Structures of the Major Capsid Proteins of the Human Karolinska Institutet and Washington University Polyomaviruses

Abstract

The Karolinska Institutet and Washington University polyomaviruses (KIPyV and WUPyV, respectively) are recently discovered human viruses that infect the respiratory tract. Although they have not yet been linked to disease, they are prevalent in populations worldwide, with initial infection occurring in early childhood. Polyomavirus capsids consist of 72 pentamers of the major capsid protein viral protein 1 (VP1), which determines antigenicity and receptor specificity. The WUPyV and KIPyV VP1 proteins are distant in evolution from VP1 proteins of known structure such as simian virus 40 or murine polyomavirus. We present here the crystal structures of unassembled recombinant WUPyV and KIPyV VP1 pentamers at resolutions of 2.9 and 2.55 Å, respectively. The WUPyV and KIPyV VP1 core structures fold into the same β-sandwich that is a hallmark of all polyomavirus VP1 proteins crystallized to date. However, differences in sequence translate into profoundly different surface loop structures in KIPyV and WUPyV VP1 proteins. Such loop structures have not been observed for other polyomaviruses, and they provide initial clues about the possible interactions of these viruses with cell surface receptors.

INTRODUCTION

The human Karolinska Institutet polyomavirus (KIPyV) and Washington University polyomavirus (WUPyV) were discovered in 2007 (2, 9). Their closest relatives among polyomaviruses are the newly discovered human polyomaviruses 6 and 7 (HPyV6 and HPyV7, respectively) (21). They are also related to simian virus 40 (SV40), murine polyomavirus (Polyoma), and the human viruses BK (BKV), JC (JCV), Merkel cell polyomavirus (MCPyV) (8), Trichodysplasia spinulosa-associated polyomavirus (TSPyV) (28), and HPyV9 (22).

KIPyV and WUPyV have been identified in populations worldwide (reviewed in reference 11). Kean et al. (14) reported that 55 and 69% of (Western) adults are seropositive for KIPyV and WUPyV, respectively, with seroconversion occurring during childhood. Both viruses were sequenced from respiratory tract samples, indicating that they might persist in the respiratory system. However, they have not yet been linked to disease.

The architecture of polyomavirus particles is known from X-ray crystal structures of SV40 and Polyoma virions (16, 26). Their icosahedral capsids consist of 360 copies of the major structural protein viral protein 1 (VP1), arranged in 72 pentamers conforming to a T=7d lattice. The core portion of VP1 adopts a β-sandwich fold with jelly-roll topology and assembles into stable ring-shaped pentamers. The N and C termini of VP1 form extensions that both emanate from the bottom of the pentamer, which corresponds to the inside of the capsid. The C-terminal extensions, termed “arms,” extend toward other pentamers in the capsid and contact them by adding a strand to one sheet of their β-sandwich cores. Each incoming C-terminal arm interacts with the N-terminal extension of the invaded VP1 monomer, which fastens the added strand, and then turns toward the interior of the virion to contact the viral DNA. These interactions are stabilized by Ca²⁺ ions. The surface of VP1 is formed almost entirely by extensive loops linking the β-strands of the core. These loops, the most variable regions of VP1, contain the receptor binding sites in other polyomaviruses and define the antigenicity of the virus (19, 20, 26).

The VP1 proteins of WUPyV and KIPyV share high sequence homology with each other but are much less similar to other polyomavirus VP1 proteins (21, 28). Since there is also high sequence homology among other polyomavirus VP1 proteins, the diverging WUPyV and KIPyV VP1 sequences point to a distant evolutionary relationship and perhaps an early evolutionary divergence from the polyomavirus family tree. It is thus likely that WUPyV and KIPyV possess structural features not present in other polyomaviruses that might provide insight into their receptor binding specificities.

In order to visualize these features, we have determined the crystal structures of the KIPyV and WUPyV VP1 proteins as pentameric capsomeres. Despite their low sequence homology to VP1 proteins of known structure, the core structures of WUPyV and KIPyV VP1 are highly similar to those of other polyomavirus VP1 proteins. Interestingly, however, the surface loops of KIPyV and WUPyV VP1 have conformations that differ profoundly from those seen in other VP1 structures. Since these loops mediate receptor interactions in several polyomaviruses, their unique structures provide insights into the possible interactions of WUPyV and KIPyV with cell surface receptors.

MATERIALS AND METHODS

Protein expression and purification.

DNA coding for amino acids 31 to 303 of KIPyV VP1 and amino acids 33 to 295 of WUPyV VP1 (GenBank numbers EF127906 and NC009539, respectively) was amplified by PCR and cloned into the pET15b expression vector (Novagen) in frame with an N-terminal hexahistidine tag (His tag) and a thrombin cleavage site. Both proteins were overexpressed in Escherichia coli BL21(DE3) and purified by nickel affinity chromatography and gel filtration on Superdex-200. The tag was cleaved with thrombin before the gel filtration step, leaving the non-native amino acids GHSM at the N terminus in both cases. The WUPyV VP1 protein was internally cleaved by thrombin after amino acid R197 due to a noncanonical thrombin cleavage site, and it therefore could not be used for crystallization. After mutation of R197 to K, the His tag could be cleaved without degradation of the protein. This construct was then used for crystallization. The mutation did not alter the overall secondary structure of the protein, as confirmed by circular dichroism (CD) spectroscopy (data not shown). For simplicity, we will refer to the R197K mutant as WUPyV VP1.

Crystallization.

After gel filtration, both proteins were initially in a buffer comprised of 20 mM HEPES (pH 7.5) and 150 mM NaCl. WUPyV VP1 was crystallized at 4°C at a concentration of 6.0 mg/ml by sitting drop vapor diffusion against a reservoir of 0.2 M trisodium citrate, 0.1 M HEPES (pH 7.5), and 20% (vol/vol) isopropanol. KIPyV VP1 was supplemented with 20 mM dithiothreitol and then crystallized at 20°C at a concentration of 8.6 mg/ml against a reservoir of 0.2 M sodium formate (pH 7.0) and 20% (wt/vol) PEG 3350. Crystals were harvested into the respective reservoir solutions and cryoprotected by soaking them for 10 s in reservoir solution containing 30% (vol/vol) glycerol. They were then flash-frozen in liquid nitrogen.

Structure determination.

Diffraction data for KIPyV VP1 were collected at beamline X06SA at SLS (Villigen, Switzerland), and data for WUPyV VP1 were collected at beamline ID14-1 at ESRF (Grenoble, France). The data were processed with xds (13), and the structures were solved by molecular replacement with Phaser in CCP4 (4, 17) using the β-sandwich core of the JCV VP1 pentamer structure (3NXD) as a search model (19). The crystals of KIPyV VP1 belong to space group P1 with two pentamers in their asymmetric unit, while the WUPyV VP1 crystals contain only one pentamer in their asymmetric unit and belong to space group P4₃2₁2 (Table 1). After rigid body and simulated annealing coordinate refinement in Phenix (1), missing parts of the model such as the surface loops were visible in electron density maps and could be built in Coot (6). Refinement proceeded by alternating rounds of restrained coordinate and isotropic B-factor refinement in Phenix or Refmac5 (18) and model building in Coot. The noncrystallographic symmetries relating the five WUPyV VP1 monomers and the 10 KIPyV VP1 monomers in the respective asymmetric units were used as restraints throughout refinement. The final models comprise residues 37 to 106 and 111 to 294 for all chains of WUPyV VP1, and 42 to 63, 69 to 84, 90 to 114, and 124 to 303 for all chains of KIPyV VP1. Additional loop residues could be built in some chains. The models have low R_free values of 22.1 and 25.0%, respectively, and good stereochemistry (3) (Table 1). Figures showing the X-ray structures were prepared with PyMol (Schrödinger, Inc.).

Table 1.

Crystallographic statistics

Parameter	VP1
	KIPyV	WUPyV
Unit cell axes (a, b, c) (Å)	70.3, 82.8, 142.1	166.2, 166.2, 127.6
Unit cell angles (α, β, γ) (°)	87.0, 98.2, 108.8	90, 90, 90
Space group	P1	P43212
Resolution (Å)	50–2.55 (2.62–2.55)	40–2.9 (2.98–2.90)
No. of reflections (total)	237,505 (16,282)	159,898 (10,139)
No. of reflections (unique)	93,982 (6,639)	39,647 (2,908)
Rmergea (%)	5.4 (60.2)	9.6 (57.8)
Completeness (%)	95.9, (91.1)	98.8 (99.4)
I/σI	11.9 (1.4)	11.1 (1.8)
Rworkb (%)	22.7 (34.5)	22.2 (33.0)
Rfreec (%)	24.4 (36.7)	25.0 (35.7)
RMSD bond length (Å)	0.009	0.009
RMSD bond angle (°)	1.2	1.2
No. of protein atoms	19,845	9,966
Avg B factor (Å2)	56.2	58.0
B factor (Wilson) (Å2)	60.6	58.3

^aR_merge = ΣhklÇI − <I>Ç/ΣhklI, where I is the intensity of a reflection hkl, and <I> is the average over symmetry-related observations of hkl.

^bR_work = ΣhklÇF_obs − F_calcÇ/ΣhklF_obs, where F_obs and F_calc are the observed and calculated structure factors, respectively.

^c5.0 and 7.5% of the reflections were not used during refinement to calculate free R factors for the KIPyV and WUPyV models, respectively.

Accession numbers.

Coordinates and structure factor amplitudes have been deposited in the RCSB Protein Data Bank (http://www.rcsb.org) under accession numbers 3S7X (WUPyV VP1) and 3S7V (KIPyV VP1).

RESULTS

Overall structure of KIPyV and WUPyV VP1.

In order to crystallize the KIPyV and WUPyV VP1 pentamers, we recombinantly expressed truncated versions of each protein, spanning amino acids 31 to 303 of KIPyV VP1 and 33 to 295 of WUPyV VP1. The expression constructs were designed based on the constructs previously used for the crystallization of SV40 and Polyoma VP1 pentamers (20, 25). In all cases, the VP1 sequences contain the predicted β-sandwich core and form pentamers but cannot assemble into viral capsids due to truncation of their C-terminal arm. The N-terminal extension was deleted because it had inhibited crystallization of Polyoma VP1 (25) and only forms an ordered structure when the C-terminal arm is present. The structures of KIPyV and WUPyV VP1 pentamers were solved at 2.55- and 2.9-Å resolution, respectively (Fig. 1A and B; Table 1). Similar to other polyomavirus VP1 proteins, both KIPyV and WUPyV VP1 adopt the antiparallel β-sandwich fold iconic for viral capsid proteins. β-Strands are named alphabetically from the N terminus. Two β-sheets, formed by strands B, I, D, and G and strands C, H, E, and F, respectively, stack against one another in each monomer, forming the hydrophobic core of each protein. Additional β-strands align with the β-sandwiches from neighboring VP1 molecules to form VP1 pentamers. Residues 35 to 40 of KIPyV VP1 and 37 to 42 of WUPyV VP1 form a short β-strand, termed A_new, that aligns with the most C-terminal part of the G strand and contacts the residues at the C terminus of the β-sandwich structure. In both KIPyV and WUPyV VP1, the β-sandwich structures are decorated with extensive loops that link the β-strands and make up most of the protein surface. The top surface of the pentamer, corresponding to the outer surface of the virion, is to a large extent formed by the long BC loop, which is divided for clarity into BC1 and BC2 loops (residues 59 to 70 and 77 to 90, in WUPyV, respectively, and residues 57 to 74 and 81 to 99 in KIPyV, respectively) that face in different directions (Fig. 1C). The sides of the pentamers are decorated with the extensive EF loops, parts of which fold into small, three-stranded β-sheets (Fig. 1). The CD loops at the bottom of each pentamer are disordered in most VP1 monomers and only become ordered when engaged in crystal contacts (Fig. 1C). They have elevated temperature factors, assume different conformations, and are poorly defined by electron density in all VP1 pentamer structures determined thus far (19, 20, 25). They also have variable conformations in structures of entire virions (16, 23, 24, 26).

Fig. 1.

Structures of KIPyV and WUPyV VP1 pentameric capsomeres. (A and B) Crystal structures of unassembled KIPyV (A) and WUPyV (B) VP1 pentamers are shown in ribbon representation. In each case, one monomer is highlighted in color, the others are in gray. β-Strands are labeled alphabetically according to the convention established for SV40 and Polyoma. The 5-fold symmetry axes are indicated with gray arrows. (C) Superposition of KIPyV and WUPyV VP1 monomers. The proteins are shown in tube representation, with a thicker tube indicating a higher temperature factor (B factor) of the respective amino acid. Protein regions for which the RMSD values for Cα atoms exceed 0.8 Å are highlighted in color. These regions align well with areas that exhibit elevated temperature factors due to flexibility. Loop structures discussed in the text are labeled.

The structures of WUPyV and KIPyV VP1 are highly similar to one another. The Cα atoms of a monomer of KIPyV and WUPyV VP1 can be superimposed with a low root mean square deviation (RMSD) value of 0.79 Å (Table 2), a finding consistent with their high sequence identity of 67.8%. The main differences between KIPyV and WUPyV VP1 locate to relatively short stretches (8 amino acids maximum) in the surface loops, namely, the tips of the BC1, BC2, and DE loops and parts of the EF loop (Fig. 1C). If only the conserved cores of VP1 monomers are superposed, the RMSD between Cα atoms is 0.42 Å, indicating that their differences map to the loop structures. With an RMSD of 0.86 Å, the Cα atoms of the entire KIPyV and WUPyV VP1 pentamers also superimpose very well onto each other (Table 2).

Table 2.

Comparison of WUPyV, KIPyV, SV40, Polyoma, and JCV VP1 structures

Structure comparisona	Cα RMSD (Å)
	Monomer (core only)	Monomer	Pentamer
WUPyV-KIPyV	0.42	0.79	0.86
Polyoma-KIPyV	0.72	1.57	1.98
SV40-KIPyV	0.76	1.48	1.90
SV40-Polyoma	0.46	1.11	1.07
SV40-JCV	0.34	0.71	0.86

^aVP1 monomers or unassembled VP1 pentamers of WUPyV, KIPyV, SV40 (pdb 3BWR [20]), Polyoma (PDB 1VPS [25]), and JCV VP1 (PDB 3NXD [19]) were superimposed on one another with Superpose in CCP4 (4) using a secondary-structure matching algorithm. Conserved core structures were identified from the structure-based sequence alignment of WUPyV, KIPyV, SV40, and Polyoma as stretches of sequence with more than 5 amino acids, for which the average RMSD of the KIPyV-SV40, KIPyV-Polyoma, and Polyoma-SV40 pairs was <0.8 Å. For SV40, these were amino acids 44 to 54, 86 to 92, 109 to 120, 144 to 166, 193 to 199, 202 to 227, 232 to 254, 257 to 269, 277 to 282, and 283 to 295. These were then superposed by using LSQMAN (15).

N termini and regulation of capsid assembly.

In both WUPyV and KIPyV VP1, a short stretch of amino acids at the N termini of the free pentamers forms the short A_new-strand, which interacts with C-terminal residues of the G-strand (Fig. 1A and B). The same conformation is seen for homologous residues in free pentamers of Polyoma, SV40, and JCV VP1 (19, 20, 25). In the context of Polyoma or SV40 virions, these residues are part of the N-terminal clamp that fix incoming C-terminal arms in place and adopt turn and helical conformations (23, 26). When the structure of Polyoma VP1 pentamers was first determined, it was unclear whether the A_new-strand conformation observed in free pentamers was relevant or not. Since the free pentamer structures of the rather diverse polyomaviruses Polyoma, SV40, WUPyV and KIPyV that all belong to different space groups all feature an A_new-strand, it is much more likely that this conformation is a common feature of free pentamers and has physiological importance. One possible function of the A_new-strand might be to contact the beginning of the C-terminal arm of the same monomer and to guide it away from the molecule to prevent self-invasion (25).

Comparison with other polyomavirus VP1 structures.

Multiple sequence alignment and phylogenetic analysis revealed that VP1 proteins from all polyomaviruses known to date can be classified into three main phylogenetic groups (21, 28): (i) one comprising SV40 and the related BKV and JCV; (ii) a large and diverse group containing Polyoma, as well as MCPyV and avian polyomaviruses; and (iii) a third group that is more distant in evolution and consists of WUPyV, KIPyV, and the newly discovered human viruses HPyV6 and HPyV7.

We therefore compared the KIPyV and WUPyV VP1 structures to two structurally known members of the other two groups, SV40 and Polyoma VP1, by superposing structures of unassembled pentamers of each protein using different residue ranges and calculating RMSD values for their Cα atoms (Table 2). The monomer superpositions for the different proteins were then manually combined into a structure-based sequence alignment (Fig. 2). For reference, the SV40-JCV VP1 pair was also included (19) because these two VP1 proteins feature a high level of sequence identity of 77% in the truncated pentamers used here.

Fig. 2.

Structure-based sequence alignment of WUPyV, KIPyV, Polyoma, and SV40 VP1. Protein regions for which all structures align with RMSD values between Cα atoms that are <0.8 Å and with RMSD values between Cα atoms that are <1.5 Å are colored blue and light blue, respectively. Regions in which only the WUPyV and KIPyV structures align with each other are colored orange and yellow for Cα RMSD values of <0.8 and <1.5 Å, respectively. Regions in which only SV40 and Polyoma align with each other are colored purple and light purple for the same Cα RMSD values. β-Strands are indicated with gray arrows. Residues shown in lowercase letters and colored gray were not included in the expression constructs and are therefore not present in the crystal structures. Every tenth amino acid in each sequence is highlighted in boldface.

The structure of the β-sandwich core is conserved among all proteins compared (Table 2). The β-sandwich cores of WUPyV and KIPyV VP1 superimpose very well onto each other (RMSD = 0.42 Å). However, they both differ significantly from the core structures of SV40 and Polyoma VP1, as evidenced by elevated RMSD values for the KIPyV-SV40 and KIPyV-Polyoma comparisons. These structural differences are more pronounced when entire monomers, including the diverse loops, are compared. The RMSDs for entire pentamers are even higher than for monomers when WUPyV or KIPyV are compared to SV40 or Polyoma. This indicates that the orientation of monomers with respect to one another is somewhat different within the WUPyV and KIPyV VP1 pentamers compared to the SV40 and Polyoma structures. Taken together, our analysis supports the conclusion that the VP1 proteins of KIPyV and WUPyV form a new group that is different from both the group represented by Polyoma and the group represented by SV40 (Table 2). The structural comparisons are also consistent with the structure-based sequence alignment (Fig. 2), which reveals larger differences in sequences among SV40, Polyoma, WUPyV, and KIPyV VP1 than between WUPyV and KIPyV VP1 alone.

Organization of surface loops.

The top surface of the pentamer, which corresponds to the outer surface of the virion, is defined almost entirely by the BC1, BC2, DE, and HI loops of each VP1 monomer (Fig. 3A). These loops are known to mediate interactions with different cellular receptors in the known cases (19, 20, 25, 26). As a result, their structures are highly diverse among polyomaviruses (Fig. 2 and Fig. 3A to E). The most profound differences between VP1 proteins from different viruses are found in the BC1 and BC2 loops. Unlike the BC1 loops of SV40 and Polyoma, the equivalent loops of KIPyV and WUPyV VP1 do not point toward the clockwise neighboring VP1 monomer in the same pentamer but face away from the 5-fold central axis of the pentamer, toward neighboring pentamers within the virion (Fig. 3A and E). Although the BC2 loops of Polyoma and SV40 VP1 lie relatively flat on the surface of the protein, the tip of the WUPyV VP1 BC2 loop forms a short α-helix that is tilted upward (Fig. 3A, D, and G). The BC2 loop of KIPyV VP1 is longer than those of the other proteins, protruding markedly from the pentamer surface (Fig. 3A and D).

Fig. 3.

Comparison of VP1 loop structures. (A) Cα traces of VP1 monomers of KIPyV (purple), WUPyV (green), SV40 (light gray), and Polyoma (dark gray). The superpositions are based on the conserved core structure of VP1. (B to E) Details of the structural differences in the DE loop (B), HI loop (C), BC2 loop (D), and BC1 loop (E). (F and G) Interactions stabilizing the BC2 loop conformations of KIPyV (F) and WUPyV (G). The Cα traces of the BC2 loops and interacting parts of the EF loop of WUPyV and KIPyV are shown in ribbon representation in green and purple, respectively. The main chains are shown as thin lines, and residues engaging in interactions are shown in stick representation. Atoms are colored according to atom type: oxygens in red, nitrogens in blue, and carbons in green (unique to WUPyV), purple (unique to KIPyV), orange (common to both WUPyV and KIPyV), or gray (from a different VP1 monomer). (H) Interactions that stabilize the common BC1 loop conformation of WUPyV and KIPyV VP1. The Cα traces of WUPyV (green) and KIPyV (purple) are shown in ribbon representation. Conserved carbon atoms are colored orange. Residues engaging in interactions are shown in stick representation.

The DE loops of WUPyV and KIPyV VP1 are shorter than those of SV40 and Polyoma and engage in different interactions with the core of the proteins (Fig. 3A and B). In addition, they contain a short helix that is not present in SV40 or Polyoma VP1. The HI loops of KIPyV and WUPyV VP1 are also significantly shorter compared to those of SV40 or Polyoma VP1 (Fig. 2). The small β-sheet consisting of strands E′, E", and E‴ and inserted into the EF loop is structurally well conserved between all four viruses (Fig. 2).

Structure of the BC loop.

The BC1 loop of KIPyV is six amino acids longer than the BC1 loop of WUPyV VP1. Most of these additional residues are disordered in all 10 copies of KIPyV VP1 in the crystal structure, suggesting that they are also flexible in solution. However, the N- and C-terminal parts of the WUPyV and KIPyV BC1 loops are conserved and feature the same conformation, which is stabilized by interactions between residues at the beginning and end of the BC1 loop (Fig. 3H). Most prominent is the conserved π-stacking interaction between N59, H72, and W74 in KIPyV VP1 (N61, H68, and W70 in WUPyV VP1). The asparagine and histidine residues in this sequence are further stabilized by conserved hydrogen bonds to main chain atoms.

Subtle differences in other parts of the protein likely serve to accommodate the different BC1 loop conformations of WUPyV and KIPyV compared to their SV40 and Polyoma counterparts. For instance, the BC1 loop conformation observed in WUPyV and KIPyV would clash with the main chain of the EF loops of SV40 and Polyoma, which are one amino acid longer.

The BC2 loops of both WUPyV and KIPyV VP1 are stabilized by salt bridges that link the beginning and the end of the loop and by main-chain hydrogen bonds to the EF loop (Fig. 3F and G). The long protrusions in the BC2 loop of KIPyV are rather flexible, as indicated by elevated B factors, but they adopt the same structure in the eight chains in the asymmetric unit that are completely defined by electron density (Fig. 3F).

BC-linker.

Although both the BC1 and BC2 loops assume different conformations in the WUPyV, KIPyV, SV40, and Polyoma proteins, a stretch of six amino acids between the BC1 and BC2 loops is structurally conserved among all four proteins and also JCV VP1 (Fig. 4A and B). The term “BC-linker” is proposed here for that region. Given its location and its interactions with surrounding residues, the BC-linker is likely important for maintaining the structure and stability of both parts of the BC loop (Fig. 4). All six BC-linker residues feature torsion angles characteristic for β-strands, and they form main chain hydrogen bonds to the I′ strand, but the side chain of the conserved serine at the third position of the BC-linker makes a hydrogen bond to the main chain of the I′ strand, disrupting the β-sheet (Fig. 4A and C). The consensus sequence of the BC-linker is G(S/T)-φ-S(T)-x-x-φ, with “φ” denoting a hydrophobic residue and “x” indicating any residue. The first residue of the linker is a glycine or another small amino acid that can tolerate divergent BC1 loop structures. The two conserved hydrophobic residues anchor the linker by making van der Waals contacts with the side chains of two consecutive rather hydrophobic residues on the I′ strand. In addition, the residue on the I′ strand preceding the two hydrophobic ones donates a hydrogen bond to the main chain of the conserved serine in all known structures. The features of the linker and the I′ strand are conserved in polyomavirus VP1 sequences from most species (Fig. 4A). For HPyV6 and HPyV7, automated sequence alignments in this region are problematic because of their low level of sequence similarity even to their closest relatives WUPyV and KIPyV VP1. However, they do possess a sequence consistent with features of the linker between the sequences for the B and C strands (data not shown). Interestingly, the two consecutive residues denoted x in the linker sequence participate in sialic acid binding in Polyoma, SV40, and JCV VP1 proteins (19, 20, 25).

Fig. 4.

Structure of the BC-linker. (A) Multiple sequence alignment of the BC-linker (left) and I′ strand (right). The sequences of HPyV6 and HPyV7 were omitted because alignments in this region were ambiguous. Residues that have the same properties are shaded teal, with the conserved S/T in dark teal. Residues that are similar among most polyomaviruses are shaded orange. The alignment was performed by using Muscle (5), and sequences were then arranged according to the structure-based sequence alignment presented in Fig. 2. (B) Overlay of the BC-linker structures of WUPyV (green), KIPyV (purple), Polyoma (dark gray), SV40 (light gray), and JCV (medium gray) VP1. The structures were superposed using only the six residues of the linker with LSQMAN (15). The backbone atoms of the BC-linkers and of the three preceding and following residues are shown as thin lines. The side chains of the BC-linker residues are shown in stick representation and labeled according to the consensus sequence of the linker. The Cα atoms of the glycines often found as first residue of the linker are indicated as spheres. (C) Interactions of the BC-linker of KIPyV VP1. The Cα traces of the BC-linker and the I′ strand of KIPyV are shown in ribbon representation and colored purple. The main chain is shown as lines, and side chains engaging in contacts are shown in stick representation. Side chains are colored as in panel A. Nitrogen atoms are colored blue and oxygen atoms are colored red.

Features of the virion surface.

The surface loops of the KIPyV and WUPyV VP1 structures differ substantially in their conformations from those seen in other polyomaviruses. Since these loops constitute the bulk of the surface-exposed area of the virus, they are primarily responsible for the appearance of the virion surface (Fig. 5). The BC1 loops of KIPyV and WUPyV form large protrusions at positions on the top and side surfaces of VP1 that differ in location from protrusions observed in SV40 and Polyoma. On the other hand, the HI loops of WUPyV and KIPyV VP1 are six amino acids shorter than their counterparts in SV40 or Polyoma. Consequently, they do not protrude as far from the capsid, instead forming a groove between other, longer loops (Fig. 3A and D, Fig. 5C and D). The long, protruding BC2 loops of KIPyV VP1 render its surface especially rugged, contributing to the formation of a deep groove that has not been observed in any other polyomavirus VP1 structure (Fig. 5A and C).

Fig. 5.

Comparison of VP1 pentamer surfaces. (A and B) Overlays of KIPyV and SV40 (A) and WUPyV and Polyoma (B) VP1 pentamers in surface representation. The KIPyV and WUPyV structures are colored purple and green, respectively, whereas the SV40 and Polyoma VP1 proteins are colored gray. The 5-fold symmetry axes are indicated with gray arrows. Surface features that can be attributed to specific loops are indicated. Loops of the VP1 monomer facing the viewer have standard labels (e.g., BC2 or HI), while loops of its clockwise and counterclockwise neighbors in the ring-shaped pentamer are designated cw and ccw, respectively. The orange boxes indicate the close-up views used in panels C to F. (C) to (F) Close-up views of the top surfaces of the KIPyV (C), WUPyV (D), SV40 (E), and Polyoma (F) VP1 pentamers, which feature the receptor binding sites in SV40 and Polyoma. Proteins are shown in surface representation and colored gray, ligands are shown in stick representation and colored by atom type. The carbon atoms of the ligands are colored yellow for the glycerol chain seen in the WUPyV VP1 structure and orange for the oligosaccharide ligands of SV40 and Polyoma.

DISCUSSION

We have determined the structures of the VP1 pentamers of two newly identified polyomaviruses, KIPyV and WUPyV. Based on their sequences, these two viruses are evolutionarily distant from other polyomaviruses. Consistent with this divergence, the two new structures hold some surprises and reveal novel features that had not been seen in other polyomaviruses. These features provide insights into possible modes of receptor engagement by KIPyV and WUPyV, and they also provide a platform for understanding essential features of the very recently described HPyV6 and HPyV7 viruses.

The receptors used by KIPyV and WUPyV are still unknown. However, inspection of the unique surface structures of the KIPyV and WUPyV VP1 pentamers provides insights into what types of receptors they might bind. Structural analyses have defined how the VP1 proteins of SV40, JCV, and Polyoma engage their receptors, which in all cases terminate in sialylated oligosaccharides (19, 20, 25, 27). The mode of interaction with sialic acid is conserved in SV40 and JCV, whereas Polyoma binds sialic acid in a different orientation and with different residues (19, 20, 25). However, none of the sialic-acid binding residues of SV40, JCV, or Polyoma are conserved in the WUPyV and KIPyV VP1 structures. In addition, the BC1 loops, which are involved in sialic acid binding in Polyoma and SV40, point into a different direction in KIPyV and WUPyV (Fig. 3 and 5). Moreover, the residues at the tip of the HI loop that contact the oligosaccharides in the SV40 and Polyoma complexes are not present in WUPyV and KIPyV VP1, and their HI loops are also much shorter. This configuration leads to surprisingly different surface structures of KIPyV and WUPyV compared to SV40 and Polyoma (Fig. 5). Whereas SV40 and Polyoma feature a wall formed by the tip of the HI loop at the back of the oligosaccharide binding site, KIPyV and WUPyV feature a groove at that position of the surface (Fig. 5C to F). We therefore believe it unlikely that KIPyV or WUPyV bind sialic acid in a manner similar to that seen in SV40, JCV, or Polyoma VP1.

It is possible that KIPyV and WUPyV bind sialic acid in a new orientation or at a different location. We note that MCPyV, which also lacks the residues that contact sialic acid in Polyoma, SV40, and JCV, has been shown to bind sialylated oligosaccharides (7), and thus there is precedent for an alternative ligand binding pocket and/or orientation in VP1 pentamers. It is also possible that KIPyV and WUPyV do not bind sialic acid at all, perhaps engaging other carbohydrates or even proteinaceous receptors. The unique structural features of the WUPyV and KIPyV VP1 pentamers provide some support for this second possibility.

Both KIPyV and WUPyV VP1 feature pronounced grooves on the top surface of VP1, which might be involved in receptor binding (Fig. 5C and D). These grooves are deeper than the shallow depressions that bind oligosaccharides in SV40 and Polyoma VP1 (Fig. 5E and F). In KIPyV VP1, the groove is flanked by the protruding BC2 loops on each side (Fig. 5C). This rugged surface might serve to accommodate a peptide receptor or a very long oligosaccharide chain. Interestingly, the electron density map for WUPyV VP1 revealed similar electron density features in the grooves of all five VP1 monomers that can best be explained by bound glycerol molecules (Fig. 5D). Glycerol is included in the cryoprotectant solution in high concentrations. Since it has three adjacent hydroxyl groups, glycerol mimics part of a monosaccharide and is even present in sialic acid. The glycerol binds near a positively charged patch on the surface of WUPyV VP1 that is formed by residues of the HI and BC2cw loops. Thus, the presence of glycerol might point to an oligosaccharide binding site, perhaps a negatively charged oligosaccharide such as a sulfated glycosaminoglycan or a sialylated oligosaccharide. We note that heparin is also a receptor for the related papillomaviruses (10, 12). Unfortunately, the structural data presented here allow few conclusions concerning specific receptor binding until the nature of their receptors are identified by other means.

The structures of WUPyV and KIPyV VP1 also shed light on likely features of the VP1 structures of the newly identified human polyomaviruses HPyV6 and HPyV7, which share homology with WUPyV and KIPyV VP1 but are more distantly related to other polyomavirus VP1 proteins (21). The HPyV6 and HPyV7 viruses likely also contain a BC-linker that anchors the endpoints of the BC1 and BC2 loops, although sequence alignments in this region are somewhat ambiguous. It is clear, however, that the BC loops of HPyV6 and HPyV7 are much shorter than their KIPyV and WUPyV counterparts, and thus unlikely to form similar protrusions. On the other hand, the HI loop sequences of HPyV6 and HPyV7 VP1 are much longer than those of all other polyomaviruses and may perhaps even fold back onto the VP1 core structure. Taken together, the surfaces of HPyV6 and HPyV7 likely have protrusions and recessions that are quite different from those seen in other polyomaviruses. The KIPyV and WUPyV VP1 structures reported here thus highlight the capacity of the conserved pentameric VP1 core to support highly diverse loop arrangements, which endow the different polyomaviruses with unique interaction surfaces.

ADDENDUM IN PROOF

After the manuscript was accepted, the international Committee on Taxonomy of Viruses (ICTV) created the genus Wukipolyomavirus, thus taking into account the evolutionary distance of WUPyV and KIPyV from other polyomaviruses.