Heat shock protein 90 (Hsp90), a molecular chaperone that plays a significant role in the stability and maturation of client proteins, including oncogenic targets for cell transformation, proliferation, and survival, is an attractive target for cancer therapy. We identified the novel Hsp90 inhibitor, CH5164840, and investigated its induction of oncogenic client protein degradation, antiproliferative activity, and apoptosis against an NCI-N87 gastric cancer cell line and a BT-474 breast cancer cell line. Interestingly, CH5164840 demonstrated tumor selectivity both in vitro and in vivo, binding to tumor Hsp90 (which forms active multiple chaperone complexes) in vitro, and being distributed effectively to tumors in a mouse model, which, taken together, supports the decreased levels of phosphorylated Akt by CH5164840 that we observed in tumor tissues, but not in normal tissues. As well as being well tolerated, the oral administration of CH5164840 exhibited potent antitumor efficacy with regression in NCI-N87 and BT-474 tumor xenograft models. In addition, CH5164840 significantly enhanced antitumor efficacy against gastric and breast cancer models when combined with the human epidermal growth factor receptor 2 (HER2)-targeted agents, trastuzumab and lapatinib. These data demonstrate the potent antitumor efficacy of CH5164840 when administered alone, and its significant combination efficacy when combined with trastuzumab or lapatinib, supporting the clinical development of CH5164840 as an Hsp90 inhibitor for combination therapy with HER2-targeted agents against HER2-overexpressing tumors. (Cancer Sci 2012; 103: 342–349)
Heat shock protein 90 (Hsp90) is a molecular chaperone that plays an important role in the maturation and stability of client proteins.(1) Oncogenic client proteins, which are the targets of many anticancer agents, and encompass dysregulated, mutated, and fusion proteins, are particularly dependent on Hsp90 for maintaining their conformation.(2–6) The inhibition of Hsp90 function is known to lead to the degradation of multiple client proteins involved in tumor progression and the simultaneous inhibition of multiple oncogenic pathways, resulting in signal transduction loss, growth inhibition, cell death, and anti-angiogenesis. Moreover, Hsp90 is present in an activated multi-chaperone complex in tumor cells,(7) and elevated Hsp90 expression is associated with poor prognosis in breast cancer patients,(8) so targeting Hsp90 is considered a promising strategy for anticancer therapy.(1,9–12)
The overexpression of human epidermal growth factor receptor 2 (HER2), a receptor tyrosine kinase that functions as an oncogene, is also associated with poor prognosis in breast(13) and gastric cancers.(14) Indeed, in HER2-overexpressing breast cancers, HER2 and HER3 frequently form a heterodimer as an oncogenic unit that plays an important role in HER2-mediated signaling.(15–17) Although both trastuzumab, a humanized mAb directed against HER2, and lapatinib, an epidermal growth factor receptor (EGFR) and HER2 dual-kinase inhibitor, have demonstrated good initial clinical response and are considered standard-of-care agents, clinical relapse with trastuzumab or lapatinib therapy has been observed in some patients.(18) Therefore, more effective treatment of HER2-overexpressing tumors is required, and forming rationalized combinations of agents appears particularly promising.(19)
As HER2 is a representative client protein that depends on the function of Hsp90 for stability,(20–22) the Hsp90 inhibitor is effective against HER2-overexpressing tumors.(23,24) However, first-generation geldanamycin derivatives have been found to pose a risk of hepatotoxicity and efflux by P-glycoprotein, and have not yet been approved for cancer therapy. To overcome these drawbacks, several second-generation Hsp90 inhibitors have been synthesized and are currently in clinical development.(25–27)
Recently, we identified CH5164840 as a novel Hsp90 inhibitor with a unique chemical structure through virtual screening based on a 3-D structure(28) (Atsushi Suda, personal communication, 2011). To examine the usefulness of the significant antitumor efficacy of CH5164840, in this study, we conducted in vitro and in vivo investigations of the efficacy of combining CH5164840 with HER2-targeted agents.
Materials and Methods
Compound. CH5164840 (Atsushi Suda, personal communication, 2011) and its biotin-labeled version and 17-dimethylaminoethylamino-17-demethoxygeldanamycin (17-DMAG) were synthesized by Chugai Pharmaceutical (Kanagawa, Japan). Lapatinib was prepared from the commercial product Tykerb (GlaxoSmithKline, Clifton, NJ, USA). Trastuzumab was obtained from Chugai Pharmaceutical/F. Hoffmann-La Roche (Basel, Switzerland).
Cells and culture. The human gastric cell line NCI-N87; breast cancer cell lines BT-474, SK-BR-3, and MDA-MB-231; ovarian cancer cell line SK-OV-3; prostate cancer cell line DU 145; colorectal cancer cell line HCT116; and NSCLC cell line NCI-H460 were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). The human gastric cell line JR-St was obtained from Immuno-Biological Laboratories (Fujioka, Japan). Adult normal human dermal (NHDF-Ad) fibroblast cells were obtained from Takara Bio (Shiga, Japan). Human mammary tumor MAXF401 was obtained from Professor Heinz-Herbert Fiebig (University of Freiburg, Freiburg, Germany). All cell lines were cultured according to the supplier’s instructions.
Cell proliferation assay. Tumor cells were seeded into microtiter plates containing compounds, and incubated at 37°C in 5% CO2. After 4-day incubation, Cell Counting Kit-8 solution (Dojindo Laboratories, Kumamoto, Japan) was added, and absorbance at 450 nm was measured with the Microplate-Reader iMark (Bio-Rad Laboratories, Hercules, CA, USA). Antiproliferative activity was calculated using the formula: (1−T/C) × 100 (%), where T represents the absorbance of drug-treated cells, and C represents that of untreated control cells at 450 nm. The IC50 values were calculated using Microsoft Excel 2007 (Redmond, WA, USA).
Surface plasmon resonance analysis. All biosensor experiments were conducted using a Biacore 2000 instrument (GE Healthcare Japan, Tokyo, Japan). The N-terminal domains of the human Hsp90α (9–236) and Hsp90β (1–221) expressed in Escherichia coli were minimally biotinylated (sulfo-NHS-LC-LC biotin; Thermo Fisher Scientific, Waltham, MA, USA) and coupled on a streptavidin-coated sensor chip (GE Healthcare, Buckinghamshire, UK), respectively. Sensorgrams were processed using SCRUBBER2 (BioLogics, Campbell, Australia) and Kd values were determined by fitting the processed data globally to the 1:1 binding model using BIAevaluation (version 3.1; GE Healthcare). The experiments for the Kd determinations were performed in 50 mM tris(hydroxymethyl)aminomethane-HCl (pH 7.6), 150 mM NaCl, 0.005% P-20, and 1% DMSO at 30 μL/min, 20°C.
Protein kinase assay. The protein kinase inhibition assay was performed as previously described.(29) The protein kinases included EGFR, KIT, MET, fibroblast growth factor receptor 2 (FGFR2), FMS-related tyrosine kinase 3 (FLT3), LTK, insulin receptor (InsR), Yes, Abl, insulin-like growth factor receptor I, vascular endothelial growth factor receptor 2 (KDR), platelet derived growth factor receptor (PDGFR)β, Aurora A, Akt1/PKBα, PKA, Cdk1/cyclin B, Cdk2/cyclin A, MEK, MAPK 2/Erk2, PKCα, PKCβ1, and PKCβ2.
Western blotting and co-immunoprecipitation. Western blotting was performed as previously described.(29) Primary antibodies were used: (pY1221-HER2, pY1068-EGFR, pY1289-HER3, pS473-Akt, Akt, pT202/Y204-ERK, ERK (Cell Signaling, Beverly, MA, USA), EGFR, HER2, HER3, and actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and Hsp70 (Stressgen, Victoria, Canada). Signals were detected with ECL Plus (GE Healthcare), followed by LAS-4000 (Fujifilm, Tokyo, Japan) or the Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA). For co-immunoprecipitation, cell lysates were incubated with or without the Hsp90 antibody (Santa Cruz Biotechnology) overnight at 4°C, and protein A-agarose beads (Roche Diagnostics, Penzberg, Germany) were added for 1 h at 4°C. Beads were then washed three times with cell lysis buffer, boiled with sample buffer solution, and analyzed by Western blotting.
Xenograft model and efficacy study. All animal studies were approved by the Chugai Institutional Animal Care and Use Committee. Cancer cells (0.5–1 × 107) were implanted subcutaneously into the right flank of athymic nude (BALB/c nu/nu) mice (CAnN.Cg-Foxn1<nu>/CrlCrlj nu/nu; Charles River Laboratories, Kanagawa, Japan). Tumor volume (TV) was calculated using the formula: TV = ab2/2, where a and b represent tumor length and width, respectively. Once the tumors had reached a volume of approximately 200–300 mm3, animals were randomized into each group (n = 4 or 5), and treatment was initiated. CH5164840 and lapatinib were orally administered; trastuzumab was administered by intraperitoneal injection. Tumor growth inhibition (TGI) was calculated using the formula: TGI = (1−[Tt−T0]/[Ct−C0]) × 100 (%), where T (TV of treated group; T0 on day 0 or Tt on day t) and C (TV of control group; C0 on day 0 or Ct on day t) represent mean tumor volume. The maximum tolerated dose (MTD) was defined as the dose that resulted in neither lethality nor more than 20% body weight loss. The ED50 was calculated from the values of TGI on the final experimental day using XLfit version 220.127.116.11 (Microsoft, Redmond, WA, USA).
Pharmacokinetic and pharmacodynamic study. Tumor xenograft (NCI N87)-bearing nude mice were prepared by the same method used for the efficacy studies. Two mice were killed after oral administration of 50 mg/kg CH5164840 for collection of the indicated organs and grafted tumors. CH5164840 was extracted with acetonitrile/water/formic acid (75/25/0.1, v/v/v) under ice-cold conditions. The CH5164840 concentration was determined using liquid chromatography/tandem mass spectrometry (LC/MS/MS) equipment (API 3000, Applied Biosystems, Foster City, CA, USA).
Real-time quantitative RT-PCR. The QuantiTect Probe RT-PCR kit (Qiagen, Valencia, CA, USA) and Taqman primers (Applied Biosystems) were used to perform all reactions. Total RNA extraction was performed with the RNeasy mini kit (Qiagen). For the data analysis, counts were normalized to a housekeeping gene (GAPDH) at the same time point and condition. Counts are reported as fold change relative to the untreated control.
Statistical analysis. SAS preclinical package software (version 5; SAS, Cary, NC, USA) was used for the statistical analyses (Tukey’s test). Statistically-significant differences are indicated with asterisks; **P < 0.01 and ***P < 0.001.
CH5164840 is a novel Hsp90 inhibitor, which induces the degradation of multiple Hsp90 client proteins and apoptosis in NCI-N87 and BT-474 HER2-overexpressing cells. Recently, we identified CH5164840 as an Hsp90 inhibitor with a novel chemical structure through virtual screening based on a 3-D structure.(28) As shown in Figure 1, the binding affinity of CH5164840 for Hsp90α was found to be 0.52 and 1.4 nM for Hsp90β, a level that is associated with a slow dissociation rate. Specificity to the ATP-binding pocket of Hsp90 was indicated by no inhibition against 22 kinases (IC50 > 20 μM) in a cell-free kinase assay and the absence of significant ATP competitive binding (DiscoveRx, Fremont, CA, USA) against 400 kinases at 10 μM (data not shown).
As the first step, we tested the sensitivity of various cancer cell lines to CH5164840 with in vitro cell growth assays, and the IC50 values of CH5164840 were found to be 66–260 nM in nine cancer cell lines (Table 1). In the second step, we examined the degradation of client proteins in cancer cells induced by CH5164840 Hsp90 inhibition. CH5164840 was found to reduce phosphorylation and protein levels of HER2, HER3, and EGFR significantly, and also inhibited downstream signals in NCI-N87 (HER2-overexpressing gastric cancer) cells and BT-474 (HER2-overexpressing breast cancer) cells. We also confirmed the induction of Hsp70 (Fig. 2a,c). In a time course study, CH5164840 and 17-DMAG caused client protein degradation and Hsp70 induction (Fig. S1). Furthermore, activated caspase-3/7 levels (Fig. 2b,d) and cleaved poly(ADP-ribose) polymerase (Fig. 2a,c) increased following exposure to CH5164840, indicating that CH5164840 induces apoptosis.
Table 1. In vitro antitumor spectrum of CH5164840
Cancer cell lines
NCI-N87 (gastric cancer)
JR-St (gastric cancer)
BT-474 (breast cancer)
SK-BR-3 (breast cancer)
MDA-MB231 (breast cancer)
SK-OV-3 (ovarian cancer)
DU145 (prostate cancer)
HCT116 (colorectal cancer)
These results indicate that CH5164840 selectively inhibited Hsp90, induced key client protein degradation, and suppressed both the PI3K/Akt survival and the Raf/MEK/ERK signaling pathways, leading to apoptosis induction and inhibition of proliferation in tumor.
CH5164840 exhibits tumor-selective Hsp90 inhibition. Based on the observation of tumor Hsp90 in multi-chaperone complexes with high ATPase activity and its high binding affinity to Hsp90 inhibitor 17-allylamino-17-demethoxy-geldanamycin (17-AAG),(7) we assessed the tumor selectivity of CH5164840 using the following procedure. First, we compared the antiproliferative activity of CH5164840 in normal fibroblast (NHDF-Ad) cells to that in NCI-N87 tumor cells and found that proliferation of NHDF-Ad cells was moderately inhibited by CH5164840 treatment with an IC50 of 1.2 μM, and the cells were less sensitive to our compound than NCI-N87 and other cancer cells (Table 1; Fig. 3a). Next, to examine whether CH5164840 binds selectively to tumor Hsp90, pull-down assays were conducted with a biotin-labeled CH5164840 (Bio-CH) probe using cell extracts prepared from NCI-N87 or NHDF-Ad cell lines. As shown in Figure 3(b), Bio-CH preferentially bound to Hsp90 when it formed a multi-chaperone complex with the co-chaperones Hsp70, Hop, and Hsp40 in NCI-N87 cells, rather than to Hsp90 in NHDF-Ad cells, which remained mainly free from these co-chaperones. These data indicate that CH5164840 selectively binds to the tumor Hsp90, which forms a multi-chaperone complex.
We then explored the tumor-selective distribution of CH5164840 after a single administration in mice. After oral administration of one dose of 50 mg/kg CH5164840 in the NCI-N87 xenograft model, greater CH5164840 retention was observed in tumor tissues (half-life = 6.3 h) compared to that in normal tissues and plasma (Fig. 3c). Finally, the phosphorylation level of Akt in tumor and normal tissues was measured to determine whether the presence of CH5164840 was associated with tumor-selective Hsp90 inhibition in vivo in NCI-N87 models. The results indicate that CH5164840 causes a reduction in levels of Akt phosphorylation in tumor tissues, but not in normal tissues, by the degradation of upstream signaling molecules and of p-Akt itself (Fig. 3d). In addition, we confirmed the induction of Hsp70 in tumor and normal tissues. Taken together, these results indicate that preferential binding to tumor Hsp90 and increased retention in tumors lead to tumor-selective Hsp90 inhibition by CH5164840.
CH5164840 demonstrates potent antitumor efficacy in xenograft models. We then examined the in vivo antitumor efficacy of CH5164840 against human tumor xenograft models in mice. The daily oral administration of CH5164840 resulted in a significant level of dose-dependent antitumor efficacy in the NCI-N87 and BT-474 xenograft models, with a maximum TGI of 160% and 152%, respectively (Fig. 4a,b), without any significant loss of body weight. In addition, the protein levels of HER2, HER3, EGFR, and phospho-AKT significantly decreased in the NCI-N87 xenograft model, resulting in CH5164840 inhibition of downstream signals (Fig. 4c). Furthermore, CH5164840 demonstrated significant antitumor efficacy against various xenograft models without any significant loss of body weight (Table S1). These results are consistent with those regarding in vitro antitumor activity and tumor-selective Hsp90 inhibition by CH5164840.
CH5164840 enhances the antitumor efficacy of HER2-targeted therapy with trastuzumab or lapatinib. We examined the combined effect of CH5164840 with a HER2-targeted therapy. When a combination of 25 mg/kg CH5164840 and 30 mg/kg trastuzumab was administered to the trastuzumab-sensitive NCI-N87 and BT-474 xenograft models, CH5164840 was observed to significantly enhance the antitumor efficacy of trastuzumab, increasing its TGI as a single agent from 56% to 167% in the NCI-N87 model (Fig. 5a), and from 136% to 240% in the BT-474 xenograft model (Fig. 5b). With lapatinib, when a combination of 100 mg/kg lapatinib and 12.5 mg/kg CH5164840 was administered to the NCI-N87 and BT-474 xenograft models, CH5164840 significantly enhanced the antitumor efficacy of lapatinib in both the NCI-N87 model (TGI = 45% lapatinib only vs 155% in combination; Fig. 5c) and the BT-474 xenograft model (TGI = 88% lapatinib only vs 214% in combination; Fig. 5d).
In all cases, the enhanced efficacy with combination treatment was statistically significant, and no gross toxicity was observed in any of the treated animals (data not shown). These results suggest that CH5164840 potentiates the efficacy of HER2-targeted agents.
CH5164840 suppresses HER3 signals activated after lapatinib treatment in vitro. Intrigued by the enhancement of lapatinib antitumor efficacy by CH5164840, we then investigated the mechanism underlying the combined efficacy of CH5164840 and lapatinib. First, we examined HER3 expression after lapatinib treatment in NCI-N87 cells, as lapatinib treatment has been reported to induce HER3 expression in SK-BR-3 cells.(30) Quantitative RT-PCR revealed that lapatinib induced HER3 mRNA expression (Fig. 6a). The HER3 protein level consistently increased in a time-dependent manner (Fig. 6b). Although HER3 phosphorylation was transiently inhibited by lapatinib, its inhibition was not long lasting, despite continuous lapatinib treatment (Fig. 6b). In addition, the withdrawal of lapatinib after 24 h of treatment, followed by lapatinib-free incubation, resulted in hyper-phosphorylation of HER3 (Fig. 6c). Next, to determine whether Hsp90 inhibition induced the degradation of lapatinib-induced HER3 and other clients, we investigated the effect of CH5164840 in NCI-N87 cells. As the results in Figure 6(d) show, HER2 and EGFR were degraded, and their phosphorylation levels and downstream signals were significantly decreased by combined CH5164840 and lapatinib treatment. Interestingly, lapatinib-induced HER3 returned to basal levels after the addition of CH5164840. Finally, we observed that HER3 was strongly co-precipitated with Hsp90 in lapatinib-treated cells compared to that in non-treated control cells, and importantly, was returned to basal levels by the addition of CH5164840 (Fig. 6e).
Hsp90 is considered an attractive target in anticancer therapy for several reasons; first, its role in the regulation of its client proteins, many of which are cancer-related proteins, such as kinases, transcription factors, and steroid receptors. Hsp90 inhibition induces degradation of these clients, leading to the inhibition of multiple signaling pathways that regulate tumor cell proliferation and survival. Such inhibition was observed in this study by the novel Hsp90 inhibitor CH5164840, which induced multiple client degradation, followed by the suppression of multiple signaling pathways.
Another attractive feature of Hsp90 as an antitumor target is the possibility of tumor selectivity. As shown in Figure 3, we also observed a higher binding affinity of bio-CH to tumor Hsp90 compared to normal Hsp90. (In this pull-down assay, the biotin label was positioned, based on the co-crystal structure of Hsp90α with CH5164840, so as to have no influence on CH5164840–Hsp90 binding.) This selective binding to Hsp90 in tumor tissues and the fact that CH5164840 also showed long retention in tumor tissues in a mouse model explain why the levels of phosphorylated Akt in our pharmacodynamic study decreased in tumor tissues, but not in normal tissues, indicating that Hsp90 inhibition by CH5164840 is tumor selective.
Furthermore, CH5164840 enhanced the antitumor efficacy of HER2-targeted therapy by trastuzumab or lapatinib in the NCI-N87 and BT-474 tumor models. In combination with trastuzumab in the NCI-N87 xenograft model, antitumor efficacy was sustained over the follow-up period without any additional administration. Furthermore, when the efficacy of intermittent dosing was examined in the form of once- or twice-weekly regimens of 50 mg/kg CH5164840 in the NCI-N87 xenograft model, CH5164840 enhanced the antitumor efficacy of trastuzumab (TGI = 69% with trastuzumab only vs 97% or 123% with combined trastuzumab and once- or twice-weekly CH5164840 administration, respectively; data not shown). The efficacy of CH5164840 in combination with trastuzumab matches the findings of a preclinical study,(23) and a phase I study in which the combination of 17-AAG and trastuzumab was well tolerated and demonstrated clinical response in patients with HER2-overexpressing breast cancer whose tumors had progressed during treatment with trastuzumab.(24) In addition, as shown in Figure 5(c,d), when a reduced-dosage regimen was evaluated by combining 12.5 mg/kg CH5164840 (one-quarter of the MTD dose for single-agent administration) with lapatinib, the reduced dosage of CH5164840 still enhanced the antitumor efficacy of lapatinib. These results suggest that CH5164840 administration at reduced frequencies and doses, which might aid in the avoidance of possible side-effects, still provides a high level of antitumor efficacy in combination with HER2-targeted agents.
As shown in Figure 6, we began examining the molecular mechanisms behind CH5164840’s enhanced efficacy of agents by looking at HER3 induction by lapatinib in the NCI-N87 cells, in the same manner as that previously reported in SK-BR-3 cells.(30) As the authors of a previous study regarding gefitinib-induced p-HER3 had done,(31) we demonstrated that cessation of lapatinib treatment causes hyper-phosphorylation of HER3. These data suggest that lapatinib-induced HER3 confers resistance to lapatinib itself in NCI-N87 cells in a preclinical setting, and possibly in a clinical setting as well. Interestingly, this lapatinib-induced HER3 is strongly degraded by CH5184840, suggesting that lapatinib-induced HER3 is a preferred client of Hsp90, and is therefore sensitive to Hsp90 inhibition. This hypothesis is strongly supported by the preferred binding of induced HER3 to Hsp90 and its suppression by CH5164840. Overall, the results suggest that the Hsp90 inhibitor CH5164840 could enhance the antitumor efficacy of lapatinib by effectively targeting lapatinib-induced HER3. This possibly unique feature of Hsp90 inhibition is strongly supported by a recent report from Chandarlapaty et al.,(32) who showed that an Hsp90 inhibitor prevented the production of Akt inhibitor-induced receptor tyrosine kinases, resulting in a high level of efficacy.
In conclusion, the characteristics of CH5164840, particularly its selective binding to tumor Hsp90 and relatively long length of retention in tumor tissues, allow it to exert potent antitumor activity, even when administered as a single agent. In this examination of the antitumor efficacy of CH5164840 both in vitro and in vivo, the combination of CH5164840 with HER2-targeted agents, especially lapatinib, was found to enhance the antitumor efficacy of the agents by degradation of lapatinib-induced HER3. These observations indicate that the Hsp90 inhibitor CH5164840 might be helpful in combination with HER2-targeted therapies for the treatment of tumors whose growth and survival depend on Hsp90.
We thank Dr Neal Rosen (Memorial Sloan-Kettering Cancer Center, NY, USA) and Dr Shige Nagahashi for their helpful discussion. We also thank Ms Fumie Sawamura, Mr Sachiya Yamamoto, Mr Yusuke Ide, Ms Yasue Nagata, Ms Ikuko Matsuo, Ms Yumiko Hashimoto, Ms Maiko Izawa, and Mr Kim ByoungJin for their technical support.
All authors are employees of Chugai Pharmaceutical Co. Ltd.