Archaeal Proteasomes Effectively Degrade Aggregation-prone Proteins and Reduce Cellular Toxicities in Mammalian Cells*

The 20 S proteasome is a ubiquitous, barrel-shaped protease complex responsible for most of cellular proteolysis, and its reduced activity is thought to be associated with accumulations of aberrant or misfolded proteins, resulting in a number of neurodegenerative diseases, including amyotrophic lateral sclerosis, spinal and bulbar muscular atrophy, Parkinson disease, and Alzheimer disease. The 20 S proteasomes of archaebacteria (archaea) are structurally simple and proteolytically powerful and thought to be an evolutionary precursor to eukaryotic proteasomes. We successfully reproduced the archaeal proteasome in a functional state in mammalian cells, and here we show that the archaeal proteasome effectively accelerated species-specific degradation of mutant superoxide dismutase-1 and the mutant polyglutamine tract-extended androgen receptor, causative proteins of familial amyotrophic lateral sclerosis and spinal and bulbar muscular atrophy, respectively, and reduced the cellular toxicities of these mutant proteins. Further, we demonstrate that archaeal proteasome can also degrade other neurodegenerative disease-associated proteins such as α-synuclein and tau. Our study showed that archaeal proteasomes can degrade aggregation-prone proteins whose toxic gain of function causes neurodegradation and reduce protein cellular toxicity.

The 20 S proteasome is a ubiquitous, barrel-shaped protease complex responsible for most of cellular proteolysis (1) and is formed by four stacked seven-membered rings (2). The ␣-type subunits, which are proteolytically inactive (3), form the outer rings, and the ␤-type subunits, which contain the active site (4), form the inner rings of the complex (5). The 20 S proteasome of archaebacteria (archaea) consists of only one type of each of the ␣and ␤-subunits and is thought to be the evolutionary ancestor of the eukaryotic proteasome (6), which is quite similar in architecture to that of archaea but is composed of seven different ␣and seven different ␤-subunits (6). Archaea do not have the ubiquitin recognition system for protein degradation and are thought to have unidentified tags in its degradation pathway (7). Like eukaryotic cells, archaea also have a regulatory complex for the 20 S proteasome, known as proteasome-activating nucleotidase (PAN) 2 (8). PAN is an evolutionary precursor to the 19 S base in eukaryotic cells and thought to be necessary for efficient archaeal 20 S proteasomal protein degradation (8). However in vitro, the archaeal 20 S proteasome has been reported to rapidly degrade polyglutamine aggregates without the help of PAN (9). This PAN-independent degradation by the archaeal 20 S proteasome inspired us to introduce and test a novel proteolytic facility in mammalian cells. We have chosen the archaeal Methanosarcina mazei (Mm) 20 S proteasome, because its optimal growth temperature is around 37°C, making it suitable to examine its proteasomal effects in mammalian cells.
The eukaryotic ubiquitin-proteasome system degrades aberrant or misfolded proteins that could otherwise form potentially toxic aggregates (10). These aggregate formations in cells are related to the pathogenesis of several common aging-related neurodegenerative diseases, including Parkinson disease (PD), amyotrophic lateral sclerosis (ALS), polyglutamine diseases (e.g. Huntington disease, some spinocerebellar ataxias, and spinal and bulbar muscular atrophy), and Alzheimer disease (AD), which are thought to be associated with the reduced activities of the proteasome (11)(12)(13)(14)(15). However, a critical cause of the accumulation of abnormal proteins remains unclear. Solving this common aspect of many neurodegenerative disorders would be a breakthrough in treating these diseases.
In the present study, we show that the Mm proteasome functions in mammalian cells to accelerate the degradation of the following aggregation-prone proteins: mutant superoxide dismutase-1 (SOD1), a causative protein of familial ALS; mutant androgen receptor (AR) with expanded polyglutamine tract, a causative protein of spinal and bulbar muscular atrophy; ␣-synuclein, an accumulated protein in PD; and tau, an accumulated protein in AD.
Glycerol Density Gradient Centrifugation-Cells grown on a 10-cm dish were lysed in 1 ml of 0.01 M Tris-EDTA, pH 7.5, by two freeze-thaw cycles, and the lysates were centrifuged for 15 min at 15 000 ϫ g at 4°C. The cleared supernatants were loaded on the top of a 36-ml linear gradient of glycerol (10 -40%) prepared in 25 mM Tris-HCl buffer, pH 7.5, containing 1 mM dithiothreitol and then centrifuged at 80,000 ϫ g for 22 h at 4°C in a Beckman SW28 rotor (Beckman Coulter Inc.). Following centrifugation, 37 fractions (1.0 ml each) were collected from the top of the tubes with a liquid layer injector fractionator (model number CHD255AA; Advantech) connected to a fraction col- The deleted sequences of the ⌬␣-subunit are depicted. The T1C ␤-subunit (m␤1) has three mutated base pairs (a to t, c to g, and a to t). B, Western blot analysis with anti-proteasome ␣-subunit, anti-proteasome ␤-subunit, and anti-His antibodies. C, Ni 2ϩ -NTA pulldown assay. Pulled down proteins run on SDS-PAGE were probed with anti-proteasome ␣-subunit. D, chymotrypsin-like activity of the Ni 2ϩ -NTA pulled down samples. This protease activity gradually became higher after transfection. Error bars, S.D. (n ϭ 3). E, glycerol gradient centrifugation experiment: Mm proteasome ␣and ␤-subunits fractionated into nearly the same fractions as did the human 20 S proteasome subunits ␣1 and ␣5, ␣␤Ϫ and ␣␤ϩ, indicating that cells were transfected with mock and Mm proteasome ␣␤, respectively.
lector. 200 l of each fraction was precipitated with acetone; the pellets were lysed with 50 l of sample buffer and then used for SDS-PAGE followed by Western blotting. The immunostained bands were quantified using ImageGauge software (Fuji Film).
Ni 2ϩ -NTA Pulldown-HEK 293 cells grown on 10-cm dishes, transfected with Mm proteasome ␣ (as a control), ␣␤, ⌬␣␤, and ␣m␤1, were lysed by two freeze-thaw cycles in 1 ml of phosphate-buffered saline buffer and centrifuged at 3000 ϫ g. Proteasome complexes were pulled down from the supernatants with 200 l of Ni 2ϩ -NTA-agarose, washed 4 times in 4 ml of 10 mM imidazole/phosphate-buffered saline buffer, and eluted in 2 ml of 250 mM imidazole/phosphate-buffered saline buffer. Samples were then boiled and subjected to Western blotting.
Measurement of the Proteasome Activity-HEK 293 cells grown on 10-cm dishes were transfected with Mm proteasome ␣ (as a control), ␣␤, ⌬␣␤, and ␣m␤1. 12, 24, and 48 h after transfection, the cells were lysed and pulled down with Ni 2ϩ -NTA. The chymotrypsin-like activity of 500 l of the Ni 2ϩ -NTA pulled down samples were assayed colorimetrically after 12-h incubations at 37°C with 100 mM Suc-LLVY-amino-4methylcoumarin (Sigma) by a multiple-plate reader (Power-scanHT, Dainippon Pharmaceutical). The assay was carried out in triplicate and statistically analyzed by one-way analysis of variance.
Immunocytochemistry-Neuro2a cells grown on glass coverslips were co-transfected with pEGFP-N1-SOD1 and Mm proteasome ␣and His-tagged ␤-subunit. 48 h after transfection, cells were fixed, blocked, and incubated with anti-His antibody overnight at 4°C. After washing, samples were incubated with Alexa-546-conjugated anti-mouse antibody (Molecular Probes, Inc.) and visualized with an Olympus BX51 epifluorescence microscope.
Cycloheximide Chase Analysis-Neuro2a cells grown on 6-cm dishes were transfected with 1 g of pcDNA3.1/MycHis-SOD1 with mock (0.6 g), Mm proteasome ␣m␤1 (0.3 g each), or Mm proteasome ␣␤ (0.3 g each). 24 h after transfection, cycloheximide (50 g/ml) was added to the culture medium, and the cells were harvested at the indicated time points. The samples were subjected to SDS-PAGE and analyzed by Western blotting with anti-SOD1 antibody.
Pulse-chase Analysis-Neuro2a cells grown on 6-cm dishes were transfected with 1 g of pCMV-Tag4-SOD1 G93A with mock (0.6 g) Mm proteasome ␣m␤1 (0.3 g each) or Mm proteasome ␣␤ (0.3 g each). 24 h after transfection, cells were pulse-labeled with [ 35 S]Cys for 60 min and harvested at the indicated time points. After the immunoprecipitation by anti-FLAG antibody (M2; Sigma), the samples were subjected to SDS-PAGE, phosphor-imaged (Typhoon 9410; Genaral Electric Co.), and statistically analyzed by one-way analysis of variance.
Cell Viability Analysis-HEK293 cells were grown on collagen-coated 96-well plates and co-transfected with pcDNA3.1/MycHis-SOD1 (WT, G93A, and G85R) and Mm 20 S proteasome ␣␤, ␣m␤1, or mock in 12 wells each. The MTSbased cell proliferation assays were performed after 48 h of transfection. Absorbance at 490 nm was measured at 37°C in a multiple-plate reader (PowerscanHT, Dainippon Pharmaceutical). The assay was carried out in triplicate and statistically analyzed by one-way analysis of variance.
To confirm protein expression of the Mm subunits, HEK293 cells transfected with mock, ␣, ⌬␣, ␤, or m␤1 were lysed, subjected to SDS-PAGE, and immunoblotted with anti-proteasome ␣-subunit, antiproteasome ␤-subunit, and anti-His antibodies. Fig. 1B demonstrates that the ␣and ␤-subunit antibodies detected the Mm proteasome ␣-subunit at 26 kDa, the ⌬␣-subunit around 25 kDa, and the ␤-subunit at 22 kDa, respectively, and faintly recognized endogenous human proteasome subunits. A Ni 2ϩ -NTA pulldown assay showed that the Mm proteasome ␣and ⌬␣-subunits cosedimented with the Mm proteasome ␤and m␤1-subunits but not with mock (Fig. 1C), and protease activity of the pulled down samples of the cells lysed 48 h after transfection showed significantly higher chymotrypsin-like protease activity in the Mm proteasome ␣␤ than in the ␣m␤1 or mock-transfected samples (Fig. 1D). This protease activity was confirmed to become gradually higher after transfection (Fig. 1D).
Glycerol density gradient centrifugation fractionated the ␣␤, ⌬␣␤, and ␣m␤1 complexes of the Mm proteasome into nearly the same fractions as those of the human 20 S proteasome subunits ␣1 and ␣5 (Fig. 1E, data not shown for ⌬␣␤ and ␣m␤1). Moreover, of the anti-His-immunoblotted bands (Fig. 1E), the density of staining in fractions 20 -25 accounts for about 80 -90% of the total anti-His staining. That these fractions constitute the majority of the anti-␣ staining as well suggests that about 80 -90% of the ␤-subunit expression is incorporated into the Mm proteasome. These results suggested that the Mm proteasome ␣-, ⌬␣-, ␤-, and m␤1-subunits could properly assemble to form four stacked seven-membered rings and that an active Mm proteasome could be reproduced in mammalian cells. The cells expressing Mm proteasome ⌬␣␤ displayed cellular toxicity, whereas the cells expressing Mm proteasome ␣␤ showed little toxicity  (data not shown); thus, further experiments were carried out with Mm proteasomes ␣␤ and ␣m␤1.
M. mazei Proteasome Degrades Specifically Mutant Superoxide Dismutase-1-We then assessed whether the Mm proteasome actually affects mutant SOD1 protein (SOD1 G85R , SOD1 G37R , SOD1 G93A , and SOD1 H46R ) expression. In cultured cells, mutant SOD1 G85R , SOD1 G37R , and SOD1 G93A are more likely to form aggregates than is SOD1 H46R (16), and cases of familial ALS expressing these mutant forms are also more severe than those expressing SOD1 H46R . Western blot analyses demonstrated that the levels of mutant SOD1 were markedly reduced as the expression of Mm proteasome ␣␤ increased (Fig. 2). However, wild-type SOD1 levels were not affected by the expression of Mm proteasome ␣␤. Furthermore, mutant SOD1 levels were not affected by the expression of Mm proteasome containing the m␤1-subunit in all mutant species, indicating that Mm proteasomal activity was important to reduce the levels of mutant SOD1 proteins. That the expression level of SOD1 H46R was less affected by Mm proteasomal expression than other mutant SOD1 species may be associated with the lower toxicity of SOD1 H46R .
To determine whether the reduced levels of mutant SOD1 protein were due to accelerated degradation of mutant SOD1 or to the reduction of mutant SOD1 expression, we examined the stability of mutant SOD1 proteins expressed in Neuro2a cells co-expressed with Mm proteasome ␣␤, ␣m␤1, or mock (Fig. 3,  A and B). Chase experiments with cycloheximide, which halts all cellular protein synthesis, demonstrated mutant species-dependent acceleration in SOD1 protein degradation, whereas the expression levels of Mm proteasome ␣and ␤-subunits did not change (Fig. 3A). The degree of wild-type SOD1 degradation was not affected by the expression of Mm proteasome ␣␤. Pulse-chase experiments further confirmed that 35 S-labeled SOD1 G93A degradation was significantly accelerated when coexpressed with Mm proteasome ␣␤ but not with Mm proteasome ␣m␤1 or mock (Fig. 3B). These facts strongly suggest that the catalytic center in the Mm proteasome ␤-subunit is important to accelerate the degradation of mutant SOD1 proteins.
M. mazei Proteasome Reduces Cellular Toxicities of Mutant Superoxide Dismutase-1-Next, we investigated the viability of HEK293 cells evoked by SOD1 (wild-type, SOD1 G93A , and SOD1 G85R ) when co-expressed with Mm proteasome ␣␤, ␣m␤1, or mock by the MTS-based cell proliferation assay (Fig.  4). We confirmed a linear response between cell number and optical density at 490 nm between 0.85 and 1.30 (data not shown). The viability of cells expressing wild-type SOD1 with Mm proteasome ␣␤ did not change as the transfected DNA doses of SOD1 and Mm proteasome ␣␤ increased (Fig. 4A). However, the viability of cells expressing mutant SOD1 was reduced as the transfected DNA dose of SOD1 increased (Fig. 4, B and C), and this reduction was prevented by the co-transfection with Mm proteasome ␣␤ but not with Mm proteasome ␣m␤1. Toxicities of mutant SOD1 proteins are associated with the activation of caspase family proteins, especially caspase-3 (21). Using fluorescent substrates of activated caspase-3/7 as markers, we analyzed caspase-3/7 activities in the cells co-transfected with SOD1 proteins and with mock, Mm proteasome ␣␤, and ␣m␤1. Mm proteasome ␣␤ suppressed the activation of caspase-3/7, resulting in reductions of cellular toxicities of SOD1 proteins (Fig. 4D). These results show that Mm proteasome ␣␤ has a protective effect against the decrease in cellular viability evoked by mutant SOD1.
M. mazei Proteasome Co-localizes with Aggregates Formed by Mutant SOD1-In the assembly process of the archaeal proteasome, ␣-subunit assembly is required for ␤-subunit incorporation into the proteasome (20), and since the anti-His-stained ␤-subunit is restricted largely to that incorporated into the Mm proteasome (Fig. 1E), we used anti-His staining to localize the transfected proteasome in Neuro2a cells. GFP-tagged wild-type and G93A mutant SOD1 vectors were transfected along with Mm proteasome ␣␤ into Neuro2a cells, which were then fixed and immunostained with anti-His antibody. Fig. 5A shows that GFP-positive SOD1 G93A aggregates are also anti-His positive, whereas the cells expressing wild-type SOD1-GFP are diffusely stained with anti-His antibody. There were no GFP-negative inclusion bodies stained with anti-His antibody, indicating that Mm proteasome co-localizes with the inclusion bodies consisting of mutant SOD1 in the vicinity of the nucleus. The percentages of aggregate-positive cells among the GFP-positive cells were determined in Fig. 5B. SOD1 G93A aggregates were significantly reduced when co-expressed with Mm proteasome ␣␤.

M. mazei Proteasome Degrades Specifically Mutant Androgen Receptor with Expanded Polyglutamine Tract and Reduces
Its Cellular Toxicity-To demonstrate the ability of the Mm proteasome to degrade aggregation-prone proteins, we examined the AR with expanded polyglutamine tract (97-repeated glutamine; 97Q) protein, the causative protein of spinal and bulbar muscular atrophy. Similar to the results obtained with SOD1 proteins, Fig. 6A shows that in Neuro2a cells, the levels of mutant AR (97Q) were markedly reduced as the expression of Mm proteasome ␣␤ increased, but they were unaffected by the expression of the Mm proteasome ␣m␤1. On the other hand, wild-type AR (24-repeated glutamine; 24Q) levels were not affected by the expression of Mm proteasome ␣␤. Cycloheximide-chasing analysis demonstrated that the half-life of mutant AR (97Q) was reduced in the presence of the Mm proteasome but not in the presence of the mutant Mm proteasome (Fig. 6B). The viability of cells expressing mutant AR (97Q) was reduced compared with wild-type AR (24Q), and this reduction was attenuated by the co-transfection with Mm proteasome ␣␤ (Fig. 6C). These results show that Mm proteasome ␣␤ can accelerate the degradation of the aggregation-prone mutant AR with expanded polyglutamine tract and possibly protect the cells from its toxicities.
M. mazei Proteasome Degrades Other Aggregation-prone Proteins but Not Non-aggregation-prone Proteins-To determine whether the Mm proteasome degrades other aggregation-prone proteins as well, we examined its effects on ␣-synuclein (wild-type, A53T, and A30P) and six isoforms of wild-type tau protein in Neuro2a cells. The six tau isoforms contained either three (3L, 3M, and 3S) or four (4L, 4M, and 4S) microtubule binding domains in the C-terminal portion and two (3L, 4L), one (3M, 4M), or no (3S, 4S) inserts of 29 amino acids each in the N-terminal portion. Similar to the results obtained with the mutant SOD1 and AR with an expanded polyglutamine tract, the expression levels of all ␣-synuclein and tau proteins were reduced in the presence of Mm proteasome ␣␤ (Fig. 7, A and B). Although the degradations of wildtype SOD1 and AR proteins were not accelerated by Mm proteasome, the expression levels of ␣-synuclein including wild-type and all of the six forms of wild-type tau were reduced.
We also examined whether Mm proteasomes degrade non-aggregation-prone proteins such as GFP or LacZ. Fig. 7C shows that the Mm proteasome does not affect the degradation of the exogenously expressed proteins, GFP and LacZ.

DISCUSSION
In this study, we showed that the archaeal Mm proteasome ␣and ␤-subunits properly assembled to have proteolytic activity and accelerate the degradation of aggregationprone, neurodegeneration-associated proteins in mammalian cells. Archaeal proteasomes contain 14 identical active sites that, although originally classified as chymotrypsin-like, were later shown to cleave after acidic and basic residues (22), and they consist of only one type of each of the ␣and ␤-subunits (6). A comparison between archaeal and eukaryotic proteasomes in vitro showed that archaeal proteasomes are far more active in degrading poly(Q) peptides than are eukaryotic proteasomes (9). We utilized this potential power and manageability of archaeal proteasomes to degrade abnormal proteins that could not be effectively degraded by eukaryotic proteasomes. This is the first report showing that archaeal proteasomes can work to accelerate degradation of aggregation-prone proteins in mammalian cells.
Mm proteasomes promoted degradation of mutant SOD1, AR with an expanded polyglutamine tract, wild-type and mutant ␣-synuclein, and six isoforms of wild-type tau. The first two proteins, mutant SOD1 and AR with an expanded polyglutamine tract, exhibit toxicity in cell culture models. Mice overexpressing these mutant proteins display abnormal aggrega- tions in their motor neurons and significant loss of motor functions, and they have been used as disease models (23,24). Mm proteasomes accelerated the degradation of only the mutant forms of these two proteins and not that of the nonaggregating wild-type forms. Furthermore, chasing studies (Fig. 3,  A and B) confirmed our belief that Mm proteasomes directly accelerate the degradation of mutant proteins.
However, both the wild-type and two mutants of ␣-synuclein as well as six isoforms of wild-type tau were also degraded by Mm proteasomes (Fig. 7). ␣-Synuclein and tau are pathogenically different proteins from SOD1 and AR, since they are known to accumulate as wild-type proteins in the affected lesions of PD and AD, respectively. Aggregation of the presynaptic protein, ␣-synuclein, has been implicated in synucleinopathies, such as sporadic and familial PD, diffuse Lewy body disease, and multiple-system atrophy (25). In sporadic PD patients, wild-type ␣-synuclein is accumulated, and increased expression of wild-type ␣-synuclein is also observed (26). Proteasomal dysfunction has been thought to impair ␣-synuclein degradation and thereby to facilitate its aggregation (27). Three-and four-repeat wild-type tau are among the proteins characteristically detected in neurofibrillary tangles formed by paired helical filaments in sporadic AD (28). Decreased proteasomal activity has been also reported in the AD brain (29). ␣-Synuclein and tau are both relatively easily misfolded, which leads to the formation of aggregates, even in their wild-type forms (30,31), thus possibly explaining why the Mm proteasomes degraded wild-type ␣-synuclein and tau. Mm proteasomes might be able to recognize a wide range of aggregationprone proteins, whereas they do not affect the degradation of exogenously expressed nonaggregating proteins, such as GFP and LacZ, or abundant endogenous proteins, such as ␣-tubulin and glyceraldehyde-3-phosphate dehydrogenase (Fig. 7).
The question raised here is what is the molecular mechanism of such selective, mutant species-dependant degradation. Archaeal 20 S proteasomes contain proteasome-activating nucleotidase, PAN, enabling substrates to enter the proteasomes easily and effectively (8). PAN has a chaperone-like activity to unfold aggregated proteins (32) and is thought to be an evolutionary precursor to the 19 S base in eukaryotic cells (8). Archaeal recognition tags (like ubiquitin tags in eukaryotic cells) have not been identified yet. However, archaeal 20 S proteasomes have been reported to rapidly degrade polyglutamine aggregates in vitro, without the help of PAN (9). Here we confirmed that this PAN-independent degradation by Mm 20 S proteasomes could occur in mammalian cells. Since the pore diameter of the closed gate in 20 S proteasomes is estimated to be much smaller than that of aggregated proteins (33), the question is, how do the unfolded substrate proteins enter the 20 S proteasomes? One hypothesis might be that the ␣-ring in Mm proteasomes has chaperone-like activity to recognize and unfold the aggregation-prone proteins or misfolded proteins. The gated channel in the ␣-ring of the archaeal 20 S proteasomes is thought to regulate substrate entry into the proteasomes and is assumed to be in either an open (34) or a closed state (2,33) in vitro. In our experiments, the gate-free Mm 20 S proteasome ⌬␣␤ substantially reduced cell viability, but the Mm proteasome ␣␤, with the "gate," had little toxic effect on the cells and, furthermore, accelerated the degradation of mutant proteins. This would be hard to explain if the gate is always in the closed state. There is a possibility that when Mm proteasomes gather, actively or passively, near aggregationprone proteins, the ␣-ring opens its gate and unfolds the aggregated proteins, enabling them to enter the proteasomes to be degraded.
Some kinds of molecular chaperones, such as Hsp90, -70, and -27, have been reported to assist in the selective degradation of mutant SOD1 and AR proteins in proteasome degradation pathways (35,17). However, neither the protein levels of molecular chaperones (Hsp90, -70, -40, and -27) nor the ubiquitylation levels of mutant SOD1 and AR were changed in the presence of Mm proteasome ␣␤ expression (data not shown), thus supporting the idea that endogenous ubiquitin-proteasome degradation pathways possibly did not play an important role in the accelerated degradation of mutant proteins. Further study is needed to elucidate the molecular mechanisms of selective recognition of misfolded aggregation-prone proteins by Mm proteasomes.
In this paper, we demonstrated that Mm proteasomes could effectively degrade neurodegenerative disease-related aggregation-prone proteins in vivo. Further studies are needed to determine whether archaeal proteasomes can be available to treat diseases in which toxic gain of proteins is causative.