Structure-Dependent Modulation of Alpha Interferon Production by Porcine Circovirus 2 Oligodeoxyribonucleotide and CpG DNAs in Porcine Peripheral Blood Mononuclear Cells

Abstract

DNA sequences containing CpG motifs are recognized as immunomodulators in several species. Phosphodiester oligodeoxyribonucleotides (ODNs) representing sequences from the genome of porcine circovirus type 2 (PCV2) have been identified as potent inducers (ODN PCV2/5) or inhibitors (ODN PCV2/1) of alpha interferon (IFN-α) production by porcine peripheral blood mononuclear cells (poPBMCs) in vitro. In this study, the IFN-α-inducing or -inhibitory activities of specific phosphodiester ODNs were demonstrated to be dependent on their ability to form secondary structures. When a poly(G) sequence was added to a stimulatory self-complementary ODN, high levels of IFN-α were elicited, and the induction was not dependent on pretreatment with the transfecting agent Lipofectin. In addition, the IFN-α-inducing ODN required the presence of an intact CpG dinucleotide, whereas the inhibitory activity of ODN PCV2/1 was not affected by methylation or removal of the central CpG dinucleotide. Of particular significance, the IFN-α inhibition elicited by ODN PCV2/1 was only effective against induction stimulated by DNA control inducers and not RNA control inducers, indicating activity directed to TLR9 signaling. The PCV2 genome as a whole was demonstrated to induce IFN-α in cultures of poPBMCs, and the presence of immune modulatory sequences within the genome of PCV2 may, therefore, have implications with regard to the immune evasion mechanisms utilized by PCV2.

Among DNA viruses, the porcine circovirus type 1 (PCV1) and PCV2 have a unique genome organization and composition insofar as they are the smallest autonomously replicating mammalian DNA viruses, characterized by their circular, single-stranded genome of only 1.7 kb (44).

PCV2 infects monocytes/macrophages and dendritic cells (DCs) persistently without obvious deleterious effects and with limited or no evidence of virus replication (2, 15, 36, 47). It is thought that the occurrence of monocytes/macrophages persistently infected with PCV2 facilitates immune evasion and enables the virus to be readily and rapidly disseminated throughout the host. It is still unknown how PCV2 becomes activated, but in clinically diseased pigs, such as those presenting with the PCV2-associated disease postweaning multisystemic wasting syndrome, a number of organs, including the liver, lungs, and kidneys, are affected, and an abundance of the PCV2 antigen can be detected in the lesions (2). In vitro studies have confirmed that PCV2 can persist in porcine myeloid DCs and porcine plasmacytoid DCs (PDCs), also referred to as natural interferon-producing cells (NIPCs) (11, 43), with no effect on their viability or expression of differentiation antigens. However, the ability of PDCs to produce the cytokines alpha interferon (IFN-α) and tumor necrosis factor alpha is clearly diminished in the presence of PCV2 (46).

It has been postulated that PCV2 elicits immunomodulatory effects through the presence of both immune stimulatory and inhibitory DNA sequences within the viral genome (19). Bacterial and viral DNA containing unmethylated CpG motifs can act as danger signals in vertebrates, activating innate immunity through Toll-like receptor 9 (TLR9) (39). The ability of CpG motifs to elicit immunomodulatory effects has been demonstrated using synthetic oligodeoxynucleotides (ODNs), and extensive research has been performed with regard to their possible utility in the field of medical therapeutics (1, 22). Earlier studies have focused mainly on human and murine systems and have preferentially utilized stable nuclease-resistant phosphorothioate ODN derivatives. However, it has been observed that multiple injections of phosphorothioate ODNs can elicit undesired effects, such as lymphadenopathy (29) and immunosuppression (20). In contrast, ODNs synthesized with a natural phosphodiester backbone do not elicit any such adverse side effects but, due to rapid degradation in vivo and/or hampered cellular uptake, most phosphodiester ODNs appear to be much less potent immune modulators (39).

In vitro studies have shown that many phosphodiester ODNs require pretreatment with a transfecting agent such as Lipofectin to induce IFN-α production in human and porcine cells (11, 33). One such phosphodiester ODN (ODN H) (42) contains the central hexamer TTCGAA, analogous to a DNA sequence found in the blood of a patient presenting with the autoimmune disease systemic lupus erythematosus (42). Modifications such as the addition of poly(G) sequences and/or double-stranded conformation of ODN H have been demonstrated to increase the IFN-α-inducing capacity elicited by ODN H, but the requirement for pretreatment with Lipofectin still remains a prerequisite for any observed IFN-α induction (11). These results are in agreement with studies using human peripheral blood mononuclear cells (PBMCs), which have shown that the secondary structure can be crucial for the cytokine induction and/or B-cell proliferation elicited by specific ODNs (9, 24).

In a previous study, five ODNs corresponding to 20-nucleotide (nt)-long sequences of the PCV2 (Imp. 1010 Stoon) genome were characterized with regard to their IFN-α modulatory capacity in porcine PBMCs (poPBMCs) (19). One of these ODNs (ODN PCV2/5) was selected based on the presence of the hexamer TTCGAA that is also present in the reference phosphodiester ODN H (33, 42). ODN PCV2/5 induced high levels of IFN-α, whereas another ODN from the PCV2 genome (PCV2/1) strongly inhibited the IFN-α production (19).

The present study therefore focuses on the importance of the central CpG dinucleotide and the ability to adopt a double-stranded DNA (dsDNA) conformation with regard to IFN-α stimulatory and inhibitory effects of phosphodiester ODNs. Specifically, derivatives of the previously characterized stimulatory (ODN H) (11) and inhibitory (ODN PCV2/1) (19) ODNs were used. The inhibition of IFN-α production was assayed in cultures of poPBMCs induced by other heterologous and reference ODNs, using viral and bacterial inducers as controls. In addition, the cumulative IFN-α stimulatory effect of the PCV2 genome as a whole was studied using a low-molecular-weight PCV2 DNA extract.

MATERIALS AND METHODS

Preparation of poPBMCs.

Yorkshire pigs (8 to 12 weeks of age) conventionally reared at the University Research Station at Funbo Lövsta, Uppsala, Sweden, were used throughout. PBMCs were isolated following density gradient centrifugation (30 min; 500 × g) on Ficoll-Paque (Pharmacia, Uppsala, Sweden) from heparinized blood collected in 10-ml evacuated test tubes (BD Vacutainer Systems, Plymouth, United Kingdom) from the vena cava cranialis. After three washes in phosphate-buffered saline (PBS), the cells were suspended at a concentration of 10 × 10⁶ cells/ml in RPMI 1640 medium (BioWhittaker; Cambrex Bioscience, Verviers, Belgium) supplemented with 20 mM HEPES buffer, 2 mM l-glutamine, 200 IU penicillin/ml, 100 μg/ml streptomycin, 0.5 μM 2-mercaptoethanol and 5% (vol/vol) heat-inactivated fetal calf serum (Invitrogen Life Technologies).

Establishment of cell cultures for IFN-α assays.

Cultures of poPBMCs (200 μl) were established in microtiter plates (Nunc; Roskilde, Denmark) at a final concentration of 5 × 10⁶ cells/ml in the presence of specific ODN preparations, control plasmid DNA, or virus preparations at concentrations previously determined as optimal for IFN-α induction (11, 19). Cell culture supernatants were collected after 20 h of incubation at 37°C, 7% CO₂, and stored at −20°C until further use.

Inducers/inhibitors of IFN-α production.

Synthetic or natural preparations of nucleic acids and control DNA or RNA viruses were used in IFN-α induction assays. Reference and modified ODNs (as specified in Table 1), were purchased desalted and dissolved in double-distilled water (Cybergene, Stockholm, Sweden). The ODNs were subsequently aliquoted and stored at −80°C until further use. The potential to form secondary structures and the corresponding ΔG values (kcal/mol) for each of the individual ODNs were determined using the IDT SciTools Oligo Analyzer 3.0 software (Integrated DNA Technologies, Coralville, IA) and are depicted in Fig. 1. In cell cultures, all ODNs were used at a final concentration of 25 μg per ml, with the exception of ODN 2216, which was used at 5 μg per ml. Double-stranded ODN DNAs were formed by heating a mixture of equimolar amounts of the respective ODNs to 95°C for 5 min and allowing them to cool slowly to room temperature (RT) followed by incubation for 30 min at RT (32). Denaturation of ODN 2216 was performed by heating to 100°C for 5 min followed by rapid cooling and storage on ice until further use. Double-stranded polyriboinosinic-polyribocytidylic acid [poly(I:C); Pharmacia, Sweden] was dissolved at a final concentration of 2 mg per ml saline in accordance with the manufacturer's instructions and stored at +4°C. In cell cultures, poly(I:C) was used at a final concentration of 5 μg per ml.

TABLE 1.

Sequence of ODNs used for induction or inhibition of IFN-α production in poPBMCs

ODN	Sequence (5′-3′)a	Origin or reference
H	TTTTCAATTCGAAGATGAAT	42
I	ATTCATCTTCGAATTGAAAA	33
H1a	GGTATTTCGAAATAGGGGGG	New modification of ODN H
H1amet	GGTATTTCMGAAATAGGGGGG	New modification of ODN H
H1b	GGGGGGTATTTCGAAATAGG	New modification of ODN H
H1c	GGGGGGTATTTCGAAATAGGGGGG	New modification of ODN H
H2a	GGTTCGAAGGGGGG	New modification of ODN H
H2b	GGGGGGTTCGAAGG	New modification of ODN H
H2c	GGGGGGTTCGAAGGGGGG	New modification of ODN H
H3b	GGGGGGTATTTCGAATAAGG	New modification of ODN H
H3c	GGGGGGTATTTCGAATAAGGGGGG	New modification of ODN H
HG-tail 2	GGTTCAATTCGAAGGGGGGG	11
IG-tail 2	GGTTCGAATTGAAGGGGGGG	New modification of ODN I
2216	ggGGGACGATCGTCgggggG	27
		23
PCV2/1	CCCCCCTCCCGGGGGAACAA	19
PCV2/1met	CCCCCCTCCCMGGGGGAACAA	New modification of ODN PCV2/1
PCV2/1a	CCCCCCTCCCAAAGGAACAA	New modification of ODN PCV2/1
PCV2/1b	CCCCCCTAAAGGGGGAACAA	New modification of ODN PCV2/1
PCV2/1c	CCCCCCTAACGAAGGAACAA	New modification of ODN PCV2/1

^aNucleotides in upper case have a phosphodiester backbone, whereas nucleotides in lower case have phosphorothioate backbone. CG sequences are underlined. C^M corresponds to methylated cytosine.

FIG. 1.

Predicted secondary structure formation of ODNs. (a) Spontaneous self-dimer formation of ODNs H1a, b, and c; H2a, b, and c; H3b and c; and 2216 and the double-stranded formation generated after hybridization of ODNs H^Gtail2 and I^Gtail2. (b) Spontaneous hairpin formation of ODN PCV2/1 and its derivatives PCV2/1a, b, and c. The double-stranded conformations were predicted using the IDT SciTools Oligo Analyzer 3.0 software. In all cases, the conformation with the lowest ΔG value (kcal/mol) is given.

For control bacterial DNA, the plasmid pcDNA3 (Invitrogen Life Technologies, Carlsbad, CA) was purified using the EndoFree Plasmid Maxi kit (QIAGEN, Hilden, Germany) and used at a final concentration of 2.5 μg per ml in cell cultures. As a control DNA virus, the Aujeszky's disease virus (ADV) strain Bartha (10⁵ 50% tissue culture infective doses [TCID₅₀]/ml) was inactivated by four cycles of UV radiation (1 J cm⁻²) and diluted 1:100 in RPMI media prior to use. As a control RNA virus, Sendai virus (SV) was propagated in eggs, and the chorioallantoic fluid was diluted 1:10 in RPMI media prior to use in IFN-α induction assays.

Where indicated, the inducers were tested both with and without preincubation with 5.0 μg/ml Lipofectin (Invitrogen Life Technologies, Carlsbad, CA) for 15 min at RT as described previously (19).

Detection of secreted IFN-α and IFN-α-producing cells.

Porcine IFN-α was quantified by dissociation-enhanced lanthanide fluoro-immunoassay based on the use of two monoclonal antibodies to poIFN-α (F17 and K9; kindly provided by Bernard Charley, Jouy-en-Josas, France). The concentration of poIFN-α (U/ml) in cell culture supernatants was determined by comparison with a laboratory standard of natural poIFN-α. The lower limit for detection of poIFN-α was 0.3 U IFN-α per ml (3).

The enzyme-linked immunospot (ELISPOT) assay used for enumeration of the IFN-α producing cells was that initially described by Nowacki et al. (38) with the following modifications. In brief, microtiter nitrocellulose plates (Multiscreen HTS; Millipore, Bedford, MA) were coated with monoclonal antibody F17 (7.5 μg/ml PBS, 100 μl per well) overnight at +4°C. After two washes in sterile water, the wells were blocked for 2 h at 37°C with RPMI culture medium supplemented with 10% (vol/vol) fetal calf serum. For all ELISPOT assays, cultures of poPBMCs were established at three different cell concentrations (0.5, 1, and 2 × 10⁶ cells/ml) in the presence or absence of IFN-α inducers in a final volume of 100 μl per well. Each type of culture was set up in triplicate.

Following incubation for 18 h at 37°C, 7% CO₂, the poPBMCs were discarded and the plates were washed six times with PBS containing 0.01% (vol/vol) Tween 20 (PBS-Tween) before incubation with 100 μl per well of horseradish peroxidase-conjugated monoclonal antibody K9 (diluted 1:2,000 in PBS plus 0.5% bovine serum albumin). The plates were then incubated for a further 2 h at 37°C, prior to three washes with PBS-Tween and followed by three washes with PBS. For subsequent horseradish peroxidase detection, 100 μl diaminobenzidine solution (Sigma-Aldrich, St. Louis, MO) was added to each well and the reaction was stopped after 5 min at RT by washing the plates thoroughly in water. The plates were then placed in the dark at RT prior to ELISPOT enumeration at low magnification using a macroscope. The number of spots (IFN-α-producing cells) is given as a mean value of the three replicates at each cell concentration.

Cell viability and apoptosis.

Cell viability and apoptosis assays were performed using flow cytometry (FACScan; BD Biosciences, San Jose, CA) following staining with propidium iodide (PI) and annexin V (BD Bioscience) as described previously (5). The proportion of poPBMCs that were negative for both PI and annexin V staining, indicating live cells, were determined using the CellQuest software (BD Biosciences).

Restriction enzyme digestion of “Hirt” extracts and Southern blot analyses.

A low-molecular-weight DNA extract containing viral DNAs was purified from PK-15A cells infected with the Imp. 1010 Stoon isolate of PCV2 using the “Hirt” method as described previously (35). The PK-15A cell line has previously been shown to be free from PCV, porcine pestivirus, adenovirus, parvovirus, transmissible gastroenteritis virus, and porcine reproductive and respiratory syndrome virus (35). Where indicated, RNase-free DNase I (QIAGEN) was used according to the method of Lövgren et al. (30) for the digestion of DNAs in the “Hirt” extract. Restriction endonuclease (RE) analyses were performed using MspI, HpaII, DpnI, MboI, and EcoRI (New England Biolabs, Inc., Beverly, MA) in accordance with the manufacturer's instructions. The isoscizomeric pair MspI and HpaII recognize the same sequence (5′-CCGG-3′), but HpaII cleavage is blocked by CpG methylation. DpnI and MboI recognize the same sequence (5′-GATC-3′), but DpnI cleavage requires methylation at the recognition site, whereas MboI cleavage is blocked by methylation at this site. For Southern blot analyses, the digested “Hirt” low-molecular-weight DNA extract was separated by agarose gel electrophoresis and blotted onto nitrocellulose filters as described by Sambrook and Russell (41). Southern blots were prehybridized prior to hybridization overnight with a full-length genomic ³²P-labeled PCV2 probe (Megaprime DNA labeling system; Amersham Biosciences, Uppsala, Sweden), washed, and then visualized following autoradiography using standard procedures.

Presentation of data.

To normalize for the expected variation in IFN-α-producing capacity between individual pigs (12), the results of the individual IFN-α induction assays in units of IFN-α/ml are expressed as a percentage of the level of IFN-α produced in response to the internal control pcDNA3. The results of the IFN-α inhibition assays are expressed as percent inhibition of IFN-α production in cultures where the inducer [ODN 2216, ADV, pcDNA3, SV, or poly(I:C)] was present alone, according to the following formula: percent inhibition = 100 − (100 × (U IFN-α per ml in cultures with inducer plus inhibitory ODN)/U IFN-α per ml in cultures with inducer alone). All data on IFN-α production given in the respective tables and figures are presented as mean values ± 95% confidence intervals, with nonoverlapping confidence intervals indicating statistically significant differences. Significance levels quoted in the text were determined following pairwise comparisons of IFN-α responses (U/ml) for PBMCs obtained from individual animals using Wilcoxon signed rank test (Statview 5.0; SAS Institute, Inc.).

RESULTS

Double-strand formation and poly(G) sequences enhance IFN-α induction by a CpG ODN.

The phosphodiester ODN H has previously been demonstrated to be an efficient inducer of IFN-α production in poPBMCs. Although hybridization with its complementary ODN (ODN I) or the addition of poly(G) sequences (ODN H^G-tail2) increased the IFN-α inducing capacity, these modifications did not circumvent the requirement for pretreatment with the transfecting agent Lipofectin for the induction of IFN-α (11).

In the present study, a complementary ODN to ODN H^G-tail2 was constructed (I^G-tail2) and used following hybridization to generate a double-stranded ODN (H^G-tail2 I^G-tail2) with an 11-nt-long complementary dsDNA sequence and poly(G) sequences at the 3′ ends. This double-stranded ODN induced high amounts of IFN-α regardless of pretreatment with Lipofectin or not (Table 2). When the two ODNs were tested separately in the absence of Lipofectin, only low amounts of IFN-α were induced, but when preincubated with Lipofectin, IFN-α levels comparable to those induced by the hybridized double-stranded ODN (H^G-tail2 I^G-tail2) were achieved. These results indicate that the presence of poly(G) sequences, as well as ability to form a double-stranded conformation, is of importance with regard to the IFN-α inducing capacity of phosphodiester ODNs.

TABLE 2.

Summary of IFN-α production induced by ODNs with different palindromes and poly(G) sequences^a

ODN	Palindrome (nt)	poly(G) sequence	IFN-α induction		n
			Without Lipofectin	With Lipofectin
HG-tail 2IG-tail 2	11	3′	114 ± 41	205 ± 66	10
HG-tail 2	6 + 5c	3′	16 ± 14	125 ± 48	10
IG-tail 2	6 + 5c	3′	0 ± 0	109 ± 36	10
H1a	12	3′	92 ± 29	151 ± 25	26
H1b	12	5′	62 ± 47	79 ± 23	6
H1c	12	5′ 3′	0 ± 0	35 ± 22	6
H1amet	12	3′	0 ± 0	0 ± 0	6
H2a	6	3′	0 ± 0	52 ± 32	6
H2b	6	5′	0 ± 0	3 ± 4	6
H2c	6	5′ 3′	24 ± 15	36 ± 22	20
H3b	6 + 6b	5′	0 ± 0	32 ± 12	24
H3c	6 + 6b	5′ 3′	20 ± 15	44 ± 22	24
2216	10	5′ 3′	151 ± 67	306 ± 113	16
2216 htd	10	5′ 3′	33 ± 26	349 ± 137	16

^aLevels of IFN-α produced are given as percentages of the levels induced by the control plasmid pcDNA3 ± 95% confidence intervals.

^bSix palindromic nt plus 6 nonpalindromic nt.

^cSix palindromic nt plus 5 nonpalindromic nt.

^dht, heat treated.

To further study the importance of double-stranded formation, a self-complementary ODN (H1a) was designed to contain a 12-nt palindromic sequence with the same central hexamer as in the immunostimulatory ODN H but with an additional 3′ poly(G) sequence (Fig. 1a). ODN H1a efficiently induced IFN-α in the absence of Lipofectin (Table 2), and the addition of a poly(G) sequence at the 5′ end (H1b) instead of at the 3′ end did not alter the levels of IFN-α induced (P > 0.05). However, the addition of dual poly(G) sequences at both the 3′ and 5′ ends (ODN H1c) instead of just at the 3′ (H1a) or 5′ (H1b) end reduced the IFN-α inducing capacity significantly (P < 0.05, n = 6). Methylation of the cytosine residue in the central CpG motif of ODN H1a (ODN H1a^met) totally abolished the IFN-α-inducing activity of the ODN, confirming the important role of CpG-motifs in IFN-α induction by phosphodiester ODNs.

Length of palindrome in PD-ODNs is important for IFN-α induction.

Effects of alterations to the palindrome on IFN-α induction are summarized in Table 2. In brief, reduction of the length of the palindrome sequence of ODNs H1a and H1b from 12 to 6 nt (Fig. 1a) (ODNs H2a and H2b) decreased the IFN-α induction significantly (P < 0.05). The addition of 6 nonpalindromic nucleotides adjacent to the palindrome of ODN H2b (ODN H3b) did not restore the IFN-α-inducing capacity of the ODN, whereas low to moderate levels of IFN-α were induced when these ODNs were supplemented with dual poly(G) sequences (ODNs H3c and H2c). Furthermore, all ODNs consisting of the hexameric palindrome with or without 6 nonpalindromic nucleotides and single or dual poly(G) sequences could induce moderate levels of IFN-α when preincubated with Lipofectin. In comparison, ODN H1c consisting of a 12-nt-long palindrome and dual poly(G) sequences did not induce any IFN-α in the absence of Lipofectin but could induce moderate IFN-α responses when preincubated with Lipofectin (Table 2). These results support the hypothesis that phosphodiester CpG ODNs need to be capable of adopting a double-stranded configuration of a certain length containing 3′ or 5′ poly(G) sequences to induce IFN-α production and that pretreatment with Lipofectin can circumvent some of these requirements.

To further elucidate the effect of Lipofectin, the phosphodiester/phosphorothioate chimera ODN 2216 was used. This ODN can form double strands spontaneously due to a 10-nt palindromic sequence that is surrounded by poly(G) sequences at both the 3′ and 5′ ends (25). ODN 2216 induced high levels of IFN-α both in the absence and presence of Lipofectin, but when ODN 2216 was heat treated to disrupt double-strand formations, the induction in the absence of Lipofectin was significantly reduced (P < 0.01, n = 16). However, this heat-treated preparation of ODN 2216 regained its full IFN-α-inducing capacity after incubation with Lipofectin (ODN 2216 with Lipofectin versus heat-treated ODN 2216 with Lipofectin; P > 0.05) (Table 2). These results further support the hypothesis that Lipofectin enhances the IFN-α-inducing capacity of ODNs via its ability to incorporate several ODNs within the same structure and that this multimeric form is important for IFN-α induction.

PCV2/1 inhibits IFN-α production induced by DNA controls but not RNA controls.

To further study the role of dsDNA sequences on the regulation of IFN-α production, the inhibitory ODN selected from the PCV2 genome (ODN PCV2/1) was used. This ODN can form secondary hairpin structures spontaneously through C-G pairing (Fig. 1b), and the importance of these formations was elucidated using modifications of the ODN with a reduced ability to form hairpins (ODNs PCV2/1a and PCV2/1c). The derivatives of ODN PCV2/1 are detailed in Fig. 1b.

For IFN-α assays, ODN PCV2/1 and its derivatives were used both individually and in combination with each of the control inducers [ODN 2216, ADV, pcDNA3, poly(I:C), or SV]. The results are expressed as percent inhibition of IFN-α production induced by the controls in the absence of PCV2/1 (Fig. 2). Pretreatment with Lipofectin was used only when it was required by the control inducer for IFN-α production, i.e., pcDNA3 and poly(I:C).

FIG. 2.

Inhibitory effect of ODN PCV2/1 and its derivatives PCV2/1a, b, and c on IFN-α production induced by the control ODN 2216, ADV, plasmid DNA (pcDNA3), poly(I:C), or SV. The IFN-α production is expressed as the percentage of inhibition of the IFN-α produced in cultures induced with the controls alone. The results are given as mean values ± 95% confidence intervals so that nonoverlapping confidence intervals indicate statistically significant differences (n ≥ 6). All derivatives of ODN PCV2/1 were used at a final concentration of 25 μg/ml. ODN 2216 and poly(I:C) were used at 5 μg/ml, and pcDNA3 was used at 2.5 μg/ml. ADV was used at a concentration corresponding to 10³ 50% infectious doses per ml before UV inactivation, and SV was diluted 1:10.

ODN PCV2/1 inhibited the IFN-α production induced by ODN 2216 and ADV completely (P < 0.001, n = 34 and 20, respectively). The IFN-α production induced by pcDNA3 was significantly inhibited by ODN PCV2/1 but to a lesser extent (P < 0.01, n = 14). In contrast, however, the IFN-α production induced by the RNA control inducers poly(I:C) and SV were not inhibited but were slightly increased (Fig. 2). The inhibitory activity of ODN PCV2/1 was not affected by pretreatment with Lipofectin or not but was strictly dependent on whether the IFN-α production was elicited by a DNA or RNA control inducer.

CpG motif is not essential for IFN-α inhibitory activity of ODN PCV2/1.

The importance of the central CpG motif for the IFN-α-inhibitory capacity of PCV2/1 was studied using modifications where the C was exchanged with an A (PCV2/1b) or altered by methylation (PCV2/1^met) as detailed in Table 1. ODN PCV2/1b retained full inhibitory capacity on IFN-α production induced by ODN 2216, ADV, and pcDNA3 compared to ODN PCV2/1 (Fig. 2) (P > 0.05). In addition, methylation of the cytosine in the CpG motif (PCV2/1^met) did not affect the IFN-α inhibitory activity (data not shown). These results clearly demonstrate that the presence of a CpG dinucleotide is not required for the inhibition of IFN-α production elicited by ODN PCV2/1.

Hairpin structure formation is essential for the IFN-α-inhibitory activity of ODN PCV2/1.

In contrast to ODN PCV2/1, the two ODNs with reduced ability to form hairpin structures (PCV2/1a and PCV 2/1c) (Fig. 1b) did not have any pronounced inhibitory effect on the IFN-α induction by ODN 2216 (P < 0.05), ADV, or pcDNA3 (Fig. 2). As shown for ODN PCV2/1, no inhibition but rather a stimulatory effect on IFN-α production induced by poly(I:C) and SV was observed (Fig. 2). These results clearly support the importance of the ability of ODN PCV2/1 to spontaneously form hairpin structures to exert its inhibitory activity on IFN-α production induced by DNA control inducers.

Inhibitory effect of ODN PCV2/1 is not due to cell death or dilution.

To exclude the possibility that the inhibitory function of PCV2/1 on IFN-α production was due to a toxic effect of the ODN, flow cytometric analyses were performed on poPBMCs cultured with ODN 2216 alone or in combination with derivatives of ODN PCV2/1 (PCV2/1, PCV2/1a, PCV2/1b, and PCV2/1c). After 20 h of culture, the proportion of cells labeled with annexin V and/or PI were enumerated. The percent viability of cells obtained from all cultures, including control untreated poPBMCs, were comparable and closely clustered within the same range (Table 3). Accordingly, it is clearly evident that PCV2/1 does not affect either cell viability or the degree of apoptosis in the cell cultures.

TABLE 3.

Proportions of viable PBMCs in cell cultures induced with ODN 2216 and different modifications of ODN PCV2/1

ODN	% of viable cells in pig no.a:
	1	2	3	4
None	79.7	64.7	62.4	59.7
2216	84.6	70.8	72.0	59.2
PCV2/1	84.9	64.0	67.6	57.7
PCV2/1 + 2216	83.0	67.2	69.0	61.7
PCV2/1a	NTb	NT	61.2	61.4
PCV2/1b	NT	NT	63.1	59.3
PCV2/1c	NT	NT	58.9	58.2
PCV2/1a + 2216	NT	NT	72.7	62.8
PCV2/1b + 2216	NT	NT	70.3	60.7
PCV2/1c + 2216	NT	NT	72.3	60.9

^aViable cells were determined as cells negative for staining with PI and annexin V using flow cytometry.

^bNT, not tested.

Furthermore, the inhibitory effect of ODN PCV2/1 cannot be explained by dilution of the IFN-α-inducing ODN in cell cultures, as an inert ODN (H1a^met) used at the same concentration as PCV2/1 (25 μg/ml) in combination with ODN 2216 (5 μg/ml) had only a slight and not significant effect on IFN-α production (data not shown).

Effect of ODN PCV2/1 on the number of IFN-α-producing cells.

The number of IFN-α-producing cells in the poPBMC cultures treated with either ODN 2216, ADV, or poly(I:C) alone or in combination with ODN PCV2/1 were enumerated by ELISPOT. In the presence of ODN PCV2/1, cultures induced by ODN 2216 or ADV showed a markedly decreased number of IFN-α-producing cells. In contrast, PCV2/1 did not affect the number of cells producing IFN-α in response to poly(I:C) (Fig. 3). Thus, the IFN-α-inhibitory activity of ODN PCV2/1 was due to a reduction of the number of IFN-α-producing cells in response to DNA control inducers rather than decreased amounts of IFN-α produced by each cell.

FIG. 3.

Effect of ODN PCV2/1 on the frequency of IFN-α-producing cells in cultures of poPBMCs induced with the control ODN 2216, ADV, or poly(I:C). The number of IFN-α-producing cells was determined by ELISPOT assay in cultures induced by the controls alone (solid lines) or in combination with ODN PCV2/1 (dashed lines). The spot frequency was determined at three cell concentrations, 0.5 × 10⁶, 1 × 10⁶, and 2 × 10⁶ cells/ml (pig no. 1, closed circles; pig no. 2, closed squares; pig no. 3, closed triangles wide base; pig no. 4, closed triangles narrow base; pig no. 5, open circle; pig no. 6, open squares).

IFN-α-inducing activity of the PCV2 genome.

To further study the intrinsic IFN-α-inducing capacity of the PCV2 genome as a whole, a low-molecular-weight “Hirt” DNA extract containing PCV2 DNAs was purified from PCV2-infected PK15A cells and assayed for its effect on IFN-α production in poPBMCs. The optimal concentration of the “Hirt” preparation was determined by titration in twofold dilutions. No IFN-α was detected in the absence of Lipofectin, whereas in the presence of Lipofectin, a plateau level of IFN-α production was found at a dilution of 1:1,000. The levels of IFN-α induced by the low-molecular-weight DNA preparation corresponded to 43% ± 33 of that induced by pcDNA3 (n = 4). The IFN-α-inducing capacity of the “Hirt” extract could be predominantly attributed to DNA content, since treatment with DNase reduced the IFN-α-inducing capacity from 90% ± 156 to 6% ± 11 of pcDNA3 (dilution, 1:1,000; n = 4). These results indicate a potential IFN-α-inducing capability associated with the PCV2 genome as a whole.

DNA methylation status of PCV2 RF DNAs.

Methylation of the CpG motif completely abrogated the IFN-α-inducing capacity (ODN H1a^met) but had no effect on the IFN-α-inhibitory activity of ODNs representing various parts of the PCV2 genome (ODN PCV2/1^met). The methylation status of PCV2 DNA was studied using the RE isochizomer pairs HpaII/MspI and MboI/DpnI which differ in their sensitivity to CpG methylation. Digestion with EcoRI was used as a control to linearize the PCV2 replicative form (RF) DNAs at a single site and provide a size reference for subsequent methylation-sensitive RE digestion.

There are two potential HpaII/MspI sites on the circular PCV2 RF with the recognition sequence 5′-CCGG-3′. Digestion with either HpaII or MspI produced a similar pattern (Fig. 4), suggesting that at neither of these two sites were any of the two C residues methylated in the predominant population of PCV2 RF molecules. In contrast, however, digestion with either DpnI or MboI produced differing restriction patterns, suggesting the occurrence of species with different methylation patterns at some of the five potential 5′-GATC-3′ recognition sites. For example, following DpnI digestion, the occurrence of a band at the position expected for the linearized form of the PCV2 RF was indicative of the lack of CpG methylation at the majority (4/5) of the DpnI recognition sites present in the PCV2 RF population as a whole. Thus, the results suggest that the PCV2 genome is methylated to a certain extent but not at all possible CpG motifs.

FIG. 4.

Southern blot analysis of low-molecular-weight DNA extracted from PCV2-infected PK15A cells. The methylation status of PCV2 DNA was studied using the RE isochizomer pairs HpaII/MspI and MboI/DpnI which differ in their sensitivity to CpG methylation as described in Materials and Methods. An asterisk indicates the specific RE of the pairs that is insensitive to methylation. Digestion with EcoRI was used as a control to linearize the PCV2 RF DNAs at a single site and provide a size reference. The positions of the linearized double-stranded RF of DNA (linearized dsDNA) and single-stranded covalently closed circular genomic DNA (circular ssDNA) are indicated by arrows.

DISCUSSION

Phosphodiester ODNs with central CpG motifs representing sequences found within the genome of PCV2 can either induce or inhibit the production of IFN-α by poPBMCs in vitro (19). In the present study, it was clearly demonstrated that a critical factor for the IFN-α regulatory role of the ODNs is the ability to form secondary structures.

Previous studies have shown that phosphodiester ODNs can be potent stimulators of IFN-α production by poPBMCs, but such ODNs generally require pretreatment with the transfecting agent Lipofectin (11, 19, 33). Cationic lipids such as Lipofectin or DOTAP that incorporate the ODNs in liposome complexes facilitate delivery to endosomes which may be crucial for the IFN-α induction by phosphodiester ODNs (21). Phosphorothioate ODNs, however, generally do not require pretreatment with a transfecting agent to elicit immune stimulation, possibly due to the resistance of the phosphorothioate backbone to degradation by endonucleases and thereby prolonging the effect of the ODN (1). In the present study, a reference phosphodiester ODN (ODN H) (42) containing the same central hexamer as an IFN-α stimulatory ODN from the PCV2 genome (PCV2/5) (19) was modified to study the importance of poly(G) sequences, CpG motifs, and self-dimer formation for the elicitation of IFN-α-inducing capacity in the presence or absence of Lipofectin.

By addition of a 3′ or 5′ poly(G) sequence to an ODN (H1a/H1b) containing a 12-nt palindrome, the requirement for pretreatment with Lipofectin to induce IFN-α was circumvented. The importance of a 3′ poly(G) sequence could be explained by the protection of the ODN from 3′ exonuclease activity in cell cultures (8). Furthermore, poly(G) sequences at the 3′ end have been suggested to contribute to the immunomodulatory activity of ODNs by the possible formation of G tetrads that enhance cellular uptake by scavenger receptors (10, 50). One scavenger receptor in particular, CXCL16, has been demonstrated to direct the uptake of ODNs containing poly(G) sequences to early endosomal vesicles rather than to lysosomal vesicles and thereby trigger IFN-α production instead of tumor necrosis factor alpha production (16). This effect would then be similar to that described for cationic lipids which direct the ODNs to early endosomes and retain the DNA there, resulting in a prolonged exposure to TLR9 (21). In the present study, an interaction between IFN-α stimulatory ODNs and TLR9 was indicated as methylation of the cytosine residue of the CpG motif in ODN H1a abolished the IFN-α-inducing capacity (H1a^met).

Phosphodiester ODNs containing an 11- to 12-nt palindromic sequence were consistently demonstrated to be the most efficient IFN-α inducers of the ODN H variants tested. When the palindrome was reduced to 6 nt, the levels of IFN-α were significantly decreased, which is in accordance with previous findings (34). The likelihood of the ODNs to form self-dimers is enhanced when the number of base pairs are increased, and consequently, these results suggest that secondary structure formation is of critical importance with regard to the IFN-α-inducing capacity of this ODN.

The importance of double strand formation for IFN-α induction was further supported from results generated using the chimeric ODN 2216 which has been described to form nanoparticle structures spontaneously (25). When the secondary structure of ODN 2216 was disrupted by heat treatment, the IFN-α-inducing capacity of the heat-treated ODN in the absence of Lipofectin was significantly reduced. This could, however, be reversed by pretreatment of the heat-treated ODN 2216 with Lipofectin, indicating that the incorporation of the ODN into a larger complex provided by cationic lipids mimics the spontaneously assembled nanoparticles present among native ODN 2216. These results are in agreement with earlier studies which have highlighted the importance of ODN aggregation for interleukin-12 production in mice (48).

Phosphodiester ODNs have not only been described as potent IFN-α inducers but certain ODNs have also been demonstrated to inhibit IFN-α production induced by other ODNs as well as by viral and bacterial inducers. In an earlier study, we showed that the IFN-α-inhibitory action of the phosphodiester ODN PCV2/1, derived from the genome of PCV2, was independent of Lipofectin pretreatment (19). In the present study, the IFN-α-inhibitory action of phosphodiester ODN PCV2/1 was demonstrated not to be dependent on the presence of a central CpG dinucleotide. Of particular significance, the fact that ODN PCV2/1 could completely inhibit the IFN-α production induced by CpG ODN 2216 as well as ADV, whereas the induction by the RNA control inducers SV and poly(I:C) remained unaffected, indicates that the inhibitory action of ODN PCV2/1 is specific for poPDCs. These cells are also referred to as NIPCs and have been demonstrated to respond with high IFN-α production to CpG ODNs (11, 17). Furthermore, the results of the ELISPOT assay showed that the inhibitory ODN PCV2/1 decreased the number of IFN-α-producing cells rather than the amount of IFN-α produced by each cell. Taken together, these results suggest that ODN PCV2/1 has an inhibitory activity that is directed to IFN-α production induced by TLR9 agonists rather than TLR3 agonists.

In the present study, it was clearly evident that modifications of the nucleotide sequence of ODN PCV2/1 that altered its ability to spontaneously form secondary hairpin structures abolished IFN-α inhibitory activity. DNA in hairpin and dumbbell secondary structures bind to transcription factors (6), and this may represent the underlying mechanism whereby the inhibitory ODN PCV2/1 functions on IFN-α production induced by TLR9 agonists. Hairpin and dumbbell structures of DNA are also more resistant to degradation by endonucleases than single-stranded DNA (ssDNA) (7), and this could prolong the inhibitory activity of ODN PCV2/1. In addition, the ability of PCV2/1 to form secondary structures could also explain why the inhibition observed was not dependent on pretreatment with Lipofectin.

Recent reports have suggested possible competition between inhibitory and stimulatory ODNs. Specifically, it has been postulated that binding of inhibitory ODNs to TLR9 causes a conformational change of the MyD88 binding sites which interrupts further signaling through TLR9 (4, 28). However, these studies were conducted using synthetic phosphorothioate ODNs, and ODNs with this backbone conformation have been demonstrated to exhibit unspecific binding to several proteins including TLRs (49). In addition, phosphorothioate ODNs also differ from phosphodiester ODNs with regard to specific sequence requirements, possibly due to structural alterations caused by the sulfur modification of the phosphorothioate ODN backbone (40). The interaction of stimulatory phosphodiester CpG ODNs with TLR9 is dependent on the presence of an unmethylated CpG motif (40), whereas the inhibitory activity of ODN PCV2/1 was unaffected by modifications of the CpG motif. It is therefore probable that ODN PCV2/1 does not elicit its effect through a simple steric binding to TLR9.

Using synthetic ODNs, it was clearly demonstrated that the inhibitory activity of ODN PCV2/1 was independent of the methylation status of the central CpG motif. Restriction enzyme analysis of the low-molecular-weight “Hirt” DNA extract containing PCV2 RF DNAs indicated the presence of both methylated and unmethylated CpG dinucleotides. The porcine circovirus virion ssDNA replicates via a double-stranded RF and consequently both double- and single-stranded PCV2 DNAs can occur naturally in cells of the monocyte/macrophage lineage, including monocyte-derived DCs and NIPCs (47). The net or cumulative effect of all of the stimulatory and inhibitory CpG motifs present within the PCV2 DNA species is likely to be dependent on the number of each type of motif and their position in relation to each other (26) and would explain our preliminary data suggesting the overall IFN-α-inducing activity in vitro of the PCV2 DNA species as a whole. Indeed, analysis of sera from piglets experimentally infected with PCV2 revealed detectable amounts of IFN-α in a majority of the pigs coinciding with PCV2 viremia (18). In contrast, a recent report using full-length cloned PCV2 DNA amplified in bacterial plasmids has been demonstrated to inhibit several functions of poPDCs, including IFN-α production, independently of virus replication (45). However, the amplification of PCV2 DNA within bacteria could affect the methylation status of the resultant DNA, and this may influence the interferogenic properties of the genome as a whole. Nevertheless, these conflicting results indicate possible IFN-α regulatory mechanisms of PCV2 that could be of importance during various stages of infection and manifestation of PCV2-related diseases in a similar manner as has been described for several other viruses (13, 14). In this context, it is of particular interest that comparisons of PCV1 sequences with PCV2 (Imp. 1010 Stoon) have revealed a number of nucleotide exchanges within the sequences representing ODNs PCV2/1 and PCV2/5, and the biological significances of these variations are currently being studied.

In addition, the results obtained in this study suggest the possible immunoregulatory activity of PCV2 DNA fragments in a similar manner as has been speculated for persistent herpes simplex virus type 1 infection (31). PCV2 enters monocytic cells by clathrin-mediated endocytosis and is dependent on endosomal acidification for infection (37). In this study, the PCV2 DNA preparation required pretreatment with Lipofectin for IFN-α induction, possibly because the cationic lipids directed the DNA to the endosomes where it could be exposed to TLR9 in the right acidic environment to elicit IFN-α induction. The occurrence of sequences within the genome of PCV2 with the potential to form immunomodulatory secondary structures may therefore represent a mechanism whereby PCV2 not only establishes clinically normal and persistent infections but also can progress to clinical disease such as postweaning multisystemic wasting syndrome.