Development and Characterization of a Conditional Mitochondrial Complex I Assembly System*

We developed a conditional complex I assembly system in a Chinese hamster fibroblast mutant line, CCL16-B2, that does not express the NDUFA1 gene (encoding the MWFE protein). In this mutant, a hemagglutinin (HA) epitope-tagged MWFE protein was expressed from a doxycycline-inducible promoter. The expression of the protein was absolutely dependent on the presence of doxycycline, and the gene could be turned off completely by removal of doxycycline. These experiments demonstrated a key role of MWFE in the pathway of complex I assembly. Upon induction the MWFE·HA protein reached steady-state levels within 24 h, but the appearance of fully active complex I was delayed by another ∼24 h. The MWFE appeared in a precomplex that probably includes one or more subunits encoded by mtDNA. The fate of MWFE and the stability of complex I were themselves very tightly linked to the activity of mitochondrial protein synthesis and to the assembly of subunits encoded by mtDNA (ND1–6 and ND4L). This novel conditional system can shed light not only on the mechanism of complex I assembly but emphasizes the role of subunits previously thought of as “accessory.” It promises to have broader applications in the study of cellular energy metabolism and production of reactive oxygen species and related processes.

The mammalian mitochondrial complex I (NADH-ubiquinone oxidoreductase) is an assembly of ϳ46 distinct proteins encoded by 39 nuclear and 7 mitochondrial genes (1). Fourteen of these subunits, including all the mtDNA-encoded ones, have orthologues in prokaryotes (2). This simpler prokaryotic complex of 14 subunits can perform all the currently known functions of mammalian complex I: NADH oxidation, electron transport from the flavin mononucleotide via several Fe-S centers to a quinone, and proton translocation across the membrane. The roles of the other 32 "accessory" or "supernumerary proteins" in the mammalian complex I are largely unknown. Many of these subunits are also found in fungal and plant complexes. Only a few of these accessory proteins, e.g. the acyl-carrier protein SDAP and the AQDQ and MWFE subunits from Neurospora crassa and mammals have been studied in more detail (3)(4)(5). The orthologue of the mammalian MWFE in N. crassa was identified recently (6). Like its mammalian counterpart it was found to be essential for complex I biogenesis.
From bacteria to humans, the overall "boot-shaped" structure of complex I is conserved with one arm protruding into the matrix and the other localized in the plane of the membrane (7)(8)(9)(10)(11). The matrix arm contains all the prosthetic groups, whereas the function of the large membrane arm is less clear. Recent findings suggest that it is involved in the binding of quinone and inhibitors and proton/ion translocation across the inner membrane (12). Studies with a N. crassa mutant indicate that complex I assembly takes place in discrete steps with independent assembly of membrane and peripheral arms (13). The attachment of the peripheral arm to the membrane depends upon the assembly of the membrane arm. Although likely, this remains to be proved explicitly for the mammalian complex I assembly. Differences in the biogenesis of complex I between N. crassa and humans have been noticed, for example, with regard to the exact role of the AQDQ subunit. In N. crassa, the AQDQ homologue is not essential for complex I assembly and rotenone-sensitive activity (5), whereas in human patients with a homozygous mutation in the NDUFS4 gene (AQDQ subunit) complex I assembly is severely affected (14,15).
Previous studies from this laboratory have characterized a Chinese hamster mutant cell completely deficient in the MWFE subunit (encoded by the X-linked NDUFA1 gene) (3,4,16). This mutant lacked rotenone-sensitive complex I activity. However, several subunits of the peripheral membrane subcomplex were present in this mutant at a normal or near normal level. Thus, the MWFE protein appears to be required for either the assembly and/or stability of complex I. At the same time, MWFE by itself is quite unstable in mutants defective either in complete mitochondrial protein synthesis or in the absence of single mtDNA-encoded ND subunits (Ref. 4

and the present study).
Experiments presented in this communication describe the development and characterization of a cell line in which complex I assembly is conditional, depending upon the induction of the MWFE subunit. The HA 1 epitope-tagged MWFE protein (MWFE⅐HA) was expressed from a doxycycline-inducible promoter. Using this system, the CCL16-B2 cells that lack complex I and were completely respiration-deficient (res Ϫ ) could be made respiration-competent (res ϩ ) after the addition of doxycycline to the growth medium. This model system permitted more detailed questions to be raised about the exact role of the MWFE protein in 1) the assembly, 2) the kinetics of assembly, 3) the maintenance of the complex I, and 4) the stability/ accumulation of MWFE subunits in relation to mitochondrial protein synthesis and the incorporation of the ND subunits into a mature complex I.

EXPERIMENTAL PROCEDURES
Cell Lines and Cell Culture-The CCL16-B2 cells with a deletion in the NDUFA1 transcript have been described (3, 4, 16 -19). 2 V79-G7 cells are also respiration-deficient (res Ϫ ) hamster cells with almost no measurable mitochondrial protein synthesis (20 -22). The res Ϫ cells grow normally in DME medium with an abundant supply of glucose (DME-Glu) to sustain glycolysis and a supplement of nonessential amino acids. Substitution of glucose with galactose (DME-Gal) in the medium represents the non-permissive condition for res Ϫ cells. Routinely the medium contained 10% fetal bovine serum and the antibiotics gentamicin and Fungizone. We thank Drs. Attardi and A. Chomyn for sending us their mouse cell lines (3A-20-4 and 4A) with null mutations in ND5 and ND6 subunits and human cell line 4TL with a null mutation in ND4 subunit (23). Cells were harvested by trypsinization after one wash with TD buffer (0.3% Tris, 0.8% NaCl, 0.038% KCl, 0.025% Na 2 HPO 4 ⅐12H 2 O, brought to pH 7.4 with HCl).
Plasmids and Genes-The elongation factor 1␣ promoter of pTRI-DENT-neo vector expressing wild type hamster MWFE⅐HA (pIS2022) (4) was replaced with doxycycline-responsive human cytomegalovirus-1 promoter from vector pTRE-2 (Clontech). The pTRE-2 vector was digested with XhoI, end-filled with Klenow polymerase, and redigested with EcoRI to excise the human cytomegalovirus-1 promoter. This XhoI-EcoRI fragment was ligated with the larger SspI-EcoRI fragment of pIS2022. The resulting inducible construct (pIS2022 i ) was identified by its smaller size and diagnostic SacII digestion. Further the pIS2022 i was modified to express the activator protein (rtTA) from the same construct. This was achieved by taking out the smaller Bst1107I-BsaI fragment (3579 bp) and ligating it with the larger Bst1107I-BsaI fragment (3884 bp) of pIS2022 i . The final construct, which can express both the regulator protein (rtTA) and the inducible proteins (MWFE⅐HA and NeoR), was named pTRE ON.
Selection of the Doxycycline-inducible Mutant-Cells were transfected with DNA using SuperFect reagent (Qiagen) essentially as described previously (4). The CCL16-B2 cells (5 ϫ 10 5 ) with the NDUFA1 null mutation were seeded in a 6-well tissue culture plate overnight and then transfected with the regulator plasmid pTet-ON (Clontech). Stable transfectants were selected in the presence of 800 g/ml Geneticin (G418). These transfectants were retransfected with the inducible construct pIS2022 i . The superscript "i" and "c" refer to inducible and constitutive expression of MWFE⅐HA from the vector. After the transfection, the cells were induced with variable concentrations of doxycycline for ϳ48 h and then selected in DME-Gal. The selective medium was routinely changed every 48 h to replenish it with fresh doxycycline.
The V79-G7 cells were transfected with the pTRE ON construct and selected with G418 in the presence of doxycycline, after 48 h of induction, in DME-Glu medium as described above. V79-G7 cells expressing the inducible MWFE⅐HA were named G7-MWFE⅐HA i .
Northern Analysis-Northern analysis was performed as described previously using a hamster NDUFA1 cDNA probe (3). A probe for human citrate synthase mRNA was made by reverse transcription PCR with oligonucleotide primers CS.F1 (5Ј-ATGGCTTTACTTACTGCGGC-3Ј) and CS.R1 (5Ј-CACATGGGAAGGCAGAGCTG-3Ј), which amplify a 453-bp cDNA beginning with the ATG start codon. Additionally Chinese hamster DDP1 (Timm81a) cDNA was also used as a probe for monitoring RNA loading.
Measurement of Respiratory Activities-The respiratory chain activities of various cells were measured as described previously (4,24). The cells were harvested by trypsinization, collected by centrifugation (350 ϫ g), and resuspended in 1ϫ HSM buffer (20 mM Hepes, pH 7.1, 250 mM sucrose, and 10 mM MgCl 2 ) at a density of 2 ϫ 10 7 cells/ml. Cells were permeabilized by digitonin (100 g/ml) treatment until ϳ90% of the cells could be stained with trypan blue. After ϳ5-min digitonin treatment at 4°C, the cell suspension was diluted 10-fold with HSM buffer, and the cells were harvested by centrifugation. Subsequently, after one wash, cells were resuspended at a density of 3 ϫ 10 7 cells/ml. The total protein content was measured by Bradford microassay, and ϳ5-6 mg of cell suspension was used per assay. Oxygen consumption was measured polarographically with a Clark oxygen electrode in a 2.2-ml metabolic chamber with a water jacket maintained at 37°C. Substrates, inhibitors, etc. could be added via a capillary opening using microsyringes as described previously (4).

Isolation of Mitochondria and Mitochondrial
Fractions-Mitochondria were isolated from cells essentially according to Ref. 25. Approximately 1 ϫ 10 9 cells were washed twice with TD buffer and harvested by trypsinization. The pellets were suspended in 5 ml of SM buffer (50 mM Tris⅐HCl, pH 7.4, 0.25 M sucrose, 2 mM EDTA) and homogenized using a tightly fitting Potter-Elvehjem homogenizer (30 -35 up/down strokes). The homogenate was centrifuged twice at 625 ϫ g for 10 min at 4°C to remove unbroken cells and nuclei. The supernatant was centrifuged at 10,000 ϫ g for 20 min at 4°C. The mitochondrial pellet was suspended in 0.1 ml of SM buffer. This fraction was designated as the mitochondrial fraction.
Immunochemical Assays and Antibodies-The lysis of cells was carried out in buffer (50 mM Tris⅐HCl, pH 7.4, 2 mM phenylmethylsulfonyl fluoride, 2 mM EDTA, with protease inhibitor mixture P8340 from Sigma) by three cycles of freeze-thaw in liquid nitrogen and three bursts of sonication. Protein samples (between 50 and 100 g) were separated by SDS-PAGE and transferred to Immobilon-P (0.45 m) or polyvinylidene difluoride (0.1 m) membranes. Anti-MWFE was used at 1:5000 dilution, whereas the other antibodies were used at 1:1000 dilution. Horseradish peroxidase-conjugated secondary antibodies (anti-rabbit or anti-mouse) were used at 1:5000 dilution, and signals on the immunoblots were detected using an enhanced chemiluminescence system (ECL Plus from Amersham Biosciences).
The anti-MWFE was developed as described previously (4). Anti-51 kDa and anti-PSST (Rhodobacter capsulatus) were generous gifts from Drs. Y. Hatefi and T. Yagi (The Scripps Research Institute). Antiserum against the mammalian SDHC subunit was generously provided by Dr. B. Ackrell (University of California, San Francisco). Sources of other antibodies were as follows. Anti-20 kDa/PSST, anti-30 kDa, anti-39 kDa, and anti-ATP5␤ (all human) were from Molecular Probes; antiporin was from Calbiochem; anti-HA was from Covance BabCo; antimouse and anti-rabbit secondary antibodies were from Bio-Rad and Amersham Biosciences, respectively.
The NADH dehydrogenase assay was carried out using substrate deamino-NADH and nitro blue tetrazolium, an artificial electron acceptor, as described previously (4,27). Gel slices were incubated at room temperature in 2 mM Tris⅐HCl (pH 7.4), 0.1 mg/ml deamino-NADH (generously provided by Dr. T. Friedrich, University of Freiburg), and 2.5 mg/ml nitro blue tetrazolium (Sigma) for 2-4 h. After complex I staining the same gels were washed twice (10 min) with 1.5 mM potassium phosphate buffer (pH 7.4) and stained for complex II activity with a similar protocol except that succinate (84 mM) was used as the substrate in the presence of phenazine methosulfate (0.2 mM), 4.5 mM EDTA, and 10 mM KCN in 1.5 mM phosphate buffer (pH 7.4). The blue background from stained gels was cleared by removing the Coomassie Brilliant Blue G-250 dye electrophoretically using standard Western transfer conditions for ϳ3 h. Elimination of the Coomassie Brilliant Blue G-250 background by transfer did not compromise the quality of stained bands but increased the sensitivity (data not shown).
Quantitative Analyses-The signals on Western blots were quantified either by using phosphorimaging systems (Amersham Biosciences STORM 860 and Bio-Rad GS525) or Kodak imaging systems equipped with ImageQuant, Molecular Analyst, and Kodak 1D softwares.
Other Reagents-Tetracycline-free fetal bovine serum (Tet System Approved) was obtained from Clontech. All other reagents were of the highest grade available.

Selection and Characterization of Cells Expressing MWFE⅐HA
Conditionally-To develop a conditional complex I assembly system, we expressed MWFE⅐HA from a doxycycline-inducible promoter in the previously described MWFE-null cells (CCL16-B2) (3,4). Stable, transfected CCL16-B2 cells, expressing rtTA activator constitutively, were retransfected with the inducible construct pIS2022 i and selected in DME-Gal medium in the presence of different doses of doxycycline (0.01-2.00 g/ml). The parental CCL16-B2 cells die within 24 h in DME-Gal medium, and this condition was used for the selection of double transfectants with the res ϩ phenotype, i.e. with a functional oxidative phosphorylation system (see "Experimental Procedures"). Selected populations or individual clones with desirable features (no background and maximal induction) were used for all the experiments in the present study. Doxycycline was added to the cells growing in DME-Glu, and MWFE⅐HA protein levels were measured by Western blotting after 72 h. A dose-response curve (Fig. 1A) showed that the maximal MWFE⅐HA induction was achieved in the presence of 2 g/ml doxycycline in the medium.
We compared the inducible cell line (B2-MWFE⅐HA i ) with the constitutive cell line (B2-MWFE⅐HA c ) with regard to the maximal MWFE⅐HA protein expression (Fig. 1B). The MWFE⅐HA protein was present only in the presence of doxycycline in B2-MWFE⅐HA i cells where its level appeared to be somewhat lower than that in the B2-MWFE⅐HA c cells. The constitutive and induced transcript levels were found to be comparable in both cell lines. In the absence of doxycycline, the constitutive MWFE⅐HA level was higher because even at 2 g/ml concentration the doxycycline interfered partially with mitochondrial protein synthesis. The inhibitory effect at 2 g/ml was more pronounced in hamster cells than in human HT1080 cells (data not shown). It should be noted that no transcript or protein was detected in the absence of inducer (Fig. 1B). Thus, there was no significant leaky expression in these cells, and the transcription was absolutely dependent on the presence of the doxycycline. The maximal doxycycline dose (2 g/ml) applied during selection of B2-MWFE⅐HA i cells was chosen based on a preliminary growth analysis of wild type (CCL16-B1) cells in DME-Gal medium. We did not find a significant reduction in the growth rate (data not shown). However, a Western blot analysis revealed that the level of MWFE was reduced in the presence of doxycycline relative to its absence. At higher doxycycline doses (10 g/ml or higher), the MWFE⅐HA induction in B2-MWFE⅐HA i cells was inhibited even further, while the level of mRNA was increased relative to that at 2 g/ml doxycycline (Fig. 1C). Upon prolonged culture, some cells appeared in the population in which MWFE⅐HA was constitutively expressed at a level sustaining growth in DME-Gal. The origin of these cells is not quite clear. After recloning, the desired phenotype was stable over a period of months. It should be noted here that the fetal calf serum from normal sources may contain traces of this antibiotic, causing induction, and the clean result observed here was obtained with either dialyzed serum or with a special serum certified to be free of residual antibiotics (Tet Ϫ ) (Clontech).
Time Course of Induction-To find out how quickly the biogenesis of the active complex I takes place in B2-MWFE⅐HA i cells, we monitored the induction kinetics of MWFE⅐HA protein by Western analysis, complex I assembly by BN-PAGE, and its activity by polarography and histochemistry. Since the activator protein (rtTA) was expressed constitutively, it was of interest first to measure the time of appearance of the inducible mRNA following the addition of inducer and second to determine how long it would take for MWFE⅐HA levels to reach steady state. Fig. 2 shows that no transcript was detectable at 0 h, a significant signal was seen at 12 h, and after 12-24 h steady-state levels of mRNA were attained ( Fig. 2A). Citrate synthase mRNA served as a loading control for the total RNA in the Northern blot. The MWFE⅐HA protein appeared at ϳ12 h and reached steady-state levels at or shortly after 24 h. During the entire induction period, the levels of porin and the 39-kDa subunit of complex I remained unchanged relative to total proteins loaded (Fig. 2B). The Western analyses were performed with whole cell extracts to monitor the entire cellular pool of MWFE⅐HA. However, our experiments showed the MWFE⅐HA to be exclusively associated with the mitochondrial membrane fraction at all times during induction, indicating that there was no delay in the import into mitochondria (data not shown).
The induction of complex I activity in B2-MWFE⅐HA i cells was measured by polarography at various times after induction (Fig. 3A). Rotenone-sensitive, glutamate ϩ malate-driven respiration (oxygen consumption) was measured in digitonin-permeabilized cells, and the activity was normalized with the KCN-sensitive succinate ϩ glyceraldehyde 3-phosphate-driven respiration, representing the downstream portion (complexes II-IV) of the electron transport chain. This approach of normalization was adopted because 2 g/ml doxycycline used for induction was partially inhibiting the mitochondrial protein synthesis and hence complex III and IV levels (Figs. 1C and 3C). It was unexpected that the appearance of complex I activity was delayed significantly relative to the attainment of steady-state levels of MWFE⅐HA seen in Western blots with whole cells. An initial small rise after 12 h was followed by a plateau, and a second wave of induction was observed beginning at about 48 h. These kinetics were reproducibly observed in independent experiments (Fig. 3A).
The appearance of complex I assembly was also measured by BN-PAGE and the identification of a ϳ900-kDa complex that can be detected either by the histochemical NADH dehydrogenase  3. A, correlation between the appearance of MWFE⅐HA protein and complex I assembly and activity. The levels of the protein and activity were normalized with porin and the succinate ϩ glycerate 3-phosphate-dependent, cyanide-sensitive respiration (complexes II-IV), respectively. The levels of protein and activity at 96 h of induction were considered as 100% after subtracting the background level values obtained from the "0 h" time point. Experimental error for the same sample is ϳ10%. Kinetics of complex I assembly as measured by BN-PAGE, histochemical assay using deamino-NADH as a substrate (B), and Western blotting (C) are shown. The average relative signals for MWFE⅐HA and 39 kDa on the BN-PAGE/Western blot are plotted in A as CI.assembly. The level of assembled complex V is also shown by ATP5␤. CI, complex I. assay or by Western blotting using anti-HA and anti-39 kDa antibodies. The data are shown in Fig. 3, B and C. Both measurements (polarography and BN-PAGE) indicate that the appearance of mature complex I was delayed significantly after the start of induction. It has been experimentally difficult to determine whether the small discrepancy between the curves describing respiratory activity and histochemical activity (at 900 kDa) is significant (Fig. 3A). The former may require incorporation of an assembled complex I into a supercomplex (28).
A second revealing experiment resembling a pulse-chase experiment was performed as follows. The inducer doxycycline was present for only 24 h ("pulse") followed by a "chase" in the absence of inducer over several days. The MWFE⅐HA levels were monitored in whole mitochondria by SDS-PAGE, while BN-PAGE was used to measure assembled MWFE⅐HA in an intact (900-kDa) complex I. As shown in Fig. 4, at the end of the pulse the total amount of measurable MWFE⅐HA was at or near steady state, while only a small fraction of it had appeared in a 900-kDa complex I. During the chase, total MWFE⅐HA levels were maintained for another 24 h, aided by the continued translation of slowly decaying mRNA; the subsequent decline is due to an arrest in synthesis and a dilution of the existing pool (Fig. 4A). Notably a significant signal for assembled complex I became apparent after 24 h of chase, and it also was diluted and possibly degraded during subsequent cell divisions (Fig.  4B). The assembly of complex I during the first 24 h of chase may also be aided by the release from the partial inhibition of mitochondrial protein synthesis due to the inducer. These results also support the previous conclusions from Fig. 3A. The level of complex I reached under these conditions (24-h pulse/ 24-h chase) is comparable or even elevated compared with continuous induction.
The presence of sufficient functional complex I can also be tested by switching the cells from DME-Glu to DME-Gal at various times after the induction by doxycycline. It was observed that cells survive well only if the switch was made at 48 h of induction or later.
Turnover of Complex I after Removal of Doxycycline-To determine whether MWFE plays a role in complex I maintenance, we carried out "turn-OFF" experiments by removing the doxycycline from the medium after induction of complex I to near steady-state levels. Cells were induced for 96 h and then transferred to medium free of inducer after washing twice with medium without doxycycline. The decays of the transgenic NDUFA1 mRNA, MWFE⅐HA protein, and complex I were measured at various times after the removal of doxycycline. One of the following two outcomes could be anticipated: either MWFE⅐HA decays faster than complex I, or both MWFE⅐HA and complex I decay simultaneously. In the first case, a large (ϳ900-kDa) complex I without MWFE⅐HA would be detectable in BN-PAGE/Western analyses. Northern analysis revealed that the MWFE⅐HA mRNA decayed moderately quickly with an apparent half-life of less than 24 h such that very little transcript remained ϳ24 h after the removal of the inducer (Fig. 5A). MWFE⅐HA levels as measured by Western analysis in whole cell lysates and within complex I decayed slowly over this time period. The decay of rotenone-sensitive complex I activity as detected by polarography was equally slow. The correlation between the decay of MWFE⅐HA protein (total and assembled), complex I, and its activity is shown in Fig. 5B. These observations are in agreement with previous studies suggesting that the half-life of complexes of the electron transport chain is relatively long (29). In particular, the data showed that the MWFE⅐HA subunit decayed in parallel with the complex I and not with significantly faster, independent kinetics. Thus, once incorporated into complex I, the MWFE might be playing a role in complex I stabilization, but its fate is linked to that of the complex as a whole. A more precise quantitative interpretation of these experiments is complicated by the fact that even if induction stopped immediately after withdrawal of the inducer, MWFE⅐HA synthesis from slowly decaying mRNA persists for some time.  24D, 48D, and 72D). B, the same samples as in A were analyzed by BN-PAGE. The signals for the complex I proteins MWFE⅐HA and 39 kDa correspond to the signals at the position of the ϳ900-kDa complex. SDHC is a signal from complex II on the same blot from an antibody against the integral membrane protein, C II-3 , of this complex (loading control).

FIG. 5. Decay kinetics after removal of the inducer doxycycline.
A, Northern blot showing mRNA for citrate synthase and transgenic NDUFA1 mRNA (NDUFA1t). B, rotenone-sensitive respiration (Activity) and its correlation with assembled MWFE⅐HA measured by BN-PAGE (Complex I) and MWFE⅐HA in whole cell extracts (MWFE⅐HA). Cells were induced with 2 g/ml doxycycline for 96 h; the inducer was removed by washing twice with doxycycline-free medium, and incubation was continued for the indicated time period. CS, citrate synthase.
Under the same conditions cell viability declines very slowly, although the doubling time in DME-Gal increases significantly as the existing complex I activity is diluted out. No precise correlation between viability and decline of complex I activity was made.
What Stabilizes MWFE in the Precomplex?-We expressed MWFE⅐HA conditionally in our previously described V79-G7 cells. MWFE levels are severely depleted in V79-G7 cells, which are incapable of mitochondrial translation (20,21). For further investigation this V79-G7 mutant was transfected with a new construct (pTRE ON) (see "Experimental Procedures") capable of expressing both the activator rtTA protein (from a constitutive promoter) and the MWFE⅐HA protein from a polycistronic mRNA transcribed from an inducible promoter. Details on its construction are presented under "Experimental Procedures." A single transfection and selection with G418 allows the conditional expression of MWFE⅐HA in any mammalian cell type. The vector was tested first in CCL16-B2 cells and shown to yield permanently transformed cells behaving exactly like the CCL16-MWFE⅐HA i cells described above (results not shown). Thus, the fate of induced MWFE⅐HA could also be investigated in V79-G7 cells. We could detect the MWFE⅐HA on Western blots from induced whole cell extracts (Fig. 6, left panel). It appeared slowly and reached a plateau after about 48 h. However, the absolute amount was much less compared with that observed in induced B2-MWFE⅐HA i cells (Fig. 6, right panel). The synthesis and levels of the endogenous MWFE were also reinvestigated in these cells. With much higher levels of loading some endogenous MWFE was also detectable, and it appeared that the endogenous MWFE levels decreased as the induced levels of MWFE⅐HA were increased over time. One or more of the ND subunits are required for the accumulation of MWFE⅐HA in the precomplex normally seen after 24 h.
The PSST subunit has been tentatively localized at the junction/neck between the peripheral and integral membrane domains (30,31), and biochemical fractionations have placed MWFE in the same region (2). PSST is present at normal levels in the CCL16-B2 mutant. The stabilization/accumulation of the MWFE and PSST subunits was also investigated in human and mouse mutant cells lacking individual ND subunits (23,32). The accumulation of MWFE depended absolutely on ND4 and ND6, while a low level of MWFE was observed in cells with a mutation in ND5 (Fig. 7). In contrast, it appeared that the presence of normal levels of PSST was strictly dependent on the presence of ND5 and ND6 but independent of the presence of ND4 (Fig. 7). In the ND5 mutants there was also a low level of complex I as shown by immunoprecipitations (23) and by blue native gels (results not shown).
The Fate of the Assembled Complex I in the Absence of Mitochondrial Protein Synthesis-To further explore the relationship between mitochondrial protein synthesis and the stability of complex I and MWFE, experiments were carried out with chloramphenicol in cells expressing MWFE constitutively. The fate of the endogenous MWFE and existing complex I in wild type CCL16-B1 cells and of the MWFE⅐HA in B2-MWFE⅐HA c cells was determined after a 72-h treatment with 100 g/ml chloramphenicol. Both MWFE and MWFE⅐HA protein levels were reduced almost 10-fold compared with untreated control cells (Fig. 8A). A proportional decline in the level of complex I was also observed (data not shown). The reduction in the signal might be expected if complex I biogenesis stops immediately upon inhibition of mitochondrial protein synthesis and the existing complex I is simply diluted by further cell divisions. However, from an estimate of the increase in cell number during the same interval it appeared that the decrease could not be accounted for entirely by dilution. Is it possible that continuous mitochondrial protein synthesis is required for the maintenance of electron transport complexes even after they have been assembled? This leads to the interesting and provocative hypothesis that one or more ND subunits in complex I may turn over faster and need to be continuously replaced.
A more detailed kinetic analysis was performed as follows. B2-MWFE⅐HA i cells were induced for 96 h and then transferred after washing to medium free of inducer. The cells were divided in several aliquots, and one-half of these were treated with 100 g/ml chloramphenicol. The decay of assembled MWFE⅐HA in complex I was monitored by BN-PAGE/Western analysis in both sets. The decay in the chloramphenicol-treated cells compared with untreated control cells was significantly accelerated (Fig. 8B). Since the increase in cell number was less in chloramphenicol-treated cells relative to the controls, a simple dilution effect cannot explain the more rapid decay, suggesting that continuous mitochondrial protein synthesis was required to maintain the assembled MWFE⅐HA and thus complex I. DISCUSSION This study describes the development of an inducible system for the assembly of the mitochondrial NADH-ubiquinone oxidoreductase (complex I). The conditional expression of the MWFE subunit leads to conditional complex I assembly and respiration in transfected CCL16-B2 cells. In the B2-MWFE⅐HA i cells described here there was no detectable expression of the MWFE⅐HA transcript or protein in the absence of inducer. The phenotypes of these cells, 1) ability to grow in DME-Gal, 2) appearance of a stable 900-kDa complex I on BN-PAGE, and 3) glutamate-malate dependent, rotenone-sensitive respiration, were strictly dependent on the presence of the inducer doxycycline. The MWFE protein is playing a crucial role in the assembly process, and the lack of complex I in CCL16-B2 cells is most likely due to interrupted assembly not to the instability of complex I. During the course of these studies it became apparent that even at low concentrations (2 g/ml) the inducer of nuclear transcription also acted as a partial inhibitor of mitochondrial protein synthesis. As a consequence, the two effects must be balanced for optimum results, and future approaches might consider other inducible systems such as the ecdysone-dependent transcriptional activation (33).
An ideal system for studying the MWFE-dependent assembly of complex I would require the instantaneous appearance of sufficient MWFE subunits to initiate complex I biogenesis. However, in the system described here, there is first a build-up of the relevant mRNA during the first 12-24 h accompanied by MWFE⅐HA synthesis leading to steady-state levels of this protein after about 24 h, consistent with the general experience with such Tet ON/OFF systems. The most interesting finding is that the formation of a mature and functional complex I requires a considerably longer time, and the appearance of MWFE⅐HA in the membrane is not a rate-limiting step in the assembly pathway (Fig. 3). The experiments with a pulse of inducer followed by a chase in the absence of inducer (Fig. 4) strongly support this idea. The MWFE⅐HA seen after the pulse on SDS-PAGE with whole mitochondria is incorporated during the 24-h chase into a stable complex I detectable by BN-PAGE. The observations therefore suggest that stable MWFE may appear in a precomplex (assembly intermediate) that is formed relatively rapidly, but its conversion into a mature complex I requires more time. However, no such high molecular weight intermediate containing MWFE⅐HA could be detected by BN-PAGE and Western analyses. It may be unstable under the experimental conditions used.
MWFE is one of the 46 subunits of complex I, raising questions about the synthesis and fate of the other 45 subunits in the MWFE-deficient CCL16-B2 mutant cells. It was shown earlier that at least six subunits from the peripheral membrane subcomplex (for which antisera were available) were present at normal or near normal levels (3). Furthermore these subunits were found to be associated with a mitochondrial membrane fraction obtained by sonication. It is likely that at least some integral membrane proteins are present and sufficient for the association of the peripheral proteins with the membrane. The ESSS protein (p17.3), 10-kDa protein (NDUFA2), and 18-kDa protein (NDUFB6) are present at near normal levels. 3 The proteins accumulated in the absence of MWFE must also include one or more of the mitochondrially encoded subunits since MWFE insertion and/or stabilization after induction are not observed in the V79-G7 mutant cells lacking all ND subunits. In the ND4 and ND6 mutants (23,32) the endogenous MWFE appears to be equally unstable. In contrast, the PSST subunit is missing in ND5 and ND6 mutants and the V79-G7 mutant but not in ND4 mutants. The ND4, ND5, and ND6 subunits are found in different subcomplexes (ND6 in 1␣ and ND4 and ND5 in 1␤) that have been defined from a specific purification protocol (see Refs. 2 and 34). PSST and MWFE are found in the 1␣ subcomplex (2). Do these 1␣ and 1␤ subcomplexes have any relationship to assembly intermediates? Both PSST and MWFE proteins are missing in ND6 mutants, consistent with such a view, but MWFE is absent, while PSST is normal in ND4 mutants, i.e. mutants in which the 1␤ subcomplex is expected to be affected. From such data it becomes apparent that the accumulation and assembly of these (and other) subunits is a coordinated, interdependent process likely involving many more subunits. One can exclude a mechanism in which these subunits are inserted into the membrane independently and maintained there in pools until needed in complex I biogenesis. They may also be subject to rapid turnover when the maturation toward a complete complex I is disturbed by mutations or inhibitors. MWFE and PSST levels may be sensitive indicators of complex I assembly, and it would be worthwhile to utilize this property in the initial characterization of human patients with isolated complex I deficiency especially when mutations in mtDNA have been established.
The ND subunits belong to the "core" of the complex I as defined for prokaryotes, and MWFE is a protein that has been added to the complex during evolution of the eukaryotic lineages. It may not have been unexpected to find the membrane insertion and stabilization of MWFE subunits to be dependent on the presence of ND subunits. A further speculation is that the conversion of the precomplex to an active complex I requires the insertion of additional subunits, including some synthesized in the mitochondrial matrix. This is suggested from a study by Hall and Hare (35) in which the mechanism of complex I assembly was examined in rat hepatoma cells by pulse-chase experiments. These authors found that mitochondrially translated proteins did not appear in an immunoprecipitable holoenzyme until after a particularly long chase, and the suggestion was made that a "scaffolding or nucleation complex" of nuclear encoded proteins was required before the ND proteins could be "inserted into the enzyme." In a related experiment it was shown that the ND proteins were not assembled when cytosolic protein synthesis was prevented by cycloheximide, and they were apparently degraded by an ATPdependent protease.
It remains to be established which NDs are inserted before MWFE and which are inserted after the addition of MWFE to a precomplex. One can entertain several hypotheses. 1) One or more critical ND subunits are being continuously made, but they are highly unstable when complex I assembly cannot proceed in the absence of MWFE. 2) Their synthesis is coupled to and dependent on simultaneous insertion into the membrane and into an assembly intermediate containing other complex I subunits. In other words, a partially assembled precomplex containing some nuclear subunits acts as receptor or docking site for the selective, co-translational insertion of (some) ND subunits. It is not clear whether MWFE is an obligatory component of such an import mechanism or whether 3 P. Potluri, unpublished observations. the subsequent fate of these subunits depends on the presence of MWFE (possibly in another, distinct precomplex). Such a mechanism may also include assembly factors (chaperones) that are transiently associated during complex I assembly (36,37).
Very interesting and more speculative is the suggestion by Carroll et al. (34) that the recently identified nuclear encoded B14.7 subunit (NDUFA11) of complex I has some homology to Tim proteins. While Tim proteins are thought to be associated with protein import from the cytosol, it is plausible that a related mechanism must be operating for proteins made in the mitochondrial matrix. One general protein that has been identified to be essential for the mitochondrial export translocase is OxaI (38,39). Additional membrane-associated and transcriptspecific proteins have been demonstrated to be required for the translation and membrane insertion of mtDNA-encoded Cox subunits in cytochrome oxidase (complex IV) in yeast (40,41).
The emerging hypothesis is that the assembly of the integral membrane subcomplex involves more than one precomplex, but the composition of each intermediate remains to be more precisely defined. It seems less likely that assembly occurs at one site with subunits added one at a time in a precise order. One of these precomplexes formed in the absence of MWFE could provide the contact site on which the peripheral membrane subcomplex is assembled as recently described by Antonicka et al. (42). These authors reported on the status of complex I assembly in several patients with mitochondrial disease and partial complex I deficiencies. By two-dimensional BN/SDS gel electrophoresis several subcomplexes with peripheral membrane subunits were identified, and a similar pattern of subcomplexes was observed in the diverse patients. It was suggested that these subcomplexes are intermediates in the assembly of the holoenzyme complex in contrast to the assembly model proposed for N. crassa (43). As a note of caution it should be mentioned that it is difficult to distinguish between assembly intermediates and breakdown products due to instability introduced by mutations (under conditions of solubilization for BN-PAGE).
In the decay experiments following the removal of inducer (after allowing complex I assembly to a certain level), we found that assembled MWFE⅐HA protein disappeared quite slowly (Fig. 5B). Under the experimental conditions described here, at no time point did we detect intermediates of complex I lacking MWFE⅐HA. Thus, the decays of the MWFE subunit and complex I are inseparable. The single transmembrane helix of MWFE is likely to be deeply embedded in the large bundle of the other (ϳ60) transmembrane helices rather than being loosely associated at the periphery of the bundle. The hydrophilic domain must also interact with hydrophilic domains of other subunits (4), but it is still an open question whether it has simply a stabilizing structural role or is a critical participant in the reaction mechanism.
Finally the relatively rapid disappearance of assembled MWFE⅐HA and complex I in the presence of chloramphenicol could be interpreted in terms of a model in which at least one or more of the ND subunit(s) may be dynamically exchanging in and out of the assembled complex I and hence subject to rapid turnover. Therefore, to keep the complex I stable, continuous mitochondrial protein synthesis would be required to replenish the ND subunit(s) needed for complex I maintenance. These conclusions require further experimental support, but it should be noted that other investigators reported heterogeneous degradation rates for inner membrane polypeptides, including ND subunits, extending to those within the same respiratory complex (29,35).
In summary, a novel model system is presented that has provided observations on the intricacies of complex I assembly.
It promises to yield further insights in future studies with additional tools/antisera and experimental approaches to this challenging problem. It should also find applications in the study of other mitochondrial properties and functions. The isogenic, induced, and uninduced B2-MWFA⅐HA i cells can serve in studying the role of complex I (and respiration) in the assembly of the rest of the electron transport chain, apoptosis, the production of reactive oxygen species, mitochondrial calcium storage/accumulation, and other problems related to mitochondria such as mitochondrial morphological changes.