Introduction
Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by the degeneration of motor neurons, progressive muscle wasting, and reduced mobility [
25,
55]. Genetic studies have linked mutations in ubiquilin2 (UBQLN2) to both ALS and frontotemporal lobar degeneration (FTLD) [
8,
10,
48]. How UBQLN2 mutations cause neuronal death remains to be determined.
A prominent feature of Ubqln2-linked diseases is protein aggregation [
8], which is well reproduced in transgenic rats and mice overexpressing mutant Ubqln2 [
10,
25,
52]. Both TDP-43 pathology and ubiquitination are common features in a variety of neurological diseases, including ALS, FTLD, and Alzheimer’s disease (AD) [
31,
37]. Ubqln2-positive inclusions exist not just in patients harboring a Ubqln2 mutation but also in chromosome 9 open reading frame 72 (C9ORF72)-linked cases [
5,
8]. Ubqln2 is an X-linked gene that consists of an ubiquitin-like domain (UBL) at the N-terminus and an ubiquitin-associated domain (UBA) at the C-terminus [
20]. UBQLN2 shuttles between the nucleus and cytoplasm to perform functions related to protein degradation via proteasomes and autophagy [
36,
41]. UBQLN2 inclusion is a well reproduced feature in in vivo models of ALS [
10,
14,
25,
52]. Another ALS-linked gene, p62/SQSTM1, co-localizes with abnormal UBQLN2 inclusions [
8,
10,
14,
52], suggesting a synergistic effect of UBQLN2 and p62 during neurodegenerative disease progression. Thus, accumulating evidence in both patients and rodent models suggests that Ubqln2-positive inclusions play an important role in proteinopathy in neurodegenerative disorders.
Although both mutant SOD1 and mutant TDP-43 reproduce typical ALS features in rodent models, overexpression of mutant SOD1 in spinal motor neurons does not lead to motor neuron death [
29,
39], whereas selective expression of mutant TDP-43 in motor neurons causes substantial motor neuron death [
16] in rats. These findings suggest different contributions of mutant SOD1 and mutant TDP43 to ALS. Previous studies have shown that mutant UBQLN2
P497H transgenic rat or mouse models as well as a UBQLN2
P520T knock-in mouse model harboring the equivalent human P506T mutation exhibited memory deficits but did not develop any phenotypes of motor neuron disease [
10,
13]. It has also been shown that selective expression of either wild-type or ALS–FTLD mutant UBQLN2
P497S or UBQLN2
P506T in the motor neurons of mice leads to both memory deficits and abnormal motor phenotypes [
25]. However, the influence of selectively expressing ALS-linked mutant UBQLN2 in the spinal motor neurons remains unknown.
As the most abundant glial cells in the central nervous system (CNS), astrocytes play an important role during central nervous system (CNS) development [
32]. Astrocytes also serve as mediators of inflammatory responses in the CNS [
45]. Recent studies suggest, however, that reactive astrocytes cause detrimental effects in neurons in several neurological disorders [
2,
19,
26,
27,
35]. The contribution of astrocytes to the pathogenesis of ALS remains controversial.
Here, we show that the selective overexpression of mutant UBQLN2P497H in the spinal motor neurons led to age-dependent impairment of motor functions in ChATtTA/UBQLN2P497H rats, including motor neuron degeneration, skeletal muscle atrophy, progressive impairment of motor function, TDP-43 pathology, ubiquitination and glial reactions, and abnormal protein accumulation. In contrast, selective overexpression of mutant UBQLN2P497H in astrocytes was not associated with motor impairment. The accumulation of p62 and ubiquitin was increased, however, in the spinal cord astrocytes of GFAPtTA/UBQLN2P497H rats.
Discussion
Mutations in UBQLN2 have been linked to ALS-FLTD, and the abnormal accumulation of UBQLN2 inclusions is a remarkable feature of pathological alterations linked to the UBQLN2 mutation [
8,
46]. Several groups have recapitulated this specific pathological change in rodent models [
10,
13,
25,
52]. No studies to date, however, have shown the effect of expressing mutant UBQLN2 either in the motor neurons or in the astrocytes to test whether mutant UBQLN2 expression leads to motor neuron degeneration in a cell-autonomous manner. To test this hypothesis, therefore, we created novel transgenic models expressing mutant UBQLN2
P497H in the spinal motor neurons or astrocytes in rats.
One recent report showed that transgenic mice expressing either the UBQLN2 P497S or P506T mutation developed both cognitive deficits and motor phenotypes, including progressive reduction of mobility, progressive loss of motor neurons in the spinal cord, and denervation of skeletal muscles as well as abnormal accumulation of UBQLN2 inclusions [
25]. Similar motor phenotypes were observed in our novel ChATtTA/TRE-UBQLN2
P497H transgenic rats. In particular, denervation atrophy of the gastrocnemius muscles was observed as early as 3 months old, but loss of motor neurons was not detected at that age. The motor phenotypes, however, appeared for even low levels of mutant UBQLN2 (about 20% of the endogenous levels). In contrast, disease did not develop in SOD1
G93A mice until the levels of mutant SOD1 were three times that of endogenous SOD1 protein [
11]. Moreover, both Gorrie et al. [
10] and Hjerpe et al. [
13] reported progressive accumulation of UBQLN2 inclusions and progressive cognitive deficits in mice expressing either the UBQLN2 P497H or P506T mutation. All these findings indicate that mutant UBQLN2 leads to neuron degeneration in rodent models. Furthermore, the abnormal accumulation of UBQLN2 inclusions is a remarkable pathological feature of UBQLN2-related diseases.
Similar findings have been reported in transgenic rats expressing mutant TDP43 (M337 V substitution) in the spinal motor neurons, which causes rapid degeneration of motor neurons and paralysis [
16]. In contrast, transgenic mice expressing mutant SOD1 in the motor neurons do not develop motor phenotypes [
29,
50]. The causes of these differences remain unknown. One possible reason for these differences is that different disease mechanisms may underlie UBQLN2, TDP-43, and SOD1 genes. For example, UBQLN2 involves protein degradation via both autophagy and the ubiquitin-proteasome pathway [
9,
13,
36,
41,
53]. The overexpression of UBQLN2
P497H in the spinal motor neurons caused the autophagy substrate p62 to progressively accumulate as well as colocalize with both UBQLN2 and ChAT inclusions in the ventral horn of spinal cord in ChATtTA/UBQLN2
P497H rats, which is similar to the results from transgenic rats expressing UBQLN2 in forebrain neurons [
14,
52]. At 12 months old, p62 accumulated predominantly in the nucleus and cytoplasmic p62 was mostly depleted. Under physiological conditions, however, p62 is commonly considered a cytoplasmic protein. p62 contains two nuclear localization signals (NLS) and a nuclear export signal, however, and it has been confirmed that p62 also shuttles between the nucleus and cytoplasm, a process that is regulated by the phosphorylation of NLS [
38]. The mislocalization of p62, as observed in our study, may be the underlying cause of the abnormal functions observed in our ChATtTA/UBQLN2
P497H rats. Total p62 was increased in ChATtTA/UBQLN2
P497H rats at 1 month old, which is similar to the findings in rats expressing mutant UBQLN2 in the forebrain [
52]. As an autophagy substrate, p62 is essential to neurons [
22], and mutations in p62 have been linked to ALS and FTLD [
42]. In mouse models, loss of p62 leads to neurodegeneration [
40]. Our finding that accumulated p62 colocalizes with UBQLN2 inclusions is similar to our previous reports in other rat models [
14,
52]. These findings imply that the two disease genes may share similar mechanisms underlying neurodegeneration. Specifically, mutant UBQLN2
P497H may compromise the functions of autophagy, leading to abnormal protein accumulations of UBQLN2, p62, and others.
Although p62 has been used as one indicator of autophagy [
3,
30,
43], autophagic flux should be measured by an LC3 turnover assay in addition to p62. LC3-II is one isoform of LC3, and is widely used to measure the autophagic process [
22,
47]. The suppression of LC3-II expression reflects impaired autophagy, and the amount of LC3-II is correlated with the extent of autophagosome formation [
18]. ATG7 is another autophagy component that is essential for autophagosome formation. The loss of ATG7 leads to the reduction of autophagy in mice [
23]. In our ChATtTA/UBQLN2
P497H rats, both LC3 and ATG7 were accumulated at 1 month old but decreased substantially after 6-month old. Similarly, the lysosomal membrane protein LAMP2a also was accumulated and colocalized with UBQLN2 inclusions in 12-month old ChATtTA/UBQLN2
P497H rats. Consistent with these findings, the selective loss of LAMP2A protein directly correlated with increased levels of α-synuclein in early Parkinson’s disease [
34]. All these results suggest that mutant UBQLN2
P497H compromises autophagy-lysosomal pathways in an age-dependent manner. Moreover, the decrease of several core ATG proteins suggests that mutant UBQLN2
P497H is more likely to suppress autophagy at upstream stages.
The hyperphosphorylated form of TDP-43 has been identified as a core component of cytosolic inclusions in sporadic ALS [
1,
37]. In ChATtTA/UBQLN2
P497H rats, relatively little phosphorylated TDP-43 was detected in the spinal cord and did not colocalize with UBQLN2 inclusions. In contrast, accumulated ubiquitin was colocalized with UBQLN2 inclusions. Ubiquitin accumulation is one of the key pathological alterations in neurodegenerative diseases, and ubiquitin accumulation may correlate with neurodegeneration in our rats. In contrast, the expression of the astrocyte marker GFAP was elevated in our rats, but no significant changes in IBa1, a marker of microglia, were observed. These findings are different from those of other rodent models, in which increased expression in both astrocytes and microglia have been observed [
16,
17,
25,
52]. This difference may be due to the slow progression of the disease and a mild loss of motor neurons in ChATtTA/UBQLN2
P497H rats compared to other rat models, indicating that astrocytes are more sensitive to stress conditions than microglia in our mutant UBQLN2
P497H rats.
We did not observe any motor deficits at 6-month old GFAPtTA/UBQLN2
P497H rats, which is not consistent with the motor phenotypes observed in mutant TDP-43 or mutant SOD1 rodent models [
26,
35,
49]. Mutant UBQLN2 protein expression was higher in the spinal cord of GFAPtTA/UBQLN2
P497H rats compared to ChATtTA/UBQLN2
P497H rats. These findings suggest that phenotypes caused by mutant UBQLN2
P497H are not dependent solely on the expression level of mutant proteins in motor neurons and astrocytes in rats. It would be important to investigate whether any disease phenotypes can be induced in older rats (up to 18 months old) expressing mutant UBQLN2
P497H in astrocytes. In addition, future studies are needed to examine whether mutant UBQLN2 will initiate motor neuron degeneration in a non-cell autonomous manner when overexpressing mutant UBQLN2 in other non-neuronal cells, such as microglia or oligodendrocytes.