Over 170 mutations in superoxide dismutase-1 (SOD1) cause familial amyotrophic lateral sclerosis (ALS), a lethal motor neuron disease. Although the molecular properties of SOD1 mutants differ considerably, we have recently shown that intracellular copper dyshomeostasis is a common pathogenic feature of different SOD1 mutants. Thus, the potentiation of endogenous copper regulation could be a therapeutic strategy. In this study, we investigated the effects of the overexpression of metallothionein-I (MT-I), a major copper-regulating protein, on the disease course of a mouse model of ALS (SOD1G93A). Using double transgenic techniques, we found that the overexpression of MT-I in SOD1G93A mice significantly extended the lifespan and slowed disease progression, but the effects on disease onset were modest. Genetically induced MT-I normalized copper dyshomeostasis in the spinal cord without influencing SOD1 enzymatic activity. The overexpression of MT-I in SOD1G93A mice markedly attenuated the pathological features of the mice, including the death of motor neurons, the degeneration of ventral root axons, the atrophy of skeletal muscles, and the activation of glial cells. Double transgenic mice also showed a decreased level of SOD1 aggregates within the glial cells of the spinal cord. Furthermore, the overexpression of MT-I in SOD1G93A mice reduced the number of spheroid-shaped astrocytes cleaved by active caspase-3. We concluded that therapeutic strategies aimed at the potentiation of copper regulation by MT-I could be of benefit in cases of ALS caused by SOD1 mutations.
Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disease characterized by the loss of motor neurons, leading to paralysis and death within 3 years after diagnosis (1). The vast majority of ALS cases are sporadic, whereas 10% are familial cases. Mutations in the superoxide dismutase-1 (Sod1) gene account for 20% of familial ALS (2). Mice overexpressing human SOD1 with ALS-causing mutations develop a motor neuron disease that replicates the clinical and pathological features of ALS (3), whereas the deletion of the Sod1 gene in mice does not (4). Thus, the SOD1 mutations are believed to cause the disease through a toxic gain-of-function, which has been largely uncharacterized.
To date, >170 different mutations in the Sod1 gene have been identified (1). The molecular properties of SOD1 mutants differ considerably (5). Nevertheless, we have recently identified that intracellular copper dyshomeostasis is a common pathogenic feature of different SOD1 mutants (6). The total amounts of copper ions are significantly elevated in the spinal cords of mutant SOD1 mice, regardless of the copper-binding abilities (6–9). Synchrotron X-ray fluorescence microscopy has shown that copper ions accumulate in the ventral horn of SOD1G93A mice (10), the most studied mouse model of ALS. These copper accumulations are associated with a disturbance of the copper-trafficking system, which regulates the copper levels in the cells (6,9). The importance of intracellular copper dyshomeostasis in pathogenesis is supported by pharmacological and genetic findings. Tetrathiomolybdate, an intracellular copper chelator, has an excellent therapeutic benefit in SOD1G93A mice (11), even when the treatment was started at a symptomatic stage (6). However, the effects of other extracellular copper chelators, such as d-penicillamine (12) and trientine (13,14), were less effective in SOD1G93A mice. Moreover, genetically decreased copper levels in the spinal cord, induced by X-linked mottled/brindled, prolong survival and prevent the loss of motor neurons in SOD1G86R mice (15). These findings suggest that the manipulation of intracellular copper dyshomeostasis via copper-regulating proteins could be a therapeutic strategy.
Metallothioneins (MTs) belong to a family of metal-binding proteins with high cysteine contents (16). Four isoforms of MTs exist in mammals, of which MT-I and MT-II are the major isoforms in the central nervous system (CNS) (16). Since these two isoforms share a high degree of homology and similar expression profiles (17), they are functionally considered as a single isoform (hereafter designated as MT-I/-II). MT-I/-II is believed to act as a copper chaperone to sequester excusive copper ions (18), although the precise role of MT-I/-II in the regulation of copper homeostasis is not fully understood. This putative function of MT-I/-II arises from the following evidence. MT-I/-II has the second highest affinity for copper ions in cells after the SOD1 active site (19). Because the Mt-I/-II gene promoter possesses a metal regulatory element that is activated by copper ions (16), the synthesis of MT-I/-II proteins is markedly increased in patients with Wilson's disease (20), a copper metabolic disease that results in the accumulation of copper in the liver and brain. The coordination of copper ions by MT-I/-II decreases its toxicity (21), and these complexes are thermodynamically stable (22). Indeed, MT-I/-II has been found to exist as a copper-binding form in patients with Wilson's disease (20).
The contributions of MT-I/-II to the disease course of ALS have been clearly demonstrated. The expression level of MT-I/-II proteins (8) and mRNA (23–25) is significantly increased in the spinal cords of SOD1G93A mice, even at pre-symptomatic stages. The upregulation of MT-I/-II in SOD1G93A mice is observed in reactive astrocytes, but not in motor neurons, in the spinal cord (23,24). The deletion of the Mt-I/-II gene in SOD1G93A mice shortens the lifespan and accelerates the loss of motor neurons (24,26). The role of MT-I/-II in the disease course is probably applicable to the pathogenesis of sporadic ALS cases. The levels of MT-I/-II-copper ion binding are significantly increased in the spinal cords of patients with sporadic ALS (27). Overall, MT-I/-II is most likely to play a protective role in the disease course that is potentially related to its copper-regulating ability. However, whether elevated levels of MT-I/-II can ameliorate the disease course in vivo remains unknown, although overexpression of the proteins in mice prevents neuronal loss in many acute models of neurodegeneration, cerebral ischemia (28).
The aim of the present study was to address whether the potentiation of copper regulation by MT-I could modify the disease course of SOD1G93A mice. To answer this question, we generated double transgenic mice expressing murine MT-I and SOD1G93A. Here, we provide the first in vivo evidence that the functional enhancement of MT-I in SOD1G93A mice normalizes copper dyshomeostasis in the spinal cord. We also show that the overexpression of MT-I in SOD1G93A mice significantly attenuates the disease course.
MT-I/-II is upregulated in astrocytes and microglia, but not in motor neurons, in SOD1G93A mice, concomitant with disease development
To extend our earlier evidence regarding the molecular behavior of MT-I/-II in high-copy SOD1G93A mice (8,25), we sought to determine the temporal expression patterns of MT-I/-II in the lumbar spinal cord, the region most affected by ALS, of SOD1G93A mice at three different disease stages: pre-symptomatic (30 and 60 days), symptomatic (90 days) and end-stage (120 days). As shown in Fig. 1A, a western blot analysis revealed a significant increase in the MT-I/-II level in SOD1G93A mice even at 30 days of age, compared with the levels in non-transgenic mice (NTG) and transgenic mice carrying human wild-type SOD1 (SOD1WT). This expression pattern in SOD1G93A mice showed an age-dependent upregulation with disease development (Fig. 1B). In the cerebellum, a region that is unaffected by the disease, the MT-I/-II levels of SOD1G93A mice did not differ from those in the NTG and SOD1WT mice (data not shown).
We next explored the temporal distribution patterns of MT-I/-II in the lumbar spinal cord of SOD1G93A mice at different disease stages (Fig. 1C). Immunofluorescence studies showed that intense immunoreactivity for MT-I/-II was mainly detected in the white matter of SOD1G93A mice even at a pre-symptomatic stage, as previously reported (23). With disease development, the distribution of MT-I/-II in SOD1G93A mice spread throughout the gray matter including the ventral horns. In contrast, the distribution of MT-I/-II in NTG and SOD1WT mice was predominantly localized in the white matter. To determine the cell types expressing MT-I in the lumbar spinal cord, we performed double immunofluorescence studies using cell type-specific antibodies against NeuN, glial fibrillary acidic protein (GFAP) and CD11b, which are markers for neurons, astrocytes and microglia, respectively (Fig. 1D–F). While MT-I/-II in NTG and SOD1WT mice was predominantly localized in the astrocytes of lumbar spinal cords, MT-I/-II was expressed in reactive astrocytes (Fig. 1E) and microglia (Fig. 1F), but not in motor neurons (Fig. 1D), in end-stage SOD1G93A mice. These results partially disagree with those of an earlier report (23), which showed that MT-I/-II is not localized in the microglia of low-copy SOD1G93A mice. A possible reason for this discrepancy is related to the different expression levels of human SOD1 protein in distinct lines of SOD1G93A mice.
Overexpression of MT-I in SOD1G93A mice delays disease onset, prolongs the lifespan and slows disease progression
To determine the contributions of overexpressed MT-I/-II to the disease course, we generated double transgenic mice harboring murine MT-I and SOD1G93A (MT-I*/SOD1G93A). Since a decline in body weight is highly correlated with denervation-induced muscle atrophy, weight loss is well established as a simple and objective indictor of the disease course (29). Thus, we assessed the disease course based on body weight alterations (Table 1). The overexpression of MT-I in SOD1G93A mice slightly delayed the disease onset by 8 days (Fig. 2A), defined as the age with the peak weight before the loss of body weight began to occur. The development of early disease was also delayed by 8 days (Fig. 2B), at which time the body weight loss had reached a reduction of 10%. The overexpression of MT-I significantly prolonged the lifespan by 24 days (Fig. 2C). The effect of overexpressed MT-I on the total disease phase was remarkable, resulting in a 16 days slower progression (Fig. 2D, left). No changes were observed in the early disease phase, a period corresponding to the time from disease onset until early disease (Fig. 2D, middle). However, the late disease phase, an interval corresponding to the time from early disease through to the endpoint, in the MT-I*/SOD1G93A mice was greater by 15 days compared with the SOD1G93A mice (Fig. 2D, right), indicating that the slowed disease progression in response to MT-I overexpression was related to a change in the late disease phase. Overall, the overexpression of MT-I in SOD1G93A mice greatly modified the disease progression, rather than the disease onset.
|Onset||104 ± 3.8||112 ± 6.2||3.6 × 10−5|
|Early disease||122 ± 2.4||130 ± 3.3||6.2 × 10−9|
|Survival||132 ± 3.2||156 ± 8.9||4.5 × 10−10|
|Total disease phase||28 ± 4.8||44 ± 9.4||3.9 × 10−8|
|Early disease phase||18 ± 3.4||19 ± 4.6||0.73|
|Late disease phase||10 ± 2.0||25 ± 7.9||2.1 × 10−10|
|Onset||104 ± 3.8||112 ± 6.2||3.6 × 10−5|
|Early disease||122 ± 2.4||130 ± 3.3||6.2 × 10−9|
|Survival||132 ± 3.2||156 ± 8.9||4.5 × 10−10|
|Total disease phase||28 ± 4.8||44 ± 9.4||3.9 × 10−8|
|Early disease phase||18 ± 3.4||19 ± 4.6||0.73|
|Late disease phase||10 ± 2.0||25 ± 7.9||2.1 × 10−10|
Results (days) were expressed as the means ± SD. n = 20 for G93A (male:female = 10:10), n = 24 for MT-I*/G93A (male:female = 12:12). The disease course of the mice was evaluated based on body weight alterations, reflecting denervation-induced muscle atrophy. The time of disease onset was regarded as the age when the mice reached their peak bodyweights. Early disease was defined as the time at which the peak body weight had decreased by 10%. The total, early, and late disease phases were determined as the duration between disease onset and end-stage, between disease onset and early disease and between early disease and end-stage, respectively.
MT-I*/SOD1G93A mice have modestly increased MT-I/-II levels in the astrocytes and microglia of the spinal cord
We confirmed the expression level of MT-I/-II in the lumbar spinal cord of MT-I*/SOD1G93A mice. A western blot analysis showed that the MT-I*/SOD1G93A mice had a significantly higher level of MT-I/-II than the SOD1G93A mice at the end-stage, representing an increase of 31% (Fig. 3A and B). We also validated the distribution of MT-I/-II in the lumbar spinal cord of MT-I*/SOD1G93A mice. MT-I* transgenic mice reportedly express the murine Mt-I transgene from its endogenous promoter (30). The construct results in a tissue distribution of MT-I/-II that resembles that of endogenous MT-I (31). However, the overexpression of MT-I in the SOD1G93A mice resulted in a different distribution pattern of MT-I/-II in the lumbar spinal cord (Fig. 3C). Immunohistochemical studies showed that robust immunoreactivities for MT-I/-II were observed in the ventral funiculus and ventral horn of MT-I*/SOD1G93A mice, whereas the immunoreactivity was mainly detected in the gray matter of end-stage SOD1G93A mice. We identified the cell types expressing MT-I/-II in the lumbar spinal cord of MT-I*/SOD1G93A mice using dual immunofluorescence studies (Fig. 3D). The cell types in the lumbar spinal cord of MT-I*/SOD1G93A mice were similar to those in SOD1G93A mice (Fig. 1D–F), showing that MT-I/-II was localized in the astrocytes and microglia, but not in the motor neurons.
Overexpression of MT-I in SOD1G93A mice can normalize intracellular copper dyshomeostasis
To address the effects of overexpressed MT-I on copper dyshomeostasis in SOD1G93A mice, we measured the total amounts of copper ions in the lumbar spinal cord (Fig. 3E). In agreement with previous reports (6–9), inductively coupled plasma-mass spectrometry (ICP-MS) showed that the total amounts of copper in SOD1G93A mice at end stage (130 days) were approximately three-fold higher than those in age-matched NTG mice. In contrast, the total copper levels in MT-I*/SOD1G93A mice at 130 days were significantly lower than those in SOD1G93A mice. Because SOD1 enzymatic activity is strictly dependent on the copper content that has bound to the SOD1 active site (32), the activity is considered to be a simple indicator of the copper metallation status of SOD1. Thus, we measured the SOD1 activity in the lumbar spinal cords of mice at 130 days of age (Fig. 3F). The SOD1 activity in MT-I* mice did not differ from that in NTG mice, suggesting that the overexpression of MT-I did not alter the copper metallation status of murine SOD1. Likewise, the overexpression of MT-I did not affect human SOD1 activities in either SOD1WT or SOD1G93A mice.
Overexpression of MT-I attenuates the histopathological features of SOD1G93A mice
We next studied whether the overexpression of MT-I could prevent the histopathological features of SOD1G93A mice (Fig. 4). We first evaluated the effects of overexpressed MT-I on motoneuronal death. Lumbar spinal cord sections were immunostained with an antibody against NeuN (Fig. 4A). The number of NeuN-positive α-motor neurons in the ventral horn was then counted in a blinded manner. In terminal SOD1G93A mice (130 days of age), only 40 ± 7.9% of the α-motor neurons, relative to those in age-matched NTG mice, had survived. Conversely, compared with the same age in terminal SOD1G93A mice, 71 ± 2.2% of the α-motor neurons were retained in the MT-I*/SOD1G93A mice (Fig. 4E). Next, to evaluate the effects of MT-I overexpression on axonal degeneration, ventral root sections were immunostained with anti-SMI-312 antibody, a marker of axons (Fig. 4B). End-stage SOD1G93A mice showed a substantial reduction in the number of axons (53 ± 3.8%), whereas the overexpression of MT-I in SOD1G93A mice significantly decreased the loss of axons in the ventral roots (Fig. 4F, 76 ± 5.6%). The overexpression of MT-I in SOD1G93A mice also resulted in a reduction of muscle atrophy in the gastrocnemius muscle. Histological staining with hematoxylin and eosin, followed by a quantitative assessment, revealed that the myofiber area in end-stage SOD1G93A mice was significantly reduced to 52 ± 2.1%, whereas the overexpression of MT-I markedly inhibited the muscle atrophy, resulting in a retention of 74 ± 2.9% (only representative pictures of the gastrocnemius muscle stained with hematoxylin and eosin are not shown). Similarly, the overexpression of MT-I in SOD1G93A mice markedly inhibited the morphological activation of astrocytes (Fig. 4C and G) and microglia (Fig. 4D and H) compared with that in SOD1G93A mice, a finding that is supported by a previous paper showing that the absence of Mt-I/-II in low-copy SOD1G93A mice exacerbates astrogliosis (26).
MT-I interacts with SOD1 aggregates in the spinal cords of SOD1G93A mice
The presence of SOD1 aggregates is a pathological hallmark of familial ALS caused by SOD1 mutations (33). We then tested the molecular relationships between MT-I and SOD1G93A aggregates (Fig. 5). Lumbar spinal cord sections from the terminal stage SOD1G93A mice and age-matched SOD1WT mice were dually immunostained with antibodies against human SOD1 and MT-I. Confocal microscopy showed that MT-I in SOD1G93A mice, but not in SOD1WT mice, was colocalized in SOD1G93A aggregates (Fig. 5A). An antibody-absorption test with recombinant MT-I protein confirmed the above observations, demonstrating that the MT-I signals within the SOD1G93A aggregates were completely eliminated (Fig. 5B). These results strongly indicate that MT-I interacts with the SOD1G93A aggregates.
Overexpression of MT-I in SOD1G93A mice decreases the amounts of insoluble SOD1 aggregates
ALS-causing SOD1 mutants acquire an aberrant propensity to form altered multimeric conformations, including detergent-insolubility (34). We examined whether MT-I overexpression could modify the detergent-insoluble SOD1G93A mutant (Fig. 6). The lumbar spinal cords from mice at 130 days of age were separated into three different fractions based on their detergent-solubility properties: whole homogenates, soluble and insoluble fractions. The extracted proteins from each fraction were analyzed using western blotting with an antibody against human SOD1 (Fig. 6A). In whole homogenates, the overexpression of MT-I in SOD1G93A mice significantly decreased the total level of SOD1G93A protein by 28%, relative to that in SOD1G93A mice (Fig. 6B, left). The soluble mutant level in MT-I*/SOD1G93A mice was slightly diminished by 5%, but the difference did not reach statistical significance (Fig. 6B, middle). However, the insoluble SOD1G93A level in MT-I*/SOD1G93A mice was significantly decreased by 22% (Fig. 6B, right). Immunofluorescence studies using an antibody against human SOD1 confirmed the above biochemical findings, showing that the overexpression of MT-I in SOD1G93A mice markedly decreased the amounts of SOD1 aggregates in the ventral horn of the lumbar spinal cord (Fig. 6C).
Overexpression of MT-I in SOD1G93A mice decreases SOD1 aggregates in astrocytes and microglia
The degeneration of motor neurons as a result of SOD1 mutations in ALS requires the expression of the mutant proteins in not only the motor neurons themselves, but also in non-neuronal cells (35). Among the non-neuronal cells, the expression of SOD1 mutations in astrocytes (36,37) or microglia (29,38) reportedly determines disease progression. Since MT-I/-II is expressed in the astrocytes and microglia of SOD1G93A mice (Fig. 1E and F) and the overexpression of MT-I in SOD1G93A mice greatly impacted disease progression rather than disease onset (Fig. 2A and D), we hypothesized that the overexpression of MT-I might influence the expression level of SOD1G93A protein within astrocytes and/or microglia. To test this hypothesis, lumbar spinal cord sections were dually immunostained with antibodies against human SOD1 and either GFAP or CD11b. As expected, the overexpression of MT-I in SOD1G93A mice markedly decreased the SOD1G93A level within astrocytes (Fig. 6D) and microglia (Fig. 6E).
Overexpression of MT-I in SOD1G93A mice reduces the degeneration of astrocytes cleaved by active caspase-3
To gain molecular insights into the effects of overexpressed MT-I on non-neuronal autonomous toxicity caused by SOD1G93A, we focused on astrocytes. Astrocytes closely interact with neurons to maintain an optimized environment, including copper homeostasis, for neuronal survival (39). However, astrocytes expressing SOD1 mutants have a compromised function, resulting in the death of motor neurons. A subpopulation of astrocytes undergoes degeneration in the spinal cords of mutant SOD1 mice (40) as well as sporadic ALS patients (41). These astrocytes display morphologically distinct features from the original astrocytes, appearing as spheroid-shaped cells with an increased diameter (40). Thus, astrocytic spheroids are a morphological marker of dysfunctional astrocytes.
We then explored whether the overexpression of MT-I in SOD1G93A mice could rescue the degenerative process of astrocytes. We first quantified the total number of spheroid-shaped astrocytes in the spinal cords of different genotypic mice. The lumbar spinal cord sections were immunostained with an antibody against GFAP (Fig. 7A). The overall number of spheroid-shaped astrocytes in the ventral horn was counted in a blinded manner. Compared with end-stage SOD1G93A mice, MT-I*/SOD1G93A mice had a significantly lower number of spheroid-shaped astrocytes, representing a reduction of 77% (Fig. 7B). No such spheroids were observed in mice that do not express SOD1G93A mutation, including NTG, MT-I*, SOD1WT and MT-I*/SOD1WT mice (Fig. 6A).
The molecular mechanism leading to the degeneration of astrocytes in SOD1G93A mice involves proteolytic cleavage by active caspase-3 (40), since murine GFAP has a caspase-3 cleavage consensus sequence, the DXXD motif, corresponding to amino acids 260–263 (UniProtKB/Swiss-Prot number: P03995). With this in mind, we next quantified the number of spheroid-shaped astrocytes that were immunopositive for active caspase-3. Lumbar spinal cord sections from end-stage SOD1G93A mice and age-matched MT-I*/SOD1G93A mice were stained using double immunofluorescence labeling for GFAP and active caspase-3 (Fig. 7C). Consistent with previous reports (40), we found that the proportion of spheroid-shaped astrocytes that were immunopositive for active caspase-3 was approximately one-third (34 ± 1.6%) in the SOD1G93A mice (Fig. 7D). In contrast, the overexpression of MT-I in SOD1G93A mice significantly decreased the number of active caspase-3-positive spheroids by 88%, as well as the proportion of these cells (18 ± 1.1%).
For the first time, we have provided in vivo evidence that the normalization of copper dyshomeostasis by the overexpression of MT-I was able to ameliorate the disease course as well as to protect motor neurons in SOD1G93A mice. Our present study is supported by previous reports (24,26), which showed that the deletion of Mt-I/-II in low-copy SOD1G93A mice significantly decreased the lifespan and accelerated the loss of motor neurons. Thus, MT-I/-II protects against the death of motor neurons induced by the SOD1G93A mutation, even in vivo.
The toxicity of SOD1 mutant protein is associated with a non-cell autonomous mechanism mediated by disrupted communication between motor neurons and the surrounding glial cell (29,35–38). We propose that the overexpression of MT-I in SOD1G93A mice attenuates the non-cell autonomous toxicity, especially within astrocytes. This idea is supported by previous in vitro evidence, demonstrating that the overexpression of MT-I in cultured motor neurons expressing the SOD1G93A mutant protein fails to rescue the death of motor neurons (42). We showed here that the overexpression of MT-I conferred several therapeutic benefits to astrocytes expressing the SOD1G93A mutant. These benefits included (i) the suppression of the morphological activation of astrocytes (Fig. 4C and G), (ii) the reduction of SOD1 aggregates within astrocytes (Fig. 6D) and (iii) the rescue of astrocytic degeneration caused by active caspase-3 (Fig. 7). We speculated that the molecular mechanism underlying the therapeutic benefits of overexpressed MT-I is closely related to the normalization of copper dyshomeostasis within astrocytes, although we did not provide direct evidence of the copper distribution in the spinal cords of MT-I*/SOD1G93A mice. Under physiological conditions, astrocytes serve as a main source of copper ions in the CNS, since these cells are the first parenchymal cells that encounter copper as it crosses the blood–brain barrier (43). Astrocytes can transmit copper ions to other surrounding cells to maintain cellular functions in the CNS (43). However, astrocytes expressing SOD1 mutant proteins probably do not exhibit this property. In vitro co-culture systems have shown that astrocytes harboring SOD1 mutants cause the selective death of motor neurons by secreting toxic factors (44,45), which we hypothesize to be copper ions. Astrocytes carrying SOD1 mutants may secrete copper ions to motor neurons in a disordered manner, exacerbating their degeneration. Therefore, the overexpression of MT-I may enhance the capacity of astrocytes to inhibit the aberrant transmission of copper ions to motor neurons, thereby promoting the survival of motor neurons (Fig. 4A and E).
To our surprise, the overexpression of MT-I in SOD1G93A mice completely restored the elevated copper ions to a normal level in the spinal cord (Fig. 3E), even though the putative function of MT-I/-II in copper regulation is to sequester copper ions (18). How does the overexpression of MT-I in SOD1G93A mice remove accumulated copper ions from the spinal cord? One possibility is that the overexpression of MT-I might function as a peripheral reservoir for copper in SOD1G93A, resulting in a correction of the maldistribution of copper ions by facilitating the removal of copper ions from the spinal cord to the peripheral tissues, such as the liver and kidney. Although exogenous murine MT-I protein is ubiquitously expressed in all the tissues of MT-I* transgenic mice (31), the highest levels of MT-I expression are found within peripheral tissues, particularly in the liver (31), with lower expression levels observed in the spinal cord (Fig. 3A–C). Therefore, we cannot exclude the possibility that the overexpression of MT-I may exert its action primarily from outside of the CNS. The sequestration of copper ions in the peripheral tissues by the overexpression of MT-I might affect the dynamics of copper efflux and/or influx, leading to the creation of a peripheral sink that draws copper ions out of the spinal cord via a concentration gradient. In this respect, it is interesting that the levels of copper bound to MT-I/-II in peripheral tissues, including the liver and kidney, are significantly elevated in patients with sporadic ALS (46). To establish how MT-I overexpression could influence copper dynamics, the copper metallation states of MT-I/-II in the CNS and periphery of double transgenic mice will be determined in the future.
We demonstrated that the overexpression of MT-I significantly decreased the amounts of insoluble SOD1 aggregates, but not soluble mutant, in SOD1G93A mice (Fig. 6A–C). These results potentially link intracellular copper dyshomeostasis and SOD1 aggregation, and strongly suggest that MT-I/-II plays a protective role in SOD1 aggregates through the normalization of copper dyshomeostasis. Our suggestion is supported by the following four findings. First, the normalization of copper dyshomeostasis by treatment with tetrathiomolybdate markedly decreased insoluble SOD1 aggregates in SOD1G93A mice (6). Second, the components of the copper-trafficking system are co-aggregated with SOD1G93A mutant protein (9). Third, MT-I/-II directly interacted with SOD1 aggregates in the spinal cord (Fig. 5). Finally, the absence of Mt-I/-II in SOD1G93A mice resulted in the further accumulation of SOD1 aggregates, accompanied by a significant elevation in the level of copper ions (our unpublished data). How does overexpressed MT-I modify SOD1 aggregates? The copper metallation status of SOD1G93A probably does not contribute to the process, since MT-I overexpression did not affect SOD1 enzymatic activity (Fig. 3F), which is consistent with previous in vitro studies (47,48). Several lines of in vitro evidence indicate that copper ions can trigger the oxidization of SOD1 mutants, causing them to become pathogenic (49,50). Therefore, the overexpression of MT-I may suppress the aberrant redox activity of copper ions caused by SOD1 mutants, leading to an attenuation in the oxidative modification of SOD1 mutants, as observed in SOD1G93A mice (51) as well as in ALS patients (52). Further studies are needed to determine the effects of MT-I overexpression on the oxidative modification of SOD1 mutants.
Another possible mechanism, other than the normalization of intracellular copper dyshomeostasis, underlying the attenuation induced by MT-I overexpression
, should be considered. Apart from its role in copper homeostasis, MT-I/-II plays various roles in protecting against oxidative stress, and apoptosis, which have been proposed as possible pathological processes resulting in the loss of motor neurons (53). Therefore, we cannot rule out the possibility that the therapeutic benefits of the MT-I overexpression in SOD1G93A mice are not dependent on the modification of intracellular copper dyshomeostasis. However, copper dyshomeostasis might interact with, and potentially exacerbate, the above-mentioned pathological processes. For example, copper ions efficiently catalyze the formation of strong oxidants from hydrogen peroxide (54). Moreover, copper ions act as ligand for death receptor, triggering caspase activation and apoptosis (55). In this regard, we would like to emphasize that the overexpression of MT-I in SOD1G93A mice markedly suppressed caspase-3-dependent apoptosis itself (Fig. 7C). These results are supported by a previous paper, which reported that MT-I* transgenic mice were resistant to neuronal apoptosis in an acute model of neurodegeneration, cerebral cryolesion (56).
What are the clinical implications of this study? We believe that MT-I/-II is a promising therapeutic target for the treatment of ALS patients with SOD1 mutations for the following three reasons. First, despite the modestly increased levels of MT-I/-II in MT-I*/SOD1G93A mice, representing a 31% upregulation (Fig. 3A and B), the double transgenic mice had a significantly extended lifespan and exhibited slower disease progression (Fig. 2C and D). These results strongly suggest that therapeutic strategies aimed at MT-I/-II induction could be of benefit in patients with ALS, even though the elevated levels of MT-I/-II would be limited. How can MT-I/-II be induced? Endogenous MT-I/-II proteins can be easily and rapidly induced by pharmacological reagents that are already used clinically (16), since the Mt-I/-II gene promoter consists of multiple regulator elements including those for metals, glucocorticoids, oxidative stress and inflammatory cytokines (16). Also, the exogenous supplementation of recombinant MT-I/-II may be a therapeutic option for ALS, since treatment with recombinant MT in mice significantly prevented the loss of neurons in the injured brain, and this effect was similar to that observed in MT-I* transgenic mice (56). Second, the overexpression of MT-I recused not only motor neurons (Fig. 4A and E), but also other components of the motor system, including motor axons (Fig. 4B and F) and skeletal muscles. Although the death of motor neurons is a pathological hallmark of ALS, the pathological process in ALS is now recognized to extend beyond the motor neurons. Indeed, the complete protection of motor neurons does not improve mortality in mouse models of mutant SOD1-linked ALS (57). Thus, therapeutic interventions must prevent pathological events that occur both inside and outside of motor neurons. Finally, the third reason is related to the current clinical situation for ALS. Since there are no effective biomarkers that can be used to predict disease onset (58), ALS patients typically visit a hospital for the first time after the onset of symptoms. Moreover, the definite diagnosis of ALS continues to be based on clinical assessments using the El Escorial criteria (58). Diagnostic certainty requires a delay of over 1 year from disease onset until diagnosis (59), during which time the patients might have progressed beyond the window of therapeutic opportunity. Thus, therapeutic interventions that can slow the progress of the disease would be of practical benefit. In this regard, it is noteworthy that the overexpression of MT-I dramatically modified the duration of the disease, especially the late phase of the disease (Fig. 3D), rather than the disease onset (Fig. 3A).
MATERIALS AND METHODS
Transgenic mice expressing human SOD1 with the G93A mutation [strain name: B6SJL-Tg(SOD1-G93A)1Gur/J, stock number: 002726] or human wild-type SOD1 [strain name: B6SJL-Tg(SOD1)2Gur/J, stock number: 002297] were purchased from the Jackson Laboratory (3). These transgenic lines were maintained as hemizygotes by crossing transgenic males with F1 non-transgenic females on a B6SJL background; therefore, B6SJL mice were used as controls. In this study, we analyzed the lumbar spinal cord and cerebellum from SOD1G93A mice at three different disease stages: pre-symptomatic (30 and 60 days), symptomatic (90 days) and end stage (120 days). The lumbar spinal cord and cerebellum from age-matched B6SJL and SOD1WT mice were also harvested. The animal protocols were approved by the Institutional Animal Care and Use Committee of the School of Pharmacy, Nihon University, in compliance with our institutional animal care guidelines.
Generation of double transgenic mice expressing murine MT-I and human SOD1s
Transgenic mice harboring murine MT-I [strain name: B6.Cg-Tg(Mt1)174Bri/J, stock number: 002210] were purchased from the Jackson Laboratory (30). The transgenic strain of mice expresses 56 copies of murine Mt-I transgenes from its endogenous promoter (30). The transgene is constructed by cloning a slightly modified murine Mt-I gene with an additional two nucleotides located nine nucleotides upstream of the translational initiation codon (herein referred to as MT-I*) (30). MT-I* mice were homozygously maintained on a C57BL/6J background. For the generation of double transgenic mice expressing murine MT-I and SOD1G93A, a two-step breeding strategy was performed to exclude differences caused by the strain background. First, homozygous male MT-I* mice were crossed with female C57BL/6J mice. Next, hemizygous male SOD1G93A mice were crossed with hemizygous female MT-I* mice to obtain hemizygous MT-I* mice carrying SOD1G93A. We also generated double transgenic mice harboring murine MT-I and SOD1WT. Genomic DNA was extracted from a tail biopsy using the DNeasy Blood and Tissue Kit (QIAGEN). The genotype of the offspring was identified using PCR with the following primers: 5′-ACG CCA GCT GGC GAA AGG G-3′ for forward Mt-I* transgene, 5′-GCC ACC GCG TGG AGC TC-3′ for reverse Mt-I* transgene, 5′-CAT CAG CCC TAA TCC ATC TGA-3′ for forward human Sod1 and 5′-CGC GAC TAA CAA TCA AAG TGA-3′ for reverse human Sod1.
Chemical modification of cysteine residues of MT-I/-II
The cysteine residues of MT-I/-II account for approximately one-third of the total amino acid content (UniProtKB/Swiss-Prot number: P02802 for MT-I; P02798 for MT-II). This high cysteine content makes it difficult to detect MT-I/-II protein using conventional immunoblots. Therefore, the cysteine residues of MT-I/-II were chemically modified with monobromobimane, a thiol-reactive reagent that acts at cysteine residues, as described previously with a slight revision (60). Briefly, the protein extracts from lumbar spinal cord and cerebellum were incubated with 20 mM monobromobimane (Molecular Probes) for 1 h at 4°C in the dark. The excess amount of monobromobimane was removed by extraction with CH2Cl2. The samples were then mixed on Vertex and centrifuged at 9000g for 3 min at 4°C. The upper phase, which contains the MT-I/-II proteins modified with monobromobimane, was carefully collected. The extraction step was repeated three times. The proteins (2 μg) were then used for a western blot analysis.
Western blot analyses were performed as described previously (9). The following primary antibodies were used: mouse anti-MT-I/-II (1:100; Dako), mouse anti-human SOD1 antibody (1:1,000; raised against full length of human SOD1, Santa Cruz Biotechnology) or mouse anti-β-Tubulin (1:10,000; Sigma).
Mice were transcardially perfused with phosphate buffered saline (PBS) followed by ice-cold 4% (w/v) paraformaldehyde in PBS (pH 7.4). The lumbar spinal cords were harvested, post-fixed in the same solution for 24 h at 4°C, and embedded in paraffin. The antigens in the transverse sections (6 μm) were retrieved using an autoclave and a retrieval solution containing 10 mM citrate acid (pH 6.0). Endogenous peroxidase activities were inhibited with 3% (v/v) hydrogen peroxide in methanol for 15 min at room temperature. The lumbar sections were immunostained with mouse anti-MT-I/-II antibody (1:400; Dako). The signals were detected using the Histofine® Mouse Stain Kit (Nichirei Biosciences, Inc.) and 3,3′-diaminobenzidine (Nichirei Biosciences, Inc.) as the chromogen. The sections were mounted in Malinole (Muto Pure Chemicals Co.). The sections were imaged using a light microscope (DP70, Olympus).
Immunofluorescence was performed as described elsewhere (9). The lumbar spinal cord sections were immunostained using the following primary antibodies: mouse anti-MT-I/-II (1:50; Dako), mouse anti-NeuN (1:400; Chemicon, Inc.), biotin-conjugated anti-NeuN (1:400; Chemicon, Inc.), goat anti-GFAP (1:50; Santa Cruz Biotechnology), goat anti-CD11b (1:50; Santa Cruz Biotechnology), mouse anti-human SOD1 (1:100; Santa Cruz Biotechnology) or rabbit anti-human SOD1 (1:100; raised against the full length of human SOD1, Santa Cruz Biotechnology).
The disease statuses of the mice were evaluated according to criteria based on body weight changes (29). Disease onset was regarded as the time when each mouse reached its peak weight before its weight began to decline. Early disease was defined as a 10% loss of the peak weight. The endpoint was defined as the age at which a mouse was unable to right itself within 30 s after being pushed onto its side. The total phase of disease progression was regarded as the period from disease onset until the endpoint. The early phase was estimated as the period between disease onset and early disease. The late phase was defined as the interval from early disease to the endpoint. All the clinical evaluations were performed by separate individuals; thus, the neurological signs were scored in a blinded manner.
Quantification of total amounts of copper ions
The total amounts of copper ions in the lumbar spinal cords were measured using ICP-MS (Agilent 7500, Yokogawa Analytical Systems), as described previously (8). The lumbar spinal cords obtained from mice at 130 days (three mice per genotype) were digested in concentrated nitric acid (Wako Pure Chemical Industries) for 1 h at 100°C. The values were expressed as micrograms per gram of wet tissue weight.
Measurement of SOD1 enzymatic activity
SOD1 enzymatic activity was measured using the SOD1 Assay Kit-WST (Dojindo Molecular Technology, Inc.), as described previously (11). One unit of SOD1 activity was defined as a 50% inhibition of the rate of water-soluble tetrazolium salt reduction. The values were expressed as unit per microgram of the total protein levels.
The histopathological features of the SOD1G93A mice were evaluated as described previously (6). All the measurements were conducted by an observer who was blinded to the genotype. For the quantification of α-motor neurons, every fifth lumbar section of the spinal cord (L4-L5) was immunostained with mouse anti-NeuN antibody (1:1,000; Chemicon, Inc.). Alpha motor neurons were molecularly and morphologically identified as described elsewhere (61). The number of α-motor neurons in the ventral horn was counted in 10 sections per mouse.
For axon counting, the ventral roots were dissected out from the lumbar spinal cord segment (L5) and immunostained with mouse antibody against SMI-312 (1: 2,000; Abcam). The number of large axons (>4 μm in diameter) was counted using the Image J software (National Institutes of Health).
To measure the muscle fiber area, transverse sections from the gastrocnemius muscles were stained with hematoxylin and eosin (Sakura Finetek Japan). At least 100 individual myofiber areas in each mouse were measured using the Image J software.
To quantify astrocytosis and the microglia, lumbar spinal cord sections were immunostained with mouse anti-GFAP (1:200; BD Pharmingen) or goat anti-CD11b (1:50; Santa Cruz Biotechnology). Fluorescent images of the astrocytes and microglia were acquired using a confocal microscope (Laser Scanning System Z510, Carl Zeiss). The fluorescence intensity in the ventral horn was measured using the Image J software.
The antibody-absorption test was performed as described previously (9). Briefly, an antibody against MT-I/-II (1:50; Dako) was pre-incubated with recombinant MT-I protein (Alexis Biochemicals) at 4°C for 24 h. After the pre-incubation, the lumbar sections from SOD1G93A mice at end-stage (120 days) and age-matched SOD1WT mice were dually immunostained using antibodies against human SOD1 (1:100; Santa Cruz Biotechnology) and against MT-I/-II bound to recombinant MT-I protein.
Measurement of human SOD1 protein level
The human SOD1 protein level was determined as described previously (6). Briefly, lumbar spinal cords from mice at 130 days of age were homogenized in 20 volumes (w/v) of ice-cold lysis buffer containing 1% (v/v) Nonidet P-40 and Complete EDTA-free protease inhibitor (Roche Applied Science) in PBS (pH 7.0). The homogenates were sonicated, yielding a whole homogenate fraction. To prepare the detergent-insoluble fractions, a subset of the whole homogenates was centrifuged at 20 000g for 30 min at 4°C. The supernatants were collected as the detergent-soluble fraction. The pellets were dissolved in lysis buffer containing 2% (w/v) sodium dodecyl sulfate. Equal amounts of proteins (5 μg) were analyzed using a western blot with mouse anti-human SOD1 antibody (1:1,000; Santa Cruz Biotechnology).
Quantification of spheroid-shaped astrocytes
A quantitative analysis of spheroid-shaped astrocytes was performed as described previously (40). Lumbar spinal cord sections from mice at 130 days (three mice per each genotype) were dually immunostained with mouse anti-GFAP (1:200; BD Pharmingen) and rabbit anti-active caspase-3 (1:200; Sigma). Fluorescent images were acquired using a confocal microscope (Laser Scanning System Z510, Carl Zeiss). The number of active caspase-3-positive spheroid-shaped astrocytes in the ventral horn was counted in 10 sections per mouse. All the measurements were conducted by an observer who was blinded to the genotype.
The results were expressed as the means ± SD. The disease onset, early disease and survival were compared using a Kaplan–Meier analysis with a log-rank test. The disease progression and the quantification of spheroid-shaped astrocytes were analyzed using an unpaired Student's t-test. Multiple group comparisons were performed using a one-way ANOVA followed by the Tukey–Kramer post-hoc test. The statistical significance was defined as P< 0.05.
This work was supported by a Grant-in-Aid for a research fellowship with the Japan Society for the Promotion of Science (JSPS) for Young Scientists from the JSPS (08J10542 to E.T.), by a Grant-in-Aid for Exploratory Research from the JSPS (21659222 to S.O.), by an Academic Frontier Project for Private Universities matching a fund subsidy from the Ministry of Education, Culture, Sports, Science and Technology (years of 2007–2009 to S.O.) and by a Joint Research Grant from Nihon University College of Pharmacy (year of 2007 to S.O.).
The authors would like to thank Drs Per Zetterström and Stefan L. Marklund (Department of Medical Biosciences, Clinical Chemistry, Umeå University, Sweden) for excellent comments regarding this manuscript.
Conflict of interest statement. None declared.