Genome-wide association mapping of transcriptome variation in Mimulus guttatus indicates differing patterns of selection on cis- versus trans-acting mutations

Pleiotropy, a feature or a bug? Toward co-ordinating plant growth, development, and environmental responses through engineering plant hormone signaling

The advent of gene editing technologies such as CRISPR has simplified co-ordinating trait development. However, identifying candidate genes remains a challenge due to complex gene networks and pathways. These networks exhibit pleiotropy, complicating the determination of specific gene and pathway functions. In this review, we explore how systems biology and single-cell sequencing technologies can aid in identifying candidate genes for co-ordinating specifics of plant growth and development within specific temporal and tissue contexts. Exploring sequence-function space of these candidate genes and pathway modules with synthetic biology allows us to test hypotheses and define genotype-phenotype relationships through reductionist approaches. Collectively, these techniques hold the potential to advance breeding and genetic engineering strategies while also addressing genetic diversity issues critical for adaptation and trait development.

Predicting gene expression responses to environment in Arabidopsis thaliana using natural variation in DNA sequence

The evolution of gene expression responses are a critical component of adaptation to variable environments. Predicting how DNA sequence influences expression is challenging because the genotype to phenotype map is not well resolved for cis regulatory elements, transcription factor binding, regulatory interactions, and epigenetic features, not to mention how these factors respond to environment. We tested if flexible machine learning models could learn some of the underlying cis- regulatory genotype to phenotype map. We tested this approach using cold-responsive transcriptome profiles in 5 diverse Arabidopsis thaliana accessions. We first tested for evidence that cis regulation plays a role in environmental response, finding 14 and 15 motifs that were significantly enriched within the up- and down-stream regions of cold-responsive differentially regulated genes (DEGs). We next applied convolutional neural networks (CNNs), which learn de novo cis- regulatory motifs in DNA sequences to predict expression response to environment. We found that CNNs predicted differential expression with moderate accuracy, with evidence that predictions were hindered by biological complexity of regulation and the large potential regulatory code. Overall, DEGs between specific environments can be predicted based on variation in cis- regulatory sequences, although more information needs to be incorporated and better models may be required.

The quantitative genetics of gene expression in Mimulus guttatus

Gene expression can be influenced by genetic variants that are closely linked to the expressed gene (cis eQTLs) and variants in other parts of the genome (trans eQTLs). We created a multiparental mapping population by sampling genotypes from a single natural population of Mimulus guttatus and scored gene expression in the leaves of 1,588 plants. We find that nearly every measured gene exhibits cis regulatory variation (91% have FDR < 0.05). cis eQTLs are usually allelic series with three or more functionally distinct alleles. The cis locus explains about two thirds of the standing genetic variance (on average) but varies among genes and tends to be greatest when there is high indel variation in the upstream regulatory region and high nucleotide diversity in the coding sequence. Despite mapping over 10,000 trans eQTL / affected gene pairs, most of the genetic variance generated by trans acting loci remains unexplained. This implies a large reservoir of trans acting genes with subtle or diffuse effects. Mapped trans eQTLs show lower allelic diversity but much higher genetic dominance than cis eQTLs. Several analyses also indicate that trans eQTLs make a substantial contribution to the genetic correlations in expression among different genes. They may thus be essential determinants of "gene expression modules," which has important implications for the evolution of gene expression and how it is studied by geneticists.

The quantitative genetics of gene expression in Mimulus guttatus

Gene expression can be influenced by genetic variants that are closely linked to the expressed gene (cis eQTLs) and variants in other parts of the genome (trans eQTLs). We created a multiparental mapping population by sampling genotypes from a single natural population of Mimulus guttatus and scored gene expression in the leaves of 1,588 plants. We find that nearly every measured gene exhibits cis regulatory variation (91% have FDR < 0.05) and that cis eQTLs are usually allelic series with three or more functionally distinct alleles. The cis locus explains about two thirds of the standing genetic variance (on average) but varies among genes and tends to be greatest when there is high indel variation in the upstream regulatory region and high nucleotide diversity in the coding sequence. Despite mapping over 10,000 trans eQTL / affected gene pairs, most of the genetic variance generated by trans acting loci remains unexplained. This implies a large reservoir of trans acting genes with subtle or diffuse effects. Mapped trans eQTLs show lower allelic diversity but much higher genetic dominance than cis eQTLs. Several analyses also indicate that trans eQTL make a substantial contribution to the genetic correlations in expression among different genes. They may thus be essential determinants of "gene expression modules", which has important implications for the evolution of gene expression and also how it is studied by geneticists.

Contributions of mutation and selection to regulatory variation: lessons from the Saccharomyces cerevisiae TDH3 gene

Heritable variation in gene expression is common within and among species and contributes to phenotypic diversity. Mutations affecting either cis- or trans-regulatory sequences controlling gene expression give rise to variation in gene expression, and natural selection acting on this variation causes some regulatory variants to persist in a population for longer than others. To understand how mutation and selection interact to produce the patterns of regulatory variation we see within and among species, my colleagues and I have been systematically determining the effects of new mutations on expression of the TDH3 gene in Saccharomyces cerevisiae and comparing them to the effects of polymorphisms segregating within this species. We have also investigated the molecular mechanisms by which regulatory variants act. Over the past decade, this work has revealed properties of cis- and trans-regulatory mutations including their relative frequency, effects, dominance, pleiotropy and fitness consequences. Comparing these mutational effects to the effects of polymorphisms in natural populations, we have inferred selection acting on expression level, expression noise and phenotypic plasticity. Here, I summarize this body of work and synthesize its findings to make inferences not readily discernible from the individual studies alone. This article is part of the theme issue 'Interdisciplinary approaches to predicting evolutionary biology'.

Deleterious mutations and the rare allele burden on rice gene expression

Deleterious genetic variation is maintained in populations at low frequencies. Under a model of stabilizing selection, rare (and presumably deleterious) genetic variants are associated with increase or decrease in gene expression from some intermediate optimum. We investigate this phenomenon in a population of largely Oryza sativa ssp. indica rice landraces under normal unstressed wet and stressful drought field conditions. We include single nucleotide polymorphisms, insertion/deletion mutations, and structural variants in our analysis and find a stronger association between rare variants and gene expression outliers under the stress condition. We also show an association of the strength of this rare variant effect with linkage, gene expression levels, network connectivity, local recombination rate, and fitness consequence scores, consistent with the stabilizing selection model of gene expression.

Supergene potential of a selfish centromere

Selfishly evolving centromeres bias their transmission by exploiting the asymmetry of female meiosis and preferentially segregating to the egg. Such female meiotic drive systems have the potential to be supergenes, with multiple linked loci contributing to drive costs or enhancement. Here, we explore the supergene potential of a selfish centromere (D) in Mimulus guttatus, which was discovered in the Iron Mountain (IM) Oregon population. In the nearby Cone Peak population, D is still a large, non-recombining and costly haplotype that recently swept, but shorter haplotypes and mutational variation suggest a distinct population history. We detected D in five additional populations spanning more than 200 km; together, these findings suggest that selfish centromere dynamics are widespread in M. guttatus. Transcriptome comparisons reveal elevated differences in expression between driving and non-driving haplotypes within, but not outside, the drive region, suggesting large-scale cis effects of D's spread on gene expression. We use the expression data to refine linked candidates that may interact with drive, including Nuclear Autoantigenic Sperm Protein (NASP^SIM3), which chaperones the centromere-defining histone CenH3 known to modify Mimulus drive. Together, our results show that selfishly evolving centromeres may exhibit supergene behaviour and lay the foundation for future genetic dissection of drive and its costs. This article is part of the theme issue 'Genomic architecture of supergenes: causes and evolutionary consequences'.

PhyloM: A Computer Program for Phylogenetic Inference from Measurement or Binary Data, with Bootstrapping

Quantitative and binary results are ubiquitous in biology. Inasmuch as an underlying genetic basis for the observed variation in these observations can be assumed, it is pertinent to infer the evolutionary relationships among the entities being measured. I present a computer program, PhyloM, that takes measurement data or binary data as input, using which, it directly generates a pairwise distance matrix that can then be subjected to the popular neighbor-joining (NJ) algorithm to produce a phylogenetic tree. PhyloM also has the option of nonparametric bootstrapping for testing the level of support for the inferred phylogeny. Finally, PhyloM also allows the user to root the tree on any desired branch. PhyloM was tested on Biolog Gen III growth data from isolates within the genus Chromobacterium and the closely related Aquitalea sp. This allowed a comparison with the genotypic tree inferred from whole-genome sequences for the same set of isolates. From this comparison, it was possible to infer parallel evolution. PhyloM is a stand-alone and easy-to-use computer program with a user-friendly graphical user interface that computes pairwise distances from measurement or binary data, which can then be used to infer phylogeny using NJ using a utility in the same program. Alternatively, the distance matrix can be downloaded for use in another program for phylogenetic inference or other purposes. It does not require any software to be installed or computer code written and is open source. The executable and computer code are available on GitHub.