Crystal structure of the NS3-like helicase from Alongshan virus

Alongshan virus (ALSV) is an emerging human pathogen that was identified in China and rapidly spread to the European continent in 2019, raising concerns about public health. ALSV belongs to the distinct Jingmenvirus group within the Flaviviridae family with segmented RNA genomes. While segments 2 and 4 of the ALSV genome encode the VP1–VP3 proteins of unknown origin, segments 1 and 3 encode the NS2b–NS3 and NS5 proteins, which are related to Flavivirus nonstructural proteins, suggesting an evolutionary link between segmented and unsegmented viruses within the Flaviviridae family. Here, the enzymatic activity of the ALSV NS3-like helicase (NS3-Hel) was characterized and its crystal structure was determined to 2.9 Å resolution. ALSV NS3-Hel exhibits an ATPase activity that is comparable to those measured for Flavivirus NS3 helicases. The structure of ALSV NS3-Hel exhibits an overall fold similar to those of Flavivirus NS3 helicases. Despite the limited amino-acid sequence identity between ALSV NS3-Hel and Flavivirus NS3 helicases, structural features at the ATPase active site and the RNA-binding groove remain conserved in ALSV NS3-Hel. These findings provide a structural framework for drug design and suggest the possibility of developing a broad-spectrum antiviral drug against both Flavivirus and Jingmenvirus.


Introduction
Alongshan virus (ALSV) is a novel tick-borne virus associated with human disease that was first identified in northeastern China . Patients infected with ALSV present with fever, persistent headache, fatigue and nausea. Most confirmed ALSV cases had a clear history of tick bites. Soon after the identification of ALSV, the RNA of the virus was detected in Ixodes ricinus ticks in Finland; these ticks are a common species across Europe (Kuivanen et al., 2019). ALSV is a positive-sense single-stranded RNA virus with a segmented genome. It was classified into the Jingmenvirus group of the Flaviviridae. Another segmented tick-borne virus capable of infecting humans is Jingmen tick virus (JMTV). JMTV was first identified approximately a decade ago in Rhipicephalus microplus ticks in China (Jia et al., 2019). The emergence and spread of these novel pathogens pose potential threats to human health and an in-depth study of the proteins encoded by these viruses is thus warranted.
Most viruses of the Flaviviridae family have an unsegmented genome. These include the Flavivirus, Pestivirus, Hepacivirus and Pegivirus groups (Qin et al., 2014). They host an $11 kb positive-sense single-stranded RNA genome containing a single open reading frame that encodes a polyprotein precursor. The polyprotein is processed by viral and cellular proteinases, yielding three structural proteins (E, PrM and C) and seven nonstructural proteins (NS1,NS2a,NS2b,NS3,NS4a,NS4b and NS5). In contrast, the segmented RNA viruses of the Jingmenvirus group (such as ALSV and JMTV) host four genomic segments, with a total genome size similar to those of the flaviviruses. Segments 1 and 3 encode proteins related to Flavivirus NS5 (containing RNA-dependent RNA polymerase and methyltransferase motifs) and the NS2b-NS3 complex (containing proteinase and RNA helicase motifs), whereas segments 2 and 4 encode the proteins VP1-VP3 that are unrelated to Flavivirus proteins and are of unknown origin (Qin et al., 2014;Wang et al., 2019). These findings suggest an unusual evolutionary link between the unsegmented and segmented viruses in the Flaviviridae family.
The Flavivirus NS3 protein is one of the most studied nonstructural proteins because of its central role in virus replication. It is the core component of the membrane-bound Flavivirus replication complex and has multiple enzymatic activities. NS3 contains an N-terminal protease domain and a C-terminal RNA helicase domain. While the proteinase domain participates in polyprotein processing, the RNA helicase domain is involved in viral RNA capping and synthesis (Ferron et al., 2005). A collection of NS3 helicase structures from unsegmented flaviviruses have been reported to date, including those from Dengue virus (DENV; Xu et al., 2005), Yellow fever virus (YFV; Wu et al., 2005), West Nile virus (WNV; Mastrangelo et al., 2007), Hepatitis C virus (HCV; Kim et al., 1998;Cho et al., 1998) and Zika virus (ZIKV; Jain et al., 2016;Bukrejewska et al., 2017;Cao et al., 2016;Fang et al., 2019;Li et al., 2018;Tian, Ji, Yang, Zhang et al., 2016;Xu et al., 2019;Yang et al., 2018). These studies demonstrate that NS3 helicases not only exhibit a common fold, but also share a similar mechanism underlying ATP hydrolysis, RNA recognition and unwinding. The putative RNA helicases identified in the NS3-like proteins of the segmented RNA viruses share limited sequence identity with that of Flavivirus NS3, raising an intriguing question as to whether the structure and function of the NS3-like helicase is preserved in segmented RNA viruses.

Protein expression and purification
The DNA encoding the C-terminal helicase domain of ALSV NS3 (NS3-Hel; residues 322-810; GenBank AXE71876.1) was synthesized and inserted into a pET-28a-SUMO vector between the BamHI and XhoI sites, expressing ALSV NS3-Hel with an N-terminal 6ÂHis-SUMO tag. The resulting vector was transformed into Escherichia coli BL21(DE3) competent cells. A single colony was picked and the bacterial culture was grown in LB medium containing 50 mg l À1 kanamycin at 37 C. Expression was induced by adding 0.5 mM isopropyl -d-1-thiogalactopyranoside (IPTG) when the OD 600 reached 1.0. The bacterial culture was rapidly cooled to 18 C and shaking was continued at 18 C overnight. The cell pellets were harvested and resuspended in lysis buffer consisting of 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 10 mM imidazole, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM -mercaptoethanol. The bacterial cells were lysed by ultrasonication on ice and the cell debris was removed by centrifugation at 20 000 rev min À1 for 30 min. The supernatant was filtered through a 0.45 mm filter and then loaded onto Ni-NTA resin pre-equilibrated with lysis buffer. The resin was washed twice with ten column volumes of wash buffer consisting of 50 mM Tris-HCl pH 8.0, 100 mM NaCl, 20 mM imidazole, 1 mM PMSF, 1 mM -mercaptoethanol to remove nonspecifically bound proteins. Subsequently, the 6ÂHis-SUMO tag was cleaved on the column by adding Ulp1 peptidase at 4 C overnight. The flowthrough containing nontagged ALSV NS3-Hel was collected and subjected to a HiTrap Q HP column (GE Healthcare) pre-equilibrated with buffer consisting of 20 mM Tris-HCl pH 8.0, 75 mM NaCl. Nontagged ALSV NS3-Hel was eluted with a linear gradient of NaCl from 75 mM to 1 M.
ALSV NS3-Hel containing l-selenomethionine residues was prepared by transforming the vector into E. coli B834 (DE3) competent cells. The bacterial cells were grown in LeMaster medium (Molecular Dimensions) supplemented with l-selenomethionine. The purification of this derivative was the same as described above for the native protein.

Crystallization and structure determination
The ALSV NS3-Hel protein was concentrated to approximately 2 mg ml À1 prior to crystallization trials. Crystallization was performed in a hanging-drop vapor-diffusion setup at 20 C. 1 ml protein sample was mixed with 1.2 ml crystallization buffer consisting of 0.16 M calcium acetate, 20%(v/v) PEG 3350, 5 mM tris(2-carboxyethyl)phosphine. The crystals were soaked in crystallization buffer containing 10% ethylene glycol and flash-cooled in liquid nitrogen. X-ray diffraction experiments were conducted on the X06DA beamline at the Swiss Light Source, Paul Scherrer Institute, Villigen, Switzerland. Highly redundant data were collected usings X-rays at a wavelength of 0.9791 Å . The crystal diffracted X-rays to 2.9 Å resolution and belonged to space group P2 1 2 1 2 1 . The data were processed using XDS (Kabsch, 2010). SHELXC/D/E were used to locate heavy atoms (Se) and to calculate an initial electron-density map. The preliminary atomic model with a single molecule in the asymmetric unit was built using the CRANK2 pipeline in the CCP4 package (Winn et al., 2011). The preliminary model was improved by manual model building using Coot (Emsley et al., 2010). The structure was refined using Phenix (Liebschner et al., 2019). All structural figures were prepared using PyMOL (Schrö dinger).

ATPase activity assay
The ATPase assay was performed as described previously (Kuo et al., 1996). Each reaction mixture (50 ml) consisted of

Results and discussion
To gain structural and functional insights into the putative NS3-like helicase identified in ALSV (GenBank AXE71876.1), we overexpressed the C-terminal portion of ALSV NS3 (residues 322-810) containing the predicted RNA helicase domain, designated ALSV NS3-Hel. A 6ÂHis-SUMO tag was fused to the N-terminus of ALSV NS3-Hel. The recombinant protein was purified and the 6ÂHis-SUMO tag was cleaved using a SUMO-specific proteinase [Figs. 1(a) and 1(b)]. To investigate the enzymatic activity of ALSV NS3-Hel, we performed an ATPase assay. We found that ALSV NS3-Hel could hydrolyze ATP with a K m value of 55 AE 8 mM and k cat = 0.61 AE 0.04 s À1 [ Fig. 1(c)]. In addition, we evaluated the UTP hydrolysis activity of ALSV NS3-Hel. The enzyme hydrolyzed UTP with a K m value of 185 AE 19 mM and k cat = 0.85 AE 0.02 s À1 (Supplementary Fig. S1). Hence, the results indicate that ALSV NS3-Hel does not have significant specificity for NTP, which is consistent with the specificities of many other Flavivirus NS3 helicases. Finally, we compared the NTP hydrolysis activity of ALSV NS3-Hel with a selection of nonsegmented NS3-Hels (Supplementary Table S1), demonstrating that the NTPase activity of ALSV NS3-Hel is comparable to those measured for characterized nonsegmented NS3-Hels (Tian, Ji, Yang, Zhang et al., 2016;Suzich et al., 1993;Warrener et al., 1993;Jin & Peterson, 1995;Kuo et al., 1996;Mancini et al., 2007;Speroni et al., 2008;Assenberg et al., 2009;Yang et al., 2018;Xu et al., 2019).
ALSV NS3-Hel shares limited sequence identity (15-28%) with Flavivirus NS3 helicases, and none of the available NS3-Hel structures could be used as a homologous model for structure determination. We therefore adopted an ab initio phasing strategy. We overexpressed ALSV NS3-Hel in E. coli B834 (DE3) cells to obtain a derivative containing selenomethionine residues. The crystals of ALSV NS3-Hel diffracted X-rays to 2.9 Å resolution. The structure was solved using the single-wavelength anomalous dispersion (SAD) method. The final structure has R work and R free values of 26.1% and 29.2%, respectively. The statistics for data collection, phasing and structure refinement are summarized in Supplementary Tables  S2 and S3.
The overall fold of ALSV NS3-Hel is similar to that of the nonsegmented viral NS3 helicases. We used the DALI server (http://ekhidna2.biocenter.helsinki.fi/dali/) to search for structural homologs of ALSV NS3-Hel (Supplementary Table  S4 Fig. 2). A structure-similarity dendrogram was derived by average linkage clustering of the structure-similarity matrix (DALI Z-scores). The Z-scores between ALSV NS3-Hel and Flavivirus NS3 helicase structures range from 24.9 to 30.4, with r.m.s.d.s of 2.9-3.2 Å . By contrast, the Z-scores between the ALSV NS3-Hel structure and HCV NS3 helicases ranges from 20.7 to 22.3, with r.m.s.d.s of 3.8-4.4 Å . The unusually high r.m.s.d. values may be attributed to the interdomain movement between the D1 and D2 domains triggered by RNA or ATP binding and the less conserved D3 domain. We therefore compared the structure of the isolated domains of ALSV NS3-Hel with their counterparts in various flaviviral NS3 structures (Supplementary Table S4). Superimposing the D1 and D2 domains of ALSV NS3-Hel onto flaviviral NS3 structures gave r.m.s.d.s of 2.0-2.7 and 1.9-2.6 Å , respectively.
By contrast, the structure of the D3 domain of ALSV NS3-Hel exhibited very limited similarity to the D3 domain of flaviviral NS3 structures, which is consistent with the high variability of the D3 domain in flaviviral NS3 helicases. These data suggest that ALSV NS3-Hel is structurally more related to Flavivirus NS3 helicases. The structure of ALSV NS3-Hel has a flattened triangular shape [ Fig. 3(a)], which can be divided into three domains: the N-terminal domains 1 and 2 (D1, residues 322-480; D2, residues 481-640) and the C-terminal domain 3 (D3). D1 and D2 are tandem RecA-like domains with an / fold. D1 is composed of six parallel -sheets (1, 2, 2A, 3, 4 and 5) sandwiched by four -helices (1-4), whereas D2 contains six parallel -sheets (1 0 , 2 0 , 3 0 , 3A 0 , 4 0 and 5 0 ) sandwiched by four -helices (1 0 -4 0 ). A -hairpin composed of a pair of antiparallel -sheets (4A 0 -4B 0 ) protrudes from the D2 domain and interacts with the D3 domain. In particular, the -hairpin is packed against a Structure-similarity dendrogram of SF2 helicases structurally related to ALSV NS3-Hel. In a search for structural homologs of ALSV NS3-Hel using the DALI server, the most related structures (DALI Z-score ! 18.0) and the structure of ALSV NS3-Hel were submitted to all-against-all structure comparison (http:// ekhidna2.biocenter.helsinki.fi/dali/) to generate the structure-similarity dendrogram. The dendrogram is derived by average linkage clustering of the structure-similarity matrix (DALI Z-scores). The protein name, PDB code and chain ID are indicated. ALSV NS3 is highlighted with a yellow background. hydrophobic patch formed by 1 00 , 2 00 and the N-terminus of 5 00 of the D3 domain.
Unlike the available NS3-Hel structures [ Fig. 3(c)], the D3 domain of ALSV NS3-Hel is partially disordered. While the first half of D3 (1 00 -5 00 ; residues 641-735) is visible in the electron-density map, the C-terminal 75 residues (736-810) are missing [ Fig. 3(a)]. The last visible residue, Arg735, at the C-terminus of ALSV NS3-Hel extends into the solvent. In the SDS-PAGE analysis of the fractions eluted from a HiTrap Q column [ Fig. 1(b)], protein-degradation products were visible. We therefore analyzed the crystals of ALSV NS3-Hel ( Supplementary Fig. S2). Multistep transfer of ALSV NS3-Hel crystals to fresh drops of crystallization buffer was carried out to remove free protein. The proteolytic products were present in the crystals and their proportion was clearly increased in comparison with the sample before crystallization. This result indicates that the missing C-terminal portion of the D3 domain might be owing to proteolytic cleavage that occurred during purification despite the presence of a high concentra-tion of proteinase inhibitor (see Section 2). This high susceptibility to proteolysis may reflect an usual intrinsic flexibility of the D3 domain. For example, the D3 domain of the DENV NS3 helicase is implicated in binding the viral RNA-dependent RNA polymerase NS5 (Johansson et al., 2001). The concave surface between the D2 and D3 domains of the DENV NS3 helicase has been proposed to bind the duplex portion of the RNA substrate ahead of the partially opened fork (Sampath et al., 2006). Therefore, it is possible that the D3 domain of ALSV NS3-Hel is also involved in the binding of other proteins or RNA, which is essential for the stabilization of its conformation. All conserved SF2 helicase motifs are located in the cleft formed between the D1 and D2 domains. While the D1 domain contains motifs I (P-loop/ Walker A), Ia, II (Walker B) and III, the D2 domain contains motifs IV, IVa, V and VI [ Fig. 3 Fig. S4(a) and Supplementary Table S5]. We found that the residues involved in the recognition of NTP, hydrolysis intermediates, catalytic water and metal ion are invariant in ALSV NS3-Hel, which suggests that the catalytic mechanism underlying ATP hydrolysis is conserved in the NS3-Hels from segmented viruses. In motif I, the structural equivalents of Gly197 and Gly199 in ZIKV NS3 are Gly349 and Gly351 in ALSV NS3, suggesting their role in the recognition of the triphosphate moiety of NTP. Thr201 and Glu286 of ZIKV NS3 coordinate an Mn 2+ ion for NTP hydrolysis, and their equivalents in ALSV NS3 are Thr353 (motif I) and Glu438 (motif II). Arg202 of ZIKV NS3 stabilizes the adenosine base of ADP, and its match in ALSV NS3 is Arg354 (motif I).  The active site and RNA-binding groove of ALSV NS3-Hel. (a) Invariant residues in the active site of ALSV NS3-Hel are shown as stick models (blue); their counterparts in the superimposed ZIKV NS3-Hel (PDB entry 5y6m; Yang et al., 2018) are shown in gray. The bound ligands ADP (orange), AlF 3 (cyan) and manganese ion (purple) are also shown. Residues from ALSV NS3-Hel are annotated in italics. (b) Structural superimposition of the P-loops of apo ALSV NS3-Hel (blue), apo DENV4 NS3-Hel (red), DENV4 NS3-Hel complexed with RNA and AMPPNP (green), apo ZIKV NS3-Hel (orange) and ZIKV NS3-Hel complexed with RNA (yellow). (c) A model of an ALSV NS3-Hel-RNA complex. The RNA was modeled into the ALSV NS3-Hel structure by superimposition with the DENV4 NS3-Hel-RNA complex (PDB entry 2jlv; Luo et al., 2008). ALSV NS3-Hel is colored by domain: D1, light blue; D2, light green; D3, pink. Residues that were predicted to contact RNA are colored blue and shown as stick models. (d) Aromatic residues located at the 5 0 end of the model RNA. Residues from ALSV NS3-Hel are shown as magenta stick models and their structural counterparts in DEVN4 NS3-Hel are shown as cyan stick models. RNA is shown a a red stick model.
-phosphate of NTP, and their analogs in ALSV NS3 are Lys352, Arg620 and Arg623, respectively. Gln455 of ZIKV NS3 coordinates a catalytic water and the nearby -phosphate (or its mimic AlF 3 ), and its match in ALSV NS3 is Gln616. While most of the catalytic residues of ALSV NS3-Hel superimpose well with their equivalents in ZIKV NS3, the P-loop of ALSV NS3-Hel adopts a different conformation. In particular, Thr353 and Arg354 of ALSV NS3-Hel do not superimpose with their counterparts in ZIKV NS3 [ Fig. 4(a)]. This might be owing to the fact that the structure of ALSV NS3 is in the apo form and ATP binding might induce conformational changes of the P-loop. This speculation is consistent with the finding that the P-loop adopts various conformations in different flaviviral NS3 structures (Yang et al., 2018), suggesting intrinsic flexibility of this region.
DENV NS3-Hel undergoes major conformational changes upon RNA binding (Luo et al., 2008). In particular, RNA binding triggers a conformational switch of the P-loop to a catalytically competent state, offering a mechanistic explanation for RNA-stimulated ATP hydrolysis. We superimposed the structure of ALSV NS3-Hel with the structures of DENV4 NS3-Hel and ZIKV NS3-Hel in apo and RNA-bound forms, and compared the structures of the P-loop [ Fig. 4(b)]. While the conformational differences of the P-loop of ZIKV NS3-Hel between apo and RNA-bound forms are small, significant conformational changes were observed between the different forms of DENV NS3-Hel. The P-loop conformation of apo ALSV NS3-Hel is more similar to the catalytically competent state than to the unusual P-loop in apo DENV NS3-Hel. A full understanding of the RNA-induced conformational changes of ALSV NS3-Hel will requires the further investigation of enzyme-RNA complexes.
It has been reported that single-stranded RNA binding affects the coordination of the divalent ion, which may affect the catalytic activity and facilitate the release of ADP-Mn 2+ (Luo et al., 2008). Thr200 and Glu285 coordinating the divalent ion in DENV NS3-Hel are invariant in ALSV NS3-Hel as Thr353 (motif I) and Glu438 (motif II). We superimposed the ALSV NS3-Hel structure with those of the DENV4 NS3-Hel-AMPPNP-RNA and the DENV4 NS3-AMPPNP complexes ( Supplementary Fig. S5). The comparison reveals that the conformation of Glu438 in ALSV NS3-Hel is similar to the conformation of Glu285 of the DENV4 NS3-Hel-AMPPNP-RNA complex, which is not an optimal conformation for ion coordination. Additionally, Thr353 of ALSV NS3-Hel is located $4.4 Å apart from its counterpart in DENV4 NS3-Hel. To understand whether ATP or RNA binding triggers the optimal P-loop conformation for divalent ion coordination requires further structure determination of ALSV NS3-Hel in complex with ATP, divalent ion and RNA.
To investigate the interaction between ALSV NS3-Hel and RNA, we modeled a single-stranded RNA into the ALSV NS3-Hel structure by superimposing it with the structure of the DENV4 NS3-RNA-AMPPNP complex (PDB entry 2jlv; Luo et al., 2008). This analysis revealed a set of residues in ALSV-Hel that may participate in RNA recognition, most of which are highly conserved [ Fig. 4(c), Supplementary Fig.   S4(b) and Supplementary Table S6]. This model demonstrates that the single-stranded RNA is accommodated in the groove separating the D1 and D2 domains from the D3 domain of ALSV NS3-Hel. The 3 0 portion of the modeled RNA strand is located on top of the D1 domain. Thr376 and Arg377 (motif Ia) may recognize the phosphodiester backbone of the RNA, whereas Pro375 and Thr416 may recognize the 2 0 -OH group and Ser443 may recognize the nucleotide base. The 5 0 portion of the modeled RNA makes contacts with residues from the D2 domain (motifs IV, IVa and V). Leu523, Arg549 and Thr572 may interact with the RNA backbone, whereas Pro521 and Ser573 may interact with the 2 0 -OH group of the RNA and Tyr593 and Tyr595 may interact with the RNA base. The -hairpin extending from the D2 domain (4A 0 -4B 0 ) forms a wall of the RNA-binding tunnel which accommodates the 5 0 end of the single-stranded RNA. It is worth noting that the structural counterparts of the two aromatic residues Tyr593 and Tyr595 in ALSV NS3 are the hydrophilic residues Pro431 and Leu429 in DENV4 NS3. Both of these residues are involved in interaction with the RNA base. The availability of Tyr593 and Tyr595 at this location suggest that they could recognize the RNA base viastacking (Fig. 4d). Our multiple sequence alignment indicates that this feature is conserved in the segmented flaviviruses but is not present in the nonsegmented flaviviruses ( Supplementary Fig. S3). While the structural counterparts of Tyr593 in the nonsegmented flaviviruses are hydrophobic residues, the counterparts of Tyr595 are mostly prolines. Hence, it is possible that the segmented viruses adopt a distinctive mechanism of RNA recognition.
In summary, we provide biochemical and structural characterizations of the NS3-like helicase of ALSV, a member of the segmented virus group in the Flaviviridae. Our findings demonstrate that ALSV NS3-Hel exhibits an ATPase activity similar to those of NS3 helicases encoded by unsegmented flaviviruses. ALSV NS3-Hel exhibits an overall fold resembling that of Flavivirus NS3 helicases. Structure-similarity analysis showed that ALSV NS3-Hel is more structurally related to Flavivirus NS3 helicases than to HCV NS3 helicases. These results reveal the unusual evolutionary link between the unsegmented and segmented RNA viruses of the Flaviviridae family from structural and biochemical perspectives. Our results provide a structural framework for the design of antiviral agents and suggest the possibility of developing widespectrum antivirals targeting the Flavivirus and Jingmenvirus groups owing to the shared structural features in their NS3-Hels.