Cucurbitacin Q: a Selective STAT3 Activation Inhibitor with Potent Antitumor Activity

Cucurbitacin Q: a selective STAT3 activation inhibitor with potent antitumor activity

Jiazhi Sun1,2, Michelle A Blaskovich1,2, Richard Jove1, Sandra K Livingston1, Domenico Coppola1 and Saı¨ d M Sebti*,1

1Departments of Interdisciplinary Oncology and Biochemistry and Molecular Biology, Drug Discovery and Molecular Oncology Programs, H Lee Mofﬁtt Cancer Center and Research Institute, University of South Florida, 12902 Magnolia Drive, MRC-DRDIS, Tampa, FL 33612-9497, USA

Constitutive activation of the JAK/STAT3 pathway is a Introduction major contributor to oncogenesis.In the present study, structure–activity relationship (SAR) studies with five Signal transducers and activators of transcription cucurbitacin (Cuc) analogs, A, B, E, I, and Q, led to the (STATs) are a family of seven proteins (STATs 1, 2, 3, discovery of Cuc Q, which inhibits the activation of 4, 5a, 5b, and 6) unique in their ability both to transduce STAT3 but not JAK2; Cuc A which inhibits JAK2 but not extracellular signals and regulate transcription directly. STAT3 activation; and Cuc B, E, and I, which inhibit the STATs transduce extracellular signals from cytokines activation of both.Furthermore, these SAR studies such as interleukin-6 and interferons or growth factors demonstrated that conversion of the C3 carbonyl of the such as platelet-derived growth factor (PDGF) and cucurbitacins to a hydroxyl results in loss of anti-JAK2 epidermal growth factor (EGF). Upon activation of activity, whereas addition of a hydroxyl group to C11 of these receptors, STATs are recruited to the plasma the cucurbitacins results in loss of anti-STAT3 activity. membrane where they become activated via phosphor- Cuc Q inhibits selectively the activation of STAT3 and ylation of conserved tyrosine residues either directly by induces apoptosis without inhibition of JAK2, Src, Akt, receptor tyrosine kinases, for example, PDGF receptor Erk, or JNK activation.Furthermore, Cuc Q induces (PDGFR) and EGF receptor (EGFR) or by nonrecep- apoptosis more potently in human and murine tumors that tor tyrosine kinases, for example, Src and JAK. contain constitutively activated STAT3 (i.e., A549, Phosphorylated STAT proteins either homo- or hetero- MDA-MB-435, and v-Src/NIH 3T3) as compared to dimerize via reciprocal phosphotyrosine–SH2 interac- those that do not (i.e., H-Ras/NIH 3T3, MDA-MB-453, tions after which the STAT dimers translocate to the cell and NIH 3T3 cells).Finally, in a nude mouse tumor nucleus where they bind DNA at STAT-specific binding xenograft model, Cuc Q, but not Cuc A, suppresses tumor sites. growth indicating that JAK2 inhibition is not sufficient to In normal cells STAT proteins have been identified as inhibit tumor growth and suggesting that the ability of important regulators of diverse physiological functions Cuc Q to inhibit tumor growth is related to its anti- such as immune response, inflammation, proliferation, STAT3 activity.These studies further validate STAT3 as differentiation, development, cell survival, and apopto- a drug discovery target and provide evidence that sis (Ihle and Kerr, 1995; Schindler and Darnell, 1995; pharmacological agents that can selectively reduce the Horvath and Darnell, 1997; Stark et al., 1998). STAT P-STAT3 levels in human cancer cells result in tumor signaling is tightly regulated in normal cells, either apoptosis and growth inhibition. through inhibition of upstream signaling proteins (e.g., Oncogene (2005) 24, 3236–3245. doi:10.1038/sj.onc.1208470 internalization of receptors) or negative regulators of Published online 21 February 2005 Src and JAK proteins, such as SOCS proteins, and Src family and JAK phosphatases (e.g., CD45 and SHP-2) Keywords: STAT3; apoptosis; antitumor activity; (Irie-Sasaki et al., 2001; Myers et al., 2001; Lefebvre cucurbitacins; JAK2 et al., 2003; Lehmann et al., 2003). STAT proteins have been demonstrated to be directly negatively regulated by SOCs proteins, by protein inhibitors of activated STATs (PIAS ), by SHP phosphatases, and recent evidence has shown both Grb2 and GRIM-19 to be novel regulators of STAT3 activation (Lufei et al., 2003; Zhang et al., *Correspondence: SM Sebti, Drug Discovery Program, H. Lee Moffitt 2003; Wormald and Hilton, 2004). However, in both Cancer Center & Research Institute, 12902 Magnolia Drive, MRC- tumor cells and tissues, disregulation and constitutive DRDIS, Tampa, FL 33612-9497, USA; activation of STATs, especially STAT3 and STAT5, E-mail: sebti@moffitt.usf.edu 2These two authors contributed equally to this work have been demonstrated to be important to the Received 14 October 2004; revised 14 December 2004; accepted 14 proliferation and antiapoptotic activity of tumor cells December 2004; published online 21 February 2005 (Bowman and Jove, 1999; Turkson and Jove, 2000). Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3237 STATs have been shown to play active roles at all In this paper, structure–activity relationship (SAR) levels of tumorigenesis. STATs are responsible for studies have identified cucurbitacin Q (Cuc Q) as generating proproliferative signals (e.g., Cyclin D1, an inhibitor of the activation of STAT3 but not survivin; Sinibaldi et al., 2000; Aoki et al., 2003) JAK2. In contrast, Cuc A was found to be an inhibitor and have been shown to upregulate antiapoptotic of JAK2 but not STAT3 activation. Furthermore, Cuc proteins (e.g., Bcl-XL, Bcl-2; Catlett-Falcone et al., Q but not A induces apoptosis and inhibits human 1999). In addition, STAT3 has been demonstrated tumor growth in mice. Finally, Cuc Q induces apoptosis to upregulate VEGF expression, which is necessary selectively in tumors that contain constitutively acti- for angiogenesis and the maintenance of tumor vascu- vated STAT3 but not in those tumors without activated lature (Niu et al., 2002b). Finally, STAT3 has been STAT3. implicated in the inhibition of immune responses to tumor growth by blocking expression of proinflamma- tory factors (Wang et al., 2004). Unregulated activation Results of STAT3 and STAT5 has been demonstrated in a variety of tumor types, including breast carcinoma, Cuc Q selectively suppresses STAT3 but not JAK2 prostate cancer, melanoma, multiple myeloma, and activation in A549 cells leukemia among others (Shuai et al., 1996; Garcia et al., 1997, 2001; Catlett-Falcone et al., 1999; Mora Having previously identified cucurbitacin I (JSI-124) as et al., 2002; Niu et al., 2002a). Various genetic a potent inhibitor of activation of both JAK2 and alterations can lead to constitutive activation of either STAT3 prompted us to carry out SAR studies to STAT3 or STAT5 (e.g., overexpression of EGFR and identify agents that are selective for inhibiting the ErbB2; Fernandes et al., 1999; Berclaz et al., 2001). activation of either JAK2 or STAT3. To this end, Autocrine and paracrine production of IL-6 results in A549 cells (a human non-small-cell lung carcinoma line) activation of STAT3 in prostate cancer and multiple were treated with either vehicle or cucurbitacin analogs myeloma (Catlett-Falcone et al., 1999; Mora et al., A, B, E, I, or Q (10 mM) for 4 h and the cell lysates 2002), while the oncogene BCR-Abl has been demon- processed for Western blotting with antiphosphotyro- strated to act through constitutive tyrosine phosphor- sine STAT3 (Y705) antibody or antiphosphotyrosine ylation of STAT5 in chronic myelogenous leukemia JAK2 (Y1007, Y1008) antibody as described under (Shuai et al., 1996). Various other tyrosine kinases, for Materials and methods. Figure 1a shows that Cuc Q example, TEL-JAK2, v-Src, and c-Kit, may require suppressed the levels of P-STAT3 but had no effect on activation of downstream signaling pathways including those of P-JAK2. In contrast, Cuc A suppressed the STAT3 and STAT5 (Yu et al., 1995; Cao et al., 1996; levels of P-JAK2 but had no effect on those of P- Ning et al., 2001; Spiekermann et al., 2002; Paner et al., STAT3. Cuc B, E, and I inhibited both P-STAT3 and P- 2003). JAK2 levels (Figure 1a). The fact that Cuc B, E, I, and On the basis of the importance of STAT3 in tumor Q, but not A, suppressed P-STAT3 levels in A549 cells progression and survival, we and others recently have indicates that addition of a single hydroxyl to carbon 11 begun to focus on STAT3 as a viable molecular target of the cucurbitacin pharmacophore results in loss of for cancer chemotherapeutics (Turkson and Jove, 2000). anti-STAT3 activity (Figure 1a; compare Cuc A to B). Several different approaches can be taken for the Similarly, the ability of Cuc A, B, E, and I, but not Q, to inhibition of the STAT signaling pathway: targeting suppress P-JAK2 levels indicates that simple conversion receptor–ligand interactions; inhibition of upstream of the carbon 3 carbonyl in the cucurbitacins to a STAT-activating receptor tyrosine kinases and nonre- hydroxyl results in loss of anti-JAK2 activity (Figure 1a; ceptor tyrosine kinases; activation of STAT phospha- compare cucurbitacin Q to B). tases and other negative regulators of STATs; and To confirm that Cuc Q decreases phosphotyrosine inhibition of STAT dimerization, nuclear translocation, levels of STAT3 without affecting total STAT3 levels, DNA binding, or DNA transcription. Studies with we treated A549 cells with either vehicle control or Cuc antisense, gene therapy, and RNA interference (siRNA) Q (10 mM) for 4 h, immunoprecipitated the lysates (Niu et al., 1999, 2002b; Konnikova et al., 2003) have against whole STAT3, then blotted with both P-STAT3 demonstrated that inhibition of STAT3 signaling and STAT3 antibodies as described under Materials and suppresses tumor growth and induces apoptosis in cell methods. Figure 1b shows that Cuc Q treatment lines and mouse models, validating STAT3 as a target suppressed P-STAT3 without affecting total STAT3 for molecular intervention. Recently, pharmacological levels. We have also shown that treatment of A549 cells approaches to STAT inhibition have resulted in the with 10 mM Cuc I and A, like Cuc Q, does not affect total identification of peptides capable of blocking STAT STAT3 levels, and none of the three compounds affects dimerization (Turkson et al., 2001, 2004) and identifica- total JAK2 levels (data not shown). As further support tion of the natural product curcumin as an inhibitor of of the specific antiphosphotyrosine STAT3, but not the IL-6/JAK/STAT signaling pathway (Bharti et al., antiphosphotyrosine JAK2, activity of Cuc Q, we next 2003). Our laboratory has identified the natural product, treated A549 cells as well as two breast carcinoma cell cucurbitacin I (JSI-124) as a dual inhibitor of phospho- lines (MDA-MB-435 and MDA-MB-468) that also JAK2 and phospho-STAT3 levels in cancer cells express constitutively activated JAK2 and STAT3, with (Blaskovich et al., 2003). Cuc Q at various concentrations, and determined IC50

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3238

Figure 1 SAR studies of cucurbitacins: Effects on signal transduction pathways in A549 Cells. (a) A549 cells were treated with either vehicle control or Cuc A, B, E, I, or Q at 10 mM for 4 h and cell lysates processed for immunoblotting with phospho-speciﬁc antibodies for STAT3, JAK2, Src, Erk1, Erk2, JNK, and Akt antibodies as described under Materials and methods. The ﬁgure also indicates data obtained from both trypan blue exclusion assay and TUNEL staining (reported as average7s.d.), as described under Materials and methods. Data are representative of at least three independent experiments. (b) A549 cells were treated with either vehicle or Cuc Q for 4 h and the lysates immunoprecipitated with anti-STAT3 antibody then immunoblotted with P-STAT3 and STAT3 antibodies as described under Materials and methods. Data are representative of two independent experiments

values of inhibition of STAT3 and JAK2 activation. Cucurbitacins are highly selective for STAT3 and JAK2 Table 1 shows that in all three cell lines, Cuc Q is a over Src, Akt, Erk, and JNK signaling selective inhibitor of STAT3 activation over JAK2 activation, with IC50 values of 3.771.7, 0.970.6, and We next determined whether the Cuc analogs are 1.470.7 mM in A549, MDA-MB-435, and MDA-MB- selective for the JAK2/STAT3 pathway over other 468, respectively. In all three cell lines, JAK2 activation signal transduction pathways. To this end, we treated was not inhibited at Cuc Q concentrations as high as A549 cells with 10 mM of the different Cuc derivatives 10 mM. Cuc A specifically inhibited JAK2 activation and processed the lysates for Western blotting with (IC50s of 1.570.7, 0.6570.05, and 0.86 mM for A549, antibodies specific for phospho-Src, phospho-Erk1/2, MDA-MB-435, and MDA-MB-468, respectively) with- phospho-JNK, and phospho-Akt as described under out affecting STAT3 activation at 10 mM. Cuc I inhibited Materials and methods. Figure 1a shows that A549 cells the activation of both STAT3 and JAK2 but was more possess constitutively phosphorylated Src, Erk1/Erk2, potent towards inhibiting JAK2 activation (Table 1). JNK1, and Akt in addition to phospho-STAT3 and Thus, in all three cell lines Cuc Q inhibits specifically phospho-JAK2. Treatment with Cuc Q, for 4 h at 10 mM STAT3 but not JAK2 activation and Cuc A inhibits significantly blocked STAT3 phosphorylation with little JAK2 but not STAT3 activation whereas Cuc I inhibits effect on phosphotyrosine levels of JAK2, Src, JNK1, or the activation of both STAT3 and JAK2. Akt. In contrast, Cuc A potently inhibited JAK2

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3239

Table 1 IC50 values of inhibition of phosphotyrosine-STAT3 and phosphotyrosine-JAK2 in human tumor cell lines Cuc Q Cuc I Cuc A

Cell line P-STAT3 P-JAK2 P-STAT3 P-JAK2 P-STAT3 P-JAK2

A549 3.771.7 >10 (n ¼ 3) 0.870.7 0.2570.09 >10 (n ¼ 4) 1.570.7 MDA-MB-435 0.970.6 >10 (n ¼ 3) 4.671.9 0.1870.07 >10 (n ¼ 3) 0.6570.05 MDA-MB-468 1.470.7 >10 (n ¼ 2) 7.571.5 0.4070.26 >10 (n ¼ 3) 0.86, 0.86 (n ¼ 2)

Data are representative of at least three independent experiments, unless otherwise indicated phosphorylation, but showed little inhibitory activity cells and human breast carcinoma MDA-MB-435 cells against STAT3, Src, JNK1, and Akt. As noted above, which express very high levels of constitutively activated the other Cuc compounds were able to inhibit both STAT3, and human breast carcinoma, MDA-MB-453, phosphotyrosine-STAT3 and phosphotyrosine-JAK2 which do not show constitutive activation of STAT3 but, like both Cuc Q and A, these compounds showed (Blaskovich et al., 2003; and data not shown), for 24 h little inhibitory effect on phosphotyrosine levels of Src, with 10 mM Cuc Q or DMSO vehicle control. Figure 2a JNK1, and Akt . Interestingly, all of the Cuc analogs shows that Cuc Q only induced apoptosis strongly in the significantly increased the levels of phosphorylated two cell lines expressing activated STAT3, but not in Erk1/2 in A549 cells. Thus, these results demonstrate MDA-MB-453 cells. In A549 cells, Cuc Q increased the that cucurbitacins are highly selective for inhibition of percentage of apoptotic tumor cells by 27.4-fold the JAK/STAT3 pathway activation. compared to vehicle-treated control cells. In MDA- MB-435 cells, Cuc Q increased the percentage of Inhibition of the activation of JAK2, Src, JNK, Akt, apoptotic cells by a 25.9-fold. However, in MDA-MB- and Erk is not required for induction of apoptosis 453 cells, Cuc Q increased this percentage by only 4.7- by cucurbitacins fold (Figure 2a). To further confirm that tumor cells that depend on The results of the SAR studies above prompted us to STAT3 for transformation are more sensitive to Cuc Q- determine whether the ability of the cucurbitacins to induced apoptosis compared to cell lines that do not induce apoptosis is dependent on suppression of P- depend on STAT3, we treated v-Src/3T3 that contain JAK2 and/or P-STAT3 levels. To this end, we treated constitutively-activated STAT3, oncogenic H-Ras/3T3, A549 cells with either vehicle control or cucurbitacins and vector-transfected NIH 3T3 cells that do not (10 mM) for 24 h, harvested the cells, and determined (Garcia et al., 1997; Blaskovich et al., 2003) with tumor cell death (trypan blue exclusion) and apoptosis 10 mM Cuc Q for 24 h. Figure 2b illustrates the results (TUNEL) as described under Materials and methods. from this experiment. As with the human tumor cell Figure 1a shows that the most potent inducer of cell lines, the v-Src/3T3 cell line, with its constitutively death and apoptosis was Cuc Q (60 and 28%, activated STAT3, showed a strong induction of respectively). The least potent was Cuc A (11 and 5%, apoptosis (from 0.870.9% in control compared to respectively). Cuc B, E, and I also induced tumor cell 39.277.3% with Cuc Q treatment, a 50.2-fold increase). death (15–33%) and apoptosis (10–19%). Taken to- In contrast, the H-Ras/3T3 cell line showed significantly gether, the results of Figure 1a demonstrate that less induction of apoptosis (from 0.671.3% in control decreasing P-JAK2 and increasing P-Erk1/2 levels are to only 7.374.7% with Cuc Q treatment, a 12.5-fold not sufficient for significant apoptosis induction, as increase). In vector/3T3 cells, Cuc Q increased the indicated by the low potency of Cuc A. Furthermore, percentage of apoptotic cells by only 4.2-fold (from the results also demonstrate that decreasing the levels of 1.771.8% in control to 7.373.9% with Cuc Q P-JAK2, P-Src, P-JNK, and P-Akt is not required for treatment) (Figure 2b). Coupled with the human tumor induction of apoptosis as indicated by the high potency cell results from Figure 2a, these results demonstrate of Cuc Q. Finally, the results also suggest that the ability that Cuc Q selectively induces more apoptosis in cell of the cucurbitacins to induce apoptosis is related to lines which express activated STAT3 compared to those their ability to suppress P-STAT3 but not P-JAK2 levels with little or no STAT3 activation. in A549 cells (compare Cuc Q to A). Cuc Q inhibits A549 and v-Src transformed NIH 3T3 Induction of apoptosis by Cuc Q is selective for cells tumor growth in nude mice that express constitutively activated STAT3 To determine the ability of the cucurbitacin analogs to Figure 1a SAR studies suggest that Cuc Q induces inhibit tumor growth in vivo, we evaluated the antitumor apoptosis by blocking the activation of STAT3 in A549 activity of the cucurbitacin analogs against both A549 cells. To give further support for this suggestion, we next and v-Src/3T3 tumors in a nude mouse xenograft model. determined whether Cuc Q induced apoptosis selectively When the tumors became palpable (at volumes of in tumor cells that have high levels of activated STAT3 approximately 100–150 mm3), the mice were treated over those that do not. To this end, we treated A549 either with vehicle control or 1 mg/kg/day of the

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3240

Figure 2 Cuc Q induces apoptosis in human tumor cell lines and oncogene-transformed NIH 3T3 cells expressing constitutively activated STAT3. (a) A549, MDA-MB-435, and MDA-MB-453 cells and (b) Vector NIH 3T3, v-Src/3T3, and H-Ras/3T3 cells were treated with either vehicle control or 10 mM Cuc Q and processed for TUNEL staining as described under Materials and methods. Cells were costained with DAPI to detect the nuclei. The table indicates induction of apoptosis by Cuc Q as determined by TUNEL assay

cucurbitacins. Tumor volumes were monitored by to grow on nude mice. These results demonstrate caliper measurement as previously described (Blaskovich that the ability of the Cuc molecules to inhibit tumor et al., 2003) and under Materials and methods. Figure 3 growth is independent of their ability to inhibit JAK2 shows the antitumor efficacy of the cucurbitacin activation. compounds. With A549 xenografts, all compounds except for Cuc A (11.1% inhibition, P ¼ 0.656) showed Immunohistochemical analysis of tumor sections statistically significant inhibition of tumor growth. Cuc for STAT3 activation and apoptosis Q was highly potent, with 73.1% inhibition (P ¼ 0.001) of A549 tumor growth in nude mice (Figure 3 and To determine whether phosphotyrosine STAT3 is Table 2). Cuc I was a potent inhibitor of A549 tumor targeted by Cuc Q in vivo, and to determine if the growth with 55.4% inhibition (P ¼ 0.011). Likewise, Cuc results seen in cell culture concerning induction of B (53.6% inhibition, P ¼ 0.010) and Cuc E (48.5%, apoptosis were occurring in tumors from animals P ¼ 0.024) were significant inhibitors of growth of A549 treated with Cuc Q, on the termination day of the adenocarcinoma in nude mice (Table 2). A549 antitumor experiment, tumors from animals In the v-Src/3T3 xenograft model, again Cuc A treated with Cuc A, I, and Q, as well as vehicle control, treatment did not result in statistically significant were extracted and fixed in 10% neutral-buffered inhibition of tumor growth (16%, P ¼ 0.35). As in formalin and then processed into paraffin blocks for A549 tumors, Cuc Q was highly potent at inhibiting tissue sectioning. These tissue sections were stained the growth of v-Src/3T3 tumors. Cuc Q inhibited 57% separately with either TUNEL for determination of of tumor growth while Cuc I, B, and E inhibited 45, apoptosis, or phosphotyrosine STAT3 to determine if 40, and 42% of tumor growth, respectively (Figure 3 the signaling protein is inhibited in the tumors. Results and Table 2). Taken together, and consistent with the in of IHC staining are summarized in Figure 4. With vitro data of Figure 1a, the results of both xenograft P-STAT3 staining (Figure 4a), it is apparent that both Cuc models show that Cuc Q is a potent and significant Q and I inhibited STAT3 activation in A549 tumors, inhibitor of tumor growth, while Cuc A shows little with Cuc Q more potent than I (22.677.3% P-STAT3 ability to inhibit tumor growth in either model. positive cells for Q and 54.774.5% for I compared to Inhibition of STAT3 activity, with or without the ability 76.571.4% for control; 70.5 and 28.5% inhibition of to inhibit JAK2 activation (as with all cucurbitacins phosphotyrosine-STAT3 with Q and I treatment, tested but Cuc A), results in antitumor activity, whereas respectively). Cuc A showed virtually equal staining inhibition of JAK2 activity, but not STAT3 activity (as for phospho-STAT3 as vehicle control (80.871.8% with Cuc A), does not hinder the ability of the tumors P-STAT3-positive cells), indicating that there was no

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3241

Figure 3 Cuc Q inhibits tumor growth in nude mice of both A549 human tumors cells and v-Src-transformed NIH 3T3 cells. Human lung adenocarcinoma A549 and v-Src-transformed NIH 3T3 cells were implanted s.c. onto the ﬂanks of athymic nude mice. When the tumors reached an average size of 100–150 mm3, the animals were randomized and treated with either vehicle control () or 1 mg/kg/ day of Cuc A (n), E (m), I (J), and Q (&) or 0.5 mg/kg/day Cuc B (B) as described under Materials and methods. **designates Po0.001 and *designates Po0.05

Table 2 Antitumor activity of cucurbitacin analogs inhibition of STAT3 activation. TUNEL staining of v-Src/3T3 A549 tissue sections (Figure 4b) revealed that, while Cuc A

a a showed virtually no induction of TUNEL staining Cucurbitacin % Inhibition P % Inhibition P (0.370.2% TUNEL-positive cells) compared to control A 16 0.35 11.1 0.656 (0.470.1% TUNEL positive), both Cuc Q (14.37 B40a 0.006 53.6b 0.010 2.7%) and I (10.573.0%) showed strong staining for E 42 0.047 48.5 0.024 TUNEL, indicating the induction of apoptosis in the I 45 0.003 55.4 0.011 A549 cells comprising the tumors. As with the cell work, Q 57 0.001 73.1 0.002 we can see that only the two compounds that inhibit aTwo-sided Student’s t-test bToxic at 1 mg/kg/day; results shown here STAT3 activation demonstrate an ability to induce are for 0.5 mg/kg/day. apoptosis.

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3242

Figure 4 Immunohistochemical analysis of tumors for phosphotyrosine STAT3 and TUNEL staining. A549 tumor sections were stained as described under Materials and methods with P-STAT3 antibody and dTd (TUNEL) enzyme for the determination of cucurbitacin activity in the target tumor in vivo. Treatment conditions were: control (C); 1 mg/kg/day Cuc Q; 1 mg/kg/day Cuc I; 1 mg/ kg/day Cuc A. (a) Cells stained positive for phospho-STAT3 were scored and percent inhibition of STAT3 activation determined by comparison to vehicle control. (b) Cells stained positive for TUNEL were scored and induction of apoptosis determined by comparison to vehicle control. For both graphs, *indicates Po0.05; **indicates Po0.005. Data were determined by counting sections from eight independent tumors. Data are representative of two independent experiments

Discussion Identifying compounds that are highly selective for either STAT3 or JAK2 allowed us to address important Over the last decade overwhelming evidence has issues concerning the involvement of STAT3 vs JAK2 in accumulated demonstrating the intimate involvement human cancer cell survival. Our studies suggest that of STAT3 in malignant transformation and tumor suppressing STAT3 activation is more detrimental to survival. This prompted us and others to develop tumor survival than blocking JAK2 activation. Indeed, inhibitors of STAT3 function as novel anticancer drugs. both in cultured cells as well as in nude mouse To this end, we have used two approaches, one targeting xenografts, Cuc A, which blocks JAK2 but not STAT3 STAT3 dimerization (Turkson et al., 2001, 2004), a step activation, was a poor inducer of apoptosis and an required for STAT3 activation and translocation to the ineffective inhibitor of tumor growth. Furthermore, all nucleus; and the other, inhibition of the activation of three cucurbitacins (Cuc I, E, and B) that inhibit the STAT3 by reducing its cellular phosphotyrosine levels activation of both STAT3 and JAK2 were less active at (Blaskovich et al., 2003). Recently, using a phosphotyr- inducing apoptosis and inhibiting tumor growth sug- osine-STAT3 cytoblot to evaluate the NCI diversity gesting that inhibition of JAK2 activation may hinder set chemical library, we discovered Cuc I, which the antitumor activity of cucurbitacins. inhibited both STAT3 and JAK2 activation (Blaskovich Cancer is a result of many genetic alterations resulting et al., 2003). In the present study, SAR studies with in numerous aberrant signal transduction pathways ﬁve cucurbitacin analogs led us to a highly selective (Hanahan and Weinberg, 2000). Although activation of STAT3 activation inhibitor, Cuc Q; a highly selective STAT3 is a major contributor to malignant transforma- inhibitor of JAK2 activation, Cuc A; and three dual tion, other pathways such as those that mediate the inhibitors, Cuc I, E, and B. From the chemical point of action of the Ras and Src oncoproteins play pivotal roles view, these are very important ﬁndings that indicate in oncogenesis and tumor survival. An important that addition of a single hydroxyl groupto carbon 11 of question is whether suppression of all aberrant pathways the cucurbitacins results in loss of anti-STAT3 activity, is necessary for inducing tumor cell death. In this paper, whereas a simple conversion of a carbon 3 carbonyl to a we have demonstrated that Cuc Q, I, E, and B induced hydroxyl leads to loss of anti-JAK2 activity (see apoptosis without inhibiting the activation of Src, Akt, Figure 1). Erk1/2, and JNK, suggesting that the suppression of

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3243 STAT3 activation is sufficient for apoptosis induction. Cucurbitacin analogs This is consistent with the notion that many genetic All cucurbitacin compounds were obtained from the National alterations need to accumulate for cancer development Cancer Institute: cucurbitacin A (NSC #94743), cucurbitacin B and consequently suppressing one of these could be (NSC #49451), cucurbitacin E (NSC #106399), cucurbitacin I sufficient for reversal of malignant transformation. (NSC #521777), cucurbitacin Q (NSC #135075). The fact that Cuc Q inhibits STAT3 activation whereas Cuc A inhibits JAK2 activation suggests that these compounds have different targets. The actual Western blotting biochemical targets for cucurbitacins are not known. Treated cell samples were lysed in 30 mM HEPES, pH 7.5, The lowering of phosphotyrosine levels, suggest that 10 mM NaCl, 5 mM MgCl2,25mM NaF, 1 mM EGTA, 1% these agents either inhibit upstream tyrosine kinases or Triton X-100, 10% glycerol, 2 mM sodium orthovanadate, activate upstream phosphotyrosine phosphatases. Pos- 10 mg/ml aprotinin, 10 mg/ml soybean trypsin inhibitor, 25 mg/ sible tyrosine kinases that could be targets are the Src ml leupeptin, 2 mM PMSF, and 6.4 mg/ml p-nitrophenylpho- family of kinases. Src kinase itself was not inhibited in sphate. Phospho-STAT3, phospho-AKT, phospho-Src, and vitro by Cuc I (Blaskovich et al., 2003) and Q (data not phospho-p42/p44 MAPK antibodies were obtained from Cell Signaling Technologies (Cambridge, MA, USA). Phospho- shown). We do not know if other Src kinase family JNK and whole STAT3 antibodies were purchased from Santa members such as Lyn and Fyn are affected. Similarly, Cruz Biotechnology (Santa Cruz, CA, USA); phospho-JAK2 although JAK2 kinase in vitro was not inhibited by Cuc antibody came from Upstate Biotechnology (Lake Placid, NY, I (Blaskovich et al., 2003) and Q (data not shown), we USA). Membranes were blocked in either 5% milk in do not know if other JAKs such as JAK1 are affected. phosphate-buffered saline (PBS), pH 7.4, containing 0.1% Phosphatases that are known to downregulate STAT3 Tween-20 (PBS-T) or 1% BSA in tris-buffered saline (TBS), such as SHP-2 as well as other downregulators of pH 7.5, containing 0.1% Tween-20 (TBS-T). Phospho-specific STAT3 such as SOCS and PIAS could also be targets. antibodies (excepting P-MAPK and P-JNK) were incubated in Cuc Q and A are chemically very similar (see 1% BSA in TBS-T while all other antibodies were diluted in Figure 1a), yet their biological and physiological effects 5% milk in PBS-T for either 2 h at room temperature or overnight at 41C. HRP-conjugated secondary antibodies are very distinct. Cuc A inhibited JAK2 but not STAT3 (Jackson ImmunoResearch, West Grove, PA, USA) were activation and was not able to induce apoptosis and diluted in 5% milk in either PBS-T or TBS-T at 1 : 1000 inhibit tumor growth of the A549 lung tumors in nude dilution for 1 h at room temperature. Western blots were mice. In contrast, Cuc Q inhibited STAT3 but not JAK2 visualized using enhanced chemiluminescence. activation and was very potent at inducing apoptosis and at inhibiting A549 tumor growth in the same animal model. Furthermore, in cultured human cancer cells and STAT3 immunoprecipitation oncogene-transformed murine cells, Cuc Q induced A549 cells were treated for 4 h with vehicle or Cuc Q, then programmed cell death much more efficiently in those lysed in 150 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM EDTA, tumors with constitutively activated STAT3. These SAR 0.5% NP-40, 10% glycerol, 5 mM NaF, 1 mM DTT, 1 mM and in vitro/in vivo studies suggest that inactivation of PMSF, 2 mM sodium orthovanadate, and 5 mg/ml leupeptin. JAK2 is not sufficient and that selective inhibition of Sample lysates were collected and cleared, then 500 mg of lysate STAT3 with pharmacological agents can lead to tumor was immunoprecipitated with 50 ng STAT3 antibody over- cell death. This is consistent with previous studies that night at 41C, then rocked with 25 ml Protein A/G PLUS 1 demonstrated that a dominant-negative form of STAT3 agarose (Santa Cruz Biotechnology) for 1 h at 4 C. Samples were washed four times with lysis buffer, then boiled in 2 Â (STAT3-beta) can induce apoptosis in human cancer SDS–PAGE sample buffer and run on 10% SDS–PAGE gel. cells (Niu et al., 1999; Turkson and Jove, 2000). Protein was transferred to nitrocellulose then blotted as above In conclusion, we have discovered compounds that for both phospho-specific STAT3 and STAT3. are highly selective for disrupting JAK2 or STAT3 signaling and that can be used as chemical probes to dissect the importance of these signal transduction Antitumor activity in the nude mouse tumor xenograft model circuits in normal and pathophysiological conditions. Nude mice (Charles River, Wilmington, MA, USA) were We have used these probes to demonstrate that maintained in accordance with the Institutional Animal Care disruption of STAT3, not JAK2, function is more and Use Committee (IACUC) procedures and guidelines. detrimental to tumor survival. These results give further A549 cells were harvested, resuspended in PBS, and injected support for the use of STAT3 as a molecular therapeutic subcutaneously (s.c.) into the right and left flank (1 Â 107 cells target to combat cancer. per flank) of 8-week-old female nude mice as reported previously (Blaskovich et al., 2003). When tumors reached about 150 mm3, animals were randomized (four animals per group; two tumors per animal) and dosed intraperitoneally Materials and methods (i.p.) either with cucurbitacin analogs (0.5 or 1 mg/kg/day, i.p.) in 20% DMSO in water or with an equal volume of vehicle Cell lines control. The tumor volumes were determined by measuring the All human tumor cell lines used were obtained from American length (l) and the width (w) and calculating the volume Type Culture Collection (Manassas, VA, USA). Stably (V ¼ lw2/2) as described previously (Blaskovich et al., 2003). transfected v-Src/NIH 3T3 cell line has been described earlier Statistical significance between control and treated animals (Turkson et al., 1999). were evaluated by using Student’s t-test.

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3244 In vitro cellular proliferation and TUNEL assays with 3% hydrogen peroxide and nonspecific binding with 2% normal goat serum in 3% BSA/PBS. Sections then were Subconfluent A549, MDA-MB-435, MDA-MB-453, MDA- incubated overnight with 1 : 400 phospho-STAT3 (Cell Signal- MB-468, v-Src transformed NIH 3T3 (v-Src/3T3), H-Ras ing Technologies) at 41C in a humidified chamber. Detection transformed NIH 3T3 (H-Ras/3T3), and vector NIH 3T3 cells was performed using the Elite ABC Rabbit kit (Vector were grown in the presence of 10 mM cucurbitacin A, I, or Q or Laboratories) and DAB chromogen (DakoCytomation Cali- DMSO vehicle control. After 24 h, cells were harvested by fornia, Inc.) according to the manufacturer’s instructions. trypsinization and counted via trypan blue exclusion assay to Slides were counterstained for 20–30 s with modified Mayer’s determine cellular viability. In all, 75 000–150 000 cells hematoxylin, dehydrated through ascending alcohol, cleared, (depending on cell line) were then spun onto glass slides using and mounted with resinous mounting medium. Quantification a Cytospin 3 centrifuge (Thermo Shandon Inc., Pittsburgh, was performed by counting both the phospho-STAT3-positive PA, USA). After fixing cells to the slides with 4% parafor- and -negative cells on slides representative of eight tumors and maldehyde in PBS, pH 7.5, for 1 h at room temperature, cells significance was determined by Student’s t-test. were labeled for apoptotic DNA strand breaks by TUNEL reaction using an in situ cell death detection kit (Roche Applied Science, Indianapolis, IN, USA) according to the TUNEL immunohistochemistry manufacturer’s instructions, then mounted in Vectashield mounting medium (Vector Laboratories, Burlingame, CA, Tumors were harvested, frozen, and dewaxed as described for USA) containing 40,6-diamidino-2-phenylindole (DAPI) to P-STAT3 immunohistochemistry. Tissue sections (5 mm) were counterstain DNA. Fluorescein-labeled DNA strand breaks digested for 10 min with 25 mg/ml proteinase K in PBS and (TUNEL-positive cells) were then visualized using a fluores- then washed thoroughly. Peroxidases were quenched with 3% cent microscope (Leica Microsystems Inc., Bannockburn, IL, hydrogen peroxide in PBS and washed. Sections were USA) and pictures taken with a digital camera (Diagnostic equilibrated with equilibration buffer, then incubated in 30% Instruments, Inc., Sterling Heights, MI, USA). TUNEL- TdT enzymes/70% digoxigenin nucleotidyl reaction buffer for positive nuclei were counted and compared to DAPI-stained 1 h at 371C in a humidified chamber. The labeling reaction was nuclei to determine the percent induction of apoptosis by the stopped in stop/wash buffer with moderate shaking. Slides different cucurbitacin compounds. Statistical significance then were placed on the Dako Autostainer and incubated with between control and treated tumors were evaluated by using antidigoxigenin-peroxidase (DakoCytomation California, Student’s t-test. Inc.) for 30 min using DAB substrate. Sections were counterstained with methyl green (Vector Laboratories), dehydrated P-STAT3 immunohistochemistry through ascending alcohol, cleared, and mounted with resinous mounting medium. The quantification was performed On the termination day of the A549 antitumor experiment, by counting both the TUNEL-positive and -negative cells on tumors were extracted and fixed in 10% neutral-buffered slides representative of eight tumors and significance was formalin for 6 h. After fixation, the tissue samples were determined by Student’s t-test. processed into paraffin blocks. Tissue sections (5 mm) were dewaxed with xylene and rehydrated through descending alcohol to deionized water and then placed in PBS. Antigens Acknowledgements were retrieved briefly with citrate buffer, pH 6.0, in a We thank Drs Edward Sausville, Jill Johnson, George microwave followed by a mild trypsinization (0.025% trypsin Johnson, Daniel Zaharevitz, Robert Schultz, and John Beutler in 50 mM Tris buffer containing 0.05% calcium chloride, pH from the NCI Developmental Therapeutics Program for 7.6). From this point, all steps were carried out in a DAKO providing us with the cucurbitacin compounds. We also thank Autostainer (DakoCytomation California, Inc., Carpinteria, the Pathology Core at the H Lee Moffitt Cancer Center and CA, USA). Sections were rinsed three times in TBS-Tween Research Institute. This work was supported by the National buffer, pH 7.6, then endogenous peroxidases were quenched Cancer Institute Grant CA78038 (SMS).

References

Aoki Y, Feldman GM and Tosato G. (2003). Blood, 101, 1535–1542. A, Gazit A, Levitzki A, Kraker A and Jove R. (2001). Berclaz G, Altermatt HJ, Rohrbach V, Siragusa A, Dreher E Oncogene, 20, 2499–2513. and Smith PD. (2001). Int. J. Oncol., 19, 1155–1160. Garcia R, Yu CL, Hudnall A, Catlett R, Nelson KL, Smithgall Bharti AC, Donato N and Aggarwal BB. (2003). J. Immunol., T, Fujita DJ, Ethier SP and Jove R. (1997). Cell Growth 171, 3863–3871. Differ., 8, 1267–1276. Blaskovich MA, Sun J, Cantor A, Turkson J, Jove R and Sebti Hanahan D and Weinberg RA. (2000). Cell, 100, 57–70. SM. (2003). Cancer Res., 63, 1270–1279. Horvath CM and Darnell JE. (1997). Curr. Opin. Cell Biol., 9, Bowman T and Jove R. (1999). Cancer Control, 6, 615–619. 233–239. Cao X, Tay A, Guy GR and Tan YH. (1996). Mol. Cell. Biol., Ihle JN and Kerr IM. (1995). Trends Genet., 11, 69–74. 16, 1595–1603. Irie-Sasaki J, Sasaki T, Matsumoto W, Opavsky A, Cheng M, Catlett-Falcone R, Landowski TH, Oshiro MM, Turkson J, Welstead G, Grifﬁths E, Krawczyk C, Richardson CD, Levitzki A, Savino R, Ciliberto G, Moscinski L, Fernandez- Aitken K, Iscove N, Koretzky G, Johnson P, Liu P, Luna JL, Nunez G, Dalton WS and Jove R. (1999). Rothstein DM and Penninger JM. (2001). Nature, 409, Immunity, 10, 105–115. 349–354. Fernandes A, Hamburger AW and Gerwin BI. (1999). Int. J. Konnikova L, Kotecki M, Kruger MM and Cochran BH. Cancer, 83, 564–570. (2003). BMC Cancer, 3, 23. Garcia R, Bowman TL, Niu G, Yu H, Minton S, Muro-Cacho Lefebvre DC, Felberg J, Cross JL and Johnson P. (2003). CA, Cox CE, Falcone R, Fairclough R, Parsons S, Laudano Biochim. Biophys. Acta, 1650, 40–49.

Oncogene Inhibition of STAT3 activation with cucurbitacin Q J Sun et al 3245 Lehmann U, Schmitz J, Weissenbach M, Sobota RM, Hortner Shuai K, Halpern J, ten Hoeve J, Rao X and Sawyers CL. M, Friederichs K, Behrmann I, Tsiaris W, Sasaki A, (1996). Oncogene, 13, 247–254. Schneider-Mergener J, Yoshimura A, Neel BG, Heinrich Sinibaldi D, Wharton W, Turkson J, Bowman T, Pledger WJ PC and Schaper F. (2003). J. Biol. Chem., 278, 661–671. and Jove R. (2000). Oncogene, 19, 5419–5427. Lufei C, Ma J, Huang G, Zhang T, Novotny-Diermayr V, Ong Spiekermann K, Pau M, Schwab R, Schmieja K, Franzrahe S CT and Cao X. (2003). EMBO J., 22, 1325–1335. and Hiddemann W. (2002). Exp. Hematol., 30, 262–271. Mora LB, Buettner R, Seigne J, Diaz J, Ahmad N, Garcia R, Stark GR, Kerr IM, Williams BR, Silverman RH and Bowman T, Falcone R, Fairclough R, Cantor A, Muro- Schreiber RD. (1998). Annu. Rev. Biochem., 67, 227–264. Cacho C, Livingston S, Karras J, Pow-Sang J and Jove R. Turkson J, Bowman T, Adnane J, Zhang Y, Djeu JY, (2002). Cancer Res., 62, 6659–6666. Sekharam M, Frank DA, Holzman LB, Wu J, Sebti S and Myers MP, Andersen JN, Cheng A, Tremblay ML, Horvath Jove R. (1999). Mol. Cell. Biol., 19, 7519–7528. CM, Parisien JP, Salmeen A, Barford D and Tonks NK. Turkson J and Jove R. (2000). Oncogene, 19, 6613–6626. (2001). J. Biol. Chem., 276, 47771–47774. Turkson J, Kim JS, Zhang S, Yuan J, Huang M, Glenn M, Ning ZQ, Li J and Arceci RJ. (2001). Blood, 97, 3559–3567. Haura E, Sebti S, Hamilton AD and Jove R. (2004). Mol. Niu G, Bowman T, Huang M, Shivers S, Reintgen D, Daud A, Cancer Ther., 3, 261–269. Chang A, Kraker A, Jove R and Yu H. (2002a). Oncogene, Turkson J, Ryan D, Kim JS, Zhang Y, Chen Z, Haura E, 21, 7001–7010. Laudano A, Sebti S, Hamilton AD and Jove R. (2001). J. Niu G, Heller R, Catlett-Falcone R, Coppola D, Jaroszeski M, Biol. Chem., 276, 45443–45455. Dalton W, Jove R and Yu H. (1999). Cancer Res., 59, 5059– Wang T, Niu G, Kortylewski M, Burdelya L, Shain K, Zhang 5063. S, Bhattacharya R, Gabrilovich D, Heller R, Coppola D, Niu G, Wright KL, Huang M, Song L, Haura E, Turkson J, Dalton W, Jove R, Pardoll D and Yu H. (2004). Nat. Med., Zhang S, Wang T, Sinibaldi D, Coppola D, Heller R, Ellis 10, 48–54. LM, Karras J, Bromberg J, Pardoll D, Jove R and Yu H. Wormald S and Hilton DJ. (2004). J. Biol. Chem., 279, (2002b). Oncogene, 21, 2000–2008. 821–824. Paner GP, Silberman S, Hartman G, Micetich KC, Aranha Yu CL, Meyer DJ, Campbell GS, Larner AC, Carter-Su C, GV and Alkan S. (2003). Anticancer Res., 23, 2253–2260. Schwartz J and Jove R. (1995). Science, 269, 81–83. Schindler C and Darnell Jr JE. (1995). Annu. Rev. Biochem., Zhang T, Ma J and Cao X. (2003). Biochem. J., 376, 64, 621–651. 457–464.

Oncogene