The contributions of TGF-β signaling to cancer are complex but involve the inflammatory microenvironment as well as cancer cells themselves. In mice encoding a TGF-β mutant that precludes its binding to the latent TGF-β binding protein (Tgfb1−/C33S), we observed multiorgan inflammation and an elevated incidence of various types of gastrointestinal solid tumors due to impaired conversion of latent to active TGF-β1. By genetically eliminating activators of latent TGF-β1, we further lowered the amount of TGF-β, which enhanced tumor frequency and multiorgan inflammation. This model system was used to further investigate the relative contribution of TGF-β1 to lymphocyte-mediated inflammation in gastrointestinal tumorigenesis. Toward this end, we generated Tgfb1−/C33S;Rag2−/− mice that lacked adaptive immune function, which eliminated tumor production. Analysis of tissue from Tgfb1−/C33S mice indicated decreased levels of P-Smad3 compared with wild-type animals, whereas tissue from Tgfb1−/C33S;Rag2−/− mice had normal P-Smad3 levels. Inhibiting the inflammatory response normalized levels of interleukin (IL)-1β and IL-6 and reduced tumor cell proliferation. In addition, Tgfb1−/C33S;Rag2−/− mice exhibited reduced paracrine signaling in the epithelia, mediated by hepatocyte growth factor produced by gastric stroma. Together, our results indicate that many of the responses of the gastric tissue associated with decreased TGF-β1 may be directly or indirectly affected by inflammatory processes, which accompany loss of TGF-β1, rather than a direct effect of loss of the cytokine. Cancer Res; 74(9); 2642–51. ©2014 AACR.

TGF-β1 was described as a stimulator of mesenchymal cell growth (1). However, subsequent studies revealed that this cytokine also has an important function as an inhibitor of epithelial cell proliferation. As such, TGF-β1 acts as a tumor suppressor, a view supported by data from numerous human and animal studies demonstrating that inactivating mutations in the TGF-β signaling pathway result in different types of carcinomas (2). The multiple and sometimes apparently contradictory actions of TGF-β1 are also evident by the fact that although TGF-β1 is a growth suppressor in early-stage tumors, it may have a promoting effect in late-stage tumors (3–6). The reasons for this difference in response to TGF-β are unclear but may relate to the acquisition of mutations by later-stage tumors that block TGF-β's growth-inhibiting activity but accentuate its ability to initiate an epithelial-to-mesenchymal transition, a process that enhances the invasive and metastatic phenotypes (7).

In addition to its growth-regulatory capacity, a second important property of TGF-β1 is the promotion of T regulatory cell maturation and the consequent suppression of inflammation (8). Thus, TGF-β1 null mutations in mice are associated with severe multiorgan inflammation and rapid death after weaning (9, 10). Because inflammation and inflammatory mediators promote tumor formation, loss of TGF-β or loss of TGF-β signaling might have a dual effect upon tumorigenesis both by relieving the growth-suppressive activity and by initiating the stimulation associated with inflammation.

There are three TGF-β genes (Tgfb1, 2, and 3). Mutations of TGF-β2 or TGF-β3 genes are not known to be associated with tumor production (11–13). Loss of TGF-β1 results in severe multiorgan inflammation and early death (9, 10). Tumor generation has also been reported in TGF-β1–null mice but has been difficult to study because of rapid death after weaning due to the strong phenotype.

To circumvent this problem, we generated TGF-β1 hypomorphic mice by permuting the mechanism used to generate active TGF-β1 (14). Although all TGF-βs are initially produced as homodimeric proproteins and the bond between the prosequence and the mature molecule is cleaved within the cell, the propeptide, named the latency-associated protein (LAP), remains associated with the signaling molecule after secretion (15). This complex of TGF-β and LAP is called the small latent complex (SLC). The continued interaction of cytokine with its propeptide renders TGF-β latent, because in this form TGF-β cannot interact with its receptor. A variety of mechanisms for the activation, i.e., the liberation of TGF-β from the SLC, of latent TGF-β have been described, and all involve the perturbation of the binding between TGF-β and LAP by integrins, proteases, thrombospondin 1, or reactive oxygen (15). The TGF-β–LAP complex is usually linked to a second gene product, a latent TGF-β binding protein (LTBP), through cysteine residues in LAP forming the large latent complex (LLC). Four LTBP genes exist, three of which, LTBP-1, -3, and -4, bind to TGF-β LAP isoforms (16). LTBPs target the latent TGF-β complex to the matrix and may participate indirectly in activation of the latent complex by the integrin αvβ6 (15). The αvβ6-mediated activation process depends upon the integrin binding to an RGD sequence in the TGF-β1 propeptide and the latent complex binding to the matrix via the LTBP. When the cell applies force to the complex by cytoskeleton contraction mediated by the integrin, tethering to the matrix via the LTBP results in distortion of the TGF-β1–LAP complex and release of active TGF-β1 (17, 18). Mutation of either the RGD to a non-integrin binding sequence (RGE) or deletion of the matrix-binding regions in LTBP blocks latent TGF-β activation (19, 20).

To perturb latent TGF-β activation, we mutated the TGF-β1 LAP cysteine (residue 33), which binds LTBP, to serine (14). Thus, TGF-β1 produced by these mutant Tgfb1C33S/C33S mice is unable to form a complex with any LTBP. With this approach, we demonstrated that the binding of TGF-β1 LAP to an LTBP is crucial for proper TGF-β1 function, as Tgfb1C33S/C33S mice resemble Tgfb1−/− mice with respect to shortened lifespan, multiorgan inflammation, and lack of epidermal Langerhan's cells (14). However, the shortened lifespan and multiorgan inflammation of Tgfb1C33S/C33S mice are not as severe as with Tgfb1−/− mice, suggesting that Tgfb1C33S/C33S animals are hypomorphic rather than null for TGF-β1. The amount of active TGF-β1 is decreased as measured by both indirect and direct assays. The mutant mice develop tumors of the stomach, colon, and rectum at an overall frequency of 20% within 12 to 16 weeks. Tumor generation is enhanced upon further reduction of the TGF-β1 level by elimination of the αvβ8 integrin, which is an activator of soluble SLC, or by generating Tgfb1−/C33S mice, which have one Tgfb1 null allele and one C33S allele (21). With this approach, the tumor incidence in Tgfb1−/C33S mice is as high as 73%. It was notable that the Tgfb1−/C33S genotype yielded high numbers of gastric tumors, an effect not observed in other models of TGF-β1 deficiency (14, 22). Concomitant with the increase in tumors, we observed a decrease in TGF-β signaling and a concordant increase in the degree of inflammation (21). Therefore, Tgfb1C33S/C33S mice provide a system in which the level of active TGF-β1 can be manipulated genetically and the properties of selective phenotypes examined over an extended period of time.

We hypothesized that we could use Tgfb1−/C33S mice to examine the contribution of inflammation to tumorigenesis by constructing mice with Tgfb1−/C33S alleles combined with Rag2−/− alleles. These animals would produce little active TGF-β1 and no mature T or B cells; therefore, by monitoring tumor frequency, we could discern the contribution of lymphocyte-mediated inflammation to tumor production. In this article, we describe the phenotypes of these compound mutant mice. Because of the paucity of genetic models of gastric cancer, we focused our analysis on gastric tumors in Tgfb1−/C33S mice, although there were additional rectal–anal tumors in the mutant animals. We show that Tgfb1−/C33S;Rag2−/− mice display decreased levels of inflammation and a severe reduction in tumorigenesis compared with Tgfb−/C33S mice. We found that in the compound mutant animals, the loss of gastric tumor production is accompanied by reductions in interleukin (IL)-1β and IL-6 cytokine production as well as decreased signaling through the hepatocyte growth factor (HGF) receptor cMet, and cRon.

Mouse lines

The generation and phenotypes of Tgfb1−/C33S mice have been described in ref. 21. Tgfb1−/C33S;Rag2−/− mice were generated by breeding Rag2−/− (TACONIC; ref. 23) on a C57BL/6 background with Tgfb1+/C33S or Tgfb1−/+ mice on a predominately C57BL/6 background. The offspring, Tgfb1+/C33S;Rag2+/− and Tgfb1−/+;Rag2+/− mice, were crossed to generate Tgfb1−/C33S;Rag2−/− mice. Tissues were collected at 3 months of age unless otherwise noted. All procedures were conducted according to the regulations of the New York University Langone Medical Center Institutional Animal Care and Use Committee.

TGF-β assays

The levels of total and active TGF-β1 were measured in sera using a mouse TGF-β1 duo-set (DY1679) according to the manufacturer's instructions (R&D Systems).

Histology

Stomach and rectal tissues were dissected, opened, and the contents flushed with ice-cold saline. The tissue was immersed and fixed in 10% formalin and embedded in paraffin. Each tissue was serial sectioned and a minimum of five hematoxylin and eosin (H&E)–stained sections were examined that were at least 100-μm apart.

Pathologic scoring of inflammation

A scoring system in which three parameters were monitored in individual histologic sections was used for quantification of inflammation (14, 21). The three parameters were: (i) inflammation, measured as cell infiltration of the mucosa by mixed populations of inflammatory cells and edema; (ii) hyperplasia, monitored by the degree of hyperplasia of the mucosal epithelium including lengthening of crypts, increased density of epithelial cells and crypts and thickening of mucosa; and (iii) necrosis/ulceration, monitored by examining for necrosis of mucosal epithelial cells with attenuation, erosion, or ulceration of the epithelial barrier. A scoring system for each parameter was used in which 0 = within normal limits, 1 = minimal to mild, 2 = moderate and 3 = severe. Two individuals monitored slides in a blinded fashion.

Immunohistochemistry

Preparation of tissues, sectioning, and antibody staining were performed as described in ref. 14. Antibodies used were purchased from the following sources: calponin (EP798Y, ab46794), CD3 (ab5690-100), F4/80 (CI:A3-1, ab6640), and phospho-SMAD3 (phospho S423+S425, ab51451) were from Abcam; HGF (N-terminal, LSB4012) was from LifeSpan BioSciences; Ki67 (NCL-Ki67p) was from Leica Biosystems; phospho-cMet (pYpYpY1230/1234/1235, 44888G) was from Invitrogen; phospho-ERK (THR202/TYR204, 4370p), and phospho-STAT3 (9145), were from Cell Signaling Technology; phospho-histone H3 (Ser10, 06-570), was from Millipore; and P-MSPR/RON (AF1947) was from R&D Systems.

KI67 and P histone H3 analysis

Staining and analysis were performed as described in ref. 21.

Image analysis

Stained slides were scanned using a Leica SCN400F. Slides were analyzed using Ariol image analysis system. The settings were adjusted in a process called “training of classifiers” to analyze objects of specific color, size, and shape, and to distinguish positive from negative cells. Only areas of glandular epithelium were included. Images of all sections were analyzed using the settings established during the training. To quantify the degree of inflammation, positive areas for F4/80 and CD3 staining in the submucosal layer and above were measured using ImageJ software 1.47v (24).

Quantitative real-time PCR analysis

Half of the tissue was immersed in RNAlater RNA Stabilization Reagent (Qiagen) overnight at 4°C. The tissue was removed, frozen in liquid nitrogen, and stored at −80°C. RNA was extracted using RNeasy Mini Kit (Qiagen). Reverse transcription was performed using the SuperScript III Reverse Transcriptase (Invitrogen; 50°C, 60 minutes) according to the instructions. The resulting cDNA was used for quantitative real-time PCR (qRT-PCR) analysis using specific primers and the QuantiFast SYBR Green PCR Kit (Qiagen) on an iCycler Thermal Cycler (Bio-Rad). Expression of each target transcript was quantified by use of calibration curves and normalized to the expression of β-actin. The primers are described in Supplementary Table S1.

Statistical analysis

Descriptive statistics were performed with StatView J-5.0 program (SAS Institute). The Kaplan–Meier method was used to estimate survival curves. The log-rank statistic was used to compare the overall survival distributions. Errors were computed in Fig. 1 as SE and in all other figures as SD.

Figure 1.

Inflammation in Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− mice. A, the weights of the stomach and colorectum standardized by total body weight were increased in Tgfb1−/C33S mice but reduced to normal upon introduction of the Rag2−/− mutation. N = 10 per group. B and C, inflammation scores in the stomach and rectum were reduced to WT levels upon introduction of the Rag2−/− gene. N = 16–20 per group. D–I, sections from Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− stomach tissue stained with H&E (D and G), antibody to macrophage marker F4/80 (E and H), and antibody to CD3, which marks T cells (F and I). Addition of the Rag2−/− mutation to Tgfb1−/C33S yields significant decreases in numbers of macrophages and T cells in stomach tissue, especially in the mucosa.

Figure 1.

Inflammation in Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− mice. A, the weights of the stomach and colorectum standardized by total body weight were increased in Tgfb1−/C33S mice but reduced to normal upon introduction of the Rag2−/− mutation. N = 10 per group. B and C, inflammation scores in the stomach and rectum were reduced to WT levels upon introduction of the Rag2−/− gene. N = 16–20 per group. D–I, sections from Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− stomach tissue stained with H&E (D and G), antibody to macrophage marker F4/80 (E and H), and antibody to CD3, which marks T cells (F and I). Addition of the Rag2−/− mutation to Tgfb1−/C33S yields significant decreases in numbers of macrophages and T cells in stomach tissue, especially in the mucosa.

Close modal

TGF-β1 production in Tgfb1−/C33S;Rag2−/− mice

When Tgfb1+/C33S;Rag2+/− and Tgfb1−/+;Rag2+/− mice were crossed, Tgfb1−/C33S;Rag2−/− mice were born at the expected frequency (Supplementary Table S2). Although only 55% of the Tgfb1−/C33S mice survived at 12 weeks, survival of Tgfb1−/C33S;Rag2−/− mice was considerably better, as by 12 weeks of age 72% of the animals were alive (Supplementary Fig. S1A). By 20 weeks of age, only 33% of the Tgfb1−/C33S mice survived, whereas approximately 68% of the Tgfb1−/C33S;Rag2−/− mice survived. Interestingly, the death of Tgfb1−/C33S mice seemed to occur in two phases. The first phase, between weeks 2–4, displayed a steep decline with 30% of the animals dying during this period. The second phase, after week 4 and continuing for at least 19 weeks, displayed a more gradual decline. During the first 4 weeks, Tgfb1−/C33S;Rag2−/− mice displayed survival kinetics similar to those of Tgfb1−/C33S animals, but were protected against the second phase of decline. Thus, two processes seem to be responsible for the death of Tgfb1−/C33S mice, an early process, which occurs even in the absence of mature T and B cells, and a second process related to inflammation, which is almost completely negated upon elimination of RAG2 activity. The actual causes of death at these two times remain unknown.

Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− animals weighed less than wild-type (WT) or Rag2−/− animals (Supplementary Fig. S1B). The weights of the spleen and thymus standardized to total body weight in Rag2−/− and Tgfb1−/C33S;Rag2−/− mice were decreased compared with those of WT mice, consistent with a lack of T and B cells, whereas the weight and size of the spleen in Tgfb1−/C33S mice were increased compared with WT, presumably because of inflammation (Supplementary Fig. S1C and S1D).

Inflammation

As expected, Tgfb1−/C33S;Rag2−/− mice displayed less inflammation than Tgfb1−/C33S animals by several measures. Tgfb1−/C33S;Rag2−/− stomachs and colorectums weighed significantly less than corresponding Tgfb1−/C33S tissues when normalized to total body weight (Fig. 1A). The degree of inflammation, as measured by a system that scored inflammatory cell infiltration and edema, hyperplasia, and necrosis, was elevated in Tgfb1−/C33S mice but was reduced to normal in Tgfb1−/C33S;Rag2−/− animals (Fig. 1B and C). There was little hyperplasia or necrosis, but this may be a feature of gastric tissue, as hyperplasia and necrosis occurred in the rectal–anal tissue (data not shown). Immunohistologic analysis of stomach tissue revealed increased numbers of macrophages and T cells in both the lamina propria and the submucosa of Tgfb1−/C33S mice compared with Tgfb1−/C33S;Rag2−/− mice (Fig. 1D–F). In the stomach of double-mutant animals, there was an absence of T cells and fewer macrophages in the tissues of Tgfb1−/C33S;Rag2−/− mice (Fig. 1D–I). Similar results were observed for the rectal–anal tissue (Data not shown). The degree of inflammation was quantified by measuring the positively stained areas in slides of tissue samples from animals of the three genotypes. The ratios of the positively stained areas for the macrophage marker F4/80 for WT, Tgfb1−/C33S, and Tgfb1−/C33S;Rag2−/− mice were 1:10.0:1.4, whereas the ratios for the T-cell marker CD3 were 1:24.3:2.1.

TGF-β levels

The amount of total TGF-β1 was decreased in Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− sera compared with WT (Supplementary Fig. S1E), as expected. However, total TGF-β1 was significantly lower than the 50% expected for these two mutant genotypes. The reason for this is not clear but was also observed in our earlier studies (21). It is possible that the SLC is less stable in serum than the LLC. The amount of active TGF-β1 in Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− sera was significantly less than that found in WT or Rag2−/− mice (Supplementary Fig. S1F), as reported in ref. 21.

To determine whether TGF-β signaling was affected by the Rag2 mutation, we performed two types of experiments. First, we measured P-Smad3 levels in stomach tissue of WT, Tgfb−/C33S, and Tgfb1−/C33S;Rag2−/− animals using immunohistochemistry (IHC). P-Smad3 levels are surrogate markers for TGF-β signaling, as phosphorylation of Smad3 is part of the TGF-β signaling cascade (25). The amount of P-Smad staining was clearly decreased in Tgfb1−/C33S compared with WT (Fig. 2A, B, and D). P-Smad3 levels were restored to the level of WT in Tgfb1−/C33S;Rag2−/− tissue (Fig. 2C and D). This result was unexpected, as we assumed TGF-β activation and, therefore, TGF-β signaling would remain low in the Tgfb1−/C33S;Rag2−/− mice. It is possible that in the absence of T and B cells, there is compensation by other ligands that might increase P-Smad3 levels.

Figure 2.

TGF-β signaling in Tgfb1−/C33S gastric tissue. To monitor TGF-β signaling, P-Smad3–positive cells and expression levels of five TGF-β responsive genes were quantified. A–C, P-Smad3 levels decreased in Tgfb1−/C33S tissue and this effect was reversed upon introduction of the Rag2−/− gene. D, quantification of positive cells indicated that there was a 3-fold difference among mice of the different genotypes. N = 3 per group. E, the expression of three genes, C-Myc, Id1, and Id2, whose expression is negatively regulated by TGF-β, displayed enhanced expression in Tgfb1−/C33S animals that was reversed in Tgfb1−/C33S;Rag2−/− mice. Two other genes, Pai1 and Ctgf, whose expression is positively regulated by TGF-β, showed no significant changes dependent upon the Tgfb1−/C33S mutation. N = 6 per group.

Figure 2.

TGF-β signaling in Tgfb1−/C33S gastric tissue. To monitor TGF-β signaling, P-Smad3–positive cells and expression levels of five TGF-β responsive genes were quantified. A–C, P-Smad3 levels decreased in Tgfb1−/C33S tissue and this effect was reversed upon introduction of the Rag2−/− gene. D, quantification of positive cells indicated that there was a 3-fold difference among mice of the different genotypes. N = 3 per group. E, the expression of three genes, C-Myc, Id1, and Id2, whose expression is negatively regulated by TGF-β, displayed enhanced expression in Tgfb1−/C33S animals that was reversed in Tgfb1−/C33S;Rag2−/− mice. Two other genes, Pai1 and Ctgf, whose expression is positively regulated by TGF-β, showed no significant changes dependent upon the Tgfb1−/C33S mutation. N = 6 per group.

Close modal

In a second approach, we examined the expression of TGF-β–sensitive genes in mice of the different genotypes (14, 26). The expression of three genes, C-Myc, Id1, and Id2, normally suppressed by TGF-β1, showed increased expression in Tgfb1−/C33S mice compared with either WT or Rag2−/− animals (Fig. 2E). Expression of Id1 and Id2 was significantly reduced in Tgfb1−/C33S;Rag2−/− mice compared with Tgfb1−/C33S, suggesting that the observed enhancement of gene expression was related to the inflammatory state. The precise mechanism for enhancement is not known, but could be the result of enhanced BMP signaling. The differences in C-Myc expression between the Tgfb1−/C33S;Rag2−/− and Tgfb1−/C33S genotypes were not significant. Two other genes, Pai1 and Ctgf, whose expression is normally stimulated by TGF-β1, displayed no increase in the Tgfb1−/C33S mutants, consistent with lack of an increase in active TGF-β1 in the mutant genotypes (Fig. 2E).

Tumor incidence

Tgfb1−/C33S mice had an overall incidence of stomach and rectal tumors of 40% between 16 and 19 weeks (Table 1). All tumors were well-differentiated adenocarcinomas (Fig. 3) and all but one of the stomach tumors were invasive. No metastases were observed. The tumor incidence was slightly less than observed previously (21). This may reflect the genetic background as the animals in this study were in a primarily C57BL/129 background, whereas animals in our earlier studies were outbred. No tumors were found in the stomach or rectum of Tgfb1−/C33S;Rag2−/− mice (Table 1). We therefore concluded that decreasing the inflammatory response impeded tumor development.

Figure 3.

Gastric and rectal–anal tumors in Tgfb1−/C33S mice. Representative gastric tissue sections stained with H&E are shown. A, gastric tumor with abnormal gland structure. The arrow points to immune cells infiltrating the stomach tissue. B, higher magnification of the boxed area in A. C, gastric tumor with submucosal invasion and polyp-like lesion. D and E, rectal–anal tumor. Invasive cells are clearly seen in the submucosa.

Figure 3.

Gastric and rectal–anal tumors in Tgfb1−/C33S mice. Representative gastric tissue sections stained with H&E are shown. A, gastric tumor with abnormal gland structure. The arrow points to immune cells infiltrating the stomach tissue. B, higher magnification of the boxed area in A. C, gastric tumor with submucosal invasion and polyp-like lesion. D and E, rectal–anal tumor. Invasive cells are clearly seen in the submucosa.

Close modal
Table 1.

Tumor incidence (16–19 weeks)

GenotypeNStomachRectumTotal (%)
WT 15 0 (0) 
Rag2−/− 15 0 (0) 
Tgfb1−/C33S 15 6a (40)b 
Tgfb1−/C33S;Rag2−/− 15 0 (0) 
GenotypeNStomachRectumTotal (%)
WT 15 0 (0) 
Rag2−/− 15 0 (0) 
Tgfb1−/C33S 15 6a (40)b 
Tgfb1−/C33S;Rag2−/− 15 0 (0) 

aIncluding one double tumor–bearing mouse.

bP < 0.05 versus Tgfb1−/C33S;Rag2−/−.

Prevention of tumors in Tgfb1−/C33S mice that harbor the Rag2−/− mutation contrasts with a report by Engle and colleagues (22), who found that Tgfb1−/−;Rag2−/− mice developed tumors. However, we found mainly gastric and rectal–anal tumors, whereas Engle and colleagues observed colon and cecal tumors. Engle and colleagues also described high levels of granulocytes in early tumor stages, but we failed to observe this. Indeed, we detected very little hyperplasia. There are several technical differences between our model and that of Engle and colleagues that might account for the apparently conflicting results. Perhaps the most significant difference is that our animals are TGF-β1 hypomorphs, whereas the animals that Engel and colleagues studied were TGF-β1 nulls. Thus, tumor type may reflect the absolute level of TGF-β1. In addition, Engle and colleagues used 129 x CF-1 mice, whereas our mice were on a primarily C57BL6 background. Individual mouse strains may exhibit unique responses to TGF-β1 loss due to differences in modifier genes.

We quantified the numbers of proliferating cells by scoring for Ki67-positive cells (Fig. 4). Gastric tissue samples from WT mice had low numbers of positive cells (Fig. 4A and D), whereas the samples from Tgfb1−/C33S mice had a high number of Ki67-reactive cells (Fig. 4B and D). Thus, decreasing TGF-β1 levels results in enhanced proliferation, a situation reversed in the Tgfb1−/C33S;Rag2−/− sample (Fig. 4C and D). To confirm these results, we also stained stomach epithelium with an antibody against phosphohistone H3 (P-HH3), a proliferation-specific antigen. Again, Tgfb1−/C33S samples contained more positive cells than WT, and this number was reduced to normal in Tgfb1−/C33S;Rag2−/− mice (Fig. 4E–H). Therefore, increased cell proliferation as measured by two markers, correlated with the degree of inflammation. The potential complexity of the contribution of inflammation to tumor production is further highlighted in a recent article that reported that tumor production in Smad3−/− mice was enhanced in Smad3−/−;Rag2−/− mice (27).

Figure 4.

Cell proliferation in Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− mice. A–C, sections of gastric tissue stained with an antibody to Ki67, which reveals proliferating cells. (A, WT; B, Tgfb1−/C33S; C, Tgfb1−/C33S;Rag2−/−). There is a dramatic increase in the number of positive cells in sample B. This increase is reduced upon introduction of the Rag2−/− mutation (C). When the percentage of positive cells was computed (D), the change in proliferation was evident. N = 3 per group. E–G, cells stained for the proliferation-specific antigen P-HH3. The number of positive cells was greater in the Tgfb1−/C33S sample (F) than in either WT or Tgfb1−/C33S;Rag2−/− samples (E and G), indicating that removal of T and B cells decreased the number of dividing cells. When the percentage of dividing cells was computed (H), the changes between the genotypes were obvious. N = 3 per group.

Figure 4.

Cell proliferation in Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− mice. A–C, sections of gastric tissue stained with an antibody to Ki67, which reveals proliferating cells. (A, WT; B, Tgfb1−/C33S; C, Tgfb1−/C33S;Rag2−/−). There is a dramatic increase in the number of positive cells in sample B. This increase is reduced upon introduction of the Rag2−/− mutation (C). When the percentage of positive cells was computed (D), the change in proliferation was evident. N = 3 per group. E–G, cells stained for the proliferation-specific antigen P-HH3. The number of positive cells was greater in the Tgfb1−/C33S sample (F) than in either WT or Tgfb1−/C33S;Rag2−/− samples (E and G), indicating that removal of T and B cells decreased the number of dividing cells. When the percentage of dividing cells was computed (H), the changes between the genotypes were obvious. N = 3 per group.

Close modal

Cytokine signaling

To determine the potential inflammatory stimuli contributing to tumor generation, we used qRT-PCR to screen stomach tissue of the different genotypes for a number of cytokines associated with tumor generation (Fig. 5A). The assay revealed increased levels of IL-1β and IL-6 expression in Tgfb1−/C33S tissue compared with WT or Rag2−/− samples. This is consistent with the diminution in Th17 cells we reported earlier (28). Both IL-1β and IL-6 have been associated with enhanced tumorigenesis (29, 30). These increases in gene expression were reversed by introduction of the Rag2−/− mutation (Fig. 5A). Heightened levels of IL-10, Mip1α, and Mip1β were also observed in Tgfb1−/C33S samples, but these changes were at the limit of statistical significance. We examined IL-1β and IL-6 protein levels in tissues by ELISA but we were unable to detect changes in either gastric tissue or in the sera of Tgfb1−/C33S mice compared with WT (data not shown).

Figure 5.

Inflammatory mediator expression and signaling in Tgfb1−/C33S mice. A, when the expression levels of several inflammatory mediators were measured, increases in IL-1β, IL-6, IL-10, Mip1α, Mip1β, and IFN-γ were detected. These were all reduced in the Tgfb1−/C33S;Rag2−/− mice. N = 6 per group. B–D, tissues were also stained for the inflammatory cytokine mediator P-Stat3. Gastric tissue from Tgfb1−/C33S mice displayed significantly more signaling than either control or Tgfb1−/C33S;Rag2−/− samples. This was more clearly seen when the percentage of P-Stat3–positive cells was computed (E). N = 3 per group.F, the expression of three genes, Mmp-13, Survivin, and Bcl-xl, which are stimulated by P-Stat3, was measured. When P-Stat3 levels were high, expression of each of these genes was enhanced, and when P-Stat3 levels were normalized in Tgfb1−/C33S;Rag2−/− mice, the expression levels of Mmp-13 and Survivin returned to normal. N = 6 per group.

Figure 5.

Inflammatory mediator expression and signaling in Tgfb1−/C33S mice. A, when the expression levels of several inflammatory mediators were measured, increases in IL-1β, IL-6, IL-10, Mip1α, Mip1β, and IFN-γ were detected. These were all reduced in the Tgfb1−/C33S;Rag2−/− mice. N = 6 per group. B–D, tissues were also stained for the inflammatory cytokine mediator P-Stat3. Gastric tissue from Tgfb1−/C33S mice displayed significantly more signaling than either control or Tgfb1−/C33S;Rag2−/− samples. This was more clearly seen when the percentage of P-Stat3–positive cells was computed (E). N = 3 per group.F, the expression of three genes, Mmp-13, Survivin, and Bcl-xl, which are stimulated by P-Stat3, was measured. When P-Stat3 levels were high, expression of each of these genes was enhanced, and when P-Stat3 levels were normalized in Tgfb1−/C33S;Rag2−/− mice, the expression levels of Mmp-13 and Survivin returned to normal. N = 6 per group.

Close modal

Because IL-6 signals through the Stat pathway (31), we monitored tissue from WT, Tgfb1−/C33S, and Tgfb1−/C33S;Rag2−/− stomachs by IHC for the product of Stat activation, P-Stat3. Tgfb1−/C33S mice displayed significantly more immunoreactivity than WT mice and the level of reactivity was dramatically reduced in samples from Tgfb1−/C33S;Rag2−/− animals (Fig. 5B–E). The expression of three genes, Mmp13, Survivin, and Bcl-xl, whose expression is enhanced by P-Stat3, was assayed (32–34). Mmp13 and Survivin displayed enhanced expression levels in Tgfb1−/C33S tissues and normal levels in Tgfb1−/C33S;Rag2−/− tissues consistent with Stat3 activation in the compound mutant mice (Fig. 5F). Bcl-xl expression was not significantly enhanced in Tgfβ1−/C33S gastric tissue. These results suggest that there is enhanced signaling through the Stat3 pathway, perhaps because of increased IL-6, in Tgfb1−/C33S gastric tissue compared with WT and this is reversed in Tgfb1−/C33S;Rag2−/− mice.

An earlier report using a conditional deletion of the TGF-β receptor (TβRII) in fibroblasts described gastric tumor formation in TβRII-deficient mice (35). In that system, tumor induction was mediated through a paracrine mechanism in which loss of TGF-β signaling in gastric stroma resulted in enhanced expression of HGF, which stimulated the growth of epithelial cells. Therefore, we examined gastric tissues for HGF levels, as measured by IHC and qRT-PCR. Staining with an antibody for HGF revealed no staining in WT tissue. In samples from Tgfb1−/C33S mice, staining occurred in a pattern corresponding to the morphology of stromal fibroblasts or smooth muscle cells (Fig. 6A–C). To validate that the HGF-positive cells were fibroblasts, we performed colocalization studies using the antibodies to HGF and calponin, a marker for stromal nonepithelial cells (36). We observed coincidence of the two markers (Fig. 6Q–S). We found that HGF expression as measured by qRT-PCR was increased in gastric tissue of Tgfb1−/C33S mice compared with WT animals and was reduced almost to normal levels in Tgfb1−/C33S;Rag2−/− mice (Fig. 6D).

Figure 6.

HGF signaling in Tgfb1−/C33S mice. Gastric tissue from animals of three genotypes was stained for HGF. Cells in the stroma were positive (A–C). DAPI (4′,6-diamidino-2-phenylindole) staining is in blue and HGF in red. The position and shape of these cells were similar to that expected for stromal fibroblasts. When we costained Tgfb1−/C33S gastric tissue for HGF and calponin, a marker for fibroblasts (Q–S), all cells positive for HGF were positive for calponin staining. We also measured HGF expression levels (D) and found a significant increase in Tgfb1−/C33S tissue, which was normalized in Tgfb1−/C33S;Rag2−/− samples. We measured the numbers of cells in the gastric epithelium positive for the activated forms of the HGF receptor c-Met (P-Met) and c-Ron (P-Ron). Both P-Met (E–G) and P-Ron (I–K) levels were increased in Tgfb1−/C33S and reduced in Tgfb1−/C33S;Rag2−/− samples. The differences in percentage positive cells for both markers are presented in H and L, respectively. We also monitored P-Erk levels in the three genotypes (M–O). P-Erk levels increased in Tgfb1−/C33S and decreased in Tgfb1−/C33S;Rag2−/− compared with WT. The percentage differences are indicated in P. N = 3 per group for experiments in D, H, L, and P.

Figure 6.

HGF signaling in Tgfb1−/C33S mice. Gastric tissue from animals of three genotypes was stained for HGF. Cells in the stroma were positive (A–C). DAPI (4′,6-diamidino-2-phenylindole) staining is in blue and HGF in red. The position and shape of these cells were similar to that expected for stromal fibroblasts. When we costained Tgfb1−/C33S gastric tissue for HGF and calponin, a marker for fibroblasts (Q–S), all cells positive for HGF were positive for calponin staining. We also measured HGF expression levels (D) and found a significant increase in Tgfb1−/C33S tissue, which was normalized in Tgfb1−/C33S;Rag2−/− samples. We measured the numbers of cells in the gastric epithelium positive for the activated forms of the HGF receptor c-Met (P-Met) and c-Ron (P-Ron). Both P-Met (E–G) and P-Ron (I–K) levels were increased in Tgfb1−/C33S and reduced in Tgfb1−/C33S;Rag2−/− samples. The differences in percentage positive cells for both markers are presented in H and L, respectively. We also monitored P-Erk levels in the three genotypes (M–O). P-Erk levels increased in Tgfb1−/C33S and decreased in Tgfb1−/C33S;Rag2−/− compared with WT. The percentage differences are indicated in P. N = 3 per group for experiments in D, H, L, and P.

Close modal

Coincident with the increased expression of HGF in Tgfb1−/C33S tissue, there was an increase (7-fold) in cells positive for activated HGF receptor (P-cMet; Fig. 6E–H). This increase was not observed in stomach tissue from Tgfb1−/C33S;Rag2−/− mice. We also examined the tissues for P-Ron, as concomitant increases in P-Ron have been reported upon decreased TGF-β signaling (37, 38). As with P-cMet, staining was increased in Tgfb1−/C33S tissue and reversed to WT levels in Tgfb1−/C33S;Rag2−/− mice (Fig. 6I–L).

We next examined tissues for the level of the downstream mediator of P-cMet, P-Erk (Fig. 6M–P). There was a large increase (∼4-fold) in the percentage of cells that were positive for P-Erk and this increase was reduced by more than 50% when Tgfb1−/C33S and Tgfb1−/C33S;Rag2−/− tissues were compared.

These results indicate a potential signaling sequence of decreased active TGF-β1, enhanced levels of inflammation, the production of inflammatory mediators, upregulation of HGF in stromal tissues, and enhanced gastric epithelial proliferation generating adenocarcinomas. The association of inflammation and proliferation in our model is perhaps surprising as in vitro experiments with gastric stromal cell lines have indicated a direct effect of TGF-β1 on HGF expression (37, 38). Our results, however, indicate that inflammatory cells primarily mediated the effect of lowered TGF-β. The difference between our results and previous reports may reflect differences in responses of cell lines versus tissues. Our data suggest that interference with HGF signaling should decrease both gastric epithelial cell proliferation and tumor production. As chemical inhibitors of HGF are available, future experiments are planned to test the dependence of tumor production on HGF. In addition, interference with IL-6 or IL-1β might prevent tumor formation.

No potential conflicts of interest were disclosed.

Conception and design: M. Ota, K. Shibahara, D.B. Rifkin

Development of methodology: M. Ota, C. Loomis

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Ota, M. Horiguchi, V. Fang, K. Kadota, C. Loomis

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Ota, M. Horiguchi, V. Fang, C. Loomis, M. Cammer, D.B. Rifkin

Writing, review, and/or revision of the manuscript: M. Ota, M. Horiguchi, V. Fang, K. Kadota, D.B. Rifkin

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Ota

Study supervision: D.B. Rifkin

The authors thank Melinda Vassallo and Joseph Ambrogio for their technical assistance.

This work was supported by NIH grants CA034282, GM083220, and CA139238 to D.B. Rifkin. The authors wish to acknowledge the use of the New York University Langone Medical Center (NYUMC) Histopathology and Transgenic Mouse Cores supported by the NYUMC Cancer Center grant P30CA016087. M. Horiguchi and K. Shibahara were supported by fellowships from the Uehara Memorial Foundation. M. Horiguchi was supported by a fellowship from Banyu Life Science Foundation International.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
de Larco
JE
,
Todaro
GJ
. 
Growth factors from murine sarcoma virus-transformed cells
.
Proc Natl Acad Sci U S A
1978
;
75
:
4001
5
.
2.
Bierie
B
,
Moses
HL
. 
TGF-beta and cancer
.
Cytokine Growth Factor Rev
2006
;
17
:
29
40
.
3.
Bachman
KE
,
Park
BH
. 
Duel nature of TGF-beta signaling: tumor suppressor vs. tumor promoter
.
Curr Opin Oncol
2005
;
17
:
49
54
.
4.
Derynck
R
,
Akhurst
RJ
,
Balmain
A
. 
TGF-beta signaling in tumor suppression and cancer progression
.
Nat Genet
2001
;
29
:
117
29
.
5.
Tang
B
,
Vu
M
,
Booker
T
,
Santner
SJ
,
Miller
FR
,
Anver
MR
, et al
TGF-beta switches from tumor suppressor to prometastatic factor in a model of breast cancer progression
.
J Clin Invest
2003
;
112
:
1116
24
.
6.
Wakefield
LM
,
Roberts
AB
. 
TGF-beta signaling: positive and negative effects on tumorigenesis
.
Curr Opin Genet Dev
2002
;
12
:
22
9
.
7.
Kalluri
R
,
Weinberg
RA
. 
The basics of epithelial-mesenchymal transition
.
J Clin Invest
2009
;
119
:
1420
8
.
8.
Burstein
B
,
Libby
E
,
Calderone
A
,
Nattel
S
. 
Differential behaviors of atrial versus ventricular fibroblasts: a potential role for platelet-derived growth factor in atrial-ventricular remodeling differences
.
Circulation
2008
;
117
:
1630
41
.
9.
Kulkarni
AB
,
Huh
CG
,
Becker
D
,
Geiser
A
,
Lyght
M
,
Flanders
KC
, et al
Transforming growth factor beta 1 null mutation in mice causes excessive inflammatory response and early death
.
Proc Natl Acad Sci U S A
1993
;
90
:
770
4
.
10.
Shull
MM
,
Ormsby
I
,
Kier
AB
,
Pawlowski
S
,
Diebold
RJ
,
Yin
M
, et al
Targeted disruption of the mouse transforming growth factor-β1 gene results in mulitfocal inflammatory disease
.
Nature
1992
;
359
:
693
9
.
11.
Kaartinen
V
,
Voncken
JW
,
Shuler
C
,
Warburton
D
,
Bu
D
,
Heisterkamp
N
, et al
Abnormal lung development and cleft palate in mice lacking TGF-β3 indicates defects of epithelial-mesenchymal interaction
.
Nat Genet
1995
;
11
:
415
21
.
12.
Proetzel
G
,
Pawlowski
SA
,
Wiles
MV
,
Yin
M
,
Boivin
GP
,
Howles
PN
, et al
Transforming growth factor-β3 is required for secondary palate fusion
.
Nat Genet
1995
;
11
:
409
14
.
13.
Sanford
LP
,
Ormsby
I
,
Gittenberger-de Groot
AC
,
Sariola
H
,
Friedman
R
,
Boivin
GP
, et al
TGFbeta2 knockout mice have multiple developmental defects that are non-overlapping with other TGFbeta knockout phenotypes
.
Development
1997
;
124
:
2659
70
.
14.
Yoshinaga
K
,
Obata
H
,
Jurukovski
V
,
Mazzieri
R
,
Chen
Y
,
Zilberberg
L
, et al
Perturbation of transforming growth factor (TGF)-beta1 association with latent TGF-beta binding protein yields inflammation and tumors
.
Proc Natl Acad Sci U S A
2008
;
105
:
18758
63
.
15.
Annes
JP
,
Munger
JS
,
Rifkin
DB
. 
Making sense of latent TGFbeta activation
.
J Cell Sci
2003
;
116
:
217
24
.
16.
Rifkin
DB
. 
Latent transforming growth factor-beta (TGF-beta) binding proteins: orchestrators of TGF-beta availability
.
J Biol Chem
2005
;
280
:
7409
12
.
17.
Annes
JP
,
Chen
Y
,
Munger
JS
,
Rifkin
DB
. 
Integrin alphaVbeta6-mediated activation of latent TGF-beta requires the latent TGF-beta binding protein-1
.
J Cell Biol
2004
;
165
:
723
34
.
18.
Munger
JS
,
Huang
XZ
,
Kawakatsu
H
,
Griffiths
MJD
,
Dalton
SL
,
Wu
JF
, et al
The integrin αvβ6 binds and activates latent TGFβ1: a mechanism for regulating pulmonary inflammation and fibrosis
.
Cell
1998
;
96
:
319
28
.
19.
Mazzieri
R
,
Jurukovski
V
,
Obata
H
,
Sung
J
,
Platt
A
,
Annes
E
, et al
Expression of truncated latent TGF-beta-binding protein modulates TGF-beta signaling
.
J Cell Sci
2005
;
118
:
2177
87
.
20.
Yang
Z
,
Mu
Z
,
Dabovic
B
,
Jurukovski
V
,
Yu
D
,
Sung
J
, et al
Absence of integrin-mediated TGFbeta1 activation in vivo recapitulates the phenotype of TGFbeta1-null mice
.
J Cell Biol
2007
;
176
:
787
93
.
21.
Shibahara
K
,
Ota
M
,
Horiguchi
M
,
Yoshinaga
K
,
Melamed
J
,
Rifkin
DB
. 
Production of gastrointestinal tumors in mice by modulating latent TGF-beta1 activation
.
Cancer Res
2013
;
73
:
459
68
.
22.
Engle
SJ
,
Hoying
JB
,
Boivin
GP
,
Ormsby
I
,
Gartside
PS
,
Doetschman
T
. 
Transforming growth factor beta1 suppresses nonmetastatic colon cancer at an early stage of tumorigenesis
.
Cancer Res
1999
;
59
:
3379
86
.
23.
Shinkai
Y
,
Rathbun
G
,
Lam
KP
,
Oltz
EM
,
Stewart
V
,
Mendelsohn
M
, et al
RAG-2-deficient mice lack mature lymphocytes owing to inability to initiate V(D)J rearrangement
.
Cell
1992
;
68
:
855
67
.
24.
Rasband
WS
.
ImageJ
Bethesda, Maryland
:
NIH
;
1997–2014 [cited 2014 Jan 17]. Available from
: http://imagej.nih.gov/ij/.
25.
Shi
Y
,
Massague
J
. 
Mechanisms of TGF-beta signaling from cell membrane to the nucleus
.
Cell
2003
;
113
:
685
700
.
26.
Kang
Y
,
Chen
CR
,
Massague
J
. 
A self-enabling TGFbeta response coupled to stress signaling: Smad engages stress response factor ATF3 for Id1 repression in epithelial cells
.
Mol Cell
2003
;
11
:
915
26
.
27.
Seamons
A
,
Treuting
PM
,
Brabb
T
,
Maggio-Price
L
. 
Characterization of dextran sodium sulfate-induced inflammation and colonic tumorigenesis in Smad3 (−/−) mice with dysregulated TGFbeta
.
PLoS ONE
2013
;
8
:
e79182
.
28.
Ivanov
II
,
Frutos Rde
L
,
Manel
N
,
Yoshinaga
K
,
Rifkin
DB
,
Sartor
RB
, et al
Specific microbiota direct the differentiation of IL-17-producing T-helper cells in the mucosa of the small intestine
.
Cell Host Microbe
2008
;
4
:
337
49
.
29.
Apte
RN
,
Dotan
S
,
Elkabets
M
,
White
MR
,
Reich
E
,
Carmi
Y
, et al
The involvement of IL-1 in tumorigenesis, tumor invasiveness, metastasis and tumor-host interactions
.
Cancer Metastasis Rev
2006
;
25
:
387
408
.
30.
Hodge
DR
,
Hurt
EM
,
Farrar
WL
. 
The role of IL-6 and STAT3 in inflammation and cancer
.
Eur J Cancer
2005
;
41
:
2502
12
.
31.
Heinrich
PC
,
Behrmann
I
,
Muller-Newen
G
,
Schaper
F
,
Graeve
L
. 
Interleukin-6-type cytokine signalling through the gp130/Jak/STAT pathway
.
Biochem J
1998
;
334
(
Pt 2
):
297
314
.
32.
Bromberg
JF
,
Wrzeszczynska
MH
,
Devgan
G
,
Zhao
Y
,
Pestell
RG
,
Albanese
C
, et al
Stat3 as an oncogene
.
Cell
1999
;
98
:
295
303
.
33.
Kanda
N
,
Seno
H
,
Konda
Y
,
Marusawa
H
,
Kanai
M
,
Nakajima
T
, et al
STAT3 is constitutively activated and supports cell survival in association with survivin expression in gastric cancer cells
.
Oncogene
2004
;
23
:
4921
9
.
34.
Solis-Herruzo
JA
,
Rippe
RA
,
Schrum
LW
,
de La Torre
P
,
Garcia
I
,
Jeffrey
JJ
, et al
Interleukin-6 increases rat metalloproteinase-13 gene expression through stimulation of activator protein 1 transcription factor in cultured fibroblasts
.
J Biol Chem
1999
;
274
:
30919
26
.
35.
Bhowmick
NA
,
Chytil
A
,
Plieth
D
,
Gorska
AE
,
Dumont
N
,
Shappell
S
, et al
TGF-beta signaling in fibroblasts modulates the oncogenic potential of adjacent epithelia
.
Science
2004
;
303
:
848
51
.
36.
Tuxhorn
JA
,
Ayala
GE
,
Smith
MJ
,
Smith
VC
,
Dang
TD
,
Rowley
DR
. 
Reactive stroma in human prostate cancer: induction of myofibroblast phenotype and extracellular matrix remodeling
.
Clin Cancer Res
2002
;
8
:
2912
23
.
37.
Cheng
N
,
Bhowmick
NA
,
Chytil
A
,
Gorksa
AE
,
Brown
KA
,
Muraoka
R
, et al
Loss of TGF-beta type II receptor in fibroblasts promotes mammary carcinoma growth and invasion through upregulation of TGF-alpha-, MSP- and HGF-mediated signaling networks
.
Oncogene
2005
;
24
:
5053
68
.
38.
Cheng
N
,
Chytil
A
,
Shyr
Y
,
Joly
A
,
Moses
HL
. 
Transforming growth factor-beta signaling-deficient fibroblasts enhance hepatocyte growth factor signaling in mammary carcinoma cells to promote scattering and invasion
.
Mol Cancer Res
2008
;
6
:
1521
33
.