PAK4 Suppresses Motor Neuron Degeneration in hSOD1G93A -Linked ALS Cell and Rat Models

Background: Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by the progressive loss of upper and lower motor neurons. The exact mechanisms underlying motor neuron death in ALS are still not fully understood, but a growing body of evidence indicates that inammatory could accentuate disease severity and accelerate disease progression. Currently, no neuroprotective strategies have effectively prevented the progression of this disease. Methods: IF, western blotting and RT-PCR were used to analyze the expression of PAK4 in vitro and in vivo models of ALS. We examined PAK4 function in ALS and the underlying mechanism by cell transfection, intraspinally injection of LV-PAK4 in hSOD1 G93A mice, ow cytometry, TUNEL staining, IF and western blotting. Results: Here, we observed that the expression and activity of PAK4 signicantly decreased in hSOD1 G93A -related cell and mouse models of ALS. In hSOD1 G93A mice (cid:0) the expression of PAK4 began to decrease at early-symptom stages of the disease. PAK4 silencing increased degeneration of motor neurons (NSC34 cells) and suppressed the CREB pathway. Overexpression of PAK4 protected motor neurons from hSOD1 G93A -induced degeneration by increasing the levels and transcriptional activity of CREB. The neuroprotective effect of PAK4 was markedly inhibited by compound 3i, a specic CREB inhibitor. In hSOD1 G93A -linked cell and mice, the CREB pathway, as the downstream target of decreased PAK4, was inhibited, and cell apoptosis increased. We also found that the expression of PAK4 was negatively regulated by miR-9-5p, and the miR-9-5p levels were upregulated in ALS. In vivo experiments revealed that PAK4 overexpression in the spinal neurons of hSOD1 G93A mice suppressed motor neuron degeneration, prolonged survival and promoted the CREB pathway. Conclusion: These results indicate that PAK4 plays a protective role for motor neurons by targeting CREB, suggesting it may be a useful therapeutic target in ALS. Immunouorescence analyses of the spinal cords of hSOD1 G93A mice and age-matched non-transgenic littermates(WT) were performed to detect differences in PAK4 levels associated with ALS. Results showed that compared with WT, signicantly reduced PAK4 (red) expression was observed in the MN (green) of hSOD1 G93A mice (Fig. 1A). Decreased PAK4 expression in ALS was conrmed by western blotting. We also found that the protein levels of pPAK4, an index of PAK4 activation, were lower in p140 hSOD1 G93A mice compared to WT (Fig. 1B and 1C). The pPAK4/PAK4 ratio was greater in hSOD1 G93A mice (Fig. 1D). These data suggested that the expression and activity of PAK4 in hSOD1 G93A mice were both downregulated. Next, we analyzed PAK4 mRNA levels in the spinal cords of hSOD1 G93A mice (n = 6) and WT (n = 6) using qRT-PCR. It was found that the gene-expression pattern was followed by protein expression. The levels of PAK4 mRNA were signicantly reduced in hSOD1 G93A mice (Fig. 1E). Results


Introduction
Amyotrophic lateral sclerosis (ALS) is a heterogeneous, fatal neurodegenerative disorder characterized by progressive loss of motor neurons (MNs) in the brain, brainstem and spinal cord, resulting in motor and extra-motor symptoms [1][2][3]. Approximately 5-10% of patients with ALS are familial (fALS), most often in an autosomal dominant pattern [4,5], whereas the rest of cases are sporadic (sALS) and may result from genetic mutations or environmental exposure [6]. Mutations in the SOD1 gene account for ~ 20% of fALS and ~ 5% of sALS [7,8,9], and transgenic mice carrying mutant SOD1 develop progressive paralysis [10]. Although the exact mechanisms underlying SOD1-induced toxicity are still mostly unknown [11], in ammation is proposed to cause MN degeneration in ALS. Currently, only riluzole and edaravone are approved by FDA for ALS clinical therapy. However, they could provide limited e cacy [12,13]. Thus, strategies to promote MN survival remain an urgent need for ALS treatment.
The serine/threonine p21-activated kinases (PAKs) are comprised of six mammalian proteins that are divided into group I (PAK1 to PAK3) and group II (PAK4 to PAK6) in terms of their structural and functional features [30]. Group II PAKs contribute to a wide range of intracellular processes including cytoskeletal dynamics, cell growth, tumorigenesis, neuronal dysfunction and cell survival [30][31][32][33][34]. Studies on PAK4 have revealed its pivotal role in cytoskeletal remodeling, neuronal development, axonal outgrowth and neuronal survival [24,35,36]. PAK4 knockout mice display dramatic defects in the heart and nervous system, with striking abnormalities observed in axonal outgrowth and neural tube formation [36]. Studies have shown that PAK4 expression in PD patients is down-regulated and that overexpressing it promotes DA neurons survival in the a-synuclein rat model [24]. Deregulation of AKT signaling is involved in ALS.
PAK4 has been shown to increase the levels and activity of CREB globally in diverse cell types [24,[43][44][45]. Given that PAK4 targets CREB and that CREB promotes MN survival, maintaining PAK4 activity is expected to play a pivotal role in preventing the progressive loss of MNs in ALS. Therefore, we hypothesized that decreased PAK4 activity might involve in the pathogenesis of ALS. We found that the expression and activity of PAK4 were deregulated in ALS models. Overexpression of PAK4 protected MNs from hSOD1 G93A -induced apoptosis, and this neuroprotective effect was alleviated by CREB inhibitor.
Finally, hSOD1 G93A mice injected with LV-PAK4 manifested a delay in disease onset, prolongation of survival and promotion of CREB signaling. These data suggest that the neuroprotective role of PAK4 in ALS MNs is mediated by the CREB pathway.

Animals
Transgenic mice carrying the hSOD1 G93A gene on the C57BL/6J strain background were obtained from Jackson Laboratory (Bar Harbor, ME, USA). All mice were bred and mainta-ined on a 12 h day/night cycle with adequate food and water. The offspring were identi ed by polymerase chain reaction (PCR) analysis of tail DNA [46]. Transgenic female mice were culled at postnatal day 75, 120 and 140, corresponding

Western blotting analysis
The whole spinal cords from mice or prepared cells were lysed in RIPA lysis buffer (Beyotime Biotechnology Co.,Ltd., Shanghai, China) containing phosphatase and protease inhibitor mixture (Roche, 4906845001 and 04693132001). Lysates were centrifuged at 14000 RPM for 20 min at 4 °C, and only the supernatants were analyzed by western blotting, as described previously [47].

qRT-PCR
For PAK4 mRNA quanti cation, total RNA was extracted from spinal cords or prepared cells using Trizol (Cwbio); cDNA was generated using ReverTra Ace qPCR RT Master Mix with gDNA Remover (TOYOBO CO.,LTD. Life Science Department OSAKA JAPAN). For miR-9-5p detection, miRNA was collected with an miRNA Puri cation Kit (Cwbio) and then reverse-transcribed using an miRNA cDNA Synthesis Kit (Cwbio).

Plasmid transfection and treatment
The cDNA plasmid (OriGene, Beijing, China) mediated overexpression of PAK4 in mSOD1 cells was performed as previous description [48] using

Disease course analysis and behavior tests
To test motor function, rotarod performance was examined in LV-PAK4 or LV-cherry injected hSOD1 G93A mice starting at 80 days of age. After 1 week of training, each mouse was placed on the rotating rod at 16 rpm every 2 days to record the latency time before the mouse would fall off. The test was repeated three times for each animal. Disease onset was de ned as the age at which a mouse could no longer stay on the rotating rod for 180 s. The date of disease endpoint was recorded when the mouse could not rise within 30 s after being placed on its side.

Statistical analyses
The data were analyzed using SPSS version 23.0 and GraphPad Prism 8.0 software (GraphPad Software, USA). Statistical signi cance was evaluated with Student's t-test or one-way analysis of variance (ANOVA). Quantitative data were expressed as mean values of at least three determinations ± SD. Tests were considered statistically signi cant at P < 0.05.

Results
Reduced PAK4 expression and activity in the spinal cord of hSOD1 G93A mice Immuno uorescence analyses of the spinal cords of hSOD1 G93A mice and age-matched non-transgenic littermates(WT) were performed to detect differences in PAK4 levels associated with ALS. Results showed that compared with WT, signi cantly reduced PAK4 (red) expression was observed in the MN (green) of hSOD1 G93A mice (Fig. 1A). Decreased PAK4 expression in ALS was con rmed by western blotting. We also found that the protein levels of pPAK4, an index of PAK4 activation, were lower in p140 hSOD1 G93A mice compared to WT ( Fig. 1B and 1C). The pPAK4/PAK4 ratio was greater in hSOD1 G93A mice (Fig. 1D).
These data suggested that the expression and activity of PAK4 in hSOD1 G93A mice were both downregulated. Next, we analyzed PAK4 mRNA levels in the spinal cords of hSOD1 G93A mice (n = 6) and WT (n = 6) using qRT-PCR. It was found that the gene-expression pattern was followed by protein expression.
To characterize the temporal expression patterns of PAK4 in ALS mice, we evaluated PAK4 levels in hSOD1 G93A and WT mice at three stages of the disease (the presymptomatic stage P75, early-sym stage P120 and late-sym stage P140) using immunohistochemistry. Results showed that in both groups, PAK4 was mainly expressed in the cytoplasm and nuclei of MN ( Fig. 2A). Compared with WT, PAK4 immunoreactivity was signi cantly lower at early-sym and late-sym stages of the disease in ALS mice ( Fig. 2A). However, at the presymptomatic stage, OD values of PAK4 were not signi cantly altered (Fig. 2B). Taking these data together, we predicted that PAK4 was involved in the pathophysiological process of ALS.
PAK4 protein levels and activity are down-regulated in mSOD1 cells Next, we explored whether the expression and activity of PAK4 were altered in vitro models of ALS. NSC34 cells exhibit many MN characteristics, including the generation of action potentials, expression of neuro lament proteins and synthesis of acetylcholine [49][50][51]. Our group has con rmed that choline acetyltransferase, an MN marker, is expressed in NSC34 cells [40]. Thus, NSC34 cells stably transfected with hSOD1 G93A (mSOD1) were used as a cell model of ALS in subsequent studies. Immunocytochemistry staining showed that PAK4 (red) was expressed both in the cytoplasm and nuclei of pLV, wtSOD1 and mSOD1 cells. Compared to pLV cells, mSOD1 cells displayed weaker uorescence intensity of PAK4, while wtSOD1 cells displayed a similar intensity (Fig. 3A). Decreased PAK4 expression in mSOD1 cells was con rmed by western blotting. We also found that compared with pLV, protein levels of pPAK4 were signi cantly decreased in mSOD1 cells, while wtSOD1 and pLV cells had similar pPAK4 protein levels ( Fig. 3B and 3C). The pPAK4/PAK4 ratio in mSOD1 cells was greater than in control pLV and wtSOD1 cells (Fig. 3D). These data suggested that the expression and activity of PAK4 in mSOD1 were both down-regulated. Next, qRT-PCR analysis suggested that PAK4 mRNA levels of mSOD1 cells were robustly downregulated to approximately 18.9% compared to pLV cells. PAK4 mRNA expression in wtSOD1 and pLV cells showed no statistical difference (Fig. 3E).
Knockdown of PAK4 promotes motor neuron degeneration and decreases the levels and activity of CREB To determine the biological role of PAK4 in MN, we selected NSC34 cells with many MN characteristics for further study. Three siRNAs targeting PAK4 were transfected into NSC34 cells, and the e ciency of PAK4 knockdown was evaluated by western blotting and qRT-PCR (Fig. 4A). We selected siPAK4-2 for subsequent experiments because of its obvious knockdown effects. Flow cytometry analysis suggested that NSC34 cells transfected with siPAK4-2 yielded a signi cantly higher rate of cell apoptotic death (49.34 ± 0.95%) than siNC control (32.07 ± 2.45%) ( Fig. 4B and Fig. 4C). Consistently, TdT-mediated dUTP nick end labeling (TUNEL) staining indicated that knockdown of PAK4 signi cantly increased the percentage of cells experiencing apoptotic death (35.67 ± 2.51%), and in the siNC group, the percentage of cell apoptosis was 20.57 ± 1.50% (Fig. 4D and Fig. 4E). These data suggested that PAK4 protected NSC34 cells from apoptotic death.
Next, we further investigated the mechanism by which PAK4 promoted cell survival. Previous studies have revealed that PAK4 is a critical regulator of CREB and that CREB promotes MN survival [52]. Thus, western blotting of PAK4, CREB, pCREB and the CREB target proteins (PGC-1a and Bcl-2) was performed with extract from siNC and siPAK4-2 groups. The results showed that the levels of CREB, pCREB, Bcl-2 and PGC-1a were decreased in PAK4 knockdown NSC34 cells. These data indicated that PAK4 might inhibit motor neuron degeneration via the CREB pathway ( Fig. 4F and 4G).

CREB pathway mediates PAK4-induced neuroprotection in mSOD1 cells
To investigate whether the neuroprotective role of PAK4 would extend to cell models of ALS, we upregulated PAK4 expression using a plasmid vector for PAK4 in mSOD1 cells. The results of western blotting and qRT-PCR showed that PAK4 plasmid signi cantly increased PAK4 expression in mSOD1 cells (Fig. 5A). To further determine whether the possible neuroprotective effect of PAK4 in mSOD1 cells was mediated by CREB signaling, we selectively inhibited CREB function by using compound 3i (a speci c inhibitor of CREB) while transfecting the PAK4 plasmid or empty-vector (EV). Accordingly, the results showed that the percentage of cell apoptotic death was lowest in the PAK4 + group (15.19 ± 1.98%). In the PAK4 + +3i group, 3i impaired the neuroprotective function of PAK4 (24.93 ± 2.27%). In the EV and EV + 3i groups, the percentages of apoptotic death were 31.70 ± 1.68% and 45.70 ± 0.87% (Fig. 5B and 5C). This effect was further con rmed by TUNEL (Fig. 5D and 5E), and administration of PAK4 improved mSOD1 cell viability (10.25 ± 1.29%). The ability of PAK4 to relieve mSOD1-induced cytotoxicity was compromised in the PAK4 + +3i group (16.28 ± 0.86%). In the EV and EV + 3i groups, the percentages of apoptotic death were 20.84 ± 2.17% and 34.39 ± 1.89%. Furthermore, western blotting studies con rmed increased PAK4, CREB, pCREB, Bcl-2 and PGC-1a levels in the PAK4 + group compared with EV control (Fig. 6A-F). In addition, 3i inhibited CREB phosphorylation and transcription activity effectively. Compared with the PAK4 + group, the levels of pCREB, Bcl-2 and PGC-1a were decreased in the PAK4 + +3i group, while the levels of PAK4 and CREB showed no signi cant changes (Fig. 6A, 6E-F). Thus, these data suggested that CREB and its target proteins BCL-2, PGC-1a mediated the neuroprotective effect of PAK4.
CREB pathway and motor neuron survival are suppressed in mSOD1 cells and hSOD1 G93A mice To explore the effects of deregulated PAK4 expression on downstream consequences and neurodegeneration in ALS models, the protein levels of CREB signaling and cleaved-caspase3 were analyzed in mSOD1 cells and the spinal cords from hSOD1 G93A mice. Immunoblotting analysis revealed that mSOD1 cells exhibited low levels of CREB, pCREB and CREB target proteins than pLV cells, while wtSOD1 cells displayed similar levels ( Fig. 7A and 7B). Down-regulated PAK4 markedly increased the levels of cleaved-caspase3 in mSOD1 cells ( Fig. 7A and 7C). In the in vivo study, the levels of CREB, pCREB, Bcl-2 and PGC-1a in the spinal cords of late-sym stage hSOD1 G93A mice were signi cantly reduced compared with WT ( Fig. 7D and 7E). Moreover, hSOD1 G93A mice exhibited higher levels of cleaved-caspase3 than control ( Fig. 7D and 7F). These data suggested that down-regulated PAK4 expression impaired the CREB pathway and promoted neurodegeneration in ALS models.
PAK4 expression is negatively regulated by miR-9-5p in ALS MicroRNAs (miRNAs), small non-coding RNAs, repress gene translation and promote target mRNA degradation [53]. It has been reported that PAK4 is target regulated by a variety of miRNAs, including miR-433 [54], miR-224 [55], miR-199a-3p [56] and miR-9-5p [57]. Moreover, the expression of miR-9-5p in the cerebrospinal uid of ALS patients is increased [58]. We speculated that PAK4 might be inversely regulated by miR-9-5p in ALS. To test our hypothesis, we rst analyzed miR-9-5p levels in ALS models using qRT-PCR. Results showed that compared with control, the expression of miR-9-5p in spinal cords of p140 hSOD1 G93A mice was increased, especially in the mSOD1 cell line (Fig. 8A). In our previous studies, we have found that the mRNA levels of PAK4 were decreased in vivo and in vitro models of ALS ( Fig. 1E  and 3D). To determine whether miR-9-5p negatively regulated PAK4 mRNA expression, miR-9-5p inhibitor transfection was conducted on mSOD1 cells. The results showed that downregulation of miR-9-5p signi cantly upregulated the expression of PAK4 mRNA and protein (Fig. 8B).
Delayed disease onset and extended lifespan of hSOD1 G93A mice injected with LV-PAK4 Since PAK4 played a neuroprotective role in vitro, we next evaluated the in vivo biological function in hSOD1 G93A mice. We injected hSOD1 G93A mice intraspinally at 60 days of age with LV-PAK4 or LV-cherry. Four weeks postinjection, the spinal cord sections of treated mice were examined for cherry expression by immuno uorescence (Fig. 9A). The data showed that transduction e ciency was high in the ventral horns of the lumbar spinal cord. Western blotting and qRT-PCR further determined the results. The protein levels of PAK4 were higher in the spinal cords of LV-PAK4-injected mice compared with LV-cherry-injected ones (Fig. 9B). Furthermore, qRT-PCR detected signi cantly higher PAK4 mRNA levels in LV-PAK4 injected mice compared with the controls (Fig. 9C). Overexpression of PAK4 protects motor neurons from degeneration and enhances CREB signaling in hSOD1 G93A mice To test the effects of PAK4 overexpression on MN degeneration, we performed immunostaining with anti-MAP-2 to examine the number of MNs in the spinal cords of LV treated hSOD1 G93A mice. We found that the number of MNs surviving in LV-PAK4-injected mice was signi cantly increased compared with LVcherry-injected mice (cervical spinal cord: 32.00 ± 3.61 vs. 12.33 ± 2.08; lumbar spinal cord: 29.67 ± 3.06 vs. 14.33 ± 2.52; P < 0.05) ( Fig. 10A and 10B). Furthermore, immunoblotting analysis was performed to determine the effects of PAK4 overexpression on downstream consequences and neurodegeneration. Results revealed that LV-PAK4 injected mice exhibited low levels of CREB, pCREB, PGC-1a and Bcl-2 compared with LV-cherry injected mice ( Fig. 10C and 10D). Overexpression of PAK4 markedly decreased the levels of cleaved-caspase3 in hSOD 1G93A mice ( Fig. 10C and 10E). These data suggested that PAK4 exerted neuroprotective effects by activating CREB signaling in the hSOD1 G93A mouse models of ALS.

Discussion
ALS is a relentless and fatal neurodegenerative disease. Development of neuroprotective treatments to prevent or delay the progression of the disease remains an unachieved goal. Here, we validate for the rst time that PAK4 is a crucial survival factor for MN. The expression and activity of PAK4 are downregulated in vivo and in vitro models of ALS. Activation or overexpression of PAK4 and its downstream signaling pathways might serve as a new potential therapeutic method for ALS.
PAK4 expresses ubiquitously and plays an essential role in multiple biological processes [36,59]. Dysregulation of PAK4 expression involves in various diseases [60]. The functions of PAK4 in diseases have principally been investigated in cancers, and overexpression of it contributes to tumorigenesis [61][62][63]. In the nervous system, PAK4 is required for neuronal development and axonal outgrowth [35,36]. In dopaminergic neurons of PD patients, the expression and activity of PAK4 are down-regulated and this decline may promote PD pathogenesis [24]. However, the relationship between PAK4 and ALS disease remains unclear. In this context, we demonstrate for the rst time that PAK4 protein expression as well as activity signi cantly declined in cells and mice carrying hSOD1 G93A . Considering that the expression of PAK4 declined at early-sym stages, PAK4 might be involved in ALS disease progression.
A series of ndings have shown that activated CREB promotes neuronal survival through a transcriptiondependent pathway [14][15][16][17]20]. Moreover, the activation of CREB-mediated downstream proteins plays a neuroprotective role in PD and AD rat models [22,24,64]. Increasing expression of p-CREB can improve MN function and prolong the life span of ALS mice [28]. In accordance with previous studies, we found that PAK4 protected MN from mSOD1-induced cytotoxicity through increasing CREB expression and activity. 3i amelior-ated the neuroprotective effects of PAK4 and increased apoptosis by inhibiting the transcription activity and phosphorylation of CREB in ALS cell models. Thus, we further con rmed the protective role of CREB in the normal nervous system and neurodegenerative diseases.
Accumulating evidence suggests that CREB is a target of PAK4. In the present study, we found that knockdown of PAK4 reduced the levels and activity of CREB in NSC34 cells. Overexpression of PAK4 increased CREB levels and activity in mSOD1 cells. In mice and cell models of ALS, the low expression of PAK4 caused inhibition of the CREB pathway and promoted MN degeneration. Cumulatively, our ndings are consistent with previous works that PAK4 sustains high CREB levels and improves CREB activity in various tissues [43,44]. However, it is unclear how PAK4 maintains high levels of CREB. Additional studies are required to further elucidate the relationship between PAK4 and CREB expression and degradation. PAK4 could not directly phosphorylate CREB in vitro, suggesting that PAK4 activated CREB transcription by maintaining a high level. Whether PAK4 increases the transcriptional activity of CREB through a phosphorylation-independent pathway in ALS needs subsequent study.
Growing evidence reveals that mitochondrial dysfunction plays a crucial role in ALS [65]. PAK4 can inactivate the pro-apoptotic protein BAD (members of the Bcl-2 family pro-apoptotic activities) by speci cally phosphorylating S112 [66]. This phosphorylation protects cells from apoptosis by inhibiting BAD localization in the mitochondrial outer membrane. siRNA treatment decreased PAK4 expression in parallel with inhibition of the CREB downstream effector proteins PGC-1a and Bcl-2. Conversely, overexpression of PAK4 promoted the levels of Bcl-2 and PGC-1a in cell models of ALS. Moreover, 3i decreased PAK4-induced expression of the CREB downstream proteins Bcl-2 and PGC-1a, thereby reducing cell survival. These data demonstrate that the up-regulation of Bcl-2 and PGC-1a mediates PAK4-induced neuroprotection. Considering the close association of PGC-1a and Bcl-2 with mitochondrial health [67,68], PAK4 may contribute to mitochondrial protection in a transcription-dependent manner.
Thus, transcription-dependent and transcription-independent theories could explain the mechanisms that PAK4 protects mitochondria in the presence of neurotoxic stimuli.
Although PAK4 is a potential therapeutic target for ALS, persistent expression of it in the spinal cord after gene therapy might contribute to tumorigenesis because PAK4 is a potent oncogene. However, we did not monitor tumor formation in LV-PAK4 injected hSOD1 G93A mice, but we did not maintain lengthy follow-up. In addition, a lower cancer risk was observed in human ALS patients after diagnosis compared with healthy individuals [69]. Thus, we speculate that a lack of tumorigenesis is likely due to a poor microenvironment for cell survival and proliferation in neurodegenerative diseases. ALS is characterized by progressive death of upper and lower MN, suggesting that the brain and spinal cord could not provide a proliferative environment. Moreover, most primary brain tumors develop from glial and meningeal cells.
Thus, although PAK4 is regarded as an oncogenic gene, if a local delivery method can guarantee MNspeci c targeting, this could prevent undesirable effects and provide a promising therapeutic intervention for ALS. We also consider CREB as another target for modulation of the PAK4 neuroprotective pathway. CREB may be less potent than PAK4 because the role of CREB is more restricted compared with PAK4, but it may speci cally enhance the expression of Bcl-2 with a reduced risk of tumor formation.
Group I of the PAK proteins plays a vital role in spinocerebellar ataxia type 1 (SCA1) [52], fragile X syndrome (FXS) [70], X-linked mental retardation [71], and neurodegenerative diseases such as HD [31,32] and AD [72]. Rescue from disease pathology in cell and mice models of SCA1 and improvement of abnormal behavior in FXS mice model by group I PAK inhibitors, offer proof of PAKs as therapeutic targets [73,74]. Recent studies have implicated PAK4 and PAK6 in PD by demonstrating that PAK4 plays a neuroprotective role in PD models and that PAK6 is a downstream regulator of PD-causing mutations [24,75]. Moreover, PAK signaling is stimulated by ALS2; mutations in the ALS2 gene cause some rare juvenile forms of ALS [76,77]. Our ndings demonstrated that PAK4 plays a neuroprotective role in ALS models. In aggregate, these studies suggest that dysregulation of PAKs is a pathogenic mechanism in a series of neurological diseases. Thus, the development of more selective and effective interventions targeting PAKs may be a promising therapeutic strategy.
In summary, the current ndings identify a neuroprotective role for PAK4 in ALS models. A limitation of this study is the absence of human postmortem brain samples; accordingly, we did not study how PAK4 expression levels correlate with ALS in human patients. Further work needs to address the mechanism for PAK4 increasing the levels of CREB and determine if there are other mechanisms for PAK4 enhancing CREB transcriptional activity in ALS. The neuroprotective pathway of PAK4 still needs to be better clari ed before targeting PAK4 as a therapeutic strategy in patients with ALS.

Conclusions
This work demonstrates that PAK4 protein levels and activity are down-regulated in hSOD1 G93A -linked ALS cell and rat models, and overexpression of it protects motor neurons from degeneration. The neuroprotective effect of PAK4 is mediated by activation of CREB sianaling pathway. Therefore, PAK4 may be a potential therapeutic target in ALS.

Figure 2
The intensity of PAK4 immunohistochemistry staining is decreased in motor neuron of hSOD1G93A mice at early-sym and late-sym stages. Para n sections of cervical spinal cords from hSOD1G93A mice and WT at three stages of the disease were subjected to immunohistochemistry. (A) PAK4-stained MNs (black arrow) were detected in the anterior horn of spinal cords from both hSOD1G93A mice and WT. Scale bar = 50 μm. (B) At stages p120 and p140, optical density (OD) values of PAK4 signi cantly decreased in hSOD1G93A mice, while there was no deregulation at stages p75 (n = 3/group, 6 sections/mouse). Data were provided as means ± SD and were tested for signi cance using Student's t test. ns ≥ 0.05, *p < 0.05.  normalized to β-actin. Data were presented as means ± SD. *P < 0.05, **P < 0.01, ***p<0.001. Student's t test (C, E, G) or ANOVA with Student-Newman-Keuls post hoc analysis (A). Scale bar = 50 μm.    The expression of miR-9-5p and PAK4 at mRNA and protein levels in mSOD1 cells transfected with anti-miR-9-5p or miR-NC were determined by qRT-PCR and western blotting. Data were presented as means ± SD. *P < 0.05, **P < 0.01, ***p<0.001 vs. Control group. Student's t test.