Limits...
Combined QTL and selective sweep mappings with coding SNP annotation and cis-eQTL analysis revealed PARK2 and JAG2 as new candidate genes for adiposity regulation.

Roux PF, Boitard S, Blum Y, Parks B, Montagner A, Mouisel E, Djari A, Esquerré D, Désert C, Boutin M, Leroux S, Lecerf F, Le Bihan-Duval E, Klopp C, Servin B, Pitel F, Duclos MJ, Guillou H, Lusis AJ, Demeure O, Lagarrigue S - G3 (Bethesda) (2015)

Bottom Line: Using new haplotype-based statistics exploiting the very high SNP density generated through whole-genome resequencing, we found 129 significant selective sweeps.We then focused on two of these QTL/sweeps.This study shows for the first time the interest of combining selective sweeps mapping, coding SNP annotation and cis-eQTL analyses for identifying causative genes for a complex trait, in the context of divergent lines selected for this specific trait.

View Article: PubMed Central - PubMed

Affiliation: INRA, UMR1348 Pegase, Saint-Gilles, 35590, France Agrocampus Ouest, UMR1348 Pegase, Rennes, 35000, France Université Européenne de Bretagne, France.

Show MeSH

Related in: MedlinePlus

Expression characterizations of PARK2. (A) Expression pattern in various tissues of chickens quantified using RT-qPCR. Results are given as expression fold change relative to the tight muscle, which exhibited the lowest level of expression. (B) Comparison of PARK2 mRNA level in the fat line (FL, n = 12) and the lean line (LL, n = 12) in the liver and the white adipose tissue (WAT). Results are expressed as the expression ratio relative to the LL ± SEM.*p < 5% and $p < 10% based on unpaired two-tailed Student t-test. (C) cDNA allelic ratio for two marker SNPs located on PARK2 for F1 birds obtained by crossing chicken FL and LL. SNP1 is located on chromosome 3 at 46,581,638 bp and SNP2 is on chromosome 3 at 46,581,695 bp. A total of five birds heterozygous on gDNA at those positions were considered for pyro-sequencing-based cDNA imbalance analyses. Each color and shape is associated with one individual. Down arrows indicate the average allelic ratio for each SNP. The top line stands for the expected allelic ratio in the case of a perfect cDNA allelic balance. The line included in the circle stands for the average value of the imbalance for a given SNP for five individuals. Significance of the allelic imbalance was assessed using a Mann-Whitney unpaired two-tailed nonparametric test to compare the observed allelic ratio with the expected one equal to 1 (i.e., in the case of a perfect balance). **p < 0.01. (D) Comparison of perigonadal fat mass (g) and PARK2 mRNA level in the liver and the white adipose tissue (WAT) of B6.V-Lepob/J mice (Ob, n = 8), KO for the gene encoding leptin, and their genetic background C57BL6/J (BL6, n = 8). For mRNA levels, results are expressed as the expression ratio relatively to the C57BL6/J ± SEM. *p < 0.05 and ***p < 0.001 based on unpaired two-tailed Student t-test. (E) Haplotype cluster frequencies for both chicken lines for the PARK2 selective sweep. The difference in color along the Y-axis gives the frequencies of each haplotype cluster. The difference in color along the X-axis has no meaning. The fixed haplotype is red here in the FL.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4390568&req=5

fig6: Expression characterizations of PARK2. (A) Expression pattern in various tissues of chickens quantified using RT-qPCR. Results are given as expression fold change relative to the tight muscle, which exhibited the lowest level of expression. (B) Comparison of PARK2 mRNA level in the fat line (FL, n = 12) and the lean line (LL, n = 12) in the liver and the white adipose tissue (WAT). Results are expressed as the expression ratio relative to the LL ± SEM.*p < 5% and $p < 10% based on unpaired two-tailed Student t-test. (C) cDNA allelic ratio for two marker SNPs located on PARK2 for F1 birds obtained by crossing chicken FL and LL. SNP1 is located on chromosome 3 at 46,581,638 bp and SNP2 is on chromosome 3 at 46,581,695 bp. A total of five birds heterozygous on gDNA at those positions were considered for pyro-sequencing-based cDNA imbalance analyses. Each color and shape is associated with one individual. Down arrows indicate the average allelic ratio for each SNP. The top line stands for the expected allelic ratio in the case of a perfect cDNA allelic balance. The line included in the circle stands for the average value of the imbalance for a given SNP for five individuals. Significance of the allelic imbalance was assessed using a Mann-Whitney unpaired two-tailed nonparametric test to compare the observed allelic ratio with the expected one equal to 1 (i.e., in the case of a perfect balance). **p < 0.01. (D) Comparison of perigonadal fat mass (g) and PARK2 mRNA level in the liver and the white adipose tissue (WAT) of B6.V-Lepob/J mice (Ob, n = 8), KO for the gene encoding leptin, and their genetic background C57BL6/J (BL6, n = 8). For mRNA levels, results are expressed as the expression ratio relatively to the C57BL6/J ± SEM. *p < 0.05 and ***p < 0.001 based on unpaired two-tailed Student t-test. (E) Haplotype cluster frequencies for both chicken lines for the PARK2 selective sweep. The difference in color along the Y-axis gives the frequencies of each haplotype cluster. The difference in color along the X-axis has no meaning. The fixed haplotype is red here in the FL.

Mentions: Having established that our approach had sufficient resolution to target a unique sweep containing a unique gene, and to identify strong candidate genes underlying QTL with a simple genetic pattern, we then worked on deciphering the more complex QTL region AF3.II. As suggested by the hapFLK profile (Figure 2), this QTL region includes three selective sweeps, AF3.II-a, AF3.II-b, and AF3.II-c, one of them presumably containing a causative mutation. Those three sweeps contained, respectively, no gene, a fraction of MLLT4 (from 1 kb upstream of the gene to intron 6), and a fraction of PARK2 (the whole exon 3 and the flanking intronic regions) (File S2). We first focused on the sweep AF3.II-a containing no gene and established that there was no unannotated gene on this region that was expressed by visualizing, at the sweep position, all tissues RNA-seq reads available in Ensembl. Also, this sweep was located quite far from the closest annotated genes, i.e., 80 kb upstream of SLC35F3 and 100 kb downstream of KCNK1. Taken together, these results were strong enough to obviate a deeper exploration of the AF3.II-a sweep, with no clear evidence that this sweep was actually carrying a variant with an impact on AF. We then explored both sweeps AF3.II-b and AF3.II-c containing, respectively, a fraction of MLLT4 and a fraction of PARK2. Focusing on polymorphisms in coding regions, we observed no indels or nonsynonymous SNPs, suggesting the existence of a causal polymorphism acting in cis on the expression of these genes. Regarding JAG2, we therefore focused on the expression of these two genes in hepatic and adipose tissues in both FL and LL lines. First, we verified that these genes are expressed in these tissues (Figure 5A and Figure 6A). We then analyzed differential expression of these two genes in those tissues between 12 FL and 12 LL. For MLLT4, this analysis revealed no differential expression (Figure 5B). For PARK2, which appeared as ubiquitously expressed with a high expression values in liver (Figure 6A) as already reported in humans (Cesari et al. 2003), we showed a suggested differential expression in WAT and a significant differential expression in liver between FL and LL (Figure 6B). This gene was significantly more expressed in liver in FL compared with LL (23.5 ± 0.3 Ct in FL vs. 24.9 ±0.5 Ct in LL; P < 0.01). We then generated F1 birds by crossing FL and LL to investigate the ASE using pyro-sequencing focusing on two marker SNPs for which assignment of the line-of-origin was possible. We found that PARK2 was a cis-eQTL; it exhibited an ASE profile in liver at both marker SNPs (LL/FL allelic ratio of 0.6 ± 0.1 for both marker SNPs; P < 0.01; n = 5) (Figure 6C), suggesting that those markers are in linkage disequilibrium with a variant acting in cis on PARK2 expression in this tissue. Based on gDNA allelic frequencies at those two markers observed in F0 individuals, we concluded that the underexpressed haplotype was characteristic of LL. In summary, with differential RT-qPCR analysis and a pyro-sequencing-based ASE approach, we obtained consistent results. First, we showed that PARK2 was significantly less expressed in liver in LL. Second, we showed that an LL-specific haplotype was significantly less expressed in F1 liver. These results are consistent with the direction of allele effects in the QTL mapping study reported by Lagarrigue et al. (2006), in which the microsatellite allele of the AF3.II QTL associated with a decrease of adiposity came from the F0 lean line. Those results point to a cis-acting variant impacting the expression of PARK2 in the liver of divergent chicken lines, emphasizing the potential causal status of this gene for adiposity.


Combined QTL and selective sweep mappings with coding SNP annotation and cis-eQTL analysis revealed PARK2 and JAG2 as new candidate genes for adiposity regulation.

Roux PF, Boitard S, Blum Y, Parks B, Montagner A, Mouisel E, Djari A, Esquerré D, Désert C, Boutin M, Leroux S, Lecerf F, Le Bihan-Duval E, Klopp C, Servin B, Pitel F, Duclos MJ, Guillou H, Lusis AJ, Demeure O, Lagarrigue S - G3 (Bethesda) (2015)

Expression characterizations of PARK2. (A) Expression pattern in various tissues of chickens quantified using RT-qPCR. Results are given as expression fold change relative to the tight muscle, which exhibited the lowest level of expression. (B) Comparison of PARK2 mRNA level in the fat line (FL, n = 12) and the lean line (LL, n = 12) in the liver and the white adipose tissue (WAT). Results are expressed as the expression ratio relative to the LL ± SEM.*p < 5% and $p < 10% based on unpaired two-tailed Student t-test. (C) cDNA allelic ratio for two marker SNPs located on PARK2 for F1 birds obtained by crossing chicken FL and LL. SNP1 is located on chromosome 3 at 46,581,638 bp and SNP2 is on chromosome 3 at 46,581,695 bp. A total of five birds heterozygous on gDNA at those positions were considered for pyro-sequencing-based cDNA imbalance analyses. Each color and shape is associated with one individual. Down arrows indicate the average allelic ratio for each SNP. The top line stands for the expected allelic ratio in the case of a perfect cDNA allelic balance. The line included in the circle stands for the average value of the imbalance for a given SNP for five individuals. Significance of the allelic imbalance was assessed using a Mann-Whitney unpaired two-tailed nonparametric test to compare the observed allelic ratio with the expected one equal to 1 (i.e., in the case of a perfect balance). **p < 0.01. (D) Comparison of perigonadal fat mass (g) and PARK2 mRNA level in the liver and the white adipose tissue (WAT) of B6.V-Lepob/J mice (Ob, n = 8), KO for the gene encoding leptin, and their genetic background C57BL6/J (BL6, n = 8). For mRNA levels, results are expressed as the expression ratio relatively to the C57BL6/J ± SEM. *p < 0.05 and ***p < 0.001 based on unpaired two-tailed Student t-test. (E) Haplotype cluster frequencies for both chicken lines for the PARK2 selective sweep. The difference in color along the Y-axis gives the frequencies of each haplotype cluster. The difference in color along the X-axis has no meaning. The fixed haplotype is red here in the FL.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4390568&req=5

fig6: Expression characterizations of PARK2. (A) Expression pattern in various tissues of chickens quantified using RT-qPCR. Results are given as expression fold change relative to the tight muscle, which exhibited the lowest level of expression. (B) Comparison of PARK2 mRNA level in the fat line (FL, n = 12) and the lean line (LL, n = 12) in the liver and the white adipose tissue (WAT). Results are expressed as the expression ratio relative to the LL ± SEM.*p < 5% and $p < 10% based on unpaired two-tailed Student t-test. (C) cDNA allelic ratio for two marker SNPs located on PARK2 for F1 birds obtained by crossing chicken FL and LL. SNP1 is located on chromosome 3 at 46,581,638 bp and SNP2 is on chromosome 3 at 46,581,695 bp. A total of five birds heterozygous on gDNA at those positions were considered for pyro-sequencing-based cDNA imbalance analyses. Each color and shape is associated with one individual. Down arrows indicate the average allelic ratio for each SNP. The top line stands for the expected allelic ratio in the case of a perfect cDNA allelic balance. The line included in the circle stands for the average value of the imbalance for a given SNP for five individuals. Significance of the allelic imbalance was assessed using a Mann-Whitney unpaired two-tailed nonparametric test to compare the observed allelic ratio with the expected one equal to 1 (i.e., in the case of a perfect balance). **p < 0.01. (D) Comparison of perigonadal fat mass (g) and PARK2 mRNA level in the liver and the white adipose tissue (WAT) of B6.V-Lepob/J mice (Ob, n = 8), KO for the gene encoding leptin, and their genetic background C57BL6/J (BL6, n = 8). For mRNA levels, results are expressed as the expression ratio relatively to the C57BL6/J ± SEM. *p < 0.05 and ***p < 0.001 based on unpaired two-tailed Student t-test. (E) Haplotype cluster frequencies for both chicken lines for the PARK2 selective sweep. The difference in color along the Y-axis gives the frequencies of each haplotype cluster. The difference in color along the X-axis has no meaning. The fixed haplotype is red here in the FL.
Mentions: Having established that our approach had sufficient resolution to target a unique sweep containing a unique gene, and to identify strong candidate genes underlying QTL with a simple genetic pattern, we then worked on deciphering the more complex QTL region AF3.II. As suggested by the hapFLK profile (Figure 2), this QTL region includes three selective sweeps, AF3.II-a, AF3.II-b, and AF3.II-c, one of them presumably containing a causative mutation. Those three sweeps contained, respectively, no gene, a fraction of MLLT4 (from 1 kb upstream of the gene to intron 6), and a fraction of PARK2 (the whole exon 3 and the flanking intronic regions) (File S2). We first focused on the sweep AF3.II-a containing no gene and established that there was no unannotated gene on this region that was expressed by visualizing, at the sweep position, all tissues RNA-seq reads available in Ensembl. Also, this sweep was located quite far from the closest annotated genes, i.e., 80 kb upstream of SLC35F3 and 100 kb downstream of KCNK1. Taken together, these results were strong enough to obviate a deeper exploration of the AF3.II-a sweep, with no clear evidence that this sweep was actually carrying a variant with an impact on AF. We then explored both sweeps AF3.II-b and AF3.II-c containing, respectively, a fraction of MLLT4 and a fraction of PARK2. Focusing on polymorphisms in coding regions, we observed no indels or nonsynonymous SNPs, suggesting the existence of a causal polymorphism acting in cis on the expression of these genes. Regarding JAG2, we therefore focused on the expression of these two genes in hepatic and adipose tissues in both FL and LL lines. First, we verified that these genes are expressed in these tissues (Figure 5A and Figure 6A). We then analyzed differential expression of these two genes in those tissues between 12 FL and 12 LL. For MLLT4, this analysis revealed no differential expression (Figure 5B). For PARK2, which appeared as ubiquitously expressed with a high expression values in liver (Figure 6A) as already reported in humans (Cesari et al. 2003), we showed a suggested differential expression in WAT and a significant differential expression in liver between FL and LL (Figure 6B). This gene was significantly more expressed in liver in FL compared with LL (23.5 ± 0.3 Ct in FL vs. 24.9 ±0.5 Ct in LL; P < 0.01). We then generated F1 birds by crossing FL and LL to investigate the ASE using pyro-sequencing focusing on two marker SNPs for which assignment of the line-of-origin was possible. We found that PARK2 was a cis-eQTL; it exhibited an ASE profile in liver at both marker SNPs (LL/FL allelic ratio of 0.6 ± 0.1 for both marker SNPs; P < 0.01; n = 5) (Figure 6C), suggesting that those markers are in linkage disequilibrium with a variant acting in cis on PARK2 expression in this tissue. Based on gDNA allelic frequencies at those two markers observed in F0 individuals, we concluded that the underexpressed haplotype was characteristic of LL. In summary, with differential RT-qPCR analysis and a pyro-sequencing-based ASE approach, we obtained consistent results. First, we showed that PARK2 was significantly less expressed in liver in LL. Second, we showed that an LL-specific haplotype was significantly less expressed in F1 liver. These results are consistent with the direction of allele effects in the QTL mapping study reported by Lagarrigue et al. (2006), in which the microsatellite allele of the AF3.II QTL associated with a decrease of adiposity came from the F0 lean line. Those results point to a cis-acting variant impacting the expression of PARK2 in the liver of divergent chicken lines, emphasizing the potential causal status of this gene for adiposity.

Bottom Line: Using new haplotype-based statistics exploiting the very high SNP density generated through whole-genome resequencing, we found 129 significant selective sweeps.We then focused on two of these QTL/sweeps.This study shows for the first time the interest of combining selective sweeps mapping, coding SNP annotation and cis-eQTL analyses for identifying causative genes for a complex trait, in the context of divergent lines selected for this specific trait.

View Article: PubMed Central - PubMed

Affiliation: INRA, UMR1348 Pegase, Saint-Gilles, 35590, France Agrocampus Ouest, UMR1348 Pegase, Rennes, 35000, France Université Européenne de Bretagne, France.

Show MeSH
Related in: MedlinePlus