Mtrr hypomorphic mutation alters liver morphology, metabolism and fuel storage in mice

Nonalcoholic fatty liver disease (NAFLD) is associated with dietary folate deficiency and mutations in genes required for one‑carbon metabolism. However, the mechanism through which this occurs is unclear. To improve our understanding of this link, we investigated liver morphology, metabolism and fuel storage in adult mice with a hypomorphic mutation in the gene methionine synthase reductase (Mtrrgt). MTRR enzyme is a key regulator of the methionine and folate cycles. The Mtrrgt mutation in mice was previously shown to disrupt one‑carbon metabolism and cause a wide-spectrum of developmental phenotypes and late adult-onset macrocytic anaemia. Here, we showed that livers of Mtrrgt/gt female mice were enlarged compared to control C57Bl/6J livers. Histological analysis of these livers revealed eosinophilic hepatocytes with decreased glycogen content, which was associated with down-regulation of genes involved in glycogen synthesis (e.g., Ugp2 and Gsk3a genes). While female Mtrrgt/gt livers showed evidence of reduced β-oxidation of fatty acids, there were no other associated changes in the lipidome in female or male Mtrrgt/gt livers compared with controls. Defects in glycogen storage and lipid metabolism often associate with disruption of mitochondrial electron transfer system activity. However, defects in mitochondrial function were not detected in Mtrrgt/gt livers as determined by high-resolution respirometry analysis. Overall, we demonstrated that adult Mtrrgt/gt female mice showed abnormal liver morphology that differed from the NAFLD phenotype and that was accompanied by subtle changes in their hepatic metabolism and fuel storage.


Introduction
A connection between one-carbon metabolism and fatty liver disease (steatosis) is well established [1]. This is largely based on studies showing that humans and rodents fed diets deficient for one-carbon methyl groups (e.g., choline, folate, methionine, betaine) and/or have specific mutations in genes involved in one-carbon metabolism (e.g., Mat1a, Pemt, Gnmt, Mthfr) develop fatty livers [2][3][4][5][6][7]. Furthermore, patients with liver disease often have abnormal expression of key genes involved in methionine metabolism and hyperhomocysteinemia [8]. However, the mechanism through which this occurs is not well understood.
One-carbon metabolism is at a major junction of several metabolic processes [9]. For instance, it has been implicated in hepatic lipid [10,11] and glycogen metabolism [5,12]. In liver, methyl groups from SAM are required for the sequential methylation of phosphatidylethanolamine (PE) to form phosphatidylcholine (PC) [13,14], the most abundant membrane phospholipid [15,16]. Liver cells have a high demand for PC synthesis, such that it requires as much as 40% of S-adenosylmethionine (SAM) for this purpose [17]. Disruption of hepatic PC production leads to steatosis in mice through accumulation of triacylglycerols (TAGs) and impaired secretion of very low-density lipoprotein particles [18,19]. Apart from hepatic lipid accumulation, glycogen storage disease is also associated with hyperhomocysteinemia [12] caused by perturbation of one-carbon metabolism. Furthermore, mice that lack the enzyme glycine N-methyltransferase (GNMT), which regulates the ratio of SAM to S-adenosylhomocysteine (SAH) [20], develop glycogen storage disease in the liver [5]. Better understanding of how one-carbon metabolism is connected to lipid and glycogen metabolism in the liver will help us gain insight into disease.
During one-carbon metabolism, the enzyme methionine synthase (MTR, also known as MS), which contains a cobalamin (vitamin B 12 ) cofactor, catalyzes the transfer of methyl groups from 5-methyltetrahydrofolate (5-methyl-THF) to cob(I)alamin to produce methylcobalamin. It donates this methyl group to methylate homocysteine to form methionine [21,22]. Methionine acts as a precursor for SAM, which then serves as a methyl donor for dozens of substrates including DNA, RNA, proteins (e.g., histones) and lipids (e.g., PE) [13,14,[23][24][25]. In mammals, the cob(I)alamin cofactor of MTR becomes oxidized by molecular oxygen forming the inactive cob(II)alamin form. Methionine synthase reductase (MTRR or MSR) provides the electron that, along with the transfer of a methyl group from SAM, returns MTR to the active methylcobalamin form [26]. MTR is the only mammalian enzyme that uses 5-methyl-THF and deficiencies in MTR or MTRR causes an accumulation 5-methyl-THF that makes folate unavailable for other reactions, such as purine and pyrimindine biosynthesis [8,21]. Therefore, MTR and MTRR are key enzymes required for the progression of one-carbon metabolism.
Mutations in human MTR and MTRR genes have not previously been associated with hepatosteatosis or glycogen storage disease. Mouse knockout mutations in Mtr and Mtrr, and other genes important for folate uptake and transport, are not conducive with life [21,[27][28][29], resulting in embryonic lethality that prevents the assessment of liver disease in adulthood. We have developed a mouse model with a hypomorphic mutation in the Mtrr gene (Mtrr gt ) that displays many key features of dietary folate deficiency or MTRR mutations in humans [30][31][32][33][34]. These include increased plasma total homocysteine concentrations [21,35], decreased plasma methionine concentrations [21], an altered hepatic SAM:SAH ratio [21], and tissue-specific dysregulation of DNA methylation [35]. These phenotypes are attributed to a substantial decrease in MTR activity [21]. We previously showed that Mtrr gt/gt mice exhibit a wide spectrum and frequency of phenotypes during fetoplacental development (e.g. neural tube and heart defects, growth phenotypes), a portion of which are associated with embryonic lethality at midgestation [35]. The Mtrr gt/gt mice that survive into adulthood display late onset macrocytic anaemia [36]. However, it is unknown whether adult livers of Mtrr gt/gt mice display liver disease similar to other mouse models of defective one-carbon metabolism. Here, we explored the effects of Mtrr gt/gt homozygosity on glycogen and lipid storage and metabolism in adult livers. In doing so, we also investigated whether there were associated alterations in mitochondrial function that often accompany NAFLD and glycogen storage disease [37][38][39][40].

Mice and genotyping
All animal work was carried out in accordance with UK Home Office regulations under the Animals (Scientific Procedures) Act 1986 and underwent review by the University of Cambridge Animal Welfare and Ethical Review Body. The Mtrr gt mouse line was generated through a gene-trap (gt) insertion into the Mtrr locus and upon germline transmission, the Mtrr gt mutation was backcrossed to the C57Bl/6J genetic background for eight generations as previously described in detail [21,35]. C57Bl/6J mice (The Jackson Laboratory, Bar Harbour, ME, USA), which are wildtype for Mtrr, were used as controls for all experiments, and were maintained in house and bred separately from the Mtrr gt mouse line. Mtrr gt/gt mice were generated via Mtrr gt/gt intercrosses. From birth, mice were housed in conventional cages in a temperature and humidity-controlled environment with a 12 h light/ dark cycle. Mice were fed normal breeding diet (Rodent No. 3 breeding chow, Special Diet Services, Essex, UK) ad libtum from weaning onwards. A full dietary breakdown of the chow was previously described [35] including folic acid at 2.99 mg/kg of diet, methionine at 0.37%, choline at 1422.4 mg/kg of diet, and vitamin B 12 at 19.2 μg/kg of diet. All female mice were virgins. DNA samples were obtained from ear tissue for PCR genotyping of the Mtrr gt allele as reported earlier [35].

Tissue collection
Mice were euthanized by cervical dislocation at 9-16 weeks of age. Prior to dissection, mouse body weight was recorded. Whole livers were removed and weighed. The left lobe of each liver was dissected immediately in cold 1× phosphate buffered saline (1× PBS) diluted in diethyl-pyrocarbonate treated water and placed into ice-cold biopsy preservation medium (2.77 mM CaK 2 EGTA, 7.23 mM K 2 EGTA, 6.56 mM MgCl 2 .6H 2 O, 20 mM taurine, 15 mM phosphocreatine, 20 mM imidazole, 0.5 mM dithiothreitol, 50 mM MES, 5.77 mM Na 2 ATP, pH 7.1) for respirometry analysis. The right lateral lobe was prepared for histological analysis. The remaining liver tissue was snap-frozen in liquid nitrogen and stored at −80°C for molecular analysis.

RNA extraction and quantitive reverse transcription PCR (RT-qPCR)
For analysis of folate metabolism and mitochondrial genes, tissue was homogenized using Lysing Matrix D beads (MP Biomedicals, Eschwege, Germany). Total RNA extraction from livers was completed using Trizol (Sigma-Aldrich) according to the manufacturer's instructions. All extracts were treated with DNase I (Thermo Fisher Scientific, Hemel Hempstead, UK). Reverse-transcription was performed with a RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Scientific) using 2 μg RNA in a 20-μl reaction. PCR amplification was conducted using MESA SYBR® Green qPCR MasterMix Plus (Eurogentec, Liege, Belgium) on a DNA Engine Opticon® 2 system (BioRad, Hercules, CA, USA). Transcript levels were normalized to Hprt or Bactin RNA levels. Fold change was quantified using the standard curve method. cDNA levels in C57Bl/6 J tissue were normalized to 1. Experiments were conducted in triplicate with at least three biological replicates.
For analysis of glycogen metabolism genes, RNA was extracted using the GenElute Mammalian Total RNA Miniprep Kit (Sigma-Aldrich) including OnColumn DNase treatment (Sigma-Aldrich).
Reverse transcription was performed using SuperScript IV Reverse Transcriptase (Thermo Scientific) and Random Hexamers (Thermo Scientific) according to manufacturer's instructions using 1 μg total RNA as template. PCR amplification was performed using 2 μl of cDNA (diluted 1:10 in water) in a 10-μl reaction containing 1× DreamTaq Buffer (Thermo Scientific), 0.2 mM of each dNTP (Thermo Scientific), 0.5 μM of each primer (Sigma-Aldrich), 0.5 U DreamTaq (Thermo Scientific) and 0.1× SYBR® Green Dye (Thermo Scientific). Reactions were run in triplicate on a DNA Engine Opticon® 2 system (BioRad). Transcript levels were normalized to the geometric mean of Hprt and Bactin levels and fold change determined using the 2 -ΔΔCT method with expression levels in C57Bl/6 J tissue normalized to 1.
Refer to Supplementary table 1 for primer sequences.

Tissue preparation and histological stains
Livers were fixed in 4% paraformaldehyde (PFA) in 1× PBS, embedded in paraffin using standard techniques and sectioned to 7 μm using a Leica RM2235 microtome (Leica Biosystems, Milton Keynes, UK). For each staining protocol, at least six histological sections from different regions of each liver (N = 3 female mice per genotype) were analysed. Tissue sections were stained with hematoxylin and eosin (H& E) or using a periodic acid-Schiff stain kit (Sigma-Aldrich, Gillingham, UK) according to standard protocols or the manufacturer's instructions. For immunohistochemistry, tissue sections were incubated in 3% H 2 O 2 in 1× PBS for 30 min, treated with porcine trypsin tablets (Sigma-Aldrich) for 10 min, then incubated with blocking serum (5% donkey serum, 1% bovine serum albumin in 1× PBS) for 1 h. Tissue was then incubated in rabbit anti-mouse Ki67 IgG (abcam, Cambridge, UK, cat. No. ab15580, RRID:AB_443209) diluted to 1:100 in blocking serum overnight at 4°C, then in donkey anti-rabbit IgG conjugated to horseradish peroxidase (abcam cat No. ab6802, RRID:AB_955445) diluted to 1:300 in blocking serum for 1 h at room temperature. Peroxidase substrate reactions were conducted with 3,3′-diaminobenzidine (DAB) chromagen substrate kit (abcam) according to the manufacturer's instructions. Sections were counterstained in Nuclear Fast Red (Sigma-Aldrich) before dehydration, clearing and coverslip mounting in DPX mountant (Sigma-Aldrich).

High-resolution respirometry analysis of mitochondrial function
From each left lobe of liver, 15-20 mg sections were cut, blotted dry and quickly weighed before immediate homogenisation in ice-cold respiratory medium (MiR05; 0.5 mM EGTA, 3 mM MgCl 2 .6H 2 O, 60 mM K-lactobionate, 20 mM taurine, 10 mM KH 2 PO 4 , 20 mM HEPES, 110 mM sucrose, 1 mg.ml −1 defatted bovine serum albumin, pH 7.1) at a final concentration of 100 μg.μl −1 . Homogenisation was carried out using an Eppendorf pestle and a DLH overhead stirrer (Velp Scientifica, Usmate, Italy) with a maximum of five pulses of no longer than 5 s at 600 rpm until a consistent homogenate was generated. Liver homogenate (20 μl, equivalent to 2 mg wet tissue) was added to each chamber of an Oxygraph-2 k (Oroboros Instruments, Innsbruck, Austria) containing 2.1 ml MiR05 at 37°C with continuous stirring. Oxygen flux (JO 2 ) was continuously recorded using DatLab software 6.1 (Oroboros Instruments). Calibration at air saturation was carried out every day prior to experimentation and all data was corrected for background instrumental oxygen flux in accordance with the manufacturer instructions.
Respiration rates were analysed using a substrate-uncoupler titration that was run in duplicate on each biological replicate. JO 2 through the N-pathway via electron transfer system (ETS) complex I was assayed in the LEAK state through addition of 1 mM malate and 10 mM glutamate (GM L ). Subsequently, N-pathway supported oxidative phosphorylation (OXPHOS state) was stimulated through addition of 10 mM ADP (GM P ). Maximal OXPHOS capacity was then assessed via addition of 10 mM succinate (GMS P ), which can supply electrons through the S-pathway via complex II. Finally, the ETS was uncoupled from F 1 F O -ATP synthase through addition of the protonophore carbonyl cyanide p-trifluoro-methoxyphenyl hydrazone (FCCP) in 0.5 μM increments until maximal ETS capacity was observed (GMS E ). Respiration rates were normalized to wet tissue mass and also to maximal ETS capacity (GMS E ).
2.6. Citrate synthase and 3-hydroxyacyl-CoA dehydrogenase (HOAD) enzyme activity assays Activities of citrate synthase (EC 2.3.3.1) and HOAD (EC 1.1.1. 35) were analysed spectrophotometrically at 37°C (Evolution 220 UV-Visible spectrophotometer, Thermo Fisher Scientific). Liver homogenate was generated from a small section of frozen tissue in 300 μl homogenisation buffer (20 mM HEPES, 1 mM EDTA, 0.1% v/v Triton X-100, pH 7.2) using a liquid nitrogen-cooled Eppendorf pestle. The supernatant was extracted following centrifugation (two spins at 380 xg for 30 s) and protein density quantified in each sample using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) as per the manufacturer instructions. For citrate synthase activity, a sample containing 10 μg of protein was measured as previously described [41]. The citrate synthase assay buffer contained 18 mM tris-base, 0.1 mM 5,5′-dithiobis (2-nitrobenzoic acid) and 0.3 mM acetyl-CoA, pH 8.0 and the reaction stimulated by addition of 0.5 mM oxaloacetate. The change in absorbance was measured at 412 nm to determine citrate synthase activity. For HOAD activity, a sample containing of 20 μg protein was measured as previously described [42]. The HOAD assay buffer contained 50 mM imidazole, 0.1% v/v Triton X-100 and 0.15 mM NADH at pH 7.4. The reaction was initiated by addition of 0.1 mM acetoacetyl-CoA and absorbance change at 340 nm was measured. HOAD activity was corrected to citrate synthase activity to provide an indication of HOAD activity per mitochondrial unit.

Lipidomic and acyl-carnitine analysis by liquid chromatography-mass spectrometry (LC-MS)
Lipid metabolites were extracted from~50 mg of frozen liver tissue using a modified Bligh and Dyer method [43]. Each sample was homogenized in 600 μl of 2:1 methanol/chloroform for 6 min at 20 Hz, and sonicated for 15 min. To each sample, 400 μl of water/chloroform (1:1) was added and samples were thoroughly vortexed before centrifugation (13,000 xg, 10 min). The aqueous and organic fractions were separated. To maximize metabolite yield, the extraction was repeated on the resultant protein pellet and the aqueous and organic fractions were separately pooled. Half of the resultant aqueous and organic fractions were combined for acyl-carnitine analysis and dried under nitrogen. To the remaining organic fraction, 250 μl of an internal standard mix containing 16 deuterated lipid standards was added before drying under nitrogen (see Supplementary table 2) for lipidomics. For lipidomics, organic metabolites were reconstituted in 100 μl of methanol/chloroform (1:1) and 10 μl of this solution was diluted to 1:20 in propan-2-ol/acetonitrile/water (2:1:1). Open-profiling of lipid species was carried out using an Ultimate 3000 UHPLC system (Thermo Fisher Scientific) coupled to an LTQ Orbitrap Elite Mass Spectrometer (Thermo Fisher Scientific). Mass spectrometry was performed in both positive and negative modes. 5 μl of each sample was injected onto a C18 CSH column (2.1 × 100 mm, 1.7 μM pore size; Waters Ltd., Elstree, UK), which was held at 55°C. The flow rate of the mobile phase gradient was 0.4 μl.min −1 with mobile phase A acetonitrile/water solution (3:2) and mobile phase B acetonitrile/propan-2-ol solution (1:9). The total run time was 20 min during which the mobile phases ran through the column in a gradient as follows: 40% B raised to 43% B at 2 min, 50% B at 2.1 min, 54% B at 12 min, 70% B at 12.1 min, 99% B at 18 min, and then 40% B for 2 min.10 mM ammonium formate (Fisher Scientific) was added to mobile phases A and B for positive mode, and 10 mM ammonium acetate (Sigma Aldrich) added for negative mode.
Metabolites were ionised by heated electrospray for entry into the mass spectrometer with a source temperature of 420°C and capillary temperature of 380°C. Ionization voltages were set to 3.5 kV in positive mode and 2.5 kV in negative mode. Data was collected using the Fourier transform mass spectrometer analyser and the full scan was performed across an m/z range of 110-2000. Peaks were annotated by accurate mass using an automated in-house R script and comparison to the LipidMaps database [44]. Where isobaric species of different classes were annotated, lipid identity was determined through manual identification of fragmentation patterns according to published methods [45][46][47].
The combined carnitine extraction fraction was reconstituted in 150 μl of a methanol/water (4:1) solution containing an internal standard mix of eight deuterated acyl carnitines at a concentration of 2 μM (see Supplementary table 2). The metabolite levels were quantified using a Vanquish UHPLC + series (Thermo Fisher Scientific) fitted with an ACE Excel-2 C18-PFP 5 μm column (100 Å, 150 × 2.1 mm; Advanced Chromatography Technologies Ltd.) conditioned at 40°C, coupled to a TSQ Quantiva Triple Quadripole Mass Spectrometer (Thermo Fisher Scientific). The mobile phase was pumped at 0.45 ml.min −1 with mobile phase A 0.1% aqueous formic acid solution and mobile phase B 0.1% formic acid/methanol solution. Mobile phase B was initially held at 0.5% for 1 min and was then subjected to a linear increase to 100% B over 9 min, held for 2 min, and then brought back to initial conditions over 0.1 min. UPLC-MS data acquisition and processing was carried out using Xcalibur software (version 2.2, Thermo Fisher Scientific) with putative recognition of the detected metabolites performed using a targeted MS/MS analysis.
The protein pellet remaining from the metabolite extraction was homogenized in 250 μl of RIPA buffer and used to spectrophotometrically quantify the protein concentration using a Pierce BCA protein assay kit (Sigma-Aldrich) as per the manufacturer's instructions. All metabolite levels were then normalized to the intensity of an appropriate internal standard and to the protein concentration of the Cebpa genes as determined by RT-qPCR analysis. N = 3-5 females per genotype. Data is presented as fold change relative to C57Bl/6J controls (normalized to 1; mean ± sd). Independent t-tests, *p < .05. Legend: Squares, males; Circles, females; black, C57Bl/6J; white, Mtrr gt/gt . sample to account for differences in initial sample size.

Hepatic glycogen concentration
Glycogen extraction from livers was carried out as previously described [48]. Glycogen was extracted from~100 mg of liver from each sample and resuspended in 1 ml of H 2 O. Glycogen content was measured with a spectrophotometer (490 nm absorbance; Elx800 microplate reader, BioTek, Winooski, VT, USA) using the phenol-sulphuric acid method [4]. A standard curve was plotted to calculate glycogen concentrations using serial dilution of rabbit liver glycogen (Sigma-Aldrich). Glycogen content was measured in duplicate with 6-8 biological replicates per genotype and sex.

Statistics, equipment and software
Statistical analyses were performed using GraphPad Prism 6 software (La Jolla, CA, USA). Differences attributable to genotype, sex, or the interaction of these two factors were assessed through a two-way analysis of variance (ANOVA) whilst differences between C57Bl/6 J and Mtrr gt/gt mice within one sex were separately analysed either through a two-tailed, unpaired Mann-Whitney test or a two-tailed, unpaired student's t-test. P < .05 was considered significant. A NanoZoomer 2.0-RS (Hamamatsu) and NDP.view2 viewing software program (Hamamatsu) were used to obtain images. Histological measurements including Ki67 + and PAS + cell counts and percentage area of PAS stain were performed using ImageJ software (NIH, Bethesda, MD, USA). All graphs were produced using GraphPad Prism 6 software. PCR primers were designed using Primer-BLAST software [49].

Misexpression of genes that encode for folate metabolic enzymes in Mtrr gt/gt female liver
The Mtrr gt allele is a knock-down mutation that leads to the disruption of one-carbon metabolism in liver including a 62% reduction in MTR activity associated with plasma hyperhomocysteinemia and tissuespecific alterations in DNA methylation [21,35]. Using RT-qPCR, we confirmed that the wildtype Mtrr transcript was significantly reduced in Mtrr gt/gt female liver to 24.1 ± 10.0% (mean ± sd) of C57Bl/6J control transcript levels (p = .0016; Fig. 1A). This result was in line with a previous report, which also showed a corresponding reduction in wildtype MTRR protein expression in Mtrr gt/gt livers [21]. Of note, Mtrr gt/gt female livers displayed elevated expression of total Mtrr transcripts (including wildtype and gene-trapped Mtrr transcripts) relative to controls (150.4 ± 37.3% of C57Bl/6J transcripts, p = .033; Fig. 1A), which was different to a previous study [21] and suggested a potential transcriptional regulatory feedback loop. Differences in total Mtrr RNA expression levels in Mtrr gt/gt livers between studies might be due to the controls used: C57Bl/6J livers [this study] versus Mtrr +/+ littermates [21]. Similarly, other genes that encode for folate metabolism enzymes in Mtrr gt/gt female livers were also up-regulated, such as Mthfr and Mtr (132.7 ± 21.0% [p = .033] and 139.9 ± 19.5% [p = .024] of C57Bl/6J transcripts, respectively; Fig. 1A). However, the expression of Bhmt and Bhmt2 genes, which provide an alternate pathway for methylation of homocysteine [50], was unchanged in Mtrr gt/gt female livers compared to controls (Fig. 1A). Altogether, these data suggest possible transcriptional compensation for Mtrr knockdown via the selective up-regulation of other genes involved in one-carbon metabolism. Whether these transcriptional changes lead to alterations in protein level or function requires confirmation.

Hepatocytes are eosinophilic in Mtrr gt/gt female mice
To explore the effects of the Mtrr gt mutation on body mass, male and female Mtrr gt/gt mice were weighed and compared to C57Bl/6J controls (N = 4-8 mice per group). Mtrr gt/gt homozygosity resulted in low body weight in male and female mice relative to controls (p < .05, Fig. 1B). Next, liver mass was assessed. While Mtrr gt/gt males showed a proportional decrease in liver and body mass (p = .343), Mtrr gt/gt female livers represented a greater percentage of body mass than controls (p = .0006; Fig. 1C-D). This result indicated that the Mtrr gt/gt mutation led to a female-specific effect on liver size. Histological analysis of female livers showed that Mtrr gt/gt hepatocytes were more densely packed than controls (Fig. 1E-F) with greater nuclear density (nuclei per 0.1 mm 2 ; p < .05) (Fig. 1I). It is possible that this was a result of increased cell proliferation since the percentage of mitotic Ki67 + cells was slightly higher in Mtrr gt/gt female livers (p = .059; Fig. 1G-H,J). However, the expression of genes that regulate cell proliferation in the liver (e.g., Myc [p = .0746] and C/ebpa [p = .0184]; Fig. 1K) was not appreciably higher (1.3-1.6-fold change) in Mtrr gt/gt female livers relative to controls. Besides being more densely arranged, Mtrr gt/gt hepatocytes were eosinophilic (Fig. 1E-F), which might indicate an aberrant accumulation of proteins [51] caused by cell stress [52] or altered glycogen or lipid content (see Sections 3.3 and 3.4). We observed that the cell stress gene Ddit3 was up-regulated by 2.2-fold in Mtrr gt/gt female livers compared to controls (p = .0271; Fig. 1K). Therefore, the Mtrr gt/gt mutation might activate stress pathways in liver.

Analysis of glycogen storage in Mtrr gt/gt livers
Next, hepatic fuel storage was investigated in Mtrr gt/gt female livers. First, periodic acid-Schiff (PAS) staining was performed on histological sections of C57Bl/6J and Mtrr gt/gt female livers to assess glycogen storage (N = 3 livers per genotype). Fewer Mtrr gt/gt hepatocytes stained positive for PAS (52.8 ± 11.2%, p < .0001) compared with controls (76.0 ± 7.0%; Fig. 2A-C), and positive cells showed less intense PAS stain in Mtrr gt/gt livers ( Fig. 2A-B,D). However, when glycogen was extracted from C57Bl/6J and Mtrr gt/gt female livers, direct biochemical quantification of liver glycogen revealed that there was no difference in hepatic glycogen concentration in Mtrr gt/gt female livers compared to controls (Fig. 2E). These results indicated a discrepancy between the histology and biochemical assay that might be explained by a degree of inter-individual variability caused by ad libitum feeding or by the fact that both the PAS stain and the phenol-sulphuric acid method are not specific for glycogen [53]. However, the discrepancy observed is unlikely to be due to major differences in glycogen content between liver lobes [54].
To better understand the effects of the Mtrr gt mutation on hepatic glycogen metabolism, expression of genes involved in glycogen synthesis (i.e., Gys2, Ugp2, Gbe1, Gyg, Gsk3a, and Gsk3b), and glycogen breakdown (i.e., Agl and G6pc) were assessed via RT-qPCR in C57Bl/6J and Mtrr gt/gt female livers. Glycogen synthesis regulator genes Ugp2 and Gsk3a were down-regulated by~2-fold in Mtrr gt/gt female livers compared to C57Bl/6J controls (p < .002; Fig. 2F, G). All other genes assessed displayed normal expression in Mtrr gt/gt female livers compared to controls (Fig. 2F-H). UGP2 is an enzyme responsible for converting glucose-1-phosphate to UDP-glucose, the precursor of glycogen. In contrast, GSK3 negatively regulates glycogen synthesis [55,56] and reduced expression of its gene in Mtrr gt/gt livers might indicate transcriptional feedback associated with reduced hepatic glycogen. Downregulation of these genes in Mtrr gt/gt female livers substantiates the histological data that showed reduced glycogen (Fig. 2B). Overall, these data revealed dysregulation of hepatic glycogen synthesis in Mtrr gt/gt females. However, to more fully understand the effect of the Mtrr gt allele on hepatic glycogen metabolism, protein levels should be assessed.

Fatty acid metabolism but not lipid storage was altered in Mtrr gt/gt livers
To further explore the effects of the Mtrr gt allele on hepatic fuel storage, lipid metabolism was investigated. Firstly, 3-hydoxyacyl-CoA dehydrogenase (HOAD) activity was measured spectrophotometrically in male and female liver to indicate whether β-oxidation of fatty acids was altered by the Mtrr gt mutation. Regardless of sex, HOAD activity was similar between C57Bl/6J control and Mtrr gt/gt livers (Fig. 3A). Interestingly, HOAD activity in Mtrr gt/gt female livers was 139.9% ± 33.3% (mean ± sd) of the activity in Mtrr gt/gt males (p = .008; Fig. 3A), yet this sexually dimorphic effect was not observed in controls (Fig. 3A). When HOAD activity was normalized to citrate synthase activity, a putative marker of mitochondrial content [57], Mtrr gt/gt female livers displayed 22.5% less activity than C57Bl/6J female livers (p = .026; Fig. 3B-C). However, no differences were apparent in males (Fig. 3BeC). Overall, these data suggested that the Mtrr gt/gt mutation in mice might cause altered hepatic fatty acid metabolism in a sexually dimorphic manner.
To determine whether reduced fatty acid metabolism in Mtrr gt/gt female livers resulted in lipid accumulation, liver lipidomic analysis was performed in C57Bl/6J and Mtrr gt/gt mice by LC-MS. Firstly, we observed that the level of total carnitines (including free carnitine) and the relative proportions of short-(C2-5), medium-(C6-12) and longchain (≥C13) acyl-carnitines of male and female Mtrr gt/gt livers were comparable to the same-sex control group (Fig. 3D-E). While further considering sex-specific differences, we observed that short-chain acylcarnitine levels were lower and long-chain acyl-carnitine levels were higher in control female livers than male livers (p < .05; two-way ANOVA, Sidak's multiple comparison; Fig. 3E). This was not apparent in Mtrr gt/gt mice (Fig. 3D), which only displayed sexual dimorphism of hepatic total carnitine, an effect that was absent in controls: Mtrr gt/gt female livers had 43.1% less total carnitine (p = .045) than Mtrr gt/gt male livers (two-way ANOVA, Tukey's multiple comparison; Fig. 3E). Taken together with the HOAD activity data (Fig. 3C), these data supported the hypothesis that fatty acid handling was differently affected in Mtrr gt/gt mice in a sexually dimorphic manner.
Next, open profiling of the hepatic lipidome was performed. Total lipid content was unchanged in Mtrr gt/gt livers compared to same-sex C57Bl/6J controls (Fig. 3F). Yet, there was a distinct sex-specific effect of total lipid content regardless of genotype (Fig. 3F) that is consistent with other studies [58,59]. For instance, C57Bl/6J female livers showed 22.0% more total lipid than C57Bl/6J males (p = .041) and similarly, Mtrr gt/gt females livers had 28.3% more total lipid content than Mtrr gt/gt males (p = .007; two-way ANOVA, Sidak's multiple comparison). Lipid species were grouped by lipid class for further analysis. Similar to total lipid content, levels of TAGs, DAGs, FFAs, cholesteryl-esters, ceramides, sphingomyelins, and phospholipids were unchanged in male and female Mtrr gt/gt livers when compared to same sex controls (Fig. 3G-H, Supplementary fig. 1A). However, there was a sex-specific difference between TAGs levels regardless of genotype whereby female livers consistently showed higher levels of specific medium-chain (48-54 carbon) and long-chain (> 54 carbon) TAGs (p < .05; two-way ANOVA with a Holm-Sidak correction for false discoveries, Sidak's multiple At least four liver regions were assessed per female. N = 3 females per genotype. Data is represented as mean ± sd. Independent t-test, ***p < .005. (E) Hepatic glycogen concentration in female C57Bl/6J and Mtrr gt/gt mice. N = 8 livers per group. Data is represented as mean ± sd. Independent ttest. (FeH) Relative mRNA expression in C57Bl/6J control and Mtrr gt/gt female livers was determined for (F) Gys2, Ugp2, Gbe1 and Gyg, (G) Gsk3a and Gsk3b and (H) Agl and G6pc as determined by RT-qPCR analysis. N = 6 livers per group. Data is presented as fold change relative to C57Bl/6J controls (normalized to 1; mean ± sd). Independent t-tests, § p = .0599, *p < .05, **p < .01, ***p < .001. Black circles, C57Bl/6J; white circles, Mtrr gt/gt . comparison; Supplementary fig. 2). The medium-chain TAGs were species rich in palmitic and oleic acid (Supplementary fig. 2A) and are associated with de novo lipogenesis [60]. This data might account for the increased total lipid content that was observed in the female livers.
In specific cases, such as in DAGs and phosphatidylinositol phosphates, there was a sex-specific effect in C57Bl/6 J livers (p < .05, two-way ANOVA, Sidak's multiple comparison) that was absent in Mtrr gt/gt livers (Fig. 3G, Supplementary fig. 1B-C). Conversely, there were several lipid species whereby a sex-specific effect in liver was generated by the Mtrr gt/gt mutation that was not apparent in controls. More specifically, Mtrr gt/gt female livers contained more TAG, phosphatidylcholine and phosphatidic acid than male Mtrr gt/gt livers (p < .05, two-way ANOVA, Sidak's multiple comparison; Fig. 3G-H,Supplementary fig. 1B). Sexual dimorphism of the hepatic lipidome suggested that male and female mice might be differently affected by the Mtrr gt mutation. Overall, these data suggested that Mtrr gt/gt homozygosity in female liver might alter fatty acid metabolism without significantly affecting in lipid content.

Mtrr gt/gt livers display normal mitochondrial electron transfer system function
Lastly, we aimed to assess whether the metabolic phenotype of Mtrr gt/gt livers was related to changes in mitochondrial respiratory function. First, the mRNA expression of a selection of nuclear-encoded genes required for mitochondrial function (e.g., Ndufs4, Ndufs5, Ndufs6, Isca1) was assessed. No significant differences in mRNA expression between male and female C57Bl/6 J and Mtrr gt/gt livers were observed (p > .114; Fig. 4A). Next, mitochondrial respiration rates were analysed in fresh C57Bl/6 J and Mtrr gt/gt liver homogenates by high-resolution respirometry (N = 4 males and N = 8 females per genotype). No difference in flux was observed in livers of male or female Mtrr gt/gt mice compared with controls, regardless of the state assayed (when corrected to wet tissue mass; Fig. 4B-C). Similarly, when rates were corrected to GMS E (maximal ETS capacity) to provide a qualitative indication of changes in mitochondrial function per mitochondrial unit, oxygen flux rates were no different to controls (Fig. 4D-E). Altogether, these data suggest that the Mtrr gt/gt mutation does not cause mitochondrial dysfunction in the liver of adult mice.

Discussion
The Mtrr gt allele in mice causes defects in one-carbon metabolism including plasma hyperhomocysteimia, decreased plasma methionine, increased tissue 5-methyl-THF, and altered cellular methylation [21,35]. Similar to folate deficiency and MTRR mutations in humans [30,[32][33][34], metabolic derangement caused by Mtrr gt/gt homozygosity in mice is associated with macrocytic anaemia [36] and a wide spectrum of developmental phenotypes (e.g., neural tube defects, cardiac defects, fetal growth restriction) that display incomplete penetrance [35]. Here, we investigated the effects of the Mtrr gt allele on liver metabolism given that NAFLD has previously been associated with defects in one-carbon metabolism [2][3][4][5][6][7][8] and to better understand the larger metabolic milieu in Mtrr gt/gt mice. We found adult Mtrr gt/gt female livers were large and displayed eosinophilic hepatocytes, the appearance of which was distinct from the lipid-filled vacuoles of hepatocytes associated with steatosis caused by dietary folate deficiency and/or Mthfr knockout mutation in mice [31,61]. Despite this, the Mtrr gt mutation leads to changes in hepatic glycogen storage and β-oxidation of fatty acids that occur independent of changes in electron transfer system capacity.
The most striking phenotype of this study was the densely packed, eosinophilic hepatocytes in Mtrr gt/gt female mice. The altered appearance of Mtrr gt/gt hepatocytes might be explained by impaired glycogen storage. Evidence for this hypothesis was reduced glycogen content in Mtrr gt/gt female livers as determined by PAS stain in combination with a down-regulation of genes involved in glycogen synthesis. Indeed, mice fed a folate-and/or choline-deficient diet exhibit alterations in hepatic DNA methylation of genomic regions that are associated with genes related to carbohydrate metabolism [62]. In another context, folate transporters (e.g., FOLR1) and metabolic enzymes (i.e., MTR and MTHFR) are highly expressed in specialized placental cells involved in glycogen storage and metabolism [63]. These observations support the hypothesis that glycogen metabolism and folate metabolism might be linked.
In conflict with the data above, glycogen concentration was unchanged in Mtrr gt/gt livers compared to controls when biochemically determined. While different liver lobes were assessed in the PAS stain and the phenol-sulphuric acid method, glycogen content is typically consistent between lobes [54] and is unlikely the cause of this discrepancy. Given that glycogen is a multi-branched polysaccharide of glucose, it is possible that the structure of glycogen is altered (e.g., branch lengths) in Mtrr gt/gt livers. As a result, PAS stain and phenolsulphuric acid method might differently recognize the glycogen pool based on these hypothetical structural changes. Alternatively, PAS stain also detects glycoproteins and proteoglycans in addition to glycogen [64]. Therefore, changes in PAS staining intensity in Mtrr gt/gt female livers might instead reflect major cellular changes, such as in connective tissue and basal laminae. More sensitive assays, such as 13 C nuclear magnetic resonance [65] or amyloglucosidase digestion followed by liquid chromatography/electrospray ionization-tandem mass spectrometry quantification of the resultant free glucose [66] might reveal more accurate changes in Mtrr gt/gt liver glycogen content. Additionally, Mtrr gt/gt mice that are challenged by an obesogenic diet might show more striking and consistent defects in glycogen synthesis and/or metabolism relative to controls. This type of increased sensitivity to dietary stress should be explored in the future.
Independent of glycogen storage, eosinophila of Mtrr gt/gt hepatocytes might be explained by aggregation of non-specific cytoplasmic proteins [51]. Endoplasmic reticulum (ER) stress, a condition where unfolded and misfolded proteins accumulate, is a feature of fibroblasts of human patients with mutations in MTR [67]. These cells display mislocalization of RNA binding proteins that prevent nucleo-cytoplasmic shuttling of RNA and splicing of mRNA [67], which likely has a Fig. 4. Mitochondrial function is unchanged in livers of Mtrr gt/gt adult mice. (A) Relative mRNA expression of genes encoding for nuclear-encoded mitochondrial genes in male (m) and female (f) livers from C57Bl/6J control and Mtrr gt/gt adult mice as determined by RT-qPCR analysis. N = 3-5 females per genotype. Data is presented as fold change relative to C57Bl/6J controls (normalized to 1; mean ± sd). (B-E) Mitochondrial respiratory function (JO 2 ) in male and female livers from C57Bl/6J and Mtrr gt/gt adult mice. GM L , malate and glutamate stimulate LEAK respiration through the N-pathway via complex I; GM P , OXPHOS supported by glutamate and malate through the N-pathway via complex I; GMS P , OXPHOS supported by glutamate, malate and succinate through the NS-pathway via complexes I and II; GMS E , maximal ET capacity supported by malate, glutamate and succinate through the NS-pathway via complexes I and II in uncoupled mitochondria. Oxygen flux rates were normalized to (B-C) wet tissue mass or to (D-E) GMS E to indicate mitochondrial respiratory capacity per mitochondrial unit. Squares, males; circles, females; black, C57Bl/6J; white, Mtrr gt/gt . profound effect on post-transcriptional regulation and normal cell function. Hyperhomocysteinemia, which is a prominent feature in Mtrr gt/gt mice [21,35], associates with ER stress in other contexts [68,69]. While we did not test specifically for protein aggregation in eosinophilic Mtrr gt/gt female livers, an up-regulation of the Ddit3 gene implicates ER stress. Whether this is a primary or secondary effect of the Mtrr gt allele is unclear. Further experimentation is required to more fully understand how the Mtrr gt/gt mutation causes eosinophilic hepatocytes and its relationship to cell stress.
Given that homocysteine aids TAG accumulation [70,71], one reason why perturbation of one-carbon metabolism alters lipid content and hepatosteatosis is the presence of hyperhomocysteinemia. Others have shown via reduced representation bisulfite sequencing analysis that Mtrr gt/gt livers have patterns of differential methylation of DNA compared to wildtype littermates [72]. Some of the affected genomic regions included genes associated with lipid metabolism [72], though the functional relevance of this differential methylation is yet-to-be determined. Despite increased plasma total homocysteine and altered liver DNA methylation in Mtrr gt/gt mice, the only detectable effect of Mtrr gt/gt mutation on lipid metabolism was reduced β-oxidation of fatty acids in female livers. While hepatic fatty acid oxidation was also impaired in rats fed a methyl-donor deficient diet, these rats also displayed carnitine deficit and liver steatosis [73], features that were absent in Mtrr gt/gt livers. In fact, lipid content and metabolism was within the normal range in Mtrr gt/gt female and male livers.
NAFLD and glycogen storage disease are often accompanied by alterations in mitochondrial OXPHOS and/or ETS function [37][38][39][40]. Furthermore, folate metabolism is essential for the methylation of mitochondrial tRNA required for successful mRNA translation in the mitochondria [74] to ensure respiratory complexes for OXPHOS are functioning [75]. No difference in JO 2 in any of the states assayed was observed in Mtrr gt/gt liver. This result is consistent with only subtle changes in lipid and glycogen metabolism and demonstrates that hepatic mitochondrial function was unaltered by the Mtrr gt/gt mutation.
It is unclear why folate deficiency and not the Mtrr gt/gt mutation cause steatosis of the liver. The Mtrr gt allele in mice is robust enough to cause many of the same phenotypes as humans with MTRR mutations or folate deficiency (e.g., hyperhomocysteinemia, anaemia, neural tube defects, fetal growth restriction) [21,30,[32][33][34][35][36]. The macrocytic anaemia phenotype in Mtrr gt/gt female mice is late-onset at 22 weeks of age [36]. It is possible that liver disease also appears later in life in Mtrr gt/gt mice and was not apparent in our study since 9-16 week-old mice were assessed. Alternatively, the Mtrr gt allele is a knockdown rather than knockout mutation, with 24% of wildtype Mtrr transcripts remaining in Mtrr gt/gt female livers. This level of expression might be sufficient for lipid metabolism to proceed and a conditional knockout of Mtrr expression specific to the liver might be required to induce more severe effects on lipid metabolism and steatosis. More severe metabolic liver phenotypes may also be uncovered through exposing the Mtrr gt/gt mouse to a high fat diet (e.g. [31]), a NASH-inducing diet containing trans-fats, (e.g [76]), or a folate-deficient diet (e.g. [77]).