A Rac1/Phosphatidylinositol 3-Kinase/Akt3 Anti-apoptotic Pathway, Triggered by AlsinLF, the Product of the ALS2 Gene, Antagonizes Cu/Zn-superoxide Dismutase (SOD1) Mutant-induced Motoneuronal Cell Death*

AlsinLF, the product of the ALS2 gene, inhibits Cu/Zn-superoxide dismutase (SOD1) mutant-induced neurotoxicity via its Rho guanine nucleotide-exchanging factor domain. We here identified Rac1, a Rho family small GTPase, as a target for the Rho guanine nucleotide-ex-changing factor activity of alsinLF. Rac1 associates with alsinLF. The amount of the GTP form of Rac1 is up-regu-lated by enforced overexpression of alsinLF. We further found not only that constitutively active Rac1 suppresses motoneuronal cell death induced by SOD1 mutants but also that the neuroprotective activity of alsinLF was completely inhibited by knocking down the endogenous Rac1 expression with small interfering RNA for Rac1, indicating that Rac1 is the major effector for alsinLF-mediated neuroprotection. Such alsinLF/Rac1-mediated neuroprotection occurs specifically against the SOD1 mutant-in-duced cell death but not against the cell death induced by any other (GAG-CACTCCAGGTATTTGACAG) for mouse Rac1, a sense primer (CCA- CACACACCCATCCTTCTG) and an antisense primer (GCACTCCA-GGTACTTGACGG) for mouse Rac3, a sense primer (CACGCTACTT- CCTCCTCAAG) and an antisense primer (CTCTGTCTTCATCAGCT-GGC) for mouse Akt1, a sense primer (CCTTCCATGTAGACTCTC- CAG) and an antisense primer (CCTCCATCATCTCAGATGTGG) for mouse Akt2, and a sense primer (CATAGGCTATAAGGAGAAACC) and an antisense primer (TTGGATAGCTTCCGTCCAC) for mouse Akt3, were synthesized. Sense and antisense primers for mouse glyceraldehyde 3-phosphate dehydrogenase were 5 (cid:3) -TCCACCACCC-TGTTGCTGTA-3 (cid:3) and 5 (cid:3) -ACCACAGTCCATGCCATCAC-3 (cid:3) , respec- tively. The data analyses were performed using Sequence Detection System software Version 1.9.1 (Applied Biosystems). To adjust the expression level of each mRNA, that of glyceraldehyde 3-phosphate dehydrogenase mRNA was used as an internal control. Pull-down Assays— For bacterial expression of glutathione S -trans-ferase (GST), GST-fused Rho family proteins or GST-fused Rab5 family proteins, the pGEX vector or pGEX vectors encoding Rho family proteins including RhoA, RhoB, RhoC, RhoG, Rac1, Rac2, Rac3, and Cdc42, and a dominant-negative Rac1 (dnRac1) or Rab5 family proteins including Rab5A, Rab5B, and Rab5C were introduced into Escherichia coli . Expressions of GST, GST-Rho family proteins, or GST-Rab5 family proteins were induced by incubating with

both upper and lower motoneurons. Similar to other neurodegenerative diseases, most ALS cases show no genetic linkage, but ϳ10% of patients have a hereditary background. From dominantly inherited families, mutations in the Cu/Znsuperoxide dismutase (SOD1) gene were identified as a cause of an autosomal-dominant ALS (1), and the discovery gave rise to a breakthrough in the investigation of the intricate pathogenesis of ALS. Coupled with outstanding advances in gene-engineering techniques, various transgenic model mice carrying SOD1 mutants have been produced (2)(3)(4), and a large quantity of information has been obtained by intensive investigation with in vitro/in vivo models. However, the most essential mechanism of progressive motoneuronal degeneration still remains unclear.
In 2001, an autosomal recessive ALS-causative gene, ALS2, was identified from inbred Arabic families (5,6). ALS2 encodes not only a large protein called alsin long form (alsinLF) but also alsin short form, an alternatively spliced short form (Fig. 1A). AlsinLF contains three guanine nucleotide exchange factor (GEF) homologous domains consisting of a regulator of chromosome condensation 1 (RCC1), Rho guanine nucleotide exchanging factor (RhoGEF), and vacuolar protein sorting 9 (VPS9) (5). All known ALS2 mutations are nucleotide deletions causing a C-terminal truncation of alsinLF (5)(6)(7)(8)(9)(10). Patients with a C-terminally truncated alsinLF show a variety of infantile-onset motoneuronal disorders, including ALS, which involves both upper and lower motor neurons, and hereditary spastic paraplegia, which involves only upper motor neurons. Motifs in the primary sequence of alsinLF including three GEFs and membrane occupation and recognition nexus (MORN) are thought to be related to signal transduction, protein sorting, and membrane localization (Fig. 1A). The variety and the degree of loss of these functions due to frameshift probably give rise to a variety of pathophysiological disorders in alsin-related motor neuron diseases.
We recently demonstrated that alsinLF has neuroprotective activity specifically effective against SOD1 mutant-induced cytotoxicity (11). Detailed characterization has indicated that the RhoGEF domain is essential and sufficient for alsinLF-mediated neuroprotection. Considering that RhoGEF is a kind of GTPase regulator that activates Rho-family small G proteins by accelerating the reaction exchanging GDP for GTP, alsinLFmediated neuroprotection against toxicity by SOD1 mutants has been hypothesized to be mediated by the RhoGEF activity of alsinLF, although target Rho family proteins of alsinLF remain unidentified.
Rho family proteins are classified into several major subfam-ilies such as Rho, Rac, or Cdc42. Rho family proteins play a number of important roles in cell biology including cellular polarity, morphology, chemotaxis, invasion, cell division, cell transformation, metastasis, and cell survival (12). Especially, it has been generally known that Rho family proteins become pro-apoptotic or anti-apoptotic, depending on the cellular conditions and the cell types. Although there still remain many uncharacterized mechanisms underlying the pro-apoptotic and anti-apoptotic activities of Rho family proteins, several studies have identified pathways leading to cell death or cell survival. Upon stimulation of specific signals, different Rho family proteins might trigger different apoptotic pathways, which have been shown to be partially mediated by c-Jun N-terminal kinase and nuclear factor B (13). Fas stimuli induce apoptosis via Rho family proteins (14). Rac1 participates in serum deprivation-induced apoptosis (15). p75 neurotrophin receptor activates Rac1, which in turn activates c-Jun N-terminal kinase in a nerve growth factor-dependent manner and causes apoptosis in neuronal cells (16). On the other hand, it has been reported that Rac1 antagonizes apoptosis in certain situations (17)(18)(19)(20)(21). Rac1 protects endothelial cells from tumor necrosis factor-␣-induced apoptosis (19) and tumor cells from apoptosis induced by anti-cancer drugs or the cell surface death receptor, CD95 (18). Rac1 and RhoG promote cell survival by the activation of phosphatidylinositol 3-kinase (PI3K) and Akt in COS7 cells (20). Most recently, Topp et al. (22) demonstrated that the Rho-GEF domain of alsinLF has the guanine nucleotide exchanging activity for Rac1 in Sf9 insect cells, although the functional consequence of activation of Rac1 by the RhoGEF domain of alsinLF has not been assessed (22). It remains completely unknown whether Rac1, activated by the RhoGEF domain of alsinLF, acts as an effector for alsinLF-mediated protection against SOD1 mutant-induced neuronal cell death. Moreover, even if this is the case, it should be assessed whether Rac1 is the sole mediator for alsinLF-mediated protection of SOD1 mutant-induced neuronal cell death.
In this study, we dissected the precise mechanism underlying neuroprotection by alsinLF against SOD1 mutant-induced neurotoxicity. We demonstrated that alsinLF is a Rac1-specific RhoGEF. Most importantly, Rac1, activated by the RhoGEF activity of alsinLF, plays a critical role in alsinLF-mediated neuroprotection against toxicity by SOD1 mutants. We further showed that alsinLF protects neuronal cells via the PI3K/Akt3 pathway.

EXPERIMENTAL PROCEDURES
DNA Constructs-The full length of alsinLF cDNA was constructed as described previously (11). T701A-alsinLF, alsinLF with an amino acid substitution of alanine for threonine at the amino acid 701 position, was constructed by site-directed mutagenesis using a sense primer (CTCCACGAGTTAGCTACTGCAGAAAGACGATTC) and an antisense primer (GAATCGTCTTTCTGCAGTAGCTAACTCGTGGAG). Several expression vectors encoding human Rho family proteins such as pEX-V-myc-G12V-Rac1 (constitutively active Rac1), pGEX-2T-RhoA, pGEX-2TK-Rac1, and pGEX-2TK-Cdc42, were kind gifts of Dr. Shu Narumiya at The University of Kyoto (Kyoto, Japan). A human RhoB cDNA, a human Rac2 cDNA, and a human Rac3 cDNA were kindly provided by Dr. Harry Mellor at The University of Bristol (Bristol, UK), Dr. Gary Bokoch at The Scripps Research Institute (La Jolla, CA), and Dr. Nora Heisterkamp at Children's Hospital Los Angeles (Los Angeles, CA), respectively. A human RhoC cDNA was PCR-amplified from cDNAs isolated from the frontal lobe of a human brain cerebrum (BioChain, Hayward, CA) with a sense primer (CGGGATCCATGGCTGCAATCC-GAAAGAAGC) and an antisense primer (GGAATTCTCAGAGAATGG-GACAGCCCC). A human RhoG cDNA was obtained from a human frontal lobe cDNA library (BioChain) by PCR with a sense primer (CGGGATCCATGCAGAGCATCAAGTGCGTG) and an antisense primer (GGAATTCTCACAAGAGGATGCAGGACC). cDNAs for RhoB, RhoC, RhoG, Rac2, and Rac3 were subcloned into the pGEX vector to produce recombinant proteins. cDNAs encoding Rab5A, Rab5B, and Rab5C were provided by the UMR cDNA Resource Center (www.cdna. org) and subcloned into the pGEX vectors. Dominant-negative and constitutively active mutants of a human Akt1 cDNA were kindly provided by Dr. Yukiko Gotoh at University of Tokyo (Tokyo, Japan). Mouse Akt1 and Akt2 cDNA were kindly provided by Dr. Satoru Sumitani at The University of Osaka (Osaka, Japan). Human and mouse Akt3 cDNAs were kindly provided by Dr. Brian A. Hemmings at The Friedrich Miescher Institut (Basel, Switzerland). Referring to the structure of the constitutively active human Akt1, we constructed a constitutively active mutant of human Akt3 by PCR with a sense primer (GGAATTCACCATGGGGAGTAGCAAGAGCAAGCCTAAGGA-CCCCAGCCAGCGCAGCGAGGAGGAAGAGATGGATGCCTC) and an antisense primer (GGGGTACCTTCTCGTCCACTTGCAGAGTAG) to delete the N-terminal PH domain and add a myristoylation site to its N terminus.
Cell Culture, Transfection, and Cell Death Assays-NSC34 cells, one of the best models for primary cultured motor neurons (11, 24 -27), were kindly provided by Dr. Neil Cashman at The University of Toronto (Toronto, Canada) and cultured in Dulbecco's modified Eagle's medium (DMEM) (Sigma) supplemented with 10% fetal bovine serum (FBS) (Hyclone, Logan, UT) and antibiotics. NSC34 cell is a hybrid cell line of motor neurons derived from embryonic mouse spinal cord cells and mouse neuroblastoma with a number of characteristics for primary cultured motor neurons, including generation of action potential, acetylcholine synthesis, storage, and release. For transient transfection, NSC34 cells were seeded in a 6-well plate at 7 ϫ 10 4 cells/well and cultured in 10% FBS, DMEM for 12-16 h and then transfected with a vector or vectors encoding alsinLF or other genes by lipofection (1 g DNA, 2 l of Lipofectamine, 4 l of PLUS Reagent) in the absence of serum for 3 h. Lipofectamine and PLUS Reagent were purchased from Invitrogen. Transfected NSC34 cells were incubated with DMEM plus 10% FBS. Twenty-four hours after the onset of transfection, their culture media were changed to DMEM plus N2 supplement (Invitrogen), and cells were incubated for additional 48 h. Seventy-two hours after transfection, cell mortality was determined by trypan blue exclusion assay performed as follows. Cells were suspended by gentle pipetting, and 50 l of 0.4% trypan blue solution (Sigma) was mixed with 200 l of the cell suspension (final concentration 0.08%) at room temperature. Stained cells were counted within 3 min after the mixture with the trypan blue solution. 100 cells/well were counted for each trypan blue exclusion assay. Mortality of cells was then determined as a percentage of trypan blue-stained cells in total cells. Therefore, cell mortality determined by this method represents the population of dead cells in total cells, including both adhesive and floating cells, at the termination of experiments. It has been established by our previously reported experiments that our cell mortality determined with the trypan blue exclusion assay displays precisely reciprocal results to the WST-8 assay, the most established cell viability assay (11). The WST-8 assay was performed with 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4disulfophenyl)-2H-tetrazolium, monosodium salt (WST-8) using Cell Counting kit-8 (Wako Pure Chemicals Industries, Osaka, Japan). Seventy-two hours after transfection, cells were suspended by gentle pipetting. One-tenth volume (100 l) of the cell suspension was incubated with 10 l of WST-8 solution in a 96-well plate for 2 h at 37°C. Absorbance of the samples at the 450-nm wavelength was measured by a Wallac 1420 ARVOsx Multi Label Counter.
Real-time PCR-To confirm the gene-silencing effect of siRNA, we performed real-time PCR to assess the amount of endogenous mRNA of Rac1, Rac3, Akt1, Akt2, and Akt3. NSC34 cells were transfected with pRNA-U6.1/Shuttle-siRNA for Rac1 (siRac1), siRac3, siAkt1, siAkt2, and siAkt3 as described above. Transfection efficiency was estimated to be about 70% (11). Seventy-two hours after transfection, cells were harvested for RNA extraction with ISOGEN reagent (Nippon Gene, Toyama, Japan). The first-strand cDNAs were synthesized using Sensiscript reverse transcriptase (Qiagen, Germany) with 0.5 g of total RNA. Real-time PCR analysis was performed using a QuantiTect SYBR Green PCR kit (Qiagen) followed by analysis with ABI PRISM7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Five sets of primers, consisting of a sense primer (GTCCCAATACTCCTATCATCCTC) and an antisense primer (GAG-CACTCCAGGTATTTGACAG) for mouse Rac1, a sense primer (CCA-CACACACCCATCCTTCTG) and an antisense primer (GCACTCCA-GGTACTTGACGG) for mouse Rac3, a sense primer (CACGCTACTT-CCTCCTCAAG) and an antisense primer (CTCTGTCTTCATCAGCT-GGC) for mouse Akt1, a sense primer (CCTTCCATGTAGACTCTC-CAG) and an antisense primer (CCTCCATCATCTCAGATGTGG) for mouse Akt2, and a sense primer (CATAGGCTATAAGGAGAAACC) and an antisense primer (TTGGATAGCTTCCGTCCAC) for mouse Akt3, were synthesized. Sense and antisense primers for mouse glyceraldehyde 3-phosphate dehydrogenase were 5Ј-TCCACCACCC-TGTTGCTGTA-3Ј and 5Ј-ACCACAGTCCATGCCATCAC-3Ј, respectively. The data analyses were performed using Sequence Detection System software Version 1.9.1 (Applied Biosystems). To adjust the expression level of each mRNA, that of glyceraldehyde 3-phosphate dehydrogenase mRNA was used as an internal control.
Pull-down Assays-For bacterial expression of glutathione S-transferase (GST), GST-fused Rho family proteins or GST-fused Rab5 family proteins, the pGEX vector or pGEX vectors encoding Rho family proteins including RhoA, RhoB, RhoC, RhoG, Rac1, Rac2, Rac3, and Cdc42, and a dominant-negative Rac1 (dnRac1) or Rab5 family proteins including Rab5A, Rab5B, and Rab5C were introduced into Escherichia coli. Expressions of GST, GST-Rho family proteins, or GST-Rab5 family proteins were induced by incubating with 0.2 mM isopropyl-1-thio-␤-Dgalactopyranoside at 30°C for 4 h. Bacterial cells, centrifuged and resuspended in phosphate-buffered saline, were lysed with lysozyme (1 mg/ml) and sonicated in the presence of Triton X-100 (final 0.1%). GST-Rho family proteins or GST-Rab5 family proteins were then precipitated with glutathione-Sepharose beads (Amersham Biosciences) at 4°C for 6 h. The beads were washed three times with phosphatebuffered saline. To remove preloaded nucleotides, GST-Rho family proteins or GST-Rab5 family proteins bound to glutathione-Sepharose beads were incubated with cell lysis buffer A (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol, 0.5% Triton X-100, 0.2% sodium deoxycholate, protease inhibitors) plus 10 mM EDTA at room temperature for 1 h. Beads were then incubated with 1% bovine serum albumin-containing phosphate-buffered saline for 3 h at 4°C. After incubation, the beads were washed with cell lysis buffer A.
To prepare cell lysates containing alsinLF or T701A-alsinLF for pull-down assays, NSC34 cells, seeded at 7 ϫ 10 5 cells/dish in a 10-cm culture dish, were transfected with 10 g of plasmids encoding alsinLF or T701A-alsinLF. Cells were cultured in 10% FBS-DMEM for 48 h after transfection and then harvested for lysis with the cell lysis buffer A. After a cycle of freeze-thaw was performed, lysates were centrifuged to remove cellular debris. Supernatants were then precleaned with recombinant GST beads at 4°C for 3 h. Precleared cell lysates (100 g/sample) were incubated with 10 l of GST-beads or GST-Rho family beads or GST-Rab5 family beads for 30 min at room temperature. Beads were then washed four times with the cell lysis buffer before immunoblot analysis with anti-Myc antibody.
To show the expression levels of each protein in cell-death assays or cell-viability assays, immunoblot analysis was performed using cells that had been simultaneously prepared with cells used for these assays. Immunoblot analysis was repeated twice or three times per each experiment.
Statistical Analysis-All experiments on neuronal cell lines described here were repeated at least three times with independent transfections and treatments, each of which showed essentially the same results. All cell-death and cell-viability experiments were done with n ϭ 3. All values in the figures of the in vitro study are the mean Ϯ S.D. Statistical analysis was performed with analysis of variance followed by post hoc test, in which p Ͻ 0.05 was assessed as significant.

RESULTS
AlsinLF Associates with Rac1-AlsinLF has been predicted to regulate activities of small GTP-binding proteins, based on the finding that it contains three GEF domains homologous to RCC1, RhoGEF, and VPS9 (Fig. 1A). Our previous study indicated that the RhoGEF domain of alsinLF is essential and sufficient for alsinLF-mediated protection against SOD1 mutant-induced neuronal cell death (11). We have further asked whether the putative RhoGEF activity is necessary for alsinLFmediated neuroprotection, although there was no direct evidence indicating that alsinLF has RhoGEF activity at this moment. It has been reported that an amino acid substitution of alanine for threonine at amino acid position 506 of Dbl, a well characterized RhoGEF, nullifies the RhoGEF activity of Dbl (28,29). Accordingly, we substituted alanine for threonine at the position 701 of alsinLF (T701A-alsinLF), which is homologous to Thr-506 of Dbl (Fig. 1B). To examine the effect of the amino acid substitution on the neuroprotective function of alsinLF, NSC34 cells, the most established motoneuronal cells, were cotransfected with the plasmid vector encoding alsinLF or T701A-alsinLF in association with the A4T-SOD1 mutant-encoding vector. Seventy-two hours after transfection, cell mortality was determined by trypan blue exclusion assay. As shown in Fig. 1C, alsinLF protected NSC34 cells from SOD1 mutant-induced neurotoxicity, whereas T701A-alsinLF did not. This finding indirectly supported the hypothesis that alsinLF has RhoGEF activity, which is necessary for alsinLFmediated neuroprotection, further prompting us to search for the target(s) of the RhoGEF activity of alsinLF.
We first examined the physical interaction between alsinLF and various Rho family proteins. Rho family proteins are classified into three subfamilies. The RhoA subfamily consists of RhoA, RhoB, and RhoC. The Rac1 subfamily includes Rac1, Rac2, and Rac3. The Cdc42 is another protein belonging to the Rho family. RhoG has a primary sequence and function similar to the Rac1 family (30).
To perform the pull-down assay to determine which small GTPases interact with alsinLF, we constructed plasmid vectors encoding GST-tagged Rho family proteins consisting of RhoA, RhoB, RhoC, RhoG, Rac1, Rac2, Rac3, and Cdc42. Recombinant GST or GST-Rho family proteins were expressed in E. coli. Nucleotides preloaded in bacteria were removed from the purified recombinant proteins by adding EDTA (see "Experimental Procedures"). Lysates from NSC34 cells, in which alsinLF was overexpressed, were then mixed with the nucleotide-free GST-Rho family proteins immobilized to glutathione-Sepharose beads followed by extensive washing. As shown in Fig. 2A, alsinLF was co-precipitated with GST-Rac1 but not with GST nor with any Rho family protein fused to GST, suggesting that Rac1 is a putative major downstream effector for the RhoGEF activity of alsinLF.
It has been reported that T506A-Dbl loses the GEF activity because this amino acid substitution attenuates the affinity of RhoGEF to its effector GTPase (29). We, therefore, speculated that the analogous amino acid substitution of alanine for threonine at amino acid 701 of alsinLF would result in loss of the affinity of alsinLF to its putative effector Rac1. To test this hypothesis, we generated recombinant GST-T17N-Rac1 protein to perform the pull-down assay. It has been established that a substitution of asparagine for threonine at the amino acid 17 of Rac1 (T17N-Rac1) results in a dominantnegative mutant of Rac1 (dnRac1), which efficiently binds to its upstream GEFs (31). As expected, alsinLF was co-precipitated with GST-dnRac1, whereas T701A-alsinLF was not apparently co-precipitated with GST-dnRac1 (Fig. 2B), supporting our hypothesis.
Recent studies indicate that alsinLF interacts with Rab5family small G proteins including Rab5A, Rab5B, and Rab5C (22). Using our pull-down assays, we were able to see direct interaction between alsinLF and any Rab5 protein (Fig. 2C). In contrast to Rac1, however, Rab5 proteins bind to both alsinLF and T701A-alsinLF.
AlsinLF Is a Guanine Nucleotide-exchanging Factor for Rac1-To prove that Rac1 is a target of the RhoGEF activity of alsinLF, we examined whether enforced expression of alsinLF activates endogenous Rac1. The amount of GTP-form (activated) Rac1 (GTP-Rac1) was estimated by performing an affinity precipitation of GTP-Rac1 with GST-PBD (67-150 residues of the p21 binding domain of human PAK1)-conjugated agarose beads. In this assay, only GTP-Rac1 and GTP-Cdc42 binds to the PBD of PAK-1, one of their effector proteins (32). Five hundred g (25 times more the input) of lysates from Chinese hamster ovary cells, in which alsinLF or T701A-alsinLF was overexpressed, were mixed with GST-PBD-conjugated agarose. Resultant precipitates containing GTP-Rac1/GTP-Cdc42 were fractionated by SDS-PAGE and then immunoblotted with anti-FIG. 1. The T701A mutation of alsinLF results in loss of neuroprotection of alsinLF. A, scheme of structure of alsinLF and alsin short form (alsinSF). AlsinLF has three GTPase-regulator homologous domains, regulator of chromosome condensation 1 (RCC1) domain, the Rho guanine nucleotide exchanging factor (RhoGEF) domain, and the vacuolar protein sorting 9 (VPS9) domain. The RhoGEF domain consists of the Dbl homology (DH) domain, responsible for guanine nucleotide exchanging activity, and the pleckstrin homology (PH) domain, which contributes to membrane localization and binding to inositol lipids and inositol phosphates. RCC1 protein is known as a GEF for a nuclear GTP-binding protein, Ran. VPS9 protein is known as a GEF for small GTPase Rab5. B, scheme of amino acid substitution of alanine for threonine residue at the amino acid 701 of alsinLF. Amino acid substitution of alanine for the threonine residue at amino acid 506 of Dbl, a well characterized RhoGEF, results in loss of RhoGEF activity. The black letters show the homologous amino acid residues. C, T701A substitution abolishes the protective function of alsinLF against neurotoxicity by SOD1 mutants. NSC34 cells were cotransfected with 1.0 g of pEF1/MycHis vector, pEF1/MycHis-alsinLF, or pEF1/MycHis-T701A-alsinLF together with 0.5 g of pEF-BOS or pEF-BOS-A4T/G85R/G93R-SOD1. Seventy-two hours after transfection, cells were harvested for trypan blue exclusion assay. This experiment was performed with n ϭ 3. All values indicate the means Ϯ S.D. Immunoblotting was performed with anti-Myc antibody for detection of alsinLF or anti-SOD1 antibody (lower panels).
Rac1 and anti-Cdc42 antibodies (Fig. 2D). The equality of the amount of applied proteins was examined with detection of actin as an internal standard. Expression of alsinLF, but not T701A-alsinLF, increased the amount of GTP-Rac1, indicating that alsinLF activates Rac1. In contrast, Cdc42 was not activated by expression of alsinLF.
In addition, we performed an affinity precipitation of GTP-RhoA with GST-RBD-agarose beads (7-89 residues of the Rho binding domain of mouse Rhotekin) to examine whether alsinLF activates endogenous RhoA. This system uses a process by which GTP-RhoA binds to its effector protein, Rhotekin, via its Rho binding domain (33). The pull-down assay was performed with the lysates of the Chinese hamster ovary cells, in which alsinLF or T701A-alsinLF was overexpressed. The precipitates containing GTP-RhoA were fractionated with SDS-PAGE and then immunoblotted with anti-RhoA antibody. As shown in Fig. 2E, expression of alsinLF or T701A-alsinLF did not affect the amount of precipitated active RhoA. These results are consistent with the notion that alsinLF has guanine nucleotide exchanging activity specific to Rac1.
A Constitutively Active Rac1 Mutant, G12V-Rac1, Suppresses SOD1 Mutant-induced Death of NSC34 Cells-We then asked whether Rac1 is involved in alsinLF-mediated suppression of SOD1 mutant-induced cell death. A Rac1 mutant harboring an amino acid substitution of valine for glycine at amino acid 12 is a widely used constitutively active form of Rac1 (caRac1). We examined whether overexpression of G12V-Rac1 prevents neuronal cell death triggered by SOD1 mutants. To this end, NSC34 cells were cotransfected with 0.3 g of the pEF-BOS vector or pEF-BOS-A4T-SOD1 together with stepwise increasing amounts of pEXV-G12V-Rac1. The backbone vector was added to keep the total amount of plasmids at 0.8 g. As shown in our earlier study (11), overexpression of A4T-SOD1 induced marked death of NSC34 cells (Fig. 3A). Unfortunately, overexpression of G12V-Rac1 itself also caused the death of NSC34 cells in a dose-responsive fashion. Considering that Rac1 is also involved in certain cell-death pathways (13,34), this result was within our expectation. When we cotransfected pEXV-G12V-Rac1 within amounts of 0.1-0.2 g/well together with pEF-BOS-A4T-SOD1 into NSC34 cells, G12V-Rac1 seemed to suppress A4T-SOD1-induced cell death in a dose-dependent fashion. Cotransfection of 0.2 g of pEXV-G12V-Rac1 with pEF-BOS-A4T-SOD1 resulted in a statistically significant suppression of neurotoxicity by A4T-SOD1 (p Ͻ 0.0001). However, when larger amounts of G12V-Rac1 plasmids were cotransfected with A4T-SOD1, total cell mortality probably consisting of both A4T-SOD1-induced cell death, suppressed by co-expression of G12V-Rac1, and G12V-Rac1-induced cell death rose to higher levels than A4T-SOD1-induced cell death because the extent of G12V-Rac1-induced cell death became larger than that of G12V-Rac1-mediated suppression of A4T-SOD1-induced cell death. This interpretation supported the notion that Rac1 protects neuronal cells from SOD1 mutant-induced neurotoxicity.
Disruption of Endogenous Rac1 Expression with a siRNA for Rac1 Abolishes Neuroprotective Function of alsinLF against Toxicity by SOD1 Mutants-To confirm that Rac1 is FIG. 2. AlsinLF is a GEF for Rac1. A, alsinLF selectively binds to Rac1. B, T701A substitution attenuates binding between alsinLF (LF) and a dominantnegative Rac1 (dnRac1). C, Rab5 family proteins bind to both alsinLF and T701A-alsinLF. GST or GST-fused recombinant small GTPases were expressed in E. coli and purified by glutathione-Sepharose beads. GST proteins, mixed with 100-g lysates of NSC34 cells expressing Myctagged alsinLF, were pulled down with 10 l of slurry of glutathione-Sepharose beads, fractionated by SDS-PAGE, and then immunoblotted with anti-Myc antibody. Pulled-down GST proteins were visualized with Coomassie Blue staining of immunoblotted membranes. D and E, overexpression of wild-type alsinLF activates endogenous Rac1 but not Cdc42 or RhoA. In addition, T701A substitution nullifies GEF activity of alsinLF for Rac1. Chinese hamster ovary cells were transfected with 10 g of pEF1/MycHis-alsinLF or pEF1/MycHis-T701A-alsinLF. Forty-eight hours after transfection, cells were harvested for an affinity precipitation of endogenous active-form Rho family proteins. Activated Rac1 and Cdc42 were precipitated by GST-PAK-agarose (D), and activated RhoA was precipitated by GST-RBD agarose (E). Precipitates were then fractionated by SDS-PAGE and immunoblotted with anti-Rac1 antibody, anti-Cdc42 antibody, or anti-RhoA antibody. Myc-tagged alsinLF and T701A-alsinLF were detected with anti-Myc antibody. All the experiments described in this figure were repeated at least three times, with independent material preparation, each of which showed essentially the same results.
involved in alsinLF-mediated neuroprotection, we next examined the effect of siRNA-mediated disruption of endogenous Rac1 on neuroprotection by alsinLF. For that purpose we constructed a plasmid, pRNA-U6.1/Shuttle-siRac1, encoding siRNA for Rac1 (siRac1) to specifically down-regulate mRNA expression of endogenous Rac1. In this assay we also used a pRNA-U6.1/Shuttle-siRac3 expressing siRNA for Rac3 (si-Rac3) as a negative control that knocks down mRNA expression of endogenous Rac3.
First, we examined whether siRac1 specifically suppresses the amount of mRNA of Rac1. After mRNA was prepared from NSC34 cells transiently transfected with siRac1 or siRac3, the quantity of mRNA of endogenous Rac1 or Rac3 was determined by real-time PCR. As shown in Fig. 3B, siRac1 reduced the endogenous mRNA level of Rac1 about to 30% that of the control level, whereas it did not reduce the mRNA level of Rac3. Reciprocally, siRac3 reduced the cellular mRNA level of Rac3 about to 30% that of the control level, whereas it did not reduce the mRNA level of Rac1. To confirm the effect of treatment of siRac1 at the protein level, NSC34 cells were transfected with the siRac1 plasmid, and the cell lysates were immunoblotted with anti-Rac1 antibody (Fig. 3C). Expression of siRac1 diminished the protein level of whole endogenous Rac1 by almost half. Considering that transfection efficiency was estimated to be 60 -70% in this particular experiment (data not shown), it was speculated that siRac1 nearly completely disrupted endogenous Rac1. By co-expressing siRac1, we examined the effect of disruption of endogenous Rac1 on the neuroprotective activity of alsinLF. After NSC34 cells were cotransfected with pEF-BOS-A4T-SOD1 and pEF1/MycHis-alsinLF together with si-Rac1 or siRac3, cell mortality was determined by trypan blue exclusion assay (Fig. 3D). When NSC34 cells were cotransfected with siRac1, neuroprotection by alsinLF was markedly attenuated down to the basal level. In contrast, cotransfection with siRac3 did not result in attenuation of alsinLF-mediated neuroprotection at all. These findings indicate that Rac1 is the major mediator for alsinLF-mediated neuroprotection against the SOD1 mutant toxicity.
PI3 Kinase and Akt Family Proteins Are Involved in alsinLFmediated Neuroprotection-Via its effector proteins, Rac1 plays roles in multiple cellular processes including cellular chemotaxis, invasion, cell division, cell transformation, metastasis, and apoptosis. It has also been reported that Rac1 antagonizes apoptosis in certain situations (17)(18)(19)(20)(21). An effector for the Rac1-mediated anti-apoptotic activity is PI3K (17,20). The principal function of PI3K is pro-survival or anti-apoptotic activity, which is mediated mainly by Akt family proteins (35). We, thus, hypothesized that alsinLF protects neuronal cells by activating the PI3K/Akt pathway. To test this possibility, we first examined whether a PI3K-specific inhibitor, wortmannin, antagonizes neuroprotection by alsinLF. NSC34 cells were transiently cotransfected with A4T-SOD1 and alsinLF. At 24 h after transfection, wortmannin was added to the culture media at a concentration of 10 nM. Seventy-two hours after transfection, cells were harvested for examination of cell mortality. Treatment with wortmannin completely abolished neuroprotective function of alsinLF (Fig. 4A). Wortmannin did not affect expression of alsinLF or A4T-SOD1. This result indi- cated that PI3K activation is necessary for neuroprotection by alsinLF.
We further examined whether Akt family proteins (35), downstream targets of PI3K, are involved in alsinLF-mediated neuroprotection. It has been reported that a number of neurotrophic factors and growth factors including glia-derived neurotrophic factor, insulin-like growth factor-1, and vascular endothelial growth factor, which activate Akt kinases, protect motoneurons from SOD1 mutant-mediated toxicity, improving symptoms and extending the life span of SOD1 mutant trans-genic mice (27, 36 -38). The Akt family has three isoforms named Akt1, -2, and -3. All three Akt proteins are expressed in NSC34 (data not shown but see Fig. 6A). Three Akt family proteins are very similar in their primary structure and signal transduction manner (39). The identity of amino acids between an Akt protein and any one of the other two Akt proteins is around 80%. Different Akt isoforms appear to be redundant in their certain part of anti-apoptotic function (35). It is, therefore, anticipated that enforced overexpression of a dominantnegative mutant of Akt1 (dnAkt1), a Akt1 mutant with three Immunoblotting was performed with anti-Myc antibody for cahAkt1 or anti-SOD1 antibody for SOD1. Right panel, NSC34 cells were cotransfected with 0.5 g of pcDNA3.1/MycHis(Ϫ) vector or pcDNA3.1/MycHis(Ϫ)-cahAkt1 together with 0.5 g of the pcDNA3 vector, pcDNA3-V642I-APP, pcDNA3-K595N/M596L-APP (NL-APP), pcDNA3-M146L-PS1, or pcDNA3-N141I-PS2 (FAD insults). At 72 h after transfection, cell viability was determined by WST-8 assay. Immunoblot analysis was performed with anti-Myc antibody for detection of Myc-tagged cahAkt1, anti-APP antibody (22C11), anti-PS1 antibody, or anti-PS2 antibody (right panels). Arrows beside the immunoblot panel of PS1 indicate the PS1 holoprotein and the PS1 N-terminal fragment (NTF) as described previously (56). Arrows beside the immunoblot panel of PS2 show aggregated PS2 and the PS2 C-terminal fragment, as described previously (57). These experiments were done with n ϭ 3. All values indicate the means Ϯ S.D. amino acid substitutions (T308A, S473A, and K179A) in the two phosphorylation sites and the catalytic site, abolished not only Akt1 function but also a portion of Akt2 and Akt3 functions common to Akt1. Thus, it is possible that enforced overexpression of a dominant-negative human Akt1 (dnhAkt1) results in disappearance not only of Akt1 function but also of certain portions of Akt2 and Akt3 function common to Akt1.
To see the effect of dnhAkt1 on neuroprotection by alsinLF, NSC34 cells cotransfected with A4T-SOD1, G85R-SOD1, or G93R-SOD1 together with or without alsinLF in association with or without dnhAkt1 were harvested at 72 h after transfection for analysis of cell mortality or cell viability. As shown in Fig. 4, B and C, dnhAkt1 completely nullified the protective activity of alsinLF against neurotoxicity by any FALS-associated SOD1 mutants. To reciprocally examine whether constitutively active human Akt1 (cahAkt1) is able to take the part of alsinLF in neuroprotection, NSC34 cells cotransfected with A4T-SOD1, G85R-SOD1, or G93R-SOD1 together with or without cahAkt1 were subsequently harvested at 72 h after transfection for analysis of cell viability using the WST-8 assay (Fig.  4D, left panel). Coexpression of cahAkt1 completely antagonized neurotoxicity induced by SOD1 mutants.
Unexpectedly, simultaneously performed immunoblot analysis indicated that co-expression of cahAkt1 seemed to down-regulate expression of some SOD1 mutants, suggesting that reduced levels of overexpression of SOD1 mutants resulted in artificial neuroprotection against toxicity by some SOD1 mutants (Fig. 4D, immunoblot part of the right panel). Although we do not know how downregulation of expression of SOD1 mutants was induced by coexpression of cahAkt1, we estimated that the down-regulated expression level of any SOD1 mutant was still sufficient for induction of death of NSC34 cells (data not shown). Such down-regulation of expression of SOD1 mutants has been also observed in the case of co-expression of SOD1 mutants and the RhoGEF domain of alsinLF, as described in detail in our earlier study (11). Based on these results we were able to conclude that Akt family proteins are involved in alsinLF-mediated antagonization of neurotoxicity induced by SOD1 mutants.
In contrast, as shown in Fig. 5, lower level overexpression of caRac1, accomplished by transfection with 0.2 g/well of the caRac1-encoding vector, reduced neurotoxicity by SOD1 mutants, whereas caRac1 did not affect neurotoxicity by FAD-or FPD-related insults, indicating that Rac1 is the mediator transmitting the alsinLF-mediated neuroprotective signal specifically effective against SOD1 mutant-induced toxicity in NSC34 cells. Thus, we have concluded that there is a certain molecular mechanism determining specificity of the alsinLF-mediated neuroprotection even at the level of Akt family proteins.
Akt3 Mediates alsinLF-induced Neuroprotection against Toxicity by SOD1-In search of the mechanism determining the specificity of alsinLF-mediated neuroprotection at the level of Akt proteins, we first asked which isoform is involved there. To identify the Akt isoform involved in alsin-LF-mediated neuroprotection, we constructed three expression vectors encoding a siRNA for each mouse Akt isoform. Endogenous mRNA expression of any Akt in NSC34 cells was markedly reduced by introduction of each vector into the cells (Fig. 6A). In accordance, protein expression of any co-expressed HA-tagged mouse Akt was strikingly down-regulated with co-expression of each siRNA (Fig. 6B). We examined how siRNA-mediated disruption of Akt proteins altered alsinLF-mediated neuroprotection against SOD1 mutant-induced toxicity (Fig. 6C). Disruption of Akt3 almost completely inhibited alsinLF-mediated neuroprotection. In contrast, disruption of Akt1 or Akt2 did not inhibit alsinLFmediated neuroprotection. These findings indicated that Akt3 is the major mediator of alsinLF-mediated neuroprotection. It has been generally accepted that Akt proteins are actually activated by phosphoinositide-dependent kinase 1-mediated phosphorylation of Thr-308 of human Akt1 or equivalent amino acids of Akt2 and Akt3 (40,41). Phosphorylation of Ser-473 of human Akt1 or equivalent amino acids of Akt2 and Akt3 also is essential for full activation of Akt family proteins. To further confirm that Akt3 is involved in alsinLF-mediated neuroprotection, we examined whether the phosphorylation level of Akt3 was up-regulated by expression of alsinLF. To obtain clearer results, we made a NSC34 cell subline stably expressing His 6 /Myc-tagged human Akt3. Because its molecular weight is slightly larger than endogenous mouse Akt family proteins, it is easily identified by immunoblotting analysis. We then examined the Ser-472 (equivalent to Ser-473 of Akt1) phosphorylation levels of His 6 /Myc-tagged human Akt3 in these subline cells, which had been transfected with an empty vector, alsinLF-coding vector, or T701A-alinLF-coding vector at 48 h before harvest. As shown in Fig. 6D, enforced expression of alsinLF, but not T701A-alsinLF, resulted in up-regulation of the Ser-472 phosphorylation level of Akt3, supporting that Akt3 mediates alsinLF-mediated neuroprotection.

Akt3-mediated Neuroprotection Is Specifically Effective against SOD1 Mutant-induced Motoneuronal Cell Death-We
further examined whether neuroprotection by Akt3 is universal or specific using a constitutively active form of human Akt3 (cahAkt3). cahAkt3, a mutant of hAkt3, in which the N-terminal PH domain is deleted and the myristoylation signal of c-src is added, was constructed based on information regarding constitutively active mutant of Akt1 (42) (Fig. 7A). When wild-type hAkt3 (wthAkt3) or cahAkt3 was transiently expressed in NSC34 cells, phosphorylation of Ser-472 of Akt3 occurred more efficiently in cahAkt3 than wthAkt3, indicating that cahAkt3 is really constitutively active (Fig. 7B). Using this cahAkt3 expression vector, we examined which neurodegenerative celldeath signaling pathways are antagonized by Akt3 (Fig. 7C). NSC34 cells were cotransfected with expression vectors encoding FALS-, FAD-, or FPD-related genes in association with the backbone vector, the cahAkt1-expressing vector, or the cah-Akt3-expressing vector. As shown in Fig. 7C, cahAkt3 specifically antagonized SOD1 mutant-induced neurotoxicity, whereas cahAkt1 universally antagonized any neurotoxic insult, indicating that the alsinLF/Rac1-mediated neuroprotec- tive signal keeps its target specificity by selectively activating Akt3. Note that simultaneously performed immunoblot analysis indicated that co-expression of cahAkt1 again seemed to down-regulate expression of some SOD1 mutants (see also Fig.  4D, immunoblot part of the right panel).
Rac1 Is Located between AlsinLF and PI3K/Akt3-Taking all findings shown above together, we are able to conclude that the neuroprotective signal generated by alsinLF is transmitted to Rac1/PI3K/Akt3 in this order. However, there is no direct evidence that Rac1 is located upstream of PI3K/Akt3. To confirm that the Rac1 is located between alsinLF and PI3K/Akt3 in the alsinLF-mediated neuroprotective signal transduction cascade, we asked whether Rac1-mediated neuroprotection against toxicity by SOD1 mutants depends on Akt3. To this end, we actually examined the effect of siRNA-mediated disruption of endogenous Akt3 on Rac1-mediated neuroprotection. As shown in Fig. 8, disruption of Akt3 almost completely nullified Rac1mediated neuroprotection (p Ͻ 0.0001), indicating that Akt3 is really located downstream of Rac1. DISCUSSION We here demonstrated that alsinLF, a RhoGEF-containing protein, protects motoneuronal cells from SOD1 mutant-induced toxicity via the Rac1/PI3K/Akt3 pathway. We showed that alsinLF specifically binds to Rac1 but not to RhoA, Cdc42, or other Rho family proteins and activates Rac1 by increasing the amount of GTP-Rac1. In NSC34 cells, overexpression of a constitutively active Rac1 induces both apoptotic and antiapoptotic effects; it inhibits neurotoxicity by SOD1 mutants while itself simultaneously inducing cell death through uncharacterized mechanisms (Fig. 3A). By the gene-overexpression study alone we are unable to conclude that Rac1 is the mediator for alsinLF-mediated neuroprotection. However, a siRNA for Rac1 that selectively decreases mRNA and protein levels of Rac1 almost completely abolished the neuroprotective activity of alsinLF, clearly indicating that Rac1 plays a major role in alsinLF-mediated neuroprotection as an effector for alsinLF.
It has been found that Rac1 is involved in multiple deathrelated pathways and becomes pro-apoptotic (13,15,16,43) or anti-apoptotic (17)(18)(19)(20)(21), possibly depending on its interaction with other signaling molecules and on the type of cells in which it is expressed. In parallel, it has also been shown that upstream RhoGEFs for Rho family proteins play important roles in determining their downstream effectors, although the precise molecular mechanism remains unknown. For instance, if stimulated by a RhoGEF, Tiam-1, both Rac1 and Cdc42 activate PAK1 more efficiently than c-Jun N-terminal kinase. On the other hand, if stimulated by the RKB3 domain of another RhoGEF, FGD1, Cdc42 activates c-Jun N-terminal kinase more efficiently than PAK1 (44). The level of superoxide anion production by NADPH oxidase, another effector for Rac1, is also determined by RhoGEF species of Rac1. For example, Vav1 is the RhoGEF that most effectively induces superoxide anion production, whereas Vav2 and Tiam-1 increase the amount of the GTP-form Rac1 more efficiently than Vav1 (45). Altogether, it is speculated that alsinLF plays an important role in the Rac1-mediated anti-apoptotic activity, for example, by determining its downstream effector.
Although numerous downstream effector proteins exist for Rac1, this study has shown that the alsinLF-mediated neuroprotective effect is transmitted to the PI3K/Akt3 pathway. Similar cytoprotective signal transduction cascades have been reported in non-neuronal cells in earlier studies (17,20). Although it has been reported that Rac1 may activate PI3K by direct or nearly direct association, the precise molecular mechanism underlying signal transmission from Rac1 to PI3K largely remains to be elucidated.
Conversely, it has been well known that during migration of cells such as neutrophils, PI3K activates Rac1 (44). Phosphatidylinositol-3,4,5-triphosphate, generated by PI3K, stimulates the activity of RhoGEFs such as Dbl by binding to the pleckstrin domains of RhoGEFs. Activated RhoGEFs then increase the amount of GTP-Rac1. GTP-Rac1 in turn stimulates the NADPH oxidase, an effector of phagocytosis (46). Although this type of signal transduction has not been experimentally proven in apoptosis or cell survival, it may also potentiate alsinLFmediated neuroprotective signal transduction by a positive feedback mechanism in the alsinLF-mediated neuroprotective signal cascade.
The Akt family consists of three very similar proteins, Akt1, Akt2, and Akt3 (Fig. 7A). Akt1 was originally cloned as an oncogene Akt8. Akt family proteins are activated by PI3K through phosphatidylinositol-3,4,5-triphosphate-mediated translocation to the cell membrane and phosphoinositide-dependent kinase 1-mediated phosphorylation of Thr-308 and recently identified phosphoinositide-dependent kinase 2-mediated phosphorylation of Ser-473 of Akt (the positions of amino acids indicate the primary sequence of Akt1) (41,47,48). They antagonize a number of cell-death signals. This study demonstrated that alsinLF is another pro-survival molecule that protects motoneuronal cells by triggering a cell-survival signal leading to activation of Akt3. It has already been shown that a number of neurotrophic factors and growth factors including glia-derived neurotrophic factor, insulin-like growth factor-1, and vascular endothelial growth factor, which activate Akt kinases, protect motoneurons from SOD1 mutant-mediated toxicity, improve symptoms, and extend the life span of SOD1 mutant transgenic mice (27, 36 -38). Altogether, it is very reasonable that loss of alsinLF-mediated neuroprotective activity against SOD1 mutant-induced toxicity predisposes to ALS because the signaling pathway triggered by alsinLF promotes motoneuronal survival.
Because the three Akt isoforms are very similar in their primary structures (Fig. 6C) and they are activated in a similar manner in vitro, it has been postulated that the different Akt isoforms are functionally redundant (35). However, recent gene-targeting techniques clarified their difference in signal transduction. Disruption of Akt1 causes growth retardation and increased apoptosis in mice, but their glucose tolerance and insulin-stimulated disposal of blood glucose is normal (48,49). In contrast, mice lacking Akt2 show insulin resistance, severe diabetes, age-dependent loss of adipose tissue, and mild growth retardation (50,51). Mice lacking Akt1 and Akt2 show complicated symptoms including dwarfism, impaired skin development, skeletal muscle atrophy, delayed bone development, and impeded adipogenesis (52). Therefore, Akt1 and Akt2 are redundant in growth control, but their functions in glucose metabolism are different. Supporting the notion that each isoform of Akt has its specific function, recent studies have revealed specificity for interaction between Akt isoforms and some of their interactors. Both Akt1 and Akt2 interacted with all three TCL (T-cell leukemia-1 oncogene) family members, whereas Akt3 specifically interacted with TCL1 (53). It has been known that c-Jun N-terminal kinase interacting protein 1 (JIP1) binds to the PH domain of Akt. Interestingly, this interaction was selective for Akt1 (54). Different Akt isoforms appear to play different roles in the cellular responses to insulin, with Akt2 predominantly involved in this metabolic pathway. The roles of individual Akt family members in various human cancers (e.g. Akt2 in stomach cancer, Akt3 in ovarian cancer), further suggests subtleties in signaling specificity based on unique protein-protein interactions (55). We here found that Akt3 is specifically involved in alsinLF-mediated neuroprotection. Furthermore, constitutively active Akt1 has universal anti-apoptotic activities against any neurodegenerative insult, whereas constitutively active Akt3 is specifically effective against SOD1 mutant-mediated neurotoxicity. Thus, effectors shared by Akt1 and Akt3 probably play a critical role in their neuroprotection against toxicity by SOD1 mutants. The downstream molecules activated by alsinLF/Rac1/PI3K/Akt3mediated neuroprotective signal remain to be elucidated.
cascade selectively activates Akt3. The neuroprotective activity of alsinLF was effectively abolished by siRNA for Akt3 but not siAkt1 or Akt2. Considering that alsinLF/Rac1-mediated neuroprotection is selectively effective against SOD1 mutant-induced toxicity while PI3K has anti-apoptotic activities universally effective against various pro-apoptotic signals, we speculate that there should be an unidentified molecular mechanism involving PI3K, leading to specific activation of Akt3.
One possible candidate belonging to such molecules is a scaffold protein that brings these proteins and Akt3 close by binding to these proteins. ALS is a motoneuron-specific neurodegenerative disease without curative therapy. Elucidation of its pathogenesis is essential for development of more effective therapy. Based on the clinicopathological finding that the abnormality of either the ALS2 gene or the SOD1 gene eventually contributes to onset of typical manifestations of ALS, it is strongly suggested that these two genes play substantial roles in a common pathomechanism for the onset of ALS. Findings presented in our earlier study (11) suggested that alsinLF acts selectively as a pro-survival factor effective against motoneuronal loss induced by ALS-related toxic insults such as mutations of SOD1. Findings presented here provide substantial clues for understanding the precise molecular mechanism underlying alsinLF-mediated antagonization of neurotoxicity by SOD1 mutants, which will further contribute to elucidation of the precise ALS mechanism as well as to development of a new therapeutic approach for ALS.