The transcriptome landscape of Prochlorococcus MED4 and the factors for stabilizing the core genome

Abstract

Background: Gene gain and loss frequently occurs in the cyanobacterium Prochlorococcus, a phototroph that numerically dominates tropical and subtropical open oceans. However, little is known about the stabilization of its core genome, which contains approximately 1250 genes, in the context of genome streamlining. Using Prochlorococcus MED4 as a model organism, we investigated the constraints on core genome stabilization using transcriptome profiling.

Results: RNA-Seq technique was used to obtain the transcriptome map of Prochlorococcus MED4, including operons, untranslated regions, non-coding RNAs, and novel genes. Genome-wide expression profiles revealed that three factors contribute to core genome stabilization. First, a negative correlation between gene expression levels and protein evolutionary rates was observed. Highly expressed genes were overrepresented in the core genome but not in the flexible genome. Gene necessity was determined as a second powerful constraint on genome evolution through functional enrichment analysis. Third, quick mRNA turnover may increase corresponding proteins' fidelity among genes that were abundantly expressed. Together, these factors influence core genome stabilization during MED4 genome evolution.

Conclusions: Gene expression, gene necessity, and mRNA turnover contribute to core genome maintenance during cyanobacterium Prochlorococcus genus evolution.

Figure 1

Operon map comparison. The operon map experimentally generated by this study compared with a bioinformatically predicted operon map generated by the Prokaryotic Operon DataBase (ProOpDB) [27]. “hit”: operons that were the same as predictions in ProOpDB; “large”: operons that contained more genes than predicted in the database; “partial”: operons that share partial genes with those in the database; “small”: operons that contain one or more genes predicted in the database; “new”: operons that were newly identified in this study.

Figure 2

Correlation between the gene expression levels and nonsynonymous substitution rates (Ka). RPKM, reads per kilobase per million mapped reads; number of pairwise protein = 1275, Spearman’s r = -0.68, P < 0.001.

Figure 3

Gene expression and molecular evolution of the core genome and flexible genome of Prochlorococcus MED4. (a) Box plot of the correlation between gene expression levels and the nonsynonymous substitution rates (Ka). The line was drawn through the median. A circle represents an outlier, and an asterisk represents an extreme data point. (b) Nonsynonymous substitution rate comparison between CEG and VEG (Mann–Whitney U Test, two-tailed). A circle represents an outlier, and an asterisk represents an extreme data point. (c) Comparison of five expression subclasses between the core genome and flexible genome (Fisher’s exact test, one-tailed). P-value ≤ 0.05 was indicated in figure. HEG, highly expressed genes; MEG, moderately expressed genes; LEG, lowly expressed genes; NEG, non expressed genes; CEG, constantly expressed genes (including four expression subclasses mentioned above); VEG, variably expressed genes.

Figure 4

Gene necessity analysis and COG functional enrichment of HEG. All coding-sequence genes were searched on the Database of Essential Genes (DEG8.0 [39]) using BLASTx (E-value = 1 × 10^-4). (a) Comparison of the DEG-hit genes in the core and flexible genomes. (b) Comparisons of gene expression subclasses between DEG-hit and DEG-miss genes. (c) COG functional enrichment of HEG in the core genome. Statistic significance was performed by Fisher’s exact test (one-tailed). P-value ≤ 0.05 was indicated in figure. COG, clusters of orthologous groups; Core, the core genome; DEG-hit, genes with homologs identified in the database; DEG-miss, genes without any known homologs; Flexible, the flexible genome; Unk, unknown function.

Figure 5

Varied expression in six cellular processes, including photosynthesis[45], carbon metabolism[46], phosphate acquisition[47], nitrogen acquisition[46], hli (high-light inducible genes), and phage infection[48]. (a) Expression profiles of six cellular processes. For each gene, the mean expression in ten samples was used as its expression value. For six functional categories, the mean and median expression values were normalized to values of ribosomal genes. (b) Enrichment analysis of four expression subclasses (HEG, MEG, LEG, and VEG) for six functional processes (Fisher’s exact test, one-tailed). Core: the core genome; Flexible: the flexible genome, HEG: highly expressed genes; MEG, moderately expressed genes; LEG, lowly expressed genes; VEG, variably expressed genes.

Figure 6

Operon distribution of different expression subclasses. (a) Comparison of nonsynonymous substitution rate between operon genes and non-operon genes in MED4 (Mann–Whitney U Test, two-tailed). A circle represents an outlier. (b) Operon rate of four expression subclasses (HEG, MEG, LEG, and VEG) or the core/flexible genomes (Fisher’s exact test, one-tailed). The operon rate was defined as the ratio of operon genes to total genes in a certain gene collection. The operon rate of each subclass was normalized by the whole genome operon rate (55.5%). P-value ≤ 0.05 was indicated in figure. Core, the core genome; Flexible, the flexible genome; HEG, highly expressed genes; MEG, moderately expressed genes; LEG, lowly expressed genes; VEG, variably expressed genes.

Figure 7

Correlation between gene expression levels and mRNA half-lives. (a) Correlation between gene expression levels and mRNA half-lives. Red line shows loess-smoothed curve. The exceptions reported by Steglich et al.[29] were indicated with arrows (b) Box plot of the correlation between gene expression levels and mRNA half-lives (Mann–Whitney U Test, two-tailed). The line was drawn through the median. A circle represents an outlier, and an asterisk represents an extreme data point.