Neil3-dependent base excision repair regulates lipid metabolism and prevents atherosclerosis in Apoe-deficient mice

Increasing evidence suggests that oxidative DNA damage accumulates in atherosclerosis. Recently, we showed that a genetic variant in the human DNA repair enzyme NEIL3 was associated with increased risk of myocardial infarction. Here, we explored the role of Neil3/NEIL3 in atherogenesis by both clinical and experimental approaches. Human carotid plaques revealed increased NEIL3 mRNA expression which significantly correlated with mRNA levels of the macrophage marker CD68. Apoe−/−Neil3−/− mice on high-fat diet showed accelerated plaque formation as compared to Apoe−/− mice, reflecting an atherogenic lipid profile, increased hepatic triglyceride levels and attenuated macrophage cholesterol efflux capacity. Apoe−/−Neil3−/− mice showed marked alterations in several pathways affecting hepatic lipid metabolism, but no genotypic alterations in genome integrity or genome-wide accumulation of oxidative DNA damage. These results suggest a novel role for the DNA glycosylase Neil3 in atherogenesis in balancing lipid metabolism and macrophage function, potentially independently of genome-wide canonical base excision repair of oxidative DNA damage.

BiKE gene expression microarray. RNA was hybridized to Affymetrix HG-U133 plus 2.0 A Genechip® arrays according to standard protocol (http://www.affymetrix.com) and mRNA levels were measured and analyzed at the Karolinska Institute Affymetrix microarray core facility as previously described 2,7 . Microarray data were pre-processed, normalized and log2-transformed using the RMA algorithm 8 . The data set is available from Gene Expression Omnibus, accession number GSE21545. The cells were differentiated into macrophages by incubation with phorbol myristate acetate (PMA, 100 nM, Sigma Aldrich) for 24 hours. Thereafter, the cells were stimulated with endotoxin-free VLDL (25 μg/ml; Calbiochem, Darmstadt, Germany) and cholesterol crystals (100 μg/ml), prepared as previously described 9 , for 6 hours before harvesting of cell pellets.
Murine tissue collection and blood sampling. Mice were anaesthetized under non-fasting conditions with intraperitoneal injection of 0.1 ml per 10 g body weight Hypnorm/Dormicum (1.25 mg/ml Dormicum, 2.5 mg/ml Fluanisone, 0.079 mg/ml Fentanyl citrate) and killed by exsanguination. Blood was collected at sacrifice by trans-thoracic cardioscentesis using a 1ml syringe without (serum) or with (plasma) coating of 0.5 M EDTA (Fluka, Sigma-Aldrich). Serum was obtained after coagulation at room temperature for 30 minutes before centrifugation at 1000g for 15 minutes. EDTA blood was immediately placed on ice and centrifuged within 30 minutes at 2000g (4 °C) for 20 minutes to obtain platelet-poor plasma. Serum and plasma were aliquoted and kept at -80°C until use. For collection of tissue specimens, the vasculature was perfused through the left ventricle with 2 ml of sterile PBS at a rate of approximately 1 ml/minute. The cranial half of the heart was placed in OCT compound (Tissue-Tek, Sakura Finetek, Torrance, CA) and frozen for subsequent cryo-sectioning. Thoracic aortas were either placed on 4% paraformaldehyde (PFA) in PBS for en face quantification of atherosclerosis, or snap frozen together with the other organs for subsequent analysis. For plaque-analyses, the proximal brachiocephalic artery, containing large occluding plaques, was excised and snap frozen.
Histological and morphometric analyses of the aorta and aortic root. The thoracic aorta was fixed in 4% PFA in PBS and dissected under an Olympus SZX12 microscope. All adventitial fat tissue was carefully resected. The aortic arch was longitudinally cut open, pinned on a black math and stained with Sudan IV (Sigma-Aldrich). En face plaque area was quantified (plaque surface area/total surface area of aortic arch  100) by using identical denominator (proximal 16.5 mm 2 area of the aortic arch) in all mice (excluding the branching vessels). Frozen hearts in OCT compound (Tissue-Tek, Sakura Finetek, Torrance, CA) were serially sectioned towards the base of the heart on a cryostat. 10 μm frozen sections were collected in cranial direction, starting at 100 μm distance after appearance of the aortic cusps. Sections were air-dried and fixated with either 4% PFA or cold acetone. PFA-fixed sections were stained with Oil Red O (Sigma-Aldrich) and hematoxylin (Vector Laboratories Inc., Burlingame, CA) and relative lesion areas (plaque area/vessel area inside external elastic lamina  100) were calculated in 8 consecutive sections at 100 μm intervals. The mean relative lesion area was calculated from 8 sections from each mouse. This method is reported to reduce the chance of miscalculation of lesion areas due to oblique sections 11 . Necrotic core was assessed in sections from the aortic root stained with hematoxylin (Vector Laboratories Inc.) and eosin (Histolab, Gothenburg, Sweden) and defined as acellular areas with or without cholesterol crystals. Necrotic core was quantified twice and mean values were used for analyses. Total collagen content, thin vulnerable fibers (green), and mature collagen (red), as assessed by the fraction: "collagen area/plaque area within internal elastic lamina  100", was stained with Picrosirius Red (Histolab). Images were captured using both bright field and polarized microscopy. Pixels of a specific red hue in bright field images were interpreted as collagen and counted with Fiji, an ImageJ-derived imaging software. Green and red fibers were isolated by using their colorimetric footprint in the polarized light microscopy images  Murine plasma and liver FA composition. Lipids were extracted using a mixture of chloroform and methanol. The extracts were trans-esterified using BF 3 (boron trifluride)-methanol. To remove neutral sterols and non-saponifiable material, extracts of fatty acyl methyl esters were heated in 0.5 M KOH in an ethanol-water solution (9:1). Recovered FAs were re-esterified using BF 3 -methanol. The methyl esters were quantified by gas chromatography as previously described 14 .
Murine lipoprotein subfractions. Lipoprotein subfractions in plasma were measured by a linear polyacrylamide gel electrophoresis system. Lipoprint HDL kit and Lipoprint LDL kit (Quantimetrix Corporation, Redondo Beach, CA) were prepared according to manufacturer's protocol. The gels were scanned and analyzed using Lipoware software (Quantimetrix).
Murine plasma carnitine derivatives. Carnitine derivatives were measured in plasma using LC-MS/MS as previously described 15 , with some modifications of the HPLC conditions 16,17 . and Cpt-2 were performed according to Bremer et al. 18 , with some modifications 19 . The activities of Fas and Gpat were measured in the post-nuclear fraction, as described previously 20 , with some modifications 19 . Acc was measured in the post-nuclear fraction as described by Skorve et al. 20 . Immunophenotyping of leukocytes. Immunophenotypic screening of blood, spleen, lymph nodes and thymus was performed by flow cytometry. Absolute leukocyte counts in blood were obtained using BD Trucount tubes, according to manufacturer's protocol (BD). Briefly, Fcreceptor blocking antibody (anti Cd16/Cd32; eBioscience) was added to 50 μl anti-coagulated whole blood in Trucount tubes. After incubation at room temperature (10 minutes) the Trucount antibody cocktail (Supplemental Table 9) was added and the antibody/blood mix incubated for another 20 minutes at room temperature. Finally, samples were analyzed after incubation with hypotonic lysis buffer. Myeloid progenitor densities and mature leukocytes were measured in spleen, and T-cell subsets were measured in blood, spleen, pooled lymph nodes, and thymus.
Single cell suspensions of blood and spleen were prepared as described 25 . Lymph nodes and thymi were crushed over a 70 μm cell strainer (BD) and the strainers were flushed with ice-cold PBS. One femur and tibia per mouse were flushed with ice-cold PBS and single cell suspensions were prepared using a 70 μm cell strainer (BD). After blocking non-specific Fc-receptor binding (anti-Cd16/Cd32 antibody; eBioscience), complete cell pools were stained with the DIF or T-cell antibody mix (Supplemental Table 9) and incubated for 30 minutes. Cells were washed once and subsequently resuspended in 200 μl buffer (1xPBS, 5% BSA, 1 mM EDTA). All samples and buffers were kept on ice throughout the experiment unless indicated otherwise. All measurements were performed on a FACS Canto II (BD) and analysis of acquired data was performed using FACS Diva software (BD) and FlowJo 7.6.5 (Tree Star, Ashland, Oregon). In RNA sequencing analyses, differentially expressed genes (DEGs) were calculated based on RPKM values, assuming a Poisson distribution 26 . DEGs were defined as having a log2 ratio≥1 and a false discovery rate (FDR) <0.001. A KEGG pathway enrichment analysis of DEGs was performed using a hypergeometric distribution test model and pathways were ranked according to their level of significance. In figure 4, enrichment is reported as Q values, which is the FDR corrected p value of the enrichment analysis. A Q value <0.05 was considered as significant.

Measurement of body composition with DEXA Lunar PIXImus
Gene list enrichment based on functional annotations and protein interactions network on differentially expressed gene signatures was done using the GATHER software package (http://gather.genome.duke.edu) 27 and Toppfun (ToppGene Suite; https://toppgene.cchmc.org) 28 .
Transcription factor enrichment was based on experimentally proven binding sites and consensus binding sequences for eukaryotic transcription factors (positional weight matrices) and known transcription factor regulated genes (as listed in TRANSFAC v7.0; GATHER). All output was FDR corrected (P<0.05), while statistical significance was calculated based on the probability of seeing a Bayes factor or gene set enrichment of a particular magnitude in a given query.    Data are presented as median and interquartile range and were analyzed using the Mann-Whitney U test (n=7-12). Immunological genome project