Background
Dendrites are the primary sites for the reception of sensory and synaptic input to neurons. This input is influenced by the architecture of the dendritic tree 12 and by the targeting of dendrites into appropriate territories 34. For example, the length and tufted architecture of dendrites in the auditory brainstem of birds and mammals influences the tuning of coincidence-detecting neurons to optimal stimulus frequencies 5. In the vertebrate spinal cord, specific targeting of motor neuron dendrites correlates with the precise matching with their presynaptic sensory afferents 4.
A current challenge for developmental neurobiologists is to uncover the cellular and molecular mechanisms that underlie the growth, branching and targeting of dendrites. The fruitfly, Drosophila melanogaster, has proven to be an effective model system for applying genetics to this issue 6789. Drosophila has neurons that are uniquely identifiable, with reproducible dendrite morphologies as intricate and diverse as those of vertebrates 2. Moreover, Drosophila dendrites are also thought to be homologous to those of vertebrate neurons 7. Within the peripheral nervous system (PNS) of Drosophila, studies to date indicate that genetic programs regulate the outgrowth, size, branching pattern and orientation of dendritic arborisation (da) sensory neurons 26. Although a range of molecules implicated in da dendrite development have been identified, including cytoskeleton-associated proteins, small GTPases, transmembrane proteins, transcription factors and translation regulators, our understanding of da neuron dendrite morphogenesis remains far from complete 6101112131415. Still less is known about the development of dendrites in the Drosophila central nervous system (CNS) 16. Dendrites in the CNS differ from PNS sensory neuron dendrites in that they are not specialised for the reception of particular external stimuli 17181920 but instead form connections with presynaptic terminals of other neurons, and develop in a highly complex environment, the central neuropile.
In view of these fundamental differences, it is important to ask whether dendrites of peripheral and central neurons are shaped by distinct or shared mechanisms. To this end, we have carried out two comparable gain-of-function genetic screens in Drosophila to identify genes influencing dendrite morphogenesis. We used GAL4 driver lines 21 that express in either the da sensory neurons or an identified central motor neuron, RP2, for which we designed a novel mosaic expression system 22. We screened a well characterized collection of 141 lines that carry independent insertions of the Gene Search (GS) P element 23, a potent UAS-based vector that can direct the expression of genes flanking the site of insertion 24. We identified genes in each screen that influenced dendritic architecture. Some were novel, some were previously characterized, and 39% of the genes were common to both screens. However, the phenotypes arising from the genes identified in these screens revealed that fundamental differences may exist in the way peripheral and central neurons grow, branch, and find their targets.
Results
A gain-of-function screen for genes that influence the morphologies of peripheral neuron dendrites
Screening in embryos
To identify genes affecting da dendrite morphogenesis, we focused on da neurons of the dorsal PNS cluster, visualised selectively by GAL4109(2)80 and a transgene encoding a membrane-targeted green fluorescent protein (GFP) reporter, UAS-mCD8::GFP 1521. We crossed each of 141 GS expression lines into this background and assayed in late stage 17 Drosophila embryos the effects on: overall PNS integrity; the number and position of GFP-positive cell bodies in the dorsal cluster; the extent of dendrite outgrowth of dorsal cluster da neurons; and their pattern of branching. The results of this screen are reported in Table 1.
<p>Table 1</p>
Dendrite morphology phenotypes observed in da neurons
Observed increases (+) or decreases (-)
Embryos (stage 17)
Larvae (third Instar)a
Class IV
Line
Cytological location
Closest gene
Growth
Branching
Class I
Class II growth
Class III
Growth
Branching
Other observations
shrub-like effects
GSd034
3R;100D1
ttk
—
o
o
o
o
o
GSd446
3R;100D1
ttk
—
o
o
o
o
o
GSd468
3R;100D1
ttk
—
o
o
o
o
o
GSd462b
3R;100D1 3R;92F1
ttk Stat92E
—
o
o
o
o
o
PcG-like effects
GSd219
2R;57A6
bl
—
GSd226
2L;21D1
cbt
—
GSd220
2R;60A6
ken
—
GSd247
2R;57A6
mir-313
—
GSd472
2R;47D6
shn
—
GSd469
3R;97E11
woc
—
cut-like effects
GSd324
2L;32E2
ab
—
—
—
—
GSd331
3R;98F13
CG11897
—
—
—
GSd454
2L;30B5
CG33298
—
—
—
—
GSd233
3R;88A4
foxo
—
—
—
—
GSd411
3R;88A5
foxo
—
—
—
—
GSd327
2L;38E3
Hr38
—
—
—
—
GFP intensity reduced
GSd500
3L;70D7
stwl
—
—
Other
GSd430
2R;59F1
apt
—
—
GSd325
3R;92B3
bnl
+/—
GSd332
3R;92F2
bon
—
—
GSd321
3R;91F4
CG11779
—
—
GFP intensity reduced
GSd239
3R;86E11
CG14709
—
—
GSd450
2L;38D5
CG2617
—
—
—
—
—
Fewer than eight multidendritic (md) neurons of dorsal cluster express GFP
GSd422
2R;42E1
CG33558
+
o
o
o
o
o
GSd211
3R;87D7
CG7518
—
—
—
—
—
GFP expressing cell bodies and dendrites appear degenerative
GSd496
2L;29F8
CG9582
—
o
o
o
o
o
Similar to effect of activated cdc42
GSd066
3L;61B3
E(bx)
+/—
GSd451
3L;61B3
E(bx)
+/—
GSd492
3L;61B3
E(bx)
+/—
GSd402
3R;99A1
EST:EN05557/EN06658
—
—
—
—
—
Fewer than eight md neurons express GFP
GSd214
2L;24A2
for
—
—
GSd328
2L;24A4
for
—
—
GSd312
3R;99E4
hdc
—
—
GSd113
2L;26B5
Kr-h1
—
—
—
GSd429
2R;47A12
lola
—
—
GSd484
3R;93D9
mod(mdg4)
+
—
—
Similar to spineless mutations, but lacks effect on class II
GSd302
3R;91F4
nos
—
—
GSd428
2L;36E3
PFE
—
—
GSd236
3R;94E13
pnt
—
—
GSd420
3R;94E13
pnt
—
—
GSd458
2R;47A13
psq
+
—
—
—
—
Fewer than eight md neurons express GFP, reduced GFP intensity
GSd231
2R;52E5
Rho1
—
+
o
o
o
o
o
Reduced dendritic field, but terminals have exuberant branching
GSd431
2R;52E5
Rho1
—
+
o
o
o
o
o
Reduced dendritic field, but terminals have exuberant branching
aBoxes marked with an 'o' indicate that larval phenotypes were not assessed for those GS lines that had severe effects in embryos.
bLine GSd462 harbours two insertions of the GS element.
Our screen identified eight GS lines that, at embryonic stages, caused severely reduced dendrite outgrowth and/or branching, or increased branching (Figure 1). For example, misexpression of GSd034 resulted in thickened lower order dendrites and reduced outgrowth of higher order dendritic branches (Figure 1c). Misexpression of GSd231 led to reduced dendritic growth as reflected by reduced area of the dendritic field (Figure 1d), though profuse branching was retained at the terminals of the shortened primary branches. A third example is misexpression of GSd422, which caused the formation of ectopic spine-like protrusions from the main branches of all dorsal cluster da neurons (Figure 1f). These examples demonstrate that gene misexpression can modulate specific aspects of dendritic development, such as growth and branching, and that the regulation of these two processes is genetically separable.
<p>Figure 1</p>
Embryonic da dendrite screen – examples of phenotypes
Embryonic da dendrite screen – examples of phenotypes.(a) Cartoon showing the relative positions of cell bodies of dorsal cluster multidendritic (md) neurons in late stage 17 embryos. The dotted box indicates the region of dendritic field chosen for enlargement in each of (b',c',d',e',f'). (b,b') In control animals, the dendritic trees of the six da neurons in the dorsal cluster can be visualized with GAL4109(2)80 driving membrane-targeted GFP (UAS-mCD8::GFP). (c,c') Misexpression of GSd034: dilation of primary and reduced outgrowth of higher order dendritic branches. (d,d') Misexpression of GSd231: reduced dendritic field size with residual branching. (e,e') Slightly younger control animal, though still late stage 17, for comparison with (f,f'). (f,f') Misexpression of GSd422: production of filopodial spine-like protrusions. All images are maximal Z-projections of stacked confocal images. Anterior is left and ventral is down in all panels. Scale bar in (b) = 20 μm and applies to (c-f) also.
![]()
Screening in larvae
35 GS lines caused mild but reproducible defects at embryonic stages. We characterised their phenotypes at a later developmental stage, analyzing the same dorsal cluster da neurons in wandering third instar larvae (Figure 2). In larvae, four da neuron classes (I-IV) of increasing dendritic complexity and size can be readily observed, and at least one representative from each class resides in the dorsal cluster. We examined whether GS misexpression influenced dendritic morphology of the class I neurons ddaD and ddaE, the class II neuron ddaB, the class III neurons ddaA and ddaF, and the class IV neuron ddaC (Figure 2a). Class I da neurons have the simplest pattern, having relatively few primary dendrites with interstitial secondary and tertiary branches oriented in the anterior-posterior direction (Figure 2e). The dendrites of class II da neurons are long and typically have symmetric bifurcations. Class III dendrites are characterized by short spine-like protrusions emanating from long main branches (Figure 2g). Class IV dendrites show highly complex branching patterns that innervate large regions of the body wall (Figure 2c) 25.
<p>Figure 2</p>
Larval da dendrite screen – examples of phenotypes
Larval da dendrite screen – examples of phenotypes. (a) In a control third instar larva (GAL4109(2)80, UAS-mCD8::GFP/+), one can visualize the eight multidendritic neuron cell bodies of the dorsal cluster and the fields occupied by da neuron dendrites. Dotted boxes indicate fields examined at higher power in (c,e,g) to illustrate regions occupied primarily by the dendrites of class IV (c) (ddaC), class I (e) (ddaE), and class III (g) (ddaA) da neurons. (b) Misexpression of GSd302: severe reduction of higher order branches in class III and class IV da neurons, though the growth of primary branches of these and other da neurons appears intact. (c) Region in control highlighting (in red) the higher order branches of the class IV ddaC neuron. (d) Misexpression of GSd239 reduced the length of higher order branches of ddaC, with no obvious reduction in branch number. (e) Control ddaE (arrowhead at cell body), a class I da neuron that ordinarily has a simple pattern of lower order dendrite branches (highlighted in red). (f) Misexpression of GSd458 caused numerous small branches to emerge from ddaE (arrowhead at cell body). (h) Compared to controls (as in (g)), misexpression of GSd236 severely reduced numbers of spine-like protrusions in the class III neuron ddaA (cell body marked with arrowhead, dendrites highlighted in red). (i-k) Misexpression GSd402 and GSd450 caused severe reduction of dendrite outgrowth and branching (i,k), often with fewer GFP-labelled da neurons and signs of neuronal degeneration (higher power in (j,l)). All images are maximal Z-projections of stacked confocal images. Anterior is left and ventral is down. Scale bar in (a) = 100 μm and applies to (b) also. Scale bar in (c) = 30 μm and applies to (d-h) also. Scale bar in (i) = 100 μm and applies to (k) also.
![]()
The majority of the 35 selected GS lines affected specific aspects of growth and branching of larval dendrites. The results of our study are catalogued in Table 1. Examples of phenotypes are provided in Figure 2 and are summarized below. Where phenotypic defects were specific to particular classes, we confirmed this with class-specific driver lines (Figure 3).
<p>Figure 3</p>
GS misexpression with class-specific Gal4 drivers in larval da neurons
GS misexpression with class-specific Gal4 drivers in larval da neurons.(a) Control class I da neuron ddaE visualized with GAL4221 driving UAS-mCD8::GFP. (b) Misexpression of GSd458 caused increased numbers of small dendritic branches, though the primary branches were unaffected. (c) C161-GAL4 drives expression in classes I-III, but not class IV, allowing better visualization of spine-like protrusions on the class III neuron ddaA. Arrowheads in (c,d) mark the cell body of ddaA. (c') Tracing of ddaA cell body and dendrites in (c). (d) Misexpression of GSd236: primary dendrites of ddaA are devoid of spine-like protrusions. (d') Tracing of ddaA cell body and dendrites in (d). (e) ppk1.9-GAL4 is a class IV da neuron driver, revealing the complex dendritic tree of ddaC. (f) Misexpression of Gsd454: reductions in the number and growth of higher order branches of ddaC. (g) Quantification of branch ends per neuron for the genotypes shown in (a-f), showing class specificity of branching defects. In class I ddaE neurons (left), GSd458 increases branching dramatically (asterisk denotes t-test, P < 1e-5), while GSd454 has no effect. In class IV ddaC neurons (right), both GSd458 and GSd454 reduce branching relative to controls (wild type (WT); asterisks denote t-tests, both P < 1e-8). In both cases, the total length of the dendritic arbor was dramatically reduced (control (WT) = 17,389 ± 422 μm versus GSd454 = 8,544 ± 657 μm (t-test P < 1e-10) or versus GSd458 = 2,650 ± 296 μm (t-test, P < 1e-16). Since higher order branches were reduced but the growth of primary dendrites was mostly unaffected, there was no effect on dendritic field area (for example, control = 304,899 ± 7,115 μm2 versus GSd454 = 301,475 ± 9,141 μm2; t-test P > 0.8). In class III ddaA neurons (middle), GSd236 dramatically reduced the number of short spine-like protrusions (t-test, P < 0.003), but had no effect on the total length of primary dendrites (control = 1,736 ± 137 μm versus GSd236 = 2,132 ± 157 μm; t-test P > 0.1). All images are maximal Z-projections of stacked confocal images. Anterior is left and ventral is down. Scale bars: (a-d) = 50 μm; (e,f) = 100 μm.
![]()
Growth (12 lines)
Dendritic arbors with overtly reduced field area were interpreted as having growth defects. In other GS lines, the field area was unaffected because the major branches extended fully, yet minor branches of higher order showed reduced growth. For example, in class IV ddaC dendrites, misexpression of GSd239 reduced the length, but not necessarily the number of higher order branches (Figure 2d).
Branch number (31 lines)
Increases and decreases of branch number were also observed. For example, misexpression of GSd302 (Figure 2b) and GSd454 (Figure 3f,g) specifically reduced the number of higher though not lower order branches.
Branch type (six lines)
In class I neurons, GSd458 increased the number of short tips on the normally bare primary and secondary branches (Figures 2f and 3b,g). The opposite effect was caused by misexpression of GSd236, which reduced the number of spine-like protrusions normally found on class III neurons (Figures 2h and 3d,d',g).
Degeneration (four lines)
Despite only mild effects on embryonic dendrite morphology, several GS lines caused severely reduced dendritic trees at larval stages, often accompanied by degeneration of the cell soma, for example, GSd402 (Figure 2i,j) and GSd450 (Figure 2k,l).
Summary of phenotypes induced in da neuron dendrites
In total, we identified 43 GS lines causing da dendrite phenotypes at embryonic (8) or larval (35) stages. The phenotypes generated by many GS lines resemble those of known mutants and fall into five categories that may relate to known molecular pathways (Table 1). In the first category, four GS lines had phenotypes that resemble genetic mutations in shrub, which leads to reduced embryonic da dendrite growth 26. The second category (six lines) resembles phenotypes of Polycomb Group genes like E(z), esc, or Su(z)12, which are involved in the maintenance of dendritic arbors of class IV neurons 27. The third category contains seven GS lines that showed effects similar to cut mutations, the levels of which regulate class-specific dendritic growth and terminal branching 28. A fourth phenotypic category is represented by GSd484, which resembled spineless mutants because it increased class I and reduced class III and IV dendrites 29. However, spineless mutants also have increased class II dendrites, which we did not observe with GSd484. Fifth, GSd496 showed reduced dendritic branching in embryonic da neurons, similar to constitutive activation of the GTPase cdc42 15.
In addition to these five phenotypic categories, we also identified 24 GS lines causing phenotypes that did not resemble known mutants. These lines suggest the existence of additional genetic pathways that underlie dendritic development and may, via their insertion sites into the genome, provide clues about their molecular nature.
A gain-of-function screen for genes that affect central neuron dendrites
To compare dendrite morphogenesis between peripheral sensory neurons and central neurons, we executed a comparable misexpression screen for central neuron dendrites. Using a new FLPout based system we expressed the same 141 GS lines discussed above in selected RP2 motor neurons, again using mCD8::GFP to reveal dendrite morphologies 22. With this system, misexpression is initiated by 14–15 hours after egg laying (AEL). At this time, RP2 neurons have begun to establish characteristic dendritic trees in a particular neuropile territory and, in the periphery, RP2 axons have made contacts with their target muscles 30. We examined RP2 dendritic trees more than 10 hours later, at 25–31 hours AEL, when the majority of dendritic branches are normally located in the lateral neuropile and only a few branches project towards the midline (Figure 4a–c).
<p>Figure 4</p>
RP2 dendrite screen – phenotypic categories
RP2 dendrite screen – phenotypic categories. Rows show examples representing the main phenotypic categories recovered from the central (RP2) neuron dendrite misexpression screen. Left and centre columns: confocal images (maximal Z-projections) of RP2 neurons at 25–31 hours AEL, visualised with UAS-mCD8::GFP. (a) Control RP2 neuron with brackets indicating the dendritic tree. (b) Control RP2 neuron in the context of a set of axon tracts visualised by anti-FasciclinII staining (magenta), with arrowheads pointing from top to bottom to the lateral, intermediate and medial FasciclinII tracts and the midline indicated by a dotted line. Dendrites between the lateral and central intermediate Fasciclin II fascicle are defined as 'lateral'; dendrites located between the central intermediate fascicle and the midline as 'medial'; the same applies to (o,p). (c) Same neuron as in (b) but with sectors of its dendritic tree pseudo-coloured to highlight branches targeted to anterior lateral (magenta), anterior medial (yellow) and posterior lateral (cyan) regions. Anterior is left and the ventral midline is down. (d,e,g,h,i,j,l,m,o,p) Experimental cells: misexpression lines are indicated in the bottom right-hand corner of each panel. Right column: (f,k,n,q) quantifications of the dendritic phenotypes shown in the left and central columns. As illustrated in (f), both dendritic tree length and number of branching events are reduced in the 'Growth' and 'Branching' categories. 'Branching' phenotypes have trees with an anterior-posterior extent comparable to controls (Additional file 2) but have an altered pattern of branching: fewer branching events and more segments that are longer (>5 μm). *P < 0.01, **P < 0.005, t-test, N = 5. Error bars indicate the standard error. Arrows in (b,o,p) point to medial branches present in controls (b) and absent/reduced in experiments (o,p). Black asterisks in (e,p) indicate the cell body of the contralateral RP2 neuron. Scale bar: 10 μm.
![]()
We identified 60 GS lines that affect specific aspects of RP2 dendrite development (summarised in Table 2; for a comparative summary of both screens see Additional file 1). The resultant RP2 phenotypes fall into two partially overlapping classes: altered dendritic growth and/or branching; and aberrant dendritic targeting.
<p>Additional file 1</p>
Summary of da and RP2 dendrite screens. Summary of da and RP2 dendrite screens.
Click here for file
<p>Table 2</p>
Summary of phenotypes observed in RP2 neurons
Observed increases (+) or decreases (-)
Targeting
Line
Cytological location
Closest gene
Growth
Branching
Anterior
Posterior
Medial
Lateral
Expression in CNS: ubiquitous (Ub) or patterned (P)
GSd324
2L;32E2
ab
+
-
Not in CNS
GSd332
3R;92F2
bon
-
+
+
P
GSd226
2L;21D1
cbt
+
-
-
+
Ub
GSd321
3R;91F4
CG11779
-
-
-
-
-
P
GSd239
3R;86E11
CG14709
-
+
-
P
GSd449
3R;86E11
CG14709
+
-
+
P
GSd440
2R;43D1
CG1602
-
Ub
GSd450
2L;38D5
CG2617
-
P
GSd454
2L;30B5
CG33298
-
-
-
-
Ub
GSd466
2L;30B5
CG33298
-
-
-
-
-
Ub
GSd486
2R;58D4
CG3624
Variable
Variable
P
GSd499
3L;74E2
CG7510
-
-
-
-
-
P
GSd211
3R;87D7
CG7518
-
-
-
-
-
P
GSd098
2R;49B12
CG8776
-
+
-
+
Ub
GSd496
2L;29F8
CG9582
-
Ub
GSd322
3R;90D1
cpo
+
-
P
GSd447
2L;36C9
dl
-
-
-
-
-
P
GSd066
3L;61B3
E(bx)
-
+
-
+
P
GSd451
3L;61B3
E(bx)
-
+
P
GSd492
3L;61B3
E(bx)
+
-
-
P
GSd017
3L;75B2
Eip75B
-
+
-
Ub
GSd436
2L;35D2
esg
-
+
-
+
P
GSd481a
2L;35D2 2R;55C8
esg imd/Dp1
-
-
-
-
-
P, imd N/D, Dp in CNS (Ub)
GSd421
3R;100C2
EST:LP08211
-
+
-
+
N/D
GSd207
2R;53D11
EST:SD02913
-
-
+
N/D
GSd233
3R;88A4
foxo