Allele-specific expression elucidates cis-regulatory logic.

Most people who tune pianos do not have perfect pitch. In the days before electronic tuners, they would accomplish this feat with amazing precision by listening for the rate at which two notes, struck simultaneously, would “beat” against each other. The contrast of two tones can allow interval tuning with a pitch error well under 0.1%. Throughout the sciences, methods that contrast two signals have provided massive improvements in precision. The interferometer in its various forms has allowed comparison of two signals to achieve precise measurement of the speed of light, the existence of gravitational waves, and quantum entanglement. In functional genomics, one of the most precise ways to compare gene expression levels is to contrast the expression of the two alleles in heterozygotes using an approach called allele-specific expression (ASE). By quantifying the relative counts of a cell’s transcripts from two alleles that differ by single nucleotide polymorphisms (SNPs), ASE has been used to contrast cis- versus trans-regulation of genes [1] and to examine genomic imprinting [2] and escape from X chromosome inactivation [3]. In this issue, we see another excellent application of this approach to explore differences in spatial patterns of expression in the early Drosophila embryo.

Advances in expression profiling in the Drosophila embryo

The Drosophila embryo provides a classic model for pattern formation during development. Much progress has been made in mapping the spatial and temporal gene regulatory circuits that pattern cell fate in the embryo. Spatial gradients of gene expression are strictly controlled along the anterior-posterior and dorso-ventral axes. This is mediated by cis-regulatory modules (CRMs) that respond to maternal and zygotic trans-acting factors that are differentially distributed along the axes [4]. Early methods of analysis relied on beautiful immunofluorescence-tagged images of fly embryos, probing one or a few genes at a time. Now, modern genomic approaches allow whole-genome gene expression measurements that enable construction of 3D gene expression maps of organisms, as done recently using single-cell RNA sequencing in late-stage fly embryos [5]. Similar attempts to enhance spatial resolution have been done by cell-specific barcoding and fluorescence activated cell sorting (FACS) of specific neurons in Caenorhabditis elegans [6] and 2D-arrayed bar-coded primers for RNA sequencing of tissue sections [7]. Concurrently, efforts have been done to more precisely delineate temporal transcriptional differences in the maternal-zygotic transition in Drosophila [8] and across tissues [9] and developmental stages in C. elegans [10]. Because these organisms have two copies of the genome, CRMs can differ between alleles of each copy. Thus, quantification of ASE can augment these studies by allowing us to ask which sequences on the same allele (in cis) are important in orchestrating gene expression during development.

Spatially resolved ASE in F1 hybrid embryos

Combs and Fraser [11] extend these methods by applying spatially resolved genome-wide ASE analysis to Drosophila embryos, to identify candidate genes involved in cis-regulatory divergence. Earlier studies often measured ASE in whole organisms, averaging transcription factor activity across tissues and cell types [12]. This can lead to an underestimation of ASE signals. To overcome this limitation, Combs and Fraser measured transcript profiles in 14-μm–thick cryo-sections that were made along the anterior-posterior axis of F1 hybrid embryos of D. melanogaster and its sister species D. simulans (Fig 1). This approach identified 66 genes whose patterns of ASE varied along the embryo. One of these genes, hunchback, encodes a transcription factor that is expressed anteriorly during Drosophila development and is required for formation of the fly’s anterior segments [13]. Anterior hunchback expression is higher in the melanogaster embryo compared to the simulans embryo, and this same melanogaster bias persists in ASE in the hybrid embryo, suggesting that the difference is cis based. Combs and Fraser then focused tightly on hunchback, seeking to identify binding site differences in the melanogaster and simulans CRMs that might explain the observed ASE bias.

Fig 1

Schematic of Combs and Fraser's model for ASE of hunchback.

In the F1 hybrid embryo (from a D. melanogaster × D. simulans cross), there is a maternal gradient of Bicoid protein, with greater concentration toward the anterior. Bicoid acts as a trans-activator of the zygotic gene hunchback, which has CRMs upstream of the transcription start-site (arrow). The canonical CRM has a cluster of binding sites for Bicoid, indicated by green triangles. The melanogaster and simulans alleles of hunchback differ in that the simulans form has an additional (sixth) Bicoid binding site in its canonical CRM, and it has two weak binding sites for the transcription factor Huckebein, which could act as a repressor. Given these differences in their CRMs, the two species’ hunchback alleles “see” the trans-acting milieu differently. In F1 hybrid embryos, there is greater expression of the melanogaster allele at the anterior end (blue [melanogaster] versus red [simulans] transcripts; detected by identifying SNPs in the RNA sequencing reads). Further posterior in the embryo, Bicoid concentration is lower, and the two alleles of hunchback are transcribed at equal levels. ASE, allele-specific expression; CRM, cis-regulatory module; SNP, single nucleotide polymorphism.

A single SNP drives species-specific spatial hunchback expression

Using a modeling approach [14], the authors narrowed down the transcription factor binding motifs in the hunchback CRMs with the largest effects on hunchback expression. The model singled out the canonical CRM (most proximal to the promoter), which has five Bicoid binding sites in melanogaster. In simulans, there is a sixth Bicoid binding site that also shows weak binding for the transcription factor Huckebein, which could act as a repressor. To functionally confirm that these differences in the canonical CRM are responsible for ASE of hunchback, Combs and Fraser edited the endogenous locus in melanogaster using CRISPR to have a simulans-like canonical CRM, with an extra Bicoid and Huckebein binding site. The edited line was crossed to a simulans line whose canonical CRM contains polymorphisms that make it more similar to melanogaster’s. In the resulting F1 hybrid, there was no longer an expression bias of the melanogaster allele when controlled by the simulans CRM, thus providing evidence that the simulans CRM reduces expression of whichever coding allele it controls. Whether this is directly due to the binding of Huckebein remains to be determined.

Implications for the evolution of development

Because ASE is the read-out of differential cis- and trans-regulation, the molecular mechanisms that drive ASE can only be elucidated if comprehensive information regarding the presence of cis-regulatory regions, transcription factor binding motifs, and epigenetic modifications is known across tissues, cell types, timing in development, and disease states. Such work is ongoing in model organisms, using modeling in silico and using large-scale functional assays, including chromatin immunoprecipitation (ChIP) sequencing and fluorescent reporters [15, 16]. However, in the meantime, ASE in hybrids between model organisms and closely related sister species can help answer important questions related to the evolution of cis-regulation.

In dissecting one instance of species-specific spatially localized ASE, Combs and Fraser confirmed CRM divergence for hunchback that results in spatially distinct expression patterns between melanogaster and simulans embryos, despite the fact that these species are virtually indistinguishable in body plan. This observation is consistent with earlier studies that found abundant differences between melanogaster and simulans in both cis- and trans-regulation, despite the endpoint having minimal phenotypic divergence [17]. Across embryos of multiple fly species, mRNA levels are much more highly conserved than individual transcription factor binding events, consistent with the existence of compensatory cis and trans changes [18]. Such compensatory cis and/or trans variation might be neutral but could be selectively favored under certain conditions [19]. In this regard, ASE analysis is a powerful tool in genome-wide studies to dissect the adaptive nature of changes in gene regulation and how those changes impact development, behavior, and disease [20]. This new study adds a spatial dimension to the analysis of ASE and opens the door for many novel applications for dissecting cis- and trans-regulation of gene expression.