Allele-specific expression patterns reveal biases and embryo-specific parent-of-origin effects in hybrid maize

Abstract

We employed allele-specific expression (ASE) analyses to document biased allelic expression in maize (Zea mays). A set of 316 quantitative ASE assays were used to profile the relative allelic expression in seedling tissue derived from five maize hybrids. The different hybrids included in this study exhibit a range of heterosis levels; however, we did not observe differences in the frequencies of allelic bias. Allelic biases in gene expression were consistently observed for approximately 50% of the genes assayed in hybrid seedlings. The relative proportion of genes that exhibit cis- or trans-acting regulatory variation was very similar among the different genotypes. The cis-acting regulatory variation was more prevalent and resulted in greater expression differences than trans-acting regulatory variation for these genes. The ASE assays were further used to compare the relative expression of the B73 and Mo17 alleles in three tissue types (seedling, immature ear, and embryo) derived from reciprocal hybrids. These comparisons provided evidence for tissue-specific cis-acting variation and for a slight maternal expression bias in approximately 20% of genes in embryo tissue. Collectively, these data provide evidence for prevalent cis-acting regulatory variation that contributes to biased allelic expression between genotypes and between tissues.

Figure 1.

Definition of Regulatory Variation Based on a Comparison of 1:1 Parental Mix RNA and F1 Hybrid RNA. Hypothetical examples of each type of regulatory variation are illustrated. The F1 genomic DNA exhibits a standardized value of 0.5 (solid black line). The first three genes are examples of differentially expressed genes such that the 1:1 parental mix RNA is statistically different from the F1 genomic DNA control. The three genes on the right are examples of genes that did not have evidence for differential expression (the 1:1 parental mix RNA is not significantly different from the F1 genomic DNA control). The first gene is an example of cis-acting regulatory variation since the 1:1 parental mix RNA and the F1 RNA both exhibit the same level of biased expression. The second gene is an example of trans-acting regulatory variation because the F1 RNA does not exhibit biased allelic variation. The third gene is an example in which the differential expression is caused partially by cis-acting variation and partially by trans-acting regulatory variation. The nondifferentially expressed genes include a gene with no expression variation between any samples, a gene with low levels of cis-variation, and a gene with variation only in the F1 RNA sample.

Figure 2.

ASE and Gene Regulatory Modes. (A) and (B) The proportion of the B73 allele detected in a 1:1 parental mix of inbred B73 and Mo17 RNA is graphed on the x axis, and the proportion of transcript derived from the B73 allele in F1 RNA is graphed on the y axis. All three tissues are displayed using different symbols to denote data derived from embryos (circles), immature ears (squares), and seedlings (triangles). Genes were divided into differentially expressed (A) or nondifferentially expressed (B) groups based on a comparison of allele-specific detection of the inbred 1:1 parental mix RNA versus the F1 DNA control samples. The genes are displayed with color coding to indicate their putative regulatory mode (see Table 2 for definition of regulatory classes). Trend lines that predict the expected location for cis- or trans-acting regulatory variation are shown in (A). Trend lines that show the predicted location for cis-acting or F1 regulatory variation are shown in (B). (C) The genes that exhibit differential expression in the 11-d-old seedling tissues of four additional hybrids are plotted in a graph similar to (A). The proportion of the B73 allele (or the Oh43 allele in the Oh43 × Mo17 cross) was determined in the F1 RNA (y axis) and in a 1:1 mix of the parental inbred lines (x axis). The shape of the symbols denotes the hybrid that was tested, while the color denotes the putative regulatory mode (see Table 2 for definition of regulatory classes). These values represent the mean derived from three biological replicates, and the values have been scaled such that the F1 genomic DNA is equal to 0.5 for each assay.

Figure 3.

Tissue-Specific Alterations in ASE of Hybrids between B73 and Mo17. (A) A graphical representation is used to display the correlation between the F1 hybrid ASE in different tissues. The symbols are used to denote the comparisons between different tissue pairs. The diagonal line indicates the position at which there is no difference in the ASE ratio between the two different tissues. The black data points represent genes with statistically significant (P < 0.05) differences in ASE ratios between the two tissues, while gray data points represent genes that do not exhibit significant differences. (B) A subset of 112 of the 233 genes that were expressed in all three tissues exhibited ASE ratio differences in at least one of the tissue-by-tissue comparisons. The number of genes with significantly different ASE ratio differences for each tissue-by-tissue comparison is illustrated using a Venn diagram. Eleven genes displayed significantly different ASE ratios across all three tissue types.

Figure 4.

Parent-of-Origin Effects on Allelic Expression. (A) Several genes exhibited maternal effects on ASE in embryo tissue; three such genes are shown. The proportion of the transcript that is derived from the B73 allele in the B73 × Mo17 hybrid (black) and the Mo17 × B73 hybrid (gray) for three different tissue types are shown. The standard deviation for each measurement is also indicated. Note that the B73 transcript proportion is highest in B73 × Mo17 embryo and lowest in Mo17 × B73 embryo tissues, indicating maternal effects in the embryo. (B) The difference in the proportion of the B73 allele transcript in B73 × Mo17 hybrids relative to Mo17 × B73 for each of expressed genes was calculated. For any given gene, a value of zero indicates no difference in the ASE of the reciprocal hybrids, whereas a value greater than zero or less than zero indicates a maternal or paternal allelic expression bias, respectively. The frequency of each observed value across all genes is shown for each of the three tissues (for graphical purposes, the data were binned for every 0.01 interval). The seedling and immature ear data show an approximately even distribution that peaks at zero, indicating no widespread trend toward parental expression biases. The embryo data are shifted toward higher maternal allele expression, indicating that the slight maternal bias is widespread and applies to many genes in the embryo tissue.