In courtroom dramas, the savvy attorney always manages an acquittal for the innocent defendant. Research dramas can be similar. While the obvious suspects for amyotrophic lateral sclerosis (ALS), aka motor neuron disease, would seem to be the motor neurons, growing evidence suggests that these specialized cells are victims more than perpetrators. In at least one form of ALS—that caused by superoxide dismutase (SOD) mutations—it is becoming increasingly clear that glia are the guilty party. In 2003, Don Cleveland and colleagues at the University of San Diego showed that glial expression of mutant SOD is sufficient to cause ALS-like symptoms in mice (see ARF related news story). Now, working independently, researchers at Harvard University and Columbia University, New York, come to similar conclusions. Their findings, together with some new data from the Cleveland lab, may help uncover how and why glia cause harm to motor neurons and may lead to new therapeutic approaches for ALS and, perhaps, other neurodegenerative diseases.

Writing in the April 15 Nature Neuroscience online, the Harvard researchers, led by Kevin Eggan and Tom Maniatis, reported that motor neurons coaxed from mouse embryonic stem cells grow fairly well when surrounded by normal glia, but degenerate more rapidly when cultured together with glia expressing the SOD G93A mutant that causes ALS. In the same issue of the journal, the Columbia group, led by Serge Przedborski, described a slightly different in-vitro approach. They reported that while expression of mutant SOD in primary motor neurons does not cause neurodegeneration, astrocytes expressing the toxic protein kill both primary neurons and motor neurons derived through stem cell differentiation. Since these researchers found that neurons grow fine when co-cultured with fibroblasts, microglia, and myocytes expressing mutant SOD, they reasoned that some toxic factor specific to astrocytes must mediate the demise of motor neurons. Przedborski and colleagues show that the degenerative process requires expression of Bax, a protein that induces programmed cell death, or apoptosis, suggesting it is or ”it might be” a key mediator of glial-neuron interactions. Cleveland and colleagues identified some other factors that might mediate toxicity. They used DNA arrays to compare gene expression patterns in neurons taken from young and adult mice expressing wild-type and mutant forms of SOD.

Maniatis, Eggan and colleagues isolated embryonic stem cells from SOD G93A mice crossed with mice expressing a green fluorescent reporter that is only expressed in motor neurons. Joint first authors Francesco Di Giorgio and Monica Carrasco were then able to identify differentiated motor neurons by their green fluorescence. Though the numbers of surviving motor neurons dropped dramatically within two weeks of taking cells from embroid bodies, some motor neurons survived as long as 54 days after plating. Overall, the researchers found that only about half as many SOD G93A-positive neurons survived as wild-type. To determine if this poor survival may be related to other cells that co-differentiate in the cell cultures, such as glia, the scientists plated differentiated neurons on primary glial cultures obtained from SOD G93A or wild-type mice. Initially, the motor neurons seemed to thrive in either environment, but by 14 days, all neurons growing on glia harboring mutant SOD had begun to falter. There was a 50 and 32 percent decrease in wild-type and SOD-mutant neurons, respectively, when grown on the toxic compared to wild-type glia.

Findings from the Przedborski lab are strikingly similar. Joint first authors Makiko Nagai, Diane Re, Tetsuya Nagata, and colleagues plated primary motor neurons from 12.5-day-old mouse embryos on either a poly-D-lysine/laminin substrate or on astrocyte monolayers. While the numbers of surviving neurons dropped by about 25 percent over 2 weeks, the loss was independent of SOD genotype. Neurons isolated from transgenic mice expressing several SOD variants that cause ALS, including G93A, G37R, and G85R SOD, survived as well as those from normal mice or transgenic mice expressing normal human SOD. However, it was a different story when the researchers plated neurons on mutant astrocytes.

Compared to those grown on normal astrocytes, the number of primary neurons, either wild-type or expressing SOD G93A, dropped by about half within 7 days of plating onto SOD G93A astrocyte monolayers. All astrocyte mutations (G93A, G37R, and G85R) had a similar detrimental effect on primary neurons and embryonic stem cell-derived motor neurons. Significantly, the combination of both mutant neurons and mutant astrocytes did not exacerbate neuronal losses when compared to wild-type neurons grown on mutant astrocytes. “Thus, these data indicate that expression of mutated SOD1 in both astrocytes and motor neurons did not exacerbate the death or the morphometric changes of PMNs caused by its expression in astrocytes alone,” write the authors.

How do mutant astrocytes inflict damage on motor neurons? It appears they may be releasing a toxic soluble factor. When Nagai and colleagues challenged primary neurons with conditioned astrocyte medium, they found that less than 50 percent survived the first 7 days compared to neurons grown in medium conditioned by normal astrocytes. Medium conditioned by several other SOD G93A cell types, including myocytes, fibroblasts, and microglia, had no effect, suggesting that only astrocytes secrete the soluble factor. The authors were able to extend survival of motor neurons on toxic astrocyte monolayers by incubating the cultures with the Bax antagonist V5 (a VPMLK pentapeptide). This suggests that Bax-induced cell death might be at least partly to blame for the neurodegeneration, at least in vitro. In support of this, the authors found that V5 also counteracted the elevation of neuronal fractin, a fragment of β-actin released during apoptosis.

Analyzing the response of motor neurons to toxic astrocytes is a goal shared by Cleveland and colleagues. As reported in today’s PNAS online, first author Christian Lobsiger and colleagues used laser microdissection to pluck out about 3,000 ventral horn motor neurons from adult mice. The researchers then used gene arrays to compare the expression of 30,000 transcripts in normal and mutant SOD neurons and also the expression profiles of purified embryonic motor neurons.

Lobsiger and colleagues found that only 12 genes appeared to be dysregulated in embryonic neurons harboring mutant SOD. The adult cell analysis painted a different picture, however. In neurons taken at 8 weeks of age while the animals were presymptomatic, only seven genes were dysregulated in G37R mutant neurons compared to wild-type, but in the period between 8 and 15 weeks that number jumped to 108. The authors chose a few of these genes for further analysis. Two, 3-phosphoglycerate dehydrogenase (Phgdh) and phosphoserine phosphatase (Psph), that were elevated at week 8, were further elevated at week 15. Both are involved in serine biosynthesis, as is a third gene, Psat1, which the researchers found was also elevated by week 15. The three cases suggest to the authors that serine biosynthesis may be askew in ALS motor neurons. The finding of elevated levels of Phgdh protein support that idea. Serine, an NMDA receptor co-agonist, could be involved in detrimental excitotoxic actions, the authors suggest.

To get a more general picture of gene dysregulation in ALS, the authors also profiled expression patterns in neurons taken from SOD G85R animals. Unlike G37R, this mutant lacks dismutase activity and causes later disease onset. Lobsiger and colleagues found 21 genes dysregulated in G85R neurons that are common to those identified in the G37R animals. The genes fall into three main groups: neuronal regeneration/injury; the complement system; and the lysosomal degradation machinery (see the paper and supplementary information online for a full list of the dysregulated genes). Again, the authors confirmed some of the transcriptional analysis by immunohistochemistry. Ventral horn motor neurons of mutant animals had elevated levels for ATF3 and Sprr1a, both involved in regeneration/injury responses, and complement C1q. “The unexpected induction of mRNAs of the classic complement pathway (C1qa, C1qb, C1qc) long before appearance of obvious clinical symptoms and before major neuroinflammation, suggests that mutant SOD1-induced upregulation of motor neuron-derived complement components is a likely aspect of a toxicity developed within motor neurons that contributes to neurodegeneration,” write the authors. C1q has also been implicated in AD pathology (see Fonseca et al., 2004). Whether astrocyte-derived soluble factors cause any of these neuronal transcriptional changes is unclear.

All told, the three papers suggest that glial/neuron interactions may set off a chain of events that specifically arises from astrocytes and is specifically toxic to motor neurons. Nagai and colleagues found that mutant astrocytes had no effect on GABAergic, dorsal root ganglion, or stem cell-derived cortical interneurons. The three papers also identify some likely pathways that might be involved, which could, perhaps, lead to novel therapeutic interventions. “The model system described here may also provide a high-throughput cell-based assay for small molecules that promote survival of mutant SOD1 motor neurons,” write Eggan and colleagues.—Tom Fagan


  1. In the recent papers from the groups of Przedborski and Eggan, provocative evidence is reported that spinal motor neurons may die in SOD1 mutant mice
    because of soluble toxic factors released by SOD1 mutant astrocytes. This
    result is surprising because previous studies with chimeric SOD1 mutant mice have shown that expression of mutant SOD1 in microglia but not
    astrocytes is implicated in the neuron death. However, profound reactive
    astrocytosis occurs very early in mouse and human motor neuron diseases.
    This is true in the SOD1 mutant mice, where reactive astrocytosis is a dramatic feature of the disease, with prominent reactive astrocytosis occurring long before much motor neuron death occurs (Carlos Pardo, personal communication).

    The new studies provide striking evidence that astrocyte-conditioned medium from SOD1 mutant astrocytes is toxic, as wild-type spinal motor neurons survive longer in culture when cultured alone or with wild-type astrocyte conditioned medium than with mutant astrocyte- conditioned medium. Thus, the lower survival of the spinal motor neurons cannot be attributed to less production of neurotrophic factors by the mutant astrocytes. Together, these in-vitro and in-vivo findings directly implicate reactive astrocytes in the pathophysiology of spinal motor neuron death in the SOD1 mutant mice. One caveat is that in the in-vitro studies, the astrocytes that were studied were obtained from neonatal spinal cords long before any reactive gliosis actually occurs. However, neonatal astrocytes in culture have a similar phenotype to reactive astrocytes in vivo, and may in fact be comparable.

    So how can these new observations of Przedborski and Eggan be reconciled with previous studies that found that chimeric SOD1 mutant mice with mutant SOD1 in microglia but not astrocytes is implicated in the neuron death? For one thing, it is unclear if mutant microglia were actually present in the astrocyte cultures used in these new studies. Steps were taken to minimize microglial contamination, but because astrocytes secrete high levels of microglial mitogens such as colony stimulating factor-1 (CSF1), microglia almost always heavily contaminate neonatal astrocyte cultures prepared by the commonly used methods of McCarthy and DeVellis. It would be good to repeat the study using more stringent methods of microglial elimination, such as immunopanning, and it would be important to confirm by immunostaining that microglia are in fact absent from the astrocyte-conditioned medium at the time it is harvested.
    It is also possible that mutant astrocytes release factors in culture that are toxic to motor neurons but that these factors are not actually secreted in vivo or are not toxic to the neurons in vivo. The only way to find out
    for sure, of course, will be to identify this toxic astrocyte factor. At
    the present time, it is difficult to think of a model that reconciles the previous in-vivo observations implicating mutant microglia with the present in-vitro observations implicating mutant astrocytes.

    Assuming astrocytes make a toxic factor, what could be its nature? The possibility that it is glutamate has already been ruled out, as have been the obvious cytokine candidates. Moreover, the motor neurons undergo
    apoptosis. One possibility is that it is a factor that binds to, inhibits,
    or proteolyzes required trophic factors or culture substrates present in the culture medium that are required for long-term motor neuron survival.

    Another possibility is that the toxic astrocytes alter the pH of the culture medium or lower antioxidant levels, which are both crucial parameters for good neuronal survival. Alternatively, a toxic factor could be released, such as a cytokine of some sort or an excitotoxin. Glutamate agonists have not been ruled out. For instance, homocysteine is an NMDA agonist that is exclusively made by astrocytes; other possibilities are aspartate and N-acetyl-aspartylglutamate, which all act on NMDA receptors. In addition, astrocytes have previously been shown to secrete high levels of NMDA potentiators such as L-glycine or D-serine, and it is possible that the mutant astrocytes secrete higher levels of these. Glutamate excitotoxicity can lead to apoptosis, so it would be important in future experiments to test whether the toxic astrocyte factor can be blocked by APV or other NMDA receptor blockers, as so far only kainate and AMPA receptor blockers have been tested.

    A very interesting new paper by Don Cleveland’s group provides evidence, using laser capture studies of mRNA expression, that spinal motor neurons in the SOD mutant mice have elevated levels of several complement proteins.
    This raises the possibility that there is complement-induced toxicity.
    However, microglia and serum, which are both rich sources of the complete set of complement proteins required for the complement cascade to function, were not present in the motor neuron cultures; therefore, this seems an unlikely possibility. Moreover, complement-mediated toxicity would be expected to cause lysis and not necessarily apoptosis (though mild toxic insults are well documented to lead to apoptosis in neurons).
    Interestingly, the new Cleveland work also provides evidence for a strong
    upregulation of the serine biosynthetic pathway. Astrocytes have
    previously been shown to preferentially use the serine synthetic pathway, whereas neurons do not (a result we have recently confirmed by gene profiling of purified neural cell populations; Cahoy and Barres, unpublished observations). It is possible that SOD1 mutant neurons upregulate these pathways as they die, but a more likely possibility is that there was some contamination by reactive glial genes in these studies, a possibility that is suggested by the presence of other upregulated well-described astrocyte genes such as CD44 and aquaporin 4. This latter possibility again raises the possibility that the toxic factor being released by mutant astrocytes is D-serine or L-glycine.

    Whatever the case, these new papers call attention to the important but still poorly understood roles of neuron-glial interactions in the pathophysiology of neurodegenerative disease.

  2. This paper by C. S. Lobsiger et al. is an experimental and conceptual tour de force. It combines laser capture microdissection (LCM) with Affymetrix oligonucleotide array analysis in superoxide dismutase 1 (SOD1) active versus inactive mutant rodent models of amyotrophic lateral sclerosis (ALS), and it provides potentially breakthrough results in the biology of ALS motor neuron degeneration. Essentially, the group isolated relatively pure populations of lumbar motor neurons from embryonic rat preparations, as well as adult motor neurons in mutant mouse models of SOD1 prior to the appearance of clinical symptoms. Importantly, the group compared embryonic motor neurons that expressed high levels of mutant SOD1 with corresponding wild-type controls overexpressing SOD1 at comparable levels that do not develop ALS-like symptoms. Similarly, adult motor neurons were acquired via LCM from mutant mice expressing dismutase-active SOD1 (the G37R mutant), dismutase-inactive SOD1 (the G85R mutant), wild-type SOD1 overexpressing mice, and non-transgenic littermate controls. The experimental design enabled a provocative microarray analysis that was validated at the protein level via immunocytochemistry for several individual candidates.

    The study presented three major observations. First, no overt transcriptome-related results were found in the embryonic motor neuron preparations, suggesting that any alterations at this early time point are not the result of SOD1-induced transcriptional alterations in vivo. Second, marked alterations in the motor neuron transcriptome were found prior to the presentation of overt clinical symptoms in the mutant mouse lines, indicating that age-related alterations in gene expression may drive the toxic process well before dysfunction becomes apparent clinically. This finding has tremendous implications for human ALS, and at-risk relatives of families with familial ALS should be investigated from an early age. Moreover, additional single cell/population cell microarray studies of motor neurons obtained postmortem are warranted in ALS cases (both sporadic and familial), even though the disease process may be at the end stages when tissues become available. Third, several common classes of transcripts appear to be involved in the pathogenesis of mutant SOD1 (both dismutase-active and dismutase-inactive forms) motor neuron degeneration. These include genes for the neuronal regenerative/injury response, the complement pathway, and the D/L serine biosynthetic pathway. All three classes are interesting, and relatively novel for the investigation of motor neuron disease within mutant SOD1 models, and are certainly worth extensive experimental follow-up.

    Importantly, the LCM-based paradigm enabled the investigators to demonstrate that induction of several classical complement-related genes was fairly neuron-specific, and not a contamination effect from proliferating microglial cells. Previous regional and spinal cord microarray evaluations in mutant SOD1 models were not able to reach this level of single cell/homogeneous population resolution, and consequently were unable to provide such in-depth analysis and interpretation. In summary, this work illustrates the power of combining state-of-the-art functional genomics approaches in well-characterized mutant animal models of neurodegeneration (along with appropriate wild-type overexpressing controls) with microarray analysis for high-throughput evaluation of relevant classes of transcripts that may have functional and/or mechanistic implications for human neurological disorders.

    View all comments by Stephen D. Ginsberg
  3. These two papers by Nagai et al. (2007) and Di Giorgio et al. (2007) independently provide strong evidence that glial cells, and perhaps specifically astrocytes, bearing SOD1 mutations are responsible for degeneration and death of motor neurons in embryonic stem cell (ESC)-based co-cultures of primary neurons and glial cells. Motor neurons bearing SOD1 mutation did not degenerate in the absence of mutant glial cells.

    While these elegant findings provide important insights into the interdependency between neurons and glial cells, and provide key data concerning the pathogenesis of human ALS associated with SOD1 mutation, their relevance to sporadic and other non-SOD1 related forms of human ALS is uncertain. Increasingly, it is becoming recognized that SOD1- associated ALS, and non-SOD1 forms of ALS may be driven through different pathogenetic cascade mechanisms. In SOD1 ALS, the accumulated protein within the conglomerated ubiquitinated inclusion bodies is mutated SOD1. In other, non-SOD1 forms of familial ALS, and sporadic ALS, the filamentous or skein-like ubiquitinated inclusions contain the TAR DNA binding protein, TDP-43 (Neumann et al., 2006; Davidson et al., 2007). Pertinently, the inclusions in SOD1-associated ALS are not TDP-43 immunoreactive (Tan et al., 2007). These latter morphological and immunohistochemical data reinforce the concept that SOD1 and non-SOD1 ALS are separate disorders even though they share a common clinical phenotype. The data, moreover, imply that a role for glial cells, as described in the work of Nagai et al., (2007) and Di Giorgio et al. (2007), may not pertain in the more common forms of ALS that are not associated with SOD1 mutation.

    Nonetheless, a potential role for glial cells in non-SOD1 ALS could, perhaps, be tested in ESC-based studies using the Q342X stop codon mutation in the intraflagellar transport protein 74 (IFT74) gene, which has been associated in one family with a frontotemporal dementia and motor neuron disease (FTD+MND) clinical phenotype (Momeni et al., 2006). One patient from this family with this mutation showed ubiquitinated pathological changes within cerebral cortex and brain stem and spinal cord detectable by TDP-43 immunohistochemistry (Cairns et al., 2007). These were typical of those seen in FTD+MND, and in ALS alone (Neumann et al., 2006; Davidson et al., 2007).


    . TDP-43 in familial and sporadic frontotemporal lobar degeneration with ubiquitin inclusions. Am J Pathol. 2007 Jul;171(1):227-40. PubMed.

    . Ubiquitinated pathological lesions in frontotemporal lobar degeneration contain the TAR DNA-binding protein, TDP-43. Acta Neuropathol. 2007 May;113(5):521-33. PubMed.

    . Non-cell autonomous effect of glia on motor neurons in an embryonic stem cell-based ALS model. Nat Neurosci. 2007 May;10(5):608-14. PubMed.

    . Analysis of IFT74 as a candidate gene for chromosome 9p-linked ALS-FTD. BMC Neurol. 2006;6:44. PubMed.

    . Astrocytes expressing ALS-linked mutated SOD1 release factors selectively toxic to motor neurons. Nat Neurosci. 2007 May;10(5):615-22. PubMed.

    . Ubiquitinated TDP-43 in frontotemporal lobar degeneration and amyotrophic lateral sclerosis. Science. 2006 Oct 6;314(5796):130-3. PubMed.

    . TDP-43 immunoreactivity in neuronal inclusions in familial amyotrophic lateral sclerosis with or without SOD1 gene mutation. Acta Neuropathol. 2007 May;113(5):535-42. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. ALS—Is It the Neurons or the Glia?

Paper Citations

  1. . Absence of C1q leads to less neuropathology in transgenic mouse models of Alzheimer's disease. J Neurosci. 2004 Jul 21;24(29):6457-65. PubMed.

Further Reading

Primary Papers

  1. . Toxicity from different SOD1 mutants dysregulates the complement system and the neuronal regenerative response in ALS motor neurons. Proc Natl Acad Sci U S A. 2007 May 1;104(18):7319-26. PubMed.
  2. . Astrocytes expressing ALS-linked mutated SOD1 release factors selectively toxic to motor neurons. Nat Neurosci. 2007 May;10(5):615-22. PubMed.
  3. . Non-cell autonomous effect of glia on motor neurons in an embryonic stem cell-based ALS model. Nat Neurosci. 2007 May;10(5):608-14. PubMed.