Mouse transgenesis identifies conserved functional enhancers and cis-regulatory motif in the vertebrate LIM homeobox gene Lhx2 locus

Abstract

The vertebrate Lhx2 is a member of the LIM homeobox family of transcription factors. It is essential for the normal development of the forebrain, eye, olfactory system and liver as well for the differentiation of lymphoid cells. However, despite the highly restricted spatio-temporal expression pattern of Lhx2, nothing is known about its transcriptional regulation. In mammals and chicken, Crb2, Dennd1a and Lhx2 constitute a conserved linkage block, while the intervening Dennd1a is lost in the fugu Lhx2 locus. To identify functional enhancers of Lhx2, we predicted conserved noncoding elements (CNEs) in the human, mouse and fugu Crb2-Lhx2 loci and assayed their function in transgenic mouse at E11.5. Four of the eight CNE constructs tested functioned as tissue-specific enhancers in specific regions of the central nervous system and the dorsal root ganglia (DRG), recapitulating partial and overlapping expression patterns of Lhx2 and Crb2 genes. There was considerable overlap in the expression domains of the CNEs, which suggests that the CNEs are either redundant enhancers or regulating different genes in the locus. Using a large set of CNEs (810 CNEs) associated with transcription factor-encoding genes that express predominantly in the central nervous system, we predicted four over-represented 8-mer motifs that are likely to be associated with expression in the central nervous system. Mutation of one of them in a CNE that drove reporter expression in the neural tube and DRG abolished expression in both domains indicating that this motif is essential for expression in these domains. The failure of the four functional enhancers to recapitulate the complete expression pattern of Lhx2 at E11.5 indicates that there must be other Lhx2 enhancers that are either located outside the region investigated or divergent in mammals and fishes. Other approaches such as sequence comparison between multiple mammals are required to identify and characterize such enhancers.

Figure 1. Lhx2 gene loci in human, mouse, chicken and fugu.

Rectangles above the line indicate genes on the forward strand, while rectangles below the line indicate genes on the reverse strand. Genes in blue are syntenic in the 4 species. Not drawn to scale.

Figure 2. Expression of Lhx2, Crb2, Dennd1a in E11.5 mouse embryos.

Expression patterns (ventral, lateral and dorsal views) as determined by whole-mount in situ hybridization at E11.5 for genes (A) Lhx2, (B) Crb2, and (C) Dennd1a. Three to four embryos were assayed for each gene and all embryos gave essentially the same results as these representative embryos. Scale bar at lower right corner denotes 1 mm in length. F: forebrain; M: midbrain; H: hindbrain; E: eye; NT: neural tube.

Figure 3. Region of conserved synteny in human, mouse and fugu used for sequence alignment.

Lhx2 locus in human, mouse and fugu used for sequence alignment. In human and mouse, Crb2 and Dennd1a overlap by ∼100 bp at their 3′ ends.

Figure 4. CNEs in the Lhx2 locus.

Human, mouse and fugu Lhx2 loci were aligned using MLAGAN and CNEs (≥65% identity over 50 bp) were predicted with VISTA. Fugu served as the base sequence. Pink peaks denote CNEs while blue peaks denote conserved coding sequences. The arrows above the blue boxes that denote exons indicate the direction of transcription. x-axis represents distance along the fugu sequence while y-axis shows the percentage identity in each pairwise alignment. The pink peaks that are not annotated do not meet the thresholds of ≥65% identity over 50 bp.

Figure 5. Location of CNEs in the human LHX2 locus.

The exons of protein-coding genes are shown in light/dark blue or green rectangles. The CNEs are located within the introns of the upstream gene DENND1A are indicated by red rectangles that extend above the exons.

Figure 6. Lhx2-associated CNEs direct expression to various tissues in E11.5 mouse embryos.

Lateral and dorsal views of a representative transgenic embryo for each construct. (A) CNE2/3; Strong lacZ expression was observed in the neural tube and dorsal root ganglia. (B) CNE5/6; (C) CNE7; lacZ expression was observed in the hindbrain and the neural tube for CNE5/6 and CNE7. (D) CNE10. lacZ expression extending from the most rostral region of the midbrain to the hindbrain and the entire length of the neural tube. Additional ectopic lacZ expression was detected in the diencephalon for this embryo. Scale bar denotes 1 mm in length. M: midbrain; H: hindbrain; NT: neural tube; DRG: dorsal root ganglion.

Figure 7. TF-encoding genes predominantly expressed in the central nervous system are enriched with CNEs.

(A, B) Tissue expression patterns of 718 human TF-encoding genes. (A) Each of the six panels displays a heat map. The rows and columns of each heat map represent human TF-encoding genes and 62 human tissues respectively. (B) Each graph represents the average gene expression levels of a cluster of TF-encoding genes. Expression cluster #3 consists of genes that predominantly express in the central nervous system. (C) The table pertains to the CNE density (CNE bases per kb of noncoding sequence in a human gene locus), the number and total length of CNEs per TF-encoding gene in different expression clusters.

Figure 8. Site-directed mutagenesis of overlapping motifs in CNE2.

(A) Wild-type and mutant sequences of overlapping motifs in CNE2. (B) lacZ expression although still detectable in neural tube and dorsal root ganglia, is greatly reduced especially in the anterior neural tube. (C, D) Faint lacZ expression in neural tube is present in two different embryos, and visible under higher magnification in (E, F). Yellow scale bar denotes 1 mm in length. Red scale bar denotes 0.2 mm in length.