Drosophila Psidin Regulates Olfactory Neuron Number and Axon Targeting through Two Distinct Molecular Mechanisms

A mutation in the psidin gene affects olfactory neuron circuit formation

Using mosaic histological screening of ∼6000 EMS mutants, we addressed the molecular mechanisms required for ORN axons to navigate to their correct targets within the AL. We used the eyFlp mosaic system, which induces mutant clones in the antenna and MP in addition to in the eye, but not in the central brain (Newsome et al., 2000). These clones usually resulted in ∼50% of all ORNs being homozygous for the respective mutant allele. Using specific reporter lines for distinct ORN classes (see Materials and Methods), we analyzed the effects that the mutations had on the synaptic targeting patterns of different ORN subclasses.

We identified a mutant IG978 displaying ORNs that failed to innervate their correct target glomeruli (Fig. 1A). Using deletion and SNP mapping, we mapped the mutant to the psidin locus (phagocyte signaling impaired; Fig. 1C). Psidin is the homolog of the yeast protein MDM20. MDM20 is the noncatalytic subunit of the N-acetyltransferase complex NatB known in yeast to acetylate ∼60% or more of all cytosolic proteins on specific N-terminal amino acid residues (Polevoda and Sherman, 2003a; Singer and Shaw, 2003). Psidin was identified previously in Drosophila in two independent screens: as a lysosomal protein in blood cells (Brennan et al., 2007; Kim et al., 2011) and as a novel actin-binding protein essential for oocyte migration (Kim et al., 2011). To validate and understand the role of Psidin during olfactory circuit formation, we analyzed three different alleles in parallel: our newly established psidin^IG978 allele (harboring the missense mutation E320K) and two expected null alleles, psidin¹ (Brennan et al., 2007) and psidin^55D4 (Kim et al., 2011) (harboring a STOP codon at K441 and K471, respectively; Fig. 1C). These analyses revealed two phenotypes. First, psidin null alleles revealed an apparent reduction in axon numbers in several, but not all, ORN classes (Fig. 1A,E; Table 1). This phenotype was caused by a reduction in neuron numbers of these ORN classes, which was already visible at around 8 h after puparium formation (APF) in developing pupae (Fig. 1E and data not shown), and therefore seemed to relate to a defect in early neurogenesis or cell survival (see below). Psidin^IG978 mutant clones showed a trend similar to psidin¹ mutants but no significant change in neuron numbers compared with controls (Fig. 1E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Psidin is required for pathfinding of a subset of olfactory receptor neurons. Adult brains were stained with anti-GFP (green) and NC82 (magenta) to visualize synaptic innervation of ORNs in the AL. In forward MARCM, only mutant cells are visible selectively labeled in green in a wild-type background. Conversely, wild-type cells are labeled in a mutant background in reverse MARCM. MARCM analysis was done using eyFlp or hsFlp to generate clones. A, Or59c and Or42a mutant axons show a strong defasciculation phenotype in the psidin¹ and psidin^55D4 background (dashed lines). Or47a psidin mutant axons show ectopic synapse formation in ventral parts of the AL (arrows). Centrally projecting Or22a mutant neurons display no mistargeting phenotype but a weaker innervation (arrows). All ORN classes show a strong cell-loss phenotype in the psidin¹ and psidin^55D4 background. B, Reverse MARCM analysis of eyFlp clones. Or59c, Or42a, and Or47a wild-type neurons project normally in the psidin¹ or psidin^IG978 mutant background. Forward MARCM analysis using hsFlp showed that small psidin^IG978 clones are sufficient to cause mistargeting in Or47a axons. C, Protein structure of Psidin and the used psidin alleles. Psidin contains a Tpr-domain and coiled coils at the C terminus. A “NatB interaction domain” is predicted in the center of Psidin. D, ORN classes are affected differently in targeting according to their projection route in psidin¹ mutants. E, Mutant cells in the adult ANT and MP were counted using Or-gal4 and UAS-mCD8-GFP. The number of neurons is significantly reduced in the psidin¹ loss-of-function mutant background in all classes. Although the neuron number is not significantly reduced in psidin^IG978 mutants, there is a trend toward a reduction. Counting of Or83b-positive cells (Orco) in the MP revealed a 35% reduction of neurons in the psidin¹ background. *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA and Bonferroni's posttest. Error bars ± SEM. Scale bar, 20 μm.

Second, all three psidin mutant alleles revealed defects in glomerulus targeting, consistently affecting the same subclasses of ORNs (Fig. 1A,D; Table 1). The nature of these phenotypes was comparable regardless of whether ORN numbers were reduced (psidin¹, psidin^55D4) or unaffected (psidin^IG978). Or47a-expressing ORNs target the dorsomedial glomerulus DM3 in control animals (Fig. 1A). In psidin mutants, they frequently innervated a ventromedial glomerulus (Fig. 1A,D). The majority of the ventromedial targeting ORNs (e.g., Or59c, Or42a, and Or92a) displayed strong mistargeting and ectopic synapse formation (Fig. 1A,D; Table 1). In contrast, centrally projecting ORNs like Or22a, Or47b, and Or71a axons, which travel a short distance at a straight angle from the AL entry point, were hardly affected (Fig. 1A,D; Table 1).

Since the targeting tissue in the AL was not genetically manipulated by eyFlp, potential phenotypes were expected to affect the ORNs but not their target cells in the brain. We sought additional evidence that Psidin was required within the mistargeting ORN by using single cell and reverse MARCM analysis to visualize selectively wild-type axons in a mutant background. In these experiments, wild-type Or47a, 42a, and 59c ORNs within nonlabeled mutant ORNs projected normally, confirming that Psidin is required in a cell-autonomous manner in ORNs (Fig. 1B). In addition, hs-Flp-mediated recombination to induce very small mutant ORN clones showed that single mutant axons mistargeted in psidin clones (Fig. 1B). To confirm the requirement of Psidin, we used the GAL4/UAS system to reexpress Psidin in ORNs homozygous for psidin (eyFLP clones; Fig. 2A). In these experiments, targeted expression of Psidin achieved successful rescue of ORN cell numbers and pathfinding phenotypes (Fig. 2A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Cell-loss phenotype can be rescued independent from the targeting phenotype. The entire ORN population was visualized using a direct-fusion Or-mCD8-GFP. Act-GAL4 was used to drive the expression of wild-type Psidin or p35 in eyFlp clones. A, A mistargeting phenotype is visible in both the psidin^IG978 and psidin¹ mutant backgrounds, displaying mildly and strongly dispersed (dashed circles) glomerular innervation, respectively. Additionally, in the psidin¹ background, axons mistarget to dorsal parts of the AL (arrowheads). The normal targeting pattern is restored after reexpression of wild-type Psidin in the respective mutant background. In contrast, expression of p35 does not rescue the mistargeting phenotype. B, We observed a strong loss of neurons in the psidin¹ background. In the psidin¹ background, the ORN number was reduced by 36% compared with wild type. Reexpression of wild-type Psidin rescues the ORN cell number in the psidin¹ background. Similarly, expression of p35 restores the neuron number in the psidin loss-of-function background. The neuron number was not significantly changed in the psidin^IG978 background. C, Quantification of the mistargeting phenotype. In the psidin¹ mutant background, 41 and 32% of the neurons display a strong or mild mistargeting phenotype, respectively. Mistargeting phenotype is completely rescued after expression of wild-type Psidin. Expression of p35 reduces the targeting phenotype only to levels comparable to psidin^IG978 mutants. Similarly, in the psidin^IG978 background, reexpression of wild-type Psidin rescues the phenotype completely. *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA and Bonferroni's posttest. Error bars ± SEM.

To understand the reason for the reduced number of ORNs, we expressed the antiapoptotic and caspase-inhibiting protein p35 (Davidson and Steller, 1998) in psidin¹ and psidin^IG978 mutant and control clones. In these experiments, the cell number defect of psidin¹ mutant ORNs was fully rescued, but axonal targeting defects persisted (Fig. 2). These findings suggest that Psidin prevents apoptosis of ORNs during development. It also strongly supported the conclusion from our anatomical studies that the role of Psidin in axon navigation represents an independent function. Therefore, we conclude that Psidin-deficient defects in olfactory circuit formation are caused by two distinct cellular mechanisms.

Psidin is expressed at the right stages in the developing olfactory organs

We next analyzed the expression of psidin during different developmental stages using in situ hybridization and antibody staining (Fig. 3). Both methods suggested that Psidin is expressed ubiquitously in all cells, including the developing olfactory organs. Using in situ hybridization, psidin expression was detected at 5–10 h APF in the eye-antennal disc, the primordia of MP and antenna (Fig. 3A,B). This expression is consistent with a cell-autonomous role of Psidin in ORNs during neuronal differentiation, since ORNs arise at 6–20 h APF (Rodrigues and Hummel, 2008). At around 24 h APF, expression reached its maximum in both the MP and the antenna (Fig. 3C,D). At 30 and 45 h, expression started to decline in the olfactory organs (Fig. 3E–H). This expression is coherent with a role in the axonal pathfinding of ORNs, known to start around 20 h APF and continue over a period of ∼30 h (Rodrigues and Hummel, 2008). We next asked whether Psidin protein was expressed in ORNs during development. ORs start to be expressed only after development of the olfactory system. Therefore, we used the pan-neuronal marker Elav to label all neurons. Analysis of protein expression using an anti-Psidin antibody showed that Psidin protein was expressed in neuronal marker Elav-positive cells likely to be developing ORNs, judging by location in the antennal disc (Fig. 3I). Thus, we conclude that Psidin is expressed during the relevant developmental times in ORNs both during neuron differentiation and during axonal targeting.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Psidin is expressed in the developing olfactory organs. A–H, In situ hybridization (ISH) for psidin at different developmental stages. A, At 6 h APF, the psidin ISH signal can be detected in the developing eye-antennal discs (arrow). C, E, The expression stays high in the developing antenna and MPs both at 24 and 30 h APF (arrows). G, Psidin expression starts to decline at 45 h APF. B, D, F, H, Control (sense) probes for psidin do not show any signals. I, Staining of a developing antennal disc in 6 h APF pupae. Psidin is widely expressed in the developing antenna at this stage and partially correlates with elav-positive cells. Scale bars, 100 μm.

Psidin promotes lamellipodia of neuronal growth cones

Recent data showed that Psidin is a novel actin-binding protein required for the migration of Drosophila oocytes (Kim et al., 2011). To unravel how Psidin might contribute to the process of axonal navigation and whether similar mechanisms are required as for oocyte migration, we turned to cultures of embryo-derived primary Drosophila neurons that are ideally suited for the genetic dissection of cytoskeletal mechanisms underpinning axonal growth (Matusek et al., 2008; Sánchez-Soriano et al., 2010; Gonçalves-Pimentel et al., 2011). We found that psidin¹, psidin^55D4, or psidin^IG978 mutant neurons displayed significantly smaller lamellipodia compared with wild-type controls (Fig. 4A,B). These findings were consistent with potential roles of Psidin in actin regulation. Furthermore, neurites were significantly longer (Fig. 4B), similar to observations for other actin-regulating proteins with known pathfinding defects in vivo (Sánchez-Soriano et al., 2010). HA-tagged Psidin partially colocalized with actin filaments in neurites and growth cones (Fig. 4C). These data were in agreement with the above-mentioned report that Psidin binds actin and regulates lamellipodia during border cell migration in Drosophila oocytes (Kim et al., 2011). In that study, Psidin was suggested to antagonize the function of the actin-stabilizing protein Tropomyosin 1 (Tm1), thus maintaining actin networks in a dynamic state. To test whether a similar mechanism might apply to Drosophila neurons, we analyzed tm1^ZCL0722 mutant neurons and found no detectable phenotypes (Fig. 4A,B). However, when the tm1^ZCL0722 mutation was combined with the psidin^55D4 null mutant, the Psidin-deficient axon length and lamellipodial phenotypes were reverted to wild-type levels (Fig. 4A,B). These data were consistent with observations in non-neuronal cells and suggested that the psidin mutant phenotype is caused by unrestrained Tm1 activity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Psidin regulates lamellipodia size in the growth cones. A, B, Growth cones (asterisks, A) are significantly smaller in psidin¹, psidin^55D4, and psidin^IG978 neurons compared with wild type. Specifically, neurons lacking Psidin have significantly smaller lamellipodia compared with wild type. Additionally, psidin mutant axons are longer than wild-type axons. Tm1^ZCL0722 neurons displayed no lamellipodia phenotype. Lamellipodia size is restored in the psidin^55D4 plus tm1^ZCL0722 double mutants. C, Primary culture of 6-h-old neurons from wild-type embryos overexpressing HA-tagged Psidin. It localizes to the growth cone together with actin and is visible in the lamellipodia. Neurons were stained with anti-tubulin, phalloidin, and anti-HA (Psidin). D, F-actin pelleting assay. Arrows indicate F-actin, and arrowheads represent the respective proteins: α-actinin, BSA, Psidin, and Psidin^IG978. The indicated proteins were incubated in the presence (+) and absence (−) of F-actin. Reaction mixtures were incubated and centrifuged at 150,000 × g. Supernatant (S) and pellet (P) were then separately loaded onto an SDS gel. The positive control, α-actinin, can be found in the pellet together with F-actin. In contrast, the negative control, BSA, can only be found in the supernatant. Both, wild-type Psidin and Psidin^IG978 are only found in the pellet in the presence of F-actin. Both proteins seem to pellet with F-actin at comparable levels. *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA and Bonferroni's posttest. Error bars ± SEM.

We asked whether the point mutation in Psidin^IG978 that causes defects in lamellipodia size and axon targeting, but not in cell survival, would interfere with actin binding and thereby explain the observed phenotype. Using an in vitro actin-binding assay, we found that Psidin^IG978 bound actin at similar levels as wild-type Psidin (Fig. 4D). Thus, unidentified mechanisms in addition to actin binding must be involved in Psidin's regulation of lamellipodia size and actin dynamics.

Roles of Psidin in maintaining dynamic actin networks can explain its in vivo roles

We next tested whether the role of Psidin in restraining Tm1 activity in growth cones is of relevance for the axon-targeting phenotype of ORNs in vivo (Fig. 5). For this, we analyzed the two independent loss-of-function mutant alleles, tm1^ZCL0722 (Morin et al., 2001; Kim et al., 2011) and tm1^su(flw)4 (Vereshchagina et al., 2004), using eyFlp analysis. Consistent with our findings in growth cones, both alleles displayed no defects, neither in axonal projection patterns nor in the number of Or59c-expressing ORNs (Fig. 5A–C). Also consistent with our findings in primary neuron cultures, ORN homozygous mutants for both psidin^55D4 and tm1^ZCL0722 showed a marked reduction in the percentage of brains that display Or59c axonal targeting defects compared with psidin^55D4 mutants alone (Fig. 5A,B; 70% vs 20%). However, the psidin^55D4, tm1^ZCL0722 double-mutant condition did not rescue the reduction in the number of Or59c-expressing ORNs typical of the psidin^55D4 or psidin¹ single-mutant backgrounds (Fig. 5C). These data suggest that Psidin functions as a regulator of actin dynamics in conjunction with Tm1 during axon targeting (see also Kim et al., 2011). In contrast, it implicates a Tm1-independent mechanism of Psidin during neuronal survival.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Psidin and Tropomyosin have opposing roles during axon targeting. A, Or59c neurons in a psidin¹ and a psidin^55D4 mutant background showed a complete loss of innervation at their innate glomerulus (dashed circles). The tropomyosin mutant tm1^ZCL0722 did not show any targeting defect (dashed circles). The targeting pattern of psidin single mutants was restored in psidin^55D4 plus tm1^ZCL0722 double mutants (dashed circles). B, Single mutants of tm1^ZCL0722 show no targeting defect, whereas psidin¹ (72%) and psidin^55D4 (71%) single mutants showed a pronounced overall targeting defect. The targeting defect of psidin^55D4 mutants was rescued in the double-mutant psidin^55D4 plus tm1^ZCL0722 (25%). C, The loss of Or59c neurons in the psidin^55D4 background (n = 25) was not rescued in the double mutant (n = 20). The tm1^ZCL0722 background did not show any significant reduction of Or59c neurons. D, E, Overexpression of constitutively active (const. act.) Cofilin partially rescues the psidin¹-targeting phenotype (72% vs 58%). Expression of inactive (inact.) Cofilin aggravates the phenotype (83%). Expression of both cofilin variants leads to a mild phenotype in the wild-type background. F, G, Overexpression of LimK in a wild-type background leads to a strong targeting defect. Axons innervate along the outer rim of the AL (red arrowheads). Knockdown and overexpression of a kinase inactive form leads to a mild phenotype. The overall targeting phenotype in psidin¹ (72%) can be further aggravated by the overexpression of LimK (88%). Knockdown or expression of DN LimK (72% vs 31% or 38%) rescues the targeting phenotype, especially the strong targeting phenotype. *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA and Bonferroni's posttest. Error bars ± SEM.

Tm1 acts primarily as a stabilizer of F-actin (Bugyi and Carlier, 2010). Another factor reported to compete with Tm1 is ADF/cofilin (Kuhn and Bamburg, 2008). Therefore, we tested whether Drosophila cofilin (called Twinstar/Tsr) has an impact on Psidin function. For this, we overexpressed a constitutive active (S3A) and a constitutive inactive version (S3E) of Tsr in wild-type and psidin¹ mutant clones. Overexpression of either of the two Tsr isoforms in wild-type clones resulted in mild mistargeting of Or59c ORN axons in ∼20% of the brains (Tsr^S3A, 20%; Tsr^S3E, 23%; Fig. 5D,E). However, constitutive active Tsr^S3A overexpressed in psidin¹ mutant Or59c ORNs reduced their mistargeting phenotype by 14% compared with psidin¹ alone, consistent with a compensatory role of cofilin through competing with Tm1 for actin. Vice versa, inactive Tsr^S3E aggravated the psidin¹ phenotype (8% increase; Fig. 5D,E), likewise consistent with the above idea.

Cofilin function in F-actin disassembly is negatively regulated by phosphorylation through LimK, which was previously implicated in ORN targeting (Ang et al., 2006). We found that expression of wild-type LimK in wild-type ORNs caused 88% of Or59c ORNs to mistarget (Fig. 5F,G). Expression of LimK RNAi or of a kinase-inactive LimK affected the targeting of only 10% of ORNs (Fig. 5F,G). Expression of wild-type LimK in psidin¹ mutant ORNs resulted in a comparable degree of targeting defects as the same experiment caused in wild-type neurons (87%). This value was slightly higher than that of psidin¹ mutant ORNs without targeted LimK expression (Fig. 5F,G). Consistent with the idea that Psidin promotes actin dynamics and instability, overexpression of LimK RNAi or kinase-inactive LimK (i.e., those conditions that promote cofilin activity) significantly rescued the psidin¹ mutant axonal targeting phenotype (Fig. 5F,G). The results confirm that Psidin's function as an actin regulatory protein, originally discovered in non-neuronal cells, can be applied in the context of neuronal guidance. Furthermore, we conclude that Psidin's role in actin dynamics is required for growth cone targeting but is not involved during ORN survival.

Neuron number but not axon targeting depend on Psidin as a subunit of the N-acetyl transferase B complex

We next addressed potential molecular mechanisms of Psidin function during the regulation of neuron number. Psidin is the homolog of yeast MDM20, the noncatalytic subunit of the evolutionarily well conserved NatB complex required for N-terminal protein acetylation (Polevoda and Sherman, 2003a). In budding yeast, natB subunit mutants show defects including mitochondrial inheritance, budding, and cell division (Polevoda and Sherman, 2003b; Singer and Shaw, 2003). A role in cell growth was also found using human cell culture (Starheim et al., 2008). To test whether Psidin's role in ORN survival reflects a function as a constituent of the NatB complex, we investigated the Drosophila gene CG14222 (referred to as dNAA20 in the following) encoding the predicted catalytic subunit of NatB. Because of the lack of dNAA20 mutants or suitable deletions, we used eyFlp in combination with a strong driver (act-GAL4) and expressed three independent RNAi constructs selectively in ORNs (Fig. 6A). All constructs showed comparable phenotypes (Fig. 6A). The RNAi used for the experiments achieves strong knockdown of dNAA20 levels (Fig. 6B), but its expression in ORNs of wild-type flies was insufficient to cause neuron loss or axon-targeting phenotypes in Or59c or Or42a neurons (Fig. 6C,E,F,H). Expression of dNAA20 RNAi in clones mutant for the hypomorphic allele psidin^IG978, however, resulted in a significant decrease in ORN numbers, reminiscent of phenotypes observed in psidin¹ mutant clones (Fig. 6D,G). This suggests that psidin^IG978 provides a sensitized background for knockdown of dNAA20, and it is consistent with the trend in neuron number reduction in hypomorphic mutant clones (see Fig. 1). Or59c and Or42a neuron numbers in psidin¹ null mutants were not further decreased by dNAA20 RNAi expression, suggesting that Psidin's role in neuron survival is NatB dependent (Fig. 6D,G). In contrast to the reduction in cell number, Or59c and Or42a ORNs did not show any enhancement in mistargeting or ectopic synapse formation when dNAA20 RNAi was expressed in psidin^IG978 or psidin¹ mutants (Fig. 6E,H). These results indicate that dNAA20 is dispensable for ORN targeting but is required for the survival of the correct number of ORNs in a Psidin-dependent manner.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Psidin interacts with dNAA20 in vivo A, The effect of dNAA20 on ORN targeting was assessed using RNAi. Act-GAL4 was used to drive the expression of RNAi in eyFlp clones. Three different RNAi lines were used to knock down dNAA20 in Or42a neurons. All three lines had the same effect. B, Quantification of dNAA20 knockdown. S2 cells were transfected with dNAA20-myc and RNAi against dNAA20 (line#2). After coexpression of the RNAi against dNAA20, protein levels were reduced by ∼70–80% compared with expression of dNAA20-myc alone. C, Knockdown using RNAi (line#2) of dNAA20 in the psidin^IG978 mutant background showed a loss of glomerular innervation of Or42a neurons similar to the psidin¹ mutant background. D, Quantification of the total Or42a cell population: knockdown leads to reduction of Or42a neurons comparable to levels of psidin¹ mutants. No further reduction was observed in the psidin¹ mutant after dNAA20 knockdown. E, Quantification of the Or42a mistargeting phenotype expression of RNAi had no effect on the targeting phenotype. F, The effect of dNAA20 on Or59c targeting. The targeted glomerulus of Or59c neurons appears not to be innervated in psidin¹ mutant flies (dashed circles). Axons display an overall random innervation pattern across the AL (arrowhead). The targeting phenotype does not change after expression of RNAi. Comparably, the expression of RNAi has no effect on psidin^IG978 mutant neurons. G, Quantification of the total Or59c cell population: knockdown using RNAi (line#2) of dNAA20 in the psidin^IG978 mutant background leads to a significant reduction of Or59c neurons, to levels comparable to psidin¹ mutants. No further reduction of the neuron number was observed after knockdown in psidin¹ mutants. The knockdown of dNAA20 had no effect on wild-type cells. H, Quantification of the targeting phenotype of Or59c axons expression of RNAi in psidin¹ and psidin^IG978 had no effect on targeting. *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA and Bonferroni's posttest. Error bars ± SEM.

The central domain of Psidin binds the catalytic NatB subunit dNAA20

We next tested whether dNAA20 and Psidin physically interact. To this aim, we used coimmunoprecipitation assays of HA-tagged Psidin with myc-tagged dNAA20, both coexpressed in S2 cells (Fig. 7A). Pulldown of wild-type Psidin with anti-HA antibody resulted in significant amounts of coimmunoprecipitated dNAA20 (Fig. 7A). These data show that Psidin and dNAA20 form a complex, as has similarly been reported for their yeast and human homologs (Singer and Shaw, 2003; Starheim et al., 2008). Comparable coimmunoprecipitation was achieved with Psidin^IG978, suggesting that this mutation does not interfere with NatB complex formation (Fig. 7A). The same physical interaction was observed in the reverse experiment using anti-myc-mediated pulldown (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Psidin interacts with dNAA20 in vitro A, S2 cells were cotransfected with Psidin-HA/Psidin^IG978-HA and dNAA20-myc. dNAA20 was coimmunoprecipitated with wild-type Psidin (arrowhead) and Psidin^IG978 (arrow) isoforms in S2 cells at comparable levels. The protein Psidin^IG978 carries the point mutation found in psidin^IG978 mutants. B, The position E320 is highly conserved in higher organisms having a glutamate at this position. Interestingly, the yeast wild type resembles exactly the mutation found in the psidin^IG978 allele. C, Representative examples of coimmunoprecipitated dNAA20-myc with various Psidin deletions. The top blot shows the different Psidin deletions (anti-HA blot). The bottom blot displays the amount of dNAA20 that was pulled down together with the Psidin variants (anti-myc blot). As a reference, wild-type Psidin pulldown of dNAA20 was used (arrowhead, left lane) to compare the pulldown of dNAA20 with Psidin deletions (arrow, right lane). Deletion of the entire NatB interaction domain in Psidin^ΔNatBfull leads to loss of dNAA20 (arrow) pulldown. Smaller deletions of this domain (Psidin^ΔNatB23, Psidin^ΔNatB2) showed a reduced pulldown of dNAA20 (arrow). Only Psidin^ΔNatB1 was able to pull down dNAA20 (arrow) at levels close to wild-type Psidin (arrowhead). D, Quantification of the Western blots showing the normalized pulldown of dNAA20 with different Psidin isoforms. Psidin^IG978, Psidin^S678A, and Psidin^ΔNatB1 pulled down dNAA20 at wild-type levels. In contrast, Psidin^S678D, Psidin^ΔNatBfull, Psidin^ΔNatB23, and Psidin^ΔNatB2 show a reduced or no pulldown of dNAA20, respectively. E, Representative examples of coimmunoprecipitated dNAA20-myc with Psidin phosphomutants. The top blot shows the Psidin phosphomutant (anti-HA blot). The bottom blot displays the amount of dNAA20 that was pulled down together with the phosphomutant (anti-myc blot). As a reference, wild-type Psidin pulldown of dNAA20 was used (arrowhead, left lane) to compare the pulldown of dNAA20 with Psidin deletions (arrow, right lane). Psidin^S678A was able to pull down dNAA20 (arrow) at levels similar to wild-type Psidin (arrowhead). In contrast, Psidin^S678D showed a reduced pulldown of dNAA20 (30% of wild-type Psidin). F, The serine residue at the position 678 in Drosophila is highly conserved among other organisms. Again, in yeast this residue seems not to be conserved.

Next we sought to map the interaction domain of Psidin with dNAA20 within the Psidin protein. In addition to the Tpr- and coiled-coil domains, a so-called putative NatB interaction domain is predicted by some software programs (Fig. 1B and SMART Heidelberg). We deleted this domain entirely (ΔNatB-full) or in parts (ΔNatB 1–3) from Psidin and used it to test coimmunoprecipitation of dNAA20 from S2 cell lysates (Fig. 7C). The deletion of the entire domain resulted in complete loss of coimmunoprecipitated dNAA20 (Fig. 7C). Subdeletions of the N-terminal, middle, or C-terminal parts of the domain resulted in partial loss of interaction (Fig. 7C,D). The strongest interaction was maintained with a deletion of the region where Psidin^IG978 carries the point mutation E320K (ΔNatB 264–378; Fig. 7C). We quantified the level of bound dNAA20 to the Psidin deletions relative to the binding of dNAA20 to the wild-type Psidin protein (Fig. 7D). We conclude that the putative NatB interaction domain is indeed required for the interaction of dNAA20 and Psidin.

Phosphorylation of a conserved serine in Psidin prevents NatB complex formation and interferes with Psidin's function during ORN survival

Our next experiments were designed to more precisely distinguish the dNatB-dependent and -independent roles of Psidin in vivo. Deleting the entire or parts of the NatB interaction domain appeared very crude and not well suited for in vivo experiments. Therefore, we addressed potential mechanisms for regulated interaction between dNAA20 and Psidin. Previously, a conserved serine (S678 in fly; Fig. 7F) has been shown to be phosphorylated in human MDM20 using mass spec analysis (Trost et al., 2009). As this serine is located just downstream to the NatB interaction domain, we generated mutants that would either mimic phosphorylation (S678D) or make phosphorylation impossible at this location (S678A). The nonphosphorylatable Psidin mutant S687A coimmunoprecipitated dNAA20 at similar levels to Psidin^IG978 and wild-type Psidin (Fig. 7D,E). In contrast, the phosphomimetic mutant S678D pulled down significantly lower amounts of dNAA20 compared with wild-type Psidin and Psidin^IG978, but similar levels to the Psidin deletion mutants (Fig. 7D,E, Psidin^S678D vs Psidin^ΔNatB2). Hence, this amino acid mutation provides an excellent opportunity to ask whether binding to dNAA20 is required for both of the in vivo functions we observed for Psidin. We generated transgenic flies and reexpressed three different point mutations within the Psidin protein in psidin mutant clones (Fig. 8A–C). First, we expressed Psidin^IG978 in psidin¹ and wild-type clones. UAS-psidin^IG978 fully rescued the cell-loss phenotype in psidin¹ clones (Fig. 8B). In contrast, the axon-targeting defect remained and was rescued only to the level of psidin^IG978 mutant clones (Fig. 8C). This result confirms that the single-point mutation in Psidin^IG978 is sufficient to interfere with axon targeting without affecting ORN numbers. Next, we expressed psidin^S678A and psidin^S678D in mutant and control clones. While expression of Psidin^S678A rescued both the cell-loss and axon-targeting phenotypes, expression of Psidin^S678D selectively rescued the axon-targeting phenotype but showed no ability to prevent cell loss (Fig. 8A–C). Given that Psidin^S678D binds dNAA20 significantly less, these data provide additional strong evidence for a dNAA20-dependent role of Psidin during neuron survival.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Conserved phosphorylated serine 678 prevents binding of dNAA20 to Psidin. A, Psidin isoforms were expressed in eyFlp clones driven by act-GAL4. Expression of Psidin^IG978 (carrying the E320K point mutation), rescued the targeting phenotype of psidin¹ mutant neurons, mimicking the psidin^IG978-targeting phenotype. Expression of both serine mutant isoforms Psidin^S678A and Psidin^S678D rescued the targeting phenotype to wild-type levels. B, Quantification of the Or59c cell population: expression of Psidin^IG978 and Psidin^S678A rescues the cell loss in the psidin¹ background. Expression of the phosphomimetic Psidin^S678D fails to rescue the cell loss. C, Quantification of the targeting phenotype. Expression of Psidin^IG978 in the psidin¹ mutant background rescues targeting to psidin^IG978 levels. Expression of both serine mutants phosphomutants rescue the targeting to wild-type levels. D, Model of Psidin's two functions during axon targeting and ORN survival. During axon guidance of olfactory neurons, Psidin function as an actin-binding protein, preventing the formation of long and stable filaments in growth cones. It acts as an antagonist to the actin filament stabilizer Tropomyosin. LimK, possibly via inactivation of cofilin, allows actin filament stabilization, and inactivation of LimK rescues Psidin's requirement in targeting. For ORN survival, Psidin is required as a noncatalytic subunit of the NatB complex, in which it interacts with the catalytic subunit dNAA20 to regulate protein acetylation (e.g., during cell cycle). *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA and Bonferroni's post-test. Error bars indicate ± SEM.