R7 Photoreceptor Axon Growth Is Temporally Controlled by the Transcription Factor Ttk69, Which Inhibits Growth in Part by Promoting Transforming Growth Factor-β/Activin Signaling

Ttk69 is required to prevent overgrowth of R7 axon terminals

To identify genes that regulate R7 axon growth and targeting, we performed a genetic screen using mosaic animals whose R7 neurons were homozygous for randomly mutagenized chromosome arms. We identified three lethal, noncomplementing mutations on 3R that caused R7 terminals to overlap with their neighbors. We mapped these by performing complementation tests with a standard set of deficiencies (Bloomington Drosophila Stock Center) and ultimately found that they failed to complement loss-of-function mutations in the ttk gene for lethality. We therefore examined whether loss of ttk caused an R7 axon defect. By 60 h after puparium formation (APF), wild-type R7 axons terminate in non-overlapping, approximately spherical boutons in the M6 layer of the medulla (Fig. 1A). In contrast, 20.8 ± 0.70% (n = 9 brains) of ttk^1e11 mutant R7 axon terminals extended laterally within the M6 layer, resulting in overlap with adjacent wild-type R7 terminals (Fig. 1B,E). Some ttk^1e11 mutant R7s extended beyond their immediate R7 neighbors while remaining in M6 (data not shown), whereas others extended back toward the M3 layer along an adjacent R7 axon (Fig. 1B). ttk^1e11 mutant R7 axons also occasionally extended short branches at the M1 or M3 layers (Fig. 1C). We noticed that even those ttk^1e11 R7 axon terminals that remained spherical had abnormally large terminals (Fig. 1C,F). We conclude that ttk is required to prevent overgrowth of R7 axon terminals within the medulla.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Ttk69 is required to prevent overgrowth of R7 axon terminals. A–C, Medullas of mosaic animals, created by GMR–FLP and MARCM, at 60 h APF. D, Medulla of a mosaic animal, created by ey3.5–FLP and MARCM, at 60 h APF. R7s homozygous for the specified chromosome arm express mCD8–GFP (green). All R7 and R8 axons are labeled with anti-Chp (red). A, Wild-type (FRT82) (WT) R7 axons terminate at the M6 layer of the medulla in spherical boutons (chevrons). B, ttk^1e11 mutant R7 axons terminate in the M6 layer but often extend laterally (arrows), resulting in overlap with neighboring R7 axon terminals. Some ttk^1e11 mutant R7 axons extend not only laterally but also return along an adjacent R7 axon toward the M3 layer (arrowheads). C, Even those ttk^1e11 mutant R7 terminals that remain spherical are abnormally large (chevron). ttk^1e11 mutant R7 axons occasionally have lateral branches at the M1 or M3 layer (8.3 ± 0.7%; n = 9; arrowheads point to an M3 example). D, babo mutant R7 axons also extend laterally within M6, resulting in overlap with neighboring R7 terminals (arrow), but those babo mutant R7 terminals that remain spherical are of normal diameter. E, Quantification of R7 axon extension within M6, scored as the frequency with which individual homozygous R7 axons, created by GMR–FLP/MARCM, overlapped with neighboring R7 axon terminals. If neighboring R7 axons were absent (see below), “overlap” was scored as overlap with the M6 regions directly below neighboring R8 axon terminals in M3. Error bars represent SEM. Zero percent of wild-type (FRT82) R7 axon terminals extend laterally within M6, whether neighboring R7 axons are present (WT; n = 5) or have been removed by means of the hypomorphic sev^V1 mutation (n = 6). The frequency with which ttk^1e11 mutant R7 axon terminals extend laterally within M6 (20.8 ± 0.70%; n = 9) is enhanced by removing neighboring R7 axons (53.3 ± 2.2%; n = 9). Expression of the Ttk69 isoform rescues ttk^1e11 mutant R7 terminal overlap (0.8 ± 0.4%; n = 13), but expression of Ttk88 does not (21.9 ± 2.0%; n = 8). Individual homozygous wild-type (FRT82) R7 axons expressing ttk–RNAi extend laterally at a frequency (18.7 ± 0.77%; n = 14) similar to that at which ttk^1e11 mutant R7 axons do so. The same is true of imp-α3 mutant R7 axons (22.3 ± 1.0%; n = 10). F, Quantification of the width of R7 axon terminals that remain spherical. Error bars represent SEM. Both ttk^1e11 (2.69 ± 0.10 μm in width; n = 9) and imp-α3 (2.84 ± 0.10 μm; n = 10) mutant R7 terminals are significantly larger than either wild-type (1.58 ± 0.05 μm; n = 5) or babo (1.67 ± 0.12 μm; n = 7) mutant terminals. ***p < 0.00001. There is no significant difference (n.s.) between wild-type and babo R7 terminals (p = 0.7) or between ttk^1e11 and imp-α3 terminals (p = 0.2). G, H, Mosaic animals created by GMR–FLP/MARCM in which R7s homozygous for imp-α3 express mCD8–GFP (green). G, G′, At 48 h APF, Ttk69 protein (red) is excluded from the nuclei of imp-α3 mutant R7s (arrows; compare with wild-type R7s, arrowheads). H, At 60 h APF, imp-α3 mutant R7 terminals extend laterally within M6 (arrows). Those imp-α3 mutant R7 terminals that remain spherical are abnormally large (chevron). Scale bars, 5 μm. n equals number of brains scored.

The ttk locus encodes two transcription factor isoforms, Ttk69 and Ttk88, which share a common BTB/POZ domain but have different zinc fingers and consequently different DNA-binding specificities (Read and Manley, 1992). The ttk^1e11 allele contains a deletion within a Ttk69-specific exon, completely eliminating Ttk69 expression but leaving Ttk88 expression intact (Lai and Li, 1999). To confirm that the Ttk69 isoform is required for proper R7 axon growth, we tested the ability of each isoform to rescue the ttk mutant R7 axon defect. We found that overexpressing Ttk88 had no effect on ttk^1e11 mutant R7s (Fig. 1E) but that overexpressing Ttk69 prevented ttk^1e11 mutant R7 axon overgrowth (Fig. 1E; Ttk69 overexpression also caused a gain-of-function phenotype, as described below). We conclude that Ttk69 is required to prevent R7 axon overgrowth and will refer to ttk^1e11 as ttk69.

Ttk69 is required in the nucleus and specifically affects R7 axon terminal morphology

Ttk69 has been shown previously to act as a transcription factor that represses neuronal development (Read and Manley, 1992; Guo et al., 1995; Giesen et al., 1997; Badenhorst, 2001; Badenhorst et al., 2002). Although some transcription factors have been found to have additional functions outside the nucleus, we did not observe Ttk69 expression in R7 axons (data not shown), and we found that excluding Ttk69 from R7 nuclei by disrupting the nuclear importer Imp-α3 precisely phenocopied loss of Ttk69 (Fig. 1E–H). We conclude that Ttk69 is likely functioning as a conventional transcription factor to regulate R7 axon terminals.

Loss of Ttk69 from the entire retina has been shown previously to disrupt adult R neurons (Lai and Li, 1999). We therefore examined whether the ttk69 mutant R7 axon defect might be secondary to a failure to differentiate properly. By 60 h APF, wild-type R7 cell bodies have undergone a change in apical–basal polarity and consequently have a characteristic wedge shape (Fig. 2A; Wolff and Ready, 1993). In addition, they express the neuronal marker Elav (Fig. 2A′), the photoreceptor marker Chp (Fig. 1A), and the R7 fate marker Prospero (Pros; Fig. 2A″), a combination of markers that is unique among retinal cells (Kauffmann et al., 1996). We found that ttk69 mutant R7 neurons exhibited all the characteristics of normal R7 neurons (Figs. 1B, 2B–B″) and so conclude that loss of ttk69 does not grossly alter R7 differentiation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Ttk69 specifically affects R7 axon terminal morphology. A, B, Retinas of mosaic animals, created by GMR–FLP/MARCM, at 60 h APF. Homozygous R7s express mCD8–GFP (green). A–A″, Wild-type (FRT82) (WT) R7 cell bodies (arrow) express the neuronal marker Elav (blue; A′) and the R7 fate marker Pros (red; A″). The GMR–FLP/MARCM technique also creates homozygous R1 and R6 neurons (arrowhead), which express Elav but not Pros. B–B″, Native R7s that are homozygous for ttk69 (ttk^1e11; arrow) correctly express Elav (B′) and Pros (B″). Cone cells homozygous for ttk69 are transformed into R7s (arrowheads) and also express Elav and Pros. Transformed ttk69 mutant cone cells can be positioned anywhere within the ommatidium. In this example, two transformed cone cells flank the native, wild-type R7 (double arrowhead), displacing the latter into a slightly different focal plane. C, Medulla of a wild-type adult animal in which R1–R8 neurons express EB1–GFP (white) under the control of chp–Gal4. The R7 axons terminate in non-overlapping boutons in M6. D, D′, Retina from a wild-type 60 h APF pupa in which R1–R8 neurons express EB1–GFP under the control of chp–Gal4. All R neurons express Ttk69 (D), and each ommatidium contains exactly one R7 neuron, as judged by Pros expression (D′). E, Medulla of a wild-type adult animal in which R1–R8 neurons express both EB1–GFP (white) and ttk–RNAi under the control of chp–Gal4. R7 axons in these animals extend laterally within M6, resulting in overlap (21.8 ± 2.6%; n = 7 brains; arrow), or have abnormally large terminals (chevrons). F, F′, Retina from a 60 h APF pupa in which R1–R8 neurons express both EB1–GFP and ttk–RNAi under the control of chp–Gal4. Ttk69 protein is absent from all R neurons (F), but each ommatidium still contains exactly one R7, as judged by Pros expression (F′); the asterisk indicates an ommatidium in which the Pros-positive R7 is present but out of the plane of focus. Scale bars, 5 μm.

Loss of Ttk69 from non-neuronal cone cells has been shown previously to transform them into R7 neurons (Siddall et al., 2009). Our GMR–FLP/MARCM method of creating ttk69 mutant R7s also creates ttk69 mutant cone cells and, as a consequence, causes some ommatidia to contain extra R7 neurons (Fig. 2B–B″). We found that 49.2 ± 3.5% (n = 7 retinas) of ommatidia that contained at least one ttk69 mutant R7 (whether the R7 itself or a transformed cone cell) also contained at least one additional R7 neuron. These extra R7s are indistinguishable from the native R7s in shape and in their expression of R7 markers (Fig. 2B–B″). However, because we cannot unambiguously trace the origin of each ttk69 mutant R7 axon, we were concerned that these ectopic R7s might somehow be abnormal and exclusively responsible for the observed axon morphology defect. To test this, we needed to remove Ttk69 from R7s without disrupting any non-neuronal cells. To do so, we used the R neuron-specific driver chp–Gal4 to drive expression of a ttk–RNAi transgene (Fig. 2C–F′). We found that this technique specifically eliminated Ttk69 expression from R neurons and therefore did not result in the formation of extra R7 neurons (Fig. 2F,F′). However, R7 axon terminals consequently extended laterally within the M6 layer, causing overlap with adjacent R7s at a frequency similar to that observed for individual ttk69 mutant R7 axons (21.8 ± 2.6%; n = 9 brains; Fig. 2E). We conclude that Ttk69 is normally required in R7 neurons to prevent abnormal axon terminal growth. In additional support of this conclusion, we found that using GMR–FLP/MARCM to drive expression of the ttk–RNAi transgene in individual homozygous wild-type (FRT82) cells did not cause non-neuronal cells to become extra R7s [data not shown; presumably this is because of the delay in expression (see below and Materials and Methods)] but did cause R7 axon terminals to extend laterally within M6 (18.7 ± 0.77%; n = 14 brains; Fig. 1E).

ttk69 mutant R7 axons are still responsive to repulsion by their neighbors

Previous studies have shown that wild-type R7 axons will extend laterally within M6 if two conditions are met: (1) there is at least one extra R7 in the same ommatidium, causing multiple R7 axon terminals to occupy the M6 space normally reserved for one (a condition we will refer to as “crowding”); and (2) neighboring R7 axons have been removed (Ashley and Katz, 1994). Under these circumstances, crowded R7 axon terminals will specifically extend into M6 regions lacking R7 innervation, leading to the conclusion that R7 axons are mutually repulsive but that a cell-intrinsic mechanism is normally sufficient to prevent R7 axon overlap (Ashley and Katz, 1994). As mentioned in the previous section, ttk69 mutant R7 axons can be crowded, satisfying condition 1. One possibility, therefore, is that ttk69 mutant R7 axons extend laterally within M6 because they are unable to detect the presence of adjacent R7 axons, satisfying condition 2. If this were the case, we would predict that removing adjacent R7 axons would have no effect on the ability of ttk69 mutant terminals to overlap with their neighbors. However, when we used a hypomorphic sev allele to eliminate most R7s, those ttk69 mutant R7 axons that remained were significantly more likely to extend laterally within M6 (Fig. 1E) We conclude that ttk69 mutant R7 axons are still responsive to repulsion by adjacent R7 axons.

ttk69 mutant R7 axons correctly recognize their synaptic target layers

In addition to being repelled by their neighbors, R7 axons presumably also receive cues from their synaptic targets. Although disrupting presynaptic assembly within R7s does not cause their axons to extend laterally (Clandinin et al., 2001; Lee et al., 2001; Maurel-Zaffran et al., 2001; Choe et al., 2006; Hofmeyer et al., 2006; Astigarraga et al., 2010; Holbrook et al., 2012), it remained possible that the ttk69 mutant R7 axon defect might nonetheless be caused by an inability to recognize synaptic targets or might reflect a failure to transition from axon growth to synaptogenesis. Indeed, although wild-type R7 terminals lack filopodia and are instead smooth, ellipsoid boutons by 60 h APF (Fig. 1A), ttk69 mutant R7 terminals failed to form normal boutons and instead were irregular in shape and had short, filopodia-like extensions (Fig. 1B). To distinguish whether ttk69 mutant R7 axons recognize their synaptic targets, we examined their ability to localize the synaptic vesicle marker Syt–GFP. In wild-type R7 axons, Syt–GFP is localized primarily below the M3 layer, consistent with the known locations of presynaptic sites (Fig. 3A,A′; Takemura et al., 2008). However, R7s lacking the presynaptic assembly proteins Syd-1 or Liprin-α mislocalize Syt–GFP to puncta along the length of the axon (Fig. 3B,B′; Holbrook et al., 2012). We found that ttk69 mutant R7s correctly localized Syt–GFP below the M3 layer (Fig. 3C,C′), although the amount of Syt–GFP in this region appeared to be increased in ttk69 mutant R7 axons (Fig. 3C,C′), perhaps because of their expanded size. This result suggests that ttk69 mutant R7s correctly identify their synaptic targets and can at least partly assemble synapses. In addition, ttk69 mutant R7 axons never extended outside the layers within which they normally form synapses: in particular, most remained within M6, and those that instead turned back along adjacent R7 axons never extended beyond M4 (Fig. 1B). This result suggests that the overgrowth of ttk69 mutant R7 terminals is restricted to regions that contain their normal synaptic targets.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

ttk69 mutant R7 axons correctly identify their synaptic target layer yet still overgrow if contact with their target layer is disrupted. A–F′, Medullas of mosaic animals, created by GMR–FLP/MARCM, at 60 h APF. R7s homozygous for the specified chromosome arm express Syt–GFP (green). All R7 and R8 axons are labeled with anti-Chp (red). A, A′, Wild-type (FRT82) (WT) R7s primarily localize Syt–GFP to the region of the axon below the M3 layer (dashed line). B, B′, syd-1 mutant R7s mislocalize Syt–GFP to regions of the axon that lie within and above the M3 layer. syd-1 mutant R7 axons often fail to contact M6. In ttk69 (ttk^1e11; C, C′) and imp-α3 (D, D′) mutant R7 axons Syt-GFP is primarily localized below M3, as in wild type. E–F′, imp-α3, syd-1 double-mutant R7 axons mislocalize Syt–GFP yet extend laterally within M6. F, Even double-mutant R7s that fail to contact M6 can still extend laterally (arrowheads). G, Quantification of R7 axons that either extend within or fail to contact M6. Error bars represent SEM. Loss of imp-α3 greatly increases the frequency with which syd-1 mutant R7 axons contact M6 [only 9.5 ± 1.3% (n = 19) of imp-α3, syd-1 R7 axons fail to contact M6 compared with 20.0 ± 1.6% (n = 13) of syd-1 mutant axons]. Loss of syd-1 moderately decreases the ability of imp-α3 mutant axons to extend laterally [15.8 ± 1.0% (n = 19) of imp-α3, syd-1 R7 axons overlap with neighboring R7 axons compared with 21.6 ± 0.93% (n = 11) of imp-α3 mutant axons]. Similarly, disrupting ttk with ttk–RNAi greatly increases the frequency with which syd-1 mutant R7 axons contact M6 [only 10.3 ± 0.84% (n = 13) of ttk–RNAi-expressing syd-1 mutant R7 axons fail to contact M6], and loss of syd-1 moderately decreases the ability of ttk–RNAi-expressing R7s to extend laterally [only 15.0 ± 0.99% (n = 13) of ttk–RNAi-expressing syd-1 mutant R7 axons extend within M6 (the percentage of ttk–RNAi-expressing wild-type R7 axons that do so is redisplayed from Fig. 1E for ease of comparison]. Finally, loss of ttk69 likewise greatly increases the frequency with which Ncad–RNAi-expressing R7 axons contact M6 [only 5.5 ± 0.44% (n = 14) of Ncad–RNAi-expressing ttk69 (ttk^1e11) mutant R7 axons fail to contact M6 compared with 16.2 ± 3.4% (n = 7) of Ncad–RNAi-expressing wild-type R7 axons]. Ncad–RNAi does not significantly affect the frequency with which ttk69 mutant R7 axons extend laterally [23.6 ± 1.3% (n = 14) of Ncad–RNAi-expressing ttk69 mutant R7 axons overlap with neighboring R7 axons (the percentage of ttk69 mutant R7 axons that do so is redisplayed from Fig. 1E for ease of comparison); p = 0.12]. Error bars represent SEM. *p < 0.01 and ***p < 0.0001. Scale bar, 5 μm. n indicates number of brains scored.

ttk69 mutant R7 axons still overgrow if contact with their M6 target layer is disrupted

The restriction of ttk69 mutant R7 axon terminals to regions containing R7 synaptic targets suggested that their ability to overgrow might require contact with M6 or the ability to form synapses. To test this, we set out to disrupt the latter processes in R7s lacking Ttk69. Loss of Ncad, LAR, Liprin-α, or Syd-1 reduces the size of R7 terminal boutons and causes many to fail to contact M6 (Clandinin et al., 2001; Lee et al., 2001; Maurel-Zaffran et al., 2001; Choe et al., 2006; Hofmeyer et al., 2006; Holbrook et al., 2012). For technical reasons, we were unable to create R7s doubly mutant for ttk69 and Ncad, LAR, or Liprin-α (see Materials and Methods). Instead, we took the following three approaches to disrupt both ttk69 and contact with M6: (1) we created imp-α3, syd-1 double-mutant R7s [loss of imp-α3 eliminates nuclear Ttk69 expression and exactly phenocopies all aspects of the ttk69 mutant R7 axon defect (Fig. 1E–H′), including normal localization of Syt–GFP (Fig. 3D,D′)]; (2) we disrupted Ttk69 protein in syd-1 mutant R7s by causing the latter to express ttk–RNAi; and (3) we disrupted Ncad protein in ttk69 mutant R7s by causing the latter to express Ncad–RNAi. We obtained similar results in all three cases. As expected, loss of syd-1 from imp-α3 mutant R7s disrupted their ability to localize Syt–GFP and to maintain contact with M6 (Fig. 3E–G), as did loss of syd-1 from ttk–RNAi-expressing R7s (Fig. 3G) or causing ttk69 mutant R7s to express Ncad–RNAi (Fig. 3G). However, even those imp-α3, syd-1 double-mutant R7 axons that failed to contact M6 were able to extend laterally (Fig. 3F,F′), suggesting that ttk69 mutant R7 axon overgrowth does not require contact with synaptic targets. In addition, significantly more imp-α3, syd-1 double-mutant and ttk–RNAi-expressing syd-1 mutant R7 axons made contact with M6 than did syd-1 single-mutant R7 axons (Fig. 3G), suggesting that loss of Ttk69 from syd-1 mutant R7s may cause their axons to overgrow not only laterally but also toward M6. Similarly, more Ncad–RNAi-expressing ttk69 mutant R7 axons made contact with M6 than did Ncad–RNAi-expressing wild-type R7 axons (Fig. 3G), again indicating that loss of Ttk69 causes R7 axons lacking Ncad to overgrow toward M6. However, we do note that loss of syd-1 from imp-α3 mutant R7s and loss of syd-1 from ttk–RNAi-expressing R7s reduces the frequency with which their axons extend laterally (Fig. 3G), indicating that, although not absolutely required, stable contact with M6 may facilitate R7 axon overgrowth.

Ttk69 is first expressed in wild-type R7s after their axons extend into the medulla; ttk69 mutant R7 axon terminal overgrowth begins shortly afterward

We have so far shown that loss of Ttk69 causes R7 axon terminals to overgrow, that this overgrowth does not require contact with the normal R7 synaptic target layer, and that this overgrowth is not caused by an inability to be repelled by neighboring R7 axons or to recognize the appropriate synaptic targets. Together, these results indicate that Ttk69 is required to downregulate some R7-intrinsic mechanism that promotes axon terminal growth. We next wanted to determine whether Ttk69 might also be sufficient to downregulate R7 axon growth and so play an instructive rather than simply permissive role in this process. To do so, we first determined the time course of Ttk69 expression to compare it with that of wild-type R7 axon outgrowth. Previous work has shown that Ttk69 is absent from cells as they differentiate into R neurons during larval development (Li et al., 1997) but is present at high levels in all R neurons at 40 h APF (Lai and Li, 1999). If Ttk69 is indeed instructive rather than merely permissive, one prediction is that it should be absent from R7 axons that are actively extending. We therefore expected that R7s would not express Ttk69 as they initially extend to their temporary target layer. Because R7s are specified by a wave of differentiation that passes across the developing retina (Wolff and Ready, 1993), the initial extension of R7 axons occurs sequentially rather than simultaneously and is complete by ∼30 h APF (Ting et al., 2005). Indeed, we found that Ttk69 was still undetectable in R7s at 24 h APF, when most R7 axons have completed their initial extension into the medulla (Fig. 4A,A′,L). However, by 30 h APF, Ttk69 was expressed at high levels in all R7s (Fig. 4B,B′,L). Ttk69 levels then decreased but persisted throughout the second phase of R7 axon targeting (Fig. 4C–E′,L).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Ttk69 is normally first expressed in R7s after their axons extend into the medulla, and ttk69 mutant R7 axon overgrowth begins shortly afterward. A–E′, Wild-type pupal retinas in which the R neurons are outlined by Phalloidin (red) and R7 nuclei express Pros (blue). Anti-Ttk69 is in green. A, A′, At 24 h APF, Ttk69 protein is absent from all R neurons including R7s (arrow). B–E′, From 30 h until at least 60 h APF, Ttk69 is expressed in all R neurons, including R7s (arrows; quantified in L). F–K, Pupal medullas in which homozygous R7 axons express mCD8–GFP (green). At 30 h APF, wild-type (F) and ttk69 (ttk^1e11) mutant (I) R7 axon terminals are indistinguishable. By 36 h APF, ttk69 mutant R7 axons extend laterally within the M6 layer (J). By 48 h APF, ttk69 mutant R7 axons have begun to extend back toward M3 (K, arrowhead). Quantified in M. L, Quantification of Ttk69 protein expression in wild-type R7 nuclei. Previous work has shown that Ttk69 protein is absent from third-instar larval R7s, suggesting that the 9.60 ± 1.5 (n = 15) units of fluorescence we observe at that time point is caused by nonspecific background staining. At 24 h APF, Ttk69 protein remains absent from R7s (9.97 ± 2.3 units; n = 11). However, by 30 h APF, R7s express a high level of Ttk69 (155 ± 17 units; n = 10). This level drops somewhat by 36 h APF (75.2 ± 6.0 units; n = 14) and further decreases by 60 h APF (57.6 ± 2.2 units; n = 14). The differences between 30 h APF and the later time points are highly significant, as is the difference between 48 and 60 h APF. The difference between 36 and 48 h APF (93.5 ± 4.5 units; n = 14) is not (p = 0.02). *p < 0.01 and ***p < 0.0001. M, Quantification of ttk69 (ttk^1e11) mutant R7 axon extension over time. Error bars represent SEM. At 36 h APF, 10.1 ± 0.8% (n = 10) of ttk69 mutant R7 axons overlap with neighboring wild-type R7 axons. The frequency of ttk69 mutant R7 axon terminals that overlap with their neighbors peaks by 42 h APF (21.2 ± 1.1%; n = 11) and remains unchanged at 48 h APF (21.6 ± 1.8%; n = 10) and 60 h APF (data for the latter time point is redisplayed from Fig. 1E for ease of comparison). The average length of the ttk69 R7 axon terminal extensions continues to increase even after the frequency of overlap has peaked: the difference between 36 h APF (2.9 ± 0.08 μm; n = 10) and 48 h APF (4.5 ± 0.3 μm; n = 10) is significant, as is the difference between 48 and 60 h APF (5.7 ± 0.1 μm; n = 9). **p < 0.001 and ***p < 0.0001. Scale bars, 5 μm. n indicates number of brains or retinas scored.

The absence of Ttk69 from R7s whose axons are actively extending is consistent with the possibility that Ttk69 might be an instructive signal for R7s to decrease axon growth. If such a model were true, we would also predict that ttk69 mutant R7 axons should exhibit increased growth shortly after Ttk69 would normally be expressed. Indeed, we found that wild-type and ttk69 mutant R7 axons were indistinguishable at both 24 h APF (data not shown) and 30 h APF (Fig. 4F,I). However, by 36 h APF, 10.1 ± 0.8% (n = 10 brains) of ttk69 mutant R7 axon terminals overlapped with their neighbors (Fig. 4G,J,M), and by 42 h APF, this overlap was as frequent as that observed at 60 h APF (Fig. 4H,K,M). The average length of ttk69 mutant lateral extensions continued to increase between 48 and 60 h APF (Fig. 4M). Together, these results are consistent with a model in which the timing of Ttk69 expression coincides with the second phase of R7 targeting, during which Ttk69 inhibits R7 axon growth.

Premature Ttk69 expression disrupts R7 axon extension

To directly test whether Ttk69 might be sufficient to inhibit R7 axon growth, we wanted to drive expression of Ttk69 prematurely, before the R7 axons have reached their temporary target layer. However, expressing Ttk69 in R7 precursors before they are committed to a neuronal fate has been shown previously to prevent their differentiation as R7s (Li et al., 1997; Siddall et al., 2009). To avoid this outcome, we therefore needed to delay ectopic expression of Ttk69 until after the R7s are specified. We chose not to use a global driver, such as the heat shock promoter, because that would result in Ttk69 induction simultaneously in all R7s despite their varying developmental ages; at any given time point, many would not yet be committed to the neuronal fate. Instead, we took advantage of the same GMR–FLP/MARCM system that we use to generate and label individual homozygous R7s. We have found that GMR–FLP/MARCM causes each homozygous R7 to express UAS-driven transgenes ∼10–15 h after its recruitment as an R7 and causes their axons to be GFP labeled before they have reached their temporary target layer (Ting et al., 2005; Miller et al., 2009). We therefore used GMR–FLP/MARCM to express UAS–ttk69 in wild-type R7s. By 24 h APF, most wild-type R7 axons have reached their temporary target layer in the medulla, although some have not (Fig. 5A; Ting et al., 2005). Although R7s forced to express Ttk69 prematurely can still reach their temporary target layer, we noticed that some appeared to terminate instead in the R8 temporary target layer and had abnormally thin terminals (Fig. 5B,B′). This result suggested that premature Ttk69 might collapse R7 growth cones and so inhibit axon growth.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Premature Ttk69 expression disrupts R7 axon extension. A–E, Pupal medullas in which homozygous R7s, created by GMR–FLP/MARCM, express both mCD8–GFP (green) and the UAS transgene indicated. All R7 and R8 axons are labeled with anti-Chp (red). A, By 24 h APF, most wild-type (FRT82) R7 axons have reached their temporary target layer in the medulla. B, B′, At 24 h APF, wild-type R7 axons forced to express Ttk69 prematurely can have abnormally thin terminals (arrow), although some appear indistinguishable from wild type (chevron). C, At 48 h APF, 17.6 ± 1.3% (n = 9) of wild-type R7 axons forced to prematurely express Ttk69 fail to contact their target layer (arrow). D, Forced Ttk69 expression rescues the overgrowth of ttk69 (ttk^1e11) mutant R7 axons (see also Fig. 1E) but also prevents 14.3 ± 1.6% (n = 10) from contacting their target layer (arrows). E, Forced Ttk88 expression neither rescues ttk69 mutant R7 axon overgrowth (see also Fig. 1E) nor prevents their axons from contacting their target layer. F–F″, At 48 h APF, the cell bodies of wild-type R7s that express Ttk69 prematurely (arrow) are normal in shape and express the R7 fate marker Pros. Scale bars, 5 μm. n indicates number of brains scored.

To test whether premature Ttk69 actually prevents R7 axons from reaching their targets, we next examined GMR–FLP/MARCM/UAS–ttk69 R7s at 48 h APF, by which time all wild-type R7 axons have reached their temporary target layer and have begun their slow advance to their final destination, M6. We found that 17.6 ± 1.3% (n = 9 brains) of wild-type R7 axons forced to express Ttk69 were absent from their normal target layer and instead terminated in the same layer as R8 axons (Fig. 5C); some of these apparently stalled axons extended thin filopodia-like processes toward their normal target layer (Fig. 5C). Even those Ttk69-expressing R7s that did reach the correct target layer had abnormally thin terminals. We observed similar results when we expressed Ttk69 in ttk69 mutant R7s (Fig. 5D). Because Ttk69, when expressed earlier, can transform R7 precursors into non-neuronal cells, we examined whether GMR–FLP/MARCM-mediated Ttk69 expression altered R7 fate. We found that Ttk69-expressing R7s underwent their normal change in polarity and continued to express the photoreceptor-specific marker Chp (Fig. 5C), the neuron-specific marker Elav (data not shown), and the R7-specific marker Pros (Fig. 5F–F″), indicating that other aspects of R7 development were unaffected. In additional support of the specificity of the effect of Ttk69, using GMR–FLP/MARCM to express the Ttk88 isoform had no effect on R7 growth (Fig. 5E), despite also being able to transform R7 precursors into non-neuronal cells if expressed earlier (Li et al., 1997; Tang et al., 1997). Together, these results suggest that premature Ttk69 expression can collapse R7 growth cones and prevent them from advancing and that Ttk69 does so independently of its effect on neuronal differentiation.

Ttk69 is required for normal Activin signaling in R7s

Previous work has shown that the Activin receptor Babo is required to downregulate R7 axon growth during the second phase of targeting (Ting et al., 2007). We therefore wanted to test whether Ttk69 might act upstream or downstream of one or more components of this pathway. If Ttk69 controlled R7 axon growth independently of Activin signaling, we would predict that disrupting both Ttk69 and the Activin pathway would have a more severe effect on R7s than disrupting either alone. The most effective way to eliminate Activin signaling in R7s is to create large clones of babo mutant tissue using ey3.5–FLP/MARCM (Ting et al., 2007). Consistent with previous reports, we found that 12.8 ± 0.68% (n = 9 brains) of babo mutant R7 axon terminals overlapped with their neighbors (Figs. 1D,6C). We found it technically easiest then to disrupt Ttk69 protein production in babo mutant clones by causing these clones to express a UAS–ttk–RNAi transgene. Expressing ttk–RNAi in wild-type clones using ey3.5–FLP/MARCM completely eliminated detectable Ttk69 in R7s (data not shown) and caused their axon terminals to overgrow at approximately the same frequency as ttk69 mutant R7 terminals (25.5 ± 0.66%; n = 14 brains; Fig. 6A,C), suggesting that this method completely disrupts Ttk69 function. We found that the frequency of axon overgrowth was not increased by expressing ttk–RNAi in babo mutant R7s (25.4 ± 1.3%; n = 12 brain; Fig. 6B,C), suggesting that Ttk69 and Babo likely function in the same pathway to regulate R7 axon growth.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Ttk69 is required for normal Activin signaling in R7s. A, B, Medullas of mosaic animals, created by ey3.5–FLP/MARCM, at 48 h APF. R7s homozygous for the specified chromosome arm express mCD8–GFP (green). All R7 and R8 axons are labeled with anti-Chp (red). A, Wild-type (FRT82) (WT) R7 axons that express ttk–RNAi extend laterally and overlap with neighboring R7 terminals at a frequency similar to that of ttk69 mutant R7 axons. B, babo mutant R7 axons that express ttk–RNAi also extend laterally and overlap with neighboring R7 terminals. C, Quantification of R7 axon extension within M6. Error bars represent SEM. babo mutant R7 axons overlap with their neighbors significantly less often (12.8% ± 0.68; n = 9) than wild-type R7 axons expressing ttk–RNAi (25.5 ± 0.66; n = 14). However, removing babo from R7s that express ttk–RNAi does not significantly increase the frequency of this defect (25.4 ± 1.3%; n = 12). D, D′, A 50 h APF retina in which homozygous wild-type (FRT82) R7s (created by GMR–FLP/MARCM; arrows) express mCD8–GFP (green) and FLAG–dSmad2 (red). All R neuron nuclei are labeled with anti-Elav (blue). Unlabeled R7 nuclei (arrowheads) do not express FLAG–dSmad2. E, Quantification of FLAG–dSmad2 levels in mutant R7 nuclei at 50 h APF. Error bars represent SEM. FLAG–dSmad2 staining in ttk69 (ttk^1e11; 1.48 ± 0.06; n = 10) and imp-α3 (1.37 ± 0.06; n = 10) mutant R7 nuclei is significantly reduced compared with that in wild-type R7 nuclei (1.82 ± 0.08; n = 10). Scale bars, 5 μm. n indicates number of brains or retinas scored. *p < 0.01 and ***p < 0.0001.

We next wanted to test directly whether Ttk69 might act upstream or downstream of the Activin/Babo pathway. We found that Ttk69 is expressed at normal levels in babo mutant R7 nuclei during the second phase of R7 axon targeting (data not shown), suggesting that Ttk69 is not regulated by Babo. We therefore tested whether, instead, Activin signaling might be regulated by Ttk69. The output of Activin signaling can be assessed by measuring nuclear accumulation of the effector molecule dSmad2. A previous comparison of wild-type and babo mutant R7s expressing a UAS–FLAG–dSmad2 transgene showed that loss of babo significantly decreased the amount of FLAG-tagged dSmad2 present in R7 nuclei (Ting et al., 2007). We therefore tested whether loss of ttk69 might have a similar effect. We expressed UAS–FLAG–dSmad2 in wild-type, ttk69, and imp-α3 mutant R7s and quantified the amount of nuclear FLAG-tagged dSmad2 in each case (Fig. 6D,D′). We found that loss of either ttk69 or imp-α3 significantly decreased nuclear dSmad2 compared with wild-type (Fig. 6E). We conclude that Ttk69 is required for normal Activin pathway signaling in R7s.

Ttk69 can also inhibit R7 axon growth independently of the Activin pathway

Although these results indicate that Ttk69 restricts R7 axon growth by promoting Activin pathway activity, disrupting the Activin pathway in R7s does not cause as severe a defect as loss of Ttk69. babo mutant R7 axon terminals overlap with neighboring terminals less frequently [12.8 ± 0.68%, n = 9 (Fig. 6C) compared with 20.8 ± 0.70%, n = 9 (Fig. 1E); p < 0.00001] and have shorter extensions [2.6 ± 0.2 μm (n = 7) at 60 h APF compared with 5.7 ± 0.1 μm (n = 9; Fig. 4M); p < 1 × 10⁻⁸], and those babo mutant R7 terminals that do not overlap with their neighbors have terminals of normal size (Fig. 1D,F). This comparison suggests that Ttk69 might additionally inhibit R7 axon growth independently of the Activin pathway. To test this directly, we examined whether premature Ttk69 misexpression could inhibit R7 axon growth in the absence of Babo. We chose to disrupt Babo expression using a UAS–babo–RNAi transgene, because expressing Ttk69 in babo mutant clones using ey3.5–FLP/MARCM causes very early expression of Ttk69 and so blocks R7 differentiation (data not shown). We expressed both UAS–babo–RNAi and UAS–ttk69 using chp–Gal4, which initiates expression in R neurons after their specification as neurons (data not shown). We found that expressing babo–RNAi alone caused R7 axon terminals to overlap with their neighbors at the same frequency as babo mutant R7 terminals (13.9 ± 0.9%; n = 8; Fig. 7A,D). Expressing Ttk69 alone using chp–Gal4 caused R7 axons to stop short of their final target layer only slightly less often than GMR–FLP/MARCM/UAS–ttk69 R7 axons [11.5 ± 0.6%, n = 11 (Fig. 7B,D) compared with 17.6 ± 1.3%, n = 9 (Fig. 5C)] and did not disrupt their neuronal identity (data not shown). We therefore examined the effects of expressing Ttk69 and babo–RNAi simultaneously. We found that forced Ttk69 expression caused only a mild reduction in the frequency with which Babo-lacking R7 axons extended laterally within M6 (Fig. 7C,D), suggesting that Ttk69 acts upstream of Babo to promote Activin signaling. However, if this were the only mechanism by which Ttk69 restricts R7 axon growth, then disrupting Babo should similarly prevent Ttk69 from causing premature arrest of R7 axons. Instead, disrupting Babo did not significantly decrease the frequency with which Ttk69-expressing R7s failed to reach M6 (Fig. 7D). Together with the observed difference between the ttk69 and babo loss-of-function phenotypes, these results suggest that Ttk69 can inhibit R7 axon growth independently of the Activin pathway.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Ttk69 also acts via an Activin-independent pathway. A–C, Medullas of adult animals in which R1–R8 neurons express EB1–GFP (white) and the UAS transgene(s) indicated under the control of chp–Gal4. A, R7 axons that express babo–RNAi extend laterally and overlap with neighboring axon terminals at a frequency similar to that of babo mutant R7 axons (13.9 ± 0.9%; n = 8; arrows; see Fig. 6C for comparison). B, R7 axons forced to express Ttk69 under the control of chp–Gal4 fail to contact the M6 layer at approximately the same frequency (11.5 ± 0.6%; n = 11; arrowheads) as R7 axons forced to express Ttk69 prematurely by GMR–FLP/MARCM (see Fig. 5 for comparison). C, R7 axons that coexpress Ttk69 and babo–RNAi exhibit both phenotypes: lateral extension within M6 (9.7 ± 0.7; n = 11; arrows) and failure to contact M6 (9.1 ± 1.4%; n = 7; arrowheads). D, Quantification of R7 axon extension within M6 and failure to contact M6. Error bars represent SEM. Expression of Ttk69 partially rescues the R7 lateral extensions caused by expression of babo–RNAi (*p = 0.002). Although expression of babo–RNAi may reduce the ability of Ttk69 to prevent R7s from contacting M6, the difference is not statistically significant (p = 0.1). Scale bar, 5 μm. n indicates number of brains scored.