OUP user menu

Gene, Cell, and Axon Changes in the Familial Amyotrophic Lateral Sclerosis Mouse Sensorimotor Cortex

Roman M. Kassa MD, PhD, Raffaella Mariotti PhD, Marta Bonaconsa PhD, Giuseppe Bertini MD, PhD, Marina Bentivoglio MD
DOI: http://dx.doi.org/10.1097/NEN.0b013e3181922572 59-72 First published online: 1 January 2009


Lower motoneuron abnormalities have been extensively documented in the murine model of familial amyotrophic lateral sclerosis, whereas information on corticospinal neurons in these mice is very limited. We investigated 1) mRNA levels of inflammation-related molecules in the deep layers in which corticospinal neurons reside, 2) corticospinal neurons labeled from tracer injections in the corticospinal tract at the cervical level, 3) axonal damage revealed by β-amyloid precursor protein accumulation, and 4) glial cell activation in the sensorimotor cortex of presymptomatic and end-stage superoxide dismutase (SOD)-1 (G93A) mice. We demonstrated induction of inflammatory gene transcripts in the deep layers, early and progressive shrinkage of corticospinal cell bodies and activation of surrounding astrocytes and microglia with upregulation of major histocompatibility complex class I antigen. Accumulation of β-amyloid precursor protein in proximal axonal swellings indicating axonal injury was also evident at the terminal stage in the motor cortex and internal capsule. Glial and axon changes were not observed elsewhere in the cortex. These data reveal that the entire motor circuit is affected in this murine amyotrophic lateral sclerosis model as it is in human amyotrophic lateral sclerosis. Sensorimotor cortical inflammation and progressive corticospinal cell body and fiber damage may reflect transsynaptic signaling of damage from lower motoneurons.

Key Words
  • Amyotrophic lateral sclerosis
  • Astrocytes
  • β-amyloid precursor protein
  • Corticospinal tract
  • Cytokines, Gene expression
  • Microglia


Progressive loss of both cortical motor cells and lower (i.e. cranial and spinal) motoneurons is the hallmark of the neurodegenerative disease amyotrophic lateral sclerosis (ALS). Amyotrophic lateral sclerosis can be sporadic or familial (FALS); the phenotype is the same in both forms. Point mutations in the gene coding superoxide dismutase 1 (SOD1) have been identified in approximately 20% of FALS cases. Mice transgenic (Tg) for human SOD1 carrying disease-linked mutations have been extensively studied as murine models of FALS because they develop a disorder with some of the clinical and pathological features of the human disease (1-3). Murine models allow the study of neuropathological processes that precede and accompany disease progression. However, at variance with the human forms of ALS, which can start at different districts innervated by cranial or spinal motoneurons, the most widely used murine models are characterized by an ascending course of disease in which degeneration of lumbar motoneurons leads to hind limb impairment that progresses to paralysis and death (1-3). Therefore, most of the experimental studies on murine FALS have focused on the fate of lower motoneurons, and little is known about the occurrence of cortical changes in SOD1-mutant mice (2,4).

Impaired glutamate reuptake without alteration of glutamate transporters, decreased glucose utilization, and increased protein nitration and oxidation have been identified in the cortex of SOD1(G93A) mutant mice (5-7). Reduction of cortical thickness detected by neuroimaging (8) and progressive loss of corticospinal and bulbospinal neurons retrogradely labeled after tracer injection at low thoracic spinal level have also been reported in these mice (9). Loss of neurons retrogradely labeled from the lower spinal cord could, however, reflect loss of axon terminals taking up the tracer at these levels with the parent corticospinal cell bodies preserved by the axon collaterals they distribute at more rostral segments (10,11). Furthermore, in a histopathological survey, no signs of degeneration of corticospinal tract fibers were found in G93A mice, leading to the conclusion that cortical motor cells and lower motoneurons degenerate independently from each other in the murine disease (12). The degree of involvement and pathology of the corticospinal tract in the murine FALS model are, therefore, still unclear.

The occurrence of glial activation in the cortex in murine FALS is also an open question. The recent histopathological survey did not detect inflammatory changes in the cerebral cortex of SOD1(G93A) mice despite evidence of marked inflammatory pathology at subcortical levels (12); by contrast, activation of glial cells in the cortex of these mice was mentioned in another study (8). Particularly in view of the key role ascribed to glia in the disease pathogenesis, therefore, this issue requires further detailed investigation. Although recent studies have proposed that motoneuron degeneration can be caused at least in part by mutant SOD1 expression restricted to spinal neuron subsets (13,14), several investigations focusing on lumbar motoneurons have indicated that the disease onset and/or progression involves interactions between SOD1-mutant motoneurons, astrocytes, and microglia (4,15).

To address these issues, we examined the cerebral cortex of presymptomatic and end-stage G93A mice using several distinct techniques. We investigated the expression of transcripts encoding immune response molecules and glial antigens in samples of the deep layers of the sensorimotor cortex (i.e. containing layer V, in which corticospinal neurons reside). Qualitative and quantitative analyses of corticospinal neurons were performed by labeling cell bodies with retrograde tract tracing from the cervical spinal cord to visualize the entire corticospinal cell population. To investigate the occurrence of corticospinal axon damage, we used immunohistochemistry for β-amyloid precursor protein (APP), which accumulates in axon segments on fiber injury (16). The response of cortical glia to the disease was examined by immunohistochemical phenotyping of astrocytes and microglia and quantitative analyses of immunosignals. The results from these investigations indicate specific and selective changes in the sensorimotor cortex of SOD1-mutant mice.

Materials and Methods

A total of 80 mice were used in the different experimental paradigms. Forty-four mice carried a mutated human SOD1 gene (strain designation: B6SJL-TgN[SOD1-G93A]1Gur) (17); 36 were wild-type (Wt) B6SJL mice. All progenitors were obtained from Jackson Laboratories (Bar Harbor, ME). Transgenic mice were identified by a polymerase chain reaction (PCR) specific for human SOD1. Time of disease onset (which in our colony is around 90 days, consistent with data in other laboratories [3[), and its progression were monitored by careful behavioral examination. The experiments were performed when Tg animals were at either 65 to 73 or 114 to 135 days of age. The former were presymptomatic, as characterized by lack of tremors and intact extension reflex, as well as by assessment with the paw-grip endurance test, which measures how long the animal is able to hang on to a metal grid before falling down. The older animals were at an advanced stage of disease, as defined by the animals' inability to right themselves within 30 seconds of being put on their sides. The number of mice used in each procedure depended on the availability of animals at a given stage of disease (preclinical or clinical) at the time of experiment and adequacy for statistical power analysis.

Animals were maintained under controlled environmental parameters with food and water ad libitum. The experiments were performed with the approval of the institutional animal care committee and the Italian Ministry of Health, following the National Institutes of Health Guide for the Use and Care of Laboratory Animals, and in accordance with the European Communities Council Directive (86/609/EEC).

Real-Time PCR

This part of the study aimed to analyze the expression of transcripts encoding interleukin (IL)-1α, IL-1β, inducible nitric oxide synthase (iNOS), nuclear factor (NF) κB, and transcripts encoding glial antigens: glial fibrillary acidic protein (GFAP) and cluster of differentiation molecule 11b (CD11b; complement receptor 3).

Mice used for these analyses (i.e. 4 presymptomatic mice of 72-73 days of age and 3 end-stage Tg mice of 130-135 days, both groups matched with Wt littermates) were killed by decapitation, and the brains were rapidly removed from the skull. To obtain a tissue block from the deep layers of the sensorimotor cortex, a coronal slab of the brain, extending approximately from 2 to −0.5 mm relative to bregma, was first dissected out. A block was then microdissected from both hemispheres under a stereomicroscope cutting off the most superficial part of the cortex and the border with the white matter. This technique was initially validated with histological verification using Nissl staining on sections from similarly prepared tissue blocks that were found to include layers III to VI.

Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Genomic DNA was then removed from the RNA extract by means of RNase-free DNase treatment using amplification grade DNaseI (Invitrogen, Paisley, UK) for 15 minutes at room temperature, inactivated by adding 2.5 mM EDTA, and, finally, incubated for 10 minutes at 65°C according to the manufacturer's protocol. Reverse transcription of the DNase-treated RNA was performed in 20-μl reaction volumes using 250 ng of random primers, 1X RT buffer, 10 mM DTT, 0.5 mM of dNTP, and 200 U of MoMLV reverse-transcriptase (Superscript II). The cDNA synthesis was performed at 42°C for 1 hour before inactivation at 70°C for 15 minutes.

Primer pairs (Table 1) were designed with PrimerExpress 2.0 software (Applied Biosystems, Weiterstadt, Germany). Amplification, data acquisition, and data analysis were carried out in the ABI Prism 7000 Sequence Detector. Dissociation curves of PCR products were run to verify amplification of the correct product. For 25-μl reaction, 4 ng of cDNA template was mixed with 400 nM of each primer and 1X QuantiTec SYBR Green PCR Master Mix (Invitrogen). The reaction was allowed to proceed for 2 minutes at 50°C for heat activation, then for 2 minutes at 95°C for denaturation, followed by 45 cycles of 95°C for 15 seconds and 60°C for 30 seconds for annealing/extension. The comparative cycle threshold (Ct) method was used for relative quantification. Data were normalized to the levels of RNA encoding glyceraldehyde 3-phosphate dehydrogenase as housekeeping gene. Relative differences were subsequently calculated using the 2−ΔΔCt method (18), where values from Wt mice were used as control. The data (ratio of mRNA levels of each marker in sample compared with Wt to mRNA levels of glyceraldehyde 3-phosphate dehydrogenase in sample compared with Wt) were analyzed using 1-way multiple analysis of variance. Statistical significance was set at p < 0.05.

View this table:

Tracer Injection

Under deep anesthesia (chloral hydrate, 50 mg/kg, intraperitoneally), the fluorescent retrograde tracer FluoroGold (FG; Fluorochrome LLC, Denver, CO), 3% in phosphate-buffered saline (PBS; 0.01 M, pH 7.4), was injected bilaterally in the upper cervical segments of the spinal cord (C2-C3) with a microsyringe. Pilot experiments were also performed with the tracer Fast Blue (19), but FG was found to result in more widespread and brilliant labeling. A total of 6 presymptomatic Tg mice of 71 to 73 days, 9 end-stage Tg mice of 114 to 126 days, and 12 Wt mice of matched ages were used in these experiments. The injected volume (4 μl) was administered through 4 distinct penetrations; each injection was made very slowly, but leakage of the tracer was also observed. The injections were placed on both sides within the same spinal cord segment targeting the posterior columns, where the main contingent of corticospinal tract fibers is located in rodents (10, 11). This strategy was adopted to enhance tracer uptake not only from axon terminals but also from damaged fibers of passage that effectively take up this tracer (20). The wound was sutured, and the animals were allowed to survive for 6 to 8 days before perfusion. Although the tracer is rapidly transported through retrograde axonal flow, this relatively long survival time is required to obtain brilliant labeling of cell bodies with fluorescent tracers that accumulate over time in the parent neuronal perikarya (19, 20).

Animal Perfusion and Tissue Preparation for Histological Procedures

Animals used for investigation of cell types by means of tract tracing and immunohistochemistry were deeply anesthetized (pentobarbital, 60 mg/kg, intraperitoneally) and perfused transcardially with phosphate buffer (0.1 M, pH 7.4). This was followed by a fixative solution consisting of 4% paraformaldehyde in phosphate buffer. Brains were then dissected out, cryoprotected overnight in 20% sucrose, and cut on a freezing microtome. Transverse 30-μm-thick sections were collected in series of every fourth or fifth section, and those destined to immunohistochemistry were processed free-floating with different antibodies as described below. One series from each of the brains was mounted on slides and stained with cresyl violet.

To observe the retrograde labeling under fluorescence, sections from FG-injected cases were mounted, air-dried, and coverslipped with 1 drop of a fluorescence antifading mounting medium (Dako, Glostrup, Germany). These sections were used for qualitative observations and morphometric analyses (see below).


In all immunohistochemical procedures, care was taken to standardize variables that may affect the outcome of immunohistochemical reactions (21, 22), including perfusion parameters at the time of animal's death, and, during tissue processing, incubation time and temperature with primary antibody for each marker. The antibodies used are in wide use, and their specificity under similar experimental conditions has been already established. Tissue from SOD1-mutant mice was always processed together with tissue from Wt littermates, and analyses were pursued in matched experiments. In each run, control sections were processed, omitting each of the primary antibodies; no immunostaining was seen in this material. Negative controls were also obtained by replacing the primary antibody with similarly diluted normal serum from the same species.

β-APP Immunohistochemistry

One series of sections from animals destined to APP immunohistochemistry (4 presymptomatic Tg mice of 69 days of age and 4 end-stage Tg mice of 126-135 days, each group matched with 4 Wt mice of corresponding ages) were mounted on slides and air-dried for microwave tissue processing, which is required for this procedure (23). After a rapid wash in bidistilled water for 5 minutes, the slides were placed in a container containing citric acid-sodium citrate buffer (0.1 M, pH 6.0), which was then placed in a beaker with water and put in a microwave oven. The sections were microwaved at 600 W for 3 cycles of 5 minutes and then allowed to cool and were rinsed in phosphate buffer (0.2 M, pH 7.4) for 10 minutes and then 3 times in bidistilled water. Subsequently, endogenous peroxidase activity was quenched with 0.3% H2O2 for 15 minutes, and after washing in PBS, the sections were preincubated for 1 hour in 10% normal goat serum and 0.2% Triton-X-100 in PBS and then incubated overnight in anti-β-APP rabbit polyclonal antibody (Zymed Laboratories, San Francisco, CA; diluted 1:2000). This was followed by rinses in PBS and incubation in biotinylated goat anti-rabbit immunoglobins (IgGs) (diluted 1:200) for 2 hours.

The last step was the same for APP visualization and for all the other immunoperoxidase procedures used for bright-field observation (see below). The sections were processed with the avidin-biotin peroxidase protocol (ABC kit, Vectastain; Vector Laboratories Inc., Burlingame, CA) using 3,3′-diaminobenzidine as chromogen and were mounted on gelatin-coated slides, air-dried, dehydrated, and coverslipped with Entellan (Merck, Darmstadt, Germany).

AAP could not be used in combination with FG labeling because FG fluorescent labeling does not withstand the APP visualization procedure, and anti-β-APP and anti-FG antibodies are raised in the same host, thus impeding their use for double immunofluorescence.

Immunohistochemistry for Glial Cells and Major Histocompatibility Complex Class I Molecules

Seven presymptomatic Tg mice (65-69 days), 9 end-stage Tg mice (126-135 days), and 14 age-matched Wt mice were used for the study of glial cell populations in bright-field microscopy. Three series of sections from each of these animals were processed with different antibodies. Anti-GFAP rabbit polyclonal antibodies (Dako) were used to identify astrocytes; CD11b rat monoclonal antibodies (Serotec, Oxford, UK) that recognize mouse complement receptor 3, and anti-H-2Kb/H-2Db mouse monoclonal antibodies (BD Pharmingen, Erembodegem, Belgium) that react with major histocompatibility complex (MHC) class I antigens (which are expressed by microglia upon activation) (24) were used. The sections were repeatedly washed in PBS and preincubated in a solution containing 5% normal goat serum, or 3% bovine serum albumin for anti-MHC class I antibodies, and 0.2% Triton X-100 in PBS for 1 hour. They were then incubated in anti-GFAP (diluted 1:500) for 48 hours, CD11b (diluted 1:1000), or MHC class I (diluted 1:200) overnight. Subsequently, the sections were washed in PBS and incubated for 2 hours in biotinylated goat anti-rabbit IgGs for GFAP, goat anti-rat IgGs for CD11b (diluted 1:200 in 1% normal goat serum and 0.2% Triton X-100), or horse anti-mouse IgGs for MHC class I (diluted 1:200 in 1% bovine serum albumin). The sections were then reacted after the avidin-biotin protocol as outlined above.

Visualization of Glial Cells and Retrogradely Labeled Neurons With Double Immunofluorescence

To visualize simultaneously retrogradely labeled corticospinal neurons and glial cells, sections from FG-injected animals (3 presymptomatic Tg mice at 65-69 days, 3 end-stage Tg mice at 114-126 days, and 3 age-matched Wt mice per group) were processed for GFAP or CD11b immunofluorescence; immunofluorescence was used to reveal FG labeling in this material. The latter procedure was necessary to visualize FG labeling in confocal microscopy.

Anti-FG rabbit polyclonal antibody (Chemicon International, Temecula, CA) diluted 1:1000, anti-GFAP mouse monoclonal antibody (Chemicon International) diluted 1:500, and CD11b antibody diluted 1:100 were used for this procedure. Goat anti-rabbit Cy-TM3 (Fab), rat anti-mouse FITC, and goat anti-rat FITC (Jackson ImmunoResearch Laboratories, West Grove, PA) were used as secondary antibodies, all diluted 1:100.

Microscopic Examination and Analysis of Histopathological Data

Image Acquisition and Analysis

Bright-field studies of Nissl-stained sections and immunohistochemical material, and examination of single fluorescent retrograde FG labeling were performed on an Olympus BX51 microscope equipped for fluorescence and with a JVC CCD KY-F58 video camera. For quantitative analyses, images were captured at a resolution of 736 × 572 pixels and transferred to a personal computer running Image Pro Plus 6.0 for Windows (Media Cybernetics, Silver Spring, MD). A 40× objective (numerical aperture 0.75) was used for all measurements and yielded 230 × 182-μm frames, except for measurements of cortical thickness for which the 4× objective (numerical aperture 0.13) and frames of 2,333 × 1,801-μm were used.

Cortical thickness, from the pial surface to the white matter, was measured in the brains of 3 animals per group belonging to groups of Wt and Tg mice in the end-stage of disease, which had been processed histologically at the same time. The measurements were done in bright-field using Nissl-stained sections in 2 sections at the same levels from each animal on both hemispheres (at 1.10 and −0.22 mm to the bregma, according to the atlas of Franklin and Paxinos [25]). The border between the medial surface and the dorsal convexity in the motor cortex was used as upper reference point and the border with the corpus callosum as lower one, with minor modifications of a previously used method (26).

Double immunofluorescence material was analyzed with a Zeiss LSM 510 confocal laser-scanning microscope and recorded simultaneously in a dual-channel acquisition setup using 488- and 543-nm excitation beams.

Measurements of Corticospinal Neuron Perikarya

Neurons visualized by single FG labeling were analyzed in fluorescence microscopy using a wideband ultraviolet excitation filter (323 nm). For this part of the study, we sampled from all groups of FG-injected animals (n = 3 per group) 1 of 4 sections throughout the section series. The sampling began with the first appearance of FG-labeled neurons, rostrally, and a total of 19 or 20 sections per animal were collected; they typically ranged between approximately 1.94 mm rostral and 0.46 mm caudal to the bregma according to the mouse brain atlas (25). For each section, multiple images were collected bilaterally throughout the regions that contained retrogradely labeled neurons, located in layer V, to acquire as many corticospinal cells as possible. The perikaryal area of all labeled neurons with the nucleus in the focal plane was then measured by manually tracing on the screen the cytoplasm outline while leaving out the dendritic stems. The nucleus, which, in FG-labeled neurons, is unlabeled or lightly labeled (20) and in which a labeled nucleolus is sometimes evident, could be easily distinguished from the intensely labeled cytoplasm.

The mean perikaryal area of labeled neurons was evaluated, and between-group differences were analyzed with 1-way analysis of variance; significance threshold was set at p < 0.05. Histograms representing the distribution of perikaryal areas were plotted for each experimental group to analyze the cell size distribution.

Densitometric Analysis of Glia Immunosignal

Immunostaining intensity of GFAP- or CD11b-labeled cells was evaluated in end-stage Tg mice (n = 4) and Wt littermates (n = 4) under constant bright-field illumination. A total of 5 regularly spaced sections per animal within the selected stereotactic coordinates (see above) were sampled, and 1 frame placed in layer V of the motor cortex of each hemisphere (where activated glial cells were evident, see further) was acquired for each section.

The zero value of uncalibrated optical density was assigned to the background in each section, evaluated in tissue devoid of specific labeling according to a standardized previously adopted protocol (27). The data (mean value of optical density units per animal ± standard deviation) were statistically evaluated with the independent sample t-test; significance was set at p < 0.05.


Expression of Inflammation-Related Genes

To determine whether mRNA levels of immune response-related molecules in deep-layer cortex change with disease progression, we compared sensorimotor cortex tissue samples from terminal and presymptomatic SOD1(G93A) mice with samples from Wt littermates. The results are shown in Figure 1, and statistics are presented in Table 2. Among the markers investigated, we found significant upregulation of IL-1α, IL-1β, iNOS, and NF-κB, relative to Wt littermates (fold changes: 3.9, 3.8, 2.9, and 1.7, respectively) in terminal SOD1-mutant mice. On the other hand, no significant change in the relative mRNA levels of GFAP and CD11b compared with Wt mice was found, possibly because of changes in a limited population of glial cells in these areas. Therefore, these transcripts were not examined in presymptomatic mice.


Bar graphs illustrating quantitative real-time polymerase chain reaction data in samples from the deep layers of the sensorimotor cortex (see Figs. 2D-F) for transcripts encoding interleukin (IL)-1α and IL-1β, inducible nitric oxide synthase (iNOS), nuclear factor (NF) κB, glial fibrillary acidic protein (GFAP), and CD11b, in presymptomatic (white bars) and terminal (hatched bars) superoxide dismutase 1 (G93A) transgenic (Tg) mice. Results are expressed as fold-changes relative to mRNA expression in age-matched groups of wild-type mice (broken line). Details of statistical analysis showing significant increases in presymptomatic and end-stage superoxide dismutase 1 mutant mice are given in Table 2. Error bars correspond to the standard error of the mean.

View this table:

Examination of the other mRNAs in the presymptomatic SOD1-mutant mice also revealed increases in mRNA levels of IL-1α, IL-1β, iNOS, and NF-κB relative to Wt mice (fold changes: 1.6, 2.0, 1.3, 1.5, respectively) (Fig. 1; Table 2). The upregulation was statistically significant for the former two.

Cortical Histology

There was a slight decrease in measured cortical thickness in SOD1-mutant mice at the end-stage of disease compared with Wt mice (mean values ± SD: anterior level, Wt: 1,203.44 ± 104.75 μm, Tg: 1,183.33 ± 52.79 μm; posterior level, Wt: 1,042.10 ± 40.26 μm, Tg: 1,021.62 ± 41.85 μm). Given the limited sample, this set of data was not subjected to statistical analyses.

Analysis of Nissl-stained sections from SOD1(G93A) mice revealed alterations in the laminar arrangement of the sensorimotor cortex (Figs. 2B, C, E, F) compared with the Wt counterparts (Figs. 2A, D), although there was some interindividual variability. At the end stage, cortical changes particularly involved layer V with disruption of the laminar organization and disappearance of large neurons (Figs. 2C, F); these also seemed to be decreased in presymptomatic animals (Figs. 2B, E). No overt abnormalities in cortical cytoarchitecture were detected in other areas of the cortex of end-stage or presymptomatic Tg mice.


Cytoarchitecture, shown in Nissl-stained sections, of the motor cortex of wild-type mice (A, D) compared with presymptomatic (B, E) and terminal (C, F) superoxide dismutase 1 mutant mice; in low-power images (A-C), the arrows point to layer V in the areas illustrated at higher power in (D, E, F), respectively. Note the alterations in superoxide dismutase 1-mutant mice with progressive shrinkage of neurons in layer V. Scale bars: (A, applies to B and C) 500 μm; (D, applies also to E and F) 100 μm.

Corticospinal Neurons

Retrograde FG labeling of corticospinal neurons was achieved in Wt and Tg (both presymptomatic and terminal) mice (Fig. 3). This allowed observations of brilliantly FG-labeled neurons in all 3 groups (Figs. 3A-C); the labeled neurons exhibited a variety of sizes (Figs. 3D-F).


Corticospinal neurons labeled by the fluorescent retrograde tracer Fluoro-Gold (FG) in wild-type (A), presymptomatic (B), and terminal (C) superoxide dismutase 1 (SOD1) (G93A) mice. (D-F) FG-labeled corticospinal neurons in the largest size range in the 3 groups of mice. (D) Wild-type (largest neurons measuring 230-370 μm2); (E) presymptomatic SOD1(G93A) (largest neurons measuring 180-320 μm2); (F) terminal SOD1-mutant (largest neurons measuring 165-300 μm2). Scale bars: (A, applies also to B and C) 140 μm; (D, applies to all images in D-F) 70 μm.

The mean numbers of labeled neurons resulting from the sampling strategy were not significantly different in the 3 groups (Wt: n = 799, presymptomatic SOD: n = 853, terminal SOD: n = 720). The unbiased sampling performed thus suggests that there is no major loss of corticospinal neurons in the Tg mice when considering the overall population of these cells labeled from the cervical spinal cord.

The measurements of the areas of FG-labeled cell bodies, however, demonstrated alterations in both presymptomatic and end-stage SOD1(G93A) compared with Wt mice (Fig. 4). In Figure 4A, the average perikaryal size is plotted for each studied animal. Overall, there was a progressive reduction in mean cell size from Wt mice (252 ± 52.9 μm2) to presymptomatic (209 ± 37.7 μm2), to end-stage SOD1(G93A) mice (196 ± 34.9 μm2) that was statistically significant by 1-way analysis of variance (F(2) = 7.67, p < 0.05).


Quantitative evaluation of corticospinal cell size changes in superoxide dismutase 1 (G93A) transgenic (Tg) mice compared with wild-type littermates. (A) Average perikaryal area (±SD) of retrogradely labeled corticospinal neurons for each of the studied animals in the 3 experimental groups. Note the progressive reduction of perikaryal area of corticospinal neurons in the Tg mice. (B) Histograms representing the distributions of cell sizes in the 3 experimental groups. Cells measured in different animals from the same group were pooled. The histograms have been subdivided in 4 large bins to emphasize the frequency shifts between groups. Numbers inside the plots indicate the percentage of neurons observed in the 100 to 200 μm2, 200 to 300 μm2, and 300 to 00 μm2 bins, respectively. Note the shift toward corticospinal cells of smaller size in presymptomatic and end-stage Tg mice.

Because the standard deviations of the mean computed for Wt mice were consistently larger than they were in the mutant mice (Fig. 4A), we investigated the cell size distributions further. The 3 panels in Figure 4B show the population histograms for all measured corticospinal cells in each experimental group. Cell sizes were normally distributed in all groups, and the means showed the expected shift toward smaller perikaryal sizes for SOD1-mutant mice compared with Wt ones. Moreover, the histograms for the presymptomatic and terminal groups compared with the Wt group reflected the observed marked reduction in cell size variance. Taken together, these results strongly indicate that the reduction of volume of corticospinal somata of mutant mice was proportional to their initial size, with larger neurons being affected more than smaller ones. This effect can be exemplified by comparing the fraction of the largest neurons (more than 300 μm2), which represented 18.4% of all cells in Wt mice, but only 1% in presymptomatic and even less in end-stage mice (Fig. 4B).

β-Amyloid Precursor Protein Immunoreactivity

No specific APP immunoreactivity was observed in Wt and presymptomatic Tg mice (Fig. 5A), whereas axon segments with punctate spheroidal or large, elongate, and multilobulated swellings were consistently observed in layers V and VI of the motor cortex in end-stage Tg mice (Figs. 5B-G). The axonal swellings did not appear to be connected to cell bodies and involved axon segments to various extents (Figs. 5C-G). Segments of APP-immunoreactive axons and axonal swellings were also observed in the internal capsule (Figs. 5H-J). The brainstem was not examined.


Immunoreactivity for β-amyloid precursor protein in the motor cortex (A-G) and internal capsule (H-J). (A) No immunoreactivity is evident in presymptomatic superoxide dismutase 1 (SOD1) (G93A) mice. (B-G) The protein accumulates in numerous axonal swellings in layer V of terminal SOD1-mutant mice, indicating axonal injury. (C, D) At higher magnification, the area indicated by the arrows in panel B. (H-J). Axonal swellings are also evident in the internal capsule of terminal SOD1-mutant mice. Scale bars: (A, applies also to B) 140 μm; (C, applies also to D-J) 20 μm.


The immunohistochemical study of glial cells revealed distinct selective alterations in the motor cortex of the Tg mice (Fig. 6) that were not detected in other cortical areas. In the presymptomatic stage, occasional activated astrocytes were observed in layers ΙΙ and ΙΙΙ as well as layers V and VI of the motor cortex. Immunofluorescence with GFAP combined with FG labeling of corticospinal neurons showed some hypertrophic and intensely stained astrocytes intermingled with these neurons (Figs. 7B, C). Such features were not observed in Wt mice (Fig. 7A), indicating that FG retrograde labeling of neurons does not represent per se an activation stimulus for the surrounding astrocytes and microglia.


Glial immunoreactivity in the motor cortex of superoxide dismutase 1 (SOD1) (G93A) mice. (A-C) Immunostaining of astrocytes in the motor cortex of terminal SOD1-mutant mice. (B) At higher magnification, the area indicated by the arrow in panel A; note the marked activation of astrocytes in layer V. (D-I) CD11b immunoreactivity (D-F) and MHC class I immunoreactivity (G-I) of microglia in terminal SOD1-mutant mice; the areas indicated with the arrows in panels D and G are shown at higher magnification in panels E, H, and I, respectively; note the marked activation of microglial cells, well evident in layer V, with giant cell bodies and intense staining of cell body and processes. Scale bars: (A, applies also to D and G) 200 μm; (B, applies also to C, E, I) 30 μm; (F) 30 μm; (H) 60 μm.


Confocal microscopy images captured through separate channels of Fluoro-Gold (FG) retrograde labeling of corticospinal neurons revealed by immunofluorescence (red), immunofluorescent labeling of glial cells (glial fibrillary acidic protein [GFAP], labeling of astrocytes, or CD11b labeling of microglia; both shown in green). In the merged images, overlap of the 2 stains is rendered in yellow. (A) In a wild-type mouse, no GFAP staining is observed in this image from the deep cortical layers. (B, C) In a presymptomatic superoxide dismutase 1 (SOD1) (G93A) mouse, overlay (C) of GFAP immunofluorescence (B) and corticospinal neuron labeling reveals occasional activated astrocytes intermingled with the neurons. (D-F) In a terminal SOD1-mutant mouse, overlay (F) of labeled corticospinal neurons (D) and astrocyte immunofluorescence (E) reveals marked activation of astrocytes, with hypertrophic processes in very close proximity (yellow; white arrow in F) to corticospinal neurons. (G-I) At an advanced stage of disease, overlay (I) of corticospinal neuron labeling (G) and microglia immunofluorescence (H) also reveals marked hypertrophy of microglial soma and processes, intermingled with labeled neurons (yellow; white arrow in I). Scale bar: (A, applies to all images) 30 μm.

In terminal Tg mice, a band of astrocytic activation characterized by hypertrophic cells with thick processes and intense immunostaining was found in layer V of the motor cortex (Figs. 6A-C). Double-labeling studies further demonstrated hypertrophic astrocytes with thick and tortuous processes (Fig. 7E) in this layer and showed that these surrounded corticospinal neurons (Figs. 7D-F).

In presymptomatic SOD1-mutant mice, microglial cells throughout the cerebral cortex had small cell bodies and highly ramified multiple thin processes typical of resting microglia, as confirmed with immunofluorescence. In end-stage mice, microglial cells scattered in layer V of the motor cortex showed marked activation, indicated by intense staining and cell hypertrophy with retracted processes; occasional cells exhibited marked increase of the soma, appearing as giant elements (Figs. 6D-I). By double immunofluorescence, microglial cells with features of activation were seen in close proximity to and intermingled with corticospinal neurons (Figs. 7G-I).

Independent densitometry measurements of immunosignal intensity in end-stage SOD1(G93A) mice and Wt littermates (Figs. 8A, B) confirmed highly significant increases in the expression of GFAP (t(1) = 5.62, p < 0.005) and CD11b (t(1) = 9.41, p < 0.005) in glial cells of layer V of the SOD1-mutant motor cortex by sample t-tests.


Bar graphs illustrating the densitometric analysis of the immunosignal intensity of glial fibrillary acidic protein (GFAP)-positive astrocytes (A) and CD11b-positive microglia (B) in the motor cortex of terminal superoxide dismutase 1 (SOD1) (G93A) transgenic (Tg) mice (black bars) compared with matched wild-type (Wt) mice (white bars). The asterisks mark the statistical significance (**p < 0.005; independent t-test); note the significantly increased immunostaining intensity expressed as optical density (OD) for both glial markers in the SOD1-mutant mice.

MHC class I immunoreactivity was not detected in the cortex of presymptomatic Tg mice but became evident in mice at an advanced disease stage. In these animals, MHC class I-positive hypertrophic and intensely stained microglial cells with retracted processes were observed in layer V of the motor cortex (Figs. 6G-I), with some interindividual variability in their density. This pattern was not evident in other layers of the same area (Fig. 6G) or elsewhere in the cortex.


This study presents novel findings in the cortex of FALS mice consisting of molecular and cellular changes specifically related to corticospinal neurons. We provide evidence for upregulation of mRNA levels of proinflammatory mediators in the deep layers of the sensorimotor cortex, corticospinal neuron damage, and activation of surrounding glia in SOD1(G93A) mice. APP accumulation also revealed damage of presumed corticospinal axons. Altogether, the findings point out cortical involvement in murine FALS.

Inflammation-Related Transcripts

The gene expression analyses show for the first time a marked and progressive increase in the levels of IL-1α and IL-1β, as well as iNOS and NF-κB transcripts in the cortex of SOD1(G93A) mice. With multiprobe ribonuclease protection assays, it has been previously reported that spinal cord tissue samples from these mice overexpress IL-1α, IL-1β, and IL-1 receptor antagonist mRNAs at 80 days and undergo significant increases by 120 days (28). No detectable levels of mRNAs encoding IL-1β and other cytokines were found with semiquantitative real-time PCR in the brain or spinal cord of this murine model (29). Our findings based on quantitative real-time PCR analyses of cortical deep layers document changes of IL-1β mRNA, thereby paralleling for this, as well as for the other analyzed inflammatory transcripts, data previously reported in the spinal cord.

The proinflammatory mediators investigated are key molecules in neurodegenerative conditions. In particular, IL-1β increases rapidly in rodents in response to a variety of insults that lead to neuronal loss (30,31). This cytokine, as well as tumor necrosis factor−α and interferon-γ, induce iNOS transcription in astrocytes and/or microglia (32). An elevated iNOS mRNA level was found in the spinal cord of early symptomatic and end-stage SOD1 (G93A) mice (33). This NOS isoform plays a pivotal role in sustained and elevated release of the free radical nitric oxide, and iNOS immunoreactivity accumulates in lumbar motoneurons of these mice (34), further supporting evidence of iNOS induction also in neurons on proinflammatory stimuli (35). Moreover, iNOS gene deletion extends the life span of SOD1(G93A) mice, indicating that mutant SOD1 toxicity is associated with nitric oxide and peroxynitrite production (34). In cases where the human amyotrophic lateral sclerosis is sporadic, evidence of nitric oxide-mediated oxidative damage in the motor system has suggested a contribution of this mechanism to selective motoneuron death (36). The present findings, therefore, suggest that similar mechanisms may operate in the cortex in FALS mice.

All brain cells, i.e. glia, neurons, and endothelial cells can synthesize IL-1α and IL-1β; these cytokines are expressed at low levels in the healthy central nervous system and are upregulated on a variety of insults (30-32). As mentioned above, iNOS can also be induced in neurons, and NF-κB is activated in neurons at the synaptic level under physiological and pathological conditions, including acute and chronic neurodegenerative processes (37). The observed significant increase in mRNA expression of these inflammatory mediators in deep layers of the sensorimotor cortex of SOD1-mutant mice may thus reflect the synthesis by multiple cell types.

By contrast, the total amounts of GFAP and CD11b mRNAs were not significantly increased in deep-layer tissue of sensorimotor cortex of SOD1-mutant mice, but GFAP and CD11b protein immunostaining was significantly and selectively enhanced in glial cell subsets of layer V, suggesting that the upregulation of glial antigen transcripts in cell subpopulations might not have been sufficient to affect total mRNA level in the sampled tissue block.

Corticospinal Neurons

Our study provides evidence for distinct degenerative changes in corticospinal neurons of SOD1(G93A) mice. In particular, morphometric analyses of FG-labeled corticospinal cells revealed early and progressive shrinkage of neuronal perikarya.

In contrast to the reduction in the number of corticospinal neurons labeled from T12 in SOD1(G93A) mice reported in a previous study (9), we did not observe a significant decrease in Tg mice in our cell counts, although minor cell loss cannot be excluded. As mentioned previously, loss of neurons retrogradely labeled from lower spinal segments, where neurodegenerative events are very marked in these Tg mice (1-3), might reflect failure of tracer uptake due to degeneration or functional impairment of corticospinal axon terminals at these levels rather than cell death. In mice (11), as in other species (10), corticospinal fibers distribute collaterals into the gray spinal matter all along their rostrocaudal course, so that parent cell bodies reaching the lower spinal cord could be preserved by axon terminals distributed at higher levels. The present investigation favors the view that damage occurs in corticospinal neurons due to at least partial target deprivation but that these neurons survive in murine FALS. This is also supported by data showing that degeneration in the lumbar spinal cord of SOD1(G93A) mice involves not only motoneurons but also interneurons of the intermediate zone (34), where corticospinal fibers terminate in rodents (10,11). Spinal interneurons degenerate even before motoneurons (34), which is likely the cause of degeneration of afferent terminals, as already described in SOD1-mutant motoneurons (38,39).

Analyses in human ALS cases have reported that upper and lower motoneurons are significantly shrunken compared with controls (40), whereas no change in neuronal perikarya was found in another investigation on the motor cortex of ALS cases (41). Variations in parameters measured, region analyzed, and differing measuring techniques could account for discrepancies in human studies, but corticospinal neuron degeneration is considered a hallmark of ALS in humans (2). The present data indicate that SOD1(G93A) mice reproduce, in part, at the cortical level the pathological features documented in human ALS victims. In support of these changes, selective alterations have also been identified in the spatial distribution and number of inhibitory interneurons in the sensorimotor cortex of FALS mice (42). Vacuolation of mitochondria and inclusions resulting from abnormal protein aggregation and ubiquitination that have been described in spinal motoneurons in murine FALS (1,3) remain to be investigated at the cortical level. The early and progressive shrinkage of corticospinal neurons we observed, however, likely is the morphological correlate of the cytotoxic properties of mutant SOD1 that have been demonstrated repeatedly in subpopulations of spinal neurons (3,14,34).

Species differences in the organization of the corticospinal tract should also be taken into account in the interpretation of cortical changes in murine FALS. In part, this tract also directly reaches motoneurons in primates, including humans, but not in rodents (10,11). Therefore, cortical changes in murine FALS may not reproduce the full spectrum of cortical abnormalities human ALS.

A central issue in the pathogenesis of ALS is the question of primary or secondary involvement of cortical motor cells. Although the present data cannot disentangle this problem, corticospinal neurons exhibited shrinkage at the earliest time point we examined (around 70 days), long before the onset of clinical disease. The earliest time point at which motoneuron loss has been documented in the lumbar spinal segments of SOD1(G93A) mice is around 60 days (1,3) and is preceded by very early loss of neuromuscular junctions (3). This has led to the hypothesis of a dying-back pathology (43). The cortical alterations could thus be related to transsynaptic retrograde transfer of signals from the spinal cord and indicate that the entire motor circuit is affected in murine FALS.

Damage of Cortical Axons Revealed by β-APP Accumulation

This study provides evidence for injury of axons, which, on the basis of their distribution, are likely to derive from layer V cell bodies in terminal Tg mice. Axonal damage was also detected in the internal capsule, further indicating damage of corticospinal fibers.

Previously, no evidence of degeneration of corticospinal fibers was found with Gallyas myelin staining in terminal SOD1(G93A) mice (12), but silver impregnation may not be sufficiently sensitive to reveal damage in small contingents of corticospinal fibers that we visualized by the present strategy. APP, a transmembrane precursor of β-amyloid protein, is normally present in neurons and transported by fast anterograde axonal transport (44), but its constitutive levels are not detectable by standard immunohistochemical techniques in formalin-fixed tissue (45). At sites of axonal injury, APP accumulates, reaching the threshold for visualization, due to failure of axonal transport (45). The present finding of APP accumulation is also consistent with previous reports of axonal flow abnormalities in murine FALS (3,46). On the other hand, in view of the relatively long postinjection survival time and consequent tracer accumulation, the present experiments based on FG retrograde axonal transport would not be suited to reveal axonal flow alterations.

In human ALS, APP immunoreactivity was found to be increased in perikarya and proximal axonal swellings of anterior horn neurons in cases with short clinical course or with mild depletion of anterior horn neurons, suggesting an early and short-lived change (47). Interestingly, APP staining does not necessarily imply neuronal death (48), but its present evidence in terminal SOD1(G93A) mice further indicates damage of corticospinal neurons. The finding of fiber degeneration at a very advanced disease stage in our study supports the occurrence of a slow and progressive damage of the corticospinal tract in murine FALS.

Activation of Astrocytes and Microglia

The inflammatory state in the motor cortex of SOD1-mutant mice we documented is consistent with data in humans in which a marked glial reaction typically accompanies damage of both upper and lower motoneurons in ALS (49). Interestingly, we observed mild activation of astrocytes in the motor cortex of presymptomatic Tg mice, which was then marked in the terminal stage of disease, when also microglia were highly activated at the same sites. In the motor cortex of SOD1(G93A) mice, increase in the number of microglial cells and a trend toward an increase of astrocytes were previously mentioned at the end stage (8), but no glial cell changes were detected in the cortex in another study (12). The present qualitative and quantitative approach has revealed selective activation of astrocytes and microglia intermingled with corticospinal neurons.

Astrocytes carrying a human SOD1(G93A) mutation exert a toxic effect and release factors selectively detrimental to motoneuron survival in vitro (50-52). In vivo data in the spinal cord of SOD1(G37R) mice also suggest that the SOD1-mutant action in astrocytes determines the timing of microglial activation and infiltration (53). The earliest time point at which astrocytic activation was detected in the lumbar spinal cord of SOD1(G93A) mice was 47 days (43), whereas microglial activation occurs later on (54). Astrocyte activation in the motor cortex of SOD1(G93A) mice is reported herein around 2 months of age (i.e. the earliest time point we investigated); microglial activation was detected at an advanced disease stage. Altogether, the findings thus suggest that the activation of glia surrounding corticospinal neurons and in the spinal cord may have a similar temporal pattern.

Concluding Remarks

The parameters analyzed in the present study reveal alterations in the sensorimotor cortex of FALS mice. Our gene expression findings support the view that induction of proinflammatory cytokines and oxidative stress at cortical level participate in a cascade of events affecting corticospinal neurons. Cortical cell changes involve corticospinal perikarya and the surrounding glial cell populations and are accompanied by axonal damage. Such data support a distinct involvement of corticospinal neurons in the murine disease and indicate that the neuron-glia crosstalk documented at spinal level in FALS mice also operates at the cortical level.


The authors are very grateful to Dr Diego Minciacchi for his help in initial tract tracing experiments and to Dr Giorgio Malpeli for his help with real-time polymerase chain reaction data.


  • This work was supported in part by the grants CariVr 2002 of the “Fondazione Cassa di Risparmio di Verona Vicenza Belluno e Ancona,” FIRB 2002 (No. RBNEO1B5WW), and COFIN 2005 (No. 2005057598) from the Italian Ministry of University and Research.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
View Abstract