Z-projections from time-lapse acquisitions (top panels; lateral views of ventral CSF-cNs with rostral to the left) and schematics (bottom panels) showing the 3 stages CSF-cNs go through during the formation of their AE. The CSF-cN soma becomes round and short actin-based protrusions (Stage 2) arise from the ring of actin (Stage 1, arrowhead) concomitantly with a subapical constriction (Stages 2 and 3; chevrons). Gradually, protrusions lengthen to form the AE (Stage 3; arrow) characteristic of differentiated CSF-cNs. Data collected over 13 time-lapse sessions for 15 ventral and 6 dorsolateral cells. The number of cells imaged transitioning from one stage to the next is indicated in the bottom legend. Scale bar, 10 μm. AE, apical extension; CSF-cN, cerebrospinal fluid-contacting neuron. hpf, hours post fertilization.

(A–C) Z-projections from lateral views of the spinal cord in 24-hpf embryos showing the colocalization of the ring of actin (LifeAct; arrowheads) with different markers of the AJCs: (A) Cdh2 for adherens junctions, (B) ZO-1 for tight junctions, and (C) Crb1 for the apical domain. (A) Double immunostaining for GFP and RFP in triple transgenic Tg(pkd2l1:Gal4,UAS:LifeAct-GFP;cryaa:V,cdh2:cdh2-RFP) embryos. (B, C) Double immunostaining for GFP and (B) ZO-1, or (C) Crb1 in Tg(pkd2l1:Gal4,UAS:LifeAct-GFP;cryaa:V) embryos. (D–F) Z-projections from lateral views of 72-hpf Tg(pkd2l1:Gal4,UAS:LifeAct-GFP;cryaa:V) larvae immunostained for GFP and (D) ZO-1, (E) Crb1, or (F) ZO-1 and Crb1. The markers ZO-1 and Crb1 are retained at the AJCs (arrowheads) in both V and DL CSF-cNs after differentiation. In DL cells, Crb1 expands to the apical extension. Scale bars, 10 μm. AJC, apical junctional complexes; Cdh2, Cadherin 2; Crb1, Crumbs 1; CSF-cN, cerebrospinal fluid-contacting neurons; DL, dorso-lateral; GFP, green fluorescent protein; hpf, hours post fertilization; RFP, red fluorescent protein; V, ventral; ZO-1, zonula-occludens-1.

(A) Z-projections from transversal sections showing V and DL TagRFP-CAAX-expressing CSF-cNs in 72-hpf larvae illustrating the smaller AE in crb1^−/− (bottom panel) compared with wild-type siblings (top panel). The central canal is outlined (dotted lines) according to ZO-1 staining. Scale bars, 5 μm. (B) Quantification of the normalized area covered by CSF-cN AEs at 72 hpf in V and DL cells in crb1^−/− (light green; N = 3 fish) compared with wild-type siblings (dark green; N = 4 fish). In both CSF-cN subtypes, the AE was significantly smaller in mutant larvae compared with wild-type (p₁ = 0.0477, p₂ = 9.9019 × 10⁻⁵, p₃ = 2.6381 × 10⁻⁵, p₄ = 0.1876). Underlying data can be found in S1 Data. AE, apical extension; Crb1, Crumbs 1; CSF-cN, cerebrospinal fluid-contacting neuron; DL, dorsolateral; hpf, hours post fertilization; n.s., not significant; V, ventral; ZO-1, zonula-occludens-1.

Expression of candidate factors known to be involved in the formation of actin-based protrusions—baiap2a, baiap2l1b, myo3b, and espin—assessed by FISH at 24 hpf (A1–4) and 72 hpf (B1–4). Expression in CSF-cNs was validated by combining FISH to IHC for RFP or GFP in Tg(pkd2l1:TagRFP) (A1–3, B1–3) or Tg(pkd2l1:GCaMP5G) (A4, B4) transgenic animals, respectively. In 24-hpf embryos, baipa2a (A1), baiap2l1b (A2), myo3b (A3), and espin (A4) were enriched in CSF-cNs. In 72-hpf larvae, expression of baiap2a (B1) and baiap2l1b (B2) was not clearly detected, whereas myo3b (B3) and espin (B4) remained strongly expressed in CSF-cNs. Scale bars, 10 μm. CSF-cN, cerebrospinal fluid-contacting neuron; FISH, fluorescent in situ hybridization; GFP, green fluorescent protein; hpf, hours post fertilization; IHC, immunohistochemistry; RFP, red fluorescent protein.

(A) Z-projections of whole-mounted spinal cords at 72 hpf showing mosaic expression of Myo3b-DN under the control of the pkd2l1 promoter. Immunostaining reveals Espin (cyan) in TagRFP-CAAX-positive CSF-cNs (magenta) expressing Myo3b-DN (yellow, arrowheads) or not (arrows). In ventral CSF-cNs expressing Myo3b-DN, Espin staining is reduced (observed in 9 cells out of 9), and the AE appears smaller compared with wild-type cells. (B) Quantification of the normalized area covered by the AE of ventral cells expressing Myo3b-DN (N = 9 cells) compared with nonexpressing neighboring cells (control; N = 10 cells). Cells expressing Myo3b-DN form a significantly smaller AE (N = 5 fish; p = 0.025). (C) The same quantification was performed in espin^−/− ventral cells at 72 hpf (N = 30 cells in 5 fish) compared with wild-type cells (N = 16 cells in 5 fish). Mutant cells display significantly smaller AEs (p₁ = 8.4010 × 10⁻⁶). In wild-type ventral cells, the overexpression of the zEspin1 was sometimes associated with abnormally long microvilli (observed in 2 cells out of 9; red chevrons; p₂ = 0.2123). In espin^−/− cells, the mutant phenotype was rescued by zEspin1 (N = 15 cells; p₃ = 0.3029 and p₄ = 3.8556 × 10⁻⁶). (D) Z-projections of whole-mounted spinal cords at 72 hpf showing mosaic expression of zEspin1 under the control of the pkd2l1 promoter. Immunostaining for Espin (cyan) in TagRFP-CAAX-positive ventral CSF-cNs (magenta) reveals the loss of Espin immunoreactivity in espin^−/− larvae, which is retrieved in mutant cells expressing zEspin1 (nuclear RFP; magenta; stars). Scale bars, 5 μm. Underlying data can be found in S1 Data. AE, apical extension; CSF-cN, cerebrospinal fluid-contacting neuron; DN, dominant-negative; hpf, hours post fertilization; Myo3b, myosin 3b;n.s., not significant; RFP, red fluorescent protein; zEspin1, zebrafish Espin isoform 1.

(A) Z-projections from transversal sections of spinal cords with V and DL TagRFP-CAAX-positive CSF-cNs at 72 hpf illustrating the smaller AEs in espin^−/− (bottom panel) compared with wild-type siblings (top panel). The central canal was outlined (dotted lines) according to ZO-1 staining. Scale bars, 5 μm. (B) Quantification of the normalized area covered by the CSF-cN AE at 72 hpf in V and DL cells in espin^−/− (light blue, N = 2 fish), espin^+/− (blue, N = 2 fish), and espin^+/+ (dark blue, N = 3 fish) siblings (1 representative experiment out of 2). In both CSF-cN subtypes, the AE size gradually decreases when cells miss 1 (espin^+/−) or 2 (espin^−/−) copies of the wild-type allele (p_V = 0.0112, df = 74, and t = 2.3311 in ventral cells; p_DL = 0.0017, df = 45, and t = 3.0935 in DL cells). (C) Overlay of calcium transients in ipsilateral DL CSF-cNs in response to passive tail bending induced by a glass probe in paralyzed control wild-type larvae versus espin^+/− and espin^−/− siblings at 5 days (120 hpf). The average across cells is shown in black (pulled data from 4 experiments). (D) The amplitude of CSF-cN calcium transients shown in (C) is represented as the ratio of peak fluorescence over baseline (ΔR/R) and is gradually reduced in espin^+/− and espin^−/− compared with wild-type siblings in a wild-type allele number-dependent manner (p = 1.1102 × 10⁻⁶, df = 1309, and t = 8.462). Underlying data can be found in S1 Data. AE, apical extension; CSF-cN, cerebrospinal fluid-contacting neuron; df, degrees of freedom; DL, dorsolateral; hpf, hours post fertilization; V, ventral; ZO-1, zonula-occludens-1.