Cryo-ET of infected cells reveals that a succession of two lattices drives vaccinia virus

37 During its cytoplasmic replication, vaccinia virus assembles non-infectious spherical 38 immature virions (IV) coated by a viral D13 lattice. Subsequently, IV mature into infectious 39 brick-shaped intracellular mature virus (IMV) that lack D13. Here, we performed cryo- 40 electron tomography of frozen-hydrated vaccinia-infected cells to structurally characterise 41 the maturation process in situ. During IMV formation a new pseudohexagonal lattice forms 42 inside IV to produce the viral core. This lattice, consisting of trimeric pillars, appears as a 43 palisade in cross-section. Our measurements suggest that the length of this core is 44 determined by the D13 lattice. During maturation, which involves a 50% reduction in 45 volume, the viral membrane becomes corrugated as it adapts to the IMV core in a process 46 that does not appear to require membrane removal. Our study suggests that the flexible 47 viral membrane is controlled by the consecutive D13 and palisade lattices, which 48 structurally define vaccinia assembly and maturation. 49 used new the and The two


123
This D13 lattice flexibility is also evident from its ability to coat the exterior of "open 124 spheres" with different membrane curvature (also known as crescents due to their shape 125 when seen in a middle view) during IV formation (Figure 2A, B and Movie 3).

127
The interior of the majority of IV, which contain viral DNA and proteins required for virion 128 assembly, are homogenous and lack any of the distinctive regions or layers seen in IMV 129 ( Figure 1B). In a few cases, however, we could observe a single dense striated structure 130 inside IV corresponding to the nucleoid (viral DNA genome) before it decondenses ( Figure   131 2C and S2), as seen in previous ultra-structural analysis of thin EM sections of vaccinia 132 infected cells (28). We also found examples of related striated structures in the cytoplasm 133 adjacent to, or in direct association with, assembling IV ( Figure 2D, E). Consistent with the 134 interpretation that the example shown in Figure 2E represents the viral genome being

145
Given the clear lattices present on the IV surface in our tomograms ( Figure 3A, Movie 4), we 146 analysed the structure of the D13 lattice in situ by subtomogram averaging. We obtained 147 maps by averaging surface sectors revealing hexamers of trimer arrangements with a lattice 148 spacing of 133 Å, in agreement with lattices described for in vitro assembled D13 and deep-149 etch EM honeycombs ( Figure 3B). Rigid-body docking of the high-resolution structure of a 150 D13 trimer (PDB: 7VFE) into the map shows the packing of D13 trimers is consistent with 151 that recently reported (19). While at much lower resolution, precluding a detailed 152 description of interface residues, the arrangement of D13 on the spherical IV surface does 153 not appear to include a relative twist of trimeric axes between adjacent trimers as observed 154 in single particle dimers or tubular assembles.

156
From the hexagonal packing of D13 in the subtomogram average, we estimate that ~5,300 157 D13 trimers will cover the surface of a spherical IV. The packing density and radius of the

171
The outer membrane of IMV has a corrugated appearance and the D13 lattice is absent.

172
There is, however, a new additional outer layer of 6.67 ± 0.11 nm on the viral membrane 173 that was not present on IV (Figure 4). Underneath the viral membrane there is a palisade-174 looking structure consisting of a series of turret-like densities in cross-section (12.5 nm 175 thick) that is associated with a inner wall that is 3.84 ± 0.09 nm thick ( Figure 4A). Depending 176 on the orientation of the virion, the interior organisation of the IMV appears very different.

177
In the mid view of its two widest dimensions (352 x 281 nm), the palisade appears in contact 178 with the viral membrane. However, in the orthogonal lateral view (352 x 198 nm) the 179 palisade structure, which only contacts the viral membrane at the ends of the virion ( Figure   180 5 Sept 2022 6 4B), forms a dumbbell shape that results in the formation of two concavities. Each cavity 181 contains a single dense and amorphous structure corresponding to a lateral body. The space 182 between the viral membrane and the palisade varies between 9.28 ± 0.18 nm and 46.53 ± 183 2.72 nm, the latter being where the lateral bodies are accommodated (see Table S1 for 184 detailed measurements). 3D tomographic analyses reveal that the palisade is a semi-regular 185 array when viewed from above ( Figure 4C, D).

187
The semi-regular organisation of the palisade is most apparent in naked cores, which lack 188 the viral membrane and are occasionally found in the cytoplasm ( Figure 5A, B).

189
Segmentation of our 3D tomograms reveals the palisade is a continuous structure without 190 fenestrations that defines the virus core, including the regions that contact the lateral 191 bodies ( Figure 4E, Movie 6). Inside the core, there are no obvious higher order structures.

192
There are, however, interconnected densities that vary between virions but tend to 193 accumulate beneath the inner wall of the core ( Figure 4A). These densities are especially 194 apparent in the compressed region underneath the lateral bodies ( Figure 4C). Such densities 195 associated with the inner wall of the core potentially represent the viral genome and its 196 associated proteins given its high contrast. Another characteristic that is evident in 197 midplane views of approximately half of the IMV (48.9%, n = 94), is that one corner of the 198 virion (and in a few cases, two corners) appears as a straight or flattened "cut corner" 199 ( Figure 1C and Figure S3). This characteristic is intrinsic to the core wall as it is also apparent

201
In addition, these naked cores lack the two densities corresponding to lateral bodies but 202 have associated spaghetti-like structures contacting the palisade surface ( Figure 5A). These 203 flexible polymers, which are 2.6 nm in diameter, form an exclusion zone of ~ 40 nm around 204 the naked core ( Figure 5A). Furthermore, it was noticeable that naked cores have randomly 205 distributed ring like structures on their surface that were not observed in IMV. These rings 206 seem to protrude 10-20 nm from the palisade surface and had inner and outer diameters of 207 4.5 ± 0.1 nm and 9.9 ± 0.1 nm, respectively ( Figure 5C).   Figure 6A). In the case of IEV, the outer membrane is also not always in close 217 contact with the underlying membrane (see Table S1 for virion measurements). The

218
dimensions and structure of the IMV are also unaffected by envelopment ( Figure 6A).

219
Notably, the palisade fully coats the viral core in all infectious virions and its organisation 220 appears unaltered ( Figure 6A-C). To study the architecture of the palisade, we performed 221 subtomogram averaging separately using IMV, IEV and CEV/EEV particles. Maps obtained 222 from the different virion types, all displayed the same organisation and lattice parameters 223 (P6 symmetry with a=b=89 ± 2 Å, θ=120°) ( Fig S4). A new combined map obtained by 224 averaging all particle types, reveals that the palisade is composed of trimeric pillars with 225 projecting lobes that interact with neighbouring pillars with local hexagonal symmetry 226 ( Figure 6D). These pillars are embedded in an unfeatured inner wall. While further details 227 are required to understand its molecular composition, this arrangement appears to be 228 flexible enough to assemble a continuous biconcave capsid structure. and IMV, following the wrinkles of the viral membrane to obtain its contour length. We 239 found that the membrane contour is virtually identical in IV and IMV, as well as in the 240 equivalent innermost mmembrane of EEV/CEV (~1100 nm, Figure 7B). This suggests the viral 241 membrane folds during maturation, which would explain the reduction in volume without 242 any detectable loss of membrane surface. In addition to reducing their volume by ~50%, IVs 243 also change their shape when they mature into IMV, becoming a triaxial ellipsoid. Strikingly, 244 the longest IMV dimension matches the diameter of IV ( Figure 7A, B), suggesting that the 245 major axis of IMV is dictated by the IV diameter. In our tomograms we also found particles 246 that may represent intermediates and/or defective examples of IV maturation. These

247
include examples where the palisade is fully formed but the IV membrane, which either 248 lacks or is partially coated with D13, is not associated with the viral core ( Figure 7C). It is 249 also interesting that we did not observe any virions with partially formed cores suggesting 250 that palisade formation is likely to be rapid or occurs en bloc.

295
The palisade, which is largely composed of p4A and A4 (31, 44, 45), consists of a trimeric 296 assembly in a pseudo-hexagonal lattice arrangement. This new viral lattice, which fully 297 covers the core and defines its boundary, also dictates the dimensions and shape of IMV.

298
The viral membrane, no longer covered by D13, wrinkles and adopts the shape and 299 dimensions of the newly formed palisade. This process may also be responsible for the 300 compression of the lateral bodies onto the virus core, causing the biconcave deformation of 301 the latter. The maturation model that emerges from our work provides a simple way by 302 which a membrane-bound particle adapts to a new internal lattice and changes its shape 303 and dimensions to become more compact without membrane removal ( Figure 7D).

304
Moreover, our data suggest the palisade length is dictated by the diameter of the IV, which    EEV. This is also the reason why we did not find enough side views of IEV to directly measure 413 their minor axis (Table S1). To estimate the IEV minor axis, we measured the thickness that

436
Initial references were generated using independent reference-free 3D classifications for 437 each virion type. This clearly showed the palisade lattice in every case, and the particles 438 for each virion type were from then on processed independently, but in an identical manner.

439
For each virion type, alternating rounds of 3D classification and refinement were used to