Development of potent small-molecule inhibitors to drug the undruggable steroid receptor coactivator-3

Significance Steroid receptor coactivator-3 (SRC-3) sits at the nexus of many intracellular signaling pathways critical for cancer formation and proliferation. Although the oncogenic role of SRC-3 has been well established in breast and other cancers, coactivators are usually considered as “undruggable” because of their large and flexible structures. Herein, we developed SI-2 as a new class of potent small-molecule inhibitors for SRC-3. SI-2 can selectively reduce the transcriptional activities and the protein concentrations of SRC-3 in cells and significantly inhibit primary tumor growth in a breast cancer mouse model. This work not only has the potential to improve breast cancer treatment, but also to provide a viable strategy to target often “undruggable but important” protein targets without ligand-binding sites. Protein–protein interactions (PPIs) play a central role in most biological processes, and therefore represent an important class of targets for therapeutic development. However, disrupting PPIs using small-molecule inhibitors (SMIs) is challenging and often deemed as “undruggable.” We developed a cell-based functional assay for high-throughput screening to identify SMIs for steroid receptor coactivator-3 (SRC-3 or AIB1), a large and mostly unstructured nuclear protein. Without any SRC-3 structural information, we identified SI-2 as a highly promising SMI for SRC-3. SI-2 meets all of the criteria of Lipinski’s rule [Lipinski et al. (2001) Adv Drug Deliv Rev 46(1-3):3–26] for a drug-like molecule and has a half-life of 1 h in a pharmacokinetics study and a reasonable oral availability in mice. As a SRC-3 SMI, SI-2 can selectively reduce the transcriptional activities and the protein concentrations of SRC-3 in cells through direct physical interactions with SRC-3, and selectively induce breast cancer cell death with IC50 values in the low nanomolar range (3–20 nM), but not affect normal cell viability. Furthermore, SI-2 can significantly inhibit primary tumor growth and reduce SRC-3 protein levels in a breast cancer mouse model. In a toxicology study, SI-2 caused minimal acute cardiotoxicity based on a hERG channel blocking assay and an unappreciable chronic toxicity to major organs based on histological analyses. We believe that this work could significantly improve breast cancer treatment through the development of “first-in-class” drugs that target oncogenic coactivators.

P rotein-protein interactions (PPIs) play a central role in most biological processes, and therefore represent an important class of targets for therapeutic development (1). Biologics-based therapeutics, such as antibodies, exemplify success in PPI regulation (2). However, antibodies usually can only be applied to protein targets on cell surfaces because of their impermeability to plasma membranes (2). Although small-molecule drugs can readily cross membranes, applying small-molecule inhibitors (SMIs) to disrupt PPIs is a challenging task because ∼750-1,500 Å 2 of protein surface area is involved at the interface of PPIs (3), which is too large for SMIs to cover. In addition, these interacting protein surfaces do not have pocket-like small-molecule binding sites (2). Therefore, these PPI sites are deemed as "undruggable" targets for SMIs. The Holy Grail of drug development is to render small molecules the power of biologics to regulate PPIs.
The current strategies for designing small-molecule PPI inhibitors primarily rely on the structural information of the protein targets (4). Clackson and Wells discovered that only a small set of residues at the PPI interface are critical for their interactions, known as "hot spots" (5). Therefore, current drug design for PPIs is mainly focused on small hot spots that can be covered by a drugsized molecule. Unfortunately, many important proteins do not have structural information available or well-defined structures, such as intrinsically disordered proteins. Alternative drug-discovery strategies are urgently needed to target this subset of proteins without knowledge of structural information.
Taking estrogen receptor-positive (ER + ) breast cancer as an example, cancer cells can use a number of mechanisms to overcome selective estrogen receptor modulators to silence the NR activity. Although breast cancer cells can become resistant to endocrine therapies, it is essential for them to recruit coactivators to survive. Earlier efforts have been focused on developing peptides and SMIs to interfere with the interactions between NRs and coactivators (12)(13)(14). A major drawback of this strategy is that overexpression of coactivators, a hallmark of endocrine resistance, often occurs regardless of the context of which NR is expressed in the cancer cell. Coactivators also partner with other transcription factors; therefore, SMIs that can directly target the overexpressed coactivators and reduce their activity or stability should be preferred for drug development.
Identification of SMIs for coactivators is challenging because coactivators are usually considered as undruggable because of their large and flexible structures (6)(7)(8)(9). We recently developed a cell-based functional assay for high-throughput screening to identify SMIs for steroid receptor coactivator-3 (SRC-3). Without any SRC-3 structural information, we identified and improved a Significance Steroid receptor coactivator-3 (SRC-3) sits at the nexus of many intracellular signaling pathways critical for cancer formation and proliferation. Although the oncogenic role of SRC-3 has been well established in breast and other cancers, coactivators are usually considered as "undruggable" because of their large and flexible structures. Herein, we developed SI-2 as a new class of potent small-molecule inhibitors for SRC-3. SI-2 can selectively reduce the transcriptional activities and the protein concentrations of SRC-3 in cells and significantly inhibit primary tumor growth in a breast cancer mouse model. This work not only has the potential to improve breast cancer treatment, but also to provide a viable strategy to target often "undruggable but important" protein targets without ligand-binding sites.
series of SMIs that can target SRC-3 (15)(16)(17). We initially reported gossypol as our first "proof-of-concept" SRC-3 SMI (17). Despite the encouraging success of gossypol as the first selective SRC-3 SMI, the IC 50 values of gossypol are in the micromolar range, which is suboptimal for drug development and may cause off-target toxicity (17). Subsequently, we reported bufalin, a cardiac glycoside, as a potent SRC-3 SMI (16). Bufalin is an active component in the Chinese medicine Huachansu, which is prepared from the skin and parotid venom glands of the Asiatic toad. Bufalin directly binds to SRC-3 in its receptor interacting domain (RID) and selectively reduces the concentration of SRC-3 in breast, lung, and pancreatic cancer cell lines without perturbing overall protein expression patterns (16). Additionally, bufalin is selectively toxic to cancer cells with IC 50 values in the low nanomolar range, and normal cell viability is not affected (16,18,19). Importantly and most excitingly, bufalin sensitizes breast and lung cancer cells to the inhibitory effects of other chemotherapeutics and inhibits primary tumor growth in vivo. Other groups also reported that bufalin and other cardiac glycosides can inhibit transcription factors and induce synergistic immune responses (19)(20)(21)(22)(23)(24)(25)(26)(27)(28). Unfortunately, cardiac glycosides are well known for their cardiotoxicity. Although we have developed bufalin nanoparticles (16) and a phospho-bufalin prodrug (29) to reduce cardiotoxicity, the potential of cardiac glycosides to cause cardiac arrest dampens the enthusiasm for their clinical use.
In this report, we describe a new class of SRC-3 SMI: SRC-3 inhibitor-2 (SI-2). Unlike gossypol and bufalin, SI-2 is an unnatural compound and identified through a combination of highthroughput screening and medicinal chemistry optimization. SI-2 meets all of the criteria for drug-like compounds. Because of the well-established roles of SRC-3 in endocrine resistance and tumor metastasis in breast cancer, we tested the therapeutic efficacy of SI-2 in a series of breast cancer cell lines and an orthotopic triple-negative breast cancer (TNBC) mouse model. Similar to bufalin, SI-2 can selectively reduce the protein concentrations and transcriptional activities of SRC-3, and selectively kill breast cancer cells with IC 50 values in the low nanomolar range but not affecting normal cell viability. Different from bufalin, SI-2 has a much improved toxicity and pharmacokinetic profile. In addition, an animal study showed that SI-2 can significantly inhibit breast tumor growth in vivo without any observable toxicity. Based on our strong data, we envision that SI-2 is a promising drug candidate as an SRC-3 SMI that could potentially expand breast cancer treatment. In addition, our efforts to identify coactivator SMIs will build confidence for other researchers to develop approaches to target relatively unstructured, regulatory proteins that are designated as "important but difficult" targets in the future.

Results
Identification of SI-2 as a Potent SRC-3 SMI. Through a highthroughput compatible luciferase assay, we screened with the Molecular Libraries Probe Production Centers (MLPCN) library of the NIH to identify compounds targeting the intrinsic transcriptional activities of SRC-3 (PubChem AID 588352). In these assays, we evaluated the effects of compounds by measuring the output of a GAL4-responsive luciferase reporter (pGL-LUC) in the presence of GAL4 DNA binding (DBD) SRC coactivator fusion proteins (pBIND-SRC) (30). Compounds that inhibit the intrinsic ability of SRC coactivators to activate transcription will lead to a decrease in expression of the luciferase gene, resulting in reduced luminescence. Additionally, active compounds also were tested in a counter screen using cells transfected with a DBD-VP16 fusion protein to exclude those perturbing pGL/pBIND systems (AID 588794). Only those compounds that reduce SRC-3 activity greater than they do toward the VP16 control were defined as potential SRC-3 inhibitors. Active compounds retrieved from the primary screens were clustered according to structure similarities in PubChem. Through dose-dependent toxicity studies, we identified SI-1 (Fig. S1) as a promising candidate for further development. SI-1 has a submicromolar IC 50 value in a MDA-MB-468 cell line, a TNBC cell line used as a model in this study.
Subsequently, we performed a series of medicinal chemistry optimizations to enhance the potency of SI-1. We discovered that SI-2 ( Fig. 1A), which is formed by introduction of N-methylation to the benzimidazole ring in SI-1, can decrease the IC 50 value by ∼60-fold in the same TNBC cell line. In addition, we also substituted the pyridine ring in SI-2 with a series of hetero-aromatic groups (Fig. S1). Unfortunately, this substitution significantly reduced the potency of these compounds in MDA-MB-468 cells. In this study, we chose SI-2 as a leading SRC-3 SMI candidate in the following in vitro and in vivo tests. Drug-Like Properties of SI-2. SI-2 is a drug-like molecule and meets all of the criteria for Lipinski's rule (31), Veber's rule (32), and Oprea's rule (33) of drug-likeness (Fig. 1B). SI-2 has a molecular weight of 265 g·mol −1 and an experimental LogP value of 0.44. In addition, SI-2 has five rotatable bonds and its numbers of hydrogen bond (H-bond) donor and acceptor are 1 and 4, respectively. The molecular polar surface area of SI-2 is calculated to be 52 Å 2 based on a topological method (34), which is well below 140 Å 2 and suggests good oral availability based on Veber et al. (32) rules. The thermodynamic and kinetic solubilities of SI-2 in PBS are 168 and 327 μg/mL, respectively. SI-2 Inhibits Intrinsic Transcriptional Activity of SRC-3. We investigated the effects of SI-2 on the intrinsic transcriptional activities of SRC-3. HeLa cells were transiently transfected with a pGL5-LUC reporter and expression vectors for pBIND and pBIND-SRC-3, followed by 24 h of treatment with different concentrations of SI-2. As shown in Fig. 1C, SI-2 significantly reduced the luciferase reporter activities in cells transfected in pBIND-SRC-3 in a dosedependent manner, but only minimally affected the activity of pBIND. This result suggests that SI-2 can selectively inhibit the intrinsic transcriptional activities of SRC-3. Similar to bufalin, SI-2 also can inhibit the transcriptional activities of SRC-1 and SRC-2 ( Fig. S2A), the other two members of the p160 family of steroidreceptor coactivators.

SI-2 Selectively Inhibits SRC-3 Protein Levels Posttranscriptionally.
We showed that the steady-state levels of coactivator proteins correlate with their transcriptional activities and with cancer progression (35). Therefore, we studied the effects of SI-2 on SRC-3 protein levels in MDA-MB-468 breast cancer cells after 24 h of incubation. As shown in Fig. 1D, SI-2 selectively reduces cellular protein levels of SRC-3, but not that of coactivatorassociated arginine methyltransferase 1 (CARM-1), which is part of a multiprotein coactivator complex with SRC-3. Furthermore, inhibition of SRC-3 levels also was observed in other breast cancer cell lines, including endocrine sensitive MCF-7, T47D cells, and endocrine-resistant BT-474 cells (Fig. 1E). In addition, SI-2 can inhibit the protein levels of SRC-1 and SRC-2, but to a lesser extent than for SRC-3, especially at lower concentrations ( Fig.  S2 B and C).
To gain insights into the mechanism of SI-2-mediated SRC-3 protein down-regulation, we assessed whether 24 h of SI-2 treatment affected the production of mRNAs for each SRC family member in MDA-MB-468 cells. Quantitative PCR (qPCR) revealed that mRNA levels for SRC-1 and SRC-2 were not significantly altered (Fig. S2D), whereas the mRNA levels for SRC-3 were actually increased upon SI-2 incubation (Fig. 1F). This result is consistent with the mRNA level changes for the SRC family in bufalin-treated cells (16) and suggests that SI-2 reduces SRC protein levels posttranscriptionally.
SRC-3 Is Required for SI-2 Activity. It is well established that targeting SRC-3 with siRNAs inhibits cell growth in many cancer types (7,36). Considering the fact that the decreased cell viability induced by SI-2 is accompanied by reduced SRC-3 protein levels, we sought to investigate the specific role of the SRC-3 protein in blocking cancer cell proliferation. Previously, we used a zinc finger nuclease to knockout both SRC-3 alleles in the HeLa cell line and developed HeLa SRC-3 KO cells (16). Compared with parental SRC-3 +/+ cells, the response of HeLa SRC-3 KO cells to SI-2 treatment is reduced by ∼fivefold (Fig. 1G). This finding supports the idea that SRC-3 protein is involved in mediating the cell response to SI-2 treatment. However, the remaining response of HeLa SRC-3 KO cells to SI-2 is likely because of the SRC-1 and SRC-2 that continue to be expressed in these cells and that also respond to SI-2 at a similar dose as SRC-3, in addition to any unknown off-target actions of the compound. SI-2 Inhibits SRC-3 Through Direct Physical Interactions. We used a fluorescence assay to demonstrate that down-regulation of SRC-3 by SI-2 proceeds through direct physical interaction with SRC-3. We found that the intrinsic tryptophan fluorescence emission spectra of the SRC-3 RID (λ ex = 278 nm) were quenched with increasing concentrations of SI-2, indicating that SI-2 binds directly to the RID of SRC-3 (Fig. 1H). In contrast, there were no changes of the intrinsic fluorescence observed for a KPC-2 β-lactamase, a negative control, upon addition of SI-2 (Fig. S3), which negates the possibility that the fluorescence changes of RID are a result of nonspecific interactions with SI-2.
SI-2 Selectively Kills Cancer Cells but Spares Normal Cells. We then evaluated the cytotoxic effect of SI-2 on cancer cells. First, MTT assays were performed on MDA-MB-468 cells treated with SI-2 at different concentrations for 72 h. SI-2 can block MDA-MB-468 cell growth with an IC 50 value of 3.4 nM ( Fig. 2A), in line with the dose of SI-2 required to reduce SRC-3 protein levels in the cell (Fig. 2A). In addition, similar low nanomolar IC 50 values also were observed for SI-2 in many other cell lines, including endocrine-sensitive, endocrine-resistant, and TNBC cells (Fig. 2B).
We performed apoptosis assays to examine the cause of the cytotoxic effect of SI-2 on cancer cells. Annexin V and propidium iodide were used to stain MDA-MB-468 cells treated with SI-2 at different concentrations for 24 h. Flow cytometry revealed that the percentages of cancer cells in the early and late apoptosis phases increase in a SI-2 dose-dependent manner (Fig. 2C), which is similar to etoposide, a positive control, caused apoptosis (Fig. S4). In addition, we confirmed the apoptotic effect of SI-2 on cancer cells by measuring the poly(ADP ribose) polymerase (PARP) cleavage using Western blotting (Fig. 2D).
It is important to note that inhibition of SRC-3 is selectively toxic to cancer cells while sparing normal cells (16,17), and prior data showed that knockout of the SRC-3 gene does not influence adult mice life span (37). Consistent with our previously reported SRC-3 SMIs, we did not observe any toxicity of SI-2 up to 500 nM (the highest concentration used) in primary hepatocytes ( Fig. 2A), demonstrating the potential of SI-2 as a targeted therapy.
SI-2 Inhibits Migration of Breast Cancer Cells. Down-regulation of SRC-3 can decrease cell motility, invasion, and tumor metastasis (38). We performed a cell migration assay of MDA-MB-468 cells in the presence and absence of SI-2 using a Cellomics cell motility kit. The bright areas of 50 images for each sample were analyzed using ImageJ. We found that SI-2 treatment can significantly reduce the motility of cancer cells (Fig. 2 E and F). It should be noted that under the same conditions, SI-2 caused minimal toxicity in MDA-MB-468 cells (Fig. 2F), indicating that the decrease of cell motility is not because of decrease of cell numbers.
SI-2 Is Nontoxic to Heart and Other Major Organs. Minimizing cardiotoxicity is an important consideration in drug development. hERG (the human ether-à-go-go-related gene) is a gene that codes for a cardiac potassium ion channel (39). Blockage of hERG can cause cardiac arrhythmia and even sudden death. Therefore, hERG screening is commonly used in drug discovery (40). We compared the hERG blocking ability of bufalin and SI-2 using a well-established in vitro method (41). HEK293 cells that stably express hERG channels (HEK293-hERG) were used as the in vitro system. HEK293-hERG cells have a more negative membrane potential than do wild-type cells, as a result of the potassium channel activity, which can be monitored using a membrane potential-sensitive fluorescent dye DiBAC 4 (3). Blocking of hERG can increase the membrane potential and thus the fluorescence of DiBAC 4 (3). HEK293-hERG cells were treated with DiBAC 4 (3) and bufalin or SI-2 at different concentrations (0.1, 1.0, and 5.0 μM). The fluorescence of DiBAC 4 (3) was monitored using a plate reader. To offset the nonspecific interactions between the drug and DiBAC 4 (3), a similar experiment was performed in parental HEK293 cells. After correcting for the background, we found a dramatic fluorescence increase if cells were treated with bufalin, indicating strong cardiotoxicity (Fig. 3A). In contrast, SI-2 did not cause appreciable changes in DiBAC 4 (3) fluorescence, suggesting that SI-2 cannot block hERG even at 5 μM (Fig. 3A).
SI-2 does not cause observable acute or chronic toxicity in vivo. Intraperitoneal or oral administration of SI-2 up to 20 mg/kg (the highest dose tested) did not cause appreciable stress in mice. In addition, continuous treatment of mice with SI-2 (2 mg/kg, twice per day) for 5 wk did not cause any loss of body weight (Fig. 4C) or observable damages to the major organs, including heart, liver, spleen, kidney, lung, and stomach based on histological analyses (Fig. 3B). Based on both the in vitro and in vivo studies of the toxicity profiles, SI-2 appears to be a relatively safe and promising candidate for anticancer drug development.

SI-2 Has an Acceptable Pharmacokinetic Profile and Is Orally
Available. We measured the pharmacokinetic (PK) profile of SI-2 (20 mg/kg) in CD1 mice (n = 3) via intraperitoneal administration. Twenty microliters of blood were collected at each time point by tail-nicking. The plasma concentration of SI-2 was quantified using liquid chromatography mass spectrometry (LC-MS). The PK data were fitted into an extravascular noncompartment model (Fig. 4A), affording half-life t 1/2 of 1 h, the maximum plasma concentration C max of 3.0 μM, and the time to reach the maximum plasma concentration t max of 0.25 h.
As a drug candidate, it is important for SI-2 to achieve a reasonable oral availability. SI-2 contains a hydrazone group (Fig. 1A), which is potentially sensitive to acid-catalyzed hydrolysis. We tested the stability of SI-2 in different simulated gastric fluids (SGFs) with different pH values to mimic fasted state (pH 1.6) SGF, and early (pH 6.4), middle (pH 5), and late (pH 3) fed-state SGFs (42). A phosphate buffer with the physiological pH 7.4 was also used as a control. We found that SI-2 only degrades slightly (less than 5%) at pH 1.6 and 3.0 within 6 h, and is stable in buffers with pH ≥ 5 (Fig. S5). Next, we measured the PK profile of SI-2 (20 mg/kg) via oral gavage in CD1 mice (n = 3). The PK profile is biphasic (Fig. 4B). We could still detect SI-2 at a 24-h time point, which approaches the lower limit of detection 10 nM. The area under the curve, an indicator of drug exposure, for oral administration is ∼30% of that for intraperitoneal administration. Based on these measurements, SI-2 showed acceptable oral availability and holds promise for drug development.

SI-2 Significantly Inhibits Breast Tumor Growth and Reduces SRC-3
Levels in Vivo. Next, we assessed the antitumor activity of SI-2 in an orthotopic MDA-MB-468 breast cancer mouse model. We determined the dosing schedule based on the PK profile of SI-2.
In the PK study (20 mg/kg, i.p.) the plasma concentration of SI-2 is 23 nM at 8 h. Therefore, if a therapeutic dose 2 mg/kg were used, the plasma concentration of SI-2 should be ∼2.3 nM at 8 h, which is close to the IC 50 value of SI-2 in MDA-MB-468 cells (3.4 nM). To maintain the plasma concentration of SI-2 above the IC 50 value in cancer cells, we decided to treat the mice twice per day. To establish orthotopic breast tumors, 2 × 10 6 MDA-MB-468 cells were injected into one of the second mammary fat pads of SCID mice (female, 5-6 wk). When tumors became palpable, 14 d after injection, mice were randomized into two groups (n = 6 for the control group; n = 8 for the treatment group). The treatment group was injected with SI-2 (intraperitoneally, 2 mg/kg per dose, two doses per day), whereas the control group received PBS. Tumor lengths and widths and mouse weights were measured once per week. Tumor sizes were calculated by (length × width 2 )/2. As shown in Fig. 4 D and E, SI-2 treatment can significantly inhibit tumor growth. After 5 wk of treatment, mice in each group were killed and tumors were harvested. To test whether SI-2 can cause SRC-3 down-regulation in vivo, the SRC-3 levels in tumor tissues were measured by immunohistochemistry. As shown in Fig. 4F, the SRC-3 levels (brown color) in SI-2-treated tumor tissues were significantly lower than the PBS treated control group. K i -67, a cell proliferation marker, also was immunochemically analyzed. Accordingly, the SI-2-treated tumor tissues have dramatically fewer K i -67 + cells (the cells with brown staining, Fig. 4F). Based on these data, we conclude that SI-2 can significantly inhibit breast tumor growth through inhibition of SRC-3.

Discussion
SRC-3 is a critical coactivator that regulates many transcriptional signaling pathways for cancer formation and proliferation. Although selective estrogen receptor modulators, such as tamoxifen, are first-line treatments for ER + breast cancer (43,44), high expression of SRC-3 is associated with recurrence and poor overall survival in both ER + and TNBC (29). Inhibition of SRC-3 can circumvent endocrine therapy resistance in ER + breast cancer. Additionally, downregulating SRC-3 can reduce proliferation and migration in TNBC cells (29). Unlike gossypol and bufalin, SI-2 is a drug-like molecule and meets all of the criteria of Lipinski's rule (31), Veber's rule (32), and Oprea's rule (33) for drug-likeness. As a SRC-3 SMI, SI-2 can selectively reduce the transcriptional activities and the protein concentrations of SRC-3 in cells through direct physical interactions with SRC-3, and can selectively induce breast cancer cell death with IC 50 values in the low nanomolar range (3-20 nM) but not affect normal cell viability. It should be noted although SI-2 treatment in cancer cells can inhibit the transcriptional activities and protein levels of all three members of the p160 family of steroid receptor coactivators, this inhibitory effect against the SRC family should be advantageous because these three SRC proteins share many redundant functions (45). Furthermore, our in vivo study demonstrated that SI-2 can significantly inhibit primary tumor growth and reduce SRC-3 protein levels in a breast cancer mouse model. In addition, in a toxicology study, we showed that SI-2 caused minimal acute cardiotoxicity based a hERG channel blocking assay and unappreciable chronic toxicity to major organs based on histological analyses.
Our study should not only pave the way to development of "first-in-class" drugs that target oncogenic coactivators and make a potential impact on breast cancer treatment, but also provide Fig. 4. Therapeutic efficacy of SI-2 in an orthotopic breast cancer mouse model. (A) PK of SI-2 (20 mg/kg) in CD1 mice (n = 3) via intraperitoneal administration. SI-2 was quantified using LC-MS. The data were fitted into an extravascular noncompartment model. C max = 3.0 μM; t max = 0.25 h; and t 1/2 = 1.0 h. (B) PK of SI-2 (20 mg/kg) in CD1 mice (n = 3) via oral gavage. Lower limit of detection (LLOD) is 10 nM. MDA-MB-468 cells were inoculated into the mammary fat pads of SCID mice (female, 5-6 wk). Treatment started 14 d after tumor inoculation. The treatment group was treated with SI-2 (2 mg/kg) twice per day for 5 wk (n = 8). The control group was treated with PBS (n = 6). (C) Body weights and (D) tumor volumes were measured once per week. SI-2 can significantly inhibit tumor growth while not affecting mouse body weights. Data represent mean ± SEM; ***P < 0.001, ****P < 0.0001 by Student t test. (E) Representative images of harvested tumors after 5 wk of treatment. (F) SI-2 down-regulates SRC-3 protein levels and reduces the number of K i -67 + cells. Multiple tumor tissues (n > 3) collected from both the control group (treated with PBS) and the SI-2 treated group were processed and immunohistochemically stained with an anti-SRC-3 antibody, or an anti-K i -67 antibody. (Magnification: 400×.)