Expression pattern of magi1 in the zebrafish embryo. (A) Early stages analysed by RT-PCR (data replicated in three separate experiments). Maternal mRNA present in the 0 hpf egg has largely disappeared by 3 hpf, with fresh (zygotically synthesized) transcripts appearing by 6 hpf. EF1α was used as a positive control. (B-E) In situ hybridization patterns. (B) Lateral view at bud stage (10 hpf), showing diffuse expression. (C) Dorsal view at 24 hpf, showing expression still diffuse but strongest in the neural tube. (D) Dorsal view of the head of a whole mount at 72 hpf, showing expression in the retinae and, most strongly, in the diencephalon and telencephalon. (E) Transverse section of embryo stained as a whole mount at 72 hpf, showing expression in the hindbrain and in sensory hair cells in the ear (red arrow)

A morpholino, MO[dlD-V], targeted to the intron-exon boundary responsible for the addition of the DeltaD terminal valine residue disrupts splicing and selectively removes the PDZ domain binding site for up to 72 hours of development. (A) The genomic structure of the zebrafish DeltaD gene surrounding the PDZ domain binding site, indicating the amino acids at the boundaries of each exon, the MO[dlDV] morpholino annealing site, and primer binding sites used to monitor the effects on splicing by RT-PCR in (B). (B) Effects on deltaD mRNA splicing, monitored by RTPCR, in embryos injected with MO[dlD-V] at the one-cell stage and left to develop until the shield stage (6 hpf). Total RNA was extracted from 20 embryos for each dose, reverse transcribed and amplified by PCR. PCR products were cloned and fully sequenced. Uninjected embryos produced a 370 bp band that corresponds to the correctly spliced transcript, coding for a protein that ends –ATEV. The amount of correctly spliced product decreases as the amount of injected morpholino increases; at a dose of 5 ng or more per embryo, no correctly spliced product is observable. Products at 469 and 269 bp were mis-spliced as shown and correspond to proteins lacking the terminal ATEV. The band at ˜330 bp was not characterized but probably represents the use of a cryptic splice donor site. The 546 bp band may result from small amounts of contaminating genomic DNA in the extracted RNA. The separate small panel below shows RT-PCR analysis of the same cDNA using oligonucleotides targeted to an upstream region of the deltaD transcript (nucleotides 389-923) that is unaffected by MO[dlD-V]. The total quantity of deltaD mRNA is not significantly altered by the MO injections. (C) Time course of the effect. Embryos were injected and allowed to develop until the indicated stages, and RT-PCR was performed as in B.

Effects of morpholino injections that block the interaction of DeltaD with MAGI1. (A-E) Dorsal views of the midbrain-hindbrain region in live embryos at 24 hpf. (A) Uninjected wild-type control. (B) Wild-type embryo injected with 5 ng of MO[dlD-V]; note narrowed third (III) and fourth (IV) ventricles and irregular texture due to dying cells in the walls of the hindbrain. (C) Similar phenotype produced by injection of 10 ng of MO[MAGI1]. (D) after eight (aei) uninjected embryo; note normal morphology. (E) aei embryo injected with 5 ng of MO[dlD-V]; note phenotype similar to that seen in (B) and (C). (F-H) Lateral views of live embryos at 24 hpf showing somite segmentation. Somite boundaries are disorganized below the eighth somite in aei (H), but are unaffected by MO[dlD-V] treatment (G). (I-K) Lateral views, anterior towards the left, of embryos at 26 hpf analysed by in situ hybridization for col2a1 expression to mark floor-plate (fp) and hypochord (hc). aei embryos exhibit a reduction in hypochord cell number. This effect is not observed upon MO[dlD-V] treatment (J). Scale bar: 50 μm for I,J,K.

Disruption of the DeltaD-MAGI interaction causes mislocalization of Rohon-Beard neurons. (A-D) Dorsal views of embryos at 16 hpf stained by in situ hybridization for islet1 (black) to mark Rohon-Beard neurons (as well as other primary neurons below the plane of focus). myoD (brown) expression serves as a reference for somite number and position. aei embryos (C,D) show a 1.6-fold increase in the number of Rohon-Beard cells compared to wild-type embryos (A), whereas the number of these cells is only slightly increased in wild-type embryos injected with MO[dlD-V] (B). The MO[dlD-V] embryos are abnormal, however, in that many of the Rohon-Beard cells stray into the midline region. The proportion of such mislocalized cells is not affected by the morpholino in aei embryos, where DeltaD is missing (C,D). (E,F) Cell counts. The distribution of the neurons was quantified for a region corresponding approximately to somites 5 to 10; 10-13 embryos were analysed for each condition. Error bars represent s.e.m. Scale bar: 50 μm for A-D