Plasmid construction

Full-length human GTPBP8 coding sequence (GenBank Accession No. NM_014170.2) in the pENTR221 vector (ORFeome Library; Biocentrum Helsinki Genome Biology Unit) was cloned with XhoI and BamHl restriction sites into the pEGFP-N1 vector (Takara Bio Inc.). GTPBP8 was cloned into the pHAT vector (provided by the EMBL Protein Expression and Purification Facility) with BcuI and BamHI cloning sites. GTPBP8-myc was constructed from GTPBP8-GFP with GFP replaced by myc. GTPBP81–46aa-GFP and GTPBP847–284aa-GFP were subcloned into the pEGFP-N1 vector. Inserts were sequenced to confirm the cloning. GTPBP8 was cloned into the pGAT2 vector using NcoI and XhoI restriction sites. LR recombination was performed between the GTPBP8 entry clone and the MAC-C destination vector [32] to generate the MAC-tagged GTPBP8 expression vector.

GTPase activity assay

The GTPase activity of purified GTPBP8 protein was assessed using the Enzcheck assay kit (E6646, Life Technologies) according to the manufacturer’s instructions. GTPBP8 (12 μM) was incubated with various concentrations of GTP (0, 0.094, 0.1875, 0.375, 0.75, 1.5, and 3 mM) in the kit's reaction buffer at 25 °C. Absorbance at 360 nm was immediately recorded every 4 min for a duration of 60 min using the VarioskanTM LUX multimode microplate reader (Thermo Fisher Scientific) with SkanIt Software 5.0. The amount of inorganic phosphate released from GTP hydrolysis at each time point was determined by extrapolation using a phosphate standard curve. Kinetic parameters were determined by fitting the data to the Michaelis–Menten equation using OriginPro 2018.

Cell culture and transient transfections

Human osteosarcoma (U2OS) cells, which regularly tested negative for mycoplasma with MycoAlert Plus (Lonza), were cultured in high-glucose DMEM containing 10% FBS (12-614F, Lonza), 10 U/mL Penicillin, 10 µg/mL Streptomycin and 20 mM L-glutamine (10378-016; Gibco) in a humidified 95% air/5% CO2 incubator at 37 °C. For transient transfections, cells were plated on culture plates and transfected the next day with DNA constructs using FugeneHD reagent (#E2311, Promega), according to the manufacturer’s instructions.

For silencing experiments, cells were transfected with 30 pmol of target gene or control siRNA directly after replating on the 6 cm culture plate using Lipofectamine® RNAiMAX (#13778-075, ThermoFisher), according to the manufacturer’s instructions. Cells were transfected again after 48 h with the same amount of siRNA and incubated for another 72 h. All the samples were subjected to mass spectrometry and western blot analysis. The following siRNA target sequences were used: GTPBP5 siRNA: 5′-CGGUGGACACGUCAUUCUGTT-3′ (134621, Ambion); GTPBP7 siRNA: 5′- GCAACACUUAGAAGGAGAAGGCCUA-3′ (1362318, Invitrogen); GTPBP8 siRNA#1: 5′-AAAGTTACTATGTAAGCCTAA-3′ (SI04257974); siRNA #2: 5′-AGCGACTGAGCCGCTATAATA-3′ (SI04232011, Qiagen); #3: 5′-AAGCATCGATAGGTAAGTTGA-3′ (SI03130008, Qiagen); #4: 5′-CCGGTTTAGCTGAAGATTCAA-3′ (SI00443779, Qiagen), GTPBP10 siRNA 5′-TTGCGTGTTGTTCAGAAAGTA-3′ (SI04308647, Qiagen), uL16m siRNA: 5′ TACGGAGTTTACAGAAGGCAA-3′ (SI00648291, Qiagen) and Negative control siRNA (SI03650318, Qiagen), MRPL45 siRNA: 5′-CTGGAGTATGTTGTATTCGAA-3′ (SI00649005, Qiagen), bL27m siRNA 5′CAGGCAGACGCCAAGGCATTA-3′, uL11m siRNA 5′-AGGAAGGAGGTCACACCAATA-3′ (SI04135684, Qiagen), mL65 siRNA 5′CAAGCUAUGUAUCAAGGAUtt3′ (s21375, ThermoFisher Scientific) bL31m siRNA 5′-CCAGGCTTATGCACGACTCTA-3′ (SI00649271, Qiagen), mL64 siRNA 5′-AAGAACGCGAATGGTACCCGA-3′ (SI02652349, Qiagen), and bL36m siRNA 5′-CGGTGGTACGTCTACTGTAAA-3′ (SI04156299, Qiagen).

In the rescue experiments, U2OS cells were treated with either control or GTPBP8 siRNA altogether for 8 days (siRNA transfection on the 1st and 3rd day on replated cells) out of which the last 48 h the GTPBP8-depleted cells were rescued with the GTPBP8-myc construct using FugeneHD reagent according to the manufacturer’s instructions. The control cells were replated after 5 days of AllStars negative-control siRNA (Qiagen) treatment, and all the control, GTPBP8 depleted, and GTPBP8 rescued cells were collected at the same time on the 8th day.

Immunofluorescence microscopy

U2OS cells were grown on coverslips and fixed with 4% paraformaldehyde for 20 min, washed three times with PBS, and permeabilized with 0.1% Triton X-100 in PBS for 5 min. For antibody staining, permeabilized cells were incubated with primary antibody for 1 h at 37 °C. Coverslips were washed with Dulbecco plus 0.2% BSA and incubated with fluorescence-conjugated secondary antibody for 1 h at RT. Cells were mounted in Mowiol supplemented with 1,4-Diazabicyclo[2.2.2]octane (DABCO). The following antibodies were used: rabbit anti-GTPBP8 (1:100, HPA034831, Sigma), mouse anti-TOM20 (1:200, sc-17764, Santa Cruz Biotechnology), mouse anti-myc (1:100, sc-40, Santa Cruz Biotechnology), rabbit anti-COXIV (1:100, MA5-15078, ThermoFisher Scientific). Cells were imaged with a DM6000B microscope (Leica Biosystems) equipped with a Hamamatsu Orca-Flash 4.0 V2 sCMOS camera and LAS X software (Leica), using 63x/1.4–0.60 HCX PL Apochromat objective and brightline filters (Semrock): GFP-4050B (excitation, 466/40; emission, 525/50), TRITC-B (excitation, 543/22; emission, 593/40).

Immunoblotting and antibodies

Cells were solubilized in PBS containing 1% dodecyl maltoside (DDM), 1 mM PMSF, 10 mM sodium azide, 10 mM sodium ascorbate, and 5 mM Trolox. Protein concentrations were measured by the Bradford assay (Bio-Rad Laboratories). Equal amounts of proteins were loaded and separated by Tris–glycine SDS-PAGE, and transferred to PVDF membranes. Primary antibodies were incubated overnight at 4 °C and detected the following day with HRP-linked secondary antibodies (1:10,000) using the ECL reagent. The following primary antibodies were used: mouse anti-ATP5A (1:1000, 14748, Abcam), rabbit anti-GFP (1:4000, 50430-2-AP, Proteintech), mouse anti-MT-CO1 (1:2000, 14705, Abcam), rabbit anti-COXIV (1:1000, MA5-15078, ThermoFisher Scientific), rabbit anti-CytB (1:3000, 55090-1-AP, Proteintech), mouse anti-CytC (1:10,000, 05-479, Millipore), goat anti-GAPDH (1:1000, sc-20357, Santa Cruz Biotechnology), rabbit anti-GTPBP5 (1:2000, 20133-1-AP, Proteintech), rabbit anti-GTPBP7 (1:4000, 13742-1-AP, Proteintech), rabbit anti-GTPBP8 (1:500, HPA034831, Sigma), rabbit anti-GTPBP10 (1:500, NBP1-85055, Novus Biologicals), mouse anti-HSP60 (1:20,000, SMC-110, StressMarq Biosciences), rabbit anti-uL1m[MRPL1] (1:2000, 16254-1-AP, Proteintech), rabbit anti-uL2m[MRPL2] (1:2000, 16492-1-AP, Proteintech), rabbit anti-uL11m[MRPL11] (1:6000, 15543-1-AP, Proteintech), rabbit anti-bL12m[MRPL12] (1:4000, 14795-1-AP, Proteintech), rabbit anti-uL16m[MRPL16] (1:500, HPA054133, Sigma), rabbit anti-bL21m[MRPL21] (1:3000, PA5-31939, ThermoFisher Scientific), rabbit anti-uL24m[MRPL24] (1:2000, 16224-1-AP, Proteintech), rabbit anti-bL27m[MRPL27] (1:1000, 14765-1-AP, Proteintech), rabbit anti-mS27[MRPS27] (1:1000, 17280-1-AP, Proteintech), rabbit anti-mL65[MRPS30] (1:1000, 18441-1-AP, Proteintech), rabbit anti-mS35[MRPS35] (1:10,000, 16457-1-AP, Proteintech), rabbit anti-bL36m[MRPL36] (1:1000, ab126517, Abcam), rabbit anti-mL40[MRPL40] (1:1000, HPA006181, Sigma), rabbit anti-mL45[MRPL45] (1:4000, 15682-1-AP, Proteintech), rabbit anti-mL46[MRPL46] (1:2000, 16611-1-AP, Proteintech), rabbit anti-bL31m[MRPL55] (1:1000, 17679-1-AP, Proteintech), rabbit anti-mL64[MRPL59] (1:500, 16260-1-AP, Proteintech), mouse anti-myc (1:1000, sc-40, Santa Cruz Biotechnology), mouse anti-NDUFA9 (1:1000, 20312-1-AP, Proteintech), mouse anti-TOM20 (1:1000, sc-17764, Santa Cruz Biotechnology), rabbit anti-TOM40 (1:1000, 18409-1-AP, Proteintech), rabbit anti-TRMT61B (1:500, 26009-1-AP, Proteintech), mouse anti-SDHA (1:1000, 14715, Abcam).

Mitochondrial isolation and submitochondrial fractionation

Mitochondria were isolated and purified as described previously [33]. Briefly, cells were collected and resuspended in a mitochondrial isolation buffer (10 mM Tris-MOPS, 1 mM EGTA, 200 mM sucrose, pH 7.4) and homogenized with a Teflon potter (Potter S, Braun). Cells that had not been lysed were sedimented with 600 ×g for 5 min at 4 °C before being discarded. The supernatant was then centrifuged for 10 min at 4 °C at 9000 ×g. The resulting pellet was dissolved in ice-cold mitochondrial isolation buffer and centrifuged at 9000 ×g for 10 min at 4 °C. The pellet includes the mitochondrial fraction.

For identification of the sub-mitochondrial localization of GTPBP8, the isolated mitochondria were sub-fractionated to obtain mitoplasts using a phosphate swelling-shrinking method [34, 35]. Briefly, purified mitochondrial pellets were resuspended in a swelling buffer (10 mM KH2PO4, pH 7.4) and incubated for 20 min with gentle mixing. To keep the mitoplasts intact, mitochondria were mixed with an equal volume of shrinking buffer (10 mM KH2PO4, pH 7.4, 32% sucrose, 30% glycerol, and 10 mM MgCl2) for another 20 min. The purified mitochondria and mitoplast were resuspended in a homogenization buffer (10 mM Tris-MOPS, 1 mM EGTA, 200 mM sucrose, pH 7.4) and treated with 0.2 mg/mL proteinase K with or without 1% NP-40 for 30 min. Proteinase K activity was quenched with 2 mM PMSF for 10 min. All steps were carried out on ice. 1% SDS was added to solubilize the mitochondrial proteins and immunoblotted as indicated.

To detect the mitochondrial membrane binding of GTPBP8, mitochondria were subjected to alkaline extraction in freshly prepared 0.1 M Na2CO3 at pH 11, 11.5, and 12. Mitochondrial membranes were pelleted at 72,000 ×g for 30 min at 4 °C using an optimaTM ultracentrifuge (Beckman) with Beckman rotor (TLA 120). The membrane pellets (P) were dissolved in Laemmli loading buffer (2% SDS, 10% glycerol, 60 mM Tris–HCl, 0.005% bromophenol blue). Supernatants (S) were further precipitated using trichloroacetic acid (TCA) to a final concentration of 13% on ice for 30 min. After centrifugation at 20,000×g for 30 min, pellets were washed twice with ice-cold acetone and dissolved in Laemmli loading buffer. Both pellets (P) and supernatant (S) were subjected to SDS-PAGE and western blot analysis.

Transmission electron microscopy

For pre-embedding immuno-EM, cells were grown on coverslips and fixed with PLP fixative (2% formaldehyde, 0.01 M periodatem, 0.075 M lysine-HCl in 0.075 M phosphate buffer, pH 7.4) for 2 h at room temperature, as described previously [36]. Cells were permeabilized with 0.05% saponin before being immunolabeled with anti-myc antibody in a 1:50 dilution, followed by Nano-gold mouse IgG in a 1:60 dilution. Nano-gold was silver enhanced using the HQ Silver kit for 5 min and gold toned with 0.05% gold chloride. After washing, the cells were further processed for embedding.

Blue native (BN)-PAGE

The abundance of respiratory chain complexes was analyzed by Blue Native polyacrylamide gel electrophoresis (BN-PAGE) using a NativePAGETM Novex® Bis–Tris Gel System according to manufacturer’s instructions. Briefly, mitochondria were suspended in 1% DDM in PBS on ice for 30 min and centrifuged at 22,000 ×g for 30 min. The supernatant was supplemented with 0.2% Coomassie G250 and loaded on a 3–12% gradient gel. Proteins were transferred to the polyvinylidene fluoride (PVDF) membrane (Millipore, USA) which was subsequently fixed with 8% acetic acid for 15 min, washed with MilliQ and methanol. The proteins were probed with the indicated antibodies.

Viability assays

Cell proliferation/viability was determined using the Thiazolyl Blue Tetrazolium Bromide (MTT) reagent. Briefly, MTT (M2128, Sigma-Aldrich) solution was added to the cell culture medium with a final concentration of 0.5 mg/ml at 24, 48, and 72 h during siRNA treatment. The medium was discarded after the incubation with MTT at 37 °C for 4 h. Subsequently, 150 µL DMSO was added to dissolve the formazan, and the quantity of formazan was measured by the absorbance at 570 nm using a Varioskan™ LUX multimode microplate reader (Thermo Fisher Scientific) with SkanIt Software 5.0.

Measurement of oxygen consumption rate (OCR) and ATP production

After treatment with control or GTPBP8 siRNA for 5 days, mitochondrial function was assessed by detecting OCR and ATP production. Cellular oxygen consumption rates were measured using a Seahorse XF96e analyzer (Seahorse Bioscience, Agilent Technologies) as described before [37]. The XF96 sensor cartridge was activated with 200 µL of XF96 calibration solution per well for 12 h at 37 °C. U2OS cells were seeded onto XF96 cell culture microplates at 1 × 104 cells per well and treated with GTPBP8 or control siRNA for 5 days. One hour prior to measurement, the culture medium was changed to serum-free and bicarbonate-free Dulbecco’s Modified Eagle’s medium supplemented with 10 mM glucose, 5 mM pyruvate and 5 mM glutamine. After incubation for 1 h at 37 °C in a non-CO2 incubator, steady-state and post-intervention analyses were performed. Respiration was assessed by the injection of oligomycin (1 µM) to inhibit the mitochondrial ATP synthase, carbonyl cyanide-p-trifluoromethoxy-phenylhydrazone (FCCP, 1 µM) to collapse the mitochondrial membrane potential, and rotenone (1 µM) and antimycin A (1 µM) to inhibit the respiratory chain. OCR was measured before and after the addition of inhibitors at the indicated time points. The OCRs were normalized to total protein amounts to account for alterations in cell density.

Cellular ATP levels were detected by an ATP determination kit (A22066, ThermoFisher Scientific) according to the manufacturer's instructions. After treatment with GTPBP8 siRNA for 5 days, cells were harvested and lysed on ice for 10 min using the DDM lysis buffer without proteinase inhibitors. The cell lysates were briefly sonicated and centrifuged at 16,000 ×g for 10 min at 4 °C. 10 µL supernatant was mixed with 90 µL reaction solution and luminescence measured using a Varioskan™ LUX multimode microplate reader (Varioskan Flash, Thermo Fisher Scientific, Inc., Waltham, MA USA). The amount of ATP was calculated from the standard curve and normalized to the total protein amount.

Radioisotope labeling of mitochondrial translation

Mitochondrial protein synthesis in cultured cells was detected through metabolic labeling with [35S] methionine/cysteine in the presence of anisomycin to inhibit cytoplasmic ribosomes [38]. After treatment with targeted or control siRNA for 5 days, cells were washed three times with PBS and pretreated with 100 µg/mL anisomycin for 5 min to inhibit cytoplasmic translation. Subsequently, [35S] methionine/cysteine (EasyTag, PerkinElmer) was added with a final concentration of 400 μCi for pulse labeling assays. After pulse labeling for 30 min, cells were washed twice either with PBS (pulse samples) or medium without [35S] methionine/cysteine (chase labeling). After 6 h of chasing, cells were scraped and treated with benzonase® Nuclease (E1014, Sigma) according to the manufacturer’s instructions. Gel loading buffer (186 mM Tris–HCl, pH 6.7, 15% glycerol, 2% SDS, 0.5 mg/mL bromophenol blue, and 6% -mercaptoethanol) was added to the samples before loading onto a 12–20% gradient SDS-PAGE gel to separate proteins. The gel was then dried for exposure to a phosphor screen and scanned with a FUJIFILM FLA-5100. Gels were rehydrated in water and Coomassie-stained to confirm equal loading.

Isokinetic sucrose gradient assays

The sedimentation properties of GTPBP8 and the ribosomal proteins in sucrose gradients were analyzed as described before [36]. Cells or mitochondria were lysed in 1% DDM lysis buffer (50 mM Tris, pH 7.2, 10 mM Mg(Ac)2, 40 mM NH4Cl, 100 mM KCl, 1% DDM, 1 mM PMSF, 6 µl/mL Chloramphenicol, and 1 mM ATP) for 20 min on ice, then centrifuged at 20,000 ×g for 20 min at 4 °C. A supernatant containing 900 g of total protein was loaded onto a 16 mL linear 10–30% sucrose gradient (50 mM Tris, pH 7.2, 10 mM Mg(Ac)2, 40 mM NH4Cl, 100 mM KCl, and 1 mM PMSF) and centrifuged for 15 h at 4 °C with 74,400 ×g (SW 32.1 Ti; Beckman Coulter). 24 equal volume fractions were collected from the top and subjected to TCA precipitation. Samples were separated by SDS-PAGE for subsequent immunoblotting. For RNaseA treatments, RNaseA (ThermoFisher Scientific) was added to the cell lysate at a final concentration of 600 U/mL at the stage of protein sample preparation, and sucrose gradients were prepared as described.

RNA and DNA isolation and real-time quantitative PCR (RT-qPCR)

Total RNA was extracted from the whole cells and sucrose fractions using the GeneJET RNA purification kit (K0731; Thermo Fisher Scientific) and Trizol reagent (Invitrogen), respectively. Single-stranded cDNA was synthesized from extracted mRNA with the Maxima first strand cDNA synthesis kit (K1671; Thermo Fisher Scientific). For quantification of mtDNA copy numbers, total DNA was extracted from cells using a total DNA, RNA, and protein isolation kit (Macherey–Nagel) following the manufacturer’s instructions, and mtDNA copy numbers were assessed using quantitative PCR analyses. β-globin was used to normalize the copy numbers of mtDNA. Quantitative PCR reactions were performed with Maxima SYBR Green/ROX (K0221; Thermo Fisher Scientific) by Bio-Rad Laboratories CFX96. GAPDH and 18S rRNA were used as controls for genes encoded in the nucleus and mitochondria, respectively. Changes in expression levels were calculated with the 2−ΔΔCt method. Primers used in this study were: GTPBP8 F: 5′- GCGGCCAGAGGTGTGTTTTA-3′, GTPBP8 R: 5′-CCATAACCTGGCATGTCCAC-3′; GAPDH F: 5′- TGCACCACCAACTGCTTAGC-3′, GAPDH R: 5′- GGCATGGACTGTGGTCATGAG-3′; MT-CO1 F: 5′- CTCTTCGTCTGATCCGTCCT-3′, MT-CO1 R: 5′-ATTCCGAAGCCTGGTAGGAT-3′; MT-CytB F: 5′- TAGACAGTCCCACCCTCACA-3′, MT-CytB R: 5′-CGTGCAAGAATAGGAGGTGGA-3′; mt-tRNAval F: 5′- TAGACAGTCCCACCCTCACA-3′, mt-tRNAval R: 5′-CGTGCAAGAATAGGAGGTGGA-3′; 12S rRNA F: 5′- TAGAGGAGCCTGTTCTGTAATCGA-3′, 12S rRNA R: 5′-TGCGCTTACTTTGTAGCCTTCAT-3′; 16S rRNA F: 5′-AGAGAGTAAAAAATTTAACACCCAT-3′, 16S rRNA R: 5′-TTCTATAGGGTGATAGATTGGTCC-3′; 18S rRNA F: 5′-CCAGTAAGTGCGGGTCATAAGC-3′, 18S rRNA R: 5′-CCTCACTAAACCATCCAA TCGG-3′; mtDNA F: 5′-ACCACAGTTTCATGCCCATCGT-3′, mtDNA R: 5′-TTTATGGGCTTTGGTGAGGGAGGT-3′; β-globin F: 5′-GGTGAAGGCTCATGGCAAGAAAG-3′ and β-globin R: 5′-GTCACAGTGCAGCTCACTCAGT-3′.

Co-immunoprecipitation analysis

Immunoprecipitation of expressed GTPBP8-myc was performed using Dynabeads™ Protein G (10004D, Invitrogen) according to the manufacturer's instructions. Briefly, 50 µL Dynabeads™ Protein G beads were first coated with 1 µg mouse anti-MYC antibody (sc-40, Santa Cruz Biotechnology) or normal mouse IgG1 (sc-3877, Santa Cruz Biotechnology). After transient expression of GTPBP8-myc in U2OS cells for 48 h, mitochondria were extracted and lysed in ice-cold 1% DDM in PBS buffer containing a complete protease inhibitor cocktail. After 60 min of incubation on ice, the insolubilized material was removed by centrifugation at 20,000 ×g for 20 min at 4 °C. The supernatants were subsequently incubated with antibody-bound Dynabeads™ Protein G overnight at 4 °C. After washing, the protein G Dynabeads-antibody-protein was eluted with 50 mM glycine (pH 2.8) to dissociate the complex at RT for 2 min. The samples were subjected to SDS-PAGE and immunoblotting.

Proximity protein purification

U2OS were transfected with the MAC-tagged GTPBP8 expression vector or a control MAC-tagged GFP vector using Fugene 6 (Invitrogen) according to the manufacturer’s instructions. Cells were treated with 50 μM biotin for 24 h before harvesting, and one biological sample contains cells derived from 5 × 150 mm fully confluent dishes (approximately 5 × 107 cells). Each sample has three biological replicates. Samples were frozen at − 80 °C until further use. For proximity purification (BioID), cells were lysed and loaded onto columns (Bio-Rad) containing the Strep-Tactin matrix (IBA, GmbH). The detailed procedures are described in [39].

Mass spectrometry

Protein lysate or pulldown samples were processed for a reduction of the cysteine bonds with 5 mM Tris(2-carboxyethyl)phosphine (TCEP) for 30 min at 37 °C and alkylation with 10 mM iodoacetamide. The proteins were then digested into peptides with sequencing grade modified trypsin (Promega V5113) at 37 °C overnight. Digested peptides were used for mass spectrometry analysis. The analysis was performed on a Q-Exactive mass spectrometer using Xcalibur version 3.0.63 coupled with an EASY-nLC 1000 system via an electrospray ionization sprayer (Thermo Fisher Scientific). Mass spectrometry was in a data-dependent acquisition mode using FTMS full scan (300–1700 m/z) resolution of 60,000 and collision-induced dissociation (CID) scan of the top 20 most abundant ions [40]. BioID samples were run in quadruplicates, GTPBP8 knockdown samples in triplicates, respectively.

Data processing for mass spectrometry

For protein identification, Thermo.RAW files were uploaded into the Proteome Discoverer 1.4 (Thermo Scientific) and searched against the Sequest search engine of the selected human part of UniProtKB/SwissProt database (http://www.uniprot.org). All reported data were based on high- confidence peptides assigned in Proteome Discoverer (Thermo Scientific) with a 5% FDR by Percolator. The high confidence protein–protein interactions (HCIs) were identified using stringent filtering against GFP control samples and the contaminant Repository for Affinity Purification (CRAPome, http://www.crapome.org/) database [41]. HCIs data were imported into Cytoscape 3.4.0 for the visualization. The known prey–prey interaction data were obtained from the IMEx database (http://www.imexconsortium.org/) [42]. Gene ontology classification analysis was based on the DAVID bioinformatics resource (https://david.ncifcrf.gov/) [43]. Hierarchical cluster was performed by centered correlation using Cluster 3.0 and the clusters were visualized with Tree View 1.1.6, and the matrix2png web server (http://www.chibi.ubc.ca/matrix2png/).

Statistical analysis

All of the experiments were done in triplicates or otherwise indicated. All data are reported as mean ± SD as indicated in the figure legends. Differences among groups were analyzed using the unpaired student t-test, one-way or two-way Anova.