Heat shock factor-1 knockout induces multidrug resistance gene, MDR1b, and enhances P-glycoprotein (ABCB1)-based drug extrusion in the heart

Abstract

Heat-shock factor 1 (HSF-1), a transcription factor for heat-shock proteins (HSPs), is known to interfere with the transcriptional activity of many oncogenic factors. In the present work, we have discovered that HSF-1 ablation induced the multidrug resistance gene, MDR1b, in the heart and increased the expression of P-glycoprotein (P-gp, ABCB1), an ATP binding cassette that is usually associated with multidrug-resistant cancer cells. The increase in P-gp enhanced the extrusion of doxorubicin (Dox) to alleviate Dox-induced heart failure and reduce mortality in mice. Dox-induced left ventricular (LV) dysfunction was significantly reduced in HSF-1^−/− mice. DNA-binding activity of NF-κB was higher in HSF-1^−/− mice. IκB, the NF-κB inhibitor, was depleted due to enhanced IκB kinase (IKK)-α activity. In parallel, MDR1b gene expression and a large increase in P-gp and lowering Dox loading were observed in HSF-1^−/− mouse hearts. Moreover, application of the P-gp antagonist, verapamil, increased Dox loading in HSF-1^−/− cardiomyocytes, deteriorated cardiac function in HSF-1^−/− mice, and decreased survival. MDR1 promoter activity was higher in HSF-1^−/− cardiomyocytes, whereas a mutant MDR1 promoter with heat-shock element (HSE) mutation showed increased activity only in HSF-1^+/+ cardiomyocytes. However, deletion of HSE and NF-κB binding sites diminished luminescence in both HSF-1^+/+ and HSF-1^−/− cardiomyocytes, suggesting that HSF-1 inhibits MDR1 activity in the heart. Thus, because high levels of HSF-1 are attributed to poor prognosis of cancer, systemic down-regulation of HSF-1 before chemotherapy is a potential therapeutic approach to ameliorate the chemotherapy-induced cardiotoxicity and enhance cancer prognosis.

Anthracyclines such as doxorubicin (Dox) and its derivatives are used as antineoplastics, either alone or in combination with other drugs, to treat many types of cancer. Although Dox is an effective chemotherapeutic drug, it has long been known to cause severe cardiotoxicity, leading to dilated cardiomyopathy and congestive heart failure (1). Despite recent improvements of the drug, clinical manifestation of cardiac dysfunction among Dox-treated cancer patients still persists as a serious side effect. Other classes of chemotherapeutics, including tyrosine kinase inhibitors (TKIs) such as imatinib mesylate and humanized antibody-based therapeutics such as trastuzumab have also been found to cause significant cardiotoxicity (2, 3). In the case of Dox, the cardiotoxicity has been considered to be primarily due to oxidative-stress–induced death of cardiomyocytes and subsequent irreversible myocardial remodeling. However, many clinical trials testing the effectiveness of antioxidants indicated no significant cardioprotection during Dox treatment (4). Thus, alternate strategies, based on other identified pathways, are warranted to overcome this deleterious effect. Recent studies have identified several pathways of Dox-induced myocardial remodeling, including the involvement of heat-shock factor 1 (HSF-1) (5–11). Here we report a unique finding that HSF-1 knockout provided cardioprotection from Dox by inducing the multidrug-resistant gene 1 (MDR1) and expression of P-glycoprotein (P-gp), an ATP-binding cassette (ABCB1) transporter, which is usually associated with multidrug-resistant cancer cells, and we found that it actively extrudes Dox from cardiomyocytes in HSF-1 knockout mouse hearts.

HSF-1 is an immediate responder of any intrinsic or extrinsic stress, and it enhances expression of heat-shock proteins (HSPs). This response is generally considered an act of stress tolerance, so that the cells can recover from the exerted stress and survive (12). Many studies have found that preinduction of Hsp25 (a member of the small heat-shock protein family that is linked to cell apoptosis), by either HSF-1 activation before inflammation or by overexpression methods, is cytoprotective, whereas Hsp25 induction postinflammation is cytotoxic (13). This “heat-shock paradox” is poorly understood at present, especially in relation to Dox-induced heart failure (13). Understanding the role of HSF-1 in the heart is further complicated by the fact that HSF-1 can indirectly regulate the transcriptional activity of another transcription factor, nuclear factor-κB (NF-κB) (14–16). The DNA-binding sites for HSF-1 and NF-κB show close homology; as such, they can mutually influence DNA binding. NF-κB has been shown to promote expression of heat-shock genes (14), but not in response to heat stress (17).

P-gp, a product of MDR1, protects multidrug-resistant cancer cells by extruding the antineoplastic agents. The MDR1 promoter has been shown to have NF-κB binding, and multiple binding sites for heat-shock transcription factors, termed heat-shock elements (HSEs). Because HSF-1 is activated by Dox in the heart (11), here we tested a unique hypothesis that HSF-1 binding to HSEs in the MDR1 promoter would repress MDR1 gene expression by antagonizing the NF-κB binding site. The dogma for HSF-1 knockout is that it should enhance Dox-induced cell death in the heart due to the expected decrease in defense against Dox-induced oxidative stress. However, our results show an atypical pathway wherein the ablation of HSF-1 indeed provided cardioprotection, by increasing NF-κB transactivation and enhanced MDR1 gene expression to establish P-gp–based drug extrusion in the heart.

Results

MDR1b Gene Expression and Dox Extrusion in HSF-1^−/− Mouse Hearts.

Fig. 1A shows the Western blots of P-gp in HSF-1 wild-type and knockout mice (Fig. S1). P-gp level (both 150 kDa and 170 kDa, corresponding to fully glycosylated and lesser glycosylated isoforms) (18) was very high in HSF-1^−/− mouse hearts. RT-PCR assessment of MDR1a and MDR1b (isoforms known to be in murine muscle tissues) mRNA showed higher MDR1b (the inducible isoform) expression in HSF-1^−/− hearts (Fig. 1B), suggesting that ablation of HSF-1 induces a multidrug-resistant phenotype in the heart. Coimmunofluorescence staining of P-gp and α-actin (cardiomyocyte specific) in HSF-1^+/+ and HSF-1^−/− mouse heart tissues (merged images in Fig. 1C) confirmed higher P-gp expression in cardiomyocytes of HSF-1^−/− mice.

Fig. 1.

MDR1/P-gp expression in HSF-1^−/− hearts. (A) Western blots of P-gp in heart tissue lysates of HSF-1^+/+ and HSF-1^−/− mice. (B) RT-PCR assessment of MDR1a and MDR1b mRNAs. (C) Multimodal fluorescence of α-actin and P-gp in Dox-treated HSF-1^+/+ and HSF-1^−/− mouse heart tissues (green, Alexa Fluor 488, α-actin; red, Alexa Fluor 555, P-gp; blue, DAPI, nuclei).

Transcriptional regulation of MDR1 gene/P-gp expression by HSF-1 was studied using a luciferase reporter gene assay. In the MDR1 promoter, a HSE is found to be upstream of a NF-κB binding site, between −174 and −153 (Fig. 2A). The proximal promoter region of the human MDR1 gene (Fig. S2) was cloned and used as the promoter for the luciferase gene in a lentiviral vector (Fig. 2A). Adult cardiomyocytes from HSF-1^+/+ and HSF-1^−/− mice were transduced with pLenti-MDR1pro-Luci virus and the luciferase activity was determined. The luminescence was close to fivefold higher in HSF-1^−/− cardiomyocytes, relative to wild type (Fig. 2B). When the HSE (GAACTTTC) in the MDR1 promoter was mutated (GACCATAC) (Fig. 2A), the luminescence increased in HSF-1^+/+ cells, but there was no change in the HSF-1^−/− cardiomyocytes (Fig. 2B). Thus, HSF-1 binding in the HSE inhibits MDR1 gene expression. Moreover, deletion of both HSE and NF-κB binding motifs in the MDR1 promoter (Fig. 2A) diminished the luminescence in HSF-1^−/− cardiomyocytes (Fig. 2B) and removed the increase observed by HSE mutation in the wild-type cells. Taken together, these data indicate that NF-κB binding drives MDR1 gene activation, whereas HSF-1 binding represses it.

Fig. 2.

MDR1 promoter activity in HSF-1^+/+ and HSF-1^−/− cardiomyocytes. (A) Schematics of luciferase lentiviral construct with human MDR1 promoter subcloned as replacement of PGK1 promoter in pLentiPGKV5Luci expression vector. WT-MDR1 promoter has overlapping HSF-1 and NF-κB binding sites (–174 to –153). Mut-MDR1 prom1 was generated by site-directed mutagenesis, to ablate the HSF-1 binding site, and Mut-MDR1 prom2 was generated by deleting both HSF-1 and NF-kB binding sites (Table S1). (B) Quantitative plots (n = 4) are the luminescence measured from HSF-1^+/+ and HSF-1^−/− cardiomyocytes transduced with the WT-MDR1 and Mut-MDR1 promoter containing luciferase viral vectors.

Experiments were carried out to determine whether silencing HSF-1 will induce P-gp in cancer cells. MCF-7, a human breast cancer cell line, was treated with HSF-1 siRNA. As seen in Fig. 3A, treatment with the siRNA significantly reduced HSF-1 expression in these cells, with almost complete ablation seen at 10 nM. However, there was no induction of P-gp at any concentration of siRNA used (Fig. 3).

Fig. 3.

HSF-1 silencing in human adenocarcinoma (breast cancer) cell line. (A) Western blots and quantitative estimation (n = 3) of HSF-1 in siRNA-treated MCF-7 cells. MCF-7 cells were treated with various concentrations of HSF-1 siRNA (21-mer sequence: sense strand, 5′-CCCUGCAGGUUGUUCAUAAtt-3′ and antisense strand, 5′-UUAUGAACAACCUGCAGGGtc-3′) for 24 h and lysed for Western blotting. (B) Western blots of P-gp in control and siRNA-treated MCF-7 cells. For positive control, MCF-7/Dox (Dox-resistant MCF-7 cells) lysate, which is known to have abundant P-gp, was used.

In Vitro P-gp Pump Action and Dox Extrusion in HSF-1^−/− Mouse Cardiomyocytes.

To demonstrate the function of P-gp as a Dox efflux pump in the heart, the Dox loading level in isolated adult mouse cardiomyocytes from HSF-1^+/+ and HSF-1^−/− mice was determined using Dox autofluorescence (18, 19). Isolated adult cardiomyocytes from HSF-1^+/+ and HSF-1^−/− mice were treated with 0.1 μM of Dox in vitro and intracellular red fluorescence due to Dox retention was determined. Fig. 4A shows the phase contrast image of an isolated myocyte, with the Dox fluorescence image (fluorescence intensity largely seen from the nuclei) superimposed. Fluorescence quantitation showed significantly reduced intensity in HSF-1^−/− cardiomyocytes (Fig. 4B). Preincubation of HSF-1^−/− mouse cardiomyocytes with 1 μM verapamil, a well-known inhibitor of P-gp activity, before treating with 0.1 μM of Dox increased Dox loading (Fig. 4B), indicating that P-gp is functionally very active in HSF-1^−/− cardiomyocytes. Survival analysis of HSF-1^+/+ and HSF-1^−/− mice against Dox was carried out. As seen in Fig. 4C, the 50% survival for HSF-1^+/+ mice was 19 ± 3 d, which is consistent with previously reported values at comparable experimental conditions (11). The HSF-1^−/− group showed improved survival compared with the HSF-1^+/+ group (34 ± 3 d) (Fig. 4D). HSF-1^−/− mice, treated with verapamil before Dox injection, showed decreased survival (50% survival 21 ± 2 d, Fig. 4D). These results demonstrate that P-gp–based drug efflux is active in HSF-1^−/− mouse hearts and that it provides increased survival.

Fig. 4.

Activity of P-gp and Dox extrusion in HSF-1^−/− cardiomyocytes. (A) Microscopic images (phase contrast, Dox autofluorescence, and merged) of HSF-1^+/+ cardiomyocytes treated with 0.1 μM DOX. (B) Quantitative analysis of Dox fluorescence from HSF-1^+/+ and HSF-1^−/− cardiomyocytes (n = 10) that were treated with Dox (0.1 μM) or Dox + verapamil (1 μM). (C and D) Survival curves of Dox (6 mg/kg for 4 wk), verapamil only (25 mg/kg for 4 wk), or Dox + verapamil (25 mg/kg, injected 2 h before Dox, 6 mg/kg for 4 wk, injection) treated HSF-1^+/+ and HSF-1^−/− mice. Arrows indicate the days of (0, 7, 14, and 21 d) 4 doses of Dox.

Attenuation of Dox-Induced Heart Failure and Reduced Systemic Toxicity of HSF-1^−/− Mice.

Fig. 5A shows representative MRI images of the midmyocardial section (sagittal axis) of Dox-treated mouse hearts (8 wk after the first dose) of HSF-1^+/+ and HSF-1^−/− at three time points of a cardiac cycle: end diastolic (ED), end systolic (ES), and again at ED (as designated in ECG, Fig. 5A). In the Dox-treated HSF-1^+/+ group, the left ventricular (LV) wall thickness at ED was 0.58 ± 0.2 mm, which was significantly smaller than the 0.73 ± 0.18 mm determined for HSF-1^−/− groups (Fig. 5B). Similarly, during ES, the LV wall thickness was 0.62 ± 0.2 mm for HSF-1^+/+ groups, whereas it was 0.97 ± 0.52 mm for HSF-1^−/− groups. The stroke volumes, computed from the bright blood pool area in the short axis images (as marked on the images in Fig. 5A) from eight equally spaced (1 mm) cross-sectional slices in the sagittal axis of the whole heart (as marked in Fig. 5C, long axis view), are summarized in Fig. 5D. End diastolic volume (EDV) for HSF-1^+/+ was 21.8 ± 0.6 mm³, which reduced to only 15.83 ± 0.58 mm³ end systolic volume (ESV) during ES, whereas the EDV and ESV for HSF-1^−/− were 19.2 ± 0.63 mm³ and 8.2 ± 0.8 mm³. These results showed that the cardiac output was less than 30% in Dox-treated HSF-1^+/+ mice, whereas in Dox-treated HSF-1^−/− mice, the cardiac output was 60%, a value comparable to controls. Independent echocardiographic evaluations of cardiac function also confirmed these results (Fig. S3). Pretreating animals with verapamil (25 mg/kg, 1 h before Dox injection) removed the cardioprotection from Dox-induced cardiac dysfunction among HSF-1^−/− mice (Fig. 5 B and D), as there was a significantly increased ESV (11.9 ± 0.6 mm³) and reduced LV wall thickness (0.81 ± 0.07 mm), compared with Dox-alone treated groups. Echocardiography also confirmed that verapamil pretreatment potentiated cardiac dysfunction in HSF-1^−/− mice (% fractional shortening, FS 48.3 ± 4.3% for Dox-alone treated group, compared with 29.3 ± 3.1% for Dox + verapamil treated group) (Fig. S4).

Fig. 5.

Cardiac MRI of failing hearts and quantitative comparison of myocardial remodeling in Dox-treated HSF-1^+/+ and HSF-1^−/− mice. (A) Short axis cardiac MRI images at midmyocardium (with slice thickness 1 mm) obtained from Dox-treated and verapamil (25 mg/kg) + Dox-treated HSF-1^+/+ and HSF-1^−/− mice (4 wk after last dose of Dox) at ED, ES, and ED. ECG describes the actual time points of the images presented. Bright blood pool and LV wall are distinguishable as marked in the images. (B) Quantitative analysis of LV wall thickness during ED, ES, and ED of a cardiac cycle in these groups at 4 wk after last dose. (C) Long axis view of the cardiac MRI showing slice selection in short axis for the stroke volume determination. (D) Stroke volumes in Dox-alone–treated and verapamil + Dox-treated HSF-1^−/− mice (n = 3).

Increased Transactivation of NF-κB in Dox-Treated HSF-1^−/− Mice.

To determine whether HSF-1 ablation enhances NF-κB expression and activation upon treatment with Dox, differences in protein level and DNA-binding activity were determined between the HSF-1^+/+ and HSF-1^−/− mouse hearts. The constitutive expression of NF-κB, (p65 Western blotting), was found to be more than threefold higher (Fig. 6A) in HSF-1^−/− mouse hearts compared with HSF-1^+/+. Both HSF-1^+/+ and HSF-1^−/− mice showed increases in p65 upon treatment with Dox (Fig. 6A). The ser-276 phospho-p65 was also found to be higher in Dox-treated HSF-1^−/− mouse hearts (Fig. 6A). IκB-α and ser-32 phospho-IκB-α levels were more than fourfold higher in HSF-1^−/− groups (Fig. 6A). Upon treating with Dox, the IκB-α level drastically decreased in HSF-1^−/− groups. There was also a similar decrease in IκB-α in HSF-1^−/− groups, which showed a very low basal level of IκB in control (Fig. 6A). Similarly ser-32 phospho IκB-α was also higher in HSF-1^−/− hearts, but upon treatment with Dox, it was completely abolished (Fig. 6A). Because phosphorylation and depletion of IκB is initiated by the IκB kinase (IKK) complexes, IKK levels were determined. The IKK-α level was higher (more than fivefold) in the HSF-1^−/− group compared with the HSF-1^+/+ group (Fig. 6A). Interestingly, the IKK-α level was very high in HSF-1^−/− mouse hearts relative to the wild type, and it was enhanced in the HSF-1^−/− hearts by Dox treatment (Fig. 6A), whereas there was no change in IKK-β. Nuclear protein extracts were used to demonstrate a Dox-induced enhancement of NF-κB translocation into the nucleus. Fig. 6B shows the Western blot of p65 in the nuclear extracts for control and Dox-treated mouse hearts. As seen, there is no significant p65 band in the nuclear extracts of Dox-treated HSF-1^+/+ mouse hearts. However, in HSF-1^−/− mice, Dox treatment greatly increased nuclear p65 (Fig. 6B). These results indicate that ablation of HSF-1 favors NF-κB activation/translocation into the nucleus upon treated with Dox. Taken together these data indicate that HSF-1 primes the NF-κB pathway for activation during Dox treatment.

Fig. 6.

Transcriptional activation of NF-κB in Dox-treated HSF-1^−/− mice. (A) Western blots of NF-κB (p65), inhibitor IκB, IKK-α, IKK-β (n = 3). (B) Western blots of p65 in the nuclear extracts of control and Dox-treated HSF-1^+/+ and HSF-1^−/− mice nuclear extracts (n = 3).

NF-κB DNA binding was studied using EMSA and the results obtained are shown in Fig. 7A. In HSF-1^−/− groups, a significant level of NF-κB binding was observed in the absence of Dox, and this binding increased with Dox treatment, consistent with the decreased IκB found in the cytosolic fraction and with the observed increase in p65 in the nuclear fraction (Fig. 6 A and B). These data demonstrate that NF-κB transcriptional activity is increased in HSF-1^−/− groups upon treated with Dox. Preincubation of the nuclear extracts with p65 monoclonal antibody before addition of oligonucleotide completely abolished the band (Fig. 7B) due to competitive inhibition, and electrophoretic mobility supershift assays (EMSSAs) demonstrated a shift only with the p65 antibody, confirming that the observed band in EMSA is due to p65.

Fig. 7.

NF-κB binding activity measurements (EMSA and EMSSA) in HSF-1^−/− mouse hearts. (A) EMSA analysis of NF-κB binding. (B) Competitive inhibition and supershift assay (EMSSA). Preincubation of the nuclear extracts with p65 antibody diminished the band intensity, whereas preincubation of oligonucleotide + nuclear extract of Dox-treated HSF-1^−/− mice with p65 antibody shifted the band up (supershift). (C and D) Quantitative plots of IL-6 and TNF-α (measured 4 wk after last dose of Dox) in plasma of vehicle or Dox-treated mice (n = 3).

Because higher NF-κB binding was observed in HSF-1^−/− mice, NF-κB–related cytokines, namely IL-6 and TNF-α, were determined to be in blood plasma, which were obtained 8 wk after the first dose of Dox. The basal IL-6 was higher in HSF-1^−/− mice relative to HSF-1^+/+ (Fig. 7C). Conversely, the basal TNF-α level was significantly lower in these mice compared with wild type (Fig. 7D). Upon treating with Dox, IL-6 increased about three times in both HSF-1^+/+ and HSF-1^−/−, relative to respective controls, but the IL-6 was significantly higher in the HSF-1^−/− setting. These data indicate that NF-κB activity is constitutively higher in HSF-1^−/−, supporting our hypothesis that HSF-1 knockout will enhances NF-κB activity, and ultimately the MDR1 expression.

Discussion

Overexpression of small heat-shock proteins, such as Hsp20 and Hsp27, two gene targets for HSF-1, has been shown to be cardioprotective against Dox-induced cardiotoxicity (13, 20). However, other studies have found that the inhibition of HSF-1 is cardioprotective, highlighting the heat-shock paradox (13, 14). As such, there must be different mechanisms of cardioprotection in the Hsp27 overexpressed hearts and in the HSF-1 knockout heart. The present work has demonstrated that HSF-1 knockout is protective against Dox-induced cardiotoxicity by enhancing NF-κB activity and inducing MDR1 gene expression in the heart. The P-gp transporters resulting from the MDR1 gene activation extrude Dox from cardiomyocytes to limit Dox-induced cardiomyocyte death in the heart. Indeed, human MDR overexpression in mice has been shown previously to protect against Dox (21), and the present work demonstrates that the stress-responsive endogenous transcription factors HSF-1 and NF-κB regulate MDR1 expression and thus P-gp function in the heart. Therefore, the present results indicate that systemic down-regulation of HSF-1, before chemotheraphy, will be beneficial in terms of both cardioprotection as well as better prognosis of cancer, because high levels of HSF-1 have been attributed to poor prognosis in human breast cancer (22).

Our data reveal that HSF-1 represses the NF-κB driven expression of P-gp. However, previous research has indicated that absence of HSF-1 enhanced NF-κB activity (14, 16). Indeed, HSF-1 is generally thought to be an activator of gene expression; nevertheless, its repression of a few other genes has also been observed (23). Moreover, activation of HSF-1 has been shown to repress NF-κB by enhancing Hsp25 expression and interaction with IKK-β (24). Wirth et al. reported that HSF-1–deficient mice were highly susceptible to cadmium-induced lung injury, due to higher NF-κB activity and a higher proinflammatory response (16). Similarly, endotoxin and heat shock were found to inhibit NF-κB in RAW264.7 cells (15). These studies did not address whether any of the HSPs that are transcriptionally regulated by HSF-1 are directly involved. Other studies, however, have found that Hsp25 regulates some of the protein kinases that regulate NF-κB activation. For example, Dodd et al. observed that IKK-β, a component of the enzyme complex IKK-α/IKK-β/IKK-γ (NEMO) that phosphorylates IκB and facilitates its degradation, is inhibited by Hsp25 and hence the canonical NF-κB activation is retarded (25). Hsp25 can increase the IKK-α activity by Hsp25-assisted Akt phosphorylation of IKK (26). The present study has yielded three major insights on the role of NF-κB and Dox-induced cardiotoxicity: (i) deletion of HSF-1 enhanced the IKK, IκB, and NF-κB expression in the heart; (ii) upon treating with Dox, IKK-α is increased and IκB is very much reduced (Fig. 6A), unlike the wild-type mouse hearts where the expression is very low and there is not much change upon treating with Dox; and (iii) NF-κB activates MDR1 gene expression and establishes P-gp pump action in the HSF-1^−/− mouse hearts, that efflux Dox efficiently (Figs. 1 and 2). Thus, HSF-1 activation and accumulation of Hsp25 (Hsp27 in human), can potentially inhibit NF-κB–dependent survival pathways in any stressed hearts, and our results, in addition to demonstrating the HSF-1/NF-κB–driven expression of P-gp, have provided evidence that activation of HSF-1 by Dox-induced stress contributes to the retention of NF-κB in the latent (repressive) mode, preventing survival signaling by NF-κB in the heart.

We and others have previously established that Hsp25 stabilizes/transactivates p53 to express proapoptotic proteins, and hence p53 inhibitors could lead to cardioprotection against Dox-induced toxicity (11, 27, 28). p53 transcriptional activity has also been found to be modulated by the HSF-1 activation and Hsp25 overexpression in Dox-treated mice (11), and pharmacological inhibition of p53 by pfithrin-α (PFT-α) was found to be cardioprotective against Dox-induced toxicity (27). However, PFT-α is a nonspecific p53 inhibitor and it has also been shown to inhibit HSF-1 (29). Thus, whether the cardioprotection by PFT-α is due to p53 inhibition or because of HSF-1 inhibition is not clear. It is also not currently known how p53 signaling is lowered in HSF-1^−/− mice, where NF-κB is favored. However, there are multiple possibilities for mutual turnoff of p53 and NF-κB (30). Although they are involved in opposing activities, p53 for cell cycle arrest and apoptosis, and NF-κB for prosurvival pathways, they do have opportunities for cross-talks between their pathways (30). Nevertheless, regardless of the involvement of these cell-fate determining pathways, our data clearly demonstrate that the HSF-1–dependent expression of P-gp is critical in the observed cardioprotection in the HSF-1^−/− setting.

The interplay between HSF-1 and NF-κB in the pathogenesis of human disease has been found to be important and that they mutually influence transcriptional activities (14, 15). Although HSF-1 activation was found to protect cells (31) it has also been found to cause disease. For example, recently HSF-1 was found to positively regulate cancer growth, and its ablation reduced the carcinogenesis from RAS mutation and hot spot mutation in tumor suppressor p53 (32). Likewise, accumulation of Hsp25 (an HSF-1 target gene) has been found to interact with and chaperone many protein kinases that are critical to phosphorylate the transiently induced and highly ubiquitous prosurvival proteins including Akt and IKK (26). However, its accumulation in cells, especially in cardiomyocytes is reported to be toxic and reported to cause conditions such as reductive stress (33). The role of HSF-1 in MDR1 induction is not completely defined. A previous study has reported that HSF-1 mutant (deletion of 202–316 residues, which is close to the range that was deleted to create the HSF-1^−/− mice, used in the present study) increased MDR1 activity in HeLa cells (34), and another study reported that mutHSF-1, which is lacking the HSPs gene induction capability, was able to constitutively induce MDR1 (35). However, neither of these studies addressed the possibility of the effect of NF-κB. Moreover, in our recent study we found that HSF-1 activity was completely absent in Dox-resistant human adenocarcinoma cells (MCF-7/Dox), where significant MDR1 was present (19). Thus, like NF-κB, which shows positive or negative regulation of MDR1, HSF-1 regulation of MDR1 expression is also dependent upon cell type and can be regulated by complex cross-interactions involving other transcription factors.

In summary, the present study has provided a number of insights on the role of HSF-1 in Dox-induced heart failure (Fig. S5–S8) that are useful for developing new preventive therapy based on HSF-1 as the target. We demonstrate a unique mechanism that ablation of HSF-1 enhances NF-κB and induces the MDR1 gene, establishing P-gp–based drug efflux in the heart. This cardioprotection by deletion of HSF-1 is by an atypical mechanism. The present study demonstrates the finding that HSF-1 represses NF-κB activation of the MDR1 gene in the heart. Although there are other possible mechanisms for the observed Dox-induced heart failure, the present work uniquely establishes that the HSF-1/NF-κB–dependent expression of the MDR1 gene is an integral part of the observed cardiac dysfunction and remodeling. Overall, our present results suggest that either gene targeting, or even more clinically relevant pharmacological inhibition of HSF-1 (36), is a promising approach that can augment cancer chemotherapy and alleviate chemotherapy-induced heart failure, enhancing cancer prognosis.

Materials and Methods

Animals and Drug Treatment.

Eight- to 12-wk-old HSF-1 wild type (HSF-1^+/+, BALB/c; Charles River) and knockout (HSF-1^−/−) mice, a kind gift from I. J. Benjamin (University of Utah, Salt Lake City, UT), were crossed and genotyped (Fig. S1). Age matched HSF-1^+/+ and HSF-1^−/− mice were treated with doxorubicin hydrochloride (6 mg/kg) once a week for 4 wk by i.p. injection and the control group was treated with vehicle (saline). All of the protocols were approved by the Ohio State University Animal Care Committee.

Cardiac Function Analysis.

A 11.7-T vertical bore magnet with the proton NMR frequency at about 500 MHz was used for MRI. T1 weighted anatomical cardiac images were obtained using 3D fast low angle shot (FLASH) gradient sequence.

pLenti-MDR1pro-Luci Viral Vectors for Luciferase Reporter Gene Assay.

For luciferease lentivirus reporter gene assay, the PGK1 promoter in pLentiPGKV5Luci vector (Addgene; 21471) was replaced with human MDR1 proximal promoter (–198 to +43, Fig. S2), obtained from PCR amplification using the primers in Table S1. Mutations were carried out using a site-directed mutagenesis kit (QuikChange II XL; Agilent Technologies). Adult cardiomyocytes isolated from HSF-1^+/+ and HSF-1^−/− mice were transduced with pLenti-MDR1pro-Luci virus and the luminescence was determined, after appropriate background correction of empty vector transduction.

Dox Fluorescence in Cardiomyocytes.

Adult cardiomyocytes from HSF-1^+/+ and HSF-1^−/− were treated with 0.1 μM Dox or 0.1 μM Dox + 1 μM verapamil and fixed using 4% (wt/vol) paraformaldehyde and used for fluorescence measurements (excited at 546 nm and emission detected at 565 nm).

Statistical Analysis.

Statistical analysis was performed using Student’s t test and one-way ANOVA. Data are presented as means ± SE. The general acceptance level of significance was P < 0.05. Kaplan–Meier curves were analyzed using a log-rank test to determine the significance of the difference in survival.