|Home | About | Journals | Submit | Contact Us | Français|
Viral replication initiator proteins are multifunctional proteins that utilize ATP binding and hydrolysis by their AAA+ modules for multiple functions in the replication of their viral genomes. These proteins are therefore of particular interest for understanding how AAA+ proteins carry out multiple ATP driven functions. We have performed a comprehensive mutational analysis of the residues involved in ATP binding and hydrolysis in the papillomavirus E1 initiator protein based on the recent structural data. Ten of the eleven residues that were targeted were defective for ATP hydrolysis, and seven of these were also defective for ATP binding. The three mutants that could still bind nucleotide represent the Walker B motif (D478 and D479) and Sensor 1 (N523), three residues that are in close proximity to each other and generally are considered to be involved in ATP hydrolysis. Surprisingly, however, two of these mutants, D478A and N523A, mimicked the nucleotide bound state and were capable of binding DNA in the absence of nucleotide. However, these mutants could not form the E1 double trimer in the absence of nucleotide, demonstrating that there are two qualitatively different consequences of ATP binding by E1, one that can be mimicked by D478A and N523A and one which cannot.
Viral initiator proteins from DNA viruses belong to the superfamily 3 (SF3) helicases (5, 9). Well-studied members of this group include the T-antigens from the polyomaviruses, the E1 proteins from the papillomaviruses, and the Rep proteins from the adeno-associated viruses. These proteins are multifunctional proteins that utilize ATP binding and hydrolysis by their AAA+ (ATPases associated with various cellular activities) modules for multiple functions in the replication of their viral genomes. AAA+ modules are ~250-amino-acid ATP binding domains that carry out numerous ATP driven functions (for reviews, see references 6 and 7). For example, the E1 protein, which plays an essential role in papillomavirus DNA replication, has multiple functions that are affected by binding or hydrolysis of ATP (14, 18, 21, 23, 24, 26). E1 is a DNA-binding protein, which binds specifically to E1 binding sites (E1 BS) in the origin of DNA replication (2, 8, 15, 22, 25). DNA-binding activity requires nucleotide binding by E1 (15). In the presence of ATP or ADP, E1 can form a specific double-trimer (DT) complex on the ori and, through ATP hydrolysis, this complex can melt the ori DNA (13, 15, 16). In a process that requires ATP hydrolysis, the DT is then converted into a double hexamer (DH), which has ATP-dependent DNA helicase activity and is the replicative DNA helicase (15, 26). Consequently, the E1 AAA+ module is utilized for ATP binding and hydrolysis in at least two different E1 complexes with different functions. An interesting question is how the same motif for ATP binding and hydrolysis is used in these different complexes to achieve their differing functions.
Structural studies of representatives from all three groups—E1 proteins, T antigens, and Rep proteins—have provided important information about how ATP is bound and hydrolyzed by these proteins and the structural consequences that result (1, 4, 7, 10-12). For example, in the recent crystal structure of a hexamer of the E1 oligomerization and helicase domains formed on single-stranded DNA, an ATP binding pocket is formed by 11 residues from two adjacent monomers of the E1 helicase domain (Fig. (Fig.1A)1A) (3). Because most of the residues thought to be involved in ATP binding and hydrolysis in these AAA+ proteins are highly conserved and form particular substructures, the specific function of the individual residues have been predicted for these proteins (6) (see Fig. Fig.1A).1A). It is well established that the conserved residues in the Walker A and Walker B motifs are involved in both binding and hydrolysis of ATP. The Sensor 1 residues are generally involved in contacting Walker B and the γ-phosphate of ATP. The Sensor 2 motif also participates in nucleotide binding and interacts directly with the γ-phosphate of ATP.
To gain a more precise understanding of the role of these particular residues in ATP binding and hydrolysis and because a systematic analysis of such residues has not been performed for the E1 initiator proteins, we performed a mutational analysis of these 11 residues. Based on the behavior of mutants in these residues, the residues can be classified into three groups. Seven of the mutations result in a protein that fails to bind nucleotide and consequently also fail to hydrolyze nucleotide. Three mutants can still bind nucleotide but fail to hydrolyze ATP. Surprisingly, two of these mutants mimic the ATP-bound state and can bind DNA in the absence of nucleotide, demonstrating that E1 utilizes ATP binding for two different modes, only one of which can be mimicked by the mutations in the ATP binding pocket.
Expression and purification of E1 and E2 proteins were carried out as described previously (19).
4% acrylamide gels (39:1, acrylamide-bis) containing 0.5× Tris-borate-EDTA were used for all electrophoretic mobility shift assay (EMSA) experiments. E1 was added to the probe (5,000 cpm) in a solution containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 0.7 mg of bovine serum albumin (BSA)/ml, 0.1% NP-40, 5% glycerol, 5 mM dithiothreitol (DTT), 5 mM MgCl2, and 2 mM ATP or ADP. After incubation at room temperature for 60 min, the samples were loaded and run for 2 h at 9 V/cm. For the E12E22-ori complex formation, EMSA was performed using the conditions described above except that no nucleotide was present. The 84-bp probe, which contains the E1 BS and the adjacent low-affinity E2 BS12, was used. The E12E22-ori complex formation requires both the E1 and the E2 binding sites and is independent of nucleotide.
The E1 oligomerization and helicase domain fragment (amino acids [aa] 308 to 605) was expressed in Escherichia coli as a glutathione S-transferase (GST) fusion, purified as previously described (19), separated from the GST tag by thrombin cleavage and further purified by ion-exchange chromatography on a Mono S column. A portion (0.5 μg) of the protein was labeled at residue S584 by incubation with 1 U of the protein kinase CK2 (New England Biolabs) in the presence of [γ-32P]ATP in a volume of 10 μl. The labeled protein was diluted 200-fold in a solution containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 0.7 mg of BSA/ml, 0.1% NP-40, 5% glycerol, 5 mM DTT, 5 mM MgCl2, and 10 μl of the 32P-labeled 308-605 protein was incubated with the GST fusion proteins (~1 μg) in the absence of nucleotide or in the presence of 2 mM ADP or 2 mM ATP. After incubation at room temperature for 1 h, 5 μl of glutathione agarose gel slurry was added, and incubation was continued for 30 min. The beads were washed in 4 × 500 μl of Tris-buffered saline with 0.1% NP-40, and the bound material was analyzed by SDS-PAGE and quantitated by using a Fuji BAS 5000 imager.
ATPase assays were performed in a 20-μl reaction containing 30 mM HEPES (pH 7.5), 30 mM NaCl, 1 mM DTT, 7 mM MgCl2, 100 μg of bovine serum albumin/ml, 100 μM ATP, and 40,000 cpm of [γ-32P]ATP and E1. Reactions were incubated for 1 h at room temperature and stopped by the addition of EDTA to a final concentration of 10 mM. Then, 2-μl portions of the reactions were spotted onto a polyethyleneimine-cellulose plate, and the plates were then developed in 1 M formic acid and 0.5 M LiCl2 for 40 min. After drying, each plate was exposed to a Fuji imaging plate, and the level of free phosphate was determined by scanning the plate using a Fuji BAS imager.
Based on the structure of the hexamer of the E1 oligomerization and helicase domains (Fig. (Fig.1A),1A), we mutated all of the residues predicted to contact ATP in a BPV1 E1 hexamer (3). These 11 substitutions were generated in the E. coli expression vector pETGST-E1 by site-directed mutagenesis. Ten of the eleven residues (K425, K439, S440, D478, D479, D497, Y499, N523, Y534, and R538) were changed into alanine. Because the alanine substitution at R493 was not expressed, this residue was mutated into Leu, Met, and Glu. After expression and purification of the mutant proteins they were first tested for the ability to bind to the origin of DNA replication, together with the BPV E2 protein. The E1 and E2 proteins bind cooperatively to adjacent sites in the origin of DNA replication (17, 20, 25, 27). The resulting complex, E12E22-ori, does not rely on nucleotide binding or hydrolysis for formation and therefore serves as a convenient control for the structural integrity of the E1 mutants and for the intrinsic ability of the mutant proteins to bind to the E1 BS in the ori.
The alanine substitutions were tested for E12E22 complex formation to ascertain that they had no structural defects due to the mutations (Fig. (Fig.1B).1B). As is well established, E2 alone binds to the E2 binding site present in the ori and forms an E2 dimer complex (Fig. (Fig.1B,1B, lane 23). In the presence of wild-type (wt) E1 a larger complex E12E22 forms through cooperative binding of E1 and E2 to adjacent E1 and E2 binding sites (Fig. (Fig.1B,1B, lanes 1 and 2). All of the mutants, with one exception, could form this complex, as well as wt E1 (Fig. (Fig.1B,1B, lanes 3 to 20). The exception, R538A, showed a slight defect in formation of the E12E22 complex (Fig. (Fig.1B,1B, lanes 21 and 22). R493E, -M, and -L, also formed wt levels of E12E22-ori complex formation (data not shown).
We next tested the mutants in an ATPase assay (Fig. (Fig.1C).1C). In this assay, ATP with a radioactively labeled γ-phosphate was incubated with either the wt or the mutant E1 proteins. Hydrolysis of the ATP would result in the appearance of radioactively labeled free phosphate, which can be separated from radioactively labeled ATP by thin-layer chromatography (Fig. (Fig.1C).1C). The 10 alanine substitutions were, with one exception, devoid of detectable ATPase activity, showing a >20-fold reduction compared to wt E1. The exception (Y534, lane 10) had ~25% of the ATPase activity of the wt E1 (compare lanes 1 and 13). This result demonstrated that all of the 10 residues play a role in ATP hydrolysis. As expected, the mutants R493E, M, and L was also devoid of ATPase activity (data not shown).
Measurement of nucleotide binding for proteins such as E1 are complicated by the fact that ATP binds between two subunits, and the level of ATP binding would therefore depend greatly on whether the protein is monomeric or oligomeric. We have demonstrated that E1 in the absence of DNA is monomeric, and methods such as filter-binding assays to measure ATP binding are therefore not practical (18). Instead, we used an indirect method to measure nucleotide binding. In the absence of nucleotide, E1 binds to DNA weakly or not at all, depending on the conditions (15). The basis for this dependence on nucleotide binding for DNA binding is unknown but is likely the result of conformational changes in E1 that exposes the E1 DNA-binding domain.
To determine whether the 10 alanine substitutions were also defective for nucleotide binding, we utilized EMSA. E1 can, in the presence of ADP or ATP, form a trimer complex on any short DNA sequence (15). We incubated the wt and mutant E1 proteins with a 39-bp ori probe that lacks E1 BS. In the absence of nucleotide, no complex is observed (Fig. (Fig.2A,2A, lanes 1 to 3), while in the presence of ADP (lanes 4 to 6) the trimer band is observed. Two of the mutants, D479A and Y534A similar to wt E1, bound as a trimer in the presence of ADP (Fig. (Fig.2A,2A, lanes 10 to 12 and lanes 16 to 18, respectively) but not in its absence (Fig. (Fig.2A,2A, lanes 7 to 9 and lanes 13 to 15, respectively). Two other mutants, D478A and N523A, showed a different behavior. These two mutants showed significant trimer formation both in the absence (Fig. (Fig.2B,2B, lanes 7 to 9 and lanes 13 to 15, respectively) and in the presence of ADP (Fig. (Fig.2B,2B, lanes 10 to 12 and lanes 16 to 18, respectively). The remaining six alanine substitutions—K425A, K439A, S440A, D497A, Y499A, and R538A—failed to form the trimer complex (Fig. (Fig.2C),2C), as did the three R493 substitutions (Fig. (Fig.2D,2D, lanes 2 to 12).
These results placed the mutants in three different categories. The largest group, seven mutants, failed to form a trimer in the absence and presence of ADP, indicating that these mutants are defective for nucleotide binding. The second group with the two mutants D479A and Y534A behaved as wt E1, i.e., these mutants formed the trimer, but only in the presence of ADP. Since D479A was defective for ATP hydrolysis, this demonstrates that this mutant can bind but not hydrolyze ATP. Y534A can also clearly bind ATP but is only slightly defective for ATP hydrolysis (Fig. (Fig.1B).1B). The third group, consisting of the mutants D478A and N523A, are the most interesting mutants, since they behaved as if they had nucleotide bound even in the absence of nucleotide.
We focused on the four mutants that were capable of DNA binding in the E1 trimer assay above and tested these mutants for the ability to form the functional double trimer (DT) and double hexamer (DH) (15) (Fig. (Fig.3).3). As expected, wt E1 failed to form a complex in the absence of nucleotide (Fig. (Fig.3A,3A, lanes 1 and 2) formed a DT (E16) in the presence of ADP (lanes 3 to 4) and a DH (E112) in the presence of ATP (lanes 5 and 6). The mutant D478A (Fig. (Fig.3A,3A, lanes 7 to 12) formed ladders in the absence of nucleotide (lanes 7 and 8). In the presence of ADP and ATP a DT was formed (Fig. (Fig.3A,3A, lanes 9 and 10 and lanes 11 and 12, respectively). N523A (Fig. (Fig.3B,3B, lanes 10 to 18) behaved similarly. These results clearly indicate that D478A and N523A are incapable of hydrolyzing ATP, a finding consistent with the ATPase assays but, because a qualitative effect of ADP addition is observed (a ladder versus a DT), these proteins can both bind nucleotide. However, since DNA binding is observed even in the absence of nucleotide these mutations clearly have an effect on the DNA binding properties of the protein in the absence of nucleotide, mimicking some aspects of nucleotide binding, as observed above.
The mutants D479A (Fig. (Fig.3A,3A, lanes 13 and 14) and Y534A (Fig. (Fig.3A,3A, lanes 19 and 20) behaved as predicted from the trimer formation and failed to form DT without nucleotide. D479A also formed the DT very weakly in the presence of ADP (Fig. (Fig.3A,3A, lanes 15 to 18) and instead formed a larger complex of unknown composition. This indicates that this mutant may have a defect in ADP binding; however, the DT was formed more efficiently in the presence of ATP (lanes 17 to 18). Y534A (Fig. (Fig.3A,3A, lanes 21 to 24), however, formed DT in the presence of both ADP and ATP, showing no indication that the residual ATPase activity of this mutant allows the formation of the DH.
The DNA binding properties of the two mutants D478A and N523A clearly indicate that they are defective for ATP hydrolysis since the DH does not form in the presence of ATP. This result is consistent with the ATPase assays, which also demonstrated a lack of ATPase activity for these mutants (Fig. (Fig.1C).1C). Furthermore, these two mutants have substantial DNA-binding activity in the absence of nucleotide, indicating that the mutations may mimic the nucleotide bound state. Interestingly, however, both of these mutants are also affected by addition of nucleotide, indicating that they can still bind nucleotide.
The behavior of D478A and N523A indicated that these particular mutations mimic a conformational change that normally is induced by nucleotide binding, giving rise to DNA-binding activity in the absence of nucleotide. Another possibility is that these particular mutant proteins already have nucleotide stably bound that survives the purification procedure. To distinguish between these two possibilities, we performed the experiment shown in Fig. Fig.4.4. We first incubated the E1 protein under EMSA conditions with 2 mM ADP at room temperature for 10 min. The sample was then diluted 20-fold to provide a final ADP concentration of 0.1 mM. Probe was then added to the reaction mixture, and the sample was divided in two. To one half no ADP was added (lane 2), and to the other half 2 mM ADP was added (lane 3). wt E1 failed to form trimer without nucleotide (lane 1) and also failed to form trimer in the sample diluted without nucleotide (lane 2) but formed a robust complex in the presence of added ADP (lane 3). This result demonstrates that 0.1 mM ADP is not sufficient for trimer formation. It also demonstrates that wt E1 does not bind ADP sufficiently well to allow dilution and therefore that the half-life of nucleotide bound to wt E1 is very short since the duration of the dilution is less than 1 min.
We performed the same experiment with N523A and D478A (Fig. (Fig.4,4, lanes 4 to 9). These mutants, which bind DNA without nucleotide, showed no difference in trimer formation without ADP and with the diluted ADP (Fig. (Fig.4,4, lanes 4 and 5 and lanes 7 and 8, respectively), demonstrating either that they like wt E1 bind nucleotide with a very short half-life or that they cannot bind nucleotide. Since we have already shown that both of these proteins can bind nucleotide in Fig. Fig.3,3, this result demonstrates that the half-life for bound nucleotide is very short. That N523A is capable of binding ADP is also clearly demonstrated by the incubation with 2 mM ADP, which resulted in a substantial increase in binding (Fig. (Fig.4,4, lane 6). The D478A mutant behaved similarly except that addition of ADP did not stimulate trimer formation (Fig. (Fig.4,4, lanes 7 to 9), a finding consistent with the results in Fig. Fig.22.
These results demonstrate that N523A clearly can bind ADP but that the resulting E1-ADP complex has a very short half-life, since after dilution, which takes less than 1 min, the effect of the ADP is completely lost (compare lanes 4 and 5 in Fig. Fig.4).4). Since the E1 purification procedure does not include ADP and takes about 48 h, it is very unlikely that ADP could still be bound to E1 after this procedure. This result demonstrates that these mutations induce conformational changes, which is similar to those induced by nucleotide binding.
Our expectation was that the majority of the residues that we mutated in E1 would affect both ATP binding and hydrolysis, and it is extremely gratifying to note the exceptional predictive value of the E1 hexamer structure. Of the 11 residues that were predicted to be involved in ATP binding and hydrolysis in the structure, when mutated 10 were devoid of ATPase activity, and one mutant (Y534A) showed a fourfold reduction in ATPase activity. Seven of the residues (K425A, K439A, S440A, D497A, Y499A, R538A, and R493M, -L, and -E) when mutated, resulted in a protein that was defective for both ATP binding and hydrolysis. These include the expected Walker A mutations (K439A and S440A), the Sensor 2 mutation (K425A), and the arginine finger (R538A), plus an additional three residues (R493A, D497A, and Y499A). The only surprises in this group were the Sensor 2 and the arginine finger mutations, which generally affect ATP hydrolysis but not ATP binding.
We also hoped that some mutations would affect ATP hydrolysis only, since these mutant proteins would be useful reagents. Three of the mutants—D478A, D479A, and N523A—are completely defective for ATP hydrolysis (Fig. (Fig.1B)1B) but still are affected by the addition of ATP, indicating that they can bind nucleotide (Fig. (Fig.3).3). Interestingly, these mutants represent the Walker B motif (D478 and D479) and the Sensor 1 (N523), three residues that are in close proximity to each other. Two of these mutants (D478A and N523A) also showed a completely unexpected phenotype since these proteins, even in the absence of nucleotide, are capable of binding DNA. A particularly striking result is the trimer formation in Fig. Fig.2B,2B, which clearly demonstrates that D478A and N523A are fully competent for trimer formation in the absence of nucleotide. It is also interesting that the phenotype that we observe in Fig. Fig.3,3, in contrast to the results in Fig. Fig.2,2, only partially mimics nucleotide binding. Here, although D478A and N523A clearly show that they can bind DNA in the absence of nucleotide, a ladder is formed instead of the DT, and DT formation requires addition of ADP. This result indicates that the nucleotide binding has at least two functions in these assays, one is to activate DNA binding in a generic way and the other is specifically to allow or to stimulate DT formation and that these two functions can be separated. Only the first of these functions is mimicked by the D478A and N523A mutations.
Currently, we cannot propose a structural model for how the D478A and N523A mutations result in a conformation that mimics the ATP-bound state for either of these mutants. It is well established that the conserved residues in Walker B (D478 and D479) are involved in hydrolysis of ATP and, consistent with such a function neither of the alanine substitutions, can hydrolyze ATP, although both can bind nucleotide. The Sensor 1 residue, N523, is generally involved in contacting Walker B and the γ-phosphate of ATP. The normal function of these residues in AAA+ proteins does not explain the ability of these mutations to mimic the nucleotide-bound state. Furthermore, neither of these residues is in a radically different conformation in the structures generated in the absence or presence of nucleotide (3).
These data therefore fail to provide a mechanism for how nucleotide binding results in generic activation of DNA binding. Our data also do not explain why the E1DBD in the context of the full-length E1 is inactive for DNA binding unless nucleotide is present. A possibility that would explain these phenomena is that the failure of E1 to bind DNA in the absence of nucleotide might be caused by a physical obstruction of the E1 DBD. Such an obstruction would require that another part of E1 could interact with the DBD and block DNA binding. Because the blockage would occur only in the absence of nucleotide, the region responsible for the blockage presumably would be affected by nucleotide binding, such as the E1 oligomerization and helicase domain, which is located immediately C-terminal to the E1 DBD. In this scenario, the two residues D478 and N523 in the helicase domain would be involved in interaction with the E1 DBD in the absence of nucleotide.
To determine whether the E1 DBD interacts physically with the E1 oligomerization and helicase domain, we used two versions of GST E1DBD (aa 142 to 308 and aa 159 to 303) to attempt to pull down the C-terminal half of E1 (aa 308 to 605), which contains the oligomerization and helicase domains and binds nucleotide (Fig. (Fig.5).5). We labeled the C-terminal half of E1 by in vitro phosphorylation using the protein kinase CK2 and [γ-32P]ATP. This results in phosphorylation on residue S584 in E1. CK2α is autophosphorylated and gives rise to a band that migrates slower than 308-605 (lane 1, input). As a negative control we performed pulldown assays in the absence and presence of ATP (lanes 2 and 3) using GST 1-60, which contains the N-terminal 60 residues of E1 fused to GST. As expected, the GST 1-60 fusion protein could bring down neither CK2α nor 308-605. We next used the two GST E1 DBD fusions, 142-308 and 159-303, and performed pulldown assays in the absence of nucleotide (lanes 4 and 7) in the presence of ADP (lanes 5 and 8) or in the presence of ATP (lanes 6 and 9). Interestingly, the 308-605 fragment was efficiently pulled down by both of the GST E1DBD fragments in the absence of nucleotide (lanes 4 and 7). In the presence of ADP the interaction was reduced twofold for both GST E1DBD fragments (lanes 5 and 8) and in the presence of ATP the interaction was reduced fourfold for both GST E1DBD fragments (compare lane 4 to lane 6 and lane 7 to lane 9). Importantly, these results demonstrate that the E1 DBD interacts efficiently with the oligomerization and helicase domains and that this interaction is affected by the presence of nucleotide consistent with the model presented above. We currently do not understand why the interaction between the E1 DBD and the 308-605 fragment is not completely abolished by the presence of nucleotide. Furthermore, the D478A mutation did not appreciably affect the interaction with the E1 DBD (data not shown). Consequently, although some of the pulldown results are consistent with the model presented above, other results such as the residual interaction in the presence of ATP and the failure of the D478A mutation to disrupt the interaction indicate that our model lacks some important components. The resolution of these questions will require further analysis of the interaction between the E1 DBD and the E1 helicase domain.
This research was supported by grant RO1 AI 072345 to A.S.
Published ahead of print on 25 November 2009.