Loss of ALS2/Alsin Exacerbates Motor Dysfunction in a SOD1H46R-Expressing Mouse ALS Model by Disturbing Endolysosomal Trafficking

Background ALS2/alsin is a guanine nucleotide exchange factor for the small GTPase Rab5 and involved in macropinocytosis-associated endosome fusion and trafficking, and neurite outgrowth. ALS2 deficiency accounts for a number of juvenile recessive motor neuron diseases (MNDs). Recently, it has been shown that ALS2 plays a role in neuroprotection against MND-associated pathological insults, such as toxicity induced by mutant Cu/Zn superoxide dismutase (SOD1). However, molecular mechanisms underlying the relationship between ALS2-associated cellular function and its neuroprotective role remain unclear. Methodology/Principal Findings To address this issue, we investigated the molecular and pathological basis for the phenotypic modification of mutant SOD1-expressing mice by ALS2 loss. Genetic ablation of Als2 in SOD1H46R, but not SOD1G93A, transgenic mice aggravated the mutant SOD1-associated disease symptoms such as body weight loss and motor dysfunction, leading to the earlier death. Light and electron microscopic examinations revealed the presence of degenerating and/or swollen spinal axons accumulating granular aggregates and autophagosome-like vesicles in early- and even pre-symptomatic SOD1H46R mice. Further, enhanced accumulation of insoluble high molecular weight SOD1, poly-ubiquitinated proteins, and macroautophagy-associated proteins such as polyubiquitin-binding protein p62/SQSTM1 and a lipidated form of light chain 3 (LC3-II), emerged in ALS2-deficient SOD1H46R mice. Intriguingly, ALS2 was colocalized with LC3 and p62, and partly with SOD1 on autophagosome/endosome hybrid compartments, and loss of ALS2 significantly lowered the lysosome-dependent clearance of LC3 and p62 in cultured cells. Conclusions/Significance Based on these observations, although molecular basis for the distinctive susceptibilities to ALS2 loss in different mutant SOD1-expressing ALS models is still elusive, disturbance of the endolysosomal system by ALS2 loss may exacerbate the SOD1H46R-mediated neurotoxicity by accelerating the accumulation of immature vesicles and misfolded proteins in the spinal cord. We propose that ALS2 is implicated in endolysosomal trafficking through the fusion between endosomes and autophagosomes, thereby regulating endolysosomal protein degradation in vivo.

ALS2 is a causative gene for a juvenile autosomal recessive form of motor neuron diseases (MNDs) [4,5,6], including amyotrophic lateral sclerosis 2 (ALS2) [7] (OMIM 205100), juvenile primary lateral sclerosis (PLSJ) [8] (OMIM 606353), and infantile-onset ascending hereditary spastic paralysis (IAHSP) [9] (OMIM 607225).These disorders are characterized by ascending degeneration of UMN with or without LMN involvement.A total of 19 independent ALS2 mutations from 17 families have been reported [4,5,6,10,11,12,13,14].They are predicted to result in either premature termination of translation or substitution of an evolutionarily conserved amino acid for the ALS2-coded protein, ALS2 or alsin, leading to loss of its function.ALS2 is a guanine nucleotide exchange factor for the small GTPase Rab5 [15] and involves in macropinocytosis-associated endosome trafficking and fusion [16,17], and neurite outgrowth [17,18,19].Loss of these functions accounts for motor dysfunction and axonal degeneration in the ALS2-linked MNDs.Logically designed mouse studies could provide persuasive evidence that loss of ALS2 triggers motor neuron degeneration.However, mice lacking ALS2 cannot recapitulate the complex disease phenotypes, despite the subclinical levels of motor dysfunction and axonal degeneration in aged animals [6,20,21].Thus, although there are potentially important clinical implications of these observations, the physiological functions of ALS2 and the molecular mechanisms underlying the motor dysfunction resulting from ALS2 deficiency remain to be clarified.
Most efforts to delineate ALS/MND pathogenesis have converged on the mutations in SOD1, encoding Cu/Zn superoxide dismutase (SOD1), which accounts for most prevalent form of the autosomal dominant familial ALS [22].More than 120 different SOD1 mutations have been identified (http://alsod.iop.kcl.ac.uk/ als), and several transgenic mouse lines expressing disease causative SOD1 mutants have been generated and thoroughly characterized [23].Nonetheless, no consensus has yet emerged as to how SOD1 mutations lead to selective death of motor neurons, except that multiple toxicity pathways including oxidative stress, endoplasmic reticulum (ER) stress, excitotoxicity, mitochondrial dysfunction, neural inflammation, protein misfolding and accumulation, and dysfunctional intracellular trafficking, are implicated in the pathogenesis of ALS/MNDs [1,24].
Recently, it has been reported that overexpression of ALS2 protects cultured motor neuronal cells from toxicity induced by mutant SOD1 [25].Further, loss of ALS2 renders neurons more susceptible to excitotoxicity [26], while cell death induced by neurotoxic stimuli, such as N-methyl-D-aspartate, is significantly suppressed by overexpression of ALS2 [27].Moreover, Als2-null mice are slightly vulnerable to oxidative stress [28].These observations imply a neuroprotective function of ALS2 against ALS/MND-associated pathological insults.On the other hand, in vivo studies have demonstrated that loss of ALS2 does not affect the motor neuron degeneration and survival of SOD1 G93A mice [29,30], which does not support the functional interaction between ALS2 and mutant SOD1-mediated toxicity in vivo.However, with the use of only a single mutant SOD1 transgenic line, such notions still remain inconclusive.Rather, it is possible that the extremely rapid progression of motor dysfunction observed in high-copy number SOD1 G93A mice could overwhelm the modest symptoms by the ALS2 deficiency [29].
To clarify these issues, we used SOD1 H46R mice, which exhibit a widespread axonal degeneration with slowly progressive motor neuron degeneration in the spinal cord [31] instead of SOD1 G93A mice, and generated SOD1 H46R mice on an Als2-null background.SOD1 H46R mutation accounts for a mild form of familial ALS that was originally identified in Japanese kindred [32].We here revealed that loss of ALS2 exacerbated the SOD1 H46R -associated disease symptoms in mice, and identified ALS2 as a novel regulator for the endolysosomal system.Our findings suggest that ALS2 plays a role in the maturation of autophagosomes through the endosome-autophagosome fusion, thereby regulating endolysosomal trafficking in vivo.

Loss of ALS2 aggravates motor dysfunction in SOD1 H46R mice
As mice expressing SOD1 H46R exhibited progressive motor dysfunction and paralysis, particularly of the hind limbs, we next assessed whether loss of ALS2 in SOD1 H46R mice affects the course of motor deficits by conducting quantitative behavioral analyses.First, to evaluate motor coordination and balance, we performed balance beam test, by which the onset of disease could be sensitively determined [34].Als2 2/2 ;SOD1 H46R mice showed an earlier motor dysfunction, in which the onset of disease (,15 weeks of age) was approximately 3 weeks earlier than those (,18 weeks of age) of Als2 +/+ ;SOD1 H46R or Als2 +/2 ;SOD1 H46R littermates (Figure 2A).Further, analyses of rearing and cage activities revealed that although there were no significant differences in their activities among all pre-symptomatic mice (12 weeks of age) with different genotypes, Als2 2/2 ;SOD1 H46R mice showed a significantly lower spontaneous motor activity than wild-type or Als2 +/2 ;SOD1 H46R littermates at a later stage of the disease (18 week of age) (Figure 2B and 2C).

SOD1 H46R mice show an axonal degeneration in the spinal tracts from an early symptomatic stage
To determine whether these motor phenotypes are associated with motor neuron degeneration, we conducted histological analyses using early symptomatic mice (16-18 weeks of age).Although there were no evidences for the motor neuron loss at this stage, a wide-spread axonal degeneration in the spinal tracts of the lateral and ventral columns was evident, particularly in Als2 2/2 ;SOD1 H46R mice (Figure 3A).
Osmiophilic granular aggregates, multivesicular bodies, and autophagosome-like structures are accumulated in the spinal cord of SOD1 H46R mice To investigate the histopathology in more detail, we next conducted an electron microscopic (EM) analysis.Although there were no observable abnormalities in soma of motor neurons (data not shown), we observed the degenerative and swollen axons with the accumulation of granular aggregates, disorganized fibrillar materials, multivesicular bodies (MVBs), and/or autophagosome-like vesicles in the spinal cord of Als2 2/2 ;SOD1 H46R and Als2 +/+ ;SOD1 H46R mice at an early symptomatic stage (Figure 3B and 3C).Further, membrane saccules containing granular/osmiophilic aggregates (Figure 3D) and autophagosome-like vesicles (Figure 3E) were observed.Astrocytes containing osmiophillic aggregates were also occasionally observed (Figure 3F).Notably, in Als2 2/2 ;SOD1 H46R mice, these pathologic phenotypes were evident even at 8 weeks of pre-symptomatic stage (Figure 3G), but not in a same stage of Als2 +/+ ;SOD1 H46R mice (data not shown).These data underscore the relevance of axonal dysfunction and/or degeneration, probably caused by the dysfunctional axonal trafficking, on the disease onset in SOD1 H46Rexpressing mice.
Loss of ALS2 promotes the progressive accumulation of insoluble proteins in the spinal cord of SOD1 H46R mice First, we confirmed that there were no significant differences in the expression levels of the SOD1 H46R transcript (Figure S4A) and soluble SOD1 H46R protein (Figure 4, S5A, and S6A) among three different groups of mice carrying the human SOD1 transgene; i.e., Als2 +/+ ;SOD1 H46R , Als2 +/2 ;SOD1 H46R , and Als2 2/2 ;SOD1 H46R .These results indicate that the exacerbation of motor dysfunction seen in SOD1 H46R mice lacking ALS2 was not simply due to the increased level of the SOD1 H46R expression.
To identify the molecular factors that were associated with behavioral and pathological features observed in Als2 2/2 ;SOD1 H46R mice, we conducted western blot analysis of the lumbo-sacral cord extracts using a panel of antibodies, and performed their quantitative analyses.Although the levels of the ALS-related neuronal intermediate filament proteins [peripherin and neurofilament heavy chain (NFH)] [35] and the ALS-causative gene product [TAR DNA-binding protein 43-kD (TDP-43)] [1,2] were unchanged, a progressive accumulation of insoluble high-molecular weight (HMW) SOD1 and poly-ubiquitinated proteins was observed in mice expressing SOD1 H46R , particularly in those lacking ALS2 (Als2 2/2 ;SOD1 H46R ) (Figure 4, S5B, S5C, S5E, S6A, and S6B).Further, from 16 weeks of pre-and early-symptomatic stage, glial intermediate filament proteins, vimentin and glial fibrillary acidic protein (GFAP), were accumulated (Figure 4, S5J, S5K, and S6E).Remarkably, the levels of two macroautophagy (hereafter referred to as autophagy)-associated proteins, polyubi-quitin binding protein p62/SQSTM1 (p62) and a lipidated-form of microtubule-associated protein 1-light chain 3 (LC3-II), both of which were selectively degraded by autophagy-lysosomal system [36], were significantly increased in insoluble fractions of Als2 2/2 ;SOD1 H46R mice (Figure 4, S5G, S5I, S6C, and S6D).Similar results were obtained in samples from the brainstem, cerebellum, and cervical cord, but not from the cerebral cortex (Figure S7).A quantitative reverse transcriptase PCR (qRT-PCR) revealed that although expression of Vim and Gfap was significantly upregulated as the disease progressed, no transcriptional activation of SOD1 H46R , Sqstm1 (p62), and Map1lc3a/Map1lc3b (LC3A/LC3B) was observed (Figure S4).These results suggest that the accumulation of insoluble mutant SOD1, p62, and LC3-II is not simply due to an increased expression of these proteins, but rather to a decrease in their degradation.Since mutant SOD1 is degraded by both the proteasome and autophagy [37,38], impairment of the ubiquitinproteasome system (UPS) and/or the autophagy-endolysosomal system could result in the accumulation of such insoluble proteins in SOD1 H46R mice, particularly in those lacking ALS2.

Proteasome activity in the spinal cord is increased as the disease progresses
To investigate the contribution of UPS impairment on the protein accumulation observed, we analyzed the catalytic activity for the 20S proteasome in the spinal cord from mice with different genotypes.Surprisingly, the proteasomal activity was induced rather than impaired in SOD1 H46R -expressing mice at a late symptomatic stage (20 and 23 weeks of age) (Figure 5).Although loss of ALS2 by itself did not affect the proteasome activity (wildtype vs Als2 2/2 at 18 weeks of age), ALS2 deficiency in SOD1 H46R mice seems to result in the earlier enhancement of proteasome activity in the spinal cord (20 weeks of age) (Figure 5), which might be correlated with the disease progression.Importantly, the levels of LMP2, a inducible subunit of immunoproteasome, was increased in symptomatic Als2 2/2 ;SOD1 H46R mice (Figure 4), consistent with recent findings [39,40].Since the levels of the constitutive subunits of 20S proteasome (a5 and C2) and one of the molecular chaperone Hsp70 were unchanged (Figure 4), the UPS was not severely affected in SOD1 H46R mice, at least, at the disease stages examined, but rather enhanced in certain cell types within the spinal cord in response to the disease progression.Thus, the accelerated accumulation of HMW SOD1 and polyubiquitinated proteins from an early symptomatic stage in SOD1 H46R mice is not simply due to impairment of the UPS.

Loss of ALS2 results in a decreased level of lysosomal clearance of autophagy-associated proteins in fibroblasts
Previously, it has been shown that ALS2 regulates the trafficking and clearance of internalized molecules such as epidermal growth factor (EGF) [20] and glutamate receptors [44] in cultured cells.To further clarify the functional interaction of ALS2 with the autophagosomal-endolysosomal protein degradation in general, fibroblasts derived from wild-type and Als2 2/2 mice were subjected to autophagic-flux analysis [45].Since the intracellular levels of LC3-II correlate with the number of autophagosomes, which is regulated by a balance between autophagosome formation and degradation, the LC3-II levels in the presence or absence of lysosomal inhibitors can be used as ''autophagomometer'' to measure the autophagic-flux [45].Under steady-state conditions (unstarved), the treatment with chloroquine (CQ), a lysosomotropic agent that inhibits the lysosomal proteases, resulted in a comparable increase in the LC3-II level in Als2 2/2 cells with those in wild-type (Figure 8A), indicating that loss of ALS2 by itself does not seem to Western blot analysis of the levels of proteins, including ALS2, SOD1, ubiquitin (Ub), polyubiquitin binding protein p62/SQSTM1 (p62), microtubule-associated protein 1-light chain 3 (LC3), peripherin, neurofilament heavy chain (NFH), TAR DNA-binding protein 43-kD (TDP-43), heat-shock protein Hsp70, 20S proteasome subunits (a5, C2, and LMP2), vimentin, and glial fibrillary acidic protein (GFAP), in the lumbo-sacral cord from 8, 12, 16, and 20 week-old mice with four distinct genotypes; wild-type (Als2 +/+ ), Als2 +/+ ;SOD1 H46R , Als2 +/2 ;SOD1 H46R , and Als2 2/2 ;SOD1 H46R .Two fractions; 1% Triton X-soluble fraction (TXsoluble; left panels) and 1% Triton X-insoluble/5% SDS-soluble fraction (TX-insoluble; right panels) were analyzed.SOD1_mono and SOD1_HMW represent monomeric and high molecular-weight (aggregated) forms of SOD1, respectively.Ub_mono and Ub_HMW represent monomeric ubiquitin and the polyubiquitinated proteins, respectively.LC-I and LC-II are cytosolic and lipidated forms of LC3, respectively.Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and b-tubulin served as controls for TX-soluble and TX-insoluble fractions, respectively.doi:10.1371/journal.pone.0009805.g004affect the formation of autophagosomes in cells.Further, a shortterm starvation in wild-type and Als2 2/2 fibroblasts resulted in decreased levels of p62 and LC3-II, indicating an enhancement of autophagy-dependent protein degradation in both cell types (Figure 8A).Notably, under such starved conditions, the LC3-II levels in Als2 2/2 cells were significantly higher than those in wildtype cells (Figure 8A), suggesting that a degree of autophagic clearance was affected by ALS2 loss.
To confirm this, a quantitative analysis of the LC3-II levels with a different inhibitor (pepstatin A) was performed.Wild-type fibroblasts showed a progressive decrease in the levels of LC3-II, which was partially restored by the pepstatin treatment.By contrast, Als2 2/2 cells exhibited the least effects after a similar starvation, in which the LC3-II levels in Als2 2/2 cells was significantly higher than those in wild-type Als2 2/2 cells, and stayed unchanged under the presence of pepstatin (Figure 8B and 8C), indicating an inefficient lysosomal clearance of LC3-II in Als2 2/2 cells.Double-immunostaining experiments also confirmed that loss of ALS2 led to sustained fluorescent signals of p62 and LC3 after starvation in fibroblasts (Figure 8D).Importantly, fibroblasts derived from ALS2 overexpressing mice (ALS2-tg line L6-2) (Figure S3) showed the opposite effects in which the clearance of LC3-II was accelerated (Figure S10).
We further performed a small interfering RNA (siRNA)mediated ALS2 knockdown in HeLa cells, and revealed that the suppression of ALS2 markedly increased a steady-state level of LC3-II (Figure S11A and S11B) with accompanying large perinuclear aggregates and/or vesicles containing LC3 (Figure S11D).Although the apparent level of p62 was unchanged, p62 was also re-distributed to the perinuclear LC3-positive compartments by suppressing the ALS2 expression (Figure S11C and S11D).Again, autophagic-flux analysis with the CQ treatment showed that ALS2 knockdown by itself did not induce autophagy in HeLa cells (Figure S11C).
Altogether, it is suggested that the accumulation of LC3-II observed in ALS2 deficient cells is due to a decreased level of clearance, rather than an increased formation, of autophagosomes and/or amphisomes.Thus, ALS2 might play an active role in the autophagosomal-endolysosomal trafficking in fibroblasts and HeLa cells.

ALS2 regulates the autophagosomal-endolysosomal trafficking in cultured spinal motor neurons
To determine whether ALS2 acts as a modulator for the endolysosomal system in neuronal cells, we investigated the changes in the level and distribution of p62 and LC3 in differentiated primary spinal motor neurons after nutrientstarvation.Since this cell type is very sensitive to lysosomal inhibitors (data not shown), we avoided the use of such inhibitors in this experiment.As in fibroblasts, diffused as well as punctated immunostainings for p62 and LC3 in soma were both decreased after a transient starvation in the wild-type neuronal cells (Figure 9A).By contrast, the same treatment induced the enlargement of LC3/p62-double positive puncta/vesicles in soma of Als2 2/2 cells (Figure 9A).Quantitative analysis revealed that the basal level of p62 in Als2 2/2 cells was higher than that of wildtype cells (Figure 9B and 9C).Further, the levels of p62 in Als2 2/2 cells was unaltered by starvation, while those in wild-type were significantly decreased (Figure 9C and S12).Collectively, loss of ALS2 in cultured spinal motor neurons compromises autophagosomal-endolysosomal trafficking.

SOD1 is partially colocalized with ALS2 and LC3 onto autophagosome-endosomal compartments in NSC-34 cells
Finally, to investigate the ALS2's contribution to the degradation of mutant SOD1, we analyzed the relative changes in the ectopically-expressed SOD1 H46R levels in cycloheximide-treated wild-type and Als2 2/2 fibroblasts by the treatment with either the UPS inhibitor (epoxomicin) or CQ according to the methods as described [37].No significant differences in the SOD1 H46R levels between wild-type and Als2 2/2 cells were detected under the experimental conditions used (data not shown), indicating that effects of ALS2 on either the UPS-or lysosome-dependent SOD1 H46R degradation should be marginal, or undetectable in short-term cell-culture experiments.
To investigate the relationship between ALS2/LC3-localizing autophagosome-endosomal compartments and neuronal SOD1 dynamics in more detail, we conducted a series of co-transfection experiments in the presence of either the UPS or lysosomal inhibitor using NSC-34 motor neuron-like hybrid cell lines.Ectopically expressed EGFP-LC3, FLAG-tagged SOD1, and ALS2 were diffusedly distributed throughout the cytosol with no colocalization in NSC-34 cells under normal cultured conditions (Figure 10A).The treatment with the UPS inhibitor (2 mM MG132, 3hr) induced a formation of small punctated cytoplasmic aggregates of SOD1 mutants, while the distribution of EGFP-LC3 and ALS2 were unchanged (data not shown).On the other hand, the CQ treatment resulted in an extensive enlargement of ALS2/ LC3-positve vesicular compartments in NSC-34 cells (Figure 10B).Intriguingly, ectopically expressed SOD1, namely SOD1 WT and SOD1 H46R , were frequently colocalized with and accumulated onto such enlarged vesicular compartments (Figure 10B, white arrows in enlarged images), supporting the notion that a portion of cytoplasmic SOD1 is indeed degraded through the autophagyendolysosomal system [37,38].

Discussion
ALS2/alsin is an activator for the small GTPase Rab5 [15], and involves not only in early endosome/macropinosome trafficking and fusion [16,17] but also in neuroprotection against MNDassociated pathological insults, such as toxicity induced by mutant SOD1 [25,46].However, molecular mechanisms underlying the relationship between ALS2-associated cellular function and its neuroprotective role remain unclear.In this study, we demonstrated that disturbance of endolysosomal trafficking by ALS2 loss exacerbated the SOD1 H46R -mediated neurotoxicity by accelerat-ing the accumulation of immature vesicles and insoluble proteins in the spinal cord.Thus, ALS2 might play a role in endolysosomal trafficking in vivo, accounting for the ALS2-assocaited neuroprotective function (Figure 11).
It has been reported that overexpression of ALS2 protects cultured motor neuronal cells from toxicity induced by SOD1 mutants; A4T, G85R, and G93R [25,46].This protection is dependent on the physical interaction between mutant SOD1 and ALS2 [25], and such ALS2-mediated neuroprotective function exerts through the activation of the Rac1/phosphatidylinositol 3kinase (PI3K)/Akt prosurvival pathway [46].However, we were unable to reconfirm the ALS2 interaction with either wild-type or SOD1 mutants; H46R, A4V, G93A, and G85R (data not shown), and the Rac1 activation by ALS2 [16].In addition, although we revealed a milder accumulation of wild-type as well as mutant SOD1 onto ALS2/LC3-double positive autophagosome-endosomal compartments in motor neuron-like cells under dysfunctional lysosomal conditions, ALS2 per se seems to play a limited role in the SOD1 degradation in cultured cells (data not shown).Nonetheless, a significant accumulation of insoluble mutant SOD1, polyubiquitinated proteins, p62, and LC3-II emerged in the spinal cord of ALS2 deficient SOD1 H46R mice at an early-or even pre-symptomatic stage.A large number of studies have suggested that overexpression of mutant SOD1 results in the increased levels of misfolded and aggregated proteins [41], which compromises the UPS for protein degradation [39], and initiates the ER-associated unfolded protein response (UPR) [47].Under such stress conditions, autophagy-lysosomal protein degradation appears to function as a safeguard for maintaining the cellular homeostasis [48,49].Thus, it is possible that the persistent and long-term excess production of SOD1 H46R , which exceeds the intracellular capacity for the clearance of misfolded proteins by the UPS and/or the autophagy-endolysosomal system, results in the accumulation of insoluble proteins in tissues, and that loss of ALS2 accelerates such phenotypes in vivo.
Intriguingly, we here showed that the proteasome activity was gradually induced rather than impaired in the spinal cord of SOD1 H46R -expressing mice as the disease progressed, and that loss of ALS2 further enhanced their activation.A recent study has demonstrated that Ub G76V -GFP reporter-expressing SOD1 G93A mice show only mild proteasome impairment in motor neurons at symptomatic stage but not before symptoms onset [39].These results suggest that the UPS impairment is not a primary cause for the accumulation of the insoluble proteins, while the possibility that the UPS is impaired by perturbing the delivery of ubiquitinated substrates to the proteasome without affecting the proteasome activity [50] should not be excluded.On the other hand, both p62 and LC3-II are localized onto autophagosomes and are preferentially degraded by the endolysosomal system [36,50,51].Thus, either the excess autophagosome formation or the overwhelmed endolysosomal system, or both of them, can account for the observed accumulation of the insoluble proteins such as p62 and LC3-II in SOD1 H46R -expressing mice.Since ALS2 appears to regulate endolysosomal protein degradation rather than autophagosome formation, we favor the hypothesis that loss of ALS2 augments the SOD1 H46R -mediated neurotoxicity, albeit its exact entity is still unclear, by slowing the fusion-mediated endolysosomal trafficking and/or autophagic clearance of misfolded protein aggregates (Figure 11).Contrary to our findings, a recent study has reported the enhanced endolysosomal degradation of glutamate receptors in Als2 2/2 cells [44].The reason for this discrepancy is currently unclear.It may be due to that the fates (degradation and/ or recycling) of membrane bound functional molecules such as receptors and misfolded/ubiquitinated non-functional protein aggregates are differently regulated within cells.Indeed, this notion is supported by the recent findings that loss of histone deacetylase 6 (HDAC6) shows seemingly opposite effects on endolysosomedependent protein degradation in cells; HDAC6 loss results in a failure of autophagosome maturation on the one hand [52], but also promotes EGF receptor degradation on the other [53].
Given the role of ALS2 on early endosomal and/or macropinosomal compartments as an activator for Rab5 [15,16], how does ALS2 regulate rather downstream endolysosomal pathway?It has been demonstrated that autophagosome maturation is essential for the degradation of abnormal proteins and organelle via the autophagy-lysosome pathway [54].This maturation is accomplished by the sequentially fusion of nascent autophagosomes with late endosomes and lysosomes under the control of many regulatory factors including endosomal sorting complex required for transport (ESCRT) and small GTPase Rab7 [54].Remarkably, it has recently been revealed that the fusion of autophagosomes with functional early endosomes and endosomal coatomer is also required for autophagy-dependent protein degradation [55].Since ALS2 is colocalized with LC3 and p62 on autophagosome/endosome hybrid membrane compartments called amphisomes [43], it is conceivable that ALS2 contributes to the much earlier phase of the autophagosome maturation, i.e. amphisome formation, via regulating the fusion between early endosomes and nascent autophagosomes (Figure 11).
An EM analysis revealed an extensive degeneration of spinal axon with accompanying the accumulation of granular/osmio- philic aggregates and autophagosome-like vesicles in early symptomatic SOD1 H46R mice, while spinal motor neurons were still preserved.This indicates that motor dysfunction seen in these mice is primarily associated with axonal degeneration rather than motor neuron loss.Importantly, loss of ALS2 appears to aggravate such pathological phenotypes in this mice model.In cultured neurons from wild-type mice, ALS2 is enriched to the membrane and vesicular compartments at the distal tip [17] as well as at the branching points (Otomo et al., unpublished) of neurites, and acts as sustenance in axonal development and function [17].Interestingly, it has been reported that Rab5 regulates an early sorting step preceding axonal transport, while Rab7, a regulator for the endosome/autophagosome-lysosome fusion, regulates long-rage retrograde axonal transport in motor neurons [56].Together with our findings, ALS2 may regulate the early step of endosome and/or autophagosome maturation through the activation of Rab5 within the axons, and loss of ALS2 results in an increased number of the immature vesicles, which precludes the normal long-range axonal vesicle trafficking by Rab7, leading to the accumulation of MVBs and autophagosome-like vesicles, axonal swelling, and degeneration in SOD1 H46R mice.
In this study, we demonstrated that loss of ALS2 resulted in a significant increase in the levels of astrocytic intermediate filaments, namely those of vimentin from a pre-symptomatic stage.It has been reported that vimentin is also expressed in motor neurons, and its expression is upregulated in a number of animal models for MNDs as the disease progresses [57].Thus, it is still possible that neuronal vimentin is responsible for such increments.However, our preliminary immunohistochemical analysis of SOD1 H46R and Als2 2/2 ;SOD1 H46R mice, in which the increased levels of vimentin were mainly observed in astrocytes rather than motor neurons in the spinal cord (Otomo et al., unpublished), did not fit this notion.As there is currently no supportive evidence that ALS2 is expressed in glial cells including astrocyte [15,19], loss of ALS2 may lead to insidious adverse effects on neuronal cells, which in turn initiates and accelerates astrogliosis through some signaling cross-talk between neurons and astrocytes.Interestingly, a recent study has shown that ALS2 depleted spinal motor neurons, but not cortical neurons, are rescued by co-cultured astrocytes, indicative of a cell-type specific neuron-glia crosstalk [19].Further investigation of such neuron-astrocyte interaction on the pathogenesis is warranted.One important question arising from this study is whether the effects of ALS2 loss are specific to H46R mutation in SOD1 or not.Previously, it has been reported that loss of ALS2 does not affect the pathological course of SOD1 G93A mice [29,30].Although the exact reasons for this discrepancy are unclear, pathological differences observed between SOD1 H46R and SOD1 G93A mice might be related.Despite that the expression levels of mutant SOD1; ,20-fold of endogenous SOD1, were comparative between SOD1 H46R and SOD1 G93A mice (data not shown), the pathology of the spinal cord was very different.In SOD1 G93A mice, a prominent distension of mitochondria forming vacuolar structures with accompanying much faster progression of disease symptoms was evident [58], while such pathological features were barely observed in SOD1 H46R mice.Rather, a widespread axonal degeneration with preserved motor neuron was noticeable in the spinal cord of SOD1 H46R mice.This suggests that molecular basis for the pathogenesis in each mutant SOD1-expressing model may not be the same.Recent findings that dynein mutants crossed with three different SOD1 mutant animals showed different outcomes [59] also support this notion.We are currently speculating that overlapping pathogenic processes between SOD1 H46R and Als2 null mice; i.e., the axonal degeneration in the spinal cord [6,21,31], are implicated in the enhancement of the additive adverse effects.By contrast, a devastating mitochondrial dysfunction in SOD1 G93A mice overwhelms the modest symptoms by the ALS2 deficiency.
In conclusions, loss of ALS2 impairs the maturation of autophagosomes, which causes the lowering of the autophagic flux, thereby accumulating immature vesicles and insoluble misfolded proteins under stress conditions.Thus, disturbance of the autophagosome-endolysosomal trafficking by ALS2 loss might be a causative of the manifestations of ALS2-linked MNDs in human.Notably, functional impairment of other endolysosomal-associated proteins such as ESCRT subunit and Rab7 also causes motor neuron dysfunction [60,61].Further characterizations will give us more clues to understanding physiological roles of autophagy-endolysosomal process in the pathogenesis for ALS and other MNDs.

Behavioral analysis
Motor coordination and balance was assessed by a balancebeam test using the fixed-stainless steel bar (45 cm long and 0.9 cm in diameter) at 12 weeks of age, and weekly thereafter until the day at which mice were unable to stay on the bar.Each mouse was given five trials, and the maximum durations (up to 60 sec) at which mice fall off from the bar were scored.To evaluate the spontaneous motor activities in mice, we conducted rearing and cage activity tests by using SUPERMEX with an infrared ray sensor monitor (Muromachi Kikai).Both rearing and cage activities were uninterruptedly monitored for 7 consecutive days starting at either 12 weeks or 18 weeks of age.The cumulative counts of rearing and cage activities for either a light-(12 hr; 7:00-19:00) or a dark-period (12 hr; 19:00-7:00) were analyzed.

Southern blot analysis
The probe DNA spanning the region between introns 3 and 5 for human SOD1 gene (product size; 1,955 bp) was prepared by PCR amplification using primer sets as follows; hSOD1_probe_L; 59-CCCCTGCTCCCAAATGCTGGAATGC-39, hSOD1_ probe_R; 59-GGGGCCTCAGACTACATCCAAGGG-39.Genomic DNA samples prepared from tail tissues were digested with FbaI, separated by electrophoresis, and blotted onto nylon membrane (Hybond-N+; Amersham Biosciences).The blot was hybridized with [a 32 P-dCTP]-labeled hSOD1_probe, detecting an ,3.3 kb restriction fragment of the transgene; SOD1 H46R .As a control, mouse Actb cDNA encoding b-actin was used as a probe.(not shown), thereby promoting the maturation (fusion and trafficking) of macropinosomes and early endosomes (EE).In the present study, we show that ALS2 is also present to the p62/LC3-positive autophagosomes in cells, suggesting that ALS2 plays a role not only in the maturation of macropinosomes and EE, but also of autophagosomes via their heterotypic fusions, generating amphisomes.The resulting amiphisomes further mature to late endosomes (LE) accompanying the recruitment of Rab7, and ultimately fuse with lysosomes, thereby cargo molecules including polyubiquitinated proteins are degraded.Poly-ubiquitinated proteins associated with p62 are also degraded by the proteasomes.(B) Loss of ALS2 results in a decrease in endosome fusion, thereby the maturation of autophagosomes and endolysosomal trafficking are disturbed.doi:10.1371/journal.pone.0009805.g011

Tissue sample preparation for western blot analysis
were centrifuged at 23,000g for 20 min at 4uC.The resultant supernatant was collected as a TX-soluble fraction.The insoluble pellet fraction was once washed with A buffer, and then suspended with Lysis buffer B [25 mM Tris-HCl; pH 7.5, 50 mM NaCl, 5% (w/v) sodium dodecyl sulfate (SDS)], sonicated, and left for 30 min at room temperature (RT).After the centrifugation at 23,000g for 20 min, the supernatant was collected as a TX-insoulble/SDSsoluble fraction.Fibroblasts and HeLa cells were harvested and lysed in Lysis buffer C [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.1% SDS, 1% TX, 0.5% sodium deoxycholate, and Complete Protease Inhibitor Cocktail (Roche)], and sonicated.Protein concentration of each fraction was determined by the Micro BCA system (Pierce).

Western blot analysis
Equal amount of protein (1-2 mg) from each fraction was subjected to SDS-polyacrylamide gel electrophoresis, and transferred onto a polyvinylidene difluoride membrane (Bio-Rad).The membranes were blocked with Blocking One (Nacalai Tesque) for 1 hr at RT, incubated with the primary antibody in TBST [20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.1% Tween 20] containing 5% Blocking One (Nacalai Tesque).After washing with TBST, membranes were incubated with HRP-conjugated secondary antibody.Signals were visualized by Immobilon TM Western (Millipore) and BioMax X-ray films (Kodak).The bands or signal intensities were quantified by analyzing the digitally-captured images using CS Analyzer ver3 (ATTO).

20S proteasome activity assay
Spinal cord tissues were homogenized in Lysis buffer D [50 mM HEPES (pH 7.5), 5 mM EDTA, 150 mM NaCl, 1% (w/v) TX], and were centrifuged at 23,000g for 20 min at 4uC.Protein concentration of the resultant supernatant was determined by the Micro BCA system (Pierce), and adjusted to 2 mg protein/ml with Lysis buffer D. A 10 ml of each sample (20 mg) was subjected to the proteasome activity assay using 20S Proteasome Activity Assay Kit (Chemicon) according to manufacturer's instructions.The lactacyctin (25 mM)-inhibitable chymotrypsin-like proteasome activity was measured by quantitating the fluorescent intensity for the fluorophore 7-amino-4-methylcoumarin (AMC) hydrolyzed from the substrate peptide; LLVY-AMC using a 360/460 nm filter set in a fluorometer (CytoFluor 4000; PerSeptive Biosystems).

Histological analysis
Mice were anesthetized with 4% halothane in a mixture of N 2 O/O 2 (70:30), and transcardially perfused with physiological saline containing 1,000U/ml heparin, followed by 2% paraformaldehyde (PFA)/2% glutaraldehyde (GA) in 0.1 M phosphate buffer (PB) (pH 7.3).Brain and spinal cord was removed and postfixed with the same fixative for 12 hr or 36 hr at 4uC and with 2% GA for 2 hr at 4uC, followed by washing with 0.1 M PB (pH 7.3).Lumbar segment was dissected out and post-fixed in 1% osmium tetroxide in 0.05 M PB (pH 7.4).After dehydration in graded alcohol, the tissues were embedded in the epoxy resin.Semi-thin sections (2-3 mm) of L5 lumbar cord were stained with 0.5% toluidine blue and examined under a computer-assisted light microscope (BZ-9000, KEYENCE).Selected areas of the spinal cord were cut into ultrathin sections, and stained with uranyl acetate and lead citrate for ultrastructural examination using electron microscopes (H-7100, Hitachi; JEOL1200EX, JEOL).

Immunohistochemical analysis
Anesthetized mice were transcardially perfused with 4% PFA in 0.1 M PB (pH 7.5).Brain and spinal cord were removed and postfixed for at least 48 hr in 4% PFA followed by paraffin embedding.For fluorescent immunohisochemistry, 6 mm paraffin embedded sections were cut on a microtome, and brain and spinal cord sections were incubated in phosphate buffered saline (PBS, pH 7.4) with 5% normal goat serum (NGS) and 0.1% TX for 1 hr at RT.For double-or triple-immunostaining, sections were incubated with primary antibodies in PBS containing 0.05% TX overnight at 4uC.Sections were incubated with secondary antibody for 3 hr at RT. Controls for all immunostainings were performed simultaneously by omitting the primary antibody.Sections were coverslipped using VectorShield (Vector Laboratories) with or without 49,6-diamidino-2-phenylindole dihydrochloride (DAPI) for nuclei counterstaining, and analyzed by a computer-assisted light microscope (BZ-9000, KEYENCE), or Leica TCS-NT system (Leica Microsystems) and processed by ImageJ 1.39u (NIH).All images presented are representative of at least n = 2-3 animals examined in each group at each time point.
Primary fibroblast cultures were established from E14-E15 mouse embryos.The skin tissues were isolated from pups, washed with ice-cold Hank's balanced salt solution devoid of calcium and magnesium ion [HBSS(-)] (Invitrogen) and treated with 0.5 ml of 0.25% trypsin-EDTA for 15 min at 37uC.Trypsin-EDTA was removed and washed several times with HBSS(-).Tissue samples were then treated with DNase I (final 50 mg/ml) in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 mg/ ml streptomycin at RT for 10 min.The dissociated cells were seeded onto a T75 flask at an appropriate cell density, and cultured in the same culture medium.
Primary hippocampal neuronal cultures were established from E18 embryos.In brief, tissues from each embryo were dissected out and immediately placed into 1 ml of ice-cold HBSS(-).After removing HBSS(-) by aspiration, 0.5 ml of 0.25% trypsin-EDTA was added and incubated for 15 min at 37uC.Trypsin-EDTA was removed and washed several times with 20% FBS/Neurobasal medium (Invitrogen).Tissue samples were treated with DNase I (final 50 mg/ml) in 20% FBS/Neurobasal medium for 10 min at RT.After the centrifugation at 150g for 15 s, the resulting tissue pellets were dissociated in a 0.6 ml of Neurobasal medium containing 20% FBS by pipeting using the fire polished Pasteur pipet.After counting the living cell numbers by the trypan blue assay, the cells were plated onto poly-D-lysine coated round glasses at a density of 10 5 cells/mm 2 (days in vivo 1; DIV1) for transfection in neuronal cell culture (NCC) media [Neurobasal medium containing 16B27 supplement (Invitrogen), 25 mg/ml insulin (Sigma), 0.5 mM L-glutamine, 50 mg/ml streptomycin, and 50 U/ml penicillin G], and cultured for 12 hr at 37uC.Medium was exchanged with the fresh one, and cultured for another 36 hr.Medium was further replaced with the fresh NCC medium containing cytosine b-D-arabinofuranoside hydrochloride (Ara-C; Sigma).
Primary spinal neuronal cultures were established from the lumbar spinal cord of E14 embryos.In brief, after removal of the dorsal root ganglia and meninges under microscopic observation, the ventral horns were dissected using microscalpels, transferred to a 1.5-ml tube containing 0.5 ml of HBSS(-)/0.025%(w/v) of trypsin, and incubated for 20 min at 37uC.After the trypsin treatment, 0.5 ml of HBSS(-) was added to the tube.The tissues were dissociated by gentle pipetting, and then centrifuged at 400g for 10 min at RT.The pellets were suspended in 0.5 ml of HBSS(-).The resulting suspension was centrifuged through a bovine serum albumin (BSA) cushion [3% (w/v) in HBSS(-)] at 700g for 10 min at 4uC.The cells obtained at this step represent mixed neuron/glia population.The cells were resuspended in Neurobasal medium containing physiological concentrations of Ca 2+ (1.8 mM) and Mg 2+ (0.8 mM) (Invitrogen) and centrifuged at 1,000g for 10 min at RT.The pellet was suspended in Neurobasal medium (,400 ml) containing 16B27, 0.5 mM L-glutamine, 2% FBS, 25 mM 2mercaptoethanol, 25 mM glutamate, 50 U/ml penicillin, 50 mg/ml streptomycin, and 10 ng/ml Brain-derived neurotrophic factor (BDNF), and seeded onto poly-D-lysine (Sigma) pre-coated glass cover slips in the wells of a 24-well plate at a density of 5,000 cells/ mm 2 .After 24 hr, the medium was replaced with 400 ml of fresh Neurobasal medium without glutamate, thereafter 100 ml of the same medium was added three times a week to the cultures for 2 weeks until the fixation.

Transfection
Transfection was performed by using Effectene Transfection Reagent (Qiagen), Lipofectoamine 2000 (Invitrogen), or Lipofectamine TM LTX Reagent (Invitrogen) according to the manufacturer's instructions.HeLa and primary neuronal cultures were transfected with plasmid DNAs as previously described [15,20].NSC-34 cells were seeded onto a 24 well-plate at a density of 2610 5 cells, cultured for 16 hr, and transfected with an appropriate amount of plasmid DNAs.After 24 hr of culture, cells were trypsinized and re-seeded onto a 11 mm round glass cover slip coated with 0.01% poly-D-lysine and 10 mg/ml laminine at a density of 5610 4 cells/well.Finally, cells were cultured in DMEM with 10% FBS for 18 hr in the presence or absence of CQ (50 mM) or MG132 (10 mM), or a solvent alone (dimethyl sulfoxide; DMSO) as a control.

Nutrient starvation
Fibroblasts and HeLa cells were seeded onto a 6 well-plate at a density of 2610 5 cells, and cultured for 24 hr.In case of primary spinal neurons, dissociated tissues were seeded onto a poly-Dlysine-coated cover slip and incubated for 14 days (DIV14) under appropriate conditions (see above).Medium was then exchanged with Earle's Balanced Salt Solution (EBSS) (Sigma) lacking FBS, and cells were cultured for another 6-8 hr either in the presence or absence of pepstatin A (20 mg/ml) or CQ (0.5 mM or 12.5 mM).

Small Interfering RNA-mediated knockdown of ALS2
To knock down the expression of the endogenous ALS2 gene in HeLa cells, we conducted the oligonucleotide-based RNA interference using small interfering RNA (siRNA).In brief, 0.4 ml siRNA (2.5 mM) [ALS2si-1 (SI00128226; QIAGEN), ALS2si-2 (SI00128219; QIAGEN), ALS2si-3 (SI00128205; QIA-GEN), ALS2si-4 (SI00128212; QIAGEN), and control (scrambled siRNA, cat#4611; Ambion)] and 0.8 ml RNAiMAX (Invitrogen) were mixed with 100 ml OPTI-MEM, and incubated for 20 min at RT. One-hundred ml of the resulting mixture was transferred onto a 24 well-plate, and mixed with 400 ml of medium containing 2.5610 4 dissociated HeLa cells with a final siRNA concentration of 1 nM.After 48 hr of culture, cells were harvested, lysed in Lysis buffer C, and sonicated.Equal amount of protein from each sample was subjected to western blot analysis.

Immunocytochemistry
Cells were washed with PBS(-) twice, fixed with 4% PFA in PBS(-) (pH 7.5) for 20 min, followed by permeabilization with 0.3% TX in PBS(-) for 20 min or with 100 mg/ml digitonin for 30 min at RT.The primary antibody, diluted in PBS(-) containing 1.5% normal goat serum and 50 mg/ml digitonin, was added to cells and incubated overnight at RT. Appropriate secondary antibodies were used for the detection of the signals of either the tag epitope or proteins of interest.Finally, images of optical sections with 0.025 mm thickness were captured and analyzed by Leica TCS_NT confocal-microscope systems (Leica Microsystems).Randomly selected images for spinal neurons were digitally-processed by ImageJ 1.39u (NIH) to demarcate the outline of the cell body, and fluorescent intensities for p62 signals within the demarcated area (corresponding to the cell body for the single cell) were measured using the fluorescence microscopy (Leica) with identical settings.

Figure 8 .
Figure 8. Loss of ALS2 results in decreased levels of the lysosome-dependent degradation of LC3 in fibroblasts.(A) Effect of ALS2 expression on the autophagic flux in fibroblasts.Fibroblasts derived from either wild-type (Als2 +/+ ) or Als2 2/2 mice were incubated in a starvation medium with or without 0.5 mM chloroquine (CQ) for 6 hr.Equal amount of protein from total lysates were analyzed by immunoblotting using antibodies as indicated.Alpha(a)-tubulin served as control.(B) Effect of ALS2 expression on the autophagic flux in fibroblasts.Fibroblasts derived from either wild-type (Als2 +/+ ) or Als2 2/2 mice were incubated in a starvation medium with or without 20 mg/ml pepstatin A (pepA) for indicated periods.Equal amount of protein from total lysates were analyzed by immunoblotting using antibodies as indicated.Alpha(a)-tubulin served as control.(C) Quantitative densitometry for the levels of LC3-II immunoreactive signals shown in B. Data were normalized by the levels of a-tubulin (LC3-II/a-tubulin).Values are mean6SEM (n = 4) in an arbitrary unit relative to control.Statistical significance is evaluated by ANOVA with Bonferroni's post hoc test (one-way, compared with respective controls; *p,0.05,***p,0.001,and two-way, compared between WT and Als2 2/2 ; +p,0.05).(D) Representative images for double immunostaining with LC3 (green) and p62 (red) in fibroblasts.Fibroblasts from wild-type (WT) and Als2 2/2 mice were either left unstarved (0 hr) (upper) or starved for 6 hr (lower).It is notable that a 6 hr of starvation leads to decreased levels of the LC3-and p62immunoreactive signals in WT cells, but not in Als2 2/2 cells.Scale bars = 50 mm.doi:10.1371/journal.pone.0009805.g008

Figure 9 .
Figure 9. Loss of ALS2 lowers the starvation-induced clearance of p62 in cultured motor neurons.(A) Representative confocal images for double immunostaining with LC3 (green) and p62 (red) in primary spinal motor neurons.The cells (DIV14) from wild-type (WT) and Als2 2/2 mice were either left unstarved (0 hr) (upper) or starved for 6 hr (lower).A 6 hr of starvation leads to decreased levels of the LC3-and p62-immunoreactive signals in WT cells.It is notable that a same treatment to Als2 2/2 cells results in the enlargement of LC3/p62-double positive puncta/vesicles in soma.Scale bars = 5 mm.(B) Representative fluorescent microscopic images used for the quantitative analysis of p62-immunostaining in primary spinal motor neurons shown in C. The cells (DIV14) derived from wild-type (WT) and Als2 2/2 mice were either left unstarved (0 hr) or starved for 6 hr.Scale bars = 10 mm.(C) Quantitation of the p62-immunoreactivity in randomly selected spinal neurons.Signal intensities for the p62-immunoreactivity relative to unit area of soma (pixel) (AU; arbitrary unit) are shown as Box-Wisker plots [wild-type (WT); 0 hr (n = 149) and 6 hr (n = 123), Als2 2/2 ; 0 hr (n = 127) and 6 hr (n = 139)].Statistical significance is evaluated by non-parametric ANOVA (Kruskal-Wallis) followed by Dunn's post hoc test.There is a significant difference in the basal levels of p62 immunoreactive intensities between WT and Als2 2/2 cells (***p,0.001).A 6 hr of starvation results in a significant decrease in the signal intensities in WT [compared 0 hr (white) with 6 hr (gray), ***p,0.001],but not in Als2 2/2 cells.doi:10.1371/journal.pone.0009805.g009

Figure
Figure S1 Copy numbers of the transgene in human SOD1 H46R transgenic mice on different Als2 genotypes used were comparable.(Upper panel) Image for the ethidium bromide-stained mouse genomic DNA digested with FbaI.Mice with five different genotypes [Als2 +/+ ;SOD1 H46R (blue), Als2 +/2 ;SOD1 H46R (green), Als2 2/2 ;SOD1 H46R (red), Als2 2/2 (orange), and wild-type (WT) (black), (each n = 4)] were analyzed.Equal amount of genomic DNA (2 mg) was loaded and separated by agarose gel electrophoresis.The positions of size-markers are shown on the left.(Middle panel) Southern blot analysis of the mouse genomic DNA.The FbaI blot was probed with the radiolabeled human genomic DNA fragment of the SOD1 gene.A 3.3 kbp of the restriction fragment originating from human SOD1 H46R transgene was specifically detected.(Lower panel) As a control, the FbaI blot was re-probed with the radio-labeled mouse Actb (b-actin) cDNA, detecting two fragments originating from mouse endogenous Actb gene (2.0 and 1.4 kbp).