| QTL Map Information |
|---|
| Chromosome: | 14 |
| QTL Peak Location: | n/a |
| QTL Span: | n/a
90.3-90.3 (Mbp) |
 |
Upper, "Suggestive": | n/a |
| Upper, "Significant": | n/a |
| Peak: | rs81451342 |
| Lower, "Significant": | n/a |
| Lower, "Suggestive": | n/a |
| Marker type: | SNP |
| Analysis type: | Association |
| Model tested: | n/a |
| Test base: | Genome-wise |
| Threshold significance level: | Significant |
 |
PostProbInc | 0.1 | | VARIANCE | 1.21 | | Dominance effect: | n/a |
| Additive effect: | n/a |
| Associated Gene: | n/a |
| Cis/Trans acting type: |  |
| Links:
Edit |
Map view
|
| Extended information: |
|---|
| (none) |
User inputs on QTL #263303
Add your annotations
Add your comments
Report problems
|
| QTL Experiment in Brief |
|---|
| Animals: | |
| Breeds associated:
|
| Design: | 2.2.1 Piglets
Detailed descriptions of the piglet trials conducted by the PHGC have been previously published (Lunney et al., 2011) and described more specifically for the trials used in the current study by Hess et al. (Hess et al., 2016). Briefly, 1792 animals from 12 of the PHGC trials were used for this study. Piglets were commercially sourced from North American breeding companies. Pig genetics varied by trial, and all piglets came from farms that were affirmed to be free of PRRSV, Mycoplasma hyponeumoniae, and swine influenza. Animals were transported to Kansas State University at approximately 21 days of age, randomly placed into 10-15 pens, and acclimatized for 7 days. Piglets were inoculated intramuscularly and intranasally with either NVSL-97-7985 or KS-2006-72109 PRRSV2 isolates; all animals used in the current study were challenged and animal was considered the experimental unit. Authors were aware of group allocation at the different stages of the experiment. A subset of phenotypic and genotypic data collected as part of the PHGC trials and stored in the secure PHGC database was used in the present study. Specifically, to study WG, body weight (BW) was measured at 0, 21, and 42 DPI. Piglet WG from 0 to 21 DPI (WG21) and WG from 0 to 42 DPI (WG42) were included in the genetic parameter analyses. PRRSV2 RNA concentration tested from sera at 11 DPI and then log10 transformed for future analyses. Pigs were euthanized at 42 DPI and, except in Trials 7 and 8, where they were euthanized at 35 DPI due to facility availability limitations. Weights from animals in trial 9 were excluded from the dataset as the animals were from the Iowa State University Residual Feed Intake selection lines (Cai et al., 2008). Animals were humanely euthanized by pentobarbital overdose following the American Veterinary Medical Association guidelines for the euthanasia of animals, and all efforts were made to minimize suffering. Serum VL was measured using a semi-quantitative Taq-Man PCR for PRRSV2 RNA described previously (Boddicker et al., 2012). Ear tissues were collected from all pigs for DNA isolation. Trials 1-9 were genotyped on the Illumina Porcine SNP60 Beadchip version 1 (San Diego, CA) at GeneSeek Inc. (Lincoln, NE) while samples from trials 10-15 were genotyped on the Illumina Porcine SNP60 Beadchip version 2 (San Diego, CA) at Delta Genomics (Edmonton, Alberta).
2.2.2 Fetuses
The experimental procedures used for the two PGM trials providing the fetal samples have been previously described in great detail (Ladinig et al., 2014c; Pasternak et al., 2020b). In PGM1, 114 pregnant gilts at 85 (± 1) days of gestation were randomly chosen to be infected with NVSL97–7895 at the University of Saskatchewan (Ladinig et al., 2015); all animals used in the current study were challenged and fetus was considered the experimental unit. In PGM2, 31 gilts were similarly inoculated at 84 (± 0.5) days of gestation (Pasternak et al., 2020b). All gilts in both PGM1 and PGM2 experiments were purebred Landrace bred to homospermic Yorkshire semen sourced from Fast Genetics Inc. (Spiritwood, Saskatchewan). Authors were aware of group allocation at the different stages of the experiment. Prior to infection all gilts were confirmed to be free of PRRSV infection (Ladinig et al., 2015). At 12 (PGM2) or 21 (PGM1) DPMI, gilts and fetuses were euthanized by intravenous barbiturate overdose (Euthanyl Forte, Bimeda MTC Animal Health, 16,200 mg/gilt, ~75-80 mg/kg) and a total of 1276 fetuses (combined PGM1 and 2) were collected. Fetal preservation status was determined based on external appearance and presence of blood and pulsations in the umbilical cord as: viable (white skin, pulsing blood in umbilical cord), meconium staining on the head only, or meconium staining on the body, decomposed (dead, largely normal skin color, minimal edema) or autolyzed (dead, discolored externally, edematous). Blood was collected from the axillary vessels of viable and meconium-stained fetuses but not from dead fetuses that were not included in the present study. Fetal BW, brain weight, and liver weight were measured, and the thymus dissected and snap frozen in liquid nitrogen. The brain:liver ratio of weights was calculated as an indicator of intrauterine growth restriction. PRRSV2 RNA concentration was quantified in fetal thymus using an inhouse probe-based RT-qPCR specific for the inoculum strain as described previously (Ladinig et al., 2014c) and then log10 transformed for future analyses. A total of 818 PGM1 fetuses were genotyped on the Illumina Porcine SNP60 Beadchip version 2 (San Diego, CA) at Delta Genomics (Edmonton, Alberta), while 458 PGM2 fetuses were genotyped on the Affymetrix Axiom Porcine Genotyping Array 650K Beadchip (Santa Clara, CA) at Delta Genomics.
2.2.3 Thyroid hormone measurements
Our investigations focused on serum thyroid hormone levels at 11 DPI in piglets, as T3 hormone levels were most extremely suppressed at this timepoint post PRRSV challenge (Pasternak, 2021). Fetal T3 and T4 levels were measured on sera collected on day of termination at 12 or 21 DPMI, depending on the animal experiment the fetuses were derived from (Pasternak et al., 2020b). There are two primary reasons we have not used the T3:T4 ratio in our analysis. The first is that this value is typically evaluated in the human clinical setting, particularly when evaluating the response to levothyroxine treatment. The value in this measure is derived from an established ratio of production (1:13) in the healthy thyroid, however such a ratio has not been effectively established in a healthy fetal pig throughout gestation. The second stems from our past investigations into fetal thyroid hormone response to PRRSV (Pasternak et al., 2020b; Ko et al., 2022) which indicates that the relative response of these two hormones is dependent on fetal phenotype. In short fetuses classified as viable show a decrease in both T3 and T4 while those classified as meconium stained primarily have a decrease in T4 while maintaining near normal levels of T3. A genome-wide association study (GWAS) analysis on the ratio is therefore likely to identify SNPs associated with the phenotype rather than those directly associated with thyroid hormone. All sera was stored at -20oC or -80°C, respectively, until used to test total T3 (piglet_T3, or fetal_T3, ng/dl) or T4 (fetal_T4, µg/dl) levels using commercial RIA kits (MP Biomedical, Irvine, CA) as previously described (Pasternak et al., 2020b)
|
| Analysis: |
SNP analyses
For the piglet genotyping, the WUR SNP (WUR10000125) genotype, associated with PRRS tolerance (Boddicker et al., 2012; Boddicker et al., 2014; Hess et al., 2016), for each animal was included in our phenotypic dataset based on animals having either 0 or 1-2 copies of the favorable and assumed dominant allele (Boddicker et al., 2012). The effect of WUR SNP on the LS mean of piglet T3 for the 0 genotype was 36.4 ± 1.4 and for the 1 genotype was 38.5 ± 1.5., with a Prob > F = 0.0233 (Supplemental Data File 1). Only SNPs that were present on both versions of the genotyping platforms were retained for downstream analyses. SNPs were removed if they were unmapped or mapped to a sex chromosome in the swine genome build 11.1 (GenBank assembly accession: GCA_000003025.6) (latest), leaving in 58,563 SNPs. Quality control filtering was completed in PLINK software (Purcell et al. 2007); an animal was removed if it had a genotyping call rate less than 90%; SNPs that had a minor allele frequency (MAF) less than 1% were also removed. A total of 1792 animals and 54,357 SNPs was retained after quality control and used for downstream analyses.
For fetal genotyping, only SNPs that were present on both the Illumina Porcine SNP60 Beadchip version 2 and the Affymetrix Axiom Porcine Genotyping Array 650K Beadchip were retained for downstream analyses. SNPs were matched between platforms using both RS identifiers and chromosomal map location, and MAF were verified between platforms. The WUR SNP was not included as a fixed effect in the fetal models as the impact of WUR on the fetal response to PRRS using the PGM model has been previously shown by our group to not been shown to be associated with response to infection (Ladinig et al., 2015).This resulted in 46,526 SNPs available for downstream analyses. SNP processing was as noted for piglets and resulted in 1267 fetuses (9 fetuses removed due to low genotyping call rate) and 38,843 SNPs used for downstream analyses.
Genetic parameter estimation
The piglet and fetal genotype data were used separately to compute a genomic relationship matrix using the Van Raden Method (VanRaden, 2008). Genetic variance components and heritabilities were estimated using ASReml version 4 (Gilmour et al., 2002), separately for the piglet and fetal data sets. The models were developed using an iterative approach, fitting statistically and biologically relevant factors, as summarized in Supplemental Table 1. All fixed effects, covariates, and random effects included in the models were significant at P < 0.05. The Iowa State University High Performance Computing Nova cluster was used for these analyses. The following univariate mixed linear animal model was used for piglet_T3;
Y_"ijklmno" " "= μ+T_i+S_j+W_k+V_l+TP_m+D_n+A_o+e_ijklmno
where Y is the phenotype of piglet T3 (ng/dl) at 11 DPI. Fixed effects included PHGC trial (T = 2, 4, 5, 7-15), sex (S = female, male, or barrow), and WUR SNP genotype (W = no copies or 1-2 copies of the favorable allele). Serum VL at 11 DPI (V) was included as a covariate to account for differences in infection levels (i.e., VL affects thyroid hormone levels). Random effects included pen nested within trial (TP), dam (i.e., litter) (D), and animal genetics (A). The residuals (e) were also included in the model.
The following univariate mixed linear animal model was used for both fetal_T3 and fetal_T4;
Y_"ijklmn" " "= μ+T_i+S_j+F_k+V_l+D_m+ A_n+e_ijklmn
where Y is the phenotype of fetal T3 (ng/dl) or T4 (µg/dl) at 12 or 21 DPMI. Fixed effects included PGM trial (T = PGM1 at 21 DPMI or PGM2 at 12 DPMI), sex (S = female or male), and fetal preservation status (F = viable, meconium staining on the head only, or meconium staining on the entire body). Thymus VL at 12 or 21 DPMI (V) was included as a covariate to account for differences in infection levels. Dam (D) and animal genetics (A) were fitted as random effects, along with residuals (e).
For all traits, the phenotypic variance was calculated as the sum of variance from dam, animal, and residuals. Heritability was calculated as the ratio of animal variance to phenotypic variance. Bivariate animal models were also run in ASReml to estimate phenotypic and genetic correlations between T3 and/or T4 traits and other important traits (i.e., WG and VL), using the same fixed and random effects as in the univariate models, and this applied to WG and VL traits as well as previously described (Hess et al., 2016).
Genome-wide association analyses
GWAS for T3 and T4 levels during PRRSV infection were performed for each trait separately using the Julia for Whole-genome Analysis Software (JWAS) (Cheng et al., 2018). The ASREML genotypic and residual variance estimates were used as starting values in the JWAS analysis. In addition, the genomic relationship matrix results were included as input into the JWAS analysis. The Bayes-B model fits all SNPs simultaneously and, therefore, accounts for relationships and population structure and does not require additional inclusion of PCA or genomic relationships. The USDA-ARS SCINet High Performance Computing Ceres Cluster was used to run these analyses. A Bayes B approach was used (Habier et al., 2011), which fits all SNPs simultaneously as random effects. We employed = 0.999, and a Monte Carlo Markov Chain (MCMC) of length 50,000, with 5,000 for burn-in. The following mixed model was used for both the piglet and fetal datasets, with the same fixed and random effects as the univariate mixed linear animal model was used for piglet_T3, fetal_T3, and fetal_T4, respectively;
Y_ " "= Xb+Wu+∑_j^k▒〖z_j α_j 〗 δ_j+e_
Where Y is a vector of phenotypes, X is an incidence matrix for fixed effects in the model, b is a vector of fixed effects, W is an incidence matrix to account for other random effects in the model, u is a vector of random effects, zj is a vector of genotypes for SNP j based on the number of B alleles (0, 1, or 2), αj is the allele substitution effect for SNP j, δj is a parameter that indicates whether SNP j was included in that sample of the MCMC, and e is the vector or residuals. The analyzed SNPs were split into 2,265 non-overlapping 1-Mb windows across the genome. Based on the infinitesimal model, each window is expected to explain 0.04% (100%/2265) of the GV. We considered windows that explained more than 1% of the GV as significant.
|
| Software: | R, ASREML, and JWAS |
| Notes: | |
| Links: | Edit |
| Reference |
|---|
| Authors: | Angelica Van Goor1,#, Alex Pasternak2, Muhammed Walugembe3, Nadya Chehab1, Glenn Hamonic4, Jack CM Dekkers3, John CS Harding4, and Joan K Lunney1* |
| Affiliation: | USDA-ARS |
| Title: | Genome wide association study of thyroid hormone levels following challenge with porcine reproductive and respiratory syndrome virus |
| Journal: | Frontiers in Genetics, 2023, 14 |
| Links: |
PubMed | Abstract | List all data
| Edit |
| Additional Information |
|---|
| Comments: | Analysis of piglet serum for the thyroid hormone levels triiodothyronine (T3, ng/dl) at 11 days post innoculation with PRRSV. |
User inputs on reference #36845393
Add your comments
Note similar findings
Note contradictions
|
