The fatty liver disease–causing protein PNPLA3-I148M alters lipid droplet–Golgi dynamics

Significance Fatty liver disease affects nearly a quarter of the world’s population and has both environmental and genetic risk factors. A mutation in the gene PNPLA3 that converts Ile 148 to Met is the strongest known genetic risk factor for developing fatty liver disease. Using a series of techniques to track endogenous PNPLA3 and PNPLA3-I148M biogenesis and localization, we reveal insights into how the mutation changes cellular dynamics. Although previous reports focus on its role on lipid droplets, we reveal that PNPLA3-I148M also functions at the Golgi apparatus, an organelle critical for protein transport into and out of the cell and lipid signaling. PNPLA3-I148M causes altered Golgi morphology and drives changes reminiscent of liver disease.

Nano-Glo HiBiT Blotting System (Promega).For detection of other proteins, membranes were blocked for 1 h at room temperature in 5% non-fat milk/PBS-T.Membranes were subsequently incubated with primary antibodies overnight at 4ºC (Table S3), followed by washes at room temperature (1 x 15 min, 2 x 5 min) and incubation with HRP-conjugated secondary antibodies (Bio-Rad) for 1 h.Following three washes at room temperature, membranes were treated with SuperSignal West Dura Extended Duration Substrate (ThermoFisher Scientific) and affixed to a development cassette.Blots were exposed to BioMax MR film (Carestream Health) in a dark room and developed in an SRX-101A film developer (Konica Minolta).
In vitro translation.Full-length DNA constructs for the coding sequences of human PNPLA3-HA, PNPLA3-I148M-HA, PNPLA3Δ42-62-HA, mouse Ii-op (2), and mouse RAMP4-op (2) were ordered as gBlocks (IDT) with a 5' T7 promoter-linker-Kozak site (5'-TAATACGACTCACTATAGGG AATATTCTTGTTCCCACCATG…). gBlocks were PCR-amplified with Q5 High-Fidelity DNA Polymerase (NEB) using the forward primer T7-Kozak-start-fwd and the reverse primers HA-PolyA-rev (for PNPLA3-HA constructs) or Op-PolyA-rev (for Ii-op and RAMP4-op; Table S4).Amplified constructs were used as inputs for reactions (final reaction concentration of 6.96 ng/µL) using the TnT Quick Coupled Transcription/Translation System (Promega).Reactions were performed as described for the TnT system.Briefly, the rabbit reticulocyte lysate master mix was supplemented with EasyTag L- 35 -Methionine (4% final reaction volume; Perkin Elmer).Canine rough pancreatic microsomes (cRMs; final A280 ~ 2.5-3.5),prepared as described (3), were supplemented co-translationally.Reactions were incubated on a benchtop Thermomixer (Eppendorf) at 30ºC for 55 min.To stop the reactions, puromycin dihydrochloride (Gibco) was added (final concentration of 2.5 mM), and samples were placed on ice.For post-translational reactions, cRMs were added following puromycin supplementation, and reactions were incubated at 30ºC for an additional 30 min.Samples (kept on ice) were added to an equal volume of 1x PSB (100 mM KOAc, 2 mM Mg(OAc)2, 50 mM HEPES, pH 7.4) with 250 mM sucrose.Samples were ultracentrifuged in thickwall ultracentrifuge tubes (Thermo Scientific) for 15 min at 200,000 x g, 4ºC (Sorvall MTX 150 ultracentrifuge with an S120-AT3 rotor).Supernatants were carefully removed and placed on ice.Pellets were resuspended in 1x PSB with 250 mM sucrose, and then mixed well with 100 mM (final concentration) sodium carbonate, pH 11.Tubes were left on ice for 30 min followed by another round of ultracentrifugation.Supernatants were carefully removed and kept on ice (wash fraction) and pellets were resuspended in PKB (1% SDS, 0.1 M Tris, pH 8.0).16% of each fraction (initial supernatant, wash, and pellet) was mixed with 2x SDS sample buffer supplemented with 5% β-mercaptoethanol.Samples were boiled for 5 min prior to loading onto NuPAGE 4-12% Bis-Tris gels (Invitrogen).After washing with water and then with 5% glycerol, gels were placed upside down on thick filter paper (Bio-Rad) and covered with saran wrap to dry in a SGD2000 slab gel dryer (ThermoFisher Scientific).Gels were exposed to film and developed as above.
Preparation of ER-derived microsomes from cells.HEK239T cells (ATCC) were split into 10-cm dishes and grown overnight in DMEM (Gibco) supplemented with 10% FBS.The following day, cells were transfected with pSNAPf-EF1α-PNPLA3-HA-HiBiT (Azenta Life Sciences) using Lipofectamine 3000 and grown for an additional 24 h.Cells were washed once in cold PBS, detached in cold PBS and centrifuged to sediment.Cell pellets were stored at -80ºC until ready for use.Prior to microsome preparation, pellets were thawed on ice, resuspended in 10 mM HEPES-KOH, pH 7.5, and incubated for 10 min on ice.The cells were then pelleted, resuspended in homogenization buffer (10 mM HEPES-KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 5 mM EGTA, 250 mM sucrose) and passed through a 27 G syringe needle 20 times.The homogenate was subjected to serial centrifugations (600 x g for 10 min, 3000 x g for 10 min, 100,000 x g for 60 min) prior to resuspension in membrane buffer (10 mM HEPES-KOH, pH 7.5, 50 mM KOAc, 2 mM Mg(OAc)2, 1 mM DTT, 250 mM sucrose), as described (5).Proteinase K digestion was done the same way as for cRMs, described above.
Protein production.Proteins were expressed in suspension HEK 293-EBNA1 cells, as described (1, 6).First, complexes of plasmid and PEI MAX (Polysciences) were generated by mixing the appropriate pTT5 expression vector with transfection reagent (ratio of 1 µg vector : 4 µg PEI MAX) in F17 Expression Medium (Gibco).Complexes were incubated for 15 min at room temperature.Plasmid-PEI solutions were added to 293-EBNA1 cells (2 x 10 6 cells/mL) in Expression Medium [F17 Medium supplemented with 0.1% Kolliphor P 188 (Sigma), 6 mM L-glutamine (Gibco), 25 µg/mL G418 Sulfate (Gibco)].Cells were shaken overnight at 36ºC, 5% CO2, 85% humidity.The next day, an equal volume of Expression Medium was added to the culture, and cells were grown in the same incubation conditions for 72 h.Cells were harvested by centrifugation at 3,700 x g for 30 min.
For PNPLA3-His constructs, proteins were purified as described (7), except that whole cell lysate (instead of just the membrane fraction) was passed over Ni-NTA.For PNPLA3-FLAG constructs, proteins were purified from the membrane fraction, as described (8).
Lipid binding assays.PIP Strips (Echelon) were blocked overnight in PBS-T with 3% fatty acid-free bovine serum albumin (GoldBio) at 4ºC.The next day, blocking buffer was removed and replaced with blocking buffer supplemented with 2 -6 µg/mL of purified proteins (PLC-δ1 PH was obtained from Echelon).Strips were incubated on a shaker at room temperature for 1 h, followed by 3 x 10 min washes in PBS-T.Strips were then incubated in blocking buffer containing secondary antibodies (Table S3) on a shaker at room temperature for 1 h.Following 3 x 10 min washes, strips were developed with SuperSignal West Dura Extended Duration Substrate and imaged on a ChemiDoc MP Imaging System (Bio-Rad).
Generating LgBiT-expressing cells.Hep3B-HiBiT cells (WT and I148M) were transfected with pCMV-LgBiT (Promega CS1956B03) using FuGENE HD (Promega).After 48 h, cells were treated with Hygromycin to select for incorporation of the plasmid.Stable cell lines were initially tested for bioluminescence using the Nano-Glo Live Cell Assay System (Promega).Bioluminescence was read on an EnVision plate reader (Perkin Elmer).
Sample preparation for electron microscopy.Cells were grown on 3 mm synthetic sapphire disks (Technotrade International) and treated with 100 µM oleic acid overnight.Cells were then pre-fixed with 3% glutaraldehyde, 1% paraformaldehyde, 5% sucrose in 0.1 M sodium cacodylate.The disks were rinsed with fresh cacodylate buffer containing 10% Ficoll, placed into brass planchettes (Ted Pella, Inc.), and rapidly frozen with an HPM-010 high-pressure freezing machine (Bal-Tec).The frozen samples were transferred under liquid nitrogen to cryotubes (Nunc) containing a frozen solution of 2.5% osmium tetroxide, 0.05% uranyl acetate in acetone.Tubes were loaded into an AFS-2 freeze-substitution machine (Leica Microsystems) and processed at -90°C for 72 h, warmed over 12 h to -20°C, held at that temperature for 6 h, then warmed to 4°C for 2 h.The fixative was removed, and the samples rinsed 4x with cold acetone, after which they were infiltrated with Epon-Araldite resin (Electron Microscopy Sciences) over 48 h.The sapphire disks with affixed cells were flat-embedded onto a Teflon-coated glass microscope slide and covered with a Thermanox coverslip (Electron Microscopy Sciences).Resin was polymerized at 60°C for 48 h and the sapphire disks were excised, leaving the cells as a monolayer within a resin wafer.
Electron microscopy and Dual-Axis tomography.Cells were observed by light microscopy and appropriate regions were extracted with a microsurgical scalpel and glued to the tips of plastic sectioning stubs.Semi-thin (170 nm) serial sections were cut with a UC6 ultramicrotome (Leica Microsystems) using a diamond knife (Diatome).Sections were placed on formvar-coated copper-rhodium slot grids (Electron Microscopy Sciences) and stained with 3% uranyl acetate and lead citrate.Gold beads (10 nm) were placed on both surfaces of the grid to serve as fiducial markers for subsequent image alignment.Sections were placed in a dual-axis tomography holder (Model 2040, E.A. Fischione Instruments) and imaged with a Tecnai T12-G2 transmission electron microscope operating at 120 KeV (ThermoFisher Scientific) equipped with a 2k x 2k CCD camera (XP1000; Gatan, Inc.).Tomographic tilt-series and large-area montaged overviews were acquired automatically using the SerialEM software package (9,10).For tomography, samples were tilted ± 62° and images collected at 1° intervals.The grid was then rotated 90° and a similar series taken about the orthogonal axis.Tomographic data was calculated, analyzed, and modeled using the IMOD software package (10,11) on iMac Pro and Mac Studio M1 computers (Apple, Inc.).
Mass spectrometry.Hep3B cells were grown in 6-well plates (500,000 cells/mL) and washed three times with PBS prior to harvesting.Cell pellets (1 x 10 6 cells/pellet) were flash frozen in liquid nitrogen and stored at -80ºC until ready for analysis.Pellets were thawed on ice and resuspended in 0.05% SDS/0.5 M TEAB.Following multiple rounds of pipetting and brief vortexing, samples were passed through a 23 G syringe needle 30 times on ice and sonicated.Lysates were clarified by centrifugation at 16,000 x g for 10 min at 4ºC.Supernatants were transferred to new tubes and total protein was quantified using a Bradford Assay (Bio-Rad).20 µg protein per sample was reduced with 3 mM TCEP (Sigma) and alkylated with 10 mM iodoacetamide (Sigma) in the dark.Samples were digested overnight at 37ºC with LysC and trypsin.5 µg peptides from each sample were labelled using the TMTpro reagents (ThermoFisher Scientific) for 2 hours, followed by quenching with 5% hydroxylamine for 15 minutes, pooling, and drying down.Labelled peptides were mixed and analyzed using two-dimensional liquid chromatography and tandem mass spectrometry, as previously described (12).The pooled samples were desalted followed by phosphopeptide enrichment through sequential use of the High-Select TiO2 and Fe-NTA phosphopeptide enrichment kits (Thermo Scientific).All native peptides not captured by the phosphopeptide enrichment kits were then separated by offline medium pH C4 peptide fractionation (Accucore 150-C4, 2.6um pore size, 150mm X 2.1mm, Thermo Scientific) using gradient mobile phase conditions as previously reported (12).Fractionated peptides were concatenated into 6 pooled fractions, then dried down and stored at −80 °C until mass spectrometry analysis.
MS1 spectra were acquired in the Orbitrap at 120K resolution with a scan range from 375-2000 m/z, an AGC target of 4e5, and a maximum injection rate of 50 ms in Profile mode.Features were filtered for monoisotopic peaks with a charge state of 2-7 and a minimum intensity of 2.5e4, with dynamic exclusion set to exclude features after 1 time for 60 seconds with a 5-ppm mass tolerance.HCD fragmentation was performed with collision energy of 32% after quadrupole isolation of features using an isolation window of 0.7 m/z, an AGC target of 5e4, and a maximum injection time of 86 ms.MS2 scans were then acquired in the Orbitrap at 50K resolution in Centroid mode with the first mass fixed at 110.Cycle time was set at 1 s.
Proteomics data analysis was performed in Proteome Discoverer 2.4 (Thermo Scientific) using the Byonic search algorithm (Protein Metrics) and Uniprot human database.Byonic search parameters for native peptide fractions were as follows: fully Tryptic peptides with no more than 2 missed cleavages, precursor mass tolerance of 10 ppm and fragment mass tolerance of 20 ppm, and a maximum of 3 common modifications and 2 rare modifications.Cysteine carbamidomethylation and TMTpro addition to lysine and peptide N-termini were static modifications.Methionine oxidation and lysine acetylation were common dynamic modifications (up to 2 each).Methionine loss on protein N-termini, methionine loss + acetylation on protein N-termini, protein N-terminal acetylation, and phosphorylation of serine, threonine, and tyrosine were rare dynamic modifications (only 1 each).Percolator FDRs were set at 0.001 (strict) and 0.01 (relaxed).Spectrum file retention time calibration was used with TMTpro addition to peptide N-termini and lysines as static modifications.Reporter ion quantification used a co-isolation threshold of 50 and average reporter S/N threshold of 10.Normalization was performed on total peptide amount and scaling was performed on all average.Peptide and protein FDRs were set at 0.01 (strict) and 0.05 (relaxed), with peptide confidence at least high, lower confidence peptides excluded, and minimum peptide length set at 6. Statistical analysis of protein abundances was performed as previously described (13).Proteinlevel output files from Proteome Discoverer 2.4 were used for analyses.A limma test, implemented in R, was used to determine significance between WT and I148M cell lines (n = 8 per group).p-values were corrected for multiple testing by a Benjamini-Hochberg procedure.
For analysis of differential enrichment of biological process keywords in I148M versus WT cell lines, proteins significantly (adj.p-value < 0.01) decreased or increased by >1.5 fold in the I148M vs. WT PNPLA3 cell lines were compared with those not significantly changing.Biological process keywords for these sets of proteins were obtained from UniProt and Fisher's exact test was used to determine a p-value, which was Bonferroni corrected, for each keyword.Only biological process keywords annotated for at least 5 proteins in at least one of the sets of proteins were considered.
A list of liver-specific genes was obtained from the Tissue-specific Gene Expression and Regulation (TiGER) database.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE ( 17) partner repository with the dataset identifier PXD046335.RNA-seq.Hep3B cells were grown overnight in 6-well plates (seeded at 500,000 cells/mL).Enough plates were seeded to allow for 5 replicates of the following conditions: WT & I148M basal, WT & I148M 30 min, 1, 2, 4, and 24 h oleic acid (200 µM) treatment.Following the specified treatment, cells (1 x 10 6 cells/well) were washed three times with PBS and lysed in Buffer RLT (Qiagen).Lysates were transferred to Eppendorf tubes on ice and stored at -80ºC until ready for RNA purification.Library preparation from purified RNA, sequencing and analysis protocols have been described previously (18).To determine the changes of gene expression, differential expression analysis between WT and I148M samples at each time point was conducted using DESeq2 (1.26.1) (19).Gene expression fold-change and adjusted p-values were generated using default parameters.The RNA-seq data have been deposited in NCBI's Gene Expression Omnibus (20) and are accessible through GEO Series accession number GSE261297.vehicle-treated cells expressing WT or PNPLA3-I148M were filtered to include only those with one defined phosphorylation site.Scatterplot depicts log2 PNPLA3-I148M vs. WT PNPLA3 fold changes from this experiment compared with log2 PNPLA3-I148M vs. WT PNPLA3 fold changes from the whole-cell proteomics experiment (see Fig. 3A).The Pearson correlation coefficient between these two independent experiments (635 peptides and 421 proteins) is shown.The top 3 peptides differentially (green = increased; red = decreased) phosphorylated are labeled.(F) Sources of liver disease-related gene sets.(G) Plot showing log10 fold change for proteins quantified in the I148M vs. WT PNPLA3 untreated cell lines.Those reported to be tissue-enriched genes in hepatocellular carcinoma (HCC) are shown in black (16).Kolmogorov-Smirnov p-value testing for a difference in fold change distribution between the two protein sets is reported.

Fig. S1 .
Fig. S1.PNPLA3 lacks an ER-targeting signal sequence.(A) Schematic (top) of the PNPLA3 primary sequence, with predicted hydrophobic segments of roughly ~ 20 amino acids in the PNPLA domain shown in yellow.Cartoon (bottom) depicts contrasting models of where PNPLA3 originates prior to localization on LDs, including a predicted ER topology.Amino acids 42-62 comprise a predicted type II signal anchor sequence.Residues S47 and D166 comprise the putative catalytic dyad.(B) Schematic of in vitro translation experimental design.(C) PNPLA3-HA and variants were translated in rabbit reticulocyte lysate in the presence of 35 S-methionine.Canine rough microsomal membranes (cRMs) were added co-or posttranslationally.Membranes were isolated, washed with sodium carbonate, then re-isolated.As controls for membrane insertion, the signal recognition particle-dependent type II signal anchor sequence-containing protein invariant chain (Ii) and the tail-anchored protein RAMP4, both containing C-terminal opsin (op) tags,

Figure S2 .
Figure S2.Membrane-bound PNPLA3 is cytosolically exposed.(A) cRMs added co-or posttranslationally were incubated with Proteinase K with or without Triton X-100.(B) HEK293T cells were transiently transfected with a PNPLA3-HA-HiBiT vector.After 48 h, microsomes were isolated from the cells and subsequently treated with Proteinase K with or without Triton X-100.(C) cRM's added posttranslationally were solublized and treated with Endoglycosidase H.Control reactions were done with Ii-op and RAMP4-op.(D) cRM's were mock-treated (M-cRMs) or trypsin-treated (T-cRMs) to remove exposed portions of membrane proteins.These cRMs were added post-translationally to translation reactions.Following a carbonate wash and re-isolation of cRMs, localization of PNPLA3-HA constructs or RAMP4-op were assessed by SDS-PAGE (left).cRMs were characterized by immunonlotting to ensure trypsin digestion of exposed proteins (right).

Fig. S3 .
Fig. S3.PNPLA3-I148M fractionates with endosomes and interacts with membrane phosphoinositides.(A) Endogenous PNPLA3-HiBiT and PNPLA3-I148M-HiBiT are enriched in the plasma membrane (PM)/larger organelle fraction and the endosomal fraction.(B) Immunoblotting and Coomassie staining of purified PNPLA3 constructs.*SDS-resistant PNPLA3 dimers.(C) Purified PNPLA3-His, PNPLA3-I148M-His, and GST-PLC-δ1 PH domain were incubated with membranes spotted with 100 pmol of different lipids (left).After washing the membranes thoroughly, they were incubated with anti-His or anti-GST secondary antibodies (HRP-conjugated) and imaged using chemiluminescence.Experiment was repeated at least 3 times for each protein, with representative images shown (right).Experiment was repeated with PNPLA3-FLAG constructs, with similar results.

Fig. S5 .
Fig. S5.PNPLA3-I148M has a distinct effect on proteomic and transcriptomic cellular changes relative to oleic acid treatment.(A) Volcano plot of proteins quantified by mass spectrometry in cells expressing WT or PNPLA3-I148M, following 16 h treatment with 200 µM oleic acid.Red and green compartments contain proteins significantly (adj.p-value < 0.01) decreased or increased, respectively, by >1.5 fold in the I148M vs. WT cell lines.(B) Comparison of the log2 fold change (I148M vs. WT PNPLA3) between two mass spectrometry experiments, in which cells were treated with or without oleic acid for 16 h.(C) Confirmation of top proteomic hits (downregulated in I148M; see Fig. 3A) in parental Hep3B cells.(D) Scatterplot of phosphopeptides quantified by mass spectrometry in vehicle-(2313 peptides) or oleic

Fig. S6 .
Fig. S6.LD-Golgi contacts in Hep3B cells.Representative TEM images from WT (A) and I148M (B) Hep3B cells pre-treated with 100 µM oleic acid for 16 h.(C) Hep3B cells expressing endogenous PNPLA3 or PNPLA3-I148M were treated with 100 µM oleic acid for 16 h, fixed and stained with LipidTOX Green to label LDs and anti-TGOLN2 to label the trans-Golgi network.Lipid droplet-TGOLN2 contact sites in ~ 45 cells per cell line were quantified using Imaris software.p-values were calculated from Student's t-tests.ns p > 0.05.

Fig. S7 .
Fig. S7.Identification of primary human hepatocyte lots that express PNPLA3-I148M.(A) Individual lots of primary human hepatocytes were assessed for PNPLA3 expression by digital PCR using a TaqMan pan-PNPLA3 primer/probe set (left) and a custom TaqMan allele-specific primer/probe set (ANTZ9CG; right) that can distinguish the reference and mutant PNPLA3 alleles.(B)Table showing key identifying information about primary human hepatocyte lots.