Dear Editor,

The outbreak of human monkeypox in May 2022 across non-endemic countries is unusual compared to previous reports.1 A recent study indicates that the basic reproduction number (R0) of monkeypox virus (MPXV) is > 1 in populations of men who have sex with men,2 and MPXV is reported to be transmitted via close contact and airborne droplets, suggesting that monkeypox poses a great threat to the global health.

MPXV, belonging to the orthopoxvirus genus of the poxviridae family, is a large enveloped DNA virus with a double-stranded DNA (dsDNA) molecule of ~197 kilobase pairs (kb), which encodes ~181 proteins.3 The life cycle of MPXV takes place entirely in the cytoplasm of infected cells, where MPXV replicates its DNA by a DNA replication machinery that consists of the catalytic subunit F8, a B-family DNA polymerase, and a processivity factor heterodimer A22 and E4 (Fig. 1a).3,4 E4, a uracil-DNA glycosylase that removes uracil from DNA, is linked to F8 by A22. Unlike RNA viruses, the dsDNA poxviruses have much lower mutation frequency as their DNA replication machineries enable coupling of DNA replication, proofreading and base excision repair.5 In addition, the DNA replication machinery is an important drug target.6 Here we report cryogenic electron microscopy (cryo-EM) structures of MPXV DNA replication machinery F8–A22–E4 in the presence and absence of DNA, providing insights into MPXV DNA replication and development of drugs against MPXV.

Fig. 1: Structure of MPXV DNA replication machinery and the potential interactions.
figure 1

a Schematic diagram of domain organizations of subunits F8, A22 and E4. The inserts 0–4 in purple are based on the previous reports.7,15 b Cryo-EM structure of MPXV DNA replication machinery shown in density map. Subunits F8, A22 and E4 are arranged in a circular structure. c Cryo-EM structure of MPXV DNA replication machinery in complex with DNA in density map. DNA is colored in red. d Interaction between F8 insert 3 and C-terminal domain of A22. F8 residues are colored in purple. e A22 interaction with E4. N-terminal residues of A22 are colored in sky blue. f E4 interaction with F8. N-terminal residues of F8 are colored in yellow. g Potential DNA binding residues of F8 for DNA synthesis. The DNA is shown in cartoon and orange. h Potential active site of F8 for DNA synthesis. As revealed by structural superimposition with bacteriophage RB69 DNA polymerase complexed with DNA (PDB code: 3NCI), the DNA and the incoming dCTP are shown in stick and orange, and the potential catalytic residues in green (bacteriophage RB69) and in yellow (F8). i Potential cidofovir diphosphate binding site. Cidofovir diphosphate is shown in purple and stick. j Potential DNA binding site for 3′-5′ exonuclease. The superimposed DNA is from T. gorgonarius DNA polymerase in complex with DNA (PDB code: 2XHB). k The superimposed DNA from T. gorgonarius DNA polymerase shown in stick and orange is located in the potential catalytic site of 3′-5′ exonuclease. The residues from T. gorgonarius are colored in green and from MPXV in yellow. l Potential DNA binding site of E4. The superimposed DNA is from VACV D4 in complex with DNA (PDB code: 4YIG). m Stereoview of E4 potential catalytic residues. The superimposed DNA from VACV D4 complexed with DNA is shown in stick and orange. The deoxyribose of the abasic product is in the active site. Residues from VACV D4 and MPXV E4 are colored in green and cyan, respectively. n Modeling DNA from VACA D4 complex into MPXV DNA replication machinery. F8 synthesizes DNA; E4 scans uracil-containg DNAs and initiates base excision repair. o Modeling DNAs from T. gorgonarius DNA polymerase complex and VACA D4 complex into the MPXV DNA replication machinery. F8 detects mismached bases of newly synthesized DNA and switches the DNA from synthesis state to proofreading state at the 3′-5′ exonuclease active site. Once proofreading is finished, F8 would switch the DNA back to the active site for synthesis, indicating the coupling mechanism.

Full size image

The MPXV polymerase subunit F8 and processivity factor subunits A22 and E4 were co-expressed and the machinery was co-purified into homogeneity (Supplementary information, Fig. S1a, b). The molar ratio of the three subunits within the replication complex is 1:1:1 with a molecular weight of 191 kDa. Primer (P) and template (T) strands were synthesized and annealed to form the P/T DNA duplex (Supplementary information, Fig. S1c). A biolayer interferometry (BLI) analysis showed that the MPXV DNA replication machinery binds the DNA duplex with a KD of 56.7 nM (Supplementary information, Fig. S1d). The cryo-EM structures of MPXV DNA replication machinery were determined at resolutions of 3.05 Å (without DNA), 2.76 Å (no visible DNA) and 3.01 Å (DNA bound) using the golden standard of Fourier Shell Correlation of 0.143 (Supplementary information, Figs. S2, S3 and Table S1).

The MPXV DNA replication machinery is beetle-shaped with a dimension of ~115 Å (length), ~92.5 Å (width) and ~131 Å (height) (Fig. 1b, c). The polymerase subunit F8 contains N-terminal, exonuclease, finger, palm and thumb domains, as well as five insertions, namely, inserts 0–4. The inserts 1–2 are located in the exonuclease domain, while inserts 0 and 3–4 are part of the N-terminal domain and palm domain, respectively. The structure of polymerase subunit F8 resembles VACV E9 crystal structure with a root-mean-square deviation (RMSD) of 2.55 Å over 723 Cα atoms (PDB code: 5N2E)7 (Supplementary information, Fig. S4). Remarkably, the three subunits (F8, A22, E4) of the DNA replication machinery are arranged in a circle and form a hole near the tail (Fig. 1b, c).

A22 has a hair-clip shape, consisting of four domains, a N-terminal domain, two central domains and a C-terminal domain (Fig. 1a, b). The C-terminal domain of A22 binds to insert 3 of F8 and its N-terminal domain binds to the C-terminal domain of E4. Interestingly, the N-terminal domain of E4 makes contact with the 3′-5′ exonuclease domain of F8 (Fig. 1b). This is in contrast to other B-family polymerase holoenzymes where processivity factors always bind to their thumb domains of catalytic subunits (Supplementary information, Fig. S4).7,8,9 Together, this reveals a unique structural feature of the MPXV DNA replication machinery that the processivity factor A22–E4 complex binds to the insert 38 and the exonuclease domain of the catalytic subunit F8.

The interface between F8 and A22 is 564.5 Å2 (Fig. 1b). The C-terminal domain of A22 forms a hydrophobic cavity composed of residues Phe354, Ile369, Val372, Asn373, Met375, Arg376, Phe377, Ile379, Cys382, Phe407, and Phe414 (Fig. 1d), whereas residues Leu578 and Ile582 extending out of insert 3 of F8 are located inside the hydrophobic cavity. In addition, Thr575, Asn576, Arg577, and Glu581 of insert 3 are involved in salt bridge or hydrogen bond formation with the A22 residues around the hydrophobic cavity (Fig. 1d).

The A22–E4 interaction is through a head (A22)-to-tail (E4) manner (Fig. 1b). The N-terminal domain of A22 and the C-terminal domain of E4 interact via an interface of 675.6 Å2. A22 residue Trp43 (top left) is sandwiched by E4 residues Arg167 and Pro173, while E4 residues Val174, Thr176, Ile177, Ile197, Leu201 and Leu204 are located to a hydrophobic groove formed by A22 residues Lys11, Leu14, Tyr42, Lys44 and Ile45 (Fig. 1e). E4 residues Arg193, Glu196 (top right), as well as A22 residues Ser40, Thr41 and Asn9, are involved in salt bridge or hydrogen bond formation (Fig. 1e).

The interface between the F8 exonuclease domain and the N-terminal domain of E4 is ~231.3 Å2, where Leu278 of F8 extends into a hydrophobic groove composed of E4 residues Val33, Trp36, Ile135 and Tyr136 (Fig. 1f). Residues Glu32 and Ser35 of E4, and residues Ser177, Phe179, Asn274 and Glu277 of F8 are also in the range of contact in this interaction interface (Fig. 1f).

In contrast to the MPXV DNA replication machinery structure (3.05 Å) in the absence of DNA, the density of the F8 thumb domain is clearly visible and the domain was built into the DNA-bound machinery structure (3.01 Å) (Fig. 1c), indicating that the F8 thumb domain is critical for DNA binding. There is an additional density near the thumb domain, which was identified as DNA with a low resolution, suggesting that the DNA is flexible in the F8–A22–E4 machinery. The DNA-bound structure is highly similar to the no visible DNA-bound structure (2.76 Å) with a RMSD of 2.34 Å over 1296 aligned Cα atoms (Supplementary information, Fig. S5a, b), while superimposition of the 3.05 Å (without DNA) structure and the 2.76 Å (no visible DNA) structure showed a RMSD of 2.40 Å over 1267 aligned Cα atoms. About 13 lysine or arginine residues (Arg302, Lys308, Lys340, Lys525, Arg674, Lys803, Lys804, Lys805, Arg832, Arg833, Lys973, Arg974 and Arg1000) line up in the positively charged groove of F8 to bind the DNA (Fig. 1g). Among these residues, Arg832, Arg833, Lys973, Arg974 and Arg1000 from the thumb domain are involved in DNA binding. Structural superimposition of bacteriophage RB69 DNA polymerase in complex with DNA (PDB code: 3NCI) and the MPXV DNA replication machinery shows that the DNA from the RB69 complex is located in the same position as that from the MPXV machinery (Supplementary information, Fig. S5c, d). The incoming nucleotide dCTP from the RB69 complex is located in the catalytic site, pairing with deoxyguanosine of the template DNA, which allowed the identification of the F8 potential catalytic residues Asp753 and Asp549, as well as potential DNA binding residues Ser552, Tyr554, Asn665, Lys661, Arg634, Lys638 and Thr752 in the palm and finger domains for the 5′-3′ polymerase activity (Fig. 1h).

Brincidofovir (Supplementary information, Fig. S6a) is a prodrug of acyclic nucleoside phosphonate cidofovir (Supplementary information, Fig. S6b), a nucleotide analog of deoxycytidine monophosphate. Brincidofovir is the lipid 3-hexadecyloxy-1-propanol conjugate of cidofovir. Once brincidofovir enters cells, its lipid moiety is cleaved and the phosphonate of cidofovir is phosphorylated to generate cidofovir diphosphate (Supplementary information, Fig. S6c).6,10 Both brincidofovir and cidofovir are reported to possess a broad-spectrum activity against dsDNA viruses and be effective against poxviruses in animal models.6,11 Compared to cidofovir, brincidofovir was made more bioavailable by virtue of an added alkoxy chain. To understand the inhibition mechanism of brincidofovir against MPXV, we modeled cidofovir diphosphate into the MPXV DNA replication machinery based on RB69 DNA polymerase in complex with DNA,12 and superimposed on the dCTP, which shows that cidofovir diphosphate is located perfectly at the active site, competing with the incoming dCTP for interaction with template deoxyguanosine, substrate binding and catalytic residues (Fig. 1h, i). Therefore, cidofovir diphosphate could be incorporated into the newly synthesized DNA and greatly reduce DNA synthesis.6

A B-family DNA polymerase from Thermococcus gorgonarius binds a DNA at the exonuclease site.13 We superimposed the structure of T. gorgonarius DNA polymerase in complex with DNA (PDB code: 2XHB) on MPXV DNA replication machinery structure with a RMSD of 3.45 Å over 404 aligned Cα atoms. Highly positively charged residues Lys174, Arg302, Lys308, Lys340, Lys345, Lys435 and Arg497 in the exonuclease domain are located in the potential DNA binding site (Fig. 1j). Interestingly, the 3′ DNA nucleotide of the newly synthesized DNA colored in forest green is located in the active site of the 3′-5′ exonuclease domain of F8, composed of potential catalytic residues Asp166, Asp462, Glu168, Asp347 and Lys435, and potential DNA binding residues Phe262 and Phe267 in the exonuclease domain (Fig. 1k).

Uracil is incorporated into DNA by misincorporation of dUTP or deamination of cytosine.14 E4 is a homolog of VACV D4 that recognizes uracil base and removes it. The D4 structure in complex with DNA (PDB code: 4YIG) was superimposed on structure of E4. By this, residues Pro71, Lys72, Lys86, Ser88, Lys90, Thr130, Thr161, Tyr180, Arg185, Asp186, and His187 in the N-terminal domain of E4 were identified to closely contact with the DNA (Fig. 1l), while the deoxyribose of the abasic product is located in the active site composed of potential catalytic residues Asp68 and His181, and potential DNA binding residues Tyr70, Phe79 and Arg185 of the uracil-DNA glycosylase (Fig. 1m).

Superimposition of DNA-complexed structures of DNA polymerases of Herpes simplex virus (HSV), bacteriophage RB69 and T. gorgonarius, as well as VACA D4 on MPXV DNA replication machinery shows that the single-stranded template DNAs are extended to a tunnel toward E4 (Supplementary information, Fig. S7a–c) in the DNA-bound MPXV DNA replication machinery (Fig. 1g, n; Supplementary information, Fig. S5a, c). Structural superimposition of the bacteriophage RB69 DNA polymerase in complex with DNA on MPXV DNA replication machinery shows that the newly synthesized DNA strand is at the active site of F8 subunit for DNA polymerization (Fig. 1h, n; Supplementary information, Fig. S7b). Structural alignment of T. gorgonarius DNA polymerase in complex with DNA to MPXV machinery reveals that the DNA is switched from the active site of DNA synthesis to the 3′-5′ exonuclease active site (Fig. 1o; Supplementary information, Fig. S7c), suggesting that the machinery might perform proofreading.

In summary, our work provides the DNA replication machinery structure of poxviruses, laying a foundation for studying the coupling mechanism of DNA replication, proofreading and base excision repair, as well as developing drugs against MPXV and other orthopoxviruses.