Heat Shock Protein 60 Regulation of the Mitochondrial Permeability Transition Pore in Tumor Cells

Disparate cell death stimuli initiate mitochondrial apoptosis (1) through a stepwise cascade that includes dissipation of the organelle inner membrane potential, swelling of the matrix, rupture of the outer membrane, and release of apoptogenic proteins in the cytosol, most notably cytochrome c (2). This process of mitochondrial permeability transition (3) is commonly subverted in tumors, resulting in enhanced cell viability, adaptation, and resistance to anticancer therapy (4). Although antiapoptotic Bcl-2 proteins oppose mitochondrial outer membrane permeability (5) and are often exploited in cancer, the existence of additional, “mitochondrial-intrinsic” pathways that counter apoptosis selectively in tumors in vivo has remained unclear.

Potential mediators of such mechanism(s) may include molecular chaperones of the heat shock protein (Hsp) family, including Hsp90, and its related molecule, tumor necrosis factor receptor-associated protein-1 (TRAP-1), which are abundantly expressed in mitochondria of tumor cells, but not most normal cells, in vivo (6). These mitochondrial Hsp90s associate with the matrix immunophilin, cyclophilin D (CypD; ref. 6), a proapoptotic component of the mitochondrial permeability transition pore (7, 8), and antagonize its function through protein (re)folding, thus preserving organelle integrity and shutting off the initiation of cell death (9).

In this study, we asked whether a broader, mitochondrial chaperone network is exploited to quench CypD activity selectively in tumors. We found that Hsp60 (10), a chaperone with critical roles in mitochondrial protein folding (11) and overexpressed in tumors (12), is a novel cytoprotective regulator of CypD, and that this pathway is required for tumor growth in vivo.

Human embryonic kidney HEK293T, normal human fibroblasts (NHF), breast adenocarcinoma MCF-7, colorectal adenocarcinoma HCT116, and glioblastoma U87 and LN229 cells were obtained from the American Type Culture Collection. CypD-directed short hairpin RNA (Open Biosystems) in pLKO.1 lentivirus vector was cotransfected (0.5 μg) in HEK293T cells with 0.33 μg pCMVdel8.2 and 0.2 μg pMDG (VSVG envelope glycoprotein) using Lipofectamine 2000 (Invitrogen). Supernatants were harvested after 48 hours, filtered through a 0.45-μm filter, and used to infect LN229 cells in the presence of 4 μg/mL of polybrene. Cells were selected in 2 μg/mL puromycin for 7 days. Alternatively, U87 cells were infected with pGIPZ lentiviral vector encoding control shRNA (RHS4349, Open Biosystems) or Hsp60-directed shRNA (RHS4531 clone ID V2LHS-191368, Open Biosystems), with selection of stable clones in the puromycin-containing medium.

Mitochondrial extracts from normal mouse liver or HCT116 cells were prepared as described (6). Lysates were fractionated by fast protein liquid chromatography (FPLC) on a Superdex 200 10/30 column (Amersham Biosciences) with elution at 4°C in 50 mmol/L Tris, 100 mmol/L NaCl, and 0.5% IGEPAL (pH 7.0; Sigma-Aldrich), at 0.4 mL/min, and collection of 0.5- or 1-mL fractions. The columns were calibrated with protein standards (GE Healthcare Biosciences), blue dextran (2,000 kDa), ferritin (440 kDa), aldolase (158 kDa), alcohol dehydrogenase (150 kDa), conalbumin (75 kDa), and ovalbumin (43 kDa).

Expression of recombinant proteins in BL21 Escherichia coli strain, pull-down assay, immunoprecipitation, or Western blotting were performed as previously described (6, 12). Gene silencing by siRNA was carried out as described (6, 12).

Mitochondrial membrane potential was quantified by JC-1 staining and flow cytometry (12). Caspase-dependent apoptosis was determined by DEVDase activity, propidium iodide staining, and multiparametric flow cytometry (6). Genetically engineered cultures (5 × 10⁴) were analyzed for cell proliferation or viability by direct cell counting or MTT (9).

All experiments involving animals were approved by an Institutional Animal Care and Use Committee. U87 cells expressing luciferase (U87-Luc) and stably transduced with scrambled shRNA or Hsp60-directed shRNA (five animals per group) were stereotactically implanted (1 × 10⁵) in the right cerebral striatum of immunocompromised nude mice (Charles River Laboratories). Tumor growth was quantified weekly by bioluminescence imaging using a Xenogen In Vivo Imaging System after i.p injection of 110 mg/kg d-luciferin. At the end of the experiment, tumor samples from the two animal groups were harvested and analyzed by immunohistochemistry as previously described (6).

Data were analyzed by two-sided unpaired t tests using a GraphPad software package (Prism 4.0) for Windows. Animal survival was quantified by a Kaplan-Meier curve. A P value of 0.05 was considered statistically significant.

We began this study by fractionating mitochondrial extracts of tumor cells (HCT116) or normal mouse liver to identify novel regulators of CypD (7, 8)–dependent permeability transition (2). Size-exclusion FPLC produced comparable elution profiles from the two mitochondrial sources (Fig. 1A) and similar protein yield (Fig. 1B). Partially overlapping mitochondrial fractions eluted from HCT116 extracts contained Hsp90, TRAP-1 (6), and Hsp60 (ref. 12; Fig. 1C). A pool of CypD coeluted in the same fractions containing Hsp60 and TRAP-1, and, to a lesser extent, Hsp90 (Fig. 1C). A second pool of CypD eluted in later fractions, potentially corresponding to free (i.e., uncomplexed) protein (Fig. 1C). As control, CypD-reactive fractions contained the adenine nucleotide translocator, which associates with CypD in a permeability transition pore (2), whereas a cytoplasmic marker, Ran, was negative (Fig. 1C). In contrast, FPLC fractions eluted from normal liver mitochondria contained low levels of Hsp90 and Hsp60 and barely detectable TRAP-1 (Fig. 1D), consistent with the low expression of these chaperones in normal mitochondria (6, 12). In addition, CypD did not coelute with any of the mitochondrial chaperones in normal liver extracts and emerged only as a free, uncomplexed protein (Fig. 1D).

In capture assays with HCT116 mitochondrial extracts, endogenous CypD bound recombinant Hsp90 (6), as well as Hsp60, whereas glutathione S-transferase (GST) was unreactive (Fig. 2A). Bacterially expressed Hsp60, but not GST, bound recombinant CypD and TRAP-1 (6) in the absence of cellular proteins, indicating that these interactions were direct (Fig. 2B). When analyzed in vivo, immune complexes precipitated with an antibody to Hsp60 from tumor mitochondria contained endogenous CypD, whereas nonbinding IgG was negative (Fig. 2C). Although Hsp60 (and TRAP-1) also associated with survivin (12) in vivo (Supplementary Fig. S1A) and in vitro (Supplementary Fig. S1B), mitochondrial survivin did not directly bind recombinant CypD in a pull-down assay (Supplementary Fig. S1B). Consistent with the assembly of an Hsp60-CypD complex selectively in tumor cells, recombinant Hsp60 did not bind CypD in NHF (Fig. 2D, left). Conversely, both normal and tumor human cell lines expressed comparable levels of CypD (Fig. 2D, right).

siRNA silencing of Hsp60 (Supplementary Fig. S2A) triggered mitochondrial permeability transition in tumor cells, with loss of membrane potential (Supplementary Fig. S2B), cytochrome c release (Fig. 3A), and caspase-dependent cell death (Fig. 3B; ref. 12). Conversely, inhibition of CypD peptidyl-prolyl cis,trans isomerase activity with cyclosporine A (6) reversed mitochondrial depolarization (Supplementary Fig. S2C), cytochrome c release (Fig. 3A), and apoptosis (Fig. 3B) induced by Hsp60 silencing. Similar results were independently obtained with siRNA knockdown of CypD, which rescued cytochrome c release (Supplementary Fig. S2A) and loss of mitochondrial membrane potential (Supplementary Fig. S2B) associated with Hsp60 depletion. Further, LN229 clones with stable shRNA knockdown of CypD (Fig. 3C) also exhibited reduced mitochondrial depolarization (Supplementary Fig. S3A) and cytochrome c release (Supplementary Fig. S3B) after Hsp60 targeting, compared with scrambled transfectants. Consistent with a role of CypD as target of Hsp60 cytoprotection, Hsp60 knockdown reduced the viability of LN229 cells transduced with nontargeting shRNA, but had no effect on cells with stable shRNA silencing of CypD (Fig. 3D).

To test the relevance of Hsp60 regulation of CypD for tumor cell survival in vivo, we next infected U87-Luc glioblastoma cells with lentivirus encoding scrambled shRNA or Hsp60-directed shRNA. Stable shRNA knockdown of Hsp60 (Fig. 4A, left) significantly inhibited U87 cell proliferation in the presence of serum (10%), compared with control cultures (Fig. 4B, right). Stereotactic injection of control U87-Luc cells in the right cerebral striatum of immunocompromised mice gave rise to exponentially growing intracranial glioblastomas (Fig. 4B, top), quantified by bioluminescence imaging (Fig. 4B, bottom), with death of all animals within 25 days (Fig. 4C). In contrast, stable silencing of Hsp60 in U87-Luc cells suppressed the growth of intracranial glioblastoma in vivo (Fig. 4B) and significantly prolonged survival of all animals in this group (Fig. 4C). Tumors in the Hsp60 shRNA-targeted group retained nearly complete loss of Hsp60 protein expression at the end of the experiment, compared with scrambled shRNA, by immunohistochemistry (Fig. 4D).

In summary, we identified Hsp60 as a novel regulator of CypD-mediated mitochondrial permeability transition (2) selectively in tumors (1). These findings point to a broad antiapoptotic function of Hsp60 in cancer, which also includes restraining p53 activity and promoting survivin stability (12). As a regulator of CypD in tumors, Hsp60 assembles in a multichaperone complex, comprising at least Hsp90 and TRAP-1 (6). This mitochondrial chaperone network is required for tumor maintenance, as silencing of Hsp60 (this study) or inhibition of Hsp90 or TRAP-1 activity (9) triggered CypD-dependent permeability transition, apoptosis, and suppression of tumor growth in vivo (9). In contrast, Hsp60 (12) or Hsp90 (9) targeting had no effect on normal cells, consistent with the low levels of these chaperones in mitochondria and their lack of physical interaction with CypD. Altogether, these data suggest that regulatory pathways of mitochondrial permeability transition may be qualitatively different in tumor versus normal mitochondria and thus provide unique molecular targets for novel cancer therapeutics (9).

No potential conflicts of interest were disclosed.

Grant Support: NIH grants CA140043, CA78810, HL54131, and CA118005 (D.C. Altieri), and Deutsche Forschungsgemeinschaft grant Si 1546/1-1 (M.D. Siegelin).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.