Pharmacological Inhibition of HSP90 Radiosensitizes Head and Neck Squamous Cell Carcinoma Xenograft by Inhibition of DNA Damage Repair, Nucleotide Metabolism, and Radiation-Induced Tumor Vasculogenesis

Abstract

Purpose:

Recent preclinical studies suggest combining the HSP90 inhibitor AT13387 (Onalespib) with radiation (IR) against colon cancer and head and neck squamous cell carcinoma (HNSCC). These studies emphasized that AT13387 downregulates HSP90 client proteins involved in oncogenic signaling and DNA repair mechanisms as major drivers of enhanced radiosensitivity. Given the large array of client proteins HSP90 directs, we hypothesized that other key proteins or signaling pathways may be inhibited by AT13387 and contribute to enhanced radiosensitivity. Metabolomic analysis of HSP90 inhibition by AT13387 was conducted to identify metabolic biomarkers of radiosensitization and whether modulations of key proteins were involved in IR-induced tumor vasculogenesis, a process involved in tumor recurrence.

Methods and Materials:

HNSCC and non-small cell lung cancer cell lines were used to evaluate the AT13387 radiosensitization effect in vitro and in vivo. Flow cytometry, immunofluorescence, and immunoblot analysis were used to evaluate cell cycle changes and HSP90 client protein’s role in DNA damage repair. Metabolic analysis was performed using liquid chromatography−Mass spectrometry. Immunohistochemical examination of resected tumors post-AT13387 and IR treatment were conducted to identify biomarkers of IR-induced tumor vasculogenesis.

Results:

In agreement with recent studies, AT13387 treatment combined with IR resulted in a G2/M cell cycle arrest and inhibited DNA repair. Metabolomic profiling indicated a decrease in key metabolites in glycolysis and tricarboxylic acid cycle by AT13387, a reduction in Adenosine 5’-triphosphate levels, and rate-limiting metabolites in nucleotide metabolism, namely phosphoribosyl diphosphate and aspartate. HNSCC xenografts treated with the combination exhibited increased tumor regrowth delay, decreased tumor infiltration of CD45 and CD11b+ bone marrow−derived cells, and inhibition of HIF-1 and SDF-1 expression, thereby inhibiting IR-induced vasculogenesis.

Conclusions:

AT13387 treatment resulted in pharmacologic inhibition of cancer cell metabolism that was linked to DNA damage repair. AT13387 combined with IR inhibited IR-induced vasculogenesis, a process involved in tumor recurrence post-radiotherapy. Combining AT13387 with IR warrants consideration of clinical trial assessment. © 2021 Elsevier Inc. All rights reserved.

Introduction

Several studies have established targeting HSP90 as a potential multitarget approach to overcome tumor adaptive/resistance mechanisms and improve tumor sensitivity to the standard treatment and some targeted therapies.^1,2 HSP90 inhibitors developed in the past have not received approval for any cancer types owing to inadequate single-agent clinical activity, off-target effects, and HSP-related toxicities.^3–7

AT13387 (Onalespib) is a non-ansamycin second-generation HSP90 inhibitor. It differs from other HSP90 inhibitors in its longer duration of target inhibition^8,9 and its favorable safety profile.^4,10 Several preclinical studies have demonstrated that AT13387 can degrade HSP90 client proteins in various cell lines^8,9,11 and tumor xenograft models with a prolonged duration of action, enabling once-weekly administration.⁸ In glioblastoma, AT13387 can inhibit HSP90 in vivo by crossing the blood-brain barrier, thereby extending survival as a single agent when combined with temozolomide in both zebrafish and glioblastoma stem cell mouse xenografts.⁹

Several clinical studies have failed to show the efficacy of AT13387 as a single agent in most solid tumors. Therefore, AT13387 is currently tested in combination with various targeted agents in different types of cancer. Potential combination therapy targeting HSP90 with standard-of-care radiation therapy is promising because several HSP90 client proteins are linked with tumor radiation (IR) response,¹² although in a cell type-dependent manner.

Previous preclinical work from others and our group has shown that first-generation HSP90 inhibitors, namely geldanamycin (17AAG), geldanamycin analogs (17DMAG), and radicicol (natural products), are potential IR modifiers.^13,14 Recently, preclinical reports have also established AT13387 as a potential radiosensitizer in head and neck squamous cell carcinoma (HNSCC), pancreatic xenograft, and colorectal cell line models,^11,15,16 in vitro and in vivo. Mechanistically, these studies established that the inhibition of DNA repair and the degradation of HSP90 client protein expression contributes significantly to enhanced tumor regression in combination with IR.

In the present study, we expanded the mechanisms of radiosensitization by AT13387. Our data confirmed AT13387 as a radiosensitizer in non-small cell lung cancer (NSCLC) and HNSCC models. Our present data suggest a novel role of AT13387 in altering tumor cell metabolism, including a significant decrease in Adenosine 5’-triphosphate (ATP) production, metabolic changes affecting glycolysis and the tricarboxylic acid cycle (TCA), and dramatically reduced levels of metabolites governing rate-limiting steps of purine and pyrimidine biosynthesis. Lastly, we demonstrate that AT13387 treatment in vivo in an adjuvant setting after fractionated IR leads to significant tumor regrowth delay by inhibiting IR-induced vasculogenesis.

Methods and Material

Details regarding procedures can be found in the Methods E1.

Cell survival studies

UMSCC1 (HNSCC) were kindly provided by one of the coauthors in this study. A549 and H460 wt and H460DNp53 (NSCLC) and normal human fibroblast (1522) cell lines were obtained and maintained as previously described.^17,18 AT13387 was obtained through the Cancer Therapy Evaluation Program, National Cancer Institute, or Selleck Chemicals. A drug stock solution was prepared in 100% dimethyl sulfoxide (Sigma) and stored at −70°C. AT13387 was added to cells for 24 hours, either before or after radiation treatment (RT) treatment at a concentration of 100 nM. Clonogenic cell survival studies were performed as described previously.^17,18 All experiments were repeated 3 times. IR dose-modifying factors (DMFs) were determined at 10% survival levels by dividing the IR dose for control by the IR dose for drug treated. DMFs >1.0 indicate enhancement of radiosensitivity.

Metabolite extraction and mass spectrometry analysis

Metabolite extraction and detailed capillary electrophoresis time of flight and tandem triple quadruple mass spectrometry mass spectrometry analysis was performed as described previously.¹⁷

Aspartate measurement

Cells were plated and treated with AT13387 (100 nM) for 24 hours. Cells were then trypsinized and counted. Approximately 1 million cells were lysed in 150 μL of RIPA buffer for 30 minutes on ice. Cell lysate was centrifuged at 13,000 rpm for 10 minutes at 4°C, and supernatant was collected and stored at −80°C. The fluorometric assay was performed using a Biovision Aspartate kit (Cat# #K552–100) as per the manufacturer’s instructions.

Xenograft studies

All animal experiments were performed in accordance with protocols approved by the National Cancer Institute Animal Care and Use Committee (ACUC). For IR regrowth delay studies, tumor cells (1 × 10⁶) were injected into the subcutaneous space of the right hind leg of athymic nude mice, 5 to 6 weeks of age. Individual mice were ear tagged and randomized on day 6 of tumor growth into the following 4 groups (8 mice/group): vehicle control, vehicle control + fractionated IR, AT13387 control, and AT13387 + fractionated IR. Tumor growth delay was calculated as the time (days) to reach a tumor size of 600 mm³ (3 × the initial tumor size) starting from the size measured at day 7 (∼200 mm³). Tumor volumes for each animal were recorded several times per week. The measured growth curves for each animal in each group were fitted with exponential growth curves using Microsoft Excel to determine the time to reach 600 mm³ tumor volume.

Statistical analyses

Statistical significance for densitometry was obtained using ImageJ and GraphPad Prism 7 software. The Student t test in Microsoft Excel was used to quantify the growth differences of the xenograft group.

Results

Inhibition of HSP90 activity by AT13387 leads to radiosensitization of HNSCC and NSCLC cells

The effect of AT13387 on in vitro IR sensitivity was assessed in NSCLC and HNSCC cell lines. Cells were treated with either dimethyl sulfoxide or 100 nM AT13387 in combination with different doses of IR, and complete IR dose response curves were generated via colony formation assay. Figure 1A to 1C shows the IR survival curves of FaDU cells treated with AT13387, 24 hours before (A and C) or for 24 hours after (B) IR treatment (RT). Cell cycle and DMF results for each cell line tested are tabulated in Table 1.

Fig. 1.

AT13387 sensitizes a panel of head and neck squamous cell carcinoma and NSCLC cell lines to radiation. Radiation survival curves for indicated cancer cell lines treated with AT13387. (A-C) Representative radiation survival curves are shown for proliferating (A and B) and plateau phase (C) FaDU cells treated with AT13387 (100 nM, dashed curve with open circles) or with dimethyl sulfoxide control (0.1%, solid curve with filled circles) for 24 hours before irradiation (A and C) and 24 hours after irradiation (B). (D) Representative radiation survival curve of normal proliferating human fibroblast (1522) cells treated with AT13387 24 hours before irradiation. Data are representative of 2 to 3 independent experiments. Error bar represents standard error of the mean (SEM). Radiation survival curves for HNSCC (FaDU and UMSCC-1) and NSCLC cell lines (A549, H460wt, and H460DNp53) were obtained upon treatment with AT13387 (100 nM) for 24 hours. Dose-modifying factors, AT13387 toxicities, and its effect on the cell cycle of indicated cell lines are shown in Table 1.

Table 1.

Effect of AT13387 on radiation response, drug toxicity, and cell cycle changes

Cell lines	24-h pre-IR treatment DMF (± SEM)	24-h post-IR treatment DMF (± SEM)	24-h AT13387 (100 nM) exposure (no IR)
			Surviving fraction (± SEM)	Cell cycle
				% G1	% S	% G2/M
A549	1.53 (± 0.08)	1.03 (± 0.01)	0.94 (± 0.01)	30	19	51 (28)
H460 wtp53	1.69 (± 0.19)	1.24 (± 0.07)	0.64 (± 0.03)	29	18	53 (21)
H460 DNp53	1.61 (± 0.21)	1.45 (± 0.14)	0.60 (± 0.13)	31	11	58 (27)
FaDU	1.50 (± 0.10)	1.75 (± 0.15)	0.25 (± 0.09)	24	22	49 (17)
UMSCC-1	1.45 (± 0.05)	1.12 (± 0.02)	0.88 (± 0.01)	22	24	54 (26)
1522	1.03 (± 0.02)	1.21 (± 0.12)	0.72 (± 0.10)	66	15	19 (21)
A549 plateau	1.03 (± 0.04)	1.10 (± 0.02)	1.0 (± 0.06)	91	6	3 (9)
FaDU plateau	1.0 (± 0.02)	1.02 (± 0.01)	1.02 (± 0.03)	88	7	5 (11)

Abbreviations: DMF = dose-modifying factor; DN = dominant negative; IR = ionizing radiation; SEM = standard error of the mean; wt = wild type.

Effect of AT13387 toxicity, cell cycle changes, and IR sensitivity in head and neck squamous cell carcinoma and NSCLC cell lines. Values in parentheses are the percent G2/M of untreated control cells (dimethyl sulfoxide treated). Toxicity was based on a 100 nM AT13387 exposure for 24 hours.

All cell lines showed an increase in IR sensitivity 24 hours after pre-IR treatment with A13387, with DMFs ranging from 1.4 to 1.7 (Table 1). Post-IR treatment sensitization with AT13387 was more variable with FaDU, H460, and H460DNp53 cells, showing enhanced radiosensitivity with DMFs of 1.75 ± 0.15, 1.24 ± 0.07, and 1.45 ± 0.14, respectively. A549 and UMSCC-1 cells were not radiosensitized (DMF = 1.03 ± 0.03 and 1.12 ± 0.02, respectively; Table 1). Notably, 24-hour preincubation with 100 nM AT13387 failed to enhance IR cell killing in plateau phase (quiescent) FaDU (Fig. 1C) and A549 cells (Fig. E1) and exponentially growing normal human 1522 fibroblast cells (Fig. 1D). However, AT13387 administration post-IR exposure modestly radiosensitized normal human fibroblasts (1522), with a DMF of 1.21 ± 0.12 (Table 1).

Both pre- and post-IR AT13387 incubation was performed for H460 (WT p53) and H460 DNp53 (dominant negative p53) to determine whether altered p53 status would affect the AT13387 effect. As shown in Table 1, inactivating p53 did not alter radiosensitization by AT13387. A 24-hour exposure to 100 nM AT13387 alone was not particularly toxic for most of the cell lines shown in Table 1 (survival >60%); however, AT13387 was toxic to the FaDU cells (25% survival). Similar to geldanamycin analog 17AGG (13), AT13387 treatment for 24 hours also led to a modest increase in the percent of cells (∼20% above the control cells) in the G2/M phase of the cell cycle (Table 1); this is a radiosensitive phase of the cell cycle, suggesting that radiosensitization could partly be due to cell-cycle redistribution into G2/M by AT13387 treatment.

Taken together, our data agree with findings from Mehta et al¹⁵ that a subcytotoxic concentration (100 nM) of AT13387 can enhance the HNSCC tumor IR response regardless of any p53 genetic alterations present in the tumor cells. Given the consistent enhancement of radiosensitivity observed in most of the HNSCC and NSCLC tumor cell lines in vitro after pretreatment with AT13387, most of the subsequent mechanistic studies were conducted with a 24-hour incubation with 100 nM AT13387 before IR exposure.

As previously reported for other HSP90 inhibitors,^10,13 including AT13387,^11,15,16 inhibition of HSP90 activity with AT13387 in FaDU cells showed an upregulation in HSP70 protein expression (Fig. E2A). HSP70 is a cochaperone protein, and its upregulation is an established surrogate biomarker of HSP90 inhibition. HSP70 protein levels were elevated as a function of increasing AT13387 concentrations (Fig. E2A). Other HSP90 client proteins, such as p-S6 (mTOR target), p-ERK1/2, Rad51, CDK4, and EGFR, were substantially reduced in a dose-dependent manner (10 nM to 1 μM) on treatment with AT13387 (Fig. E2A).

Because our data suggested the importance of cell cycle movement as a mechanism of AT13387 radiosensitzation, we examined whether mitotic catastrophe was enhanced after drug treatment. We sought to determine whether features of mitotic catastrophe such as micronucleation, altered nuclear morphology, and multinuclei formation^19,20 occurred on AT13387 treatment alone or in combination with IR (3 Gy). As shown in Figure E2C, neither AT13387 alone nor IR treatment resulted in increased mitotic catastrophe during 24 to 48 hours of treatment. In contrast, we observed a number of multipolar (Fig. E2B panel B) versus bipolar metaphases (Fig. E2B panel A) and fragmented nuclei (micronuclei) formation in cells after the combination of AT13387 + IR treatment (Fig. E2B panel B) significantly as a function of time. Taken together, these results indicate that HSP90 inhibition by AT13387 leads to alterations of mitosis in irradiated cells, leading to improper or incomplete cell division that favors aneuploidy or polyploidy or possible mitotic cell death.

HSP90 inhibition impairs de novo nucleotide biosynthesis

Efficient DNA synthesis is required during DNA damage repair.²¹ However, how DNA damage repair in cancer cells is linked to metabolic reprogramming remains elusive. To determine whether reduced DNA damage repair after AT13387 + IR treatment may be related to alterations in tumor metabolism, we performed a global metabolic profiling of FaDU cells treated with AT13387 alone and in combination with IR.

We quantified steady-state levels of 116 metabolites involved in several metabolic pathways, namely the TCA cycle, glycolysis, pentose phosphate pathway, and amino acid metabolism. Overall, changes observed in the levels of metabolites studied were primarily due to AT13387 treatment; IR alone or IR combined with AT13387 incubation did not show additional changes in metabolites levels compared with the AT13387 treated cells (data not shown). Notably, a significant reduction in several key metabolites important in nucleotide biosynthesis and subsequently in DNA damage repair was observed (Fig. 2A and B). Significant inhibition of key rate-limiting metabolites (ie, aspartate, N-carbamoyl-L-aspartate, and phosphoribosyl diphosphate, which are known to be involved in pyrimidine biosynthesis (Fig. 2A), and glycine, adenylo-succinate, adenine, and inosine monophosphate, which are involved in the purine salvage pathway) was observed (Fig. 2A and C). Figure 2C shows that although the levels of several amino acids (eg, alanine, serine, and glutamate) were reduced by 25% compared with control, aspartate was the only amino acid showing a significant 75% reduction compared with untreated control cells. Figure E3B suggested a possible reason for the aspartate reduction, as both fumarate and malate were reduced by 90% compared with untreated control cells. Malate provides multiple carbons for the synthesis of aspartate. Therefore, its reduction is consistent with the decrease in overall aspartate levels.²²

Fig. 2.

Effect of AT13387 on tumor metabolism. (A-D) Liquid chromatography−Mass spectrometry analysis showing fold expression of steady state levels of key metabolites involved in pyrimidine (A) and purine (B) biosynthesis in FaDU cells treated in the presence (red bars) and absence (black bars) of AT13387 for 24 hours. (A) Schematic of the de novo pyrimidine synthesis pathway. Pyrimidine synthesis enzymes: CAD = carbamoyl-phosphate synthetase 2 (E1); aspartate transcarbamylase (E2); and dihydroorotase (E3); DHODH = dihydroorotate dehydrogenase; UMPS = uridine monophosphate synthetase. (B) Schematic of the de novo and purine salvage pathways. Purine synthesis enzymes: ADSL = adenylosuccinate lyase; ADSS = adenylosuccinate synthase; APRT = adenine phosphoribosyltransferase; ATIC = 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase; GART = glycinamide ribonucleotide transformylase; GMPS = guanine monophosphate synthase; HPRT = hypoxanthine phosphoribosyltransferase; IMPDH = inosine monophosphate dehydrogenase; PAICS = phosphoribosylaminoimidazole carboxylase and phosphoribosylamino-imidazolesuccinocarboxamide synthase; PFAS = phosphoribosylformylglycinamidine synthase; PPAT = phosphoribosyl pyrophosphate amidotransferase. Adapted and modified from Villa et al.²² (C) Fold expression of key amino acid changes upon inhibiting HSP90 activity by AT13387. (D) Fold change in the levels of total ATP, GTP, total AN (total adenylate [ADP + AMP + ATP]), and total GN (total guanylate [GDP + GMP + GTP]). The levels of indicated metabolites in the supernatants were normalized to concentration (μg) of protein in the sample. (E) Fold change in the total intracellular ATP levels measured using ATP assay kit (Methods E1) with AT13387, IR, and AT13387 + IR combination. All experiments were done in triplicate, and error bars represent mean ± standard deviation. *P < .01, **P < .001, and ***P < .0001.

Other TCA cycle members, such as citrate and succinate, were also reduced with the inhibition of HSP90, suggesting that the TCA cycle was also significantly affected by AT13387 treatment. Given the effect of AT13387 on the TCA cycle and adenosine and guanosine levels (Fig. 2D), it was unsurprising to find that both ATP and GTP levels were reduced by 45% to 55% compared with untreated cells (Fig. 2D, E). Furthermore, fluorometric assessment of intracellular aspartate in FaDU cells treated with AT13387 and the combination of AT13387 + IR also produced additional decreases in the ATP levels (Fig. 2E). Metabolites involved in the glycolytic pathways were also reduced (although not significantly), suggesting that glycolysis was negatively affected by AT13387 incubation (Fig. E3A).

In summary, global metabolic profiling of several metabolites suggests that AT13387 treatment leads to a reduction in the metabolic flux toward nucleotide biosynthesis for efficient DNA repair and thus contributes to increased sensitivity of cancer cells to IR.

Supplementation of aspartate partially restored AT13387 induced inhibition of cell proliferation

Sabatini et al²³ discovered that the vital function of the mitochondrial electron transport chain (ETC) is to enable the synthesis of aspartate.²⁴ Aspartate is a proteogenic amino acid required for cell proliferation and purine and pyrimidine biosynthesis. On ETC inhibition, aspartate levels become limited, leading to inhibition of cell proliferation. However, extracellular supplementation of aspartate or overexpression of glutamate aspartate transporter SLC1A3 allows cells with defective ETC activity to proliferate.²³ Because pharmacologic or genetic inhibition of the ETC can greatly suppress cell proliferation,²³ we studied the effect of prolonged AT13387 incubation on the cell growth of FaDU cells (Fig. 3A).

Fig. 3.

Effect of AT13387 on cell proliferation and intracellular aspartate biosynthesis. (A) AT13387 treatment significantly reduced cell number over 5-day prolonged treatment. Cells were trypsinized, and total cell numbers were counted every 24 hours. Changes in absolute cell number were plotted over time. AT13387 inhibits intracellular aspartate biosynthesis in a dose- (B) and time-dependent fashion (C). *P < .05, **P < .01, ***P < .001.

Figure 3A shows that prolonged exposure of FaDU cells to AT13387 significantly halted cell growth and was highly toxic to cells after 24 hours of incubation. Figure 3B shows the corresponding dose-dependent reduction in intracellular aspartate levels measured after 24 hours of AT13387 exposure. Figure 3C shows that when excess extracellular aspartate (2.5 and 5 mM) was added to the AT13387-treated FaDU cells, there was a partial reversal in the AT13387 cell toxicity, as seen by the modest increase in cell numbers. In addition to the effects of AT13387 on intracellular aspartate levels, AT13387 also reduced the expression of mitochondrial aspartate transporter AGC-1 (Fig. E3C).

Collectively, these results imply that inhibition of HSP90 activity by AT13387 may impair mitochondrial ETC function, leading to reduced aspartate biosynthesis in FaDU cells and contributing to the inhibition of IR-induced DNA repair and a corresponding reduction in cell growth.

AT13387 inhibits IR induced vasculogenesis and tumor regrowth in combination with fractionated IR

Our in vitro data clearly demonstrated that AT13387 can cause significant radiosensitization of FaDU cells by altering the cell cycle distribution and inhibiting nucleotide biosynthesis, leading to reduced DNA repair. To translate these findings in vivo, we conducted a pilot FaDU xenograft study combining fractionated IR (3 × 2 Gy fractions—Monday, Wednesday, Friday) with daily intraperitoneal (Monday to Friday) AT13387 (55 mg/kg) treatments over a 1-week period. AT13387 was administered 2 to 3 hours before each 2 Gy fraction.

Figure 4A shows that although both AT13387 and 2 Gy IR did slow tumor growth, the combination of IR + AT13387 was essentially additive but not statistically significant (P = .246) compared with IR alone. Immunoblot analysis of tumors taken from each group (n = 2) at the end of the first week of treatment is shown in Figure 4B. HSP70 was elevated in the AT13387 groups and CDK4 expression was reduced, similar to the levels observed in the in vitro findings (Fig. E2A). Collectively, these data suggest that the HSP90 target engagement and effects of AT13387 were present during the first week of IR treatments.

Fig. 4.

Pilot FaDU xenograft study combining AT13387 and fractionated IR (2 Gy × 3). (A) Mice with tumors measuring approximately 200 mm³ were given AT13387 intraperitonially 2 to 3 hours before exposure of the tumor-bearing leg to local fractionated IR (2 Gy) on Monday, Wednesday, and Friday (indicated by broken arrows). AT13387 was given once daily from Monday to Friday (indicated by bold arrows). Tumor regrowth kinetics over time were plotted. Statistical significance of tumor regrowth delay of IR + AT13887 combination was P = .246 (compared with IR), P < .001 (compared with AT only), and P < .0001 (compared with control). (B) Western blot analysis of FaDU xenograft (n = 2) showing HSP90 target inhibition (reduced CDK4 expression) and increased HSP70 expression in week 1 resected tumors.

Recent studies have suggested that IR-induced tumor vasculogenesis is a crucial player in the resistance of solid tumors to radiation therapy.^25,26 Essentially, it is both a molecular and physiologic response by irradiated tumors to overcome IR-induced growth arrest by initiating genes and proteins involved in angiogenic, metabolic, and immune suppression pathways.^27–30 Because HSP90 inhibitors are well-known modulators of these pathways, we designed a second in vivo tumor study wherein mice bearing FaDU tumors were given IR treatment for week 1 (2 Gy; Monday, Wednesday, Friday) with 5 daily intraperitoneal AT13387 treatments (55 mg/kg; Monday to Friday) followed by 5 daily AT13387 intraperitoneal injections (Monday to Friday) for week 2 (Fig. 5). As shown Figure 5A, both AT13387 and (3 × 2 Gy) IR significantly delayed growth to 600 mm³ (9.5-day delay [P < .001] and 2.1-day delay [P = .08] compared with control tumor growth, respectively). The combination gave a greater than additive growth delay compared with control untreated tumors (18.2 days; P < .001). Clearly, the second week treatment of AT13387 significantly altered tumor regrowth kinetics.

Fig. 5.

Effect of AT13387 on IR-induced vasculogenesis. (A) FaDU xenograft showing enhanced tumor regrowth delay with daily administration of 55 mg/kg body weight of AT13387 through intraperitoneal injections for 5 days 2 to 3 hours before (2 Gy × 3) IR exposure. IR dose was administered every alternate day (Monday, Wednesday, and Friday), as indicated by dashed arrows. The drug dosing was continued for another 5 days (second week), indicated with bold arrows. The significance for tumor regrowth delay was measured when tumor size reached 800 mm³. Statistical significance is derived from 8 animals per group. Statistical significance of tumor regrowth delay with the IR + AT13387 combination was P < .0001 (compared with IR), P < .0001 (compared with AT13387 only), and P < .0001 (compared with control) (B). Confocal microscopy images showing expression of Ki-67 (measure of mitotic index) and infiltration of bone marrow−derived cells (shown in green), such as CD45+ lymphocytes, CD11b+ monocytes, and SDF-1 secretion from tumor cells in the frozen section (shown in red) of FaDU xenograft derived from week 2 treatment of AT13387 with fractionated IR. The nucleus is stained with DAPI (shown in blue). All images were taken at a magnification of 200 ×. Scale bar is 10 μm. (C) Mean fluorescence intensity (MFI) of immune-positive cells of indicated proteins was measured for approximately 6 random fields of the stained section using Image J. The fold expression was calculated by normalizing the MFI to that of dimethyl sulfoxide−treated tumor sections. (D) Immunoblot analysis showing expression of HIF-1 in FaDU xenograft (n = 2) for week 1 and week 2 of AT13387 treatment administered 2 to 3 hours before fractionated IR. (E) Densitometric quantitation of the expression of HIF-1 expression as shown in (D).

Studies have indicated that irradiated cancer cells result in release of factors that lead to the recruitment of circulating endothelial precursor cells or bone marrow−derived hematopoietic cells to initiate vasculogenesis.^24,25 The switch from angiogenesis to vasculogenesis in the irradiated tumor is supported by the activation of the stromal cell−derived factor-1 (SDF-1) and its receptor (CXCR4) and stabilization of HIF-1.²⁶

To determine whether the treated tumors were also expressing proteins involved in the vasculogenesis program, several marker proteins (SDF-1, CD11b+, CD45, HIF-1) and Ki67 were quantitated in frozen sections (Fig. 5B–E). As reported by Kioi et al²⁶ and a previous study combining Abemaciclib with IR in NSCLC,^17,31 we found that immediately after the last fraction of 2 Gy IR in week 1, HIF-1 levels were increased in the IR-treated groups and that cotreatment with AT13387 reduced the HIF-1 level (Fig. 5D and E). In addition, HIF-1−regulated expression of SDF-1²⁵ also was shown to be elevated after the 3 × 2 Gy IR fractions, and cotreatment with AT13387 reduced SDF-1 expression to control levels (Fig. 5B, C). Finally, marker proteins for the influx of macrophages and myeloid-derived hematopoietic lymphocytes (CD11b+ and CD45, respectively) were also seen to be elevated by IR treatment and reduced by the AT13387 cotreatment (Fig. 5B and C).

Collectively, our data indicate that the IR and AT13387 combination inhibited multiple marker proteins involved in vasculogenesis, resulting in the inhibition of IR-induced vasculogenesis and a concomitant increase in tumor regrowth delay.

Discussion

HSP90 is a conserved molecular chaperone involved in proper folding of more than ∼200 client proteins. Therefore, it stands to reason that a potent HSP90 inhibitor would exert a variety of effects on tumor cells.^32,33 Studies have shown that several HSP90 inhibitors (geldanamycin, 17-AAG, and SNX5422/2112) when combined with IR can enhance tumor radiosensitivity.^13,34–36 Recently, 3 reports have demonstrated that a second-generation HSP90 inhibitor, AT13387, can also enhance tumor IR response in different types of tumors.^15,16,37 These studies emphasized the role of HSP90 inhibition on the IR-induced DNA damage response, as shown by increased YH2AX expression, delayed clearance of YH2AX foci, and degradation of HSP90 client proteins involved in DNA repair pathways. The present study strongly supports these previous findings; the DNA damage repair in the FaDU cell line was reduced after AT13387 incubation, as observed by YH2AX formation and resolution (Fig. E4A), and multiple DNA repair proteins were found to be reduced (Fig. E4B).

The present study also demonstrates the importance of movement of cells through the cell cycle to achieve radiosensitization after AT13387 incubation (Table 1). The importance of cell cycle movement for HSP90 inhibition IR sensitization was further corroborated by mitotic catastrophe formation (Fig. E2B and C). Progression through mitosis was disrupted and IR treatment was enhanced after AT13387 treatment. In cell lines in which movement through the cell cycle was reduced (plateau phase cells), no AT13387 enhancement of IR was observed (Fig. 1C and Fig. E1). These observations are in contrast to those of Mehta et al, who did not observe a G2/M block in MiaPaca cells and reported that cell cycle movement did not appear to be involved in HSP90 inhibition and IR sensitization.¹⁵ The effect of AT13387 on the cell cycle could be cell type dependent. Overall, our data suggest that partial synchronization of FaDu and other cell lines into more radiosensitive parts of the cell cycle may be an important aspect of HSP90 inhibition on IR sensitization.

In addition to effects on proteins associated with DNA damage repair, our results clearly indicate a unique role of AT13387 in altering tumor metabolism (Figs. 2 and 3 and Fig. E3). To our knowledge, our study is the first to explore the global metabolic changes observed with AT13387. Inhibition of HSP90 activity by AT13387 significantly inhibited levels of rate-limiting metabolites involved in de novo nucleotide metabolism, such as phosphoribosyl diphosphate (first and rate-limiting intermediate in purine biosynthesis) and aspartate, a key amino acid essential for tumor cell proliferation and pyrimidine biosynthesis (Fig. 3). In addition, AT13387 affected metabolic intermediates of glycolysis and TCA cycle (Fig. E3A and B, respectively). Strikingly, AT13387 reduced aspartate steady state levels in FaDU cells, which was confirmed by measuring intracellular aspartate and its functional implication in rescuing AT13387-induced cell proliferation (Fig. 3). These metabolic changes in tumor cells are promising from a clinical perspective because supporting aspartate biosynthesis is a vital function of ETC in proliferating cells.^23,24 AT13387 treatment also resulted in a significant decrease in total cellular ATP levels (Fig. 2D and E). DNA damage repair is an energy-consuming process, and decreases in ATP levels directly affect repair of the damaged DNA strands breaks.³⁸ Taken together, AT13387’s negative impact on maintaining the nucleotide pool and total ATP levels is an important contributor to its role as a tumor IR modifier.³⁸

Another unique finding of the study was the role of AT13387 in inhibiting HIF-1 and IR-induced tumor vasculogenesis. Studies from Brown et al²⁵ during the past decade have established that targeting tumor vasculogenesis, and not angiogenesis, can lead to reduced tumor recurrence after radiation therapy.³¹ In murine tumor models, reconstitution of the tumor vasculature post-IR, leading to tumor recurrence, is set into motion after the first week of IR due to additional hypoxia induction and an influx of bone marrow−derived cells into irradiated tumors. This process does not involve active recruitment of vascular growth by localized secreted tumor angiogenic factors such as VEGF.^5,26,31 Mobilization of tumor-associated macrophages involves the upregulation of the CXCR4 receptor, which, when binding SDF-1, switches on a proangiogenic program leading to increased vascularization of tumors.^39,40 This receptor, and the corresponding downstream pathways that are activated, can be blocked by several mechanisms, such as preventing HIF-1α activation or blocking the CXCR4 receptor with drugs such as plerixafor.^25,40,41 Blocking these pathways led to an enhancement of IR effects in several preclinical tumor models.^24,39,40 Recently a CDK4/6 inhibitor (Abemaciclib) was shown to inhibit IR-induced vasculogenesis in NSCLC xenograft by significantly inhibiting HIF-1 and SDF-1 activation.¹⁷ Given that HIF-1α is a client protein of HSP90,⁴¹ we reasoned that AT13387 may inhibit IR-induced vasculogenesis. Our data clearly show that AT13387 inhibited key proteins in the vasculogenesis pathway (HIF-1α, SDF-1) and decreased bone marrow−derived cell (CD45, CD11b+) influx into tumors, which contributes to IR-induced vasculogenesis (Fig. 5B–E) and results in a concomitant enhancement in tumor regression (Fig. 5A).

Despite the promise of HSP90 inhibitors as a class of therapeutic agents, they have shown no encouraging single-agent activity. Our study therefore provides a strong case for AT13387 combination studies with IR because it will not only affect enhanced tumor regression but also may reduce the challenge of tumor recurrence after radiation therapy owing to its involvement in inhibition of IR-induced vasculogenesis. Our study provides well-characterized surrogate biomarkers such as HIF-1α and SDF-1 expression as predictive biomarkers of vasculogenesis inhibition and correlates it with positive responses in patients.

Conclusions

The findings of the present study indicated that AT13387 is a promising IR modifier for HNSCC and NSCLC in vitro while exhibiting little IR sensitization in normal cells. The AT13387 and IR combination would be a reasonable way to manage HNSCC and NSCLC patients in early and advanced disease settings with respect to local tumor control, with far-reaching implications on minimizing IR-induced tumor recurrence after radiation therapy. Taken together, our findings suggest that this novel chemoradiation combination should be considered for human clinical trials, not only with conventional IR fractionation but also perhaps for stereotactic body radiation therapy.