The Serine Protease Domain of Hepatitis C Viral NS3 Activates RNA Helicase Activity by Promoting the Binding of RNA Substrate*

Nonstructural (NS) protein 3 is a DEXH/D-box motor protein that is an essential component of the hepatitis C viral (HCV) replicative complex. The full-length NS3 protein contains two functional modules, both of which are essential in the life cycle of HCV: a serine protease domain at the N terminus and an ATPase/helicase domain (NS3hel) at the C terminus. Truncated NS3hel constructs have been studied extensively; the ATPase, nucleic acid binding, and helicase activities have been examined and NS3hel has been used as a target in the development of antivirals. However, a comprehensive comparison of NS3 and NS3hel activities has not been performed, so it remains unclear whether the protease domain plays a vital role in NS3 helicase function. Given that many DEXH/D-box proteins are activated upon interaction with cofactor proteins, it is important to establish if the protease domain acts as the cofactor for stimulating NS3 helicase function. Here we show that the protease domain greatly enhances both the direct and functional binding of RNA to NS3. Whereas electrostatics plays an important role in this process, there is a specific allosteric contribution from the interaction interface between NS3hel and the protease domain. Most importantly, we establish that the protease domain is required for RNA unwinding by NS3. Our results suggest that, in addition to its role in cleavage of host and viral proteins, the NS3 protease domain is essential for the process of viral RNA replication and, given its electrostatic contribution to RNA binding, it may also assist in packaging of the viral RNA.

Nonstructural protein (NS3) 3 is an essential component of the hepatitis C virus (HCV) replication complex. It is a nucleic acid-stimulated ATPase and DEXH/D-box protein that is structurally similar to other DNA and RNA helicases from the phylogenetically conserved superfamily 2. Like these relatives, NS3 contains two RecA fold domains (domains 1 and 2) that comprise the ATPase site and the platform for RNA binding, which is supplemented by domain 3, which sits atop D1 and D2 (1). The three-lobed triangle comprised of D1-D3 is considered the "helicase" of NS3 (NS3hel), and it has been studied extensively in isolation (2)(3)(4)(5). However, the full-length NS3 protein differs from other SF2 helicases in that it possesses a fourth domain at the N terminus that displays robust serine protease activity (the protease domain) (6,7). This protease domain is located at the vertices of D1-D3, in an ideal location to influence the catalytic activities of the helicase region.
NS3 serine protease activity is crucial for cleaving the viral polyprotein into individual, nonstructural (NS) proteins. It also facilitates viral pathogenicity by cleaving host proteins and down-regulating the innate immune response of the cell (7,8). NS3 helicase activity may be required to assist the viral polymerase during replication of the HCV RNA genome. NS3 may unwind complex RNA structures within the genome, whereas the NS5B viral polymerase copies the RNA (9). There may also be a role for NS3 in viral capsid assembly and the packaging of the RNA genome within (10). However, a specific function for the helicase has not yet been clearly demonstrated.
NS3hel constructs have been extensively studied, and analyses of ATPase activity, nucleic acid binding, and duplex unwinding have been reported (2-4, 11, 12). Moreover, NS3hel has been used to test potential anti-HCV compounds (13). However, much of the previous work comparing NS3hel with the full-length NS3 protein has been performed on DNA (2,4). In another study, RNA unwinding, but not RNA binding was examined (2). A diverse array of NS3 constructs has been utilized, and in many cases the proteins were appended to positively charged histidine tags (2)(3)(4)(5)14). To develop a clear understanding of the protease domain and its influence on activities of the NS3 helicase, we set out to do a comparative analysis of full-length NS3 (which preserves the protease domain) and a diversity of truncated NS3hel constructs.
Understanding the role of the protease domain is not only important for gaining mechanistic insights into the viral life cycle. It is also important for the field of helicase enzymology, as cofactors frequently enhance the re-modeling activities of DEXH/D proteins. For example, RhlB ATPase activity is greatly stimulated by the presence of ribonuclease E (15) and Dbp5 ATPase activity and RNA binding are enhanced by Gle1 (16). Moreover, Dbp8 ATPase activity is enhanced by Esf2 (17) and Prp43 RNA unwinding activity is enhanced by Ntr1 (18). Given these findings, we reasoned that RNA binding and unwinding by NS3hel may be enhanced by the presence of its appended protease cofactor domain.
Here we show that the serine protease domain greatly enhances both the direct and functional binding of RNA to the NS3 protein. Whereas electrostatics play an important role in these processes, there is a specific allosteric contribution from the interaction interface between NS3hel and the protease domain. Most importantly, we show that the serine protease domain is required for NS3 unwinding of RNA and that NS3hel constructs are not functionally representative of NS3 behavior.

EXPERIMENTAL PROCEDURES
Materials-RNA oligonucleotides described in this work were obtained from Dharmacon and DNA oligonucleotides were obtained from Invitrogen. All other reagents were obtained from Fisher unless otherwise indicated.
NS3(1a)ϩ was cloned into pET-SUMO using ExTaq PCR (Takara) followed by ligation with linear pET-SUMO according to the manufacturer's protocol (Invitrogen). The oligonucleotides used were the same as those described above for the pET15b cloning, except that the 5Ј PCR oligonucleotide was "SUMO start": 5Ј-TCAGCGCCCATCACGGCGTACG-3Ј.
Protein Purification-All proteins were purified according to the nickel column/gel-filtration procedure described in Beran et al. (19) except that in all cases yield was increased by adding Triton X-100 to the purification buffers. 2% Triton X-100 (final concentration (v/v)) was added to purification buffer A and 0.2% Triton X-100 (final concentration (v/v)) was added to buffers B, C, and D. The His 6 -SUMO fusion proteins were treated with SUMO protease to create untagged protein as previously described (19).
The filter binding assays were performed essentially according to Wong and Lohman (21). In our case, we incubated various concentrations of protein in a volume of 100 l for 1 h with 0.2 nM labeled RNA oligonucleotide at 37°C in NS3 helicase assay buffer (25 mM MOPS-NH 4 ϩ (pH 6.5), 3 mM MgCl 2 , 1% glycerol, 2 mM dithiothreitol, 30 mM NaCl, 0.2% Triton X-100 (v/v)) before loading onto a membrane sandwich. The membrane sandwich was composed of a top layer of nitrocellulose membrane (0.45 m) (Pierce) and a bottom layer of Nytran N nylon membrane (0.2 micron) (Whatman). We loaded 50 l of each incubation mixture per lane on a dot-blot apparatus while a vacuum was being applied. We subsequently washed the samples in each lane using 150 l of ice-cold NS3 helicase assay buffer while a vacuum was being applied. The blots were exposed to a phosphorimager plate for 1 h before being scanned by a PhosphorImager (GE Healthcare). The percentage of the total RNA bound to protein was subsequently calculated. The RNA binding data were fit to the Hill equation ATPase Assays-ATPase assays were performed at 37°C using 20-l reactions composed of 30 nM protein suspended in NS3 helicase assay buffer (25 mM MOPS-NH 4 ϩ (pH 6.5), 3 mM MgCl 2 , 1% glycerol, 2 mM dithiothreitol, 30 mM NaCl, 0.2% Triton X-100 (v/v)) containing varying amounts of TS34 RNA oligonucleotide, and ϳ1 mM ATP final concentration. The ATP used in these reactions was prepared by mixing 4 l of 100 mM ATP (Promega) with 35 l of water and 1 l of [␥-32 P]ATP (150 Ci/l, 6000 Ci/mmol) (PerkinElmer). 2 l of this ATP mixture was added to each incubation mixture to fill them to 20 l total. All tubes with protein and RNA were preincubated at 37°C for 1 h before the addition of the ATP mixture and the subsequent reactions were also incubated at this same temperature. 1 l of each reaction was spotted after the appropriate incubation times along an axis 1 inch from the bottom of a thin-layer chromatography (TLC) plate (Analtech). After the spots dried, the TLC plates were placed in a glass chamber with a lid in which the bottom 1 cm of each plate was submerged in TLC solution (1 M formic acid, 0.75 M LiCl). After about 30 min of incubation at room temperature, the TLC buffer migrated ϳ10 cm up each plate. The plates were removed from the chamber at this point and allowed to air dry. The plates were subsequently exposed to a PhosphorImager plate (GE Healthcare) for 5 min before scanning with a PhosphorImager (GE Healthcare). The initial reaction velocities versus RNA concentration were plotted and fit to the Michaelis-Menten equation

RESULTS
The Serine Protease Domain Enhances Direct RNA Binding by the NS3 Helicase Domain-As a first step toward evaluating the effects of the serine protease domain on NS3 helicase activity, we examined the RNA binding affinity of full-length NS3 and compared it with helicase domain constructs that lacked the protease domain. It was necessary to create two different protease deletion constructs of the helicase domain: NS3hel(⌬N166) and NS3hel(⌬N188), which are missing 166 and 188 amino acids, respectively, from the N terminus of the NS3 protein ( Fig. 1). NS3hel(⌬N166) was expressed and purified because it is the most commonly studied form of NS3hel (2,4,23), thereby facilitating comparisons with previous work. However, NS3hel(⌬N166) is not a "clean" deletion of the protease domain as it retains extra N-terminal amino acids that may not fold appropriately in the truncated context (6). To generate the helicase domain in isolation, we created NS3hel(⌬N188) (Fig. 1), which lacks additional N-terminal residues. All proteins used in this study were expressed and purified in the pET-SUMO system (Invitrogen) (19) (Fig. 1), which efficiently removes all non-native N-terminal amino acids and His tags after purification. Untagged proteins were essential for this study because His tags add positive charges to the NS3 N terminus, which may influence affinity for RNA.
Using filter binding assays (21) under buffer salt conditions typical for monitoring RNA-protein interactions (see "Experimental Procedures" and Refs. 14, 19, 24, and 25), and in the absence of ATP, we measured direct binding of the NS3 variants to a 5Ј-end labeled, single-stranded RNA that was 34 nucleotides in length (TS34). Whereas both NS3 and NS3hel(⌬N188) bind TS34 RNA, RNA binding by NS3hel(⌬N166) was not detectable in this assay ( Fig. 2 and Table 1). Filter binding data for NS3 and NS3hel(⌬N188) were fit to the Hill equation to determine binding constants and to estimate the degree of binding cooperativity ("Experimental Procedures" and Fig. 2). The binding data for NS3hel(⌬N188) displays an elongated sigmoidal curve relative to the hyperbolic behavior of full-length NS3. The best-fit Hill coefficients for NS3hel(⌬N188) and NS3 are 2 and 1, respectively, indicating that NS3hel(⌬N188) binds RNA in a highly cooperative manner, whereas NS3 does not.  (6). The three versions of NS3 purified for this work (NS3, NS3hel(⌬N166), and NS3hel(⌬N188)) were subjected to electrophoresis for 1 h at 200 V on a NuPAGE 4 -12% BisTris gradient gel (Invitrogen) and subsequently stained with Coomassie Blue (C). The predicted molecular masses were: NS3 ϭ 67 kDa, NS3hel(⌬N166) ϭ 50 kDa, and NS3hel(⌬N188) ϭ 48 kDa. The identity of these purified proteins was confirmed through anti-NS3 Western blotting (not shown).
Full-length NS3 binds TS34 RNA with a much greater affinity than NS3hel, having a dissociation constant of 9 Ϯ 3 nM relative to 43 Ϯ 17 nM for NS3hel(⌬N188) ( Table 1). Full-length NS3 with extra positive charge in the form of an N-terminal His tag (His-NS3) binds single-stranded RNA even more tightly than untagged NS3 as expected (K d ϭ 1.0 Ϯ 0.4 nM) ( Table 1). The NS3hel constructs are likely to display such poor affinity for RNA because the helicase domain carries a net negative charge. Indeed, NS3hel(⌬N166) possesses slightly more negative charge (calculated pI ϭ 5.7) than NS3hel(⌬N188) (calculated pI ϭ 6.1) under our binding conditions (pH 6.5) (26).
NS3 binds a much greater fraction of the available RNA than NS3hel. NS3 binds ϳ100% of the available RNA at saturating protein concentrations, whereas NS3hel(⌬N188) binds less than half of the RNA substrate at saturation (Fig. 2, Table 1). Similar results were also observed for the binding of these proteins to two other single-stranded RNAs (data not shown, see "Experimental Procedures" for the RNA oligonucleotide sequences used). Given that most "single-stranded RNAs" collapse into various weak secondary structures, the results indicate that full-length NS3 (even in the absence of ATP) may effectively melt any structural features and bind the RNA quantitatively. NS3hel may not have this capability, resulting in a substantial fraction of unbound RNA.
The Serine Protease Domain Enhances Functional RNA Binding by the NS3 Helicase Domain-Although direct binding studies are useful, they cannot differentiate between specific and nonspecific associations and they may not reflect the productively bound form of a helicase. Functional RNA binding assays are therefore important metrics for monitoring helicasenucleic acid interactions (14). For most DEXH/D proteins, such as NS3, ATPase activity is stimulated by the binding of RNA (16,(27)(28)(29). We therefore used RNA-stimulated ATPase assays as one metric of productive RNA association with the NS3 and NS3hel constructs. When monitoring the initial rate of ATPase activity as a function of RNA concentration, we observed that the K m for RNA-stimulated ATPase activity by NS3 (K m ϭ 0.100 Ϯ 0.001 M) is 20-fold lower than that of NS3hel(⌬N188) (K m ϭ 2.0 Ϯ 1.1 M) and 151-fold lower, respectively, than that of NS3hel(⌬N166) (K m ϭ 15.1 Ϯ 3.6 M) (Fig. 3A and Table 2). The k cat values (nmol of ATP/min/pmol NS3 variant) did not vary greatly between these three NS3 constructs (1.4 Ϯ 0.2, 2.7 Ϯ 0.3, and 0.80 Ϯ 0.03, respectively) ( Table 2). Thus, the serine protease domain radically increased functional binding of RNA by the NS3 helicase, allowing it to become a more efficient RNA-stimulated ATPase (in terms of k cat /K m ) ( Table 2). Furthermore, the addition of positive charge in the form of a His tag (His-NS3) further increased functional RNA binding affinity relative to untagged NS3 (K m for His-NS3 ϭ 0.003 Ϯ 0.001 M) ( Fig. 3b and Table 2).
It is interesting that the serine protease domain does not increase the V max for ATPase activity of NS3. Indeed, the NS3hel constructs exhibited a 2-3-fold greater V max than NS3 (Fig. 3 and Table 2) (NS3hel(⌬N166) V max ϭ 1.6 Ϯ 0.2 nmol of ATP hydrolyzed/min; NS3hel(⌬N188) V max ϭ 0.84 Ϯ 0.14; NS3 V max ϭ 0.48 Ϯ 0.05). Thus, provided that RNA is fully bound, the protease domain does not influence the function of the ATPase active site. To examine this possibility in more detail, we compared the unstimulated, basal ATPase activity of the NS3 constructs in the absence of RNA. Under these conditions, ATPase  To address this question, we performed comparative unwinding experiments on two duplex substrates that differ in length and stability: RNA 12/30, which is a 12-base pair RNA duplex with an 18-nucleotide single-stranded 3Ј-overhang (14); and RNA 34/54, which is a 34-base pair RNA duplex with a 20-nucleotide 3Ј-overhang (19) (Fig. 4). Note that we refer to the shorter RNA strand as the top strand and it is this strand that is 5Ј 32 P-end labeled (19). Unwinding experiments were conducted with the NS3, NS3hel(⌬N188), and NS3hel(⌬N166) constructs under singlecycle conditions in which excess, unlabeled top strand RNA is added together with ATP as the "trap" (19). Under these conditions, any helicase molecules that fall off during the course of unwinding are "trapped," so they cannot rebind or reinitiate unwinding (14,30,31). The use of a top strand trap maximizes unwinding amplitudes by preventing reannealing of the labeled top strand RNA (19). With full-length NS3 we have reported highly processive RNA unwinding under these conditions (19), even at relatively low concentrations of protein (10 nM) (data not shown).
Neither of the NS3hel constructs showed any detectable activity for unwinding of the long, 34-base pair RNA duplex (data not shown), whereas it was unwound efficiently and quantitatively by NS3 (19). To determine whether the defect in NS3hel unwinding was due to reduced processivity or defects in translocation (which are evident only on long substrates), and to determine whether the NS3hel retains the ability to catalyze single rounds of unwinding, we examined unwinding of a short duplex: substrate RNA 12/30. However, only trace amounts of unwinding were observed for the NS3hel constructs (Fig. 4). This unwinding experiment was repeated under single-cycle conditions in which a generic DNA oligonucleotide was added in excess to the reactions (14) and under multiple cycle conditions in which only ATP was added to the reactions. However, even when examined under a variety of highly permissive conditions, only full-length NS3 displayed robust activity for unwinding of the short 12/30 substrate (data not shown). Based on these results, we conclude that the serine protease domain is required for NS3hel to unwind even a short RNA duplex.
It has been reported that a His-tagged NS3 helicase domain (His-NS3hel(⌬N166)) can unwind duplex RNA and DNA (2,5). Thus, we purified His-NS3hel(⌬N188) and His-NS3hel(⌬N166), each with 6 histidines linked to the N terminus (see "Experimental Procedures"), and we evaluated the unwinding activity of these constructs. These proteins unwind the short substrate 12/30 to a final amplitude of about 30%, with a rate constant that is unusually slow (k obs ϭ 0.003 s Ϫ1 ), in a buffer that contains 30 mM NaCl (Fig. 4). The addition of the positively charged His tag may therefore enhance RNA binding and rescue some degree of RNA unwinding by NS3, as reported previously (2)(3)(4)11). However, the His tag is insufficient for restoring efficient helicase activity, which underscores the important and specific role played by the protease domain in the function of the NS3 helicase.
We identified another explanation for previous reports of unwinding activity by NS3hel constructs. Previous studies were conducted under conditions of exceptionally low ionic strength. Under the reported conditions in which no salt is added to the preincubation or reaction mixtures (2,5), we also observe that His-NS3hel(⌬N166) unwinds an 18-base pair RNA duplex and an 18-base pair DNA duplex as previously described (2,5). In the absence of salt, we observe that His-NS3hel(⌬N166) unwinds an 18-base pair RNA duplex to 60 Ϯ 3% with a rate constant of 0.010 Ϯ 0.001 s Ϫ1 and it unwinds an 18-base pair DNA duplex to 94 Ϯ 5% with a rate constant of 0.010 Ϯ 0.002 s Ϫ1 . This activity is similar to that of full-length NS3 for unwinding of these same duplexes in the absence of salt   (for NS3 unwinding of the RNA duplex, 86 Ϯ 3%, k obs ϭ 0.020 Ϯ 0.002 s Ϫ1 ; for NS3 unwinding of the DNA duplex, 98 Ϯ 5%, k obs ϭ 0.010 Ϯ 0.002 s Ϫ1 ). The relevance of activity in the absence of salt is unclear, however, because physiological conditions involve relatively high ionic strength (ϳ1 mM Mg 2ϩ , 150 mM K ϩ ) (22). Furthermore, it is well established that helicase activity becomes highly permissive in the absence of salt due to an increase in nonspecific, electrostatic interactions between the helicase and nucleic acid (32). Finally, the stability of RNA duplexes is strongly and linearly dependent on ionic strength, such that duplex integrity and normal structural features are both contingent on the presence of counterions. For these reasons, most mechanistic studies of helicases involve ionic strengths similar to those employed here (24,25,32,33). Fulllength NS3 efficiently unwinds RNA at concentrations of buffer salts that approach physiological (Ͼ75 mM) (data not shown), which suggests that this construct is fully functional. It is notable that the activity of full-length NS3 can be further enhanced by the addition of N-terminal His tags, which enhance both the direct and functional binding of His-NS3 by an order of magnitude (Figs. 2 and 3B). However, the addition of charged tags does not affect unwinding rate constants or the final amplitudes for unwinding relative to untagged NS3 (Fig.  4). Moreover, His tags were not observed to influence NS3 unwinding of RNA substrates containing nicks or non-nucleic acid linkers (19). Therefore, on the full-length protein, an N-terminal His tag does not strongly influence mechanistic features of the NS3 helicase.

DISCUSSION
Here we show that the serine protease domain confers strong, specific RNA binding affinity to NS3, thereby enabling the helicase domain (NS3hel) to recognize and unwind RNA substrates. Our results extend the functional importance of the protease domain and imply a role for this motif in diverse aspects of the viral life cycle. Although the protease domain has previously been implicated only in polyprotein processing and the degradation of host defense proteins, its strong influence on NS3 unwinding suggests that it is essential for HCV RNA replication. Furthermore, given its importance for RNA binding by NS3, the protease domain may also assist in viral packaging, which has recently been shown to require NS3 (10) and is likely to involve protein-RNA interactions.
The Separate Enzymatic Functions of NS3 Are Codependent-Although NS3 possesses two separable protein domains, each with a different enzymatic activity, the two regions of the protein are functionally coupled. Given that the protease domain is required for RNA unwinding, it will be interesting to determine whether protease activity is stimulated by the helicase. If both enzymatic activities are functionally codependent, it will suggest that some phase of the viral life cycle requires coordination between proteolysis and RNA unwinding by the NS3 protein. Based on what is presently known about the HCV replication mechanism, it is unclear why this would occur. However, it is notable that both helicase and protease activities are required very early in the process of replication and that established replicons are inhibited by serine protease inhibitors (34 -36).

The Mechanism of NS3hel Activation by the Protease Domain-
The protease stimulates NS3 helicase activity through a variety of mechanisms, but one of the most important involves simple electrostatics. The protease domain is the most positively charged region of the NS3 protein (Fig. 5) and it provides an electrostatic potential that exceeds even that of the positively charged ATPase site on the protein. It is remarkable that the major electrostatic determinants for RNA affinity have been shunted from the helicase domain (which contains the actual ATP and nucleic acid binding sites) to an appended domain. This arrangement may enable NS3 helicase activity and RNA affinity to be turned on and off, with the protease fold behaving as an electrostatic switch. Given its role in stimulating RNA binding by NS3hel, one might assume that the protease domain can bind RNA by itself. However, the purified, untagged serine protease domain was not observed to bind RNA independently (data not shown). Therefore, the protease domain appears to act allosterically, stimulating substrate binding and unwinding by the helicase domain.
Despite the important role of electrostatics in the mechanism of stimulation, the simple addition of charged tags to NS3hel is insufficient for restoring full activity. This suggests that the protease domain plays a specific structural role in NS3 function. The interface between NS3hel and the protease domain may help tighten and optimize the RNA binding site within NS3hel, enabling NS3 to form a strong grip on substrate. Indeed, it has been previously noted that the protease domain is highly electropositive and it has been hypothesized to comprise part of a secondary RNA binding cleft (37). However, the exact conformation of the protease domain relative to the helicase domain is far from clear at this time and the existence of a rigid cleft has not been established. Whereas our data clearly show that the protease domain stimulates RNA binding in some way, the mechanism by which this occurs is not yet known. The helicase-protease interface is likely to be more extensive and functionally important than previously thought or implied by the existing crystal structure of the full-length protein (6). Recent genetic studies on viral replicons show that mutations in domain 2 of NS3 suppress deleterious mutations in the C terminus of NS4A (which is integrated within the protease domain) (38). Thus, the protease domain may form a network of functional interactions with the helicase domain.
A mutual allosteric activation by protease and helicase domains of NS3 is not only interesting from an enzymological point of view; it may have important practical applications. NS3 is an important and relatively undeveloped target for the development of anti-HCV therapeutics. Small molecules that disrupt the interaction interface between the helicase and protease domains may provide a powerful new therapeutic strategy that is likely to be specific for the viral helicase. Host cells contain many SF2 helicases that are essential for both DNA and RNA metabolism, and whereas they share conserved ATPase residues with the NS3 helicase domain, none of them is appended to a motif that resembles the protease domain. Drugs that target the unique protease-helicase interface would be highly specific for the viral replication complex.
Implications for the Helicase Mechanism-Our findings have important implications for the physical mechanism of RNA unwinding by the NS3 helicase. Two models for NS3 unwinding that have recently been proposed: a Brownian motor model (3) and a spring-loaded unwinding model (39,40). The first model is based on the observation that the NS3 helicase domain binds DNA tightly in the absence of ATP, but not in the presence of ATP. Thus, ATP hydrolysis acts as a switch between high and low affinity states, enabling NS3 to grab and release the tracking strand and thereby move forward. In the springloaded mechanism, ATP hydrolysis causes the RecA folds of NS3 to move like a pincer, enabling NS3 to track along the backbone one nucleotide at a time, ultimately ripping the duplex apart in 3-base pair increments. Unlike the Brownian motor mechanism, the spring mechanism requires that NS3 maintain contact with the lattice in both the ATP-bound and ATP-free states. Here we show that these seemingly inconsistent models can be fully reconciled. We observe that full-length NS3 binds RNA with high affinity in both the ATP bound and ATP free states (as required for the spring-loaded model). However, the affinity of NS3 for the lattice decreases slightly in the ATP-bound state, enabling a shifting in binding register, as predicted by the Brownian motor model. Here we show that for NS3 binding to RNA, K m ϭ 100 Ϯ 1 nM in the presence of ATP and K d ϭ 9 Ϯ 3 nM in the absence of ATP. Furthermore, for RNA binding by NS3hel(⌬N188), K m ϭ 2000 Ϯ 1100 nM in the presence of ATP and K d ϭ 43 Ϯ 17 nM in the absence of ATP. These results indicate that the protease domain provides the appropriate level of ATP-dependent affinity for alternating between strong and loosely bound states of the RNA polymer, thereby facilitating stepwise motion down the backbone without complete loss of affinity for the lattice.
Activation Cofactors and the Function of DEXH/D Proteins-Many DEXH/D proteins require the binding of protein cofactors to stimulate ATPase activity, RNA binding, helicase or translocation activities. The major difference between these proteins and NS3hel (with which they are structurally similar) is that their "activation cofactors" are provided in trans, rather than through covalent attachment. For example, the ATPase and RNA binding activities of RhlB, Dbp5, and Dbp8 are all activated by cofactor proteins (15)(16)(17). However, the electrostatic contribution differs in these cases from the strict delegation of charges observed for NS3hel and protease domains. For example, Dbp5 is predicted to lack a net charge at neutral pH and Dbp8 and RhlB are predicted to be positively charged at neutral pH (26). Only the DEXH/D protein DbpA has been shown to have the RNA unwinding activity of its catalytic domain activated by a domain in cis. The DbpA unwinding domain is activated by the binding of its positively charged C-terminal domain to the 23 S rRNA loop 92 (41). In this manner, the C-terminal domain specifies the unwinding substrate for DbpA. Thus, unlike the serine protease domain-activated NS3hel that has no observed sequence specificity for RNA unwinding (19), DbpA must be activated by an appended domain to unwind the 23 S rRNA (41).
Additional layers of regulatory complexity are added by the binding of other proteins to NS3, which functions in the context of a large complex that includes the NS4A and NS5B proteins (42,43). NS3 binds directly to NS4A, which is required for complete folding of the protease domain and for protease activity (7). Based on its role in stabilizing the protease fold, it will be interesting to determine whether the NS4A cofactor further enhances NS3 RNA binding and ATPase activity. The NS3-4A complex has been studied in our laboratory; however, previous work compared purified NS3 and NS3-4A with different N termini (14). In light of our current findings, it will be interesting to examine the behavior of NS3 in complex with 4A, 5B, and perhaps other proteins, with particular emphasis on the utilization of untagged constructs.