Human and zebrafish cranial motor nerves and study workflow. (A) In the human brainstem, motor neurons cluster into cranial motor nuclei, whose axons form the CNs. The oCCDDs can arise from defective formation, identity, and/or axonal projections of three of these motor neuron populations (CN3, CN4, and CN6) that collectively innervate seven extraocular muscles to orchestrate eye/eyelid movement. CN3 innervates the IO, IR, LPS, MR, and SR muscles. CN4 innervates the SO muscle, and CN6 innervates the LR muscle. CN5 normally innervates the muscles of mastication (not shown) but can also aberrantly innervate the LPS muscle in MGJWS. (B) Summary of CN pathology in humans or mice with specific oCCDDs, and the zebrafish reporter lines used to analyze them. To model oCCDDs in zebrafish, our study leveraged the HGj4A mnr2b/hlxb9lb line for DRS candidate genes and the Tg(isl1:GFP) line for CFEOM, congenital ptosis, and MGJWS genes. (C) Zebrafish wild-type ocular and trigeminal motor nerves visualized in the Tg(isl1:GFP) (a) or HGj4A reporter line (b). Left-lateral view; right-dorsal view. (c) Summary of oCCDD-relevant CNs that are present in zebrafish and labeled by each reporter line. (a, c) As in humans, zebrafish CN3 innervates the IO, IR, MR, and SR muscles. However, the CN3 branch to the IO muscle cannot be visualized with the Tg(isl1:GFP) transgenic line. Additionally, unlike humans, zebrafish lack the LPS muscle and its corresponding CN3 subdivision. Zebrafish have bilateral CN4 motor nuclei and nerves (left and right). CN4 exits the brainstem dorsally and travels ventrally to innervate the contralateral SO muscle. At 72 hpf, wild-type zebrafish CN4 is variably defasciculated. Zebrafish have anterior and posterior trigeminal motor nuclei bilaterally, which extend axons for the motor trigeminal (CN5) nerve that innervates the muscles of mastication (not shown). (b, c) The HGj4A line labels the CN6 motor neurons and their axons. Zebrafish have anterior and posterior CN6 motor nuclei bilaterally, which extend CN6 nerves that target the LR muscles. (D) Workflow for the identification and screening of known and novel oCCDD candidate genes and variants. We reported steps 1-3 in our previous study describing exome/genome sequencing of our human oCCDD cohort³ and performed steps 4–7 in the present study. (E) Demographics of the 46 human probands whose 43 distinct human candidate genes were tested in the zebrafish screen. Percentages of probands are shown with oCCDDs that are (a) syndromic or isolated, (b) sporadic or familial, (c) fitting various oCCDD subdiagnoses, or (d) fitting various modes of inheritance. (e) The types of variants fitting modes of inheritance defined in d are provided. AD, autosomal dominant; AR, autosomal recessive; CFEOM, congenital fibrosis of the extraocular muscles; CN, cranial nerve; CN3, oculomotor/cranial nerve 3; CN4, trochlear/cranial nerve 4; CN5, trigeminal motor/cranial nerve 5; CN6, abducens/cranial nerve 6; DNV, de novo variant; DRS, Duane retraction syndrome; hpf, hours post-fertilization; IO, inferior oblique muscle; IR, inferior rectus muscle; LPS, levator palpebrae superioris muscle; LR, lateral rectus muscle; MGJWS, Marcus Gunn jaw-winking syndrome; MR, medial rectus muscle; oCCDD, ocular congenital cranial dysinnervation disorder; Ptosis, congenital ptosis; SO, superior oblique muscle; SR, superior rectus muscle; XLR, X-linked recessive. Pink, CN3; light green, CN4; dark blue, CN5; dark green, CN6; partially transparent brown, extraocular muscle; dashed line, nerve that is present in zebrafish but not labeled with the transgenic reporter line.

G0 zebrafish targeting of phox2a. Representative images obtained by stereomicroscope (A–C) and confocal microscopy (D–F) of uninjected (A, D), scramble-injected (B, E), or phox2a-targeting guide-injected G0 Tg(isl1:GFP) zebrafish at 72 hpf. Stereomicroscope images for each treatment group correspond to the confocal images obtained from the same fish in the panel below. Because residual motor neurons from CN3 and/or CN4 could not be definitively assigned to either of these specific motor neuron nuclei, CN3/CN4 were labeled as a single entity in Guide-targeted fish. Note apparent absence (C) or paucity (F) of CN3/CN4 motor neurons following phox2a G0 mosaic knockout. (G) Barplots showing the percentages of zebrafish exhibiting wild-type-like (“WT-like”), malformed, or absent CN3/CN4 motor nuclei, as scored under the dissecting stereomicroscope in G0 zebrafish at 72 hpf. Total numbers of fish in each group are given above the corresponding bar. Data are shown from three experimental replicates. Pearson's 3 × 3 χ² test with 4 degrees of freedom; P values and χ² values provided for each replicate. (H) Kaplan-Meier survival curves demonstrating the relative survival probabilities of scramble-injected (blue line) and guide-injected (pink line) zebrafish over the first 72 hours of life. “Number at risk” below the plot provides the counts of surviving embryos in each group taken every 24 hpf over a 72 hpf period. Relative survival probabilities of phox2a-targeting and scrambled gRNA-injected embryos were compared by the log-rank test. Displayed data were derived from a single experimental replicate, but measurements were taken for three experimental replicates, all of which showed the same trend.

G0 zebrafish targeting of mafba. (A–H) Representative images obtained by stereomicroscope (A–D) and confocal microscopy (E–H) of uninjected (A, E), scrambled guide-injected (B, F), or mafba-targeting guide-injected (C, D, G, H) G0 HGj4A zebrafish at 72 hpf. Two images are provided for guide-injected fish, demonstrating variability in G0 targeting outcomes. Stereomicroscope images for each treatment group correspond to the confocal images obtained from the same fish in the panel below. Motor neurons in the abducens nucleus (dashed white circled region) appear absent by stereomicroscopy (C, D) or variably reduced (G, H) by confocal microscopy, and abducens nerves appear thin (G;blue arrow) or absent (G, H;red arrows) compared to normal (E, F; white arrows) by confocal microscopy. (I) Barplots showing the percentages of zebrafish exhibiting wild-type-like (WT-like), malformed, or absent CN6 motor neuron nuclei, as scored under the dissecting stereomicroscope in G0 zebrafish at 72 hpf. Total numbers of fish in each group are given above the corresponding bar. Data are shown from three experimental replicates. Pearson's 3 × 3 χ² test with 4 degrees of freedom; P values and χ² values provided for each replicate. (J) Kaplan-Meier survival curves demonstrating the relative survival probabilities of scramble-injected (blue line) and guide-injected (pink line) zebrafish over the first 72 hours of life. Number at risk below the plot provides the counts of surviving embryos in each group taken every 24 hpf over a 72 hpf period. Relative survival probabilities of mafba-targeting guide-injected and scrambled-injected embryos were compared by the log-rank test. Displayed data were derived from a single experimental replicate, but measurements were taken for three experimental replicates, all of which showed the same trend.

G0 mosaic and F2 germline zebrafish targeting of sema3fa. (A–E) Initial sema3fa targeting experiments were performed in G0 embryos. Representative images obtained by confocal microscopy (A–C) of uninjected (A), scrambled guide-injected (B), or sema3fa-targeting guide-injected (C) G0 Tg(isl1:GFP) zebrafish at 72 hpf. Note defasciculated CN3 (yellow arrow) and failed CN3 nerve extension toward extraocular muscles (yellow stars) relative to uninjected and scramble-injected CN3 (white stars). CN3, cranial nerve 3 (oculomotor); yellow arrow, increased defasciculation of CN3 nerve; yellow star, failed extension of CN3 nerve toward target extraocular muscles; white star, typical CN3 branches toward extraocular muscles. (D) Barplots showing the percentages of zebrafish exhibiting wild-type-like (WT-like) or defasciculated CN3 nerve(s), as scored under the dissecting stereomicroscope in G0 zebrafish at 72 hpf. Total numbers of fish in each group are given above the corresponding bar. Data are shown from three experimental replicates. Pearson's 2 × 3 χ² test with 2 degrees of freedom; for each of three experimental replicates, χ² = 129.19, P < 2.2e⁻¹⁶; χ² = 74.61, P < 2.2e⁻¹⁶; χ² = 52.5, P = 4.0e⁻¹². (E) Kaplan-Meier survival curves demonstrating the relative survival probabilities of scramble-injected (blue line) and guide-injected (pink line) zebrafish over the first 72 hours of life. Number at risk below the plot provides the counts of surviving embryos in each group taken every 24 hpf over a 72 hpf period. Relative survival probabilities of sema3fa-targeting and scrambled gRNA-injected embryos were compared by the log-rank test. Displayed data were derived from a single experimental replicate, but measurements were taken for three experimental replicates, all of which showed the same trend. (F, G) Representative images from wild-type or F2 germline sema3fa^−/− (c.869_873delTGAGA, p.(Glu290GlyfsTer8)) mutants (arrows and stars as per A–C).

G0 mosaic and F2 germline zebrafish targeting of olig2. (A–H) Initial olig2 targeting experiments were performed in G0 embryos. Representative images obtained by stereomicroscope (A–D) and confocal microscopy (E–H) for Uninjected (A, E), Scrambled guide injected (B, F), or olig2-targeting Guide-injected (C–D, G–H) G0 HGj4A zebrafish at 72 hpf. Two images are provided for guide-injected fish, to demonstrate variability in G0 targeting outcomes. Stereomicroscope images for each treatment group correspond to the confocal images obtained from the same fish in the panel below. Motor neurons in the abducens nucleus (dashed white circled region) appear absent by stereomicroscopy (C, D) or variably reduced (G, H) by confocal microscopy, and abducens nerves appear thin (G;blue arrow) or absent (G, H; red arrows) compared to normal (E, F; white arrows) by confocal microscopy. (I) Barplots showing the percentages of zebrafish exhibiting wild-type-like (WT-like), malformed, or absent CN6 motor neuron nuclei, as scored under the dissecting stereomicroscope in G0 zebrafish at 72 hpf. Total numbers of fish in each group are given above the corresponding bar. Data are shown from three experimental replicates. Pearson's 3 × 3 χ² test with 4 degrees of freedom; χ² = 114.34, 95.77, 102.24 for each of three experimental replicates; P < 2.2e⁻¹⁶. (J) Kaplan-Meier survival curves demonstrating the relative survival probabilities of scramble-injected (blue line) and guide-injected (pink line) zebrafish over the first 72 hours of life. Number at risk below the plot provides counts of surviving embryos in each group taken every 24 hpf over a 72 hpf period. Relative survival probabilities of olig2-targeting and scrambled gRNA-injected embryos were compared by the log-rank test. Displayed data were derived from a single experimental replicate, but measurements were taken for three experimental replicates, all of which showed the same trend. (K, L) Representative images from wild-type or F2 germline olig2^−/− mutants (c.496_499delAGTT, p.(Ser166ValfsTer93)); key as noted for A–H).

G0 mosaic and F2 germline zebrafish targeting of frmd4bb. (A–E) Initial frmd4bb targeting experiments were performed in G0 embryos. Representative images obtained by confocal microscopy (A–C) of uninjected (A), scrambled guide-injected (B), or frmd4bb-targeting guide-injected (C) G0 HGj4A zebrafish at 72 hpf. Motor neurons in the abducens nucleus (dashed white circled region) have reduced dispersion through the columns of the abducens nuclei (C), and abducens nerves appear stalled and fail to target the lateral rectus muscles (C;white arrows and yellow stars) compared to uninjected and scrambled (A, B;white arrows and white stars). (D) Barplots showing the percentages of zebrafish exhibiting wild-type-like (WT-like) or malformed CN6 motor neuron nuclei, as scored under the dissecting stereomicroscope in G0 zebrafish at 72 hpf. Total numbers of fish in each group are given above the corresponding bar. Data are shown from three experimental replicates. Pearson's 2 × 3 χ² test with 2 degrees of freedom; for each of three experimental replicates, χ² = 52.6, P < 2.2e⁻¹⁶; χ² = 61.8, p < 2.2e⁻¹⁶; χ² = 34.8, P = 3.0e⁻⁸. (E) Kaplan-Meier survival curves demonstrating the relative survival probabilities of scramble-injected (blue line) and guide-injected (pink line) zebrafish over the first 72 hours of life. Number at risk below the plot provides counts of surviving embryos in each group taken every 24 hpf over a 72 hpf period. Relative survival probabilities of frmd4bb-targeting and scrambled gRNA-injected embryos were compared by the log-rank test. Displayed data were derived from a single experimental replicate, but measurements were taken for three experimental replicates, all of which showed the same trend. (F, G) Representative images from wild-type or F2 germline frmd4bb^−/− mutants (c.1687_1697delGATGAAATGAA, p.(Asp563ProfsTer25)).

The oCCDD proband-derived candidate variants in transcription factors disrupt DNA binding. (A–C) Two-dimensional structural mapping of human variants. (A) Variants in PHOX2A associated with CFEOM. (B) Variants in MAFB associated with DRS and/or co-occurring phenotypes. (C) Variant in OLIG2 associated with DRS. References for Figure 7 are provided in Supplementary Methods and Results. Asterisks represent variants above schematics that were identified and reported in our sequenced human oCCDD cohort¹ and are functionally tested for the first time in this work: PHOX2A p.(Trp137Cys), MAFB p.(Glu223Lys), and OLIG2 p.(Arg156Leu). Variants below schematics are previously reported as follows: (A) Filled circles represent previously reported PHOX2A variants associated with CFEOM: IVS1, G>A, +1 (c.217+1G>A)²; IVS2, G>A, −1(c.406-1G>A)^2,3; p.(Ala72Val)^2,3; p.(Asn76Lys)⁴; p.(Gln90Ter).⁵ (B) -Previously reported MAFB variants associated with DRS and colored as follows: blue denotes isolated DRS⁶; magenta denotes DRS ± hearing impairment ± intellectual disability^6,7; black denotes DRS + focal segmental glomerulosclerosis ± hearing impairment⁸; green denotes focal segmental glomerulosclerosis ± DRS⁹; purple denotes contiguous gene deletions with variable neurodevelopmental anomalies ± DRS.¹⁰ Variants are mapped using the following transcripts: ENST00000298231.5 (PHOX2A), ENST00000373313.3 (MAFB), and ENST00000382357.4 (OLIG2). (D–F) Transcription factor binding site motif logos for PHOX2A (D), MAFB (E), and OLIG2 (F). Top: logos from the JASPAR database of transcription factor binding profiles derived from high-throughput sequencing SELEX (HT-SELEX) experiments for either the wild-type human protein (PHOX2A and OLIG2) or orthologous mouse protein (Mafb). Middle and bottom: Protein binding microarray (PBM) experiment motifs derived from universal protein binding microarrays for the reference DNA binding domain (PBM REF, middle) and for the mutant DNA binding domain (PBM MUT, bottom). (G–I) For 8-mers resembling the wild-type motif, E-score comparison between reference and mutant DNA binding domain for PHOX2A (G), MAFB (H), and OLIG2 (I). Red dots correspond to 8-mer sequences that contain the labeled International Union of Pure and Applied Chemistry code-based k-mers, which are selected to closely resemble each motif logo. Y, C or T nucleotides; R, A or G nucleotides; W, A or T nucleotides; M, A or C nucleotides; K, G or T nucleotides; N, any base. (J–L) The 8-mer E-score comparison for the k-mers labeled in G–I across the replicate protein binding microarrays. (J) Reference versus p.(Trp137Cys) PHOX2A variant. PHOX2A p.(Trp137Cys) led to a significant drop in E-scores for 8-mers resembling the wild-type motif (P < 10⁻⁹; one-sided Mann-Whitney U test) to a level indistinguishable from E-scores of the GST negative control (P > 0.9; two-sided Mann-Whitney U test). (K) Reference versus p.(Glu223Lys) MAFB variant. The E-scores for the 8-mers recognized by the wild-type MAFB showed significantly lower values for the mutant DNA binding domain (P < 10⁻⁷; one-sided Mann-Whitney U test), but the E-score distribution for the mutant was still higher than that of the GST-negative control (P < 10⁻⁶; two-sided Mann-Whitney U test), suggesting partial loss of binding. (L) Reference versus p.(Arg156Leu) OLIG2 variant. OLIG2 p.(Arg156Leu) led to a significant reduction in mutant E-scores for 8-mers recognized by wild-type OLIG2 (P < 10⁻¹⁰; one-sided Mann-Whitney U test) to a level similar to that of the GST-tagged negative control (P > 0.1; two-sided Mann-Whitney U test). Orange, reference protein; gray, mutant/nonreference protein; yellow, GST-tagged negative control. * P < 1 × 10⁻⁷; one-sided Mann-Whitney U test. ** P < 1 × 10⁻⁶; two-sided Mann-Whitney U test. BHLH, basic helix-loop-helix domain; BZIP, basic leucine zipper domain; CFEOM, congenital fibrosis of the extraocular muscles; CTRL, GST-negative control protein; DRS, Duane retraction syndrome; GS, glutathione S-transferase tagged negative control protein; HD, homeodomain; MUT, mutant/nonreference; N-ter, N-terminal region; PBM, protein-binding microarray; REF, reference (nonmutant) sequence; rep, replicate.

The oCCDD pedigrees with functionally validated candidate genes/variants and brain MR images of probands with homozygous PHOX2A and SEMA3F variants. (A, G, H, T, U) Schematics of pedigrees segregating variants of uncertain significance in known or novel candidate oCCDD genes. (B−F) Images from a 1.5T Siemens brain MRI of the pedigree 160 proband at four months of age that has CFEOM and harbors a homozygous PHOX2A variant. (B) Sagittal T1 fluid–attenuated inversion recovery 4 mm–thick image reveals abnormal anatomy of the corpus callosum with a somewhat down-slanted posterior body and splenium (long yellow arrow). (C) Axial turbo spin echo (TSE) T2-weighted 4 mm–thick image shows diminutive medial rectus muscles (short double-headed white arrow). (D) Axial TSE T2 weighted 4 mm thick image at the level of the midbrain and interpeduncular cistern does not show oculomotor nerves. (E, F) Coronal TSE T2-weighted 4 mm–thick images with and without fat-suppression show asymmetric positioning of the optic nerves, higher on the right (long white arrow), diminutive medial rectus muscles (short white arrows), and small superior oblique muscles (short yellow arrows). Because of slice thickness and slice angle, the superior rectus muscles were not as readily assessed but also appeared slightly small. (I–S) The ENG_CMK proband that has CFEOM and harbors a homozygous SEMA3F had MR imaging at 12 months of age obtained on a 3T Siemens Skyra (I–M, O–S) and three months of age obtained on a 1.5T Siemens unit (N) variant. I, Midline sagittal T1 magnetization prepared rapid gradient echo (MPRAGE) 1 mm–thick image demonstrates abnormal anatomy of the corpus callosum with a somewhat down-slanted posterior body and splenium (long white arrow). (J) Axial 1 mm reformatted MPRAGE and (K) 2.5 mm–thick TSE T2-weighted images show a small protuberance off or along the right side of the tectum (short white arrows) that appears hypointense on T1 and heterogeneous on T2 with central high-signal and peripheral low-signal intensity. This lesion is of uncertain etiology but was present in retrospect on an examination the previous year. (L) Axial 2.5–mm thick TSE T2-weighted image demonstrating a diminutive anterior commissure (long yellow arrow) and slight underdevelopment of the right frontal and temporal opercula (short yellow arrow). (M) Reformatted 1 mm coronal MPRAGE image shows asymmetry of the hippocampal formations and medial temporal lobes with the right side appearing mildly misshapen (yellow asterisk). N, Axial 2.5 mm–thick T2-weighted image at the level of the midbrain does not show oculomotor nerves at the level of the interpeduncular cistern; O, Axial 0.44 mm–thick T2 Sampling perfection with application optimized contrasts using different flip angle evolution (SPACE) image shows that the cisternal segments of the vestibulocochlear nerves are present but small (long white arrows) as they course posterior and parallel to the facial nerves. (P, Q) T2 SPACE images through the inner ears show dysmorphic, thickened cochlear modioli with stenotic cochlear apertures, more pronounced on the left (short yellow arrows). The apices of the cochleae appear mildly flattened. (R) Sagittal oblique T2 space MRI shows marked stenosis of the left internal auditory meatus (white circle) with only one cranial nerve visible instead of the expected four (facial, cochlear, and superior and inferior vestibular nerves). (S) Coronal 5 mm–thick T1-weighted image of the brain at the level of the orbits shows small medial rectus (short white arrows) and inferior rectus muscles (short yellow arrows). The superior rectus muscles are likely also small but are suboptimally assessed because of slice thickness.