This study shows how the over-expression of the G93A-SOD1 gene mutation modifies the molecular response to a mild mechanical compression in the spinal cord tissue from a pre-symptomatic rat. Using a histopathological assessment of the compression injury, we have found no significant differences between the WT and the G93A-SOD1 animals in the post-injury immunoresponse or in the amount of spared white matter following the trauma (Figures 10 - 11). Hence, despite delivering comparable spinal cord compression injuries to pre-symptomatic WT and G93A-SOD1 rats, G93A-SOD1 rats displayed a significantly lower level of post-injury locomotor recovery than their WT littermates. We observe that the significant differences in the post-injury locomotor recovery between the WT and the G93A-SOD1 rats are not due to differences in the induction of the injury, but rather to an intrinsically different molecular response to trauma in the two genetic variants of rats. A significantly low level of functional recovery has also been reported in G93A-SOD1 transgenic mice after sciatic nerve injury, with an acceleration of the disease progression so that 90 day old mice show deficits that are normally only seen at the end stage in uninjured G93A-SOD1 mice .
Whilst mutations of the SOD1 gene account only for approximately 2% of the whole ALS population, other not yet identified predisposing or causative genetic factors may act along the same lines, reducing the neuroregenerative potential and escalating the level of disability after trauma. These observations may partially explain the observed higher occurrence of ALS cases among individuals that have been exposed to repetitive mechanical trauma or that have been subjected to other forms of mechanical stress or surgical procedures [2, 3, 5].
For the purpose of our study, we have employed a well-characterised experimental platform of mechanical stress, which is known to cause only a mild damage to the spinal cord and to be associated with a good outcome in terms of post-traumatic locomotor recovery in rodents [29, 19, 30]. Other forms of injury cause a significant level of tissue destruction and a more florid profile of gene expression changes, masking subtle transcriptional changes that may be induced by the SOD1 gene mutation in an otherwise macroscopically healthy spinal cord. Under the experimental conditions of mild SCI, the G93A-SOD1 spinal cord displays a higher number of up-regulated genes compared to WT tissue at the 4 hour post-injury time point (Figure 2A-B). These include a broad range of genes involved in the modulation of macrophages, Toll-response, T and B-cell proliferation and in the production of interleukin-6 and TNF-alpha, featuring an early inflammatory response that surfaces only at a later time point in WT spinal cord (Figure 3A-B). Other gene expression changes unique to the G93A-SOD1 animal are detected in two later time points within the experimental period. Following the 4-hour time point, the G93A-SOD1 spinal cord displays characteristic apoptotic signals, like the activation of genes regulating cytochrome C release from the mitochondria. Cytochrome C release is known to occur when the inner mitochondrial membrane becomes excessively permeable to ions, leading to energy failure and apoptosis [31, 12]. The inflammatory milieu and the activation of cell-death signals in the spinal cord at this stage of the animal recovery may reduce the tissue repair potential contributing to the observed poor functional recovery observed in the G93A-SOD1 rat.
The post-injury molecular profile of the G93A-SOD1 spinal cord does not show the down-regulation of a large number of genes 7 days from compression SCI demonstrated in WT rats compared to naïve controls (Figures 2A-B; 3D). For example, those genes that control cholesterol and isoprenoid biosynthesis remain up-regulated in the G93A-SOD1 spinal cord at this time-point after compression injury. Isoprenoids are biological precursors of carotenes and of coenzyme-Q, retinol and tocopherol (vitamin E), compounds known to modulate mitochondrial activity and oxidative stress [32, 33]. A previous report has documented a state of systemic dyslipidemia in individuals with ALS and this state of altered lipid homeostasis has been linked to a potential neuroprotective effect . The presence of a SOD1 gene mutation or of a genetic predisposition to develop ALS may be associated to a state of dysregulation of molecular pathways of cholesterol and isoprenoid biosynthesis, both during the natural development of the disease and in response to a stressful situation like a mechanical injury.
Also down-regulated in WT spinal cord at the 7-day post-injury time point are many genes that like Hcn2 control cell-cell signalling and ion/neurotransmitter transport (GO:0019226; GO:0006836, Figures 3D and 6). Previous experimental observations in animal models of ALS and brain ischemia have reported increased neuronal excitability in affected tissues, generated by the up-regulation of genes capable of modulating ion currents [3, 13]. Hence, the acquisition of a state of tissue hyperexcitability may represent a distinguishing feature of SOD1-induced ALS, both during the natural development of the disease and also in a pre-symptomatic phase as a result of a stressful condition.
At the 7 day time-point, both the WT and the G93A-SOD1 spinal cords up-regulate genes involved in retinol (vitamin A) metabolism including RBP1 and CRABP2, whilst only the G93A-SOD1 spinal cord tissue presents the additional up-regulation of ADH1 and ALDH1 (Figure 4B). A link between the vitamin-A down-stream effects and the pathological changes observed in ALS has already been suggested [35–38]. Early vitamin-A deprivation, for example, causes motor cell loss in rodents . RBP1 and CRABP2 are over-expressed in spinal cord from the end-stage G93A-SOD1 rat, whilst surviving small motor neurons show selective immunostaining for specific retinoid receptors [36–38]. Our observations in the acute post-injury phase may inspire potential neuroprotective therapeutic strategies, since both the alteration of ion current regulation and the activation of retinoid signalling can be pharmacologically modified.
We and others have already reported the post-injury down-regulation and/or accumulation of phosphorylated and non-phosphorylated neurofilaments and of synaptophysins (SYN), the latter appearing to be linked to post-traumatic motor and cognitive deficits in different models of neurotrauma [24, 25, 6, 39]. Given the reported trauma-induced differential regulation of Nfh and SYN expression along the motor cell/axon/synapsis axis, we have evaluated the Nfh and SYN differential regulation in our injury model. At a RNA level, Nfh expression appears to be higher in injured G93A-SOD1 spinal cord compared to injured WT tissue from the 4 hour time point onward (Figure 4A). Post-injury protein expression of Nfh measured by immunohistochemistry is also tendentially higher in the G93A-SOD1 spinal cord at all the time points, with a significant level of differential expression at the 24 hour time-point, which was further confirmed with Western blotting (Figures 4, 5, 6, 7, 8). This observation may indicate that neurofilaments expression is altered in a situation of genetic susceptibility to develop ALS, both under conditions of stress and during the natural development of the disease. Intermediate neurofilaments like peripherin within spheroid-like structures are already known to accumulate in affected tissues from animal models of ALS and to possibly interfere with axonal transport . We did not observe any significant differential regulation or any genotype-specific redistribution of SYN in spinal cord after compression SCI. Failure to detect a post-injury change in SYN expression or re-distribution in our study may be related to the relatively low force of impact or to the fact that changes in synaptic vesicle transportation as a consequence of a mild compression SCI may have not yet occurred in its full scale in the 7-day time period. Another indication that SOD1 gene defects may act on the homeostasis of neurofilaments when nervous tissue is under stress is the up-regulation of Mapb1 and possibly of Mast1 in the post-injury G93A-SOD1 spinal cord, both genes appearing to be down-regulated in WT spinal cord after compression injury (Figure 6). Mapb1 binds to microfilaments linking them with the microtubule system, whilst remaining associated to the plasma membranes. Mapb1 exerts a central role in the process of axonal elongation and branching . The observed selective activation of Map1b in G93A-SOD1 spinal cord could be part of a compensative neuroprotective response that counters the intrinsic SOD1-mediated vulnerability under conditions of stress. This survival program may be mediated by the activation of molecules involved in the maintenance of the cytoskeletal integrity as well as by the production of cholesterol and isoprenoids derivates as reported above.
Microarray studies in spinal cord have shown that sham operation alone can induce gene expression changes similar to those caused by mild injury, but only in the first few hours after trauma [8, 22]. Our study confirms that simple laminectomy triggers the transcriptional regulation of a wide range of gene categories in both the WT and the G93A-SOD1 spinal cords. However, gene categories involved in transcription, differentiation, ion homeostasis, apoptosis, organisation of mitochondrial function and interleukin-6 release become activated only in the G93A-SOD1 spinal cord at 4 hours from laminectomy. It is clear from our experiments that both laminectomy and mild spinal cord injury provoke a mixture of pro-survival and pro-apoptotic expression changes, the latter being dominated by mitochondrial cell-death signals and interleukin-6-mediated inflammatory changes.
In this study, we have attempted to correlate the poor functional recovery observed in the G93A-SOD1 rats during the acute post-injury phase with changes at the level of individual cell types in the affected spinal cord. We have shown only a trend towards an increase in microglia, macrophages and astrocytes in a cord segment caudal to the injury epicenter in the G93A-SOD1 spinal cord compared to WT littermates under the same experimental conditions, but no significant changes in the number of these cells between the two groups. This observation is in line with a recent study that has analysed the effects of a longitudinal stab injury of the lumbar spinal cord region of the same pre-symptomatic animal model of ALS utilised in our study . In this experimental paradigm of more invasive spinal cord injury, the level of host glial activation and the motor cell numbers at the site of the lesion in a 2-week post-injury period were not significantly different in the injured G93A-SOD1 rats compared to the injured WT littermates . Our study demonstrates that the size of the anterior horns motor cells in the G93A-SOD1 animals after compression SCI is significantly reduced when compared to the WT motor cell population. Previous studies on animal models of ALS have also shown a loss of the largest spinal motor neurons with the disease progression . Various attempts to clarify whether the late stage loss of large motor neurones is due to a higher vulnerability of these neurones, which may die or atrophy earlier than the small ones or whether motor cells never reach their maximal size before disease onset have not come to any conclusion . It is also not known whether the loss or the atrophy of large motor neurons is the main determinant of the functional decline and ultimately of the loss of motor units observed in the overt phase of the disease. Clearly, the same type of uncertainties may apply to the interpretation of our results. The presence of smaller motor neurones in the G93A-SOD1 spinal cord one week after compression injury may play a part in the poor locomotor performances observed in the G93A-SOD1 rats compared to their WT littermates under to the same experimental conditions. The overall molecular changes we have observed in the G93-SOD1 spinal cord in the post-injury phase may also be responsible of the poor functional recovery described in the G93-SOD1 rats.
Similarly to a number of published investigations, our study has used WT littermates as control for the human G93A-SOD1 rats, in order to ensure the maximum level of genetic homogeneity between the groups of animals. This choice is of particular importance when studying changes at a genetic level [6, 36, 41, 44–48]. Previous studies have shown that the human G93A-SOD1 transgene does not affect the endogenous rat SOD1 protein levels . However, it has also been shown that the total SOD1 activity in the G93A-SOD1 rats is increased to 200-300% of the control level, as the result of the combined endogenous SOD1 activity of the rat and of the added mutant G93A-SOD1 activity . This observation is clearly important in the interpretation of our results. In future studies, it will be useful to include rats with elevated non-mutant SOD1 activity as an extra control group, in order to normalise whatever experimental approach to comparable levels of SOD1 activity .