- resequencing reveal the association of a deletion in the complex interferon alpha gene cluster with hypothyroidism in dogs. - Genetic and environmental factors are both known to play important roles in the disease development. - In this study, we sought to identify the genetic risk factors potentially involved in the susceptibility to the disease in the high-risk Giant Schnauzer dog breed.. - The deletion is located between two predicted Interferon alpha ( IFNA ) genes and it may eliminate functional elements potentially involved in the transcriptional regulation of these genes. - Nonetheless, the extreme genomic complexity of the associated region on CFA11 warrants further long-read sequencing and annotation efforts in order to ascribe functions to the identified deletion and to characterize the canine IFNA gene cluster in more detail.. - The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. - Full list of author information is available at the end of the article. - One of the most common endocrinopathies affecting both dogs and humans is hypothyroidism [7, 8]. - Determining the genetic etiology of hypothyroidism is of major interest due to the high prevalence and the im- pact of the disease in both humans and dogs. - The identification of this locus implicates a noteworthy link with the human counterpart of the disease, thus confirming the validity of employing the domestic dog as a disease animal model.. - A thorough examination of the phenotypic data revealed that five of the outlying samples had a dif- ferent coat colour compared with the rest of the dogs, resulting in a separate cluster. - Out of the ~ 170,000 single nucleotide polymorphisms (SNPs) genotyped, 112,683 passed the marker-based QC. - The minor allele frequency (MAF) of the top SNP was 0.12 across all samples, whereas 0.05 in cases and 0.26 in controls. - The odds ra- tio (OR) of the top SNP suggests that the associated locus is protective in this breed. - 1), based on pairwise linkage disequilibrium (LD) estimates (r 2 ≥ 0.8) of the top SNP to the rest of the SNPs on CFA11. - Conditional analysis confirmed the in- dependence of the association signal (Fig. - More than 92% (SD = 0.4) of the reads aligned to the dog reference genome in both groups of resequenced samples (Table S1).. - In the HC samples we identified 18,470 SNPs in the ex- tended associated interval bp) on CFA11. - In the first pruning step we removed 5264 vari- ants with identical genotypes between the two HC cases and the HC control. - The selection of SNPs with functional potential (n = 740) was genotyped using Sequenom MassARRAY in 96 dogs out of the initial GWA analysis cohort of 107 individuals, leaving out 11 individuals due to poor quality or lack of DNA specimen. - 0.001) that likely represented false variant calls in the HC indi- viduals. - The remaining 638 SNPs were subsequently combined with the SNPchip variants covering the ex- tended region of association, while discarding one of the duplicated control SNPs (i.e. - Out of the total 7293 imputed genotypes, 4876 were retained for further analyses (67%) after the appli- cation of the imputation likelihood-based filters.. - The statistical significance of the test largely exceeded the empirical CI 95 levels, and was additionally suggestive towards the empirical genome-wide threshold calculated after 1000 permutations (p-value Fig. - The MAF of the fine-mapping top SNP was 0.09 in the whole sample set, with 0.02 and 0.22 in cases and controls, respectively. - This conserved element overlaps with nei- ther a protein-coding gene sequence nor any predicted regulatory element, but it is located approximately 121 kbp downstream of the ELAVL2 gene and within a pre- dicted long non-coding RNA according to the Broad Im- proved Canine Annotation v1 [27]. - The new fine- mapped candidate locus with a protective effect against disease development was defined based on pairwise LD estimates (r 2 ≥ 0.8) of the fine-mapping top SNP to SNPs on CFA11. - After applying the stringent filtering criteria described in the Methods (section “Copy number variation (CNV). - The coverage of the two groups statistically significantly devi- ated in a window with coordinates CFA . - Figure 2d shows the fold coverage dif- ferences between the LC case and control groups based on M-values (see Methods, section “Copy number vari- ation (CNV) analysis”) in the CNV-overlapping win- dows. - It is worth noting that, according to M-values, the CNV might start upstream of the predicted Del3. - The predicted Del3 is present in two copies only in the HC control, which is homozygous for the protective haplo- type previously defined through GWA analysis and shows zero coverage in the region of the predicted deletion. - The putative CNV maps to the type I Interferon (IFN) gene family cluster and, ac- cording to the canine RefSeq annotation [28], overlaps with the potential promoter, 5 ′ UTR and first protein coding co- dons of the IFNA7 gene.. - The deletion overlapping two predicted interferon alpha genes emerges as a plausible functional candidate Considering the indications that the predicted Del3 may start further upstream of the estimated start, we subse- quently defined the deletion coordinates by aligning the HC case and control sequences to the wolf genome [30].. - By combining long-range PCR and Oxford Nanopore MinION sequencing of the PCR products from an individual heterozygous for the deletion, we de- termined the exact size of the deletion to 8875 bp (CFA including three gen- omic gaps in the CanFam3.1 genome assembly (Fig. - However, the fusion IFNA is miss- ing its potential regulatory upstream elements and instead gains the putative regulatory elements from the unannotated IFNA gene located at the beginning of the deletion. - Screening for the presence of the deletion in the Giant Schnauzer study cohort identified a statistically signifi- cant enrichment of the deletion allele in the control. - The discrepancies in allelic loads and the corresponding power of associa- tions shown in Table S5 might be due to technical dis- similarities during genotyping experiments caused by the extreme complexity of the target region’s genomic land- scape, the variable number of imputed genotypes charac- terizing the examined variants and their final corresponding accuracy, as well as to differences in the models used for the individual statistical analyses, re- spectively. - As ex- pected, none of the wolves and the majority of the add- itional dog breeds did not show the deletion. - Unfortunately, we did not have any information on thyroid status in these dogs and can thereby only postulate the potential protective effect of the variant in the breed where hypothyroidism does occur [32].. - 0.8) in the fine-mapped region colored according to their r 2 (darker colors indicating higher LD) in respect to the fine-mapping top SNP, location of genomic gaps, and location of RefSeq protein coding genes. - c Plot showing the – log10( p -value) of the difference in coverage between the LC case and control groups in the fine-mapped region. - The green boxes represent the three CNVs (Del1, Del2 and Del3) predicted by CNVnator in the HC individuals. - P -value: p -value of the mean normalized read depth value difference from genomic average. - The small sample size of our GWA analysis is likely to have hampered the statistical power of our study, as reflected by the magni- tude of the association. - The prediction of the deletion location is based on the alignments of the same HC individuals (NoDel/NoDel and Del/Del) to the wolf genome, as shown in the bottom panel. - Furthermore, based on the fixation of the hypothyroidism (non-protective) allele in the wolf population studied by Axelsson and colleagues, we pos- tulate that this allele (allele C) represents the ancestral allele (http://genome.ucsc.edu, public track hub: Broad Improved Canine Annotation v1, track: Axelsson SNPs) [27, 36]. - This could either reflect recombination events between this risk tagging variant and the actual causative allele in the Giant Schnauzer dogs, or the ab- sence of the causative risk allele in this breed.. - The MAF of the tagging risk variant did not sig- nificantly change in either cases or in controls, suggesting that this locus does not contribute to hypothyroidism sus- ceptibility in the Giant Schnauzer, thus implicating alter- native disease risk loci yet to be discovered. - A comparison of canine genome assembly (CanFam3.1), the allele without the deletion (NoDel) and with deletion (Del). - The grey dashed lines between NoDel and Del alleles indicate high sequence similarities in the deletion 5 ’ and 3 ′ breakpoints. - The arrows indicate the transcriptional direction of the genes. - The 5 ′ deletion breakpoint is located in the gap on CanFam3.1 assembly, indicated with ‘ n ’ s. - The prioritized vari- ants were breed-specific and potentially causative, thus be- ing instrumental in our fine-mapping approach, as neither the Giant Schnauzer nor any other Schnauzer breed was included in the design of the 170 K Illumina SNPChip.. - The CNV analysis detected a putative structural event (Del3, CFA segregating with the genotype of the fine-mapping top SNP (CFA in all the resequenced samples. - Moreover, it was detected in the Leonberger breed, which is prone to develop hypothyroidism. - Therefore, al- though we could not a priori exclude the possibility of the fine-mapping top SNP being the putative causative mutation, the hypothesis of the identified CNV as the causative variant would certainly be plausible, given that conditioning for its genotype eliminates the association on CFA11 due to the strong LD. - According to RefSeq annotation [28], the most plaus- ible causative variant bioinformatically identified in this study (Del3) overlaps with the region spanning both the potential promoter, 5 ′ UTR and first protein coding co- dons of the IFNA7 gene, coding for the IFNA7 cytokine.. - However, we could identify the breakpoints of the dele- tion as located in two different IFNA genes, namely an unannotated IFNA (IFNA?) and the RefSeq annotated IFNA7. - Based on the above-mentioned evi- dences and the hypothesis that type I IFNs can boost autoimmunity by altering the function of the immune system effector cells [59], the deletion characterized in this study emerges as a plausible candidate for protec- tion against canine hypothyroidism. - Due to high genomic complexity in the region, high se- quence similarity between single exon genes (IFNA) and presence of numerous gaps in the reference sequence, we were unable to determine the exact annotation of the IFNA genes located at the deletion breakpoints. - As a conse- quence, we were not able to perform in silico comparative analyses of the regulatory elements presumably abolished. - Thus, the use of alternative long-read se- quencing approaches is advisable, especially in conjunction with the improvement of the assembly and the annotation of the corresponding reference genome. - However, we were neither able to assign a specific function nor a definitive annotation to our candidate variant due to the extremely high complexity of the as- sociated genomic region.. - We also detected an association of an evolutionarily conserved SNP with protection against development of hypothyroidism and its high linkage to the candidate structural variant, which makes it extremely arduous to bridge the gap between genetic association and the reve- lation of the actual causative functional variation(s).. - Nevertheless, the knowledge gained in this study might contribute to the development of breeding strategies, via the adoption of a marker-assisted selection eventually in- creasing the frequency of the candidate protective allele(s) in the population. - Considering that cases and controls shared the same geographical origin, as well as appeared as a homogenous population and uniformly in- terspersed in the MDS two-dimensional space after the removal of the outliers, we used a linear mixed model including genomic kinship as random effect, without the inclusion of any population-defining vector as fixed ef- fect. - The statistical significance of the obtained results from the GWA analysis was evaluated as described previously [1]. - The candidate locus was defined based on pairwise LD estimates (r 2 ≥ 0.8) of the most significantly associ- ated SNP (top SNP) to the rest of the SNPs on the chromosome. - The independence of the association sig- nal was tested by a conditional analysis in which the genotype of the top SNP was included as a covariate in the statistical model. - Conditional analysis determines the independence of additional association signals from the leading top SNP and the variants in LD with it com- pared to the remaining variants in the candidate locus.. - Sequencing libraries were sequenced as paired-end reads (2 × 101 bp) with HiSeq2500 (Illumina, San Diego, CA, USA) by using the services of the National Genom- ics Infrastructure at Science for Life Laboratory, Stockholm, Sweden.. - Moreover, we used Integrative Genome Viewer (IGV) [77] to confirm the calling reliability of the resulting set of prioritized SNPs.. - The regenotyped SNPs were subsequently phased and im- puted in the few missing samples (n cases = 8, n controls = 3) using Beagle v as well as employing a reference dataset comprising of the Illumina SNPChip (see Methods, section “Genotyping and quality con- trol”) and Sequenom MassARRAY regenotyped SNPs, in which variants were pruned based on MAF <. - We then performed fine-mapping of the determined genome-wide associated region using. - We filtered CNV calls in the fine- mapped region of association by the mean RD value difference from genomic average (p-value <. - 0.001) and the length of the CNV (>. - Finally, we discarded all CNV calls overlapping with gaps in the reference genome.. - In the individuals with available DNA (n = 101), genotyping of the deletion was performed using a three-primer approach (Fig. - In the remaining individuals lacking DNA specimen (n = 6), the deletion genotype was imputed with the same method as described earlier (see Methods, section “SNP genotyping, imputation and fine-mapping. - The reference dataset for this imputation comprised of the PCR-typed deletion ge- notypes, along with the Illumina SNPChip and Sequenom MassARRAY regenotyped SNPs covering the extended region of association. - (a) QQ plot showing the observed versus expected SNPs p -value distribution of the final complete dataset including both GWA and fine-mapping SNPs. - Table showing the genotypes of the GWA analysis top SNP (CFA average coverage and proportions of reads mapping to Canfam3.1 for the sequenced HC and LC case (highlighted in dark and light red) and HC and LC control (highlighted in dark and light blue) samples. - Table showing all CNVs detected in the fine-mapped region of association by CNVnator. - length of the CNV (Event Length). - p -value of the mean normalized read depth value difference from genomic average ( p -val1). - same as p -val1, but for the middle of the CNV ( p -val3). - same as p -val2, but for the middle of the CNV (p-val4). - Table showing the genotypes of the GWA study top SNP (CFA the predicted CNV (Del3) inferred genotypes (CFA and genotypes of the fine- mapping top SNP (CFA of the sequenced HC and LC case (highlighted in dark and light red) and HC and LC control (highlighted in dark and light blue) samples. - MB contributed to the experimental design and results interpretation, performed all the bioinformatic and genetic analyses, and drafted the initial version of the manuscript, NR contributed to the CNV analysis, ÅK performed serological measurements, EM performed PCR genotyping, CJR performed Nanopore MinION sequencing, KS performed the sample collection and phenotyping, contributed to the experimental design and genetic analyses, GA and OK contributed to the experimental design and results. - Epidemiology and estimated population burden of selected autoimmune diseases in the United States.. - Thyroid function and infertility in the dog: a survey in five breeds. - comparative analysis and haplotype structure of the domestic dog. - Leader of the pack: gene mapping in dogs and other model organisms. - Unexpectedly high allelic diversity at the KIT locus causing dominant white color in the domestic pig.. - A copy number variation map of the human genome. - The importance of the type I interferon system in autoimmunity. - Exacerbation of thyroid autoimmunity by interferon alpha treatment in patients with chronic viral hepatitis: our studies and review of the literature. - An integrated encyclopedia of DNA elements in the human genome. - Resolving the complexity of the human genome using single-molecule sequencing.
Xem thử không khả dụng, vui lòng xem tại trang nguồn hoặc xem
Tóm tắt