RNA in situ hybridisation reveals fgf3 expression in the developing hypothalamus. (A,B) At 10 somites, fgf3 transcripts are detectable in the forebrain (Fb), at the mid- hindbrain boundary (MHB) and in rhombomere 4 (R4). (C–F) At 20 somites and 30 hpf, fgf3 expression is present in the hypothalamic primordium (arrows), and (G–J) at 36 and 68 hpf in the posterior hypothalamus (pH). Left column depicts lateral views and right column ventral views. Insets are high magnifications of boxed areas. Anterior to the left. (K–N) Double in situ hybridisation for fgf3 with rx3 or emx2 at 30 and 36 hpf shows that fgf3 is concentrated to the posterior hypothalamus. Anterior to the left. (O–V) 20 µm frontal cryosections at levels indicated in G and I of embryos hybridised for fgf3, dusp6 and etv5b at 36 or 68 hpf. O′–V′ are high magnifications of boxed areas. Dashed lines in O′, P′, S′ and U′, and arrowheads in T′, Q′, R′ and V′ indicate ventricle. Dashed circles in T′, Q′, R′ and V′ outline borders of posterior hypothalamus. At 36 hpf, expression of fgf3 spans the entire medial to lateral dimension of the posterior hypothalamus. dusp6 and etv5b are similarly broadly expressed. At 68 hpf, the expression of fgf3 is restricted to cells located medially at the ventricle (arrowheads), while dusp6 and etv5b are present laterally, in cells around the posterior recess. Scale bars: 30 µm.

fgf3^t24152 mutants and fgf3 morphants exhibit a reduced expression of etv5b in the posterior hypothalamus. (A–E) Light microscopic pictures of wild type (+/+), heterozygous (+/−) and homozygous (−/−) fgf3^t24152 mutant siblings, and uninjected control (UC) and fgf3 morphant (MO) siblings processed for RNA in situ hybridisation for etv5b at 36 hpf. Dotted line indicates posterior hypothalamus where etv5b is expressed. Notably, the hypothalamic etv5b expression is weaker in +/− than in +/+ embryos, and strongly reduced in −/− embryos. Similarly, etv5b expression is weaker in MO embryos than in UC embryos. Lateral views, anterior to the left. Scale bars: 30 µm. (F) Read counts of etv5b obtained by RNA sequencing of dissected hypothalami of +/+ and fgf3^t24152 −/− mutants at 3 and 7 dpf. Tukey boxplots show median, 25–75% percentile, IQR whiskers and outliers. n=number of analysed replicates.

Quantification of the number of serotonergic cells in the intermediate (i.)/posterior (p.) clusters and of dopaminergic cells in the DC 4/5/6 and DC 7 clusters in the hypothalamus at 72 hpf after fgf3 impairment. (A–P) Confocal maximum intensity projections from wild-type controls (Ctr), homozygous fgf3^t24152 mutants (−/−), fgf3 morphants (MO) or fgf3 CRISPR/Cas9-injected embryos (CR) immunostained for 5-HT (green) and TH1 (magenta) shown as single channels and merged. C,G,K and O show boxed areas in B,F,J and N, respectively, with adjusted brightness and contrast to reveal faint TH1 immunoreactive cells of the DC 7 cluster. Ventral views, anterior to the left. Scale bars: 10 µm. (Q–Y) Quantifications of 5-HT and TH1 positive cells after fgf3 impairment and in control siblings. The number of serotonergic cells was counted in the i./p. clusters as indicated by the line in A. The number of dopaminergic cells was counted in the DC 4/5/6 and DC 7 clusters as indicated by the lines in B and C. Tukey boxplots show median, 25–75% percentile, IQR whiskers and outliers. n=number of analysed individuals. +/−, heterozygous fgf3^t24152 mutants; UC, uninjected siblings; C9C, injected with Cas9 only. *P>0.05, **P>0.01, ***P>0.001.

Quantification of the number of oxt-,avp- and cort-expressing cells in the hypothalamus of fgf3^t24152 mutants and fgf3 morphants at 72 hpf. (A–L) Light microscopic pictures of wild type (+/+) and homozygous fgf3^t24152 mutant (−/−) siblings as well as fgf3 morphants (MO) and uninjected control siblings (UC) processed for RNA in situ hybridisation. The posterior cells expressing avp are more affected by impaired fgf3 than the anterior ones. Ventral views, anterior to the left. Scale bar: 30 µm. (M–R) Quantifications of oxt-, avp- and cort-positive cells in fgf3^t24152 mutants, fgf3 morphants and control siblings. Cell clusters used for the analyses are indicated by the lines in A, E and I. Tukey boxplots show median, 25–75% percentile, IQR whiskers and outliers. n=number of analysed individuals, +/−, heterozygous fgf3^t24152 mutants. *P>0.05, **P>0.01, ***P>0.001.

The size of the hypothalamic nkx2.4b domain is significantly reduced in fgf3^t24152 mutants and fgf3 morphants. (A–L) Light microscopic pictures showing expression of nkx2.4b in wild type (+/+) and homozygous fgf3^t24152 mutant (−/−) siblings as well as fgf3 morphants (MO) and uninjected control siblings (UC) at 36, 48 and 72 hpf visualised by RNA in situ hybridisation. Outlines of semi-automated measurements of hypothalamic area are highlighted in blue. Ventral views, anterior to the left. Scale bars: 30 µm. (M–R) Area measurements (pixels) in fgf3^t24152 mutants, fgf3 morphants and control siblings. Tukey boxplots show median, 25–75% percentile, IQR whiskers and outliers. n=number of analysed individuals +/−, heterozygous fgf3^t24152 mutants. *P>0.05, ***P>0.001.

fgf3 impairment results in reduced proliferation and increased cell death in the posterior hypothalamus at 36 hpf. (A–F,I,J) Confocal maximum intensity projections of uninjected control (UC) and fgf3 morphant (MO) siblings immunostained for BrdU and phospho-histone H3 (phH3) counterstained with DAPI, or for cleaved caspase 3 (cCasp3) at 36 hpf. Dashed lines indicate ventricle in A and D, and outer posterior border of the hypothalamus in I and J. Examples of cCasp3 positive cells are indicated (arrowheads). Anterior is to the left. Scale bars: 10 µm. (G,H,K) Quantifications of BrdU, phH3 and cCasp3 positive cells after fgf3 impairment. Tukey boxplots show median, 25–75% percentile, IQR whiskers and outliers. n=number of analysed individuals. *P>0.05, **P>0.01.

Ribbon representation of models of zebrafish Fgf3 isoforms bound to a FGF1 receptor. (A) Ribbon representation of a zebrafish Fgf3 model (blue and pink) bound to a FGF1 receptor (yellow). The overall part affected by the fgf3^t24152 mutation or the fgf3 morpholino is coloured in pink. (B) The deleted region of the fgf3^t24152 mutant isoform is omitted. The part that remains (green) as compared to the morpholino knockdown isoform still folds. Notably, receptor binding would be still possible if the variant maintains partial folding, since most of the interfaces with the FGF1 receptor would stay intact. (C) Model of the morpholino knockdown. The exchanged sequence is coloured in red. In comparison to the fgf3^t24152 mutant model it lacks even more structural elements rendering a correct folding and residual receptor interactions less likely.

Hierarchical clustering of hypothalamic Fgf-signalling pathway genes and downstream targets analysed in fgf3^t24152 mutants (−/−) and wildtypes (+/+) at 3 and 7 dpf. (A) Scheme illustrating sample preparation and groups used for RNA sequencing. Micrographs show the dissected hypothalamic area (ventral view, anterior to the left) collected from wild types (blue tubes) and fgf3^t24152 mutants (green tubes) at 3 (light tubes) and 7 dpf (dark tubes). For each group, three independent tissue collections were performed resulting in a total of 12 samples processed for RNA sequencing. Scale bar: 500 µm. (B,D) List of fgfs, fgfrs, Fgf downstream target genes and Fgf-signalling components expressed (base mean ≥10) in the hypothalamus of −/− and +/+ embryos. All genes except shc2 pass the base mean threshold in all four groups. shc2 does not pass the base mean threshold in the 3 dpf −/− group. (C,E) Heat maps of differentially regulated (base mean ≥10, fold change ≥1.5) fgfs, fgfrs and Fgf downstream target genes (C) as well as Fgf-signalling components (E) in the hypothalamus of −/− and +/+ embryos. Columns represent z-score of mean values of replicates for each analysed group and rows depict individual genes. Colour key displays z-score ranging from ≥−1.5 to ≤1.5. Colour intensity represents expression levels of a gene for each group with blue or yellow indicating low or high expression.

Live images showing morphology of 72 hpf embryos after fgf3  impairment with characteristic ear and craniofacial malformations. (A,B) Uninjected control (UC) and fgf3 morpholino injected (MO) siblings.  (C,D) Cas9 injected control (C9C) and fgf3 CRISPR/Cas9 injected (CR) siblings.  (E,F) Wildtype (+/+) and homozygous fgf3t24152 mutant (-/-) siblings. Boxes indicate magnified area shown in  A’-F’. After fgf3 impairment the two otoliths fuse (arrow) and the lower jaw bones are malformed (arrow heads). Lateral views, anterior to the left. Scale bar in  E, 100 μm; in  E’, 50 μm.

Fluorescence pictures showing cell death in fgf3  morphants. (A,B) Live images of acridine orange stained uninjected control (UC) and morphant (MO) siblings at 24 hpf.  A’ and  B’ are high magnifications of boxed areas in  A and  B.  (C,D) Cleaved caspase 3 (cCasp3) immuno stained control and morphant siblings at 36 hpf. Lateral views, anterior to the left. Scale bar in  A, 100 μm; in  A’ and  C 50 μm

Quantification of the number of serotonergic cells in the intermediate (i.)/posterior (p.) clusters and of dopaminergic cells in the DC 4/5/6 and DC 7 clusters in the hypothalamus of fgf3  morphants at 4 dpf. (AH) Confocal maximum intensity projections from uninjected control (UC) and morpholino injected (MO) siblings immuno stained for 5-HT (green) and TH1 (magenta) shown as single channels and merged.  C and  G show boxed areas in  B and  F, respectively, with adjusted brightness and contrast to reveal the faint TH1 immunoreactive cells of the DC 7 cluster.  (L, M) Light microscopic pictures of fgf3 morphants and uninjected control siblings processed for RNA in situ hybridisation for th2 expressed by dopaminergic cells intermingled with TH1 positive cells in the DC7 cluster. Insets show high magnifications of boxed areas. Ventral views, anterior to the left. Scale bars in  A,E,C,G, 10 μm; in  L, 30 μm.  (I-K, N) Quantifications of 5-HT, TH1 and th2 positive cells in control and morphant siblings. The number of serotonergic cells was counted in the i./p. clusters as indicated by the line in  A. The number of dopaminergic (TH1 and th2) cells was counted in the DC 4/5/6 and DC 7 clusters as indicated by the lines in  B and  C, respectively. Tukey boxplots showing median, 25-75% percentile, IQR whiskers and outliers. n = number of analysed individuals.

The shape of the hypothalamic nkx2.4b  domain is altered in fgf3  morphants. (A-F) Light microscopic pictures showing expression of nkx2.4b in fgf3 morphants (MO) and uninjected control siblings (UC) at 36, 48 and 72 hpf visualised by RNA in situ hybridisation. Outlines of semi-automated measurement of hypothalamic area are highlighted in blue. Lateral views, anterior to the left. Scale bar = 30 μm.  (G-I) Area measurements (pixels) in fgf3 morphants and control siblings of the nkx2.4b domain at 36, 48 and 72 hpf.  (J) Total length measurements of fgf3 morphants and control siblings at 72 hpf. Tukey boxplots showing median, 25-75% percentile, IQR whiskers and outliers. n= number of analysed individuals.  (K) Scheme illustrating the lateral and ventral silhouette of the nkx2.4b expressing hypothalamic domain at 36, 48 and 72 hpf. Dark grey indicates location of monoaminergic populations around the posterior recess (PR). Dashed lines show hypothalamic ventricular system with the (LR) lateral recess and the PR.