyte derived chemokine (KC), interleukin (IL)–6, and IL-1β (Fig. 4 and fig. S4), all of which are components of the tissue repair response and have been either shown to be, or implicated as, positive regulators of tumorigenesis. Thus, MyD88-dependent signaling is critically involved in the tumor-specific induction of modifier genes of intestinal tumorigenesis, as well as genes involved in intestinal tissue repair. Interestingly, MyD88-dependent genes in ApcMin/+ tumors included both classical NF-κB–regulated genes (KC and IL-6) and genes that are regulated by other signaling pathways (FGF10, RegIIIβ, IGF1, and IGFBP5).

Fig. 4.

MyD88 regulation of modifiers of tumor progression. RNA was prepared from the distal small intestine of normal intestine of WT and MyD88–/– mice and size-matched tumors of ApcMin/+ and ApcMin/+ MyD88–/– mice. cDNA was prepared, and quantitative polymerase chain reaction was performed. Expression of candidate genes was normalized to hypoxanthine-guanine phosphoribosyltransferase (HPRT) and relative induction compared with levels in the intestines of WT mice. Data are representative of five (tumor) tissue samples each isolated from different mice. Error bars, ± SEM. *, P < 0.05; **, P < 0.01 (compared to ApcMin/+); Mann-Whitney U test.

The tumor modifier MMP7 is expressed only in dysplastic epithelium (19), whereas another tumor modifier, COX-2, is expressed by multiple cell types in ApcMin/+ tumors, including macrophages, fibroblasts, endothelial cells, and epithelium (22). Overexpression of both MMP7 and COX-2 is MyD88-dependent in tumors (Fig. 4 and fig. S5). This suggests that MyD88 regulates intestinal tumor progression by induction of tumor modifiers in both epithelial and stromal cells, similar to the role of IKKβ-dependent signaling in chronic inflammation-dependent intestinal tumorigenesis (4).

We observed similar frequencies of leukocytes, such as macrophages, both in normal mucosa and different sized tumors of ApcMin/+ mice (23) (fig. S6), indicating that differences in gene expression observed in ApcMin/+ and ApcMin/+ MyD88–/– tumors did not result from differences in leukocyte recruitment.

In another model of intestinal (colonic) tumorigenesis [the administration of the carcinogen azoxymethane (AOM) (24)], incidence of tumor formation was again decreased in MyD88–/– mice compared with that in wild-type (WT) mice (fig. S7). Thus, the MyD88-dependent signaling pathway critically contributes to carcinogeninduced colonic tumorigenesis as well.

MyD88 functions in signaling downstream of both TLR and IL-1 receptor families. Although both types of receptors may contribute to the phenotype of ApcMin/+ MyD88–/– mice, the function of IL-1 is likely to be downstream of TLRs because expression of IL-1 is dependent on TLR-MyD88 signaling (25) (fig. S4). The nature of the trigger for TLR signals in ApcMin/+ tumors is currently unknown.

The MyD88 signaling pathway triggered by the TLR/IL-1R family of receptors functions in the control of inflammation both in host defense from infection and in tissue repair (8, 9, 11, 26). Many genes involved in tissue repair (27) are induced in a TLR and MyD88-dependent manner upon intestinal injury (8, 9) (fig. S8). Thus, the requirement for the MyD88 signaling pathway in intestinal tumor progression could result from the essential role of this pathway in the induction of a tissue-repair program. This would support the notion that tumor growth is an abnormal form of a continuous and unregulated state of tissue repair (12, 28, 29). Indeed, unlike the self-limiting repair in normal tissues, the tumor-associated tissuerepair response both promotes and is triggered by the tumor growth, thus falling into the self-perpetuating and uncontrolled state of tissue repair.

Our findings establish that the MyD88 signaling pathway downstream of members of the TLR and IL-1R family has a critical role in intestinal tumorigenesis. Tumor-associated expression of key modifier genes, as well as many other positive regulators of intestinal tumor progression and tissue repair, are critically dependent on the MyD88 signaling pathway. It should be noted however, that there are substantial differences between human FAP and ApcMin/+ tumorigenesis. In addition, the ApcMin/+ model is sensitive to a variety of genetic alterations. Therefore, it will be important to determine in future studies whether the MyD88 pathway contributes to intestinal tumorigenesis in humans.

Supporting Online Material

Materials and Methods

Figs. S1 to S8


References and Notes

Stay Connected to Science

Navigate This Article