To express a C-terminally His-tagged Dm NTD (amino acids Met¹-Ala³³³), Drosophila S2 cells were cultured and a stable cell line was generated as described previously (Matsushima and Kaguni, 2009). The stable cell line was grown in suspension culture in Insect-XPRESS™ protein-free insect cell medium (Lonza), and protein expression was induced with 0.2 mM CuSO₄. After 3 days of induction, cells were harvested and mitochondria were isolated by differential centrifugation. A mitochondrial extract was prepared and purified further by nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography as described earlier (Stiban et al., 2014).

The CG3292 cDNA (Berkeley Drosophila Genome Project Drosophila Gene Collection clone, RE72832) was purchased from the Drosophila Genomics Resource Center. Because the RE72832 clone lacks 16 base pairs in its ORF, the missing nucleotides (5′-GCA GTT AAT TTT GCC T-3′) were inserted by two cycles of site-directed mutagenesis using the following primers: LSK MS11-1F 5′ GGA AAA AGC ACC GTG TTT TGC CTG CAG CTT GGC AAA AC-3′, LSK MS11-1R 5′-GTT TTG CCA AGC TGC AGG CAA AAC ACG GTG CTT TTT CC-3, LSK MS11-2F 5′-GGA AAA AGC ACC GTG GCA GTT AAT TTT GCC TGC AGC-3′, LSK MS11-2R 5′-GCT GCA GGC AAA ATT AAC TGC CAC GGT GCT TTT TCC-3’. The FL ORF was confirmed by DNA sequencing and designated as pFlc-NUBPL.

The pMt-EGFP and pMt-HA vectors were constructed by insertion of PCR-amplified EGFP (enhanced GFP, 209 amino acids) or HA fragments (hemagglutinin, 9 amino acids), respectively, into the Drosophila expression vector, pMt (Matsushima and Kaguni, 2009). The following primers were used for the pMt-EGFP construct (LSK MS1F 5′-GGA GGA TCC ATG GTG AGC AAG GGC GAG GA-3′ and LSK MS1R 5′-TCC ACT AGT TTA CTT GTA CAG CTC GTC CAT GC-3′) and for the pMt-HA construct (LSK MS3F 5′ GGA GGA TCC TAC CCA TAC GAT GTT C-3′ and LSK MS3R 5′-AAC ACT AGT CTA CTA CAA GCT AGC-3′).

For both C-terminally EGFP-tagged Dm NUBPL (M¹-H²⁹³) and C-terminally HA-tagged Dm NUBPL (M¹-H²⁹³), the ORF of the FL CG3292 was amplified by PCR from pFlc-NUBPL with following primers: LSK MS13F, 5′-GGA CTC GAG ATG GAG CGT CTA TTG ATC-3′ and LSK MS13R, 5′- CCC GGA TCC ATG TGC ACT GTT ATT TTG-3.’ The PCR products were ligated to pMt-EGFP to obtain pMt-EGFP-NUBPL, and pMt-HA to obtain pMt-HA-NUBPL.

S2 cells were transiently transfected with pMt-EGFP-NUBPL and pMt-HA-NUBPL by subculture and transfection as in the generation of stable cell lines of the Dm NTD, except that the selection procedure was omitted. The recombinant Dm NUBPL proteins were induced for 3 days with 0.2 mM CuSO₄ at 48 h after transfection.

For E. coli cell expression of the N-terminally His-tagged NUBPL (Met²⁵-His²⁹³), the ORF was amplified by PCR from pFlc-NUBPL with the following primers: LSK MS25F 5′-TTA GGA TCC ATG GCG CGG GGA TTG C-3′ and LSK MS17R 5′-CCA GTC GAC CTA CTA CAA GCT AGC GTA ATC-3’. The PCR product was then ligated into the pET28a vector (Novagen), generating pET28a-NUBPL. An ISC coordinating-deficient variant (C214A/C217A) was constructed by site directed mutagenesis of pET28a-NUBPL using following primers: LSK MS19F 5′-GAG AAC ATG AAG TAC ACC ATT GCT CAG AAC GCT AAT CAA CGA TTG GAG TTT TTT AAA G-3′ and LSK MS 19R 5′-CTT TAA AAA ACT CCA ATC GTT GAT-3’.

The N-terminally His-tagged Dm mtDNA helicase NTD (Asn²⁴-Ala³³³) and the ISC deficient variants (C68A/C71A) and (C102A/C105A) were produced by overexpression in E. coli BL21 cells and purified by Ni-NTA affinity chromatography as described previously (Stiban et al., 2014), except that Superdex 200 HR 10/30 (GE Healthcare) gel filtration chromatography was substituted for the glycerol gradient sedimentation.

The N-terminally His-tagged NUBPL (Met²⁵-H²⁹³ is) and its ISC-deficient variant (C214A/C217A) were produced by overexpression in E. coli BL21 cells (Novagen) in auto-induction media as described (Gajewski et al., 2016). Briefly, transformed cells were cultured at 37°C with aeration until an OD₆₀₀ of 1 was reached. Then, the cultures were shifted to 25°C until an OD₆₀₀ of ∼15–20 was reached (∼48–60 h).

Cell harvest and protein purification were performed as described (Stiban et al., 2014). Briefly, after harvest, cells were suspended in cold Tris-sucrose buffer (50 mM Tris-HCl, pH 7.5, 10% sucrose, 250 mM NaCl, 1.5% n-dodecyl β-D-maltoside, 2 μg/ml of leupeptin, 1 mM phenylmethylsulfonyl fluoride, 5 mM β-mercaptoethanol, and 10 mM sodium metabisulfite) and lysed by a freeze-thaw cycle. The resulting lysate was centrifuged 17,000 × g for 50 min at 4°C and the recombinant Dm NUBPL proteins were purified by Ni-NTA affinity chromatography. Similar approaches were taken to purify FL Dm and human Hs mtDNA helicase and the RNA polymerase domain (RPD) of mtDNA helicase.

After SDS-PAGE, gels were immersed in potassium ferricyanide solution (100 mM potassium ferricyanide, 50 mM Tris-HCl, pH 7.5, 100 mM NaCl) in the dark for 1 h. The gels were transferred to freshly-prepared color-developing solution (10% methanol, 10% trichloroacetic acid) until the protein bands were visible.

To produce the apo-NTD, the protocol of Kennedy and Beinert (Kennedy and Beinert, 1988) was modified as follows. EDTA and potassium ferricyanide were added to the NTD in a 100-fold and 50-fold molar excess in 1X replacement buffer (10 mM NH₄HCO₃, pH 7.5, 12% glycerol). The mixtures were kept on ice for 10 min to allow oxidation and extraction of the ISC. The apo-NTD was then desalted on a Zeba™ spin desalting column (40 kDa MWCO) and the buffer was exchanged with 2X replacement buffer. Then, the UV-vis absorption spectrum of the apo-NTD was measured in buffer A (250 mM NaCl, 35 mM Tris-HCl pH 7.5, 12% glycerol, and 5 mM β-mercaptoethanol) in a Hewlett-Packard 8453 spectrophotometer with a quartz cuvette (1 cm path length). For Zn²⁺ replacement, the apo-NTD was incubated with a 50-fold molar excess of ZnSO₄ on ice for 1 h. Unbound Zn²⁺ was removed by desalting, and the UV-vis spectrum of the protein was measured.

Because removal of the ISC led to precipitation of the protein, a modified protocol was performed to achieve ISC removal and metal replacement at the same time. The NTD was incubated in the presence of a 100-fold excess of ZnSO₄, a 100-fold molar excess of EDTA and a 50-fold molar excess of potassium ferricyanide. The mixture was incubated on ice for 1 h. Unbound Zn²⁺ was removed by desalting and the UV-vis spectrum of the protein was measured.

Purified NTD, metal-substituted NTD and purified NTD variants (C68A/C71A) and (C102A/C105A) were incubated on ice after their initial absorption spectra were taken. At the indicated times, the samples were centrifuged at 20,000 × g for 30 s to remove precipitated material and absorption spectra were measured from supernatant fractions. Protein concentration was calculated as described previously (Stiban et al., 2014).

After incubation of the metal-substituted NTD for 71 h on ice, protein concentration was determined by A280 measurement, and 1.5 and 0.8 nmol protein from different preparations were centrifuged for 1 h at 20,000 × g to remove protein precipitates. Both pellet and supernatant samples were boiled in the presence of 0.5 ml nitric acid (15.7 N) and 0.5 ml H₂O₂ (30%) until all liquid was evaporated. A control sample was prepared using 1× replacement buffer without protein. After evaporation, all samples were resuspended in 2 ml of 2% nitric acid. Fe and Zn standards were prepared using (Fe(NO₃)₃·9H₂O and ZnSO₄·7H₂O in 2% nitric acid solutions of 50, 100, 150, 200, 300, 400, 500, 750, and 1,000 ppb of both ions. Samples and standards were analyzed in an ICP-OES Varian 710-ES Axial spectrometer (Agilent Technologies) using the ICP Expert^TM II software.

Transiently-transfected, C-terminally EGFP-tagged NUBPL-expressing S2 cells and wild type S2 cells were grown to 70% confluence on a cover glass inside a six-well tissue culture plate (Corning). Mitochondria within the cells were stained with MitoTracker Red (10 μM final concentration, Molecular Probes) for 1 h in the dark before observation in an Olympus FluoView FV1000 confocal laser scanning microscope. The instrument was equipped with 488 nm argon and 543 nm helium-neon lasers for GFP and MitoTracker Red excitation, respectively. Fluorescence emission detected with a 505–525 nm band-pass filter for GFP, and 560 nm long-pass filter for MitoTracker Red. Images were obtained using a 60X PlanApo oil objectives.

Dm NUBPL (0.5 mg/ml, total 1 mg) was loaded onto a Superdex 75 HR 10/30 gel filtration column (GE healthcare) and chromatographed in gel filtration buffer (35 mM Tris-HCl pH 7.5, 350 mM NaCl, 10% glycerol, 5 mM β-mercaptoethanol) in the presence or absence of EDTA. Target protein eluted in the peak fractions was analyzed by 12% SDS-PAGE. Band intensities were measured using the ImageQuant 5.2 software after Coomassie blue staining.

Asolectin (Sigma) in the presence or absence of additional lipids (cholesterol (Sigma) or cardiolipin (Sigma)) were dissolved in chloroform (Baker) and dried under nitrogen gas. The residual chloroform was eliminated by vacuum desiccation for 8 h. In the preparation of multilamellar liposomes, the dry lipids were hydrated for 30 min in hydration buffer (10 mM NH₄HCO₃, pH 7.4, 0.1 mM EDTA). The solution was mixed by vortexing until visible clots disappeared. Then, four cycles of freezing, thawing, sonication, and vortexing were performed. Large unilamellar liposomes were prepared by extrusion through a polycarbonate membrane (100 nm pore size) using a mini-extruder (Avanti Polar Lipids). After incubation of the protein and liposomes for 15 min under the stated conditions, the reaction mixtures were centrifuged at 47,000 × g in an S120-AT3 rotor (Thermo Scientific) for 1 h. Equal portions of supernatant and pellet were analyzed by 12% SDS-PAGE followed by immunoblotting or silver staining. The intensities of the protein signals on the membrane were quantified using the ImageQuant 5.2 software. The fraction of bound protein was determined dividing the band intensity of the pellet (bound protein) by those of the pellet and the supernatant together (total protein).

Unilamellar liposomes containing asolectin and 0.5% BODIPY-TMR-PI(4,5)P₂ (mol%) were prepared as previously stated. Liposomes were hydrated in LH2 buffer. The fluorescence emission spectrum of a 10 μM liposome suspension was measured at 574 nm following an excitation at 542 nm. Following sequential additions of purified NTD, the fluorescence spectra were measured. NTD was added from stock of >50 μM. Buffer controls were done by adding the same amount of buffer to the cuvette. Fluorescence measurements were performed in PTI QW4 spectrofluorimeter equipped with a temperature-controlled cuvette holder kept at 20°C.

Both purified FL Hs mtDNA helicase and the RPD were used to test liposome binding using BODIPY-TMR-PI(4,5)P₂ quenching and cosedimentation assays. Fluorescence quenching experiments were performed in a similar manner as presented earlier. Cosedimentation was also repeated using 16% CL-asolectin liposomes and 8 μg of protein.

In order to measure the activity of the FL Hs mtDNA helicase, an ATPase/GTPase activity assay (Sigma cat. # MAK113) was performed according to the manufacturer’s instructions. In triplicate wells of a 96-well flat bottom plate, FL Hs helicase was incubated without or with increasing 16% CL-asolectin liposome concentrations and the volume was adjusted to 10 μl with assay buffer (40 mM Tris, 80 mM NaCl, 8 mM Mg acetate, 1 mM EDTA, pH 7.5). Background control was 10 μl of H₂O. Phosphate standards were also prepared in triplicate wells. Master reaction mix was prepared by mixing 20 μl of assay buffer with 10 μl of 4 mM ATP per well. 30 μl of the reaction mix was added to each well, and the plate was incubated for 30 min at room temperature. To each well, 150 μl of Reagent and 50 μl H₂O were added, and the plate was further incubated at room temperature for 30 min, followed by absorbance measurement at 650 nm using a SpectroMax M2/M2^e multi-well spectrophotometric plate reader, and data analysis by SoftMax Pro 53 software.

Calf thymus DNA was used to stimulate FL Hs helicase activity in an independent experiment following the same protocol, but with the addition of increasing amounts of DNA. Similarly, as a positive control, the ATPase activity of ΔN-helicase, termed P66 in Ziebarth et al. (2007), was measured in separate experiments.

To obtain a homology model of Dm NUBPL, the Dm NUBPL sequence was submitted to Phyre2 using the default parameters (Kelley et al., 2015). The dimer structure was derived by Dali (Holm and Rosenström, 2010) using the superposition of the NUBPL monomer model on the dimer crystal structure of nucleotide-binding protein AF2269 from A. fulgidus as a template (PDB# 3KB1). The electrostatic surface potential map of Dm NUBPL was generated using the APBS Tools in PyMol with the default parameters. Dm NUBPL was also modelled using AlphaFold software (Jumper et al., 2021).

To generate helical wheel projections, predicted secondary structures were obtained first using PsiPred (Buchan et al., 2013) from the primary amino acid sequences of the proteins. The corresponding sequence of the last predicted helix at the C-terminus of each protein was submitted to NetWheels (http://lbqp.unb.br/NetWheels/) to obtain a helical wheel projection.

In order to align human and Drosophila mtDNA helicases, the protein sequences (PDB accession numbers: Q96RR1 (Hs) and AAF52820 (Dm)) were subjected to BLAST analysis using the BLAST and Multiple Sequence Alignment tools on NCBI website (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

The 2DAGE experiments were performed as described previously (Ciesielski et al., 2018), using mitochondrial nucleic acids (mtNA) isolated from Drosophila S2 cells transfected transiently with pMt-HA-NUBPL, with or without the CuSO₄ induction. mtNA were treated with ClaI restriction endonuclease, and DNA replication intermediates (RI) were detected using the radiolabeled probe 6 that hybridizes to nts 6801–7378.

We demonstrated previously the presence of an ISC in Dm mtDNA helicase (Stiban et al., 2014), though the possibility remained that cluster insertion might be an artefact of the E. coli overexpression system. To verify the presence of an ISC in Dm mtDNA helicase, a recombinant form of the NTD was extracted from mitochondria isolated from Drosophila S2 cells and purified by Ni-NTA affinity chromatography. The partially-purified, NTD-containing fraction was separated by 15% SDS-PAGE and stained with potassium ferricyanide to detect non-heme iron ions such as those in an ISC. In this protocol, protein-bound ferrous iron reacts with potassium ferricyanide to form a royal blue complex, for which the intensity is directly proportional to the amount of iron present in the gel (Leong et al., 1992). The NTD proteins from S2 cells (Figure 1A, lane 1) and from E. coli BL21 cells (Figure 1A, lane 2) were detected by both Coomassie (ii) and potassium ferricyanide (i) staining. A recombinant FL Dm mtDNA helicase from an E. coli BL21 cell lysate was also detected at a position corresponding to ∼70 kDa by both staining procedures (Figure 1A, lane 3). These data indicate that the ISC is a bona fide component of the Dm mtDNA helicase.

We showed previously that double-substitution variants that lack the ability to bind an ISC (C68A/C71A and C102A/C105A) are unstable and precipitate during purification (Stiban et al., 2014). To examine further the role of the ISC in protein stability, a metal replacement strategy was pursued (Kennedy and Beinert, 1988). The ISC was fully oxidized and removed by incubating the holo-NTD with excess EDTA and potassium ferricyanide, as shown by a time-dependent decrease in the characteristic ISC peaks at 325 and 420 nm upon UV-visible spectroscopy (Figure 1B). The apo-protein however, was destabilized and precipitated (Figure 1C). We then evaluated Zn²⁺ substitution before and after ISC removal. While it had no effect after ISC removal (Figure 1C), the addition of Zn²⁺ with potassium ferricyanide to achieve ISC removal and replacement at the same time resulted in the disappearance of the characteristic peaks of the ISC in the absorbance spectrum (Figure 1B). The appearance of 0.77 ± 0.1 zinc molecule per NTD was detected by Inductively Coupled Plasma Optical Emission Spectrometry (Table 1). The effect of replacing the ISC with Zn²⁺ was probed by examining protein stability over time of incubation at 0°C. The Zn²⁺-substituted NTD showed a decay curve that was nearly identical to that of the ISC-containing wild-type NTD. By comparison, the ISC binding-deficient variants showed a rapid decay (Figure 1D).

The mtDNA helicase is an integral part of the minimal mitochondrial replisome, and initiation of mtDNA replication has been suggested to involve attachment of the DNA-protein apparatus to the mitochondrial inner membrane (MIM) (Rajala et al., 2014; Gerhold et al., 2015). We evaluated the possibility of membrane binding by mixing Dm NTD with unilamellar soybean asolectin liposomes and subjecting them to lipid vesicle cosedimentation to demonstrate protein-liposome complexes. Asolectin is a mixture of phospholipids extracted from soybean containing ∼45% phosphatidylcholine, 22% phosphatidylethanolamine, 18% phosphatidylinositol and 7% phosphatidic acid that mimics closely the composition of biological membranes. It is used widely in model membrane preparation and in liposome generation for sedimentation and membrane permeability assays (Nomura and Kurihara, 1987; Stiban et al., 2006; Avanti-Polar-Lipids, 2021; Sigma-Aldrich, 2021). We observed that in the absence of liposomes, Dm NTD was predominantly in the supernatant fraction (S) after ultracentrifugation, whereas in their presence, the protein was found in the pellet (P) (Figure 2A), indicating that the Dm NTD binds to liposomes. Liposome binding was inhibited with increasing NaCl concentrations (Figure 2B). The loss of cosedimentation of Dm NTD and liposomes by increased salt concentration suggests that the interaction with asolectin liposomes is electrostatic.

In order to confirm binding, intrinsic tryptophan fluorescence was measured. Dm NTD contains three tryptophan residues, which exhibit intrinsic fluorescence. The quantum yield of the fluorescence increases when the tryptophan residues are subjected to a hydrophobic environment, whereas it decreases in hydrophilic conditions (Kraft et al., 2009). Upon incubation of NTD with increasing liposome concentrations, we found that the quantum yield of its tryptophan fluorescence decreased in a dose-dependent manner, indicating binding to liposomes and exposure of the residues to a hydrophilic environment (Figure 2C), whereas the addition of buffer alone had no effect on tryptophan fluorescence (data not shown). Quenching was reversed by the addition of NaCl indicating salt-dependent binding (data not shown). Additionally, using fluorescently-labelled liposomes, the NTD was shown to induce the rearrangement and sequestration of BODIPY-TMR-PI(4,5)P₂ lipids in asolectin liposomes in a dose-dependent manner (Figure 2D). Upon sequestration, BODIPY-TMR-PI(4,5)P₂ monomers self-quench their fluorescent signal. Thus, NTD binding to liposomes was confirmed by three methods.

A recent report showed that mtDNA replicative proteins in a mitochondrial lysate of human HEK293 cells cosedimented with cholesterol in flotation gradients, suggesting that the replisome binds a cholesterol-rich membrane (Gerhold et al., 2015). To investigate whether the Dm NTD binds to cholesterol in cholesterol-containing liposomes, we diluted standard asolectin-only liposomes with those containing cholesterol to generate variable concentrations of cholesterol-containing liposomes. The increase in cholesterol concentration had no significant effect on the bound protein fraction (Figure 3A), indicating that Dm NTD does not bind to cholesterol in the asolectin liposomes. In contrast, when cardiolipin (CL), an abundant lipid in the MIM (Horvath and Daum, 2013; Rappocciolo and Stiban, 2019), was added to standard asolectin liposomes, the fraction of bound Dm NTD increased in a CL concentration-dependent fashion: binding increased from 0.1 to 0.6 as CL concentration increased from 2 to 20% w/w, and it was saturated when liposomes contained more than 20% w/w CL (Figure 3B). This data suggests that the Dm NTD binds specifically to MIM as a consequence of their CL content. The results also argue that cholesterol is not involved in membrane attachment of the Dm NTD. In order to confirm the specificity of CL binding, liposomes were enriched with 30% phosphatidylserine (PS) the cosedimentation analysis was repeated. CL has two negative charges while PS has one, so to control for the charge effect of CL, asolectin liposomes were supplemented with 15% CL or 30% PS prior to Dm NTD binding. We found that enrichment of liposomes with CL enhanced markedly the binding efficiency of Dm NTD as compared to PS-enriched or unenriched asolectin liposomes (Figure 3C), indicating that the protein may interact with CL microdomains that are elevated in the inner membrane.

The physiological relevance of liposome binding to the Dm NTD is unclear because it has no identified activity, and we have been unable to produce a stable recombinant form of purified FL Dm mtDNA helicase. Thus, we evaluated lipid binding by the human FL homolog. We observed that both the FL Hs mtDNA helicase and its RPD domain (the RNA polymerase-like subdomain of the NTD) bound to CL-asolectin liposomes in a dose-dependent manner in the vesicle cosedimentation assays (Figure 4A). Moreover, both proteins caused quenching of BODIPY-TMR-PI(4,5)P₂ fluorescence to the same extent (Figure 4B) arguing that the RPD subdomain of the NTD is the site of attachment to the bilayer.

We investigated the implications of binding by assessing the ATPase activity that was previously documented in the human FL helicase (Ziebarth et al., 2007) in the presence of increasing liposome concentrations. We observed that ATPase activity was stimulated by liposome binding (Figure 4C), suggesting a potentially important role for membrane binding to enhance the activity of the helicase in initiation of mtDNA replication. To confirm that the ATPase assay was indicative of helicase activity, ATP hydrolysis of FL helicase was increased with added calf thymus DNA in a dose dependent fashion (Figure 4D). A variant of helicase lacking the N-terminus (ΔN) was used as a positive control because it exhibits heightened ATPase activity (Ziebarth et al., 2007).

Proteins containing ISCs require ISC transfer partners to insert the cluster. ISC biogenesis can occur through the cytosolic ISC assembly (CIA) or the mitochondrial ISC assembly pathways (Lill et al., 2012; Braymer and Lill, 2017). Interestingly, in the D. melanogaster proteome, an interaction between the protein CG3262 and the mtDNA helicase was found by high-throughput co-affinity purification coupled with mass spectrometry (Guruharsha et al., 2011). The CG3262 protein shows homology to human NUBPL, with a 44% amino acid sequence identity predicted by BLAST, and a 78.1% sequence identity to Drosophila rhopaloa NUBPL. NUBPL in humans and yeast is an ISC transfer protein known to be required specifically for ISC insertion into mitochondrial NADH dehydrogenase (complex I) (Bych et al., 2008; Sheftel et al., 2009; Lill et al., 2012).

To investigate the possibility that the CG3262 protein (herein termed Dm NUBPL) serves as the transfer protein for Dm mtDNA helicase, we first examined its mitochondrial localization. Primary sequence analyses with MitoProt II (Ison et al., 2016) and iPSORT (Bannai et al., 2002) predicted its mitochondrial localization with a high probability (0.7036 and positive, respectively) via a canonical N-terminal mitochondrial targeting sequence. To demonstrate mitochondrial localization in vivo, a C-terminal EGFP tag was inserted into the ORF containing CG3262, and the recombinant protein expressed in Drosophila S2 cells was evaluated by confocal microscopy (Figure 5A). The green fluorescent signal in successfully transfected cells overlapped completely the red fluorescent signal from mitochondria stained with MitoTracker Red. Moreover, subcellular fractionation of a C-terminally HA-tagged Dm NUBPL showed localization to the mitochondrial fraction (M) but not the cytosolic fraction (C) when proteins derived from equal cell equivalents were analyzed by 12% SDS-PAGE and immunoblotting (Figure 5B). In composite, these in silico, in vivo and in vitro experiments demonstrate that Dm NUBPL is a mitochondrial protein.

The Dm NUBPL monomer contains only a single CXXC motif, a common cysteine-containing motif for coordinating metal ions (Supplementary Figure S2A). Homodimerization of Dm NUBPL was expected because four cysteines are required typically to coordinate an ISC (Supplementary Figure S2B; Maio and Rouault, 2016). We investigated the oligomeric state of Dm NUBPL by gel filtration of a N-terminally His-tagged recombinant form in the absence of cofactors (Figure 5C), and in the presence of the metal chelator EDTA to reduce the possibility that it would contain an ISC or a zinc ion. We found that Dm NUBPL both in the presence or absence of EDTA produced a single chromatographic peak with a Stokes radius of 37 Å and an estimated molecular mass of 66 kDa, corresponding to the theoretical molecular mass of a dimer (62 kDa). This data suggests that the presence/insertion of an ISC in Dm NUBPL is not obligatory for its dimerization. Furthermore, we produced a Dm NUBPL variant lacking the ISC-coordinating capability by alanine substitution of the two cysteines in the CXXC motif. The C214A/C217A variant also demonstrated a single peak in the same fractions as the dimer-form, wild-type Dm NUBPL, confirming that an ISC or a metal ion is not required for dimerization. Instead, the Dm NUBPL dimer may form disulfide bridges in the absence of binding its metal cofactor in the presence of EDTA. We observed that the C214A/C217A variant had significantly lower solubility than the wild-type form, suggesting that as with the Dm NTD, the coordinating cysteines are likely important for protein stability.

Because NUBPL is an ISC transferring protein specific for respiratory complex I, it is expected to associate with the MIM where complex I resides. We investigated membrane binding by Dm NUBPL by mixing the purified protein with unilamellar soybean asolectin liposomes, and protein-liposome complexes were evaluated by vesicle cosedimentation. We found that in the absence of liposomes, Dm NUBPL was predominantly in the supernatant fraction (S) after ultracentrifugation whereas in their presence, it was found in the pellet (P) (Figure 5D, closed circles), indicating that Dm NUBPL binds to liposomes. The fraction of bound Dm NUBPL (0.8) was saturated at a concentration of 100 μM of liposomes. In contrast to Dm NTD liposome binding, Dm NUBPL binds to standard or CL-containing liposomes similarly (Figure 5D, open circles), and more tightly than Dm NTD, with a K _d of ∼50 μM for Dm NUBPL as compared to ∼100 μM for Dm NTD. Interestingly, we found that the bound fraction of Dm NUBPL at equilibration was four times higher than that of Dm NTD, and binding increased when the NaCl concentration was increased up to 50 mM, but it was inhibited at higher concentrations (Figure 5E). As for Dm NTD, the loss of cosedimentation of Dm NUBPL and liposomes by increased salt concentration indicates that its interaction with asolectin liposomes is electrostatic. However, we observed a residual fraction of bound Dm NUBPL remaining at elevated salt concentration (∼0.2), suggesting an additional type of interaction. Hydrophobic interaction seems a likely possibility because a C-terminal amphipathic helix is predicted in human, Drosophila and Y. lipolytica NUBPL by the helical wheel projection program NetWheels (Supplementary Figure S2C). Our data suggest that Dm NUBPL is a peripheral membrane protein, and its membrane binding likely facilitates access to its membrane-bound ISC recipient, complex I.

To investigate a possible functional relationship between Dm NUBPL and the Dm replicative mtDNA helicase, we evaluated the effect of Dm NUBPL overexpression on mtDNA replication in S2 cells, by two-dimensional agarose gel electrophoresis. To do so, total mitochondrial nucleic acids (mtNA) obtained from sucrose gradient-purified mitochondria of the control and HA-NUBPL -overexpressing cells were cleaved by ClaI restriction endonuclease, and the resulting fragment bearing the previously identified replication pause site that defines replication slow zone 2 (Joers and Jacobs, 2013) was probed by hybridization with a homologous radiolabeled fragment. We observed that the signal from the prominent replication pause site was reduced substantially in samples from cells overexpressing Dm NUBPL (Figure 6), a result that is strikingly similar to the effect of the helicase overexpression in S2 cells that we have reported previously (Ciesielski et al., 2018).

These data suggest a functional interaction between Dm NUBPL and mtDNA helicase that is relevant to the efficiency of mtDNA replication. We propose that in the environment of the MIM, NUBPL provides mtDNA helicase with its ISC, enhancing its stability and activity, and in turn regulates the mtDNA replication process (Figure 7).