The Structure of the Herpes Simplex Virus DNA-Packaging Terminase pUL15 Nuclease Domain Suggests an Evolutionary Lineage among Eukaryotic and Prokaryotic Viruses

Abstract

Herpes simplex virus 1 (HSV-1), the prototypic member of herpesviruses, employs a virally encoded molecular machine called terminase to package the viral double-stranded DNA (dsDNA) genome into a preformed protein shell. The terminase contains a large subunit that is thought to cleave concatemeric viral DNA during the packaging initiation and completion of each packaging cycle and supply energy to the packaging process via ATP hydrolysis. We have determined the X-ray structure of the C-terminal domain of the terminase large-subunit pUL15 (pUL15C) from HSV-1. The structure shows a fold resembling those of bacteriophage terminases, RNase H, integrases, DNA polymerases, and topoisomerases, with an active site clustered with acidic residues. Docking analysis reveals a DNA-binding surface surrounded by flexible loops, indicating considerable conformational changes upon DNA binding. In vitro assay shows that pUL15C possesses non-sequence-specific, Mg²⁺-dependent nuclease activity. These results suggest that pUL15 uses an RNase H-like, metal ion-mediated catalysis mechanism for cleavage of viral concatemeric DNA. The structure reveals extra structural elements in addition to the RNase H-like fold core and variations in local architecture of the nuclease active site, which are conserved in herpesvirus terminases and bear great similarity to the phage T4 gp17 but are distinct from podovirus and siphovirus orthologs and cellular RNase H, delineating a new evolutionary lineage among a large family of eukaryotic viruses and simple and complex prokaryotic viruses.

INTRODUCTION

Herpes simplex virus 1 (HSV-1) is the prototypical member of the herpesvirus family and belongs to the alpha-herpesvirus subfamily (1). HSV-1 infects more than 60% of the United States population and is responsible for oral cold sores and rare but severe encephalitis (1). HSV-1 can establish latency in neurons and can replicate in brains. The virion comprises three major structural elements: a nucleocapsid containing the 152-kbp genome, an envelope consisting of a lipid bilayer with embedded glycoproteins, and a proteinaceous tegument in between (2). The capsid is an icosahedral shell of 1,250-Å diameter enclosing a double-stranded DNA (dsDNA) genome (3).

HSV-1 produces branched concatemeric viral DNA in the nuclei of infected cells. Individual genomes are generated by virally encoded machinery that recognizes packaging signals in concatemeric DNA and endonucleolytically cleaves the DNA within these signals, and the cleavage is tightly coupled to insertion of DNA into a preformed protein shell called procapsid (4–8). The first of these cleavages generates the long component terminus, while the second generates the terminus of the short component and frees the cleaved genome from the concatemer (9). The viral DNA is inserted into procapsid through a unique portal vertex, which is composed of a dodecameric ring of pUL6 protein with an internal diameter wide enough to accommodate passage of dsDNA (10–12). As the DNA is packaged into the capsid, it replaces the internal proteinaceous shell or scaffold, and the outer shell undergoes a dramatic and stabilizing conformational change (13, 14). Capsids containing DNA are termed nucleocapsids or type C capsids (15). In a default reaction that occurs in the absence of terminase components, immature capsids (termed procapsids or large-cored B capsids) undergo the conformational change and retain the internal shell (13, 14). These forms are believed to represent dead-end products and are termed B capsids or small-cored B capsids. In instances in which the scaffold is expelled but DNA is absent, type A capsids are produced. Such capsids are believed to result when the DNA-packaging process initiates but is subsequently aborted and the DNA slips out of the capsid.

DNA packaging of HSV-1 requires seven protein-encoding genes, namely, UL6, UL15, UL17, UL25, UL28, UL32, and UL33, as identified through genetic analysis (7). Mutations within these genes result in defects either in the cleavage of concatemeric DNA or DNA encapsidation (8, 16, 17). Like many tailed dsDNA bacteriophages, genome cleavage and packaging in HSV-1 are tightly coupled and require a terminase enzyme that harbors the DNA recognition, endonuclease, and ATPase activities (7, 18). Unlike bacteriophage terminases that consist of two subunits, evidence suggests that the herpes simplex virus terminase consists of three protein components encoded by the UL15, UL28, and UL33 genes (19, 20). It was reported that the HSV-1 terminase complex assembled in the cytoplasm, eventually interacting with the portal vertex of the procapsid in infected cell nuclei (21). Trafficking of HSV-1 terminase proteins into nuclei depends on a nuclear localization signal in the product of the UL15 gene (pUL15) (21). pUL28 has been shown to bind viral DNA-packaging sequences (22), while pUL33 associates with pUL28 and enhances the interaction between the pUL28-pUL33 complex and pUL15 (20).

The UL15 gene is composed of two exons separated by an intron of 3,587 bp, and it encodes a 735-residue protein (23, 24). The UL15 gene family of herpesviruses predicts a highly conserved ATPase motif that, at least in the herpes simplex virus, is required for viral DNA cleavage and packaging (25, 26). The structure of the C-terminal domain of the pUL15 ortholog (designated pUL89C) from human cytomegalovirus (HCMV), a beta-herpesvirus, exhibits an RNase H-like nuclease domain fold that is responsible for cleavage of concatemeric DNA into unit-length genomes (27, 28). Here, we report the 2.46-Å resolution crystal structure of the pUL15 C-terminal domain (pUL15C) arranged as a novel trimer. The structure shows a fold resembling those of RNase H, integrases, DNA polymerases, and topoisomerases, indicating that pUL15 utilizes a similar metal ion-mediated catalytic mechanism. Docking analysis enabled by better-defined surface loops shows a putative DNA-binding surface comprising numerous positively charged residues. Structural comparison with viral terminase nuclease domains and RNase H-like nucleases reveals conserved and variable features in the fold cores and the nuclease active sites, suggesting an evolutionary lineage among eukaryotic and prokaryotic viruses. As HSV-1 is the prototype of herpesviruses and much has been known regarding cis-acting elements on the viral genome required for DNA packaging and composition and cellular localization of the terminase components, the pUL15C structure can enable further studies to understand assembly of the HSV-1 terminase and the DNA-packaging mechanisms in herpesviruses.

MATERIALS AND METHODS

Protein expression and purification.

The pUL15C gene encoding residues 471 to 735 was cloned into the pET28b (Novagen, Madison, WI) expression vector between the NdeI and XhoI restriction sites, resulting in pUL15C with an N-terminal His tag. The protein was expressed in Escherichia coli B834(DE3) cells overnight at 15°C to an optical density at 600 nm (OD₆₀₀) of 2.2. Protein expression was induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to a final concentration of 1.0 mM at an OD₆₀₀ of 0.59, and the cells were kept to grow overnight. Cells were harvested at 5,000 rpm, resuspended in a buffer containing 20 mM Tris-HCl (pH 8.5), 500 mM NaCl, and 10 mM 2-mercaptoethanol and sonicated. Insoluble materials were sedimented by centrifugation (15,000 rpm, 4°C, 60 min), and the supernatant was passed through a 0.45-μm-pore-size filter. Proteins were purified by Ni²⁺ affinity chromatography followed by size exclusion chromatography on a Sephacryl S-300 column (GE Healthcare) equilibrated in 20 mM Tris-HCl (pH 8.5), 150 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol (DTT).

Crystallization and data collection.

The proteins were concentrated using centrifugal filters with a molecular weight cutoff of 10 kDa to 10 mg/ml in 20 mM Tris-HCl (pH 8.5), 150 mM NaCl, 1 mM EDTA, and 1 mM DTT. Native crystals of pUL15C were obtained at 20°C by vapor diffusion in hanging drops containing equal volumes of the protein solution and a reservoir solution with 1 M ammonium citrate and 0.1 M sodium acetate at pH 4.6. The crystals were flash-cooled in 15% polyethylene glycol 400 as the cryoprotectant. The X-ray data were collected at 100 K at the Advanced Photon Source (APS) beamlines GM-CA/CAT 23ID-B and 23ID-D and at the Stanford Synchrotron Radiation Lightsource (SSRL). The data for final structure refinement were collected at SSRL beamline BL 11-1 and were processed with the HKL2000 (29). X-ray data processing statistics are summarized in Table 1.

Table 1.

Data collection and refinement statistics

Statisticb	Valuea
Data collection
Beamline	SSRL BL 11-1
Wavelength (Å)	0.97945
Resolution (Å)	50–2.46 (2.50–2.46)
No. of measurements	421,384
No. of unique reflections	34,432 (1,635)
% completeness	99.6 (95.2)
I/σ	42.9 (2.6)
% Rmerge	8.6 (48.9)
Space group	P43212
Unit cell (Å)	a = 96.9, c = 194.0
Refinement
Resolution (Å)	20–2.46 (2.53–2.46)
No. of reflections	34,293
% Rwork	20.8 (26.4)
% Rfree	24.5 (30.6)
RMS deviation
Bond length (Å)	0.01
Bond angle (°)	1.42
Chiral vol (Å3)	0.096
Mean B values (Å2)	55.7
Ramachandran plot
Most favorable regions (%)	93.4
Allowed region (%)	5.8
Outlier (%)	0.8

^aValues in parentheses are for the outmost resolution shells.

^bR_merge = Σ_hklΣ_i|I_i(hkl) − <I(hkl)>|/Σ_hklΣ_iI_i(hkl), where I_i(hkl) is the observed intensity of reflection hkl and <I(hkl)> is the averaged intensity of symmetry-equivalent measurements; R_work = Σ||Fobs| − |Fcal||/Σ|Fobs|; R_free has the same formula as R_work, except that calculation was made with the structure factors from the test set representing 5% of all data.

Structure determination.

The structure was determined by molecular replacement with the structure of the HCMV pUL89 C-terminal nuclease domain (47% sequence identity) as the initial search model using the program Phaser incorporated in PHENIX (30). The structure was first determined with X-ray data collected at APS and was later refined with higher-resolution data collected at SSRL. Crystal belonged to space group P4₃2₁2, with three molecules in the asymmetric unit, and the cell dimensions are a = 96.9 and c = 194.0. The structure was built manually using the program COOT (31), and the refinement was performed with the program PHENIX using noncrystallographic symmetry restraints. The model was improved by alternating cycles of refinement. The final refinement cycles included TLS refinement. Refinement statistics are summarized in Table 1. In the final refined structure, residues 476 to 732 were modeled in A, B, and C molecules. Residues 512 to 519 and 544 to 547 were invisible in B and C molecules, and residues 604 to 613 where invisible in all the three chains. Residues 686 to 704 were invisible in A and B molecules, but partial density at a 0.70 sigma cutoff were found in C molecules, and alanine residues were added from 686 to 688 and 693 to 698. In the pUL15C DNA-binding model, a model for the disordered residues in loop L4 (residues 686 to 704) was built based on the C molecule using the COOT program.

Nuclease activity assay.

The purified pUL15C at various concentrations were incubated with ∼400 ng of supercoiled plasmid DNA (pET20b; 3,716 bp) in a reaction solution containing 20 mM Tris-HCl (pH 7.8), 10 mM NaCl, and 1.0 mM MgCl₂ for 1 h at room temperature. The reactions were terminated by the addition of EDTA to a final concentration of 50 mM. To analyze the effects of EDTA and MgCl₂ on pUL15C nuclease activity, different concentrations of pUL15C were incubated with the same plasmid DNA in the presence of 5 mM EDTA or 5 mM MgCl₂ in a 10-μl reaction mixture containing 20 mM Tris-HCl (pH 7.8), 100 mM NaCl, and 1 mM DTT. The reactions were terminated as described above and analyzed by 1.0% (wt/vol) agarose gel electrophoresis followed by ethidium bromide staining.

Protein structure accession number.

The coordinates and reflection data have been deposited with RCSB Protein Data Bank with the accession code of 4IOX.

RESULTS AND DISCUSSION

Overall structure of pUL15C.

The pUL15C crystallizes with three molecules per crystallographic asymmetric unit (Fig. 1A; Table 1). The root mean square (RMS) deviations after pairwise superposition are 0.45 Å, 0.50 Å, and 0.61 Å for 211, 212, and 214 equivalent C-α, respectively, indicating that they are essentially identical. The A molecule is more complete compared to the other two molecules. The Phe512-Gly520 loop (L1) and Lys543-Gly548 loop (L2) are well defined in the electron density map for the A molecule but are missing in B and C molecules (Fig. 1B). The Lys604-Ser613 loop is invisible in the electron density map for all three molecules and thus is disordered. The loop Glu686-Ala704 (L4) is disordered in A and B molecules but displays weak main-chain electron density for C molecule and thus is modeled as poly-alanine between 686 to 688 and 693 to 698 in the C molecule. These loops are located in close proximity to the putative nuclease active site and presumably make contact with the DNA substrate, indicating that these loops are flexible and may become ordered upon DNA binding (Fig. 1B). The three molecules are arranged about a noncrystallographic, approximate 3-fold rotational symmetry axis, and the pairwise rotations between molecules are 119.50 Å, 120.02 Å, and 120.51 Å, respectively. The buried solvent accessible surface areas are 639.3, 871.9, and 974.5 A², consistent with slight discrepancy from a proper 3-fold symmetry. Intermolecular interactions involve the N-terminal (α1), C-terminal, α2-β4, and β6-α5 regions, in which loop 3 (residues 665 to 672) plays a major role. There is a channel at the center of the crystallographic trimer that is of ∼10 Å in diameter and is unlikely for passage of DNA.

Fig 1.

The overall structure of pUL15C. (A) Ribbon representation of the pUL15C trimer viewed down the noncrystallographic 3-fold axis. A, B, and C molecules are shown in green, blue, and pink, respectively. Conserved acidic residues in the active site, D509, E581, D706, and D707, are labeled. (B) Overall structure of pUL15C in a ribbon representation. The α-helices, β-strands, and loops are in green, pink, and yellow, respectively. The loops L1, L2, L3, and L4 and the N terminus are indicated. (C) The electrostatic potential surface of pUL15C in the same view as shown in panel B. The positive potential is shown in blue, whereas the negative potential is in red.

The pUL15C exists mainly as a monomer in solution, although a very small fraction that corresponds to a potential trimer is observed (data not shown). The biological implication of the pUL15C crystallographic trimer is unclear. Terminase large subunits of phages T4 and phi29 assemble into ringlike pentamers upon binding to the procapsid (32, 33), and gpA of phage lambda forms a tetramer in complex with the terminase small-subunit gpNu1 (34). The pUL89 nuclease domain structure displays a dimer of dimers, and the intermolecular interactions involve regions different from those in pUL15C and is mediated through the central β-sheet (27). Nevertheless, it is interesting to note that the active sites in the crystallographic trimer of pUL15C all face outside (Fig. 1A), making them accessible for potential DNA binding and cleavage.

The pUL15C belongs to the RNase H-like endonucleases and polynucleotidyl transferases.

The pUL15C molecule measures about 48 by 47 by 42 ų and consists of a seven-stranded β-sheet, with parallel and antiparallel strands sandwiched between six α-helices (Fig. 1B). The central β-sheet (β1 to β5) curves around the helix α6, which is situated on the concave face of the central β-sheet. Negatively and positively charged areas disperse along the surface of pUL15C (Fig. 1C). The active-site groove is largely negatively charged, consistent with recruiting of metal ions for DNA binding and cleavage. The core of pUL15C, that is, the central β-sheet (β1 to β5) surrounded by α-helices, exhibits a characteristic fold similar to those of the RNase H-like superfamily of nucleases and polynucleotidyl transferases (35), despite the lack of an apparent amino acid sequence identity (Fig. 2). Structural superposition shows RMS deviations of 2.8 Å for 80 equivalent C-α with 9% identity with RNase H (36), 3.0 Å for 104 equivalent C-α with 7% identity with the Holliday junction resolvase RuvC (37), 3.0 Å for 79 equivalent C-α with 5% identity with RNase H1 (38) (Fig. 2A), and 3.0 Å for 90 equivalent C-α with 4.4% identity with the Tn5 transposase (39) (Table 2). The core folds in these proteins display a conserved 5-stranded mixed β-sheet arranged as β5-β4-β1-β2-β3, with four parallel strands and an antiparallel β2.

Fig 2.

Structural comparison of pUL15C with RNase H family nucleases and viral terminase nuclease domains. The pUL15C (green) is superposed with RNase H1 (A; PDB code 2QKB), phage SPP1 (B; PDB code 2WC9), phage T4 gp17 (C; PDB code 3CPE), and pUL89C (D; PDB code 3N4P). In all panels, the other protein is in gray. In panels C and D, the right views are 90° from the left ones. The pUL15C active-site residues are shown as stick models for side chains and are labeled in panel A. The β7 of pUL15C is shown in blue and the hairpin 620 to 633 is shown in pink, both of which are indicated with an arrow. In panel D, the C-terminal region encompassing residues 720 to 732 of pUL15C and the corresponding region of pUL89C are shown in red and cyan, respectively. The amino and carboxyl termini are indicated with N and C, respectively, in panels C and D.

Table 2.

Structural superposition of pUL15C with pUL89C, RNase H family nucleases, and bacteriophage terminases

Protein family	PDB ID	No. of amino acid	No. of equivalent C-α atoms	RMS deviation (Å)
Herpesvirus
pUL89C	3N4P	216	202	1.3
RNase H
RNase H	1ZBI	135	80	2.8
RNase H1	2QKB	143	79	3.0
RuvC resolvase	1HJR	131	104	3.0
Tn5 transposase	1MM8	164	90	3.0
Bacteriophage
T4 gp17	3CPE	201	157	2.7
RB49 gp17	3C6H	198	162	2.8
SPP1 G2P	2WBN	178	149	2.9
P22 gp2	4DKW	194	145	2.4

The pUL15C structure is readily superimposable with those of phage terminase large-subunit nuclease domains despite <15% sequence identity, which shows RMS deviations of 2.8 Å for 162 equivalent C-α with 12% identity for RB49 gp17, 2.7 Å for 157 equivalent C-α with 11% identity for T4 gp17, 2.9 Å for 149 equivalent C-α with 2% identity for SPP1 G2P, and 2.4 Å for 145 equivalent C-α with 9% identity for P22 gp2 (Table 2). The pUL15C structure superimposes well onto the HCMV pUL89C, with RMS deviation of 1.3 Å for C-α atoms (Fig. 2D). The surface loops L1 and L2 are observed in pUL15C, which were invisible in HCMV pUL89C. These loops are located next to the nuclease active site, indicating that they may undergo conformational changes upon DNA binding. A major difference between the pUL15C and HCMV pUL89C structures is that the C-terminal region encompassing residues 720 to 732 of pUL15C adopts a remarkably different conformation, resulting in a 31-Å distance between C-α atoms of visible C-terminal residues of the two proteins (Fig. 2D). In pUL15C, the C-terminal region is involved in intermolecular interactions in the crystallographic trimer (Fig. 1A). The C-terminal portions of terminase large subunits of phages T4, T3, and lambda were suggested to interact with the portal proteins (40–43), although the C-terminal nuclease domain of phage T4 gp17 was positioned distal to the portal in an assembly model based on low-resolution electron cryo-microscopic studies (32). The C-terminal portions in phage terminase large subunits were disordered and invisible in X-ray structures (32, 44, 45). These observations in herpesviruses and phages suggest that the C-terminal regions of terminase large subunits are mobile and may adopt defined conformations upon interactions with other components of the viral DNA-packaging machinery, such as the portal proteins.

Structural elements in pUL15C exhibit variations to phage orthologs.

Comparative structural analysis reveals two structural elements in pUL15C that exhibit variations compared to phage orthologs and cellular RNase H1 (arrows in Fig. 2). The first structural element is an extended surface hairpin structure formed by residues 620 to 633, which follows β5 in the RNase H-like fold core. This hairpin is absent in RNase H1 and SPP1 G2P (45) (Fig. 2B) and P22 gp2 (44), but a similar hairpin structure, albeit shorter, is observed in phage T4 gp17 (Fig. 2C). This hairpin is in proximity to DNA in our DNA-binding model (see below) and thus presumably helps properly position the DNA for cleavage. The second structural element is the segment consisting of residues 661 to 665, which forms a β-strand (β7), adding an extra parallel strand to β3 of the RNase H-like fold core. Like the hairpin 620 to 633 in pUL15C, this region is absent in RNase H1, SPP1 G2P, and P22 gp2. However, a similar β-strand is present in phage T4 gp17 but adopts an opposite orientation (Fig. 2C). Interestingly, segment β7 is located between α4 and α5 in pUL15C, but the corresponding region in phage T4 gp17 is located immediately downstream to the α-helix at the N-terminal proximity of the nuclease domain (corresponding to helix α1 in pUL15C). In the T4 gp17 nuclease domain, the N-terminal α-helix is followed by that β segment, which then extends to β1 of the RNase H-like fold core. In comparison, in pUL15C, the N-terminal helix α1 is followed by a loop that extends to β1 of the RNase H-like fold core, and that β-strand is located between α4 and α5. This can be viewed as if this β7 segment jumped from between the N-terminal helix α1 and β1 in the case of T4 gp17 to a position between helices α4 and α5 in the case of pUL15C, which might result from transferring of the genetic fragments that encode those polypeptides. These results indicate that the terminase large-subunit nuclease domains of herpesviruses and myoviruses are closely related but are only remotely related to cellular RNase H family nucleases, while siphovirus and podovirus orthologs are closer relatives of RNase H family nucleases.

The nuclease active site.

The active site is located at one side of the central five-stranded β-sheet, forming a cleft region clustered with acidic residues (Fig. 1B; Fig. 3). The local structure of the pUL15 nuclease active site resembles those in HCMV pUL89, phage terminase large subunits, and the RNase H family nucleases, while variations are clearly observed (Fig. 3). Three acidic residues, Asp509, Glu581, and Asp707, in pUL15C form the conserved triad in the active site, corresponding to residues Asp463, Glu534, and Asp651 in HCMV pUL89C, respectively. Metal ions, usually Mg²⁺, are required for the functioning of RNase H family nucleases, and a two-metal-ion catalysis mechanism has been proposed (35, 46, 47). Two Mn²⁺ ions were observed in the HCMV pUL89C structure, the first coordinated by Asp463 and Glu534 and the second by Asp463 and Asp651 (27). No Mg²⁺ ion density was observed in the nuclease active center of pUL15C, but water molecules were observed. These acidic residues are strictly conserved among human herpesvirus pUL15 orthologs (data not shown), suggesting that they are essential for nucleolytic catalysis. These acidic residues generate a strongly electronegative environment in the active site, consistent with the metal ion-mediated mechanism for DNA binding and cleavage.

Fig 3.

Comparison of the pUL15C active-site region with those of HCMV pUL89C (A), phage terminases (B), and RNase H family nucleases (C) showing conserved features and structural variations. Conserved residues in the active sites are labeled with side chains shown as stick models. Metal ions in RNase H family nucleases are shown as green spheres. All structures are in the same view after superimposition.

In RNase H1, a conserved acidic residue triad (D210-D145-D274) plus a fourth acidic residue (E186) located on the fold core is responsible for metal ion-mediated nucleolytic catalysis (38) (Fig. 3C). In particular, the D210-D145-D274 triad helps coordinate an Mg²⁺ ion, which is presumed to form the nucleophile, while D210, D145, and E186 help coordinate a second Mg²⁺ ion, which is thought to stabilize the transition state (35, 47). Structural comparison of pUL15C, phage terminases, and RNase H family nucleases shows that the active-site residues are readily superimposable, indicating conserved catalytic mechanisms. The central residue of the triad (D509 in pUL15C, D401 in T4 gp17, D266 in SPP1 G2P, and D145 in RNase H1) is invariably an aspartic acid, indicating a central role of this residue in catalysis and strict conservation during evolution. However, interesting structural variations are observed in other parts of the active site. In Siphovirus SPP1 G2P, a catalytic triad comprised of residues D321, D266, and D403 superimposes well with that of RNase H1, while the fourth acidic residue corresponding to E186 in RNase H1 is missing and, instead, residue H400 on the opposite side of the catalytic triad participates in coordination of a second metal ion (45). In phage T4 gp17, further variations are observed: (i) the residue corresponding to the first one of the RNase H1 catalytic triad is a glutamic acid (E458) instead of an aspartic acid; (ii) the residue corresponding to the third aspartic acid of the RNase H1 catalytic triad is a methionine (M545) that is conserved in a related phage RB49; and (iii) an additional acidic residue D542 was proved to be essential for catalysis, may participate in coordination of a metal ion (32), and occupies a position similar to that of residue H400 in SPP1 G2P. The local architecture of the pUL15C active site closely resembles that of T4 gp17, in which a valine (V710) occupies the same position as residue M545 in T4 gp17. Position 710 is invariably a hydrophobic residue in human herpesviruses (data not shown). Based on these analyses, it becomes clear that there is a trajectory of accumulative structural variations in the nuclease active sites, which changes from (E186)-D210-D145-D274 in RNase H1 to D321-D266-D403-(H400) in SPP1 G2P, then to E458-D401-(M545)-D542 in T4 gp17, and finally to E581-D509-(V710)-D707 in HSV-1 pUL15. HSV-1 pUL15 and T4 gp17 are closely related to each other but are remotely related to SPP1 G2P and RNase H1.

The pUL15c shows metal ion-dependent, non-sequence-specific nuclease activity.

In vitro assay shows that pUL15C possesses non-sequence-specific nuclease activity (Fig. 4A). Both nicked and linearized dsDNA are observed as reaction products. At lower protein concentrations, there is an accumulation of nicked DNA, while the fraction of linearized DNA increases when the protein concentration is higher (Fig. 4A). At higher protein concentrations, smeared DNA appears, representing cleavage of DNA into variable, smaller-size fragments. Similar behavior was observed for HCMV pUL89C (27). These data are implicative of a stepwise course of the nucleolytic catalysis, that is, pUL15C may bind to DNA and make a cut on one strand of the dsDNA before cutting the other strand, thus distinguishing pUL15C from dimeric nucleases, such as restriction endonucleases that cooperatively cut the two strands of dsDNA. In vivo cleavage of HSV-1 genomic DNA by pUL15 occurs within the DR1 motif, generating a 3′ single-base overhang (48). Cleavage of the two DNA strands likely requires two copies of the pUL15 molecules. As pUL15C exists as a monomer in solution, molecular determinants for oligomerization of pUL15C may be located elsewhere in the HSV-1 DNA-packaging machinery or in the N-terminal domain of pUL15. EDTA eliminates the nuclease activity of pUL15C, whereas Mg²⁺ shows stimulatory effect (Fig. 4B). These results indicate that the pUL15C nuclease activity is dependent on metal ions, consistent with structural homology to RNase H family nucleases.

Fig 4.

In vitro nuclease activity assay. (A) Non-sequence-specific nuclease activity of pUL15C. Lane 1, 1 kb plus DNA ladder (Invitrogen); lane 2, supercoiled plasmid DNA; lane 3, plasmid DNA linearized with restriction enzyme; lanes 4 to 7, pUL15C at a concentration of 17.5 μM, 35 μM, 70 μM, and 140 μM, respectively, was incubated with supercoiled plasmid DNA at room temperature for 1 h. Positions for the nicked, linearized, and supercoiled DNA are indicated. (B) Effects of EDTA or MgCl₂ on pUL15C nuclease activity. Lane 1, 1 kb plus DNA ladder (Invitrogen); lane 2, plasmid DNA linearized with restriction enzyme; lane 3, supercoiled plasmid DNA; lanes 4 and 5, pUL15C at 35 μM and 70 μM, respectively, was incubated with supercoiled plasmid DNA in the presence of 5 mM EDTA; lanes 6 to 9, pUL15C at 17.5 μM, 35 μM, 70 μM, and 140 μM, respectively, was incubated with supercoiled plasmid DNA in the presence of 5 mM MgCl₂.

The DNA binding mode of pUL15.

To understand how DNA binds to pUL15C for cleavage, we modeled dsDNA on the pUL15C nuclease active site by superimposing the pUL15C structure onto that of the RNase H1 in complex with a DNA-RNA hybrid, which was then replaced with dsDNA (38) (Fig. 5). In this model, the dsDNA spans the nuclease active site, approaching the catalytic active residues. Loops L1 and L4 make contact with the DNA, in which positively charged residues R517, R695, K700, and R701 may interact with the DNA phosphate backbones. Additional positively charged residues K638 and K640 may also contribute to DNA binding. The hairpin (residues 620 to 633) also appears to interact with DNA, presumably helping hold the DNA in an appropriate position and orientation with respect to the nuclease active site. The loop L4 is located at a position similar to that of a β-hairpin structure in the SPP1 G2P nuclease domain (45) and in T4 gp17 (32). In the SPP1 G2P structure, it was proposed that the conformational flexibility of this β-hairpin might mediate DNA binding and thus the nuclease activity. In a T4-related phage, RB49 gp17, this β-hairpin is disordered (32). While it is ordered in the T4 gp17 structure (32), this β-hairpin also showed flexibility and was proposed to interact with DNA in a later molecular dynamics study of T4 gp17 (49). In the pUL15C structure, the loop L4 is disordered in the A and B molecules but is partially ordered in the C molecule, indicative of conformational flexibility. In our DNA binding model, the loop L4 is adjacent to DNA and the positively charged residues R695, K700, and R701 are in close proximity from DNA to make direct interactions. Additionally, the loop L1 is well defined in the electron density map for the A molecule but is disordered in B and C molecules, suggesting flexibility of this loop and induced fit upon DNA binding.

Fig 5.

A DNA-binding model for pUL15C. A model of 25-bp dsDNA docked onto the pUL15C structure. The docking was performed by superimposing the pUL15C structure onto that (PDB code 2QKB) of the RNase H1 in complex with a DNA-RNA hybrid, which was replaced with dsDNA. The loops L1 and L4 and the hairpin 620 to 633 are indicated. The active-site residues are shown as stick models for side chains and are labeled with D509, E581, D706, and D707. The positively charged residues that potentially interact with DNA are shown as blue stick models for side chains and are labeled.

An evolutionary lineage among the herpesviruses and bacteriophages.

Our comparative structural analysis reveals two structural elements in pUL15C that are in addition to the RNase H-like fold core, that is, the hairpin formed by residues 620 to 633 and β7 (Fig. 2). Both structural elements show close similarity between pUL15 and phage T4 gp17 but are absent in terminase large subunits of simple phages such as SPP1 and P22 and RNase H family nucleases. Likewise, a trajectory of accumulative structural variations in the nuclease active sites is observed from the (E186)-D210-D145-D274 motif in RNase H1 to D321-D266-D403-(H400) in SPP1 G2P to E458-D401-(M545)-D542 in T4 gp17 to E581-D509-(V710)-D707 in HSV-1 pUL15. These structural variations of the protein folds and the local architectures of the nuclease active sites consistently indicate a relationship among these proteins, that is, the terminase large-subunit nuclease domains of herpesviruses and myoviruses are closely related to each other but are only remotely related to cellular RNase H family nucleases, while siphovirus and podovirus orthologs are closer relatives of RNase H family nucleases. Thus, an evolutionary lineage can be delineated based on the structures of the viral DNA-packaging nuclease modules, which might originate from cellular RNase H family nucleases by, for example, gene acquisition, and first evolve to simple phages such as podoviruses and siphoviruses and then to complex phages such as myoviruses and to herpesviruses. Evolutionary relationships between herpesviruses and tailed dsDNA bacteriophages have been proposed based on similarity of the virus assembly pathways, ringlike dodecameric portal proteins, and folds of major capsid proteins (50, 51). Genome packaging is among the most fundamental processes in virus life cycles and may represent an evolutionarily ancient event. Thus, the relationship observed in our comparative structural analysis of pUL15 and phage terminases and the lineage revealed thereby provide new, strong evidence for an evolutionary path from simple phages such as podoviruses and siphoviruses to complex phages such as myoviruses and then to eukaryotic herpesviruses. It is worth pointing out that amino acid sequence identity of proteins between herpesviruses and tailed dsDNA phages is usually too low to detect apparent homologies, and comparative analysis based on three-dimensional structures can provide exceptionally valuable information for understanding evolutionary relationships among viruses.