The VP2 of BuV1, BuV2, and BuV3, produced using the baculovirus/
Sf9 protein expression system and purified using sucrose cushions and gradients, assembled VLPs suitable for structure determination by cryo-reconstruction (
Figure 1A). Movie frame micrographs collected for these VLPs, at two different cryo-EM resources (West/Midwest Consortium for High-Resolution Cryo Electron Microscopy and SECM4) yielded capsid structures to 2.84, 3.79, and 3.25 Å resolution from 29,596, 7564, and 5234 particle images, respectively, for BuV1, BuV2, and BuV3 (
Table 1,
Figure 1B). Differences in resolution of the BuV structures are likely due to micrograph quality and the different Å/pixel used for the different data collections, for example, 1.07 Å for BuV1, 1.06 Å for BuV3, and 1.22 Å for BuV2 (
Table 1). The capsid density morphology exhibited the known surface features of other
Parvovirinae subfamily members. These included shallow depressions at the icosahedral 2-fold axes and surrounding the 5-fold axes, and three protrusions surrounding each 3-fold axis (
Figure 1B). A raised capsid region between the depressions at the 2- and 5-fold axes is referred to as the 2/5-fold wall (
Figure 1B). Qualitative visual inspection of the capsid surfaces indicated differences on raised regions, likely due to amino acid side-chain sequence and structure variation. At these resolutions, the VP2 sequence of each BuV genotype was readily interpretable within the cryo-reconstructed density maps for core and surface loop regions (
Figure 2). For each BuV, the first 32 N-terminal residues of VP2 were disordered, and thus residues modeled for BuV1, BuV2, and BuV3 were 33–568, 33–567, and 33–572, respectively, with the last residue being their C-terminus (
Figure 3). This N-terminal disorder is consistent with observations for all structures previously determined for members of the
Parvoviridae with the exception of B19 for which the N-terminus of VP2 was exposed on the capsid surface of a medium resolution cryo-reconstructed density map [
15,
40]. The N-terminal disorder has been proposed to be due to flexibility of this VP region arising from a long stretch of glycine residues in the parvoviruses (
Figure 4A) [
24]. Thus, density ordering, regardless of the method utilized for structure determination (X-ray crystallography or cryo-reconstruction), generally begins towards the end of this glycine stretch (
Figure 4A).
The model building for the ordered density was conducted at a σ ≥ 2, however, two surface loops, residues 381 to 392, 381 to 392, 384 to 395 and 496 to 503, 495 to 502, and 499 to 506 in BuV1, BuV2, and BuV3, respectively, were less ordered and were built at 1σ. For BuV2, at the lowest resolution of 3.79 Å, only the main chain for the residues 381–391 could be interpreted with confidence. Notably, as has been reported for other cryo-reconstructed density maps, a number of the acidic residue side-chains were less ordered because of their increased sensitivity to electron induced radiation damage (e.g., Glu366 in
Figure 2B). The final capsid models were refined with a good correlation coefficient (relative to the cryo-reconstructed maps) and geometries consistent with other virus structures, including parvoviruses, determined to a similar resolution (
Table 1) [
35].
As previously mentioned, the VP2 sequence of the BuVs shares 64%, 73%, and 65% sequence identity between BuV1 and BuV2, BuV1 and BuV3, and BuV2 and BuV3, respectively. This level of identity results in a VP2 topology that is highly superposable with 498/536 (~92.9%), 528/536 (98.5%), and 508/535 (95.0%), respectively, of their pairwise Cα positions aligned consistent with an anticipated high structure similarity. The structures differ from each other at seven surface loop regions located within previously defined VRs (
Figure 3B). These loops are located at the 2-fold wall (VR-VI), 2/5-fold wall (VR-IV, VR-VII, and VR-IX), and the protrusions surrounding the 3-fold axes (VR-I and VR-III). Minor changes were observed in the 5-fold region at the tops of the DE loops forming the 5-fold channel (VR-II) and the HI loop that lines the floor of the channel surrounding this channel. VR-IX is mostly structurally conserved among all three viruses, with only the BuV2 residue 540 being conformationally different, VR-VI and VR-VII adopt different conformations, with BuV1 differing from BuV2 and BuV3 at VR-VI and BuV2 differing from BuV1 and BuV3 at VR-VII. In VR-I, two additional residues are inserted in BuV3 relative to BuV1 and BuV2 and VR-III contains conformational difference among all three BuV resulting in Cα position differences of up to 5.0 Å (
Figure 3B). These two VRs extend from the 5-fold region to assemble the 3-fold protrusions along with VR-IV and VR-VIII that are structurally conserved. These four loops are contributed from two VP2 monomers with one contributing VR-I and VR-III and the other VR-IV and VR-VIII. At the top of the 5-fold DE loop, BuV2 residue 160 adopts a different conformation to BuV1/BuV3 residues 161/163 that are structurally similar. This structural difference of BuV2 likely stems from four amino acid changes in the DE loop compared to BuV1 and BuV3 that share the same amino acid composition. The HI loop adopts a different conformation in all three viruses. Although notably, as mentioned above, the HI loop residues are less ordered than other VP2 regions and side chain density was difficult to interpret for residues 498–506. The main chains were interpreted with confidence. All the loop differences give rise to localized capsid surface morphology differences between the BuVs at the 2/5-fold wall and the regions surrounding the 3-fold and 5-fold axes (
Figure 1B). BuV2, which is the most different, has more rounded 3-fold protrusions due to VR-I and VR-III and 2/5-fold wall is thicker and more pronounced than BuV1 and BuV3 due to the larger VR-VII (
Figure 1B). Significantly, the 2/5-fold wall and 3-fold protrusions are reported antigenic hotspots for the parvoviruses [
41,
42,
43,
44]. These structural differences thus support the reported lack of serum cross-reactivity for these viruses and the suggestion that they are serotypes [
12].