Dynamic Patterns of Transcript Abundance of Transposable Element Families in Maize

Abstract

Transposable Elements (TEs) are mobile elements that contribute the majority of DNA sequences in the maize genome. Due to their repetitive nature, genomic studies of TEs are complicated by the difficulty of properly attributing multi-mapped short reads to specific genomic loci. Here, we utilize a method to attribute RNA-seq reads to TE families rather than particular loci in order to characterize transcript abundance for TE families in the maize genome. We applied this method to assess per-family expression of transposable elements in >800 published RNA-seq libraries representing a range of maize development, genotypes, and hybrids. While a relatively small proportion of TE families are transcribed, expression is highly dynamic with most families exhibiting tissue-specific expression. A large number of TE families were specifically detected in pollen and endosperm, consistent with reproductive dynamics that maintain silencing of TEs in the germ line. We find that B73 transcript abundance is a poor predictor of TE expression in other genotypes and that transcript levels can differ even for shared TEs. Finally, by assessing recombinant inbred line and hybrid transcriptomes, complex patterns of TE transcript abundance across genotypes emerged. Taken together, this study reveals a dynamic contribution of TEs to maize transcriptomes.

Keywords: Zea mays; expression; transposable elements.

Figure 1

Assessment of expressed TE families in B73. A. Schematic representation of reads assigned to genes or TEs in four categories. The black boxes represent exons of a gene while the colored triangles indicate TEs. Triangles of the same color represent different TEs that are members of the same family. The colored lines represent aligned RNAseq reads. B. For each library, the percent of assigned reads (to TEs or genes) that are assigned to TEs in each of the four categories is shown, with libraries labeled by tissue type and ordered within each tissue by TE contribution. C. The percent of TE reads (unambiguous unique or multi) assigned to the top 10 most highly expressed families, with libraries ordered as in B, demonstrating that much of the variation in total TE contribution across libraries results from expression of the top few families. D. The TE orders for TE families expressed or not expressed in B73 is shown. Less than 15% of all TE families have transcripts.

Figure 2

Dynamic transcript abundance for TE families. A. The tissue-specificity of gene and TE transcript abundance was estimated with the tau metric, where low values indicate constitutive expression and high values indicate tissue-specific expression. B. Heat map showing the transcript abundance of TE families that are expressed across the developmental subset or expressed specifically in one tissue type. Transcript abundance is normalized by row. Columns are labeled by dataset (S = Stelpflug et al. 2016, W = Walley et al. 2016, Z = Zhou et al. 2019). C. The TE families in each group are broken down by TE order (left) and family size (right), showing differences between tissue-specific and constitutive groups.

Figure 3

TE transcripts in ear tissue are dynamic across genotypes. A. The number of TE families expressed (RPM >= 1) is similar in B73 and the other NAM founder lines. B. TE families expressed in B73 also tend to be expressed in a large number of lines, while families not detected in B73 tend to be expressed in few lines. C. Breakdown of TEs detected in any genotype, with rare families detected in < 5 genotypes and common families detected in at least 20 genotypes.

Figure 4

Segregation of TE expression in three LTR families across recombinant inbred lines (RILs). A. The number of annotated and shared members in B73 and Mo17 are shown. B. The relative expression of TE families in B73 (red dot), Mo17 (orange triangle), and the distribution of values across 107 RILs (violin plot). C. RPM values for each library are ranked by total expression and colored by unique or multi-mapping. D. Unique mapping reads for each library are ranked as in C. and colored by the element where the read mapped.

Figure 5

TE expression dynamics in maize hybrids. A. For TE families that are differentially expressed between B73 and Mo17 in four tissues, the distribution of hybrid expression values divided by the mean of the parental values is shown. B. TE families showing non-additive expression in each tissue, defined by the hybrid expressed higher or lower than both inbred parents. The amount of non-additive expression varies across tissues, with the highest counts in reproductive tissues. C-D. Expression profiles of two example LTR families where non-additive expression is restricted to a single tissue.