Thin section EM confirms that assembly is blocked in the absence of H7
When H7 is not synthesized the formation of IVs and MVs is blocked; dense inclusions accumulate collecting viral core proteins that are coated with structures resembling short viral crescents. A second structure accumulates close by, that collects D13 (Satheshkumar et al., 2009).
Conventional EM of HeLa cells fixed at 12 hrs post-infection largely confirmed these observations. In the presence of isopropyl-b-D-thiogalactopyranoside (IPTG), structures typical of a wild-type infection (crescents, IVs and MVs) were observed (figure 1A and B). Without H7-synthesis IVs and MVs were absent; large areas of lower electron-density, the VACV-replication site, collected electron-dense aggregates (figure 1C; Vi) that were occasionally coated with arcs resembling short crescents (figure 1E). At the periphery of the replication sites a second prominent structure, not described before and absent in the presence of IPTG, was seen. Electron-dense spots were arranged in a regular pattern and associated with membranes reminiscent of the ER (figure 1D; NS).
EM immuno-labeling confirmed the expected localization of VACV proteins in the presence of IPTG; anti-D13 to the surface of the IV membrane (figure 2A) and anti-A17 to the membrane of both the IVs (figure 2A) and the MVs (figure 2B). The electron-dense aggregates in the absence of IPTG collected the core protein A3 (not shown) while the short arcs were labeled with anti-D13 and A17 (figure 2D), confirming that they are short crescent-membranes. The network labeled prominently for anti-D13 showing that the scaffold protein accumulated in these aberrant structures. Anti-PDI confirmed that the membranes were derived from the ER (not shown). A17, the binding partner of D13, localized to the surrounding ER membranes (figure 2C) where it was not particularly enriched when compared to random pieces of ER (not shown).
The collective data suggested that H7 was involved in the proper formation of D13 in infected cells, resulting in the dense network associated with the ER. The D13 structures were next analyzed in 3D by ET.
Scanning- and transmission electron tomography reveal an aberrant organization of D13
Sections, 750 nm in thickness, of conventionally embedded samples were analyzed by scanning transmission electron tomography (STEM-ET), focusing on the 3D organization of the D13-network structure. In 3D the intimate association of the D13-network with the ER was readily observed (figure 3; movie 1 and 2). The rendering shown in movie 2 illustrated how the ER cisternae surrounded, and moved perpendicular into, the network (figure 3; movie 2). Slice by slice inspection of the tomograms revealed that the spots were connected to each other by thin tubes of distinct length, altogether contributing to the regular network appearance (figure 3 white arrowheads; movie 3). The electron-dense spots were seen to pull on the ER membrane to form a tube, suggesting that the connections were derived from the ER-membrane (figure 3; see also below). Rendering of the electron-dense D13-spots in z, within the 750nm volume of the section, displayed them as short hollow tubes, roughly 22nm in width (movie 2 and 3). Dual-axis transmission electron tomography of thawed cryo-sections, 150nm in thickness, labeled with anti-D13 confirmed first that the scaffold protein accumulated within the network (figure 4 Z:5). Second, the D13-labeled structures showed a massive accumulation of short membrane fragments (figure 4 Z:127), that appeared as white lines due to the negative contrasting used in this method (figure 4, movie 4). Their abundance strongly suggested these to correspond to the membrane tubes connecting the D13 units shown by STEM-ET. The electron-dense D13-spots were not readily visible in the tomograms of the thawed cryo-sections, likely because of the negative contrasting and the smaller volume analysed.
The collective 3D-EM data argued that in the absence of H7, D13 formed short hollow tubes of discrete length and diameter, interconnected by short membrane tubes, forming a regular pattern, intimately associated with ER-cisternae. The structure of the D13-spots was next analyzed with a hybrid cryoET method.
Refrozen Tokuyasu sections show D13 trimers but an absence of hexagons.
Under wild-type infection conditions D13 forms hexamers of D13-trimers. The trimers measure 7 to 9 nm in diameter, while the hexamers measure roughly 20-22 nm from vertex to vertex, depending on the EM method used (Heuser, 2015; Chlanda et al., 2009). The average width of 22 nm of the D13-spots, measured in the STEM-tomograms, suggested that D13 might form hexamers but that lattice formation was impaired in the absence of H7. However, the contrasting used displayed D13 as (uniform) electron-dense dots and failed to reveal trimers or hexamer-formation.
The technique of refrozen Tokuyasu sections (Bos et al., 2014) was applied to analyse the D13-structure. Cryo-sections, 70nm in thickness, were thawed and labeled with anti-D13, the sections were vitrified by plunge-freezing and imaged by cryoET. In the presence of IPTG the honey-comb lattices were readily observed in 3D, located on the surface of the IVs or as small discrete patches next to the IVs, as shown before (figure 5A, movie 5; Chlanda et al., 2009). By inverting the contrast, the organization of D13 trimers into hexamers, together forming the honey comb lattice was readily seen (figure 5B and C, movie 5). In the absence of IPTG the D13-network could not be unequivocally identified without prior immuno-labeling. Areas labeled for D13 on the surface of the section were subjected to tilt-series-acquisition and structures analyzed (movie 6). Structures with an average diameter of 8-9nm, were readily observed in these areas (figure 5D and E). By superimposing the known x-ray structure (Garriga et al., 2018) these could be identified as D13-trimers (figure 5F). Although the trimers arranged in a regular pattern, we failed to observe a hexagonal arrangement (figure 5F). Instead D13 collected in a pattern of 3-5 trimers separated from each other by roughly 10nm (figure 5E and F). By cryoET the D13-positive area also displayed many short membrane-tubes but their structure and relation to the D13 units was not readily displayed.
We propose that the electron-dense spots seen by RT-EM are composed of several D13-trimers, lacking the hexagonal arrangement. The trimers are arranged to form a regular pattern in x-y by cryo-ET that form the ~22nm electron-dense spots seen by conventional EM and arrange in hollow tubes in z (STEM-ET). What connects that D13-trimers to form a regular pattern remains elusive (see discussion).
Localization of H7
A putative role for H7 in proper formation of the D13 into honey-comb patches was surprising and prompted us to re-investigate its localization in infected cells, previously proposed to be predominantly cytoplasmic (Satheshkumar et al., 2009). We took advantage of the fact that recombinant H7-protein expressed in the presence of IPTG was tagged with an HA-epitope. In addition, infection with VACV-H7ind without IPTG and transfected with HA-tagged full-length H7 was used to over-express H7 using a VACV synthetic early/late promotor. Transfection efficiency was roughly 90%, based on anti-HA labeling by light microscopy. Expression of the full-length H7 lead to an efficient rescue of the phenotype observed without H7-synthesis, showing that transfected HA-H7 was functional (supplemental figure 2A). HA-tagged H7 expressed in the presence of IPTG or upon transfection displayed a general cytoplasmic labeling both by LM (supplemental figure 1) and EM (data not shown) with no concentration on, or close to, the viral membranes. While surprising, they confirm previous results (Satheshkumar et al., 2009).
Residues required for assembly
Expression of HA-tagged H7 in trans produced a phenotype indistinguishable from infection in the presence of IPTG; infected/transfected cells displayed the full complement of crescents, IVs and MVs, while the electron-dense virosomes and ER-associated networks were absent (supplemental figure 2A; Table I). The efficient rescue of the H7-phenotype prompted us to analyse residues required for membrane assembly. While single residues within H7 required for infectivity were previously analysed (Kolli et al., 2015), we focused on rescue of assembly scoring for IV and MV-formation by EM. In first instance we used truncated constructs of H7, expressing amino acid 1 to 114 (N-terminus) or 119 to 146 (C-terminus) of H7. Both constructs failed to rescue assembly, implying that both the N- and C-terminal part of the protein is essential for D13 organization (Table I).
Previous experiments implied an important role for positively charged amino acids of the H7 protein for infectivity (Kolli et al., 2015). Specifically, substituting the lysins at positions 108, 128 and 143 individually, affected the production of infectious virus, as well as a triple mutant substituting lysine 108, arginine 109 and lysine 112. The latter three amino acids map in the putative PX, PIP-binding domain of H7. We substituted all positively charged amino acids individually, expressed the mutant proteins as described above and quantified the different viral forms, IVs and MVs (Table I). None of the single point mutations affected assembly and the full complement of viral forms was made to the same extend in transfected cells (Table I; supplemental figure 2B and data not shown). Mutating all three positively charged amino acids in helical domain 7 failed to rescue assembly (Table I; supplemental figure 2D). In addition, a double mutant substituting lysine 128 and 143 also affected assembly, IVs were observed, whereas MVs were absent (Table I; supplemental figure 2C).
The data thus show, and confirm, an important role for the helical domain 7, in particular its three positively charges amino acids, its putative phosphoinositide binding domain.