The Novel Triterpenoid C-28 Methyl Ester of 2-Cyano-3, 12-Dioxoolen-1, 9-Dien-28-Oic Acid Inhibits Metastatic Murine Breast Tumor Growth through Inactivation of STAT3 Signaling

Signal transducer and activator of transcription 3 (STAT3) has frequently been found in human breast cancer samples (1). STAT3-mediated up-regulation of c-Myc levels was proposed to be one of the important mediators of STAT3 oncogenesis (2). c-Myc is a potent oncoprotein that has been found to play a critical role in tumorigenesis (3). The use of an inducible c-Myc model showed tumor growth or suppression, respectively, by controlling c-Myc expression in a murine hepatocarcinoma model (4). Our previous studies and that of others have shown that c-Myc expression is regulated by constitutive activation of STAT3 in mouse breast cancer and other types of cancer cells (5–7). As the immediate upstream regulator of STAT3, Src has been considered to play an important role in breast carcinogenesis. However, we found that Src can be inactivated by STAT3 knockdown in 4T1 cells (5). We have also shown that Akt can be inactivated following STAT3 knockdown in 4T1 cells (5), suggesting that STAT3 is required for maintaining the activation status of Src and Akt. As a downstream target of Src, Akt plays an important role in regulating antiapoptotic proteins and proteins that regulate cell cycle progression (8). However, it is unclear how these intracellular signal transduction proteins interact with each other, and the biological consequences of these interactions remain uncertain. Recent reports provided convincing evidence that constitutively activated STAT3 in breast cancer cells is associated with impaired cell-mediated immunity (9) and that STAT3 activation in cancer cells may allow them to evade host immune surveillance (10). In agreement with these observations, we reported that knockdown of STAT3 completely abolished breast tumor formation in an immunocompetent mouse model (5).

The 4T1 cell line was originally derived from a spontaneous mammary carcinoma in BALB/c mice (11). It has been reported that 4T1 cells mimic the effects of human mammary carcinoma in that morbidity is due to the outgrowth of spontaneous micrometastatic tumor cells that migrate to distant organs relatively early during primary tumor growth (12, 13) and are notoriously chemoresistant (14). Therefore, 4T1-driven breast cancer in BALB/c mice constitutes an appropriate in vivo system for modeling human breast cancer with regard to tumor growth and metastasis. As such, the 4T1/BALB/c mouse system also provides a useful experimental model for exploring the biological effects in mammary tumorigenesis and for studying the effects of small molecules that target signal transduction pathways in vivo.

2-Cyano-3, 12-dioxoolen-1, 9-dien-28-oic acid (CDDO) and its derivative CDDO-Me are synthetic triterpenoids proven to be effective agents in controlling cancer cell growth in preclinical models of leukemias, multiple myeloma, lymphomas, and solid tumors, such as breast, pancreatic, and colon cancer (15–18). CDDO-Me has been shown by us to (a) promote apoptosis in several leukemic cell lines, (b) induce proapoptotic Bax protein as a prelude to caspase activation, and (c) suppress the activation of extracellular signal-regulated kinase (ERK) 1/2 with concomitant inhibition of Bcl-2 phosphorylation (18). In the present study, we intended to examine early changes induced by CDDO-Me in mouse breast cancer cells. Our findings indicate that CDDO-Me effectively inhibits STAT3 signaling, which contributes to profound antitumor efficacy of this compound in the in vitro and in vivo models of highly metastatic 4T1 breast cancer cells. These observations hence highlight the novel mechanism of CDDO-Me activity in breast cancer, which is potentially applicable to other types of tumors driven by the activation of STAT3 signaling.

Cell line, antibodies, and reagents. The 4T1 cell line was obtained from the American Type Culture Collection. The cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS). Anti-phosphorylated Tyr-STAT3 (phosphorylated Tyr⁷⁰⁵), anti-STAT3, anti-phosphorylated Src (phosphorylated Tyr⁵²⁷), anti-Src, anti-phosphorylated Akt (phosphorylated Ser⁴⁷³), and anti-c-Myc were purchased from Cell Signaling Technology, Inc.; 33D1 and MHC class II antibodies were obtained from eBioscience; anti-β-actin was obtained from Sigma Life Sciences. The Cell Invasion kit was purchased from Chemicon International, Inc. Coelenterazine was obtained from Biotium, Inc.; Trizol was purchased from Invitrogen, Inc. Primers for c-Myc real-time PCR were purchased from Applied Biosystems, Inc. CDDO-Me was provided by Reata Pharmaceuticals, Inc.

Animals. Six- to eight-week-old female BALB/c mice were purchased from Charles River Laboratories, Inc. and maintained in M. D. Anderson conventional animal facility.

Construction of lentivirus for Renilla luciferase cDNA transduction. The plasmid was created by inserting the 2-kb Renilla luciferase cDNA fragment from the pMOD-LucSH plasmid (InvivoGen). The lentiviral gene transfer vector pWIP/GFP was a generous gift from Dr. Didier Trono (Geneva University, Geneva, Switzerland). Lentivirus preparation, gene transduction by lentiviral infection, and cell sorting were done as described previously (5).

Western blotting. Western blotting was done as described previously (19).

Cell proliferation assay. rLu/GFP-tagged 4T1 cells were seeded into six-well plates at a concentration of 5 × 10⁵ per well in triplicate, and 500 nmol/L CDDO-Me was added to each well. The total viable cell number in each well was determined using an automated analyzer (Vi-CELL, Beckman Coulter, Inc.).

Cell cycle analysis and real-time PCR. Cell cycle analysis and real-time PCR were carried out as described previously (20).

Matrix gel invasion assay. The matrix gel invasion assay was conducted in matrix chambers according to the manufacturer's guidelines; the Cell Invasion kit was purchased from Chemicon International.

Mouse tumor formation assay. Various doses of 4T1/rLu/GFP cells were injected into mice to establish the levels needed for tumor formation. For the in vivo efficacy study, 7 × 10³ of Renilla luciferase/GFP tagged 4T1 cells were injected into the mammary fat pads of 8-week-old female BALB/c mice, and CDDO-Me treatment was initiated at either day 1 or day 5 after tumor implantation. CDDO-Me at 200 μg/mouse was given i.v. at 2-day intervals. CDDO-Me was formulated in liposomes as described elsewhere (15). Control mice were injected with empty liposomes as described previously (15). Tumor formation at the inoculation site was monitored using an In Vivo Imaging System 100 from Xenogen Corp.

Live animal imaging assay. Mice were injected i.p. with 20 μg coelenterazine and imaged using the Xenogen In Vivo Imaging System 100.

Intratumoral CDDO-Me measurements. The tumor samples were placed in 1 mL methanol after being weighed. The samples were homogenized and centrifuged at 4,000 rpm for 10 min. The supernatants containing CDDO-Me were saved, lyophilized, and further reconstituted with 200 μL of 1:1 methanol/water. Samples were spun for 10 min at 13.2 relative centrifugal force, and CDDO-Me concentrations were determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS; ref. 15).

Histologic analysis. For the histologic analysis, breast tumor and lungs from mice treated with/without CDDO-Me were excised and fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned, and stained with H&E. The slides were analyzed under a Nikon Optiphot microscope with a digital capture camera (Microscopy Documentation System 290, Eastman Kodak).

Renilla luciferase is stably expressed in 4T1 cells after lentiviral infection. A lentivirus gene transfer plasmid was constructed as indicated in Fig. 1A, in which Renilla luciferase and green fluorescent protein (GFP) were designed in a bicistronic fashion, which allows GFP to function as the marker of gene transduction. After a single exposure to the lentivirus that was carrying cDNAs of Renilla luciferase and GFP, 4T1 cells expressed both Renilla luciferase and GFP. As shown in Fig. 1B and C, 4T1 cells started expressing GFP 72 h after lentiviral infection. Meanwhile, the cells were also expressing high levels of Renilla luciferase (Fig. 1D). After fluorescence-activated cell sorting for selection of GFP(+) cells, the population of Renilla luciferase expressing 4T1 cells was enriched to >90% and used for further studies after expansion.

CDDO-Me blocks 4T1 intracellular signal transduction pathways. We previously showed that, as a major oncogenic regulator, STAT3 was constitutively activated in 4T1 cells (5). In this study, we first examined whether CDDO-Me had an effect on constitutively activated STAT3. As shown in Fig. 2A, we found that STAT3 activity was significantly inhibited by 500 nmol/L CDDO-Me after 2 h, whereas total STAT3 levels remained unchanged. We also examined the effect of CDDO-Me on Src and Akt. We found that both Src and Akt were inactivated, but total Src and Akt levels were not affected. We then examined effects of CDDO-Me on the mRNA level of c-Myc, a known downstream target of STAT3. We found that the level of c-Myc in treated cells was only one third of that in untreated 4T1 cells (Fig. 2B). Using small interfering RNA (siRNA), we observed similar effects, specifically which knocking down STAT3 led to the inactivation of both Src and Akt and reduction of the expression of c-Myc (Fig. 2C).

CDDO-Me does not induce apoptosis in 4T1 cells but blocks cell proliferation in vitro. To determine the functional consequences of STAT3 inhibition by CDDO-Me, we examined the ability of CDDO-Me to induce apoptosis and/or growth inhibition in 4T1 cells in vitro. We observed no significant apoptosis induction in 4T1 cells after CDDO-Me treatment as assessed by Annexin V flow cytometry (Fig. 3A). Furthermore, no significant apoptosis was detected even at higher doses of CDDO-Me after 6, 24, and 48 h (Fig. 3B; data not shown). However, we found that CDDO-Me completely blocked cell growth at the 500 nmol/L dose level as indicated in Fig. 3C. Therefore, we further analyzed the changes in cell cycle distribution of 4T1 cells after treatment with 500 nmol/L CDDO-Me. As indicated in Fig. 3D and E, 4T1 cells were blocked in G₂-M after treatment with CDDO-Me for 24 and 48 h. Rather unexpectedly, when STAT3 was depleted in 4T1 cells using siRNA technology, CDDO-Me induced a profound apoptotic response, with no viable cells remaining at 48 h (Fig. 4F).

CDDO-Me inhibits invasion of 4T1 cells. It has been reported that 4T1 cells are highly invasive, which reflects their metastatic ability in vivo. To determine whether CDDO-Me can affect this property of 4T1 cells, we used an in vitro matrix gel invasion assay to evaluate effects on 4T1 cells in the presence of 500 nmol/L CDDO-Me. As shown in Fig. 4, 500 μmol/L CDDO-Me virtually abrogated the invasion of 4T1 in the assay.

Effect of CDDO-Me on 4T1 breast cancer growth and metastasis in immunocompetent mice. It has been well established that 4T1 cells generate aggressive primary breast tumors and metastatic tumors in BALB/c immunocompetent mice (21). To study the effect of CDDO-Me on breast tumor formation, we first established the 4T1/rLu breast tumor model to prove that 4T1 cells expressing Renilla luciferase would not be rejected by the host. As shown in Fig. 5A and B, the ability of 4T1/rLu cells to form breast tumors was indistinguishable from those formed by parental 4T1 cells (5). We then examined the effect of CDDO-Me on the ability of 4T1 cells to form tumors. CDDO-Me treatment was initiated 1 day after inoculation of mice with 4T1 breast tumor cells (7 × 10³ per mouse), at the dose of 200 μg/mouse at 2-day intervals, for a total of five injections. The results presented in Fig. 5B showed that 4T1 breast tumor formation was completely inhibited by CDDO-Me in five mice (Fig. 5B,, right). In contrast, four mice without CDDO-Me treatment developed aggressive breast tumors at the injection site (Fig. 5B,, left), which required sacrifice at 35 days. No tumors or tumor metastases were observed in CDDO-Me–treated mice over a 90-day observation period, whereas numerous metastatic tumors in the lung were observed in control animals (Fig. 5C). In addition, the lungs from four untreated mice and from five CDDO-Me–treated mice were analyzed histologically. The results indicated that extensive (50%) metastatic lesions were seen in lungs that were isolated from all four control mice, whereas no metastasis was detected in lungs isolated from five CDDO-Me–treated mice (Table 1; Fig. 6). These results show striking antitumor and antimetastatic activity of CDDO-Me treatment in this aggressive murine model of breast tumorigenesis.

Next, we extended our observations by delaying initiation of CDDO-Me treatment from day 1 to day 5 after the breast tumor inoculation. As shown in Fig. 5D, five control mice developed aggressive breast tumors (Fig. 5D,, left). In contrast, at day 30 after tumor inoculation, breast tumors were significantly smaller in 10 of 15 CDDO-Me–treated mice in comparison with controls (P < 0.0001; Fig. 5E). Further, 5 of 15 mice were completely tumor-free after CDDO-Me treatment for up to 95 days. CDDO-Me was given only nine times at 2-day intervals (200 μg/mouse). No significant toxicity was observed in CDDO-Me–treated animals. We measured intratumor CDDO-Me concentration from a total six tumor samples, which averaged from 3.8 to 6.4 μmol/L (Table 2).

The effect of CDDO-Me on splenic mature dendritic cells. To assess potential immunologic effects of CDDO-Me on tumor growth, we analyzed mature dendritic cells isolated from the spleens of control or CDDO-Me–treated tumor-bearing mice. As shown in Fig. 7, the number of mature dendritic cells from untreated tumor-bearing mice was two thirds lower than that from normal mice (Fig. 7C). In contrast, the number of mature dendritic cells in spleens from CDDO-Me–treated mice (Fig. 7B) was almost identical to that of normal untreated control mice (Fig. 7A) without tumors. These results suggest that, in the immunocompetent murine 4T1 breast cancer model, CDDO-Me maintained the mature dendritic cell population, which possibly contributed to the potent antitumor and antimetastatic effect of this agent.

Our results indicate that CDDO-Me effectively inhibits growth and metastases of the aggressive mouse breast tumor cell line 4T1 both in vitro and in vivo (Figs. 3–7). Mechanistic studies showed that CDDO-Me inactivated STAT3 in 4T1 cells after 2 h (Fig. 2A). In our previous studies, we were able to show that CDDO-Me affected ERK and other intracellular components 6 h after treatment (22). Our present data suggest that STAT3 may be the earliest intracellular target to be affected by CDDO-Me. To our knowledge, this is the first report to show that CDDO-Me is capable of inactivating STAT3.

Our investigations also showed that inhibition of phosphorylated STAT3 by CDDO-Me was associated with inactivation of two other potent oncoproteins, Src and Akt (Fig. 2A). The finding that STAT3 inactivation in breast cancer cells resulted in the inactivation of Src and Akt is consistent with findings of our previous study (5), in which knockdown of STAT3 led to the inactivation of Src and Akt in 4T1 cells. We hypothesize that this phenomenon is associated with the requirement of STAT3 in the complex formed by Src and Akt. It is believed that STAT3 relocalizes from the cytoplasm into the nucleus after the protein is phosphorylated (16). However, phosphorylated STAT3 may also remain and function in the cytoplasm, interacting with other signal transducers, such as Src and Akt, and maintain their function. The mechanisms of the activation of Src and Akt by phosphorylated STAT3 or by STAT3 remain largely unknown. Clearly, further studies are needed to clarify this coactivation and its biological sequelae on breast carcinogenesis.

In addition, we showed that CDDO-Me reduces c-Myc mRNA levels. Overexpressed c-Myc has been reported in many types of cancers and cancer cell lines, and the reduction of c-Myc has been shown to suppress tumor growth in vivo (4). In our model, 4T1 cells expressed a detectable amount of c-Myc, and the mRNA of c-Myc was reduced 4-fold by 500 nmol/L CDDO-Me, suggesting that CDDO-Me down-regulated c-Myc through the STAT3 pathway. In parallel experiments, we observed similar results using siRNA technology, whereby a depletion of STAT3 in 4T1 cells resulted in the inactivation of Src, Akt, and c-Myc down-regulation (Fig. 2C). Curiously, in STAT3-depleted 4T1 cells, CDDO-Me induced profound apoptosis as shown in Fig. 3F, suggesting that either (a) complete inactivation of STAT3 by combined CDDO-Me and siRNA exposure is required for the execution of the apoptosis program or (b) CDDO-Me has additional intracellular targets, inactivation of which combined with STAT3 blockade causes synergistic apoptotic response. Some of these targets have been elucidated (e.g., nuclear factor-κB and peroxisome proliferator-activated receptor-γ; refs. 23–25). Constitutively activated STAT3 has been widely reported to function as a potent oncoprotein for human breast cancer (1, 21, 26). Clinical data indicate that a large percentage of human breast cancer samples contain activated STAT3 (1). It has been reported that STAT3 promotes the expression of proteins that are directly involved in cancer cell invasion and metastasis, such as glycogen synthase kinase (GSK)-3β and matrix metalloproteinases (MMP; refs. 28, 29). We here provide evidence that CDDO-Me inactivates STAT3 and blocks breast tumor metastases. The mechanisms of this antimetastatic effect of CDDO-Me require further investigation; GSK and MMPs are potential candidate targets. In addition, cumulative evidence indicates that constitutively activated STAT3 in cancer cells prompts cancer cells to produce factors that impair the host immune response to cancer cells, suggesting that constitutively activated STAT3 plays an important role not only in cell transformation but also in the interaction between cancer cells and host immune cells (30, 31). A recent study indicated that dendritic cell maturation was inhibited by breast cancer development (9, 32). It has been documented that immature dendritic cells suppressed by cancer cells further induce immune tolerance to tumors. However, surgical removal of the tumor can restore the maturation of dendritic cells (9). These findings strongly suggest that the interactions between tumor and immune cells are critically important in tumorigenesis. In our previous study, abolishing STAT3 virtually eliminated the tumor-forming ability of 4T1 cells in immunocompetent mice (5). The results of the current study showed that 4T1 cells were unable to form primary breast and lung metastatic tumors after CDDO-Me treatment, suggesting that CDDO-Me–inactivated STAT3 in 4T1 cells may result in the inhibition of factors that impair dendritic cell maturation, hence reversing immune tolerance to cancer cells. In support of this hypothesis, splenic dendritic cell analysis indicated that the number of mature dendritic cells in 4T1 breast tumor-bearing mice was only one third of that in normal control splenic tissue; in contrast, in CDDO-Me–treated and tumor-free mice, the mature dendritic cell population was the same as in normal control splenic tissue. These results in conjunction with our previous studies suggest that breast tumors can be suppressed by inactivating STAT3, in part, via restoration of dendritic cell maturation.

The antitumor efficacy of CDDO-Me has been extensively investigated in several other studies (15, 18, 25, 33). In our murine model, we observed no induction of tumor cell apoptosis by CDDO-Me but complete inhibition of tumor cell growth, most likely as a consequence of cell signaling and cycle blockade (Fig. 3). Preclinical toxicity study results indicated that CDDO-Me is well tolerated in monkeys and rodents. This finding is supported by our study where no apparent toxicity was observed despite profound antitumor effects. Notably, intratumor CDDO-Me concentrations were above 3 μmol/L levels (Table 2), which by far exceeded concentrations of CDDO-Me required for growth-inhibitory effects in vitro, hence showing excellent tumor tissue accumulation of the compound without obvious systemic toxicity.

To our knowledge, the results presented here are the first evidence that CDDO-Me can inactivate STAT3 in breast cancer cells and can abolish tumor development and metastasis in immunocompetent mice. These results have potential clinical implications because STAT3 is an important target for breast cancer therapy and because CDDO-Me would be an effective agent for inactivating STAT3 and further inhibiting breast tumor growth in vivo. CDDO-Me is currently in phase I clinical trial in patients with solid tumors at M. D. Anderson Cancer Center, and the results of our study strongly suggest breast cancer patients as prime candidates for this novel therapy. Because STAT3 plays an important role in tumorigenesis in different tumor types, we propose that inactivation of STAT3 by CDDO-Me may have therapeutic benefit in other tumors driven by the activation of STAT3 signaling.

Grant support: Susan G. Komen Foundation grant BCTR0504364 and Breast Cancer Specialized Program of Research Excellence grant P50 CA116199 (M. Konopleva), Leukemia Specialized Program of Research Excellence grant 1 P50 CA100632, NIH Cancer Center Support Grant CA16672 (M. Andreeff), and Pathology CORE grant CA16672.

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.

We thank Dr. Didier Trono for providing lentivirus vectors pWIP/GFP, pMD.G2, and p8.91.