SUG1, a Component of the 26 S Proteasome, Is an ATPase Stimulated by Specific RNAs*

SUG1 is an integral component of the 26 S proteasome. Belonging to a novel putative ATPase family, it shares four conserved motifs characteristic of ATP-dependent DNA/RNA helicases. Recombinant rat SUG1 (rSUG1) produced in Escherichia coli was highly purified and characterized in terms of its biochemical properties. The rSUG1 exhibited a Mg2+-dependent ATPase activity. TheK m for ATP and V max of rSUG1 were 35 μm and 7 pmol of ATP/min/μg of protein, respectively. Both ATPase activity to release [32P]monophosphate and [32P]ATP-labeling activity were coordinately affected by cold ATP severely, GTP and UTP moderately, and CTP little. Interestingly, the rSUG1 ATPase activity was stimulated by poly(U) and poly(C), but not by poly(A), poly(G), or by any forms of DNAs tested. A UV cross-linking assay also indicated poly(U)- and poly(C)-stimulated labeling of rSUG1 with [α-32P]ATP. Moreover, the ATPase activity was facilitated by cellular poly(A)+ RNA, but not by poly(A)− RNA. RNA transcribed in vitro from cDNA encoding a b-Zip protein could stimulate the ATPase activity. This is the first report to demonstrate a specific RNA requirement for ATPase with respect to the proteasomal ATPases. Our present work suggests that SUG1 can specifically interact with protein-coding RNA (mRNA) and play some roles in mRNA metabolism.

The 26 S proteasome is a huge protease complex that degrades short-lived proteins related to metabolic regulation and cell cycle progression (1,2). It is composed of the 20 S catalytic core and an ATPase-containing 22 S regulatory complex (3,4). The human 22 S complex contains at least five highly related putative ATPases, i.e. TBP1, TBP7, S4, MSS1, and p45 (a homolog of yeast SUG1), which are members of a novel ATPase family named the AAA family (ATPases associated with a variety of cellular activities) (5). The family members have a highly conserved ATPase module with 200 amino acids and fulfill a large diversity of functions. It was demonstrated that the five ATPases contain four conserved motifs characteristic of putative ATP-dependent RNA/DNA helicases (6 -8). We previously reported constant distances between each motif among proteasomal ATPases, suggesting the functional importance of these motifs (6).
Of the above ATPases, SUG1 (11) has been shown by various studies to be involved in transcriptional regulation in addition to proteolytic function. The yeast SUG1 was reported to be a component of the RNA polymerase II (pol II) holoenzyme (12,13), and Trip1, a human homolog of the SUG1, interacts with the thyroid hormone receptor (14). Baur et al. (15) also reported that mouse SUG1 interacts with various nuclear receptors. However, it is still controversial as to whether or not SUG1 acts as an intrinsic transcription factor because one cannot eliminate the possibility that those ATPases are involved in degradation of transcription factors mediated by the 26 S proteasome. Actually, some transcription factors are regulated by proteasome-dependent proteolysis (1). Biochemical study is required to resolve issues such as how the ATPases are involved in RNA metabolism such as transcriptional regulation. In this study, we report the purification and characterization of rat SUG1. We found that rSUG1 1 exhibited ATPase activity that was specifically stimulated by particular RNA molecules.

EXPERIMENTAL PROCEDURES
Expression and Purification of Rat SUG1-SUG1 cDNA was cloned from a rat liver cDNA library as described previously (6). Histidinetagged rSUG1 was overexpressed in Escherichia coli by use of the pET vector system (16). Insoluble recombinant rSUG1 was purified by Ni 2ϩagarose under denaturing conditions according to the instructions supplied by Qiagen. The resulting proteins were further subjected to a preparative SDS-PAGE and recovered from the excised gels. The protein was redissolved in a urea-containing buffer (25 mM Tris-HCl (pH 7.5), 0.3 M NaCl, 1 mM 2-mercaptoethanol, 0.1% Nonidet P-40, 10% glycerol, and 8 M urea), and the urea was gradually removed by dialysis. These proteins were analyzed by 10% SDS-PAGE and stained with Coomassie Brilliant Blue.
Assay for ATP Hydrolysis-In the case of Fig. 3A, the ATPase activity was assayed by thin layer chromatography as described (32). Reactions (20 l) contained 0.5 g of purified rSUG1 in buffer A (20 mM Tris-HCl (pH 7.5), 70 mM KCl, 2.5 mM MgCl 2 , 1.5 mM dithiothreitol, 500 M ATP, and 1.25 Ci of [␥-32 P]ATP). ATP hydrolysis reactions were allowed to proceed at 37°C for 30 min. Radioactive phosphate released from [␥-32 P]ATP was separated on a polyethyleneimine plate (Macherey-Nagel) using 1 M formic acid and 0.5 M lithium chloride. The released phosphates were visualized by autoradiography.
In other cases, the ATPase activity was assayed using activated charcoal (Sigma) as described by Armon et al. (17). The purified recombinant SUG1 (0.5 g) was incubated at 37°C for 30 min in buffer A. On the basis of ⑀ max , amounts of RNA homopolymers were precisely determined from their A 260 . Control reactions without rSUG1 were carried out in parallel tubes, and the control value (radioactivity) was subtracted from each experimental one. Each assay was done in triplicate, and the results were presented as a simple arithmetic average.
ATP Cross-linking Assay-ATP cross-linking assays were performed as described by Pause and Sonenberg (18). Reaction mixture (20 l) containing 0.5 g of rSUG1 in a buffer (20 mM Tris-HCl (pH 7.5), 70 mM * This work was supported by grants-in-aid for scientific research on priority areas from The Japanese Ministry of Education, Science, Sports, and Culture, and by a grant from CREST Japan Science and Technology. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
KCl, 5 mM magnesium acetate, 1.5 mM dithiothreitol, 10% glycerol, and 5 Ci of [␣-32 P]ATP) was irradiated by UV cross-linker LS1500 (Funakoshi) from a distance of 2 cm at 4°C for 20 min. The samples were subjected to 10% SDS-PAGE, and then the gels were stained with Coomassie Brilliant Blue and subjected to autoradiography. Radioactivities were measured by a BAS 1500 phosphoimager (Fuji Film).
Preparation of Cellular RNAs-Cellular total RNA was prepared from rat liver by use of cesium trifluoroacetate as described (19). The total RNA was fractionated into poly(A) ϩ and poly(A) Ϫ RNAs by oligo(dT)-Latex (Takara Shuzo) (20).
cDNA Clones and in Vitro Transcription-The cDNAs encoding rat HTF, TIP120, SUG1, and MSS1 were cloned in pBluescript, and in vitro RNA synthesis was conducted with T7 or T3 polymerase according to the instructions supplied by Promega. Quality of the synthesized RNAs was checked by agarose gel electrophoresis. HTF is a rat b-Zip transcription factor closely related to the human X-box-binding protein/Taxresponse element-binding protein 5 and is activated in hepatocellular carcinoma as well as during normal hepatic cell growth in rats (30). TIP120, having partial homology to Drosophila TAF80, is a novel rat TBP-binding protein (31). MSS1 is a rat homolog of the human MSS1 and is a part of the 26 S proteasome (6).

Expression and Purification of Recombinant Rat SUG1-We
cloned rat SUG1 (rSUG1) cDNA and found it to encode a protein of 406 amino acids (6) (Fig. 1) and to have exactly the same structure as human p45 (21), the human homolog of yeast SUG1. The central regions of rSUG1 and the other four proteasomal ATPases share greater than 60% identity and are designated as the ATPase module ( Fig. 1). This domain contains a putative ATP-binding motif, GX 4 GKT, and ATP hydrolysis motif, DEID, which is analogous to the DEXD box proteins including many ATP-dependent RNA/DNA helicases (22)(23)(24)(25). Two additional sequences in rSUG1, SAT and (H/Q)RXGRXXR, are also characteristic motifs for RNA/DNA helicases (Fig. 1). Strikingly, we found the distance between individual motifs to be conserved in the five proteasomal ATPases, i.e. 51 residues between GX 4 GKT and DEID, 40 residues between DEID and SAT, and 10 residues between SAT and (H/Q)RXGRXXR ( Fig.  1) (6). The above findings suggest that these four motifs have some essential roles in the functioning of those ATPases.
To investigate the enzymatic nature of rSUG1, we subcloned its cDNA into the pET3a vector and expressed rSUG1 as a fusion protein with a histidine-tag appended to its N terminus in E. coli. Although rSUG1 protein was able to be overexpressed in E. coli (Fig. 2, lane 2), most of the protein was obtained in an insoluble fraction. We attempted to dissolve the insoluble protein in a urea-containing buffer and to purify it under denaturing conditions. Purification by Ni 2ϩ -agarose resulted in Ͼ40% pure protein as judged by inspection of the Coomassie Brilliant Blue-stained SDS-acrylamide gel (Fig. 2,  lane 3). To further remove contaminating E. coli proteins, we subjected the protein sample to a preparative SDS-PAGE and subsequently excised the rSUG1 and extracted it from the gel. The recovered proteins were dissolved in a urea-containing buffer and then refolded by dialysis to gradually remove urea (see "Experimental Procedures"). As expected, the final preparation was apparently pure as judged by SDS-PAGE (Fig. 2,  lane 4).
rSUG1 Protein Exhibits ATPase Activity-The ATPase activity of the purified rSUG1 was tested by its ability to release radioactive phosphate from [␥-32 P]ATP as described under "Experimental Procedures." A commercial ATPase from dog kidney was used as a positive control (Fig. 3A, lane 1). We found that rSUG1 hydrolyzed ATP to release monophosphate (Fig. 3A, lane 3) but that the enzyme activity could not be detected in a parallel sample prepared from control E. coli (data not shown). It is thus evident that rSUG1 was able to hydrolyze ATP. Release of labeled monophosphates was inhibited by cold ATP and EDTA (Fig. 3A, lanes 4 and 5), suggesting that the rSUG1 is a Mg 2ϩ -dependent ATPase. Kinetic analyses of the rSUG1 determined the K m for ATP to be 35 M and the V max to be 7 pmol of ATP/min/g of protein (Table I).
To examine the nucleotide specificity for the SUG1-catalyzed hydrolysis, we employed unlabeled nucleotides as competitors. Expectedly, we found ATP was the most potent competitor; however, CTP competed little with ATP for hydrolysis. GTP and UTP significantly competed with ATP for hydrolysis by rSUG1 to some extent (Fig. 3B). Thus, rSUG1 is suggested to interact efficiently with all nucleotides except for CTP. This specificity of rSUG1 is similar to that of Yhs4p, a yeast homolog of the human S4 (26).
UV-mediated Labeling of rSUG1 with [␣-32 P]ATP-Using a photolabeling technique, we next studied whether polyribopyrimidines affect the interaction of rSUG1 with ATP. This technique, in which ATP is photolyzed by UV light in the presence of ATP-binding protein, yields a covalent adduct between ATP and proteins. By use of [␣-32 P]ATP, specific radioactive proteins can be visualized by SDS-PAGE and autoradiography. As shown in Fig. 4A, illumination with UV light resulted in the labeling of the rSUG1 protein (lane 1), and the labeling was efficiently inhibited by cold ATP (lane 2). Incubation of [␣-32 P]ATP for 20 min did not cross-link the ATP to the protein in the absence of UV light (Fig. 4A, lane 3).
To further confirm the labeling specificity, we carried out the UV-labeling reaction in the presence of cold nucleotides and then estimated the radioactive protein by a phosphoimager. Addition of CTP to the reaction had little effect on the UV labeling of rSUG1 with [␣-32 P]ATP, whereas GTP and UTP significantly reduced the labeling (Fig. 4B). Taken together, the effects of nucleotides on the UV labeling of rSUG1 were consistent with those on the ATP hydrolysis (Fig. 3B), suggesting that the labeled rSUG1 is attributed to the specific interaction with ATP.
Effects of Polyribopyrimidines on the rSUG1 Functions-Because SUG1 contains a DEID motif, the protein is considered to belong to a subfamily of the DEAD-box proteins. This family carrying a putative DEAD-box helicase motif includes more than 100 proteins with diverse functions, and some of these proteins have an RNA-stimulated ATPase activity (27,28). So we investigated whether or not nucleic acids would affect the ATPase activity of rSUG1. The enzyme activity was not affected by single-stranded or double-stranded DNA, whereas the rSUG1 ATPase activity was stimulated 6-fold by poly(U) (Fig.  5A). When the other RNA homopolymers were tested, we observed significant enhancement of the ATPase activity by polyribopyrimidines: poly(U) and poly(C). It is noteworthy that the homoribopolymer stimulation of the rSUG1 ATPase activity is different from that of Yhs4p, whose activity is enhanced by single stranded-DNA, double-stranded DNA, and RNA (26). Moreover, we analyzed the effect of RNA homopolymers on the UV labeling of rSUG1. Fig. 5B shows that the cross-linking was also highly stimulated by poly(U) and poly(C), but not by poly(G) and poly(A). On the basis of these results, we demonstrated that polyribopyrimidines specifically faclitate the pho-tolabeling of rSUG1 as well as its ATPase activity.
We also determined the kinetic parameters of the rSUG1 ATPase in the presence of poly(U). The K m and V max values showed similar affinity (K m ) for ATP in the hydrolysis reaction with or without poly(U) ( Table I). In contrast, addition of 1 g of poly(U) increased the V max 4-fold (Table I). When poly(C) was used, similar results were obtained (data not shown). The increased ATP cross-linking of rSUG1 by poly(U) and poly(C) correlated with their higher rate of ATP hydrolysis, but not with their ATP binding ability, since the K m values of rSUG1 were similar in the presence or absence of the polyribopyrimidines. We suppose that the cross-linking assay used here detects two molecular species, i.e. ATP-bound rSUG1 and ADPbound rSUG1. The ATP cross-linking of rSUG1 is suggested to reflect the abilities to both bind and hydrolyze ATP.
Stimulation of rSUG1 ATPase Activity by Particular Types of RNA-On the basis of the above results, it was important to examine whether natural RNA can stimulate the ATP hydrolysis reaction of rSUG1. Cellular total RNA, poly(A) ϩ RNAs, and poly(A) Ϫ RNAs were added to the ATPase reaction. As shown in Fig. 6A, total RNA slightly stimulated the ATPase activity (1.5-fold). It was most striking that the ATPase activity was highly enhanced (6-fold) by poly(A) ϩ RNA, whereas TABLE I Kinetic parameters of ATP hydrolysis ATPase activity was measured using activated charcoal for 10 min as described under "Experimental Procedures" with ATP concentrations ranging from 5 to 500 M. Experiments were carried out in duplicate, and kinetic constants were calculated from a Lineweaver-Burk plot. poly(A) Ϫ RNA exhibited apparently no effect (Fig. 6A). This finding suggests that the rSUG1 activity is specifically stimulated by mRNA.
We therefore examined the effect of various kinds of mRNA-type RNAs on the ATP hydrolysis. Four kinds of cDNAs encoding rat SUG1, MSS1 (6), HTF (30), and TIP120 (31) in pBluescript were transcribed by T7 or T3 RNA polymerase in vitro to produce sense and antisense RNAs as described under "Experimental Procedures." Each kind of RNA was then added to the reaction mixture for ATP hydrolysis by rSUG1. In these experiments, sense HTF RNA apparently stimulated ATP hydrolysis in a dose-dependent manner, and TIP120 sense RNA had a weak but significant effect (Fig. 6B). However, sense SUG1 and MSS1 RNAs, and antisense HTF RNA enhanced the ATPase activity little (Fig. 6B). These results suggest that, in a physiological environment, the ATPase activity of rSUG1 can be specifically stimulated by particular species of protein-encoding RNAs. DISCUSSION In this paper, we have provided evidence that rat SUG1 exhibits ATPase activity. Our rSUG1 preparation was apparently pure, and Sf9 cell-expressed soluble rSUG1 protein also showed the same enzyme properties (data not shown) as obtained with the renatured bacterially expressed protein. This agreement suggests that the nature of rSUG1 presented in this study reflects intrinsic enzyme activities. Importantly, biochemical characterization of the recombinant rSUG1 yielded the following significant findings. (i) The rSUG1 ATPase activity is specifically stimulated by particular types of RNAs including poly(C) and poly(U), but not by single-stranded and double-stranded DNAs. (ii) Polyribopyrimidines also enhance the labeling of rSUG1 with [␣-32 P]ATP. (iii) Cellular poly(A) ϩ RNAs and in vitro transcribed specific RNAs enhance ATP hydrolysis.
Although several DEAD-box family proteins show ATPase activity facilitated by RNA, the observation that rSUG1 exhibits a specific RNA requirement is particularly interesting. rSUG1 is the first example of a proteasomal ATPase having such a requirement. Yhs4p, a yeast homolog of human S4, is activated by unspecified RNAs and DNAs commonly (26). It is of interest to investigate the effect of nucleic acids on the ATPase activity of human S4 and to compare the results between the human and yeast ATPases. Other mammalian pro-teasomal ATPases might also exhibit such a nucleic acid requirement, since members of the ATPase family are highly homologous to each other. The DEAD-box proteins such as eIF-4A and p68 exhibit an RNA-unwinding activity as well as RNA-stimulated ATPase activity (27,28). Therefore, we speculate that rSUG1 also could have an RNA-dependent helicase activity; however we unable to detect it in our rSUG1 preparation. A particular RNA substrate or some co-factor(s) might be required for nucleic acid unwinding.
We have reported here that rSUG1 activity is efficiently stimulated by synthetic RNAs and mRNA-type RNAs. How does HTF cDNA-derived RNA specifically activate the ATPase activity of rSUG1? We found that HTF cDNA contained clustered TC-rich stretches, whereas other cDNAs tested here did not (data not shown). Thus it is reasonable to assume that these stretches are important in such activation (30), since polyribopyrimidines highly stimulated ATP hydrolysis by rSUG1 (Fig. 3B). Additionally, it may be possible that the RNA requirement for rSUG1 is determined by unknown secondary/ tertiary structure in mRNA-type RNAs. We suspect that more suitable substrates for SUG1 than RNA encoding HTF may be present among cellular mRNAs.
The observation that some mRNAs are specifically required to stimulate ATPase activity by rSUG1 raises an interesting question. SUG1 is one of the regulatory subunits of the 26 S proteasome (21,29). Moreover, the protein has been reported to be an integral component of the RNA polymerase II complex and to bind to TATA-binding protein, as well as to several nuclear hormone receptors (12,14,15). However, it is still unclear that SUG1 functions as a factor responsible for transcription. Our data that ATPase activity of rSUG1 is facilitated by particular mRNAs suggest that rSUG1 interacts with mRNA and can play a role in mRNA metabolism in addition to one in proteolysis. This assumption may be supported by the fact that the ATPase activity of the rat 26 S proteasome (9) was not stimulated by poly(C) or poly(U) (data not shown). Most recently, McCracken et al. (10) demonstrated that the C-terminal domain of the RNA polymerase II large subunit associates with 3Ј-processing factors and splicing factors. These findings suggest that transcription, splicing, and processing (cleavage, polyadenylation, etc.) of mRNA are intimately coupled in vivo. Other investigations employing staining with antibodies against nuclear factors also support our idea (33). The RNA pol II holoenzyme is likely to contain all the general transcription factors and mRNA-processing factors. If SUG1 is, in fact, associated with RNA pol II holoenzyme, our findings might have profound implications for the role of SUG1 in mRNA processing, i.e. transcription or post-transcriptional events. However, we cannot exclude an alternative possibility that SUG1 in the 26 S proteasome is responsible for the degradation of some mRNA-binding proteins. Antizyme, a noncompetitive inhibitory protein, is responsible for degradation of ornithine decarboxylase by the 26 S proteasome (34,35). The binding of antizyme induces a conformational change in ODC that promotes recognition of ornithine decarboxylase by the 26 S proteasome to stimulate its proteolysis (36,37). Similarly, it is possible that some mRNA-binding proteins that undergo conformational changes induced by interaction with mRNAs may be targets of SUG1 for proteasome-mediated proteolysis.