Transcriptome-wide investigation of genomic imprinting in chicken

Abstract

Genomic imprinting is an epigenetic mechanism by which alleles of some specific genes are expressed in a parent-of-origin manner. It has been observed in mammals and marsupials, but not in birds. Until now, only a few genes orthologous to mammalian imprinted ones have been analyzed in chicken and did not demonstrate any evidence of imprinting in this species. However, several published observations such as imprinted-like QTL in poultry or reciprocal effects keep the question open. Our main objective was thus to screen the entire chicken genome for parental-allele-specific differential expression on whole embryonic transcriptomes, using high-throughput sequencing. To identify the parental origin of each observed haplotype, two chicken experimental populations were used, as inbred and as genetically distant as possible. Two families were produced from two reciprocal crosses. Transcripts from 20 embryos were sequenced using NGS technology, producing ∼200 Gb of sequences. This allowed the detection of 79 potentially imprinted SNPs, through an analysis method that we validated by detecting imprinting from mouse data already published. However, out of 23 candidates tested by pyrosequencing, none could be confirmed. These results come together, without a priori, with previous statements and phylogenetic considerations assessing the absence of genomic imprinting in chicken.

Figure 1.

Experimental design. (A) Heatmap of genetic similarities between lines. Colors correspond to the inbred level of tested individuals (red is inbred). Lines 6 and R⁻ (red arrows) were chosen to be as inbred and as genetically distant as possible. (B) Reciprocal cross between the two selected lines. In case of genomic imprinting, embryos are preferentially expressing one of their parents’ allele (this figure shows a case of paternally expressed gene). Detection of such event is possible with polymorphisms differentiating the lines.

Figure 2.

Genomic contexts of the candidate loci on the chicken genome.

Figure 3.

Comparison of results on a candidate gene obtained through HiSeq, droplet digital PCR and pyrosequencing. Colors are differentiating both alleles in sequencing results and ddPCR. (A) Hiseq counts of both alleles at the candidate locus (A allele in green, G allele in blue). (B) Pyrosequencing results from one sample of each cross (chosen as representative of average results). Analyses were made on the reverse strand. Allele A is overexpressed in both crosses (left: Cross 1 results from one embryo's cDNA, right: Cross 2 results from one embryo's cDNA).The highlighted peaks correspond to the SNP where the relative proportion is quantified. (C) Droplet digital PCR on the candidate locus from eight samples of each cross (A allele in green, G allele in blue).

Figure 4.

Candidate SNP positions among sequenced reads. Boxplot of SNP localization among reads for each candidate ordered by median. Orange color represents candidates with a median position in the 20% extremities of reads.

Figure 5.

Sequencing depth distribution for all SNPs (top), with classes containing candidate SNPs in gray, and for candidate SNPs (bottom), with tested SNPs in black.