Dendritic diversification through transcription factor-mediated suppression of alternative morphologies

Abstract

Neurons display a striking degree of functional and morphological diversity, and the developmental mechanisms that underlie diversification are of significant interest for understanding neural circuit assembly and function. We find that the morphology of Drosophila sensory neurons is diversified through a series of suppressive transcriptional interactions involving the POU domain transcription factors Pdm1 (Nubbin) and Pdm2, the homeodomain transcription factor Cut, and the transcriptional regulators Scalloped and Vestigial. Pdm1 and Pdm2 are expressed in a subset of proprioceptive sensory neurons and function to inhibit dendrite growth and branching. A subset of touch receptors show a capacity to express Pdm1/2, but Cut represses this expression and promotes more complex dendritic arbors. Levels of Cut expression are diversified in distinct sensory neurons by selective expression of Scalloped and Vestigial. Different levels of Cut impact dendritic complexity and, consistent with this, we show that Scalloped and Vestigial suppress terminal dendritic branching. This transcriptional hierarchy therefore acts to suppress alternative morphologies to diversify three distinct types of somatosensory neurons.

Keywords: Axon targeting; Dendrite; Drosophila; Homeodomain; Neuronal morphogenesis; POU domain; Transcription factor.

Fig. 1.

Cut function in class II and III dorsal cluster sensory neuron dendrite and axon morphogenesis. (A) Representative tracings and descriptions of peripheral sensory neuron subtypes found in the dorsal cluster of third instar Drosophila larvae. (B) Wild-type class III neuron ddaA MARCM clone. Arrowheads indicate axons here and in subsequent panels. (C) Example of a cut^c145 mutant ddaA clone showing stunted dendritic arbor (arrow). (D) Example of a cut^c145 mutant ddaA clone displaying a bipolar dendritic morphology (arrows). (E) Wild-type class III ddaF clone. (F,F′) cut^c145 mutant ddaF clone (green) showing targeting of stunted dendrites (white arrows) to the v'ch1 chordotonal organ (labeled with HRP, magenta; yellow arrows). (G) Wild-type class III ddaA axons project to the ventral medial (VM) fascicle of the VNC. (Top) Confocal projection of VNC showing ddaA axon (arrow) terminating near the medial fascicles (Fas2, magenta). (Bottom) Transverse view shows axon termination near the VM fascicle (arrow). Fascicle labels: L, lateral; I, intermediate; M, medial. (H) Axon projection of cut^c145 mutant ddaA neuron. Top: Axon (arrow) projects to the medial fascicles. Bottom: Transverse view shows aberrant termination near the dorsal medial (DM) fascicle (arrow). (I) Classification of cut mutant axon phenotypes. The percentage of ddaA, ddaB, ddaF and ldaB clones that projected to either dorsal or ventral neuropil regions is shown (classes displayed as group; n=20). (J) Schematic summary of cut axon phenotypes. Wild-type (WT) ddaA, ddaF, ldaB and ddaB axons (green) normally terminated near the VM fascicles. When mutant for cut the axons of these cells (red) instead terminated near the DM fascicles. Fas2⁺ tracts are shown in magenta. Scale bars: 50 µm in B-F′; 10 µm in G,H.

Fig. 2.

Morphology of wild-type Pdm1/2-expressing cells and cell-autonomous suppression of Pdm1/2 by Cut in dorsal cluster neurons. (A) Wild-type morphology of dmd1 and dbd in third instar larvae as shown by labeling with mCD8::GFP driven by pdm1-Gal4 (green). Anti-HRP (magenta) labels all neuronal membranes. (Aʹ) GFP channel only. Arrow indicates axon. Dorsal is to the top and anterior is to the left in all panels. (B) Orthogonal view shows dmd1 dendrites (arrowhead) extending away from the epidermis to the intersegmental nerve (ISN). (Bʹ) GFP channel only. (C) Axon projection of wild-type dmd1 MARCM clone extends to the VNC medial fascicle (arrow; fascicles labeled in magenta). (D,Dʹ) Orthogonal views at the positions indicated by the dashed lines in C show termination near the DM fascicle (D, arrow) and the trajectory through the neuropil (Dʹ, dashed arrow). (E) Cut and Pdm2 immunoreactivity in embryonic md neurons (labeled by the E7-2-36-lacZ enhancer trap). In cut heterozygotes, Pdm2 is limited to dmd1 and dbd. (F) Pdm2 immunoreactivity is expanded to additional dorsal da neurons in cut homozygous mutant embryos. (G) Control ddaA MARCM clone labeled with anti-GFP (green) lacks Pdm2 expression (magenta). Inset is a magnified view of the Pdm2 channel. (H) cut^c145 ddaA MARCM clone shows ectopic Pdm2 expression. Inset is a magnified view of the Pdm2 channel. (I) Wild-type ldaB (class III) clone does not show immunoreactivity for Pdm2 (inset). (J) cut^c145 ldaB clone showing reduced terminal branching but intact primary scaffold does not show immunoreactivity for Pdm2 (inset). (K) cut^c145 ldaB clone with a stunted primary outgrowth shows immunoreactivity for Pdm2 (inset). (L) Quantification of Pdm1/2 expression in cut^c145 clones reveals a correlation between dendritic transformation and the misregulation of Pdm1/2. (M) Wild-type dorsal cluster. Md sensory neurons are labeled with 109(2)80-Gal4, UAS-mCD8::GFP (green) and anti-Pdm1 antibody (magenta). (M′) Anti-Pdm1-labeling is seen exclusively in the dmd1 and dbd neurons. (N) Dorsal cluster in which UAS-cut and UAS-mCD8::GFP are driven by 109(2)80-Gal4. (N′) Absence of anti-Pdm1 labeling in dmd1 and dbd when UAS-cut is driven in these cells. (O) Wild-type morphology of dmd1 FLP-out clone. (P) dmd1 FLP-out clone expressing UAS-cut shows dendritic overgrowth. Arrows indicate dendritic overgrowth along the epidermis. Scale bars: 50 µm in A,G-K,M-P; 10 µm in C-F.

Fig. 3.

Genetic deletion of the pdm1/2 region leads to dendritic overgrowth. (A) Wild-type dmd1 MARCM clone. Dendrites project to the ISN (yellow arrow). Arrowhead in this and subsequent panels indicates the axon. (B) Dendrites of nub^R5 dmd1 MARCM clone arborize on the epidermis (white arrows). (C) Tracings of a wild-type (boxed, top left) and mutant nub^R5 dmd1 MARCM clones. Red arrow indicates normal dendritic pattern and black arrows indicate ectopic growth along the epidermis. (D) Classification of nub^R5 dmd1 dendrite phenotypes (see Materials and Methods for classification criteria). (E) Wild-type dbd MARCM clone showing simple bipolar morphology (yellow arrows). (F) Aberrant growth and branching of dendrites (white arrows) in a nub^R5 dbd MARCM clone. Bipolar dendrites are still seen (yellow arrows). (G) Tracings of a wild-type (top) and nub^R5 dbd clones. Red arrows indicate typical branches and black arrows indicate extra branches. (H) Classification of nub^R5 dbd dendrite phenotypes (see Materials and Methods for classification criteria). (I) Control dorsal cluster from a nub^R5/+ first instar larvae. Animals were live imaged using clh8-Gal4, UAS-mCD8::GFP. Arrows indicate the normal bipolar morphology of dbd. Asterisks mark cell bodies of other labeled neurons. (J) Tracing of cells in panel I to illustrate normal morphology. dbd displays a characteristic bipolar morphology in control animals. (K,L) Examples of dbd dendrite overgrowth in nub^R5/Df(2L) ED773 heteroallelic animals. Abnormalities in dbd morphology included branching of longitudinal dendrites (white arrows) and growth of dorsally extending arborized dendrites (red arrow). Scale bars: 50 µm in A-C,E-G; 25 µm in I-L.

Fig. 4.

Overexpression of Pdm1 or Pdm2 inhibits dendritic growth and branching in class III neurons. (A) Wild-type class III neuron ddaA, visualized using the FLP-out approach with 109(2)80-Gal4, UAS-FRT-rCD2-FRT-mCD8::GFP. (B,C) Class III ddaA neuron FLP-out clones expressing (B) UAS-pdm1 or (C) UAS-pdm2. (D,E) Total dendrite length (D) and total branch point number (E) were significantly reduced upon Pdm1 (***P<0.0001) or Pdm2 (***P<0.0001) overexpression. (F) Branch points per total dendrite length (µm) was significantly reduced upon Pdm1 (**P=0.003) but not Pdm2 (P=0.0838) overexpression. n.s., not statistically significant. Scale bars: 50 µm.

Fig. 5.

Sd is required in ddaB to repress a class III-like morphology. (A) Labeling of third instar larvae with anti-HRP (red), anti-GFP (green) and anti-Cut (blue). Sd::GFP expression is strong in the class II neuron ddaB and the class III neuron ddaF in larval stages (arrows). (B) Labeling of third instar larvae with anti-Cut (green), anti-Sd (red) and anti-Elav (blue). Anti-Sd labeling is seen in the class II neuron ddaB and class III neuron ddaF (arrows). (C) Dendrites of wild-type ddaB MARCM clone (arrowheads). (D) sd^ΔB ddaB MARCM clone showing extensive terminal branching. (E) Enlarged views of the arborization patterns of wild-type ddaB (from C), sd^ΔB ddaB (from D), and wild-type ddaA MARCM clones. (F) Quantification of branch points per total dendrite length. sd^ΔB ddaA (P>0.9999) and sd^ΔB ddaF (P=0.0714) were not significantly different from wild-type clones. sd^ΔB ddaB clones showed significant (***P<0.0001) increases in branches per length, rendering them statistically indistinguishable from ddaA on this measure. (G) Quantification of total branch points for wild-type and sd^ΔB ddaA (P=0.9555), ddaB (***P<0.0001) and ddaF (*P=0.0224) clones. Scale bars: 25 μm in A,B,E; 50 μm in C,D.

Fig. 6.

Sd and Vg are required for low Cut expression levels in ddaB. (A) Anti-HRP labeling of control dorsal cluster. Individual da neurons are named. (A′) Anti-Cut labeling of dorsal cluster neurons in grayscale (left) and pseudo-color image (look-up table, LUT; right). ddaA (arrowhead) and ddaB (arrow) neurons are outlined by a green box. (B) Anti-HRP labeling of dorsal cluster. ddaB neuron is mutant for sd. (B′) Anti-Cut labeling of dorsal cluster neurons in grayscale (left) and pseudo-color image (right). ddaA (arrowhead) and ddaB (arrow) neurons are outlined by a green box. (C) Wild-type MARCM clone of ddaB (arrow) showing location of cell body relative to the class III neuron ddaA (arrowhead). Clone is labeled by anti-GFP and neurons are labeled by anti-HRP. (D) Tracing of clone showing wild-type ddaB neuron (green, arrow) and terminal extensions of class III neurons (magenta, arrowhead). (E) Anti-Cut labeling of wild-type clone in C showing immunoreactivity in ddaB (arrow) and ddaA (arrowhead). Lower panel shows pseudo-coloring of anti-Cut label. (F) vg⁻ ddaB MARCM clone (arrow) shows a class III-like branching morphology. (F′) Arbors of vg clone show numerous fine terminal processes. (G) Tracing of vg⁻ clone (green, arrow) and the class III neuron ddaA (magenta, arrowhead). (H) Cut labeling of vg⁻ ddaB MARCM clone (arrow) and ddaA (arrowhead). Lower panel shows pseudo-coloring of anti-Cut label. (I) Quantification of Cut labeling in wild-type and sd ddaB neurons. Ratios of ddaB:ddaA Cut labeling in control genotypes (sd^+/+ ddaB:sd^+/+ ddaA, and sd^+/− ddaB:sd^+/− ddaA) were statistically equivalent. The ratio of ddaB:ddaA Cut labeling between sd^−/− ddaB clones and sd^+/− ddaA was significantly increased compared with both wild-type controls (**P=0.0017) and heterozygous controls (***P=0.0004). Significance was determined by Kruskal–Wallis with Dunn's multiple comparisons test. (J) Quantification of Cut labeling in wild-type and vg ddaB neurons. The ratio of ddaB:ddaA Cut labeling is significantly increased for vg ddaB clones (***P=0.0006). Scale bars: 50 μm in A-D,F,G; 10 μm in E,H.

Fig. 7.

Sd driven by the tubulin promoter simplifies ddaF dendrites and this effect is partially reversed by overexpression of Cut. (A) Wild-type proximal arbor of a dorsal cluster class III neuron ddaF, visualized using the FLP-out approach with 109(2)80-Gal4, UAS-FRT-rCD2-FRT-mCD8::GFP. Terminal branches are indicated by arrows. (B) A ddaF arbor in tub>HA-sd larva. Terminal branches are indicated by arrows. A branch from the ddaA neuron, which is less affected than ddaF, is visible at the bottom (arrowhead). (C) A ddaF arbor when UAS-cut is expressed using 109(2)80-Gal4 together with tub>HA-sd. (D) Quantification of the effects of tub>HA-sd on branch points per total dendrite length for ddaA (*P=0.0108). (E) Quantification of effects of tub>HA-sd on branch points per total dendrite length for ddaB (P=0.278). (F) Quantification of effects of tub>HA-sd and UAS-cut on branch points per total dendrite length for ddaF. Branch points per total dendrite length was unchanged in UAS-cut compared with control (P=0.7015), but significantly decreased in tub>HA-sd compared with control (***P<0.0001). Expression of UAS-cut together with tub>HA-sd significantly increased branch density compared with tub>HA-sd alone (*P=0.0126) but did not restore wild-type density (***P<0.0001). (G) Quantification of effects of tub>HA-sd on total number of branch points in ddaA (*P=0.0351). (H) Quantification of effects of tub>HA-sd on total number of branches in ddaB (P=0.8431). (I) Quantification of effects of tub>HA-sd and UAS-cut on total number of branch points in ddaF. Compared with wild-type, branch points were significantly reduced in tub>HA-sd larvae (***P<0.0001) and in tub>HA-sd, UAS-cut larvae (*P=0.0213). Expression of UAS-cut with tub>HA-sd did not significantly affect branch point number compared with tub>HA-sd alone (P>0.9999). Significance in panel I was determined by Kruskal–Wallis with Dunn's multiple comparisons test. Scale bars: 50 μm.

Fig. 8.

Cut, Pdm1/2, Sd and Vg interact to diversify sensory neuron morphology. (A) Summary of the proposed regulatory interactions between Cut, Pdm1/2, Sd and Vg that regulate morphogenesis of dorsal cluster neurons. Multiple neurons have a capacity to express Pdm1/2, but expression is blocked by Cut (see Fig. 2). Pdm1/2 limit dendrite growth in dbd and dmd1 (see Fig. 3). Cut promotes growth by repressing Pdm1/2 and by promoting dendritic growth and branching. At high levels, Cut promotes the development of fine terminal branches characteristic of class III da neurons (Grueber et al., 2003). Sd and Vg repress Cut to low levels in ddaB and prevent acquisition of class III features (see Figs 5 and 6). Unlike in ddaB, Cut expression remains high in the presence of Sd and Vg in ddaF neurons. These cells develop class III-like terminal branches, although Sd limits the number of terminal branches (not shown in model; see Fig. 5). In addition to acting through Cut regulation, Sd may also act in parallel to, or downstream of, Cut to restrict terminal branching (see Fig. 7). (B) The regulatory interactions depicted in A promote specific transcription factor expression patterns as determined by antibody labeling (see Figs 2, 5, Fig. S4). (C) The interactions summarized in A contribute to the mature dendritic morphology of individual sensory neurons.