A worm model for polyglutamine protein aggregation was developed by Saytal et al. 12 and later converted into an integrated strain as described 15. This strain of worms expresses GFP that has been fused to 82 glutamines. The unc-54 promoter is used to drive expression of this fusion protein, resulting in expression in body wall muscle cells of the worm. We refer to this strain of worms as the Q82:GFP strain.

Worms from the Q82:GFP strain were subjected to RNAi of various Ubc's as well as ubiquitin (ubq-1) and a proteasome subunit (rpt-1). Following RNAi, worms were fixed and analyzed with respect to the number and size of GFP aggregates. Figure 1 shows the results of the RNAi treatments. In these experiments, embryos were collected from gravid adults and then placed directly onto RNAi feeding plates. Thus, exposure to the dsRNA began as soon as these L1 larvae hatched from the egg shell.

RNAi of most Ubc's did not cause a discernable difference in aggregate size or number, however, Figure 1 shows that knockdown of ubc-2 and ubc-22 resulted in a significant reduction in the number of aggregates per worm and an increase in the average size of aggregates. Knockdown of ubc-1, ubc-13, uev-1, and ubq-1 resulted in a reduction in the size of aggregates. RNAi of ubc-1 and ubq-1 also caused an increase in the number of aggregates. No other phenotypes in these worms were noticed besides a marked growth arrest and immobility in the ubq-1 and rpt-1 RNAi worms after 48 hours.

Because ubc-1 showed an opposite phenotype from ubc-2 and ubc-22, this suggests that there are at least two different ubiquitination events and that these different ubiquitinations may have separate effects on the formation or metabolism of the aggregates. RNAi of ubq-1, a gene encoding a polyprotein consisting of 11 ubiquitin peptides, was used to knockdown ubiquitin levels. Although the phenotype with ubq-1 RNAi is severe 28, it may not completely eliminate ubiquitin as there is a second gene, ubq-2, which also encodes ubiquitin. RNAi knockdown of ubq-1 showed a phenotype similar to that of ubc-1, suggesting that the UBC-1 mediated ubiquitination event is more prevalent or proceeds ubiquitination by UBC-2 or UBC-22.

Further evidence for the epistasis of ubc-1 came from combined RNAi experiments. When ubc-1 was knocked down in combination with ubc-2, or ubc-22, the ubc-1 phenotype was observed (Figure 1D). RT-PCR confirmed knockdown of both ubc-1 and ubc-2 in the combined RNA treatments (Figure 1E). The results from the combined RNAi experiments are consistent with a model where ubc-1 functions first in the cascade leading to ubiquitination of aggregating proteins.

There are several possible scenarios to explain how knockdown of Ubc's might affect the size and number of aggregates. First, the resulting Ubc deficiency could affect the total amount of Q82:GFP protein in the cell. Second, the solubility of Q82:GFP could be altered. Third, the composition or density of the aggregate could be affected.

In order to test the first of these three possibilities, we examined levels of Q82:GFP using western blot analysis (Figure 2A). GFP is detected as a doublet that migrates in the gel to a molecular weight of approximately 33–36 kDa. This weight agrees with the predicted weight of the Q82:GFP fusion protein (37 kDa). The nature of the difference between the two bands in the doublet is unknown. However, there does appear to be some variation in the relative abundance of the two species in some RNAi samples. Future studies will address the significance of these differences. The difference in molecular weight between the two species is too small to be explained by differences in ubiquitination and is more likely to be the result of a smaller modification such as phosphorylation. Quantitative analysis of the levels compared to an actin control show that the absolute levels of Q82:GFP are not significantly altered by the RNAi treatments. Therefore, the differences in aggregate size and number that are seen in RNAi of ubc-1, ubc-2, ubc-22, ubc-13, and uev-1 are most likely not explained by changes in the expression or stability of the Q82:GFP protein.

Differences in aggregate size and number could also be explained by changes in the solubility of the polyglutamine proteins. This has been observed in several instances. For example, overexpression of chaperone proteins can increase the solubility of polyglutamine proteins and reduce aggregation 1529. As a simple test for solubility, we analyzed the amount of soluble GFP in images of live worms, either RNAi treated or control (Figure 2B). Some soluble GFP is observed in Q82:GFP worms at this stage. However, there is no apparent change in the solubility upon any of the RNAi treatments. In contrast, examination of a strain containing a stretch of only 19 glutamines fused to GFP (Q19:GFP) clearly shows soluble GFP (diffuse green fluorescence).

The knockdown of certain Ubc's has an effect upon aggregate size or numbers. Therefore, we wanted to test whether ubiquitin localization to aggregates is affected. Worms were subjected to RNAi, then, fixed and probed with antibodies to either ubiquitin or a proteasome subunit. Figure 3 shows the results of the colocalization experiments. Staining of the control worms shows that both ubiquitin and proteasome are localized to aggregates. Knockdown of ubiquitin results in the removal of ubiquitin as well as proteasomes from aggregates. This result suggests that ubiquitin localization to aggregates may be required for proteasome localization. Proteasome localization, however, is not required for localization of ubiquitin to aggregates since ubiquitin still localizes to aggregates after rpt-1 RNAi.

RNAi of ubc-1, ubc-13 and uev-1 all resulted in smaller aggregates. Interestingly, these RNAi treatments also showed decreased localization of ubiquitin to aggregates. In addition, proteasome localization was abrogated in these knockdowns (Figure 3), consistent with the hypothesis that ubiquitin localization is a prerequisite for proteasome localization. Loss of ubiquitin from aggregates after knockdown of ubc-1, ubc-13 or uev-1 cannot be explained solely by the decrease in size since even smaller aggregates in control worms also stain positively with the ubiquitin and proteasome antibodies (Figure 3, top row).

Knockdown of ubc-2 or ubc-22, which both showed a dramatic increase in aggregate size, shows no effect on localization of ubiquitin or proteasome. Since ubiquitin localization is unaffected by RNAi of ubc-2 or ubc-22, it is less likely that either of these Ubc's is involved in directly ubiquitinating aggregate proteins. ubc-1, ubc-13, and uev-1 are more likely candidates for enzymes involved in ubiquitinating aggregate proteins.

Our results in C. elegans suggest that ubiquitination plays an important role in the aggregation pathway. In order to investigate whether this might also be true in human cells, we performed shRNA experiments in HEK293 cells.

The worm genes showing the most dramatic phenotypes included ubc-1, ubc-2 and ubc-22. We searched for human orthologs of these Ubc's and identified two very close orthologs of ubc-1 and ubc-2: Ube2A and UbcH5b, respectively. The human and worm enzymes are extremely similar with Ube2A and UBC-1 showing 84% identity and UbcH5b and UBC-2 having 94% identity. There was no close ortholog for ubc-22, however E2-25K (also known as HIP2) was the closest human enzyme with 32% identity. E2-25K is a well studied Ubc enzyme and many activities have been attributed to it such as polyubiquitin chain synthesis, cyclization of ubiquitin chains, neddylation, NF-κB processing, and interaction with Huntingtin 30313233. Although E2-25K and UBC-22 show a relatively low level of identity, they both contain a C-terminal UBA or ubiquitin associated domain as does Ubc1 of yeast.

shRNA plasmids for the human orthologs were used to knockdown Ubc levels in HEK293 cells. Two different shRNA constructs were used for each gene. shRNA plasmids were transfected together with a plasmid expressing Q81:YFP. Control and shRNA treated cells are shown in Figure 4A. Knockdown of the Ubc's was confirmed by western blot (Figure 4B).

After transfection, cells were examined for the size of aggregates as well as the number of aggregates per cell. Knockdown of Ube2A, the UBC-1 homolog, resulted in a decrease in the average size of aggregates (Figure 4C). In addition, more cells overall had aggregates and a greater percentage of cells had multiple aggregates (Figure 4D). These results are consistent with the effect of ubc-1 RNAi in worms. Therefore, the role that this enzyme plays in aggregate formation or metabolism is conserved.

Knockdown of UbcH5b, the UBC-2 homolog, resulted in a significant increase in the size of aggregates but did not affect the number of aggregates per cell (Figure 4C &4D). E2-25K (HIP2) is the Ubc that showed the lowest homology with its worm counterpart, UBC-22. E2-25K showed a modest affect on aggregate size and no affect on numbers of aggregates per cell. The RNAi phenotypes of UbcH5b and E2-25K suggest that any ubiquitination event(s) mediated by these enzymes somehow limits the size of aggregates.

Antibodies to the human Ubc's were used to probe cells expressing Q81:YFP. Figure 5 shows the localization of the Ube2A, UbcH5b and E2-25K in HEK293 cells with polyglutamine aggregates. Ube2A and UbcH5b both show strong colocalization to aggregates, whereas E2-25K does not colocalize.

Ube2A is a closely related human ortholog of the worm protein UBC-1. Although we do not know the subcellular localization of UBC-1 in worms, we have shown that UBC-1 affects ubiquitin localization to aggregates. Thus, it is not surprising to find that Ube2A localizes to aggregates. On the other hand, UBC-2, the worm ortholog of UbcH5b, did not affect ubiquitin localization. Therefore, it is curious that it would localize to aggregates. Perhaps it is involved in a minor ubiquitination event on aggregate proteins. Alternatively, UBC-2 might localize to aggregates due to the high concentration of ubiquitin and proteasomes that are present there.

Knockdown of Ubc's caused changes in the size or number of polyglutamine protein aggregates. Our results indicate that there are, at minimum, two different ubiquitination events that affect aggregates. This conclusion is made because there are two groups of phenotypes with opposite characteristics: one set of knockdowns showing larger and less numerous aggregates, and one set showing smaller aggregates. Therefore, these two proposed ubiquitination pathways appear to have opposing affects on aggregation. A model for how the Ubc's might be involved in protein aggregation is presented in Figure 6.

Both ubc-2 and ubc-22 had a similar phenotype: reduction of aggregate number and increase in aggregate size. This indicates that these two ubiquitinating enzymes may be involved in a common process that limits the size of aggregates. It is possible that ubiquitination by ubc-2 and ubc-22 is important for degrading aggregating proteins or for ridding the cell of aggregates that reach a certain size. However, removal of aggregates does not seem a likely mechanism for ubc-2 and ubc-22, since their knockdown results in fewer aggregates than normal. Therefore, ubc-2 and ubc-22 may be involved in limiting the coalescence of smaller aggregates into larger aggregates.

Since they have the same knockdown phenotype and the combined knockdown is not additive (Figure 1D), the two enzymes, UBC-2 and UBC-22, might participate in the same ubiquitination pathway or might function as a dimer. There is evidence from both in vitro and in vivo studies suggesting the existence of Ubc dimers and Ubc-Uev dimers 3435. One interesting aspect of UBC-22 is that it has a non-canonical catalytic site with an awkwardly placed cysteine residue in a context lacking all consensus residues 36. Thus, it is possible that UBC-22 may not be catalytically active as a Ubc enzyme but rather partners with another functioning Ubc.

ubc-1, ubc-13, and uev-1 all showed a reduction in the size of aggregates. This reduction in size came without a reduction in the overall levels or a change in solubility of Q82:GFP protein. RNAi of ubc-1 had the most dramatic effect on size and also caused aggregates to be more numerous. UBC-1 could be involved in the process of smaller aggregates joining together into larger aggregates (Figure 6).

Since RNAi of ubc-1, ubc-13 and uev-1 all cause a reduction in aggregate size and loss of ubiquitin from aggregates, it is tempting to speculate that these enzymes might also function in the same pathway. In fact, it has been shown that the yeast homologs of these three enzymes function together in ubiquitination of PCNA. Rad6, the yeast homolog of ubc-1, is responsible for adding the initial ubiquitin molecule onto PCNA, and the ubiquitin chain is subsequently extended by the Ubc13/Uev1 dimer 37.

Knockdown of ubc-1, ubc-13 or uev-1 could cause aggregates to be smaller due to a decrease in overall bulk of the aggregate. Along with many other proteins, it is known that the proteasomes localize to aggregates and, thereby, could contribute to aggregate volume. Proteasome localization is disrupted after RNAi of ubc-1, ubc-13, or uev-1. Therefore, one possible scenario is that ubiquitination of Q82:GFP or some other protein attracts proteasomes to the aggregate. In the absence of that ubiquitination, proteasomes or other ubiquitin binding structures do not colocalize and thus the overall size of the aggregate is reduced.

We have used the nematode as a model for understanding the aggregation of polyglutamine proteins. The validity of this model was tested by examining the effects of Ubc RNAi in human cells expressing polyglutamine aggregates. Our results indicate that both worm cells and human cells deal with polyglutamine proteins using a conserved mechanism.

Using shRNA plasmids, we achieved knockdown of three human Ubc's that have homologs in our group of worm Ubc's with strong effects on aggregates. The human Ubc's with strong similarity to the worm enzymes showed phenotypes that mimicked the RNAi phenotype in worms. E2-25K, which has only weak identity to worm UBC-22, showed a less pronounced aggregate phenotype.

Worms were maintained at 20°C on nematode growth media with OP50 E. coli according to standard methods 38. The integrated Q82:GFP nematode strain is described elsewhere 15. The OP50 and HT115 bacterial strains as well as N2 worms were obtained from the Caenorhabditis Genetics Center.

The GATEWAY system (Invitrogen, Inc.) and the vector pL4440GTWY 39 were used to generate RNAi feeding plasmids 40. RNAi clones for ubc-1, ubc-2, ubc-7, ubc-8, ubc-13, ubc-14, ubc-16, ubc-20, ubc-22, ubc-24, uev-1, uev-2, ubq-1, and rpt-1 were made using recombinational cloning with pL4440GTWY and entry clones obtained from Mark Vidal and Open Biosystems, Inc. 41.

For aggregate expression in HEK293 cells, the Q81-YFP plasmid was generously provided by Dr James R. Burke (Duke University Medical Center Durham, NC). Expression Arrest™ human shRNA plasmids were from Open Biosystems. Expression Arrest™ contained either the non-silencing sequence which has no homology in mammals (control) or sequences specifically targeting Ube2A (1, RHS1764-9391619; 2, RHS1764-9690190), UbcH5b (1, RHS1764-9392787; 2, RHS1764-9685283), or E2-25K (1, RHS1764-9700185; 2, RHS1764-9100963).

RNAi was performed as previously described 39. Plasmids described above were transformed into HT115 E. coli. 35 mm NGM plates were supplemented with 1 mM isopropyl-β-thiogalactopyranoside (IPTG) and 100 mg/ml ampicillin and then seeded with 125 μl of saturated overnight culture. Q82:GFP nematode populations were synchronized by using a hypochlorite treatment to yield embryos. Embryos were placed on RNAi-seeded plates 2–4 hours after seeding. Worms were removed from plates after 24 or 48 hours for analysis. Control RNAi was performed using the pL4440 empty vector in HT115.

Nematodes were fixed in methanol onto polylysine-coated slides and mounted in Vectashield plus DAPI (4,6-diamidino-2-phenylindole). Aggresomes were counted by observation on a Nikon E600 microscope equipped with a QiCam digital camera (Q Imaging). A minimum of 10 worms were analyzed for each condition. Sizes of individual aggregates were determined using Image Pro software. At least 100 aggregates were sized for each RNAi condition. Soluble GFP was observed by anesthetizing nematodes with 25 mg/mL tetramisole (Sigma) in PBS.

Colocalization of ubiquitin and proteasome to aggregates was determined using mouse-anti-ubiquitin (1:200; P4D1 from Santa Cruz Biotechnology) and rabbit-anti-proteasome (1:200; ab2943 from Abcam). Nematodes were fixed with methanol onto polylysine-coated slides and incubated in primary antibody overnight followed by incubation with secondary antibody for 2 hours. Secondary antibodies were goat-anti-mouse Rhodamine and goat-anti-rabbit Rhodamine (Jackson ImmunoResearch) Worms were observed and analyzed using the microscopy tools mentioned above.

Cell culture reagents were purchased from Mediatech. HEK293 cells were transiently transfected using Arrest-In™ transfection reagent (Open Biosystems) following the manufacturer's protocol. Cell culture was done in DMEM with 10% FBS in an incubator with 5% CO₂.

shRNA plasmids were isolated using a Qiagen kit and verified using restriction digests. Transfections were done in 6 well plates. For each well, 2 μg of shRNA plasmid and 100 ng of Q81:YFP in 50 μl of serum free DMEM was added to 50 μl of diluted Arrest-In™ and left at room temperature for 15 min. The transfection mixture was added drop wise to the wells and incubated at 37°C; 5% CO₂. Media was changed 1 day following transfection. Cells were analyzed 72 hrs after transfection.

Images were taken of cells containing fluorescent aggregates 72 hrs after transfection with shRNA plasmids. Each experiment was repeated three times and each time 150 transfected cells were examined. The number of aggregates per cell was counted, and area of each aggregate was measured using SPOT software (Diagnostics Instruments). Area of aggregates was converted to μm² by calibrating with a micrometer (1 μm was equivalent to 39.6 pixels).

For localization of Ubc proteins in aggregate-containing cells, HEK293 cells were cultured in 24 well plates and transfected with 100 ng of Q81:YFP in each well. Cells were incubated for 48 hrs, then trypsinized and transferred to 4-chambered slides (NUNC). After another 24 hrs, cells were washed once with PBS and fixed with 4% paraformaldehyde (in PBS) at 4°C for 30 min. Cells were checked for fixation under a light microscope. After fixation, cells were washed three times with Tris buffered saline (TBS; 50 mM Tris, 150 mM NaCl pH 7.6) followed by blocking (1% BSA, 0.3% Triton-X-100 in TBS) for 1 hr at room temperature. Primary antibody (rabbit-anti-E2-25K from Boston Biochem 1: 200 or goat-anti-Ube2A polyclonal 1: 100 or goat anti-Ubch5B polyclonal 1:100 both from Santa Cruz Biotechnology) was added in 1% BSA, 0.3% Triton-X-100 and incubated overnight at 4°C. Slides were then washed three times with TBS followed by incubation with secondary antibody (goat-anti-rabbit Rhodamine from Jackson ImmunoResearch or rabbit-anti-goat Rhodamine from Open Biosystems) in 1% BSA, 0.3% Triton-X-100 and incubated for 6 hrs at 4°C in the dark. Cells were washed three times with TBS. Chambers were removed from the slides very carefully until no gum was left on the slide. Prolong Gold Antifade reagent with DAPI (Molecular Probes) was added and was allowed to cure overnight at 4°C in dark. Cells were examined using a Leica DMIRB inverted microscope equipped with a SPOT RT slider color digital camera. Images were obtained with blue, green, red fluorescence filters. Images were processed by C IMAGING software (SIMPLE PCI Inc.).

Worms were grown on RNAi plates for 48 hours, collected by centrifugation, washed 5 times with PBS, and frozen at -80°C. Frozen pellets were ground using a mortar and pestle in liquid N₂. RNA extractions were performed using the QiaShredder and RNeasy miniprep kits (Qiagen). cDNA was synthesized from the RNA samples with the Transcriptor First Strand cDNA kit (Roche) using 10 μl of RNA as template. PCR reactions used 2 μl of cDNA and the following primers: UBC1F CAAATAAACCGCCAACCGTCAA, UBC1R TCTCATATTCCCGTCGATTTTCTT, UBC2F ACTTCCCAACAGACTATCCATTCA, UBC2R AGCGTACTTTTGCGTCCATTCTCT, ACT1F CCGCCGGAATCCACGAGACT, ACT1R GTGGAGAGGGAAGCGAGGATAGAT.