A two-step mitochondrial import pathway couples the disulfide relay with matrix complex I biogenesis

Abstract

Mitochondria critically rely on protein import and its tight regulation. Here, we found that the complex I assembly factor NDUFAF8 follows a two-step import pathway linking IMS and matrix import systems. A weak targeting sequence drives TIM23-dependent NDUFAF8 matrix import, and en route, allows exposure to the IMS disulfide relay, which oxidizes NDUFAF8. Import is closely surveyed by proteases: YME1L prevents accumulation of excess NDUFAF8 in the IMS, while CLPP degrades reduced NDUFAF8 in the matrix. Therefore, NDUFAF8 can only fulfil its function in complex I biogenesis if both oxidation in the IMS and subsequent matrix import work efficiently. We propose that the two-step import pathway for NDUFAF8 allows integration of the activity of matrix complex I biogenesis pathways with the activity of the mitochondrial disulfide relay system in the IMS. Such coordination might not be limited to NDUFAF8 as we identified further proteins that can follow such a two-step import pathway.

Graphical abstract

Figure 1.

NDUFAF8 is dually localized to IMS and matrix. (A) Domain layout of NDUFAF8. NDUFAF8 contains four highly conserved cysteines in a twin CX₉C motif. These cysteines reside in two alpha-helices. Four arginine residues and no negatively charged amino acid are found in the N-terminal 19 amino acids of NDUFAF8. According to TargetP, this region does not serve as MTS; however a sliding TargetP score prediction indicates a certain propensity of this region to serve as MTS. MTS, mitochondrial targeting signal; aa, amino acid. (B) Immunofluorescence analysis to localize NDUFAF8. NDUFAF8-HA localizes to mitochondria. HEK293 cells stably and inducibly expressing the indicated NDUFAF8-HA variants were fixated, permeabilized, and stained using a primary antibody against the HA epitope (HA) and mitotracker. Cells were analyzed by fluorescence microscopy. Bar corresponds to 20 µm. (C) Submitochondrial fractionation to detect the localization of NDUFAF8-HA. Mitochondria isolated from HEK293 cells expressing NDUFAF8-HA were exposed to three different buffer conditions, an isotonic buffer which leaves the OMM intact, a hypotonic buffer (hypo), which leads to swelling of the matrix and thereby disruption of the OMM, and a TX-100 containing buffer that solubilizes mitochondria completely. Afterward, mitochondria were treated with proteinase K (PK). TOM20, MIA40/CHCHD4, and NDUFAF5 served as control proteins localizing to OMM, IMS, and matrix, respectively. The signal of NDUFAF8 becomes weaker in the IMS fraction (lane 4) and disappears in the matrix fraction (lane 6) indicating NDUFAF8 resides in IMS and matrix as a dually localized protein. (D) Split-GFP assay to detect the localization of NDUFAF8. GFP can be split into two non-fluorescent parts, a larger part (GFP1-10) and a smaller part (GFP11). If these parts come together in the same compartment, they can self-reassemble to reconstitute a fluorescent GFP. GFP1-10 was equipped either with an MTS for the matrix (SU9^MTS) or for the IMS (SCO2^MTS). GFP11 was C-terminally fused to full-length NDUFAF8 and as controls for matrix, IMS, and cytosol to SOD2, MIA40/CHCHD4, and DHFR, respectively. Combinations of GFP11 and GFP1-10-containing constructs were transfected and analyzed by fluorescence microscopy. For SOD2 and MIA40/CHCHD4, fluorescence could only be observed for matrix and IMS respectively (see “split GFP” signal). DHFR-GFP11 co-expression did not result in the reconstitution of GFP with any of the mitochondria localized GFP1-10s. In the case of NDUFAF8, GFP reassembled for both SU9^MTS-GFP1-10 and SCO2^MTS-GFP1-10, indicating NDUFAF8 to be a protein localized to IMS and matrix. Bar corresponds to 20 µm. (E) Model. NDUFAF8 is dually localized to IMS and matrix. Source data are available for this figure: SourceData F1.

Figure S1.

(Related to Figs. 1 and 8). NDUFAF8 is representative of a class of dually localized proteins. Lack of an antibody against endogenous NDUFAF8 necessitated the use of HA-tagged NDUFAF8 throughout this study. A–D assess whether the tag influences import kinetics and (sub)-mitochondrial localization of NDUFAF8 and the similar behaving protein CHCHD1. E–F show the complete data sets for the split-GFP assays performed in this study (Figs. 1 and 8). (A) Submitochondrial fractionation to detect the localization of NDUFAF8 and NDUFAF8-HA after in organello import into mitochondria isolated from HEK293 cells. Experiment was performed as described in Fig. 2 G. NDUFAF8 and NDUFAF8-HA behave similarly in this assay, indicating that the HA tag does not influence the distribution of NDUFAF8 within mitochondrial subcompartments. (B) In organello import assay with NDUFAF8 and NDUFAF8-HA. Experiment was performed as described in Fig. 2 A. NDUFAF8 and NDUFAF8-HA behave similarly in this assay indicating that the HA tag does not influence the import kinetics of NDUFAF8. (C) Submitochondrial fractionation to detect the localization of CHCHD1 and CHCHD1-HA in mitochondria isolated from HEK293 cells. Experiment was performed as described in Fig. 1 C. CHCHD1 and CHCHD1-HA behave similarly in this assay, indicating that the HA tag does not influence the distribution of CHCHD1 within mitochondrial subcompartments. (D) Immunofluorescence analysis to localize CHCHD1 and CHCHD1-HA. CHCHD1 and CHCHD1-HA localize to mitochondria. Experiment was performed as described in Fig. 1 B. The HA tag did not influence the mitochondrial localization of CHCHD1. (E) Split-GFP assay to detect the localization of NDUFAF8. GFP can be split into two non-fluorescent parts, a larger part (GFP1-10) and a smaller part (GFP11). If these parts come together in the same compartment they can self-reassemble to reconstitute a fluorescent GFP. GFP1-10 was equipped either with an MTS for the matrix (SU9^MTS) or for the IMS (SCO2^MTS). GFP11 was C-terminally fused to full-length NDUFAF8 and as controls for matrix, IMS, and cytosol to SOD2, MIA40/CHCHD4, and DHFR, respectively. Combinations of GFP11 and GFP1-10-containing constructs were transfected and analyzed by fluorescence microscopy. For SOD2 and MIA40/CHCHD4, fluorescence could only be observed for matrix and IMS, respectively (see “split GFP” signal). DHFR-GFP11 coexpression did not result in the reconstitution of GFP with any of the mitochondria-localized GFP1-10s. In the case of NDUFAF8, GFP reassembled for both SU9^MTS-GFP1-10 and SCO2^MTS-GFP1-10, indicating NDUFAF8 to be a protein localized to IMS and matrix. DAPI stain and merge serve as orientation. Bar corresponds to 20 µm. (F) Split-GFP assay to detect the localization of different disulfide relay substrates. GFP1-10 was equipped either with an MTS for the matrix (SU9^MTS) or for the IMS (SCO2^MTS). GFP11 was C-terminally fused to full-length NDUFAF8 (AF8), CHCHD2 (D2), CHCHD1 (D1), CHCHD10 (D10), or COA6 isoform 2. Combinations of GFP11 and GFP1-10-containing constructs were transfected and analyzed by fluorescence microscopy. For all selected disulfide relay substrates, GFP reassembled indicating them to be proteins localized to IMS and matrix. DAPI stain and merge serve as orientation. Bar corresponds to 20 µm. Source data are available for this figure: SourceData FS1.

Figure 2.

NDUFAF8 relies on its weak MTS for import into the matrix. (A) In organello import assay with NDUFAF8. In vitro translated radioactive NDUFAF8 was incubated with mitochondria isolated from HEK293 cells. Non-imported proteins were removed by treatment with Proteinase K. An import reaction was performed on mitochondria treated with CCCP to dissipate the mitochondrial membrane potential (–ΔΨ). Imported proteins were analyzed by reducing SDS-PAGE and autoradiography. Signals were quantified using ImageQuantTL, and the amount of imported protein was plotted. NDUFAF8 can be imported into mitochondria and relies on the mitochondrial membrane potential for import. N = 3 replicates; error bars indicate SD. (B) In organello import assay with NDUFAF8-DHFR fusion constructs. Experiment was performed as described in A. To test the capacity of the N-terminal 19 amino acids of NDUFAF8 (AF8^MTS) to serve as MTS, they were fused to the cytosolic proteins DHFR or DHFR^mut. The latter protein carries mutations that prevent it from stable folding and thereby enable mitochondrial import also by weaker MTS. AF8^MTS facilitated the import of both forms of DHFR albeit with different efficiencies. DHFR alone was not imported into mitochondria. N = 3 replicates; error bars indicate SD. (C) In organello import assay with NDUFAF8 to test for its dependence on the TIM23 import pathway. The experiment was performed as described in A. To test for the dependence of NDUFAF8 on the TIM23 channel, mitochondria were incubated with 100 µM Mitoblock-10 (MB-10) to inhibit this pathway. NDUFAF8 import into mitochondria was strongly impaired. N = 1–3 replicates; error bars indicate SD. (D) In organello import-BN-PAGE assay with NDUFAF8-variants to analyze the binding of the DHFR-fused NDUFAF8 variants in the TOM-TIM23 supercomplex during import. The indicated radiolabeled NDUFAF8-DHFR variants and as control the cytochrome b₂ (1-167)-DHFR fusion construct were imported into wild-type yeast mitochondria. The import was performed in the absence of the presence of methotrexate (MTX) as indicated. Protein complexes were analyzed by BN-PAGE and autoradiography. NDUFAF8-DHFR but not the extended precursor fusion constructs established the TOM-TIM23 supercomplex during import. N = 4 replicates. (E) In organello import assay with NDUFAF8 to test for its dependence on the disulfide relay import pathway. Experiment was performed as described in D. To test for the dependence of NDUFAF8 on the mitochondrial disulfide relay, import into mitochondria isolated from wild-type or MIA40 knockdown (MIA40 KD) was analyzed. COX19 served as a control protein that is highly dependent on MIA40/CHCHD4. NDUFAF8 import into mitochondria was not affected by lack of MIA40/CHCHD4. N = 3 replicates; error bars indicate SD. (F) In organello import assay with NDUFAF8-variants to test for its dependence on its cysteines. Experiment was performed as described in A. COX19 served as a control protein that is highly dependent on its cysteine residues for import. To test for the dependence of NDUFAF8 on its cysteines, a wild-type and a cysteine-to-alanine mutant of NDUFAF8 (NDUFAF8-4CA, mutation of all four conserved cysteines) were compared. Initial NDUFAF8 import appears to be independent of its cysteines, but amounts are lower after 8 min of import. Wild-type quantification are the data shown in A. N = 3 replicates; error bars indicate SD. (G) Submitochondrial fractionation to detect the localization of NDUFAF8 after in organello import into mitochondria isolated from HEK293 cells. Experiment was performed as described in A and Fig. 1 C. The signal of NDUFAF8 becomes weaker in the IMS fraction (lane 4) and disappears in the matrix fraction (lane 6) indicating NDUFAF8 resides in IMS and matrix as a dually localized protein. Conversely, NDUFAF8 equipped with a strong SU^MTS does only disappear in the matrix fraction indicating its sole localization to the matrix. (H) Model. NDUFAF8 is imported into the matrix independently of the mitochondrial disulfide relay but “detours” through the IMS because of its weak MTS. Source data are available for this figure: SourceData F2.

Figure S2.

(Related to Fig. 2). NDUFAF8 import does not rely on OMM receptors. (A) In organello import assay with NDUFAF8 into mitochondria devoid of OMM proteins facing the cytosol. Experiment was performed as described in Fig. 2 A except that mitochondria were pretreated with proteinase K (PK) to remove surface receptors at the OMM. Import of NDUFAF8 into mitochondria was thereby not affected. N = 2 biological replicates. (B) In organello import assay with NDUFAF8-DHFR fusion constructs. Data are from Fig. 2 B. Quantification also includes the data for the SU9MTS-DHFR construct. N = 3 replicates; error bars indicate SD. Source data are available for this figure: SourceData FS2.

Figure 3.

The weak NDUFAF8 MTS allows the mitochondrial disulfide relay to introduce disulfide bonds into NDUFAF8 en route to the matrix. (A) Modeled structure of NDUFAF8 highlighting two disulfide bonds between the four conserved cysteines in NDUFAF8. (B) Protein levels in HEK293 cell lines expressing different NDUFAF8 variants. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using ImageLab, and the amount of protein was plotted. NDUFAF8 lacking its four conserved cysteines is present at very low levels. N = 3 replicates; error bars indicate SD. (C) Assessment of MIA40/CHCHD4–NDUFAF8 interaction. NDUFAF8-HA was immunoprecipitated (IP) under native and denaturing conditions after stopping thiol-disulfide exchange reactions by NEM incubation. Precipitates were tested for MIA40/CHCHD4, HA, and as negative control the IMS protein CPOX by reducing SDS-PAGE and immunoblotting. 10% of the total lysate was loaded as input control for the HA blot while 2.5% input was loaded for the MIA40/CHCHD4 and CPOX blots. Under both precipitation conditions, NDUFAF8-HA coprecipitates MIA40/CHCHD4 indicating both proteins to interact via a covalent interaction. (D) Assessment of MIA40/CHCHD4–NDUFAF8 interaction during mitochondrial import. In vitro translated radioactive NDUFAF8 was incubated with mitochondria isolated from HEK293 cells expressing Strep-MIA40/CHCHD4. After import, MIA40/CHCHD4 was affinity-purified (AP) using streptactin beads under native conditions. Precipitates were analyzed by reducing (+DTT) and non-reducing SDS-PAGE and autoradiography. 0.25% of the total lysate was loaded as input control. During import NDUFAF8-HA and MIA40/CHCHD4-Strep form a disulfide linked complex that is reduced by the addition of reductant. (E) Assessment of MIA40/CHCHD4–NDUFAF8 interaction during mitochondrial import. In vitro translated radioactive NDUFAF8-WT and NDUFAF8-Y32A (mutant of the potential MISS/ITS site) were incubated with mitochondria isolated from HEK293 cells expressing MIA40/CHCHD4-Strep or MIA40/CHCHD4-F68E-Strep. After import, MIA40/CHCHD4 was affinity purified using streptactin beads (AP) under native conditions. Precipitates were analyzed by reducing (+DTT) and non-reducing SDS-PAGE and autoradiography. 0.25% of the total lysate was loaded as input control. During import, NDUFAF8-HA and MIA40/CHCHD4-Strep form a disulfide-linked complex only when the MISS/ITS site in NDUFAF8 is intact. (F) Redox state analysis of NDUFAF8. To test for the redox state of NDUFAF8-HA, cells were lysed and either treated with the strong reductant TCEP (lanes 1 and 3) or left untreated (lane 2). Then lysates were incubated with the maleimide mmPEG₁₂ that modifies free thiols but not thiols in disulfide bonds as indicated. Lysates were analyzed by SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of reduced and oxidized protein was plotted. NDUFAF8-HA is mainly present in the reduced state. Approximately 20% of the protein is oxidized at steady state. N = 3 replicates; error bars indicate SD. (G) Redox state analysis of NDUFAF8 over time. Experiment was performed as in E except that cells were pretreated with the ribosome inhibitor emetine before the redox state determination. Over time, the fraction of oxidized NDUFAF8-HA increases under these conditions indicating either further oxidation of NDUFAF8 or specific degradation of the reduced form of the protein. N = 3 replicates; error bars indicate SD. (H) Assessment of the MIA40/CHCHD4–SU9^MTS-NDUFAF8 interaction. NDUFAF8-HA and SU9^MTS-NDUFAF8-HA were immunoprecipitated (IP) under native conditions after stopping thiol-disulfide exchange reactions by NEM incubation. Precipitates were tested for MIA40/CHCHD4 and HA. 10% of the total lysate was loaded as input control for the HA blot while 2.5% input was loaded for the MIA40/CHCHD4 blot. While NDUFAF8-HA coprecipitates with MIA40/CHCHD4, SU9^MTS-NDUFAF8-HA cannot interact with MIA40/CHCHD4. (I) Redox state analysis of NDUFAF8-variants. Experiment was performed as in E except that cells expressing either NDUFAF8-HA or SU9^MTS-NDUAF8 were analyzed. Equipping NDUFAF8 with a strong MTS (SU9^MTS) results in a completely reduced protein at steady state, indicating that the weak MTS of NDUFAF8 is required to allow at least partial disulfide bond formation to occur. N = 3 replicates; error bars indicate SD. (J) In organello import assay with preoxidized or reduced NDUFAF8 into mitoplasts (mitochondria without the OMM). Experiment was performed as described in Fig. 2 A for 8 min import time. Preoxidized NDUFAF8 can be imported into mitoplasts. N = 2 replicates; error bars indicate SD. PK, proteinase K. (K) Model for the two-step import of NDUFAF8. The weak MTS of NDUFAF8 allows for TIM23-dependent import of NDUFAF8 into the mitochondrial matrix. It also allows for interaction with the mitochondrial disulfide relay component MIA40/CHCHD4 in the IMS that introduces two disulfide bonds into a fraction of NDUFAF8 molecules. A stronger MTS would not allow this interaction and would result in the accumulation of completely reduced NDUAF8 in the matrix. Source data are available for this figure: SourceData F3.

Figure 4.

Mitochondrial proteases monitor NDUFAF8 levels in IMS and matrix depending on their redox state. (A) Protein levels in HEK293 cell lines expressing different NDUFAF8 variants. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. NDUFAF8-HA-variants lacking its four conserved cysteines or equipped with a SU9^MTS are present at very low levels. N = 3 replicates; error bars indicate SD. (B) Assessment of stability of different NDUFAF8 variants in HEK293 cells. Cells were pretreated with the ribosome inhibitor emetine for the indicated times and then lysed. Lysates were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. NDUFAF8-HA-variants lacking its four conserved cysteines or equipped with a SU9^MTS are very unstable compared to NDUFAF8-HA. N = 3 replicates; error bars indicate SD. (C) Assessment of stability of different NDUFAF8 variants after import into isolated mitochondria. In vitro-translated radioactive NDUFAF8-variants were incubated with mitochondria isolated from HEK293 cells. Non-imported proteins were removed by treatment with Proteinase K. An import reaction was performed into mitochondria treated with CCCP and valinomycin to dissipate the mitochondrial membrane potential (−ΔΨ). Imported proteins were analyzed by reducing SDS-PAGE and autoradiography. Signals were quantified using ImageQuantTL and the amount of imported protein was plotted. NDUFAF8-variants lacking its four conserved cysteines or equipped with a SU9^MTS are very unstable compared to NDUFAF8. N = 3 replicates; error bars indicate SD. (D) Protein levels in HEK293 wild-type and CLPP knockout cell lines expressing different NDUFAF8 variants. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab and the amount of protein was plotted. NDUFAF8-HA-variants lacking its four conserved cysteines or equipped with a SU9^MTS were stabilized by the loss of CLPP. This was not the case for wild-type NDUFAF8. This indicates that CLPP degrades reduced NDUFAF8. N = 3 replicates; error bars indicate SD. (E) Protein levels in HEK293 wild-type and YME1L knockout cell lines expressing NDUFAF8-HA. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. NDUFAF8-HA levels are increased in YME1L knockout cells. N = 3 replicates; error bars indicate SD. (F) Assessment of stability of NDUFAF8 in HEK293 wild-type and YME1L knockout cells. Cells were pretreated with the ribosome inhibitor emetine for the indicated times and then lysed. Lysates were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. NDUFAF8-HA became stabilized by the loss of YME1L. N = 3 replicates; error bars indicate SD. (G) Protein levels in HEK293 wild-type and YME1L knockout cell lines expressing different NDUFAF8 variants. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. NDUFAF8-HA and NDUFAF8-4CA-HA but not SU9^MTS-NDUFAF8-HA were present at increased levels in YME1L knockout cells. Thus, only NDUFAF8 variants that are exposed to the IMS become stabilized by the loss of YME1L. N = 3 replicates; error bars indicate SD. (H) Redox state analysis of NDUFAF8 in HEK293 wild-type and YME1 knockout cells. Experiment performed as described in Fig. 3 F. Adenylate kinase 2 (AK2) served as control for an IMS protein with a disulfide bond. NDUFAF8-HA is mainly present in the oxidized state in YME1L knockout cells. This indicates that YME1L targets oxidized NDUFAF8 in the IMS. N = 3 replicates; error bars indicate SD. (I) Model for protease surveillance of NDUFAF8 import. In the wild-type situation, NDUFAF8 becomes at least in part oxidized by the mitochondrial disulfide relay. A large fraction of the oxidized protein is constantly degraded by YME1L, presumably because import mediated by the weak MTS of NDUFAF8 is slow and accumulation of oxidized NDUFAF8 in the IMS has to be prevented. If no oxidation of NDUFAF8 takes place, reduced NDUFAF8 is targeted by CLPP in the matrix, strongly decreasing the half-life of reduced NDUFAF8. Source data are available for this figure: SourceData F4.

Figure S3.

(Related to Fig. 5). NDUFAF8 or NDUFAF5 loss results in an isolated complex I deficiency. (A) Strategy to generate NDUFAF8 knockout HEK293 cell lines using CRISPR Cas. A guide (#3) directed against the first exon of NDUFAF8 gave rise to multiple clones. Successful targeting of the gene was confirmed by immunoblotting against NDUAF5 (due to the lack of a suitable antibody against NDUFAF8) and by sequencing. (B) Strategy to generate NDUFAF5 knockout HEK293 cell lines using CRISPR Cas. A guide (#1) directed against the first exon of NDUFAF5 gave rise to multiple clones. Successful targeting of the gene was confirmed by immunoblotting against NDUAF5 and by sequencing. (C) Effects of NDUFAF5 and NDUFAF8 knockout cells. Levels of proteins were assessed using immunoblotting and quantitative label-free mass spectrometry. Subunits of complex I but not of other respiratory chain complexes are present in lowered amounts in both knockout cell lines indicating an isolated complex I deficiency. N = 4 biological replicates, an unpaired one-sample two-sided Student’s t test was applied (P < 0.05, fold change > 2). Source data are available for this figure: SourceData FS3.

Figure 5.

The NDUFAF8–NDUFAF5 interaction stabilizes and activates NDUFAF5 to allow NDUFAF5 to fulfill its function in complex I assembly. (A) Complex I subunit levels in NDUFAF5 and NDUFAF8 knockout cells. Levels of indicated proteins were assessed using immunoblotting. Subunits of complex I are present in lowered amounts in both knockout cell lines. This is in line with data from quantitative label-free mass spectrometry (Fig. S3, C and D). Moreover, NDUFS5 and NDUFV2 levels can be complemented by re-expressing NDUFAF5 and NDUFAF8 in the respective knockout cells. N = three biological replicates. (B) Assessment of NDUFAF5–NDUFAF8 interaction. NDUFAF8-HA was immunoprecipitated (IP) under native conditions. Precipitates were tested for NDUFAF5, HA (NDUFAF8), and, as negative control, the protein PDH by reducing SDS-PAGE and immunoblotting. 10% of the total lysate was loaded as input control for HA blot and 2.5% was loaded as input for the NDUFAF5 input control. NDUFAF8-HA coprecipitates NDUFAF5. (C) Proteomic analysis to assess the interactomes of NDUFAF5 and NDUFAF8. HEK293 cells expressing either NDUFAF5-HA or NDUFAF8-HA were lysed, proteins were immunoprecipitated using the HA-tag, and precipitates were analyzed using quantitative label-free proteomics. Both NDUFAF8 and NDUFAF5 coprecipitate the respective other partner as well as subunits of complex I but not of other complexes of the respiratory chain. Remarkably, NDUFAF5 precipitates almost all subunits of the Q-module which contains NDUFS7, the protein on which NDUFAF5 acts during complex I maturation (highlighted in blue). This might indicate that NDUFAF5 acts on the complete Q-module and not on monomeric NDUFS7 during complex I assembly (Guerrero-Castillo et al., 2017). N = 4–5 biological replicates, an unpaired one-sample two-sided Student’s t test was applied (P < 0.05, fold change >2). (D) Protein levels in HEK293 wild-type cells, NDUFAF8 knockout cells, and NDUFAF8 knockout complemented with NDUFAF8 and NDUFAF5 (expression for 3 d). Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. The loss of complex I subunits in NDUFAF8 knockout cells cannot be complemented by re-expressing NDUFAF5-HA despite strongly increased NDUFAF5 levels. This indicates that the role of NDUFAF8 is not solely in NDUFAF5 stabilization but also in activation. N = 3 replicates. (E) Assay to assess growth of NDUFAF8 knockout cell lines in galactose-containing medium. Cells were seeded in glucose-containing medium, and after 1 d, medium was exchanged to a galactose-containing medium, and cell number was scored daily. Growth on galactose necessitates a functional respiratory chain. NDUFAF8 knockout cells and NDUFAF8 knockout cells complemented with NDUFAF5 were not capable to grow on galactose. N = 3 replicates; error bars indicate SD. (F) Predicted structure of the NDUFAF5-NDUFAF8 complex by Alpha Fold Multimer. For the prediction, ChimeraX software was run in AlphaFold 2. The complex structure was visualized by Pymol. NDUFAF8 and NDUFAF5 appeared to have a very extended interaction interface that spans one complete face of NDUFAF8. NDUFAF8 “snuggles” in an L-shaped conformation onto NDUFAF5. Two regions mediate the interaction between NDUFAF5 and NDUFAF8. In region 1, two perpendicular helices of NDUFAF5 and NDUFAF8 come in close proximity, and small aliphatic amino acids face each other in this region. Region 2 encompasses the MTS of NDUFAF8. Specifically, two arginine residues (R12 and R16 of NDUFAF8) are involved in contact with a glutamate residue and a phenylalanine in NDUFAF5 (E256 and F68 in NDUFAF5). Notably, NDUFAF8 and the other interaction partner of NDUFAF5, PYURF, appear to occupy distinct interaction sites on NDUFAF5 that are distal from each other excluding interference of these two proteins in their function (Pei et al., 2022). (G) Protein levels in NDUFAF8 knockout cells complemented with SU9^MTS-NDUFAF8 and different MTS mutant variants of SU9^MTS-NDUFAF8. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. Removing the MTS of NDUFAF8 or mutating the arginine residues important for interaction with NDUFAF5 impairs complementation. N = 3 replicates. (H) Model. NDUFAF8 is critical for NDUAF5 stability and activity. Interaction of NDUFAF8 and NDUFAF5 involving arginine residues in the MTS of NDUFAF8 is required for the functionality of NDUFAF5 and proper assembly of the Q module of complex I. Source data are available for this figure: SourceData F5.

Figure S4.

(Related to Fig. 5). NDUFAF8 or NDUFAF5 interact with subunits of the mitochondrial ribosome. Proteomic analysis to assess the interactomes of NDUFAF5 and NDUFAF8. HEK293 cells expressing either NDUFAF5-HA or NDUFAF8-HA were lysed, proteins were immunoprecipitated using the HA-tag, and precipitates were analyzed using quantitative label-free proteomics. Both NDUFAF8 and NDUFAF5 coprecipitate mitochondrial ribosomal subunits which might indicate a close coordination of their role in Q module assembly and ND1 synthesis. N = 4 biological replicates, an unpaired one-sample two-sided Student’s t test was applied (P < 0.05, fold change > 2).

Figure 6.

NDUFAF8 levels determine the efficiency of NDUFAF5 activation and complex I assembly. (A) Protein levels in NDUFAF8 knockout cells complemented with NDUFAF8 variants. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab and the amount of protein was plotted. Equipping NDUFAF8 with a SU9^MTS allows complementation of NDUFAF8 knockout cells. Mutating the four conserved cysteines in NDUFAF8 (4CA) resulted in very low steady-state levels of NDUFAF8 and did not allow complementation of the NDUFAF8 knockout. N = 3 replicates; error bars indicate SD. (B) Assay to assess the growth of NDUFAF8 knockout cell lines in a galactose-containing medium. Cells were seeded in glucose-containing medium, and after 1 d, medium was exchanged to a galactose-containing medium and cell number was scored every 6 h NDUFAF8 knockout cells and NDUFAF8 knockout cells complemented with NDUFAF8-4CA were not capable to grow on galactose. N = 4 replicates; error bars indicate SD. (C) Protein levels in NDUFAF8 knockout cells complemented with NDUFAF8 variants. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab and the amount of protein was plotted. Equipping NDUFAF8-4CA with a SU9^MTS allows complementation of NDUFAF8 knockout cells indicating that the cysteine of NDUFAF8 is dispensable for its function. Mutating the cysteines impacts the stability of NDUFAF8 in the matrix and only an efficient import of NDUFAF8-4CA by the SU9^MTS ensures a sufficiently high supply of new NDUFAF8-4CA to overcome the high degradation rate of the protein in the matrix. N = 3 replicates; error bars indicate SD. (D) Assay to assess growth of complemented NDUFAF8 knockout cell lines in galactose-containing medium. Cells were seeded in glucose-containing medium, and after 1 d, medium were exchanged to galactose-containing medium and cell number was scored every 6 h NDUFAF8 knockout cells complemented with SU9^MTS-NDUFAF8-4CA were able to grow on galactose. N = 4 replicates; error bars indicate SD. (E) Assay to assess the growth of complemented NDUFAF8 knockout cell lines in a galactose-containing medium. Cells were seeded in glucose-containing medium, and after 1 d, medium were exchanged to galactose-containing medium and cell number was scored every 6 h. Titration of NDUFAF8 variant levels was achieved by induction of their expression with different amounts of doxycycline. Upon reaching a certain threshold, knockout cells complemented with NDUFA8 or SU9^MTS-NDUFAF8-4CA were able to grow on galactose. SU9^MTS-NDUFAF8-4CA expressing cells required increased amounts of doxycycline to allow growth on galactose. N = 4 replicates; error bars indicate SD. (F) Model. The amounts of NDUFAF8 control NDUFAF5 levels and complex I assembly. Low amounts of NDUFAF8 as found in the NDUFAF8 knockout or in NDUFAF8 knockout cells expressing NDUFAF8-4CA do not activate and stabilize NDUFAF5 thereby preventing Q module assembly. High amounts of NDUFAF8 can complement NDUFAF8 knockout irrespective of whether they contain disulfide bonds or not. The formation of disulfide bonds in NUFAF8 is, however, important to stabilize the protein in the matrix. Source data are available for this figure: SourceData F6.

Figure S5.

(Related to Fig. 5). Translation of ND1 is impaired in NDUFAF5 and NDUFAF8 knockout cells. (A) Assessment of mitochondrial translation in different knockout cell lines. Cells were treated with emetine to block cytosolic translation. Under these conditions, only the 13 proteins encoded in the mitochondrial genome are synthesized. Their synthesis was assessed by radioactive pulse-labeling. Subsequently, stability of the mitochondrial translation products was assessed using chase times of 1 and 2 h, respectively. Cells were lysed and proteins were analyzed by reducing SDS-PAGE and autoradiography. Except for ND1 no other mitochondrial translation product was affected by the loss of NDUFAF5 (AF5 KO) or NDUFAF8 (AF8 KO). ND1 was completely missing even directly after the pulse period. This is in line with earlier findings that Q module assembly is directly linked to the biogenesis of the mitochondria-encoded subunit ND1 which is rapidly degraded if Q module assembly is impaired (Zurita Rendon and Shoubridge, 2012), and is also fitting to our findings with an NDUFS7 knockout (S7 KO) cell line. ND1 was not lost in NDUFS5 knockout (S5 KO) cells in which complex I is also lost, but assembly is affected at a later stage (Salscheider et al., 2022). (B) Assessment of mitochondrial translation in NDUFAF8 knockout cell lines complemented with NDUFAF8 variants. Experiment was performed as described in A. Both NDUFAF8 and SU9^MTS-NDUFAF8-4CA could complement the loss of NDUFAF8 with respect to ND1 synthesis indicating again their full functionality despite the absence of disulfide bonds. Signals were quantified using Image Lab and the amount of imported protein was plotted. N = 3 replicates. Source data are available for this figure: SourceData FS5.

Figure 7.

Functionality of the mitochondrial disulfide relay and redox conditions in the IMS determine the efficiency of NDUFAF5 activation and stability. (A) Protein levels in NDUFS5 knockout cells. Lysates from different cells were analyzed by reducing SDS-PAGE and immunoblotting. Signals were quantified using Image Lab, and the amount of protein was plotted. Loss of NDUFS5 does not affect NDUFAF5 levels. Since NDUFS5 is a structural subunit of complex I, its loss results in complex I loss. Thus, complex I loss does not affect NDUFAF5 levels. N = 3 replicates; error bars indicate SD. (B) Protein levels in MIA40/CHCHD4 knockdown cells. Experiment was performed as described in A. Depletion of MIA40/CHCHD4 results in depletion of NDUFAF5. This effect is independent of complex I levels and most likely mediated by the depletion of NDUFAF8 in these cells. N = 3 replicates; error bars indicate SD. (C) Protein levels in AIFM1 knockout cells. Experiment was performed as described in A. Loss of AIFM1 results in depletion of NDUFAF5. Like for MIA40/CHCHD4, this effect is independent from complex I levels and most likely mediated by the depletion of NDUFAF8 in these cells. N = 3 replicates; error bars indicate SD. (D) Protein levels in HEK293 cells expressing IMS-targeted LbNOX. Experiment was performed as described in A. Shifting the NADH/NAD⁺ ratio toward NAD⁺ results in the depletion of NDUFAF5. N = 3 replicates; error bars indicate SD. (E) In organello import assay with NDUFAF8 to test for its dependence on IMS NADH levels. Experiment was performed as described in Fig. 2 A. Import was performed into mitochondria isolated from wild-type or AIFM1^MTS-LbNOX expressing cells. NDUFAF8 import into mitochondria with low NADH levels was reduced. N = 1 replicate. (F) Model. The two-step import pathway of NDUFAF8 combined with the close surveillance of its levels in IMS and matrix allows control of NDUFAF5 levels and activity and efficiency of complex I assembly dependent on the activity of IMS redox processes. Both impairment of the mitochondrial disulfide relay as well as NADH depletion directly has an impact on NDUFAF5 levels in the matrix. Source data are available for this figure: SourceData F7.

Figure 8.

Additional substrates of the mitochondrial disulfide relay carry weak N-terminal targeting sequences that can drive matrix import. (A) iMTS prediction of different disulfide relay substrates. NDUFAF8, CHCHD1, CHCHD2, CHCHD9, CHCHD10, and COA6 isoform 2 all have very low TargetP scores, but a positive iMTS score in their first 19 amino acids. All proteins contain in their N-terminal amino acid stretch arginine and lysine residues but not negatively charged amino acid residues. MTS, mitochondrial targeting signal. (B) In organello import assay with disulfide relay substrate-DHFR fusion constructs. The experiment was performed as described in Fig. 2 A. To test the capacity of the N-terminal amino acids of NDUFAF8, CHCHD1, CHCHD2, CHCHD9, CHCHD10, and COA6 isoform 2 to serve as MTS, these amino acids (usually the first 19–24 amino acids) were fused to the cytosolic protein DHFR^mut. All N-terminal stretches facilitated the import of DHFR^mut albeit with different efficiencies. DHFR alone was not imported into mitochondria. N = 3 replicates; error bars indicate SD. (C) Split-GFP assay to detect the localization of full-length NDUFAF8, CHCHD1, CHCHD2, CHCHD9, CHCHD10, and COA6 isoform 2. The experiment was performed as described in Fig. 1 C. GFP1-10 was equipped either with an MTS for the matrix (SU9^MTS) or for the IMS (SCO2^MTS). GFP11 was C-terminally fused to full-length NDUFAF8, CHCHD1, CHCHD2, CHCHD9, CHCHD10, or COA6 isoform 2 and as controls for matrix, IMS, and cytosol to SOD2, MIA40/CHCHD4, and DHFR, respectively. For SOD2 and MIA40/CHCHD4, fluorescence could only be observed for matrix and IMS respectively (see “split GFP” signal). DHFR-GFP11 co-expression did not result in the reconstitution of GFP with any of the mitochondria localized GFP1-10 s. In the case of NDUFAF8, CHCHD1, CHCHD2, CHCHD9, CHCHD10, or COA6 isoform 2, GFP reassembled for both SU9^MTS-GFP1-10 and SCO2^MTS-GFP1-10 indicating for all proteins that at least a small fraction localized also to the matrix. Bar corresponds to 20 µm. (D) Protein levels in HEK293 cells expressing IMS-targeted LbNOX. Experiment was performed as described in Fig. 7 D. Shifting the NADH/NAD⁺ ratio towards NAD⁺ results in depletion of CHCHD1 and COA6. N = 3–5 replicates; error bars indicate SD. (E) Model. The import pathway combining the TIM23 and mitochondrial disulfide relay pathways allows the simultaneous readout of the functionality of both pathways and thereby potentially the energy state of matrix and IMS, respectively. In both cases, NADH levels influence import, either through establishing the mitochondrial membrane potential or by allowing the reconstitution of the MIA40/CHCHD4-AIFM import receptor complex. Source data are available for this figure: SourceData F8.