Intestinal Atp8b1 dysfunction causes hepatic choline deficiency and steatohepatitis

Choline is an essential nutrient, and its deficiency causes steatohepatitis. Dietary phosphatidylcholine (PC) is digested into lysoPC (LPC), glycerophosphocholine, and choline in the intestinal lumen and is the primary source of systemic choline. However, the major PC metabolites absorbed in the intestinal tract remain unidentified. ATP8B1 is a P4-ATPase phospholipid flippase expressed in the apical membrane of the epithelium. Here, we use intestinal epithelial cell (IEC)-specific Atp8b1-knockout (Atp8b1IEC-KO) mice. These mice progress to steatohepatitis by 4 weeks. Metabolomic analysis and cell-based assays show that loss of Atp8b1 in IEC causes LPC malabsorption and thereby hepatic choline deficiency. Feeding choline-supplemented diets to lactating mice achieves complete recovery from steatohepatitis in Atp8b1IEC-KO mice. Analysis of samples from pediatric patients with ATP8B1 deficiency suggests its translational potential. This study indicates that Atp8b1 regulates hepatic choline levels through intestinal LPC absorption, encouraging the evaluation of choline supplementation therapy for steatohepatitis caused by ATP8B1 dysfunction.

choline chloride (150 μCi/kg body weight; PerkinElmer, Waltham, MA).Blood samples were collected from the tail vein into EDTA-coated tubes at 30 and 60 min after the dosing.The plasma was separated by centrifugation at 1700g for 15 min and stored at −80 °C.The mice were sacrificed 120 min after the dosing, and their liver and SI were excised.The SI lumen was flushed with cold PBS containing 0.5 mM taurocholic acid, and the flushing flow was collected.Each sample was processed using SOLVABLE (PerkinElmer) according to the manufacturer's protocol.The amount of radioactivity was measured by a liquid scintillation counter (Tri-Carb 3110TR, PerkinElmer), and the data were expressed as disintegrations per minute (DPM).

Metabolomic and lipidomic analysis HILIC/MS/MS analysis
Metabolites were extracted from plasma, liver, and IEC with methanol.After centrifugation at 20,000g for 5 min at 4 °C, the supernatant was mixed with 400 mM ammonium formate at a ratio of 19:1.The mixture was centrifuged at 20,000g for 5 min at 4 °C.The supernatant was applied to an LC/MS/MS system, which consisted of a UHPLC Nexera liquid chromatography system (Shimadzu Co., Kyoto, Japan) and a 5500QTRAP mass spectrometer (AB Sciex Pty. Ltd., Toronto, Canada).The analytes were separated through a ZIC-cHILIC column (2.1100 mm, 3 μm; Merck Millipore) at a temperature of 30 °C with a gradient elution of the mobile phases, 10 mM ammonium formate aqueous solution and acetonitrile at a flow rate at 0.4 mL/min.The eluent was ionized using electrospray ionization and scanned with multiple reaction monitoring (MRM) mode, as described previously 1 .The MRM data were processed using MultiQuant 3.0 (AB Sciex).Each MRM peak was assigned to a target molecule in comparison with its authentic standard.For identified metabolites, MRM peak areas were calculated and relative comparisons were made between samples.

GC/MS/MS analysis
Metabolites were extracted from plasma, liver, and IEC with methanol.After centrifugation at 20,000g for 5 min at 4 °C, internal standards labelled with stable isotopes were added to the supernatant and dried under a stream of nitrogen gas.The dried samples were derivatized by oximation and trimethylsilylation.The derivatized metabolites were injected into an Agilent 7890A series gas chromatography system.Chromatographic separation was performed in a J&W Scientific DB-5MS-DG column (30 m0.25 mm i.d., df = 0.25 μm; Agilent Technologies Inc., Santa Clara, CA) by a temperature gradient with helium gas flow at a rate of 1 mL/min.Eluted metabolites were introduced into an Agilent 7010B triplequadrupole mass spectrometer for electron impact ionization and scanned in MRM mode.The MRM data were processed by MassHunter (Agilent Technologies Inc.).For identified metabolites, MRM peak areas were calculated and relative comparisons were made between samples.

Non-targeted lipidomic analysis
Lipidomic analysis was performed as previously described 2 .A sample was extracted with ethanol supplemented with 0.002% butylated hydroxytoluene and centrifuged at 21,500g for 5 min.The supernatant was injected into a CAPCELL PAK ADME column (2.1100 mm, 2.7 μm; Shiseido, Kyoto, Japan) and maintained at a temperature of 60 °C, and the lipids were separated by gradient elution of aqueous mobile phase (MilliQ water with 0.01% acetic acid, 1 mM NH3, and 10 μM EDTA-2Na) and organic mobile phase [0.001% acetic acid and 0.2 mM NH3 in ethanol/isopropanol (1:1)], with the flow rate set to 0.7 mL/min.The eluents were introduced to a Q Exactive HF-X Mass Spectrometer (Thermo Fisher Scientific).Their mass spectra were acquired in the data-dependent mode.Precursor and product ion spectra by higherenergy collisional dissociation were scanned with the orbitrap analyzer at a resolution of 120,000 and 7,500 full width at half maximum at 200 m/z.The raw LC/MS data were processed by Expressionist Refiner MS software (ver.8.2; Genedata AG, Basel, Switzerland).Each MS peak was compared with the in-house lipid database including information on retention time, exact mass, and the preferred adduct ion species, and its structure was estimated.For identified lipids, MS peak areas were calculated and relative comparisons were made between samples.

Targeted lipidomic analysis of lysophospholipids
Analysis was performed with the targeted lipidomic method according to the previous report 3 .Lipids were extracted from a sample with methanol in the presence of internal standards.The extract was further purified with chloroform/water (5:4, v/v).The organic layer was collected into a clean tube and dried under nitrogen stream.After its reconstitution with methanol, the sample was injected into a supercritical fluid chromatography/triple-quadrupole mass spectrometry system consisting of an ACQUITY Ultra-Performance Convergence Chromatography system and a Xevo TQ-XS triple-quadrupole mass spectrometer coupled with an electrospray ionization probe (Waters, Milford, MA).Lipids were loaded on an ACQUITY UPC2TM Torus diethylamine column (100 × 3.0 mm inner diameter (i.d.), particle size: sub-1.7 μm, Waters) using an autosampler and eluted under the gradient of CO2 and ammonium acetate/methanol/water (0.001:95:5, w/v/v, modifier).The eluates were ionized with modifier and scanned with selected reaction monitoring mode.MS data obtained was processed with TargetLynx (Waters).The concentration of each lysophospholipid in a sample was calculated with a single-point calibration.

Visualization of lipid-subtype enrichment in lipidomic analysis
The result of lipidomic analysis was evaluated using Kolmogorov-Smirnov (KS) running sum statistic to determine the enrichment of lipid subtypes.For each tissue, the change in each lipid content between Atp8b1 IEC-KO mice and Atp8b1 flox/flox mice was scored as follows: where  , is the amount of lipid  in the  th sample of the Atp8b1 IEC-KO mouse group, and  , and  , are mean and standard deviation of the amount of lipid  in the Atp8b1 flox/flox mouse group.KS statistics were calculated for each lipid subtype and are plotted as running sum at the top of Figure 4A and E. The middle heatmap and the bottom barcode represent the magnitude of difference in each lipid content and the positions of the lipids classified into the indicated subtypes, respectively.

Pathway-level analysis for choline metabolites
Pathway-level analysis for choline metabolites was performed using the generally applicable gene set enrichment (GAGE) method 58 .Briefly, metabolites in the choline pathway and overall metabolites were subjected to Welch's t-test, comparing all sample combinations of Atp8b1 IEC-KO mice (line #12) and the littermate Atp8b1 flox/flox mice.Negative log-sum values of resultant p-values were adjusted based on control group dependencies, and the integrated p-value was computed on a Gamma distribution with K degrees of freedom and a scale of 1.0, where K represents the number of samples of Atp8b1 IEC-KO mice (line #12).Data was analyzed with scipy (ver 1.11) modules of python 3.

Measurement of choline metabolites
The plasma concentrations of choline, betaine, and DMG were determined with an LC/MS/MS assay.For the calibration curve, working solutions were prepared by dilution of a stock solution to the concentrations of 0.5, 1, 2.5, 5, 10, 25, 50, 100 and 250 μmol/L for choline and betaine, and 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10 and 25 μmol/L for DMG in ethanol.Ten microliters of a working solution was mixed with 10 μL of the IS solution and water, and 970 μL of ethanol and then processed as described above.The linear calibration curve was drawn with a weighting of 1/ and confirmed by assessing the accuracy within a range of ± 20%.The SRM data were processed by MultiQuant 3.0.2(AB Sciex).

Measurement of methionine metabolites
The plasma concentrations of methionine, homocysteine, SAM, and S-adenosyl homocysteine were determined  8.The MRM data were processed by MultiQuant 3.0.2.A plasma concentration of each metabolite was quantitated with a calibration curve prepared with an authentic standard.

Lpcat activity assay
Acyltransferase activity was determined by measuring the incorporation of 18:1 (n9) oleoyl coenzyme A (Avanti Polar Lipids, Alabaster, AL) into 12:0 LysoNBD-PC (Avanti Polar Lipids) as described previously 4 .Briefly, 110 6 IEC were prepared from 4-week-old Atp8b1 IEC-KO mice (line #12) and littermate Atp8b1 flox/flox mice, suspended in 100 μL of homogenization buffer [75 mM Tris-HCl, pH 7.5 containing 1 mg/ml BSA (Sigma-Aldrich)], and sonicated.The debris was removed by centrifugation at 1000g for 5 min, and the supernatant was used as the source of Lpcat.Each reaction mixture contained 2 μg of LysoNBD-PC, 10 μg of 18:1 (n9) oleoyl coenzyme A, and the prepared specimens in 100 μL of homogenization buffer.The reaction tube was kept at room temperature for exactly 10 min, and the reaction was terminated by adding 375 μL of chloroform/methanol (1:2, v/v) and mixing vigorously.Lipids were extracted by the Bligh and Dyer method 5 .The organic phase was transferred to a 1.5-ml tube and dried using a centrifugal concentrator (TOMY SEIKO, Tokyo, Japan).Lipids were redissolved in 40 μL of chloroform and applied to a TLC plate.The running solvent was chloroform/methanol/water (65:25:4, v/v).The fluorescence signal was detected by ImageQuanta LAS 4000 (FUJIFILM), and the intensity was measured by ImageJ software (ver.1.53c; National Institutes of Health).

Flippase assay
The incorporation of NBD-phospholipids was analyzed by flow cytometry as described previously 6,7 .Forty-eight hours after transfection, the transfected CHO-K1 cells and HEK293T cells were detached from dishes in PBS containing 5 mM EDTA and collected by centrifugation.Eight-week-old Atp8b1 Tax-iIEC-KO mice and littermate Atp8b1 flox/flox mice were treated with 1 mg Tax intraperitoneally daily for 4 days and subjected to IEC preparation.The cells (210 6 cells/sample) were washed and equilibrated at 15 °C for 15 min in 500 μL of Hanks' Balanced Salt Solution (HBSS; Thermo Fisher Scientific).An equal volume of 1 μg/mL 18:1-06:0 NBD-PC (Avanti Polar Lipids) or 10 μg/mL 12:0 LysoNBD-PC was added to the cell suspension and incubated at 15 °C.At each time point, 300 μL of cell suspension was collected and mixed with 300 μl of ice-cold HBSS containing 5% fatty acid-free BSA (Sigma-Aldrich) to extract NBD-lipids incorporated into the exoplasmic leaflet of the plasma membrane as well as unincorporated NBD-lipids.Next, 10,000 cells were analyzed with a BD FACSCelesta (BD Bioscience, San Jose, CA) to measure fluorescence of NBD-lipids translocated into the cytoplasmic leaflet of the plasma membrane.The mean of the fluorescence intensities per cell was calculated.Cells positive for propidium iodide (DOJINDO, Kumamoto, Japan) were excluded from the analysis.

Cytotoxicity assay
Ten mg/mL LPC (FUJIFILM Wako Pure Chemical) in ethanol/water (1:1, v/v) and 5 mM edelfosine (Cayman, Ann Arbor, MI) in ethanol were prepared as stock solutions.The transfected CHO-K1 cells and HEK293T cells were seeded on a 96-well plate at a cell density of 50,000 cells/well.After 24 h, the cells were incubated with fresh medium supplemented with different concentrations of edelfosine or LPC for 24 h.Lactate dehydrogenase release from damaged cells was measured using a Cytotoxicity LDH Assay Kit-WST (DOJINDO), in accordance with the manufacturer's instruction.

Immunoblotting
Specimens were loaded into wells of Mini-PROTEAN TGX Gels (BIO-RAD, Hercules, CA), electrophoresed, and analyzed by immunoblotting as described previously 8 .Fusion Solo7S with FusionCapt17 software (Vilber Lourmat, Collégien, France) was used at high resolution and auto exposure to detect immunoreactivity using the WESTAR C ULTRA 2.0 or the WESTAR SUPERNOVA (Cyanagen, Bologna, Italy).

LTx (n=27) Other cholestasis post-LTx (n=20) Supplementary Table 3. Degree of hepatic steatosis and diarrhea and plasma choline levels in PFIC1 patients in Table 1
LTx, liver transplantation; NA, not applicable.