Singar1, a Novel RUN Domain-containing Protein, Suppresses Formation of Surplus Axons for Neuronal Polarity*♦

Although neuronal functions depend on their robust polarity, the mechanisms that ensure generation and maintenance of only a single axon remain poorly understood. Using highly sensitive two-dimensional electrophoresis-based proteomics, we identified here a novel protein, single axon-related (singar)1/KIAA0871/RPIPx/RUFY3, which contains a RUN domain and is predominantly expressed in the brain. Singar1 expression became up-regulated during polarization of cultured hippocampal neurons and remained at high levels thereafter. Singar1 was diffusely localized in hippocampal neurons and moderately accumulated in growth cones of minor processes and axons. Overexpression of singar1 did not affect normal neuronal polarization but suppressed the formation of surplus axons induced by excess levels of shootin1, a recently identified protein located upstream of phosphoinositide-3-kinase and involved in neuronal polarization. Conversely, reduction of the expression of singar1 and its splicing variant singar2 by RNA interference led to an increase in the population of neurons bearing surplus axons, in a phosphoinositide-3-kinase-dependent manner. Overexpression of singar2 did not suppress the formation of surplus axons induced by shootin1. We propose that singar1 ensures the robustness of neuronal polarity by suppressing formation of surplus axons.

Although neuronal functions depend on their robust polarity, the mechanisms that ensure generation and maintenance of only a single axon remain poorly understood. Using highly sensitive two-dimensional electrophoresis-based proteomics, we identified here a novel protein, single axon-related (singar)1/KIAA0871/RPIPx/RUFY3, which contains a RUN domain and is predominantly expressed in the brain. Singar1 expression became up-regulated during polarization of cultured hippocampal neurons and remained at high levels thereafter. Singar1 was diffusely localized in hippocampal neurons and moderately accumulated in growth cones of minor processes and axons. Overexpression of singar1 did not affect normal neuronal polarization but suppressed the formation of surplus axons induced by excess levels of shootin1, a recently identified protein located upstream of phosphoinositide-3-kinase and involved in neuronal polarization. Conversely, reduction of the expression of singar1 and its splicing variant singar2 by RNA interference led to an increase in the population of neurons bearing surplus axons, in a phosphoinositide-3-kinase-dependent manner. Overexpression of singar2 did not suppress the formation of surplus axons induced by shootin1. We propose that singar1 ensures the robustness of neuronal polarity by suppressing formation of surplus axons.
Most neurons develop polarity by forming a single long axon and multiple short dendrites (1,2). The ability of neurons to develop and maintain polarity is essential for their basic functions. The polarization processes have been extensively studied in cultured hippocampal neurons (1,3). These cells first form several minor processes, any of which appears capable of becoming either an axon or a dendrite, during the first 12-24 h after plating (stage 2). One of the neurites then rapidly elongates to acquire axonal characteristics (stage 3), whereas the others later become dendrites (stage 4). Hippocampal neurons must use a robust internal mechanism that ensures polarization as they regenerate a single axon and multiple dendrites even when polarity is altered by axonal amputation (4,5).
In contrast to the progress in defining the mechanisms of axon specification, our knowledge of how neurons generate and maintain only a single axon is limited. Reduction in the expression or activities of glycogen synthase kinase-3␤, PTEN, or microtubule affinity-regulation kinase (MARK)2 in hippocampal neurons leads to formation of multiple axons (11,12,23), suggesting that these molecules may be involved in inhibiting the formation of surplus axons. Recently, we identified the protein shootin1, which is located upstream of PI 3-kinase and involved in neuronal polarization (24). During polarization, shootin1 accumulated in a single neurite while being depleted in its sibling neurites, through competitive transport to multiple neurites. This in turn induced formation of a single axon, leading us to propose that the depletion of shootin1 from the sibling neurites of the nascent axon by competition prevents the formation of surplus axons (24).
We describe here a novel brain-specific protein, named single axon-related (singar)1. Singar1 became up-regulated during polarization of cultured hippocampal neurons and remained at high levels thereafter. Reduction of singar1 expression by RNAi led to an increase in the population of neurons bearing surplus axons, as has been reported for glycogen synthase kinase-3␤, PTEN, and MARK2. Furthermore, overexpression of singar1 suppressed the formation of surplus axons induced by excess amounts of shootin1, without affecting normal polarization processes. These data suggest that singar1 suppresses the formation of surplus axons, thereby contributing to the robustness of neuronal polarity.

EXPERIMENTAL PROCEDURES
Cultures and Metabolic Labeling-Hippocampal neurons prepared from E18 rat embryos were seeded on coverslips coated with poly-D-lysine (Sigma) and laminin (Iwaki) and cultured in Neurobasal medium (Invitrogen) supplemented with B-27 supplement (Invitrogen) and 1 mM glutamine without glia feeder layer as described (14). For quantitative two-dimensional electrophoresis (2-DE), stage 2 and stage 3 neurons were metabolically labeled with the culture medium containing 13% of the normal levels of methionine and cysteine plus Pro-mix  (25), using a composite gel series comprising a total 93 ϫ 103-cm combined gel system (26). For differential 2-DE, neurons were metabolically labeled as above, and protein spots separated by 2-DE gels were visualized by autoradiography. For protein identification, unlabeled protein samples from 2-weekold rat brains were separated by the 2-DE gel and visualized by silver staining. The protein spots corresponding to the radiolabeled ones were then excised from gels and in-gel-digested as described (27). Matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) was performed using a Voyager TM Elite equipped with delayed extraction (Applied Biosystems). Data base searches were conducted using the Mascot program (Matrix Science) and National Center for Biotechnology Information databases.
Cloning of Singar-Full-length cDNAs of rat singar1 and sin-gar2 were amplified by PCR from a rat brain cDNA library obtained from P8 Wistar rat using an RNeasy mini kit (Qiagen) with the primers 5Ј-GCGGATCCATGTCTGCCCTGACGC-CTCCGAC-3Ј and 5Ј-GCGGATCCCTAATGATGTTTTGG-GATCAGTTTATTTG-3Ј. The cDNAs were then subcloned into the BamHI site of pCMV (Stratagene) -Myc, pCMV-FLAG, pCAGGS-Myc with a ␤-actin promoter (provided by J. Miyazaki, Osaka University, Osaka, Japan) (28), and pGEX-6P-1 (GE Healthcare) vectors. These plasmids were designed to generate the proteins fused to Myc, FLAG, and glutathione S-transferase (GST) tags at the N terminus. We did not use epitope tag repeats.
Protein and Antibody Preparation-Recombinant full-length singar1 was expressed in Escherichia coli BL21(DE3)codonPlus (Stratagene) as a GST fusion protein by induction with 0.1 mM isopropyl-␤-D-thiogalactopyranoside at 20°C for 4 h and purified on a glutathione-Sepharose column (GE Healthcare), after which GST was removed from singar1 by PreScission protease (GE Healthcare). Rabbit polyclonal anti-singar antibody was raised against the recombinant singar1 and affinity-purified before use.
Transfection and RNAi-Neurons were transfected with cDNA or RNA by the calcium phosphate method (14), Nucleofector (Amaxa), or Lipofectamine 2000 (Invitrogen) before or after plating. Namely, for immunoblot analysis of the effects of small interfering RNA (siRNA) on singar expression, transfection of neurons was carried out before plating using Nucleofector. For analysis of the effects of siRNA on neuronal morphology after polarization, neurons were transfected by the calcium phosphate method on day in vitro (DIV) 3. For the other analysis, neurons were transfected with DNA or RNA using Lipofectamine 2000 before plating. Namely, neurons suspended in Neurobasal medium containing 10% fetal calf serum were incubated with lipofectamine2000/DNA (or RNA) complex for 1.5 h. Then, they were washed by centrifugation and resuspending in Neurobasal/10% fetal calf serum and plated on coverslips. Three h after the plating, the medium was replaced by the culture medium. HEK293T cells were transfected with cDNA by the calcium phosphate method. The siRNA sequence 5Ј-GGCAAAAGUAGAUGCGUUA-3Ј targeting rat singar1 and singar2 corresponds to nucleotides 846 -864 in the coding region of singar1 (nucleotides 900 -918 in the coding region of singar2). The scramble RNA (scRNA) 5Ј-AGCGAAGGUU-GACGCUCUU-3Ј encodes the same amino acid sequence as the siRNA but lacks significant sequence similarity to any gene. The RNAs were obtained from Dharmacon.
Immunoprecipitation-HEK293T cells were transfected with the expression plasmids and grown in 60-mm diameter culture dishes. Forty-eight h after transfection, cells were washed once with ice-cold PBS and harvested in the cell lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2.5 mM MgCl 2 , 2 mM EGTA, 1 mM dithiothreitol, 1 mM Na 3 VO 4 , 10 mM NaF, 0.5% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride). After incubation on ice for 30 min, the extracts were then centrifuged at 100,000 ϫ g for 30 min at 4°C. The supernatant were incubated with 1.5 g of anti-FLAG antibodies, and immunocomplexes were precipitated with protein G-Sepharose 4B. After three times of washing with the cell lysis buffer, the beads were boiled in the SDS sample buffer and applied to immunoblot analysis.
Statistics-In the statistical analysis for Fig. 4A, the neurons with one long neurite longer than 100 m and over twice as long as the other neurites were counted as stage 3 neurons (these cells do not include neurons with multiple axons). Significance was determined by the unpaired Student's t test.

Identification of Singar1 by a Proteome Screening of Stage 2 and Stage 3 Hippocampal
Neurons-To identify proteins involved in neuronal polarity formation, we previously performed proteome analysis of cultured rat hippocampal neurons using a 93 ϫ 103-cm large gel 2-DE (26). We screened about 6,200 protein spots on 2-DE gels and detected 277 that were consistently up-regulated during the transition between stages 2 and 3 (24). Tryptic digestion and mass spectrometry of one of them, located at a molecular mass of 55 kDa and pI ϭ 5.2 in gels ( Fig. 1A), identified six peptides whose sequences corresponded to human (BC051716), rat (EF577045), and mouse (BC049127) cDNA sequences of KIAA0871/RUFY3/ Rap2-interacting protein X (RPIPx) in genome databases ( Fig. 1B and supplemental Fig. 1). Rat, human, and mouse KIAA0871/RUFY3/ RPIPx encode proteins of 469 amino acids and predicted molecular masses of 52.9, 52.9, and 53.0 kDa, respectively. Domain searching revealed that KIAA0871/RUFY3/RPIPx contains one RUN (RPIP8, UNC-14, and NESCA) domain (29) and two coiled-coil domains (Fig. 1C). Recently, the crystal structure of its RUN domain was reported (30). We designated it singar1, together with Dr. Osamu Ohara (Kazusa DNA Research Institute), who initially reported its full cDNA sequence as KIAA0871 (AB020678) and Dr. Takahiro Nagase (Kazusa DNA Research Institute). It shows homology to RUN domain-containing proteins, such as RPIP8, RPIP9, RUFY1/Rabip4, RUFY2, UNC-14, and NESCA (Fig. 1D). Data base searches also identified Pan troglodytes, Macaca mulatta, Bos taurus, Canis familiari, and Gallus gallus orthologs of singar1, and partial open reading frames in Xenopus laevis, Danio rerio, Fugu rubripes, and Drosophila. In addition, a splicing variant of 487 amino acids was identified in rat (EF538802) and mouse (BC058259) cDNAs and termed singar2 ( Fig. 1C and supplemental Fig. 1).
Singar1 and Singar2 Are Predominantly Expressed in the Brain and Up-regulated during Polarization-We raised an antibody against recombinant singar1 to investigate its tissue and intracellular distribution. The antibody recognized three

Singar1 for a Single Axon
bands at 55, 60, and ϳ56 kDa in immunoblots of rat brain ( Fig.  2A). The lower band corresponds to the apparent M r of recombinant singar1, whereas the upper band corresponds to that of recombinant singar2 (Fig. 2B), suggesting that the two splicing variants are expressed in the brain. Treatment of brain lysate with -phosphatase led to disappearance of the middle 56-kDa band and increase in the intensity of the lower band (Fig. 2C), whereas treatment of HEK293T cells expressing exogenous singar1 with the phosphatase inhibitor calyculin A increased the intensity of the middle band (Fig. 2D). Therefore, we concluded that the middle band corresponds to the phosphorylated form of singar1. The three bands were not detected in various peripheral tissues ( Fig. 2A), suggesting that singar1 and singar2 are predominantly expressed in the brain. Consistent with the 2-DE data for the metabolically labeled protein (Fig. 1A), the level of singar1 expression increased remarkably during the transition from stage 2 to stage 3 in cultured hippocampal neurons and remained high until DIV28 (Fig. 2E). Consistently, expression of singar1 in rat brain was relatively low on embryonic day 15 (E15), peaked around postnatal day 4 (P4), and decreased in the adult brain (Fig. 2F). Similar expression patterns were observed for singar2 and phosphorylated singar1 in general, except that the expression of singar2 in hippocampal neurons returned to a low level at DIV14 and up-regulation of phosphorylated singar1 was relatively slow in the cultured neurons and brain (Fig. 2, E and F).
Singar1 and Singar2 Accumulates in Growth Cones of Minor Processes and Axons-Next, we examined the localization of singar in cultured hippocampal neurons. Immunocytochemical analysis showed a diffuse staining of endogenous singar in stage 2, stage 3, and stage 4 neurons (Fig. 3, A-C). In stage 2, singar immunoreactivity showed a moderate accumulation in growth cones of minor processes (arrowheads); 52% of growth cones showed accumulation (n ϭ 120 growth cones). In stage 3, singar accumulated in growth cones of axons in addition to those of minor processes (Fig. 3B, arrows and arrowheads); 77% of axonal growth cones showed accumulation (n ϭ 56 growth cones), whereas 52% of minor process growth cones showed accumulation (n ϭ 83 growth cones). In stage 4, singar accumulated in growth cones of axons (Fig. 3C); 31% of axonal growth cones showed accumulation (n ϭ 55 growth cones). Singar immunoreactivity was also localized in dendrites but did not show accumulation there (Fig. 3C). We used a green volume marker, CMFDA, as an internal standard to measure relative concentration of singar (singar immunoreactivity/CMFDA staining) (24). The relative concentrations of singar in growth cones of minor processes and axons were 2.3 Ϯ 0.2 times (n ϭ 15 growth cones) and 2.9 Ϯ 0.2 times (n ϭ 15 growth cones) higher than that in cell bodies, respectively (Fig. 3B). In growth cones, singar immunoreactivity was preferentially localized in F-actin-enriched filopodia and lamellipodia (Fig. 3D). As the anti-singar antibody recognized both singar1 and singar2, we also examined the localization of Myctagged singar1 and singar2 (Myc-singar1 and Myc-singar2) exogenously expressed in hippocampal neurons. Localizations of Myc-singar1 and Myc-singar2 were essentially similar to that FIGURE 2. Expressions of singar1 and singar2. A, immunoblot analysis of singar in adult rat tissues. B, immunoblot analysis of singar in adult brain and singar1 and singar2 exogenously expressed in HEK293T cells. C, protein lysates from P5 rat brain were treated with 4 units/l of -phosphatase for 30 min at room temperature and applied to immunoblot analysis with anti-singar antibody. D, HEK293T cells expressing Myc-singar1 were treated with 100 nM calyculin A or Me 2 SO for 20 min at 37°C. The cells were collected, and the extracted proteins were applied to immunoblot analysis with anti-Myc antibody. E and F, immunoblot analysis of singar in rat cultured hippocampal neurons (E, 3 g of proteins were loaded) and in rat brains (F, 5 g of proteins were loaded) at various developmental stages. The densities of singar immunoreactivity were measured by densitometer and expressed as the percentage of the densities of singar1 in stage 3 (E) and on P1 (F), respectively. Values are S.E. for four independent experiments except DIV28 (n ϭ 2) (*, p Ͻ 0.05; **, p Ͻ 0.01).

Singar1 for a Single Axon
of the endogenous singar immunoreactivity (supplemental Fig.  2, A and B).

Overexpression of Singar1 Suppresses the Formation of Surplus Axons Induced by Excess Levels of Shootin1, without
Affecting Normal Polarization Processes-To analyze the functions of singar1 and singar2 in neuronal polarization, we overexpressed Myc-tagged singar1 and singar2 in hippocampal neurons under the ␤-actin promoter. Neurons were transfected with cDNAs before plating, and the percentage of polarized neurons was counted at 62 h in culture. Fifty-three Ϯ 7.8% of neurons overexpressing Myc-singar1 (n ϭ three independent cultures, 239 neurons examined) and 58 Ϯ 5.6% of neurons overexpressing Myc-singar2 (n ϭ three independent cultures, 289 neurons examined) polarized, as in the case of control neurons with overexpressed Myc-GST (56 Ϯ 5.8%, n ϭ three independent cultures, 216 neurons examined) (Fig. 4A). Axons of similar lengths were formed from neurons with overexpressed Myc-singar1, Myc-singar2, and Myc-GST (Fig. 4A). We further cultured the neurons with overexpressed Myc-singar1, Myc-singar2, and Myc-GST until DIV6. We did not find differences in either the length or the number of axons and dendrites among these neurons (Fig. 4, B and C). Thus, we conclude that overexpression of singar1 or singar2 does not affect normal polarization processes.
Recently, we identified the protein shootin1, which is located at an upstream position in polarity signals involving PI 3-kinase (24). Overexpression of shootin1 in hippocampal neurons leads to formation of surplus axons. Surprisingly, co-transfection of singar1 led to a reduction in the percentage of neurons with surplus axons induced by shootin1 overexpression (n ϭ four independent cultures, 493 neurons examined; p Ͻ 0.01, when compared with control neurons overexpressing Myc-GST) (Fig. 4, D and E). On the other hand, overexpression of singar2 did not suppress the formation of surplus axons induced by shootin1 (n ϭ four independent cultures, 455 neurons examined). Immunoblot analysis showed that overexpression of singar1 did not suppress the expression of shootin1 (Fig. 4F), thereby ruling out the possibility that the reduction in the percentage of neurons with surplus axons is caused by a decrease in the level of overexpressed shootin1.

Reduction of Singar Expression by RNAi Leads to an Increase in the Population of Neurons Bearing Surplus
Axons-To further analyze the roles of singar in neuronal polarization, we reduced singar expression using an siRNA. Transfection of hippocampal neurons with an siRNA directed against singar before plating resulted in a significant decrease in the levels of endogenous singar1 and singar2 when compared with those seen after transfection with the corresponding scRNA (Fig. 5A). Suppression of singar in hippocampal neurons by the siRNA was also confirmed immunocytochemically (Fig.  5B, arrowheads). Transfection of neurons with the siRNA led to a 67% reduction in singar immunoreactivity when compared with the level seen after transfection with the scRNA (28 neurons examined). Remarkably, 15 Ϯ 2.1% (n ϭ three independent cultures, 150 neurons examined; p Ͻ 0.01, when compared with scRNA) of the neurons bore more than one (two or three) axons that were immunostained by axon-specific markers tau-1 (Fig. 5C) and anti-synaptophysin (Fig. 5D) antibodies but were immunonegative for a dendrite-specific marker anti-MAP2 antibody (Fig. 5E), on DIV6 (Fig. 6A). In contrast, only 5.9 Ϯ 0.2% (n ϭ three independent cultures, 159 neurons examined) of control neurons transfected with scRNA formed supernumerary axons. We also quantified the length and number of the neurites. Neurites labeled by axonal markers were markedly longer than dendrites (Fig. 6B). The sum of the lengths of neurites in the neurons bearing surplus axons was longer than that in control neurons (Fig. 6C). However, total neurite number in the neurons with surplus axons was similar to that of control neurons (Fig. 6D). Therefore, reduction of singar expression appears to induce surplus axons without changing the total number of neurites. In addition, reduction of singar expression enhanced the formation of surplus axons induced by shootin1 overexpression (Fig. 6E). Thirty-three Ϯ 4% (n ϭ three independent cultures, 426 neurons examined; p Ͻ 0.004, when compared with scRNA) of the neurons co-transfected with shootin1 and siRNA bore multiple axons. In contrast, 24 Ϯ 6% (n ϭ three independent cultures, 362 neurons examined) of control neurons co-transfected with shootin1 and scRNA formed supernumerary axons.
We next reduced singar expression after hippocampal neurons established polarity. Neurons were cultured until DIV3 to polarize (stage 3), after which they were transfected with the siRNA or scRNA, and their morphology was examined on DIV6. Twenty Ϯ 2.1% (n ϭ three independent cultures, 131 neurons examined; p Ͻ 0.01, when compared with scRNA) of the neurons bore multiple axons that were immunostained by tau-1 (Fig. 7A) and anti-synaptophysin (data not shown) antibodies but were immunonegative for anti-MAP2 antibody (data not shown) (Fig. 7B). In contrast, 6.3 Ϯ 0.3% (n ϭ three independent cultures, 160 neurons examined) of control neurons transfected with siRNA formed supernumerary axons. Together with the data of singar1 up-regulation and overexpression, these results suggest that singar plays an important role in suppressing the formation of unnecessary axons for neuronal polarity.

Singar1 Binds to PI 3-kinase and Singar RNAi-induced Surplus Axon Formation Is Suppressed by the PI 3-kinase
Inhibitor LY294002-Shootin1 is reported to induce axons through the PI 3-kinase pathway (24), and shootin1-induced multiple axon formation was suppressed by singar1 overexpression (Fig. 4, D and E) and enhanced by singar RNAi (Fig.  6E). We finally examined whether singar1 interacts with the PI 3-kinase pathway. Singar1 was coexpressed with the p85 or p110␣ subunit of PI 3-kinase in HEK293 cells. When sin-gar1 was immunoprecipitated, co-precipitation of the p85 and p110␣ subunits was detected (Fig. 8, A and B). Singar1 was also reciprocally co-immunoprecipitated with the p85 and p110␣ subunits, indicating that it associates with PI 3-kinase. We further examined whether the formation of surplus axons induced by reduction of singar levels depends on PI 3 kinase activity. Inhibition of PI 3-kinase activity by 20 M LY294002 led to a reduction in the percentage of neurons with multiple axons induced by singar RNAi (Fig. 8C). Overexpression of constitutively active PI 3-kinase (Myr-PI 3-K p110) is reported to induce formation of multiple axons as in the case of shootin1 overexpression (24,31). Interestingly, in contrast to the case of shootin1 overexpression (Fig. 6E), singar RNAi did not affect the formation of surplus axons induced by the active PI 3-kinase (Fig. 8D).
These results indicate that formation of surplus axons induced by the reduction of singar is dependent on PI 3-kinase activity, but surplus axon formation induced by the active PI 3-kinase is not influenced by singar reduction.

DISCUSSION
Using highly sensitive 2-DEbased proteomics, we have identified here a novel protein, singar1, which is predominantly expressed in the brain. Singar1 expression became up-regulated during polarization of cultured hippocampal neurons and remained at high levels thereafter. Overexpression of sin-gar1 suppressed the formation of surplus axons induced by excess levels of shootin1. Conversely, repression of singar by RNAi led to an increase in the population of neurons bearing surplus axons. Reduction of singar expression also enhanced the formation of surplus axons induced by shootin1 overexpression. Our siRNA decreased the levels of both singar1 and singar2, and we could not design specific siRNAs for singar1 or singar2. However, since only singar1 showed the activity to suppress the formation of surplus axons induced by shootin1, we consider that the effect of RNAi results from the reduction of singar1, not singar2. Interestingly, singar2 was found in rat and mouse databases but not in human databases. The functions of singar2 in rat and mouse hippocampal neurons remain unknown. Taken together, the present results suggest that singar1 plays an important role in suppressing formation of surplus axons.
Recent studies have reported that reduction in the expression or activities of glycogen synthase kinase-3␤, PTEN, or MARK2 in hippocampal neurons leads to formation of multiple axons (11,12,23). Although these phenotypes are sim-  ilar to that of singar RNAi, their overexpression phenotypes differ. Overexpression of constitutively active and wild-type glycogen synthase kinase-3␤ (11,12), wild-type PTEN (11), and wild-type MARK2 (23) inhibited formation and elongation of axons during polarization processes. On the other hand, singar1 overexpression did not affect the formation or elongation of axons during normal polarization processes but inhibited the formation of surplus axons induced by shoootin1 overexpression. Thus, our results suggest that singar1 may ensure the robustness of neuronal polarity by specifically suppressing formation of surplus axons. As observed in vitro, small populations of neurons with surplus axons are reported in the central nervous system (32)(33)(34). It remains to be determined whether singar1 suppresses the formation of surplus axons in situ.
At present, the mechanisms by which singar1 inhibits the formation of surplus axons are unclear. Singar1 overexpression inhibited the formation of surplus axons induced by shootin1 overexpression, whereas singar1 RNAi enhanced it. During polarization, shootin1 accumulates in a single neurite, whereas being depleted in its sibling neurites (24). This in turn generates polarized activity of PI 3-kinase (24), which is essential for neuronal polarization (6,9). On the other hand, shootin1 overexpression induces its own ectopic accumulation in the growth cones of minor processes, which in turn results in ectopic accumulation of PI 3-kinase activity there and formation of surplus axons (24). In the case of singar1, a substantial amount of singar1 was observed in minor processes and dendrites. In addition, reduction of singar expression by RNAi led to an increase in the population of neurons bearing surplus axons in a PI 3-kinase-dependent manner. Therefore, singar1 in minor processes and dendrites may inhibit the formation of surplus axons by acting upstream of PI 3-kinase. Probably, locally distributed sho-otin1 and globally distributed sin-gar1 may coordinately contribute to the maintenance of a single axon in neurons. In addition, sin-gar1 accumulated in the axon might not be involved in suppressing axon formation. Interestingly, the present study detected a phosphorylated form of singar1, raising the possibility that the functions of singar 1 may be altered by protein kinases.
In conclusion, the present study has identified a novel RUN domain-containing protein singar1. Our results suggest that singar1 may ensure the robustness of neuronal polarity by suppressing formation of surplus axons without affecting normal neuronal polarization. The in situ functions of singar1 and the molecular details of how singar1 inhibits the formation of surplus axons remain important issues for future investigation.