A mobile insulator system to detect and disrupt cis-regulatory landscapes in vertebrates

Abstract

In multicellular organisms, cis-regulation controls gene expression in space and time. Despite the essential implication of cis-regulation in the development and evolution of organisms and in human diseases, our knowledge about regulatory sequences largely derives from analyzing their activity individually and outside their genomic context. Indeed, the contribution of these sequences to the expression of their target genes in their genomic context is still largely unknown. Here we present a novel genetic screen designed to visualize and interrupt gene regulatory landscapes in vertebrates. In this screen, based on the random insertion of an engineered Tol2 transposon carrying a strong insulator separating two fluorescent reporter genes, we isolated hundreds of zebrafish lines containing insertions that disrupt the cis-regulation of tissue-specific expressed genes. We therefore provide a new easy-to-handle tool that will help to disrupt and chart the regulatory activity spread through the vast noncoding regions of the vertebrate genome.

Figure 1.

The expression disruption system. (A) Representation of a genomic landscape composed of two enhancers (gray boxes) that drive expression of their target gene in the eye and hindbrain (gray pattern in embryo). (B) An ED insertion in this genomic landscape. ED is composed of two enhancer traps, one that has GFP (green box) as a reporter gene and the other RFP (red box). An insulator is present in between these two enhancer traps (purple box). Two loxP sequences (yellow triangles) flank the RFP enhancer trap and the insulator. In this insertion, the upstream enhancer is detected exclusively by the GFP enhancer trap cassette, and the downstream is detected only by the RFP cassette, resulting in the nonoverlapping expression of these two reporters (green and red patterns in the lower embryo). The block of the upstream enhancer by the insulator results in a regulatory mutation (higher embryo).

Figure 2.

Expression disruption lines. (A–C) The lines ED185b (A), ED186 (B), and ED170 (C) show separated expression of GFP (green) and RFP (red). ED185b has expression of GFP in the neural crest (arrowhead) and RFP in the notochord (arrow). Also, both reporters show coexpression in the forebrain (asterisk). ED186 shows predominant expression of RFP in the eye (arrowhead) and central nervous system, while expression of GFP is restricted to very mild levels in the eye (arrowhead) and forebrain (asterisk). ED170 shows predominant expression of GFP in the notochord (arrow). (D) ED162 shows mostly coexpression of GFP and RFP in the central nervous system, having an exclusive domain of GFP expression in the otic vesicle (arrowhead).

Figure 3.

Example of insertions that cause mutations in nearby associated genes. (A) ED25 GFP expression recapitulates the expression pattern of klf4b detected by in situ hybridization, in the blood island (white and black arrowheads). (B) ED27 shows expression of GFP in the spinal cord (white arrow), which coincides with the expression of dacha (black arrow). (C) ED170 shows a strong expression of GFP in the notochord (blue arrowhead), recapitulating the expression of ptrfb (blue arrowhead). (A–C) In all three cases, homozygous embryos for the insertions show decreased levels of transcripts for the associated genes (third column), and the endogenous expression is recovered when Cre recombinase is injected in this homozygous mutant background (fourth column). Asterisks mark the expression of egr2b, a marker used as an internal control for the in situ hybridization. (D) ED186 shows strong RFP expression in the central nervous system (blue arrow) and eye (white dotted circle). (E) This line is an ED insertion near mir124-5 oriented with the RFP enhancer trap upstream of this noncoding gene. (F) In ED186, an assay for enhancer activity of the sequence 4.7 kb upstream of the insertion point reveals expression in the central nervous system (blue arrow) and eye (white dotted circle). (G) Homozygous embryos for ED186 insertion show a decrease of >70% in the transcription levels of mir124-5, detected by qPCR.

Figure 4.

The genomic landscape of the ED170 insertion. (A) ED170 is an insertion in an exon of atp6v0a1b, 10 kb away from the transcription start site of its associated gene, ptrfb. This insertion is oriented with the GFP enhancer trap upstream of ptrfb. Four candidate enhancer sequences were selected, 1 to 4 (black boxes). (B) Enhancer activity assays show that sequences 1, 2, and 3 are enhancers that drive expression in the notochord, as observed by expression of GFP. (C) When performing a 3C assay comparing the levels of interaction of the enhancer 1 and two control sequences with the promoter of ptrfb, a significant difference is observed; (**) P < 0.01. This same assay was performed using homozygous embryos for ED170, and the difference in the levels of interaction of the enhancer 1 and the two control sequences is not statistically significant (ns, not significant). The primers used for the 3C experiment are represented in A as an orange triangle for the enhancer 1, blue triangles for control sequences, and a magenta triangle for the ptrfb promoter sequence.

Figure 5.

Phenotype associated with the ptrfb/ED170 mutant line. (A) In homologous ptrfb/ED170 mutant embryos, indentations in the notochord are observed (arrows). Transmitted light (top) and GFP fluorescence (bottom) images are presented. (B) Graph representing the percentage of embryos, wild-type (WT), heterozygous (Het), or homozygous (Hom) for the ED170 insertion, presenting the described phenotype (blue) in uninjected (No Cre; n = 354) and in Cre mRNA-injected backgrounds (Cre; n = 126). (C) Indentations in the ED170 mutant background (ED170; top panels) correspond to nuclei-enriched regions in the notochord (asterisk) that in controls remain nuclei-free (Cont.; lower panels). (D) In morpholino-injected embryos targeting ptrfb, indentations (black arrowhead) that correspond to nuclei-enriched regions in the notochord (asterisk) are also detected. Confocal images are marked in green for GFP, red for rodamin-phalloidin/actin staining, and blue for DAPI/nuclei staining.

Figure 6.

Mouse transgenic ED lines. Eight mouse transgenic embryos were generated using the ED system. (A) In this example, GFP expression is detected in the dorsal periocular mesenchyme (asterisk; inset), and coexpression of GFP and RFP is observed in the ventral periocular mesenchyme (arrow; inset). (B) Strong GFP expression is detected in the heart (asterisk), while GFP and RFP coexpression is detected at the base of the forelimb (arrow), the branchial arches (arrowhead), and at low levels in the somites. (C) GFP expression is detected in the trigeminal ganglion (arrow) and the midgut (arrowhead), while RFP expression is not detected. (D) This embryo expresses GFP throughout the embryo in migratory cells that probably correspond to dermis (asterisk), while RFP is found ubiquitously at low levels. (E) RFP expression is detected in the olfactory pits (arrowhead) and the second branchial arch (arrow), possibly in the mesodermal core, while GFP is detected at low levels ubiquitously. (F) RFP expression alone is detected in the forebrain (arrowhead), somites (arrow), and cells migrating into the limb buds; RFP and GFP coexpression is strongly detected in the proximal region of the forelimb (asterisk), while unique GFP expression is detected at low levels ubiquitously. (G) GFP and RFP are coexpressed ubiquitously at low levels. (H) Expression of GFP alone is observed superficially in the mandibular component of the first branchial arch (arrowhead), in tissues surrounding the fore and hindlimbs (arrow), and RFP expression is not detected. The stages of the embryos shown are 11.5 d post-coitum (dpc; A,B) and 10.5 dpc (C–H). The first column presents transmitted light images, the second an overlay of GFP and RFP channels, the third the GFP channel alone (green), and the fourth the RFP channel alone (red).