ABCC4 reduces the liver’s ability to clear LDL cholesterol by lowering LDL receptor levels

Our findings demonstrate that both genetic ablation (Fig. 3) and pharmacological inhibition of hepatic ABCC4 (Fig. 4) effectively elevate hepatic LDLR protein levels and reduce plasma LDL-C levels by approximately 40% in mice. Notably, ABCC4 protein is highly conserved between humans and mice, with 100% sequence identity, suggesting functional conservation across species^39,41,42. To further explore the association between ABCC4 and plasma LDL-C levels in larger populations, we analyzed publicly available datasets from the UK Biobank and the Global Lipid Genetics Consortium^68,69. We identified additional variants at the ABCC4 gene locus that were significantly associated with plasma LDL cholesterol levels (Fig. S5a). Moreover, analysis of public data from the STARNET (Stockholm-Tartu Atherosclerosis Reverse Networks Engineering Task) study⁷⁰ revealed that ABCC4 belongs to a liver gene co-expression network module significantly linked to coronary artery disease and other cardiometabolic outcomes (Fig. S5b).

Obviously, knockout of Abcc4 in AML12 cells effectively inhibited cAMP efflux and promoted activation of the intracellular cAMP signaling pathway, as revealed by RNA-seq analysis. These findings were further validated using ELISA, which demonstrated elevated intracellular and reduced extracellular cAMP levels in Abcc4-deficient cells (Fig. 5g). Similarly, Ceefourin-1 treatment altered cAMP distribution in AML12 cells, supporting the role of ABCC4 in cAMP transport (Fig. 5h). It is widely acknowledged that clinically approved therapies, such as statins and PCSK9 inhibitors, offer mechanistic insight into LDLR regulation by altering cholesterol metabolism and arresting LDLR lysosomal degradation. Building on our earlier findings that Abcc4 deficiency in hepatocytes enhances LDL uptake and accelerates its lysosomal degradation, without affecting sterol sensing and cholesterol levels (Fig. 2g–j; Figs. S2c, S2f, S2h-S2j), we hypothesized that ABCC4 inhibition might offer complementary or synergistic benefits. Then, we investigated whether increased surface LDLR abundance observed with hepatic Abcc4 deficiency was interrelated to PCSK9 regulation. As anticipated, we observed a corresponding decrease in both intracellular and secreted PCSK9 protein levels without significant changes in Pcsk9 mRNA expression in hepatocytes lacking Abcc4 (Fig. 5i–l). Importantly, Ceefourin-1 treatment recapitulated these effects on PCSK9 protein, with no significant changes in Pcsk9 transcript levels (Fig. 5m–p). In animal models, both liver-specific disruption and pharmacological inhibition of ABCC4 significantly resulted in decreased hepatic PCSK9 protein expressions and lower plasma PCSK9 levels compared to the control group (Fig. S7a-S7e). Given that Abcc4 deficiency did not alter Pcsk9 mRNA levels (Fig. 5l), yet resulted in reduced intracellular and secreted PCSK9 levels (Fig. 5i-k), these finding suggest that ABCC4 exerts a posttranscriptional regulatory effect on PCSK9 expression. Therefore, to determine whether ABCC4 influences PCSK9 protein stability, we treated cells with cycloheximide (30 min and 90 min) to block protein synthesis. This treatment (90 min) led to a marked reduction in PCSK9 protein levels in Abcc4-KO cells, suggesting increased degradation (Fig. 5q). To further explore whether PCSK9 mediates the effect of ABCC4 on LDLR, we incubated AML12 cells with recombinant human PCSK9 protein (rhPCSK9; 500 ng/ml). Remarkably, rhPCSK9 treatment completely abolished the effect of Abcc4 silencing on LDLR surface expression and LDL uptake (Fig. 5r, s), implicating that ABCC4 plays a critical role in modulating PCSK9 levels and, consequently, influencing LDLR protein expressions and LDL-C clearance. Given previous reports linking glucagon signaling to PCSK9 regulation^45,71, our RNA-seq data also revealed enrichment of glucagon pathway genes in Abcc4-KO cells (Fig. 5c). In follow-up experiments, we indeed found that glucagon (Glu) treatment significantly reduced both intracellular and secreted PCSK9 levels, mimicking the effect of Ceefourin-1. Combined treatment with glucagon and Ceefourin-1 further suppressed PCSK9 levels, suggesting a synergistic effect and reinforcing the mechanistic role of ABCC4 in PCSK9 regulation (Fig. S8a and S8b). Together, our findings establish that activation of the intracellular cAMP signaling pathway is essential for regulating PCSK9 and LDLR expression in Abcc4-deficiency hepatocytes. Targeting ABCC4, therefore, could be a promising therapeutic strategy for hypercholesterolemia and lipid-associated metabolic disorders.

EPAC2 protein not only serves as a key intracellular effector of cAMP but also acts as an upstream activator of the small GTPase RAP1, as supported by our RNA sequencing data (Fig. 5). To further elucidate the role of RAP1 in regulating LDLR abundance, we generated Abcc4/Rap1a double-knockout (DKO) AML12 cells. Double-knockout of Abcc4/Rap1a in AML12 cells depleted LDLR protein expression levels on cell membrane, as evidenced by immunoblotting data (Fig. 6h, i). In line with the observed decline in LDLR surface abundance, Abcc4/Rap1a DKO cells impeded LDL uptake by LDLR (Fig. 6j, k). Furthermore, Abcc4/Rap1a DKO hepatocytes displayed a marked increase in both intracellular and secreted PCSK9 protein levels, while Pcsk9 mRNA expression remained unchanged (Fig. 6l–n). Overall, these data highlight the pivotal role of the ABCC4-cAMP-Epac2/Rap1a signaling pathway in modulating surface LDLR availability and LDL-C uptake in hepatocytes.

Murine hepatocyte cell line AML12 (Alpha Mouse Liver 12, ATCC CRL-2254) and human embryonic kidney (HEK) 293 T cell line (ATCC CRL-3216) were originally obtained from American Tissue Culture Collection Biobank (Manassas, VA, USA) and were incubated at 37°C with 5% CO2. Human hepatocyte cell line LO2 were originally purchased from QuiCell Biotechnology (QuiCell-L090, Shanghai, China). AML12 cells, LO2 cells and 293 T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco, 11995065) supplemented with 10% heat-inactivated FBS (ExCell, FSP500), penicillin/streptomycin 100 U/ml (Gibco, 15140122) and 2 mM L-glutamine (Gibco, 25030081).

The genome-wide CRISPR library (Liu Mouse CRISPR Knockout Library in lentiCRISPRv2-blast, Addgene#1000000173) was generated by Xiaole Shirley Liu²⁸. Libraries were packaged to produce lentiviruses using HEK 293 T cells as described previously²⁸. Briefly, HEK293T cells with 90% confluency were co-transfected using 50 μl X-TremeGene Transfection Reagent (Roche, 41106502) diluted in 2.5 ml of Opti-MEM (ThermoFisher Scientific, 11058021) that was combined with 10 μg of the plasmid pool DNA and a mixture of 7.5 μg psPAX2 (Addgene#12260) and 3 μg pMD2.G (Addgene#12259). After 12 hours, the medium was replaced, and 72 hours post-transfection, virus-containg supernatant was collected and filtered through a 0.45-μm filter. The virus was aliquoted and stored at −80°C until needed. Functional viral titer was determined using hepatocyte cell lines AML12 cells and LO2 cells by measuring survival rates post-transduction with different viral dilutions under blasticidin selection. The required lentivirus quantity, corresponding to a multiplicity of infection (MOI) of 0.3, was calculated for subsequent genome-wide screening transductions.

Wild-type (WT) murine hepatocyte cell line AML12 were independently screened. For an unbiased approach, approximately 100 million cells were transduced with the Mouse CRISPR Knockout Library viruses at an MOI of ~0.3. For lentiviral transduction, cells were spinfected at 260 rpm for 60 min at 37 °C with 8 μg/mL polybrene (Beyotime, C0351) in medium. Blasticidin S (Beyotime, ST018) was added at 2 μg/mL two days after transduction to eliminate uninfected cells, continuing until selection was complete. Cells were trypsinized, and a partial sample was collected for evaluating the sgRNA coverage. Remaining cells were cultured to maintain the growth phase for the focused screens, ensuring a sufficient cell count. Fourteen days post-transduction, 20 million edited cells were harvested, stained for surface LDLR quantification and sorted using fluorescence-activated cell sorting (FACS) with a BD FACS Melody system. Flow cytometry gating for LDLR expression was defined relative to IgG Isotype control or LDLR-targeted wide-type cells, using Mouse LDLR PE-conjugated Antibody (R&D Systems, FAB2255P). For the focused screens, the top 10% LDLR-positive AML12 cells (LDLR^high subpopulation) were collected. Focused sorted samples were stored at -80 °C for DNA extraction.

Genomic DNA of LDLR^high cells and unsorted control cells were extracted respectively using cell lysis buffer (400 mM Sodium chloride, 10 mM Tris, 2 mM EDTA, 0.5% SDS, pH 8) and then subjected to the established procedures¹⁰⁰. After barcoding of PCR amplicons, sgRNA sequences were amplified and sequenced using the Illumina HiSeq 2500 platform. The sequencing was performed by Azenta Life Sciences Company (Suzhou, China). Data quality, read counts and identification of significant hits in this screen were analyzed using the Model-based Analysis of Genome-wide CRISPR-Cas9 Knockout (MAGeCK) algorithm^29,30.

To verify our screening results, two independent guide sequences were designed targeting Abcc4 gene. The sequence of the two Abcc4 sgRNA are: 5' -ATGCTGCCGGTGCACACCG-3', 5' -CCATGGGGAAGACAACCAC-3'. Each pair of oligonucleotides were annealed and ligated into the lentiCRISPR v2 vector (Addgene#52961). Lentiviruses were produced and used to infect AML12 cells as mentioned above. Following infection with lentivirus for 48 h, the cells were selected by at a concentration of 2 μg/mL puromycin. Knockout editing of gene was confirmed by immunoblotting experiments. ABCC4 genome editing was performed similarly in the human hepatocyte cell line LO2. The sequence of the two ABCC4 sgRNA are as follows : 5' -CCATGGGGAAGACAACCAC-3', 5' -GGCTGTGATCACACTGCCG-3'.

To investigate the downstream signals affected by altered intracellular cAMP level, we relatively generated double knockout of Abcc4/Epac2 and Abcc4/Rap1a murine hepatocyte cell line AML12 by CRISPR Cas9 genome editing. The CRISPR Cas9 genome editing in the two genes were performed as above mentioned. The single guide RNA sequences of two genes were listed in the Supplementary Table 3. Each pair of oligonucleotides were annealed and ligated into the lentiCRISPR v2-Blast vector (Addgene#83480). Lentiviruses were produced and used to infect Abcc4-knockout AML12 cells and the infected cells were selected by Blasticidin S at a concentration of 2 μg/ml. Knockout editing of gene was eventually confirmed by immunoblotting experiments.

All animal experimental procedures were approved by the Ethics Committee of Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. We have complied with all relevant ethical regulations for animal use. Six-week-old male wide type (WT) C57BL/6 mice were purchased from Shanghai Laboratory Animal Center (SLAC). Mice were fed with either a normal-chow diet (NCD) or 60% high-fat diet (HFD) for 4 weeks, 60% HFD (5.24 kcal/g with 20% energy derived from protein, 60% from fat, and 20% from carbohydrate; D12492; Research Diets, Inc., New Brunswick, New Jersey). A rodent diet containing 60 kcal% fat is typically used for studies related to obesity, metabolic disorders, and lipid-related disorders, as it closely mimics high-cholesterol diets found in human populations^101–103. Animals were housed at approximately 21°C with 55% humidity under a 12-hour light/12-hour dark cycle. Ceefourin-1 (Medchem Expression, HY-101453) or vehicle (DMSO and corn oil) was administered intraperitoneally three times weekly at 10 mg/kg body weight for 4 weeks. All mice were euthanized (cervical dislocation) at the end of the experiment, and liver tissues were extracted for further investigation.

Adeno-associated virus 8 serotype (AAV8), packaged by Genechem (Shanghai, China), were used to manipulate ABCC4 expression in mice livers. We chose recombinant AAV serotype-8 expressing the Abcc4 RNAi under the liver-specific apolipoprotein E/human alpha 1-antitrypsin promter [AAV8-ApoE/hAAT; 2 × 10¹¹ viral genomes (VG)] for suppressing the expression of ABCC4 in hepatocytes, which were administered through tail vein injection. The ApoE/hAAT promoter is known for its strong and liver-specific expression, which has been demonstrated in previous studies to provide effective gene regulation specifically within hepatocytes^56,57,104. We and technicians believe that this promoter is particularly well-suited for models where a hepatocyte-specific expression pattern is required for investigating liver-specific gene functions, especially when combined with RNA interference strategies. The Abcc4 gene sequence (NM_001033336) was targeted using synthesized small pieces of interfering double-stranded RNA (RNAi) construct containing the sequence: GAGTTTCTGAAATCTGGTGTA (AAV_Abcc4_RNAi). A control RNAi construct targeting green fluorescent protein (GFP; AAV_GFP_RNAi) contained the sequence: TTCTCCGAACGTGTCACGT. Six-week-old Wild Type (WT) C57BL/6 male mice were administered either AAV_Abcc4_RNAi (AAV-ApoE/hAAT-Abcc4_RNAi) or negative control AAV_GFP_RNAi (AAV-ApoE/hAAT-GFP_RNAi) through tail vein injection.

At the end of the 4th week of the animal experiment, mice were subjected to glucose and insulin tolerance tests. For the intraperitoneal glucose tolerance test (IPGTT), mice were fasted for 6 hours, and baseline blood glucose levels were measured via tail vein using a glucometer. Glucose was then administered intraperitoneally at a dose of 2 g/kg body weight, and blood glucose levels were recorded at 0, 15, 30, 60, 90, and 120 minutes post-injection. For the insulin tolerance test (ITT), mice were fasted for 4 hours and subsequently injected intraperitoneally with insulin (0.75 IU/kg). Blood glucose measurements were taken at 0, 15, 30, 60, 90, and 120 minutes following insulin administration to assess insulin sensitivity.

Blood and tissue samples were collected immediately following the completion of the animal experiments for biochemical analysis. Serum was isolated by centrifugation of blood samples at 12,000× g for 10 min at 4°C. Liver tissue (approximately 100 mg) was homogenized and then centrifugated at 2500× g for 10 minutes at 4°C to collect the supernatant. The concentrations of total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C) and triglycerides (TG) in serum and liver TC, TG levels were measured using commercial kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Additionally, serum levels of PCSK9 (Abcam, ab215538), as well as serum insulin (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) were quantified using commercially-available detection kits, according to the manufacturers' protocols. Lipids were extracted from the AML12 cells by adding 0.2 mL of phosphate buffer for homogenization, followed by ultrasonic disruption in an ice water bath. The resulting homogenate was directly used to measure cellular cholesterol contents using commercial kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). The PCSK9 content in the cell supernatant was measured using the same commercial kit (Abcam, ab215538).

Hematoxylin and eosin (H&E) and Oil Red O staining was performed on formalin-fixed or frozen liver sections. LIver tissues were fixed in 4% paraformaldehyde (Servicebio, Wuhan, China, G1101), embedded in paraffin, and sectioned. Then Hematoxylin and eosin (H&E) staining was performed to observe tissue morphology and pathological changes. Another part of liver tissues were embedded in OCT, frozen, sectioned, and stained with Oil Red O to assess lipid deposition.

Whole lysates and membrane fractions from cultured AML12 cells, LO2 cells and liver tissue were extracted by using RIPA lysis Buffer (Beyotime, P0013B) and Membrane and Cytosol Protein Extraction Kit (Beyotime, P0033) respectively. For the western blotting experiment, equivalent protein samples were separated by 7.5% or 10% SDS-PAGE and transferred onto polyvinylidene fluoride (PVDF) membranes. Membranes were blocked with 5% non-fat milk at room temperature for 1 hour, followed by incubation with primary antibody overnight at 4°C. After 3 washed with TBST, membranes were incubated with secondary antibody at room temperature for 1 hour and visualized using an electrochemiluminescence (ECL) system. The primary antibody used were as follows: anti-LDLR at 1:1000 (Abmart, T55235); ABCC4 at 1:1000 (Abcam, ab15602); PCSK9 at 1:1000 (Proteintech, 27882-1-AP); EPAC2 at 1:1000 (Proteintech, 19103-1-AP); RAP1A at 1:1000 (Proteintech, 16336-1-AP); Vinculin at 1:5000 (Sigma, V9131); Na/K-ATPase at 1:5000 (Abmart, T55159).

Total RNA was extracted from mouse livers or cultured AML12 cells and LO2 cells using TRIzol reagent (Sigma-Aldrich, T9424) following standard protocols. RNA concentrations and purity were assessed by UV/VIS Nano Spectrophotometer (Cobra, Oxford). mRNA was reverse transcribed to cDNA using the HiScript II Reverse Transcriptase Reagent kit with gDNA eraser (Vazyme, China, R222-01). Quantitative RT-PCR were performed with AceQ Universal SYBR qPCR Master Mix (Vazyme, China, Q511-03) on a LightCycler® 96 Instrument (Roche, USA). Primers sequences are provided in Supplementary Table . 4

AML12 cells were used for RNA isolation, library construction, and RNA sequencing. Total RNA was extracted using TRIzol reagent (Sigma-Aldrich, T9424) following the manufacturer's protocol. RNA quality and integrity was assessed with a Bioanalyzer 2100 (Agilent, CA, USA; RIN > 7.0) and confirmed by denaturing agarose gel electrophoresis. cDNA libraries were constructed and sequenced on an illumina Novaseq™ 6000 (LC-Bio Technology CO., Ltd., Hangzhou, China). Pair-end FASTQ files were mapped to the Mus_musculus.GRCm38.101 reference genome using the HISAT2 package. FPKM values were calculated with StringTie and Ballgown (http://www.bioconductor.org/packages/release/bioc/html/ballgown.html↗). Differential gene expression was analyzed using DESeq2, with genes having a P value ≤ 0.05 and absolute fold change ≥2 recognized as differentially expressed (DEGs). Heatmaps were generated, and Gene Ontology (GO) and KEGG pathway analyses were performed on DEGs. Gene set enrichment analysis (GSEA) was conducted using GSEA (v4.1.0) and MSigDB to identify gene sets involved in specific pathways.

Cultured AML12 cells were precooled to 4°C for 30 minutes and incubated with 20ug/ml Dil-LDL (3,3'-dioctadecylindocarbocyanine LDL; Yiyuan biotechnology, YB-0011) at 4°C for 1 hour to allow binding. The cells were then washed three times with ice-chilled PBS and shifted to 37 °C for 3 hours to enable uptake. Afterward, cells were again washed, trypsinized into single cell suspension, and analyzed by flow cytometry with appropriate excitation and emission wavelengths following the suggested protocol. Immunofluorescence microscopy imaging was performed as above. Cells were plated on 35 mm Confocal Dishes, imaged using a 63× lens under oil immersion, and captured with a Nikon A1 Confocal Microscope System.

Intracellular and extracellular cAMP levels were measured in cultured lysates and medium of Abcc4_KO and Control_KO AML12 cells based on the competition between HRP-labeled cAMP and free cAMP according to colorimetric ELISA cAMP Assay Kit (Abcam, ab234585). Altered cAMP distribution in AML12 cells treated with DMSO or Ceefourin-1 was also evaluated as above.

Continuous data are presented as means ± SEM. Statistical analysis was conducted using Prism 10 (GraphPad Software). For comparison between two groups, an unpaired two-tailed Student's t-test was used, assuming equal variance. If variances were unequal, t test with Welch's correction was applied. Similarly, when data were compared among more than 3 groups, a ordinary one-way ANOVA was followed by Bonferroni's multiple comparisons test based on the equal variance. Providing that there were significantly equal variance among more than 3 groups, Brown-Forsythe and a Welch ANOVA test was performed, using the Tamhane T2 method. Statistical significance was defined ad P values < 0.05. Representative images displayed in figures were selected from at least three biologically independent experiments.

Further information on research design is available in the linked to this article. Nature Portfolio Reporting Summary