Genome-wide analysis of allele-specific expression of genes in the model diatom Phaeodactylum tricornutum

Abstract

Recent advances in next generation sequencing technologies have allowed the discovery of widespread autosomal allele-specific expression (aASE) in mammals and plants with potential phenotypic effects. Extensive numbers of genes with allele-specific expression have been described in the diatom Fragilariopsis cylindrus in association with adaptation to external cues, as well as in Fistulifera solaris in the context of natural hybridization. However, the role of aASE and its extent in diatoms remain elusive. In this study, we investigate allele-specific expression in the model diatom Phaeodactylum tricornutum by the re-analysis of previously published whole genome RNA sequencing data and polymorphism calling. We found that 22% of P. tricornutum genes show moderate bias in allelic expression while 1% show nearly complete monoallelic expression. Biallelic expression associates with genes encoding components of protein metabolism while moderately biased genes associate with functions in catabolism and protein transport. We validated candidate genes by pyrosequencing and found that moderate biases in allelic expression were less stable than monoallelically expressed genes that showed consistent bias upon experimental validations at the population level and in subcloning experiments. Our approach provides the basis for the analysis of aASE in P. tricornutum and could be routinely implemented to test for variations in allele expression under different environmental conditions.

Figure 1

Characterization of allele-specific expression in the model diatom P. tricornutum. (a) Number of genes in allele-specific expression categories as a percentage of total genes in P. tricornutum. (b) Examples of the most specific and abundant gene ontologies (biological processes) in ASE, BAE and MAE genes based on TopGO analysis (see also Supplementary Dataset 2). Org.cpd.met = Organic nitrogen compound metabolism. (c) Box plot representation of obsAEB (cDNA) and obsAFB (gDNA) values as per pyrosequencing results for a subset of MAE, BAE and ASE SNV. For each SNV, AEB(AFB) was extrapolated from observed difference allelic frequencies. We also named and labelled by red dots the MAE genes tested in subclonal experiments (see Fig. 2).

Figure 2

Allele-specific expression is stable upon clonal propagation in P. tricornutum. Pyrosequencing results for MAE genes in Pt18.6 subcloned populations (A to H). We represented allele frequency % for each SNV as given by pyrosequencing values. By default, alleles are named ‘allele 1’ and ‘allele 2’, and ‘allele 1’ is the most expressed allele. Error bars represent the standard deviation between 2 technical replicates. SNVs are the same as in table of Supplementary Dataset 1 for the corresponding genes. For each SNV we represent the observed allele frequency (extrapolated from obsAEB) at the population level (Fig. 1) in cDNA by a red line. J45638: 6-pyruvoyl tetrahydropterin synthase; EG01655: unknown function; J49883: glycosyltransferase; J18911: aminoacyl-tRNA synthetase; J46938: NnrU domain containing enzyme.

Figure 3

Association of epigenetic marks and allele-specific expression in P. tricornutum. Percentage overlap (‘Ov’) between annotated genes within the MAE, ASE, and BAE categories and H3K27me3, H3K9me3, H3K9me2, H3K4me2 and H3K9-14AC peaks. ‘R’ represents mean %overlap over 100 iterations between each gene categories and a random histone peak positioning. Overlap were calculated using R software R version 3.6.1 (2019-07-05) (https://cran.r-project.org/bin/windows/base/old/3.6.1/) and the genomation R package computation.