Gene Knockout Using Transcription Activator-like Effector Nucleases (TALENs) Reveals That Human NDUFA9 Protein Is Essential for Stabilizing the Junction between Membrane and Matrix Arms of Complex I*

Background: Human mitochondrial complex I is composed of 44 subunits with an elaborate assembly mechanism. Results: TALENs were used to knock out complex I subunit NDUFA9 in HEK293T cells, revealing a novel membrane arm subcomplex. Conclusion: NDUFA9 is critical for stabilizing the junction between different arms of complex I. Significance: TALEN-mediated knock-out represents an accessible approach to address protein function in human cells. Transcription activator-like effector nucleases (TALENs) represent a promising approach for targeted knock-out of genes in cultured human cells. We used TALEN-technology to knock out the nuclear gene encoding NDUFA9, a subunit of mitochondrial respiratory chain complex I in HEK293T cells. Screening for the knock-out revealed a mixture of NDUFA9 cell clones that harbored partial deletions of the mitochondrial N-terminal targeting signal but were still capable of import. A cell line lacking functional copies of both NDUFA9 alleles resulted in a loss of NDUFA9 protein expression, impaired assembly of complex I, and cells incapable of growth in galactose medium. Cells lacking NDUFA9 contained a complex I subcomplex consisting of membrane arm subunits but not marker subunits of the matrix arm. Re-expression of NDUFA9 restored the defects in complex I assembly. We conclude that NDUFA9 is involved in stabilizing the junction between membrane and matrix arms of complex I, a late assembly step critical for complex I biogenesis and activity.

Transcription activator-like effector nucleases (TALENs) represent a promising approach for targeted knock-out of genes in cultured human cells. We used TALEN-technology to knock out the nuclear gene encoding NDUFA9, a subunit of mitochondrial respiratory chain complex I in HEK293T cells. Screening for the knock-out revealed a mixture of NDUFA9 cell clones that harbored partial deletions of the mitochondrial N-terminal targeting signal but were still capable of import. A cell line lacking functional copies of both NDUFA9 alleles resulted in a loss of NDUFA9 protein expression, impaired assembly of complex I, and cells incapable of growth in galactose medium. Cells lacking NDUFA9 contained a complex I subcomplex consisting of membrane arm subunits but not marker subunits of the matrix arm. Re-expression of NDUFA9 restored the defects in complex I assembly. We conclude that NDUFA9 is involved in stabilizing the junction between membrane and matrix arms of complex I, a late assembly step critical for complex I biogenesis and activity.
Defects in complex I activity and assembly are a major cause of mitochondrial disease (1,2). In humans, complex I is composed of 44 subunits, seven of which are encoded by mitochondrial (mt) DNA and the remaining 37 of which are encoded by nuclear genes, synthesized in the cytosol, and imported into the organelle (1,3). Complex I is an L-shaped structure with a matrix and membrane arm that undergoes conformational coupling to facilitate proton pumping (4). Its holoenzyme form is ϳ1 MDa, but within mitochondria, it assembles into higher ordered supercomplexes with respiratory chain complexes III and IV (5). In recent years, the assembly pathway for complex I has been refined, and various assembly factors critical for complex I biogenesis have been identified (1). Briefly, mtDNA-encoded subunits are inserted into the inner membrane and associate in distinct modules with core subunits and specific assembly factors. Newly imported subunits of the matrix arm are incorporated at different times and can exchange with preexisting ones, presumably to maintain the structural and functional integrity of the complex (6,7). Still under debate is the role of so-called accessory (or supernumerary) subunits that are present in mitochondrial complex I but not in the bacterial counterpart (8,9).
As the subunit composition of mitochondrial complex I is different between fungi and humans (8), studies into complex I assembly and disease most often use cultured mammalian cells such as patient cell lines. These primary cell lines, however, often grow poorly, whereas isogenic control cell lines are most often missing. Although RNA interference has been utilized to study the effect of down-regulating expression of assembly factors and complex I subunits, there are limitations with this technique due to the lack of complete knockdown and potential off-target effects.
In this study, we tested the feasibility of TALEN 2 -mediated gene disruption (10 -15) for the study of complex I biogenesis in human cells. We chose to target NDUFA9, an accessory subunit found in all eukaryotes but having an unclear role in complex I function (8,9). Homozygous disruption of the NDUFA9 gene resulted in loss of complex I activity and the appearance of a novel ϳ600-kDa complex I subcomplex, which was rescued by re-expression of NDUFA9. Analysis of the subcomplex revealed it to contain subunits located in the membrane arm, but not marker subunits of the matrix arm. These results suggest that NDUFA9 is important in stabilizing the junction between matrix and membrane arms of complex I.

EXPERIMENTAL PROCEDURES
TALEN Design and Construction-The first exon and its immediate flanking regions of human NDUFA9 (NG_032124.1) were scanned for putative TALEN binding pairs using the TAL Effector Nucleotide Targeter 2.0 (16) (11). Cell Culture and Screening-HEK293T cells grown in DMEM (Invitrogen) containing 10% (v/v) FBS at 37°C under 5% CO 2 , 95% air supplemented with 50 g/ml uridine were transfected with a mixture of pTALEN-NDUFA9-L, pTALEN-NDUFA9-R, and pEGFP-N1 (Clontech) at a ratio of 9:9:1. Cells were trypsinized 2-6 days after transfection and resuspended in PBS containing 10% (v/v) FBS, and single cells were isolated using a BD Biosciences FACSAria III gated on GFP fluorescence. Clonal populations were expanded, and whole cell extracts were analyzed by SDS-PAGE and immunoblotting. To confirm NDUFA9 disruption, the targeted exon was PCR-amplified from genomic DNA isolated from individual clones and analyzed by Sanger sequencing. For NDUFA9 complementation, HEK293T or NDUFA9 Ϫ/Ϫ cells were transfected with pNDUFA9. The following day, media were refreshed or exchanged with glucose-free DMEM (Invitrogen) containing 10% (v/v) FBS, 5 mM galactose, and 50 g/ml uridine. Cells were grown an additional 24 h prior to harvesting.
In Vitro Protein Import into Isolated Mitochondria-Precursor proteins labeled with [ 35 S]methionine/cysteine were in vitro translated as described previously (6). Precursors were incubated with isolated mitochondria at 37°C for various times as indicated. Proteinase K and dissipation of the mitochondrial membrane potential (⌬⌿ m ) were performed as described (19).

RESULTS
Targeted Gene Disruption of NDUFA9-TALEN binding pairs for gene knock-out are designed to target directly downstream of the translation start codon, creating double-strand breaks that are repaired by nonhomologous end joining yielding out-of-frame deletions or insertions (10 -15). Like the majority of nuclear encoded mitochondrial proteins, NDUFA9 harbors an N-terminal cleavable presequence that targets the protein to mitochondria and is removed by specialized proteases in the mitochondrial matrix (6,20). We targeted this region of NDUFA9 for TALEN-mediated gene disruption, generating the TALEN binding pair shown in Fig. 1A (see "Experimental Procedures" for additional information) using the "TALE toolbox" described by Sanjana et al. (11). We chose HEK293T cells due to their ease of culturing and high transfection efficiency. The cells were transfected with plasmids encoding the TALEN binding pair along with a limiting amount of marker plasmid expressing GFP for cell sorting. Clonal populations were screened by immunoblot analysis using NDUFA9 antibodies. As can be seen (Fig. 1B), varying levels of NDUFA9 were evident between clones. Sequencing of a clone with no detectable NDUFA9 (Fig. 1B, lane 12, clone K; hereafter NDUFA9 Ϫ/Ϫ ) revealed loss of 17 nucleotides in both alleles that disrupted NDUFA9 expression (Fig. 1C).
Interestingly, analysis of another clone (Fig. 1C, Clone D) also revealed biallelic TALEN-mediated disruption of the NDUFA9 gene; however, NDUFA9 was still expressed (Fig. 1B,  lane 5). In this case, a second in-frame translation start codon (corresponding to Met-15; Fig. 1A) might be used to initiate translation, thereby producing NDUFA9 with a truncated presequence (NDUFA9 ⌬1-14 ) leaving the mature protein unchanged. As shown by our in vitro import experiment, [ 35 S]NDUFA9 ⌬1-14 is equally capable of mitochondrial import as full-length NDUFA9 (Fig. 1D, lanes 1-7), and when analyzed by BN-PAGE, it efficiently assembled into complex I (Fig. 1D,  lanes 8 -13). Taken together, these results demonstrate the importance of correct site selection in targets containing an N-terminal cleavable presequence and the verification of protein loss by immunoblotting.
NDUFA9 Ϫ/Ϫ Cells Harbor a Complex I Assembly Defect Reversible by Re-expression of NDUFA9-We next assessed the effect of NDUFA9 gene disruption on complex I activity. Mitochondria were isolated from HEK293T and NDUFA9 Ϫ/Ϫ cells and solubilized in digitonin, and complexes were resolved by BN-PAGE. Using an in-gel activity assay, we could detect complex I activity within a supercomplex that contains complexes III and IV (CI/CIII 2 /CIV) in HEK293T mitochondria but not in NDUFA9 Ϫ/Ϫ mitochondria (Fig. 1E, lanes 1 and 2). Immunoblot analysis using an antibody specific to NDUFB6, a membrane arm subunit of complex I, showed the absence of the CI/CIII 2 /CIV supercomplex in NDUFA9 Ϫ/Ϫ mitochondria and the presence of faster migrating species (Fig. 1E, lanes 3 and 4). These species still contained complexes III and IV (Fig. 1E,  lanes 5-8). The migration of complexes II and V, which do not form supercomplexes with complex I, also appeared normal in NDUFA9 Ϫ/Ϫ mitochondria (data not shown). We next used Triton X-100 to dissociate the majority of supercomplexes (17) and analyzed the migration of the holocomplexes. Although NDUFB6 was seen in the ϳ1-MDa mature complex I form in HEK293T mitochondria, it was found to migrate in a subcomplex of ϳ600 kDa in NDUFA9 Ϫ/Ϫ mitochondria (Fig. 1E). Furthermore, the migration of complexes III and IV was unaffected in NDUFA9 Ϫ/Ϫ mitochondria, pointing to an isolated defect in complex I assembly.
Cells with severe respiratory chain defects do not grow in the presence of galactose as the sole carbon source due to their dependence on oxidative phosphorylation for ATP generation (21). Consistent with a loss of complex I activity, we were unable to culture NDUFA9 Ϫ/Ϫ cells in galactose-containing medium. However when we transfected NDUFA9 Ϫ/Ϫ cells with a plasmid expressing NDUFA9, growth in galactose medium was restored (data not shown). Analysis of mitochondria from these cells revealed that the levels of complex I were partially restored with a concomitant reduction in the amount of ϳ600-kDa assembly (Fig. 1F).
The ϳ600-kDa Subcomplex Contains All mtDNA-encoded Complex I Subunits-To understand how the ϳ600-kDa subcomplex is formed, we analyzed the assembly of newly synthesized mtDNA-encoded subunits into complex I using pulsechase analysis (17). Mitochondria were isolated, solubilized in Triton X-100, and subjected to BN-PAGE analysis. In contrast to HEK293T mitochondria, in NDUFA9 Ϫ/Ϫ mitochondria, the mtDNA-encoded subunits accumulated in a ϳ600-kDa complex instead of mature complex I ( Fig. 2A). To identify the components within this complex, we used two-dimensional PAGE analysis of the 20-h chase. Following the labeling (0-h chase), the assembly profile of mtDNA-encoded complex I subunits between HEK293T and NDUFA9 Ϫ/Ϫ mitochondria was indistinguishable with the presence of early assembly intermediates at ϳ400 and ϳ460 kDa and a later stage ϳ830-kDa intermediate (Fig. 2B). However, at a 20-h chase, the mtDNA-encoded complex I subunits accumulated in the ϳ600-kDa complex in NDUFA9 Ϫ/Ϫ mitochondria instead of the mature complex I seen in HEK293T mitochondria. The ϳ600-kDa complex contained all mtDNA-encoded complex I subunits (ND1-6 and ND4L). The identity of the ϳ600-kDa complex and of mature complex I in control mitochondria was confirmed by immunoblotting for NDUFB6 (Fig. 2B, lower panels). Taken together, we conclude that NDUFA9 Ϫ/Ϫ cells lack mature complex I, instead harboring a ϳ600-kDa subcomplex containing at least all mtDNA-encoded subunits as well as other subunits including NDUFB6.
NDUFA9 Stabilizes the Matrix Arm of Complex I-We analyzed the steady state levels of various nuclear encoded complex  1-7), or solubilized in Triton X-100 and analyzed by BN-PAGE (lanes 8 -13). Input lysate is also shown. p, precursor; m, mature. E, mitochondria isolated from HEK293T or NDUFA9 Ϫ/Ϫ cells were solubilized in digitonin or Triton X-100, and protein complexes were separated by BN-PAGE. Gel strips were either stained for complex I in-gel activity (IGA; lanes 1 and 2) or analyzed by immunoblotting with the indicated antibodies. Complexes CI-CV are indicated. * indicates complex I subcomplexes detected in NDUFA9 Ϫ/Ϫ mitochondria. F, mitochondria were isolated from mock-or pNDUFA9-transfected HEK293T or NDUFA9 Ϫ/Ϫ cells grown in the indicated medium and solubilized in Triton X-100 as well as complexes analyzed by BN-PAGE and immunoblotting with the indicated antibodies.
I components in NDUFA9 Ϫ/Ϫ mitochondria by SDS-PAGE and immunoblotting (Fig. 2C, lanes 1 and 2). In comparison with other subunits and control proteins, the level of matrix arm subunits NDUFS2 and NDUFS3 appeared reduced. Indeed both subunits were missing from the ϳ600-kDa subcomplex in NDUFA9 Ϫ/Ϫ mitochondria (Fig. 2C, lanes 3-8). In contrast, NDUFS5, which has been proposed to associate with the membrane arm at a late stage in the assembly pathway (22)(23)(24), was detected in the ϳ600-kDa subcomplex. Due to the absence of antibodies against other subunits, we analyzed the import and assembly of matrix arm subunits NDUFS4 and NDUFA12 following their in vitro import. As can be seen (Fig. 2D), [ 35 S]NDUFS4 and [ 35 S]NDUFA12 could assemble into complex I in HEK293T mitochondria, but in NDUFA9 Ϫ/Ϫ mitochondria, they failed to assemble into a complex, consistent with their absence in the ϳ600-kDa subcomplex. We conclude that NDUFA9 is a critical subunit for the function of complex I and is required for stability of a later stage assembly, involving the connection between the membrane and matrix arms.

DISCUSSION
Routine genome modification has the potential to revolutionize the use of cultured human cells as a model for the study of intracellular processes and for understanding disease. Although targeted methods such as zinc finger technology have existed for some years, widespread use has been prohibited by high costs associated with their often proprietary nature (11,25,26). TALEN-mediated gene disruption is a promising technique due to the open-source approach adopted by its establishing groups; reagents are freely available by request or deposited in nonprofit plasmid repositories (10,11,27). TALENs can be designed to generate out-of-frame deletions in the first coding exon of the targeted gene (10 -15, 27); however, targets with multiple splice variants, especially those with internal splicing, pose significant TALEN design problems. As most nuclear encoded mitochondrial proteins are directed to mitochondria by a cleavable N-terminal presequence (28), we postulated that by disrupting the presequence and thus mitochondrial import, consideration of splice variants could be avoided. We sought to apply TALEN-mediated gene disruption to the understanding of mitochondrial complex I assembly. NDUFA9 is an accessory subunit that is conserved in all eukaryotes that harbor mitochondrial complex I. The subunit was chosen based on (i) the availability of specific antibodies for this subunit to facilitate screening of clones; and (ii) the presence of an N-terminal, cleavable targeting signal for TALEN targeting. Although targeting of the NDUFA9 presequence resulted in efficient generation of out-of-frame deletions (we estimate Ͼ60% following sorting of transfected cells), we found that instead of a loss of detectable NDUFA9, many clones exhibited a reduced level of NDUFA9 expression. Our analysis suggested translation from a second in-frame start codon present within the NDUFA9 presequence to be the causative factor because the truncated presequence remained competent for mitochondrial import. Although our results show disruption of an N-terminal targeting signal to be sufficient to ablate protein expression, they also demonstrate that care must still be taken in target site selection and that appropriate experiments for expression studies are conducted.
Consistent with a recent patient study reporting loss of complex I activity in a patient harboring a point mutation in NDUFA9 (29), we found that NDUFA9 is critical for complex I activity. NDUFA9 Ϫ/Ϫ cells were also unable to utilize galactose as a carbon source, indicating that oxidative phosphorylation was disrupted. Although it was shown that complex I was deficient in NDUFA9 patient fibroblasts, it was not established how loss of NDUFA9 affected assembly (29). Furthermore, previous studies have reached different views into NDUFA9 biogenesis with suggestions that it is incorporated either at an early stage or at a late stage in complex I assembly (30 -32). Pulse-chase analysis of membrane-integrated, mtDNA-encoded subunits revealed the presence of early complex I intermediates in NDUFA9 Ϫ/Ϫ cells, thus suggesting that NDUFA9 is required at a later stage of complex I assembly. A novel, steady state ϳ600-kDa subcomplex accumulated in NDUFA9 Ϫ/Ϫ cells that contained marker subunits of the complex I membrane arm, including all mtDNA-encoded subunits, but not marker subunits of the matrix arm. However, the earliest evidence for the incorporation of all mtDNA-encoded subunits appears to be when a late stage ϳ830-kDa intermediate is formed that is missing subunits of the matrix NADH dehydrogenase module (1). The pulse-chase studies indicated that in NDUFA9 Ϫ/Ϫ cells, the mtDNA-encoded subunits transiently assembled into a higher form before forming the stable ϳ600-kDa subcomplex. We suggest that the higher form represents the ϳ830-kDa intermediate, which, in the absence of NDUFA9, is unstable, leading to the dissociation of matrix subunits such as NDUFS2 and NDUFS3 (present in early intermediates (31,32)) to produce the ϳ600-kDa membrane arm. These results would be consistent with the suggested location of NDUFA9 at the junction between the matrix and membrane arms of complex I. As the ϳ600-kDa subcomplex identified in this study could also be found in supercomplexes with complexes III and IV, our study also suggests that the membrane portion of complex I is involved in making major contacts with subunits of complex III and IV.
In summary, NDUFA9 is an essential subunit required for human complex I assembly. Although NDUFA9 is classified as an accessory subunit due to its absence in bacteria, it is nevertheless required for stabilizing the link between the membrane and matrix arms of mitochondrial complex I. We suggest that TALEN technology can also be used to determine the importance of other accessory subunits and assembly factors involved in complex I biogenesis and disease.