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

Significance

The rapid spread of ASFV in China and other Asian countries has made the development of effective drug and vaccine target proteins extremely urgent. pA104R, an ASFV nucleoid-associated protein, shares sequence similarity with HU/IHF family members and is essentially involved in viral genome packing and replication. Here, we uncover the molecular basis of pA104R-DNA interaction by determining the structures of both apo-pA104R and pA104R-DNA complex. pA104R exhibits a conserved architecture as HUs/IHFs, while presenting a unique DNA binding pattern as its β-ribbon arm residues insert into the major groove of DNA. More importantly, our demonstration that stilbene derivatives SD1 and SD4 can disrupt the pA104R-DNA binding and inhibit ASFV replication in swine macrophages casts a new light on ASFV control.

Abstract

African swine fever virus (ASFV) is a highly contagious nucleocytoplasmic large DNA virus (NCLDV) that causes nearly 100% mortality in swine. The development of effective vaccines and drugs against this virus is urgently needed. pA104R, an ASFV-derived histone-like protein, shares sequence and functional similarity with bacterial HU/IHF family members and is essential for viral replication. Herein, we solved the crystal structures of pA104R in its apo state as well as in complex with DNA. Apo-pA104R forms a homodimer and folds into an architecture conserved in bacterial heat-unstable nucleoid proteins/integration host factors (HUs/IHFs). The pA104R-DNA complex structure, however, uncovers that pA104R has a DNA binding pattern distinct from its bacterial homologs, that is, the β-ribbon arms of pA104R stabilize DNA binding by contacting the major groove instead of the minor groove. Mutations of the basic residues at the base region of the β-strand DNA binding region (BDR), rather than those in the β-ribbon arms, completely abolished DNA binding, highlighting the major role of the BDR base in DNA binding. An overall DNA bending angle of 93.8° is observed in crystal packing of the pA104R-DNA complex structure, which is close to the DNA bending angle in the HU-DNA complex. Stilbene derivatives SD1 and SD4 were shown to disrupt the binding between pA104R and DNA and inhibit the replication of ASFV in primary porcine alveolar macrophages. Collectively, these results reveal the structural basis of pA104R binding to DNA highlighting the importance of the pA104R-DNA interaction in the ASFV replication cycle and provide inhibitor leads for ASFV chemotherapy.

Emerging and reemerging pathogens pose a great threat to the public health, both for humans and animals (1). African swine fever (ASF), an acute highly contagious viral hemorrhagic disease in domestic pigs and wild boars has a mortality rate approaching 100% and is arguably the most severe reemerging disease threat to the swine industry worldwide (2). The disease was first recognized in Kenya in the 1920s and traditionally has been confined to sub-Saharan Africa. However, it has spread to Europe, South America, and Asia, causing enormous loss to the global swine industry (3–6). To date, there are no approved vaccines or drugs for the control of the disease, and the growing threat of the disease is overwhelming. Thus, new virus control strategies are urgently needed.

The etiological agent of ASF is the African swine fever virus (ASFV), a member of the highly complicated nucleocytoplasmic large DNA virus (NCLDV) (7). ASFV consists of a DNA-containing central nucleoid, a core shell, an inner lipid envelope, an icosahedral capsid, and an outer lipid envelope (8–10). The nucleoid is a high electron dense structure of 80 nm containing the viral genome, transcriptional machinery for the synthesis and modification of early RNAs, and nucleoproteins responsible for viral genome packing and nucleoid assembly (11).

The ASFV-encoded pA104R protein is located in the nucleoid and is related to nucleoid assembly (12). This protein, also named LMW5-AR, is encoded by the A104R gene and translated into a 104 amino acid polypeptide. Sequence analysis using the Swissprot database revealed that it shares an ∼25–30% sequence identity with bacterial HUs and IHFs as well as Bacillus phage SPO1 transcription factor 1 (TF1) (13). It is expressed in the late stages of the viral infection cycle, coincident with viral DNA replication and progeny assembly and, therefore, was thought to be a DNA binding protein involved in nucleoid formation (13).

Both HUs and IHFs belong to the bacterial DNA binding protein DNABII family (14). These proteins are nucleoid-associated proteins that play an architectural role in DNA supercoiling and compaction (15). HUs are also involved in gene replication, DNA recombination and repair, cell division, and the regulation of gene transcription (16). IHFs assist CRISPR–Cas spacer acquisition in Escherichia coli (17, 18). HUs and IHFs are different in the sequence specificities of the DNA they bind: HUs bind to random DNA sequences with a K_D of 200–2500 nM, whereas IHFs bind tightly (2–20 nM) to cognate sites represented by the consensus WATCARXXXXTTR (W is A or T; X is A, T, C, or G; R is A or G) (19). Generally, IHFs are obligate heterodimers, whereas HUs can be either heterodimers or homodimers (20, 21). HU and IHF are widely distributed in different bacteria and some bacteriophages (22).

A recent report by Frouco et al. provides evidence that pA104R binds to both single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) of ∼14–16 nt/bp, and it displays DNA-supercoiling activity in the presence of ASFV topoisomerase II (pP1192R) (23). Their results also show that knockdown of pA104R reduces viral infection, suggesting that it is required for viral DNA replication and transcription (23). The importance of pA104R in the replication cycle of ASFV is further evidenced by the fact that the recombinant ASFV lacking the gene encoding pA104R cannot be isolated from even pA104R-expressing cell lines (24).

With regard to understanding the essential role of pA104R in ASFV nucleoid compaction and genome replication, it is of great interest to characterize the structural basis of how ASFV pA104R binds to its DNA ligand. In the present study, we report the crystal structure of ASFV pA104R in its apo state and in complex with DNA. More importantly, we show that two stilbene derivatives, SD1 and SD4, which were previously shown to inhibit Mycobacterium tuberculosis (Mtb) HU (25), can inhibit ASFV replication. These results provide an important basis for the development of ASFV chemotherapies.

Results

Characterization of Recombinant pA104R Protein.

Recombinant ASFV pA104R was expressed in E. coli cells with a N-terminal His₆ tag. The Ni-affinity column purified pA104R protein was eluted from a Hiload 16/60 Superdex 75 PG column (GE Healthcare) with an elution volume of ∼65 mL, indicating an approximate molecular weight of 24 kDa. The fractions were further analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS/PAGE), showing a monomeric pA104R band of ∼12 kDa (Fig. 1A). Sedimentation velocity analytical ultracentrifugation assay further confirmed that the pA104R protein exits as a homodimer (24 kDa) in solution (Fig. 1B).

Fig. 1.

Biochemical characterization of ASFV pA104R. (A) The gel filtration profiles of pA104R protein. (B) Sedimentation velocity profile of pA104R. (C and D) pA104R binds to dsDNA (C) and ssDNA (D) fragments. Serially diluted pA104R proteins were detected by electrophoretic mobility shift assay (EMSA) assays. (E and F) pA104R binds to dsDNA (E) and ssDNA (F) fragments with different lengths as indicated.

We next characterized the DNA binding ability of pA104R and the minimum length of bound DNA. To assess DNA binding, increasing concentrations of pA104R protein were incubated with 5′-FAM–labeled ssDNA or dsDNA of 30 nt or bp in length. The EMSAs demonstrate that pA104R binds both dsDNA and ssDNA at the lowest concentration of 0.1 and 0.4 μM, respectively (Fig. 1 C and D). These results indicate that pA104R possesses a higher affinity for dsDNA than ssDNA. To define the minimum length of bound DNA, 5′-FAM–labeled oligonucleotides with different lengths were used in EMSA. As shown in Fig. 1 E and F, pA104R can bind oligonucleotides with lengths of 6–12 bp/nt, while no DNA binding was detected when the length of the oligonucleotide was reduced to 5 bp/nt. This result suggests that the minimal DNA length required for pA104R binding is 6 bp/nt. Furthermore, the DNA-pA104R affinity as indicated by the relative density of the bound DNA bands to the free DNA bands increases with the oligonucleotide length.

The Apo-pA104R Structure.

The structure of apo-pA104R was solved by the single-wavelength anomalous diffraction (SAD) method to a resolution of 2.80 Å with R_work and R_free values of 25.3% and 29.9%, respectively (SI Appendix, Table S1). The overall structure of pA104R is highly similar to other bacterial HU/IHF homologs. It presents a homodimer that folds into two domains: a largely α-helical “body” (α-helical region [AHR]) capped by the β-strand DNA binding region (BDR),, which is characterized by a “bottom” and two β-ribbon “arms” (Fig. 2 A and B). The secondary structural elements of a protomer of the pA104R dimer include two α-helixes (α1 and α2) followed by five β-sheets (β1–5) and end with a final α/3₁₀-helix (α3/η1). The β1, β2, and β5 strands form an antiparallel β-sheet, which constitutes the bottom of the BDR, while the β3 and β4 strands comprise the arms of the BDR (Fig. 2C). The electrostatic surface of pA104R shows that positively charged amino acids are densely distributed in the BDR region, which are potential binding sites of the negatively charged DNA backbone (Fig. 2D).

Fig. 2.

The structure of pA104R. (A) The overall structure of dimeric pA104R. (B) One protomer of the dimeric pA104R. (C) Topological diagram of the secondary structural elements of the pA104R protomer. (D) Electrostatic surface views of pA104R. (E) Structure-based sequence alignment of pA104R with other HU/IHF members. The two sets of extra amino acids in pA104R, residues P53-G55 and A90-K92, are highlighted by cyan boxes. The lengthened β2-β3 loop and β4-β5 loop are highlighted with red stars. (F) The uncommon β-ribbon arms within the structure of pA104R. Structural superposition of 10 structure-solved HU/IHF members (corresponding to those in the sequence alignment) and pA104R, demonstrating the unique protrusion of the β-ribbon arms of pA104R (highlighted by red dashed circles). The PDB codes of these structures are 1IHF, 1P71, 1HUU, 5L8Z, 1B8Z, 2NP2, 4QJN, 4PT4, 5FBM, and 2O97.

Despite sharing highly similar architectures with HU/IHF homologs, both sequence alignment and structural superimposition unexpectedly revealed that the β-ribbon arms of pA104R are lengthened by two sets of extra amino acids: residues P53-G55 and A90-K92 (Fig. 2 E and F). As shown in Fig. 2E, sequence alignment of pA104R with HU/IHF family members shows that the β1, β3, and β4 strands are well aligned with the homologs. The six extra residues in pA104R, however, rearrange the end of the β2-strand and the start site of the β5-strand. Therefore, the number of amino acids in the β2-β3 loop and β4-β5 loop of pA104R varies from two to five, which extends the length of the loops, yielding the longer arms of pA104R.

Structure of the pA104R-DNA Complex.

To determine the mechanism by which pA104R binds to its DNA ligand, we attempted to solve the crystal structure of pA104R in complex with a partially mismatched 20-bp DNA substrate (SI Appendix, Table S2). The structure was solved at a resolution of 2.80 Å with R_work and R_free values of 25.3% and 27.5%, respectively (SI Appendix, Table S1). Overall, the complex structure reveals that the DNA substrate sits within one flank of the BDR and is clamped cooperatively by the two protomers of pA104R, one with its β-ribbon arm while the other uses its base region (Fig. 3A). Structural comparison of apo-pA104R and substrate-bound pA104R presented a rigid AHR, whereas a significant conformational change in the β-ribbon arms occurred upon DNA binding (SI Appendix, Fig. S1A). In the superimposition, the tip residues from each turn interconnecting the β3- and β4-strands differ in position by a long distance of ∼17 Å, indicating the flexibility of the β-ribbon arms. As for the DNA duplex in the complex structure, high electron density of 17 pairs of bases could be observed, covering nucleotides T1 to A19 with the exception of T4 and T7 from both strands. The residual electron density map of the sugar-phosphate backbones of nucleotides T4 and T7 suggests that these thymine bases flip out of the duplex (SI Appendix, Fig. S1B). Accordingly, the pattern of how the DNA duplex formed was deduced of which there are 14 A-T or G-C base pairs, along with three wobble base pairs formed by T-T mismatches (T5-T16, T11-T11, and T16-T5) (SI Appendix, Fig. S1C).

Fig. 3.

Atomic interactions between the pA104R and the DNA in the pA104R-DNA complex. (A) The complex structure of pA104R binding to 20 bp of the DNA duplex. The binding site is shown in a closeup view. (B and C) Zoomed-in views of the detailed interactions between the DNA duplex and the base region of the BDR bottom (B) and the β-ribbon arms (C). The hydrogen bonds and salt bridge interactions are highlighted by red dashed lines, and the related residues are shown as sticks. (D) Schematic of the detailed interactions between the pA104R and the DNA duplex. All of the contacts are highlighted by black lines, and the hydrogen bonds or salt bridge interactions are marked by red dashed lines. (E–H) The binding dynamics between wide-type pA104R (E), the K83D and K85D double mutant (F), the R69D and H72D double mutant (G), and the K92D, R94D, and K97D triple mutant (H) and the DNA substrates, respectively. BLI data were analyzed using a 1:1 Langmuir binding model, and black lines represent curve fits.

Detailed interactions within the binding interface were further characterized. Structural analysis showed that pA104R uses the bottom region (the β2′- and β5′-strands and the α3′-helix) of one protomer to contact the sugar-phosphate backbone of the minor groove of DNA substrate, while curling around the major groove via the arm region (β3- and β4-strands) of the other protomer. In the bottom region, five residues (K57, K92, R94, L96, and K97) are involved in the interactions (Fig. 3B), among which three basic residues K92, R94, and K97 contribute the majority of the tight hydrogen bonds and salt bridges with the DNA backbone (Table 1 and Fig. 3D). Accordingly, another five residues in the arm region (R69, H72, P74, K83, and K85) also participate in the binding process (Figs. 3 C and D and Table 1) such that they recruit the flexible β-ribbon arms to bend toward the DNA substrate.

Table 1.

Interaction between the pA104R and the 20-bp DNA duplex

pA104R	Contacts*	DNA duplex
K57	1, 5	T9, A10
R69	1	C3
H72	52	T4
P74	6	T4
K83	1, 3	T12′, T13′
K85	4	T14′
K92	6	T16′
R94	1, 14, 5, 1	G14′, T15′, A10, T11
L95	3, 5	A10, T11
K96	6, 9	T11, T12

^*Numbers represent the number of atom to atom contacts between the pA104R residues and the nucleotides of the DNA duplex, which were analyzed by the Contact program in the CCP4 suite (the distance cutoff is 4.5 Å). To avoid the confusion of the equivalent nucleotides located at two strands of the DNA duplex, single quotes are marked on the nucleotide numberings of one strand that is named “chain C” in the complex structure data.

Protein-based biolayer interferometry (BLI) assays were performed to evaluate the key residues involved in the pA104R-DNA interaction and clarify the DNA binding mode of pA104R. The structural analyses above revealed that the β-ribbon arm residues R69, H72, K83, and K85 as well as the base region residues K92, R94, and K97 play major roles by contributing most of the interactions. Thus, we substituted these basic amino acids with the acidic amino acid aspartate (D), including double substitutions (K83,85D and R/H69,72D) and triple substitution (K/R92,94,97D). The BLI assay indicated that the double mutations in the β-ribbon arm have no (K83,85D, Fig. 3F) or little (R/H69,72D, Fig. 3G) if any effect on DNA binding compared to the wild-type (WT) pA104R (pA104R-WT, Fig. 3E). The triple mutant in the base region (K/R92,94,97D), however, abolished binding to the DNA substrate (Fig. 3H). Taken together, the mutagenesis work indicates that, within the BDR, the bottom region—rather than the flexible arm—contributes more to DNA binding.

Crystal Packing Reveals an Underlying DNA Bending Mode Mediated by pA104R.

To elucidate the mechanism of DNA condensation by pA104R, we generated a pseudocontinuous DNA helix along with the bound pA104R by showing the neighboring pA104R-DNA complex in the crystal packing. As presented in Fig. 4A, a DNA duplex is simultaneously held by two pA104R dimers via one of their two β-ribbon arms. Under the mediation of the DNA helix, the pA104R proteins are closely arranged along the DNA molecule. Interestingly, E79 at the start of the β4 strand from the arm of one pA104R forms polar contacts with K11 and Q12 of the α1 helix from the neighboring pA104R (Fig. 4A). These polar forces further stabilize the interactions between multiple pA104R molecules, making them a firm continuous scaffold for holding DNA.

Fig. 4.

Crystal packing of the pA104R-DNA complex. (A) DNA fragments form a pseudocontinuous DNA helix, and neighboring pA104R molecules form polar contacts in crystal packing. (B) The angle between neighboring DNAs (represented lines a and b, the principal axes of the two DNAs) may reflect the overall bending angle induced by pA104R. (C) The overall bending angle of DNA (represented by the angle between lines a and b, the principal axes of the two DNA fragments outside the P74 intercalation sites) induced by Anabaena HU (PDB ID: 1P71). (D) The overall bending angle of DNA (represented by the angle between lines a and b, the principal axes of the two DNA fragments outside the P74 intercalation sites) induced by E. coli IHF (PDB ID: 1IHF).

Previously reported structures of HU/IHF in complex with DNA show that DNA bends at the site where a conserved proline at the arm tip (P74 of pA104R shown in Fig. 2E) intercalates the minor groove, disrupting base stacking and inducing/stabilizing DNA bending (19, 26–28). Due to the unique binding pattern between the pA104R and the DNA substrate, the proline-mediated intercalation occurs at the major groove of the DNA strand where T4 of one DNA chain flips out of the DNA groove (Fig. 3C). This P74-mediated intercalation does not induce obvious DNA bending. This may be because that the P74 intercalation sites are close to the DNA ends, and the bending tension is released from the free DNA ends.

However, the two DNA strands from the neighboring pA104R form an angle of 93.8°, which is close to the overall bending angle (106°) of the DNA in the Anabaena HU-DNA complex rather than to the overall bending angle (160°) of the DNA in the E. coli IHF-DNA complex (Fig. 4 B–D). Whether this bending angle is related to the P74 intercalation as in the case of the HU-DNA complex requires further investigation.

Stilbene Derivative Compounds Bind to pA104R and Inhibit the pA104R-DNA Interaction.

Two stilbene derivatives (SD1 and SD4) were previously reported to perturb the interaction between Mtb HU and DNA, disrupt the Mtb nucleoid architecture, and reduce Mtb growth (25). Here, we examined whether SD1 and SD4 can inhibit the interactions between pA104R and DNA (Fig. 5A). Docking tests demonstrated that SD1 posing with a top score binds to one of the BDR arms, while SD4 binds to the BDR bottom. These binding patterns obviously occupy the binding position of DNA at pA104R and, thus, may interrupt the interaction between DNA and pA104R (Fig. 5 B and C). In EMSAs, SD1 inhibited binding of dsDNA to pA104R in a dose-dependent manner with the inhibitory concentration to reduce 50% of DNA binding (IC₅₀) of 275 μM (Fig. 5D). SD4 also inhibited binding of dsDNA to pA104R in a dose-dependent manner, but the IC₅₀ was much lower (6.1 μM) (Fig. 5E), suggesting that SD4 is a more potent inhibitor of DNA-pA104R interaction.

Fig. 5.

SD1 and SD4 bind to pA104R and inhibit the pA104R-DNA interaction. (A) Chemical formulas of stilbene derivative SD1 and its methoxy derivative SD4. (B and C) Representative binding poses of SD1 (B) and SD4 (C) in the pA104R structure based on energy-minimized docking tests. (D and E) Effects of SD1 (D) or SD4 (E) on the inhibition of pA104R-DNA binding in EMSAs and their corresponding inhibition curves.

Due to the essential role of pA104R in ASFV replication and with the evidence of the inhibitory effects of SD1 and SD4 on the DNA-pA104R interaction, we further tested whether SD1 and SD4 can also inhibit replication of ASFV replication in swine macrophages. The cytotoxicity of stilbene derivative compounds was tested by a CCK-based assay. As shown in SI Appendix, Fig. S2, SD1 inhibited cell growth at concentrations >50 μM with a 50% cytotoxic concentration (CC₅₀) of 102 μM. The cytotoxic effect of SD4 was much lower with a CC₅₀ of 263 μM (SI Appendix, Fig. S2). Then, we selected 50 μM as the optimal nontoxic concentrations for both SD1 and SD4 in the subsequent analysis of the antiviral effect since there is only weak (12.7% for SD1) or no cell growth inhibition at this concentration.

The effect of these compounds on viral production was tested by both real-time qPCR and fluorescence assay-based TCID₅₀ tests. As shown in Fig. 6 A and B, both SD1 and SD4 displayed strong inhibition of ASFV DNA levels in a dose-dependent manner with a reduction of 77.58% for SD1 and 95.19% for SD4 at concentrations of 50 μM. The inhibitory concentration to reduce 50% of virus production (IC₅₀) was estimated to be 2.93 μM for SD1 and 0.48 μM for SD4 (Fig. 6C). Meanwhile, we also tested the titers of infectious progeny virus in the supernatants by fluorescence assay-based TCID₅₀ tests (Fig. 6 D and E). In accordance with the data above, the viral titers were reduced by both compounds in a dose-dependent manner. The IC₅₀s of SD1 and SD4 were 3.85 and 0.34 μM, respectively (Fig. 6F). These results suggest that both SD1 and SD4 can inhibit ASFV replication in swine macrophages, and SD4 has a more potent inhibitory effect than SD1.

Fig. 6.

Inhibition of ASFV replication by SD1 and SD4. (A and B) The inhibitory effect of SD1 (A) or SD4 (B) treatment on ASFV replication reflected by ASFV DNA levels. (C) The dose–response curves of SD1 and SD4 on relative ASFV DNA levels. (D and E) The inhibitory effects of SD1 (D) or SD4 (E) treatment on ASFV replication reflected by TCID₅₀. (F) The dose–response curves of SD1 and SD4 on relative ASFV TCID_50.

Discussion

As ASFV continues to spread, the unavailability of effective prevention and treatment regimens highlights the importance of studying the structures and functions of critical viral proteins that may be used as targets for vaccine and drug design. The crystal structures of some critical ASFV proteins have been reported, including pE165R, a dUTPase essentially involved in maintaining genome fidelity during viral replication (29), but the structures and functions of most ASFV-encoded proteins remain elusive. pA104R, an ASFV nucleoid-associated histonelike protein, was previously shown to be indispensable for successful viral replication (23). In the present study, we demonstrated the DNA binding properties of pA104R with EMSAs and solved the crystal structures of pA104R both in its apo state and in complex with DNA, revealing the structural basis for pA104R binding to DNA.

By evaluating the DNA binding properties of pA104R, we found that pA104R is capable of binding to both dsDNA and ssDNA while possessing a higher affinity with dsDNA. This finding indicated that pA104R might be more likely to compact the dsDNA genome of ASFV than other ssDNAs presented in the complicated environment of the cytoplasm of infected cells.

pA104R was previously shown to have sequence identity with HU and IHF family proteins. Although HU and IHF are analogous in function to both the histones and the HMG box proteins of eukaryotes, they show structural homology to neither (15). The 3D structures of HUs from Thermotoga maritima (30), Geobacillus stearothermophilus (31), E. coli (32–34), Staphylococcus aureus (35), Borrelia burgdorferi (36), Anabaena PCC7 120 (27), Streptococcus mutans (37), M. tuberculosis (25), mycoplasma Spiroplasma melliferum KC3 (38), the HU homologous protein TF1 encoded by bacteriophage SPO1 (39), as well as crystal structures of IHFs from E. coli (26, 40) all share a common AHR plus BDR architecture. Our crystal structures of the apo-pA104R and pA104R-DNA complex confirm that pA104R adapts the same fold with other HU/IHF members.

The HU-DNA and IHF-DNA cocrystal structures reported to date suggest that HUs and IHFs share largely conserved DNA binding patterns and a common DNA bending mechanism. In all of these HU/IHF-DNA complex structures, DNA contacts the β-sheets of both arms. The contacts occur mainly at the minor groove of DNA and the prolines, equivalent to P74 in pA104R at the tip of both arms intercalate between base pairs to induce DNA bending, although the bending angles vary (26, 27, 35). However, pA104R shows several unique DNA binding characteristics compared to other HU/IHF members: 1) the short DNA present in the complex structure binds to only one half of pA104R, 2) the BDR bottom contacts the minor groove of the DNA while the BDR arm contacts the major groove of the DNA, and 3) the basic residues in the BDR bottom rather than those in the BDR arm play a major role in DNA binding affinity than those (Fig. 3).

An interesting finding in our pA104R-DNA complex structure is that the proline at the tip of arm does not induce visible DNA bending, such as other HU/IHF members. However, the slant pose of the DNA in the asymmetric unit makes it form an angle of 93.8° with the DNA in the neighboring asymmetric unit. It is possible that DNA bending occurs at this angle if the two DNAs are continuous. Another remarkable difference between pA104R and other HU/IHF members is that the proline at the tip of the pA104R intercalates into the major groove of the bound DNA, whereas those of other HU/IHF members intercalate into the minor grooves of the bound DNAs. The arm tip proline intercalation does not induce a visible DNA bend. Therefore, whether the overall bending angle is induced by the proline intercalation is very intriguing and needs further study.

Although pA104R shares higher identity with HU than with IHF (13), there is no convincing evidence demonstrating that it is a HU-like or IHF-like protein. We showed that pA104R forms homodimers in solution and in the crystal structures, which is a property of HU rather than IHF. In addition, the ASFV genome does not encode any protein with a sequence similar to pA104R, which means that there is little chance of pA104R forming a heterodimer, such as IHFs. Furthermore, phylogenetic analysis with 23,368 HU/IHF protein sequences explicitly revealed that pA104R belongs to the HU clade (SI Appendix, Fig. S3). Therefore, all these structural and sequence evidences suggest that pA104R is a HU-like protein instead of an IHF-like protein.

Inspired by the report that stilbene derivatives (SD1 and SD4) can disrupt Mtb HU binding to DNA and inhibit Mtb growth, we, indeed, observed that both SD1 and SD4 disrupted pA104R binding to DNA and reduced ASFV replication in a dose-dependent manner. SD4 showed higher potency than SD1. These results further confirmed the vital role of the pA104R-DNA interaction in the ASFV replication cycle and highlight the potential of these two compounds for use as drugs or lead compounds for ASFV control and therapy.

Materials and Methods

Protein Expression and Purification.

To express the WT pA104R (pA104R-WT) protein, the cDNA encoding residues 1–104 of pA104R (GenBank: AYW34006.1) was synthesized and codon optimized for expression in E. coli. The gene was added with a His₆ tag encoding sequence and a stop codon at its 5′- and 3′-termini, respectively, and then cloned into the NdeI and XhoI sites of the pET-21a vector (Invitrogen). The pA104R mutants were constructed by site-directed mutagenesis. These recombinant proteins were expressed in E. coli strain BL21 (DE3) as soluble proteins after inducing with 0.5-mM isopropyl-β-d-thiogalactopyranoside at an OD₆₀₀ of 0.6–0.8 and expressing at 16 °C. The cells were harvested after 16 h and lysed by sonication in lysis buffer (20-mM Na₃PO₄, 500-mM NaCl, and 100-mM imidazole, pH 7.4). After centrifugation, the supernatants were then purified by metal affinity chromatography using HisTrap HP 5-mL columns (GE Healthcare). The partially purified proteins were eluted with elution buffer (20-mM Na₃PO₄, 500-mM NaCl, and 500-mM imidazole, pH 7.4) and further purified by size-exclusion chromatography using a Hiload 16/60 Superdex 75 PG column (GE Healthcare) equilibrated with binding buffer (10-mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [Hepes]-HCl and 500-mM NaCl, pH7.4).

To obtain the complex of pA104R-WT bound to dsDNA, the pA104R-WT protein and 20-bp dsDNA (SI Appendix, Table 2) were mixed in vitro at a molar ratio of 1:5 and incubated at room temperature for 0.5 h before purification by gel filtration. The fractions containing the complexes were collected and concentrated to ∼10 mg/mL for crystal screening. The target dsDNA was synthesized as single-stranded oligonucleotides and prepared by annealing.

Crystallization and Data Collection.

All of the crystals were obtained using the sitting drop vapor diffusion method with 1-μL protein mixed with 1-μL reservoir solution and then equilibrating against 100-μL reservoir solution at 18 °C (41). Native crystals of apo pA104R protein were obtained in 0.1-M citric acid, pH 4.5 and 2.4-M ammonium sulfate at a protein concentration of 5 mg/mL. Derivative crystals were obtained by soaking native crystals overnight in mother liquor containing 1-mM KAuCl4•xH2O. For the pA104R-DNA complex, the diffractable crystals were obtained in a reservoir solution of 0.2-M imidazole malate, pH 7.0 and 15% (wt/vol) PEG 4,000 with a protein concentration of 10 mg/mL.

For data collection, all crystals were briefly soaked in their individual reservoir solutions supplemented with 20% (vol/vol) glycerol and flash cooled in liquid nitrogen. The apo pA104R (wavelength, 0.979 Å) and Au derivative (wavelength, 1.034 Å) data sets were collected at the Shanghai Synchrotron Radiation Facility BL19U1 where the complex dataset (wavelength, 0.979 Å) was collected at BL17U1 at a temperature of 100 K and processed with HKL2000 software (42–44).

Structure Determination.

The crystal structure of apo pA104R was solved by the SAD method as previously described (45). In brief, the Au sites were first located by SHELXD (46) for the Au-SAD data. The identified positions were then refined, and the phases were calculated with the SAD experimental phasing module of Phaser (47). The real space constraints were further applied to the electron density map in DM (48). After using Autobuild in the Phenix package (49) to build the initial model, additional missing residues were added manually in COOT (50). Rounds of refinement were performed using phenix.refine in Phenix (49) with energy minimization, isotropic ADP refinement, and bulk solvent modeling. The complex structure was solved by the molecular replacement method using Phaser (47) from the CCP4 program suite (51) with the solved pA104R structure as the search models. The atomic model was completed with COOT (50) and refined with phenix.refine (49), and the stereochemical qualities of the final models were assessed with PROCHECK (52). Data collection, processing, and refinement statistics are summarized in SI Appendix, Table 1. All structural figures were generated using Pymol (https://pymol.org/2/).

Biochemical Characterization of the pA104R Protein.

The purified pA104R protein was analyzed with an analytical gel filtration assay with a calibrated Hiload 16/60 Superdex 75 PG column (GE Healthcare). The pooled proteins were analyzed on a 15% SDS/PAGE gel and stained with Coomassie blue.

The sedimentation velocity experiments were performed using a Beckman Optima XL‐A analytical ultracentrifuge. The pA104R protein was prepared with binding buffer at a concentration of A280 = 0.8 absorbance units. The protein sample (400 μL) and reference solution (equal volume of binding buffer) were loaded into a conventional double-sector quartz cell and mounted in a Beckman Coulter 4-hole An-60 Ti rotor. Before the run, the rotor was equilibrated for ∼1 h at 20 °C in the centrifuge. The experiments were then performed at 20 °C and 60,000 rpm with a continuous scan mode and a radial spacing of 0.003 cm. Scans were collected in 3-min intervals at 280 nm. The protein partial specific volume was calculated based on the amino acid composition of pA104R using Sednterp software (http://bitcwiki.sr.unh.edu/index.php/Downloads), and the fitting of absorbance versus the cell radius data were performed using SEDFIT software (https://sedfitsedphat.nibib.nih.gov/software/default.aspx). The molecular weight was estimated from the sedimentation velocity data using the c(M) model after fitting the frictional ratio (f/f0).

Electrophoretic Mobility Shift Assay.

The sequences of the oligonucleotides used in EMSA assays are given in SI Appendix, Table 2. For DNA fragment preparation, the 5′-FAM labeled single-strand oligonucleotides with different lengths (5, 6, 8, 10, 12, and 30 nt) were synthetized by Synbio Technologies, Inc. and dissolved in binding buffer. The corresponding double-stranded oligonucleotides were generated by annealing to the unlabeled complementary strands at a molar ratio of 1:1.

The binding reactions were performed with 20 pmol of ds/ss DNA fragments and serially diluted pA104R proteins (the target concentrations are shown in Fig. 1). After a 30-min incubation at room temperature, the reaction systems were supplemented with 5×loading buffer and electrophoresed in 10% native-PAGE with 0.5×Tris-borate-EDTA (TBE) buffer. The gels were then visualized with a VILBER FUSION FX5 imaging system.

For the inhibition assays, SD1 and SD4 were purchased from Chembridge, Inc. and dissolved in dimethyl sulfoxide. The 30-bp dsDNA (20 pmol) and pA104R protein (0.4 μM) were preincubated at room temperature for 30 min. Different concentrations of the two compounds were then added to the reaction systems, incubated for another 5 min, and electrophoresed by 10% native-PAGE as mentioned above. After visualization, the images were analyzed by ImageJ software, and the inhibition curves together with IC₅₀ values were determined using GraphPad Prism 5 software.

BLI Binding Assay.

The binding kinetics between the WT or mutant pA104R proteins and the DNA fragment were determined using the Octet RED96 platform (ForteBio, Inc.). The entire experiment was performed at 30 °C with the plate shaking at a speed of 1,000 rpm. All of the proteins used in the BLI assay were exchanged into BLI buffer that contained 10-mM Hepes-HCl (pH 7.4), 150-mM NaCl, and 0.02% (vol/vol) Tween-20 via gel filtration. The 30-bp dsDNA (SI Appendix, Table 2) used in the BLI assay was synthetized as 5′-biotin labeled single-strand oligonucleotides and annealed as described above. The SA sensors were loaded with 50-nM dsDNA for 90 s and then exposed to serially diluted WT or mutant pA104R proteins. Double reference subtraction was applied to correct the errors of sensor drifting. All data were analyzed with the data analysis software 9.0 (ForteBio) using a 1:1 Langmuir binding model and plotted with the Origin 8.0 program.

Compound Docking.

Docking of the compounds to pA104R was performed using the Auto-Dock Vina package (53). Preparation of parameter files for grid and docking was conducted using Auto-DockTools-1.5.6 (54). The grid center was designated at dimensions (x, y, and z): 16.701, −13.969, and 18.090. The size for the compounds docking was given as 62, 62, and 72 Å on X, Y, and Z coordinates, respectively, which makes sure that the search space is large enough to cover the whole structure of pA104R dimer.

Cells and Viruses.

Primary porcine alveolar macrophages (PAMs) were collected from 4-wk-old specific-pathogen-free pigs and maintained in Roswell Park Memorial Institute (RPMI) 1640 Medium (Gibco) supplemented with 10% fetal bovine sera (Gibco), 200-mg/mL streptomycin, and 200-IU/mL penicillin at 37 °C. The ASFV Pig/SY18 strain was isolated and further propagated in PAMs for amplification culture. Viral titration was performed by hemadsorption (HAD) assays and expressed as HAD₅₀/mL. All experiments with live virus were conducted in biosafety level 3 facilities in the Harbin Veterinary Research Institute of the Chinese Academy of Agricultural Science.

In Vitro Antiviral Assays.

PAMs were seeded on 24-well plates at a density of 5 × 10⁵ cell/well. The compounds were 10-fold diluted ranging from 50 μM to 0.5 nM before being incubated with an equal volume of ASFV suspension containing 2.5 × 10⁵ HAD₅₀/mL viruses, and the mixtures were incubated for 1 h at 37 °C. After removal of the supernatants, cells were washed twice with phosphate-buffered saline (PBS) and incubated with medium containing compounds at the indicated concentrations for 48 h. The supernatants of each well were collected, and the virus titers were determined by both real-time qPCR and fluorescence assay-based TCID₅₀ tests.

For the real-time qPCR, viral DNA was obtained from the supernatants of the infected cells using TRIzol Reagent (ThermoFisher Scientific, USA), and quantitative RT-PCR was performed using Premix Ex Taq TM (Probe qPCR) (TaKaRa, Japan) on a QuantStudio 3 Real-Time PCR System (Applied Biosystems, USA) according to the manufacturer’s protocol. The real-time PCR methods were recommended by the World Organization for Animal Health using fluorescent hybridization probes targeting the p72 gene (55).

For the fluorescence assay-based TCID₅₀ test, PAMs in 96-well plates were infected with 10-fold serially diluted supernatant. At 48 h p.i., cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. Then, anti-p72 mAb (Ingenasa) was used as the primary antibody and fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG was used as a secondary antibody. The fluorescence detected in each well was defined as ASFV positive. The TCID₅₀ was calculated using the Spearman and Karber algorithm (56).

Data Availability.

The atomic coordinates of apo-pA104R and pA104R-DNA complex have been deposited into the Protein Data Bank, https://www.wwpdb.org/ (PDB ID codes 6LMH and 6LMJ, respectively).