|Home | About | Journals | Submit | Contact Us | Français|
Deformylases are metalloproteases in bacteria, plants, and humans that remove the N-formyl-methionine off peptides in vitro. The human homolog of peptide deformylase (HsPDF) resides in the mitochondria, along with its putative formylated substrates; however, the cellular function of HsPDF remains elusive. Here we report on the function of HsPDF in mitochondrial translation and oxidative phosphorylation complex biogenesis. Functional HsPDF appears to be necessary for the accumulation of mitochondrial DNA-encoded proteins and assembly of new respiratory complexes containing these proteins. Consequently, inhibition of HsPDF reduces respiratory function and cellular ATP levels, causing dependence on aerobic glycolysis for cell survival. A series of structurally different HsPDF inhibitors and control peptidase inhibitors confirmed that inhibition of HsPDF decreases mtDNA-encoded protein accumulation. Therefore, HsPDF appears to have a role in maintenance of mitochondrial respiratory function, and this function is analogous to that of chloroplast PDF.
The human mitochondrial protein peptide deformylase, HsPDF, is a metalloprotease that removes the formyl moiety on the methionine of N-formyl-methionine peptide substrates in an enzymatic assay (24, 35). Despite the slow kinetic properties of HsPDF in an in vitro deformylation assay (24, 29, 35), we have shown that small interfering RNA (siRNA) interference of HsPDF decreases human cancer cell proliferation. Similarly, pharmacologic inhibition with the PDF antibiotic inhibitor actinonin and its analogs results in mitochondrial membrane depolarization and promotes cell death or proliferation arrest in a wide variety of cancer cell lines (18, 25). However, the cellular function of HsPDF remains elusive, and others have proposed that it has none (29). In bacteria, deformylation of nascent peptides is necessary for removal of the N-terminal methionine (36) and posttranslational processing of at least a subset of proteins that contribute to cell growth and viability (28). Prokaryotic PDF thus fulfills a role in cotranslational processing (7) and in protein degradation (41).
In mammals, N-terminal formylation of proteins is only known to occur during mitochondrial translation initiation, as in prokaryotic protein translation (6). In contrast to bacteria, where the entire proteome is formylated for translation initiation, formylation in eukaryotes is limited to the 13 mitochondrial DNA (mtDNA)-encoded proteins. Formylation is important for mitochondrial translation, because formyl-Met-tRNA, but not Met-tRNA, is recognized by initiation factor 2 as the initiator tRNA (26, 37, 39). Therefore, the participation of HsPDF in protein post- or cotranslational processing can be narrowed down to these mitochondrial translation products.
Despite the current understanding of the function of formyl-methionine in the initiation of protein synthesis in mammalian mitochondria (38, 39), the functional relevance of the downstream processing of nascent mitochondrial translation products has remained unexplored. Furthermore, it has been assumed that human mitochondria-encoded proteins, like those of bovine origin, are generally not deformylated after synthesis (45).
The mammalian mitochondrial genome-encoded proteins are all subunits of four of the five oxidative phosphorylation respiratory chain enzyme complexes (I, III, IV, and V) (2, 40, 42). Respiratory complexes are comprised of multiple proteins. With the exception of complex II, which is comprised entirely of nuclear DNA-encoded subunits, all other complexes include both nuclear and mitochondrial DNA-encoded proteins. Synthesis of key mtDNA-encoded protein subunits, and the assembly of these proteins with multiple nuclear-encoded subunits within the mitochondria, is necessary for the function of each individual complex (16, 30, 44). Moreover, a functional interdependence among stably assembled respiratory complexes has been demonstrated (1). Mutations in human mtDNA that affect protein-coding regions or nuclear DNA mutations that affect expression of respiratory complex subunits cause disease (13), including Parkinson's disease, for example, in which decreased respiratory function and compromised cell viability have been demonstrated (5, 21, 23). Therefore, the importance of properly assembled mitochondrial respiratory complexes suggests that their disruption, by inhibition of mtDNA-encoded protein processing, could have significant effects on cellular function.
We hypothesized that HsPDF-mediated processing of mtDNA-encoded proteins is necessary for proper function of the respiratory chain complexes. To determine how the human deformylase activity contributes to cellular function, we used pharmacologic inhibition of HsPDF activity with the hydroxamic acid peptidomimetic inhibitor of PDF, actinonin, and confirmed our findings with a variety of other structurally different inhibitors. PDF has been shown to be a target of actinonin in bacteria (9), human cells (24), and plants (17).
Here we show that inhibition of HsPDF function in mitochondria of human cell lines reduces mtDNA-encoded protein accumulation, new respiratory complex assembly, and energy production by the mitochondria. Aerobic glycolysis-dependent cell survival ensues upon disruption of HsPDF function. Therefore, HsPDF appears to fulfill a function in the mitochondria and to have a role in mtDNA-encoded protein-containing oxidative phosphorylation (OXPHOS) complex biogenesis.
PC9, a non-small cell lung cancer cell line (Sloan-Kettering Institute) and SKLC-4, a lung adenocarcinoma cell line (Ludwig Institute), as well as Ramos Burkitt's lymphoma cells (ATCC), were grown in RPMI medium supplemented with 10 mM HEPES, nonessential amino acids, l-glutamine, penicillin-streptomycin, and 10% fetal bovine serum. MDA-MB-231 cells were grown in Dulbecco's modified Eagle's medium high glucose with nonessential amino acids and 10% fetal bovine serum.
The inhibitors actinonin (242.3 ± 1.2 nM [mean ± standard error of the mean]), chloramphenicol (100 μM), and bestatin (100 μM) were purchased from Sigma. Phenoxychromanone [3-(4-fluorophenoxy)-7,8,-dihydroxy-2-methyl-4H-chromen-4-one; 10 to 59 μM] was obtained from Chembridge (San Diego, CA). Actinonamide (40 μM) was synthesized at the Sloan-Kettering Institute Organic Synthesis Core Facility. SK-BF-13 (1 to 5 μM) was provided by the High Throughput Core Facility at Sloan-Kettering Institute. CHR-2863 (>30 μM) was a gift from Chroma Therapeutics. Fifty percent inhibitory concentration values for HsPDF inhibition in vitro (3) are shown in parentheses for the PDF inhibitors. Oligomycin A and carbonyl cyanide 3-chlorophenylhydrazone (CCCP) were purchased from Sigma-Aldrich.
For denaturing gel Western blotting, cells were treated with vehicle (dimethyl sulfoxide [DMSO], 0.1%), 40 μg/ml chloramphenicol (Sigma), 40 μM actinonin, 40 μM actinonamide (Organic Chemistry Core Facility at Sloan Kettering Institute), or 40 μM bestatin (Sigma Aldrich) for 24 h. Cell pellets were lysed in 1× radioimmunoprecipitation assay buffer (50 mM Tris [pH 8.0], 0.3 M sodium chloride, 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate). Western blotting was performed as instructed by the primary antibody vendors' instructions. Native respiratory complex isolation was carried out as described elsewhere (31), using 0.75 g of lauryl-maltoside/g of mitochondrial protein and a Brilliant Blue G-250 concentration, for sample electrophoresis, of one-fourth of the detergent concentration. Native PAGE was carried out on a 3-to-12% bis-Tris resolving gel (Invitrogen). Antibodies used included anti-complex IV (Mitosciences MS404, MS405, and MS407), anti-complex II 70-kDa subunit (Invitrogen A-11142), anti-complex III subunit core 2 protein (Mitosciences MS304), anti-complex V subunit alpha (Mitosciences MS507), and anti-beta-actin (MABTECH). Goat anti-mouse horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnhology) was used for all primary antibodies. Western blot detection was carried out with SuperSignal West Pico and Femto reagents (Pierce).
Confluent cells (90%) were treated with various inhibitors for 24 h as described for Western blotting. Incorporation of [35S]Met-Cys into mitochondrial translation products, and pulse-chase experiments were carried out as described previously (10). The pulse and chase were carried out in the presence of actinonin or vehicle.
HsPDF activity was measured as described previously (3).
The rate of complex V-dependent ATP synthesis was measured as indicated elsewhere (43). Luminescence was measured with a Fluoroscan Ascent FL luminometer (Thermo Scientific). The optimal digitonin concentration for permeabilization was determined first by measuring the ATP synthesis rate of 1 × 106 cells treated with various digitonin (Sigma-Aldrich) concentrations. The amount of ATP produced was determined from the luminescence signal of ATP standards.
Complex I activity was measured in 20 μg of mitochondrial extracts from cells treated for 24 h, as indicated for Western blotting and as described elsewhere (8). NADH oxidation was measured from the disappearance of the 340-nm absorbance after 10 min of incubation with the mitochondrial lysate for each triplicate sample in the presence or absence of the complex I inhibitor rotenone. The rotenone-resistant absorbance after 10 min was subtracted from each sample, and the pmol of NADH per reaction mixture was calculated from the molar absorption coefficient for NADH and the reaction volume. Data are presented as pmol NADH/min/mg of protein.
For the complex IV activity assay, cells were treated as described for the Western blot samples. Complex IV activity was measured from the colorimetric oxidation of cytochrome c by complex IV, followed by a decrease in absorbance at 550 nm, using the MitoProfile MS443 rapid microplate assay kit for human complex IV activity, and quantititation (Mitosciences) according to the manufacturer's instructions for whole-cell lysates.
PC9 or SKLC-4 cells were treated with 0.1% DMSO, 40 μM actinonin,or 40 μg/ml chloramphenicol (CAM) for 24 h. Potassium cyanide-sensitive oxygen consumption was measured in 2 × 106 cells as described previously (19).
For proliferation assays triplicate samples of 10,000 cells were treated with half-log dilutions of actinonin (≥99%; Biochemika Fluka), starting at 100 μM, for 24 h. Cell proliferation was determined by incorporation of [3H]thymidine (11). For ATP depletion, an ATPlite assay was performed with 500 to 1,000 cells per replicate, and ATP levels were assessed from an ATP standard curve. For the actinonin treatment of cells in the absence of glucose, cells were allowed to attach, and the medium was changed to RPMI, or RPMI glucose-free supplemented with 1.5 mM sodium pyruvate, nonessential amino acids, and 10% dialyzed fetal bovine serum (10), prior to treatment. Propidium iodide (Invitrogen) staining was done as described elsewhere (12). A total of ~1.5 × 104 PC9 cells per replicate were treated with 0.1% DMSO or 40 μM actinonin, in the presence or absence of glucose as described above, and cell numbers were determined at 0, 24, 48, and 72 h posttreatment.
Lactic acid production was measured with the lactic acid kit (Trinity Biotech) as described in reference 4. PC9 and SKLC-4 cells were seeded at 60 to 70% confluence in 12-well plates, with the same cell number per well. Triplicate wells were treated with vehicle (DMSO, 0.1%) or 40 μM actinonin for 24 h, and lactic acid in the supernatant was measured.
All statistical comparisons were done with the GraphPad Prism version 4.0 program for Macintosh (GraphPad Software). Mean pair-wise comparisons were done with the two-tailed t test and 95% confidence intervals. Multiple comparison of means were done with the one-way analysis of variance test and the Bonferroni multiple comparison test of paired means. P values in bar graphs correspond to the comparison of the treatment group with the vehicle treated control.
Native, assembled complex IV was isolated from 600 × 106 to 900 × 106 PC9 cells by solubilization of isolated mitochondria with n-dodecyl-β-d-maltoside (Sigma-Aldrich) and immunoprecipitation with a cross-linked anti-complex IV antibody (Mitosciences MS401). Mitochondria were isolated as described elsewhere (31). Immunoprecipitation was done according to the antibody vendor instructions. Captured complex IV was eluted with 1% SDS, resolved in a 12% gel by SDS-PAGE, and stained with GelCode Blue dye (Pierce). In-gel protein digestion and mass spectrometry were carried out at the Sloan-Kettering Microchemistry and Mass Spectrometry Core Facility. Peptides obtained from endoproteinase Lys C in-gel digestion (34) of the 37-kDa protein band, corresponding to cytochrome oxidase subunit 1, were cleaned up and concentrated (14) and then analyzed by matrix-assisted laser desorption-reflectron time of flight (MALDI-reTOF).
We previously showed that HsPDF cleaves the formyl group from at least 10 of its putative formylated substrates, derived from the N-terminal amino acids of the mitochondrial DNA-encoded proteins (GenBank Refseq accession identifier NC_001807) in an in vitro deformylation assay (15), although no direct evidence yet exists for HsPDF-mediated deformylation in a live cell. Three of the formylated peptide substrates of HsPDF correspond to the N terminus of complex IV (cytochrome c oxidase [COX]) subunit proteins I, II, and III (COI, COII, and COIII, respectively). These three COX subunits are all mitochondrial translation products and, therefore, possible substrates of HsPDF in the mitochondria. COX is the most abundant respiratory complex in heart mitochondria and has the highest ratio (3:10) of mtDNA to nuclear DNA-encoded subunit components (32). We hypothesized that if HsPDF participates in the processing of mitochondrial translation products, then inhibition of PDF activity should affect the production of COI, -II, and -III.
To test our hypothesis, we asked whether inhibition of HsPDF with actinonin had an effect on the expression of a representative respiratory complex, complex IV, nuclear DNA, and mtDNA-encoded subunits in the PC9 non-small cell lung cancer line and SKLC-4 lung adenocarcinoma cell line, in which 24-h treatment with actinonin results in dose-dependent loss of proliferation (PC9 50% effective concentration [EC50], 12.02 ± 1.35 μM [mean ± standard error of the mean]; SKLC-4 EC50, 10.80 ± 1.50 μM, as measured by [3H]thymidine incorporation). COI and COII protein levels were measured as representatives of mtDNA-encoded COX subunits, whereas COIV is a nuclear-encoded COX component. Levels of the complex II 70-kDa subunit were determined as a reference, since this complex does not have any mtDNA-encoded components. Both COI and COII decreased in SKLC-4 cells treated with actinonin, whereas no changes were observed in the levels of COIV or the complex II 70-kDa subunit (Fig. (Fig.11 A). CAM inhibits the translation of mtDNA-encoded proteins by blocking peptide elongation (33). As expected, CAM produced an effect on protein accumulation similar to that in cells treated with actinonin. The reduced accumulation of subunits I and II of complex IV, but not of subunit IV after 24 h treatment with chloramphenicol or actinonin, suggests that whereas these inhibitors have an effect on mitochondrial translation, they do not impair import of the nuclear-encoded subunit IV into the mitochondria at the time point studied.
Aminopeptidase N (APN) is a nonmitochondrial protein target of actinonin. In order to rule out that the observed effect of actinonin on mitochondrial DNA-encoded COX subunits was due to a possible unknown indirect effect following inhibition of APN, we treated cells with the APN inhibitor bestatin, which does not inhibit PDF (25). Bestatin did not have an effect on the COX or the complex II 70-kDa subunit levels. Another control inhibitor, the actinonin analog actinonamide, in which the hydroxamic acid moiety necessary for binding to the metal in the PDF active site has been removed, was also tested. This far less active actinonin analog did not affect COX subunit levels. Similar effects on COX and complex II subunit levels were observed in PC9 cells treated with actinonin, CAM, bestatin, or actinonamide (Fig. (Fig.1B1B).
In order to determine whether deformylation affects the synthesis of all mitochondrial translation products, we selectively labeled newly synthesized mtDNA-encoded proteins with a short pulse of [35S]Met-Cys, in the presence of the cytoplasmic translation irreversible inhibitor emetine, in PC9 cells that had been pretreated with vehicle, CAM, actinonin, actinonamide, or bestatin for 24 h (Fig. (Fig.2A).2A). Because of the brief length of the pulse, incorporation of the [35S]Met-Cys is a reflection of the rate of protein synthesis (10), rather than an indication of protein turnover. Approximately 9 out of 13 mitochondrial translation products were observed in the vehicle-treated sample, with the lack of detection of the remaining four possibly reflecting differences in the limit of detection. Consistent with our previous findings, deformylase inhibition with actinonin resulted in decreased [35S]Met-Cys incorporation comparable, but slightly less effective, than that observed with CAM. No difference was observed in [35S]Met-Cys incorporation in control cells pretreated with bestatin or actinonamide relative to vehicle. These results were confirmed in SKLC-4 cells (data not shown). Actinonin also decreased the accumulation of mtDNA-encoded proteins in other cell lines (Fig. (Fig.2B2B).
In order to determine whether inhibition of deformylation increases mitochondrial translation product degradation, pulse-chase experiments were conducted. We labeled the mitochondrial translation products of PC9 cells that had been pretreated for 24 h with actinonin, by using a longer (1-h) pulse with [35S]Met-Cys mix, in the presence of cycloheximide, followed by a chase (Fig. (Fig.2C).2C). While total proteins were significantly reduced compared to untreated cells, deformylase inhibition did not appear to affect the half-life of the few formed mitochondrial translation products. Consistent with a decrease in the normal accumulation of mtDNA-encoded proteins, but not an increase in their degradation upon HsPDF inhibition with actinonin, assembled respiratory complex steady-state levels were also decreased by actinonin. Blue native electrophoresis and Western blotting of solubilized PC9 cell mitochondria showed that cells treated with actinonin or CAM for 24 h had less assembled complexes III, IV, and V, but not of complex II, than vehicle-treated cells (Fig. (Fig.2D2D).
Because small molecules may act as uncouplers of electron transport, we sought to rule out that actinonin is an uncoupler. We measured complex V-dependent ATP synthesis in digitonin-permeabilized PC9 cells treated with actinonin, vehicle, or a known uncoupler, CCCP (Fig. (Fig.2E).2E). CCCP (1 μM), but not actinonin (40 μM) or vehicle, caused a sharp decrease in complex V-dependent mitochondrial ATP synthesis. Therefore, the decrease of mitochondrial translation by actinonin is not likely due to its action as an uncoupler. The decrease in mitochondria-encoded protein accumulation by actinonin was reversed as soon as 4 h after removal of the drug, further ruling out nonspecific toxic effects of actinonin on mitochondrial translation.
To confirm that the decrease in mtDNA-encoded protein accumulation resulting from actinonin treatment is mediated by HsPDF inhibition, we turned to two additional small molecule HsPDF inhibitors. The first, phenoxychromanone (FC), a flavonoid, is structurally different from the peptidomimetic inhibitor actinonin. Despite its lower potency than actinonin in an in vitro assay of HsPDF activity, FC is more specific for HsPDF, as it does not inhibit APN, MMP-1, MMP-2, MMP-9, or ADAM10, all of which are also targets of actinonin. PC9 cell treatment (24 h) with FC results in dose-dependent loss of proliferation (EC50, 10.52 ± 1.27 μM, measured by [3H]thymidine incorporation).
Consistent with HsPDF inhibition by actinonin, FC decreased the steady-state levels of COI and COII, but not of COIV, relative to vehicle- and CAM-treated control PC9 cells (Fig. (Fig.3A).3A). Similarly, FC decreased the accumulation of mitochondrial translation in both PC9 and SKLC-4 cells, albeit less effectively than actinonin (Fig. (Fig.3B).3B). The decreased suppression of mitochondrial translation by FC compared to actinonin is a reflection of the decreased potency of this compound.
We next tested the effects of a third, structurally unrelated benzofuranone HsPDF inhibitor, SK-BF-13, on mitochondrial protein synthesis. PC9 cell treatment (24 h) with SK-BF-13 resulted in dose-dependent loss of proliferation (EC50, 30 ± 1.73 μM, measured by [3H]thymidine incorporation). PC9 cells were pretreated with vehicle, CAM, actinonin, or SK-BF-13 for 24 h and labeled with a short [35S]Met-Cys pulse in the presence of emetine (Fig. (Fig.3C).3C). SK-BF-13 inhibition of HsPDF resulted in decreased mitochondrial translation in PC9 cells at both 40 μM and 80 μM. The same effect of three structurally unrelated HsPDF inhibitors, actinonin, FC, and SK-BF-13, on mtDNA-encoded protein accumulation showed that the action of these inhibitors is mediated by inhibition of HsPDF and not another target.
As an additional control to further confirm that the inhibition of the accumulation of mtDNA-encoded proteins observed with actinonin is not related to another hydroxamic acid-sensitive activity in the mitochondria, we tested the hydroxamate-containing ester peptidomimetic compound CHR-2863. CHR-2863 belongs to the CHR-2797 family of nanomolar antiproliferative intracellular peptidase inhibitors (22). Whereas CHR-2863 is equipotent to CHR-2797 in its nanomolar inhibition of intracellular peptidases such as human puromycin-sensitive aminopeptidase (PuSA) and human leucine aminopeptidase (LAP), it does not potently inhibit HsPDF in vitro (Fig. (Fig.3D).3D). Consistent with the poor targeting of HsPDF by CHR-2863, this compound did not cause a decrease in mtDNA-encoded protein accumulation in PC9 cells (Fig. (Fig.3E).3E). These data provide additional evidence that hydroxamate inhibition of other peptidase targets is not likely to account for the decreased accumulation of mtDNA-encoded proteins induced by actinonin.
To determine if OXPHOS complex activity is also decreased by deformylase inhibition, we measured the activity of complex I, which is a rate-limiting complex in the respiratory chain, as well as COX. As expected, complex I activity was reduced in mitochondria from PC9 cells treated with actinonin for 24 h compared to the activity in mitochondria from cells treated with the vehicle alone (Fig. (Fig.4A).4A). The decrease in complex I activity in mitochondria of cells treated with actinonin was comparable to that observed with inhibition of mitochondrial translation by CAM. After 24 h of treatment with actinonin, COX activity was also decreased in both PC9 and SKLC-4 cells in the presence of actinonin, although to a lesser degree than complex I (Fig. (Fig.4B4B).
We have shown that inhibition of deformylase function decreases the accumulation of the components of the main energy-producing machinery of the cell, as well as the function of representative mitochondria-encoded subunit-containing complexes. Therefore, we sought to determine whether inhibition of deformylase function in turn affects mitochondrial respiratory function, as assessed by oxygen consumption in actinonin- or vehicle-pretreated cells.
The effect of actinonin on oxygen consumption was striking. Actinonin induced a significant decrease in coupled (Fig. 4C and D) and uncoupled cyanide-sensitive respiration (data not shown), compared to vehicle-treated cells, in PC9 and SKLC-4 cells. This decrease in mitochondrial respiration was comparable to that observed in the positive control CAM-treated cells.
We next asked whether HsPDF inhibition is sufficient to impair mitochondrial respiration-dependent cell survival. To test if the disruption in mitochondrial function elicited by HsPDF inhibition is sufficient to decrease cell survival, we measured PC9 cell viability in the presence of actinonin when glucose was available or absent from the medium, in order to produce ATP through oxidative phosphorylation. In the absence of glucose, we substituted pyruvate, in order to allow cells to produce ATP through oxidative phosphorylation, but not through aerobic glycolysis. We reasoned that if mitochondrial respiratory function, thus ATP production, is impaired by actinonin, cells might be able to sustain viability by producing ATP through aerobic glycolysis, as long as glucose was available, according to the Warburg effect.
We measured cell viability by the flow cytometry-based propidium iodide (PI) cell staining assay in PC9 cells. This method allowed simultaneous quantitation of both live cells and dead cells. Over a 3-day period in the presence of glucose, the cells stopped growing in the presence of actinonin; cells treated with vehicle continued to double every 24 h (Fig. (Fig.5A).5A). At the same time, the percentage of dead cells remained low in both vehicle- and actinonin-treated cells. These data confirm that, in this cell line, actinonin has a cytostatic effect in the presence of glucose. On the other hand, in the presence of pyruvate, but without glucose, a sharp decline in total cell number accompanied by an increase in the percentage of dead cells was observed in actinonin-treated cells (Fig. (Fig.5B).5B). Similar results were observed for PC9 cell treatment with phenoxychromanone (data not shown).
These data indicate that inhibition of HsPDF activity with actinonin disrupts the capacity of the cell to generate energy through mitochondrial oxidation and suggest that HsPDF is necessary for energy generation by the mitochondria. These data also imply that HsPDF inhibition induces a switch from oxidative mitochondrial ATP production to aerobic glycolytic metabolism. It appears that when HsPDF is inhibited with actinonin in the absence of glucose, the cell is not able to utilize the OXPHOS precursor pyruvate for energy production, and because extramitochondrial ATP production is unavailable, cells die. This is presumably due to a defect in the oxidative phosphorylation capacity of the cell, evidenced by the loss of viability when energy production depends solely on mitochondrial respiratory function.
We confirmed that the cell utilizes its cytoplasmic compensatory mechanism to produce ATP in the presence of actinonin, by increased production of lactic acid, a by-product and marker of aerobic glycolysis. Consistent with our previous observation, actinonin-treated PC9 cells showed a significant increase in lactate production, relative to the vehicle-treated controls, as soon as 8 h of treatment (data not shown). The increase in production of lactic acid remained constant for up to 24 h of treatment in PC9 cells (Fig. (Fig.5C).5C). There was also a trend of increasing lactic acid production over 24 h in SKLC-4 cells treated with actinonin compared to those treated with vehicle, but statistical significance was not reached.
Because energy production through aerobic glycolysis is less efficient than oxidative phosphorylation, increased utilization of aerobic glycolysis in the presence of actinonin should also be accompanied by cellular ATP depletion. A significant decrease in the cellular ATP levels was observed in actinonin relative to vehicle-treated PC9 and SKLC-4 cells. A smaller decrease was observed in PC9 compared to SKLC-4 cells (Fig. 5D and E). Together, the data suggest that inhibition of HsPDF results in a defect in the mitochondrial oxidative phosphorylation capacity of the cell with a resultant shift to cytoplasmic energy production through aerobic glycolysis, a reduction in ATP, less oxygen use, and an increase in lactate production.
Translation of proteins in human mitochondria is poorly understood. In this organelle translation initiation is dependent on the methionine formylation of nascent peptides; however, the function of HsPDF has remained unexplored. The need for HsPDF function for cancer cell proliferation and viability, however, strongly supports its necessary role in the cell. Furthermore, a recent report on the importance of N-terminal acetylation as a specific degradation signal (20) requires consideration of deformylase function in the cell. Here we showed that inhibition of deformylase function affects mitochondrial protein synthesis and OXPHOS complex formation, as well as the energy homeostasis of the cell, with ATP depletion and increased aerobic glycolysis triggered by defects in mitochondrial function that affect the cellular mitochondrial oxidative phosphorylation capacity. Off-targeting of other protease activities by actinonin was accounted for with various extra- and intracellular peptidase inhibitors, which did not decrease mtDNA-encoded protein accumulation. Additionally, our findings were confirmed with two structurally distinct HsPDF inhibitors, as well as several control compounds.
Although it has been suggested that HsPDF has no function (29), the decrease in mtDNA-encoded protein accumulation resulting from HsPDF (type 1A PDF) inhibition reported in this paper resembles the consequences of plant PDF (type 1B) inhibition by actinonin in the chloroplast (17). However, inhibition of human mitochondrial and chloroplast PDF differ in their consequences for protein translation, in the corresponding organelle, in two aspects: (i) inhibition of HsPDF causes a global decrease in the accumulation of mtDNA-encoded proteins, whereas inhibition of chloroplast PDF affects only a subset of chloroplast-encoded proteins; (ii) HsPDF inhibition does not decrease the half-life of the few mtDNA-encoded proteins formed in the presence of actinonin, whereas inhibition of chloroplast PDF decreases the half-life of the affected subset of chloroplast-encoded proteins.
The mechanism by which inhibition of HsPDF results in decreased accumulation of mtDNA-encoded proteins remains to be elucidated. N-terminal analysis of bovine heart mitochondria-encoded proteins suggested that in this mammalian tissue subunit III of complex IV is deformylated (45). Furthermore, in an attempt to determine whether deformylation occurs in human mitochondria, we discovered that deformylated subunit I of complex IV exists in assembled complex IV isolated from PC9 cells (Fig. (Fig.6).6). Generalization of these findings to all human mitochondria-encoded proteins suggests that complete deformylation is not necessary for complex assembly. However, systematic N-terminal analyses of mitochondria-encoded proteins of human origin do not exist. Whether deformylation is a function of the mitochondrial translation product amino acid sequence, tissue type, or transformed state is also unknown. However, prior data from our laboratory suggest that HsPDF is capable of deformylating peptides containing the N-terminal sequences of 10 of 13 mitochondria-encoded proteins (15). The functions of formylated versus deformylated proteins cannot be discerned from our data, which rather suggests that our understanding of mitochondrial N-terminal processing is still incomplete. HsPDF function may include other activities beyond its role in catalysis, for example, as a chaperone capping molecule.
Deformylation of formyl-methionine-tRNA by HsPDF would provide a mechanism by which active site inhibition of deformylase leads to mitochondria-encoded protein synthesis deficiency. Formyl-methionine-tRNA is necessary for mitochondrial translation initiation (39), and therefore inhibition of HsPDF could decrease the available pool of the aminoacyl-tRNA necessary for new protein synthesis. However, Livingston and Leder (27) have shown that formyl-methionine-puromycin-tRNA, but not formyl-methionine-tRNA, is a substrate of bacterial deformylase.
The loss of respiratory function measured upon HsPDF inhibition stems from the impaired assembly of new respiratory complexes that results from the decreased accumulation of mtDNA-encoded proteins. Although actinonin did not decrease complex IV activity to the same extent as complex I, the levels of assembled complex IV were markedly decreased as confirmed by blue native gel electrophoresis and Western blotting detection of this complex. The decrease in mtDNA-encoded protein accumulation by actinonin after 24 h may not cause a large decrease in COX activity due to the complex abundance. Also, the activity of COX is highly regulated by mechanisms other than its abundance (16).
The decrease in accumulation of mtDNA-encoded proteins by structurally different inhibitors of HsPDF points at the effect being a result of HsPDF inhibition and not off-target activities. The lack of inhibition of mitochondrial translation by a hydroxamic acid-containing analog compound (CHR-2863) also suggests that hydroxamate inhibition of other peptidases is unlikely to account for the effect of actinonin on mitochondrial translation. Although downregulation of HsPDF through siRNA (25) is possible, complete and sustained downregulation of HsPDF protein cannot be achieved, with a maximum of ~50% downregulation in HsPDF. The remaining 50% HsPDF appears sufficient to sustain function, or the half-life and abundance of mitochondria-encoded proteins may also mask any effects that HsPDF downregulation may have on mitochondrial translation. Western blotting comparisons of HsPDF protein levels in various cell lines to a recombinant HsPDF standard suggest that HsPDF is highly abundant in the cell's mitochondria; therefore, despite downregulation of HsPDF by siRNA, sufficient HsPDF protein may be available to fulfill a function in mitochondrial translation.
We have also excluded the possibility that the effects of actinonin may be due to uncoupling of mitochondrial membrane potential by this compound. Although we showed that mitochondrial translation is impaired by actinonin in multiple cell lines, this effect was not universal across all cell lines. Actinonin did not decrease accumulation of mtDNA-encoded proteins in other human lines, such as human coronary artery endothelial cells (HCAEC), WI-38 lung fibroblasts, and the thyroid oncocytoma line XTC-1 (data not shown). The difference in sensitivity does not appear to be simply related to PDF expression, because the presence of HsPDF (measured by Western blotting) appeared similar across “actinonin-sensitive” and “actinonin-insensitive” cell lines (data not shown). This points to differences in inhibitor uptake by the mitochondria across different cell lines, although other possible underlying biological differences are unknown.
As we have shown, HsPDF inhibition results in survival based on the capacity of the cell to carry out aerobic glycolysis. Thus, in the presence of actinonin, cells survive for a finite number of days, as long as glucose is available. Therefore, differences in cellular adaptation to aerobic glycolysis and in respiratory capacities across cancer cell lines may both contribute to their sensitivity to inhibition of HsPDF.
This work was supported by NIH grant CA 55349, the Cancer Research and Treatment Foundation, and by the Experimental Therapeutics Center and the Geoffrey Beene Cancer Research Center at Memorial Sloan-Kettering Cancer Center.
We thank Nikola Pavletich and Charles Sawyers for their comments and input during the preparation of the manuscript. We thank Anissa Igoudjil, Rebeca Acín-Pérez, Christopher Antczak, and Hakim Djaballah for their useful advice. Thanks go to David Krige of Chroma Therapeutics for providing CHR-2863 and Hediye Erdjument-Bromage for MS sample analysis.
Published ahead of print on 30 August 2010.