|Home | About | Journals | Submit | Contact Us | Français|
Neuregulin1 (NRG1) is a neuron-derived trophic molecule that supports axoglial and neuromuscular development through several alternatively spliced isoforms; its possible role in the pathogenesis and progression of amyotrophic lateral sclerosis (ALS) is not known. We analyzed the relationship of NRG1 isoform expression to glial cell activation and motor neuron loss in spinal cords of ALS patients and during disease progression in the superoxide dismutase 1 (SOD1) ALS mouse model. Microgliosis, astrocytosis and motor neuron loss were observed in the ventral horns in ALS patients and were increased in SOD1 mice along with disease progression. Type III (membrane-bound) NRG1 expression was reduced in parallel with motor neuron loss but type I (secreted) NRG1 increased and was associated with glial activation. Increased NRG1 erbB2 receptor activation was observed on activated microglia in both ALS patients and in SOD1 mice. This activation was observed at the time of disease onset and prior to upregulation of NRG1 gene expression in the mice. The downregulation of membrane-bound type III NRG1 forms may reflect motor neuron loss, but increased signaling by secreted type NRG1 isoforms could contribute to disease pathogenesis through glial cell activation. NRG1 might, therefore, represent a novel therapeutic target against disease progression in ALS.
The pathogenesis of amyotrophic lateral sclerosis (ALS) is poorly understood and there are presently no effective therapies to stop its insidious progression. How and where the disease begins and the molecular mechanisms of disease progression are not known. While motor neuron loss in the spinal cord has been a central area of research focus for ALS (1, 2), extensive alterations involve neuromuscular synapses, the ventral spinal cord, the lateral corticospinal tract (CST), and the motor cortex (3-9). A major therapeutic strategy has been to try to rescue spinal motor neurons in the ventral horn. Not surprisingly, some of the first clinical trials for patients with ALS focused on neurotrophic growth factors such as brain-derived neurotrophic factor (BDNF) (10, 11), ciliary neurotrophic factor, glial cell-line derived neurotrophic factor (GDNF), insulin-like growth factor-1, and vascular endothelial growth factor (11, 12). Unfortunately, the promise of neurotrophic growth factor treatment was not matched by any clear efficacy in human clinical trials (11). Among the reasons that these clinical trials may have failed are an incomplete knowledge of how and where these growth factors work locally (which is needed for optimal drug delivery) and how they interact with other cell types (10).
ALS research has traditionally been “neuron-centric,” but the protective as well as deleterious roles that astrocytosis and microglial activation glial cells have received increased attention (13-15). “Gliotrophic” factors derived from alternatively spliced forms of the neuregulin 1 (NRG1) gene have been shown to be critical for peripheral nerve development, myelination and, more recently, microglial activation in diverse diseases ranging from peripheral nerve diseases (16, 17) to schizophrenia (18, 19). The NRG1 gene encodes for both secreted (type I) and membrane-bound (type III) forms that are highly expressed in spinal motor neurons (20, 21). Several studies suggest a positive feedback loop between this “gliotrophic” factor and neurotrophic factors; for example, BDNF promotes both transcription as well as the regulated, localized release of soluble NRG1 from neurons at the neuromuscular junction and the peripheral nerve (17, 22). More recently, neuron-derived NRG1 has been implicated in the activation of spinal cord microglia responsible for the generation of chronic pain (23). To date, this important reciprocal signaling pathway has not been explored in ALS or the superoxide dismutase 1 (SOD1) ALS mouse model in which there is also marked glial activation (13-15, 24-26).
In this study, we analyzed NRG1 in the spinal cords of ALS patients and in ALS-SOD1 (G93A) mice at different stages of disease progression. We compared motor neuron loss and glial activation, NRG1 expression patterns, and measured the activation of NRG1 receptor signaling as a function of disease progression.
Fresh frozen thoracic and lumbar spinal cords from 6 sporadic and 2 familial ALS subjects (5 males, ages 41-84 years; 3 females, ages 64-77 years) and 6 control patients (4 males. ages 67-84 years, 2 females, ages 58 and 76 years) with no pathological evidence of neurological disease) were provided by one of the authors (J.R.) (n = 10) (2), or were obtained from the Human Brain and Spinal Fluid Resource Center (VA West Los Angeles Healthcare Center, Los Angeles, CA) (n = 4). Postmortem time intervals ranged from 2 to 25 hours (mean = 13.5 hours). Genetic testing, including of SOD1 mutations, was not performed on the patient samples.
Breeding pairs of SOD1 (G93A) transgenic (Tg) mice were obtained from Jackson Laboratories (Bar Harbor, ME) and genotyped as described (27, 28). SOD1 mutant mice were tested weekly for movement behavior. Thoracic and lumbar spinal cord regions were collected from each mouse at days 35 or 56 (preclinical), days 90-100 (disease onset), and days 117-126 (end-stage disease); 4-8 mice (SOD1 Tg and non-Tg littermates) were used per group at each time point (4 time points, 32 mice total).
Fresh frozen human and fresh mouse spinal cords were fixed in 4% paraformaldehyde for 24 hours, washed overnight in PBS, and immersed in 30% sucrose until saturated, all at 4°C. The spinal cords were processed and embedded in OCT (Tissue-Tek, Sakura Finetek USA, Inc., Torrance, CA). Frozen sections were cut transversely at 20 μm thickness and placed on Superfrost slides (Thermo Fisher Scientific), and then stained with either Luxol fast blue/periodic acid Schiff (Poly Scientific, Bay Shore, NY) for the presence of myelin, or Cresyl violet (Diagnostic Biosystems, Pleasanton, CA) for motor neuron counts.
Identification and quantification of microglia in human and mouse spinal cord tissue sections were performed using antibodies specific for human CD68 (mouse IgG1, 1:20, DAKO Cat # N1577, Carpinteria, CA) and mouse microglia CD11b (rat IgG2b, 1:100, Millipore, Billerica, MA). Astrocytes were labeled with anti-glial fibrillary acidic protein (GFAP) antibodies against human (rabbit polyclonal antibody, 1:100, DAKO) and mouse (mouse IgG 1:100, Chemicon, Temecula, CA). Detection of activation of the NRG1 erbB2 receptor was performed with either a rabbit anti-phospho-erbB2 antibody (p-Neu, Tyr1248, 1:50 Santa Cruz Biotechnology, Santa Cruz, CA) for both human and mouse sections or a rabbit anti-erbB2 antibody (Neu, C-18, 1:50 Santa Cruz Biotechnology) for mouse sections. Each primary antibody was diluted in blocking solution (10% normal goat serum, 0.05% Triton X-100 in PBS) overnight at 4°C, followed by incubation with goat anti-mouse or rabbit Alexa Fluor 488 (1:100, Invitrogen, Carlsbad, CA). For CD11b, phospho-erbB2 and CD68 immunostaining, biotin-conjugated goat anti-rat, rabbit or mouse was used as a secondary antibody; the signal was amplified using a tyramide signal amplification kit (1:250, Invitrogen) following the manufacturer’s instructions.
Monoclonal antibodies specific for the type III human NRG1 isoform (also referred to as the cysteine-rich domain [CRD]) were developed in collaboration with the UC Davis/NIH NeuroMab Facility, Davis, CA. To test the specificity of the type III NRG1 antibody, 1.5×105 Chinese hamster ovary cells were seeded in each well of a 24-well plate and co-transfected with either human type I NRG1 or human type III NRG1 cloned into pFLAG-Myc-CMV™-20 (E8783, Sigma, St. Louis, MO) with enhanced green fluorescent protein cloned in the pTriex vector (Novagen, Gibbstown, NJ) using Lipofectamine 2000 (11668-019,Invitrogen). After 24 hours, the cells were fixed in 4% paraformaldehyde for 30 minutes and stained for type III NRG1 (1:500; NRG1-CRD, N126B/31, 20100712, 73-226, UC Davis/NIH NeuroMab Facility). Nuclei were counterstained with 4’, 6-diamidino-2-phenylindole (DAPI).
Neuregulin-CRD (cysteine-rich domain, Type III) (mouse IgG, 1:200, clone N126B/31) or panNRG1 (SC348, rabbit IgG, 1:100 Neuregulin-1α/β1/2, C-20, Santa Cruz Biotechnology) were incubated overnight at 4°C, followed by a biotinylated goat anti-mouse or anti-rabbit IgG secondary antibody (1:200), streptavidin-horseradish peroxidase, and DAB Vectastain reagents according to the manufacturer’s recommendations (Vector Laboratories, Burlingame, CA).
1.5×105 Chinese hamster ovary cells were seeded in each well of 24-well plate and co-transfected with either human type I NRG1 or human type III NRG1 cloned into pFLAG-Myc-CMV-20 (Sigma) or enhanced green fluorescent protein cloned in the pTriEx-1.1 vector (Novagen, 70840-3) using Lipofectamine 2000 (Invitrogen). After 24 hours the cells were harvested and total protein was extracted using RIPA lysis and extraction buffer [25 mM Tris pH 7.6, 150 mM NaCl, 1% NP40, 1% sodium deoxycholate, 0.1% SDS, and Halt protease and phosphatase inhibitor cocktail (Thermo Scientific, 78442, Rockford, IL) for further immunoblotting analysis. Total protein of one mouse spinal cord and one piece of human spinal cord were also extracted using RIPA buffer. The NRG1-CRD antibody was used at a dilution of 1:500. SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific, 34077) was used as directed.
Lumber spinal cords were harvested from either SOD1 (G93A) (n = 5) or non-Tg littermates (n = 5) at disease end stage. Total protein was extracted using RIPA buffer. Protein samples from 5 of the above pairs were used for immunoblotting analysis. Antibodies were utilized as NRG1-CRD (mouse IgG, 1:500, clone N126B/31), β-actin (1:1000; A5441; Sigma), anti-phospho-ErbB2 (Y1248) (0.5 μg/mL; AF1768; R&D Systems, Minneapolis, MN), and anti-erbB2 Neu (C-18) (1:500; SC284; Santa Cruz). SuperSignal West Pico Chemiluminescent Substrate was used as directed. Each blot was probed, stripped and reprobed with a different antibody. Quantification of band intensity was performed using Metamorph image analysis software (Universal Imaging Corporation, Buckinghamshire, UK), as previously described (29-31).
Digital images were obtained with a Nikon Eclipse 600 epifluorescence microscope with a Princeton Instruments Micromax 5-MHz cooled CCD camera or a QImaging color digital camera (Micro Publisher 5.0 RTV). Mouse motor neurons (i.e. large neurons with clear nucleoli and distinctly labeled cytoplasm on Cresyl violet staining) were counted manually by 3 individuals, as described (32). Mouse microglia (CD11b-positive) (33), mouse reactive astrocytes (GFAP-positive), and mouse phospho-erbB2-positive cells were quantified based on the presence of both a glial cell shape and a nucleus labeled with DAPI. For each cell type counted, 6 to 10 sections at the lumbar spinal cord levels were used to analyze each condition in each animal. Metamorph image analysis software (Molecular Devices) was used to quantify the degree of colocalization by creating thresholded objects in one wavelength and measuring the percentage of pixel overlap for each in the second thresholded wavelength. Thresholds were set visually using the criteria of identifying the threshold that best outlined the individual cell types. For each measurement, the average counts of each group for each time point were made from 4 to 8 animals. Cell counting was done manually and blinded by 3 different individuals. For motor neuron counts in the SOD1 animals there were differences in total numbers for each of the counters, but the relative differences between the groups were consistent. Results from only one of the counters is presented due to the variations.
Mouse spinal cords were dissected and thoracic spinal cords (T1-T12) were rapidly frozen and stored at −80°C for RNA; the lumbar levels were placed in 4% paraformaldehyde for histology as described above. RNA was extracted using the Qiagen RNeasy Lipid Tissue Mini Kit (Qiagen, Valencia, CA) (34). Quantification of RNA was carried out using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE). The quality of RNA was determined on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) using an RNA 6000 Nano chip kit, RNA ladder and Agilent analysis software (Agilent Technologies). All samples had RIN (RNA integrity number) values above 8.0 and 260/280 ratios near 2.0.
The relative expression of mouse type I and III NRG1, BDNF and GDNF were measured relative to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) using Taqman Assays-On-Demand primers (Applied Biosystems, Foster City, CA). 1.5 μg of total RNA was used in a 20 μl reverse transcription synthesis reaction primed with oligo-dT primers (Superscript First Strand Synthesis System, Invitrogen). PCR was performed in triplicate using 1X Taqman Universal PCR master mix (Applied Biosystems) with the DNA Engine Opticon Continuous Fluorescence Detection System (MJ Research, Waltham, MA) utilizing the following primers and Taqman probes: type I NRG1: Mm00626552_m1; type III NRG1: Mm01212129_m1; BDNF: Mm01334047_m1; GDNF: Mm00599849_m1 and GAPDH: Mm99999915_g1). Cycle threshold (Ct) values were calculated using Opticon monitor software, with the threshold set at 40 standard deviations above background. The relative expression was calculated by normalizing the expression of individual genes to GAPDH and using the 2-ΔΔCt method (35).
For cell counts and gene expression, a non-parametric ANOVA with Tukey’s post hoc test was performed between non-Tg control and SOD1 at each stage of the disease. In some experiments, comparisons were made between end-stage SOD1 mice and the same gender and age non-Tg littermates. Groups were considered significantly different at p < 0.05 by Student t test.
SOD1 mice have been the most widely used animal model for ALS (36), but there are both pathologic similarities and differences between the human and mouse diseases (9, 37). There is severe lower motor neuron loss in the ventral horn in both, but there are differences in the descending central motor tracts, that is, the lateral corticospinal tract (LCST) in humans and dorsal corticospinal tract (DCST) in mice (Fig. 1). Quantification of motor neuron cell number loss in the ALS patients was described previously (2). The motor neuron loss in SOD1 mice in the present study (Fig. 1) is consistent with published reports (36).
In both the ALS cases and end-stage SOD1 mice there are numerous small cells in the ventral horns that are not present in controls; these small cells consist of a combination of astrocytes and activated microglia (13-15, 24-26). Large numbers of microglia were consistently seen in the ventral horns in the 8 ALS cases (Fig. 2), but (although still increased compared to controls) there were generally fewer reactive astrocytes compared to microglia, and the numbers were quite variable from patient to patient (data not shown). In more than 20 end-stage SOD1 mice there were similar levels of both microglia and astrocytes (data not shown). The finding of differences in relative proportions between ALS and end-stage SOD1 mice, with astrocytosis being more prominent in the latter, is similar to that in other reports (36)
In SOD1 mice, there was progressive motor neuron loss and activation of both microglia and astrocytes (Fig. 3A, B). Type I (soluble) NRG1 mRNA dramatically increased in the latest stage of the disease (days 112–126) (Fig. 3C). However, during this time period there was a significant downregulation of type III (membrane-bound) NRG1 mRNA expression (Fig. 3C). There were changes in both BDNF and GDNF expression but these were not as consistent as the NRG1 mRNA changes, where a modest increase in GDNF mRNA at end-stage did not reach statistical significance (Fig. 3D).
To determine whether the reduction in type III NRG1 was due to motor neuron loss we co-developed an isoform-specific monoclonal antibody (with UC Davis/NIH NeuroMab Facility) for type III NRG1 CRD (Fig. 4A). This antibody does not cross-react with type I NRG1 in transfected Chinese hamster ovary cells by Western blotting (Fig. 4B) and immunostaining (Fig. 4C) and labels an approximately 50-kDa band in normal human and mouse spinal cord tissues (Fig. 4B). We then compared total NRG1 expression using an established pan-NRG1 antibody (sc348, epitope shown in Fig. 4A) to type III NRG1 expression in both ALS and end-stage SOD1 mice spinal cords (Fig. 5). Both antibodies labeled ventral horn motor neurons in ALS patients and SOD1 mice, even in advanced disease in which motor neurons show significant pathological changes. This suggested that type III NRG1 is not appreciably reduced in individual motor neurons and that the reduction in mRNA might be explained by motor neuron loss. Indeed, the reduction of type III NRG1 mRNA correlated significantly (R = 0.995) with motor neuron loss (Fig. 5C). This is corroborated by Western blotting in Figure 5D showing a significant downregulation of type III NRG1 in SOD1 mice (30% reduction, 5 mice/group (Fig. 5D and data not shown).
Although the main sources of type I and III NRG1 isoforms in normal spinal cord are ventral motor neurons (20, 21, 38), the pan-NRG1 antibody also labeled many small cells in the ventral horn (Fig. 5A) that were not labeled by the type III specific antibody (Fig. 5B) in both ALS and SOD1 mice (n = 6 samples per group). Based on their shapes they might be glia but efforts to double stain these cells did not identify them as either reactive astrocytes (GFAP, data not shown) or microglia. One other possibility not yet explored is that they represent NG-2-positive oligodendroglial progenitor cells that have been shown to be increased in ALS (39, 40). Although the identities of these smaller cells are not clear, they could, along with the remaining motor neurons, be a source of soluble NRG1 signaling in both ALS and the SOD1 mouse model that could contribute to the increase in identified glial cells observed.
Whereas type III NRG1 remains tethered to the membrane even after proteolytic processing (41), type I NRG1 is secreted in a regulated fashion, after which it becomes concentrated in regions rich in heparan sulfate proteoglycans due to its unique heparin-binding domain (Fig. 4A) (29). Because there are no highly reliable type I isoform-specific antibodies, we determined the spatial localization of type I NRG1 activity by looking for activation of its membrane-bound receptors (42, 43). NRG1 promotes microglial activation in a peripheral nerve injury model of chronic pain through binding to either erbB3 or erbB4 receptors that then heterodimerize with erbB2 to mediate signal transduction through tyrosine phosphorylation (23, 44). Therefore, we determined the level of both erbB2 receptors and activated erbB2 receptors (p-erbB2) in the SOD1 mice and in ALS. Both erbB2 and p-erbB2 expression are increased in the SOD1 mouse ventral horn (Fig. 6A). While erbB2 is expressed primarily only in motor neurons of non-Tg littermates, the increase in erbB2 expression is due to its presence in many additional cells. However, activated p-erbB2 was seen in a subset of these cells in the SOD1 mice, but not in non-Tg littermate controls. The specificity of these 2 antibodies is shown by Western blotting in Figure 6B.
Double labeling of p-erbB2 with microglial and astrocytic markers revealed that activated erbB2 receptors were predominately on microglia (70% of CD11b-positive microglia), and, to a lesser extent, on astrocytes (22% of GFAP-positive astrocytes) (Fig. 7A, B). Quantification of cells showing erbB2 receptor activation as a function of disease progression in the SOD1 mouse showed that increased NRG1 receptor activation starts from day 56 and increases over time (Fig. 7C), correlating well with the pattern of microglial activation (Fig. 3B). These changes precede the induction of type I NRG mRNA, which occurs later in the disease course (day 90 to end-stage).
There was also a marked increase in erbB2 receptor activation in the ventral spinal cord in ALS patients that mostly colocalizes with activated microglia (167 ± 44 CD68+p-erbB2+ cells/mm2, n = 3) compared to control spinal cords (18 ± 8 CD68+p-erbB2+ cells/mm2, n = 3; p < 0.01) (Fig. 8). erbB2 receptor activation was present at multiple levels (thoracic and lumbar) in 6 out of 6 ALS patients but not in any of 4 control patients. Thus, the results suggest that aberrant NRG1 signaling could underlie microglial activation in both the SOD1 model and in ALS ventral spinal cord.
Our data support the hypothesis that NRG1 released from injured neurons and other cells in the spinal cord induce microglial activation, which, in turn, could lead to progressive motor neuron degeneration in ALS. Our parallel findings in both human tissues and the SOD1 mouse model show progressive activation of the NRG1 erbB2 receptor on activated microglia raising the possibility that disease progression in ALS could be targeted therapeutically by disrupting NRG1 signaling. Recently, NRG1 signaling through activation of erbB2 receptors has been shown to be critical for microglial activation in the dorsal spinal cord of rats that develop chronic pain following peripheral nerve injury (23). Using a novel NRG1 antagonist (45) injected intrathecally, the number of activated microglia was reduced together with mechanical and cold pain-related hypersensitivity (23). These findings suggest that NRG1 is an important, local signal for microglial activation in the spinal cord after neural injury and that blocking NRG1 signaling could be a novel approach to block the untoward effects that ensue.
Microglial activation has long been argued to be important in the pathogenesis of ALS. In fact, minocycline, an antibiotic drug, showed promise in blocking microglial activation in the SOD1 model (46); however, clinical trials with minocycline have not been encouraging (47, 48). One possible reason for this failure is its many off-target effects on cytochrome-C, MAPK, nitric oxide, and caspases that could promote apoptosis and obscure any beneficial effects (49). Another possibility for this failure is that subsets of activated microglia play neuroprotective roles in the different stages of the disease (15). Boillée et al reported that diminishing mutant SOD1 levels in microglia slow disease progression in mutant SOD1G93A mice during later stages (50); however, wild-type donor-derived microglia promote neuroprotection and extend survival in mutant SOD1G93A/PU.1 knockout mice (51). Therefore, a better understanding of the local signals between neurons and glia that both activate and block different subtypes of microglia will be important to modulate their neurodegenerative and neuroprotective roles in ALS.
Persistent activation of NRG1 receptors on activated microglia in both ALS patients and SOD1 mice was observed at relatively early stages of the disease (Figs. 7, ,8).8). As the disease progresses, a reciprocal increase in type I NRG1 mRNA together with a decrease in type III NRG1 was observed in the spinal cord; therefore, early NRG1 receptor activation on microglia cannot be explained by increased NRG1 mRNA levels. One possibility is that increased soluble forms of NRG1 are released through post-transcriptional mechanisms at the early disease stages. Type I NRG1 is synthesized as a transmembrane precursor in neurons (22, 30, 52). The precursor is transported down axons and can be released at sites where it is needed in response to local gradients of neurotrophins, such as BDNF (31). Once Type I NRG1 is released, it becomes concentrated at non-random sites by binding to specific heparan sulfate proteoglycans in the extracellular matrix (29). Thus, during the degenerative disease process, a combination of local gradients of neurotrophic factors as well as changes in the extracellular matrix composition could be important variables, other than transcription, that contribute to increased NRG1 signaling in disease progression.
Exactly which cells secrete NRG1 is also not entirely clear. Using a pan-NRG1 antibody, NRG1 staining was seen both in injured motor neurons, which also express type III forms, and in smaller cells in the ventral horn that did not co-label with type III NRG1 or the glial or microglial markers tested; they might be NG-2-positive oligodendroglial progenitor cells. Sorting out the precise alternatively forms of NRG1 is further complicated by a lack of highly specific anti-NRG1 antibodies together with a low abundance and short regions of specific DNA sequences in these forms that makes it difficult to detect them by in situ hybridization.
A second possibility is that other ligands may “cross-talk” with the erbB2 receptors producing their activation. While there are very few ligands known to activate the erbB2 receptors (53), there are a number of homologous NRG genes, some of which are also expressed in the nervous system. For example, NRG2 shares sequence homology to NRG1, activates erbB receptors, and is expressed in brain and spinal cord (54).
The expression and signaling of soluble NRG1 forms increases at end stage disease, but membrane-bound type III NRG1 is significantly reduced both in ALS and the SOD1 model. Using an isoform-specific antibody against its intracellular domain, continued protein expression of type III NRG1 was seen in surviving human and mouse motor neurons. In fact, the reduction of type III NRG1 mRNA correlated with the loss of motor neurons, suggesting that this isoform may not be critical in disease pathogenesis. Type III NRG1, also referred to as the CRD form because of its cysteine-rich hydrophobic domain, has been shown to have important roles in peripheral nerve myelination (17, 55), but requires direct cell-cell contact (26) and thus may have limited opportunities to activate receptors on microglial cells unless they are in direct contact with motor neurons.
To date, translating therapeutics from the SOD1 model to ALS patients has been disappointing (9, 37). One of the greatest pathological differences between ALS and this model is the presence of marked myelin loss in the LCST of ALS patients (56), but not in the equivalent DCST, (in which approximately 80% of corticospinal tract axons are located [57, 58[). There are, however, reports of neuronal and descending cortical axon loss by others (59-61).
On the other hand, both ALS patients and SOD1 mice showed motor neuron loss together with increased cellularity in the ventral horn. Some of this cellularity is due to the presence of activated microglia; however, astrocytosis was more prominent in ventral horn of the mice, but relatively less remarkable in the ALS tissues. Schiffer et al reported the presence of astrogliosis in the ventral horns from 70 autopsied ALS cases, but did not quantify the data nor compare this to the SOD1 mouse (62, 63). Astrogliosis in the SOD1 mice was extremely prominent in our mice and has been frequently observed by others (64-66), but very few of these astrocytes showed NRG1 receptor activation. In summary, while gliosis is present in the ventral horns of both ALS patients and SOD1 mice, extensive microglial activation and sustained NRG1 receptor signaling are common to both and may, therefore, be amenable to targeted therapeutic strategies.
We thank Drs. G. Acsadi and W. Kupsky for expert advice, Dr. J. Benjamins for reviewing the manuscript, Ms. X. Li, Ms. H. Deol and Dr. Z. Ma for technical support.
The work was supported by grants from the Hiller ALS Center at Wayne State University (F.S. and J.A.L.) and the National Institutes of Health R01 NS059947 (J.A.L. and F.S.).