Report

Onset and Progression in Inherited ALS Determined by Motor Neurons and Microglia

See allHide authors and affiliations

Science  02 Jun 2006:
Vol. 312, Issue 5778, pp. 1389-1392
DOI: 10.1126/science.1123511

Abstract

Dominant mutations in superoxide dismutase cause amyotrophic lateral sclerosis (ALS), a progressive paralytic disease characterized by loss of motor neurons. With the use of mice carrying a deletable mutant gene, expression within motor neurons was shown to be a primary determinant of disease onset and of an early phase of disease progression. Diminishing the mutant levels in microglia had little effect on the early disease phase but sharply slowed later disease progression. Onset and progression thus represent distinct disease phases defined by mutant action within different cell types to generate non–cell-autonomous killing of motor neurons; these findings validate therapies, including cell replacement, targeted to the non-neuronal cells.

Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disease that selectively kills upper and lower motor neurons. Dominant mutations in the gene encoding the ubiquitously expressed superoxide dismutase (SOD1) are the most prominent known causes of inherited ALS (1). Several hypotheses to explain motor neuron degeneration have been proposed, including mitochondrial dysfunction, protein aggregate formation, excitotoxicity, axonal transport malfunction, mutant-derived oxidative damage, lack of growth factors, and inflammation (2). Ubiquitous expression of mutant SOD1 in rodents leads to a progressive selective degeneration of motor neurons because of an acquired toxic property or properties (35). However, the contribution of damage mediated by mutant SOD1 to disease onset and progression in specific cell types of the spinal cord is not established. Expression of mutated SOD1 selectively in either motor neurons (6, 7) or astrocytes (8) has failed to cause ALS-like disease in mice. Although reducing SOD1 mutant accumulation within motor neurons by viral-delivered small interfering RNA can slow disease onset (9, 10), disease progression was not affected except in one case in which it was accelerated (10). Indeed, analyses of chimeric mice composed of mixtures of normal and SOD1 mutant–expressing cells have offered evidence that motor neuron death is non–cell-autonomous, with normal non–motor-neuronal cells having the ability to reduce or eliminate toxicity to mutant-expressing motor neurons (11).

To identify which cell types are damaged by mutant SOD1 and how this damage might influence the initiation and propagation of the course of the disease, transgenic mice (LoxSOD1G37R) were generated that carried a mutant human SOD1G37R gene flanked at both ends by 34–base pair LoxP sequences allowing recognition and regulated deletion by the Cre recombinase (fig. S1A) (12). Mice developed fatal progressive motor neuron disease, including progressive weight loss from denervation-induced muscle atrophy and paralysis that was essentially indistinguishable from that seen in previously described SOD1G37R lines (5). The LoxSOD1G37R line with highest mutant accumulation reached end-stage disease most rapidly (between 8.5 and 11 months), which was accompanied by the death of 55% of spinal motor neurons (fig. S1, B to F) (12). No human SOD1 protein was expressed in any progeny from the LoxSOD1G37R females that also carried a transgene encoding the Cre recombinase in their oocytes (13), demonstrating efficient in vivo gene removal in the presence of Cre (fig. S1, G and H) (12).

To examine the contribution to the disease of mutant SOD1 toxicity within motor neurons, LoxSOD1G37R mice were mated to mice carrying a Cre-encoding sequence under control of the promoter from the Islet-1 transcription factor (14). Cre recombinase in this line is expressed in the nervous system exclusively in progenitors of motor and dorsal root ganglion neurons and was sufficient to substantially reduce mutant SOD1 accumulation in most motor axons of L5 motor roots and lumbar motor neurons of presymptomatic Isl1-Cre+/LoxSOD1G37R animals (Fig. 1, A to E, and fig. S2) (12).

Fig. 1.

Selective Cre-mediated gene inactivation shows that mutant SOD1 action within motor neurons is a primary determinant of an early disease phase. (A and B) Mutant SOD1 levels in motor axons of Lox-SOD1G37R mice in the (A) absence or (B) presence of Islet1-Cre expression. L5 ventral root sections were stained simultaneously with antibodies to (green) human SOD1 and (red) myelin basic protein. Insets show magnified images of boxed regions. (C and D) Histograms of relative intensities of mutant SOD1 fluorescence in each L5 motor axon measured from three (C) LoxSOD1G37R or (D) LoxSOD1G37R/Isl1Cre+ mice. (E) Total accumulated mutant SOD1 levels measured as relative total fluorescence intensity in motor axons from entire L5 roots in each animal (n = 3 for each genotype). (F to H) Ages of (F) disease onset, (G) progression through an early disease phase (to 10% weight loss), and (H) disease end stage of (red) LoxSOD1G37R/Isl1Cre+ mice and (blue) LoxSOD1G37R littermates. Insets show ventral root motor axons (stained with toluidine blue) in LoxSOD1G37R mice at (F) onset, (G) early disease, and (H) end stage. (I and J) Duration of (I) an early disease phase (from onset to 10% weight loss) and (J) a later disease phase (from 10% weight loss to end stage) for (red) LoxSOD1G37R/Isl1Cre+ and (blue) LoxSOD1G37R littermates. Scale bars in (A), (B), and (F) to (H), 50 μm.

A simple objective measure of the earliest onset of disease was defined by the peak of the weight curve (12, 15). This age coincides with initial axonal retraction from neuromuscular synapses but occurs before substantial axonal degeneration or loss proximally in motor roots emerging from the spinal cord (Fig. 1F, inset). An early stage of disease, accompanied by hindlimb weakness and obvious axonal degeneration (Fig. 1G, inset), was defined to be the period from onset until denervation-induced muscle atrophy decreased maximal weight by 10%. Reduction of SOD1G37R in motor neurons slowed disease onset in a minority of LoxSOD1G37R/Isl1Cre+ mice, yielding an average delay of 18 days (Isl1Cre+, 216 ± 11.2 days; Isl1Cre, 198 ± 6.1 days) (Fig. 1F). Progression from onset through early disease was delayed in all of the mice, with a mean extension of 31 days (Isl1Cre+, 95 days; Isl1Cre, 64 days) (Fig. 1, G and I); there was also a more modest slowing of later disease progression, with a mean extension of 15 days (Fig. 1J). Overall survival was extended by 64 days (Isl1Cre+, 357.5 ± 17.2 days; Isl1Cre, 293.5 ± 8.7 days) (Fig. 1H).

Microglia are the resident immune cells of the central nervous system and are the primary mediators of neuroinflammation (16). In the normal adult nervous system, these cells exist in a resting state and are characterized by a small cell body and fine ramified processes. However, neuronal damage can rapidly activate the release of cytotoxic and inflammatory mediators, including oxygen radicals, nitric oxide, and cytokines, that affect neighboring neurons and astrocytes (16, 17); in ALS, strong activation and proliferation of microglia occur in regions of motor neuron loss (18, 19). Minocycline, an antibiotic that can inhibit microglial activation (20), extends the survival of SOD1 mutant mice (2123), as does the inhibition of cyclooxygenase 2 (COX-2) (24), a key enzyme in prostaglandin synthesis. Both findings have raised the possibility of direct microglial involvement in ALS.

To test the role of the SOD1 mutant acting within microglia, we generated mice expressing Cre selectively in these cells using the promoter for CD11b (25), an integrin expressed exclusively in the myeloid lineage (26) (fig. S3A) (12). Cell-type specificity of Cre expression was verified by mating those mice to the Rosa26 mouse line that ubiquitously expresses a β-galactosidase (β-Gal) transgene that can be translated into functional β-Gal only if Cre-mediated recombination removes a premature translation terminator (27). Peritoneal macrophages and microglial cells expressed β-Gal in Rosa26/CD11b-Cre+ mice, whereas no cells of either type expressed β-Gal in animals without the CD11b-Cre gene (fig. S3, B and D to G) (12). Although both neurons and astrocytes from Rosa26 mice showed high levels of β-Gal activity after germline Cre expression (Fig. 2B), only small microglia-like cells expressed β-Gal in mice with the CD11b-encoded Cre and no β-Gal expression was detectable in neurons or astrocytes (Fig. 2A).

Fig. 2.

CD11b-Cre–directed excision of SOD1G37R exclusively in macrophage and microglial lineages. (A and B) β-Gal activity in lumbar spinal cord sections of (A) Rosa26/CD11b-Cre+ mice or (B) Rosa26 mice after systemic Cre-mediated gene excision. Arrowheads point to small, β-Gal–expressing, microglia-like cells (blue). X-Gal, 5-bromo-4-chloro-3-indolyl-d-galactoside. (C) LoxSOD1G37R transgene levels in peritoneal macrophages from LoxSOD1G37R/CD11b-Cre+ (n = 4) and LoxSOD1G37R (n = 4) mice determined by QPCR (fig. S3C). (D) Human and mouse SOD1 accumulated in peritoneal macrophages of LoxSOD1G37R/CD11b-Cre+ (n = 5) and LoxSOD1G37R (n = 5) mice, determined by immunoblotting. (E) SOD1G37R gene content determined by QPCR in DNA from microglial cells isolated from postnatal day 1 of LoxSOD1G37R/CD11b-Cre+ (n = 3) or LoxSOD1G37R (n = 2) mice. (F) Human and mouse SOD1 protein content in isolated adult microglia from the brains of 7-week-old LoxSOD1G37R/CD11b-Cre+ (n = 4) or LoxSOD1G37R (n = 4) mice. (G and H) SOD1G37R content in astrocytes isolated from the same mice as in (E), determined by QPCR (G) or immunoblotting (H). In (C) to (H), Cre+ indicates the LoxSOD1G37R/CD11b-Cre+ line and Cre indicates the LoxSOD1G37R line.

CD11b-Cre expression significantly diminished SOD1G37R accumulation in peritoneal macrophages of LoxSOD1G37R animals (Fig. 2, C and D; n = 4 or 5 per group). Quantitative real-time fluorescence polymerase chain reaction (QPCR), capable of distinguishing as small as a 20% difference in human SOD1 transgene DNA number (fig. S3C) (12), confirmed that macrophages from CD11b-Cre+ animals retained only half of the mutant SOD1G37R genes as did the Cre animals (Fig. 2C; n = 4 per group). Microglia, purified from 1-day-old LoxSOD1G37R/CD11b-Cre+ mice or their Cre littermates and then cultured for 2 weeks, showed a 25% Cre-dependent decrease in the SOD1G37R transgene levels (Fig. 2E). A similar reduction in mutant SOD1 was seen in adult microglia (Fig. 2F). Mutant SOD1 transgene content was unchanged in purified astrocytes (Fig. 2, G and H).

Microglial activation begins at or before disease onset in mutant SOD1 mice (21, 28, 29), with the number of activated cells escalating during progression, as measured with antibodies to CD11b or Iba1 (Fig. 3, A to C). No differences in microglia (Fig. 3, D to F) or astrocyte (Fig. 3, G to I) activation were observed in disease-matched LoxSOD1G37R/CD11b-Cre+ and LoxSOD1G37R mice. Nevertheless, lowering mutant SOD1 expression within microglia significantly extended the survival of LoxSOD1G37R mice, with a longer mean survival of 99 days relative to the cohort of LoxSOD1G37R littermates (Fig. 3J). Half of the Cre-expressing cohort survived more than 100 days past the mean survival of LoxSOD1G37R mice, and most of this extension was derived from slowing disease progression after onset. Indeed, although early disease progression was unchanged (Fig. 3K), the progression of later disease in CD11b-Cre+ mice was slowed by an average of 75 days (Fig. 3L). This slowing of later disease may derive in part from gene inactivation not just in microglia but also in peripheral macrophages or their progenitors and/or from the migration of those cells into the central nervous system after initial damage to motor neurons.

Fig. 3.

Selective gene excision shows that mutant SOD1 action within the microglial and macrophage lineages is a primary determinant of a late phase of disease progression. (A to C) Microglial activation in the lumbar spinal cord of a LoxSOD1G37R mouse at (B) disease onset and (C) an early disease stage (defined as 10% weight loss) compared to (A) that in an age-matched normal littermate. (D to F) Microglial activation in the lumbar spinal cord of (D) a normal mouse and of disease-matched (E) LoxSOD1G37R or (F) LoxSOD1G37R/CD11b-Cre+ mice, detected with antibodies to Iba1. (G to I) Astrocytes in the lumbar spinal cord of the same mice as in (D) to (F), detected with antibodies to glial fibrillary acidic protein (GFAP). (J) Survival times of LoxSOD1G37R/CD11b-Cre+ (Cre+) and littermate LoxSOD1G37R (Cre) mice. (K and L) Disease duration of LoxSOD1G37R/CD11b-Cre+ mice compared to LoxSOD1G37R mice for (K) an early phase or (L) a late phase of disease.

The potential for different mechanisms underlying disease initiation and progression has previously been proposed in human ALS from observations of disease spread from an initially affected region. Our use of partial, selective gene inactivation offers direct evidence for mutant SOD1 damage within different cell types to underlie an initiating phase of disease caused by mutant SOD1 damage within motor neurons, and a mechanistically divergent later phase encompassing the progression to complete paralysis that is linked to the inflammatory response of microglia and mutant toxicity within these cells. These findings have important implications for the development of successful therapies for mutant SOD1–mediated as well as sporadic ALS. Although the mechanistic linkage between familial and sporadic ALS has not been unambiguously established, the two forms of disease are clinically indistinguishable, affect the same neurons, are characterized by ubiquitinated aggregates as hallmarks, display a loss of astrocytic glutamate transporters, and are accompanied by microgliosis and astrocytosis (2). In our study, limiting mutant damage to microglia robustly slowed the disease's course, even when all motor neurons were expressing high levels of a SOD1 mutant. Thus, although the initiation of the disease requires damage to motor neurons and probably to additional cell types, disease therapy might be successful by targeting only a single non-neuronal cell type.

Supporting Online Material

www.sciencemag.org/cgi/content/full/312/5778/1389/DC1

Materials and Methods

SOM Text

Figs. S1 to S3

References

References and Notes

View Abstract

Navigate This Article