ATP hydrolysis and RNA unwinding assays
Flavivirus helicases have both ATP hydrolysis and RNA unwinding activities. For structural studies, we have made two constructs (helicase172–617 and helicase180–617) to express the ZIKV helicase. The kinetic parameters of ATP hydrolysis for the long form of the ZIKV helicase172–617 were determined using the Malachite green assay as reported previously (Lanzetta et al., 1979). The resulting data show that ZIKV helicase displays ATPase activity with K
= 191 ± 26 μmol/L and k
= 2.3 ± 0.1 s−1 (Fig. 1A). The short form of the ZIKV helicase180–617 can also hydrolyze ATP and the activity difference between the short form and the long form are negligible. The RNA unwinding activity was assayed using radiolabeled double-stranded (ds) RNA, in the presence of Mg2+, ATP, and various concentrations of enzyme (Fig. 1B). It demonstrated that ZIKV helicase displayed strand displacement activity for dsRNA as other flavivirus helicases.
To elucidate the molecular mechanisms of ZIKV helicase in recognizing ATP/Mn2+ and RNA, we determined the crystal structures of ZIKV helicase180–617 complexed with ATP/Mn2+ and ZIKV helicase172–617 complexed with a 7-mer RNA (5′-AGAUCAA-3′) at 2.2 Å and 1.7 Å, respectively (Table S1).
The structure of the ZIKV helicase in complex with ATP and Mn2+
Due to the nucleotide hydrolysis activity, there is no structure reported for any flavivirus helicase complexed with ATP. Instead, the nucleotide analog 5′-adenylyl-β, γ-imidodiphosphate (AMPPNP) has been used to study helicase-nucleotide interaction (Luo et al., 2008). Fortunately, we captured the ZIKV helicase180–617 in an ATP-bound state, which is the first structure of any flavivirus helicase bound to ATP, even though it displays NTPase activity. The overall structure of the ZIKV helicase180–617 in complex with ATP/Mn2+ is similar to that of its apo-form (overall RMSD 0.557 Å), except for the movement of the P-loop (residues 193–203) towards the inner core and lateral movement of main-chain residues 411–416 to better accommodate ATP (Fig. 1A and 1B). As we have reported previously, the P-loop, which is critical for NTP binding and catalysis, has a variety of structural conformations among flavivirus apo-helicases (Tian et al., 2016). It is worthwhile to note that upon nucleotide binding, ATP/Mn2+ induced marked conformational change of the P-loop, also seen in DENV4 helicase (Luo et al., 2008) (Fig. 2B). Interestingly, we found that the P-loop and other elements which constitute the NTP binding pockets of ZIKV and DENV4 helicases undergo different local rearrangements, but then adopt an identical mode to recognize ATP/Mn2+. However, their apo-conformations are distinct from each other. This suggests that flavivirus helicases have evolved a conserved molecular engine to convert chemical energy into mechanical energy for unwinding viral RNA during replication.
In the ZIKV helicase180–617-ATP-Mn2+ tertiary structure, ATP/Mn2+ are located at the cleft between Domain I and II (Fig. 2A and 2C). Substrate binding causes an inward reorientation of side-chain of K200 to stabilize the triphosphate moiety of ATP and a flipping of the side-chain of R202 towards the solvent to leave room for sugar moiety. The triphosphate moiety of ATP adopts an extended conformation as seen in DENV4 helicase in complex with AMPPNP (Luo et al., 2008). The Mn2+ ion is coordinated in an octahedral geometry by side chains of E286 (motif II) and T201, two ordered water molecules and two oxygen atoms from the β/γ phosphate groups of the ATP molecule, which stabilize the nucleoside triphosphate. The ATP molecule makes additional contacts with G197, K200, R202 (P-loop), R459, R462 (motif VI) and other ordered water molecules (Fig. 2C). Among them, K200 is responsible for interacting with the γ-phosphate of the nucleotide during transition state. The 3′-OH group of the ribose forms hydrogen bonds with the carbonyl oxygen of R462 and the side chain amide group of N330. The ribose group of ATP bulges out from the binding pocket and no clear electron densities are observed for the adenine group, suggesting that the ZIKV helicase may not have nucleotide specificity for its NTPase activity.
The structure of ZIKV helicase in complex with RNA
In the structure of the ZIKV helicase172–617 in complex with a 7-mer RNA, the single-stranded (ss) RNA runs through Domain II to Domain I in an extended conformation with the bases stacked against each other, separating these two domains from Domain III. The 3′ end of ssRNA binds to Domain I, while the 5′ end mainly interacts with Domain II. Nucleotides 1–5 are well ordered and the electron densities are mostly invisible for nucleotides 6–7 (Fig. 3A).
Conformational changes upon RNA binding
Compared with its apo-form, the ZIKV helicase undergoes obvious conformational changes, largely due to a rotation of Domain II and Domain III, once it binds to ssRNA. Domain II rotates about 9° away from Domain I in a rigid-body rotation mode along axis II in the direction as noted in Fig. 3B and 3D. However, Domain III rotates about 9° away from Domain I in the opposite direction along axis III, which is approximately parallel to axis II (Fig. 3B and 3D). This rotor domain rotation caused two α-helices (residues 365–379, and residues 390–400) in Domain II and two α-helices (residues 525–537, and residues 602–615) in Domain III to move away from the RNA binding groove in an opposite direction, enlarging the groove to accommodate the ssRNA. This natural design functions like a double-leaf swing gate with each leaf opening in a reverse direction to the other (Fig. 3B and 3D). Interestingly, the motor domain rotation mode in the ZIKV helicase is distinct from that in the DENV4 helicase structure (Luo et al., 2008). In the DENV4 helicase, the rotation axis for Domain II, however, is almost vertical to that for Domain III, but the rotation directions are identical (Fig. 3C and 3D).
At first glance, the residues for RNA binding are well conserved in both the ZIKV and DENV apo-helicases (Fig. 4A). Additionally, similar to the DENV helicase, the ZIKV helicase binds to ssRNA by a positively charged tunnel identified along the domain boundary of Domain III, which directly interacts with Domain I and Domain II as well (Figs. 3–5). However, to our surprise, the exact RNA recognition mode differs markedly between these two structures due to the distinct motor domain rotation upon ssRNA binding. Compared with the structure of the DENV4 helicase complexed with a 12-mer RNA (Luo et al., 2008), the sugar-phosphate backbone of nucleotides 1–3 is more extended in the ZIKV helicase. The sugar group of nucleotide 1 (A) in the ZIKV helicase is ~5 Å away from that in the DENV4 helicase (Fig. 4B). This causes different conformations of subsite 1 in the ZIKV and DENV4 helicases to better fit the adenine. In particular, the side chain of K431 in the ZIKV helicase points to the inner core, forming a salt bridge with the side chain of D410 and a weak hydrogen bond with N3 atom of the adenine base (Fig. 4C). The corresponding residue (K430) in the DENV4 helicase, however, projects its side chain towards the solvent (Fig. 4D). In addition, as seen in both the ZIKV and DENV4 helicase structures, a complex network of water molecules is important for ssRNA binding, yet these water molecules may play different roles in recognizing an individual nucleotide. The specificity of the ZIKV helicase for RNA relies on multiple hydrogen bonds between the 2′-OH moieties from the ssRNA and the carbonyl oxygen of D410, side chain oxygen of T265, and six water molecules (W1–6) (Fig. 5B and 5C). In the DENV4 helicase, however, P363, P223, D409, T264 and the other three water molecules are responsible for interacting with 2′-OH moieties in the RNA, suggesting that the ZIKV helicase might depend more on the water network in discriminating between RNA and DNA than the DENV4 helicase (Fig. 5B and 5C). The detailed difference between ZIKV and DENV4 helicases for RNA interaction is shown in Fig. 5C.
Because of its essential role for replication, a viral helicase is an attractive target whose accurate mechanism is still largely unknown. Flavivirus helicases possess nucleoside triphosphatase activity, which enables the enzyme to convert chemical energy to unwind viral RNA replication intermediates. Our structures presented here can help deepen our understanding of this process and provide structural basis for rational drug design. Interestingly, although there exists conformational variety in the NTP binding pocket of apo-helicases among different flaviviruses, they undergo conformational changes to adopt an identical mode to bind NTPs, which may result from the natural selection for the same function of hydrolyzing NTPs. On the other hand, to our surprise, although the residues are well conserved for RNA binding between different flavivirus apo-helicases, distinct rotations of motor domains would cause different manners to recognize their individual RNAs during replication. These findings suggest that flaviviruses could have evolved a conserved engine to convert chemical energy to mechanical energy, but variable RNA recognition modes to adapt to their individual replication.