Fhit–Fdxr interaction in the mitochondria: modulation of reactive oxygen species generation and apoptosis in cancer cells

Fhit protein is lost in cancers of most, perhaps all, cancer types; when restored, it can induce apoptosis and suppress tumorigenicity, as shown in vitro and in mouse tumor models in vivo. Following protein cross-linking and proteomics analyses, we characterized a Fhit protein complex involved in triggering Fhit-mediated apoptosis. The complex includes the heat-shock chaperonin pair, HSP60/10, which is likely involved in importing Fhit into the mitochondria, where it interacts with ferredoxin reductase, responsible for transferring electrons from NADPH to cytochrome P450 via ferredoxin, in electron transport chain complex III. Overexpression of Fhit protein in Fhit-deficient cancer cells modulates the production of intracellular reactive oxygen species, causing increased ROS, following peroxide treatment, with subsequent increased apoptosis of lung cancer cells under oxidative stress conditions; conversely, Fhit-negative cells escape ROS overproduction and ROS-induced apoptosis, likely carrying oxidative damage. Thus, characterization of Fhit-interacting proteins has identified direct effectors of a Fhit-mediated apoptotic signal pathway that is lost in many cancers. This is of translational interest considering the very recent emphasis in a number of high-profile publications, concerning the role of oxidative phosphorylation in the treatment of human cancers, and especially cancer stem cells that rely upon oxidative phosphorylation for survival. Additionally, we have shown that cells from a Fhit-deficient lung cancer cell line, are sensitive to killing by exposure to atovaquone, thought to act as a selective oxidative phosphorylation inhibitor by targeting the CoQ10 dependence of the mitochondrial complex III, while the Fhit-expressing sister clone is resistant to this treatment.


Introduction
The~2-MB FHIT genomic locus straddles an active common fragile site at chromosome band 3p14.2 1,2 . Partially due to this genomic fragility, Fhit mRNA and/or protein expression is lost or reduced in large fractions of almost all types of human tumors, due to allelic loss, genomic rearrangement, promoter hypermethylation, or combinations thereof [3][4][5] . Fhit knockout mice show significantly increased susceptibility to cancer development 6,7 , and FHIT gene therapy prevents or reduces tumor burdens in carcinogen-exposed Fhit-deficient mice 8,9 . Fhit restoration by stable transfection in cancer cells has little effect on cell growth in vitro, unless cells are exposed to stress, including the stress of the nude mouse environment in vivo; 10 viral-mediated Fhit restoration, a process that simultaneously supplies stress and Fhit expression, suppresses tumorigenesis in vivo and triggers apoptosis of many types of malignant cells in vitro [11][12][13][14] , including lung cancer cells. In lung hyperplastic lesions, DNA damage checkpoint genes are already activated, in parallel with, or following, DNA alteration at FRA3B within FHIT; in the hyperplastic lesions, selection for mutations in checkpoint proteins can then lead to neoplastic progression 15,16 . Evidence of the loss of FHIT alleles occurs in normal-appearing bronchial epithelial cells of smokers, prior to pathologic changes or alterations in the expression of oncogenes or other suppressor genes [17][18][19] . Similarly, normal-appearing breast epithelial cells of mammalian ducts adjacent to cancer or even far from an invasive cancer, often show loss of Fhit protein 20 . Fhit expression is downregulated by exposure to DNAdamaging agents 21 and plays a role in response to such agents 22,23 , with Fhit-deficient cells escaping apoptosis and accumulating mutations. To identify proteins that interact with Fhit to affect downstream apoptotic pathways, we cross-linked proteins within cells, after induced or viral-mediated Fhit overexpression in lung cancer cells, or endogenous expression in colon cancer cells, and characterized proteins associated with Fhit, and pathways affected by them.

Isolation of a Fhit protein complex
A549 lung cancer-derived cells, expressing low-level endogenous Fhit, and susceptible to apoptosis on exogenous Fhit expression 11 , were infected with AdFHIT or AdFHIT-His6 24 and treated with dithiobis(succinimidyl propionate (DSP)), a cross-linker that crosses membranes and fixes proteins in complex in vivo. Cells were lysed and proteins isolated with nickel beads avid for the His6 epitope tag. Purified proteins were treated with dithiothreitol (DTT) to cleave DSP and dissociate the complex, and digested by trypsin; protein constituents were identified by liquid-chromatography tandem mass spectrometry (LC-MS/MS) ( Table 1 and Supplemental material) and six proteins were identified, all with mitochondrial localization: HSP60 and 10, ferredoxin reductase (Fdxr), malate dehydrogenase (Mdh), electron-transfer flavoprotein (Etfb), and mitochondrial aldehyde dehydrogenase 2 (Aldh2); HSP60 and HSP10 are also distributed in the cytosol 25 . Since HSP60/10 complex acts as a chaperonin, we thought that interaction with this heat-shock protein complex might chaperone Fhit to the mitochondria, where Fdxr is important in electron transport. We thus confirmed the interaction of Fhit protein with these three candidate interactors.

Fhit subcellular localization
Since candidate Fhit interactors are mitochondrial proteins, we determined if Fhit, which lacks a mitochondrial localization signal, localized in these organelles, as reported previously 26 . Fhit-negative H1299 lung cancer cells carrying an inducible FHIT cDNA (D1 cells; refs. 27 ) were treated with the inducer,  (Fig. 1b, right panel), while AdFHITtransduced cells showed sparse reactivity (Fig. 1b, left panel). To assess Fhit submitochondrial localization, the mitochondria were purified from H1299D1 (inducible Fhit expression) and HCT116 (endogenous Fhit) cells (Fig. 1c). Fhit protein was detected in both the cytosolic and mitochondrial fractions.

Fhit interacts with HSP60, HSP10, and Fdxr
Among candidate interactor proteins, we focused on HSP60 and HSP10 chaperonins, and on Fdxr, a mitochondrial respiratory chain protein known to be transcriptionally activated by p53 and involved in responses to therapeutic drugs 28 . To validate interactions, we performed GST pulldown with lysates from p53-null 29 H1299D1 cells transfected with HA-FDXR, and immunoprecipitation (IP) with lysates from PonA-treated, Fhitinduced H1299D1 cells and HCT116 cells that express endogenous Fhit. HSP60 and Fdxr were detected after pulldown with GST-Fhit (Fig. 2a). IP with anti-Fhit coprecipitated HSP10 and HSP60 in DSP-treated H1299D1/ Fhit and HCT116 cells (Fig. 2b). Fdxr-Fhit interaction was also confirmed in HCT116 cells using the proximal ligation assay Duolink. This method uses binding of individual primary antibodies to the putative interacting proteins. Under a microscope, we observed punctate fluorescent spots, corresponding to antigens in closer proximity, only when the secondary antibodies were concomitantly recognizing the anti-Fhit and Fdxr antibodies; this effect was not observed when only one antigen was targeted (Fig. 2c) 30 .

Fhit expression induces the generation of reactive oxygen species
Fdxr, a 54-kDa flavoprotein, is located on the matrix side of the inner mitochondrial membrane, and is responsible for transferring electrons from NADPH, via the single-electron shuttle ferredoxin-cytochrome P450, to substrates 31 . Under substrate-limiting conditions, electrons leak from this shuttling system and generate reactive oxygen species (ROS) 32 . Fdxr mediates p53dependent, 5-fluorouracil (FU)-induced apoptosis in colorectal cancer cells, through generation of ROS 33,34 , potent intracellular oxidants, and regulators of apoptosis 35 . The discovery of the Fhit-Fdxr interaction prompted us to determine if ROS production could be involved in Fhit-mediated apoptosis. Overexpression of Fdxr increases the sensitivity of tumor cells to apoptosis following H 2 O 2 treatment, through ROS production 33 (Fig. 3a).
A similar experiment was performed with p53-null, Fhit-negative lung cancer-derived H1299 D1 and E1 clones carrying PonA-inducible FHIT and empty vector expression plasmids, respectively; the cells were treated with 5 µM PonA and at 48 h treated with 0.5 and 1.0 mM H 2 O 2 ; the percentage of ROS-positive cells was higher in Fhit-positive D1 cells than in E1 control cells (20% vs. 3.5% at 0.5 mM H 2 O 2 , and 78% vs. 25% at 1.0 mM H 2 O 2 , respectively) (Fig. 3b).
To further examine the generation of ROS after Fhit reconstitution during oxidative stress, dichlorofluoresceindiacetate (DCFH-DA) was used to measure the redox state of Fhit-overexpressing cells. Peroxidases, cytochrome C, and Fe 2+ can oxidize DCFH-DA to fluorescent 2′-7′dichlorofluorescein (DCF) in the presence of H 2 O 2 ; 36 thus, DCF indicates H 2 O 2 levels and peroxidase activity. Increased DCF fluorescence was detected in D1 cells compared with E1 cells under stress conditions (Fig. 3d).
To determine if H 2 O 2 treatment with or without Fhit could affect cell viability or cell cycle kinetics, we performed flow cytometry (Fig. 3c); when Fhit was present Atovaquone, an FDA-approved antimalarial drug, mainly used for the treatment of pneumocystis pneumonia or toxoplasmosis, is a hydroxy-1,4-naphthoquinone analog of ubiquinone, known as Co-enzyme Q10 (CoQ10). It acts as a potent, selective oxidative phosphorylation (OXPHOS) inhibitor through targeting of CoQ10 dependence of the mitochondrial electron transport complex (ETC) III 37 . Fiorillo et al. 37 have recently shown that atovaquone has anticancer activity that is directed against cancer stem cells. Additionally, this study showed that atovaquone treatment of breast cancer cells inhibited oxygen consumption and induced aerobic glycolysis as well as oxidative stress. Since another very recent report 38 showed that an inhibitor of OXPHOS, IACS-010759, a small-molecule inhibitor of ETC complex I, inhibited proliferation and induced apoptosis of brain cancer and AML cancer cells, which are dependent on OXPHOS, we were encouraged to try treatment of our Fhit-positive and Fhit-negative lung cancer cells with atovaquone, the ETC III inhibitor, with the idea that Fhitnegative cells may be dependent on OXPHOS for energy metabolism and survival. Recent evidence has revealed that cells harboring an unstable genome can be sensitized with reagents that enhance ROS production to induce cell death 39 .
Thus, we used our H1299 sister clones, D1 and E1 (carrying an inducible Fhit plasmid and the inducible empty vector, respectively 27 ) to test this idea. The cells were first induced with 5 μM Pon A for 24 h to activate the inducible plasmid promoters, and 30 h later, the cells were treated with vehicle (DMSO) or vehicle plus atovaquone, as described 37 , and cell viability was assessed on day 3 following atovaquone treatment. As illustrated in Fig. 4, the atovaquone-treated Fhit-deficient E1 lung cancer cells showed significantly reduced viability, while the Fhit-positive cells were unaffected by atovaquone (see Fig. 1c for Fhit expression in these cells after Pon A treatment). These findings suggest that OXPHOS inhibitors could selectively target Fhit-negative tumor cells.

Discussion
Earlier searches for Fhit-interacting proteins pointed to several candidate proteins, none of which we could confirm as interactors (unpublished data) by co-IP experiments, including Ubc9, α-tubulin, and Mdm2 [40][41][42] . To readdress the question of Fhit protein interactors, we used adenovirus-transduced Fhit-His6 for Fhit complex purification after cross-linking, and Fhit-linked proteins, HSP60, HSP10, and Fdxr, were identified; the subcellular location of these proteins suggested that the mitochondria might be a focus of Fhit activity. HSP "stress proteins" as molecular chaperones perform functions, such as protein translocation, folding, and assembly 43 . The finding that Fhit interacts with HSP60/HSP10, suggests that the HSP complex may be important for Fhit stability, and possibly for its correct folding to import it into the mitochondria, prior to activation of an apoptotic pathway.
Targeted disruption of the FDXR gene in HCT116 colon cancer cells showed that it was essential for viability; reduction of the gene copy number resulted in decreased sensitivity to 5-FU-induced apoptosis 33 and FDXR is a target gene of the p53 family 34 . Overexpression of Fdxrsensitized colon cancer cells to H 2 O 2 , 5-FU, and doxorubicin-induced cell death, indicates that Fdxr contributes to p53-mediated apoptosis through generation of oxidative stress in the mitochondria. Thus, activated p53 induces apoptosis in response to cellular stresses in part through ROS, and simultaneously p53 increases transcription of the FDXR gene, which in turn enhances p53 function by increasing ROS-induced apoptosis 33,34 .
We have shown the presence of Fhit in the mitochondrial fraction. In H1299 cells, missing both Fhit and p53, Fdxr overexpression increases the sensitivity to ROSinduced cell death and H1299 cells expressing inducible Fhit or p53, are sensitive to ROS-induced cell death; cancer cells missing Fhit, p53, or both would lack ways to increase Fdxr expression, would be less sensitive to oxidative damage, and would survive.
Discovery of the mitochondrial function of Fhit in apoptosis, through interaction with Fdxr, extends functional parallels of tumor suppressors, Fhit and p53, lost sequentially in most cancers and involved in response to DNA damage, with p53 acting as a transcriptional and Fhit a posttranscriptional Fdxr regulator. Membranebound Fdxr accepts two electrons from NADPH, yielding NADP+ ; these electrons are passed to the iron-sulfur cluster of ferredoxin (Fdx), which donates the electrons to the heme prosthetic group of mitochondrial cytochrome p450 (CYP), which uses protons and molecular O 2 to hydroxylate its target substrate, yielding the final hydroxylated product. The electron donor, Fdx, is a 14-kDa mitochondrial matrix protein containing a Fe-S cluster 44 . Figure 5 depicts a model of the Fdxr-Fhit interaction, as a modulator of the Fdxr-Fdx interaction, and thus a modulator of the leakage of electrons from ETC III, which would contribute to ROS production and apoptosis in normal Fhit-positive cells, a modulation that would be absent in Fhit-deficient cells. Delineation of direct downstream effectors of the Fhit suppressor pathway will lead to further mechanistic studies of Fhit function that may influence preventive and therapeutic strategies for the many Fhit-deficient cancers of many types. The finding that ROS modulation is involved in Fhit-mediated apoptosis emphasizes the importance of Fhit loss as a negative prognostic factor in various clinical settings; for example, assessment of Fhit status in preneoplastic or neoplastic conditions may be predictive of the responses to antioxidant treatments.
There is a need for additional cellular targets for cancer prevention and treatment. Fhit expression is reduced or lost early in neoplasia development, leading to loss of a key ETC complex III OXPHOS interaction that participates in the production of protons and molecular O 2 through the electron donor ferredoxin. Fig. 4 Cell viability of Fhit-negative H1299 cells is decreased after atovaquone treatment. H1299 E1 and D1 cells were treated with atovaquone (10 μM) for 3 days. Cell viability was assessed by trypan blue staining. Data from three independent experiments. Note that the percentage of cell viability of atovaquone-exposed cells is normalized to the control DMSO-treated cells It has very recently been observed that "drugging mitochondrial oxidative phosphorylation might be relevant for cancer treatment" as reported by Molina et al. 38 . The authors noted that "this area has been unexplored, owing to an incomplete understanding of tumor contexts in which OXPHOS is essential". This report characterized an inhibitor of complex I of the mitochondrial ETC, IACS-010759, and stated that treatment of cancer cells with it inhibited proliferation and induced apoptosis in brain cancer and acute AML models that rely on OXPHOS. Brain cancer and AML tumor growth was potently inhibited in vivo following IACS-010759 treatment and the authors noted that IACS-010759 is under evaluation in phase 1 clinical trials in relapsed/refractory AML and solid tumors. Thus, it is feasible that such Complex I inhibitors may prove useful, with further exploration, in cancer treatment. The authors concluded that: "Through development and characterization of IACS-010759", they have provided evidence that there are tumor contexts that are dependent on OXPHOS for survival. In these tumor cells, "OXPHOS disruption created an environment of energy and macromolecule depletion that led to cell cycle arrest, apoptosis".
Fhit protein is a modulator of ROS and thus of apoptotic responses to ROS 45,46 . Because Fhit expression is lost in most cancers 3,4 , including nearly all AMLs 5 , most cancers have lost this important ROS modulator and an important apoptotic signal pathway, allowing inappropriate survival and growth of cancers. Does this suggest that the absence of Fhit expression in AMLs could have been contributing to the OXPHOS dependency of the AML and brain cancers examined 38 , and contributing to the responses of these cancers to the Complex I inhibitor examined in this report? We do know that the H460 lung cancer cell line used in the Molina et al. 38 experiments is completely negative for Fhit expression 10 . Studies of Fhit-deficient cancers for dependency on OXPHOS for survival, as described 38 , will identify Fhitdeficient cancer cell lines and cancers that can be examined for stratification by OXPHOS dependency and for responses to inhibitors of the mitochondrial electron transport chain complexes.

Generation of recombinant adenovirus
The recombinant adenovirus carrying the wild-type FHIT cDNA (AdFHIT) was prepared as previously described 12 . A His-tagged FHIT cDNA was generated by PCR with the following oligonucleotides: 5′-ACgTggATCCCTgTgAgg ACATgTCgTTCAgATTTggC-3′(forward) and 5′-TTgTgg ATCCTTATCAgTgATggTgATggTgATgCgATCCTCTC TgAAAgTAgACCCgCAg-3′. These primers were designed with a BamHI restriction site for subcloning into the transfer vector pAdenoVator-CMV5-IRES-GFP. The AdFHIT-His6 was generated with the AdenoVator TM kit (Qbiogene, Carlsbad, CA), following the manufacturer's procedure. AdGFP, used as a control, was purchased from Qbiogene.

Cells
H1299D1 and E1 lung cancer cell clones were maintained as previously described 47 . A549 lung cancer expressing a reduced level of endogenous Fhit protein, and HCT116 colon cancer cells, expressing normal Fhit expression level, were maintained in RPMI-1640 medium plus 10% FBS and penicillin/streptomycin (Sigma, St. Louis, MO). HCT116/FDXR+/-/-cells were obtained from Bert Vogelstein and retain only one copy of the FDXR gene vs. three copies in the parental HCT116 colon cancer cell line 33 .

Sample digestion and LC-MS/MS analysis
Proteins isolated with Ni-NTA beads were precipitated with cold acetone and resuspended in 6 M urea buffered ) and iodoacetamide (final concentration 25 mM, 1 h of incubation at 37°C). After neutralizing excess iodoacetamide with DTT (additional 5 mM), urea concentration was lowered to 1.5 M by dilution with 1 mM CaCl 2 .
Overnight digestion was carried out using 50 ng of TPCKtreated trypsin (Sigma). The total digestion solution volume was 100 μl.
Chromatography was performed on an Ultimate nano- MS/MS spectra were searched by interrogating the Swiss Protein database on the Mascot search engine (www.matrixscience.com). The following search parameters were used. MS tolerance: 50 ppm; MS/MS tolerance: 1 Da; methionine oxidized (variable modification); cysteine carbamidomethylated (fixed modification); enzyme: trypsin; maximum missed cleavages: 1. Protein lists obtained from, respectively, both A549 infected with Ad FHIT-His6 and control were compared, and proteins exclusively present in the A549-Ad FHIT-His6 list were kept for further validation. As a first validation procedure, LC-MS/MS raw data were inspected using selected ion chromatogram (SIC) displaying mode. By SIC comparison, it could be assessed for the exclusive presence of the peptides of interest, identified as belonging to the six candidate proteins under examination, in the Ad FHIT-His6 sample (data not shown). These findings were further validated by biochemical and functional assays. Additional LC-MS/ MS experiments and analyses were performed in the Ohio State University shared protein core and interactions were confirmed by IP experiments (data not shown).

Immunoprecipitation
Proteins were extracted in M-PER (Pierce) or RIPA buffer. Co-immunoprecipitation (Co-IP) experiments, with or without dithiobis(succinimidyl propionate) (DSP) (Thermo Scientific, Waltham, MA) were performed by incubating 1 mg of total proteins with HSP60 (Abcam, ab109660), HSP10 (Abcam, ab108600), Fhit 48 , and HA antisera conjugated with protein A/G ultralink resin (Thermo Scientific) overnight at 4°C; after TBST washing, beads were boiled in 1× reducing sample buffer and proteins were separated on polyacrylamide gel, transferred to a membrane, and probed with specific antisera.

GST pulldown
In total, 150 μg of GST or GST-Fhit protein was bound to glutathione agarose (50% slurry; Thermo Scientific) following the manufacturer's instructions. Equal amounts of protein lysate were added to the columns and rotated at 4°C overnight. The resin was washed five times in a 1:1 mixture of TBS (50 mM Tris-Cl, 150 mM NaCl) and RIPA (EMD Millipore, Burlington, MA) buffers and eluted in 1× RSB buffer (2% SDS, 60 mM Tris-Cl, 5% glycerol, and 3% β-mercaptoethanol) for western blotting.

Subcellular localization of Fhit protein
Fhit was sublocalized in Pon A-induced, Fhit-expressing H1299 D1 cells by indirect immunofluorescence detection using anti-Fhit serum and by detection of FhitHis6 in A549 AdFHIT-His6 infected cells in immuno-electron micrographs using anti-penta-His. See Supplementary information for details. In fractionation studies, mitochondria were isolated with the Mitochondria/Cytosol Fractionation kit (Thermo Scientific).

Duolink in situ proximity ligation assay (PLA)
HCT116 cells were cultured on poly-L-lysine-coated glass coverslips for 24 h. Cells were fixed in 4% paraformaldehyde in PBS (pH 7.4) for 15 min at room temperature, and permeabilized in 0.5% Triton X-100 in PBS for 15 min. The PLA procedure followed the manufacturer's recommended protocol (Sigma). Rabbit polyclonal anti-Fhit 48 was used at 1:400 and mouse monoclonal anti-Fdxr at 1:40 (Santa Cruz, sc-374436). Cells were viewed and photographed with an Olympus spectral FV1000 confocal microscope with 100× objective lens and images were processed with the Fiji ImageJ program.

Flow cytometry
PonA-induced H1299 D1 and E1 cells were treated with H 2 O 2 and incubated for varying times, as indicated in the text and figures. Cells were collected, washed with PBS, and resuspended in cold 70% ethanol. For analysis, cells were spun down, washed in PBS, and suspended in 0.1 mg/ml propidium iodide/Triton X-100 staining solution (0.1% Triton X-100, 20 mg/mL, 0.2 mg/mL DNase-free RNase A) for 30 min at room temperature and analyzed by flow cytometry.

Cell viability
H1299 E1 and D1 cells were seeded at a density of 45,000 cells/well in a six-well plate. To induce Fhit in D1 cells, PonA was added to all cells and 30 h later, cells were treated with DMSO (vehicle) or atovaquone (10 μM). Cell viability was measured 3 days later by trypan blue staining and automated cell counting. The percentage of cell viability of atovaquone-exposed cells is normalized to the value of control DMSO-treated cells. Experiments were performed three times. P-value was calculated by Student's t test.