On the importance of evolutionary constraint for regulatory sequence identification.

Regulation of gene expression relies on the activity of specialized genomic elements, enhancers or silencers, distributed over sometimes large distance from their target gene promoters. A significant part of vertebrate genomes consists in such regulatory elements, but their identification and that of their target genes remains challenging, due to the lack of clear signature at the nucleotide level. For many years the main hallmark used for identifying functional elements has been their sequence conservation between genomes of distant species, indicative of purifying selection. More recently, genome-wide biochemical assays have opened new avenues for detecting regulatory regions, shifting attention away from evolutionary constraints. Here, we review the respective contributions of comparative genomics and biochemical assays for the definition of regulatory elements and their targets and advocate that both sequence conservation and preserved synteny, taken as signature of functional constraint, remain essential tools in this task.

Creation of zebrafish knock-in reporter lines in the nefma gene by Cas9-mediated homologous recombination.

CRISPR/Cas9-based strategies are widely used for genome editing in many organisms, including zebrafish. Although most applications consist in introducing double strand break (DSB)-induced mutations, it is also possible to use CRISPR/Cas9 to enhance homology directed repair (HDR) at a chosen genomic location to create knock-ins with optimally controlled precision. Here, we describe the use of CRISPR/Cas9-targeted DSB followed by HDR to generate zebrafish transgenic lines where exogenous coding sequences are added in the nefma gene, in frame with the endogenous coding sequence. The resulting knock-in embryos express the added gene (fluorescent reporter or KalTA4 transactivator) specifically in the populations of neurons that express nefma, making them convenient tools for research on these populations.

An in vivo translation-reporter system for the study of protein synthesis in zebrafish embryos.

Control of gene expression at the translation level is increasingly regarded as a key feature in many biological processes. Simple, inexpensive and reliable procedures to visualize sites of protein production are required to allow observation of the spatiotemporal patterns of mRNA translation at subcellular resolution. We present a method, named SPoT (for Subcellular Patterns of Translation), developed upon the original TimeStamp technique ( Lin et al., 2008), consisting in the expression of a fluorescent protein fused to a tagged, self-cleavable protease domain. The addition of a cell-permeable protease inhibitor instantly stabilizes newly produced tagged protein allowing us to distinguish recently synthesized proteins from pre-existing ones. After a brief protease inhibitor treatment, the ratio of tagged versus non-tagged forms is highest at sites where proteins are the most recent, i.e. sites of synthesis. Therefore, by comparing tagged and non-tagged proteins it is possible to spotlight sites of translation. By specifically expressing the SPoT cassette in neurons of transgenic zebrafish embryos, we reveal sites of neuronal protein synthesis in diverse cellular compartments during early development.

BMP- and neuropilin 1-mediated motor axon navigation relies on spastin alternative translation.

Functional analyses of genes responsible for neurodegenerative disorders have unveiled crucial links between neurodegenerative processes and key developmental signalling pathways. Mutations in SPG4-encoding spastin cause hereditary spastic paraplegia (HSP). Spastin is involved in diverse cellular processes that couple microtubule severing to membrane remodelling. Two main spastin isoforms are synthesised from alternative translational start sites (M1 and M87). However, their specific roles in neuronal development and homeostasis remain largely unknown. To selectively unravel their neuronal function, we blocked spastin synthesis from each initiation codon during zebrafish development and performed rescue analyses. The knockdown of each isoform led to different motor neuron and locomotion defects, which were not rescued by the selective expression of the other isoform. Notably, both morphant neuronal phenotypes were observed in a CRISPR/Cas9 spastin mutant. We next showed that M1 spastin, together with HSP proteins atlastin 1 and NIPA1, drives motor axon targeting by repressing BMP signalling, whereas M87 spastin acts downstream of neuropilin 1 to control motor neuron migration. Our data therefore suggest that defective BMP and neuropilin 1 signalling may contribute to the motor phenotype in a vertebrate model of spastin depletion.

Initiation of cyp26a1 Expression in the Zebrafish Anterior Neural Plate by a Novel Cis-Acting Element.

Early patterning of the vertebrate neural plate involves a complex hierarchy of inductive interactions orchestrated by signalling molecules and their antagonists. The morphogen retinoic acid, together with the Cyp26 enzymes which degrade it, play a central role in this process. The cyp26a1 gene expressed in the anterior neural plate thus contributes to the fine modulation of the rostrocaudal retinoic acid gradient. Despite this important role of cyp26a1 in early brain formation, the mechanisms that control its expression in the anterior neural plate are totally unknown. Here, we present the isolation of a 310-base-pair DNA element adjacent to cyp26a1 promoter, displaying enhancer activity restricted to the anterior neural plate of the zebrafish gastrula. We show that unlike that of cyp26a1, expression driven by this cyp26a1 anterior neural plate element (cANE) is independent of retinoic acid. Through deletion analysis, we identify a 12-nucleotide motif essential for cANE activity. A consensus bipartite binding site for SoxB:Oct transcription factors overlaps with this motif. Mutational analysis suggests that SoxB binding is essential for its activity. We discuss the contribution of this study to the elucidation of the regulatory hierarchy involved in early neural plate patterning.

Zebrafish embryonic neurons transport messenger RNA to axons and growth cones in vivo.

Although mRNA was once thought to be excluded from the axonal compartment, the existence of protein synthesis in growing or regenerating axons in culture is now generally accepted. However, its extent and functional importance remain a subject of intense investigation. Furthermore, unambiguous evidence of mRNA axonal transport and local translation in vivo, in the context of a whole developing organism is still lacking. Here, we provide direct evidence of the presence of mRNAs of the tubb5, nefma, and stmnb2 genes in several types of axons in the developing zebrafish (Danio rerio) embryo, with frequent accumulation at the growth cone. We further show that axonal localization of mRNA is a specific property of a subset of genes, as mRNAs of the huc and neurod genes, abundantly expressed in neurons, were not found in axons. We set up a reporter system in which the 3' untranslated region (UTR) of candidate mRNA, fused to a fluorescent protein coding sequence, was expressed in isolated neurons of the zebrafish embryo. Using this reporter, we identified in the 3'UTR of tubb5 mRNA a motif necessary and sufficient for axonal localization. Our work thus establishes the zebrafish as a model system to study axonal transport in a whole developing vertebrate organism, provides an experimental frame to assay this transport in vivo and to study its mechanisms, and identifies a new zipcode involved in axonal mRNA localization.

DeltaC and DeltaD interact as Notch ligands in the zebrafish segmentation clock.

We describe the production and characterisation of two monoclonal antibodies, zdc2 and zdd2, directed against the zebrafish Notch ligands DeltaC and DeltaD, respectively. We use our antibodies to show that these Delta proteins can bind to one another homo- and heterophilically, and to study the localisation of DeltaC and DeltaD in the zebrafish nervous system and presomitic mesoderm (PSM). Our findings in the nervous system largely confirm expectations from previous studies, but in the PSM we see an unexpected pattern in which the localisation of DeltaD varies according to the level of expression of DeltaC: in the anterior PSM, where DeltaC is plentiful, the two proteins are colocalised in intracellular puncta, but in the posterior PSM, where DeltaC is at a lower level, DeltaD is seen mainly on the cell surface. Forced overexpression of DeltaC reduces the amount of DeltaD on the cell surface in the posterior PSM; conversely, loss-of-function mutation of DeltaC increases the amount of DeltaD on the cell surface in the anterior PSM. These findings suggest an explanation for a long-standing puzzle regarding the functions of the two Delta proteins in the somite segmentation clock--an explanation that is based on the proposition that they associate heterophilically to activate Notch.

Setting the tempo in development: an investigation of the zebrafish somite clock mechanism.

The somites of the vertebrate embryo are clocked out sequentially from the presomitic mesoderm (PSM) at the tail end of the embryo. Formation of each somite corresponds to one cycle of oscillation of the somite segmentation clock--a system of genes whose expression switches on and off periodically in the cells of the PSM. We have previously proposed a simple mathematical model explaining how the oscillations, in zebrafish at least, may be generated by a delayed negative feedback loop in which the products of two Notch target genes, her1 and her7, directly inhibit their own transcription, as well as that of the gene for the Notch ligand DeltaC; Notch signalling via DeltaC keeps the oscillations of neighbouring cells in synchrony. Here we subject the model to quantitative tests. We show how to read temporal information from the spatial pattern of stripes of gene expression in the anterior PSM and in this way obtain values for the biosynthetic delays and molecular lifetimes on which the model critically depends. Using transgenic lines of zebrafish expressing her1 or her7 under heat-shock control, we confirm the regulatory relationships postulated by the model. From the timing of somite segmentation disturbances following a pulse of her7 misexpression, we deduce that although her7 continues to oscillate in the anterior half of the PSM, it governs the future somite segmentation behaviour of the cells only while they are in the posterior half. In general, the findings strongly support the mathematical model of how the somite clock works, but they do not exclude the possibility that other oscillator mechanisms may operate upstream from the her7/her1 oscillator or in parallel with it.

The vertebrate segmentation clock.

In vertebrate embryos, somite segmentation is controlled by a molecular clock, in the form of a transcriptional oscillator that operates in the presomitic mesoderm. Most of the genes implicated in the oscillator belong to the Notch pathway; a recently discovered exception is the Wnt pathway gene Axin2. Experiments have revealed several negative feedback loops that might generate oscillations, leading to at least four different theories. The simplest of these is based on direct autoinhibition of certain members of the hairy/E(spl) family of Notch target genes--Hes7 in the mouse, and her1 and her7 in the zebrafish. A mathematical account of this mechanism explains some surprising observations and suggests that the period of oscillation is chiefly determined by the transcriptional and translational delays--the times required to make a molecule of the mRNA and a molecule of the protein.

c-jun regulation and function in the developing hindbrain.

Hindbrain development is a well-characterised segmentation process in vertebrates. The bZip transcription factor MafB/kreisler is specifically expressed in rhombomeres (r) 5 and 6 of the developing vertebrate hindbrain and is required for proper caudal hindbrain segmentation. Here, we provide evidence that the mouse protooncogene c-jun, which encodes a member of the bZip family, is coexpressed with MafB in prospective r5 and r6. Analysis of mouse mutants suggests that c-jun expression in these territories is dependent on MafB but independent of the zinc-finger transcription factor Krox20, another essential determinant of r5 development. Loss- and gain-of-function studies, performed in mouse and chick embryos, respectively, demonstrate that c-Jun participates, together with MafB and Krox20, in the transcriptional activation of the Hoxb3 gene in r5. The action of c-Jun is likely to be direct, since c-Jun homodimers and c-Jun/MafB heterodimers can bind to essential regulatory elements within the transcriptional enhancer responsible for Hoxb3 expression in r5. These data indicate that c-Jun acts both as a downstream effector and a cofactor of MafB and belongs to the complex network of factors governing hindbrain patterning.

Novel activities of Mafb underlie its dual role in hindbrain segmentation and regional specification.

The bZip transcription factor Mafb is expressed in two segments of the developing vertebrate hindbrain: the rhombomeres 5 and 6. Loss of Mafb expression in the mouse mutant kreisler leads to elimination of r5 and to alterations of r6 regional identity. Here, we further investigated the role of Mafb in hindbrain patterning using gain-of-function experiments in the chick embryo. Our work has revealed novel functions for Mafb, including a positive autoregulatory activity, the capacity to repress Hoxb1 expression, and the capacity to synergise with or antagonise Krox20 activity. These different activities appear to be spatially restricted in the hindbrain, presumably due to interactions with other factors. Reinvestigation of the kreisler mutation indicated that it also results in an ectopic activation of Mafb in rhombomere 3, accounting for the previously described molecular alterations of this rhombomere in the mutant. Together, these data allow us to refine our view of the dual function of Mafb in both segmentation and specification of anteroposterior identity in the hindbrain.

Krox20 and kreisler co-operate in the transcriptional control of segmental expression of Hoxb3 in the developing hindbrain.

In the segmented vertebrate hindbrain, the Hoxa3 and Hoxb3 genes are expressed at high relative levels in the rhombomeres (r) 5 and 6, and 5, respectively. The single enhancer elements responsible for these activities have been identified previously and shown to constitute direct targets of the transcription factor kreisler, which is expressed in r5 and r6. Here, we have analysed the contribution of the transcription factor Krox20, present in r3 and r5. Genetic analyses demonstrated that Krox20 is required for activity of the Hoxb3 r5 enhancer, but not of the Hoxa3 r5/6 enhancer. Mutational analysis of the Hoxb3 r5 enhancer, together with ectopic expression experiments, revealed that Krox20 binds to the enhancer and synergizes with kreisler to promote Hoxb3 transcription, restricting enhancer activity to their domain of overlap, r5. These analyses also suggested contributions from an Ets-related factor and from putative factors likely to heterodimerize with kreisler. The integration of multiple independent inputs present in overlapping domains by a single enhancer is likely to constitute a general mechanism for the patterning of subterritories during vertebrate development.