Early upregulation of cytosolic phospholipase A2α in motor neurons is induced by misfolded SOD1 in a mouse model of amyotrophic lateral sclerosis

Background Amyotrophic lateral sclerosis (ALS) is a fatal multifactorial neurodegenerative disease characterized by the selective death of motor neurons. Cytosolic phospholipase A2 alpha (cPLA2α) upregulation and activation in the spinal cord of ALS patients has been reported. We have previously shown that cPLA2α upregulation in the spinal cord of mutant SOD1 transgenic mice (SOD1G93A) was detected long before the development of the disease, and inhibition of cPLA2α upregulation delayed the disease’s onset. The aim of the present study was to determine the mechanism for cPLA2α upregulation. Methods Immunofluorescence analysis and western blot analysis of misfolded SOD1, cPLA2α and inflammatory markers were performed in the spinal cord sections of SOD1G93A transgenic mice and in primary motor neurons. Over expression of mutant SOD1 was performed by induction or transfection in primary motor neurons and in differentiated NSC34 motor neuron like cells. Results Misfolded SOD1 was detected in the spinal cord of 3 weeks old mutant SOD1G93A mice before cPLA2α upregulation. Elevated expression of both misfolded SOD1 and cPLA2α was specifically detected in the motor neurons at 6 weeks with a high correlation between them. Elevated TNFα levels were detected in the spinal cord lysates of 6 weeks old mutant SOD1G93A mice. Elevated TNFα was specifically detected in the motor neurons and its expression was highly correlated with cPLA2α expression at 6 weeks. Induction of mutant SOD1 in primary motor neurons induced cPLA2α and TNFα upregulation. Over expression of mutant SOD1 in NSC34 cells caused cPLA2α upregulation which was prevented by antibodies against TNFα. The addition of TNFα to NSC34 cells caused cPLA2α upregulation in a dose dependent manner. Conclusions Motor neurons expressing elevated cPLA2α and TNFα are in an inflammatory state as early as at 6 weeks old mutant SOD1G93A mice long before the development of the disease. Accumulated misfolded SOD1 in the motor neurons induced cPLA2α upregulation via induction of TNFα. Supplementary Information The online version contains supplementary material available at 10.1186/s12974-021-02326-5.


Background
Amyotrophic lateral sclerosis (ALS) is a severe degenerative disorder, mainly affecting the motor neurons. Most of the cases (about 90%) are sporadic. Familial cases have been linked to mutations in several genes, including chromosome 9 open reading frame 72 (C9ORF72) repeat expansions, Cu/Zn superoxide dismutase (SOD1), TAR DNA binding protein (TDP-43) and others [1]. Mutant SOD1 is the best characterized form of familial ALS, accounting for 20% of familial cases [2]. It is generally believed that sporadic and familial ALS may share pathological mechanisms. The pathophysiology of the multifactorial-multisystemic ALS disease includes various mechanisms. Although ALS is not primarily considered an inflammatory or immune-mediated disease, inflammation appears to play a role in the pathogenesis of the disease in both ALS patients and animal models, inflammatory responses have been observed [3][4][5]. Microglia [6] and astrocytes [7] are activated during the progression of the disease, and evidence suggests that they contribute to neuronal death.
Previous findings suggested that cytosolic phospholipase A 2 α (cPLA 2 α) is regarded as an essential source of inflammation. cPLA 2 α specifically hydrolyzes phospholipids containing arachidonic acid at the sn-2 position [8,9] and is the rate-limiting step in the generating eicosanoids and a platelet activating factor. These lipid mediators play critical roles in the initiation and modulation of inflammation and oxidative stress. cPLA 2 α is ubiquitous in all cells and is essential for their physiological regulation. However, elevated cPLA 2 α expression and activity were detected in the inflammatory sites in a vast array of inflammatory diseases, including neurodegenerative diseases [10][11][12]. Increased expression and activity of cPLA 2 α has been detected in neurons, astrocytes and microglia in the spinal cord, brainstem and cortex of sporadic ALS patients [13] and in the spinal cord of mutant SOD1 G93A transgenic mice [14], suggesting that cPLA 2 α may have an important role in the pathogenesis of the disease in all ALS patients. Our previous study [15] demonstrated that cPLA 2 α is upregulated in the spinal cord of 6 weeks old SOD1 G93A mice long before the appearance of the disease symptoms, neuronal death or gliosis, and remained elevated during the whole life span of the mice. Prevention of cPLA 2 α upregulation shortly before the onset of the disease symptoms, significantly delayed the loss of motor neuronal function, suggesting that cPLA 2 α upregulation in the spinal cord plays a role in the disease pathology. The mechanism that induces cPLA 2 α elevation in the spinal cord of as early as 6 weeks old mice, is not yet clear. SOD1 insoluble protein complexes (IPCs) were detected in motor neurons of 30 days old SOD1 G93A mice [16], before the manifestation of ALS pathology, and several months before the appearance of inclusion bodies. The present study aims to determine whether accumulated misfolded SOD1 triggers cPLA 2 α upregulation in motor neurons in the spinal cord of 6 weeks old ALS mice.

Animals
B6.Cg-Tg(SOD1G93A)1Gur/J hemizygous transgenic male mice were obtained from Jackson Laboratory (Bar Harbor, ME, U.S.A). The hemizygous transgenic male mice were also obtained by mating hemizygous transgenic males with C57BL/6J females (Jackson Laboratory). Each litter would generate hemizygous SOD1 G93A transgenic mice and littermate wild type controls as done before [15]. Transgenic male offspring were genotyped by PCR assay of DNA obtained from tail tissue (according to Jackson Laboratory). The study included male mice to avoid the estrogen effect. The study was approved by Ben-Gurion University Institutional Animal Care and Use Committee (IL-40-07-2016) and was conducted according to the Israeli Animal Welfare Act following the Guide for Care and Use of Laboratory Animal (National Research Council, 1996).
Motor function measurement by Rotarod A Rotarod test was used to evaluate the motor performance of the mice using an accelerating paradigm of 0.12 rpm/s as described before [15]. After a learning period of several days, mice were able to stay on the Rotarod (Rotamex-5, Columbus instruments, Columbus, OH, USA) for up to 150 s. Each mouse was given 3 trials and the best performance was used as a measure for motor function ability. Mice were tested twice a week from age of the 7 weeksold until they could no longer perform the task.
Spinal cord tissue preparation Mice were deeply anesthetized and transcardially perfused with 20 ml of PBS [17].
For immunoblot analysis or immunoprecipitation Spinal cords were harvested in Lysis buffer containing 20 mM Tris pH7.5, 150 mM NaCl, 0.5% Sodium deoxycholate, 0.1% SDS, 0.1% Triton, 1 mM Phenylmethylsulfonyl fluoride and 1% protease inhibitors (Roche, Mannheim, Germany). The suspensions were sonicated 3 times for 20 s with Microsom Heatsystem Sonicator and centrifugated at 13,000×g for 20 min at 4 °C. Immunoprecipitation of cPLA 2 α or misfolded SOD1 was performed as described earlier [18,19] [15] in paraformaldehyde 4%/PBS solution overnight at 4 °C. The spinal cords were then transferred to PBS containing 30% sucrose for 24 h and then embedded in a 1:2 mixture of 30% sucrose in PBS:Tissue-Tek OCT (VWR, Radnor, PA), frozen in liquid nitrogen and stored at − 80 °C. Sections were made by cryostat (Leica Biosystems, Vienna, Austria) at 12 µm thickness, washed in PBS/tween 0.05%, incubated in PBS/Glycine 0.1% for 5 min and incubated in blocking solution (3% normal donkey serum and 2% BSA) at room temperature for 1 h. Then, these sections were incubated with primary antibodies diluted in blocking solution overnight at 4 °C. The primary antibodies used in the study were: 1:100 rabbit anti-cPLA 2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-pcPLA 2 α (Cell Signaling Danvers, MA USA), 1:100 mouse anti-misfolded human SOD1 (Medimabs, Quebec, Canada), 1:100 mouse anti-TNFα (Novus Biological USA, CO, USA), 1:100 goat anti-choline Acetyl-transferase (ChAT) (Millipore, CA, USA), 1:1000 rabbit anti-Iba-1 (Wako Pure Chemical Industries, Osaka, Japan), 1:500 rabbit anti-GFAP (Dako Glostrup Denmark), 1:100 mouse anti-GFAP (Millipore Darmstadt, Germany). Sections were washed with PBS/tween 0.05%, and incubated with 1:200 Cy3 or 1:100 anti-mouse Alexa488 or anti-rabbit Dylight conjugated secondary antibodies (Jackson Immunoresearch Laboratories, West Grove, PA, USA) for 1 h at room temperature. The staining of samples from the different treatments was performed in parallel. For each treatment, a negative control was prepared by omitting the primary antibody. Sections were mounted with anti-fading mounting medium (Electron Microscopy Sciences (EMS), Hatfield, PA, USA) and photographed in a blinded fashion using a fluorescent microscope (Olympus, BX60, Hamburg, Germany) or with confocal microscopy (Olympus, FluoView 1000, Tokyo, Japan). Using a confocal microscope, Z-sections were taken at 0.5 μm intervals and the results present Z-stack images. Fluorescence intensity was determined for cPLA 2 α using CellProfiler program. The % of fluorescence intensity of cell area was determined for the different cell types using CellProfiler program. LSM880 inverted laser-scanning confocal microscope (Zena, Germany) equipped with an Airyscan high-resolution detection unit and under identical acquisition conditions was also used. A Plain-Apochromat 63x/1.4 Oil DIC M27 objective was used, and parameters were set to avoid pixel intensity saturation and to ensure Nyquist sampling in the XY plane. Excitation of lasers for DAPI, Alexa 488 and Cy3 were 405 nm, 488 nm and 561 nm, respectively.

Statistical analysis
Data were expressed as mean ± standard error of the mean (SEM). Statistical significance was determined by either one-or two-way analysis of variance (ANOVA) followed by a posteriori Bonferroni's test for multiple comparisons provided by GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA, USA). Pearson coefficient correlation (r) was used to study the relationships between the variables.

Results
In our previous study we reported that cPLA 2 α is elevated in the spinal cord of 6 weeks old mutant SOD1 G93A mice but not at 3 weeks. To study whether cPLA 2 α is affected by the accumulation of misfolded SOD1 in the cells, cPLA 2 α and misfolded SOD1 proteins expression and accumulation were analyzed in the spinal cord of SOD1 G93A mice. Immunofluorescence staining and quantitation showed a significant (p < 0.001) elevation of cPLA 2 α protein expression in the spinal cord sections ( Fig. 1A) of 6 weeks old SOD1 G93A mice, as shown in our previous study [15]. Immunofluorescence staining and quantitation of misfolded SOD1 showed that it was significantly (p < 0.001) detected in the spinal cord at 3 weeks old SOD1 G93A mice, before the elevation of cPLA 2 α. The expression of cPLA 2 α and mutant SOD1 G93A was also determined by western blot analysis and showed that mutant SOD1 G93A was detected at 3 weeks preceding the elevation of cPLA 2 α (Fig. 1B). Moreover, misfolded SOD1 determined by immunoprecipitation with anti B8H10 was detected at 3 weeks in the spinal cord of SOD1 G93A mice and gradually increased at a later age (Fig. 1C). Activation of cPLA 2 α analyzed by immunostaining of phosphor-cPLA 2 α was detected at the spinal cord section of 6 weeks old SOD1 G93A mice but not at 3 weeks (Fig. 1D).
The elevated accumulation of misfolded SOD1 and the elevated cPLA 2 α protein expression in the spinal cord sections of 6 weeks old mutant SOD1 G93A mice were detected specifically in motor neurons as determined by co-immunofluorescence staining of misfolded SOD1 and the marker of motor neurons ChAT ( Fig. 2A). Using specific antibodies against misfolded SOD1 showed that elevated misfolded SOD1 was already detected at 3 weeks (Additional file 1: Fig. S1), and co-staining with ChAT showed that it is expressed in the motor neurons (Additional file 2: Fig. S2). Co-immunofluorescence staining of misfolded SOD1 and Iba1 or GFAP, the markers of microglia or astrocytes, respectively, showed that misfolded SOD1 is not accumulated in these cells at 6 weeks ( Fig. 2A). As shown by co-immunofluorescence staining, elevated cPLA 2 α expression was also detected specifically in the motor neurons and not in microglia or astrocytes (Fig. 2B). Co-immunostaining of cPLA 2 α and misfolded SOD1 showed co-localization and overlapping between cPLA 2 α and misfolded SOD1 in the motor neurons in the spinal cord of 6 weeks old mutant SOD1 G93A mice (Fig. 2C). cPLA 2 α upregulation showed some variation that highly correlated with the accumulation of misfolded SOD1 with a correlation coefficient of 0.92 between both proteins, suggesting that the level of misfolded SOD1 in the motor neurons defines the level of cPLA 2 α protein expression (Fig. 2D). To determine whether the accumulation of misfolded SOD1 in the motor neurons is responsible for cPLA 2 α upregulation, human SOD1 WT or mutant SOD1 G93A were expressed in primary motor neurons isolated from the spinal cord from C57BL/6J mouse embryos as shown in the western blot analysis (Fig. 3A). The expression of human SOD1 did not affect the morphology of the motor neurons and their number (Fig. 3B). Double staining with anti -B8H10 and anti-cPLA 2 α showed elevated cPLA 2 α protein expression in motor neurons expressing mutant SOD1 G93A and accumulating misfolded SOD1 (Fig. 3C).
In motor neurons that did not accumulate misfolded SOD1 (as shown in Fig. 3A), cPLA 2 α expression did not change and was similar to that detected in control motor neurons (without induction). These results clearly indicate that misfolded SOD1 induced cPLA 2 α upregulation.
To study whether the elevated cPLA 2 α protein expression is triggered or stabilized by an interaction between misfolded SOD1 and cPLA 2 α, the binding between both proteins was determined by co-immunoprecipitation experiments. As shown in Fig. 4A, immunoprecipitation of cPLA 2 α in the spinal cord lysates at disease onset (13 weeks), symptomatic stage (18 weeks) or end stage (Fig. 4B) resulted in a significant co-immunoprecipitation of mutant SOD1 G93A (Fig. 4A), suggesting a binding between them. In contrast, there was no co-immunoprecipitation of mutant SOD1 G93A and cPLA 2 α in the spinal cord lysates of 6 weeks old mutant SOD1 G93A mice, suggesting that there is no binding between the proteins in that early stage. We then used an Airyscan detector, a sub-diffraction high-resolution laser-scanning confocal microscope to examine the binding between both proteins. The Airyscan detector used in the current study provides improved lateral resolution (~ 150 nm) and signal to noise ratio, as compared with conventional confocal microscopes. Under these conditions, accurate and straight forward analysis of the interaction between misfolded SOD1 and cPLA 2 α in the motor neurons in the spinal cord section was allowed. Airyscan high-resolution detection showed that there is only partial overlapping indicating partial binding between both proteins at 6 weeks (Fig. 4C, D).
Proinflammatory cytokines such as TNFα [23,24] are shown to induce elevation of cPLA 2 α protein expression, thus, we examined whether increased levels of TNFα could be detected in the spinal cord of 6 weeks old mutant SOD1 G93A mice. As shown in Fig. 5A, there is a significant (p < 0.001) elevation of TNFα in the spinal cord lysates of 6 weeks old mutant SOD1 G93A mice in comparison with spinal cord lysates of WT or 3 weeks old SOD1 G93A mice (130.0 ± 3.0 pg/ml compared with 85.5 ± 13.3 and 73.6 ± 6.7 pg/ml, respectively). To determine which type of cell produces TNFα, co-immunofluorescence staining of TNFα using anti-TNFα antibodies that show specific staining (Additional file 3: Fig. S3) and the different cell markers was performed in the spinal cord sections of 6 weeks old SOD1 G93A mice. Co-immunofluorescence staining of TNFα and ChAT clearly showed that TNFα was detected in the motor neurons in the spinal cords of 6 weeks old mutant SOD1 G93A mice (Fig. 5B). Immunofluorescence staining of the motor neurons in the spinal cord of 6 weeks old SOD1 G93A mice showed elevation of both TNFα receptors expression in comparison to WT mice (Additional file 4: Fig. S4). Coimmunofluorescence staining of TNFα and either Iba1 or GFAP showed that TNFα was not detected in the microglia or astrocytes at 6 weeks (Fig. 5B). A time course of co-immunofluorescence staining of TNFα and the different cell markers in the spinal cords of SOD1 G93A mice clearly shows elevated TNFα levels in the motor neurons at 6 weeks which was gradually increased at a later stage (Fig. 5C). TNFα was not detected in glia cells at 6 weeks, but was detected in microglia at 15 weeks and in astrocytes at 18 weeks (Fig. 5C). In accordance with these results, immunofluorescence staining of Iba1 or GFAP, to determine glia activation, showed no significant activation in the spinal cord section of mutant SOD1 G93A mice at these early stages (3 and 6 weeks), although misfolded SOD1 was already accumulated in the spinal cord, but did show a significant activation in the spinal cord sections of 17 weeks old SOD1 G93A mice (Additional file 5: Fig. S5) as reported in our previous study [15]. Coimmunofluorescence staining and densitometry analysis of cPLA 2 α and TNFα in the spinal cord sections of 6 weeks old SOD1 G93A mice showed as above that there is a variation of elevated cPLA 2 α protein expression in the different mice (Fig. 6A), which was highly correlated with TNFα (coefficient correlation, r = 0.81) in the motor neurons (Fig. 6B), suggesting that TNFα produced in the motor neurons is responsible for cPLA 2 α upregulation. We next studied the effect of misfolded SOD1 accumulation on TNFα expression using motor neurons that were viral induced with human SOD1 WT or mutant SOD1 G93A (described in Fig. 3). As shown in Fig. 6C, double staining of cPLA 2 α and TNFα showed that motor neurons with accumulated mutant SOD1 G93A (Fig. 3C) expressed elevated levels of both cPLA 2 α and TNFα. Over expression of SOD1 WT did not affect cPLA 2 α and TNFα compared with the control cells. The comparison between motor neurons expressing SOD1 WT and those expressing the mutant SOD1 G93A clearly shows that of the accumulation of misfolded SOD1 induces the elevation of cPLA 2 α and TNFα (Additional file 6: Fig. S6). To determine whether elevated TNFα is responsible for cPLA 2 α upregulation, human SOD1 WT or SOD1 G93A were expressed in the motor neurons like NSC34 cells. NSC34 is an hybrid cell line produced by the fusion of motor neurons from the spinal cords of mouse embryos with mouse neuroblastoma cells N18TG2 that exhibit properties of motor neurons after differentiation [21] as presented by shape change (Fig. 7A). As clearly shown in the western blot analysis (Fig. 7B), accumulation of mutant SOD1 caused significant elevation of cPLA 2 α, while accumulation of SOD1 WT did not affect on cPLA 2 α expression in NSC34 cells compared with the control (untransfected cells).
The presence of neutralizing TNFα antibodies (50 ng/ml) prevented cPLA 2 α upregulation, indicating that TNFα is responsible for cPLA 2 α upregulation by an autocrine mechanism. In accord with these results, although very low levels of TNFα were detected in the supernatant of the cells, they were significantly higher in supernatant of cells transfected with human SOD1 G93A compared with cells transfected with SOD1 WT or control cells (Additional file 7: Fig. S7). The differentiated neuron motors NSC34 cells express both TNF receptors as shown by immunofluoresnce staining (Fig. 7C). The addition of TNFα to differentiated NSC34 cells caused cPLA 2 α upregulation in a dose dependent manner, as shown by western blot analysis (Fig. 7D) and immunofluorescence staining cPLA 2 α (Fig. 7E).

Discussion
The present study clearly demonstrates that misfolded SOD1 is significantly detected in the spinal cords of 3 weeks old mutant SOD1 G93A mice preceding the elevated expression of cPLA 2 α and its activation at 6 weeks. The elevation of cPLA 2 α in the spinal cords at 6 weeks, was detected specifically in motor neurons and not in microglia or astrocytes. While, at the symptomatic stage, elevated cPLA 2 α was also detected in the glia cells in accordance with our and others previous studies [14,15]. Similar to our results, activation and elevation of cPLA 2 α protein expression were mainly detected in motor neurons in other pathological conditions such as after spinal cord injury and in spinal inflammatory hyperalgesia [25][26][27][28][29]. cPLA 2 α is a major inflammatory enzyme-producing arachidonic acid, a substrate for the formation of eicosanoids and a platelet-activating factor which are well-known mediators of inflammation and tissue damage implicated in pathological states of several acute and chronic neurological disorders [26,[30][31][32]. The detection of elevated and activated cPLA 2 α in the motor neurons at 6 weeks old mutant SOD1 G93A mice and long before any neuronal damage or sign of the disease is evident, indicates an inflammatory state of the motor neurons at this very early stage. Misfolded SOD1 accumulation in the spinal cord of 3 weeks old SOD1 G93A mice was detected in motor neurons but not in astrocytes or microglia. The significant accumulation of misfolded SOD1 in the motor neurons before the appearance of elevated cPLA 2 α expression and the significant correlation (r = 0.92) between both proteins at 6 weeks shown in the present study raised the possibility that the accumulation of misfolded SOD1 in the cells dictates the expression of cPLA 2 α. Indeed, the accumulation of misfolded SOD1 (determined by B8H10 staining) in primary motor neurons isolated from the mouse spinal cord or expression of mutant SOD1 in NSC34 motor neuron-like cells caused a significant elevation of cPLA 2 α protein expression. In contrast, expression of human SOD1 WT did not affect cPLA 2 α expression, indicating that intracellular misfolded SOD1 induced cPLA 2 α upregulation. Although the role of glial cells in neuronal damage and disease progression is well established [33], we show here that the elevation of cPLA 2 α in the motor neurons at 6 weeks is independent of glia cells and occurs long before any neuronal damage. In accordance with our results, the effect of accumulated mutant SOD1 on cPLA 2 α in NSC34 cells was reported recently [34]. They showed that expression of SOD1 G93A in motor neuron cell line NSC34 for long time induced cell death mediated by cPLA 2 α. The elevated levels of cPLA 2 α in familial and sporadic ALS [13] together with the observations that inclusions containing misfolded SOD1 are regularly present in motor neurons of ALS patients, both with and without SOD1 mutations [35], support the notion that misfolded SOD1 accumulated in the motor neurons contributes to the elevated cPLA 2 α expression. Induction or stabilization of proteins by other proteins within the cells via an interaction mechanism such as induction of P53 by elevated amyloid beta [36] and stabilization of EGFR or Kit C by HIP1 binding have been reported [37,38]. In addition, binding and co-precipitation of oligomeric or misfolded SOD1 with other proteins including voltage dependent anion channel (VDAC1) [39,40], macrophage migration inhibitory factor (MIF) [22,41], heat shock protein [42], glutathione peroxidase 1 [43] or Bcl-2 [44] have been documented. Taken together, these observations and the results demonstrating high overlapping of misfolded SOD1 and cPLA 2 α in the motor neurons in spinal cord sections by confocal microscopy could suggest that cPLA 2 α upregulation is induced by its interaction with misfolded SOD1 at 6 weeks old mutant SOD1 G93A mice. Co-immunoprecipitation of both proteins was detected only at the symptomatic stage but not at 6 weeks, indicating no direct interaction at this stage. Using the Airyscan detector, a sub-diffraction high-resolution laser-scanning confocal microscope [45,46], we showed that only partial binding between both proteins in motor neurons at 6 weeks, explains the absence of coimmunoprecipitation at this stage, and questioning the possibility that the interaction between misfolded SOD1 and cPLA 2 α induced the elevation of cPLA 2 α expression. cPLA 2 α was shown to be induced by different proinflammatory mediators and insults through specific receptors or scavenger receptors [23,[47][48][49][50]. Ours and other studies reported that TNFα induced cPLA 2 α upregulation in various systems [23,51,52] and in motor neuronlike NSC34 cells, as demonstrated in the present study. In the neural environment, constitutive physiological levels of TNFα regulate synaptic plasticity, modulates dendritic maturation, pruning, and synaptic connectivity to respond to alterations in sensory stimuli to maintain homeostatic plasticity [53,54]. Overexpression of TNFα has been associated with neuronal excitotoxicity, synapse loss, and propagation of the inflammatory state [55]. TNFα elicits its wide range of biological responses by activating two distinct receptors, TNF-R1 and TNF-R2 [56][57][58]. Antigenic TNFα and its soluble receptors measured by ELISA were significantly higher in ALS patients than in healthy controls [59]. To our knowledge, the present study is the first to show a significant elevation of TNFα protein in the spinal cord of mutant SOD1 G93A mice as early as at 6 weeks, that was about 150% of the levels of TNFα in the spinal cord of WT mice and of 3 weeks old mutant SOD1 G93A mice. Our results are supported by other studies reporting that TNFα was detected in the spinal cord of late pre-symptomatic stage ALS mice. TNFα was the sole cytokine whose mRNA could be observed in the spinal cord of young pre-symptomatic SOD1 G93A mice [60]. A microarray survey of 1081 gene products expressed in spinal cords of SOD1 G93A mice reported that TNFα was the only inflammatory cytokine found to be differentially expressed [4]. Upregulation of TNFα and its proapoptotic receptors mRNA were detected at late pre-symptomatic stages and preceded transcriptional upregulation of other proinflammatory gene products and temporally correlates with the progression of the disease in SOD1 G93A mice [61,62]. TNFα was reported to be elevated in the spinal cord of SOD1 G93A transgenic mice in the early life span, at 80 days [63]. We also show here for the first time that elevated TNFα is expressed specifically in spinal motor neurons of 6 weeks old SOD1 G93A mice but not in microglia or astrocytes, although glia cells are reported to be the major cell type to secrete TNFα [64]. In agreement with our results, immunohistochemical analysis showed little TNFα immunoreactivity in motor neurons from 60 days old SOD1 G93A transgenic mice with a healthy appearance [65] and FasL as early as day 40 [65]. Since FasL is upregulated by TNFα [66], it is possible that due to the methodology sensitivity TNFα was not detected in the spinal cord of 40 days old SOD1 G93A mice but at 60 days in their study [65]. We show here a high correlation (r = 0.81) between cPLA 2 α and TNFα expressed in the spinal motor neurons of 6 weeks old SOD1 G93A mice, suggesting that misfolded SOD1 induced the elevation of cPLA 2 α via production of TNFα. Indeed, as we show in the present study, that expression of mutant SOD1 but not SOD1 WT in motor neurons induced both cPLA 2 α and TNFα upregulation. Moreover, the presence of neutralizing TNFα antibodies prevented the elevation of cPLA 2 α expression NSC34 like motor neurons, indicating that TNFα is responsible for cPLA 2 α upregulation, acting via its autocrine effect. Likewise, addition of TNFα (10-100 pg/ml) to differentiated NSC34 cells for 24 h caused cPLA 2 α upregulation in a dose dependent manner similar to the concentration detected in the spinal cord of 6 weeks of SOD1 G93A mice. In agreement with our report, addition of soluble TNFα (acting through a reverse signaling) for 6 days affected motor neurons, inducing a marked motor neuron loss in SOD1-G93A monocultures [33]. Since elevated TNFα receptors were detected in motor neurons in the spinal cord of 6 weeks old SOD1 G93A mice, in accordance with others that reported elevated TNFR in the pre-symptomatic stage [67], the elevated TNFα in the spinal cord at this time point probably acts through its receptors to induce cPLA 2 α upregulation.
Our results, suggesting that motor neurons have a crucial role in inflammatory state (demonstrating elevated both cPLA 2 α and TNFα) during the early stage of the disease, are in accordance with the specific activation of motor neurons but not glia cells in the pre-symptomatic stage (at 8 weeks) of the disease in mutant SOD1 G93A mice evident by the increased p38MAPK [67,68], activation of ASK1, MKK3,4,6, overexpression of both TNFα receptors (TNFR1 and TNFR2) [67] and TNFα accumulation in transgenic motor neurons [33]. In addition, motor neurons were reported as a primary determinant of disease onset and early disease progression by selective mutant gene inactivation within the cells [69]. Moreover, it was shown [70] that neuronal expression of mutant SOD1 was sufficient to cause motor neuron degeneration and paralysis in transgenic mice with cytosolic dendritic ubiquitinated SOD1 aggregates as the dominant pathological feature. Crossing neuron-specific mutant SOD1 mice with ubiquitously wild-type SOD1-expressing mice led to dramatic wild-type SOD1 aggregation in oligodendroglia after the onset of neuronal degeneration suggesting that mutant SOD1 in neurons triggers neuronal degeneration, which in turn may facilitate aggregates formation in surrounding glial cells. In contrast, cell-specific deletion of mutant SOD1 in genetically altered mice has implicated microglia and astrocytes as contributors to the late disease progression but not the onset of disease [71][72][73].

Conclusions
We show here that elevated protein expression of both cPLA 2 α and TNFα were detected specifically in motor neurons and not in glial cells in the spinal cord of 6 weeks old SOD1 G93A mice, indicating the inflammatory state of the motor neurons long before the development of signs (See figure on next page.) Fig. 6 Elevated cPLA 2 α protein expression in the motor neurons is highly correlated with TNFα. A Double immunofluorescence staining of cPLA 2 α (green) and TNFα (red) proteins in the lumbar spinal cord sections of WT and 6 weeks old mutant SOD1 G93A mice. Scale bars = 100 μm. B The Pearson coefficient correlation (r) between cPLA 2 α and TNFα in the spinal motor neurons of mutant SOD1 G93A mice was analyzed. Florescence intensity is expressed in arbitrary units of immunostaining as presented in the representative results in A. Four fields in each of the 8 different mice analyzed. C Elevated cPLA 2 α and TNFα in primary motor neurons expressing mutant SOD1 G93A . Double immunofluorescence staining of cPLA 2 α (green) and TNFα (red) in primary motor neurons expressing human SOD1 WT , mutant SOD1 G93A and control cells described in Fig. 3. Two upper panels, scale bars = 50 μm and two lower panels, scale bar s = 20 μm. 3 different independent experiments were analyzed and showed similar results. The means ± SEM fluorescence intensity for cPLA 2 α and TNFα is presented in the bar graphs as arbitrary units. Five fields in each of the 3 different treatments of motor neurons in each experiment was analyzed. Significance compared to control ***p < 0.001, n.s. non-significant ). cPLA 2 α protein expression was determined by dividing the intensity of each cPLA 2 α with the intensity of the corresponding human SOD1 or mouse SOD1 after quantitation by densitometry and expressed in the bar graph as arbitrary units. The bar graphs are the means ± SE of three experiments. Significance compared to control ***p < 0.001, n.s. non-significant. C Representative confocal pictures of immunofluorescence staining of TNFRI and TNFRII in undifferentiated and differentiated motor like NSC34 cells, scale bars = 20 μm. Two other experiments showed similar results. D A representative immunoblot analysis of a dose dependent effect of TNFα (0.005-0.1 ng/ml) for 24 h on cPLA 2 α expression in differentiated motor neuron-like NSC34 cells. cPLA 2 α protein expression was determined by dividing the intensity of each cPLA 2 α with the intensity of the corresponding actin after quantitation by densitometry and expressed in the bar graph as arbitrary units. The bar graphs are the means ± SE of 3 experiments. Significance compared to control ***p < 0.001, n.s. non-significant. E A representative immunofluorescence staining of a dose dependent effect of TNFα (0.05-0.1 ng/ml) for 24 h on cPLA 2 α expression in differentiated motor neuron-like NSC34 cells. Scale bars = 50 μm. The bar graph is the mean ± SE of 3 different experiments. Significance compared to control ***p < 0.001, n.s. non-significant