Structures of the portal vertex reveal essential protein-protein interactions for Herpesvirus assembly and maturation

Herpesviridae is a large family of double-stranded DNA (dsDNA) viruses that cause a variety of human diseases ranging from cold sores and chicken pox to congenital defects, blindness and cancer (Chayavichitsilp et al., 2009; Wang et al., 2018). In the past 70 years, substantial advances in our knowledge of the molecular biology of herpesviruses have led to insights into disease pathogenesis and management. However, the mechanism for capsid assembly that requires the ordered packing of about 4,000 protein subunits into the hexons, pentons and triplexes remains elusive. It is still a puzzle how initially identical subunits adopt both hexameric and pentameric conformations in the capsid and select the correct locations needed to form closed shells of the proper size. Biochemical and genetic studies have shown that the portal is involved in initiation of capsid assembly (Newcomb et al., 2005) and functions akin to a DNA-sensor coupling genome-packaging achieved by a genome-packaging machinery—“terminase complex” (Chen et al., 2020; Yunxiang Yang, 2020) with icosahedral capsid maturation (Lokareddy et al., 2017). Structural investigations of the herpesvirus portal have proven challenging due to the small size of this dodecamer, which accounts for less than 1% of the total mass of the capsid protein layer and the technical difficulties involved in resolving non-icosahedral components of such large icosahedral viruses (diameter is ∼1,250 Å). Efforts of many investigators over two decades have made to reconstruct the cryo-electron microscopy (cryo-EM) structure of herpesvirus portal vertex and more recently near-atomic structures of two herpesvirus (herpes simplex virus type 1 (HSV-1) and Kaposi’s sarcoma-associated herpesvirus (KSHV)) portal vertices were reported (McElwee et al., 2018; Gong et al., 2019; Liu et al., 2019). Here we show asymmetric reconstructions of herpes simplex virus 2 (HSV-2) B(containing scaffold proteins inside, putative capsid assembly intermediate), C(DNA-filled capsid, derived from the enveloped virion by using detergent treatment to remove viral membrane proteins and outer tegument proteins) and virion-capsids (enveloped intact virion) at 8.1, 9.0 and 10.2 Å respectively (Figs. S1–3). The asymmetric reconstructions reveal a unique portal vertex in the context of the icosahedral capsid, including 1 portal, 11 pentons, 3 types of hexons (P, peripentonal; E, edge; C, center) with the hexameric rings formed by VP26s, 320 triplexes (Ta-Tf) and 12 pentagram-shaped capsid vertex specific component (CVSC) densities (composed of a UL17 monomer, a UL25 dimer and a UL36 dimer, presence in Cand virion-capsids) (Fig. 1A–C). The validation of our asymmetric reconstructions is based on previous studies in which several herpesvirus capsid structures were determined at near-atomic resolution by cryo-EM and icosahedral averaging (Yu et al., 2017; Jialing Wang, 2018; Yuan et al., 2018). Such averaging ignores the fact that one pentameric vertex is occupied by the portal vertex in the capsid and eliminates the density for any non-icosahedral structural features. Central slices through the three asymmetric reconstructions show the features of the portal vertex (Fig. 1A). The presence of scaffold proteins, some of which possibly associate with the portal, is observed in the B-capsid. Despite distinct organizations of the CVSC at the portal and penton vertices, the packaged dsDNA is visible in the Cand virion-capsids (Fig. 1A). Radially colored representations of the three reconstructions reveal the dodecameric portal situated beneath one of the pentameric vertices; but, the density surrounding the portal displays a five-fold symmetrical assembly, highlighting the symmetry-mismatch structural features (Fig. 1B). In contrast to B-capsid, small tail-like density, previously named Portal Vertex Associated


ONLINE METHODS
Virus culture and purification HSV-2 virus genotype MS was propagated in Vero cells at a multiplicity of infection (MOI) of 0.2 at 37 °C. Capsid and virion production and purification have been described previously (Jialing Wang, 2018;Wang et al., 2012;Wang et al., 2015;Yuan et al., 2018).

Negative stain
Purified B-, C-capsids and virions were diluted in PBS (pH 7.4) and concentrated to a suitable concentration. A 3-μL aliquot of purified capsids was added to a freshly glow-discharged grid, washed twice with PBS, and strained with phosphotungstic acid (pH 7). All samples were checked on a 120 KV electron microscope.

Cryo-EM and data collection
For cryo-grid preparation of three types of capsids, a 3-μL aliquot of capsids was applied to a fresh glow-discharged copper grid. Grids were blotted for 3.5s in 80% relative humidity for plunge-freezing (Vitrobot; FEI) in liquid ethane. Titan Krios microscope with Falcon3 detector was used to collect data. Movies were recorded as 25 frames at a total does of 25 e − Å −2 and defocus range from 0.8 to 2.3 μm. The magnification under these conditions was 59,000 x, which yielded a final pixel size of 1.38 Å.

Data Processing
A total of 5,567 micrographs for B-capsid; 8,413 micrographs for C-capsid and 12,730 micrographs for virion were recorded. Frames 3-22 were used and corrected for beam-induced drift by aligning and averaging the individual frames of each movie using MOTIONCORR2 (Zheng et al., 2017). The contrast transfer function (CTF) parameters for drift corrected micrographs were estimated by Gctf (Zhang, 2016). Particles were picked manually by EMAN2 package. A total of 41,956 particles from 4,600 micrographs for B-capsid, 56,901 particles from 7,023 micrographs for Ccapsid and 29,221 particles from 12,700 micrographs for virion were selected for the two-dimensional alignment and three-dimensional reconstruction using the blockbased reconstruction (Wang et al., 2019;Wang et al., 2017a;Yuan et al., 2018;Zhu et al., 2018a). Briefly, in our case, the parameters of the icosahedral orientation and the center of each particle determined by Relion were used to guide extraction of components of the blocks (~50% bigger than the penton) to refine and reconstruct separately with their local mean defoci. In each boxed cryo-EM image, there were 12 icosahedral-symmetry related copies (440,844 blocks in total for B-capsids; 682,812 blocks in total for C-capsids and 191,856 blocks in total for virion-capsids) for the blocks. After knowing the rotation and translation parameters of the virus, the distance d between the center of one copy in the 3D virus and the center of the virus along the Z axis (parallel to the incident electron beam) was calculated to solve the gradient in defocus through the capsid. This local defocus of each copy instead of the uniform defocus obtained by fitting was used to reconstruct the vertices. After refinement and reconstruction of the blocks in Relion (Scheres, 2012), two cycles of 3D classification with no-alignment were used to enrich the portal vertices from the blocks for B-, C-and virion-capsids. For the B-capsid, two cycles of 50 iterations of 3D classifications using a ball with a diameter of 250 Å as a mask were performed. ~80% of blocks which are not likely to contain the portal vertices could be separated from the first cycle of 3D classification and the rest (~20%) blocks from one major class were then used for the second cycle of 3D classification with 3 classes. After that, 22,158 blocks (~5%) from one class exhibiting the portal features were subjected to one cycle of automatic 3D refinement with C5 symmetry, which yielded a reconstruction for the B-capsid portal vertex at 4.05 Å resolution. For the C-and virion-capsids, we applied a similar strategy to that of the B-capsids, except a column mask with a diameter of 120 Å and height of 300 Å was used. ~35,698 blocks (5.2%) and ~12,001 blocks (6.2%) with the portal features were enriched and subjected to 3D refinement with C5 symmetry, leading to a 4.5 Å and 5.36 Å resolution density map for the C-and virion-capsid portal vertex, respectively.
All Euler angles parameters of selected blocks were applied to the bin2 particles. For an icosahedral virus, each vertex has a C5 symmetry. The 5 equivalent orientations of the vertex can be obtained by common-line algorithm (Liu and Cheng, 2015;Wang et al., 2017b;Zhu et al., 2018b). We assumed that the portal structure has a fixed orientation related to the five-fold symmetric vertex. The correct orientation of the portal can be determined by searching for the 5 equivalent orientations of the vertex. We randomly chose one of the 5 equivalent orientations of its vertex as the initial orientation for each raw particle image. Then the following 3 steps were performed to determine the exact orientation information. (1) We reconstructed a full particle structure (3D model) using the raw particle images.
(2) For each image, we projected the 3D model to generate 5 projection images according to the 5 equivalent orientations of the vertex. Meanwhile, we boxed out images (160 × 160 pixel) around the vertex of the portal to calculate the correlation between corresponding original image and projection images. We searched the best matched projection image with the original images. (3) We performed iterations of step (1) and (2) until the orientations of individual original images were stable and no further improvement of the portal structure could be obtained. For the B-capsid, we performed a reconstruction with C1 symmetry when the iteration came to the 20 th cycle. We obtained a portal structure with a C12 symmetry. The C-and virion-capsids portal structures with a C12 symmetry were observed at the 34 th cycle of iteration. Finally, asymmetric reconstructions of the B-, C-and virion-capsids were obtained at 8.07 Å, 9.03 Å and 10.19 Å, respectively. The overall resolution for all reconstructions were evaluated based on gold-standard Fourier shell correlation (FSC) = 0.143 criterion (Scheres, 2012).

Model building and refinement
The structures of P-Hex and triplexes of HSV-2 B-capsid (PDB code: 5ZAP) (Yuan et al., 2018) and P-Hex, triplexes and CVSC of HSV-2 C-capsid (PDB code: 5ZZ8) (Jialing Wang, 2018) were initially fitted into the five-fold averaged EM maps of the B-, C-capsid and viron portal vertices, respectively, with CHIMERA and further corrected manually by real-space refinement in COOT. The poly-alanine model of the five coiled coils was built de novo into density using COOT. These models were further refined by positional and B-factor refinement in real space using Phenix and rebuilt in COOT iteratively. Only the coordinates were refined keeping the maps constant. The data set and refinement statistics are summarized in Table S1.

Data availability
The atomic coordinates of the portal vertex from HSV-2 B-, C-and virion-capsids have been submitted to Protein Data Bank with accession numbers PDB: 6M6I, 6M6H and 6M6G, respectively. Cryo-EM density maps of asymmetrically reconstructed HSV-2 B-, C-and virion-capsids and five-fold symmetrically reconstructed portal vertices from HSV-2 B-, C-and virion-capsids have been deposited with the Electron Microscopy Data Bank: EMD-30120, EMD-30122, EMD-30121, EMD-30125, EMD-30124, and EMD-30123 respectively. The data that support the findings of this study are available from the corresponding author upon request.

Fig. S3 Density maps and atomic models of the portal vertices from B-(A), C-(B) and virion-capsids (C).
The polypeptide backbones, many bulky side chains, and extensive subunit contacts are revealed clearly by the resulting density maps.

Fig. S4 Ribbon diagram representation of the atomic models of the asymmetric unit of B-(A), C-(B) and virion-capsids (C) portal vertices.
Ribbon diagram representation of the atomic models of the asymmetric unit of B-, Cand virion-capsids portal vertices. The cartoon models show the individual protein and conformers (major conformational changes are marked by dashed lines). The color scheme is same as in Fig. 1C.

Fig. S7 Interactions between five coiled coils and their neighboring subunits in B-(A) and virion-capsids (B).
Color scheme is same as in Fig. 2C. Insets show the close contacts between the coiled coils and their surrounding subunits.