Genetic polymorphisms in vitamin E transport genes as determinants for risk of equine neuroaxonal dystrophy

Abstract Background Equine neuroaxonal dystrophy/equine degenerative myeloencephalopathy (eNAD/EDM) is an inherited neurodegenerative disorder associated with vitamin E deficiency. In humans, polymorphisms in genes involved in vitamin E uptake and distribution determines individual vitamin E requirements. Hypothesis/Objectives Genetic polymorphisms in genes involved in vitamin E metabolism would be associated with an increased risk of eNAD/EDM in Quarter Horses (QHs). Animals Whole‐genome sequencing: eNAD/EDM affected (n = 9, postmortem [PM]‐confirmed) and control (n = 32) QHs. Validation: eNAD/EDM affected (n = 39, 23‐PM confirmed) and control (n = 68, 7‐PM confirmed) QHs. Allele frequency (AF): Publicly available data from 504 horses across 47 breeds. Methods Retrospective, case control study. Whole‐genome sequencing was performed and genetic variants identified within 28 vitamin E candidate genes. These variants were subsequently genotyped in the validation cohort. Results Thirty‐nine confirmed variants in 15 vitamin E candidate genes were significantly associated with eNAD/EDM (P < .01). In the validation cohort, 2 intronic CD36 variants (chr4:726485 and chr4:731082) were significantly associated with eNAD/EDM in clinical (P = 2.78 × 10−4 and P = 4 × 10−4, respectively) and PM‐confirmed cases (P = 6.32 × 10−6 and 1.04 × 10−5, respectively). Despite the significant association, variant AFs were low in the postmortem‐confirmed eNAD/EDM cases (0.22‐0.26). In publicly available equine genomes, AFs ranged from 0.06 to 0.1. Conclusions and Clinical Importance Many PM‐confirmed cases of eNAD/EDM were wild‐type for the 2 intronic CD36 SNPs, suggesting either a false positive association or genetic heterogeneity of eNAD/EDM within the QH breed.


| INTRODUCTION
Equine neuroaxonal dystrophy/equine degenerative myeloencephalopathy (eNAD/EDM) is an inherited neurodegenerative disease associated with vitamin E deficiency during the first year of life. 1,23][4] Even after 1 year of age, serum α-tocopherol concentrations typically remain lower in eNAD/ EDM horse as compared to age-matched healthy controls. 5Equine NAD/EDM can be prevented in genetically susceptible foals by supplementing dams with high doses of water-soluble RRR-α-tocopherol during the last trimester of gestation, with continued supplementation in these foals through the first 2 years of life. 3Currently, the only way to conclusively diagnose eNAD/EDM is through postmortem histologic evaluation of the brainstem and spinal cord at necropsy. 2 While a recently developed biomarker test for phosphorylated neurofilament heavy chain (pNfH) has demonstrated some specificity for a diagnosis of eNAD/EDM in serum, the overall sensitivity is low. 6 humans, ataxia with vitamin E deficiency (AVED), an inherited disease caused by deleterious variants in tocopherol transfer protein (alpha; TTPA), shares clinicopathologic features with eNAD/EDM. 7,8P(A) is the major protein involved in transferring α-tocopherol into liver secreted plasma lipoproteins. 9Equine NAD/EDM is not associated with genetic mutations in TTPA. 10 However, polymorphisms in another gene involved in vitamin E uptake, distribution and metabolism could potentially modulate the risk of a horse developing eNAD/ EDM.Vitamin E absorption 11 and transport to target tissues 3,12 are not altered with eNAD/EDM.However, vitamin E metabolism is increased in Quarter Horses (QHs) with eNAD/EDM. 13Additionally, there is greater hepatic expression of CYP4F2, the major metabolizer of vitamin E, in eNAD/EDM horses compared to unaffected horses. 13us, while we elected to profile all known vitamin E candidate genes, we hypothesized that genetic polymorphisms in genes involved in vitamin E metabolism, specifically CYP4F2, would be associated with increased risk of eNAD/EDM.

| Animals-whole-genome sequencing
All animal procedures were approved by the University of California-Davis Institutional Animal Care and Use Committees (protocols #22427 and 22477).DNA samples for molecular work were available from a biorepository of eNAD/EDM affected and control horses, with written owner consent obtained for all postmortem sample collections of clientowned horses.Whole-genome sequencing was performed on n = 9 eNAD/EDM affected and n = 32 control QHs All horses for whole-genome sequencing were QHs and QH-related breeds (Paint, Appaloosa) and eNAD/EDM affected horses were postmortemconfirmed, as previously described. 2,14,15All horses underwent neurologic evaluation using the modified Mayhew scale 16 prior to study enrollment.Control horses for whole-genome sequencing were QHs, Paints and Appaloosa horses, maintained as part of the UC Davis Teaching and Research herd at the Center for Equine Health, with ataxia scores of 0/5 and no known history of neurologic signs.

| Whole-genome sequencing
Genomic DNA from the 41 horses were sequenced on the Illumina HiSeq4000 at approximately 30Â coverage.Whole genome sequences were deposited in the NCBI Sequence Read Archive (https://ncbi.nlm.nih.gov/subs/sra/;SUB13501748).Fastq files trimmed for quality and reads mapped to the EquCab3.0 equine reference sequence 17 using BWA mapping program. 18Mapping quality was assessed using Samtools. 19SNP, INDEL discovery and genotyping across all samples was performed using GATK HaplotypeCaller. 20,21To detect larger structural variants, DELLY 22 was run on each sample and output files were merged.

| Candidate gene evaluation
SNPSift 23 was first used to filter the resulting vcf file created by GATK by quality, using a variant Phred threshold of 30 (Q ≥ 30).For both GATK and DELLY vcf files, case/control status was assigned using SNPSift CaseControl. 23Candidate genes involved in vitamin E absorption, transport and metabolism in humans were identified. 24PSift 23 was then used to filter whole-genome vcf files into candidate gene regions using EquCab3.0 coordinates for each of the 28 vitamin E candidate genes (Table 1).We included 1 kb up-and down-stream from the annotated start and stop codons using the Ensembl annotation for EquCab3.0 (http://m.ensembl.org/Equus_caballus/Info/Annotation). Lastly, candidate genetic variants were filtered based on an allelic P value of <.01 (CC_ALL), as determined by a Fisher's Exact Test for alleles, and annotated using SNPEff. 25In addition to the variant callers used, raw bam files were visually inspected using Integrative Genome Viewer 26 in the candidate gene regions for any additional structural variants that may have been missed with DELLY, including duplications, inversions and large deletions or insertions.Candidate genetic variants identified via GATK were validated in the Integrative Genome Viewer 26 before the validation study.

| Animals-validation study
To validate putative genetic associations in vitamin E candidate genes, a validation study was performed with 39 eNAD/EDM affected QHs, of which 23 were postmortem confirmed, and 68 control QHs, of which 7 were postmortem confirmed.The additional n = 16 eNAD/ EDM affected QHs were phenotyped as part of a previous study from a single farm, with ataxia scores ≥2.

| MassARRAY genotyping
Thirty putative genetic variants in vitamin E candidate genes were genotyped using the MassARRAY platform (Agena Bioscience, San Diego, California, USA) through Neogen Corporation (Lincoln, Nebraska, USA).
Twenty-two variants were filtered out via quality control analysis in plink 27 ; 1 failed genotyping and the other 21 had minor allele frequencies below 5%.Case-control allelic association testing using plink 27 was T A B L E 1 Candidate genes involved in vitamin E absorption, transport, and metabolism in humans. 24hen performed on the remaining 9 putative genetic variants.A Bonferroni correction was applied to account for multiple testing (P Bonf = .006).

Gene
Two analyses were performed: (a) postmortem affected cases only vs controls and (b) all cases (clinical and postmortem) vs controls.
Significantly associated SNPs were evaluated using Ensembl (https:// useast.ensembl.org/Equus_caballus/Info/Index),with Genomic Evolutionary Rate Profiling (GERP) scores 28 and minor allele frequencies (MAF) reported.A GERP score greater than 2 was considered to have higher evolutionary constraint and therefore a prioritized putative functional variant. 28

| Allele frequency-public database
Allele frequencies for putative genetic variants were obtained from publicly available data from 504 horses across 47 breeds. 29 3 | RESULTS

| Whole genome association and validation study
From the whole-genome sequencing data of 9 postmortem-confirmed affected QHs and 32 control QHs, 43 variants were identified in 15 vitamin E candidate genes that were significantly associated with eNAD/ EDM (P < .01;Table S1).Of these, 13 were determined to be incorrectly classified as insertions and deletions because of repeats using IGV 26 (Table S1, red).The remaining 30 variants were genotyped in the validation cohort of horses (n = 39 eNAD/EDM and n = 68 control).
For chr4:726485 (rs1140908279), the SNP is predicted to also be intronic in 6 annotated CD36 transcripts (Table S2) and has a GERP score of 1.00.Minor allele frequency scores were not available in Ensembl for either of these SNPs.
In the clinical eNAD/EDM case analysis only (ie, not solely postmortem confirmed cases), a third SNP at chr26:39083023 (rs1140829829), which is intronic in ABCG1, was also significantly associated with eNAD/ EDM (P = .003;Table 3).This SNP is predicted to be intronic in 4 annotated ABCG1 transcripts (Table S2), has a GERP score of À1.27 and the highest population MAF in Ensembl is 0.83.

| DISCUSSION
In QHs, a postnatal vitamin E deficiency and genetic predisposition are both required to develop eNAD/EDM. 2,3Abnormalities in vitamin E metabolism have been identified in QHs with eNAD/EDM, defined primarily by increased α-tocopherol metabolism in both serum and urine, after administration of an oral dose of RRR-α-tocopherol. 13In that same study, there was greater hepatic expression of CYP4F2, the major metabolizer of vitamin E, in eNAD/EDM horses compared to unaffected horses. 13We therefore profiled genetic variants in genes associated with vitamin E uptake, transport and metabolism, including CYP4F2, in this current study.Polymorphisms in these vitamin E genes are associated with varying vitamin E levels in humans. 24While we did not identify any associated genetic variants in CYP4F2, 2 intronic CD36 SNPs were significantly associated with eNAD/EDM in QHs in both the clinical and postmortem-confirmed cohorts.A third SNP intronic in ABCG1 on chr26 was significantly associated with eNAD/EDM in the clinical cohort only.
Cluster determinant 36 (CD36) plays a key role in fatty acid uptake in many tissues.This transporter is a member of the scavenger receptor family and involved in uptake of oxidized low-density lipoproteins from the bloodstream.Alpha-tocopherol downregulates transcription of CD36, both in vitro 30 and in vivo. 31,32In humans, a single intronic CD36 SNP (rs1527479) is associated with plasma α-tocopherol concentrations, and it is suggested that CD36 is involved in either intestinal absorption or tissue uptake of vitamin E. 33 Thus, multiple lines of evidence support that CD36 participates, either directly or indirectly, in vitamin E uptake.Since intestinal absorption of eNAD horses is comparable to unaffected horses 11 and tissue uptake into liver and CNS is also comparable after vitamin E supplementation, 3,12 variants in CD36 are unlikely be implicated in eNAD.Additionally, the 2 eNAD/EDM associated SNPs (chr4:731082 and chr4:726485) are intronic variants in CD36, with GERP scores less than 2.These GERP scores do not indicate high levels of evolutionary conservation.However, there are limitations to using this approach to identify deleterious mutations, particularly in noncoding sites, since regulatory elements are not typically highly sequence conserved. 28ile we identified a strong association between eNAD/EDM and these 2 intronic CD36 SNPs, many postmortem-confirmed cases did not have the alternate allele and allele frequencies were quite high for these variants in other breeds.In particular, Thoroughbreds are less likely to have eNAD/EDM 34  A limitation of the current study is the biased preselection of candidate genes.While we attempted to include as many documented vitamin E candidate genes as possible, genes that warrant investigation may have been missed.Our selection of candidate genes was based on the most comprehensive review from the human literature. 24Additionally, although we used both an algorithm and visual inspection of reads to identify larger structural variants, errors in the reference assembly could result in missing larger genomic rearrangements within these candidate genes that could be associated with disease.Long-range sequencing efforts are required to appropriately assemble genomes from affected and unaffected horses to identify these potential rearrangements.Despite the failure of this study to demonstrate a strong association for eNAD/EDM with any candidate variants, a genetic etiology for eNAD/EDM remains strongly supported by clinical cases in families 1,36,37 and large-scale breeding farm investigations. 2,10,14,38,39 conclusion, 2 intronic CD36 SNPs were significantly associated with eNAD/EDM in QHs.However, since many PM-confirmed cases of eNAD/EDM were wild-type for these variants, we either identified a false positive association or genetic heterogeneity exists for eNAD/ EDM within the QH breed.

CONFLICT OF INTEREST DECLARATION
Authors declare no conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION
Approved by the University of California-Davis IACUC (protocols #22427 and 22477).

HUMAN ETHICS APPROVAL DECLARATION
Authors declare human ethics approval was not needed for this study.

2
Within these 16 affected horses, n = 9 were half-siblings by 1 stallion, n = 3 were half-siblings by a second stallion, n = 3 were half-siblings by a third stallion and 1 was sired by a unique fourth stallion.Control horses were either part of the previous study, with ataxia scores of 0 (n = 21), 2 maintained as part of the UC Davis Teaching and Research herd at the Center for Equine Health, with ataxia scores of 0/5 and no known history of neurologic signs (n = 39) or client-owned with ataxia scores of 0 (n = 8), with a total of n = 7 confirmed as unaffected via postmortem examination.

F I G U R E 1
Within-breed allele frequencies for the 3 top putative genetic variants in vitamin E candidate genes (A: chr4:731082, B: chr4:726485, and C: chr26:39083023) for equine neuroaxonal dystrophy/equine degenerative myeloencephalopathy in Quarter Horses.
This project was supported, in part, by the Center for Equine Health with funds provided by the State of California pari-mutuel fund and contributions by private donors.Support for this work was provided by the National Institutes of Health (NIH) to Carrie J. Finno (L40 TR001136).The authors acknowledge the large animal internal medicine residents, veterinary students and staff at the School of Veterinary Medicine that assisted in recruiting cases and controls for this project.
and allele frequencies for these