Tumor necrosis factor deficiency inhibits mammary tumorigenesis and a tumor necrosis factor neutralizing antibody decreases mammary tumor growth in neu/erbB2 transgenic mice

Introduction

In women, breast cancer is the most common form of cancer, and is the second leading cause of cancer deaths. Although advances in therapy have increased survival, advanced disease remains difficult to treat, necessitating new treatment options. One possibility is to use a biological approach, in which a naturally occurring growth-promoting molecule in the tumor microenvironment is targeted. A potential candidate is tumor necrosis factor-α (TNF-α), which is expressed in the mammary epithelium and stroma (1, 2), as well as the immune cells which are recruited to the mammary gland stroma under various physiologic or pathologic conditions. This cytokine has been shown to have pleiotropic effects on cells, with the outcome of cell death, growth, or differentiation being concentration-dependent and context-dependent (3). For example, at high doses, TNF has antitumor properties, which result, at least in part, from its cytotoxic effects on the tumor vasculature (4). At more physiologic concentrations, however, TNF exerts a growth-promoting effect on certain normal epithelia, including that in the mammary gland (5), intestine (6), ovary (7), and liver (8). Pharmacologically, high doses of TNF can be exploited clinically, although its use is limited to regional treatment, for example, isolated limb perfusion for soft tissue sarcoma (4) because of systemic toxicity at high doses. The opposite approach, i.e., a TNF blockade, should be considered to combat the growth-promoting effects of locally produced TNF on the tumor epithelium.

TNF has been shown to increase the growth of both normal and malignant mammary epithelial cells in experimental models. In a physiologically relevant three-dimensional model system, TNF stimulated the proliferation and branching morphogenesis of rat mammary epithelial organoids, and modulated functional differentiation (1, 5, 9–12). These effects appear to be physiologically relevant because expression of TNF and its two receptors is regulated throughout normal mammary gland development (1). Furthermore, TNF also stimulated the growth and morphogenesis of organoids derived from 1-methyl-1-nitrosourea–induced rat mammary tumors (13), demonstrating the potential for TNF-induced growth of mammary tumors in vivo. TNF may also play a role in metastasis, as it stimulates matrix metalloproteinase-9 secretion from mammary epithelial cells (10, 11) as well as the surrounding stroma (14). Clinically, TNF expression is increased in the stroma of invasive breast carcinoma tissue compared with benign tissue (2). Moreover, the number of cells expressing TNF increased with tumor grade (2), and serum TNF concentration showed a positive correlation with breast tumor stage (15).

Based on the above studies, we hypothesized that TNF might promote the growth of breast cancers in vivo. Two approaches were used to test this hypothesis. In the first, we crossed mice which overexpress activated neu/erbB2 in the mammary epithelium, a mouse model which gives rise to spontaneous mammary tumors (16), with TNF null mice (17), and compared mammary tumor development in TNF null and wild-type bigenic mice. ErbB2 has been shown to be overexpressed in 20% to 30% of human breast cancers, and is associated with poor outcome (18). Secondly, we asked whether anti-TNF therapy would inhibit the growth of established mammary tumors. Data from both studies support the notion that anti-TNF therapy might play a role in the treatment of breast cancer.

Results

TNF Stimulates the Growth of erbB2-Overexpressing Mammary Tumor Cells In vitro

We previously showed that TNF stimulated the growth of normal rat and mouse mammary epithelial cells, as well as 1-methyl-1-nitrosourea–induced rat mammary tumor cells, in a three-dimensional primary culture model (1, 5, 13, 24). These observations suggested that TNF might stimulate mammary tumorigenesis in vivo. To examine this, we chose a transgenic mouse model in which neu/erbB2 was overexpressed in the mammary epithelium. We further chose to use a model, NDL2-5, in which neu/erbB2 is constitutively active because of an in-frame deletion (16), reasoning that because we would be breeding the NDL2-5 mice into the relatively tumor-resistant C57BL/6 strain (25), tumorigenesis would be greater than if we were to use the somewhat less aggressive wild-type neu/erbB2 model (16). Prior to undertaking the extensive breeding required to test our hypothesis, we first asked whether mammary tumor cells from NDL2-5 mice were sensitive to TNF. Mammary tumor organoids were suspended within the Engelbreth Holmes Swarm sarcoma–reconstituted basement membrane, and cultured in serum-free medium in the absence or presence of 100 ng/mL of TNF. As can be seen in Supplementary Fig. S1, the mammary tumor cells were growth-stimulated at all time points tested (up to 3 weeks in culture). Based on these data, we then proceeded with developing the TNF null and wild-type erbB2-overexpressing mice.

Mammary Tumorigenesis Is Decreased in TNF Null Mice

NDL-2 neu/erbB2 transgenic mice were crossed with TNF null or TNF wild-type mice to generate mice that overexpressed neu/erbB2 in the mammary epithelium, and which were TNF+/+, TNF+/−, or TNF−/− in all tissues. As seen in Fig. 1, the development of palpable mammary tumors was significantly delayed in TNF−/− mice when compared with TNF+/+ mice (P < 0.05). Palpable tumor development in the two TNF+/− groups was comparable, and fell between that of the TNF null and wild-type groups, although neither TNF+/− group was significantly different from TNF wild-type or null mice. Compared with the parental FVB NDL-2 transgenic mice in which the average age of mammary tumor onset in nulliparous mice is 23 weeks (16), tumor development was slower in the TNF+/+ FVB:C57BL/6 mice of the current study, with palpable tumor incidence of only 44.4% at 78 weeks of age (Table 1). The corresponding percentages for the other groups at this age were 10.0% for the TNF−/− group (P < 0.015 compared with TNF+/+), and 20.7% and 22.2% for the TNF+/− 2A and TNF+/− 2B crosses, respectively. The number of visible tumors per mouse at sacrifice was also significantly different between the TNF−/− and TNF+/+ groups (0.10 and 0.67 tumors/mouse, respectively; P < 0.019), with the two TNF+/− groups falling between the homozygous mice (Table 1). Together, these data suggest a potential dose-dependent effect of TNF for primary tumor development. In contrast, no significant differences in lung metastasis were noted among the groups (Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Mammary tumorigenesis is significantly decreased in TNF null mice. Neu/erbB2-overexpressing mice which were TNF−/−, TNF+/−, or TNF+/+ were generated in the 2A and 2B crosses as described in Materials and Methods, and the time at which a palpable tumor was detected in each mouse was recorded. Kaplan-Meier plot showing the percentage of tumor-free mice at different ages in each group. The TNF null group (open circles) is significantly different from the TNF wild-type group (closed circles; P < 0.05); however, neither of the TNF+/− groups (dotted triangles, cross 2A or dotted squares, cross 2B), is significantly different from the TNF−/− or TNF+/+ groups. The 2A and 2B identifiers refer to the cross number, as described in Materials and Methods. The number of mice in each group was as follows: 2A (TNF+/−), 29; 2A (TNF−/−), 30; 2B (TNF+/−), 27; and 2B (TNF+/+), 27.

Because of the low palpable tumor incidence, we considered the possibility that microscopic mammary lesions had developed which had not progressed to palpable tumors. We therefore prepared whole mounts from mammary glands 1 to 3 and 4 to 5 on one side of each mouse to determine the extent to which these putative lesions had developed, and whether TNF status altered their number. Extensive hyperplasia was noted in the mammary glands of mice from all groups (see example in Fig. 2A and B). We quantified these hyperplastic and putative preneoplastic lesions within three different size groupings (0.2–0.5, 0.5–1.0, and >1.0 mm in diameter), but found no statistically significant effect of TNF gene dosage (Table 1). In cross-section, large hyperplastic ductal-lobular structures were observed in both TNF+/+ and TNF−/− mice, the majority of which were composed of a single layer of epithelial cells surrounding a lumen (TNF−/−, Fig. 2C; TNF+/+, Fig. 2D). Additionally, areas of both lobular carcinoma in situ (Fig. 2E), as well as ductal carcinoma in situ (Fig. 2F) were observed, some of which were within the large structures, as is illustrated in Fig. 2E.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Hyperplasia as well as microscopic carcinomas are observed in mammary glands of FVB:C57BL/6 mice which overexpress neu/erbB2 in the mammary epithelium. A and B, whole mount of the extensive hyperplasia that developed in these mice. Black arrow, a putative preneoplastic lesion from a TNF−/− mouse. Photos were taken under the 1× objective (A) or the 4× objective (B). C and D, H&E-stained paraffin sections of ductal-lobular hyperplasia in mammary glands of TNF−/− (C) and TNF+/+ (D) mice. Both photos were taken under the 10× objective (inset in D is shown enlarged in E). E, H&E-stained paraffin section of lobular carcinoma in situ within the large ductal-lobular structure in D (TNF+/+ mouse, 20× objective; arrows in D and E point to the same structure). F, H&E-stained paraffin section of ductal carcinoma in situ from a TNF−/− mouse (20× objective). Overall, similar morphologies were seen in the other groups.

Finally, it should be noted that all the transgenic mice were healthy, although the TNF−/− mice were slightly smaller than the TNF+/+ mice (Table 1). Spleen weights were not significantly different among the groups, although in the TNF+/+ group, tumor-bearing mice had larger spleens than non–tumor-bearing mice (Table 1). Because of the small number of mice developing tumors in the TNF−/− group, it was not possible to definitively determine whether this biological response was observed in TNF-deficient mice.

Proliferation Is Decreased in the Mammary Epithelium of TNF Null Mice

Ki67 immunohistochemistry was used as a preliminary step to investigate the mechanism by which TNF status might alter mammary tumorigenesis. We first looked at the Ki67 immunohistochemistry score (see Materials and Methods) of the mammary tumors, but found no significant difference between the TNF wild-type (n = 10) and TNF null (n = 4, three palpable and one microscopic) tumors (immunohistochemical indices of 10.8 ± 0.4 and 11.3 ± 1.0, respectively). Because this analysis was limited by the small number of mammary tumors in the TNF null mice, we therefore examined the proliferative status of the mammary epithelium per se. As seen in Table 2, the proliferative rate was high in both the TNF null and wild-type groups as compared with what would be expected in the epithelium of a nontransgenic mouse of the same age. Notably, however, the percentages of Ki67-positive nuclei was significantly decreased in both the ductal and lobular mammary epithelium of the TNF null mice, when compared with that of the TNF wild-type mice (Table 2). Apoptosis was concurrently decreased in the lobular, but not the ductal mammary epithelium of TNF−/− mice, but overall, the percentages of cells undergoing apoptosis was only ∼15% of those which were undergoing proliferation (Table 2).

Effect of TNF Status on Carcinogen Sensitivity of Mammary Glands in Organ Culture

The increased tumor latency in the TNF null mice could result from the systemic absence of TNF because TNF is deleted in all tissues, and/or from the absence of TNF in the mammary gland. Because TNF is expressed in both the epithelial and stromal compartments of the mammary gland (1),⁴ the role of mammary gland TNF on tumorigenic sensitivity is not readily addressable in an in vivo study. We therefore used the in vitro mammary gland organ culture model described in the Supplementary Materials to ask whether mammary glands from TNF null mice had an altered sensitivity to the carcinogen 7,12-dimethylbenz(a)anthracene (DMBA). Previous studies in BALB/c mice (22) had shown that the culture conditions used permitted the normal lobuloalveolar epithelium which develops during the first 9 days of culture to regress when placed in an involution medium; however, preneoplastic nodule-like alveolar lesions arising in the DMBA-treated mammary glands did not regress, and thus, could be quantified. In our studies, in which the glands from one side of the mouse were treated with DMBA, and those from the other side with the DMSO vehicle, we were unable to identify nodule-like alveolar lesions (n = 9 TNF−/− mice and n = 7 TNF+/+ mice). Specifically, although all the glands underwent extensive morphologic development, in each case, the regressed mammary glands from the control and DMBA-treated cultures could not be distinguished (Supplementary Fig. S2, A versus B and C versus D). This failure to detect preneoplastic lesions might be related to the decreased sensitivity of C57BL/6 mice (the strain background of our TNF null and wild-type mice) to carcinogen-induced mammary tumors, when compared with BALB/c mice (26). We did not test higher DMBA concentrations due to its reported cytotoxicity (27). In spite of the lack of effect of DMBA in these studies, these experiments did show that the regressed mammary glands were morphologically similar in the TNF null and wild-type mice (Supplementary Fig. S2; data not shown; n = 7–9).

Mammary Tumor Growth Is Inhibited in Mice Treated with a TNF-Neutralizing Antibody

As a result of our observation that mammary tumorigenesis was delayed and decreased in TNF null mice, we hypothesized that anti-TNF therapy would inhibit the growth of established mammary tumors. To examine this, mice overexpressing wild-type neu/erbB2 in the mammary epithelium were randomized into two groups and treated with the TNF-neutralizing antibody or the control antibody for 4 weeks. As seen in Fig. 3, the anti-TNF antibody significantly inhibited the growth of established mammary tumors, when compared with controls. This treatment did not seem to be deleterious to the mice because neither body weight (Table 3) nor liver weight (data not shown) were altered. Interestingly, however, the tumor-induced increase in spleen weight (see above), was partially, although significantly, reversed, suggesting a biological effect of the TNF antibody on this organ. A similar observation was made in a previous study in which the same TNF-neutralizing antibody was shown to partially block the increase in spleen weight resulting from immunization with sheep RBC (23).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

The TNF-neutralizing cV1q antibody inhibits the growth of neu/erbB2-overexpressing mouse mammary tumors. FVB/N-Tg (MMTVneu)202Mul/J mice were randomly divided into two groups when they developed a tumor of 3 to 4 mm in diameter, and treated with either the neutralizing anti-TNF (cV1q, □) or control (cVam, ▪) antibodies, at a dose of 1 mg/wk, i.p., for 4 wks. Points, mean of 20 mice per group; bars, SE. The two groups are significantly different from the 2-wk time period onward.

The proliferative index in the tumors, as measured by Ki67 immunohistochemistry, was not significantly different in the two groups, with immunohistochemical indices of 12.0 ± 0.3 (n = 19) and 12.4 ± 0.3 (n = 20) in the control antibody-treated and TNF antibody-treated groups, respectively. It is important to note, however, that Ki67 staining was quite variable in different regions of the tumor, and the immunohistochemical index for each tumor represents an average for the section. In contrast to the tumor, TNF antibody therapy did have an inhibitory effect on proliferation in the surrounding mammary epithelium in the tumor-bearing mice. As seen in Table 3, mice treated with the neutralizing TNF antibody had significantly decreased proliferation in the lobular epithelium (P < 0.008). Proliferation was also decreased in the ducts, although this did not achieve statistical significance (P = 0.123). Unexpectedly, no change in proliferation was noted within the microscopic hyperplasias and ductal carcinoma in situ/lobular carcinoma in situ lesions. We also used TUNEL immunohistochemistry to determine whether the TNF antibody treatment altered the apoptotic index within the tumors. However, essentially all of the tumor cells were TUNEL-negative, even though TUNEL-positive cells could be seen in the adjacent “normal” epithelium (data not shown). TNF antibody treatment had no significant effect on the apoptotic index within the ductal and lobular epithelium, nor on the small neoplastic (ductal carcinoma in situ/lobular carcinoma in situ) or hyperplastic lesions (Table 3).

Discussion

The present study provides strong support for the hypothesis that TNF plays a stimulatory role in the growth of breast cancer. Specifically, mammary tumorigenesis was decreased in TNF null mice which overexpressed activated neu/erbB2 in the mammary epithelium, when compared with the corresponding TNF wild-type mice. Moreover, treatment of mammary tumor–bearing mice with an antibody that neutralizes TNF inhibited tumor growth. Although these studies were conducted using the neu/erbB2 transgenic mouse model, based on our previous data demonstrating that TNF increases the proliferation of primary malignant mammary cells from NMU-induced rat mammary tumors, as well as normal rat and mouse mammary epithelial cells (1, 5, 13, 24), we believe that the observed effects are more general in nature, and are not limited to neu/erbB2-overexpressing breast cancers.

Growth of Several Cancer Types Is Modified by TNF Status

TNF is not required for normal mouse development, but splenic follicular architecture fails to develop normally in TNF null mice, and these mice do not develop a normal humoral immune response, are susceptible to infection, and have an altered inflammatory response (17, 28). This phenotype suggested that TNF status might affect tumorigenesis, as indeed was found to be the case. Most of the work in this regard has been done using the DMBA/12-O-tetradecanoylphorbol-13-acetate two-stage skin carcinogenesis model, in which a significant decrease in tumor development was observed in TNF−/− mice (29–31). In this model, TNF did not exert its effect on initiation, but rather as a tumor promoter (32); moreover, although both TNF receptors (TNFR) contributed to the effect of TNF, TNFR1 was found to be more critical (33). Mice deficient in TNFR1 or TNFR2 were also shown to be resistant to UV-induced skin carcinogenesis (34). In addition to its stimulatory role in skin carcinogenesis models, TNF promotes the growth of ovarian cancer. Unlike normal ovarian epithelium, ovarian tumor cells secrete TNF (35), which acts in an autocrine or paracrine manner by stimulating an increased release of chemokines, cytokines, and vascular endothelial growth factor, all potentially involved in tumor growth and neovascularization (36). Furthermore, ovarian tumors from TNF knockdown cells had reduced growth, metastasis, and vascularization in vivo, compared with TNF+/+ or mock-transfected cells (36). Finally, hepatic carcinogenesis induced by a choline-deficient, ethionine-supplemented diet, as well as carcinogen-induced colorectal carcinogenesis, were found to be decreased in TNFR1 null mice (37, 38). We suspect that TNFR1 plays a predominant role in TNF-stimulated mammary tumorigenesis as well, based on our previous studies demonstrating that TNFR1, but not TNFR2, mediated the stimulatory effect of TNF on the proliferation of normal mammary epithelial cells (1).

At least two studies have examined the role of TNF in models of inflammation-linked cancer. In a transgenic mouse model in which cyclooxygenase-2 and microsomal prostaglandin E synthase-1 are overexpressed in the gastric epithelium, producing an inflammatory phenotype, Oshima et al. showed that TNF deficiency decreased both inflammation and gastric cancer incidence (39). Moreover, in the Mdr2-null mouse model of inflammation and hepatocellular carcinoma, TNF-neutralizing antibodies inhibited the activation of the RelA subunit of nuclear factor-κB (NFκB) and induced the apoptosis of dysplastic hepatocytes (40). Because tumorigenesis in the neu/erbB2 mammary model is driven by the overexpressed neu/erbB2, it is unlikely that TNF plays a proinflammatory role in driving tumor development in the current experiments. Rather, based on our previous studies (24), the higher proliferative rate we observed in the mammary epithelium of TNF wild-type mice is most likely driven, at least in part, through TNF induction of the NFκB2/RelB or NFκB1/RelB transcription factors, and subsequent binding of the NFκB complex to the cyclin D1 promoter.

Inhibitors of TNF Activity Have Anticancer Efficacy

The above studies show that physiologic levels of TNF are stimulatory to the growth of certain types of cancer, and suggest that agents which interfere with TNF or with TNF signaling, may be a useful therapeutic option. Indeed, we found that an antibody which neutralizes TNF significantly inhibited the growth of established neu/erbB2-overexpressing mammary tumors. Although this effect was relatively modest, it is likely that more frequent dosing (e.g., more than the once per week regimen we used here), or a higher dose, would have a greater effect. To our knowledge, only one other group has examined the effect of TNF neutralization on the growth of mammary tumors. In that study, anti-TNF treatment was initiated 1 day prior to transplantation of 410.4 mouse mammary tumor cells, and continued on a weekly basis (31). Tumor growth was significantly inhibited in this model, although it should be pointed out that because treatment was started prior to tumor transplantation, it is possible that neutralization of TNF could have inhibited tumor take, as well as tumor growth. In contrast, our study used a strictly therapeutic approach, in which treatment began only after a tumor was present, such as would be the case in the clinic.

Neutralization of TNF during 12-O-tetradecanoylphorbol-13-acetate promotion was shown to inhibit skin papilloma development in DMBA-treated mice (31), suggesting a role for endogenous TNF in this model of skin carcinogenesis. Interestingly, the TNF antibody was equally effective whether it was given once weekly for only the first 6 weeks of promotion, or whether it was given for the full 15 weeks, implying that TNF is involved in the early stages of tumor promotion. Additionally, a recent report showed that the TNF antagonist, etanercept, a soluble TNFR2 fusion protein, inhibited colitis-associated colon tumorigenesis when administered after completion of the tumor induction protocol, suggesting that it may have partially reversed tumorigenesis at a time when tumors were already present (38). Our own data suggest a promotion role for TNF both during neu/erbB2-driven mammary tumorigenesis, as well as during the growth of established mammary cancers. The mechanism by which the promotion occurs remains to be determined. Although it is clear in the current study that TNF can stimulate the proliferation of the preneoplastic neu-erbB2–overexpressing mammary epithelium, neither the number of neu/erbB2-driven hyperplasias nor the proliferative rate of the tumors per se were different between the TNF+/+ and TNF−/− mice. A possible explanation for this observation is that TNF may stimulate vasculogenesis, as shown by Kulbe et al. (36) and Li et al. (41) in other models, thus providing the hyperplasias and tumors with an environment more favorable for expansion. Furthermore, in the absence of TNF, or in the presence of the TNF-neutralizing antibody, the vasculature may lose this positive stimulus, and thus, tumor growth is slowed, as we show here.

TNF Inhibitors Are in Use Clinically

Currently, there are three TNF inhibitors in routine clinical use, with more in development (42). These include the TNF-neutralizing antibodies infliximab (Remicade) and adalimumab (Humira), and the TNFR2 fusion protein etanercept (Enbrel), which are used in the treatment of inflammatory diseases including rheumatoid and psoriatic arthritis; infliximab and adalimumab are also effective in treating Crohn's disease (42). Preliminary reports have also shown that infliximab and etanercept are safe and well-tolerated in patients with advanced cancer; moreover, there was some indication of biological efficacy in these phase I/II studies, including disease stabilization (43–46). To our knowledge, only one study has examined the effect of a TNF inhibitor in patients with breast cancer. In that study, 1 out of 16 patients with metastatic breast cancer treated with etanercept had a brief period of disease stabilization (44). It is tempting to speculate that treatment might be more effective in patients with earlier disease stage. Although these first-generation drugs are not without problems (42), our data demonstrating the role of TNF in mammary tumorigenesis, as well as the efficacy of a TNF-neutralizing antibody in inhibiting the growth of established aggressive mammary tumors, suggests that TNF inhibitors may be effective in at least a subset of patients with breast cancer. Moreover, we suggest that appropriate candidates in this regard may be selected based on an elevated TNF concentration in their breast cancer and/or in the adjacent stromal and epithelial tissues.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.