Characterization of early pathogenesis in the SOD1(G93A) mouse model of ALS: part II, results and discussion

Abstract

Pathological events are well characterized in amyotrophic lateral sclerosis (ALS) mouse models, but review of the literature fails to identify a specific initiating event that precipitates disease pathology. There is now growing consensus in the field that axon and synapses are first cellular sites of degeneration, but controversy exists over whether axon and synapse loss is initiated autonomously at those sites or by pathology in the cell body, in nonneuronal cells or even in nonmotoneurons (MNs). Previous studies have identified pathological events in the mutant superoxide dismutase 1 (SOD1) models involving spinal cord, peripheral axons, neuromuscular junctions (NMJs), or muscle; however, few studies have systematically examined pathogenesis at multiple sites in the same study. We have performed ultrastructural examination of both central and peripheral components of the neuromuscular system in the SOD1(G93A) mouse model of ALS. Twenty percent of MNs undergo degeneration by P60, but NMJ innervation in fast fatigable muscles is reduced by 40% by P30. Gait alterations and muscle weakness were also found at P30. There was no change in axonal transport prior to initial NMJ denervation. Mitochondrial morphological changes are observed at P7 and become more prominent with disease progression. At P30 there was a significant decrease in excitatory axo-dendritic and axo-somatic synapses with an increase in C-type axo-somatic synapses. Our study examined early pathology in both peripheral and central neuromuscular system. The muscle denervation is associated with functional motor deficits and begins during the first postnatal month in SOD1(G93A) mice. Physiological dysfunction and pathology in the mitochondria of synapses and MN soma and dendrites occur, and disease onset in these animals begins more than 2 months earlier than originally thought. This information may be valuable for designing preclinical trials that are more likely to impact disease onset and progression.

Keywords: Axons; NMJs; cytoplasmic vacuoles; glia; mega-mitochondria; mitochondria; motoneurons; motor function.

Figure 1

Motoneurons in the TA and soleus motor pools were identified by fluorescent CTB retrograde transport that was injected at P30 and the retrogradely labeled MN soma area was determined at P34. Both SOD1 motor pools were significantly smaller as compared to WT. The number of animals for each condition is indicated in the bars of the graph. **P ≤ 0.01; statistical significance determined by one-way analysis of variance (ANOVA) followed by Bonferroni's multiple comparison test.

Figure 2

Motoneurons contain numerous cytoplasmic vacuoles that are an early sign of impending degeneration. (A) Photomicrograph of 1 μm section through P60 lateral motor column of SOD1^G93A mice. MNs are easily identified by their large size, nucleus, and prominent nucleolus (arrows). Many MNs contain numerous cytoplasmic vacuoles that can be viewed by light microscopy (arrowheads). (B) MNs were counted using well-established criteria that have been previously described (Clarke and Oppenheim 1995). At P60 in the SOD1^G93A mouse spinal cord, however, many MNs meet some or all of these criteria even though they contain numerous cytoplasmic vacuoles that are an early sign of impending degeneration viewed by light microscopy. However, if the presumptive degenerating MNs containing cytoplasmic vacuoles are included in cell counts, total numbers are comparable to WT. If vacuolated MNs are excluded from the counts, then there is a 20% decrease at P60 and a 30% decrease at P70 in mutant spinal cords. By P115–140, there are few remaining vacuolated MNs in mutant mice and the number of surviving MNs at this age is reduced by approximately 50%. The number of animals for each condition is indicated in the bars of the graph; *P ≤ 0.05; **P ≤ 0.01; statistical significance determined by t-test with Bonferroni correction.

Figure 3

Ultrastructural analysis reveals profound mitochondrial and cytoplasmic pathology by P75. At P75, there is no cytoplasmic pathology in SOD1 γMNs (A; area = 225 μm²) but profound pathology in the cytoplasm of αMNs in the same section (B; area = 2414 μm²), which can be distinguished on the basis of their size. Arrows in (B) point out areas enlarged in a and b, where one can see swollen and vacuolated mitochondria (m) and accumulated vacuoles; note that rough endoplasmic reticulum (rer) appears normal. (C) There is also extreme disruption in axons at P75 (Ax), which are often engorged with large vacuoles (*). Dendrites d also exhibit similar vacuoles (arrows), in this case resulting from an apparent distension of the outer mitochondrial membrane. (D) Cytoplasm of αMN are filled with huge vacuoles (*) that appear to reflect distended outer mitochondrial membranes as seen more clearly at arrows. Representative images are shown. Three animals of each genotype were examined.

Figure 4

Interneurons also undergo degeneration in SOD1^G93A spinal cord. Because we observed mitochondrial swelling and vacuolization in presynaptic terminals on MN soma and dendrites (see Fig. 3), we also counted interneurons to determine if they are decreased in mutant mice. At P75, the number of interneurons in the SOD1 lateral motor column was significantly decreased versus WT. The number of animals for each genotype is indicated in the bars of the graph; *P ≤ 0.05 Statistical significance determined by t-test with Bonferroni correction.

Figure 5

Ventral roots atrophy, but absolute number of axons are not reduced by P75. Ventral root axons from L3, L4, and L5 segments of spinal cord were counted to determine if there was a degeneration of ventral roots associated with the apparent degeneration of MNs and interneurons at day 75. There was no significant difference in the area of (B) or number of (C) axons in SOD1 versus WT ventral roots (t-test with Bonferroni correction). Many axons in the SOD1 mouse, however, exhibited alterations indicative of ongoing or impending demyelination/degeneration (see enlarged boxed regions in A). n = 6 animals/genotype.

Figure 6

Denervation in SOD1^G93A FF muscles (TA) occurs between P14 and 30. (A and B) Photomicrographs are shown of P30 TA muscles from WT (A) and SOD1^G93A (B) mice. Alexa fluor 555-α-BTX was used to identify postsynaptic terminals (red) and antibodies to VAChT and neurofilament (NF) were used to identify presynaptic terminals and axons, respectively, and were visualized with Alexa fluor 488 secondary antibodies (green). NMJs that exhibited an overlap of red and green were considered innervated, while those that exhibited only α-BTX expression were considered denervated. Note the obvious denervation in the SOD1^G93A TA in B. (C) Muscle innervation was examined in the P14, P30, and P100 TA in SOD1^G93A and WT mice. While significant denervation occurs at P30, there is no denervation at P14, but by P100, 70% of TA NMJs were denervated. The results are presented as % denervated of total NMJs/muscle (mean ± SEM). (D) Muscle innervation was examined in the P30 and P100 soleus in SOD1^G93A and WT mice. There was little denervation in SOD1 animals even at P100. The number of animals for each condition is indicated in the bars of the graph; *P ≤ 0.05 as compared to WT as determined by unpaired T-test.

Figure 7

Loss of axons in intramuscular nerve branches in the TA muscle of SOD1 mutant mice at P30. Intramuscular nerve branches (black axon bundles) and postsynaptic sites (acetylcholine esterase, blue) were labeled, and axons in intramuscular nerve branches (arrows) that could be followed to NMJs were counted as described in the methods. A total of 22–25 nerve branches were counted in each muscle. Results are expressed as mean number of axons in intramuscular nerve branch ± SEM. The number of animals for each condition is indicated in the bars of the graph; *P ≤ 0.05 as compared to WT as determined by unpaired T-test.

Figure 8

Endplate morphometry (form factor) was assessed for NMJs (postsynaptic a-BTX-positive endplates; A) in tibialis anterior of wild-type and SOD1 mice at P14 (B) and P30 (C). In both cases there is a shift to the right that indicates that in SOD1 mice, endplates have a less elongated shape than in WT animals. There was a significant difference between SOD1 and WT TA at P30 (P ≤ 0.01) and at P14 (P ≤ 0.001). There was no difference between SOD1 and WT soleus (not shown). n = 4 animals/genotype/age; 90 NMJs/muscle/animal were examined. Statistical significance was determined by Mann–Whitney test.

Figure 9

Retrograde transport in MNs innervating the tibialis anterior (TA) and soleus muscles was examined in mice at P20 using Alexa Fluor®555 and ®488 conjugated with cholera toxin B-subunit (CTB), respectively. (A) There was no statistically significant difference in retrograde transport in MNs innervating the TA (A) or soleus (B) between SOD1^G93A and WT mice. Individual time points are shown with symbols and a best-fit line was drawn. Three to eight animals/genotype were examined at each time point. Statistical significance was determined by ANOVA followed by Tukey–Kramer post hoc test.

Figure 10

At P30, mitochondria (arrows) in intramuscular axons of all SOD1 muscle types examined had swollen mitochondria as compared with WT (A), although this effect most prominent in outer TA (B). A node of Ranvier (N) can be seen in (A), while myelin sheath aberrations are in SOD1 material (arrowheads in B). (C) On P53, axons in TA intramuscular axons exhibit frank degeneration (*) as well as myelin sheath aberrations. Representative images are shown. Four animals of each genotype were examined.

Figure 11

(A and B) Normal NMJ appearance in the inner TA (types IIa and IIb) and outer TA (primarily type IIb), respectively in P30 animals. Arrows point to normal compact terminal mitochondria. (C) In SOD1 animals, alterations in inner TA, enlarged in a and b, while (D) shows outer TA, enlarged in a'. Asterisks in a and a' are inside of large vacuoles (>100 nm diameter), a common feature in SOD1 animals. Arrows point out swollen mitochondria, and arrowheads point out abnormalities that appear autophagic-like. Double arrows in (E) point to abnormal whorls in a terminal from the outside portion of the TA. Overall, abnormalities of all types were seen 2–3 times more frequently in SOD1 than WT. Representative images are shown. Five animals of each genotype were examined.

Figure 12

P30 SOD1 soleus NMJ exhibit slightly swollen mitochondria in the terminal (m in A) compared with WT (B), although mitochondria in the muscle fibers (mf) and sarcoplasm (arrowheads) remain normal. Numerous junctional folds are present. Note the occasional abnormality (arrows) in A. Magnification marker = 1 μm for both A and B; N = nucleus of muscle cell. Representative images are shown. Five animals of each genotype were examined.

Figure 13

Fewer, but larger mitochondria are present in presynaptic NMJs from SOD1 animals versus WT. The number and area of mitochondria was determined as described in Materials and Methods. (A) The absolute number of mitochondria is decreased in SOD1 versus WT presynaptic NMJs. (B) While there is a decrease in the number, there is an increase in the area of mitochondria in P30 SOD1 versus WT. Results are expressed as mean ± SEM. Thirty NMJs/animal were examined (n = 5 animals/group). Significant difference between groups was determined using unpaired t-tests. *P ≤ 0.05; **P ≤ 0.01.

Figure 14

(A) On P53 presynaptic terminals of NMJs in the outer component of the TA show advanced degeneration. Three areas of the terminal are enlarged in a, b, and c: (a) illustrates a region of the presynaptic terminal that contains vesicles (v), but has a large autophagic-like body (arrowhead); (b) illustrates a region of the same terminal that is completely devoid of vesicles (*); (c) illustrates another portion of the terminal that contains vesicles that appear normal. (B) Other presynaptic terminals in this region have a range of appearances and include some that are relatively normal except for swollen mitochondria (arrows). (C) Presynaptic terminals in the inner portion of the TA of SOD1 animals show less severe effects, and often appear normal except for swollen mitochondria (arrows). (D) A NMJ from the outer portion of the TA from a WT animal shows normal morphology of synaptic vesicles (v) and mitochondria (arrows). n = 4 animals for SOD1 and four animals for WT.

Figure 15

At P7, SOD1 TA NMJ often show slightly swollen mitochondria as compared with WT (A = WT; B = SOD1; arrows). The NMJ presynaptic terminal is shaded gold and lies between a terminal Schwann cell (SC) and the postsynaptic muscle (M). Representative images are shown. Three animals of each genotype were examined.

Figure 16

Enlarged mitochondria and cytoplasmic vacuoles are observed in P30 SOD1 MNs. At P30, αMNs have a similar appearance in both WT (A) and SOD1 (B) animals. However, upon closer inspection, mitochondria (arrows in C–F) are larger in SOD1 (D) as compared with WT (C). There is also an accumulation of numerous small vacuoles in the cytoplasm of SOD1 MNs (* in D and F). In the neuropil, these effects are more striking, as seen in SOD1 (F) versus WT (E). Axons (Ax) and dendrites (D) contain quite swollen mitochondria (arrows) and larger, more numerous vacuoles (*), as well as mega-mitochondria (mg). We examined ultrastructural changes in MNs in the L3–L4 segments of the spinal cord at day 30 of SOD1^G93A mice (n = 5, total of 38 MNs) and compared these findings with age-matched wild-type (WT) animals (n = 5, total of 48 MNs).

Figure 17

Distal dendrites in SOD1 MNs exhibit vacuolated mitochondria and large cytoplasmic vacuoles, and axons have fragmented myelin sheaths. Synapse types on MN distal dendrites (d) at P30 in white matter adjacent to VH showed a decrease in type I (r) synapses similar to that seen for MNs, with no significant decrease in type II (p). Note distension of mitochondrial cristae (arrows) in SOD1 (B) compared with WT (A) in both axons (Ax) and dendrites, as well as poor condition of myelin sheath (arrowheads) and presence of large vacuoles (*) in SOD1 (n = 5 animals each, WT and SOD1). Double arrows in D point to large autophagic-like bodies in dendrites. For distal dendrites, using the same levels as used for MN soma evaluation in Figure 16, 20–30 dendrites with synapses from each animal used above were located in the ventro-lateral white matter and photographed at 16,000×. Only synapses with a clear synaptic density and presynaptic vesicles were scored.

Figure 18

Fewer, but larger mitochondria are present in MNs from SOD1 animals versus WT. The number and area of mitochondria was determined as described in Materials and Methods. (A) There is a decrease in the number of mitochondria in P30 SOD1 MNs versus WT MNs. (B) While the number of mitochondria was decreased in SOD1 MNs, there was an increase in the number of larger sized of mitochondria in SOD1 versus WT MNs (B). Mitochondria from the MNs examined in Figure 16 were examined (n = 5 animals/group for a total of 38 MNs in SOD1 and 48 MNs in WT). Significant difference between groups was determined using unpaired t-tests (*P < 0.05; **P < 0.001).

Figure 19

Illustrations of synapse types in WT (A–C) and SOD1 (D–G) MNs. C-terminals, which are restricted to αMNs and are characterized by subsynaptic cisterns and organelles (arrows in A, G) and contain irregularly shaped and densely packed vesicles, showed classic appearance in both WT (A) and SOD1 (D) mice; however, some C-terminals in SOD1 spinal cord were atypical, with smaller areas and less distinctive vesicle density (G, H). Typical type I (R) and type II (P) synapses were observed on WT (B, C) and SOD1 (E, F) MNs, although terminals in SOD1 often exhibited swollen mitochondria (arrows, E and F, and mega-mitochondria, mg) and vacuoles (*) both pre- and postsynaptically. Arrowheads point to synaptic densities; mn, motoneuron; mag marker = 500 nm.

Figure 20

There is a decrease in type I and increase in C-type synapses on SOD1 MNs. Synapses meeting ultrastructure criteria for type I, type II or C-type were counted on MN soma (A) and distal dendrites (B). The total number of synapses was not different between WT and SOD1; however, there was a significant decrease in type I and increase in C-type axo-somatic synapses on SOD1 versus WT MNs. There was a decrease in the number of total axo-dendritic synapses on distal dendrites. There was a significant decrease of type I synapses on distal dendrites. Axo-somatic synapses on the MNs examined in Figure 16 were examined (n = 5 animals/group for a total of 38 MNs in SOD1 and 48 MNs in WT). For distal dendrites, using the same levels as use for MN evaluation, 20–30 dendrites with synapses from each animal were located in the ventro-lateral white matter. *P < 0.05; a mixed models approach was initially used to identify differences between WT and SOD1 while accounting for the repeated measures within each mouse and correlation between measures within each mouse. Statistical difference was determined using a least square means table.

Figure 21

Mega-mitochondria are prominent at P7 and P14. (A and B) Images from the P7 ventral horn of SOD1^G93A mice show mega-mitochondria (MG, double arrows) in both dendrites (D) and MN soma (MN). Single arrows indicate normal mitochondria. (C and D) Mega-mitochondria (MG), slightly swollen mitochondria (arrowheads) and normal mitochondria (single arrows) are present in dendrites and MNs of P14 mutant spinal cords. Mega-mitochondria were not observed in WT spinal cord. Representative images are shown. Three animals/genotype/time point were examined.

Figure 22

Glia cells do not exhibit cytoplasmic abnormalities. (A and B) At P30, both astrocytes (A) and oligodendrocytes (O) appear to have normal mitochondria (arrowheads) and cytoplasm, although swollen mitochondria can be seen in surrounding neuropil. (C and D) At P110 (n = 5 animals/genotype), an astrocyte (A) still shows normal ultrastructural elements, though increased filament bundles (arrows) indicate activation. The oligodendrocyte (O) in D likewise has normal elements and mitochondria (arrowheads). Although not quantified, at P110, oligodendrocytes appear to be more abundant. In both C and D, some ghost-like remnants of dendrites (d) and axons (a) are present. MN, motoneuron; scale bar = 1 μm.

Figure 23

SOD1^G93A mice exhibit deficits in motor function that correlate with early muscle denervation. (A) Schematic of forelimb and hindlimb stance width in WT and SOD1^G93A mice walking 40 cm/sec up an incline (∼15 degrees). Forelimb stance width is typically more narrow than hindlimb stance width in B6 mice, common to both WT and mutant mice. Narrowing of stance width, a robust and significant phenotype by 7 weeks of age, was evident in SOD1^G93A mice by P32. Data were pooled from two cohorts of mice (n = 14 per group) and are expressed as mean ± SEM. #P < 0.06; *P < 0.05; **P < 0.01. (B) The step-to-step placement of the paws during peak loading of the hindlimbs was significantly more variable in SOD1^G93A mice than in WT mice, suggesting aberrant motor function control of the ankles (gastrocnemius and tibialis anterior muscles). *P < 0.05. ANOVA was used to test for statistical differences among WT and SOD1 G93A mice at each age. When the F-score exceeded Fcritical for α = 0.05, we used post hoc unpaired Student's two-tailed t-tests to compare group means. Gait indices between forelimbs and hindlimbs within groups were compared using paired Student's two-tailed t-tests.

Figure 24

Performance of WT and SOD1 mutant mice on the loaded grid test. Performance is based on the duration of time in seconds (sec) before the loaded grid was dropped. (A) Each mouse was tested twice with a 15 g weight and allowed unlimited time before dropping the weight with a 10-min intertest interval. The values represent the average of two trials/mouse. Results are expressed as mean ± SEM (n = 10 animals each phenotype). *P < 0.02, statistical significance was determined using unpaired t-tests. (B) Performance of WT and SOD1 mutant mice. Each mouse was tested twice with each weight for a 30-sec fixed duration with 15 sec between each weight and 10 min between the two tests. The values represent the average of the two tests/mouse. Results are expressed as mean ± SEM; the number of animals for each condition is indicated in the bars of the graph. *P < 0.05 as determined using unpaired T-test.

Figure 25

(A) A summary of pathological events in central and peripheral components of the neuromuscular system of SOD1^G93A mice and the time of their appearance is shown (see accompanying paper (doi: 10.1002/brb3.143) for description of pathology in the spinal cord). Triangles show either increases or decreases in the pathology over time. (B) A proposed sequence of initial pathology in MNs in SOD1^G93A mice is shown. Altered synaptic input either alone or with a perturbed capillary supply and extracellular stress results in an imbalance of Ca²⁺ within the MN (1). This imbalance results in intracellular generation of reactive oxygen species (ROS) and together with the presence of the mutant SOD1 protein creates an environment that results in the formation of mega-mitochondria (2). The mitochondria are not able to resolve the “stress,” possibly because of their impairment by mutant SOD1, and thus become vacuolated (3). The imbalance of intracellular Ca²⁺ and generation of ROS can also initiate the unfolded protein response and ER stress resulting in vacuolization of smooth ER and Golgi (4). Together, these events result in MN dysfunction. MNs innervating fast muscle fibers also encounter events at the NMJ including extracellular stress and ROS, fast fiber-specific physiology and possibly a reduced supply of trophic factors or increased toxic factors that lead to muscle denervation. The combination of the denervation and MN dysfunction eventually leads to MN degeneration and death. While not shown in the diagram, the events contributing to MN dysfunction in the spinal cord most likely contribute to glial cell activation and/or dysfunction (e.g., decrease in astrocyte glutamate transporter) that further enhances disease pathology. With the exception of the vascular system, also not shown here are noncell (MN) autonomous contributions that may contribute to disease onset (e.g., astrocytes, Schwann cells).