Aggregation and Motor Neuron Toxicity of an ALS-Linked SOD1 Mutant Independent from Wild-Type SOD1

See allHide authors and affiliations

Science  18 Sep 1998:
Vol. 281, Issue 5384, pp. 1851-1854
DOI: 10.1126/science.281.5384.1851


Analysis of transgenic mice expressing familial amyotrophic lateral sclerosis (ALS)–linked mutations in the enzyme superoxide dismutase (SOD1) have shown that motor neuron death arises from a mutant-mediated toxic property or properties. In testing the disease mechanism, both elimination and elevation of wild-type SOD1 were found to have no effect on mutant-mediated disease, which demonstrates that the use of SOD mimetics is unlikely to be an effective therapy and raises the question of whether toxicity arises from superoxide-mediated oxidative stress. Aggregates containing SOD1 were common to disease caused by different mutants, implying that coaggregation of an unidentified essential component or components or aberrant catalysis by misfolded mutants underlies a portion of mutant-mediated toxicity.

After the discovery (1) that dominant missense mutations in Cu/Zn superoxide dismutase (SOD1) were the primary cause of 15 to 20% of familial ALS cases, experiments with transgenic mice expressing ALS-linked mutations showed that selective motor neuron degeneration arises from toxic properties of the ubiquitously expressed mutant enzymes (2–5). This has fueled proposals that (i) mutant SOD1s have an altered substrate affinity that leads to high levels of toxic reaction products (6–8); (ii) the reduced SOD activity arising from unstable or imperfectly folded enzymes causes toxic oxidative stress through an imbalance in oxidative defenses (9, 10), which in turn may exacerbate aberrant catalysis; or (iii) poorly or unstably folded mutants mitigate SOD1-containing aggregates that are toxic to motor neurons (5, 11).

To test these hypotheses, the influence of wild-type SOD1 polypeptide and activity levels on the onset and progression of SOD1 mutant toxicity was determined by eliminating or elevating wild-type SOD1 in mice developing motor neuron disease due to expression of an ALS-linked mutant with glycine substituted to arginine at position 85 (SOD1G85R). Because in vivo the SOD1G85R mutant polypeptide appears to retain only about 10% of the activity of the wild-type SOD1 (at least when expressed in yeast) (12) and disease arises with low levels of the mutant protein (5), neither SOD1 activity nor polypeptide levels are markedly elevated in the SOD1G85Ranimals.

SOD1G85R mice were mated to mice expressing wild-type human SOD1 at six times the level of endogenous mouse SOD1 (13). In addition, a two-stage breeding strategy (13) was used with mice carrying a targeted deletion of the endogenous SOD1 gene (14) to generate mice expressing SOD1G85Rin the absence of endogenous SOD1, along with littermates with one or both endogenous SOD1 alleles. Total spinal cord extracts from two animals from each genotype were immunoblotted with an antipeptide antibody (15) that reacts exclusively with human, but not mouse, SOD1. Neither a sixfold increase in wild-type SOD1 (Fig. 1A) nor its complete elimination (Fig. 1B) affected the accumulated levels of mutant SOD1G85Rprotein. Thus, despite a decreased stability of SOD1G85Rrelative to wild-type SOD1 in these mice (5), the wild-type protein did not stabilize mutant SOD1. Moreover, the presence of SOD1G85R had no effect on the level (Fig. 1A) or the activity of wild-type SOD1.

Figure 1

The presence or absence of wild-type protein does not affect the stability of mutant protein. (A) Levels of wild-type human SOD1 or mutant SOD1G85R determined by immunoblotting spinal cord extracts from normal mice (control), transgenic mice expressing wild-type human SOD1 to six times the level of endogenous mouse SOD1 (hSOD), transgenic mice expressing SOD1G85R (G85R), or mice carrying both the hSOD1 and SOD1G85R transgenes (G85R, hSOD). Each pair of slots represents analyses from duplicate animals. Human SOD1 was detected with a peptide antibody specific for human, but not mouse, SOD1 (5). (B) Total spinal cord extracts from mice transgenic for SOD1G85R and carrying both endogenous SOD1 alleles (G85R, SOD+/+) or carrying a deletion of one (G85R, SOD+/–) or both (G85R, SOD–/–) endogenous SOD1 alleles. Top panel, Extract proteins detected with Coomassie blue; middle and bottom panels, immunoblots with antibodies specific for human SOD1 (middle) or mouse SOD1 (bottom) (5,25).

Survival plots comparing transgenic mice expressing both SOD1G85R and wild-type human SOD1 (Fig. 2A, solid squares, n = 20) or SOD1G85R alone (Fig. 2A, open squares, n = 14) indicated that increasing wild-type SOD1 activity and polypeptide level had no beneficial effect. Age distributions at end-stage disease were indistinguishable (with the exception that a small fraction of mice with normal SOD1 levels survived up to 20 days longer). There were no differences in age of disease onset, and duration after onset was also constant (a median of 18 days for both genotypes). Similarly, no differences in age at end-stage disease could be seen among mice expressing SOD1G85R in the presence of normal levels of mouse SOD1 (Fig. 2B, triangles, n = 5) when levels of mouse SOD1 were reduced to 50% of normal levels (Fig. 2B, open circles,n = 10) or in the complete absence of endogenous mouse SOD1 (Fig. 2B, solid circles, n = 5).

Figure 2

Disease onset and progression mediated by SOD1G85R are independent of wild-type SOD1 protein or activity levels. (A) Survival plots of SOD1G85Rtransgenic mice (▪) with or (□) without a human wild-type SOD1 transgene to elevate wild-type SOD1 levels to six times the normal amount. (B) Survival plots of mice with the SOD1G85R transgene and (▴) both endogenous mouse SOD1 genes or (○) heterozygously or (•) homozygously deleted for endogenous mouse SOD1.

A prominent pathology found in all mouse lines that develop ALS-like disease from expressing SOD1G85R is neuronal and astrocytic inclusions that appear in the SOD1G85R mice before clinical signs, increase markedly in abundance during disease progression (5), and are highly immunoreactive for SOD1 that is concentrated either at the periphery of the inclusion or throughout it (Fig. 3B, arrowheads) or more diffusely localized in aggregates (Fig. 3B, arrows), many of which (for example,Fig. 3B, lower right arrow) are not identified by conventional histological stains (Fig. 3A). Examination of mice expressing other SOD1 mutants revealed that this is a common pathology of SOD1 mutant–mediated ALS in mice, because diffuse aggregates that are strongly immunoreactive for SOD1 (Fig. 3D, arrows) and SOD1-containing inclusions (Fig. 3D, arrowheads) observable by conventional histology are also found in both astrocytes and neurons of lines expressing high levels of SOD1G93A (Fig. 3C), as well as one line expressing low levels of SOD1G93A (16) and in neurons in several SOD1G37R lines (17). Neither type of SOD1 aggregate was found at any age in mice expressing wild-type human SOD1. Similar SOD1-containing aggregates have also been reported for some human sporadic (18–20) and familial (21–24) ALS cases, including a familial ALS patient with a 2–base pair deletion at codon 126 of SOD1 (Fig. 3, E and F). Because this latter mutation truncates the final 27 amino acids of SOD1 (23), strong immunostaining of these aggregates with an antipeptide antibody against SOD1 amino acids 125 through 137 (25) showed that wild-type SOD1 is an apparently prominent component of these aggregates (Fig. 3F).

Figure 3

Inclusions that are strongly immunoreactive with SOD1 antibodies are common features in transgenic mice expressing SOD1G93A, SOD1G85R, and SOD1G37R and resemble those seen in human patients with familial ALS. (A) Hemotoxylin and eosin (H&E) staining of the ventral portion of a spinal cord from an end-stage mouse expressing mutant SOD1G85R. (B) The identical section as in (A) (35), reacted with an antibody recognizing both mouse and human mutant SOD1. (C and D) A similar pair of sections to those in (A) and (B), except that they are from a mouse (obtained from Jackson Laboratories) developing motor neuron disease from expressing mutant SOD1G93A(2). (E) An inclusion in the spinal cord, detected with H&E, from a patient who died from disease mediated by a frameshift mutation at SOD1 position 126. (F) The identical section as in (E) (35), reacted with an antipeptide antibody that recognizes human wild-type SOD1 but not the truncated mutant. Arrowheads in (B), (D), and (F) point to inclusions that are visible in (A), (C), and (E) and are intensely immunoreactive to SOD1 antibodies. Arrows in (B), (D), and (F) point to SOD1 immunoreactive aggregates that appear to contain less dense accumulations of SOD1, which may be the precursors of the more intensely stained aggregates. Scale bar, 10 μm.

To examine whether wild-type SOD1 exacerbated the aggregates, pathology arising during disease progression was determined in SOD1G85R mice with varying levels of wild-type SOD1. Before the onset of symptoms (at ages up to 7.5 months), there were no SOD1-positive inclusions or reactive astrocytes in spinal cords from the any of the mice and no axonal degeneration in dorsal or ventral roots. A 70% loss of large (>5 μm) caliber motor axons in the L5 ventral root was seen at end-stage disease in SOD1G85Rtransgenic mice independent of the wild-type SOD1 content, whereas the small-caliber axons were almost completely spared (Table 1). Similarly, in all genotypes the numbers of motor and sensory axons degenerating at the end stage were comparable, as were the numbers of aggregates in motor neurons and astrocytes (Fig. 4A), each of which was immunoreactive for SOD1 (Fig. 4B, arrowheads), as well as more diffuse SOD1-containing aggregates (Fig. 4B, arrows).

Figure 4

SOD1-immunoreactive aggregates form in the absence of endogenous mouse SOD1.

(A) H&E staining of the ventral portion of a spinal cord from an end-stage mouse expressing mutant SOD1G85R and deleted in both endogenous SOD1 alleles. (B) The identical section as in (A) (35), reacted with an antibody recognizing both mouse and human mutant SOD1. The central arrowheads in (A) and (B) point to an inclusion that reveals a halo of intense immunoreactivity to SOD1 antibodies. Arrows point to inclusions that have uniform SOD1 immunoreactivity and are not readily identifiable with H&E. Scale bar, 10 μm.

Table 1

Surviving large and small motor axons at end-stage disease in the L5 ventral roots of mice with normal, reduced, or elevated levels of wild-type SOD1. Statistical tests (analysis of variance, P = 0.826) revealed no significant diference between the numbers of large surviving axons in G85R; G85R, SOD+/−; G85R, SOD−/−; and G85R, hSOD1 mice. For all genotypes, three mice were analyzed, except that marked with an asterisk, for which two animals were measured.

View this table:

Thus, changing the levels of wild-type SOD1 polypeptides and activity between none and six times the normal level of SOD1 did not affect the pathology, onset, or progression of disease mediated by familial ALS mutant SOD1G85R. These findings are incompatible with several proposed toxic properties of the mutants and bear directly on possible therapeutic strategies.

First, contrary to earlier in vitro evidence (6,7, 26, 27), these results showed that motor neuron degeneration almost certainly did not arise from more efficient use of either peroxynitrite or H2O2. Peroxynitrite production [proposed to form through the spontaneous reaction of nitric oxide with superoxide (6)] should be substantially elevated by the absence of wild-type SOD1 and probably lowered by a more rapid removal of superoxide in mice with a sixfold elevation of wild-type SOD1, thereby accelerating or slowing disease, respectively. This was clearly not the case for mutant SOD1G85R. When coupled with the absence of biochemically detectable protein nitration in SOD1G37R(28) and SOD1G93A (29) mice, it seems unlikely that such enzyme-mediated nitration is a key toxic property, despite apparent increases in nitration that have been reported using immunocytochemistry in some cases of familial and sporadic ALS (19, 30) and in SOD1G93A mice (29). Similarly, although we cannot directly measure the levels of intracellular H2O2, one might predict an elevation or lowering of H2O2 caused by increasing or decreasing levels of wild-type SOD1. This in turn would accelerate or slow disease were the mutant to use H2O2 more efficiently than the wild type. On the other hand, it may be that the H2O2produced in the active site from superoxide is the preferred substrate for aberrant catalysis. A sixfold molar excess of wild-type SOD1 would obviously reduce the access of the mutant to superoxide (through competition or by accelerated catalytic destruction of it, or both), whereas elimination of wild-type SOD1 would strongly enhance it. Both would predict changes in the rate of damage. But our evidence, along with the absence of evidence for predicted damaged products in SOD1G37R (28) and SOD1G93A(29) mice and in vitro evidence questioning whether the mutants do use H2O2 as a substrate (31), showed that this too was not the case.

Second, the fact that disease was not accelerated in the complete absence of wild-type SOD1 argues directly against the hypothesis (9, 32) that toxicity arises even in part from a reduction in normal SOD1 activity. Third, although we show that wild-type SOD1 was a component of aggregates in human disease, the wild-type protein was not required for aggregation, either for stabilizing the mutant or for contributing to aggregates nucleated by mutant SOD1.

Finally, the failure of increased SOD1 activity to ameliorate disease argues strongly against potential therapeutic attempts to slow disease by increasing superoxide scavenging activities using SOD1 mimetics. Indeed, the absence of influence of SOD1 activity on mutant toxicity challenges the idea that toxicity derives from (or is even linked to) oxidative stress arising from superoxide. Rather, mutant SOD1–containing aggregates that coincide with disease onset and that increase in abundance as disease progresses were found to be common to disease caused by different mutants. However, this cannot be the whole story concerning toxicity, because mutant-mediated defects (including a selective slowing of some cargoes of slow axonal transport) are seen at least 5 months before the appearance of aggregates (33); and a mating experiment strategy similar to that used here has proven that in the absence of neurofilaments, SOD1G85R mediates disease with a substantially delayed onset (34) and a markedly reduced frequency of SOD1-containing aggregates (33). Nevertheless, the present efforts support the idea that one aspect of toxicity may arise through an as-yet-unidentified chemistry mediated by the misfolded aggregated mutants or through loss by coprecipitation of an essential component or components.

  • * Present address: Pharmaceutical Research Institute, Bristol-Myers Squibb, Wallingford, CT 06492–7660, USA.

  • To whom correspondence should be addressed. E-mail: dcleveland{at}


Stay Connected to Science

Navigate This Article