Torsion dystonia is an autosomal dominant movement disorder characterized by involuntary, repetitive muscle contractions and twisted postures. The most severe early-onset form of dystonia has been linked to mutations in the human DYT1 (TOR1A) gene encoding a protein termed torsinA. While causative genetic alterations have been identified, the function of torsin proteins and the molecular mechanism underlying dystonia remain unknown. Phylogenetic analysis of the torsin protein family indicates these proteins share distant sequence similarity with the large and diverse family of AAA+ proteins. We have established the nematode, Caenorhabditis elegans, as a model system for examining torsin activity. Using an in vivo assay for polyglutamine repeat-induced protein aggregation in living animals, we have determined that ectopic overexpression of both human and C. elegans torsin proteins results in a dramatic reduction of polyglutamine-dependent protein aggregation in a manner similar to that previously reported for molecular chaperones. The suppressive effects of torsin overexpression persisted as animals aged, whereas a mutant nematode torsin protein was incapable of ameliorating aggregate formation. Antibody staining of transgenic animals indicated that both the C. elegans torsin-related protein TOR-2 and ubiquitin were localized to sites of protein aggregation. These data represent the first functional evidence of a role for torsins in effectively managing protein folding and suggest that possible breakdown in a neuroprotective mechanism that is, in part, mediated by torsins may be responsible for the neuronal dysfunction associated with dystonia.
Inherited movement disorders of the nervous system represent a significant health problem for which the connection between known molecular defects and pathophysiology remains largely undetermined. Dystonia is one such disease that is characterized by sustained muscle contractions that frequently cause twisting or repetitive movements or abnormal postures (1,2). Therapeutic options are currently limited and involve either surgery or non-specific approaches such as chemodenervation via directed injection of botulinum toxin. Dystonia, as an ‘umbrella’ term for a series of related disorders with common clinical dispositions, affects about 30 people out of 100,000, rendering it more common than Huntington's disease, amyotrophic lateral sclerosis (ALS) or muscular dystrophy. However, dystonia remains the most poorly understood of these disorders, probably owing to the fact that there is no apparent neuronal degeneration associated with the disease, making it difficult to diagnose or pathologically define. Oppenheim's dystonia, also referred to as early-onset general torsion dystonia, is the most common heritable form of this disease and affects about 50,000 people in North America alone. It is present among a variety of ethnic groups, but is more prevalent (∼1 in 2000) in Ashkenazi Jews (3).
Dystonia is a complex neurological affliction. This statement not only holds true for its clinical manifestations, but also with regard to what little is understood about its molecular nature. Oppenheim's dystonia is transmitted in an autosomal-dominant manner with reduced penetrance (30–40%). It has been elegantly linked to a specific deletion of a codon (GAG) in a gene termed DYT1 (TOR1A) (4,5) and more recently to an 18 bp deletion in this same gene (6). In determining causative mutations, clinical researchers have generated a focal point and genetic foundation for subsequent investigation into the cellular mechanisms underlying dystonia using model organisms. Evolutionary conservation of the DYT1 family of gene products, termed torsins, is evident in the genomes of metazoans, including pigs, rats, mice, zebrafish, fruit flies, and nematodes (5,7).
Aberrant protein deposition has been consistently linked to etiologically diverse neurodegenerative diseases such as Alzheimer's, ALS, prion diseases, and those caused by polyglutamine expansion (CAG repeats) including spinocerebellar ataxias, spinal and bulbar muscular atrophy, and Huntington's Disease (8). Protein inclusions termed Lewy bodies are a clinical characteristic of post-mortem brain tissues from patients with Parkinson's disease (PD). Failure of proteins to adopt their proper structure is a common cause of cellular dysfunction. Quality control mechanisms within cells serve to promote accurate protein folding and breakdown of these cytoprotective processes can result in aberrant protein aggregation and associated disease states. Immunocytochemical analyses of Lewy bodies uncovered intense reactivity for the product of the human DYT1 gene, torsinA (9). Moreover, this protein has been shown to physically associate in Lewy bodies with aggregates of the PD-associated gene product α-synuclein (10). The absence of obvious neurodegeneration in dystonia argues that more subtle causes of cellular dysfunction are responsible for the symptomatic features of the disease.
While progress toward delineating torsin expression and localization has been reported, the cellular function of these proteins remains undefined (9,11,12). Caenorhabditis elegans is a rapidly cultured, genetically amenable transparent animal with a completely defined cell lineage and anatomy. This microscopic nematode has been successfully employed toward investigations into a variety of locomotive and other neurological disorders (13,14). Here we describe the first evidence of a cellular activity for torsins, in that these proteins have the capacity to reduce the formation of protein aggregates when functioning normally, whereas an altered torsin gene product loses this potential. Intracellular responses to protein misfolding and aggregation represent an important and poorly defined biological mechanism with significant consequences for therapeutic intervention of disease mechanisms (8,15,16). These data establish C. elegans as model system for the analysis of cellular deficits related to torsin activity and their relationship to dystonia and associated disorders.
The identification of a nematode torsin-like protein was first reported in the breakthrough paper on positional cloning of human DYT1 (4). This gene product was subsequently shown to be encoded by the ooc-5 gene of C. elegans and has been implicated in spindle positioning during mitosis (17,18). Two additional torsin-related genes have been subsequently uncovered in the complete genome sequence of C. elegans and correspond to predicted open-reading frames, Y37A1B.12 and Y37A1B.13 (19); using appropriate nomenclature, we named these genes tor-1 and tor-2, respectively. Torsins share amino acid sequence similarity with members of the large and functionally diverse AAA+ family of proteins that includes heat-shock proteins, Clp proteases, dynein, and molecular chaperones (20,21). All three nematode torsin-like gene products contain Walker A, Walker B, and other sequence motifs common to AAA+ proteins (Fig. 1).
As a starting point to our investigation of torsin function in C. elegans, we isolated a cDNA corresponding to the worm tor-2 gene. This gene corresponds to the nematode torsin homolog that shares highest global sequence identity to human torsinA (∼40%). The 1239 bp tor-2 cDNA encodes a protein of 412 amino acids; DNA sequencing of both strands confirmed the predicted open-reading frame on C. elegans cosmid Y37A1B, including all exon and intron boundaries. A single protein band of approximately the correct molecular weight is recognized in C. elegans extracts by TOR-2-specific peptide antisera that we have generated (Fig. 2).
While members of the AAA+ superfamily have not previously been associated with human disease mechanisms, they have been suggested to function as oligomeric complexes involved in the conformational assembly and disassembly of protein complexes (7,20). Recent studies have shown that members of the heat-shock family of molecular chaperones can alleviate neurotoxicity and suppress formation of protein inclusions in cells (22,23). We hypothesize that torsin proteins function in a similar capacity to molecular chaperones in facilitating the proper cellular management of misfolded proteins. To experimentally discern this prospect, we adopted an elegant in vivo assay for examining states of intracellular protein aggregation in living animals (24).
Effect of wild-type and mutant TOR-2 overexpression on protein aggregation
Transgenic nematodes containing fluorescent protein aggregates were generated by ectopic expression of gene fusions between stretches of DNA encoding different lengths of polyglutamine-repeats fused to the green fluorescent protein (GFP) in the body wall muscle cells of C. elegans using the well-characterized unc-54 promoter (24,25). Aggregation of the GFP reporter protein is readily visible in these transparent animals and is dependent on the length of the polyglutamine tract. For example, expression of a fusion of 19 glutamines to GFP (referred to as Q19::GFP) exhibits a normal cytoplasmic and diffuse GFP localization pattern (Fig. 3A). However, expansion to a tract of 82 glutamines fused to GFP (Q82::GFP) results in a distinct change in fluorescence, wherein discrete cellular aggregates are clearly evident in all animals, while the former smooth evenly distributed pattern is absent (Fig. 3B).
An analysis of the effect of wild-type tor-2 expression on Q82::GFP aggregates was performed following generation of stable transgenic animals producing the TOR-2 protein under the control of the same high-level constitutive unc-54 promoter by microinjection. The addition of TOR-2 dramatically reduced both the number and size of GFP-containing aggregates in animals containing Q82::GFP (Fig. 3D). In fact, diffuse body wall muscle fluorescence reappeared in many of these animals as well. In contrast, co-expression of TOR-2 with Q19::GFP did not alter the normal cytoplasmic distribution of GFP in these animals (Fig. 3C).
Oppenheim's dystonia is dominantly linked to specific amino acid deletions at the carboxy end of torsinA, therefore implying this region is critical to protein function or interaction within a multimeric complex (4–6). While the specific glutamic acid deletion associated with torsion dystonia is not strictly conserved across species, it is found within a stretch of very high sequence identity, including conserved cysteine residues that may represent a structurally significant region of these proteins (Fig. 1). To examine the consequences of altering this portion of the TOR-2 protein, we used site-directed mutagenesis to generate a mutant tor-2 cDNA lacking a codon for amino acid 368 [TOR-2 (Δ368)]. This results in a deletion of a serine residue in a position of TOR-2 that aligns with one of a pair of diacidic glutamic acids at an analogous position in torsinA (Fig. 1).
Co-expression of TOR-2(Δ368) with Q19::GFP does not alter the general cytoplasmic expression of GFP from what is found in Q19::GFP animals alone (Fig. 3E). However, the TOR-2(Δ368) mutation, when co-expressed with Q82::GFP, was completely incapable of restoring the diffuse body-wall fluorescence in these animals (Fig. 3F). Using our TOR-2 specific antibody, we proceeded to determine if the observed change in torsin activity corresponded to a difference in protein expression or stability. Immunoblotting of extracts of C. elegans indicated that all lines of transgenic animals, including TOR-2(Δ368), clearly contained the TOR-2 protein at roughly equivalent levels above wild-type extracts, as would be expected from constitutive unc-54 overexpression of tor-2 cDNAs (Fig. 2). Therefore, these data are indicative of a loss of torsin activity that is associated with the TOR-2(Δ368) mutation rather than a change in torsin protein expression or stability.
Levels of polyglutamine proteins (both Q19::GFP and Q82::GFP) appeared consistent across all lines of transgenic nematodes tested (Fig. 2B). Perhaps a very slight decrease in polyglutamine expression was seen in lanes corresponding to extracts from wild-type or mutant tor-2 animals co-expressing polyglutamine::GFP fusions. This is probably due to an increased transcriptional load on the unc-54 promoter element as it is titrated during the simultaneous co-overexpression of both torsin and polyglutamine genes from integrated transgenes. Interestingly, immunoblots of Q82::GFP extracts either in the absence of TOR-2 or in the presence of mutant TOR-2 reproducibly revealed the appearance of an extra lower molecular weight band that cross-reacted with a polyglutamine monoclonal antibody (Fig. 2B). In contrast, Q19::GFP lines, with or without TOR-2, and the Q82::GFP line of isogenic animals expressing a functional TOR-2 protein, did not exhibit this extra polyglutamine band. It is possible this band represents a specific degradation product or perhaps the accumulation of a polyglutamine protein intermediate.
TOR-2 alters the size of protein aggregates
As an extension of these studies, we have performed a detailed quantitative analysis of TOR-2 suppression of protein aggregation in isogenic chromosomally-integrated transgenic lines of animals containing differing combinations of polyglutamine and torsin proteins. Comparative and quantitative analysis of aggregate sizes between isogenic animals indicated that there was a significant difference in the size of Q82::GFP aggregates in the presence or absence of wild-type and mutant TOR-2. These differences were readily noticeable and are easily observed in photomicrographs, as shown in Fig. 4. The average size of every aggregate from 30 animals was recorded for each of Q82::GFP, Q82::GFP+TOR-2, and Q82::GFP+TOR-2(Δ368). The average size of aggregates from Q82 animals was 2.7 µm (SE±0.123) compared with 1.6 µm (SE±0.069) from Q82::GFP+TOR-2. This difference is significant (P<0.001) using ANOVA. Furthermore, the difference in aggregate size between Q82::GFP and Q82::GFP+TOR-2(Δ368) animals was also significant (P<0.001) with an aggregate size of 4.8 µm (SE±0.286) for Q82::GFP+TOR-2(Δ368) animals (compared with 2.7 µm for Q82).
Additionally, the amount of variability in aggregate size differed among the animals carrying the various transgenic constructs (Table 1). When aggregates are classified by size, delimited by 3 µm intervals, 63% of aggregates from Q82::GFP animals were in the smallest range (0–3 µm) and 25% were 3–5 µm in size. Animals co-expressing Q82::GFP and TOR-2 demonstrated far less variability in aggregate size with 90% of the aggregates in the smallest size group (0–3 µm). Conversely, aggregates from animals co-expressing Q82::GFP and TOR-2(Δ368) demonstrated a large degree of size variability in the distribution across all categories.
Overexpression of mutant TOR-2 adversely affects growth of C. elegans
A generalized growth defect was apparent within the Q82::GFP strain. This strain exhibited a reduced growth rate (as judged by larval staging at specific time points) in comparison to wild-type animals (24). Co-expression of both wild-type and mutant torsin with Q82::GFP was performed in order to determine if TOR-2 potentially alleviated this apparent homeostatic burden (Table 2). While co-expression of the wild-type tor-2 gene product had no discernable effect on the growth inhibition associated with Q82::GFP animals, TOR-2(Δ368) co-expression significantly exacerbated the growth inhibitory effect. Some 71% of animals containing the transgene encoding TOR-2(Δ368) were still at the L1/L2 stage of development compared with 46% of Q82::GFP animals at 48 h after parental egg laying (Table 2). Neither wild-type TOR-2 nor TOR-2(Δ368) co-expression with Q19::GFP changed the growth rate of animals significantly.
TOR-2 dependent suppression of aggregation is maintained as animals age
Upon closer examination of our transgenic integrated Q19::GFP line, we discovered that Q19::GFP animals also developed aggregates when they reached adulthood. Adult worms expressing Q19::GFP, Q19::GFP+TOR-2, or Q19::GFP+ TOR-2(Δ368) were analyzed each day for 6 days post-adulthood and aggregate sizes were scored (Fig. 5). Throughout the assay period aggregate sizes in these isogenic Q19::GFP animals or Q19::GFP+TOR-2 animals remained constant as the Q19::GFP line had an average aggregate size of 7.4 µm on day 1 post adulthood and 8.1 µm on day 6 post adulthood (Fig. 5, squares). Likewise, Q19::GFP co-expressing TOR-2 had an average aggregate size of 4.4 µm on day 1 and 3.5 µm on day 6 (Fig. 5, circles). Notably, worms co-expressing wild-type TOR-2 had significantly smaller aggregates (P<0.001) throughout the course of the analysis, with an overall average aggregate size of 3.9 µm (SE±0.66) compared with an average aggregate size of 8.2 µm (SE±1.20) for Q19::GFP animals. Likewise, Q19::GFP aggregates in worms co-expressing TOR-2(Δ368) were significantly larger than either other group of worms (P<0.001) with an overall average aggregate size of 12.8 µm (SE±1.59). Furthermore, these aggregates exhibited a tendency toward an increase in size as time progressed (Fig. 5, triangles). On the first day, the average aggregate size was 10.3 µm; by day 4 this increased to 12.8 µm, and on the final day of analysis the aggregates averaged 15.0 µm in size. Thus, the suppressive effects of wild-type TOR-2 overexpression qualitatively and quantitatively persisted as animals aged whereas the mutant torsin was ineffective at containing the size of protein aggregates over time.
TOR-2 localizes to the sites of protein aggregation
Using a TOR-2-specific affinity-purified antibody, we proceeded to examine the localization of this protein in animals containing Q82::GFP aggregates. As shown in Fig. 6 (A–C), TOR-2 clearly localizes to the sites of protein aggregation in a tight ring-like formation completely surrounding the inclusions. This is similar to the manner by which human torsinA co-localizes to sites of α-synuclein aggregation in Lewy bodies of Parkinson's patients (9,10). Furthermore, just as is the case with Lewy bodies, we have shown that the sites of TOR-2/polyglutamine association are also sites of concentrated ubiquitin staining (Fig. 6, G–I). Immunolocalization of TOR-2 in animals expressing TOR-2(Δ368) did not lead to any discernible change in the cellular distribution of either TOR-2 (Fig. 6, D–F) or ubiquitin (not shown).
The subcellular localization of Q82::GFP aggregates has been previously demonstrated to be perinuclear (24). Both human torsinA and C. elegans OOC-5 have also been localized to the ER (18,26). To examine the intracellular localization of TOR-2 in the absence of polyglutamine proteins in our expression system, we generated transgenic nematodes expressing both tor-2 and a C. elegans ER-marker protein, TRAM (a protein involved with ER translocation), under the control of body-wall muscle-specific promoters (27). Strong co-localization of TOR-2 with an ER-marker protein was clearly seen in these animals (Fig. 7). These observations, taken together with our data directly implicating torsins in protein folding, strongly imply that these proteins are ER-resident molecules. In this regard, it is also notable that both the predicted nematode tor-1 gene product and the tor-2 cDNA use isolated encode proteins that end in an amino acid sequence (NDEL) reminiscent of an ER-retention signal (Fig. 1).
Conservation of chaperone activity among torsin proteins
We extended our studies of torsin activity to two additional members of this protein family, the C. elegans ooc-5 gene product and human torsinA. Co-expression of the human DYT1 cDNA in body-wall muscles of C. elegans with a Q82::GFP fusion resulted in a marked decrease in aggregates and a return to diffuse GFP fluorescence (Fig. 8A). This indicates that the sequence homology shared among torsins proteins may extend to functional homology and is conserved across species. Interestingly, ooc-5 overexpression yielded more variable results. In general, we observed that production of the OOC-5 protein alone was less effective in reducing Q82::GFP aggregate size than either TOR-2 or torsinA. Partial restoration of diffuse body-wall GFP fluorescence was apparent in some animals (Fig. 8B), but a majority of worms still displayed aggregates of polyglutamine–GFP that were quantitatively larger than Q82::GFP+TOR-2 or Q82::GFP alone (data not shown). In contrast, when tor-2 and ooc-5 cDNAs were jointly expressed in combination, a qualitatively larger decrease in aggregate size was observed and a return to predominantly diffuse GFP expression was evident (Fig. 8C). These data indicate a cumulative effect of torsin co-expression may serve to enhance their overall activity in suppressing misfolded protein formation. This may be due to a strict dosage effect or perhaps putative complex formation between torsins may facilitate their activity.
As defined, the AAA+ (ATPases associated with diverse cellular activities) protein superfamily consists of a functional assortment of molecules such as proteosome subunits, molecular motors, heat-shock proteins, transcriptional regulators, and more (20,21). While the distant sequence similarity between torsins and molecular chaperones is suggestive of a common functionality, a priori experimental evidence of a role for torsins in mediating protein folding has not been previously demonstrated.
Given the role of molecular chaperones in mediating protein conformations, it is significant that they have the ability to abrogate cellular and behavioral symptoms of polyglutamine-induced protein aggregation. This has been shown to be the case in vivo using multiple genetic model systems including S. cerevisiae (28), D. melanogaster (22,29) and C. elegans (24), as well as different mammalian cell types (30,31). One key distinction between these studies and those we describe herein is that, while both torsins and specific molecular chaperones have the capacity to suppress polyglutamine-dependent aggregation, unlike HSPs, mutations in torsinA itself result in a neurological disorder. Therefore, in the case of torsion dystonia, an apparent change in the neuroprotective activity of a chaperone-like protein may be sufficient to provoke the disease state. Increasing scrutiny of toxic misfolded protein intermediates as causative agents in neurological disorders underscores the significance of functional characterization of novel and disease-associated effectors of protein aggregation (8,15,16).
It has been demonstrated that the absence of the glutamic acid residue in torsinA (ΔE302/303) that causes dystonia results in altered subcellular distribution of this protein and the formation of membraneous cytoplasmic anomalies (26,32). It remains unclear if cytoplasmic manifestations associated with overexpression of mutant torsin gene products have a direct consequence on protein activity. Our data indicate that a single amino acid base deletion in the C-terminus of the nematode TOR-2 protein [TOR-2(Δ368)] results in a dramatic change in the ability of this protein to suppress protein aggregation without an apparent change in TOR-2 localization or cellular morphology. This suggests that changes in torsin activity may be functionally dissociated from cytoplasmic aberrations. Further evidence for a separation of torsin function from associated intracellular inclusions comes from a recent report demonstrating that the 18-bp deletion (ΔF323-Y328) in torsinA linked to early-onset dystonia does not alter the localization of torsinA or result in the formation of ER-derived inclusions in transfected human cell cultures (33). While these combined results point to the functional significance of this overall region of torsin-related proteins, they do not preclude the likelihood of differential effects being caused by alternate mutations in distinct regions of torsin proteins. Mutations in the C-terminal regions of some HSPs have also been shown to block complex formation and it is possible similar deficits are associated with torsin dysfunction (34). Subsequent analyses of TOR-2 activity, coupled to detailed examination of systematic alterations in protein structure, will allow for direct linkage of genetic abnormality to both changes in cytoplasmic architecture and protein function via this nematode assay system.
The demonstrated conservation of activity between nematode and human torsin proteins in suppressing protein aggregation implies these proteins share a common function. As other AAA+ family proteins have been shown to function as multimers, it is also not surprising that combinations of torsin proteins may efficiently work together as well (21). Interestingly, the tor-1 and tor-2 genes lie adjacent to each other in the same direction on chromosome IV of C. elegans separated by just 348 bases of sequence. This is particularly intriguing for two reasons: (i) the genes encoding human torsinA and a related protein, torsinB, are also adjacently positioned on human chromosome 9q34, but lie in opposite directions (4,5,35); and (ii) C. elegans genes often lie in operons (36). In this regard, it is likely that tor-1 and tor-2 of C. elegans are co-expressed as a functional unit and may act in tandem to carry out their cellular functions. Unfortunately, multiple attempts on our part to isolate the worm tor-1 cDNA have been unsuccessful and preclude such analyses at the present time. Preliminary expression analyses indicate that these genes are lowly expressed in just a limited subset of neurons of C. elegans (Cao, S., and Caldwell, G., unpublished data), perhaps accounting for the difficulty in cDNA isolation.
In lieu of this, our evidence that combined ooc-5 and tor-2 co-expression appears to enhance the suppressive effects of torsins on protein aggregation is possibly indicative of a cooperative nature between torsin gene products. While this may be simply due to an additive effect of expressing two putative chaperones, a similar functional relationship may also hold true for human torsinA and torsinB given the overlapping expression of these gene products in neuronal tissues (37). Basham and Rose have also shown a requirement of C. elegans OOC-3, a novel protein of unknown function, for the proper localization of OOC-5 to the ER (18). The absence of OOC-3 activity in our ectopic expression assay may account for differences between OOC-5 and other torsins in ameliorating protein aggregation. Genetic screens aimed toward the identification of additional effectors of torsin protein function will clearly provide further insights into the mechanism by which these proteins act.
The co-localization of C. elegans TOR-2 with protein aggregates and ubiquitin is reminiscent of cellular bodies that have been designated ‘aggresomes’ (38). These inclusions, which also contain ubiquitinated proteins, proteosome subunits, and chaperones, are formed in response to excess misfolded proteins. It is intriguing to speculate a role for torsins in the ATP-dependent retrotranslocation of misfolded proteins at the ER, perhaps as functional components of the translocon complex (39). Whereas microarray studies in yeast have pinpointed genes that are up- and down-regulated with regard to their expression in the unfolded protein response pathway (40), no similar global analyses have been performed for metazoan species, wherein the complexity of such pathways is likely greater. Torsins may represent an example of one such difference, as they are a protein family strictly found in multicellular organisms.
Dystonia is not a polyglutamine disease and the experiments performed in this study are not designed to elucidate a putative role for torsins in CAG repeat disorders. As it happens, these results do serendipitously define yet another specific family of proteins that might serve a neuroprotective function for such maladies, in addition to others involving protein misfolding. These data have taken on greater significance given the documented co-localization of human torsinA with α-synuclein in Lewy bodies and a report that heat-shock proteins (HSP70) can suppress neurotoxicity associated with misfolding of α-synuclein in Drosophila (23). The direct association between torsinA and α-synuclein inclusions in Lewy bodies of Parkinson's diseased brains warrants further evaluation of the relationship between torsins and synuclein protofibillar toxicity. In this regard, α-synuclein represents a far more natural target of torsin activity than polyglutamine repeat-containing proteins or for heat-shock proteins.
Taken together, these data represent the first experimental evidence of a cellular function for members of the torsin family and correlate well with the concept that these gene products may work as molecular chaperones as part of a cellular mechanism for the management of protein misfolding. Defects in this mechanism may have significant consequences on neuronal activity. In this regard, dystonia is possibly a consequence of the inability of neurons to properly respond to either environmental or physiological stress-induced changes in protein structures. For example, proteins implicated in the secretion of dopamine may be altered in subtle ways to interfere with their activity. Evidence of a functional role for torsins in facilitating protein conformational change may also have clinical implications considering only 30–40% of individuals carrying DYT1 (TOR1A) mutant alleles display symptoms of dystonia. This suggests that the protein targets of torsin function may be more directly responsible for the cellular malfunction associated with the disease. Continued investigation of torsin activity in C. elegans and other animal models will undoubtedly provide additional insights into the molecular mechanisms underlying this movement disorder and perhaps related conditions involving the cytoplasmic toxicity associated with protein aggregation.
MATERIALS AND METHODS
Plasmid constructs and mutagenesis
The tor-2 cDNA was isolated from whole worm mRNA using RT-PCR with the following primers: primer 1 (5′-AACGCGTCGACAATGAAAAAGTTCGCTGAAAAATGGTTTCTATTG-3′) and primer 2 (5′-AAGGCCTTCACAACTCATCATTAAACTCTTTCTTCG-3′). Briefly, total RNA was isolated from a mixed population of C. elegans using TriReagent (Molecular Research Center) followed by mRNA isolation using the PolyATtract mRNA Isolation System III (Promega) and cDNA synthesis using SuperScriptII (Invitrogen). The predicted ORF [Y37A1B.13 on cosmid Y37A1B corresponding to a portion of C. elegans Chromosome IV (19)] was confirmed by DNA sequencing. Mutant versions of the tor-2 cDNA were generated using site-directed PCR as follows. To obtain the Δ368 mutant form of tor-2, an initial round of PCR was performed to generate an approximately 1 kb cDNA (corresponding to amino acids 1–367 of TOR-2) using primers 1 and 3 (5′-GGGAAAAATTCAAGATCAAGAACTCTTTGCATG-3′). In parallel, an ∼200 bp fragment (corresponding to amino acids 369–412) was amplified with primers 2 and 4 (5′-CATGCAAAGAGTTCTTGATCTTGAATTTTTCCC-3′). The two fragments were then combined, diluted and amplified using primers 1 and 2 to reconstruct the complete cDNA. The mutagenesis of tor-2 was confirmed by DNA sequencing. The DYT1 cDNA for encoding human torsinA (a gift of Ben Cravatt) was amplified using the following primers: TORSINAc-Nhe-5 (5′-CTAGCTAGCATGAAGCTGGGCCGGGCCGTG-3′) and TORSINAc-Kpn-3 (5′-GGGGTACCTCAATCATCGTAGTAATAATCTAAC-3′). Partial ooc-5 cDNA fragments cloned into plasmids, pSEB18 and pSEB20, were generously provided by Lesilee Rose (18). To generate a full-length cDNA for ooc-5, two overlapping fragments from these two pieces were individually amplified using primers OOC5cNhe-5 (5′-CTAGCTAGCATGAAACTTGATTATGTTCTTCTCC-3′) and OOC5cXX-3 (5′-TCGAATAATAGTCGAGGACGG-3′) for the 5′ region of the gene in pSEB18 and amplified a region ∼600 bp in length. Similarly, the 3′ end of the cDNA was amplified from pSEB20 using primers OOC5cXX-5 (5′-TGTGCTTTGTCAACACGGAATTG-3′) and OOC5cKpn-3 (5′-GGGGTACCTTATAGTTCATCGTCGAAATGAAC-3′) and yielded a 1 kb product. These two amplified products of ooc-5 were then diluted back and mixed together for one more round of PCR using the external OOC5cNhe-5 and OOC5cKpn-3 primers. All torsin cDNAs were subcloned in between the NheI and KpnI sites of the C. elegans unc-54 promoter-containing expression vector pPD30.38 (41) and completely sequenced to verify the final products prior to expression. Plasmids unc-54::Q19-GFP and unc-54::Q82-GFP were obtained from the Morimoto lab (24).
C. elegans protocols
Nematodes were maintained using standard procedures (42). A mixture of the plasmids encoding the polyglutamine-GFP fusions and torsin constructs were co-injected into the gonads of early-adult hermaphrodites at a concentration of 0.1 µg/ml. The injection mixtures contained vectors pPD30.38-Q19::GFP or pPD30.38-Q82::GFP, and/or either pPD30.38-tor-2, pPD30.38-tor-2(Δ368), pPD30.38-ooc-5, or pPD30.38-DYT1. pRF4 [rol-6 (su1006) dominant marker (43)] was used to select for successful transformation in all transgenic animals generated in this study (as was GFP fluorescence when appropriate). For each combination of injected DNAs, multiple worm lines expressing stable extrachromosomal arrays were obtained. Following stable transmission of the arrays, multiple stable lines for each injection mixture were compared and analyzed before a consensus line was integrated into the genome. Integration was performed using gamma irradiation with 3500–4000 rad from a Cobalt 60 source (44). Integrated transgenic lines were named according to appropriate C. elegans nomenclature as follows: Q19::GFP [strain UA3 (baIn3)], Q82::GFP [strain UA4 (baIn4)], Q19::GFP+TOR-2 [strain UA5 (baIn5)], Q82::GFP+TOR-2 [strain UA6 (baIn6)], Q19::GFP+TOR-2(Δ368) [strain UA7 (baIn7)], Q82::GFP+TOR-2(Δ368) [strain UA8 (baIn8)]. Extrachromosomal arrays of transgenes were maintained as stable lines for the following: Q82::GFP+OOC-5 [strain UA9 (baEx9)], Q82::GFP+torsinA (DYT1) [strain UA10 (baEx10)], Q82::GFP+OOC-5+TOR-2 [strain UA11 (baEx11)], and TOR-2+CFP-TRAM (no polyglutamine) [strain UA12 (baEx12)]. Synchronized populations of worms for growth assays were performed by collecting embryos laid by adult animals during 4 h periods. For each line analyzed a minimum of ∼300 animals were scored at the L3 larval stage following ∼48 h of growth at 25°C.
Worms were examined using a Nikon Eclipse E800 epifluorescence microscope equipped with Endow GFP HYQ and Texas Red HYQ filter cubes (Chroma Inc.). Images were captured with a Spot RT CCD camera (Diagnostic Instruments Inc.). MetaMorph Software (Universal Imaging Inc.) was used for pseudocoloration of images, image overlays and aggregate size quantitation. For each worm line analyzed, average aggregate size was determined by capturing images of all aggregates in the posterior region of 30 L3-staged animals (for Q82::GFP aggregates) or all aggregates in 30 adult animals/day (for Q19::GFP analyses) at 1000× magnification. Pixel area was converted to µm in the MetaMorph software system and was directly downloaded to Excel spreadsheets for further analysis. Statistical analysis of aggregate size was performed by ANOVA using Statistica (SPSS Software).
Preparation of C. elegans extracts for immunoblotting
Extracts were prepared following growth of Bristol N2 worms to near-confluency on five 150 mm plates. Worms were collected by using M9 buffer and concentrated by centrifugation in a 1.5 ml microcentrifuge tube. The worm pellet was resuspended and lysed in 0.5 ml worm lysis buffer (100 mm Tris, pH 6.8, 2% SDS, 15% glycerol) by boiling for 5 min. This lysate was centrifuged again for 10 min and the supernatant was collected and concentrated using a Centricon YM-50 column (Millipore). Protein concentration was determined by using the BCA kit (Sigma).
TOR-2 antibody production, SDS–PAGE, and western blot analysis
A polyclonal peptide antibody was generated to a unique N-terminal portion of the TOR-2 protein (ETDIFNY-HALYKDFDNK) and used to immunize rabbits (ResGen). The specificity of the TOR-2 antibody was confirmed by western blot analysis using whole worm extracts (in all cases pre-immune and secondary antibody only exhibited no significant immunoreactivity) using standard procedures (45). Specifically, 5–15 µg of total proteins from each sample of nematode extract were loaded on a 10% pre-cast Tris–glycine SDS–PAGE gel (Bio-Rad) and separated for 35 min at 160 V. Proteins were transferred to Hybond nylon membranes using a semi-dry transfer cell (Bio-Rad). For detection of TOR-2 protein, non-affinity purified TOR-2 antisera (undiluted) was used to probe the blot, whereas mouse anti-polyglutamine monoclonal antibody (Chemicon) was used (1:2000 dilution) for polyglutamine-containing proteins. Membranes were blocked for 1 h in blocking buffer [2.5% non-fat dried milk, 0.5% NaCl, 0.5% Tris (1 m) pH 8.0, 0.025% Tween-20], probed for 30 min with 8 ml diluted primary antibody [8 µl TOR-2 antibody, 80 µl blocking buffer, 7.9 ml TBS-T (242 g Tris-base, 8 g NaCl, 3.8 ml 1 m HCl, 1 ml Tween-20, pH 7.6 per liter) and then washed with blocking buffer 4×5 min. Following these washes, membranes were probed for 30 min. with 8 ml diluted (1:2000) secondary antibody [4 µl horseradish peroxidase-conjugated anti-rabbit IgG for anti-TOR-2 detection, (Amersham-Pharmacia), 80 µl blocking buffer, 7.9 ml TBS-T]. A 1:10,000 dilution of conjugated anti-mouse IgG (Amersham-Pharmacia) was used as a secondary antibody in the polyglutamine blots. Following probing, membranes were washed again with blocking buffer (4×5 min), then once again with TBS-T for 5 min. Signals were detected by pipetting 2 ml of mixed SuperSignal (Pierce) solution onto membranes and incubating for 5 min in the dark, wrapped in Saran Wrap, exposed to film (Hyperfilm ECL) for 1 min and then immediately developed. Mouse anti-actin antibody staining (1:4000 dilution; IGN) was used as control for protein loading. Basham and Rose have shown that a related worm torsin protein, OOC-5, was highly expressed and localized in early embryogenesis of C. elegans (18). Wild-type embryos stained with our TOR-2-specific anti-peptide antibodies did not exhibit such expression. Moreover, a concomitant increase in a single band on a western blot, approximately the size of the predicted TOR-2 protein was observed only when the tor-2 gene is overexpressed in multiple transgenic lines of worms under unc-54 promoter control (Fig. 2). This is highly indicative of a specific increase in TOR-2 protein only, as the ooc-5 gene is not overexpressed in these animals.
Worm strains were washed from worm plates with water, rinsed twice and then resuspended in a small volume of water. An ∼30 µl drop of worms was placed on a subbed poly-lysine slide and covered with a 18 mm2 coverslip. The slide was placed on a frozen metal block for 10 s and then immediately moved to a 95°C block for 3 s. The heating/cooling process was then repeated once or twice. Excess water was removed from the sides of the coverslip and the slide was immersed in liquid nitrogen for 15 min. The coverslip was pried off with a razor blade and the slide placed in fixative (80 mm KCl, 20 mm NaCl, 10 mm Na2EGTA, 5 mm spermidine HCl, 15 mm NaPIPES, pH 7.4, 25% methanol, 1% paraformaldehyde) for 1 h. All steps following worm fixation onto microscope slides performed according to the Finney–Ruvkun antibody staining procedure (46). The fixed worms were incubated with affinity-purified anti-TOR-2 primary antibody at a concentration of 1:200, whole serum anti-ubiquitin primary antibody (Sigma) at 1:200 or anti-GFP monoclonal antibody (Clontech) at 1:100 dilution (to detect TRAM-CFP). Secondary Alexa Fluor 594 goat anti-rabbit IgG (Molecular Probes) was used at 1:800 to detect TOR-2 and ubiquitin and Alexa Fluor 488 goat anti-mouse IgG (1:800) was used to detect GFP in fixed animals. To facilitate the ER co-localization study in Fig. 7, anti-GFP monoclonal antibodies were used to detect CFP-TRAM due to very rapid photobleaching of CFP in these animals. In all cases, fixed, stained worms were mounted with Hydromount (National Diagnostics) and imaged with a Nikon E800 research microscope, Spot RT digital camera and Metamorph software. Images were processed with Adobe Photoshop 7.0.
We would like to acknowledge the contributions of the following individuals to this work: Scott Reese for his assistance with the statistical analyses, John DeModena for advice on worm immunocytochemistry, Chad Rappleye for advice on protein biochemistry, Harriett Smith-Somerville for assistance with graphics, Clay Blanton for his assistance in isolation of the tor-2 cDNA, Lesilee Rose for discussions and for sharing ooc-5 partial cDNA fragments, Ben Cravatt and Karen Kustjedo for the cDNA encoding human torsinA, Melissa Rolls and Tom Rapoport for the CFP-TRAM ER-marker vector, Lynn Boyd for discussions, Shelli Williams for assistance with worm stocks, and all members of the Caldwell Lab for their collegiality and teamwork. Special thanks go to Rick Morimoto and Jim Morely for generously sharing with us the worm polyglutamine vectors and strains. We gratefully acknowledge the Caenorhabditis Genetics Center, supported by a grant from the NIH National Center for Research Resources, for providing nematode strains. This work was funded by grants from the Dystonia Medical Research Foundation (G.A.C.) and an Undergraduate Research Program Grant from the Howard Hughes Medical Institute to The University of Alabama. G.A.C. is a Basil O'Connor Scholar of The March of Dimes Birth Defects Foundation. This research is dedicated in friendship and with hope to the Gebeloff Family of West Caldwell, New Jersey.
To whom correspondence should be addressed at: Department of Biological Sciences, The University of Alabama, Box 870344, Tuscaloosa, AL 35487-0344, USA. Tel: +1 2053489926; Fax: +1 2053481786; Email: email@example.com
|Aggregate size (µm)||Aggregatesa (%)|
|Aggregate size (µm)||Aggregatesa (%)|
aPercentage of aggregates analyzed from the posterior region in 30 L3-staged larvae of each isogenic strain.
|Worm strain||Developmental stage of animals (%)|
|Worm strain||Developmental stage of animals (%)|
Gravid adults were placed on freshly seeded plates and allowed to lay embryos for 3 h; 48 h later the animals were developmentally staged. Sample number (n) represents the total number of scored animals from four separate trials. All animals hatched and continued development through adulthood.