Quantifying whole transcriptome size, a prerequisite for understanding transcriptome evolution across species: an example from a plant allopolyploid

Abstract

Evolutionary biologists are increasingly comparing gene expression patterns across species. Due to the way in which expression assays are normalized, such studies provide no direct information about expression per gene copy (dosage responses) or per cell and can give a misleading picture of genes that are differentially expressed. We describe an assay for estimating relative expression per cell. When used in conjunction with transcript profiling data, it is possible to compare the sizes of whole transcriptomes, which in turn makes it possible to compare expression per cell for each gene in the transcript profiling data set. We applied this approach, using quantitative reverse transcriptase-polymerase chain reaction and high throughput RNA sequencing, to a recently formed allopolyploid and showed that its leaf transcriptome was approximately 1.4-fold larger than either progenitor transcriptome (70% of the sum of the progenitor transcriptomes). In contrast, the allopolyploid genome is 94.3% as large as the sum of its progenitor genomes and retains > or =93.5% of the sum of its progenitor gene complements. Thus, "transcriptome downsizing" is greater than genome downsizing. Using this transcriptome size estimate, we inferred dosage responses for several thousand genes and showed that the majority exhibit partial dosage compensation. Homoeologue silencing is nonrandomly distributed across dosage responses, with genes showing extreme responses in either direction significantly more likely to have a silent homoeologue. This experimental approach will add value to transcript profiling experiments involving interspecies and interploidy comparisons by converting expression per transcriptome to expression per genome, eliminating the need for assumptions about transcriptome size.

FIG. 1.—

A comparison of transcriptome-normalized expression data versus genome-normalized expression data. Gray circles represent cells, and wavy lines represent transcripts, with the diploid cell having a total of four transcripts in its transcriptome. Black circles represent nuclei, squiggly lines represent gDNA, and white boxes represent the genes encoding the white transcripts. (A) Transcriptome-normalized expression. Expression of the white transcript, measured on a per transcriptome basis, is 0.25 (1 transcript out of a total of 4 transcripts) in the diploid. The same transcriptome-normalized expression values are obtained in two tetraploids showing different expression levels per cell, illustrating that transcriptome-normalized measurements do not provide information on transcript abundance per genome (dosage response), or per cell. (B) Genome-normalized expression. If the expression of the white transcript is instead normalized to genome copy number (1 for the diploid, 2 for the tetraploid), differences in transcript abundance per cell become apparent, and dosage responses can be determined. Relative expression per cell in the tetraploid is simply two times the genome-normalized expression. (C) Relative transcriptome size. Tetraploid transcriptome size (relative to the diploid transcriptome) can then be estimated by dividing relative expression per cell by relative expression per transcriptome.

FIG. 2.—

qRT-PCR based estimates of transcripts per genome (gray; ± SE; N = 3) and RNA-Seq based estimates of transcripts per transcriptome (blue; N = 1) in T2 relative to the midparent values for seven genes or gene families. Values are ordered by relative expression per genome. The relative number of transcripts per cell in T2 versus midparent is equal to 2× the relative number of transcripts per genome.

FIG. 3.—

T2 transcriptome size relative to the transcriptomes of its diploid progenitors. (A) Seven individual gene-based estimates of relative transcriptome size (T2 vs. the diploid midparent transcriptome).“DE” designates the expected value if the T2 transcriptome experienced genome-wide 1:1 dosage effects. “DC” designates the expected value if the T2 transcriptome experienced genome-wide dosage compensation. (B) Average estimate of tetraploid transcriptome size relative to the transcriptomes of each diploid progenitor and to the midparent diploid transcriptome (±SE; N = 7).

FIG. 4.—

Genome-wide distribution of gene dosage responses and fraction of genes within each dosage response category exhibiting homoeologue silencing in the T2 allotetraploid. (A) Number of genes from the RNA-Seq data set with ≥10 unique RPM in at least 1 of the 3 species showing specified dosage responses in T2. (B) Of the genes from panel A for which homoeologue expression could be estimated, number of genes showing specified dosage responses (gray bars) and the fractions of each for which one homoeologue is silenced under these conditions (♦). Dosage responses are expressed as relative expression per genome (T2/midparent): 0.5× = dosage compensation, 1.0× = 1:1 dosage effect.