Diversity of Internal Sensory Neuron Axon Projection Patterns Is Controlled by the POU-Domain Protein Pdm3 in Drosophila Larvae

td neuron sensory dendrites project along specific tracheal branches

There are two adjacent td neurons in each abdominal hemisegment from A1-A6, named v′td1 and v′td2 for their location in the ventral′ cluster of sensory neurons, and one td neuron in A7 (Fig. 1A,B). Fitting with the naming of individual sensory neurons according to their ventral to dorsal position, v′td1 is the more ventral of the two neurons. We screened GAL4 and LexA lines from the InSITE (Gohl et al., 2011), Janelia (Jenett et al., 2012), and OK (courtesy of Dr. Brian McCabe, unpublished data) collections for drivers for peripheral sensory neurons, and identified several lines that labeled td neurons (Table 1). GAL4 and LexA lines typically showed expression in characteristic subsets of td neurons, suggesting that td neurons may have diverse identities and functions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

td sensory dendrites associate with tracheal branches. A, Schematic of larva showing segmentally repeated td cell bodies (green) relative to the tracheal network. Blue represents only major tracheal branches. B, Schematic of v′td1 and v′td2 neurons in one hemisegment, showing cell bodies (green circles), dendrites (solid green lines), and axons (dashed green line), and association with specific tracheal branches (blue). C, Dual-color labeling of v′td1 (C′; labeled by R35B01-LexA>mCD8::GFP) and v′td2 (C″; labeled by 260-Gal4>mCherry) neurons in segment A3. D, Dendrite lengths and node count quantifications for v′td1 (labeled by R35B01-LexA) and v′td2 (labeled by 260-Gal4) neurons in segments A2–A4. Dendrite length was reduced in v′td2 relative to v′td1 in segment A2 (Bonferroni adjusted *p = 0.012, Mann–Whitney U test, U < 0.0001), but not in segments A3 (Bonferroni adjusted p = 1, U = 17) or A4 (Bonferroni adjusted p = 1, U = 16). Similarly, dendrite nodes were reduced in v′td2 relative to v′td1 in segment A2 (Bonferroni adjusted *p = 0.018, Mann–Whitney U test, U = 1), but not in segments A3 (Bonferroni adjusted p = 1, U = 16.5) or A4 (Bonferroni adjusted p = 0.558, U = 10). n = 6 animals. E, td dendrites project along tracheal branches. Coracle labeling shows septate junctions of tracheal epithelial cells. E′–E‴, Enlarged image of area boxed in E showing enlargements of td dendrites near tracheal fusion cell (yellow arrow). F, Electron micrographs of tracheal tube cross section showing tracheal cell (tc; blue), putative td dendrites (d; green), tracheal lumen (tl), and muscle (ms). F′, Higher-magnification image of boxed area in F, showing two neuronal processes closely associated with the tracheal cell, enclosed by basement membrane. Boxplots represent median (middle line) and 25th to 75th percentile, with whiskers extending to the most extreme data point within 1.5 times the interquartile range of the hinge. Scale bars: C–C″, E, 50 μm; E′–E‴, 10 μm; F, 2 μm; F′, 250 nm.

To characterize v′td1 and v′td2 dendrite morphologies, we used GAL4/LexA dual-color labeling. We resolved peripheral arbors with single-cell resolution and characterized their relationship to each other and to tracheal branches (Fig. 1C–E). We used antibody against Coracle, a septate junction resident protein, to visualize tracheal cell structure. td cell bodies reside next to the ganglionic branch of the larval tracheal system (nomenclature following Manning and Krasnow, 1993) and extend dendrites along the ganglionic branch and dorsally (Fig. 1C). Major dendrites bifurcate and extend along the lateral trunk (LT). Branching along the LT tubules was sparse; however, there were local areas of branching and dendrite enlargements, most consistently at tracheal fusion cells (Fig. 1E, arrow), which are located along LT where tracheal tubes fuse during development. Dendrites terminate either on the LT or lateral group branch, sometimes with enlargements near or at the termination (Fig. 1E″). We did not observe differences between the trajectories of v′td1 and v′td2 in abdominal segments A3 and A4, although v′td2 was relatively simplified in segment A2 (Fig. 1D). v′td1 and v′td2 dendrites were often fasciculated along the lateral tracheal branches. Dendrites were also tightly associated with tracheal branches along most of their length (Fig. 1E). Using electron microscopy, we identified fasciculated processes on the basal surface of tracheal cells, within a thin basement membrane (Fig. 1F). Thus, td neurons have mostly similar dendritic morphologies, although GAL4 expression patterns suggested distinct classes of td neurons. We therefore examined td neuron diversity as manifest in axon projection patterns.

td neuron axons terminate in distinct regions of the VNC and the subesophageal zone (SEZ)

Sensory neuron diversity is reflected in axon projection patterns (Merritt and Whitington, 1995; Vosshall et al., 2000; Grueber et al., 2007; Kwon et al., 2014), so we examined td axons in greater detail using the collection of GAL4 and LexA lines. Although functions for td neurons are not known, they have been proposed to function as sensors of tracheal status or as proprioceptors based on the location of their dendrites (Bodmer and Jan, 1987). In the fly CNS, the terminal location of axon projections correlates with sensory modality. Within the VNC, nociceptive and touch sensing axons project to ventromedial regions of the neuropil, proprioceptors to more dorsal neuropil, and vibration sensing chordotonal axons to intermediate neuropil (Merritt and Whitington, 1995; Grueber et al., 2007; Zlatic et al., 2009). These somatosensory axons also project primarily within the same VNC neuromere after they enter the CNS (Fig. 2A) (Merritt and Whitington, 1995). By contrast, in embryonic stages, td axons from each segment project together along a far-lateral VNC tract, shift to a medial tract when they reach the thoracic neuromere, and extend further anteriorly (Merritt and Whitington, 1995), although the final targets and mature td axon patterns in the third instar larva have not been described. We labeled a subset of td neurons using R31D10-LexA>13XLexAop2-mCD8-GFP, and visualized their axon projections in the third instar CNS with major Fasciclin 2 (Fas2)-positive axonal tracts as landmarks (Landgraf et al., 2003; Grueber et al., 2007). After entering the VNC, td axons project anteriorly along the ventrolateral tract to the T3 neuromere (Fig. 2A′,B). At T3, axons turned medially and dorsally to the dorsomedial fascicle. Thus, td axons make a series of shifts along both the mediolateral and dorsoventral axes as they project to anterior parts of the CNS. Based on these patterns, td axon projections are different from all other characterized types of sensory neurons that innervate the body of the larva (Fig. 2A,A′).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

td axons project to the VNC and the SEZ. A, Schematic of CNS showing axon trajectories of typical abdominal mechanosensory neurons (A), and td neurons (A′). A′, Dashed line and question mark indicates unresolved terminal trajectory of td axons. Purple band indicates T3 neuromere. B, td axons labeled by R31D10-LexA>GFP fasciculate and project anteriorly in the CNS (outlined in white). Roman numerals and dotted cyan lines indicate position of cross sections on the right, showing that td axons initially travel anteriorly along the ventrolateral fascicle (Bi), then shift to the dorsomedial fascicle (Biv). C, Two subtypes of td axon projections to the VNC and the SEZ. A subset of td neurons labeled by R31D10-LexA>mCD8::GFP targets the SEZ (C). Another subset of td neurons labeled by R22C07-Gal4>mCherry targets the VNC (C′). Tsh-Gal80 was used to suppress VNC expression in the R22C07-Gal4 driver. R22C07-Gal4 also labels other sensory neurons from the head that project axons to the SEZ (asterisk). D, VNC- and SEZ-targeting axons from the same segment initially fasciculate with each other as they enter the VNC (Di) but segregate into separate bundles in the ventrolateral tract (Dii). Roman numerals and dotted cyan lines indicate position of cross sections on bottom. E, SEZ-targeting td axons (green) terminate in close proximity to axon terminals of gustatory sensory neurons (magenta). F, G, Presynaptic sites of VNC-targeting td axons (F) and SEZ-targeting td axons (G) labeled by brp.short.cherry reveal en passant synapses along the td axon bundle tract (labeled by GFP in green). Scale bars: B, C–C″, F, G, 50 μm; D, E, 20 μm.

Prior studies on embryos showed that some td axons extended anteriorly beyond the T3 medial turn (Merritt and Whitington, 1995), but axon terminal locations were not known. To determine their ultimate targets, we examined td axon projections using GAL4 and LexA drivers that labeled subsets of td neurons. The subset of td neurons labeled by R31D10-LexA extended axons beyond the VNC to target the SEZ in the brain (Fig. 2B,C). By contrast, a different subset of td neurons labeled by R22C07-Gal4 extended axons only to the T3 neuromere in the VNC (Fig. 2C′). Colabeling of both populations revealed that despite their different targets, VNC and SEZ targeting td axons project along the same initial path in the VNC (Fig. 2C″). Thus, td sensory neurons can be segregated into at least two classes based on axon projections that remain entirely within the VNC or project anteriorly to the SEZ.

We asked whether SEZ and VNC td axons have different patterns of association or fasciculation as they project through the VNC to their targets, which could help to explain distinct targeting. Notably, although SEZ-targeting and VNC-targeting td axons fasciculate as they enter the VNC from the periphery, once they arrive in the ventrolateral tract axons segregate according to their final targets, such that SEZ-targeting td axons group together and VNC-targeting td axons similarly group together (Fig. 2D). These data suggest that td axons show hierarchical axon–axon associations as they target to the SEZ and VNC. Initial associations between axons from the same segment may be suppressed as axons encounter other “like-type” axons, which segregates cohorts into distinct subtype-specific bundles en route to their terminal targets.

Given the projection of td axons to the SEZ, we asked how their terminal targeting relates to gustatory receptor (GR) neuron axons, which strongly innervate the SEZ. Colabeling of td neurons (with R31D10-LexA) and gustatory sensory neurons (with Gr32a-Gal4) showed that SEZ-targeting td axons terminate in a similar location as gustatory sensory axons (Fig. 2E). Compartmentalization of the SEZ is still poorly understood so we cannot eliminate modality-specific fine scale axon organization; nevertheless, the terminal location of these projections raise the possibility of chemosensory roles for td neurons.

td neurons could conceivably make synaptic contacts with targets in an en passant fashion along their entire length, or only in axon terminal regions. To determine whether SEZ and VNC td neurons might have entirely distinct or potentially similar synaptic targets, we expressed the active zone marker Brp.short in different td neuron classes. VNC td neurons labeled by R22C07-Gal4 showed presynaptic zones in VNC neuropil along the ventrolateral tract and near the dorsomedial fascicle (Fig. 2F). Notably, SEZ td neurons labeled by R31D10-LexA showed Brp.short labeling at the axon terminals in the SEZ, and also along the length of the axons in the VNC (Fig. 2G). These data are consistent with presence of en passant synapses along the length of td axons.

Based on these axon projections and putative presynaptic sites, we identify at least two distinct projection patterns, patterns of axon–axon association, and output sites, for td neurons to the VNC and SEZ. td neuron circuitry appears to be distinct from other known classes of mechanoreceptors, proprioceptors, and nociceptors but may share neuropil targets with other chemosensory neurons.

td neurons express GAL4 drivers for chemosensory receptors

Given that td axons project to chemosensory recipient regions of the SEZ, we examined whether td neurons express chemosensory receptor genes. A large battery of GR genes have been identified and the expression patterns of many of them can be assessed using available GAL4 lines. A previous analysis of the expression patterns of 68 GRs using Gr-Gal4 drivers identified 7 GRs that were expressed outside of larval head taste organs (Kwon et al., 2011). We found that 4 of these Gr-Gal4s are expressed in td neuron subsets (Fig. 3; Table 1). In Drosophila, functional GR complexes require the expression of multiple GR subunits. Almost all GR-expressing td neurons show expression of multiple GR drivers, suggesting GR expression in td neurons could be functionally relevant. We considered the caveat that Gr-Gal4 promoter fusions may not reflect GR expression in td neurons. Arguing against this possibility, a Gr33a^Gal4 knock-in line (Moon et al., 2009) showed expression in td neurons in the same pattern as the promoter fusion. In addition to GRs, the IR20a clade of ionotropic receptors (IRs) are proposed to be gustatory and pheromone receptors (Bray and Amrein, 2003; Koh et al., 2014; Joseph and Carlson, 2015). A previous analysis of the expression patterns of these IR drivers identified IR56a-Gal4 expression in neurons associated with the trachea (Stewart et al., 2015). We confirmed the expression of IR56a in a specific subset of td neurons (Table 1). We also identified another IR driver, Ir76b-Gal4, that is expressed in a subset of td neurons. Overall, these data suggest that chemosensory receptor genes are expressed in td neurons.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

GR and IR GAL4 reporters label td neurons. A–F, CNS images showing expression of Gr-Gal4 and Ir-Gal4 lines that label td neurons. A′–F′, Same images dual-labeled with R31D10-LexA, which labels a subset of td neurons. In addition to td neurons, most of the GAL4 lines labeled gustatory neurons from the head (A–D, F), and a few lines labeled terminal sensory neurons (A, D), cIV sensory neurons (B), external sensory neurons (F), or CNS neurons (B, E). Scale bars, 50 μm.

td axon projections and GR/IR GAL4 driver expression correlate with cell identity and segment of origin

The GAL4 lines we identified label cohorts of fasciculated td axons, obscuring single-axon projection patterns. To determine the axonal morphologies of all individual td neurons, we generated Flybow and MCFO clones (Hadjieconomou et al., 2011; Nern et al., 2015) (Fig. 4A–M). We examined multiple clones from each of the 13 v′td neurons in the larva (two neurons per hemisegment from A1 to A6, and one neuron per hemisegment in A7). Clonal analysis revealed two main subtypes of v′td neurons with different axon projections in the CNS. All axons project along the canonical td tract described above (Fig. 2B) but terminated at different positions along this tract. Among v′td2 neurons, axon projections were heterogeneous. Neurons in the more anterior abdominal segments A1–A3 projected to the SEZ (Fig. 4A–C), neurons in A4 and A5 projected just short of the dorsomedial fascicle (Fig. 4D,E; see also Fig. 2C′), and neurons in A6 projected only along the ventrolateral fascicle (Fig. 4F). Segment A7 contains only one td neuron and its axon terminated along the ventrolateral fascicle, supporting a v′td2 identity (Fig. 4G). We also noted that the A7 td axon often sprouted collateral branches that projected medially (Fig. 4G). By contrast, all v′td1 neurons extend axons along the td tract to the dorsomedial fascicle and then anteriorly to the SEZ (Fig. 4H–M). At the SEZ, the axons turn slightly laterally, defasciculate, and terminate as a tuft of processes (Fig. 2B). td identity (i.e., v′td1 or v′td2) and axon projections correlated with GR and IR driver expression patterns (Table 1). v′td1 neurons generally expressed IR drivers (Ir56a and Ir76b) but did not express GR drivers, with the exception of the A1 v′td1, which expressed Gr28a-Gal4. v′td2 neurons that project to the SEZ and VNC shared expression of Gr28b.c-Gal4 and Gr89a-Gal4 but showed divergent expression of Gr28a-Gal4 (in SEZ-projecting v′td2s) and Gr33a-Gal4 (in VNC-projecting v′td2s).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

td axon projections correlate with cell identity and segment of origin. A–M, Individual central axon projections of all 13 td sensory neurons. Labeled single td neurons were generated using the MCFO technique. Magenta asterisks indicate the approximate location of the T3 neuromere. Arrowhead indicates axon termination. For all image panels, axon of interest is shown on the left side of the CNS. Blue dashed lines outline the CNS. G, White arrow indicates collateral branches from the A7 v′td2. N, O, Schematic summarizing the axon projections (N) of SEZ-targeting td neurons (green; SEZ-TDs) and VNC-targeting td neurons (red; VNC-TDs), and the segment of origin and identity of these neurons in the larva body (O). Magenta asterisks indicate the approximate location of the T3 neuromere, as in A–M. Scale bars: A–M, 50 μm.

In summary, these data reveal distinctions between v′td1 and v′td2 axon targets and gene expression patterns. In addition, we find distinctions in the projections of v′td2 depending on segment of origin. Our results reveal diversity of v′td neurons and support the possibility of multiple functions.

OK282-Gal4 is expressed in SEZ targeting td neurons and maps to the pdm3 locus

Given our identification of distinct morphologies and gene expression patterns among td neurons, we next examined factors that diversify subtypes. As a starting point, we examined enhancer trap GAL4 lines that showed td neuron expression. Lines in our GAL4 collection rarely labeled all td neurons and instead showed cell-specific and segment-specific expression. We examined one line, OK282-Gal4, in more detail. OK282-Gal4 drove expression of UAS-mCD8::GFP in v′td2 in A1, both td neurons in A2 segments, and v′td1 in segments A3-A6 (Fig. 5A,E). Other than these td neurons, the only other body sensory neuron labeled was a nearby external sensory neuron. Thus, most, but not all, SEZ-projecting td neurons showed OK282-Gal4 expression, but no VNC-projecting td neurons showed expression. We performed splinkerette PCR (Potter and Luo, 2010) to identify the location of the GAL4 insertion. Molecular mapping placed the insertion 27 bp upstream of the first exon of the pdm3 gene (Fig. 5B). Pdm3 is a POU-homeodomain transcription factor that is important for sensory axon targeting and olfactory receptor gene expression in the olfactory system, and targeting of ring neuron axons in the ellipsoid body (Tichy et al., 2008; Chen et al., 2012). The expression pattern of OK282-Gal4 was highly specific and appeared to faithfully report the expression pattern of Pdm3 in both the PNS and CNS based on anti-Pdm3 labeling of OK282-Gal4, UAS-mCD8::GFP larvae (Fig. 5C,D). Importantly, anti-Pdm3 labeled each td neuron marked by OK282-Gal4-driven reporter expression (Fig. 5C). We therefore refer to OK282-Gal4 as pdm3^OK282-Gal4. Based on expression and known roles in axon targeting, we considered Pdm3 as a strong candidate regulator of td neuron morphogenesis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Pdm3 is expressed in SEZ-targeting td neurons. A, Expression of OK282-Gal4>UAS-mCD8::GFP in td neurons (arrow) and a nearby external sensory neuron in the periphery (arrowhead). Image spans three abdominal hemisegments. B, OK282-Gal4 is inserted upstream of the pdm3 locus. C, D, OK282-Gal4 expression pattern matches anti-Pdm3 labeling in td neurons (C) and in CNS (D). Yellow arrows indicate some areas of the CNS with high overlap between the two expression patterns. E, Schematic summarizing expression pattern of Pdm3 in td neurons, as determined by pdm3^OK282-Gal4 expression pattern and anti-Pdm3 labeling. Pdm3 (P3) is expressed in most of the SEZ-targeting td neurons (green circles), but not in the VNC-targeting td neurons (red circles). Scale bars: A, D, 50 μm; C, 20 μm.

pdm3 regulates terminal targeting of td neurons

The expression pattern of Pdm3 within td neurons correlates well with axon targeting. Specifically, Pdm3 is expressed in most SEZ-targeting td neurons, but not in VNC-targeting td neurons (Fig. 5E). We performed gain- and loss-of-function experiments to test the sufficiency and necessity of Pdm3 in td axon targeting. First, we asked whether ectopic Pdm3 expression in VNC-targeting td neurons alters axon targeting. R22C07-Gal4 labels v′td2 neurons in A4-A6, which are Pdm3-negative and target the VNC (Fig. 6B). Misexpression of Pdm3 under the control of R22C07-Gal4 promoted ectopic extension of bundled td axons anteriorly beyond the VNC and into the SEZ (Fig. 6C). The maximum distance of extension of the axons was not significantly different from Pdm3⁺ SEZ-targeting v′td1 neurons in A4-A6 (labeled by R31D10-LexA; Fig. 6A,D). We observed thinning at the anterior parts of the retargeted axon bundle, suggesting that some, but not all, overextended td axons project all the way to the SEZ. These data suggest that Pdm3 expression is sufficient to switch td axon targeting from the VNC toward the SEZ. In addition, we wondered whether this switch is accompanied by changes in synaptic connectivity. Because VNC axons do not normally make synapses with targets in the SEZ, coexpression of Brp.short and UAS-pdm3 provided a simple assay for altered synapse distribution. We observed Brp signal in the SEZ and in the thoracic neuromeres of the nerve cord in overextended axons, suggesting that VNC neurons that have been retargeted are competent to form synapses in anterior regions of the VNC and in the SEZ (Fig. 6E). Thus, Pdm3 is sufficient for formation of long-range axon projections in td neurons, and for establishing sensory connectivity in the upper VNC and SEZ.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Misexpression of Pdm3 in VNC-targeting td neurons promotes SEZ targeting. A–C, td axons in control (A, B) and upon ectopic Pdm3 expression (C) conditions. Arrowhead indicates termination of td axons. Below each confocal image are keys showing SEZ-targeting td neurons labeled by R31D10-LexA (A; green) or VNC-targeting td neurons labeled by R22C07-Gal4 (B; red). P3, Pdm3 expression. Ectopic Pdm3 expression using R22C07-Gal4 switches td axon targeting from the VNC to the SEZ (C). D, Quantification of the length of the axon bundle beyond the axons' anterior turn at T3. Kruskal–Wallis test showed a significant difference of means (H₍₂₎ = 24.40, p < 0.0001). Post hoc pairwise comparisons showed that VNC-targeting td neurons ectopically expressing Pdm3 have increased axon lengths (***p < 0.0001, Mann–Whitney U test, U < 0.0001), which did not differ significantly from the length of SEZ-targeting td neurons (p = 0.332, Mann–Whitney U test, U = 48). n = 12 animals per condition. E, Ectopic expression of Pdm3 in R22C07-Gal4 neurons promotes presynaptic structures in the overextended axon segments. F, G, Ectopic expression of Pdm3 in non-td sensory neurons (cIV nociceptive neurons) causes defasciculation and axon wandering. Boxplots represent median (middle line) and 25th to 75th percentile, with whiskers extending to the most extreme data point within 1.5 times the interquartile range of the hinge. Scale bars, 50 μm.

To determine whether Pdm3 expression was sufficient to promote SEZ projections in non-td axons, we drove expression in nociceptive Class IV (cIV) dendritic arborization neurons, where Pdm3 is not normally expressed, using ppk-Gal4. cIV sensory axons normally project directly to ventromedial neuropil and remain largely within their segment of origin (Grueber et al., 2007; Gerhard et al., 2017). Misexpression of Pdm3 did not disrupt expression of the ppk-Gal4 reporter. However, we observed extensive axon-targeting defects in cIV neurons, most prominently defasciculation, axon wandering away from the ventral neuropil, and thinning of the axon scaffold near the midline (Fig. 6F,G). cIV axon terminals still targeted the medial neuropil and did not make an initial “td-like” anterior turn at the lateral part of the neuropil, suggesting that Pdm3 does not on its own determine initial td-like axon trajectories (Fig. 6F,G). This conclusion is also consistent with expression of Pdm3 in only a subset of td neurons, although all td neurons take the same initial trajectory. Some stray cIV axons projected anteriorly beyond the T1 neuropil and toward the SEZ (Fig. 6F,G). Thus, Pdm3 promotes projection toward the SEZ most robustly in td neurons, suggesting context dependence of axon guidance activity rather than a strict sufficiency of Pdm3 for SEZ projection.

We next examined the necessity of Pdm3 for td axon targeting. We used R31D10-LexA to label three Pdm3⁺ td neurons that project to the SEZ (Fig. 7A). In pdm3 ^f00828 mutants, td axons took apparently normal projections along the lateral tract, made a normal dorsomedial turn at T3 and projected anteriorly (Fig. 7B), but axon terminals collapsed around the midline (Fig. 7B). This result suggests that Pdm3 is not necessary for projection to the SEZ but is required for correct axon targeting within the SEZ.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Pdm3 is required for correct axon targeting within the SEZ. A–C, td axons in control and Pdm3 loss-of-function conditions. In control condition (A), td neurons labeled by R31D10-LexA project axons to the SEZ, where the axons turn slightly laterally away from the midline. In pdm3 mutant (B) and pan-neuronal Pdm3 RNAi knockdown conditions (C), td axon terminals extend toward the SEZ but incorrectly contact the midline (yellow arrowheads). D, Quantification of midline contacts by td axon terminals. Increased axon midline contacts were observed in pdm3 mutant (χ² (1, N = 26) = 19.07, ***p < 0.0001) and pan-neuronal Pdm3 RNAi knockdown (χ² (1, N = 26) = 19.07, ***p < 0.0001) conditions. n = 13 animals per condition. E, Confirmation of Pdm3 knockdown in td neurons. F, The VNC is abnormally elongated in pdm3 mutant larvae, but not upon pan-neuronal knockdown of Pdm3. Kruskal–Wallis test showed significant difference of means (H₍₂₎ = 9.89, p = 0.007). Post hoc pairwise comparisons showed increased VNC lengths in pdm3 mutants (***p < 0.0001, Mann–Whitney U test, U < 0.0001), but not pan-neuronal Pdm3 RNAi knockdown animals (p = 0.590, Mann–Whitney U test, U = 7.5). n = 5 animals per condition. G, Loss of Pdm3 does not lead to morphological defects in td dendrites as measured by total dendrite lengths (p = 0.602, Mann–Whitney U test, U = 10) and node counts (p = 0.190, Mann–Whitney U test, U = 6.5). Analysis of A5 v′td1. n = 5 animals per condition. Boxplots represent median (middle line) and 25th to 75th percentile, with whiskers extending to the most extreme data point within 1.5 times the interquartile range of the hinge. Scale bars: A–C, 50 μm; E, 10 μm.

In addition to expression in td neurons, Pdm3 is also expressed in CNS glial cells and neurons and, consistent with prior results (Chen et al., 2012), the VNC showed an elongation phenotype in pdm3^f00828 mutants. To reduce Pdm3 function more selectively in neurons, we expressed UAS-pdm3RNAi using a pan-neuronal elav-Gal4 driver, and used R31D10-LexA to label SEZ td axons. We reasoned that elav-Gal4 would allow early enough expression in postmitotic neurons to assess roles in axon guidance or targeting. Consistent with this, we confirmed knockdown of Pdm3 expression in larval td neurons (Fig. 7E). In addition, neuronal knockdown using elav-Gal4 did not lead to the elongated VNC phenotype (Fig. 7F). In elav-Gal4>pdm3RNAi animals, td axons in the SEZ collapsed around the midline (Fig. 7C). Thus, in both pdm3 mutants and pan-neuronal RNAi knockdown animals, td axon terminals exhibited mistargeting, as seen by incorrect contact of the midline by axons (Fig. 7D). We did not observe defects in the morphogenesis of td dendrites in pdm3^f00828-null mutants (Fig. 7G), suggesting that Pdm3 is dispensable for td dendritogenesis. Overall, our results suggest that Pdm3 expression in td neurons is sufficient to promote SEZ axon targeting but might normally function in a redundant fashion. By contrast, Pdm3 appears to be essential for fine targeting of td axons in the SEZ.

Ectopic Pdm3 expression suppresses Gr33a-QF reporter expression

Pdm3 could be regulating axon projection identity or neuronal fate in general. We therefore asked whether Pdm3 regulates broader aspects of td neuronal identity. Given that SEZ-projecting and VNC-projecting td neurons differentially express some GRs and IRs, we manipulated Pdm3 expression and asked whether chemosensory receptor expression in td neurons was altered. We focused on the expression of Ir76b, Gr28a (both expressed within SEZ-td population), and Gr33a (expressed within VNC-td population). We did not observe a change in the expression of Ir76b-Gal4, Gr28a-Gal4, or Gr33a^Gal4 reporter lines in pdm3 mutants, indicating that Pdm3 is not necessary for GR/IR reporter expression (Fig. 8A). We next used R22C07-Gal4 to drive ectopic Pdm3 expression in VNC-projecting A4-A6 v′td2 neurons, and used QF reporters (Zhang et al., 2013; Choi et al., 2016) to assess GR/IR expression. In control background, Ir76b-QF expression matched Ir76b-Gal4 expression; Gr28a-QF2 and Gr33a-QF expression largely matched their respective GAL4 reporter expression, although expression differed slightly (Fig. 8A,B). We found that ectopic Pdm3 expression in VNC-td neurons using R22C07-Gal4>UAS-pdm3 decreased the frequency with which Gr33a-QF reporter labeled A4-A6 v′td2 neurons from 100% to 31% (Fig. 8B,C). Ectopic Pdm3 expression did not affect Ir76b-QF or Gr28a-QF2 expression (Fig. 8B). Thus, Pdm3 expression is sufficient to suppress VNC-td GR reporter expression, although other inductive factors may be necessary to induce SEZ-td GR/IR expression. Together, these results suggest that Pdm3 regulates both axon targeting and GR expression aspects of td neuronal identity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Misexpression of Pdm3 in VNC-tds suppresses Gr33a-QF reporter expression. A, B, Summary of GR/IR reporter expression in wild-type, pdm3 mutant (A), and ectopic Pdm3 expression (B) conditions. Green and red cells represent SEZ- and VNC-targeting td neurons, respectively. +, Cells with expression (100% of cell labeled, n > 3 animals); −, cells with reduced expression (n > 3 animals). B, Cells with blue outline indicate td neurons misexpressing Pdm3 through R22C07-Gal4>UAS-pdm3. Gr33a-QF labeled v′td2 in A4–A6 with 100% frequency (n = 7 animals, 42 total cells examined) in control condition and only 31% frequency (n = 7 animals, 42 total cells examined) in ectopic Pdm3 condition. C, Expression of Gr33a-QF>QUAS-mtdTomato in td neurons in control and ectopic Pdm3 expression conditions. Gr33a-QF also labels glial cells, near td and adjacent nerves. Anti-Elav labels td nuclei. Images show td neurons in A5. Scale bars, 10 μm.