In silico analysis of surface structure variation of PCV2 capsid resulting from loop mutations of its capsid protein (Cap)

Abstract

Outbreaks of porcine circovirus (PCV) type 2 (PCV2)-associated diseases have caused substantial economic losses worldwide in the last 20 years. The PCV capsid protein (Cap) is the sole structural protein and main antigenic determinant of this virus. In this study, not only were phylogenetic trees reconstructed, but variations of surface structure of the PCV capsid were analysed in the course of evolution. Unique surface patterns of the icosahedral fivefold axes of the PCV2 capsid were identified and characterized, all of which were absent in PCV type 1 (PCV1). Icosahedral fivefold axes, decorated with Loops BC, HI and DE, were distinctly different between PCV2 and PCV1. Loops BC, determining the outermost surface around the fivefold axes of PCV capsids, had limited homology between Caps of PCV1 and PCV2. A conserved tyrosine phosphorylation motif in Loop HI that might be recognized by non-receptor tyrosine kinase(s) in vivo was present only in PCV2. Particularly, the concurrent presence of 60 pairs of the conserved tyrosine and a canonical PXXP motif on the PCV2 capsid surface could be a mechanism for PXXP motif binding to and activation of an SH3-domain-containing tyrosine kinase in host cells. Additionally, a conserved cysteine in Loop DE of the PCV2 Cap was substituted by an arginine in PCV1, indicating potentially distinct assembly mechanisms of the capsid in vitro between PCV1 and PCV2. Therefore, these unique patterns on the PCV2 capsid surface, absent in PCV1 isolates, might be related to cell entry, virus function and pathogenesis.

Introduction

Porcine circoviruses (PCVs) are non-enveloped and icosahedral single-stranded DNA animal viruses. PCVs, members of the genus Circovirus in the family Circoviridae, contain two main genotypes: Porcine circovirus 1 (PCV1) and Porcine circovirus 2 (PCV2). These two viruses have a high degree of nucleotide identity and the same genomic organization (Olvera et al., 2007; Segalés et al., 2013). PCV1 was first isolated from a porcine kidney cell line (PK15) in 1953 by Tischer et al. (1974) and labelled as a non-pathogenic agent, whereas PCV2 was identified as the aetiological agent of PCV2-associated diseases (PCVADs). PCV2 is now prevalent in most major swine-producing countries, and responsible for huge economic losses in the global swine industry (Alarcon et al., 2013; Gillespie et al., 2009).

Three major ORFs have been identified in the PCV2 genome. The first, ORF1, contains 945 nucleotides and encodes replication-associated proteins (Rep and Rep′) essential for replication of viral DNA in host cells (Cheung, 2003; Finsterbusch & Mankertz, 2009). The second, ORF2 (702 or 705 nucleotides), encodes the sole structural protein (Cap) of the capsid (molecular weight approximately 28 kDa) and the main antigenic determinant of the virus (Horlen et al., 2008; Nawagitgul et al., 2000; Pogranichnyy et al., 2000). The PCV2 Cap expressed and harvested from an insect baculovirus expression system (Baculovirus Expression Vector System, BEVS), can self-assemble into virus-like particles (VLPs) in vitro. Furthermore, VLPs have been exploited as a key component in vaccine formulation and provided protection from PCV2 infection under field conditions (Blanchard et al., 2003; Fachinger et al., 2008; Liu et al., 2008; Nawagitgul et al., 2000; Seo et al., 2014). The crystal structure and cryo-electron microscopy (cryo-EM) image reconstruction of the PCV2 Cap has been elucidated at two laboratories (Khayat et al., 2011; Liu et al., 2016). The PCV2 Cap has been shown to possess the canonical viral jelly-roll structure common in many icosahedral viruses (Khayat et al., 2011). Furthermore, a PCV2 capsid structure comparison between the crystal structure and the cryo-EM image indicated that the nuclear localization signal (NLS) of PCV2 Cap is close to the icosahedral fivefold axes of the PCV2 capsid (Khayat et al., 2011). The NLS is enriched with positively charged arginine (R) residues. In general, a poly-arginine peptide is recognized as a cell-penetrating peptide (CPP), capable of disrupting cell membranes and carrying foreign proteins and chemicals (and even beads) into cells (van den Berg & Dowdy, 2011). Although the NLS is hidden inside the PCV2 capsid, it may be transiently externalized for virus infection when the capsid is induced by cell factors (i.e. cell receptor and pH change) in host cells (Khayat et al., 2011), as has been reported for other virus infections (Suomalainen & Greber, 2013; Yamauchi & Helenius, 2013). However, precise functions of these regions are unknown. Finally, ORF3 encodes a protein that is not essential for PCV2 replication, but may have apoptotic activity (Juhan et al., 2010; Liu et al., 2005).

On the basis of phylogenetic analysis of whole-genomic DNA or the cap gene, PCV2 is divided into three genotypes (PCV2a, PCV2b and PCV2c). In addition, PCV2a has been further subdivided into five clusters (2A, 2B, 2C, 2D and 2E) and PCV2b into three clusters (1A, 1B and 1C), whereas PCV2c has only been reported in Denmark (Cortey et al., 2011; Grau-Roma et al., 2008; Olvera et al., 2007; Segalés et al., 2008). It is noteworthy that PCV2 is continuously evolving through a series of point mutations in the area of ORF2 and genome recombination between PCV2a and PCV2b (Cai et al., 2012; Mu et al., 2012). Currently PCV2b is the predominant genotype worldwide (Wei et al., 2013). Whether there is a virulence difference between PCV2a and PCV2b remains unclear, although there are indications that PCV2b produced a higher virus titre than PCV2a in cultured cells (Cheung & Greenlee, 2011). Two signature motifs of the PCV2 Cap have been characterized to differentiate PCV2a from PCV2b (Cheung & Greenlee, 2011; Cheung et al., 2007). Motif 1 is located in Loop CD (⁸⁶TNKISI⁹¹ in PCV2a versus ⁸⁶SNPRSV⁹¹ in PCV2b) whereas Motif 2 contains distinct residues (¹⁹⁰SR¹⁹¹, ²⁰⁶K and ²¹⁰D in PCV2a versus ¹⁹⁰AG¹⁹¹, ²⁰⁶I and ²¹⁰E in PCV2b) that are discontinuously distributed in Loop GH, Loop HI and one β-strand of the PCV2 Cap. When Motif2 of PCV2b was substituted for that of PCV2a in the Cap, subsequent viral replication produced a distinct cytoplasmic pattern in infected PK15 cells (Cheung & Greenlee, 2011). Recently, it has been reported that a novel emerging PCV2b mutant (mPCV2b) with an extra lysine (K) at the end of carboxyl terminus (CT) of the Cap produced a higher virus titre in sera compared with common PCV2b stains (Opriessnig et al., 2014).

It is well known that PCV2 utilizes distinct and cell-dependent pathways for entry and infection. In the monocytic cell line 3D4/31, entry of PCV2 into cells (for infection) is clathrin-dependent (Misinzo et al., 2005). Conversely, inhibiting clathrin-mediated endocytosis enhanced PCV2 infection in epithelial cell lines of PK15, SK and ST (Misinzo et al., 2009), although the mechanisms are unknown. Three critical steps, namely virus attachment to the host cell, internalization and the viral genome reaching the replication site, are essential for PCV2 and other viruses to complete replication of their viral genomes. To date, no specific internalization receptor has been reported to be involved in cell entry of PCV2, although heparin sulfate and chondroitin sulfate B have been identified as cell attachment receptors of PCV2 (Misinzo et al., 2006). Engagement of PCV2 capsid with factors from host cells might induce a conformational change/disassembly of the PCV2 capsid, which is required for viral genome translocation to the replication site. Factors including a decrease in pH and presence of serine protease in host cells could be used for disassembly of the PCV2 capsid in the monocytic cell line and in epithelial cells, respectively (Nauwynck et al., 2012), although details are unknown. It was noteworthy that a unique cysteine in the PCV2 Cap had a critical role in integrity and assembly of the PCV2 VLPs, indicating that the redox state in cells might affect conformation change and capsid disassembly during PCV2 infection (Wu et al., 2012). In general, these processes are mainly dependent on interplays of PCV2, especially its capsid, and factors of host cells. As a non-enveloped virus, structure and surface patterns of the PCV2 capsid predominantly determine which pathway will be employed to complete PCV2 infection (by coordinating a series of factors in target cells). In that regard, PCV2 may exploit new factor(s) via capsid mutations to benefit the viral life cycle and infection of host cells. In addition, groups at several labs have reported that PCV2 causes immune suppression by inducing expression of IL-10 in host cells, whereas PCV1 does not (Doster et al., 2010; Du et al., 2016; Kekarainen et al., 2008). Further study demonstrated that the PCV2 Cap, but not the Rep or ORF3, is able to upregulate IL-10 expression through activating PI3K/Akt, p38 MAPK and ERK signalling pathways. In contrast, the PCV1 Cap is insufficient to activate these pathways for enhancement of IL-10 expression (Du et al., 2016). Therefore, investigation of the unique structure of the PCV2 Cap, which is absent in the PCV1 Cap, will greatly facilitate understanding of PCV2-related molecular pathogenesis.

In this study we exploited PCV1, a non-pathogenic agent as a reference, to identify unique surface patterns of the PCV2 capsid, which may be associated with engagement of virus with factors in host cells. It is important to elucidate molecular mechanisms underlying the viral life cycle, PCVADs and immunological recognition. Furthermore, functionality and pathogenicity caused by loop variations of PCV2 Caps are also discussed.

Results

Phylogenetic trees reconstructed based on the sequence of PCV Caps

Fifty distinct PCV cap gene sequences (43 from published records, and seven novel sequences from our laboratory) were incorporated into a phylogenetic tree. The cap gene-based tree matched the tree reconstructed from the concomitant genomic DNA sequences (Fig. S1a, b, available in the online Supplementary Material). Similar results have been reported previously (An et al., 2007; Olvera et al., 2007); overall, it appeared that variations of the PCV cap gene probably reflected PCV evolution.

PCV Cap sequence variations were predominantly present in loops located on the capsid surface

Cap is the sole structural protein of PCV1 and PCV2. The seven loops and the CT of the PCV2 Cap crystal structure were uniquely coloured in silico (Khayat et al., 2011) (Fig. 1a, b). Since the CT lacked a typical secondary structure in the crystal structure of the PCV2 Cap, we designated the CT as Loop CT. Sixty of these labelled Cap subunits were assembled to generate the capsid observed in the crystal structure (Fig. 1e). Within this complete structure, individual loops were readily apparent on the capsid surface (Fig. 1e). Loops of the PCV2 Cap (with the exception of Loop FG, which was completely buried inside the capsid) constituted the majority of the PCV2 capsid surface (Fig. 1e), and most of these residues were exposed on the capsid surface (Figs 1e and2). Residues of loops contributed 75 % of the exterior surface area of the capsid. Next, we conducted a sequence alignment of Cap residues from the 50 PCV isolates, and strictly conserved residues in the alignment are highlighted in red (Fig. 2). The 3D structure of the PCV2 Cap was used to identify loops in the 50 PCV Caps (Fig. 2). Residues on the capsid surface are indicated by empty circles at the bottom of the alignment; otherwise they are indicated by filled circles (Fig. 2). Results of comparative alignments demonstrated that among these loops, residues exposed on the capsid surface (indicated by empty circles at the bottom in Fig. 2), were highly divergent (labelled with blue in Fig. 2). Detailed sequence variations at each position in Caps are summarized (Table 1). Furthermore, there were two gaps in the NLS and Loop BC (Fig. 2). The NLS of the PCV capsids contained several positively charged residues; in that regard, 21 out of 43 (in PCV2) or 20 out of 45 residues (in PCV1) of the NLS were positively charged. The NLS was located in the interior of the capsid and therefore predicted to be involved in packaging virus genomic DNA. There was an insert containing two residues (AF) in the NLS of the PCV1 Cap, compared with PCV2. Finally, the numbers of residues in Loop CT varied from eight (in PCV1) to nine or ten (in PCV2). Therefore, in the following studies, we focused on variations of the PCV capsids, particularly alterations of the capsid surface due to mutations within hypervariable loops.

Fig. 1.

3D structures of the Caps and capsids of PCVs (PCV2 and PCV1). (a) Crystal structure of the PCV2 Cap (PDB accession number: 3R0R). Loops were labelled with distinct colours. (b) Crystal structure of the PCV2 Cap subunit with a surface view. (c) 3D structure of the PCV1 Cap. Structure was simulated by structure-homology modelling, using the PCV2 Cap as a template. Loops were labelled with distinct colours. (d) 3D structure of PCV1 Cap subunit with a surface view. Capsids of PCV2 (e) and PCV1 (f) with surface views. The icosahedral fivefold axes of PCV2 and PCV1 are shown in the centre of each image. Note: red strands for Loop BC; yellow for Loop CD; blue for Loop DE; orange for Loop EF; pink for Loop FG; cyan for Loop GH; magenta for Loop HI; sky blue for Loop CT; and grey for β-strands.

Fig. 2.

Comparative amino acid sequence alignment of the PCV Caps. Amino acid residues of the Caps from 50 typical PCV isolates were collected and aligned. Each Cap sequence was retrieved from GenBank and its accession number is indicated in the left column in the alignment. Red represents strictly conserved residues, whereas other colours represent amino acid variations at positions in the alignment. A hyphen (‘-’) indicates a gap of one residue at the position in the alignment. The numbers at the top of the alignment only indicate residue positions of the PCV2 Cap. All loops and CT were identified and marked by means of the crystal structure of the PCV2 Cap (PDB accession number: 3R0R). The residues of Caps, exposed on the capsid surface, are labelled with open circles on the bottom, other residues by filled circles (using the 3D structure of the PCV2 capsid as a reference). Residues of the two motifs (Motifs 1 and 2) that were used to distinguish genotypes PCV2a and PCV2b, and two conserved residues (¹⁰⁸CS¹⁰⁹) in Loop DE of the PCV2 Cap are marked with stars and filled red circles at the top of the alignment, respectively. Other typical motifs (i.e. a conserved PXXP motif among the PCV2 Caps) are labelled at corresponding positions at the bottom of the alignment.

Table 1. Summary of amino acid variation of the Caps within the 50 PCV isolates.

Note: (a) positions of residues located in loops of the PCV Cap (see Fig. 2) are marked with red in the first column, otherwise with black. (b–e) Amino acids exclusively present within isolates of PCV2a, PCV2b-1A1B, PCV2b-1C, or PCV1 are shown in yellow, green, blue and purple type, respectively. Amino acids located in Loops BC, DE and HI are shown on a grey background.

Position  (PCV2)	PCV2a (25 isolates)	PCV2b-1A/1B (10 isolates)	PCV2b-1C (12 isolates)	PCV1 (3 isolates)
3a	Y	Y	Y	We
4	1Q/24Pb	P	1S/11Pd	P
8	12F/13Y	Y	3Y/9F	Y
13	1L/1R/23H	1N/9H	H	T
21	1H/24Q	Q	Q	N
28	W	W	W	Y
30	4L/21V	1L/9V	V	A
34	1Q/24 H	H	4Y/8 H	N
42	N	N	N	T
47	1S/4A/20T	T	T	1C/2S
52	T	T	T	E
53	1I/24F	F	4F/8I	F
54	G	G	G	V
55	Y	Y	Y	L
57	V	V	V	I
59	4R/21A	1A/9Rc	1R/2A/9K	G
60	2S/23T	1S/9T	T	G
62	V	V	V	1F/2Y
63	7T/6S/12R	2R/8K	2S/10R	S
64	T	T	T	Q
68	A	A	3A/9N	1N/2H
70	D	D	D	N
71	M	M	M	H
72	10L/15M	M	M	L
73	R	R	R	1K/1T/1R
75	1T/10K/14N	N	N	N
76	8L/17I	I	I	I
77	5N/20D	N	N	G
78	D	D	D	Q
80	1L/24V	L	L	L
83	G	G	G	S
86	T	S	S	T
88	K	P	P	P
89	I	R	L	L
90	S	S	T	P
91	1V/24I	V	V	L
94	E	E	E	Q
101	1I/24V	V	V	A
103	V	V	V	Y
106	W	W	W	Y
108	C	C	C	R
109	S	S	S	D
113	Q	Q	Q	S
114	G	G	G	1K/2M
115	D	D	D	1Q/2E
121	10S/15T	S	T	T
122	A	A	A	V
123	5I/20V	V	V	V
127	D	D	D	A
130	5F/20V	V	V	V
131	4I/9P/12T	T	T	T
132	K	K	K	P
133	5S/20A	A	A	S
134	1P/24T	T	2T/10N	T
135	A	A	A	N
136	4Q/21L	L	L	L
137	T	T	T	A
142	V	V	V	I
151	5T/20P	T	3P/9T	R
155	S	S	S	T
166	V	V	V	E
169	4G/21S	S	6G/6R	Q
173	Y	Y	Y	W
175	Q	Q	Q	H
183	1I/24L	L	1I/11L	L
186	R	R	R	H
188	Q	Q	Q	N
190	4T/21S	A	2S/10T	H
191	1K/4G/10R/10A	G	2A/10G	T
194	D	D	D	E
196	V	V	V	T
200	T	T	T	Y
202	F	F	F	L
203	E	E	E	Q
205	S	S	S	A
206	2I/23K	I	3K/9I	A
207	Y	Y	Y	T
208	D	D	D	A
210	D	E	D	M
212	N	N	N	V
213	I	I	I	V
215	V	V	1V/11I	L
217	M	M	M	I
225	N	N	N	I
232	4N/21K	1K/9N	1K/11N	N

Surface around the icosahedral fivefold axes of the capsid exhibits high divergences throughout PCV evolution

A typical viral jelly roll was apparent in structure-homology modelling of the PCV1 Cap, using crystal structure of the PCV2 Cap as a template (Fig. 1c). However some loops of the PCV1 Cap differed significantly from those of PCV2 Cap (Fig. 1b, d). The distribution and conformation of loops on surfaces of the PCV2 and PCV1 capsids are shown (Fig. 1e, f).

Loops BC, HI and DE determined surface patterns around the fivefold axes of the capsid

Loops BC, HI and DE were three of the shortest loops among all loops of the PCV Caps (Fig. 2). Although the three loops remained separate from one another in the primary structure, they were oriented close to neighbouring Cap subunits and decorated the icosahedral fivefold axes of the PCV2 capsid in the assembled capsid (Figs 1e, f and3). Loops BC featured the fivefold axes and extended furthest from the base of the PCV2 capsid. Five Loops BC, together with Loops HI formed a small ‘crown’ on each fivefold axis of the capsid (Fig. 3m). Comparative sequence alignment indicated that Loop BC, the most divergent loop, was different between PCV1 and PCV2 not only in reside number, but also in composition (Fig. 2). Therefore, this gave rise to a distinct surface feature on the Cap subunits (Fig. 3a–c), and further generated distinct surface patterns around the icosahedral fivefold axes of the capsid after the Caps assembled. Loop BC of PCV2 contained nine residues, one more than Loop BC of PCV1 (Fig. 2). There was no sequence match from positions 59 to 64 in this loop between PCV1 and PCV2 (Fig. 2). However, with the exception of one isolate (PCV2b-1C: AEN03330, Fig. 2), three residues at positions ⁵⁸K and ⁶⁵PS⁶⁶ in the loop were conserved among all isolates. Notably, within PCV2 subtypes, four or five (of nine) residues of Loop BC were either T or S. Positions 61 and 64 were T, position 66 was S and position 60 was S or T (Fig. 2 and Table 1). These T and S residues were located on the capsid surface (Fig. 4a, yellow). In addition, Loop BC in the cluster of PCV2a contained one or two positively charged residue(s) versus two or three in PCV2b. Notably these positively charged residues were distributed on the capsid surface and surrounded the icosahedral fivefold axes of the PCV2 capsid (Fig. 4a, red). However, there was no negatively charged residue in this loop (in either PCV1 or PCV2).

Fig. 3.

The icosahedral fivefold axes of the capsid decorated by Loops BC, HI and DE of the PCV2 Cap exhibit unique features. (a–c) Distribution of surfaces formed by Loops BC on 3D structures of Caps of PCV2 and PCV1. (a) Red indicates the surface of Loop BC on the PCV2 Cap. (b) Green indicates the surface of Loop BC on the PCV1 Cap. (c) Overlap of the Caps of PCV1 and PCV2. Panels (e–g) and (i–k) are the same as (a–c), with the exceptions of distinct loops. (e–g) Loops HI; (i–k) Loops DE. Surface distributions of Loops BC, HI and DE on the PCV2 capsid are displayed in (d) red, (h) magenta and (l) green, respectively. Loops BC (red), together with Loops HI (magenta) and DE (green) determined surface patterns of the icosahedral fivefold axes of the PCV2 capsid (m) and one magnified fivefold axis is displayed with colour-labelled Loops BC, HI and DE (n).

Fig. 4.

Loops coordination around the icosahedral fivefold axis of the PCV2 capsid. (a) The positively charged residues (red) and a number of serines or threonines (yellow) of Loops BC were distributed around the surface of the icosahedral fivefold axis of the PCV2 capsid. Blue indicates the surface composed of other residues of Loop BC. (b–d) Loop BC may interact with Loop HI via an electrostatic interaction between residue ²⁰⁸D (green) of Loop HI and ⁵⁸K (red) or ⁵⁹R (red) of Loop BC. The ²⁰⁷Y (magenta) residue stands out via interaction between Loops BC and HI. Conformations of the side chains of ²⁰⁸D206, ²⁰⁷Y207, ⁵⁸K and ⁵⁹R in the PCV2 Cap are visible in (b) and surface distributions of these residues on the PCV2 Cap and the capsid are shown in (c) and (d). (e, f) Loops CT (sky blue) distribution around Loops BC (red) decorated fivefold axes of the PCV2 capsid. Conformations of two residues (²³⁰PN²³¹) in Loops CT on the surface of the PCV2 capsid indicated in yellow (f). Note: the last two residues (²³²NP²³³) of the Cap were absent from the crystal structure.

Loops HI decorated the icosahedral fivefold axes along with Loops BC and Loops DE (Fig. 3h, m, n). This loop only contained five residues (Fig. 2) and was located between Loop BC and Loop DE on the capsid surface (Fig. 3m, n). For Loop HI, there was no match between PCV1 and PCV2, except for position ²⁰⁴N, but the loop was better conserved within the respective genotypes (PCV1 and PCV2; Fig. 2). Isoleucine (²⁰⁶I) in Loop HI, along with three other key residues adjacent to this loop, was recognized as a signature motif (Motif 2) that distinguished PCV2b from PCV2a (Fig. 2, residues labelled with stars). Deep sequence mining of Loop HI revealed that evolution of the signature motif from PCV2a to PCV2b was associated with an elevated potential for tyrosine phosphorylation at position ²⁰⁷Y in a conserved ²⁰⁶IYD²⁰⁸ motif (Figs 7a and S3c). Some pathogen structural proteins of this motif required phosphorylation for cell entry and dissemination (Backert et al., 2008). Furthermore, an isoleucine (²⁰⁶I) at position −1 was strictly conserved in substrates of non-receptor tyrosine kinase (Fig. 7a) (Songyang et al., 1995). The conserved, negatively charged ²⁰⁸D at position +1 in the motif was oriented between two positively charged residues (⁵⁸KR⁵⁹) in Loop BC (Fig. 4b, c and d) in our model.

Fig. 7.

Identification and characterization of the motifs of the PCV Caps. (a) A signature motif of the PCV2 Cap is similar to the conserved motifs of the pathogen proteins which are substrates of the non-receptor tyrosine kinase (c-Src and Abl). (b) Amino acid sequence comparison of the NLS-A and -B with known CPPs.

The distance from ⁵⁸K to ²⁰⁸D (6.1 Å) is insufficient to form an effective salt bridge in the 3D structure; however ⁵⁹R may interact with ²⁰⁸D via a salt bridge within an appropriate distance of 3.9 Å (Fig. 4b). Notably, the positively charged residue (R/K) at position 59 was exclusively present in the PCV2b genotype, not in either PCV2a or PCV1 (Fig. 2). We inferred that electrostatic interaction between the two charged residues distributed in distinct loops might have a crucial role in stabilization of loops in PCV2b. Based on the 3D structure, the conserved motif of IYD was located on the surface around the icosahedral fivefold axes of the PCV2 capsid (Fig. 4d). In contrast, this motif was absent in the non-pathogenic PCV1 (Fig. 2).

Loop DE contained 10 residues (Fig. 2); five of the loops aligned together to form a sealed base on the icosahedral fivefold axis of the PCV2 capsid (Fig. 3l–n, blue). Loop DE within PCV2 was strictly conserved and the consensus sequence in this loop was ¹⁰⁸CSPITQGDRG¹¹⁷. However, within PCV1 isolates, Loop DE was slightly divergent (¹⁰⁸RDPITS(K/N)(E/Q)RG¹¹⁷) compared with PCV2 (Fig. 2). Therefore, the surface composed of Loops DE on the Caps of PCV1 and PCV2 had slight differences due to sequence differences in the loop (Fig. 3i–k). Notably, a unique and conserved cysteine (¹⁰⁸C) in Loop DE of the PCV2 Cap, recognized as a key residue responsible for integrity of PCV2 VLP through a disulfide bond (Wu et al., 2012), was substituted by another conserved positively charged arginine in the PCV1 Cap (Fig. 2). Perhaps PCV1 uses an assembly/disassembly mechanism distinct from that of PCV2.

Loop CT contained a conserved PXXP motif in PCV2 isolates

The number of residues among Loop CT of the PCV Caps varied from eight residues in PCV1 to nine or ten residues in PCV2 (Fig. 2). Loops CT were evenly distributed around the Loop BC-decorated fivefold axes (Fig. 4e, f; sky blue). A strictly conserved PXXP motif present in Loop CT within all the PCV2 isolates was not apparent in PCV1 (Fig. 2), representing a significant difference between PCV1 and PCV2. In general, the canonical PXXP motif was recognized as a binding motif being capable of interacting with an SH3-domain-containing protein, and thereby it was involved in some crucial signal cascades or cellular processes in vivo. Based on the 3D structure, we inferred the ²³⁰PXXP²³³ motif projected from the PCV2 capsid surface (Fig. 4f), although the last two residues of ²³²NP²³³ were missing in the structure. However, an extra lysine (²³⁴K) at the end of the loop, exclusively present in a newly emerging PCV2b mutant (PCV2b/1C, Fig. 2), definitely contributed more features to the PCV2 capsid surface, compared with traditional PCV2b strains. Before the PXXP motif, there were four strictly conserved residues of ²²⁶LKDP²²⁹ between PCV1 and PCV2, which were also exposed on the surface of the PCV2 capsid, with the exception of ²²⁶L (Fig. 4f, sky blue).

Capsid surface around the icosahedral threefold axes is composed of Loops GH and exhibits a similar pattern between PCV1 and PCV2

Loop GH, containing 33 residues, was the largest and most conserved loop between the PCV Caps. Most residues in this loop were invariable between PCV2 and PCV1 (Fig. 2), so they exhibited a very similar surface pattern on both Caps (Fig. 5a–c). The icosahedral threefold axis of the capsid was stabilized by extensive interactions of three Loops GH from neighbouring Cap subunits (Fig. 5c). Based on the amino acid distributions of the loop on the PCV2 capsid, Loop GH could be separated into two parts (Fig. 6a, c). Part I decorated the capsid surface (Fig. 6a), and contained two lobes: Lobe 1 (¹⁶⁶VLDSTIDY¹⁷³; Fig. 6a, red) and Lobe 2 (¹⁸⁶RLQTTGNVD¹⁹⁴; Fig. 6a, green). Lobe 1 constituted the centre of the icosahedral threefold axis by interacting with two homogeneous lobes from neighbouring subunits; Lobe 2 was partially around the first lobe and near Loop CD and Loop EF (Fig. 6b). In contrast, part II was buried inside the capsid (Fig. 6c) and contained two peptides (¹⁶²TPKP¹⁶⁵ and ¹⁷⁴FQPNNKRNQLW¹⁸⁵), both of which were strictly conserved between PCV1 and PCV2 (Fig. 2). Three of the peptides (¹⁷⁴FQPNNKRNQLW¹⁸⁵) from neighbouring Loops GH extensively interacted with one another beneath the surface and formed a buried centre around the icosahedral threefold axis of the PCV2 capsid (Fig. 6c). Notably, two residues at positions 190 and 191 in Loop GH, together with another two residues at positions 206 and 210, were recognized as a signature motif (Motif 2) to differentiate PCV2 subtypes (PCV2a and PCV2b). Clearly, both residues were localized on the PCV2 capsid surface and in Lobe 2 of Loop GH (Fig. 6d, blue).

Fig. 5.

3D structure comparisons of Loops GH, CD and EF between PCV2 and PCV1, and their distributions on the PCV2 capsid surface. Surface patterns of Loops GH (Cyan), CD (yellow) and EF (orange) on the Caps of PCV2 (a, d, g) and PCV1 (b, e, h). Surface distributions of Loops GH (c), CD (f) and EF (i) on the PCV2 capsid. Loops BC in (c, f, i) are highlighted in red to reveal the icosahedral fivefold axes of the PCV2 capsid.

Fig. 6.

Features of Loop GH, Loop CD and Loop EF on the PCV2 capsid. Three Loops GH from neighbouring Caps constituted the centre of the icosahedral threefold axis of the capsid (a–d). (a) Lobes 1 (red) and Lobes 2 (green) of Loops GH formed the exterior surface around the icosahedral threefold axis of the capsid. (b) Extensive interaction of Lobes 2 with Loops CD and EF on the capsid surface. (c) The second part of Loop GH was buried inside the capsid and three of the portions from the neighbouring Caps interacted with one another. (d) Two signature residues (blue) of Loop GH, which distinguished PCV2a and PCV2b, were localized in Lobes 2 and exposed on the surface of the capsid. (e) One pair of Loops CD (yellow) and CTs (sky blue) from neighbouring Caps decorated the surface in an anti-parallel arrangement. (f) The PPGGGS motifs (blue) of Loops CD formed two ridges on the surface around the icosahedral twofold axis of the capsid. (g) The middle segment (orange) of Loop EF contributed most of the surface composed of Loop EF and only a few residues from the first (green) and the last segments (blue) were exposed on the surface of the capsid. (h) Loops EF (orange) were evenly distributed along with Loops BC (red) on the surface of the capsid.

Loops CD and EF

Loop CD was the third largest loop of the Caps with 18 residues. One-third of the sequence was invariable between PCV1 and PCV2 (Fig. 2). Several prolines (P) and glycines (G), recognized as ‘helix breakers’, were present in this loop. In general, three to five P and two or three uninterrupted G were in this loop (Fig. 2). Surface patterns of the Caps contributed by the residues of Loop CD were similar between PCV1 and PCV2 (Fig. 5d, e). Most residues of Loop CD were distributed on the capsid surface (Fig. 2) and decorated the icosahedral twofold axes of the PCV2 capsid (Fig. 5f). Two Loops CD from neighbouring subunits, along with two Loops CT, were tightly aligned by means of an anti-parallel arrangement on the PCV2 capsid surface (Fig. 6e), whereas the ⁸¹PPGGGS⁸⁶ motifs in Loops CD formed two ridges on the capsid surface (Fig. 6f). A motif (Motif 1) containing six residues (⁸⁶TNKISI⁹¹ in PCV2a versus ⁸⁶SNPRSV⁹¹ in PCV2b) in the loop was also recognized as a signature motif to differentiate PCV2a from PCV2b (Fig. 2a, residues labelled with stars in Loop CD; Fig. 6f, red).

Loop EF was the second largest loop in the Cap and contained 24 residues. In this loop, residues exposed on the PCV2 capsid surface, were highly divergent (Fig. 2). Therefore, based on sequence conservation, Loop EF might be divided into three segments. The first segment, containing six residues (124–129), was conserved between PCV1 and PCV2, except position of D127A (Fig. 2); the last three residues of this segment were exposed on the capsid surface (Fig. 6g, green). The second segment, containing eight residues from positions 130 to 137, was entirely exposed on the PCV2 capsid surface (Fig. 6g, orange); amino acid residues of this segment were highly divergent between PCV1 and PCV2, but they were strictly conserved within PCV1 isolates (Fig. 2). The third segment was almost buried inside the capsid, with the exception of ¹³⁸Y (Fig. 2). In this segment, residues were invariable between PCV1 and PCV2 (Fig. 2). The surface contributed by Loops EF was also evenly scattered around Loops BC-decorated fivefold axes of the PCV2 capsid (Fig. 6h, orange).

Nuclear localization signal (NLS)

Amino acid residues of the NLS were conserved between PCV1 and PCV2; a unique feature was that the NH₂ terminal end of the PCV Caps contained a large amount of positively charged residues (Arg). Overall, half of these residues were positively charged (Fig. 2). The NLS of the PCV1 Cap had two more residues than that of PCV2 Cap (Fig. 2). Although there was no 3D structure available for the NLS, the secondary structure was confidently predicted (Fig. S2). Based on sequence analysis, the NLS might be split into two stretches, which were spaced by a predicted α-helix (¹⁸HLGQILR²⁴; Figs 2 and S2). The two stretches were designated NLS-A and B, both of which were enriched with several arginines. The NLS was predicted to package viral genomic DNA in the capsid and might have a capacity to interrupt the cell membrane or intracellular membrane during cell entry of virus through its transient externalization. Therefore, we compared NLS-A and NLS-B with typical cell-penetrating peptides (CPPs), e.g. TAT in HIV and artificial polyarginines; both NLS-A and -B were similar to known CPPs (Fig. 7b).

Discussion

In a non-enveloped virus, the capsid not only functions to protect and package the viral genome from degradation or denaturation, but also determines cellular tropism via engagement of the virus with various factors of host cells. The capsid contains all information regarding how a virus binds host cells and how/when/where the viral genome is dissociated from its protective protein shell (capsid) to enter the precise site for replication via a stepwise, spatiotemporal engagement with host cell factors. Since PCV2 is a non-enveloped, single-stranded DNA virus and the Cap is the sole structural protein, the simplest nano-machine might be exploited as a model to study mechanisms of how a capsid is self-assembled from 60 subunits, how capsid rearrangement is induced by host cell factors, and how a viral genome escapes from the metastable capsid and reaches the viral replication site in vivo.

PCV2 capsid structures were determined by both X-ray diffraction and cryo-EM (Khayat et al., 2011; Liu et al., 2016), and both of the structures are quite consistent. Therefore, we inferred that the icosahedral fivefold axes determined by Loops BC, HI and DE produced dramatic differences between PCV1 and PCV2 on the capsid surface, due to highly divergent sequences in these loops, even though both genotypes share 68–76 % nucleotide identity in genomic DNA (Fenaux et al., 2004; Hua et al., 2013). Although the three loops are distant in the primary structure of the Cap, they are close in the 3D structure (Fig. 1a–d). Furthermore, Loops BC, HI and DE from five neighbouring Cap subunits orient around the icosahedral fivefold axis on the capsid. Thus, residues of the three loops determine surface features around the fivefold axes of the capsid. In addition, another divergent Loop CT is also near the fivefold axis on the capsid surface. Therefore, surface patterns determined by Loops BC, HI, DE and CT have great differences between PCV1 and PCV2, and even within PCV2, as genotype sequences of Loops BC, HI and CT exhibited divergences in PCV2 evolution (Fig. 2 and Table 1). Results of previous studies have indicated that the surface composed of Loops BC, HI, DE and CT could form conformational epitopes recognized by several neutralizing antibodies (NAs) (Huang et al., 2011; Lekcharoensuk et al., 2004; Liu et al., 2013; Shang et al., 2009). Recently, a newly emerging PCV2b mutant (PCV2b/1C) with an extra lysine at the end of Loop CT has been frequently isolated on PCV2-vaccinated farms worldwide (Reiner et al., 2015; Seo et al., 2014; Xiao et al., 2012). More recently, this mutant was fitted into a new subtype (PCV2d) within phylogenetic trees (Franzo et al., 2016). The extra lysine changed the capsid surface pattern in this mutant, and might decrease avidity of NAs to epitopes on the capsid surface(Zhan et al., 2016). Indeed, it has been reported the novel mutant, isolated from a farm with PCV2 vaccination failure, was eradicated by changing the vaccination programme to two doses of vaccine (Xiao et al., 2012).

Loops DE from the five neighbouring Cap subunits constitute the base of the fivefold axis, which looks like a ‘gate’ formed by five triangles (Fig. 1e, f). A cysteine (¹¹⁰C) of Loop DE was strictly conserved within PCV2 genotypes (Fig. 2). It has been reported that the PCV2 Cap subunits can be dimerized via a disulfide bond in the absence of β-mercaptoethanol in vitro (Wu et al., 2012). Therefore, this unique cysteine is recognized as a critical residue responsible for assembly and integrity of the PCV2 capsid. Notably, the conserved cysteine in PCV2 was replaced with an arginine in PCV1 (Fig. 2), indicating that PCV1 capsid might exploit an assembly mechanism distinct from that of PCV2. Both Loops HI and BC projected from the base and determined surface patterns around the icosahedral fivefold axes of the PCV capsid. Loop BC of PCV2 Cap was T- and S-enriched. Phosphorylation prediction by software indicated that ⁶¹T, ⁶⁴T and ⁶⁶S had high potentials to be phosphorylated (Fig. S3a, b). A conserved motif (²⁰⁶IYD²⁰⁸) of Loop HI, only present in PCV2b isolates, is a preferential substrate of many non-receptor tyrosine kinases such as members of the Src kinase family (SKF) and Abl (Backert et al., 2008). This tyrosine in the motif was also predicted to have a high possibility of being phosphorylated (score>0.9; Fig. S3c). Khayat et al. (2011) suggested that the NLS of the PCV2 Cap localizes around the fivefold axis of the capsid. It is located inside and may be responsible for packaging the viral genomic DNA. Meanwhile, the positively charged NLS might also have a membrane-disrupting role during virus infection in order to release viral genome into the cytoplasm from its protein shell (capsid) during replication. Although the NLS is localized inside the capsid in the native virus, it might be externalized when capsid conformation switches into a metastable state induced by factors from host cells during cell entry. These factors include receptors, proteases, pH, oxido-reducing state, mechanical force, etc., all of which can cause conformation change of capsids in many non-enveloped viruses (Suomalainen & Greber, 2013). Results of previous studies have indicated that PCV2 can enter host cells via an actin-dependent endocytosis and it has also been demonstrated that PCV2 exploited distinct pathways to enter host cells (Misinzo et al., 2005; 2009). However, how, when and where the viral genome escapes from the capsid shell into a DNA replication factory in cells for infection are still unknown. In that regard, previous investigations mainly focused on tracking the PCV2 Cap during cell entry, and it is very hard to label the viral genome and the Cap simultaneously to discover how/when the viral genome escapes from the capsid in vivo. Since the NLS of the PCV Cap was located proximally to the fivefold axis of the capsid, perhaps the NLS is externalized through the site. Notably, both Loops BC and HI contained several potential phosphorylation sites (Figs 4a–c and S3) and the capsid might undergo another conformation alteration if these sites are phosphorylated. Moreover, the phosphorylated, negatively charged S, T and Y might also mimic phosphates in DNA to interact and stabilize the conformation of the externalized NLS. Further investigations need to be designed and conducted to elucidate underlying mechanisms of cell entry and viral DNA release.

To date, no research group has, to our knowledge, reported any post-translation modification of PCV2. Notwithstanding, it is still possible that during cell entry, some kinase(s) in host cells may phosphorylate specific residues of the capsid, or PCV2 might exploit/activate kinase(s) to modify its capsid or hijack some cell signalling pathways in host cells in order to complete or benefit its life cycle (entry, replication and egress). A potential tyrosine phosphorylation site was predicted in the PCV2 Cap in our model and the conserved tyrosine (²⁰⁷Y) was only present in PCV2 genotype, not in PCV1. Notably, the tyrosine-containing motif of PCV2b Cap (²⁰⁶IYD²⁰⁸) is present in other pathogenic proteins such as the surface protein A36R in vaccinia virus, Tir in enteropathogenic E. coli (EPEC) and CagA in Helicobacter pylori (Backert et al., 2008). Furthermore, the tyrosine residue in Loop HI is localized on the surfaces of both the PCV2 Cap subunit and the capsid. The side chain of the tyrosine stands out from the capsid surface around the icosahedral fivefold axis, so the tyrosine might be recognized, accessible and phosphorylated by specific tyrosine kinase(s) at the site. Furthermore, Loops HI and BC might be stabilized via electrostatic interaction between a conserved, negatively charged amino acid residue (²⁰⁸D) in Loop HI and a positively charged amino acid residue (⁵⁹K/R) in Loop BC. Remarkably, Loop BC extended furthest from the capsid surface and electrostatic interaction between the two loops may draw Loop HI further from the surface, which would facilitate the tyrosine access into a catalytic site of kinase(s) to be phosphorylated. However, the motif in the PCV2a Cap exhibited a slight variation at position −1 (²⁰⁶K in PCV2a versus ²⁰⁶I in PCV2b). Notably, this residue at position 206 has also been identified as one of the key residues in Motif 2 to differentiate PCV2a from 2b subtypes (Cheung & Greenlee, 2011). Indeed, based on results of chimera mutation experiments, PCV2 containing the ²⁰⁶IYD²⁰⁸ motif in Loop HI had a different distribution in vivo and produced a higher titre of virus progeny in cultured cells (Cheung & Greenlee, 2011). Regardless, the precise mechanism is unknown. It has been reported that tyrosine-phosphorylated pathogenic proteins can activate/change cell signalling pathways in host cells. For example, in vaccinia virus, phosphorylation of the surface protein A36R is required for actin polymerization and pathogenicity (Backert et al., 2008). Actin-based motility of vaccinia virus in vivo is triggered via the pathway of A36RPY-Nck/Grb2-WASP and treatment with STI-571 (a tyrosine-kinase inhibitor) dramatically reduced virus dissemination. Therefore, if the tyrosine residue in Loop HI of PCV2 can be phosphorylated in vivo, there will be the following important questions: (1) What is the specific tyrosine kinase(s) in host cells? (2) How does PCV2 activate its specific tyrosine kinase(s)? (3) Can the specific tyrosine kinase phosphorylate both the Cap subunit and the assembled capsid? (4) If it can, does the phosphorylation lead to change of the capsid conformation and facilitate viral genome escape from the capsid? (5) Is PCV2 pathogenicity associated with tyrosine phosphorylation?

The PXXP motifs are often present in proteomes of infectious organisms, such as Mycobacterium tuberculosis, Plasmodium falciparum, human immunodeficiency virus 1 (HIV-1) and hepatitis E virus. The PXXP motif at the end of Loop CT of the PCV2 Cap is strictly conserved within all PCV2 isolates, but is absent in the PCV1 genotype (Fig. 2). In many cases, there is molecular ‘cross-talk’ between a PXXP motif and an SH3 domain, and the interactions are required for regulation of cellular signal transduction, membrane trafficking and cytoskeletal rearrangement (Pagano et al., 2013). A virus may manipulate cellular protein complexes to interact with SH3-containing target proteins of host cells for completing the viral life cycle. For example, the consensus PXXP motif of HIV-1 Nef can interact with the SH3 domains of the Src-family tyrosine kinases, especially Hck and Lyn (Saksela, 2011). However, alteration of the PXXP motif to AXXA destroyed this interaction and compromised HIV replication in primary human peripheral blood mononuclear cells (PBMCs). Thus, the conserved PXXP motif at the end of Loop CT may have a critical role in molecular ‘cross-talk’ with factors of host cells. Notably, the simultaneous presence of the conserved PXXP motif in Loop CT and the tyrosine in Loop HI on the surface of the PCV2 capsid strongly indicated that the PCV2 Cap or the capsid might act as true substrates of non-receptor tyrosine kinase(s) in vivo, since these kinases usually bind their substrates via interaction between their SH3 domain and a conserved PXXP motif of substrates and then phosphorylate specific tyrosine residue(s) in substrates (Backert et al., 2008). In particular, simultaneous presence of both PXXP and IYD motifs with high-frequencies (60 pairs) on a small spherical surface of the PCV2 (only 17 nm in diameter) will form a robust and unique ‘molecular platform’ in vivo, and might interact with factors of host cells, hijack cell signal pathways and benefit the viral life cycle, whereas these features on the PCV1 capsid surface are all absent.

Altogether, a possible model of how PCV2 enters host cells and how its genomic DNA escapes from the capsid were proposed using in silico analyses and comparisons of primary and 3D structures of the Caps of PCV1 and PCV2. A metastable conformation of the PCV2 capsid might be induced by engagement of factors (i.e. receptor) in host cells during virus entry. In the course of PCV2 penetration into cells, the icosahedral fivefold axes of PCV2 capsid may be destabilized, due to conformation changes induced by phosphorylation of Loops BC and/or HI, and then the NLS from five neighbouring PCV2 Caps at one icosahedral fivefold axis of the capsid may be transiently externalized. The externalized NLS conformation could be further stabilized through electrostatic interaction between the positively charged NLS and a number of phosphorylated amino acid residues of Loops HI and BC. Finally, the membrane could become permeable to PCV2 genomic DNA via penetration of the externalized NLS into the membrane. Further investigations need to be designed and conducted to answer all the question arising from these results.

Methods

Collection of PCV genomic DNA sequences from 50 isolates.

Genomic DNA of 50 PCV isolates was collected and applied to reconstruct phylogenetic trees in our study. Forty-three representative PCV isolates were originally collected from different countries and areas, and their genotypes had been determined in previous studies (An et al., 2007; Olvera et al., 2007; Wang et al., 2013; Wei et al., 2013; Zhan et al., 2016). These genomic DNA sequences were obtained from GenBank (http://www.ncbi.nlm.nih.gov/). In addition, seven newly emerging PCV2 isolates, currently prevalent in southern China, were recovered from diseased pigs with field cases of porcine multisystemic wasting syndrome. Genomic DNA sequences of the seven isolates were analysed in our laboratory and also submitted to GenBank (accession numbers: KJ867553, KJ867554, KJ867555, KJ867556, KM235961, KP112484 and KP112486).

Reconstruction of phylogenetic trees.

Phylogenetic trees, based on whole genomic DNA or cap gene sequences of PCVs, were reconstructed using mega5 with 500 bootstrap replicates and a Kimura two-parameter nucleotide substitution model (identified as the best-fitting substitution model implemented in mega5). Each cluster in the two phylogenetic trees is designated with its corresponding accession number in GenBank.

Comparative amino acid sequences alignment of Caps.

Fifty Cap amino acid sequences were derived from ORF2 of each PCV genome. ORF2 identification and translation were facilitated by DNAStar molecular software (version 7.10). Amino acid sequences of the 50 Caps were aligned with an online tool (http://multalin.toulouse.inra.fr/multalin/).

Three-dimensional (3D) mapping of Cap amino acid residues in PCV2 capsid.

Crystal structure data for the PCV2 Cap (subunit) and VLP were retrieved from a protein data bank (PDB ID: 3R0R, http://www.rcsb.org/) (Khayat et al., 2011). All 3D structures of Cap and capsid in the study were viewed and prepared using Pymol (Version 1.7.4.4) molecular graphics software.

Protein 3D structure-homology modelling of PCV1 Cap.

A representative PCV1 isolate (accession number AGI99550) was identified via Cap amino acid sequence alignment within all the PCV1 isolates deposited in GenBank (accession numbers AHL83506, AHL83504, AGI99550, AEN19720, ACL35743, ABP98816, AAN77864, AAT72756, AAO61774 and U49186). This PCV1 Cap amino acid sequence was employed for protein 3D structure-homology modelling using the crystal structure of the PCV2 Cap (PDB accession number: 3R0R) as a template. Swiss-Model online server (http://swissmodel.expasy.org/) and Modeller (http://salilab.org/modeller/) were used for protein modelling (Biasini et al., 2014).

3D structure-homology modelling of PCV1 capsid.

The icosahedral structure of the PCV1 capsid was generated based on the monomeric structure of the PCV2 Cap by applying a VMD 1.9 matrix transformation using Tcl script (Humphrey et al., 1996). Histidine residues were protonated at pH 6.5 using the PDB2PQR server 6 (Greenidge et al., 2013). The input files of the structures were processed using XLeap from the AMBER 12 program (Case et al., 2005), using AMBER99SB force fields. A truncated octahedral box of pre-equilibrated TIP3P water was added to solvate each system with a pad of 10 Å. Counter ions were added to neutralize each system.

Prediction of serine (S), threonine (T) and tyrosine (Y) phosphorylation.

Phosphorylation of S, T and Y was predicted by online server NetPhos 2.0 (http://www.cbs.dtu.dk/services/NetPhos/). The amino acid sequence of the PCV2 Cap for phosphorylation prediction was the same as that for crystal structure.