key: cord-0976199-7q0pr04g
authors: Yang, Yan; Zhang, Changhui; Li, Xuehui; Li, Li; Chen, Yanjuan; Yang, Xin; Zhao, Yao; Chen, Cheng; Wang, Wei; Zhong, Zhihui; Yang, Cheng; Huang, Zhen; Su, Dan
title: Structural Insight into Molecular Inhibitory Mechanism of InsP(6) on African Swine Fever Virus mRNA-Decapping Enzyme g5Rp
date: 2022-04-28
journal: Journal of virology
DOI: 10.1128/jvi.01905-21
sha: 5c3197da203888dcbe5899c62aee43a4a6fe543c
doc_id: 976199
cord_uid: 7q0pr04g

Removal of 5′ cap on cellular mRNAs by the African swine fever virus (ASFV) decapping enzyme g5R protein (g5Rp) is beneficial to viral gene expression during the early stages of infection. As the only nucleoside diphosphate-linked moiety X (Nudix) decapping enzyme encoded in the ASFV genome, g5Rp works in both the degradation of cellular mRNA and the hydrolyzation of the diphosphoinositol polyphosphates. Here, we report the structures of dimeric g5Rp and its complex with inositol hexakisphosphate (InsP(6)). The two g5Rp protomers interact head to head to form a dimer, and the dimeric interface is formed by extensive polar and nonpolar interactions. Each protomer is composed of a unique N-terminal helical domain and a C-terminal classic Nudix domain. As g5Rp is an mRNA-decapping enzyme, we identified key residues, including K(8), K(94), K(95), K(98), K(175), R(221), and K(243) located on the substrate RNA binding interfaces of g5Rp which are important to RNA binding and decapping enzyme activity. Furthermore, the g5Rp-mediated mRNA decapping was inhibited by InsP(6). The g5Rp-InsP(6) complex structure showed that the InsP(6) molecules occupy the same regions that primarily mediate g5Rp-RNA interaction, elucidating the roles of InsP(6) in the regulation of the viral decapping activity of g5Rp in mRNA degradation. Collectively, these results provide the structural basis of interaction between RNA and g5Rp and highlight the inhibitory mechanism of InsP(6) on mRNA decapping by g5Rp. IMPORTANCE ASF is a highly contagious hemorrhagic viral disease in domestic pigs which causes high mortality. Currently, there are still no effective vaccines or specific drugs available against this particular virus. The protein g5Rp is the only viral mRNA-decapping enzyme, playing an essential role in the machinery assembly of mRNA regulation and translation initiation. In this study, we solved the crystal structures of g5Rp dimer and complex with InsP(6). Structure-based mutagenesis studies revealed critical residues involved in a candidate RNA binding region, which also play pivotal roles in complex with InsP(6). Notably, InsP(6) can inhibit g5Rp activity by competitively blocking the binding of substrate mRNA to the enzyme. Our structure-function studies provide the basis for potential anti-ASFV inhibitor designs targeting the critical enzyme.

the sole member of the Asfarviridae, a family of African swine fever-like viruses that are relatively independent of the host cell transcriptional machinery for viral replication (3, 4) . The ASFV infection of domestic swine can result in various disease forms, ranging from highly lethal to subclinical depending on the contributing viral and host factors (5) . Since 2018, ASFV has spread into China and led to a high mortality rate in domestic pigs (6, 7) . Currently, there are still no effective vaccines or specific drugs available against this particular virus (8, 9) .

During an ASFV infection, protein synthesis in the host cell is inhibited as a result of a massive degradation of host cellular mRNAs in the cytoplasm of infected cells (10, 11) . As part of its strategy to inhibit host cellular translation and promote viral protein synthesis instead, the virus targets the mRNAs of the host cell using specific enzymes (12) . Hydrolysis of the 59 cap structure (m 7 GpppN) on eukaryotic mRNAs, a process known as decapping, is considered to be a crucial and highly regulated step in the degradation of mRNA (13) . Some viruses including ASFV and vaccinia virus (VACV) can harbor decapping enzymes for control of viral and cellular gene expression (14) . Two poxvirus Nudix hydrolases, D9 and D10, have been confirmed with intrinsic mRNA-decapping activity, although the two decapping enzymes appear to have some differences in substrate recognition (15, 16) .

Nudix hydrolases (nucleoside diphosphate-linked moiety X) are widely present in bacteria, archaea, and eukarya, where they belong to a superfamily of hydrolytic enzymes that catalyze the cleavage of nucleoside diphosphates and the decapping of the 59 cap of mRNAs, the latter of which plays a pivotal role in mRNA metabolism (17, 18) . Mammalian cells have about 30 different genes with Nudix motifs, including Dcp2, Nudt16, and NUDT3/DIPP1, which cleaves mRNA caps in mRNA degradation by the 59-39 decay pathway in vivo (19) (20) (21) . The mRNA-decapping enzyme g5R protein (g5Rp), which is the only Nudix hydrolase in ASFV, shares sequence similarity to the mRNAdecapping enzymes Dcp2 in Schizosaccharomyces pombe and D9 or D10 in VACV (22) (23) (24) . However, g5Rp and its Nudix homologs D9 and D10 exhibit higher hydrolytic activity toward diphosphoinositol polyphosphates and dinucleotide polyphosphates than toward cap analogs (25, 26) . Similar to Dcp2, these Nudix hydrolases cleave the mRNA cap attached to an RNA moiety, predicating that RNA binding is crucial for performing its mRNA-decapping activity (16) . Recently, structural study has confirmed that the Nudix protein CFI m 25 has a sequence-specific RNA binding capability (27) . The requirement of RNA binding for the majority of the Nudix decapping enzymes suggest that the members of the Nudix family also belong to RNA binding proteins.

The viral mRNA-decapping enzyme g5Rp is expressed in the endoplasmic reticulum from the early stage of ASFV infection and accumulates throughout the infection process, playing an essential role in the machinery assembly of mRNA regulation and translation initiation (23) . Like other members of the Nudix family, g5Rp has a broader range of nucleotide substrate specificity, including that for a variety of guanine and adenine nucleotides and dinucleotide polyphosphates (25) . Generally, g5Rp has two distinct enzymatic activities in vitro (viz., diphosphoinositol polyphosphate hydrolase activity and mRNA-decapping activity), implying that it plays roles in viral membrane morphogenesis and mRNA regulation during viral infections (28) . In light of these biochemical observations, the elucidation of the structure of g5Rp is of fundamental importance for our understanding of the molecular mechanisms through which it degrades cellular RNAs and regulates viral gene expression.

Here, we report the crystal structure of g5Rp and its complex structure with InsP 6 . Combined with biochemical experiments, the dimeric form of g5Rp and three RNA binding surfaces on each protomer are critical to substrate RNA binding of g5Rp. The g5Rp-InsP 6 complex structure shows that two of the RNA binding surfaces are occupied by InsP 6 , indicating that InsP 6 may play a role in its ability to inhibit g5Rp-RNA binding activity. Meanwhile, we evaluate the inhibitor effect of InsP 6 on the mRNA-decapping enzyme activity of g5Rp. Therefore, we proposed that such inhibition could be caused by the competition of InsP 6 with substrate mRNA for binding to g5Rp. Furthermore, we show in detail how InsP 6 inhibits g5Rp activity by occupying the RNA binding interfaces on g5Rp, thereby competitively blocking the binding of substrate mRNA to the enzyme. These results suggest InsP 6 or its structural analogs may be involved in the manipulation of the mRNAdecapping process during viral infections and provide an essential structural basis for the development of ASFV chemotherapies in the future.

Characterization of recombinant ASFV g5Rp. Recombinant wild-type (WT) ASFV g5Rp (residues 1 to 250) was expressed in Escherichia coli with an N-terminal His 6 tag. The purified g5Rp was eluted from a Superdex 200 column (GE Healthcare) with a major elution volume of 15.6 mL, indicating an approximate molecular weight of 32.1 kDa (Fig. 1A) . The fractions were further analyzed by sodium dodecyl sulfate-polyacrylamide gel (Fig. 1B) . A cross-linking assay confirmed that g5Rp exists as a stable homodimer in solution (Fig. 1C) .

We first characterized the nucleic acid binding ability of g5Rp with different lengths of single-stranded RNA (12-mer and 26-mer ssRNA). Electrophoretic mobility shift assay (EMSA) results demonstrated that g5Rp binds ssRNA (0.25 mM) at the lowest concentration of 0.5 mM (Fig. 1D and E) . Furthermore, we measured the binding affinity of wild-type (WT) g5Rp for ssRNA by using surface plasmon resonance (SPR) (Fig. 1G and H) . The enzyme exhibited a stronger binding affinity to ssRNAs with the following equilibrium dissociation constants: 12-mer K D = 164.0 nM and 26-mer K D = 44.8 nM. The kinetic analysis of the binding experiments is shown in Table 1 . These results indicate that g5Rp possesses a higher affinity with long ssRNA. Next, we reevaluated the decapping activity of recombinant g5Rp by incubating the protein with a 32 P-cap-labeled RNA substrate in a reaction. The products of the reaction were resolved by polyethyleneimine (PEI)-cellulose thin-layer chromatography (TLC) and detected by autoradiography (23) . As shown in Fig. 1F , the recombinant g5Rp in the decapping reaction released 7-methylguanosine cap (m 7 GDP) product efficiently. In contrast, the 32 P-cap-labeled RNA substrate as control remained at the origin of the plate. These results suggest that the recombinant g5Rp possesses efficient mRNA-decapping enzyme activity.

Overview of the ASFV g5Rp structure. To investigate structural insights into the catalytic mechanism of g5Rp, we determined its dimeric structure by single-wavelength anomalous diffraction (SAD) phases using selenomethionine (SeMet)-labeled protein. As shown in Fig. 2A , the g5Rp dimer is composed of two protomers that each adopt a "boxing glove" shape with a distinct helical domain and Nudix domain (Fig.  2D ). The helical domain (residues 36 to 124) forms a globin-fold-like feature composed of six a-helices (a1 to a6) that connects to the Nudix domain by two hinge linkers (linker I, residues 32 to 35; linker II, residues 119 to 139). The Nudix domain (residues 1 to 35 and 125 to 250) consists of a central curved b-sheet (b1, b2, b3, b4) surrounded by five a-helices (a7 to a11) and several loops, thereby forming a classic a-b-a sandwich structure. Linker II splits the top of the b-sheet to connect a6 and a7 (Fig. 2E) . The Nudix motif located in the center of the Nudix domain is highly conserved and comprises the loop-helix-loop architecture that contains the Nudix signature sequence extending from residues 132 GKPKEDESDLTCAIREFEEETGI 154 in g5Rp (Fig. 2F ). The sequence of the g5Rp Nudix motif matches the classic pattern of the Nudix motif in the Nudix hydrolase superfamily, that is, GX 5 EX 7 REUXEEXGU, where X is any residue and U is Ile, Leu, or Val (29, 30) . Using the Dali server (31), we compared the structure of g5Rp with that of other proteins in the Protein Data Bank (PDB), whereupon 46 structures were found to be likely homologous to the enzyme, with Z-scores in the range of 8 to 20 (data not shown). However, all the listed protein structures shared high architectural similarity only with the Nudix domain located in the C terminus of g5Rp. Therefore, a search on the Dali server was carried out for the helical domain alone, whereupon no homologous structure with a Z-score above 4 was found, suggesting that the helical domain of g5Rp adopts a novel fold. Compared with the structures of Dcp2 in a number of different conformations, g5Rp shows a unique globin-fold-like domain ( Fig. 2B and C) .

A previous study showed that the helical domain of g5Rp is the major mediator of RNA interaction (28) . However, the positively charged surface of the g5Rp structure overlaps both the helical domain and the Nudix domain that may exhibit RNA binding activity (Fig. 3A) . We proposed that both positively charged regions could contribute to g5Rp-RNA interaction. To test the hypothesis, we measured the binding of the truncation variants g5RpDC (helical domain, residues 36 to 124) and g5RpDN (Nudix domain, connecting residues 1 to 35 and 125 to 250 directly) to ssRNAs (12-mer and 26-mer), respectively. Our EMSA results showed that both the helical domain (g5RpDC) and Nudix domain (g5RpDN) of g5Rp are involved in ssRNA interaction (Fig. 3B ). The helical domain exhibited K D values of 39.0 and 50.7 nM for the surface-immobilized 12-and 26mer ssRNAs measured by SPR, respectively ( Fig. 3C and D) . In contrast, the K D values of wild-type g5Rp for the ssRNAs (12-mer K D = 164.0 nM, 26-mer K D = 44.8 nM) are slightly lower than that of the helical domain with ssRNAs, indicating that both full-length and truncated g5Rp associated with RNA with high affinity.

The dimeric structure of g5Rp. When recombinant g5Rp was subjected to gel filtration chromatography to estimate molecular weight, it migrated as a single population of molecules at a molecular mass consistent with a monomer. However, g5Rp dimerization was consistent with cross-linking experiments ( Fig. 1A and C). To obtain more information about the interfaces and likely biological assemblies of g5Rp, we analyzed its structure using the PDB-related interactive tool Proteins, Interfaces, Structures and Assemblies (PDBePISA) (32) . The results suggested that g5Rp forms a stable symmetric dimer in crystal packing. The dimer was composed of two protomers (A and B) positioned in an orientation similar to two boxing gloves stuck together back to back ( Fig. 2A) . The dimer interfaces were stabilized mainly by hydrophobic interactions. Furthermore, a network of hydrogen bonds conferred additional stability on the interface. One interface was composed of four a-helices (a3 and a4 from each A and B protomer) from the N terminus of each protomer. To determine the multimeric state of g5Rp in solution and to examine which of its termini is critical for its dimerization, we measured the multimerization of two g5Rp truncation variants (g5RpDN and g5RpDC) using cross- linking experiments. The results showed that the wild type, N terminus, and C terminus of g5Rp all formed a dimeric conformation in solution ( Fig. 4B and C). The g5Rp mutant I84A/ I116A/L200A/I206A/F222A that prepared to dissociate the dimeric form of g5Rp was successful in altering a monomeric state, even the dimeric total buried area of 3,050 Å 2 . Wildtype g5Rp and mutants were subjected to gel filtration chromatography, showing that the mutant I84A/I116A/L200A/I206A/F222A has a larger retention volume, corresponding to a lower molecular weight (Fig. 5A ). The protein cross-linking experiment showed that the dimeric conformation was significantly reduced in solution for the mutant (Fig. 5B ). The ssRNA binding ability of the monomeric mutant has been measured by SPR and EMSA. The monomeric mutant with analyte concentrations was passed over immobilized ssRNA. The resultant sensorgrams are shown in Fig. 5C and D, and kinetic analysis is shown in Table 1 . EMSA data are shown in Fig. 5E . Both measurements produced consistent results indicating that the g5Rp mutant I84A/I116A/L200A/I206A/F222A partially impaired the RNA binding ability. Therefore, we proposed that the dimeric g5Rp is preferred for efficient RNA binding. Meanwhile, mRNA-decapping assays showed that the decapping activity of mutant I84A/I116A/L200A/I206A/F222A dropped greatly (Fig. 5F ).

Structure of the g5Rp-InsP 6 complex. g5Rp was originally characterized through its ability to dephosphorylate 5-PP-InsP 5 (InsP 7 ) to produce InsP 6 (25). We were surprised to find a tight interaction between InsP 6 and g5Rp by microscale thermophoresis (MST) (Fig. 6A ). To gain insight into the molecular basis of the interaction, we determined the crystal structure of the g5Rp-InsP 6 complex and found that each asymmetric unit contained one g5Rp-InsP 6 complex in space group P4 1 22. PDBePISA analysis revealed that an identical dimeric conformation exists in the g5Rp-InsP 6 complex structure (Fig. 6B ). Two InsP 6 molecules were situated on the edge of the b1 strand of each g5Rp protomer through interactions with residues Gln 6 , Lys 8 , and Lys 133 (Fig. 6C ). Due to the 2-fold symmetry in the crystal, each of the g5Rp protomers shared two InsP 6 molecules (InsP 6 and InsP 6 asym ) with its neighboring g5Rp protomer in the crystal lattice. Besides the InsP 6 binding on the b1 strand located on the edge of the Nudix domain, an extra InsP 6 molecule from the neighboring molecule also interacted with g5Rp through residues Lys 94 and Lys 98 on the a5 helix in the helical domain (Fig.  6C ). In this way, each InsP 6 molecule is surrounded by four Lys residues in complex structure. The solvent-accessible surface of the InsP 6 binding region of g5Rp was calculated according to the electrostatic potential. It was apparent that both InsP 6 molecules were situated on the highly positively charged area located in the protein cleft between the helical domain and Nudix domain of g5Rp (Fig. 7A ). The local conformational changes of g5Rp in the complex structure induced by its interaction with InsP 6 are illustrated in Fig. 7B . In the complex structure, the b1, b3, and b5 strands located in the Nudix domain had moved closer to the helical domain, and a2 was pushed away from the InsP 6 binding sites. These changes rendered the g5Rp conformation more stable in the complex.

To assess their relative importance in g5Rp-InsP 6 interaction, amino acid residues involved in InsP 6 binding pockets were replaced by single point mutation (Q6A, K8A, K94A, K98A, K133A). Each mutant was tested for its binding affinity for InsP 6 by MST. Figure 6D showed that mutants resulted in a notable decrease in g5Rp-InsP 6 interaction, and furthermore, the quintuple mutant Q6A/K8A/K94A/K98A/K133A totally lost the binding ability with InsP 6 . Taken together, the mutagenesis work indicates that positively charged residues Lys8, Lys94, Lys98, and Lys133 form a cluster to mediate the g5Rp-InsP 6 interaction.

Analysis of residues involved in g5Rp-RNA interfaces. To characterize RNA binding surface on g5Rp, we analyzed the electrostatic potential at the surface of g5Rp, which indicated that three highly positively charged areas (areas I to III) may play roles in g5Rp-RNA interaction (Fig. 3A) . Area I is located on the helical domain, containing residues Lys94, Lys95, Lys98, Arg100, and Lys101 located on helix a5. Area II is composed of residues Lys8, Lys131, Lys133, Lys135, Arg146, Lys175, Lys179, and His180 mostly located on the b1 and b3 strands, which are close to the Nudix motif; area III is located at the very end of the C terminus of g5Rp, comprising residues Arg 221 , Lys 225 , Arg 226 , Lys 243 , and Lys 247 on helices a10 and a11 (Fig. 8A) . To identify the mRNA binding surfaces on g5Rp further, the residues mentioned above located in three positively charged areas of g5Rp were mutated, respectively. The EMSA pattern showed that some mutants reduce the RNA binding affinity of g5Rp. Specifically, residues Lys 8 , The binding ability between various g5Rp mutants and InsP 6 was measured by MST. The dissociation constants between g5Rp mutants and InsP 6 were calculated from three independent replicates (shown as mean 6 standard deviation). InsP 6 is shown as a stick model. (Fig. 8B and C) , implying that the g5Rp-RNA interaction interfaces are mainly located at areas I, II, and III. These results also agree with our hypothesis that residues Lys 8 , Lys 94 , Lys 98 , and Lys 133 of g5Rp are involved in both RNA and InsP 6 interaction. We further explored whether these key residues were responsible for cap cleavage in a manner dependent on the RNA moiety interaction. Mutant proteins including Q6E/K8E, K94E, K95E, K98E, K175E, R221D, and K243E were expressed and purified. Consistent with our previous data, incubation of the 32 P-cap-labeled RNA substrate with wild-type g5Rp resulted in cap cleavage, as observed by m 7 GDP release. When equivalent amounts of the mutants of g5Rp were included in the decapping reaction, the amount of m 7 GDP released was reduced variously in each lane. Mutant K95E decreases the decapping activity almost 50% ( Fig. 8D and E) , indicating that these residues of g5Rp play a pivotal role in mRNA decapping by interacting with substrate mRNA.

Residues Gly 132 , Lys 133 , and Glu 147 in the Nudix motif impact the decapping activity. The Nudix motif of hydrolases contains crucial residues involved in catalytic activity. However, the residues in the catalytic pocket of g5Rp are still elusive from the viewpoint of structure. To elucidate the function of the key residues in g5Rp, three substrate binding structures from the Nudix superfamily were selected to identify homologous domains with high similarity at the potential catalytic pockets (Fig. 9A to C) , as shown in Table 2 , viz., Ap4A hydrolase (Aquifer aeolicus, PDB accession no. 3I7V) (33), Nudix hydrolase DR1025 (Deinococcus radiodurans, PDB accession no. 1SZ3), and MTH1 (Mus musculus, PDB accession no. 5MZE) (34, 35) (all belonging to the Nudix superfamily). Superposition of the C terminus of g5Rp with that of MTH1, Ap4A hydrolase, and Nudix hydrolase DR1025 resulted in Ca backbone root mean square deviation values of 0.50, 3.08, and 5.6 Å, respectively, despite the low sequence identities among these proteins (Fig. 9D) . Therefore, the potential substrate binding site of g5Rp was proposed on the basis of the superpositions of these substrate binding protein structures of the Nudix superfamily. Residues Gly 132 , Lys 133 , and Glu 147 located on the Nudix motif of g5Rp may be responsible for cap cleavage.

To investigate the potential roles of these key residues located on the Nudix motif in the decapping activity, we replaced g5Rp residues G132, K133, and E147 from the Nudix motif ( Fig. 9D and Fig. 10A) with Ala, Glu, and Gln, respectively. As expected, the replacement of the residue K133 with glutamate resulted in a 30% decrease in the decapping activity. And the replacement of the residues G132 and E147 by alanine and glutamine, respectively, inactivated the decapping function of g5Rp completely ( Fig.  10B and C) . No m 7 GDP was observed when the two mutants of g5Rp were included in the decapping reaction, validating that the decapping activity was dependent on these two key residues located in the Nudix hydrolase motif. Interestingly, EMSA results showed that mutant K133E reduces g5Rp's binding affinity to RNA, which suggests that the loop region of the Nudix motif takes part in substrate mRNA binding (Fig. 10D and E).

InsP6 inhibits the decapping activity by disrupting g5Rp-mRNA interaction. The finding that residues located on mRNA binding regions of g5Rp are also playing pivotal roles in g5Rp-InsP 6 interaction suggests that InsP 6 may inhibit the g5Rp decapping activity through preventing g5Rp from binding to its mRNA substrate (Fig. 11A) . This prediction was confirmed by decapping and EMSAs using recombinant g5Rp, InsP 6 , and RNA substrates in vitro. Increasing amounts of InsP 6 were added to the decapping reactions to analyze its effect on RNA decapping by g5Rp. As shown in Fig. 11B , the addition of InsP 6 significantly affected g5Rp cleavage, suggesting that InsP 6 can inhibit the decapping activity of g5Rp in vitro. To investigate if this inhibitory mechanism of InsP 6 on g5Rp is due to inositol phosphate competitively inhibiting mRNA binding to the g5Rp, we further measured the competition of InsP 6 with nucleic acids for the binding to g5Rp by using EMSA. As expected, the amount of free single-stranded nucleic acids increased with an increasing concentration of InsP 6 , demonstrating that InsP 6 interrupts the g5Rp-mRNA interaction through directly binding to g5Rp (Fig. 11C and D) . In addition, all residues involved in InsP 6 interaction in g5Rp were mutated into alanine at the same time. The quintuple mutant (Q6A/K8A/K94A/K98A/K133A) of g5Rp lost most of its ability to bind with both InsP 6 and RNA ( Fig. 6D and see Fig. 13A ). These mutations also significantly affected the catalytic ability of g5Rp in vitro (see Fig. 13C and D), suggesting that InsP 6 inhibits the mRNA-decapping activity of g5Rp through competing for the substrate mRNA binding surface in g5Rp.

Transient expression of g5Rp decreases levels of mRNA substrates in 293T cells. The above data provide strong in vitro evidence for g5Rp-mRNA interaction being a critical step for the decapping enzyme process. To determine whether changes in g5Rp-mRNA interaction were directly related to the stability of cellular mRNAs in vivo, representative cellular mRNA (eIF4E, eIF4EA, and TP53) levels were tested by quantitative real-time PCR (RT-qPCR) in cells. In 293T cells, the Flag-tagged g5Rp and the g5Rp mutants (K8E, K94E, K95E, K98E, G132A, K133E, E147Q, K175E, R221D, and K243E) were overexpressed, respectively. As shown in Fig. 12A , the g5Rp-WT and mutant proteins were detected by Western blotting. The mRNA levels of target genes (eIF4E, eIF4EA, and TP53) were decreased in 293T cells overexpressing g5Rp-WT. There were no obvious changes in mRNA levels in the catalytic destructive mutants Q132A and E147Q. The overexpression in cells of truncated version g5Rp-DN and mutants K95E and R221D, mutants which significantly lost the RNA binding ability in vitro, had no effect on the mRNA levels of target genes in 293T cells. Mutants K8E and K133E, which had reduced RNA binding in vitro, had various degrees of increase compared with the mRNA levels of the g5Rp-WT group. However, the changes in mRNA levels of target genes observed in mutants K94E, K98E, K175E, and K243E did not have statistical differences from those in g5Rp-WT (Fig. 12B to D) . Taken together, these results suggest that key residues K8, K95, K133, and R221, playing pivotal roles in g5Rp-RNA interaction, are also important to the g5Rp-related cellular RNA degradation in vivo. 

Journal of Virology

Given that an ASFV outbreak in China would potentially devastate the world's largest pork producer, significant efforts have been made to determine the structures and functions of essential viral proteins that may be used as targets for new anti-ASFV drugs. Several structures of ASFV-encoded enzymes and associated proteins that are involved in viral transcription and replication have been reported, including AP endonuclease (36) , the histone-like protein pA104R (37), pS273R protease (38) , DNA ligase (39) , and dUTPase (40, 41) . However, the structures and functions of some critical ASFV proteins remain elusive, including those of g5Rp, a decapping enzyme that is crucial for viral infection (23) . Our structures of g5Rp alone and in complex with InsP 6 provide the molecular basis for g5Rp substrate recognition and reveal that inositol phosphate was involved in the regulation of cellular mRNA degradation through direct interaction with the ASFV decapping enzyme g5Rp. Three potential RNA binding regions are identified, including a novel folding domain located on the helical domain of g5Rp and the Nudix motif on its C terminus. More importantly, identification of the major nucleic acid binding surfaces as well as the binding pocket of InsP 6 on g5Rp provides important structural information and a novel strategy for future anti-ASFV drug design.

To explore the nucleic acid binding properties of g5Rp, we conducted a series of nucleic acid binding experiments. Results indicated that an intact dimeric interface is efficient for g5Rp-RNA interaction. Meanwhile, the helical domain and Nudix domain of g5Rp are both involved in ssRNA interaction. Our EMSA and SPR measurements show that the helical domain of g5Rp can bind with ssRNA with equally high affinity as the full-length protein. Six a-helices form a globin-fold-like helical domain, which is different from the traditional RNA binding domain that prefers to adopt the alpha/beta topologies (42) (43) (44) (45) . According to the g5Rp structure, the surface electrostatic potential characteristics of the N terminus present a highly positively charged area on helix a5. The single point mutations of positively charged residues in the N terminus significantly reduced the nucleic acid binding activity of g5Rp with ssRNA ( Fig. 8B and C) . Furthermore, there are two positive areas located on the C terminus of g5Rp, including the Nudix motif, participating in the substrate RNA interaction. We mutated the two positively charged regions (K8A/K131A/K133A/K135A and R221A/K225A/R226A/ K243A/R247A) located in the Nudix domain; the EMSA data showed that the nucleic acid binding ability of these two mutants was significantly reduced (Fig. 13B) , and the Fig. 13E and F data showed that the substantial decline in capacity of K8A/K131A/ K133A/K135A removed the m 7 Gppp RNA cap. These results predicated that the Nudix motif of g5Rp possesses substrate selectivity at the step of mRNA binding. Previously, studies revealed that the Nudix motif (residues 132 to 154) is an essential component of the a-b-a sandwich in the catalytic center of g5Rp. Several of the conserved catalytic amino acids and glutamate residues (E 147 , E 149 , E 150 , and E 151 ) located on the a-helix of the Nudix motif of g5Rp have been found to be important for the activity of Nudix hydrolases (23, 28) . However, the function of the loop region within the Nudix motif is exclusive, leading us to predicate that the loop region may contribute to binding with mRNA. Therefore, we mutated several residues in this loop region, including the mutations K133E and G132A, and examined the effects on the protein's interaction with single-stranded nucleic acids. It is interesting to find that substitutes for the conserved residues K 133 and G 132 are highly sensitive to g5Rp-RNA interaction. Compared with glutamate residues located on the a-helix of the Nudix motif of g5Rp involved in mRNA cap structure interaction, residues K 133 and G 132 are important for binding with the RNA moiety on the substrate. In this way, we provided a demonstration that the short loop in the Nudix motif is required for g5Rp-RNA interaction. Including the Nudix motif, three positively charged patches on the g5Rp surface were mapped as mRNA binding regions. Furthermore, we also investigated the importance of the residues involved in mRNA interaction in g5Rp-mediated decapping. The g5Rp mutants K8E, K94E, K95E, K98E, K175E, and R221D showed a strong reduction in decapping activity, demonstrating the importance of the mRNA binding residues for catalysis. The dimeric form of g5Rp is also important to the decapping activity. We constructed mutant g5Rp-I84A/I116A/L200A/I206A/F222A in which the dimerization surface was destroyed. mRNA-decapping assays showed that the decapping activity of mutant g5Rp-I84A/I116A/L200A/I206A/F222A decreased drastically (Fig. 5F ). It will be of profound interest to elucidate the structural basis of the enzymatic activity of g5Rp by solving the structure of g5Rp in complex with mRNA in the future.

The other important finding in this study was that InsP 6 is able to inhibit the decapping activity of g5Rp. As we know, InsP 6 is widespread in cells with diverse biological functions (46) (47) (48) (49) . Here, we found that InsP 6 competes with mRNA substrates for binding to g5Rp and inhibits its decapping activity. A previous study reported that g5Rp is a diphosphoinositol polyphosphate phosphohydrolase (DIPP), which preferentially removes the 5-b-phosphate from InsP 7 to produce InsP 6 with unclear functional significance (25) . Later, Parrish and colleagues identified that g5Rp can hydrolyze the mRNA cap when tethered to an RNA moiety in vitro (23) . Our results show that InsP 6 as the product of g5Rp playing the role of DIPP can directly inhibit the mRNA-decapping activity of g5Rp. To illustrate the structural basis of the inhibitory mechanism of InsP 6 for the decapping activity of g5Rp, we solved the structure of the complex of g5Rp with InsP 6 and also the enzyme-product complex in the Nudix superfamily. To our surprise, InsP 6 is located on the mRNA binding region instead of in the catalytic center of the g5Rp. Furthermore, we superposed the catalytic domain of g5Rp-InsP 6 complex with the structures of human DIPP1 in complex with the substrate InsP 7 (50, 51) . The visualized result showed that the substrate InsP 7 is located in the catalytic center of DIPP1, unlike InsP 6 , which sits on the edge of the catalytic domain of g5Rp (Fig. 14A) . Therefore, the structure of the g5Rp-InsP 6 complex may represent an intermediate in the release of the product of the enzymatic reaction (52) . We also noticed that InsP 6 decreased the temperature value (B factor) around the binding sites compared with B factor in the same regions of the g5Rp wild-type structure, suggesting that the flexible loop closed to the catalytic center is locked in place by InsP 6 (Fig. 14B ). InsP 6 itself was refined with a correspondingly high B factor that exceeded the average B factor of the protein in complex. Considering that the g5Rp-InsP 6 interaction has a dissociation constant (K d ) in the 22.5 mM range, the ligand achieves only a reasonable occupancy of 70% (53) . To avoid an instance of overenthusiastic interpretation of ligand density, we tested the InsP 6 binding site by using single point mutations. Residues involved in the InsP 6 binding surface of g5Rp replaced by alanine (Q6A/K8A/K94A/K98A/K133A) reduced its InsP 6 binding capacity and RNA interaction, indicating the destructive InsP 6 binding site has the capability to abolish the substrate RNA binding ability of g5Rp (Fig. 13A, C, and D) .

Our study raises the possibility that g5Rp hydrolyzes InsP 7 to upregulate the level of InsP 6 , which is a key regulator of g5Rp-mediated mRNA decapping during ASFV infection in vivo (54) . Very recently, Sahu and colleagues reported that InsP 7 regulates the NUDT3-mediated mRNA decapping and also observed the phenomenon that InsP 6 inhibits mRNA decapping by NUDT3 (54) . There are emerging signs that the functions of InsP 6 are associated with mRNA transportation and degradation in ASFV-infected cells. Further studies on the function of InsP 6 and the regulation mechanism in the inositol-based cell signaling family during viral infection are required.

Cell culture. The human 293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (HyClone) supplemented with 10% fetal bovine serum (FBS) (Gibco), 100 U/mL penicillin, and 100 mg/ mL streptomycin at 37°C under a humidified 5% CO 2 atmosphere (Thermo).

Plasmid construction, protein expression, and purification. The gene encoding ASFV g5Rp (D250R) was synthesized and subcloned into pSMART-1 and pcDNA3.1, respectively. The amino acid sequence of Table 3 . The recombinant plasmids were confirmed by sequencing (Sangon Biotech, China) before being introduced into E. coli BL21(DE3) (Invitrogen, USA) or human 293T cells. The bacterial cells were cultured in Luria broth medium at 35°C until the optical density at 600 nm reached 0.6 to 0.8. Protein expression was then induced by the addition of isopropyl-b-D-1-thiogalactopyranoside for 16 h at 16°C. The g5Rp molecules were purified by Ni-nitrilotriacetic acid (NTA) (Qiagen, Germany) affinity chromatography, followed by heparin affinity chromatography (GE Healthcare, USA). The peak fractions containing the target proteins were pooled, concentrated to 1 mL, and finally loaded onto a Superdex 75 column (GE Healthcare, USA) for further purification and characterization. Selenomethionine-labeled g5Rp (SeMet-g5Rp) was then prepared using a previously described protocol (55) . The purity of all proteins was above 95% on the SDS-PAGE gel.

Protein crystallization and optimization. The prepared SeMet-g5Rp was concentrated to 12 mg/mL for the crystallization trials. The crystals were grown using the hanging-drop vapor diffusion method at 16°C in a reservoir solution containing 0.1 M sodium citrate tribasic dihydrate (pH 5.8), 0.54 M magnesium formate dihydrate, and 10% (vol/vol) 1,2-butanediol as an additive reagent. The g5Rp-InsP 6 complexes were prepared by mixing g5Rp with InsP 6 at a stoichiometric ratio of 1:3. Then, using the hanging-drop vapor diffusion method, crystals of the complexes were grown from 1 M imidazole (pH 7.0) at 16°C. All crystals were transferred into solutions containing 20% (vol/vol) glycerol prior to being frozen and stored in liquid nitrogen.

Data collection, processing, and structure determination. The single-wavelength anomalous dispersion (SAD) data were collected using synchrotron radiation of an 0.98-Å wavelength under cryogenic conditions (100 K) at the BL18U1 beamline, Shanghai Synchrotron Radiation Facility. All diffraction data sets including g5Rp-WT and the complex with InsP 6 were indexed, integrated, and scaled by using the HKL-2000 package (56) . The selenium atoms in the asymmetric unit of SeMet-g5Rp were located and PyMOL. The Ca B factors are depicted on the structure in dark blue (lowest B factor) through to red (highest B factor), with the radius of the ribbon increasing from low to high B factor. The lower B factor is observed in the overall structure of g5Rp-InsP 6 , with the InsP 6 binding sites also displaying lowerthan-average B factors, consistent with the InsP 6 contacts stabilizing this region of g5Rp relative to the overall structure. refined, and the SAD data phases were calculated and substantially improved through solvent flattening with the PHENIX program (57) . A model was built manually into the modified experimental electron density using the model-building tool Coot (58) and then further refined in PHENIX. The model geometry was verified using the program MolProbity (59) . Molecular replacement was used to solve the structure of the g5Rp-InsP 6 complex, using Phaser in the CCP4 program suite with an initial search model of SeMet-g5Rp (60). Structural figures were drawn using PyMOL (DeLano Scientific). The data collection and refinement statistics are shown in Table 1 . 

Surface plasmon resonance analysis. The SPR analyses were carried out using the Biacore 8K system with a streptavidin-coated (SA) chip (catalog no. BR-1005-30

) ligands (given in relative units [RU]) were immobilized on the SA chip. Given the apparent variation in RU of immobilized ligands used in different binding studies (61, 62), the average RU of immobilized 26-mer RNA and 12-mer RNA in this study are approximately 200 to 400 RU. The blank channel served as a negative control. Protein solutions with various concentrations are run across the chip at a rate of 30 mL/min and are then dissociated by running buffer

0) was added. The mixtures were loaded into capillaries and measured at ambient temperature by using 20% LED (light-emitting diode) and medium MST power in a Monolith NT.115 system (NanoTemper Technologies). The data were analyzed using NanoTemper analysis software (v.1.2.101). Electrophoretic mobility shift assay. EMSAs were performed to determine the nucleic acid affinity of the wild-type and g5Rp mutants. The single-stranded nucleic acids used were 6-carboxyfluoresceinlabeled ssRNA (59-GCUUUGAUUUCGUGCAUCUAUGGAGC-39) and 6-carboxyfluorescein-labeled ssRNA

All samples were incubated on ice for 30 min and then electrophoresed on 4.0% native PAGE gels for 45 min at a voltage of 100 V. The results were determined with a Bio-Rad ChemiDoc MP imaging system

1 mM DTT, and 2 mM MnCl 2 . Generally, 100 ng of wild type or g5Rp mutants was used in each 5-mL reaction mixture. To determine the correlation between InsP 6 and the decapping ability of g5Rp, we also tested the decapping ability of the enzyme mixed with different concentrations of InsP 6 (0.03 mM, 0.06 mM, 0.12 mM, 0.25 mM, 0.5 mM, 1 mM, 2 mM, and 4 mM). The decapping reaction was carried out at 37°C for 60 min and then stopped with 25 mM EDTA. The products of the reaction were separated by PEI-cellulose thin-layer chromatography developed in 0.45 M (NH 4 ) 2 SO 4 and detected with autoradiography. Cross-linking assay. A cross-linking assay was carried out by incubating 1 mg/mL of wild-type g5Rp or each g5Rp truncation variant and mutant I84A/I116A/L200A/I206A/F222A in a buffer containing 20 mM HEPES

After the membranes were washed three times for 10 min with Tris-buffered saline with Tween 20 (TBST) and incubated with secondary antibodies for 2 h at room temperature, they were washed three times again in TBST. Then, the blots were detected with enhanced chemiluminescence reagents (Millipore) using MiniChemi (Sagecreation, China) RNA extraction and quantitative real-time PCR. Briefly, 100,000 cells were seeded in a 12-well plate (Nest) and cultured for 24 h before transfection with plasmid (1 mg/well). Total RNA was extracted using RNAiso reagent (TaKaRa), and 500 ng of total RNA was reverse transcribed using the PrimeScript REFERENCES 1

Architecture of African swine fever virus and implications for viral assembly

The African swine fever virus transcriptome

African swine fever virus transcription

The genetics of life and death: virus-host interactions underpinning resistance to African swine fever, a viral hemorrhagic disease

Replication and virulence in pigs of the first African swine fever virus isolated in China

African swine fever epidemiology and control

Current status and evolving approaches to African swine fever vaccine development

Antiviral agents against African swine fever virus

African swine fever virus replication and genomics

African swine fever virus controls the host transcription and cellular machinery of protein synthesis

Two putative African swine fever virus helicases similar to yeast 'DEAH' pre-mRNA processing proteins and vaccinia virus ATPases D11L and D6R

mRNA stability and control of cell proliferation

Decapitation: poxvirus makes RNA lose its head

Characterization of a second vaccinia virus mRNAdecapping enzyme conserved in poxviruses

Vaccinia virus D10 protein has mRNA decapping activity, providing a mechanism for control of host and viral gene expression

New insights into decapping enzymes and selective mRNA decay

Principles of RNA and nucleotide discrimination by the RNA processing enzyme RppH

Multiple Nudix family proteins possess mRNA decapping activity

The hDcp2 protein is a mammalian mRNA decapping enzyme

Nudt3 is an mRNA decapping enzyme that modulates cell migration

Structure of the active form of Dcp1-Dcp2 decapping enzyme bound to m(7)GDP and its Edc3 activator

The African swine fever virus g5R protein possesses mRNA decapping activity

The D10 decapping enzyme of vaccinia virus contributes to decay of cellular and viral mRNAs and to virulence in mice

The g5R (D250) gene of African swine fever virus encodes a Nudix hydrolase that preferentially degrades diphosphoinositol polyphosphates

Insights into the molecular determinants involved in cap recognition by the vaccinia virus D10

Structural basis of UGUA recognition by the Nudix protein CFI(m)25 and implications for a regulatory role in mRNA 39 processing

Characterization of the African swine fever virus decapping enzyme during infection

The Nudix hydrolase superfamily

A cryptic activity in the Nudix hydrolase superfamily

DALI and the persistence of protein shape

Inference of macromolecular assemblies from crystalline state

Free and ATP-bound structures of Ap4A hydrolase from Aquifex aeolicus V5

Structural studies of the Nudix hydrolase DR1025 from Deinococcus radiodurans and its ligand complexes

Crystal structures and inhibitor interactions of mouse and dog MTH1 reveal species-specific differences in affinity

A unique DNA-binding mode of African swine fever virus AP endonuclease

The structural basis of African swine fever virus pA104R binding to DNA and its inhibition by stilbene derivatives

Crystal structure of African swine fever virus pS273R protease and implications for inhibitor design

Structure of the error-prone DNA ligase of African swine fever virus identifies critical active site residues

Structural insight into African swine fever virus dUTPase reveals a novel folding pattern in the dUTPase family

Crystal structure of African swine fever virus dUTPase reveals a potential drug target

The crystal structure of the Methanocaldococcus jannaschii multifunctional L7Ae RNA-binding protein reveals an induced-fit interaction with the box C/D RNAs

Structure and RNA binding of the third KH domain of poly(C)-binding protein 1

Crystal structure of activated HutP; an RNA binding protein that regulates transcription of the hut operon in Bacillus subtilis

The archaeal sRNA binding protein L7Ae has a 3D structure very similar to that of its eukaryal counterpart while having a broader RNA-binding specificity

Inositol phosphates and cell signaling: new views of InsP5 and InsP6

Inositolhexakisphosphate (InsP6): an antagonist of fibroblast growth factor receptor binding and activity

Nucleus-associated phosphorylation of Ins(1,4,5)P3 to InsP6 in Dictyostelium

Increased levels of inositol hexakisphosphate (InsP6) protect HEK293 cells from tumor necrosis factor (alpha)-and Fas-induced apoptosis

Vip1 is a kinase and pyrophosphatase switch that regulates inositol diphosphate signaling

Crystal structure of human diphosphoinositol phosphatase 1

Structure of a low-population intermediate state in the release of an enzyme product

Validation of protein-ligand crystal structure models: small molecule and peptide ligands

InsP7 is a small-molecule regulator of NUDT3-mediated mRNA decapping and processing-body dynamics

Selenomethionine and selenocysteine double labeling strategy for crystallographic phasing

Processing of X-ray diffraction data collected in oscillation mode

PHENIX: a comprehensive Python-based system for macromolecular structure solution

Coot: model-building tools for molecular graphics

MolProbity: all-atom structure validation for macromolecular crystallography

Overview of the CCP4 suite and current developments

Mass spectroscopic characterization of the coronavirus infectious bronchitis virus nucleoprotein and elucidation of the role of phosphorylation in RNA binding by using surface plasmon resonance

Kinetic studies of RNA-protein interactions using surface plasmon resonance

We thank the staff at the State Key Laboratory of Biotherapy, Sichuan University, who assisted with our research work during the period of the COVID-19 epidemic. The X-ray diffraction experiments were carried out at the Shanghai Synchrotron Radiation Facility (SSRF) at BL18U1. We also thank the beamline staff for their technical help during the data collection. We appreciate helpful advice from and discussion with Xiaoyu Xue from Texas State University. We declare that we have no conflicts of interest surrounding the contents of this article.