Nucleoside diphosphate kinases 1 and 2 regulate a protective liver response to a high-fat diet

The synthesis of fatty acids from acetyl–coenzyme A (AcCoA) is deregulated in diverse pathologies, including cancer. Here, we report that fatty acid accumulation is negatively regulated by nucleoside diphosphate kinases 1 and 2 (NME1/2), housekeeping enzymes involved in nucleotide homeostasis that were recently found to bind CoA. We show that NME1 additionally binds AcCoA and that ligand recognition involves a unique binding mode dependent on the CoA/AcCoA 3′ phosphate. We report that Nme2 knockout mice fed a high-fat diet (HFD) exhibit excessive triglyceride synthesis and liver steatosis. In liver cells, NME2 mediates a gene transcriptional response to HFD leading to the repression of fatty acid accumulation and activation of a protective gene expression program via targeted histone acetylation. Our findings implicate NME1/2 in the epigenetic regulation of a protective liver response to HFD and suggest a potential role in controlling AcCoA usage between the competing paths of histone acetylation and fatty acid synthesis.


Mass spectrometry (MS)-based proteomic analyses of the CoA bound proteins
The eluted proteins solubilized in Laemmli buffer were stacked in the top of a 4-12% NuPAGE gel (Invitrogen).After staining with R-250 Coomassie Blue (Biorad), the proteins were digested in-gel using trypsin (modified, sequencing purity, Promega), as previously described (66).The resulting peptides were analyzed by online nanoliquid chromatography coupled to MS/MS (Ultimate 3000 and LTQ-Orbitrap Velos Pro, Thermo Fisher Scientific) using a 124 min gradient.For this purpose, the peptides were sampled on a precolumn (300 μm x 5 mm PepMap C18, Thermo Scientific) and separated in a 75 μm x 250 mm C18 column (PepMap C18, 3 μm, Thermo Fisher Scientific).The MS and MS/MS data were acquired by Xcalibur (Thermo Fisher Scientific).
The peptides and proteins were identified by Mascot (version 2.7.0,Matrix Science) through concomitant searches against the Uniprot database (Mus musculus taxonomy, June 2021 download), and a homemade classical database containing the sequences of classical contaminant proteins found in proteomic analyses (human keratins, trypsin, etc.).Trypsin/P was chosen as the enzyme and two missed cleavages were allowed.Precursor and fragment mass error tolerances were set respectively at 10 ppm and 0.6 Da.Peptide modifications allowed during the search were: Carbamidomethyl (C, fixed), Acetyl (Protein N-term, variable) and Oxidation (M, variable).The Proline software (67) was used for the compilation, grouping, and filtering of the results (conservation of rank 1 peptides, peptide length ≥ 6 amino acids, false discovery rate of peptide-spectrum-match identifications < 1%, and minimum of one specific peptide per identified protein group (68).Proline was then used to perform MS1-based label free quantification of the identified protein groups based on razor and specific peptides.The proteins from the contaminant database were discarded from the final list of identified proteins.To calculate fold changes between CoA and control eluates, missing values were imputed in ProStaR (69) using the DetQuantile algorithm (imputation with a value corresponding to the first percentile of the sample).To be considered as a potential CoA interactor, a protein must be quantified with a minimum of two peptides and be enriched at least 10 times in the CoA eluates compared to the control eluate.The relative abundance of the different proteins in each sample was evaluated through calculation of their intensity-based absolute quantification (iBAQ) (57) values.

Crystallization and crystal structure determination
Protein crystallization was performed by the hanging drop vapor diffusion method at 20°C (for ADP-and CoA-bound NME1) or 4°C (for SucCoA-bound NME1) by mixing 1 μl of the NME1/ligand complex with 1 μl of reservoir solution.ADP-bound NME1 was crystallized by mixing a solution of 6.1 mg/ml NME1, 1 mM ADP, 250 mM NaCl and 25 S2.Data for ADP-and SucCoA-bound NME1 were automatically integrated with XDS (70) which is included in the automatic software toolbox auto-PROC (71) and scaled with AIMLESS (72).Data for CoA-bound NME1 was manually integrated with XDS and scaled with AIMLESS.Structures were solved by molecular replacement in Phaser (73) using the structure of human NME1 bound to ADP (PDB 2HVD) (26) or imidazole fluorosulfate (PDB 5UI4) (74) after removing ligands and water molecules, as search models.Iterative rounds of refinement and model building were performed using PHENIX (75) and Coot (76), resulting in the final R-values shown in table S2.In the ADP-bound structure, ADP is observed bound to 5 of the 6 subunits in the NME1 hexamer, whereas the remaining binding site is empty because of crystal packing constraints.For CoA-and SucCoA-bound NME1 all ligand-binding sites are occupied.Weak or missing electron density precluded reliable modeling of the CoA/SucCoA pantetheine moiety and the SucCoA succinyl group, which are consequently omitted from the final refined atomic coordinates.

LC/ESI mass spectrometry
Liquid Chromatography Electrospray Ionization Mass Spectrometry (LC/ESI-MS) was performed on a 6210 LC-TOF spectrometer coupled to an HPLC system (Agilent Technologies).All solvents used were HPLC grade (Chromasolv, Sigma-Aldrich).Trifluoroacetic acid (TFA) was from Acros Organics (puriss., p.a.).Solvent A was 0.03% TFA in water, solvent B was 95% acetonitrile-5% water-0.03%TFA.Just before analysis NME1 samples (57 mM in sample buffer: 20 mM Tris-HCl pH 8.8, 5 mM MgCl2, 1 mM DTT) were diluted with sample buffer to a final concentration of 5.7 µM and 4 ml were injected for MS analysis.Protein samples were first desalted on a reverse phase-C8 cartridge (Zorbax 300SB-C8, 5 mm, 300 µm ID´5mm, Agilent Technologies) for 3 min at a flow rate of 50 ml/min with 100% solvent A and then eluted with 70% solvent B at a flow rate of 50 ml/min for MS detection.MS acquisition was carried out in positive ion mode in the 300-3200 m/z range.MS spectra were acquired and the data processed with MassHunter workstation software (v.B.02.00, Agilent Technologies) and with GPMAW software (v.7.00b2, Lighthouse Data, Denmark).

Native mass spectrometry
Mass Spectrometry under native conditions was performed to detect NME1 species following incubation with either CoA, dephospho-CoA, AcCoA, SucCoA, or a mixture of both CoA and ATP (all from Sigma-Aldrich).NME1 (50 µM) was incubated with the ligands (5 to 2500 µM) in a buffer containing 250 mM ammonium acetate, 0.5 mM magnesium acetate and DTT 1 mM for 15 min at room temperature and subsequently loaded by a nanoflow platinum-coated borosilicate electrospary capillary (Thermo Electron SAS; Courtaboeuf, France) into a quadrupole time-offlight mass spectrometer (nano-ESI-Q-TOF instrument, Q-TOF Ultima, Waters Corporation, Manchester, U.K.).The following instrumental parameters were used: capillary voltage = 1.2-1.3kV, cone potential = 40 V, RF lens-1 potential = 40 V, RF lens-2 potential = 1 V, aperture-1 potential = 0 V, collision energy = 30-140 V, and microchannel plate (MCP) = 1900 V.The pressure in the collision cell was set to ~2 × 10 −2 mbar.All mass spectra were calibrated externally using a solution of cesium iodide (6 mg/mL in 50% isopropanol) and were processed with the Masslynx 4.0 software (Waters Corporation, Manchester, U.K.) and with Massign software package.

RNAseq overall design
Transcriptomic analysis was performed on mouse liver samples obtained from 4 conditions, corresponding to two genotypes, Nme2 WT (Nme2 +/+ ) and KO (Nme2 -/-), and two experimental diets, normal diet (ND) and high fat diet (HFD).For each condition, RNAseq was performed in 4 replicates corresponding to 4 individual mice.

Extraction protocol
Total RNA was isolated and purified from the liver tissue by NucleoSpin RNA kit (Machery-Nagel-740955.50).Four independent RNA extractions were performed for each condition.

Library construction protocol
For each sample, one µg of RNA (RIN>8) was used for library preparation with the Illumina Stranded total RNA Prep Ligation with Ribo-Zero Plus kit (Illumina) according to the manufacturer's instructions.Each library was quantified on Qubit with the Qubit® dsDNA HS Assay Kit (Life Technologies) and the size distribution was examined on the Fragment Analyzer with High Sensitivity NGS Fragment Analysis kit (Agilent).

Library strategy
The libraries were then sequenced on the Illumina NextSeq 500 (paired-end 75) at the TGML Platform of Aix-Marseille University (France).

Alignment
The sequenced reads from the raw sequence (.fastq files) were aligned on the UCSC mm10 genome using the STAR software (2.7.1a) (77) to produce bam files.

Normalization and differential analysis
The normalization, pseudo-log transformation of the read counts, and the differential analyses were performed using the R software (80), DESeq2 (1.22.2) (81,82) and SARTools (83) packages.

Grouping genes/features according to their quartile of expression
The features corresponding to non-coding transcripts and/or with zero counts were filtered out, in order to retain all non-zero count protein-coding genes from the NCBI Reference Sequence Database (RefSeq).The corresponding genes and features were ranked according to their DESeq2 normalized expression mean value in the samples corresponding to the 4 replicates of liver from wild-type mice under normal diet, and allocated into four subsets corresponding to the quartiles of this expression and exported as .bedfiles.

Identification of up-and down-regulated genes and selection for heatmap representation
Supervised transcriptomic analyses were performed to identify genes significantly up-and downregulated between two conditions using thresholds of a Student t-test p-value <0.01 and a fold change absolute value of 2.
The normalized, pseudo-log transformed and standardized read counts of the up-and downregulated genes in wild type mice between those submitted to a high fat diet (HFD) and those with a normal diet (ND) were used to generate the heatmap presented in Fig. 6A.
The reaction was stopped by adding 5 mM EDTA.Small aliquots of mononucleosome solutions were collected for input.Digested mononucleosomes were diluted with LSDB buffer (50 mM HEPES pH 7.0, 3 mM MgCl2, 500 mM KCl, 20% glycerol, protease cocktail inhibitor containing 10 mM sodium butyrate) to achieve the final KCl concentration of 350 mM and incubated with antibody-coupled beads overnight at 4°C for 16 hours.On the third day, the beads were washed four times with LSDB 350 mM KCl and once in TE buffer (10 mM Tris-HCl containing 1 mM EDTA, pH 8.0).ChIP samples were eluted in 150 µl of the buffer with SDS 1% for 20 min at 65°C.The DNA was purified from the eluted ChIP samples as well as from the input samples by phenolchloroform extraction and ethanol precipitation.

ChIPseq overall design
ChIPseq analysis was performed in mouse liver samples obtained from 4 conditions, corresponding to two genotypes, Nme2 WT (Nme2 +/+ ) and KO (Nme2 -/-), and two experimental diets, normal diet (ND) and high fat diet (HFD).For each condition, chromatin immunoprecipitation was performed twice in the respective livers of two independent mice using anti-H3K9ac antibody of mNase digested chromatin and both the immune-precipitated (chip) and input materials were sequenced.

Extraction protocol
As previously reported, the DNA was purified from the eluted ChIP samples as well as from the input samples by phenol-chloroform extraction and ethanol precipitation.

Library construction protocol
For sequencing, ChIP libraries were prepared using MicroPlex Library Preparation Kit v3 (Diagenode) according to manufacturer's instructions.Each library was quantified on Qubit with Qubit® dsDNA HS Assay Kit (Life Technologies) and size distribution was examined on the Fragment Analyzer with High Sensitivity NGS Fragment Analysis kit (Agilent).

Library strategy
The ChIP libraries were sequenced on a High-output flow cell (400M clusters) using the NextSeq® 500/550 High Output v2.5 150 cycles kit (Illumina), in paired-end 75/75nt mode, according to manufacturer's instructions at the TGML Platform of Aix-Marseille University (France).Base calling was performed using RTA version 2.*

Data processing step: Alignment
The trimmed fastq files were aligned on the UCSC Mus_musculus mm10 genome using the Bowtie2 aligner (83), with options -end-to-end, -no-mixed, -no-discordant.

Genome build: UCSC Mus_musculus mm10 genome
Processed data files format and content: big wig files (.bw) containing normalized integrated aligned read count signals.

Selection of features corresponding to TSS and normalization
Raw ChIP counts corresponding to TSS +1500/-800bp of the RefSeq genes were computed using featureCounts (84), with option -a mm10_pc_tss8001500.saf-F SAF -s 0 -Q 30 -T 8 -o mm10_pc_tss8001500_count_30.txt.
For each sample, scale factors were computed on these features using DESeq2 (80,81) assuming that the global level of ChIP signal value (corresponding lysine acetylation in all genes promoters) should be equal in all samples.
The scale factors are given in the "Deseq2_sf " table shown below.

Highly expressed gene TSS profiles
The normalized ChIP signal were converted into a 10bp bin matrix of the signal 1.5Kb upstream and downstream 25% more expressed (see Mat & Meth RNA-seq) protein-coding genes TSS, using computeMatrix.
The ChIPseq profiles were generated using the computed computeMatrix outputs and a custom R script (79).

Expression vs. Acetylation
The normalized ChIP signals were converted into a 25 bp bin matrix of the signal 750 bp upstream and 1500 bp downstream protein-coding genes TSS, using computeMatrix.
For each gene and condition the mean signal over this region in all replicates was plotted.
In Fig. 7C, the x-axis represents the log2 (fold changes) of the differential expression (DESeq2 normalized values) between HFD and ND respectively for Nme2 WT (upper panel) and KO (lower panel).(B) In silico modeling of the T94D mutant.The Asp residue is modeled in its nine preferred sidechain conformations.The labels g + , g -and t refer to the gauche + , gauche -and trans conformations of the c1 angle.Left: Replacing the Thr residue in the CoA-bound NME1 crystal structure by an Asp residue and modeling its sidechain conformations revealed that all Asp rotamers point their negatively charged carboxylate group towards one or more CoA phosphate groups located within unfavourably close proximity.Right: Repeating the same exercise with the ADP-bound structure revealed several Asp rotamers (all with c1 in a g + conformation) that point their carboxylate away from the ADP diphosphate, located over 5 Å away.This suggests that replacing Thr94 by an Asp residue should yield an NME1 mutant that could accommodate ADP in the active site by adopting a permissive Asp94 rotamer, whereas no Asp94 conformation could avoid an electrostatic repulsion with a bound CoA ligand.(C) Upper panel.The NME1 T94D mutant retains notable NDPK activity.Both WT and mutant forms of GSTtagged NME1 are phosphorylated on histidine after incubation with ATP (100 µM) and become dephosphorylated upon incubation with GDP (200 µM).Proteins were detected using antiphosphohistidine (pHis) and anti-NME1 antibodies (NME1).Lower panel.The T94D mutant is defective for CoA-binding activity.WT and mutant NME1 were incubated with CoA beads, eluted with free CoA after a pull-down as described in Fig. 1B and the blot was probed with an anti-NME1 antibody.(D) The T94D mutant has lower NDPK activity than WT.Both WT and mutant forms of NME1 are phosphorylated on histidine following incubation with ATP.The degree of phosphorylation observed at 1-10 µM ATP is approximately an order of magnitude lower for the mutant compared to the WT.Proteins were detected using anti-phosphohistidine (pHis) and anti-NME1 antibodies (NME1).Native MS spectra shown are the same as the corresponding spectra in Fig. 4A.Primary peaks in the spectra were modelled as a Gaussian function convoluted with a binomial distribution (blue and orange curves) to describe the distribution of CoA-bound NME1 states, as described in Fig. S3.Insets.A magnified view of individual peaks reveals sub-peaks corresponding to the presence of phosphoryl groups on NME1.The distribution of phosphorylation states was fitted as a Gaussian function convoluted with a binomial distribution (green curves) in a manner analogous to that used for primary peak fitting.The reasonable agreement between the recorded spectra and the fitted curves confirms that phosphorylation occurs independently at the six catalytic sites of the NME1 hexamer, allowing one to estimate the fraction of NME1 monomers that are phosphorylated [Occ(PO3 2-)].(A) GSEA plots showing that gene sets corresponding to genes involved in the indicated signaling pathways are enriched in the hepatocytes of WT mice treated with HFD for 6 weeks compared to ND (left panels).These same gene sets are depleted in the liver of Nme2 -/-mice under HFD compared to Nme2 +/+ mice, demonstrating that, in the absence of NME2, the corresponding signaling pathways are not activated under HFD.(B) The levels of H3K9 acetylation and histone H3 in the liver extracts from mice treated as above were visualized by immunoblots using the corresponding antibodies.Two samples per condition were collected from independent mice.The normalized ratio of H3K9ac/H3 signals intensity for each sample is shown.(C).The cellular concentrations of CoA and AcCoA were respectively measured in liver extracts from Nme2 +/+ and Nme2 -/-mice.For each measurement, livers from independent mice were used as follows: Nme2 +/+ ND, N = 9; Nme2 +/+ HFD, N = 4; Nme2 -/-ND, N = 9; Nme2 -/-HFD, N = 5.The graphs show the average of the concentration values (in nmol/g of liver) and ± standard deviation.Student's t-test was used to calculate the p-values for HFD vs ND in all conditions.Statistical significance is indicated by the symbols *, ** and *** for p-values < 0.05, < 0.01 and < 0.001, respectively.  Certain monomers in some of the crystal structures have an empty ligand-binding site.In most of these cases the residues defining the binding site differ considerably in conformation between the ligand-bound and unbound monomers.

!
2 Number of structural alignments excluding those involving a monomer with an empty binding site.
3 Root-mean-square deviation of aligned structures. 4Values represent the mean !"S.D. calculated over the corresponding number of pairwise alignments.
Table S4.Intersections between genes presented as GSEA (fig.S12a) and genes that could be regulated by H3K9 acetylation/deacetylation around their TSS (Fig. 7C) mM Tris pH 7.4 with 0.2 M NaCl, 0.1 M Tris pH 8.0 and 30% (w/v) PEG 400.CoA-bound NME1 was crystallized by mixing a solution of 5.5 mg/ml NME1, 1 mM CoA, 150 mM NaCl and 0.1 M phosphate buffer pH 7.4 with 0.2 M Mg(NO3)2 and 36% PEG 3350.SucCoA-bound NME1 was crystallized by mixing a solution of 5.5 mg/ml NME1, 1 mM SucCoA, 150 mM NaCl and 0.1 M phosphate buffer pH 7.4 with 0.1 M citric acid pH 3.5 and 4-10% PEG 1000.Harvested crystals were flash-cooled in liquid nitrogen.Diffraction data were collected at beamline ID30A-1 of the European Synchrotron Radiation Facility (ESRF).All crystals were monoclinic with space group P21 and contained either one (ADP, CoA) or two (SucCoA) NME1 hexamers in the asymmetric unit.Data collection statistics are summarized in table -differential scanning fluorimetry (nano-DSF) 10 µL samples of 0.5 mM NME1 prepared in the absence or presence of either CoA, AcCoA, SucCoA or ADP at a concentration of 0.62 mM in a buffer containing 25 mM Tris pH 7.4, 250 mM NaCl were loaded into nanoDSF Grade Standard Capillaries (NanoTemper, #PR-C002) and analysed on a Prometheus NT 48 instrument (Nanotemper).Assay samples were heated from 20°C to 95°C at a rate of 0.5°C/min.Intrinsic tryptophan fluorescence was measured at 330 nm and 350 nm using an excitation power of 10%.The 350/330 fluorescence ratio was plotted against temperature and the Tm value determined from the inflection point using the instrument software.Isothermal titration calorimetry (ITC) Calorimetric experiments were performed in triplicate on a MicroCal iTC200 calorimeter (Malvern Panalytical) at 20°C while stirring at 330 rpm.The syringe and cell were respectively filled with 900 µM CoA and 30 µM NME1, both in 0.1 M phosphate buffer pH 7.4, 150 mM NaCl.Titrations consisted of 50 identical injections of 6 µl made at time intervals of 450 s.ITC data were corrected for the heating of CoA injection into buffer and analyzed with the MicroCal PEAQ-ITC Analysis Software (Malvern Panalytical).

Figure S1 .
Figure S1.NME1/2 are major CoA-binding factors in mouse spermatogenic cells.CoA-pull downs were performed in the presence of 500 mM KCl on extracts from total germ cells, pachytene spermatocytes, round spermatids, and elongating and condensing spermatids.

Figure S3 .
Figure S3.Native MS spectra reveal a binomial distribution of CoA/acyl-CoA-bound hexameric NME1 species.Spectra shown are the same as in Fig.2C.Peaks in the spectra were modelled as a simple Gaussian function whose amplitude was scaled by the values given by the binomial distribution, Bin(N,p), where N is the number of binding sites on the NME1 hexamer (N=6) and p is the fitted probability of a single site being occupied by a ligand.The fraction of NME1 hexamers bound to r ligands (r = 0 to 6) is given by C(6,r)p r (1-p) 1-r where C(6,r)= 6!/[r!(6-r)!].Recorded spectra are shown in black and fitted curves for the 22+ and 21+ charge states are in cyan and orange, respectively.The good agreement between the recorded spectra and the fitted curves confirms that CoA/acylCoA ligands bind independently to the six binding sites on each NME1 hexamer, allowing one to estimate the fraction (Occ, equal to p) of NME1 monomers bound to each ligand, which for these experiments ranged between 0.45 and 0.58.

Figure S5 .
Figure S5.Omit map density for (A) ADP and (B) SucCoA bound to NME1.The electron density (mesh in blue) shows Fo-Fc omit maps contoured at 3.0s (ADP) and 2.0s (SucCoA) where ligands were omitted from the map calculation.No strong density is observed for the pantetheine and succinyl moities of the SucCoA ligand.

Figure S6 .
Figure S6.The CoA b-phosphate occupies the position of the ATP g-phosphate prior to nucleophilic attack.Alignment of the CoA-bound NME1 structure with that of DdNDPK bound to ADP-AlF3, solved at 2.8 Åresolution (86).AlF3 mimics the transition state of the ATP g phosphate.The ADP ligand bound to DdNDPK is shown with a thin stick radius.The CoA 3' phosphate is within 1.5 Å of the aluminum ion in AlF3.Compared to the Al 3+ ion, the CoA 3' phosphate is farther from the histidine's nucleophilic Nd nitrogen and closer to the b-phosphate of ADP-AlF3, suggesting that the CoA 3' phosphate occupies the position of the ATP g-phosphate prior to nucleophilic attack.

Figure S7 .
Figure S7.The a-and b-phosphates of the CoA nucleotide are poorly recognized by NME1.(A) Structure of the NME1-bound CoA nucleotide with atoms colored according to their average crystallographic B factor, showing that the aand b-phosphates are the most mobile atoms in the ligand.In contrast, the 3' phosphate is highly constrained by numerous H bond interactions with the protein, explaining the low values of its B factors.The average B factors shown were calculated by averaging over the 12 independent copies of the ligand in the crystal structure of SucCoA-bound NME1.(B) Alignment of the 12 crystallographically independent subunits of NME1 in the crystal structure of SucCoA-bound NME1.Active site residues adopt a uniform sidechain conformation in all 12 subunits except for residue Arg58, which is located close to the CoA a-phosphate and exhibits considerable variability.Arg58 is within H bonding distance of the CoA a-phosphate in only 4 of the 12 monomers.H bonds are indicated as red dashed lines.

Figure S8 .
Figure S8 .Positional shift of Thr94 and analysis of the T94D mutant.(A) Structural alignment of the CoA-and ADP-bound NME1 illustrating the shift of position of residue Thr94.Only one ligand, CoA or ADP, is shown in the left and right views, respectively.Thr94 is the only active site residue whose backbone atoms shift considerably between the CoA-and ADP-bound structures, resulting in a ~2 Å displacement of the sidechain hydroxyl and methyl groups.The lower view shows that these groups sit directly below the diphosphate of CoA but are more peripherally located with respect to that of ADP.

Figure S9 .
Figure S9.Structural analyses predict that CoA binding and histidine phosphorylation are mutually inhibitory.(A) Alignment of SucCoA-bound NME1 and PAPS-bound DdNDPK (PDB 1BUX).PAPS is chemically identical to the CoA nucleotide except that its sulfate group replaces the CoA b-phosphate.In the crystal structure, the PAPS sulfate group was observed to occupy two positions, each with partial occupancy, labeled S and S'.The latter is shown with a thinner stick radius.Residue numbering is that of mammalian NME1.(B) Structural alignment of NME1:SucCoA with DdNDPK (the drosophila NDPK) phosphorylated on His119 (equivalent to murine His118) (PDB 1NSP) predicts a steric clash and electrostatic repulsion between the histidine phosphate and the CoA 3' phosphate, located only 1.4 Å apart (red oval).!

FigureFigure
Figure S11.ATP-sensitive CoA binding by NME1 and NME2 and expression levels in Nme2 ko mouse liver.(A) Competitive CoA and ATP binding by NME1 and NME2.Extracts from bacteria expressing GST-NME1 and GST-NME2 were used in CoA pull-down experiments in the presence of increasing amounts of ATP as indicated.CoA-bound NME1/2 were then visualised after immunoblotting using an anti-NME1/2 antibody.(B) Nme1 and Nme2 expression levels in wild-type and Nme2 ko liver under either ND or HFD.The level of Nme1 and Nme2 expression from the transcriptomic data presented in Fig. 6 were analyzed and shown as boxplots.The values correspond to RNA-seq read counts normalized by DESeq method and then logtransformed with log2(DESeq+1).

Figure
Figure S13.AMPK signaling is not involved in the response to a HFD challenge and the NME2-dependent control of fatty-acid synthesis.(A) Extracts from livers from three independent Nme2 +/+ or Nme2 -/-mice under ND or from four independent Nme2 +/+ or Nme2 -/-mice under HFD for 6 weeks were probed with the indicated antibodies.Band intensities are shown as boxplots.(B) Cellular concentrations of ATP, ADP and AMP measured in liver extracts from Nme2 +/+ and Nme2 -/- mice.For each measurement, livers from independent mice were used as follows: Nme2 +/+ ND, N = 9; Nme2 +/+ HFD, N = 4; Nme2 -/-ND, N = 9; Nme2 -/-HFD, N = 5.The graph shows the average values (in µmol/g of liver) and SD.

Table S1 . MS-based characterization of CoA-binding proteins.
See accompanying auxiliary file.