Hyperthermia enhances 17-DMAG efficacy in hepatocellular carcinoma cells with aggravated DNA damage and impaired G2/M transition

Due to the lack of effective treatment, hepatocellular carcinoma (HCC) is one of the malignancies with low survival rates worldwide. Combination of hyperthermia and chemotherapy has shown promising results in several abdominal tumours, but high expression of HSP90 in tumours attenuated the efficacy of hyperthermia. Thus a combination of hyperthermia and inhibition of HSP90 might be a feasible therapeutic strategy for HCC. One hepatic cell line (L02) and two HCC cell lines (Huh7 and HepG2) were heated at 42 °C for 0, 0.5 or 4 h with or without 100 nM 17-dimethylaminoethylamino-17-demethoxygeldanamycin (17-DMAG). HCC cells of the combination group exhibited more G2/M arrest and higher apoptotic rates which might result from suffering from more reactive oxygen species and serious DNA damage. Heat shock/17-DMAG co-treatment of HCC cells also destabilized CDK1, Cyclin B1 and CDC25C with a concomitant decreased proportion of cells in the M phase. Furthermore, co-treatment impaired the interaction of HSP90α with CDC37 and with CDK1, accompanied with decreased soluble CDK1. Combination of 17-DMAG with a 1.5-h whole body hyperthermia treatment attenuated tumour growth in xenograft mice models. These results suggest hyperthermia sensitize HCC to 17-DMAG, and combination of hyperthermia with 17-DMAG might be a potential therapeutic strategy for HCC.


17-DMAG/hyperthermia co-treatment inhibits HCC growth in vivo with decreased CDK1, Cyclin B1 level.
To explore the effect of co-treatment of hyperthermia and 17-DMAG in vivo, HCC cell lines Huh7 cells were injected into nude mice. Xenograft tumour growth curves are shown in Fig. 1a and images of tumours are shown in Fig. S1. In the combination group, the mean tumour volume was the smallest from the 2 nd week until the end. At the end of the treatment, cell cycle regulation protein CDK1 and Cyclin B1 were detected by Western Blot (Fig. 1b). Expressions of CDK1 and Cyclin B1 in the combination group were significantly lower than the control group (Fig. 1c). The result showed hyperthermia sensitized HCC to 17-DMAG in vivo, and which might correlate with cell cycle regulation and the changing of the CDK1-Cyclin B1 complex.

Heat shock increased ROS levels and aggravated 17-DMAG induced double strand breaks.
In cells treated with heat shock for 0, 0.5, and 4 h in absence and presence of 17-DMAG, reactive oxygen species (ROS) were determined by flow cytometry immediately after treatment (time point I, Fig. 2a). Both HCC cells, Huh7 and HepG2, had a much higher basal level of ROS than L02 cells (Fig. 4a). In the absence of 17-DMAG a 4-h heat shock induced an increase in ROS production in all 3 cell lines (Fig. 4b). In the presence of 17-DMAG, the ROS production was increased already in the absence of heat shock (HepG2), after 0.5-h heat shock (Huh7) and after a 4-h heat shock (L02, Huh7 and HepG2), respectively. Since ROS induces DNA damage, we quantified proteins associated with the DNA damage and repair response by western blot 24 h after heat shock treatment with or without 17-DMAG (time point II, Fig. 2a). DNA double-strand breaks (DSBs) activate DNA-activated protein kinase (DNA-PK), which is shown to phosphorylate T5 and T7 residues of HSP90α 18,19 . Therefore, we also detected p-HSP90α (T5/T7) as an indicator of DNA damage. In all three cell lines a synergistic effect of 17-DMAG and heat shock on γ -H2AX and p-HSP90α (T5/T7) indicating more severe DNA damage through the combination treatment. The DNA damage response-induced G2/M arrest activates p53 through ATM/ATR, which further transactivates p21 20 . In turn, p21 interferes with the activation of CDK1 and thus prevents the activation of the CDK1-Cyclin B1 complex 21 . Therefore, we checked p53 and p21 in the 3 cell lines. Although p53 did not change considerably, p21 in HepG2 increased after co-treatment. Huh7 is a p53 mutant and p21 defective cell line 22 . The p21 status and consequently different CDK1-Cyclin B1 complex activation might contribute to the different response to DNA damage observed in Huh7 and HepG2 cells.
Heat shock enhanced 17-DMAG induced G2 arrest with decreased Cyclin B1 levels in HCC. We previously found that 17-DMAG alone induced Cyclin B1 accumulation in HCC cells 23 . We next studied how co-treatment of heat shock and 17-DMAG would influence cell cycle regulator proteins. A 4-h heat shock increased the proportion of cells in G2/M phase in L02 cells but no obvious synergistic effect of heat shock with 17-DMAG was observed (Fig. 3b). In contrast, Huh7 and HepG2 cells were much more sensitive to the HSP90 inhibitor 17-DMAG, as evidenced by the higher G2/M proportions. Furthermore, compared to the treatment with 100 nM 17-DMAG alone, more HepG2 cells arrested in G2/M after co-treatment with 0.5-h heat shock   (Fig. 3b). To analyse the effects of heat shock and 17-DMAG on the cell cycle in more detail we detected key cell cycle regulator proteins for cell transiting through G2/M phase 24 , such as Wee1, CDC25C, Cyclin B1 and CDK1, by Western blotting. In all three cell lines Wee1, CDC25C and CDK1 decreased after 4-h heat shock and 17-DMAG co-treatment. Interestingly, Cyclin B1 nearly disappeared in Huh7 and HepG2 after the 4-h heat shock in the presence of 17-DMAG, but increased clearly in L02 cells (Fig. 5a). Cyclin B1 mRNA showed the same trends as the changes of protein (Fig. 5b). The decrease of Cyclin B1 was further confirmed by immunofluorescence staining in Huh7 and HepG2 cells (Fig. 5d). Cell phases were determined by nuclei morphology under confocal laser scanning microscope as previous report 25 . Compared to the normal cell cycle in control Huh7 and HepG2 cells (Fig. 5c), the co-treatment of 4-h heat shock with 24-h 17-DMAG remarkably decreased Cyclin B1, and cells in M phase were hardly seen (Fig. 5d). Combined with the results of flow cytometry, these data suggest co-treatment impairs the G2/M transition and arrests HCC cells in early G2 phase with lower cyclin B1 level. Directly after heat shock CDK1 was found in the insoluble pellets in absence and presence of 17-DMAG. If cells were allowed to recover for 20 h after heat shock, only after combined treatment with heat shock and 17-DMAG CDK1 was still detected in the insoluble fraction (Fig. 6a, the right panel). As many kinases are HSP90 clients and docking to HSP90 needs the assistance of CDC37, we further investigated if the accumulation of insoluble CDK1 is associated with the dynamics of the HSP90α -CDC37 interaction. Huh7 cells were heated for 4 h at 42 °C with or without 100 nM 17-DMAG and subsequently allowed to recover at 37 °C for 20 h. CDK1 and CDC37 were co-immunoprecipitated with HSP90α (Fig. 6b). Though CDC37 did not change clearly after heat shock and 17-DMAG treatment (Fig. 6b), more CDC37 and CDK1 bound to HSP90α after the recovery from heat shock. 17-DMAG attenuated the interaction of CDC37 and HSP90α , but did not affect the interaction of HSP90α and CDK1. A combination of heat shock and 17-DMAG resulted in less CDC37 and little CDK1 binding to HSP90α (Fig. 6b). These results showed that CDK1 tends to aggregate during heat stress, and needs to be stabilized by HSP90α . When cells were heated, CDC37 disassociated from HSP90α after CDK1 binding to HSP90α . In the absence of 17-DMAG, CDK1 could be refolded to a mature soluble state during the recovery phase. In contrast, in the presence of 17-DMAG, CDK1 could not bind to HSP90α , thus CDK1 could not progress on its folding and maturation pathway and partitioned into the insoluble fraction. Combination of heat shock and 17-DMAG resulted in a decrease of soluble CDK1 due to its instability and the loss of HSP90 chaperone function.

Discussion
As HSP90 has been considered as an anti-cancer target for more than 20 years 15 , a series of HSP90 inhibitors have been discovered, including the first generation inhibitor Geldanamycin and its derivatives (17-DMAG, 17-AAG, etc.), and the second generation inhibitors STA-9090, NVP-AUY922 26 . However, activities of their metabolites are different. Geldanamycin derivatives are metabolized by CYP3A4, both the primary compounds and the main metabolites inhibit CYP3A4, thereby slowing down the degradation of Geldanamycin derivatives in the body 27 . In contrast, STA-9090 and its related compound NVP-AUY922 are excreted quickly in cells with high expression  17-DMAG primarily inhibits HSP90 by binding to the ATP binding site in the N-terminal domain of HSP90 16 , but due to benzoquinone reduction and GSH conjugation in its metabolism 30 , it also causes ROS production. Moreover, heat shock also induces ROS generation and DSBs directly, though the underlying mechanism has not been fully understood 31 . Huh7 and HepG2 exhibited an increased level of intrinsic ROS stress as compared to L02 (Fig. 4a). Using this biochemical feature to develop a novel therapeutic strategies might preferentially kill cancer cells through ROS-mediated mechanisms 32 . Furthermore, HSP90 in tumour cells exhibits a higher affinity to HSP90 inhibitors 33 that could explain why HCC cells are more sensitive to HSP90 inhibitors especially in combination with heat shock treatment.
In the p53 mutant human liver carcinoma cell line Huh7 the 17-DMAG/heat-shock combination treatment did not reduce cell viability and did not increase the apoptosis rate and G2/M arrest as prominently as in the case of the p53 wild type human liver carcinoma cell line HepG2. Since p53 mutant cells like Huh7 have always higher expression levels of antioxidants genes than p53 wild-type cells 34 , the ROS generation, induced by 17-DMAG/ heat shock co-treatment, was more efficiently counteracted by the cellular redox buffering systems in these cells as compared to the p53 wild-type cells HepG2 (Fig. 4b). Thus less DNA double-strand damage was produced as indicated by reduced γ H2AX (Fig. 4c). The p53 mutant phenotype in Huh7 cells is also related to p21 deficiency, and p21 prevents the activation of the CDK1-Cyclin B1 complex, explaining the lower proportion of G2/M arrest and decreased apoptosis rates in Huh7 cells after the co-treatment.
ROS and DSBs instantly activate the DNA damage response signalling pathway, which involves several HSP90 clients, in particular kinases and transcription factors, such as p53, DNA-PK, Chk1, Wee1, CDC25C, CDK1 and Cyclin B1 (http://www.picard.ch/downloads/HSP90interactors.pdf). On the other hand, client proteins could also be damaged by ROS-inducing conditions like heat shock or by ROS directly, compromising DNA damage repair; for example, CDK1 forms insoluble aggregates after heat shock 35 . Maturation and stability of most kinases need the interaction of HSP90 and its cochaperons (HSP70, HOP, CDC37, p23, etc.) 36 ; in particular, CDC37 is necessary for recognition of kinase clients by HSP90 16 . Although CDC37 transiently stabilizes kinase clients, interaction with HSP90 is essential to prevent client aggregation and degradation 37 . Thus, inhibition of HSP90 promised to attenuate kinases that depend on the function of HSP90, even if CDC37 were unaltered by the treatment. Heat causes misfolding, aggregation and inactivation of many cellular proteins. Misfolded and aggregated client proteins need to be disaggregated and refolded by the Hsp70/Hsp110 system 38,39 and depend on HSP90 for maturation, or they are degraded rapidly 40 . Since most cancer cells have defective G1 checkpoints, they depend on S and G2 checkpoints for survival from DNA damage 41 . In addition, G2/M transition requires the activation of the Cyclin B1-CDK1 complex. Co-treatment of heat shock and 17-DMAG induced more DNA damage, attenuated the interaction of HSP90 and CDC37 and inhibited HSP90 function which led to increased aggregation and thus decreased soluble CDK1. In addition, co-treatment of heat shock and 17-DMAG significantly lowered Cyclin B1 mRNA and protein. Hence, less CDK1-Cyclin B1 complex could be formed in cancer cells, leading to G2/M arrest and apoptosis, especially in HepG2 cells with less activated CDK1-Cyclin B1 complex. Similar cell viability results were obtained when heat shock was combined with STA-9090, a structurally unrelated HSP90 inhibitor (Fig. S3).
Due to inefficient blood flow through the newly formed immature blood vessels within solid tumours, including HCC, tumour cores usually experience hypoxic conditions, which induces HIF-1α , a trancription factor known to promote chemotherapy resistance by several different mechanism 42,43 . Hyperthermia also increases the blood flow in solid tumours 6 and thereby may increase delivery of 17-DMAG to the poorly vascularized tumour cores. In combination with the increased ROS levels generated by hyperthermia, this might be the reason for the more significant decrease of Cyclin B1 and CDK1 in tumours co-treated with hyperthermia and 17-DMAG.
Hyperthermia can be practiced as whole body hyperthermia and local hyperthermia, and both of them show a well tolerance up to 6 h 44 . Though the earlier clinical trial results of hyperthermia alone presented diverse sensitivity in various tumours 45 , nowadays there appears to be a renewed interest, thanks to several randomized studies demonstrating that the improvements in treatment outcome by combined hyperthermia therapy 46 . More and more evidence supports thermal therapy increases the effectiveness of radiotherapy or chemotherapy treatment. Till now, 601 clinical trials, which are combined with different hyperthermia treatments, have been carried on www. clinicaltrial.gov. By itself, thermal therapy can also kill cancer cells 47,48 . Prostate cancer xenograft tumour sizes were about 60% of control group on day 30 after mice were heated at 42-43 °C for 10 min using gold nanorod 49 . Comparing with local hyperthermia, whole body thermal therapy is a systemic treatment, and whole-body fever-range thermal therapy can safely treat cancer cells wherever they are throughout the entire body 50,51 . The mechanism is still unclear but more evidence presents that the phenomena is not only related to the thermal stress the cancer cells encountered 47,48 , but also related to the regulation of whole body immunity 52 . Whole-body hyperthermia combined with HIPEC for the advanced gastric cancer also achieved an encouraging result, which showed 1-year survival rate was 38.5% compared to the chemotherapy alone group rate of 19% 53 . In this study, a 1.5-h whole body hyperthermia with or without 17-DMAG for 5 weeks exhibited a 70.3% and 37.9% tumour growth inhibition, respectively. This results suggest hyperthermia and 17-DMAG might be also feasible to HCC. By using HIPEC with 17-DMAG after cytoreductive surgery 53 , 17-DMAG-conjugated metal nanoparticle to heat tumours 6 , and 17-DMAG intravenous injection combined with heated by regional hyperthermia applicator 44 are the methods of clinical application of the combination therapy against HCC.
Taken together, our study demonstrated that heat shock and 17-DMAG synergized in anti-proliferative treatment of HCC, both in vitro and in vivo, by inducing more ROS and DNA damage. In addition, 17-DMAG led to a decrease of mitosis promoting factor-related HSP90 client proteins, like CDK1, Cyclin B1, CDC25C, and triggered G2/M arrest and apoptosis. Notwithstanding, several open questions still remain. What is the reason for the decrease of Cyclin B1 mRNA during the co-treatment; decreased transcription or enhanced degradation? Is a whole body hyperthermia feasible when using this strategy in patients? Or could the tumour tissue be heated locally, for example by application of specific tumour-targeted gold nanoparticles? Nevertheless, our approach of combining hyperthermia with HSP90 inhibitors as a therapeutic strategy for treating HCC achieved satisfying results both in vitro and in vivo. Whether such a combination therapy can achieve similar therapeutic effects in patients remains to be explored in clinical trials.

Materials and Methods
In vivo experience. 5-week old male BALB/c nude mice were obtained from Southern Medical University Laboratory Animal Centre. All procedures in this animal study protocol were approved by the Institutional Animal Care and Use Committee of Southern Medical University and carried out according to the principles of the NIH Guide for the Care and Use of Laboratory Animals. 5 × 10 6 Huh7 cells in 200 μ L DMEM were injected subcutaneously into the flanks of the mice. After 10 days, the mice were divided into four groups to accept treatment. Mice in the 17-DMAG group and the combination group were injected intraperitoneally with 17-DMAG (dissolved in saline) at a dose of 25 mg/kg three times a week, while mice in the control group and the hyperthermia group were injected intraperitoneally with saline. 2 h latter, mice in the hyperthermia group and the combination group were transferred into a 42 °C thermostatic cabinet and kept there for 1.5 h. The weights of the mice and the longest and shortest diameters of the xenograft tumours were measured before each treatment. After 5 weeks, all mice were sacrificed. Xenograft tumours were isolated. The tumour volume was calculated according to the following formula: longest diameter × (shortest diameter) 2 × 0.5. Cell culture. Hepatic cell line L02 54 , hepatocellular carcinoma (HCC) cell lines Huh7 and HepG2 55 , hepatic cell line QGY-7703 and HCC cell line QSG-7701 which were isolated from the same hepatocellular carcinoma patient 56 , were purchased from the Type Culture Collection of the Chinese Academy of Sciences, Shanghai, China. All cells were cultured in Dulbecco's Modified Eagle Medium (Gibco) containing 10% fetal bovine serum (Gibco), 100 μ g/mL streptomycin sodium and 100 U/mL penicillin G sodium (MP Biomedicals) at 37 °C and 5% CO 2 .
Immunofluorescence. Cells were fixed using 4% paraformaldehyde for 10 min at room temperature and then washed 3 times with PBS. Cells were permeabilized by incubation in methanol at − 20 °C for 15 min. Cells were blocked with 3% BSA, and incubated with mouse anti-Cyclin B1 antibodies (Santa Cruz Biotechnology, sc-245) at 4 °C overnight. Subsequently, cells were incubated with Alexa Fluor 555 donkey anti-mouse IgG (Life technologies) at room temperature for 2 h. Finally, the nuclei were stained with DAPI. Images were captured on an Olympus FV1000 Confocal Laser Scanning microscope.
Co-Immunoprecipitation. Cells were lysed as described above. The supernatant was rotated end-over-end for 3 h at 4 °C with rat anti-HSP90α (Assay Designs, ADI-SPA-840) or non-specific rat IgG followed by additional 3 h rotation at 4 °C with protein G agarose beads (Santa Cruz Biotechnology, sc-2002). Beads were washed 3 times with fresh lysis buffer, and boiled with 2 × SDS-PAGE sample buffer. Statistical analysis. Data were analysed by one-way ANOVA or factorial design ANOVA followed least-significant difference post-hoc test for homogeneity of variance or Welch test followed Games-Howell post-hoc test for heterogeneity of variance for normally distributed data, and Kruskal-Wallis test and Mann-Whitney U test for skewed data. All analyses were performed in SPSS 19.0 (IBM). A P < 0.05 was considered statistically significant.