Rab2A-mediated Golgi-lipid droplet interactions support very-low-density lipoprotein secretion in hepatocytes

Deletion of hepatic Ras-related protein Rab-2A (Rab2A) diminishes serum triglyceride and cholesterol levels

To delineate the complicated relationship between hepatic Rab2A and serum lipid levels, we generated a mouse model with hepatocytes-specific Rab2A deletion, termed LCK, with their littermates served as controls and designated as Flox (Appendix Fig. S1). Our findings discovered that hepatic Rab2A deletion markedly reduced serum triglyceride (TG) levels (by ~28.9% under “Random feed” condition and 37.9% under “Fasted” condition) (Fig. 1A) and total cholesterol (TC) levels (by ~41.7% under “Random feed” condition and 48.5% under “Fasted” condition) (Fig. 1B), particularly TG within the very low-density lipoprotein (VLDL) particles (Fig. 1C) and TC within the high-density lipoprotein (HDL) particles (Fig. 1D). Conversely, TG and TC levels in liver (Appendix Fig. S2A,B) exhibited mild corresponding accumulation in the LCK mice, especially under the “Random feed” condition.

Lipoproteins primarily encompass TG, TC, and apolipoproteins, including Apo B-100, Apo B-48, Apo-E, Apo-AI, and Apo-CIII. We evaluated their secretion levels in serum and found them comparable in both genotypes (Fig. 1E,F). Subsequently, we examined the expression levels of apolipoproteins and several lipid transporters in the liver. Apo B levels, comprising Apo B-100 and Apo B-48, were notably increased in the LCK mice (Appendix Fig. S2C,D). Apart from Apo B, protein levels including Apo-E, Apo-AI, Microsomal triglyceride transfer protein (MTP), Low-density lipoprotein receptor (LDLR), and CD36 were all similar in the livers of Flox and LCK mice (Appendix Fig. S2C,D).

Building upon the observed relationship between Rab2A and serum lipid under both “Random feed” and “Fasted” conditions, we further investigated whether Rab2A deficiency in hepatocytes could also confer resistance against hyperlipidemia induced by high-fat-high-cholesterol diet (HFHCD). Remarkably, LCK mice exhibited pronounced rescue in serum TG levels (by ~29.8%) (Fig. 1G) and TC levels (by ~53.4%) (Fig. 1H), particularly evident in TG within the VLDL particles (Fig. 1I) and TC within the HDL particles (Fig. 1J), compared to the similar patterns of hepatic TG and TC (Appendix Fig. S2E,F). In contrast, subsequent detection of apolipoproteins uncovered marked accumulation of Apo B-100 and Apo B-48 proteins in the serum (Fig. 1K,L) and liver (Appendix Fig. S2G,H) of LCK mice.

In summary, our comprehensive analysis indicates that hepatic Rab2A deletion significantly diminishes serum TG and TC levels, but not apolipoproteins levels, suggesting that Rab2A deficiency may attenuate the degree of lipoproteins lipidation.

Absence of Rab2A hinders the lipidation of VLDL₂

In light of the profound impacts of hepatic Rab2A on serum lipid homeostasis, we delved into the mechanistic underpinnings. Our primary objective was to discern whether Rab2A deficiency primarily influenced lipid uptake or secretion. Initially, we evaluated fatty acid absorption rates in mice and primary hepatocytes, observing comparable TG absorption rates between two genotypes (Appendix Fig. S3A–C). We further analyzed the lipid-lowering effects of Rab2A deficiency in LDLR knockout mice, a model characterized by hyperlipidemia due to impaired hepatic lipid uptake (Appendix Fig. S3D). The results demonstrated that Rab2A inhibition still dramatically reduced serum TG levels (by ~26.7%) and TC levels (by ~32.5%) (Appendix Fig. S3E,F), with a notable reduction in the VLDL particles contents (Appendix Fig. S3G,H). These observations suggest that lipid uptake in the liver does not primarily account for the lipid-lowering effects of Rab2A inhibition.

To ascertain the impact of Rab2A on lipid secretion dynamics, we initially assessed lipid secretion rates in Flox and LCK mice using tyloxapol, a lipoprotein lipase inhibitor. Our findings revealed that Rab2A deficiency potentially reduced VLDL lipidation rates in both male and female mice, as evidenced by decreased TG level (Figs. 2A and EV1A,B), with no significant changes in the apolipoproteins (Fig. 2B,C). This suppression of VLDL lipidation level was also validated in the primary hepatocytes (Fig. EV1C,D). To advance our understanding of Rab2A’s integral role in VLDL secretion, we employed adeno-associated virus-mediated Apob knockdown experiments in both genotypes (Fig. EV1E,F). Apob knockdown in the Flox mice led to a substantial decrease in TG (by 59.6%) and TC levels (by 78.4%), whereas in the LCK mice, the reductions were more moderate (only 26.7% for TG and 50.0% for TC) (Fig. EV1G,H). Notably, Apo B suppression aligned the serum lipid profiles of both genotypes, irrespective of Rab2A expression (Fig. EV1G,H). These compelling findings suggest that Rab2A functions downstream of Apo B, playing a crucial role in the secretion of VLDL.

Purification and fractionation of endoplasmic reticulum (ER) and Golgi compartments enabled the assessment of VLDL lipidation levels through Apo B-48 immunoblotting (Li et al, 2012). Subsequently, Golgi apparatus and ER fractions from liver tissues were isolated and purified through sucrose gradient density centrifugation (Appendix Fig. S4). Our investigations revealed that while Apo B lipidation remained consistent in the ER, Rab2A deficiency notably impaired VLDL₂ lipidation and increased its density within the Golgi apparatus (Fig. 2D–F). Subsequent analyses of serum lipoproteins’ density disclosed a shift towards higher-density lipoproteins in the LCK mice (Fig. 2G,H). Moreover, evaluation of VLDL size demonstrated that Rab2A deletion predisposes LCK mice to secrete smaller-diameter VLDL compared to their wild-type counterparts (Fig. 2I,J).

Collectively, these findings elucidate that Rab2A deficiency in hepatocytes selectively disrupts VLDL₂ lipidation within the Golgi apparatus, culminating in the secretion of VLDL particles with diminished lipid contents.

Rab2A orchestrates Golgi-lipid droplet (LD) interactions

Elucidating the precise subcellular positioning of Rab2A within hepatocytes is pivotal for unraveling its role in VLDL₂ lipidation. Previous studies have demonstrated that Rab2A is primarily situated at Golgi apparatus, as evidenced by staining with endogenously genomic-labeled Rab2A (Götz et al, 2021; Lund et al, 2018) and exogenously overexpressed Rab2A (Ding et al, 2019; Gillingham et al, 2014). Our findings corroborated these observations, showing that GTP-bound Rab2A predominantly localized to the Golgi apparatus rather than the ER or ER-Golgi intermediate compartment (ERGIC) compartment, as confirmed by co-localization studies of exogenously expressed Rab2A in Huh7 cells (Appendix Fig. S5). Further analysis with endogenous Rab2A also confirmed its predominant localization to the Golgi apparatus in Huh7 cells, potentially suggesting that Rab2A mainly existed in the GTP-bound form (Fig. 3A). In contrast to the concentrated Golgi morphology observed in Huh7 cells, primary hepatocytes exhibit a more dispersed Golgi distribution (Sherman et al, 2024) (Figs. 3B and  EV2B; Appendix Fig. S6A). Moreover, the localization of Rab2A in primary hepatocytes appears more complex. While Rab2A still remained primarily in the Golgi apparatus, signals outside the Golgi apparatus were also observed, which may reflect the GDP-bound form of Rab2A at lysosomes (Ding et al, 2019; Yin et al, 2017) (Fig. 3B and EV5E). Additionally, studies in fly larval ventral nerve cords (Götz et al, 2021) and Hela-S3 cells (Aizawa and Fukuda, 2015) have highlighted Rab2A’s role in maintaining Golgi apparatus integrity. Our investigations extended these findings to primary hepatocytes, indicating that Rab2A deficiency had a negligible impact on the structural integrity of the Golgi apparatus (Appendix Fig. S6), suggesting varied roles of Rab2A in different cell types.

To delineate the mechanistic contribution of Golgi-anchored Rab2A to VLDL₂ lipidation, we isolated and purified Golgi apparatus and ER compartments from the livers of Flox and LCK mice. Subsequent quantitative evaluations were conducted to measure TG and TC levels, as well as the abundance of Apo B protein within these compartments. Our studies revealed that Rab2A deficiency led to a significant reduction in TG and TC levels in the Golgi compartment but not within the ER fractions (Fig. 3C–F). Conversely, the Apo B protein exhibited an obvious enrichment in the Golgi compartment, without notable alterations in the ER (Fig. 3G). These findings led us to hypothesize that Rab2A may act as an important molecule, facilitating lipid transfer from LDs to the Golgi apparatus by mediating their interactions.

Initial observations in Huh7 cells and primary hepatocytes confirmed a pronounced interplay between Golgi membranes-resident Rab2A and LDs, indicating frequent communications (Fig. EV2). Subsequently, we analyzed the potential contact points between the Golgi apparatus and LDs, and the underlying role of Rab2A. Under oleic acid (OA) incubation, an average of approximately 22 points between the Golgi apparatus and LDs was observed per primary hepatocyte (Fig. 3H,I), a frequency higher than that reported in a previous study within the Hep3B cell line (~5 points) (Sherman et al, 2024). Rab2A deficiency significantly reduced these potential contact points to about 17 (Fig. 3H,I). The finding was further corroborated by a LDs pull-down assay (Appendix Fig. S7), which elucidated that Rab2A suppression significantly decreased the interface between LDs and Golgi, as evidenced by the reduced levels of Golgi markers in Rab2A-depleted samples (Fig. 3J). This effect was specific to the Golgi-LD interface, with Rab2A inhibition showing a comparable impact on the interactions between ER and LDs (Fig. 3J).

Taken together, our findings suggest that Rab2A orchestrates the interaction between the Golgi apparatus and LDs, critically modulating the lipid transfer from LDs to the Golgi apparatus, thus pivotal in the lipidation of VLDL₂.

17-beta-hydroxysteroid dehydrogenase 13 (HSD17B13), an LD-localized protein, binds with Rab2A to mediate Golgi-LD interactions and VLDL secretion

In pursuit of unraveling the underlying molecular mechanisms by which Rab2A influences Golgi-LD interactions, our initial approach entailed co-immunoprecipitation assays aimed at identifying potential Rab2A binding partners. HSD17B13, a protein localized to LDs (Fig. 4A), emerged as a key candidate (Abul-Husn et al, 2018; Ma et al, 2019; Ma et al, 2020; Su et al, 2014; Wang et al, 2022). The binding between Rab2A and HSD17B13 was robustly confirmed across liver tissue samples, cell lines, and through in vitro assays (Figs. 4B and EV3A–C). Moreover, this binding was dependent on the Rab2A’s activity and Golgi localization (Fig. EV3F; Appendix Figs. S8A–C and S9), as well as the LD localization of HSD17B13 (Appendix Fig. S8D,E). Subsequent analysis utilizing immunofluorescence further revealed partial colocalization between endogenously expressed Rab2A and HSD17B13 in primary hepatocytes (Fig. 4C), with ~16.2% of the HSD17B13 signal contacting with Rab2A (Fig. EV3D,E).

Exploring the functional implications of HSD17B13 in Golgi-LD communications and VLDL secretion, we observed that inhibition of HSD17B13 in primary hepatocytes resulted in a decreased number of Golgi-LD interactions, subsequently decreasing TG secretion (Fig. 4D–G), a finding further validated by the liver-specific knockdown of HSD17B13 using adeno-associated viruses (Appendix Fig. S10A; Fig. 4H). Remarkably, HSD17B13 deficiency significantly reduced Golgi-LD interfacing, while leaving ER-LD interactions partially increased, as demonstrated by the LD-pulldown assay (Fig. 4I). Following HSD17B13 inhibition, serum TG and TC levels were attenuated (Fig. 4J,K), with a paradoxical increase in liver TG levels, likely due to compensatory mechanisms (Appendix Fig. S10B,C). Additionally, analysis of serum apolipoproteins revealed that HSD17B13 knockdown diminished the secretion of Apo B-48, in conjunction with disrupted lipid secretion (Appendix Fig. S10D,E), suggesting alternative mechanisms through which HSD17B13 influenced VLDL secretion beyond its binding with Rab2A.

Further exploration through molecular mapping pinpointed critical amino acids within Rab2A, specifically residues 32–42, as crucial for its binding with HSD17B13 (Appendix Fig. S11). Mutation of these residues abrogated the complex formation (Fig. EV4A) and significantly impaired TG secretion (Fig. EV4D,E), without altering Rab2A localization and activity (Fig. EV4B,C). These results underscore the indispensable role of Rab2A-HSD17B13 interaction in modulating TG secretion.

Collectively, our comprehensive dataset elucidates a novel regulatory axis wherein the proteins binding between Golgi-localized Rab2A and LD-associated HSD17B13 mediates Golgi-LD interactions, thereby facilitating VLDL secretion.

AMP-activated protein kinase (AMPK) signaling attenuates Rab2A activity and its role in Golgi-LD interactions and VLDL secretion

To deepen our understanding of Rab2A’s role in the Golgi-LD interactions and subsequent VLDL secretion, we focused on the regulatory effects of AMPK on Rab2A activity (Chen et al, 2022). Fasting, a physiological condition known to markedly increase AMPK activity (Zong et al, 2019), has been extensively studied for its inhibitory effects on VLDL secretion (Cheng et al, 2016; Rai et al, 2017). Our data confirmed these findings, demonstrating that fasting significantly inhibited VLDL secretion (Appendix Fig. S12). More importantly, fasting also reduced Rab2A activity and its Golgi localization upon AMPK activation (Fig. 5A,B; Appendix Fig. S13), suggesting Rab2A’s involvement in VLDL secretion under normal feeding conditions.

To further investigate Rab2A activity modulation, we employed A769662, a well-known agonist of AMPK, which effectively reduced Rab2A activity (Figs. 5C and EV5A,D,F), leading to a notable relocation from the Golgi apparatus to the cytosol, potentially to the lysosomes (Fig. EV5B,C,E). Subsequently, A769662 stimulation significantly disrupted Rab2A-HSD17B13 complex (Fig. EV5G), and then attenuating Golgi-LD interactions (Fig. 5D–F). Additionally, we validated that AMPK activation via A769662 dramatically reduced TG secretion by about 40% in wild-type, but not in Rab2A-deficient, primary hepatocytes, demonstrating that AMPK suppresses TG secretion primarily through Rab2A (Fig. 5G).

In summary, our findings further highlight the crucial role of Rab2A, likely modulated by AMPK signaling, in binding with HSD17B13 to facilitate dynamic Golgi-LD interactions and promote VLDL secretion.

Animals

Male C57BL/6JGpt mice (Strain No. N000013), Rab2a-flox/flox mice (Strain No. T018874), Ldlr-KO mice (Strain No. T001464), and Alb-iCre mice (Strain No. T003814) were generated and procured from Gempharmatech (Nanjing, China). Rab2a hepatocytes-specific knockout mice were generated by crossing Rab2a-flox/flox mice with Alb-iCre mice. while Ldlr-KO/ Rab2a-flox/flox mice were produced by crossing Ldlr-KO mice with Rab2a-flox/flox mice. Genotyping primers were listed: Rab2a-flox/flox mice: 5′-CACTCACAGACACATTCCCACACA-3′ and 5′-AGCAAGCCTTGGTCTTTCCAAC-3′; Alb-iCre mice: 5′-TGGATGCCACCTCTGATGAAGTC-3′ and 5′-TCCTGGCATCTGTCAGAGTTCT CC-3′; Ldlr-KO mice: 5′-CTCCCAGGATGACTTCCGAT-3′ and 5′-CGCAGTGCTCCTCATCTGAC -3′. Mice with diet-induced hyperlipidemia were fed with either Western diet (WD) (No. D12079B, Research Diets) or high-fat-high-cholesterol diet (HFHCD) (No. D09100310, Research Diets) starting from 8 weeks.

Adeno-associated virus serotype 2/8 (AAV2/8) -mediated gene knockdown in the mouse liver was conducted as previously described (Chen et al, 2022). The AAV2/8 viruses pAAV-U6-shRNA (NC)- Cbh-EGFP-WPRE, pAAV-U6-shRNA (Apob) -Cbh-EGFP-WPRE and pAAV-U6-shRNA (Hsd17b13) -Cbh-EGFP-WPRE were obtained from Obio Technology (Shanghai, China). The sequences of shRNA-Apob and shRNA-Hsd17b13 were as follows: shRNA-Apob (Cheng et al, 2016): (5′-ACCGCAGACAAGCACCTGGAAATTCTCGAGAATTTCCAGGTGCTTGTCTGCTTTTTTG-3′ and 5′-CTAGCAAAA AAGCAGACAAGCACCTGGAAATTCTCGAGAATTTCCAGGTGCTTGTCTGC-3′); shRNA-Hsd17b13 (Wang et al, 2022): (5′-ACCGGCGTCATCATCTACTCCTACCCTCGAGGGTAGGAGTAGATGATGACGCTTTTTTG-3′ and 5′-CTAGCAAAAAAGCGTCATCATCTACTCCTACCCTCGAG GGTAGGAGTAGATGATGACGC-3′).

Male mice were randomized into groups for each experiment. Eight-week-old wild-type C57BL/6 J mice received intravenous injections of 5 × 10¹¹ vg AAV2/8 virus via the tail using a 29-gauge insulin syringe (BD). All assays were conducted two weeks post-AAV2/8 virus injection. Mice were housed in a pathogen-free environment with a 12-hour light/12-hour dark cycle and had ad libitum access to water and food. All animal breeding, husbandry, care, and use procedures adhered to the guidelines outlined by the Ethics Committees of Anhui Medical University (Approval number LLSC20200327 and LLSC20241078).

Cell culture, transfection, knockdown and plasmids

Human embryonic kidney HEK293T cells, human liver carcinoma Huh7 cells were sourced from the Cell Resource Center, Chinese Academy of Medical Sciences and Peking Union Medical College (China). Generally, transient transfection followed established protocols (Chen et al, 2022), where cell seeding occurred on day 0, plasmid transfection with linear polyethylenimine (LPEI) was performed on day 1, and detailed experiments were conducted on day 3.

The plasmids in this study were constructed using standard molecular cloning techniques, incorporating site mutations and sequence truncations via quick-change mutagenesis. Key plasmids are listed in “Reagents and Tools” table.

Isolation of mouse primary hepatocytes was conducted as previously described (Chen et al, 2017a; Chen et al, 2016). Plasmids transfection and siRNA knockdown (siHsd17b13: GCGTCATCATCTACTCCTACC; siRab2a: GCCTATCTCTTCAAGTACATC) in primary hepatocytes utilized Lipo3000 reagents or viruses (Lenti-virus or AAV2/8-virus). Lentiviral transfection was constructed in our laboratory. The AAV2/8 viruses pAAV-TBG-Rab2A-Flag-P2A-GFP and pAAV-TBG-Rab2A (∆032-042)-Flag-P2A-GFP were provided by the ChuangRui Bio (Lian Yungang, China). Cells were cultured at 37 °C with 5% CO2 in DMEM (Biological Industries) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin sulphate (Thermo Fisher), and 10% fetal bovine serum (FBS, Biological Industries).

Immunoprecipitation, immunoblotting and antibodies

Proteins immunoprecipitation (IP) and immunoblotting were conducted in accordance with established protocols (Chen et al, 2022). Briefly, tissue or cell samples were promptly collected and homogenized in the respective RIPA buffer or IP buffer containing proteinase inhibitors. For immunoprecipitation assay, quantified samples were primarily incubated with immune-beads for several hours, followed by removal of non-specific binding proteins through washing. Subsequently, the aliquots were quantified, subjected to SDS-PAGE, transferred to PVDF, and incubated with relevant antibodies. Western blotting signals were then captured using an autoradiography machine (Tanon-5200). Quantification of protein levels was performed using Image J (National Institutes of Health, https://imagej.nih.gov/ij/) and normalized to the internal reference. The primary antibodies for LDLR and EGFP were generously gifted by Professor Song (Zhou et al, 2023). Detailed information regarding primary antibodies and secondary antibodies are listed in “Reagents and Tools” table.

Rab2A binding protein immunoprecipitation and LC-MS/MS

Liver protein samples (5 mg) were extracted from wild-type mice using IP buffer containing proteinase inhibitors. The primary antibody against Rab2A (5 μl) was incubated with the liver samples overnight, and binding proteins were collected with protein A/G affinity beads (No. L-1004, Biolinkedin, China). Subsequently, the beads were prepared for mass spectrometry at Bioprofile Company (Shanghai, China).

In Brief, the bound proteins were extracted from IP beads using lysis buffer (4% SDS, 100 mM DTT, 100 mM Tris-HCl pH 8.0). The IP beads samples were boiled for 3 min and further ultrasonicated. Undissolved beads were removed by centrifugation at 16,000 × g for 15 min, and the supernatant containing proteins were collected. Finally, the protein suspension was digested with 2 μg trypsin (Promega) overnight at 37 °C. The peptides were collected by centrifugation at 16,000 × g for 15 min and desalted with C18 StageTip for further LC-MS analysis.

LC-MS/MS experiments were performed on a Q Exactive Plus mass spectrometer coupled to Easy nLC1200 (Thermo Scientific). Peptides were initially loaded onto a trap column in buffer A (0.1% Formic acid in water). Reverse-phase high-performance liquid chromatography (RP-HPLC) separation was carried out using a self-packed column at a flow rate of 300 nl/min. The RP-HPLC mobile phase A was 0.1% formic acid in water, and B was 0.1% formic acid in 95% acetonitrile. The gradient was set as following: 2–4% buffer B from 0 min to 2 min, 4% to 30% buffer B from 2 min to 47 min, 30% to 45% buffer B from 47 min to 52 min, 45% to 90% buffer B from 52 min to 54 min, 90% buffer B was maintained until 60 min. MS data was acquired using a data-dependent top20 method dynamically choosing the most abundant precursor ions from the survey scan (350–1800 m/z) for HCD fragmentation. The full MS scans were acquired at a resolution of 70,000 at m/z 200, and 15,000 at m/z 200 for MS/MS scan. The maximum injection time was set to for 50 ms for MS and 25 ms for MS/MS. Normalized collision energy was 28 and the isolation window was set to 1.6 m/z. Dynamic exclusion duration was 30 s.

The MS data were analyzed using MaxQuant software version 1.6.1.0. MS data were searched against the UniProtKB Mus musculus database. The database search results were filtered and exported with <1% false discovery rate (FDR) at peptide-spectrum-matched level, and protein level, respectively. The summary of Rab2A-specific binding proteins is presented in Appendix Table S1.

Immunofluorescence staining and imaging

Huh7 cells were transfected with related plasmids expressing Rab2A, HSD17B13, RCAS1, ERGIC-53, each tagged with a fluorescent protein epitope. Endogenous Rab2A was stained with a primary antibody (Proteintech technology, 67501-1-Ig, 1:100). Golgi apparatus labeling was achieved using a GM130 primary antibody (Proteintech technology, 11308-1-AP, 1:200) or BFP-RCAS1. ER-Golgi intermediate compartment (ERGIC) was labeled with mCherry-ERGIC-53 fluorescent protein. Endoplasmic reticulum (ER) was visualized using a KDEL expression plasmid provided by Professor Baoliang Song. Lipid droplets were stained with BODIPY ((4,4-Difluoro-1,3,5,7,8-Pentamethyl-4-Bora-3a,4a-Diaza-s-Indacene), No. D3922, Thermo Fisher) or HCS LipidTOX (No. H34477, Thermo Fisher), and lysosome was labeled with lysoTracker (Invitrogen, LysoTracker™ Red DND-99, L7528).

In mouse primary hepatocytes, the Golgi was labeled with GM130 (Proteintech technology, 11308-1-AP, 1:100) or Golgin-97 (Cell signaling technology, 13192, 1:100) primary antibodies, lipid droplets were visualized using BODIPY or HCS LipidTOX, and cellular nuclei were stained with DAPI (No. BL105A, Biosharp). Endogenous Rab2A localization was evaluated with a mouse anti-Rab2A primary antibody (Proteintech technology, 67501-1-Ig, 1:100), and endogenous HSD17B13 localization was detected using a rabbit anti-HSD17B13 primary antibody (Thermo Fisher, PA5-109834, 1:20).

For fatty acid uptake in mouse primary hepatocytes, the cultured cells were pretreated in serum-free medium for 4 h, followed by treatment with 2.5 μM bovine serum albumin (BSA)-conjugated fatty acid (BODIPY™ FL C12, Thermo Fisher, D3822) for 1 h according to the standard protocol (Hao et al, 2020).

Generally, standard procedures, including washing, fixation (4% paraformaldehyde (PFA) for 30 min at room temperature or ice-cold methanol for 15 min at 4 °C), quenching (50 mM NH₄Cl for 15 min), permeabilization (0.1% Triton X-100 for 10 min or 50 µg/ml digitonin for 5 min), blocking (3% BSA for 30 min), staining primary antibodies, secondary antibodies, and mounting, were performed. Subsequently, images were captured using a Zeiss confocal microscope (LSM800 and LSM980). The representative results are presented and the corresponding quantification was performed using Image J (National Institutes of Health, https://imagej.nih.gov/ij/).

Measurement of lipids level in liver and serum

Triglyceride (TG) and total cholesterol (TC) quantification were performed on frozen liver and serum samples as previously described (Chen et al, 2022). For liver samples, lipids extraction involved weighing, homogenizing, saponifying, and extracting. TG levels in liver were determined using the free glycerol reagent (No. F6428, Sigma-Aldrich) with glycerol (No. G7793, Sigma-Aldrich) as the standard. TC levels in the liver was measured using a LabAssay Cholesterol kit (No. 294-65801; Wako Chemicals USA, Inc.) following the standard protocol. The quantification of parameters in serum samples were determined using TG kits (No. A110-1-1, Nanjing Jiancheng Bioengineering Institute, China; No. 290-63701, Wako Chemicals USA, Inc.) and TC kits (No. A111-2-1, Nanjing Jiancheng Bioengineering Institute, China; No. 294-65801, Wako Chemicals USA, Inc.).

Fast protein liquid chromatography (FPLC)

Lipoprotein profiling analysis was conducted as previously described (Zhang et al, 2019). In brief, 500 μL of pooled serum was loaded onto a Superdex 200 Increase gel filtration column (10/300 GL, GE Healthcare, No. 28-9909-44) using an ÄKTA puro FPLC system, separated at a flow rate of 0.3 ml/min in standard phosphate-buffered saline buffer supplemented with 5 mM EDTA, and 200 μl per fractions were collected for TG and TC detection (Fraction 30-Fraction 95).

Transmission electron microscope

Primary hepatocytes, as detailed in the preceding methodology, were immersed in 1 ml of 2.5% (v/v) glutaraldehyde for 4–5 h at 4 degrees Celsius. Subsequently, the samples underwent three 15-minute washes with 0.1 M phosphate buffer (PB buffer, pH 7.0). Following this, fixation was performed using a 1% osmium tetroxide solution for 1–2 h. After meticulous removal of excess osmium tetroxide, an additional three 15-minute washes with PB buffer were carried out. Dehydration was systematically performed using a series of ethanol solutions (30%, 50%, 70%, 80%, 90%, and 95%) for 15 min each, followed by two treatments with anhydrous ethanol for 20 min each. Finally, the samples underwent a 20-minute treatment with pure acetone. A one-hour treatment with a mixture of embedding agent and acetone (v/v = 1/1) preceded a three-hour treatment with a mixture of embedding agent and acetone (v/v = 3/1). Subsequently, the samples were treated overnight with pure embedding agent. Following infiltration overnight, the samples were embedded and heated at 70 °C overnight to achieve well-embedded specimens. Using a LEICA EM UC7 ultramicrotome, 70 nm sections were obtained. These sections were stained with a lead citrate solution and a 50% ethanol-saturated solution of uranyl acetate for 5 min each, before being observed under a transmission electron microscope.

VLDL-TG secretion in mice and primary hepatocytes

For the in-vivo VLDL-TG secretion assay, mice were intraperitoneal injected with tyloxapol (an inhibitor of lipoprotein lipase (LPL)) at a dose of 500 mg/kg following a 16-h fasting. Subsequently, the tail vein serum was collected at 0, 1, 2, and 4 h, individually, and subjected to TG detection. In the ex-vivo VLDL-TG secretion assay with primary hepatocytes, cultured hepatocytes following plasmids or siRNA transfection were initially incubated with 200 μM BSA-conjugated oleate acid containing ATGL and HSL inhibitors for 2 h. The cells were then washed with phosphate-buffered saline (PBS) and then incubated with Opti-MEN adding 1% fatty acid-free BSA. The medium was consecutive collected for TG measurement at indicated hours.

Fatty acids absorption in mice

Mice were subjected to a 4-h fasting period and subsequently administered olive oil (6 μl/g) via oral gavage, following the established protocol (Chen et al, 2017b). The tail vein serum was then collected at the specified time points for the detection of TG.

Subcellular fractionation and lipids measurement

Subcellular fractionation was conducted according to established procedures (Li et al, 2012). In brief, Fresh liver samples were promptly harvested and homogenized in ice-cold lysis buffer (10 mM HEPES, pH 7.4; 150 mM sucrose; 0.5 mM DTT; and 1× cocktail inhibitors) using a loosely fitted Dounce homogenizer for ~20 cycles. The homogenates were then centrifuged at 1900 × g for 10 min, and the resulting supernatant underwent ultracentrifugation at 100,000 × g for 90 min using a Beckman SW60Ti rotor. The lipids fraction was floated at the top, and the pellet was collected, resuspended in 800 μl of an 8.58% sucrose solution, and then loaded onto the top of a sucrose density gradient with layers at the following concentrations (from the top to bottom): 20% (160 μl), 30% (480 μl), 35% (800 μl), 40% (800 μl), 45% (480 μl), 50% (320 μl), and 60% (160 μl) sucrose. Following ultracentrifugation at 43,900 rpm for 18 h in a Beckman SW60Ti rotor, 12 fractions (~324 μl per fraction) were collected from top to bottom. The distribution patterns of the subcellular compartment markers were assessed through Western blotting, utilizing GRP94 and PDI as markers for endoplasmic reticulum (ER) and GM130, Golgin-97 as markers for the Golgi. Golgi fractions (Fraction 3–5) and ER fractions (Fraction 8–12) were combined and centrifugated at 100,000 × g for 90 min. The resulting pellets were then resuspended in lysis buffer (20 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton-X100, and protease inhibitors, at pH 7.4). TG and TC levels were measured using kits (TG, No. 290-63701; TC, No. 294-65801, Wako Chemicals), with protein mass serving as the internal control.

Sucrose gradient separation of Apo B-containing lipoproteins in Golgi, ER and serum

The isolation of Apo B-containing lipoproteins from Golgi, endoplasmic reticulum (ER), and serum was performed as previously described (Li et al, 2012). Briefly, Golgi and ER pellets obtained from the preceding centrifugation step were lysis with a 1.6 ml solution buffer containing 0.1 M sodium carbonate (pH 11.0) and deoxycholic acid (0.025%) for 30 min at room temperature. Subsequently, BSA was added to achieve a final concentration of 5 mg/ml, and the sample was centrifugated at 50,000 rpm for 1 h in a Beckman SW60Ti rotor. The resulting supernatant was adjusted to PH 7.4 with addition of 10% acetic acid, brought to a total volume of 1.8 ml with PBS, and adjusted to a sucrose concentration of 12.5% (w/v). The supernatant was layered in a centrifugation system, composed from top to bottom: PBS (982 μl), sample (1.686 ml), 20% sucrose (666 μl), and 49% sucrose (666 μl). All solutions were supplemented with protease inhibitors. Following centrifugation at 33,500 rpm for 43 h in a Beckman SW60Ti rotor, 12 fractions (each 324 μl) with densities ranging from 0.988 g/ml to 1.162 g/ml were collected from the top of the tube. The density of Lipoproteins in each fraction was calculated by comparing with the weight of water (VLDL, 0.96 g/ml–1.006 g/ml; IDL/LDL, 1.006 g/ml–1.063 g/ml; HDL, 1.063 g/ml–1.21 g/ml). The percentage of lipoproteins was evaluated through immunoblotting using Apo B as the marker protein.

Serum (200 μl) collected after tyloxapol injection for 4 h was diluted into 1.686 ml with PBS, and also adjusted to a sucrose concentration of 12.5% (w/v). Following the same system and method as described above, 12 fractions were separated after centrifugation and Apo B was also selected as the marker for lipoproteins.

VLDL isolation and transmission electron microscopy analysis

Serum isolation and transmission electron microscopy analysis were conducted as previously detailed (Wang et al, 2020). In summary, 110 μl of serum collected from mice after tyloxapol injection for 4 h was initially diluted 1:30 with 3.19 ml PBS. Subsequently, 1.0 ml iodixanol (60% density) was added to 3.0 ml mixture to achieve a 15% concentration of iodixanol. The final mixture underwent ultracentrifugation with SW60Ti at 350,000 × g for 3 h. The top 20 μl of very-low-density lipoprotein (VLDL) were collected for negative staining using phosphotungstic acid. Images were captured with a JEOL JEM-1400Plus. The diameter of VLDL particles was quantified by Image J software and subclassified into larger VLDL (60–200 nm), medium VLDL (35–60 nm) and small VLDL (27–35 nm), with analysis conducted on more than 200 VLDL particles (Nikolac, 2014; Wojczynski et al, 2011).

Lipid droplets pulldown assay

LDs in liver tissues (shNC vs. shHsd17b13 or wild-type) were purified according to a well-constructed protocol with some modifications (Ding et al, 2013). In brief, about 500 mg of liver was homogenized on ice ten times with a loose-fitting Dounce in buffer A solution (20 mM tricine, 250 mM sucrose (pH 7.8) and 0.2 mM PMSF) and then centrifugated at 100 × g for 10 min at 4 °C to remove almost unbroken tissues. Subsequently, the suspensions from liver were disrupted with syringe (22 G) and then centrifugated at 3000 × g for 10 min to remove nuclei, cell debris and unbroken cells. The resulting supernatant (1 ml) were transferred to new Eppendorf tubes for further centrifugation (2000 × g for 30 min at 4 °C) after being loaded with 200 μl of buffer B solution (20 mM HEPES, 100 mM KCl and 2 mM MgCl2 (pH 7.4)) on the top. LDs were carefully collected from the top band of gradient and washed 3 times with buffer B (20,000 × g for 5 min at 4 °C).

The organelle pulldown assay was also modified from a previous procedure (Krahmer et al, 2018). Briefly, mouse primary hepatocytes, either with wild-type, Rab2A deficiency, or stimulated with A769662 (100 μM, 4 h), were directly collected and homogenized in lysis buffer (200 mM Tris pH 7.4, 0.5 mM EDTA, 5 mM KCl, 3 mM MgCl₂, protease inhibitor, phosphatase inhibitor cocktail), and then centrifugated at 1000 × g for 10 min to remove nuclei, cell debris and unbroken cells. LDs in the cell lysate were further removed by centrifugation at 20,000 × g for 15 min. Subsequently, purified LDs from the upper steps (about 400 μg proteins) were isolated and incubated with the respective cellular organelles mix (about 800 μg proteins) at 37 °C for 1.5 h. Following this incubation, LDs and interacting organelles (ER and Golgi) were collected via centrifugation at 20,000 × g for 15 min, and then LDs on the top band of gradient were washed twice with buffer B (20,000 × g for 5 min at 4 °C). The results were analyzed using western blotting.

Statistical analysis

All statistical data are expressed as the mean ± standard error of the mean (s.e.m.). Each experiment was independently replicated at least thrice without specific statements, yielding consistent results. Statistical analyses were conducted using GraphPad Prism 7 (GraphPad Software). Unless otherwise indicated, unpaired two-tailed Student’s t test and two-way ANOVA were employed for data analysis, and significance was determined at P < 0.05.

The authors declare no competing interests.