phox2a loss-of-function mutants fail to develop nIII/nIV motor neurons and vertical eye rotation behavior.

Associated with Figure S1.

(A) Schematic of vestibulo-ocular reflex circuitry and the genetic loss-of-function approach used to perturb motor-derived signals.

(B) Fluorescent in situ hybridization showing phox2a transcript expression in statoacoustic ganglion sensory afferents (left), central projection neurons in the tangential nucleus (middle), or nIII/nIV extraocular motor neurons (right) at 5 days post-fertilization (dpf). Top: probe only, nuclei outlined with dashed lines. Bottom: probe (green) merged with somata, labeled by Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede) (sensory, central) or Tg(isl1:GFP) (motor).

(C) Schematic of CRISPR/Cas9 mutagenesis approach. Top: Red star shows location of guides against phox2a DNA. Bottom: RNA sequence in wildtype and phox2ad22 alleles. Red dashed lines show deleted sequence; “STOP” box shows predicted premature stop codon due to deletion. Right shows predicted protein sequence.

(D) Transmitted light image of a 5 dpf wildtype (top) and phox2a null mutant (bottom). White arrows point to a normally inflated (top) or absent (bottom) swim bladder.

(E) Images of nIII/nIV motor neurons in one hemisphere, labeled by Tg(isl1:GFP), in wildtype siblings (left) and phox2a null mutants (right) at 5 dpf. Scale bar, 20 µm.

(F) Quantification of the number of Tg(isl1:GFP)+ neurons in nIII/nIV from N=6 wildtype siblings and N=10 phox2a null mutants.

Motor neurons are dispensable for proper connectivity between utricular sensory afferents and projection neurons.

Associated with Table 1.

(A) Schematic of pitch vestibulo-ocular reflex circuitry. Dashed lines outline projection neurons as calcium imaging target. Nose-down/eyes-up channel represented with blue; orange, nose-up/eyes-down.

(B) Schematic of tonic pitch-tilt stimulus delivered with Tilt-In-Place Microscopy (TIPM). Shaded regions show calcium imaging windows when fish were oriented horizontally immediately following tilts. Inset shows timecourse of the rapid step to restore horizontal position after tilts. Imaging experiments used larvae from the Tg(isl1:GFP);Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:GCaMP6s) line.

(C) Proportion of subtypes observed in sibling controls and phox2a null mutants. Blue: nose-down. Orange: nose-up. Grey: Neurons without directional tuning (criteria in Methods).

(D/E) Soma position of nose-down (blue) and nose-up (orange) neurons in sibling controls (left) and phox2a null mutants (right). Soma size scaled by strength of calcium response (ΔFF), normalized by max observed ΔFF.

(F/I) Heatmaps showing example tilt responses from nose-down (F) or nose-up (I) neurons in sibling controls (top) and phox2a null mutants (bottom). n=10 neurons with strongest ΔFF responses to tilts shown. Each row shows an individual neuron. Shaded bars show calcium imaging window immediately following restoration from eccentric position. Black arrow points to first second of tilt response used for analyses.

(G/J) Distributions of maximum ΔFF responses to tilts for nose-down (G) or nose-up (J) neurons in sibling controls (black) and phox2a null mutants (red). Solid and shaded lines show mean and standard deviation, respectively, of bootstrapped data (Methods)

(H/K) Distributions of directional tuning score to tilts for nose-down (H) or nose-up (K) neurons in sibling controls (black) and phox2a null mutants (red). Tuning score ranges from 0 (equal responses to both tilt directions, no tuning) to 1 (responses to one tilt direction only); criteria detailed in Methods. Solid and shaded lines show mean and standard deviation, respectively, of bootstrapped data.

Statistical comparisons of tilt responses across genotypes.

WT (sampled) refers to an n=125 neuron subset, sampled with replacement from a reference dataset of wildtype projection neurons. Data shown is mean/standard deviation unless otherwise noted. p val generated from a one-way ANOVA with multiple comparisons. Associated with Figure 2 and Figure 3.

Motor neurons are dispensable for proper connectivity between semicircular canal sensory afferents and projection neurons.

Associated with Table 1.

(A) Schematic of impulse tilt experiments. Yellow dashed lines outline projection neurons as calcium imaging target. Impulse-responsive neurons (ventrally-localized) shown with purple; unresponsive neurons, grey.

(B) Schematic of impulse stimuli delivered with TIPM. Shaded regions show calcium imaging windows at horizontal immediately following impulses. Inset shows timecourse of impulse stimulus. Imaging experiments used larvae from the Tg(isl1:GFP);Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:GCaMP6s) line.

(C) Proportion of impulse-responsive (purple) and unresponsive (grey) neurons observed in sibling controls and phox2a null mutants.

(D) Soma position of impulse-responsive neurons in sibling controls (left) and phox2a null mutants (right). Soma size scaled by strength of calcium response (ΔFF), normalized by max observed ΔFF.

(E) Heatmaps showing example impulse responses from neurons in sibling controls (left) and phox2a null mutants (right). n=10 example neurons shown. Each row shows an individual neuron. Shaded bars show calcium imaging window immediately following impulse delivery. Black arrow points to first second of tilt response used for analyses. Note smaller scale (0-0.75) of impulse responses relative to Figures 2F and 2I.

(F) Distributions of maximum ΔFF responses to impulses in sibling controls (black) and phox2a null mutants (red). Solid and shaded lines show mean and standard deviation, respectively, from bootstrapped data.

(G) Distributions of directional tuning score to impulses in sibling controls (black) and phox2a null mutants (red). Tuning score ranges from 0 (equal responses to both tilt directions, no tuning) to 1 (responses to one tilt direction only); criteria detailed in Methods. Solid and shaded lines show mean and standard deviation, respectively, from bootstrapped data.

Projection neurons are anatomically and molecularly poised to assemble with motor neuron partners in phox2a mutants.

(A) Schematic of retrograde photofill experiments. Projection neuron axons expressing the photolabile protein Kaede are targeted at the midbrain-hindbrain boundary with ultraviolet light. Converted protein (magenta) retrogradely diffuses to the soma, while the unconverted nucleus remains green.

(B) Projection neuron somata in sibling controls (left) and phox2a null mutants (right) after retrograde photolabeling. Experiments performed at 5 dpf. Neurons visualized in Tg(isl1:GFP);Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:E1b-Kaede).

(C) Projection neuron axons at the hindbrain (inset) and midbrain-hindbrain boundary in sibling controls (top) and phox2a null mutants (bottom). Axons visualized using Tg(isl1:GFP);Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:E1b-Kaede). White dashed outline shows arborization fields in nIII/nIV. MHB and yellow dashed line, midbrain-hindbrain boundary. nucMLF: nucleus of the longitudinal fasciculus. Inset: Spatial segregation between early-born (magenta+green) and late-born (green only) axons. White dashed line reflects separation between dorsal (nose-up, early-born) and ventral (nose-down, late-born) axon bundles. Image at 5 dpf in sagittal view.

(D) Projection neuron axon bundle in a phox2a null mutant at 3 dpf. White arrows point to single collateral to two remaining nIII/nIV neurons.

(E) Fluorescent in situ hybridization against RNA for three pre-synaptic markers: synaptophysin a (sypa; left), synaptic vesicle glycoprotein 2 (sv2, middle), and synapsin I (syn1, right). Top row, sibling controls. Bottom row, phox2a null mutants. For each panel set, left images show in situ probe expression (green) and right images show merge with projection neurons labeled in Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:E1b-Kaede). Dashed lines outline the projection nucleus. Cell and transcript expression outside the projection nucleus is removed for visual clarity. Images taken at 5 dpf in sagittal mount. All scale bars, 20 µm.

Motor neurons are dispensable for normal transcriptional profiles of projection neurons.

Associated with Figure S2-Figure S5, Table 2.

(A) Schematic of sequencing approach. Central projection neurons (Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:E1b-Kaede)) are harvested from 3 dpf larvae. Flow cytometry is used to exclude neurons not labelled by Tg(-6.7Tru.Hcrtr2:GAL4-VP16). Bulk RNA sequencing is performed to compare the profiles of projection neurons in siblings and phox2a null mutants.

(B) Example of projection neurons before (left) and after (right) harvesting. Neurons visualized with Tg(isl1:GFP);Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS:E1b-Kaede). Dashed lines outline projection neurons in the tangential nucleus; dotted lines, medial vestibular nucleus. Yellow region shows margin of harvesting error: non-projection neurons that may be included in bulk sequencing dataset.

(C) Number of differentially expressed genes in projection neurons at 3 dpf after applying progressive filters based on gene expression in a reference single-cell dataset. Data shown on logarithmic scale. Solid, dashed, and dotted lines represent differentially-expressed gene with p adjusted<0.5, p adjusted<0.01, or p adjusted<0.001 significance, respectively.

(D) Volcano plot showing differentially expressed genes in projection neurons between control and phox2a null larvae at 3 dpf. Dashed lines represent significance cutoffs: horizontal line, p adjusted>0.05; vertical line, Log2 Fold Change > 2.0. Each circle is a gene. Genes to the left and right of 0 on the horizontal axis show downregulated and upregulated genes, respectively. Colors indicate percent of reference cells that express a given gene. Grey-colored genes are below both significance thresholds.

(E) Same data as Figure 5D. Colored genes show eight candidates evaluated with fluorescent in situ hybridization: red, upregulated; blue, downregulated; yellow, highly-expressed controls (evx2.

(F) Fluorescent in situ hybridization against candidate genes that met projection neuron filter criteria. Top row shows sibling controls; bottom row, phox2a null mutants. For each gene, left panels show RNA probe (green) and right panels show merge with projection neurons labeled by Tg(-6.7Tru.Hcrtr2:GAL4-VP16) (grey). Dashed lines outline the projection nucleus. Cell and transcript expression outside the projection nucleus is masked for visual clarity. Arrows denote whether genes are upregulated (red), downregulated (blue), or not significantly changed (yellow). Percentage refers to fraction of cells in a single-cell RNA sequencing reference atlas (Methods) with detected transcript. Candidates: itga9 (log2 fold change=23.0, p adj.=3.9×10−6), twf1b (log2 fold change=5.9, p adj.=0.024, p4hb (log2 fold change=5.1, p adj.=0.04), mapk6 (log2 fold change=5.1, p adj.=0.06), rxfp2a (log2 fold change=-8.5, p adj.=1.1×10−5), bckdhbl ((log2 fold change=-9.1.0, p adj.=0.001) satb1a ((log2 fold change=-3.0, p adj.=0.001), evx2 (log2 fold change=0.46, p adj.=0.99). All scale bars, 20 µm.

Differentially expressed genes in projection neurons.

Star indicates a gene was evaluated using fluorescent in situ hybridization. # symbol indicates a gene was also differentially expressed in adjacent medial vestibular neurons (see Figure S6). “% of projection neurons with expression” refers to detection in a filtered subset of projection neurons from a single-cell reference atlas of neurons labeled in Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede) (Methods, Figure S3). Putative origin inferred from gene expression in the annotated 10x dataset (Methods, Figure S3)). Genes sorted by p adjusted value. Data associated with Figure 5.

Top 50 expressed genes in an unfiltered bulk RNA sequencing dataset of phox2a siblings.

“% of unfiltered 10x neurons” refers to gene detection in a single-cell atlas of neurons labeled in Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede) (n=1,468 neurons). “% of projection neurons” refers to gene detection in a subset of the single-cell atlas containing projection neurons in the tangential nucleus (n=473 neurons). Data associated with Figure 5.

Top 50 differentially expressed genes in an unfiltered bulk RNA sequencing dataset of phox2a siblings and null mutants.

One star indicates a gene was retained in a filtered subset of projection neurons; %, evaluated using fluorescent in situ hybridization. “% of unfiltered 10x neurons” refers to gene detection in an unfiltered single-cell reference atlas of neurons labeled in Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede) (n=1,468 neurons). Putative origin inferred from gene expression in the annotated 10x dataset (Methods, Figure S3)). Genes sorted by p adjusted value. Data associated with Figure 5.

Key Resources Table (associated with Methods).

phox2a specifies nIII motor neuron fate in a dose- and birthdate-dependent manner.

Associated with Figure 1.

(A) Images of nIII/nIV motor neurons, labeled in Tg(isl1:GFP), in wildtype siblings (left) and phox2a heterozygotes (middle) at 5 dpf. Wildtype image same as in Figure 1E. One hemisphere shown. White dashed lines outline the dorsal extent of nIII, which contains inferior rectus and medial rectus neurons 33. Scale bar, 20 µm.

(B) Location of the earliest-born neurons in nIII/nIV (left, magenta) against all nIII/nIV neurons labeled in Tg(isl1:Kaede) (right, grey). Larvae birthdated at 34 hpf (Methods). One hemisphere shown. White dashed lines outline the dorsal extent of nIII. Scale bar, 20 µm.

(C) Quantification of the number of Tg(isl1:GFP)+ neurons in nIII/nIV from N=6 wildtype siblings (grey) and N=8 phox2a heterozygotes (teal). Wildtype data same as Figure 1F.

(D) Distributions showing probability of nIII/nIV soma location across each spatial axis in wildtype (black) and heterozygous (teal) phox2a larvae. Solid and shaded lines show mean and standard deviation, respectively, from bootstrapped data. Data from same fish quantified in Figure S1C. ns, not significant; star, significant at the p<0.001 level.

Flow cytometry gating strategy to sort fluorescently-labeled neurons for bulk RNA sequencing.

Associated with Figure 5.

(A) Sequential gates used to sort fluorescent neurons labeled with Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede);Tg(isl1:GFP). Gate A excluded presumptive debris (small cells). Gate B isolated single cells and excluded large cells and doublets. Gate C excluded DAPI+ (dead or unhealthy) neurons. Gate D isolated fluorescent (GFP or Kaede+) neurons; neurons in this gate were sorted. Gates were set using negative controls (not shown; Methods). Gates shown for one of four experimental repeats.

Molecular identification of projection neurons using a reference single-cell RNA sequencing atlas.

Associated with Figure 5.

(A) UMAP visualization of a single-cell RNA sequencing atlas of n=1,468 neurons labeled in Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede), generated with 10x Genomics (Methods). Each circle is a single neuron. Neurons are clustered (colors) according to their transcriptional identity. Annotations are based on validated marker genes (data not shown). TAN, tangential nucleus; MVN, medial vestibular nucleus; r, rhombomere; MNs, motor neurons; inhib, inhibitory neurons.

(B) Heatmap showing genes unique to each annotated cluster. Each row is a gene; names unlisted for clarity. Columns show distinct clusters. Color bar on top reflects clusters in Figure S3A. Yellow and purple reflect stronger or weaker gene expression, respectively.

(C) Fluorescent in situ hybridization against three markers (pou4f1, penkb, chrna2b) that are negative for tangential nucleus projection neurons and positive for medial vestibular nucleus neurons. Top row shows RNA expression (green); bottom row, merge with neurons labeled in Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede). Dashed lines outline the tangential nucleus (TAN) and medial vestibular nucleus (MVN). Data from 72 hpf larvae. Images shown in an axial view.

(D) Fluorescent in situ hybridization against a positive marker (evx2) for both tangential nucleus and medial vestibular nucleus neurons. All scale bars, 20 µm.

(E) Heatmap showing genes unique to tangential and medial vestibular neurons. Clusters identified using positive and negative fluorescent in situ data from Figure S3C-Figure S3D and unpublished data.

Visualization of transcripts in siblings and phox2a null mutants with fluorescent in situ hybridization is (1) consistent across larvae and (2) scales with predicted detection in projection neurons.

Associated with Figure 5.

(A-A’) Fluorescent in situ hybridization against itga9 for three sibling (A) or phox2a null mutant (A’) larvae (72 hpf), imaged with identical conditions. Left column shows RNA (green); right column, merge with projection neurons visualized with Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede) (grey). Dashed lines outline the projection nucleus. Cell and transcript expression outside the projection nucleus is removed for visual clarity. Percentage (1.9%) refers to fraction of cells in a single-cell RNA sequencing reference atlas (Methods) with detected transcript. All scale bars, 20 µm.

(B-B’) Fluorescent in situ hybridization against htt, 17%, for three sibling (B) and phox2a mutant (B’) larvae (72 hpf).

(C-C’) Fluorescent in situ hybridization against evx2, 46%, for three sibling (C) and phox2a mutant (C’) larvae (72 hpf).

Differential gene expression in an unfiltered bulk sequencing dataset of siblings and phox2a mutants.

Associated with Figure 5.

(A) Volcano plot showing differentially expressed genes across an unfiltered bulk RNA sequencing dataset. Dashed lines represent significance cutoffs: horizontal line, p adjusted>0.05; vertical line, Log2 Fold Change > 2.0. Each circle is a gene. Genes to the left and right of 0 on the horizontal axis show downregulated and upregulated genes, respectively. Red color shows genes that are differentially expressed in a filtered subset of projection neurons (Figure 5). Grey-colored genes are below both significance thresholds.

(B) Same data as Figure S5A, now highlighting candidate genes evaluated by fluorescent in situ (Figure 5) with red. One candidate (yellow) that did not meet projection neuron filter criteria (Methods) is shown in Figure S5D; remaining candidates (included in filtered data) shown in Figure 5F.

(C) Same data as Figure S5A-Figure S5B, showing the number of differentially expressed genes at progressive significance thresholds (p adjusted). Red and blue lines show the number of significantly upregulated and downregulated genes, respectively.

(D) Fluorescent in situ hybridization against a candidate gene, slc22a7a (log2 fold change=10.2, p adj.=1.6×10−4), that did not meet projection neuron filter criteria. Percentage refers to fraction of projection neurons from a single-cell sequencing dataset with expression (Methods). Left columns show RNA (green); right columns, merge with projection neurons labeled with Tg(-6.7Tru.Hcrtr2:GAL4-VP16);Tg(UAS-E1b:Kaede) (grey). Dashed lines outline the projection nucleus. Cell and transcript expression outside the projection nucleus is removed for visual clarity. All scale bars, 20 µm.

phox2a expression in the medial vestibular nucleus may underscore differential gene expression phenotypes in bulk data.

Associated with Figure 5.

(A) Fluorescent in situ hybridization against phox2a in a 5 dpf larvae (axial view). Top panel shows phox2a RNA (green); bottom panel, merge with neurons visualized with Tg(isl1:GFP);Tg(-6.7Tru.Hcrtr2:GAL4-VP16(;Tg(UAS-E1b:Kaede) (grey). White dashed lines outline three nuclei of interest: projection neurons in the tangential nucleus (TAN), the medial vestibular nucleus (MVN), and the facial nucleus (nVII). All scale bars, 20 µm.

(B) Volanco plot showing differentially expressed genes in medial vestibular nucleus neurons between control and phox2a null larvae at 3 dpf. Dashed lines represent significance cutoffs: horizontal line, p>0.05; vertical line, Log2 Fold Change > 2.0. Each circle is a gene. Genes to the left and right of 0 on the horizontal axis show downregulated and upregulated genes, respectively. Colors indicate percent of reference medial vestibular neurons (Methods) that express a given gene. Grey-colored genes are below both significance thresholds.

(C) Same data as Figure S6B. Color shows genes that are differentially expressed in both medial vestibular nucleus neurons and projection neurons.