|Home | About | Journals | Submit | Contact Us | Français|
Heat shock protein 90 (HSP90) is a chaperone protein that stabilizes proteins involved in oncogenic and therapeutic resistance pathways of epithelial cancers, including head and neck squamous cell carcinomas (HNSCCs). Here, we characterized the molecular, cellular, and preclinical activity of HSP90 inhibitor SNX5422/2112 in HNSCC overexpressing HSP90. SNX2112 inhibited proliferation, induced G2/M block, and enhanced cytotoxicity, chemosensitivity, and radiosensitivity between 25 and 250 nM in vitro. SNX2112 showed combinatorial activity with paclitaxel in wild-type (wt) TP53-deficient and cisplatin in mutant (mt) TP53 HNSCC lines. SNX2112 decreased expression or phosphorylation of epidermal growth factor receptor (EGFR), c-MET, v-akt murine thymoma viral oncogene homolog 1 (AKT), extracellular signal-regulated kinases (ERK) 1 and 2, inhibitor κB kinase, and signal transducer and transcription factor 3 (STAT3), corresponding downstream nuclear factor κB, activator protein-1, and STAT3 reporter genes, and target oncogenes and angiogenic cytokines. Furthermore, SNX2112 enhanced re-expression of TP53 and targets p21WAF1 and PUMA, while TP53 inhibitor Pifithrin or siRNA attenuated the antiproliferative activity of SNX2112 in wtTP53 HNSCC in vitro. Prodrug SNX5422 similarly down-modulated key signal targets, enhanced TP53 expression and apoptosis, and inhibited proliferation, angiogenesis, and tumorigenesis in a wtTP53-deficient HNSCC xenograft model. Thus, HSP90 inhibitor SNX5422/2112 broadly modulates multiple key nodes within the dysregulated signaling network, with corresponding effects upon the malignant phenotype. Our data support investigation of SNX5422/2112 in combination with paclitaxel, cisplatin, and radiotherapy in HNSCC with different TP53 status.
Biologic and small molecule inhibitors selectively targeting growth factor receptors, signal kinases, and angiogenesis factor receptors have undergone extensive preclinical and clinical studies in different cancers. Relative to the success of targeted therapy against prevalent genetic drivers in some hematologic malignancies (e.g., BCR/ABL), the clinical activity of these agents in solid tumors has often been modest or transient. Head and neck squamous cell carcinomas (HNSCCs) have been the subject for clinical or preclinical studies of several targeted agents against a few commonly activated or inactivated pathways, including epidermal growth factor receptor (EGFR; cetuximab, gefitinib, and erlotinib) [1–5], proteasome-dependent nuclear factor κB (NF-κB) activation (bortezomib) [6,7], signal transducer and transcription factor 3 (STAT3; decoy oligonucleotide) , and disrupted tumor suppressor TP53 (TP53 adenovirus) . These agents have demonstrated significant target-specific and preclinical activity but modest clinical activity at achievable concentrations as monotherapy or in combination with conventional therapies [1–9].
Studies of the signal and transcriptional mechanisms mediating pathogenesis and resistance of HNSCC demonstrate frequent alteration and complex cross talk among multiple pathways. This includes co-activation of growth factor and cytokine receptors and their ligands [EGFR, c-MET, tumor necrosis factor-α (TNF-α), interleukin-1 (IL-1), and IL-6], intermediate signal kinases [sarcoma Rous cellular homolog (SRC), MAP or extracellular signal-regulated kinase (ERK) kinase (MEK)-ERK1/2, phosphatidylinositol 3-kinase (PI3K)-v-akt murine thymoma viral oncogene homolog 1 (AKT), and inhibitor κB kinase (IKK)], and downstream transcription factors [activator protein-1 (AP-1), NF-κB, and STAT3] that promote the malignant phenotype [3,5–8,10–12]. Additionally, dysregulation of the key tumor suppressor TP53 may result from transcriptional inactivation or mutation in HNSCC subsets, which exhibit differences in response to therapy [13–15]. Recent next-generation sequencing studies in HNSCC support the prevalence of genetic alterations affecting multiple growth factor receptors, PIK3CA and PTEN that regulate AKT, components of the NF-κB pathway, and tumor suppressor TP53 [16,17]. Thus, it has become increasingly clear that the complex genetic and epigenetic alterations in HNSCC and other solid cancers contribute to intrinsic or acquired resistance to targeted and standard cytotoxic therapies [5,7,10–14]. Therefore, combinations or agents targeting multiple components of the dysregulated network in HNSCC and other solid cancers may have potential for broader activity [18–23].
Heat shock protein 90 (HSP90) is a molecular chaperone required for the proper folding, stabilization, and function of many proteins, including multiple overexpressed, mutated, or activated signal components and transcription factors that serve as key nodes in the network of pathways that promote cancer cell proliferation and survival . The ATP-dependent conformational state of HSP90 provides a selective target for natural toxins (e.g., geldanomycins) and synthetic inhibitors. Targeted inhibition of HSP90 leads to destabilization and proteasomal degradation of a diverse array of its client proteins, conveying the potential to simultaneously modulate many signaling pathways that synergize to promote cancer progression and reduce the development of resistance seen with more selective molecular targeted agents [24–26].
Prior studies provide evidence that HSP90 activation by interferon may contribute to EGF-mediated protection against the apoptotic effects of interferon in HNSCC cells . We recently showed that wild-type (wt) EGFR is stabilized by HSP90 in HNSCC . Enhanced activity of HSP90 inhibitor geldanomycin is observed in HNSCC with increased HSP90 and RAS activity . In addition to the ability of HSP90 inhibitors to concurrently modulate multiple key molecular targets, they can enhance standard cytotoxic modalities such as chemotherapy and radiation therapy in cancers, including HNSCC [25,26,28–31].
SNX5422 (also known as PF-04929113) is a water-soluble and orally bioavailable prodrug of SNX2112 (PF-04928473), a potent and highly selective small molecule inhibitor of HSP90 [32,33]. SNX2112 competitively binds to the N-terminal ATP pocket of HSP90 family members (HSP90α, HSP90β, Grp94, and Trap-1) and is highly potent against various cancers in vitro and in vivo [34–37]. SNX5422 has recently completed phase I testing, which defined tolerated dosages, and demonstrated prolonged disease stabilization of >150 days with various schedules in 22% to 36% of subjects with treatment-refractory cancers [38–40]. However, preclinical studies of the molecular effects on the broadly dysregulated signal and transcriptional network and therapeutic activity of 5422/SNX2112 have not been reported in HNSCC. In the present study, we examined the effects of SNX2112 and prodrug SNX5422 on the broad network of dysregulated pathways and targets and therapeutic effects alone and in combination with radiation and standard chemotherapies in preclinical models of HNSCC.
SNX5422 is a water-soluble and orally bioavailable prodrug of SNX2112, a potent and highly selective small molecule inhibitor of HSP90 [32,33]. Both were provided initially by Pfizer Inc and subsequently by Esanex Inc. Cisplatin was obtained from APP Pharmaceuticals (Lake Zurich, IL; #100351). Paclitaxel and TP53 inhibitor Pifithrin-α were from Sigma-Aldrich Inc (St Louis, MO; #T7191; P4359).
Nine HNSCC cell lines (UMSCC) obtained from Dr T. E. Carey (University of Michigan, Ann Arbor, MI) were recently characterized by genotype and TP53 status [13,41,42]. Normal human epidermal keratinocytes (HEKA) were obtained commercially (Invitrogen, Carlsbad, CA). The characteristics and culture conditions for UMSCC cell lines and HEKA cells were previously described  (see Supplementary Methods).
Quantitative reverse transcription-polymerase chain reaction (RT-PCR) was performed as described  (see Supplementary Methods).
Western blot was performed with antibodies indicated as described  (see Supplementary Methods).
A human tissue array with 20 HNSCC sections and 6 normal mucosa sections  was used for staining for HSP90. Immunohistochemical staining of this tissue array and aHNSCCxenograftmodel for HSP90 client proteins were performed as described [43,44] (see Supplementary Methods).
Cell proliferation was measured by standard 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium (MTT) or 2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanalide (XTT) assay (Roche Diagnostics, Indianapolis, IN ; see Supplementary Methods).
DNA cell cycle distribution analysis was performed by flow cytometry as described  (see Supplementary Methods).
UMSCC-1 cells were plated in 100-mm2 plates and treated the following day with either 0.01% DMSO control or 50, 100, or 200 nM SNX2112. After drug treatment, cells were expanded for 48 hours, collected at 70% to 80% confluence, and stained with Annexin V-fluorescein isothiocyanate and propidium iodide according to the manufacturer's instructions (Sigma APOAF).
TP53 or pooled control siRNA was constructed by Thermo Scientific (Lafayette, CO) and transfected in UMSCC-1 using Lipofectamine RNAiMax according to the manufacturer's instructions (Invitrogen; see Supplementary Methods).
Clonogenic assay was performed as described previously  and in the Supplementary Methods.
NF-κB, AP-1, STAT3, IL-8, and BCL-XL reporters and assays were performed as described [22,23] (see Supplementary Methods).
Cell culture supernatant samples were centrifuged at 14,000 rpm (10 minutes) to clear residual cellular debris. Individual bead kits for IL-6, IL-8, and vascular endothelial growth factor (VEGF) were purchased from Biosource (Invitrogen), and multiplex assays were performed and read by Luminex, as described . Each sample was assayed in triplicate, and data were presented as means ± SD.
Immunodeficient BALB/c severe combined-immunodeficiency (SCID) mice were obtained from the Frederick Cancer Research and Development Center, National Cancer Institute (Frederick, MD) and housed in a specific pathogen-free animal facility. Animal care and HSP90 inhibitor studies were performed under National Institutes of Health (NIH) Animal Care and Use Committee-approved Protocol 1102-07. Tumor xenografts were established by injecting 5 x 106 UMSCC-1 cells subcutaneously into the left flank. Tumor-bearing animals were randomized into experimental groups indicated (n = 10 each). PF-04929113 suspended in 0.5% carbomethylcellulose/0.5% Tween 80 was administered through oral gavage. Tumor size was measured three times weekly, and tumor volumes were calculated as V (cm3) = (L x W2)/2. Tumors were harvested for immunohistochemical analysis for selected HSP90 client proteins, proliferation (Ki-67), apoptosis (terminal deoxynucleotidyl transferase nick end labeling (TUNEL)], and microvessel density (CD31) [5,7], as described in the Supplementary Methods.
Significance of comparisons was determined where indicated by Fisher's Exact or Student's t test where indicated.
Detection of HSP90 has previously been reported in oral HNSCC . We showed that HSP90 immunostaining was detected or increased in a majority (18 of 20) of HNSCC tumors from different anatomic sites in a tissue microarray (Figure W1, A and B). Similarly, increased expression of HSP90 mRNA and protein was detected by RT-PCR and by Western blot in nine of nine HNSCC (UMSCC) cell lines compared to human keratinocytes (Figure W1, C–E). The panel included UMSCC lines derived from patients with aggressive HNSCC (survival < 15 months), and exhibiting underexpression of wtTP53 (UMSCC-1, -6, -9) or overexpression of mutant (mt) TP53 (UMSCC-11A, -11B, -22B, -38, -46) [13,42]. Thus, the pattern of increased expression of HSP90 in tumor tissue specimens and cell lines supports the rationale for investigation of the molecular, cellular, and antitumor activity of HSP90 inhibitors in HNSCC models.
Next, the effect of selective HSP90 ATPase inhibitor SNX2112 [32,33] on proliferation of a subpanel of six of these UMSCC lines was examined over 5 days by MTT assay (Figures 1A and W2). SNX2112 inhibited cell density in a dose-dependent manner with an inhibitory concentration 50% (IC50) range of 22 to 68 nM (Figure 1A), demonstrating activity over a similar range in six distinct UMSCC cell lines, without apparent relationship to HSP90 overexpression (Figure W1) or TP53 genotype. Cell cycle and cytotoxic effects of SNX2112 HSP90 inhibition were further examined by propidium iodide DNA flow cytofluorometric analysis and compared in UMSCC-1 (wtTP53) and UMSCC-38 (mtTP53) cells (Figures 1B, left panels, and W3). In wtTP53 UMSCC-1, SNX2112 at 50 nM inhibited cell cycle, with increased accumulation in the G2/M phase compared to untreated controls, 49% versus 23% at 48 hours (Figure 1B, left panels) and 29% versus 17% at 72 hours (not shown). There was also increased accumulation of fragmented sub-G0 DNA (15%) compared to untreated controls (1%) at 48 hours (Figure 1B, left panels) and 28% versus 2%, respectively, at 72 hours (not shown), indicative of cell death. We performed Annexin V-fluorescein isothiocyanate staining, providing evidence for increased apoptosis with concentration (Figure 1B, right panels). Treatment of mtTP53 UMSCC-38 cells with 50 nM SNX2112 for 48 hours showed a similar accumulation in the G2/M phase (48% vs 18%) but a lesser increase in sub-G0 cells (6% vs 4%; Figure W3), consistent with the higher IC50 observed in Figure 1A. This trend was also seen after 72 hours of drug treatment. The results suggested that despite similarities in G2/M cell cycle block, the UMSCC-1 line deficient for wt TP53 and lower IC50 was more sensitive to apoptosis than the UMSCC-38 line with mtTP53 and highest IC50. To examine the potential contribution of wtTP53 in UMSCC-1 to antiproliferative effects of SNX2112, cells were pretreated with TP53 inhibitor Pifithrin or TP53 siRNA before SNX2112. Pifithrin and TP53 siRNA attenuated the antiproliferative effects and doubled the IC50 of SNX2112, supporting a role for wtTP53 in effects of the HSP90 inhibitor in UMSCC-1 (Figure 1, C and D).
The observed effects of SNX2112 on sub-G0 cell death as well as G2/M phases of the cell cycle sensitive to radiation, led us to further investigate the effects of HSP90 inhibition on clonogenic cell survival without and with radiation in UMSCC-1 in vitro. While MTT assay indicated the IC50 for proliferation of UMSCC-1 was ~30 nM, pilot clonogenic survival assays with SNX2112 alone indicated that inhibitory concentrations for clonogenic cell survival were in a higher range, between 100 nM (90% survival) and 500 nM (59% survival), consistent with the relatively modest effects of the lower concentration of HSP90 inhibitor on cell survival seen by DNA flow cytometry. Pre-exposure of cells to an intermediate concentration of SNX2112 of 250 nM alone for 24 hours had a modest effect on clonogenic tumor cell survival [surviving fraction of 73.5 ± 9.5% (SD)], but this concentration significantly enhanced radiosensitivity, with a dose modification factor (DMF) of 1.55 ± 0.11 (fold ± SD) at 10% survival (Figure 2A). Post-radiation treatment for 24 hours with 250 nM SNX2112 showed no enhancement of radiosensitivity (data not shown).
We next examined the combinatorial effects of HSP90 inhibitor with chemotherapeutic agents paclitaxel and cisplatin, which are active in HNSCC, using two wtTP53 and two mtTP53 UMSCC lines, at subtherapeutic inhibitory concentrations of SNX2112 or chemotherapy drugs alone (Figure 2B). Interestingly, HSP90 inhibitor and paclitaxel showed combined activity in UMSCC-1 and UMSCC-9 deficient for wtTP53, while combined activity with cisplatin was observed in cell lines UMSCC-38 and UMSCC-46 expressing mtTP53. Conversely, sensitization to paclitaxel was not observed in UMSCC-38 and UMSCC-46, or to cisplatin in UMSCC-1 or UMSCC-9 (data not shown), indicating that response to these chemotherapy drugs in combination with HSP90 inhibitor was associated with differences in TP53 status. Together, the above results indicate that HSP90 inhibition has antiproliferative, cytotoxic, radiation, and chemosensitizing activity, associated with TP53 genotype in HNSCC lines in vitro. The SNX2112 concentrations of 22 to 68 nM having antiproliferative and chemosensitizing activity in MTT assay, and 250 nM having radiosensitizing activity in clonogenic assay, are clinically relevant, as they are within the range of concentration maxima achievable (Cmax~77-4098 nM) in patient serum between dosages of 4 and 133 mg/m2, where disease stabilization was observed in a recent NIH phase I clinical trial .
Western blot analysis of UMSCC-1 (deficient wtTP53) and UMSCC-38 (mtTP53) cells treated over a range of inhibitory concentrations of SNX2112 revealed modulation of expression of several important signaling molecules, transcription factors, and target genes previously shown to be altered and promote the malignant phenotype of HNSCC (Figures 3 and W4). Lower SNX2112 concentrations of 25 to 50 nM tested initially inhibited expression of hepatocyte growth factor (HGF) receptor c-MET and phosphorylation of downstream kinases AKT (Ser473 and Thr308) and ERK1/2 in both cell lines (Figures 3A and W4, left panels). AKT phosphorylation exhibited an oscillatory pattern of inhibition and reactivation between 24 and 72 hours in both lines, similar to that observed with PI3K-mTOR inhibitors, as reported with loss of mTOR-mediated negative feedback inhibition of PI3K . Higher concentrations of 100 to 200 nM for 24 to 72 hours were required for decrease of growth factor receptor EGFR, prosurvival pathway components phospho-STAT3 and total STAT3, IKKα and β, phospho-NF-κB/p65, NF-κB- and STAT3-co-regulated prosurvival protein BCL-XL, and cleavage of cell death marker poly (ADP-ribose) polymerase (c-PARP Figures 3A and W4, right panels). Further examination of the SNX2112-sensitive UMSCC-1 line deficient for wtTP53 revealed that the inhibitor also upmodulated expression of TP53 and its target proteins, p21WAF1 and isoform PUMAα, in a time- and dose-dependent manner over 72 hours (Figure 3A, left panels). SNX2112 also enhanced expression of p21WAF1 in UMSCC-38 (Figure W4), previously shown to overexpress a missensemtTP53 and TAp73, which can also potentially activate TP53 target genes .
We further assessed the functional effects of HSP90 inhibitor on reporter activity for downstream proto-oncogenic transcription factors and expression of several of their target genes and proteins implicated in phenotype and resistance of HNSCC. Overall, SNX2112 significantly inhibited NF-κB, AP-1, and STAT3-specific reporter genes (Figure 3B) over a concentration range within which inhibition of known upstream signal molecules was observed above. However, weaker inhibition of STAT3 reporter activity was detected beginning at lower concentrations than required for inhibition of STAT3 phosphorylation and incompletely inhibited at higher concentrations where minimal STAT3 phosphorylation was detected at 48 hours in Figure 3A, suggesting the drug may have more complex effects involving other determinants of STAT activation or STAT family members. IL-8 cytokine and BCL-XL prosurvival gene reporters sharing enhancer elements for two or three of these transcription factors were also inhibited (Figure 3B). Furthermore, SNX2112 inhibited expression of NF-κB- and AP-1-co-regulated IL-6, IL-8, and VEGF proteins implicated in tumor angiogenesis [49–52] in culture supernatants (Figure 3C). Consistent with inhibition of ERK and AP-1, AP-1 target VEGF was decreased by treatment with SNX2112 (50 nM). Significant inhibition of IKK-NF-κB and dependent gene IL-8 required intermediate drug concentrations ≥100 nM. Inhibition of STAT3 inducer IL-6, STAT3, and coregulated target BCL-XL required drug concentrations 200 nM. The cumulative effects of increasing concentrations of SNX2112 on key components of these signal transduction pathways and targets are consistent with the ranges in which significant inhibition of cell density in MTT assay (25–50 nM), angiogenesis factor expression (>100 nM), and decreased radiation clonogenic cell survival (250 nM) were observed.
We next examined the effects of the prodrug SNX5422 on tumor growth, survival, and angiogenesis and a panel of these molecular markers in wtTP53-deficient UMSCC-1 xenografts in vivo. After tumors were established, SNX5422 was administered for 21 days on two different dose schedules, 20 mg/kg daily (Monday to Friday) or 50 mg/kg every other day (Monday, Wednesday, and Friday). The concentration average for free drug for 50 mg/kg every other day used in mice is approximately equivalent to that achieved with 100 mg/m2 in human phase I studies ( J. Kan, N. Brega, and Pfizer, unpublished pharmacokinetic data). Biologic outcomes, including tumor growth, survival, and body weight, were measured and compared to placebo-treated controls (Figure 4). The rate of tumor growth was significantly and similarly inhibited in both 20 mg/kg (Monday to Friday) and 50 mg/kg (Monday, Wednesday, and Friday) treatment groups compared to controls while on treatment for 3 weeks. Once treatment was discontinued, both experimental groups demonstrated tumor growth at a rate similar to that seen in controls (Figure 4A). Suppression of tumor growth corresponded with a survival benefit for treated animals versus controls of ~2 weeks (Figure 4B). Toward the end of 3 weeks of treatment with SNX5422, 50 mg/kg (Monday, Wednesday, and Friday), gastrointestinal and systemic toxicities (diarrhea, piloerection, and lethargy) were observed and associated with death in 2 of 10 animals. The toxicities resolved in the remaining animals once the drug was discontinued. No systemic toxicity was apparent in the lower dose treatment group. A second controlled study of treatment with the lower dose of SNX5422, 20 mg/kg (5 days per week), for up to 6.5 weeks revealed sustained tumor stabilization while on treatment, an overall survival benefit exceeding 2 weeks, and minimal toxicity (Figure W5). In both studies, fluctuation in body weight was seen during weekdays of treatment in both vehicle and drug-treated mice, likely due to hyperosmotic or other effects of the methylcellulose/Tween 80 vehicle used for drug delivery.
Tumor specimens from extra control and mice treated with both concentrations/schedules during the 3-week in vivo study were harvested on days 16, 21, and 25 after tumor inoculation (days 9, 14, and 18 of treatment) for immunohistochemical comparison of selected HSP90 clients characterized above and markers of proliferation (Ki-67), apoptosis (TUNEL), and angiogenesis (CD31). Maximal effects on individual tumor markers were observed in specimens by day 16 or day 21, and representative sections demonstrating treatmentinduced modulation are shown in Figure 5A. HNSCC tumors from SNX5422-treated mice showed decreased immunostaining for progrowth and prosurvival signal molecules c-MET, SRC, p-AKT, and p-ERK1/2 and transcription factor p-STAT3, while that for proapoptotic transcription factor TP53 and target PUMA was increased. Corresponding to the well-established cellular functions of these pathways, SNX5422 inhibited proliferation, as indicated by reduced Ki-67, and induced apoptosis, as evidenced by an increase in tumor apoptotic index, measured by TUNEL staining (Figure 5, A and B). Consistent with the inhibition of tumor angiogenic cytokine production observed in vitro (Figure 4C), tumors treated with SNX5422 also showed an overall reduction in CD31+ vessel density (Figure 5, A and B). Together, these results provide evidence that SNX5422 treatment suppresses tumor growth in vivo by modulating the expression of known clients of HSP90 and their targets mediating proliferation, survival, and angiogenesis in HNSCC.
Here, we show that HSP90 inhibitor SNX5422/2112 inhibits proliferation of a panel of six UMSCC lines (Figure 1) that exhibit increased HSP90 (Figure W1), along with dysregulation of a network of prooncogenic signal components and tumor suppressor TP53 found in HNSCC (Figures 3, ,5,5, and and6).6). HSP90 inhibition increased G2/M cell cycle blockade, sub-G0 DNA, and apoptotic marker Annexin (Figures 1B and W3) and showed combined activity with radiation or cisplatin or paclitaxel chemotherapies active in HNSCC (Figure 2). HSP90 inhibition broadly suppressed the dysregulated network of prooncogenic signal components and transcription factors in vitro and in vivo (Figures 3, W4, 5, and 6) and prolonged progression-free survival in an HNSCC xenograft model in vivo (Figures 4 and W5). Inhibition of prosurvival pathway components EGFR, NF-κB, STAT3, and their antiapoptotic target BCL-XL was associated with PARP cleavage and decreased clonogenic survival in vitro (Figures 2A, ,3A,3A, and W4A). TP53 inhibition by Pifithrin or siRNA partially attenuated the inhibitory effects of SNX2112 observed in UMSCC-1 with wtTP53 (Figure 1, C and D), supporting a role for TP53. Further, SNX2112 induced sub-G0 DNA and TUNEL activity was associated with increased TP53 and target proapoptotic protein PUMA in UMSCC-1 in vitro and in vivo (Figures 3A and and5).5). Interestingly, sensitivity to HSP90 inhibitor in combination with chemotherapy agent cisplatin or paclitaxel (Figure 2B) also appeared to differ with TP53 status. SNX2112 enhanced paclitaxel sensitivity of UMSCC-1 and UMSCC-9 with wtTP53 but not UMSCC-38 and UMSCC-46 with mtTP53 (Figure 2B; not shown). Conversely, SNX2112 showed enhanced activity with cisplatin in UMSCC-38 and UMSCC-46, with mtTP53. We observed potent suppression of tumorigenesis in an HNSCC xenograft model with inhibition of EGFR, c-MET, p-AKT, p-ERK, SRC, and p-STAT3 and, conversely, the re-expression of proapoptotic TP53 and PUMA (Figures 4 and and5).5). Corresponding reductions in proliferation (Ki-67) and angiogenesis (CD31) and increased apoptosis by TUNEL assay were observed. Together, these preclinical studies support further investigation of SNX5422/2112 and these molecular targets, which are implicated in sensitivity to HSP90 inhibitor alone or in combination with cytotoxic chemotherapy and radiotherapy in patients with HNSCC.
We have previously shown that HNSCC tumors and UMSCC lines exhibit concurrent activation or disruption of several signaling pathways, which contain multiple HSP90 clients important in cancer (Figure 6). These include EGFR, c-MET, and TNF receptors and their ligands that contribute to cross-activation of theMEK-ERK-AP-1 and PI3K-AKT axes [22,49–52], TNF and IL-1 that co-activate the IKK-NF-κB axis [52,53], and EGF, IL-6, or both that can variably induce MEK- or Janus activated kinase-mediated STAT3 pathway activation . Conversely, we showed that tumor suppressor TP53 is inactivated or mutated in subsets of HNSCC  and inversely co-modulated with these other pathways, suggesting a potential linkage . The altered signaling pathways in this network have been individually shown to contribute to the promotion of cell proliferation, survival, expression of inflammatory and angiogenesis factors, and therapeutic resistance that comprise important hallmarks of cancer [13,22,44–53]. However, intrinsic and acquired resistance to therapies targeting individual components, such as EGFR, proteasome-dependent activation of NF-κB, or mutated or inactivated TP53, has been attributed to limited modulation or compensatory activation of other signal molecules and pathways in experimental studies and clinical trials [5,7,12]. The concurrent modulation of multiple signaling pathways by HSP90 inhibition uniquely differentiated its effects from previous targeted agents. Specifically, EGFR inhibitors often failed to simultaneously block AKT, ERK, and SRC kinases and downstream transcription factors NF-κB, AP-1, and STAT3 [5,12]. While proteasome inhibitor bortezomib effectively inhibited the activation of NF-κB p65, it did not inhibit other signaling molecules, such as noncanonical NF-κBs, p-ERK1/2, or STAT3 . MET inhibition also affected downstream signal inhibition in only a subset of HNSCC . In addition to the ability of HSP90 inhibitors to concurrently modulate multiple key molecular targets, they can enhance standard cytotoxic modalities such as chemotherapy and radiation therapy in HNSCC (Figure 2).
In our study, we observed differential activity of SNX2112 on cell growth arrest, apoptosis, combination therapy, and HSP90 client expression based on dose and duration of exposure to drug and the cell line examined. At lower concentrations (25-50 nM SNX2112), inhibition of c-MET, p-AKT, and p-ERK and induction of TP53 and p21 were associated with decreased cell proliferation in both UMSCC-1 and UMSCC-38 (Figures 1, ,3,3, and W2–W4). These findings are consistent with our previous demonstration that HGF-c-MET contributes to activation of p-ERK and p-AKT and proliferation, which can be partially blocked by MEK or PI3K inhibition . In the prior study, we further showed that HGF-c-MET induced MEK-ERK and PI3K-AKT activation contributes to co-expression of angiogenesis factors IL-8 and VEGF, which were also inhibited at lower concentrations of SNX2112 in the present study (Figure 3, B and C). However, oscillation and re-phosphorylation of ERK and AKT were observed at lower concentrations over the time course between 48 and 72 hours. Of potential relevance, higher concentrations of SNX2112 (≥200 nM) were required for inhibition of EGFR, which can co-activate ERK, AKT, as well as STAT3, and for inhibition of IKK-NF-κB pathway components. Similarly, high concentrations of SNX2112 were required for inhibition of STAT3 and NF-κB-regulated antiapoptotic gene BCL-XL, cleavage of apoptotic marker PARP (Figures 3 and W4), and decreased clonogenic cell survival (Figure 2A). wtEGFR was recently shown to be a client but a relatively resistant target for HSP90 inhibitors . Both STAT3 and IKK proteins appeared to be increased at 100 nM and required 200 nM SNX2112 for inhibition. Together, these observations suggest that incomplete HSP90 or target inhibition may enhance phosphorylation of ERK and AKT and expression of STAT3 and IKKs at concentrations below those required for destabilization and degradation of these proteins. Determination of the mechanism(s) for the compensatory enhanced expression and activation of these pathways that are critical to HNSCC cell survival could potentially lead to other combinations with HSP90 inhibitors that would enhance inhibition and cytotoxicity.
A potentially important finding of this study is that TP53, PUMA, and p21 were upmodulated by HSP90 inhibition in HNSCC with wtTP53 (Figure 3A). AKT has been reported to be a key node inhibiting TP53 gene transactivation by phosphorylation and modulation of cofactors, CBP/p300 and MDM2 [54–57]. Recently, another role of HSP90 inhibitors in releasing HSP90-RNA polymerase complexes paused at promoters has been implicated in upregulating repressed genes including TP53 . We also observed combined effects of SNX2112 with cytotoxic chemotherapy agents paclitaxel and cisplatin that were related to TP53 status. Paclitaxel has been reported to induce p21, inhibit microtubule-dependent cell division and G2/M block, and induce cell death enhanced by TP53 , potentially explaining the combined effect observed in HNSCC with wtTP53. Cisplatin cytotoxicity can be mediated by degradation of TP53 family antagonist ΔNp63 and activation of TAp73, which we recently showed are overexpressed with mtTP53 in the subset of UMSCC that were sensitive to combined cisplatin and SNX2112 [60,61]. Confirmation of these combined effects in wider subsets could potentially help in enhancing the limited cytotoxicity observed with these chemotherapeutics and in selecting optimal combinations for patient therapy.
HSP90 inhibition potently suppressed tumorigenesis in a preclinical HNSCC xenograft model in concert with potent inhibition of prooncogenic signaling and induction of expression of proapoptotic TP53. Supporting the clinical potential of HSP90 inhibition and SNX5422, a recent NIH phase I study found SNX5422 administered twice weekly was well tolerated with mild adverse events . Fifteen of 32 patients (47%) achieved disease stabilization, with 22% exceeding 150 days after progression of a variety of treatment refractory solid tumors and lymphomas. SNX5422 administered every other day at doses ranging from 9 to 100 mg/m2 was also tolerated, with one complete response, one partial response, and seven patients with stable disease for >150 days . Diarrhea was dose limiting, as found in our murine study. However, subjects receiving 50 to 89 mg/m2 SNX5422 daily exhibited dose-limiting ocular toxicity, and thus, every other day schedules have been recommended for future studies [39,40]. Importantly, the molecular and anticancer effects observed in HNSCC in the 25 to 250 nM range in vitro and concentration average for free drug for 50 mg/kg every other day used in mice in our preclinical studies were within the range attained in serum in patients who demonstrated disease-stabilizing activity and acceptable toxicity.
Regarding clinical potential of HSP90 inhibitors in squamous cell carcinoma, we previously observed prolonged stabilization of pulmonary metastasis for 8 months in a heavily pretreated patient in an NIH phase I study using geldanomycin 17-DMAG . Geldanomycin derivative IPI-504 in combination with docetaxel was reported to induce partial responses or prolonged stable disease, particularly in patients with squamous cell carcinomas of the lung . Together, these clinical data and our preclinical data support further investigation of SNX5422 and HSP90 inhibitors in combination with chemotherapy or radiation therapy in HNSCC.
Nine HNSCC cell lines (UMSCC) were obtained from Dr T. E. Carey (University of Michigan, Ann Arbor, MI) and were extensively characterized in previous studies [13,36,37]. These cell lines have been found to reflect many of the molecular and phenotypic alterations important in the pathogenicity of HNSCC. Three cell lines exhibited decreased expression of wtTP53 (UMSCC-1, -6, and -9) and six express mtTP53 (UMSCC-11A, -11B, -22A, -22B, -38, and -46). Normal primary human epidermal keratinocytes (HEKA; Invitrogen) were isolated from skin of different individual adults, established as primary cell cultures under low calcium conditions, and used as a nonmalignant control within five passages.
RNA isolation and cDNA synthesis were performed as previously described . Real-time RT-PCR for HSP90 used the Assays-on-Demand Gene Expression Assay from Applied Biosystems (Foster City, CA). Amplification conditions were given as follows: activation of enzymes for 2 minutes at 50°C, then 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. Thermal cycling and fluorescence detection were done using an ABI Prism 7700 Sequence Detection System (Applied Biosystems). Measured HSP90 mRNA was normalized to measured 18S rRNA. Each sample was assayed in triplicate, and data were reported as means ± SD.
Goat polyclonal anti-human TP53 and p21, mouse monoclonal anti-human STAT3, rabbit polyclonal anti-human PARP, Bcl-xL, and c-MET were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit monoclonal anti-human p-ERK1/2 and p-AKT ser473, rabbit polyclonal anti-ERK1/2, AKT, anti-human IKKα, p-p65 (Ser536), p-p65 (Ser276), p65, PUMA, EGFR, p-STAT3, and β-actin were obtained from Cell Signaling Technology (Danvers, MA). Rabbit polyclonal anti-human IKKβ was obtained from Abcam (Cambridge, MA). Rat monoclonal anti-mouse CD31 was obtained from BD Pharmingen (San Jose, CA), rabbit monoclonal anti-human Src from Cell Signaling Technology/Millipore (Billerica, MA), and mouse monoclonal anti-human Ki-67 from Dako (Carpinteria, CA). Goat and rabbit IgGs were used for isotype controls (Vector Laboratories Inc, Burlingame, CA).
Whole-cell lysate was extracted from tumor cells using a Nuclear Extraction Kit from Active Motif (Carlsbad, CA) and mixed with NuPAGE LDS sample buffer and NuPAGE sample reducing agent from Invitrogen. The lysate mixture was heated at 70°C for 10 minutes and subjected to electrophoresis (200 V, 50 minutes) using Bis-Tris precast gels (Invitrogen). Proteins were transferred from the Bis-Tris gel onto a 0.45-µm nitrocellulose membrane using the Invitrogen Gel Blot Module for 70 minutes at 30 V and 4°C. Primary and secondary antibodies (see above) were diluted in either 5% nonfat powdered milk or 5% BSA solution prepared from Tween 20-TBS. Membranes were incubated in primary antibody overnight at 4°C, washed in buffer, and incubated in secondary antibody for 30 minutes at room temperature. Each blot was incubated with Pierce SuperSignal West Pico substrate (Pierce Biotechnology Inc, Rockford, IL) and exposed to Kodak X-OMAT film. Protein density for Western blots was measured using ImageJ 1.45k software (NIH). Individual SNX2112-treated protein band densities were compared against their respective actin loading control and 0-hour untreated condition. Relative densities are reported as numbers relative to 1, the arbitrary value chosen for each control.
A human tissue array  from de-identified donors (Cybridi Inc, Gaithersburg, MD) was used for studies under exemption from Institutional Review Board review by the Office of Human Subjects Research, NIH. The array included 20 HNSCC sections in triplicate and 6 normal mucosa sections in duplicate that were verified by a certified pathologist by histologic hematoxylin and eosin (H&E) and immunohistochemical pan-cytokeratin staining. Each tissue section was 1.5 mm in diameter and 5 µm in thickness. The sections were deparaffinized in xylene and rehydrated in serial dilutions of ethyl alcohol. Antigens were retrieved using a standard heat-citrate method. Endogenous peroxidase activity was blocked using 5% hydrogen peroxide. Sections were washed in phosphate-buffered saline, incubated for 30 minutes in 2.5% BSA blocking solution, and incubated overnight at 4°C in a 1:100 dilution of rabbit polyclonal anti-human HSP90 antibody (Cell Signaling Technology). Sections were incubated for 30 minutes in biotinylated anti-rabbit secondary antibody (Vector Laboratories Inc) at a 1:400 dilution. ABC reagent (Vector Laboratories Inc) was added for 30 minutes after phosphate-buffered saline wash, then sections were developed in DAB (Vector Laboratories Inc). Sections were counterstained using hematoxylin (Sigma-Aldrich Inc). Sections were dehydrated using graded alcohol and xylene washes and mounted in Permount (Fisher Scientific Inc, St Louis, MO). Staining intensity, representative of HSP90 expression, was quantified on a Likert score scale of +1 (low), +2 (moderate), and +3 (high) by inspection under light microscope [high-power field (HPF), x400] by two blinded reviewers.
Frozen sections were fixed in 4% paraformaldehyde and the endogenous tissue peroxidase activity was blocked using 0.3% hydrogen peroxide. Sections were incubated with primary antibody overnight at 4°C and then incubated in biotinylated secondary antibody (Vector Laboratories Inc) for 1 hour. ABC reagent was prepared as per the manufacturer's instructions (Vector Laboratories Inc) and added for 30 minutes, and slides were then developed in stable DAB (Vector Laboratories Inc). The reaction was stopped by washing in distilled water. Mayer's hematoxylin (Sigma, St Louis, MO) was used as counterstain. The slides were dehydrated using graded alcohol and Safeclear and mounted in Permount (Fisher Scientific, Rockville, MD). Photomicrographs were taken under light microscopy at x400 magnification. TUNEL assay was performed with the DeadEnd Colorimetric TUNEL System from Promega (San Luis Obispo, CA). Vessel area for CD31-stained sections and apoptotic index for TUNEL-stained sections were measured per HPF by converting pictures taken by light microscope (x40) to grayscale and setting a consistent threshold using ImageJ software.
Cell proliferation was measured by standard MTT or XTT assay (Roche Diagnostics), according to the manufacturer's instructions. UMSCC-1 cells were plated at 5 x 103 cells per well in 100 µl of complete minimum essential medium (MEM) in 96-well plates and incubated overnight at 37°C. Exponentially growing cells were exposed to SNX2112 as indicated and vehicle control samples were assayed in parallel. For experiments examining the potential role of TP53, UMSCC-1 cells were pretreated with TP53 inhibitor Pifithrin-α (10 µM; Sigma-Aldrich Inc) or reverse transfected with ON-TARGETplus Human TP53 (3 pmol/well; Thermo Scientific, 7157) or nontargeting pool (3 pmol/well; Thermo Scientific, 1810) SMARTpool siRNA using Lipofectamine RNAiMax (Invitrogen) in 96-well plates according to the manufacturer's instructions. Each sample was assayed in quadruplicate. Absorbance was measured at a wavelength of 570 nm, and data were reported as mean optical density ± SD.
DNA cell cycle distribution analysis was performed by flow cytometry. UMSCC-1 cells were plated onto a six-well plate at 1 x 105 and 2 x 104 cells per well in 100 µl of complete MEM and incubated overnight at 37°C. Exponentially growing cells were exposed to 50 µM SNX2112 or media with vehicle as control, incubated for 24, 48, or 72 hours, then stained with propidium iodide (Cycletest Plus DNA Reagent Kit; BD Biosciences, San Jose, CA). Samples were run on the FACS Canto machine, and the data were analyzed using Diva Flow Software (BD Biosciences).
UMSCC-1 cells were plated (5 x 105 cells/100-mm culture dishes) and incubated for 16 hours at 37°C in complete MEM. SNX2112 (250 nM final concentration) was added to the exponentially growing cells, incubated for 24 hours, and then irradiated. A range of radiation doses was delivered to cell samples using an Eldorado 8 - cobalt-60 teletherapy unit (Theratronics International Ltd, Kanata, Ontario) at dose rates of 2.0 to 2.5 Gy/min. Vehicle control radiation survival curves were conducted in parallel. Following radiation treatment, cells were trypsinized, counted, plated, and incubated for 10 to 14 days. Colonies were counted and scored; cell survival was determined after correcting for the plating efficiency and for SNX2112 cytotoxicity alone, and survival curve data were fit using a linearquadratic model. Survival curves were repeated twice. The DMF was determined by taking the ratio of radiation doses at the 10% survival level (control radiation dose divided by the SNX2112-treated radiation dose). DMF values > 1 indicate enhancement of radiosensitivity.
UMSCC-1 cells were plated in triplicate onto a 24-well plate at 5 x 104 cells per well in 500 µl of MEMwithout antibiotics. The next day, cells were transfected with Lipofectamine 2000 Transfection Reagent (Invitrogen) containing reporter plasmids in Opti-MEM reduced serum media (Invitrogen) for 5 hours. The NF-κB-luciferase reporter vector was obtained from Stratagene (La Jolla, CA). The AP-1-luc reporter vector was obtained from Dr Nancy Colburn (National Cancer Institute). The STAT3-luc reporter vector was obtained from Panomics (Freemont, CA). The IL-8 reporter (133 bp) has been described previously [22,23]. The Bcl-xL promoter sequence (1498 to +427) was PCR-amplified, sequence-confirmed, and constructed with luciferase reporter gene plasmid using pBV-Luc by Dr Lin Zhang (University of Pittsburgh, Pittsburgh, PA). After transfection, cells were exposed to fresh medium without or with SNX2112. The cell lysates were harvested and reporter gene activities were assayed. Each sample was assayed in triplicate, and data were presented as means ± SD.
Technical advice of J. Kan and N. Brega of Pfizer Inc and Steve Hall and Eric Orlemans of Esanex Inc are gratefully acknowledged. The authors thank Ning T. Yeh, Xinping Yang (National Institute on Deafness and Other Communication Disorders, National Institutes of Health), and Chris Silvin (National Human Genome Research Institute, National Institutes of Health) for their technical assistance and Shivaani Kummar and Barbara Conley (National Cancer Institute, National Institutes of Health) for reading and providing helpful comments on the manuscript.
1This work was supported by National Institute on Deafness and Other Communication Disorders (NIDCD) intramural research program projects ZIA-DC-000073 and 000074. S.C.W., M.H., C.G., R.V.B., and V.R.K. were supported by the Medical Research Training Program, a public-private partnership jointly sponsored by the National Institutes of Health (NIH) and Pfizer Inc (through a grant to the Foundation for NIH from Pfizer Inc). Synthetic HSP90 inhibitor SNX5422 (PF-04928473) and its oral prodrug SNX2112 (PF-04929113) were provided to NIDCD and C.V.W. for research under a Material Transfer Agreement with Pfizer Inc or Esanex Inc. C.V.W. has no financial interests to disclose.