Trans-regulatory variant network contributes to missing heritability

In a recent Cell Genomics article, Tsouris et al.1 analyze the transcriptomes of a large diallel panel of hybrids from Saccharomyces cerevisiae natural isolates to study cis- and trans-regulatory changes underlying gene expression variation. Vanessa Pereira and Elena Kuzmin discuss the authors’ findings and the wider context in missing heritability research in this preview.

Gene expression regulation has important consequences on organismal phenotypes. 2 Variation of gene expression among individuals of a natural population is useful for understanding the genotypeto-phenotype relationship.Regulatory variants involving cis-acting factors (mutations in promoter regions) or transacting factors (mutations in transcription factors) and their interplay underlie heritable gene expression variation in a population.Large-scale genomic and transcriptomic analyses are used to measure statistical associations between genetic variants and gene expression levels, thereby identifying cis-and trans-regulatory elements.Detection of expression quantitative trait loci (eQTL), including those using human genomewide association studies, requires a large sample size to achieve sufficient statistical power. 3This limits the discovery of trans-compared to cis-regulatory effects since the former requires many more possible positions to test.However, failing to account for these trans factors and their interactions with cis factors can contribute to the observed missing heritability. 4,5llele-specific expression (ASE) analysis using purebred parental lines and their F1 hybrids to quantify the relative expression of two alleles in a diploid individual offers an alternative approach to detect cis-and trans-regulatory variants.Compared to eQTL studies, ASE-based studies focus on gene-level regulatory changes and thus are statistically well powered.By focusing on parent-hybrid trios, ASE-based studies provide an advantage since they use parentoffspring regression to infer heritability and capture both additive and non-additive variation (Figure 1).A previous study in yeast used a large biparental segregant panel and found evidence to suggest a significant contribution of non-additive effect on gene expression. 3ASE-based studies have been extensively conducted in other organisms; however, they were often limited to one or a few parent-hybrid trios.
To identify trans-eQLTs in a large-scale ASE-based study, Tsouris et al. 1 conducted transcriptomic analysis of a large diallel panel of 323 F1 hybrids from 26 S. cerevisiae isolates.They were obtained from the 1,011-yeast collection of natural isolates capturing the genomic diversity of the species. 6RNA sequencing was carried out on the heterozygous hybrids and homozygous parental lines obtaining overall and allele-specific expression levels for 6,186 genes.The authors analyzed core genes that are present in all these parental lines and accessory genes, most of which originate from S. paradoxus introgression.
The diallel experimental design enabled authors to calculate the narrow-sense heritability (h 2 ), which refers to the phenotypic variance due to additive effects of genes, and broad-sense heritability (H 2 ), which refers to total genetic contribution to the phenotypic variance of a population, for each expression trait.This design thus allowed them to measure the additive contribution from the parental lines or the non-additive contribution from the parental combination in the hybrid.Nonadditive variance was, on average, responsible for 36% of phenotypic variance controlling the expression of onethird of the genes, whereas additive variance controlled one-tenth of the genes and the variation of expression of one-fifth of the genes was stochastic.Non-additive genes were enriched for biological processes such as translation, ribosome biogenesis, and sulfur amino acid biosynthesis, while additive genes were enriched for protein transport to the vacuole.No enrichment was seen for genes with stochastic variation, indicating functional preference for a specific type of regulation.Genes annotated to terms that were enriched for non-additive variance showed higher correlated expression profiles across hybrids than highly additive genes, suggesting that non-additive genes are co-regulated across hybrids.These findings also suggest that the non-additive variance component is more functionally significant than additive.
The authors then sought to determine whether regulatory variation was in cis or in trans by comparing allelic expression in the hybrid to the expression levels in two parental lines (hybrid-parent trio).A cis-regulatory change was identified when the parental lines that differed in expression resulted in an allele-specific expression in the hybrid.alleles equally in the hybrid.Examining 1.2 million sites across 285,777 gene-trio combinations revealed that one-fourth of cases showed allelic expression difference between the hybrid and the parents, with84% due exclusively to trans effects, 4% due exclusively to cis effects, and 12% due to both cis and trans effects.
Four distinct regulatory patterns were identified: the ''attenuating'' group in which trans factors decrease cis effects in the hybrid compared to parental expression levels, the ''reinforcing'' group in which trans factors increase cis effects, the most common ''compensatory'' group trans factors cancel out the cis effects, and the ''reverse'' group with extreme cis-trans interactions.Overall, regulatory variation in trans was more common than in cis, which was consistent with previous studies. 3,7,8Most cis effects are modulated by trans effects, which often act in the opposite direction, suggesting geneexpression-buffering mechanisms.Compensatory patterns were enriched for genes annotated to translation, indicating a mechanism for the buffering of this process.Trans-regulated genes were also more highly expressed, less dispersed, and more highly connected on the coexpression network than cisregulated genes, highlighting their functional significances.Compensatory and trans-regulated genes also showed more non-additive variance compared to cis-controlled genes.
Approximately one-third of S. cerevisiae accessory genes arise from S. paradoxus introgression. 6Since the diallel panel included two Alpechin isolates with introgressed genes, the authors compared regulatory variation within (S. cer vs. S. cer) and between species (S. cer vs. S. par).Between-species allele pairs show more cis-regulatory variation than within-species allele pairs largely due to many cis-regulatory changes between species.Most genes had conserved regulatory patterns, and the majority of those showed trans-regulation for the introgressed allele and S. cer allele.The authors highlighted that trans factors controlled the S. cer allele of the cis-regulated introgressed alleles.Introgressed genes with conserved regulatory patterns showed a higher connectivity in the gene expression network and lower additive variance.Thus, heritable variation of introgressed genes may be influenced not only by interspecies differences but also by their global connectivity.
In conclusion, the study by Tsouris et al. 1 represents a significant advance in our understanding of the regulatory landscape of trait heritability.The identification of trans factors as the main driver underlying non-additive variance highlights the importance of considering trans-regulatory changes in future genetic association studies.Their integration can help uncover additional genetic factors contributing to the heritability of gene expression variation and thus improve our understanding of phenotypic variation.Although further research is needed to elucidate the precise mechanisms, the implication of genes with high non-additive variance being highly connected on the global gene expression network and being trans-regulated suggests that non-additive variance and its buffering effect contribute to missing heritability.Furthermore, together with previous research, 9 the Tsouris et al. study highlights two major routes for transregulation for gene expression variance through either cis-trans interactions or coordinated expression change.As sequencing technologies continue to advance and become more accessible, we can anticipate that analysis of additional S. cer isolates and growth conditions, as well as more nuanced integration of regulatory networks with genetic interaction networks, 10 will greatly enhance our understanding of the biology of inheritance.

DECLARATION OF INTERESTS
The authors declare no competing interests.
On the other hand, a trans-regulatory change showed no allele-specific expression because the trans-acting factor influences both Cell Genomics 4, 100470, January 10, 2024 ª 2023 The Author(s). 1 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Figure 1 .
Figure 1.Cis-and trans-regulatory effects across a population Illustration of cis-, trans-, and cis-trans-regulatory variation and the resulting allele-specific expression patterns across parent-hybrid trios.Top: singlenucleotide polymorphisms (SNPs) in the (top) local regulatory elements (cisregulatory change) manifest as different expression levels in both parents and hybrids.Middle: SNPs in the distant regulatory genes (trans-regulatory change) manifest as different parental expression levels but have no difference in allele expression levels in the hybrid.Bottom: SNPs in both local and distant regulatory elements show a complex pattern of expression due to cistrans interactions.Compensatory effects are shown as an example.