https://orcid.org/0000-0002-7506-627X
Abstract
G-quadruplexes (G4s), as an important type of non-canonical nucleic acid structure, have received much attention because of their regulations of various biological processes in cells. Identifying G4s-protein interactions is essential for understanding G4s-related biology. However, current strategies for exploring G4 binding proteins (G4BPs) include pull-down assays in cell lysates or photoaffinity labeling, which are lack of sufficient spatial specificity at the subcellular level. Herein, we develop a subcellular selective APEX2-based proximity labeling strategy to investigate the interactome of mitochondrial DNA (mtDNA) G4s in living cells. By this method, we have identified several mtDNA G4BPs. Among them, a previously unrecognized mtDNA G4BP, DHX30 has been selected as an example to explore its important biofunctions. DHX30 localizes both in cytoplasm and mitochondria and can resolve mtDNA G4s. Further studies have demonstrated that DHX30 unfolds mtDNA G4 in living cells, which results in a decrease in glycolysis activity of tumor cells. Besides, RHPS4, a known mtDNA G4 stabilizer, will reverse this inhibition effect. Benefiting from the high spatiotemporal resolution and the ability of genetically encoded systems to perform the labeling with exquisite specificity within living cells, our approach can realize the identification of subcellular localized G4BPs. Our work provides a novel strategy to map protein interactions of specific nucleic acid features in subcellular compartments of living cells.
Introduction
G-quadruplexes (G4s) are specific nucleic acid secondary structures widely prevalent in genomes, transcriptomes and telomeres, playing crucial roles in various essential cellular processes (1–3). The interactions between G4s and G4s binding proteins (G4BPs) affect numerous biological events, such as genome stability, DNA replication and transcription (4–8). The methods that can identify these interactions will promote our understanding of G4s-related physiological processes. Currently, most methods for identifying G4BPs either are low-throughput or are operated in lysed cell samples (9,10). As the loss of original structure and composition in lysed cell samples, these methods may lead to false positive results. Recently, Zhang et al. described a co-binding-mediated protein profiling strategy to map nuclear DNA G4BPs in cellular environment, enabling high-throughput capture of cellular G4BPs (11). Su et al. described a similar strategy, G4-LIMCAP, based on photoactive G4s ligands to identify multiple G4BPs with extensive sequence tolerance in living cells (12). Despite these approaches have made great progress in G4BPs identification, these photoaffinity labeling (PAL) strategies have multiple shortcomings, including off-target localization and lack of spatial specificity. Given that the carbenes generated tend to react with water rather than the target, employing diazirine-based probes has been difficult in living cell contexts (13,14). Meanwhile, the consumed probes will inhibit the tagging of unreacted molecules and impede the efficiency of labeling.
Enzyme-based proximity labeling (PL) is an efficient approach for interrogating biomolecular interaction in cellular environment (15). This method utilizes engineered enzymes [e.g. biotin ligases (16) or peroxidases (17)] to convert small molecule probes into highly reactive species which covalently tag neighboring endogenous species. PL has been widely applied to interrogate spatial proteomes (18), transcriptomes (19) and molecular interactions (20–24). Recently, Lu et al. developed a PLGPB method for the identification of G4BPs (25). PLGPB utilized miniTurbo (a biotin ligase) to label G4BPs in living cells and identified several unknown G4BPs. Therefore, we hypothesized that PL strategies could be compatible for identifying subcellular localized G4BPs, such as mitochondrial DNA (mtDNA) G4BPs.
Here, we reported a novel APEX2-based protein profiling strategy for the investigation of mtDNA G4BPs in cellular environment. By employing the strategy, we identified 206 mtDNA G4-associated proteins. Among them, we identified a previously unrecognized mtDNA G4BP, DHX30. We revealed that DHX30 could resolve mtDNA G4 in vitro and in cells. Finally, we demonstrated the tight connection between mtDNA G4s, DHX30 and tumor cell glycolysis.
Materials and methods
Materials
Biotin-phenol (BP) was purchased from Aladdin (Shanghai, China). RHPS4 and 2-Deoxy-D-glucose were purchased from MedChemExpress. Oligomycin was purchased from TargetMol. Oligonucleotides were synthesized by Sangon Biological Engineering Technology & Services (Shanghai, China). DHX30 protein was obtained from OriGene Wuxi Biotechnology Co., Ltd. (Wuxi, China).
Cell lines and cell culture
HEK293T cell line (CCTCC, Wuhan, China) was cultured in DMEM (Gibco, USA) supplemented with 10% FBS (HyClone, USA). MCF-7 cell line (CCTCC, Wuhan, China) was cultured in DMEM supplemented with 10% FBS. All the cells were maintained at 37°C in a humidified incubator containing 5% CO2.
Vectors construction
The cDNAs encoding the MTS-SG4-APEX2 and MTS-mSG4-APEX2 fused protein were PCR-amplified and subcloned into pcDNA 3.1 vectors, and were named as SG4-APEX2 and mSG4-APEX2.
Proximity labeling and enrichment of biotinylated proteins
APEX2-mediated PL experiments were conducted (26). Cells expressing APEX2 fusion construct were treated with 1 mM BP probe for 1 h. Then 1 mM H2O2 was used to treat the cells for 1 min. All cells were washed with quencher buffer (Supplementary Table S1) four times. Finally, the cells were either fixed for immunofluorescence imaging or lysed directly. For enrichment of biotinylated proteins, 40 μl of streptavidin magnetic beads (#P2151; Beyotime, Shanghai, China) were incubated with ∼100 μl cell lysates overnight at 4°C on a rotator. Then all samples were washed with various buffers (Supplementary Table S1) to remove nonspecific binders. The enriched proteins were eluted by heating all streptavidin magnetic beads samples at 100°C for 10 min in 25 μl 5 × SDS-PAGE loading buffer. Finally, all eluates were stored at −20°C.
In vitro G4 pull-down assay
The in vitro G4 capture experiments were performed as previously described (27) with slight modification. The Anti-Flag Magnetic Beads (Beyotime) were washed three times with Tris–HCl buffer containing 10 mM Tris–HCl and 100 mM KCl (pH = 7.4). Translation proteins or purified proteins were mixed with 5′-labeled oligonucleotides (FAM-mito0.5–22, FAM-mito0.5–22m, FAM-mito55 or FAM-mito55m) and Anti-Flag beads (#P2115; Beyotime) in the same Tris buffer (500 μl final volume) and incubated on a rotator for 4 h at room temperature. After incubation, the beads were immobilized (magnet) and the supernatant was removed. The solid residue was resuspended in 100 μl of TBS 1 × buffer, heated for 10 min at 90 °C (gentle stirring 250 rpm) and then centrifuged for 5 min at 10 000 rpm. The supernatant was taken up for analysis (magnet immobilization). FAM fluorescence was recorded at excitation/emission wavelengths of 483 nm/530 nm, respectively.
Mitochondrial DNA immunoprecipition
We performed mtDNA immunoprecipitation (mtDIP) assays using BeyoChIP™ ChIP Assay Kit (#P2080S, Beyotime) according to previous published work (28) with slight modification. The antibodies against the Flag (TA-05; ZSGB-Bio) were used for ChIP assays. Coprecipitated and purified DNA was detected by qRT-PCR. Isotype controls were assayed simultaneously to confirm that the detected signals were from DNA specifically binding to fusion protein with Flag tag. Data represent the mean ± SD of three independent experiments.
Mitochondria isolation assays
Mitochondrial fractions from cells were performed by using the Cell Mitochondria Isolation Kit (#C3601, Beyotime). Briefly, mitochondria were extracted in a Dounce homogenizer in mitochondrial buffer, followed by centrifugation at 1000 g for 10 min at 4°C. The supernatant was further centrifuged at 3500 g for 10 min at 4°C to pellet the mitochondria. The crude mitochondrial fraction was resuspended for washing and centrifuged at 12 000 g for 10 min at 4°C. The pellets were collected as the mitochondrial fraction.
In vitro transcription and translation
The coupled in vitro transcription and translation were performed according to the TNT Quick-coupled Transcription/Translation procedures described in the manufacturers manual (#L1170, Promega, Madison, WI). After the translation was terminated, synthesized proteins in vitro were purified by Anti-Flag Magnetic Beads (#P2115; Beyotime) according to the manufacturer’s instructions.
Enzyme-linked immunosorbent assay
Enzyme-linked immunosorbent assays (ELISAs) were conducted to assess the binding affinity, following a modified protocol described previously (29). Briefly, for investigating the dissociation constant (Kd) of DHX30 with G4s (mito0.5–22 and mito55), biotinylated G4 oligonucleotides were immobilized onto streptavidin-coated plates (#22351; Beaverbio). The plates underwent incubation with Flag-tagged DHX30 protein, anti-DHX30 antibody (#sc-390663; SANTA) and HRP-conjugated anti-mouse antibody (#A0216; Beyotime). Finally, the plates were incubated with ADHP (10-acetyl-3,7-dihydroxyphenoxazine, HRP substrate) and fluorescence emission was recorded at excitation/emission wavelengths of 545 nm/600 nm, respectively. The nucleotides are provided in Supplementary Table S2.
G4-unfolding assays
A mixture of 1μM Dabcyl-labeled oligonucleotide (S-mito0.5–22-Dabcyl) and 0.85 μM FAM-labeled oligonucleotide (F-Short-FAM) was prepared in 20 mM Tris–HCl buffer (pH 7.5) containing 5 mM MgCl2, 1 mM KCl and 99 mM NaCl. The mixture was heated to 95°C for 5 min and cooled to room temperature slowly. Unfolding assays were carried out as previously described (30) with slight modification. In brief, helicase reactions were carried out in triplicate in 96-well plates at 25°C and fluorescence monitored in a CLARIOstarplus microplate reader (BMG-Labtech, Germany). Each replicate consisted of a 40 μl solution containing 40 nM FAM-dabcyl system (pre-annealed), 200 nM Trap oligonucleotide (unlabeled and complementary to the FAM-labeled strand). Subsequently, varying amounts of DHX30 protein were added to corresponding well. Fluorescence emission was recorded every 10 s at excitation/emission wavelengths of 483 nm/530 nm, respectively. Once the maximum emission was reached and the signal was stable, 4 μl (2 μM) of a strand complementary to the dabcyl-labeled sequence (C-mito0.5–22) were added to every well, the plates were stirred for 10 s, and emission was monitored every 10 s.
Extracellular acidification rate assays
The extracellular acidification rate (ECAR) was measured by means of a glycolysis assay (E-BC-F069, Elabscience) according to manufacturer’s manual. Briefly, Cells were seeded onto a 96-well plate at a density of 2 × 104/well. Prior to measurement, cells were equilibrated for 2 h at 37°C without CO2 supply. ECAR was measured by adding 100 μl working solution (containing relevant compounds) to each well. Fluorescence emission was recorded every 2 min at excitation/emission wavelengths of 490 nm/535 nm, respectively for 3 h.
Quantitative real-time polymerase chain reaction (qRT-PCR) assays
Total RNAs were extracted from cultured cells using RNApure tissue&cell kit (CWBIO, China) according to the manufacturer’s protocol. The RNA concentrations were measured using a UV-vis spectrophotometer (Agilent). The cDNAs were synthesized from 500 ng of total RNAs using PrimeScript RT reagent kit (Takara, Japan) according to the manufacturer’s protocol. cDNA (10 ng) of each sample was used for quantitative PCR by using SYBR FAST qPCR kit (Kapa Biosystems, USA). RT-PCR of the indicated genes was performed at the BioRad IQ5 real-time PCR system (BioRad, Hercules, CA, USA). The setting procedures were as follows: 95°C for 10 min and then 95°C for 15 s, 60°C for 1 min with a total of 40 cycles. Data were collected by using qPCRsoft 4.1 (Analytik Jena AG). The relative expression levels of RNAs were calculated using the comparative Ct method. Melting curve analyses were performed on all PCRs to rule out the non-specific amplification. Actb (encoding β-actin) was used as the reference gene for accurate normalization of qPCR data. The primers were described in Supplementary Table S2.
Statistical analysis
All data reported in this work are from at least three independent experiments. The sample number per experiment and information on statistical methods and reproducibility are shown in each corresponding figure and figure legends. All statistical analyses were performed by GraphPad Prism software (La Jolla, CA, USA). Student’s t-test was performed. P < 0.05 was considered to be statistically significant.
Results
Design of an APEX2-based proximity labeling strategy to target APEX2 to mitochondrial DNA G4
We developed a mtDNA G4BPs profiling strategy by combining engineered peroxidase APEX2 (26) and a novel G4-recognizing nanobody SG4 (31). First, we engineered a SG4-APEX2 fusion construct carrying APEX2, SG4 and a mitochondrial targeting sequence (MTS) to bring APEX2 into close proximity to the mtDNA G4BPs. Upon activation by H2O2, APEX2 converts surrounding BP into biotin-phenoxyl radicals, which covalently conjugate with electron-rich amino acids residues (e.g. tyrosine, tryptophan, histidine and cysteine) of nearby G4BPs (Figure 1A). To help minimize false positives, two negative control (NC) experiments were designed, where either APEX2 enzyme was omitted or cells expressed mSG4-APEX2 fusion construct (i.e. lacking G4 targeting ability) (Figure 1B and Supplementary Figures S1 and S2). Moreover, we assessed the expression, co-localization, and activity of SG4-APEX2 and mSG4-APEX2 by western blotting and immunofluorescence staining assays (Figure 1C, D and Supplementary Figure S3). Western blotting assays revealed that both SG4-APEX2 and mSG4-APEX2 could be expressed in HEK293T cells (Figure 1C). Detailed analysis using confocal microscopy imaging with colocalization of TOMM70 (a mitochondrial marker) revealed that SG4-APEX2 and mSG4-APEX2 were localized in the mitochondria of HEK293T cells (Figure 1D). Meanwhile, the fusion of SG4 or mSG4 to APEX2 did not affect the activity of APEX2 in the mitochondria (Supplementary Figure S3). To verify the binding specificity of SG4-APEX2, we conducted a G4 pull-down assay (32). The nucleotide sequences used in the G4 pull-down assay were detailed in Supplementary Table S2. The G4 Pull-down assay indicated that SG4-APEX2 could efficiently capture G4s in vitro, but not mSG4-APEX2 (Figure 1E–G and Supplementary Figure S4A). Competitive pull-down experiments showed that unlabeled mitoG4 would inhibit the binding of SG4-APEX2 with FAM-labeled mitoG4. However, unlabeled ssDNA or dsDNA had little effect on the binding of SG4-APEX2 with FAM-labeled mitoG4s (Supplementary Figure S4B). To verify the interactions of SG4-APEX2 fusion protein and DNA G4s within mitochondria, we performed a mtDIP assay. Specifically, three distinct regions of mtDNA known to harbor potential G-quadruplex forming sequences (PQSs) (33,34) were amplified. The results showed that all three mtDNAs, containing PQSs, were enriched by mtDIP-based pull-down of SG4-APEX2, while two NCs mtDNAs (mtDNA 165–267 and mtDNA 4726–4876) were not enriched (Figure 1H). As expected, all five mtDNAs were not enriched by mtDIP-based pull-down of mSG4-APEX2 (Figure 1I). Furthermore, we tested the biotinylation specificity of SG4-APEX2 by the in vivo validation strategy (35). As a proof-of-concept study, the G4 antibody BG4 was selected to mimic the interaction of mitoG4s and G4BPs in vivo. As shown in the IP blotting result, the biotin labeling signal of BG4 was much stronger in the SG4-APEX2 group than in the mSG4-APEX2 group (Supplementary Figure S5A and B). Overall, our results suggest that we can leverage SG4-APEX2 to profile proteins that interact with mtDNA G4s under physiological conditions. Next, we employed the SG4-APEX2 to identify mitochondrial G4BPs via APEX2-based PL activity in HEK293T cells. To this end, we expressed SG4-APEX2 in HEK293T cells. The cells were treated by BP probes and H2O2. Imaging by immunofluorescence indicated that biotinylated proteins were colocalized with the SG4-APEX2 construct (Figure 2A). And western blot revealed that endogenous proteins in cell lysates were biotinylated (Figure 2B).
Design and implementation of APEX2-based proximity strategy to identify mtDNA G4BPs. (A) Schematic diagrams of SG4-APEX2 mediated mtDNA G4BP proximity strategy. (B) Design of two vectors for expressing SG4-APEX2 and mSG4-APEX2. (C) Western blotting analysis of HEK293T cells that expressed SG4-APEX2 and mSG4-APEX2. NC: no APEX2 expression. (D) Fluorescence imaging of SG4-APEX2 and mSG4-APEX2 localization to mitochondria. Fusion constructs were visualized by anti-Flag staining. Mitochondria were visualized by anti-TOMM70 staining. Nuclei were stained with Hoechst 33 342. Pixel intensity plots of the arrows are shown at right. Scale bars, 10 μm. (E) In vitro translation of SG4-APEX2 and mSG4-APEX2. (F) Schematic diagrams of G4 pull down assays. (G) Validation of the mitoG4 binding ability of SG4-APEX2 and mSG4-APEX2 constructs in vitro by G4-pull down assay. (Hand I) Validation of the mitoG4 binding ability of SG4-APEX2 (H) and mSG4-APEX2 constructs (I) in cells by mtDIP and quantitative real-time PCR (mtDIP-qPCR) assays. Two-sided Student’s t-test was performed and values represent means ± SD from three biological replicates.
Characterization and analysis of PL by the fusion constructs in HEK293T cells. (A) Confocal microscope imaging of biotin-labeled proteins after APEX2 labeling. Biotinylated proteins were visualized by streptavidin (SA)-coupled Cy3. APEX2 were detected by Flag staining. The nuclei were stained with Hoechst 33 342. Pixel intensity plots of the arrows are shown at right. Scale bars, 10 μm. (B) Gel analysis of biotinylated proteins. Lysates were analyzed by streptavidin blotting (left) or Coomassie stain (middle). Total eluted proteins after streptavidin bead enrichment were visualized by Coomassie stain (right). Substrates are BP and H2O2.
LC-MS/MS proteomic approach
To identify mitochondrial G4BPs captured by streptavidin beads, a quantitative liquid chromatography (LC)-MS proteomics approach was employed. After APEX2 PL, proteins were pulled down by streptavidin beads, followed by quantitative LC-MS/MS proteome analysis. In order to reduce false positive results, the proteins exclusively detected in the SG4-APEX2 groups, but not in the mSG-APEX2 and NC groups, were identified as putative G4BPs. We obtained 206 candidate G4BPs in HEK293T (Figure 3A). Gene Ontology (GO) analysis of the cellular component revealed that two of the significantly enriched cell components are mitochondrial matrix and mitochondrial protein complex (Figure 3B and C). The analysis of annotated biological processes and molecular function showed that the identified G4BPs candidates are implicated in mitochondrial biological processes and functions (Supplementary Figures S6 and S7). Among the enriched proteins from mitochondrial matrix and mitochondrial protein complex classes, we identified one previously reported G4BP LONP1 (36). Importantly, we identified numerous novel mtDNA G4BPs (Supplementary Table S3), such as DHX30 and a series of mitochondrial ribosomal proteins.
Profiling of mtDNA G4s interactome. (A) The Venn diagram shows the proteins exclusively detected in the SG4-APEX2 groups. (BandC) GO pathway enrichment analyses (cellular component) of G4BP candidates. Visualization of GO pathway enrichment results with network planning (B). Visualization of GO pathway enrichment results with bubble charts (C).
Characterization of candidate proteins in vitro and in cells
To verify the binding ability of candidate mtDNA G4-interacting proteins identified by our SG4-APEX2 PL system in vitro, we employed an optimized version of mitoG4 pull-down protocol (32). FAM-labeled mitoG4-forming oligonucleotides were incubated for 4 h at RT with candidate proteins in the presence of Anti-Flag beads in Tris–HCl buffer (10 mM Tris, pH = 7.4) containing 100 mM K+ and 10 mM Mg2+. The mitoG4-forming sequences used were mito0.5–22 (FAM-mito0.5–22), and the corresponding mutated sequences (FAM-mito0.5–22m) were used as controls (33). The FAM-G4/protein/beads assemblies were isolated and the FAM-G4 was resuspended in solution after a thermal denaturation step (15 min at 95°C). The binding efficiency of candidate proteins was quantified by the FAM emission of the resulting solution. As shown in Supplementary Figure S8A–C, LONP1, MRPS34 and DHX30 could capture FAM-mitoG4 rather than the corresponding mutated single-stranded sequences, indicating their G4 binding ability. Among these newly identified mtDNA G4BPs, we were particularly interested in DHX30. As a member of the DExH-box helicase family (37), DHX30 is mainly present in mitochondrial matrix (38) and possesses adenosine triphosphate (ATP)-dependent RNA unwinding activity (39), which might implicated in RNA/DNA conformational change. Firstly, immunofluorescence indicated that DHX30 localized partially to mitochondria, which was consistent with previous work (Figure 4A) (38). Isolated mitochondria and whole cell lysates were analyzed by western blot, also confirming mitochondria localization of DHX30 (Figure 4B and C). To further assess the binding affinities for DHX30 with mtDNA G4 in vitro, we carried out ELISAs (Figure 4D). DHX30 displayed selective and high-affinity binding to G4s (e.g. mito0.5–22 and mito55) with Kd = 109.1 ± 1.8 nM and Kd = 165.8 ± 16.1 nM, respectively (Figure 4E and F). Consistent with mtDNA G4 pull-down experiments, weaker binding was observed towards the control oligomers.
Validation of DHX30 as a novel mtDNA G4BP. (A–C), DHX30 partially localized to mitochondria. (A) Fluorescence imaging of endogenous DHX30 localization. DHX30 was detected by DHX30 antibody, Anti-Tomm20 antibody stains represented a mitochondria marker and the nuclei were detected by Hoechst 33 342. Pixel intensity plots of the white arrow were shown on the right. (scale bar, 10 μm). (B) Relative DHX30 protein levels in whole cell lysates and mitochondrial fractions. One of three independent experiments was shown. (C) Relative DHX30 protein levels obtained from densitometric quantification of all the blots and relative to the reference proteins, TOMM20 (n = 3); *P < 0.05. (D) Schematic representation of DHX30 recognition by antibodies and chromogenic detection in vitro (ELISA, ADHP = Amplex Red). (E) Binding curves determined by an adapted ELISA assay for mito0.5–22 as example of mtDNA G4, mito0.5–22m as example of G-rich sequences not able to fold in a G4 structure and dsDNA. (F) Binding curves determined by an adapted ELISA assay for mito55 as example of mtDNA G4 and mito55m as example of ssDNA. Dissociation constants (Kd) estimated from curve fitting. Error bars represent the SD calculated from three replicates.
To further validate G4 binding interactions of DHX30 in cells, immunofluorescence assays and mtDIP coupled with qPCR analysis (mtDIP-qPCR) were performed. Firstly, the presence of DHX30 on G4 did not hinder the binding of BG4 antibody to G4 (Figure 5A and B). Therefore, we transiently transfected Flag-tagged DHX30 in HEK293T cells and stained G4 structures in DHX30-OE cells using the G4-specific antibody BG4 (2). As expected, we observed green DHX30 puncta colocalized with red BG4 puncta (Figure 5C), indicating that DHX30 could interact with G4 structures in living cells. Western blotting indicated that transgenic Flag-tagged DHX30 accumulated to approx. 2-fold higher levels compared to endogenous DHX30 in HEK293T cells, which excluded the false positive caused by high overexpression of DHX30 (Supplementary Figure S9). Besides, we also observed the colocalization between endogenous DHX30 and G4s (Figure 5D). The overexpression of DHX30 would lead to more DHX30 to be biotinylated by SG4-APEX2 (Figure 5E). Figure 5F showed that all three mtDNA G4 sequences were enriched by mtDIP-based pull-down of Flag-tagged DHX30, while two NC mtDNAs were not enriched. We therefore summarized that DHX30 could specifically bind mitoDNA G4 in vitro and in cells.
DHX30 binding mtDNA G4 in cells. (Aand B), The BG4/mitoG4 complexes were captured by DHX30 through Anti-Flag Magnetic Beads (A), detected by western blotting assays (B). (C) Immunofluorescence image of Flag-tagged DHX30 and BG4 in HEK293T cells. The colocalized foci were indicated by white arrows. Pixel intensity plots of the yellow arrow were shown on the right. Scale bars, 10 μm. (D) Immunofluorescence image of endogenous DHX30 and BG4 in HEK293T cells. The colocalized foci were indicated by white arrows. Pixel intensity plots of the yellow arrow were shown on the right. Scale bars, 10 μm. (E) Western blot analyses for DHX30 OE and wild type cells. The IP band was to detect biotin-labeled DHX30, and the input band was to detect total DHX30. (F) Validation of the mitoG4 binding ability of Flag-DHX30 in cells by mtDIP-qPCR assays. Error bars indicate SD.
DHX30 is a mtDNA G4 resolvase
Furthermore, to investigate DHX30’s potential as a mtDNA G4s helicase, we performed an in vitro fluorescence-based helicase assay using purified DHX30 proteins (30). In this assay, a dabcyl quencher containing-mitoG4 oligonucleotide (S-mito0.5–22-dabcyl) is hybridized with a 5′ 6-FAM labeled short complementary oligonucleotide that has the potential to emit fluorescence only when unpaired from the quencher (Figure 6A). The FAM fluorescence increase was observed in a DHX30 concentration-dependent manner in the absence of ATP, indicating that DHX30 led to the unwinding of the mitoG4-containing quencher (Figure 6B). The unfolding percentage reached to 73% at high DHX30 concentrations (up to 80 nM) (Figure 6C). Besides, the addition of ATP did not appreciably alter the mitoG4 unfolding activity of DHX30 (Figure 6D and E). These data suggested that the ATP hydrolysis was not directly coupled to mitoG4-unfolding by DHX30. Recent studies have shown similar ATP-independent activities of DDX5 helicase (40) and BLM helicase (41) when they unfold DNA G4 structures. The mechanism by which DHX30 unwinds mtDNA G4s in the absence of ATP remains unexplored. Exploring protein structures of DHX30 in complex with mtDNA G4 structures in the future holds the potential to yield mechanistic insights. Next, given that DHX30 functions in resolving mtDNA G4, we hypothesized that the DHX30 overexpression (DHX30 OE) would lead to a reduction in the formation of mtDNA G4 in living cells. To confirm this hypothesis, we stained HSP60 (a mitochondrial marker) and G4 in wild-type (WT) and DHX30 OE cells. A significant decrease of BG4 signal in mitochondria in DHX30 OE cells was observed, indicating that DHX30 could unfold mitoG4 in living cells (Figure 7A).
DHX30 was involved in resolving of mtDNA G4 in vitro. (A) Schematic representation of the helicase assay developed by Mendoza et al (30,49). (B) Representation plots of emission as a function of time for unwinding of mtDNA G4 system (S-mito0.5–22-dabcyl). DHX30 was added to begin the reaction (t = 1200 s); the complementary strands (C-mito0.5–22) were added once the reactions reached a plateau (t = 8400 s). (C) Quantitation of DHX30 helicase activity; error bars indicate SD. (D) The addition of 1 mM ATP has no effect on the helicase activity of DHX30. (E) Quantitation of DHX30 helicase activity with or without ATP; error bars indicate SD.
DHX30 was involved in resolving mtDNA G4s in living cells. (A) DHX30 overexpress lead to decreased mitochondrial G4 signal, revealed by labeling of WT and DHX30 overexpress HEK293T cells with the BG4 antibody. HSP60 served to mark the mitochondria. Scale bars, 10 μm. Left, quantification of the HSP60/BG4 foci number (n = 47). Two-tailed Student’s t test. ****P < 0.0001. (B) DHX30 was involved in resolving mitoG4 in MCF-7 cells. DHX30 overexpress lead to decreased mitoG4 signal. While mtDNA G4 stabilizer, RHPS4, could elevate mitoG4 levels. Left, quantification of the HSP60/BG4 foci number (n = 30). Two-tailed Student’s t test. **P < 0.01.
Mitochondrial DNA G4 binding protein affects mtDNA G4s mediated glycolysis
Given that endogenous mtDNA G4s were linked to enhanced tumor cellular glycolysis (42,43), we speculated that DHX30 might affect cell glycolysis by resolving mtDNA G4s. To confirm this hypothesis, we first detected G4 structures in WT and DHX30 OE MCF-7 cells. We also observed a significant BG4 signal from the mitochondria in MCF-7 cells, consistent with HEK293T cells, that decreased upon DHX30 OE. Whereas treating DHX30 OE cells with RHPS4, a known mtDNA G4s stabilizer (44,45), resulted in increased BG4 signal in mitochondria (Figure 7B). Next, we measured glycolytic activity in MCF-7 cells using the glycolysis stress test in which lactic acid production is determined by the ECAR. Remarkably, the overexpression of DHX30 led to a significant reduction of glycolysis and glycolytic capacity (Figure 8A and B). Conversely, when RHPS4 was employed to stabilize and rescue the mtDNA G4 structures in MCF-7 cells with DHX30 OE, the ECAR levels, as well as glycolysis and glycolytic capacity, were partially reinstated. Furthermore, we focused on several glycolysis-related genes and found that these genes were markedly decreased after DHX30 OE and partially rescued after treatment with RHPS4 (Figure 8C), also suggesting a decrease in the glycolytic activity of MCF-7 cells after DHX30 OE. Collectively, these results indicated that DHX30 could unfold mtDNA G4 structures to regulate mtDNA G4s mediated glycolysis activity. Besides, considering that DHX30 is a mitochondrial nucleoid protein and is required for mitochondrial ribosome assembly (46), it may affect glycolysis in a mtDNA G4s-independent manner. In future research, it is necessary to explore the relationship among DHX30, mtDNA G4s and tumor cellular glycolysis in more detail.
The impact of DHX30 OE on the glycolysis of MCF-7 cells. (Aand B), The impact of DHX30 OE on glycolytic rates in MCF-7 cells treated with or without RHPS4. (C) The expression of glycolysis-related genes in DHX30 OE MCF-7 cells versus WT cells, detected by qRT-PCR assays. Data were shown in three repeated experiments. Error bars represent SD. Two-tailed Student’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.
Discussion
In summary, we developed a new subcellular selective APEX2-based PL strategy to identify the interactome of mtDNA G4 structures in living cells. In our design, MTS and DNA G4 nanobody SG4 were employed to specifically target mtDNA G4s in living cells and APEX2-mediated PL was used to capture the endogenous mtDNA G4BPs. By using the approach, we identified many mtDNA G4-associated proteins. Among them, we found that DHX30 could unfold mtDNA G4 structures to regulate mtDNA G4s mediated cellular glycolysis, further indicating the close relationship between mtDNA G4s and cell glycolysis. Compared with G4BPs identification methods carried out in vitro or G4 ligand-based PAL strategies, the in situ enzymatic catalysis PL promotes amplification of target-tag crosslinking. Besides, benefiting from the high spatiotemporal resolution and the ability of genetically encoded systems to perform the labeling with exquisite specificity within living cells, our approach could label subcellular localized G4BPs. In reviewing the results of this study, the limitation about the spatial specificity should be kept in mind. GO analysis of the cellular component revealed that many of these proteins were also associated with non-mitochondrial structures. The primary factor could be attributed to the affinity of SG4 towards cytoplasmic RNA G4, thereby facilitating the recruitment of cytoplasmic RNA G4-binding proteins into mitochondria. Nevertheless, the combination of this approach with experimental validation could still effectively facilitate the identification of novel mitochondrial G4-binding proteins. And we would also endeavor to optimize the technique in order to enhance its spatial specificity. We envisage that future studies with other subcellular regions targeting sequences fused to SG4-APEX2 may employ a similar approach to explore DNA G4BPs in other subcellular regions, such as nucleolus, the promyelocytic leukemia nuclear bodies and Cajal body. As PL has been used to identify interacting proteins of certain nucleic acid second structures [e.g. G4s (25,47) and R-loops (48)], we anticipate that similar PL approaches should be generally applicable to map endogenous interactomes of other nucleic acid structural features (e.g. i-motif, Z-DNA and triplex DNA).
Data availability
The data that support the findings of this study are available from the authors on reasonable request.
Supplementary data
Supplementary Data are available at NAR Online.
Funding
National Key Research and Development Program of China [2019YFA0709202, 2021YFF1200700]; National Natural Science Foundation of China [22437006, 22237006, T2495262, 22107098, 22122704]. Funding for open access charge: National Key Research and Development Program of China; National Natural Science Foundation of China.
Conflict of interest statement. None declared.
Comments