Acquisition of Temozolomide Chemoresistance in Gliomas Leads to Remodeling of Mitochondrial Electron Transport Chain*

Temozolomide (TMZ) is an oral alkylating agent used for the treatment of high-grade gliomas. Acquired chemoresistance is a severe limitation to this therapy with more than 90% of recurrent gliomas showing no response to a second cycle of chemotherapy. Efforts to better understand the underlying mechanisms of acquired chemoresistance to TMZ and potential strategies to overcome chemoresistance are, therefore, critically needed. TMZ methylates nuclear DNA and induces cell death; however, the impact on mitochondria DNA (mtDNA) and mitochondrial bioenergetics is not known. Herein, we tested the hypothesis that TMZ-mediated alterations in mtDNA and respiratory function contribute to TMZ-dependent acquired chemoresistance. Using an in vitro model of TMZ-mediated acquired chemoresistance, we report 1) a decrease in mtDNA copy number and the presence of large heteroplasmic mtDNA deletions in TMZ-resistant glioma cells, 2) remodeling of the entire electron transport chain with significant decreases of complexes I and V and increases of complexes II/III and IV, and 3) pharmacologic and genetic manipulation of cytochrome c oxidase, which restores sensitivity to TMZ-dependent apoptosis in resistant glioma cells. Importantly, human primary and recurrent pairs of glioblastoma multiforme (GBM) biopsies as well as primary and TMZ-resistant GBM xenograft lines exhibit similar remodeling of the ETC. Overall these results suggest that TMZ-dependent acquired chemoresistance may be due to a mitochondrial adaptive response to TMZ genotoxic stress with a major contribution from cytochrome c oxidase. Thus, abrogation of this adaptive response may reverse chemoresistance and restore sensitivity to TMZ, providing a strategy for improved therapeutic outcomes in GBM patients.

Temozolomide (TMZ) is an oral alkylating agent used for the treatment of high-grade gliomas. Acquired chemoresistance is a severe limitation to this therapy with more than 90% of recurrent gliomas showing no response to a second cycle of chemotherapy. Efforts to better understand the underlying mechanisms of acquired chemoresistance to TMZ and potential strategies to overcome chemoresistance are, therefore, critically needed. TMZ methylates nuclear DNA and induces cell death; however, the impact on mitochondria DNA (mtDNA) and mitochondrial bioenergetics is not known. Herein, we tested the hypothesis that TMZ-mediated alterations in mtDNA and respiratory function contribute to TMZ-dependent acquired chemoresistance. Using an in vitro model of TMZ-mediated acquired chemoresistance, we report 1) a decrease in mtDNA copy number and the presence of large heteroplasmic mtDNA deletions in TMZ-resistant glioma cells, 2) remodeling of the entire electron transport chain with significant decreases of complexes I and V and increases of complexes II/III and IV, and 3) pharmacologic and genetic manipulation of cytochrome c oxidase, which restores sensitivity to TMZ-dependent apoptosis in resistant glioma cells. Importantly, human primary and recurrent pairs of glioblastoma multiforme (GBM) biopsies as well as primary and TMZ-resistant GBM xenograft lines exhibit similar remodeling of the ETC. Overall these results suggest that TMZ-dependent acquired chemoresistance may be due to a mitochondrial adaptive response to TMZ genotoxic stress with a major contribution from cytochrome c oxidase. Thus, abrogation of this adaptive response may reverse chemoresistance and restore sensitivity to TMZ, providing a strategy for improved therapeutic outcomes in GBM patients.
Chemotherapeutics play an important role in the combined treatment of gliomas, but ultimately chemotherapy generally fails to demonstrate a sustainable beneficial clinical outcome due to the acquired chemoresistance to drugs such as temozolomide (TMZ). 2 This compound is an alkylating agent that has shown significant initial benefit in the treatment of highgrade gliomas, especially when combined with radiotherapy. Although TMZ is commonly used in the adjunctive treatment of gliomas, eventually chemotherapy becomes impaired by development of chemoresistance. This phenomenon presents the most challenging barrier in the successful treatment of cancer and is the principal reason of chemotherapy failure (1,2).
TMZ is known to target nuclear DNA (nDNA) and generate several nuclear DNA adducts that block the cell cycle and lead to cell death through apoptosis. Mitochondria contain their own extra-nuclear DNA (mtDNA). This DNA is potentially a prime target for mutagens due to 1) the lack of protective histones associated with mtDNA (3,4) although mitochondrial nucleoids provide some protection (5), 2) the fact that mtDNA polymerase ␥ replicates the DNA with poor fidelity (6), 3) the presence of respiratory chain enzymes that can generate reactive oxygen species in close proximity to the mtDNA (7), and 4) the high lipid content of mitochondria, which makes the organelle particularly sensitivity to lipophilic xenobiotics and lipid peroxidation products that modify DNA (8). Consequently, the mutation rate of mtDNA is much higher when compared with nuclear DNA (4,9). Although mtDNA mutations have been suggested as factors in carcinogenesis, little is known about their role in the acquisition of chemoresistance in glioma. Recently, several lines of evidence obtained from mtDNA-depleted cancer cells (i.e. 0 cells) have revealed that mitochondrial DNA deletion results in resistance to apoptosis (10,11). Cisplatin cytotoxicity is significantly reduced in 0 head and neck squamous cancer cells (12) as is the cytotoxicity of doxorubicin and vincristine for 0 colon cancer cells (13).
In this study we tested whether TMZ targets mtDNA and modifies mitochondrial function using TMZ-resistant glioma cells and xenografts. We showed that TMZ reduced mtDNA copy number, increased mitochondrial heteroplasmy, and induced profound changes in the activities of the mitochondrial ETC and cellular bioenergetic function. Most strikingly, these changes are likely to be clinically relevant since they were recapitulated in patient biopsies after adjuvant therapy with TMZ.

EXPERIMENTAL PROCEDURES
Cell Culture-TMZ-sensitive U251 cells and their TMZ-resistant counterparts (UTMZ) were grown in DMEM F-12 medium plus L-glutamine supplemented with 7% heat-inactivated FBS, penicillin, and streptomycin. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO 2 . The resistant cell line was obtained by progressive adaptation of the parental sensitive cells (U251) to increasing concentrations of TMZ (see supplemental Fig. 1A).
Primary and TMZ-resistant GBM Xenograft Lines-The establishment and maintenance of the Mayo GBM xenograft panel has been described (14). TMZ resistance models were developed by subjecting mice with established flank tumors to successively higher doses of TMZ until tumor growth was no longer inhibited by 120 mg/kg/day TMZ for 5 days. The resulting TMZ-resistant lines were maintained by serial passage. The efficacy of TMZ (66 mg/kg/day X 5 days) in orthotopic xenografts was evaluated as described (15). All animal studies were approved by the Mayo Clinic Institutional Animal Care and Use Committee.
Preparation of Mitochondria-Mitochondria were prepared according to Higuchi and Linn (16). Cells were washed twice in PBS. The pellet was resuspended in MgRSB buffer (10 mM NaCl, 1.5 mM MgCl 2 , 10 mM Tris-HCl, pH 7.5), incubated at 4°C for 10 min, and then disrupted with a Dounce glass homogenizer. The homogenate was diluted with 1.3 volumes of MSB buffer (0.525 M mannitol, 175 mM sucrose, 12.5 mM EDTA, 12.5 mM Tris-HCl, pH 7.5) and centrifuged at 1000 ϫ g for 10 min to remove cell debris. The supernatant was further centrifuged at 20,000 ϫ g for 20 min. The mitochondrial pellets were then digested with DNase I (Sigma) at 37°C for 30 min to digest nuclear DNA. After digestion, the mitochondrial pellets were washed 3 times with MSB, deep-frozen in liquid nitrogen, and stored at Ϫ80°C.
Mitochondrial Complex Activities-Mitochondrial complex activities were determined as previously described (17,18). All activities were normalized to citrate synthase activity.
Western Blot Analysis-Protein expression levels of cytochrome c oxidase (CcO) subunits in mitochondria extracts from U251 and UTMZ cells were determined by Western blot analysis. Ten g of mitochondrial protein was loaded on 4 -20% SDS-polyacrylamide gels. Western blot were performed as we previously described (19,20). Antibodies against the catalytic subunits of CcO were from Mitosciences (Eugene, OR). All the antibodies against the nuclear-encoded subunits were from Proteintech Group Inc. (Chicago, IL) except the monoclonal antibody against COX-IV-1 (Abcam, Cambridge, MA).
Cellular Bioenergetic Analysis-XF24 Extracellular Flux Analyzer (Seahorse Biosciences) was used to determine the bioenergetic profile of intact cells as previously described (21)(22)(23). Cellular conditions were first optimized independently for both U251 and UTMZ cells. Cells were seeded at increasing densities (20,000 -60,000 cells/well) and allowed to recover for 24 h, and the oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were determined. A seeding density of 20,000 cells/well was selected for both cell lines and used in all other bioenergetics experiments, as it gives reproducible base-line ECAR and OCR values of 50 -75 mpH/min ECAR and 300 -500 pmol O 2 /min, respectively. The mitochondrial function assay was then used to assess bioenergetic function for both U251 and UTMZ cells (21). In this analysis sequential injections of oligomycin, FCCP, and antimycin A were added to the cells to define a basal OCR, ATP-linked OCR, proton leak, maximal respiratory capacity, reserve respiratory capacity, and non-mitochondrial oxygen consumption. To calculate these parameters we have assumed that the oligomycin-insensitive OCR is attributable to proton leak; however, oligomycin has been shown to hyperpolarize the mitochondrial membrane (24), and therefore, the resulting OCR is likely an upper estimate of the contribution from proton leak. All reagents were purchased from Sigma.
Semiquantitative PCR-Isolation and purification of genomic DNA from U251 and UTMZ cells was performed as previously described (25). mtDNA copy number was determined by semiquantitative PCR as previously described (26). Briefly, products amplified were the nuclear-encoded 18 S rRNA gene and a fragment of the catalytic subunit 1 of CcO. The primers were as follows: 18 S sense, AAGCTTGCGTT-GATTAAGTCC; 18 S antisense, TAATGATCCTTCCGCAG-GTTC; CcO-1 sense, GCCGACCGTTGACTATTCTC; CcO-1 antisense, GGGTTCTTCGAATGTGTGGT. The PCR products for 18 S rRNA and CcO-1 were then submitted to electrophoresis on 1.5% agarose gel and stained with EtBr. Arbitrary density was plotted against PCR cycles for the two amplification products. One cycle in each curve was selected for quantification. The relative quantification of mtDNA/nDNA ratio was determined by mtDNA CcO-1/nDNA 18 S rRNA densitometry.
shRNA Transfection-The plasmids carrying shRNAs specific for human COX-IV-1 were purchased from Open Biosystems (ATCC Integrated Molecular Analysis of Genomes and their Expression (IMAGE) catalog no. RHS4533). To generate stable cell lines, 5 g of TRCN0000046288, 0000046289, 0000046290, 0000046291, and 0000046292 or empty vector (pLKO.1) were transfected into UTMZ cells with the FuGENE 6 transfection reagent. Two days after transfection, cells were selected with 5 g/ml puromycin for 2 weeks. Clones were assayed for protein expression by immunoblotting with monoclonal antibodies against COX-IV-1 (Abcam, clone ab14744). Several positive clones were expanded, and clone 15 was chosen for further use in this study.
Statistics-Data were evaluated using GraphPad program. Statistical differences were determined using the Wilcoxon test. All reported p values are two-sided, and p values of less than 0.01 were considered to indicate statistical significance. Experiments were performed by triplicate and were performed twice or more to verify the results except for human tissues biopsies. Data are shown as the means Ϯ S.E.

RESULTS
To understand the mechanisms underlying TMZ chemoresistance, we generated and characterized a TMZ-chemoresistant stable glioma cell line for use as a model; we induced TMZ resistance in an established human glioma cell line, U251. We isolated TMZ-resistant clones via stepwise selection of U251 glioma cells as previously described (27). Resistant clones were isolated at an early stage of drug treatment (2.5 M TMZ) and then successively exposed them to incremental doses of TMZ (up to 1 mM) (supplemental Fig. 1A) to yield clones that were operationally resistant to doses of TMZ that would uniformly kill all U251 parental cells; these clones were designated as UTMZ. UTMZ cells were maintained in 160 M TMZ. For more than 50 passages, the resistance to TMZ was retained.
TMZ is known to induce cell cycle arrest and apoptosis (28,29). Accordingly, we carried out cell cycle analysis of U251 (TMZ-sensitive) and UTMZ (TMZ-resistant clone) cells by using DNA flow cytometric analysis. Cell-cycle distributions after 48 h of exposure to 500 M TMZ were analyzed. Acute exposure to TMZ decreased the number of cells in G 1 from 53 to 15% and increased cells in S phase from 34 to 58% and in G 2 /M from 12 to 27%. As predicted, TMZ did not affect cell cycle profile and distribution in UTMZ cells (supplemental Fig. 1B).
The effect of TMZ on the induction of apoptosis was examined using flow cytometry. Two-color flow cytometric analysis using phycoerythrin-annexin V and 7-aminoactinomycin D discriminated viable cells (live), early apoptotic cells (apoptotic), and both late apoptotic and necrotic cells (dead) (supplemental Fig. 1C). Compared with treatment with vehicle, U251 cells treated with 500 M TMZ exhibited a time-dependent increase in the percentage of apoptotic and non-viable cells. After 7 days of TMZ treatment, the percentage of apoptotic and non-viable cells increased to 37 and 50%, respectively. In contrast, neither apoptosis nor cell death was induced in UTMZ cells (supplemental Fig. 1D). The DNA repair protein O 6 -methylguanine-DNA methyltransferase (MGMT) plays an important role in cellular resistance to alkylating agents, including TMZ (30). Because more than 70% of gliomas express MGMT (31), we compared MGMT expression in U251 and UTMZ cells by Western blot. supplemental Fig. 1E shows that MGMT expression is down-regulated in UTMZ, indicating that MGMT status is not associated with response to TMZ in UTMZ cells.

Acquired TMZ Resistance Glioma Cells Carry Heteroplasmic Mitochondrial Deletions and Decrease mtDNA Copy Number-
The cytotoxicity of TMZ is mediated by the addition of methyl groups in genomic DNA (32)(33)(34). Among the lesions produced in DNA by TMZ, the most common is methylation at the N 7 position of guanine followed by methylation at the O 3 position of adenine and the O 6 position of guanine (35). Because mtDNA lacks the protective effects of histones and is less efficiently repaired compared with nuclear DNA (36,37), we hypothesized that mtDNA may be susceptible to damage by chronic TMZ treatment. To test this, we isolated total DNA from U251 and UTMZ cells and analyzed mtDNA integrity and mtDNA copy number. Quantitative PCR was performed to assess mtDNA copy number. Segments from both mtDNA and nDNA were amplified, and the relative mtDNA to nDNA ratio was assessed in UTMZ cells compared with that of U251 controls. MtDNA copy number in UTMZ cells was reduced by 75% compared with levels in U251 (p Ͼ 0.01) (Fig. 1A).
To compare mtDNA integrity, mitochondrial DNA from U251 and UTMZ cells (Fig. 1B, lanes 2 and 3) was amplified using primers oriented in opposite directions in the cytochrome b gene using long-extension PCR (3). The PCR products were resolved by agarose gel electrophoresis to determine whether deleted mtDNA species are present. Full-length mtDNA is indicated by the amplification of a 16.3-kb product, and mtDNAs harboring deletions are indicated by the presence of shorter fragments. Although it appears that U251 parental cells contain predominantly full-length mtDNA, deleted mtDNAs were the only species detected in this assay in UTMZ cells (Fig. 1B). Because UTMZ cells were cultured under the continuous presence of TMZ and to exclude the possibility that the failure to obtain full-length mtDNA amplification was due to PCR inhibition by TMZ, two controls were performed. mtDNA isolated from U251 cells exposed to acute treatment with TMZ (160 M, 18 h) contained predominantly full-length mtDNA after long-extension PCR, whereas PCR amplification of mtDNA from UTMZ cells cultured for more than 30 passages without TMZ displayed only species of deleted mtDNAs (data not shown).
Differential Changes in ETC Activities in TMZ-resistant Glioma Cells-We found that the profile of ETC activities was significantly different in the TMZ-resistant glioma cells (UTMZ) when compared with the parental, TMZ-sensitive glioma cell line (U251) (Fig. 2A). The activity of complex I was significantly lower in UTMZ cells (53.2 Ϯ 5.1 nmol/min/mg) than for U251 cells (154.4 Ϯ 8.2 nmol/min/mg) (p ϭ 0.0005). Similarly, complex V activity was decreased by 2-fold in UTMZ cells (472.6 Ϯ 0.6 versus 834.6 Ϯ 35 nmol/min/mg, p ϭ 0.0005). In contrast, complexes II-III activity was significantly increased   (38,39), it is possible that the reduced activities of Complexes I and V are due to direct inhibition by TMZ. To exclude this possibility, we measured mitochondrial complex activities from U251 cells exposed to acute treatment with TMZ (160 M for 24 h). There were no significant differences in the specific activities of the complexes after acute TMZ treatment related to control, untreated cells (supplemental Fig. 2). These results suggest that resistance to TMZ is associated with the remodeling of all ETC complexes, including the ATP synthase.
Switch in COX-IV Isoform in TMZ-resistant Glioma Cells-Because CcO activity influences the overall rates of mitochondrial respiration and electron transport and CcO activity is elevated in TMZ-resistant glioma cells, we carried out analyses of the expression of CcO subunits. Western blots of the subunits involved in the catalytic activity of CcO (I, II, and III) showed similar expression of COXI in both cells lines. However, using monoclonal antibodies, we were not able to detect COXII in UTMZ cells or COXIII in U251 cells (Fig. 2B). Because the catalytic subunits are essential for CcO activity, our data would be consistent with a selective loss of epitopes necessary for the recognition by the antibodies or that the epitopes are affected by posttranslational modifications. Analysis of the 10 nuclear-encoded subunits that are important for the regulation of CcO (40) indicated that the other major change between cell lines is the switch in COX-IV isoforms. Fig. 2B shows that COX-IV-1 is the main isoform detected in UTMZ cells (41). In contrast, U251 cells express mostly the COX-IV-2 isoform.

TMZ-resistant Glioma Cells Display Enhanced Bioenergetic Reserve
Capacity-Because the relative activities of the major respiratory chain complexes are altered in TMZ resistance, we next tested whether this has a significant impact on cellular bioenergetics. The bioenergetic profile of U251 and UTMZ cells was determined according to a previously developed mitochondrial function assay represented schematically in Fig. 3A (24). Briefly, after a base-line OCR was established, oligomycin, FCCP, and antimycin A were injected sequentially, and measurements of OCR were made after the addition of each specific inhibitor. Using this mitochondrial function assay, we have determined the basic bioenergetic profile of U251 parental and UTMZ cells (Fig. 3B). There was no  A, a schematic diagram demonstrating the use of specific inhibitors to determine the sites of cellular oxygen consumption is shown. After three base-line OCR measurements, oligomycin (Oligo), FCCP, and antimycin A (Ant A) were injected sequentially with OCR measurements recorded after each injection. ATPlinked oxygen consumption (ATP) and the OCR due to proton leak (Proton) can be calculated using the basal and the oligomycin-sensitive rate. Injection of FCCP is used to determine the maximal respiratory capacity. Injection of antimycin A allows for the measurement of oxygen consumption independent of Complex IV (Other). The reserve respiratory capacity (Reserve) is calculated by subtracting the basal from the maximal rate of oxygen consumption. B, bioenergetic profiles were measured using sequential injection of oligomycin (0. significant difference between the two cell lines in basal OCR, suggesting that each cell possesses sufficient mitochondrial capacity to satisfy basic the energy demands of the cell (Fig. 3C). In addition, there were no significant differences with respect to ATP-linked oxygen consumption or proton leak between the two cell lines (data not shown). The addition of the uncoupler FCCP induces electron transport unconstrained by the requirement to synthesize ATP. The resulting OCR is then an estimation of the maximal respiratory capacity that can be sustained with available cellular substrates and the respiratory chain. Interestingly, the reserve respiratory capacity was significantly increased in UTMZ cells when compared with the parental U251 cells (Fig. 3D). In addition to the OCR, the ECAR of the medium was also determined. As shown in Fig. 3E, ECAR decreased in UTMZ cells when compared with the parental U251 cells. These findings suggested that TMZ-resistant glioma cells are less glycolytic than the sensitive U251 cells.
Pharmacological Inhibition of Complexes II-III and IV Activities Abrogates TMZ Chemoresistance-Based on the results showing increased activities of complexes II-III and IV in chemoresistant UTMZ cells, we hypothesized that suppression of the activities of one or more of these two complexes should restore susceptibility to TMZ.
To examine whether enhanced CcO activity is linked to TMZ resistance to cell killing, we treated UTMZ cells with N-methyl mesoporphyrin IX (NMP), an inhibitor of ferrochelatase that decreases the activity of CcO by preferentially inhibiting heme a synthesis (42,43). Heme a is an essential component of CcO assembly and function with inhibition-blocking CcO activity. Inhibition of heme a was established by treating UTMZ with 5 M NMP for 3 days. NMP treatment of UTMZ cells decreased CcO activity by 98% (p Ͻ 0.0001) without any associated cell death (data not shown). Complex II-III activities were also decreased by 60%. As expected the activities of complexes I and V were unchanged (Fig. 4A).
It is important to note that NMP-treated UTMZ cells lost resistance to TMZ and progressively died after 6 days of exposure to TMZ. After 4 days of treatment with 300 M TMZ, the percentage of annexin V-positive cells was 34.7 Ϯ 2.4% for the NMP-treated UTMZ cells, but almost no apoptotic cells were found in UTMZ cells not treated with NMP (Fig. 4B). Apoptosis in NMP-treated UTMZ cells was accompanied by PARP cleavage (Fig. 4C). These data suggest that pharmacological alteration of ETC in UTMZ cells reversed chemoresistance.
Genetic Inhibition of COX-IV-1 Reduces CcO Activity and Abrogates TMZ Chemoresistance-We used shRNA stable transfections to test whether COX-IV-1 expression is responsible of the overall increase in CcO activity in UTMZ cells. Western blots revealed significantly decreased (60%) of COX-IV-1 expression in UTMZ cells transfected with expression vector encoding a short-hairpin RNA that blocks the expression of COX-IV-1 (Fig. 5, A and B). If the protection against the cytotoxic effect of TMZ is mediated by expression of COX-IV-1, then genetic inhibition of this subunit should decrease CcO activity and reduce tolerance to TMZ. Indeed, COX-IV-1 shRNA decreased COX activity by 65% (Fig. 5C) but also diminished cell viability under TMZ treatment. After 4 days of treatment with 300 M TMZ, the percentage of annexin V positive cells was 25.7 Ϯ 1.8% for the shRNA-transfected UTMZ cells, but almost no apoptotic cells were found in UTMZ cells transfected with the empty vector (pLKO.1) (Fig. 5D). Apoptosis in shRNA-UTMZ cells was accompanied by PARP cleavage (Fig.  5E). These data suggest that genetic inhibition of COX-IV-1 in UTMZ cells reversed chemoresistance.
Mitochondrial Function in Glioma Patients-To study whether the TMZ-dependent alterations in ETC complex activity occurs in GBM patients, tumor specimens were obtained from patients undergoing surgical resection for primary (newly diagnosed) or recurrent (post TMZ-radiation failure) at the University of Alabama Hospital, Birmingham, AL. Paired biopsies (primary and recurrent) were obtained from several patients before and after administration of conventional   DECEMBER 17, 2010 • VOLUME 285 • NUMBER 51 TMZ and radiation, generally followed by at least 1 cycle of maintenance TMZ.

Temozolomide Resistance and Mitochondrial Function
Altered Activities of ETC Complexes of Human GBM Specimens after Tumor Recurrence-Mitochondrial fractions for enzymatic activity analysis were extracted from all paired GBM samples. Reproducibility of the methodology was assessed by comparing two or more separate pieces of tumor from the same patients. Analysis of primary versus recurrent profiles in each individual patient demonstrated evidence for altered respiratory chain enzymes activities after chemotherapy. Scatter plots that include the activity of each ETC complex for all primaries and recurrent GBMs were analyzed. Comparison of complex activities in a group of eight paired samples demonstrated that the enzymatic activity of complex I was significantly lower in the recurrent tissues; complex II-IV activities increased significantly The enzymatic activity of complex V was significantly lower in the recurrent tissues (p ϭ 0.0005) (Fig. 6A). These results indicated that ETC remodeling similar to that observed with the TMZ-resistant cell lines occurs in patients after combined TMZ and radiotherapy.
TMZ Alters the Activities of ETC Complexes of Human GBM Specimens from Flank Xenografts-As the samples analyzed from recurrent GBM cases were from patients treated with a combination of radiotherapy and TMZ, we extended our studies to determine whether TMZ treatment alone triggered the changes in ETC complexes. Previously we described a panel of serially transplantable GBM xenograft lines established by direct subcutaneous injection of patient tumor tissue in the flanks of nude mice (14). Using two GBM xenograft lines, we developed TMZ resistance models by subjecting mice with established flank tumors (14) to successively higher doses of TMZ.
Tumor samples from the TMZ-sensitive xenograft lines (GBM12 and GBM22) and from the derived TMZ-resistant xenograft lines (GBM12-TMZ and GBM22-TMZ) were evaluated for ETC complex activities. Similar to the clinical samples (Fig. 6) and TMZ-resistant glioma cells (Fig. 2), TMZ-resistant xenograft lines were found to have higher complex II-IV activities and decreased activities of complex I and V (Fig. 6B). In both xenograft lines, the variation in complexes I, II-III, and IV was similar in magnitude. Complex I activity decreased 2-fold in TMZ-resistant xenograft lines, and complexes II-III and IV showed 2-3-fold increases in their activities. In both TMZresistant xenograft lines, Complex V activity decreased but to different extents. For GBM22 the activity decreased from 353 Ϯ 44 to 79 Ϯ 7 nmol/min/mg (4.4-fold); however, for GBM12, the decrease is more pronounced, from 219 Ϯ 5 to 7 Ϯ 5 nmol/ min/mg (29-fold). These results suggested that the changes in complex activities are triggered by TMZ.

DISCUSSION
TMZ has shown therapeutic benefit in the treatment of highgrade gliomas, particularly in newly diagnosed patients (44). However, drug resistance is still a major problem for many patients who are either unresponsive to TMZ or relapse and develop a resistance to treatment. Despite extensive research, the mechanisms leading to acquired TMZ chemoresistance in gliomas are not completely understood. Assessing chemoresistance mechanisms in vivo is challenging (45), so as a first approach we generated an in vitro model of acquired TMZ-dependent chemoresistance using human glioma cell lines. Because TMZ targets nuclear DNA, we reasoned that mtDNA may also be a target of TMZ. Our data show that the mtDNA of TMZ-resistant glioma cells have decreased copy number and mitochondrial heteroplasmy. However, we demonstrate that these changes are not associated with a decrease in mitochondrial function but, rather, remodeling of the ETC complex activities and an increase the bioenergetic reserve capacity.
In recent years a decrease in mtDNA copy number has been reported in many types of cancer, including ovarian, gastric, hepatocellular, breast, and kidney (46 -50), suggesting that decreased mtDNA copy number may contribute to tumorigenesis. Several studies have presented evidence that mtDNA plays an important role in cellular sensitivity to cancer therapies (36). We and others have suggested that the mitochondrion could be the integrator of many signals that have a potential effect on tumor-related growth (13,19,36,(51)(52)(53)(54). Little is known however about the effect of long term exposure of an alkylating agent on mtDNA copy number and heteroplasmy in cancer cells, One possible explanation for the decreased mtDNA copy number is that expression of DNA polymerase-␥, the polymerase responsible for the replication and repair of mtDNA, is substantially reduced in TMZ-resistant cells, impairing replication of the mitochondrial genome. In support of this concept, it has been established that DNA polymerase-␥ is susceptible to different genotoxic agents (55) and oxidative damage (56). Another interesting finding of this work was that large heteroplasmic deletions in the mitochondrial genome were induced by TMZ exposure. Next we examined the bioenergetic status of cancer cells with acquired TMZ-dependent resistance. To our knowledge this is the first report that looks at the mitochondrial bioenergetic status in acquired chemoresistance cancer models in vitro, in vivo, and in patient biopsies before and after treatment. All three models show the same pattern of remodeling of the ETC. We have shown that acquisition of resistance to TMZ is correlated with decreased activities of complexes I and V of the respiratory chain and increased activities of complexes II-III and IV. The finding of similar alterations in ETC complex activities between recurrent human GBM specimens in patients subjected to TMZ-radiotherapy emphasizes the clinical significance of the findings and the primary role of TMZ in mediating these effects.
The activities of complexes II-III and IV (CcO), the terminal complex of the electron transport chain, are significantly higher in all of our models of acquired chemoresistance. TMZ-resistant glioma cells through an adaptive remodeling of the ETC activities may have built a more efficient coupling and may have a greater ability to increase mitochondrial electron transport during conditions of bioenergetic demand. This property may confer a selective advantage during the progression of the tumor, in particular under nutrient and hypoxic conditions (21,43,(57)(58)(59)(60). Our data also indicate that FCCP-linked respiration relative to the basal levels (reserve capacity) is higher in TMZresistant glioma cells compared with TMZ-sensitive ones. This parameter is an estimate of the maximal respiration that can be sustained with available substrates and the electron transport chain. In previous studies we have shown that this reserve capacity is utilized by cells when they are placed under oxidative or metabolic stress (21,58).
The mechanisms that lead to increased reserve capacity are complex and can involve increased substrate availability but would also be consistent with increased activity of Complex IV and II, III in the TMZ-resistant cells. Indeed, it has been shown that neurons in the cerebral cortex have an increased reserve capacity linked to increased CcO activity. For example, when exposed to cyanide these neurons mount an exquisite adaptive mechanism to apoptosis. On the other hand, neurons in the corpus callosum have low reserve capacity and low CcO activity, which predisposes them to injuries and oxidative stress (61). The findings of a higher respiratory reserve capacity in TMZresistant cells also have relevance to the phenomenon of inhibition of an early release of cytochrome c into the cytosol, which initiates apoptosis (62,63). Considering that the cytochrome c/CcO molar ratio is the minimum needed for CcO activity (64 -66), an increase in CcO activity will likely decrease the pool of cytochrome c available for release from the mitochondrion to initiate apoptosis.
TMZ-resistant glioma cells are less glycolytic than the sensitive U251 cells. This decreased rate of extracellular acidification in UTMZ cells suggests a decreased requirement for glycolysis, whereas basal and ATP-linked oxygen consumption does not differ significantly from the wild-type U251 cells. Overall, these data point toward a "reversal of the Warburg effect." Currently the mechanisms responsible for this response are not known. However, one possible explanation may be that whereas mitochondrial dysfunction has long been reported and was hypoth-esized to contribute to cancer development, cancer cells that survive the genotoxic stress can engage their "remodeled" ETC. Moreover, it is predicted that decreased complex I activity combined with increased of complexes II-III and IV activities may increase ETC coupling, decrease mitochondrial proton leak, and decrease the generation of mitochondrial reactive oxygen species. This may be particularly relevant as reactive oxygen species are primarily generated at complex I of the ETC, with CcO not normally directly involved in reactive oxygen species generation. This mechanism could be a part of a previously unrecognized adaptive chemoresistance mechanism linking both oxidative stress and drug resistance in cancer leading to suppressed apoptotic signaling (67).
In this study we have found that elevated CcO activity in UTMZ cells is associated with a switch in the regulatory subunit COX-IV. Although parental U251 cells express COX-IV-2 isoform, UTMZ cells express COX-IV-1 isoform. It has been suggested that COX-IV subunit switching provides a mechanism to maintain the efficiency of respiration under conditions of reduced O 2 availability and may represent the initial adaptive response to hypoxia (41). Additionally, it has been shown that the COX-IV is a key regulatory subunit of the mammalian CcO. It inhibits the catalytic activity of the enzyme allosterically by binding of ATP to the matrix domain (68), suggesting that CcO is able to sense the cellular energy level, such as the ATP:ADP ratio (69,70). On the basis of our data, it is possible that expression of COX-IV-2 provides U251 cells with a mechanism to switch from oxidative phosphorylation to an anaerobic ATP production. On the other hand, expression of COX-IV-1 in UTMZ may permit the allosteric inhibition of CcO by ATP, thus, playing an important role in adjusting energy production to match cellular energy requirements in the cancer cell.
Our results of pharmacological and genetic inhibition of CcO activity are important as they show that both approaches reverse chemoresistance to TMZ. Partial suppression of COX-IV-1 expression in UTMZ cells with shRNA leads to a decrease in CcO activity and a parallel increase in the sensitivity of UTMZ cells to TMZ. These results do not, however, rule out the possibility that changes in other CcO subunits might have an effect on TMZ tolerance. Previous studies have demonstrated that NMP is a partially selective inhibitor of CcO because of the metabolic complexity of maturation of protoheme to heme a compared with that of other hemes; e.g. hemes b and c (42). Under conditions of heme deficiency (i.e. NMP treatment), less heme a is formed due to the low affinity of the heme a maturation pathway toward protoheme relative to the pathways that incorporate hemes b and c into their apoproteins (42). Although our data clearly indicate that NMP restored TMZ-induced apoptosis, we cannot rule out the possibility that NMP may exert this effect by targeting other heme-containing proteins such as catalase and complex III.
The results of this study suggest that there may be a close correlation between acquired chemoresistance, altered metabolism, and changes in cellular metabolic machinery at the level of the mitochondrion. Our data show a dramatic remodeling of ETC with cells switching from glycolytic to OXPHOS metabolism. Of particular interest is the significant increased in CcO in the TMZ-resistant models as well as in patient biopsies after treatment and at the time of recurrence of tumors. One possible explanation for this acquired chemoresistance may be linked in part to a change in the assembly and/or the composition of the mitochondrial and nuclear-encoded subunits of CcO. We propose that modulation of the regulation of bioenergetics in the cancer cell plays an essential role in tumor progression after failure of chemotherapy.