Conjugation of a self-antigen to papillomavirus-like particles allows for efficient induction of protective autoantibodies (original) (raw)
Generation of SA–TNF-α fusion proteins. We took advantage of the strong interaction between biotin and SA to develop a flexible system allowing the biochemical conjugation of self-peptides onto preassembled VLPs (Figure 1a). Mouse TNF-α peptides were fused to the carboxy-terminus of a truncated form of SA using a plasmid and an expression system developed by Sano and Cantor (27). This segment of SA, which spans aa 16–135 of the mature protein, is less prone to aggregation than is the full-length protein, yet retains the ability to form tetramers and bind biotin (26). We generated three bacterially expressed SA proteins fused to mouse TNF-α peptides representing, respectively, aa 3–22, 29–47, and 3–54 of the mature protein. These mouse TNF-α peptides were chosen because they are homologous to regions of human TNF-α that were reported to interact with its receptor (31, 32).
Binding of SA–TNF-α fusion protein to biotinylated VLPs. (a) Drawing summarizing conjugated vaccine construction. BPV L1 VLPs were coated with activated biotin and then purified on a linear sucrose gradient. Biotinylated particles were incubated with purified SA–TNF-α fusion protein to generate conjugated particles. (b) Binding of SA–TNF-α fusion protein to VLPs. Binding of recombinant SA or purified SA–TNF-α(3–22) to wild-type or biotinylated VLPs was measured by ELISA. (c) Cosedimentation of biotinylated VLPs with SA–TNF-α(3–22). Biotinylated VLPs were incubated with SA–TNF-α(3–22) and then separated on a 24–54% sucrose gradient. Gradient fractions were analyzed by Western blot analysis, using an anti-L1 mAb (mAB837) (top panel) or an anti-SA Ab (bottom panel). Fraction 1 represents the bottom and fraction 10 represents the top of the gradient.
The biotin-binding activity of the SA–TNF-α fusion proteins was assessed by ELISA using biotinylated VLPs as the target antigen. The TNF-α peptides had no effect on the ability of the fusion proteins to bind biotin (Figure 1b and data not shown). The three fusion proteins bound biotinylated VLPs at a similar level as wild-type SA, and none bound to unbiotinylated VLPs. To determine the assembly state of the complexes, biotinylated VLPs were reacted with SA–TNF-α(3–22) and then subjected to analytical sucrose gradient centrifugation. Fractions were analyzed by slot-blot Western analysis using Ab’s to BPV-1 L1 and SA. As shown in Figure 1c, SA–TNF-α(3–22) cosedimented with L1 in the higher molecular-weight fractions of the gradient, confirming that the fusion protein binds to biotinylated particles.
L1: SA–TNF-α stoichiometry in conjugated VLPs. To examine the location of the fusion protein on conjugated particles, we measured the ability of two well-characterized mAb’s (33, 34) to bind conjugated particles by ELISA. Binding of an Ab that attaches to the sides of hexavalent capsomeres (mAb 5B6) was decreased by only twofold, relative to nonconjugated VLPs. However, conjugation led to a 10,000-fold decrease in the binding of mAB 9, which binds to the tips of hexavalent and pentavalent capsomeres. These results suggest that the SA–fusion protein bound primarily to capsomere tips and that the exterior of the conjugated particles was well coated with fusion protein. Using a spectrophotometric assay to measure SA binding (35), we estimated that there are approximately 1.5 SA tetramers bound per L1 molecule, which translates to approximately 540 tetramers per VLP (data not shown).
Humoral responses to conjugated VLPs. To examine the experimental induction of anti–TNF-α Ab’s, groups of two to six C57Bl/6 mice were vaccinated with the SA–TNF-α fusion protein either alone or conjugated to biotinylated BPV-1 VLPs. Mice were vaccinated three times at 2-week intervals. Mice that were immunized with conjugated particles received 5 μg doses of VLPs that were bound with a saturating amount of SA–TNF-α fusion protein (representing approximately 7.5–10 μg of protein). Mice injected with SA–TNF-α alone received doses of 15 μg of the fusion protein. Control mice were inoculated with VLPs conjugated to recombinant SA or with unmodified VLPs mixed either with full-length recombinant mouse TNF-α or a chemically synthesized TNF-α peptide representing aa 3–22. Injections were performed both with and without selected adjuvants. Table 1 summarizes the anti–TNF-α and anti-SA geometric mean titers obtained with sera taken 2 weeks after the final boost as determined by IgG-specific ELISA using recombinant full-length TNF-α or SA, respectively, as the target antigen.
IgG Ab titers against TNF-α and SA in vaccinated C57Bl/6 mice
VLPs linked to SA–TNF-α(3–22) (Table 1, line 1) or to SA–TNF-α(3–54) (line 2) consistently induced high-titer anti–TNF-α Ab’s. Conjugated SA–TNF-α(29–47) failed to generate anti–TNF-α Ab’s (line 3), suggesting that the TNF-α domain of this polypeptide may be malfolded. Our analysis focused on particles conjugated to SA–TNF-α(3–22) because this preparation induced the highest autoantibody titers and it was easier to generate large quantities of this protein. VLPs conjugated to SA–TNF-α(3–22>) generated high-titer IgG Ab responses even in the absence of exogenous adjuvant (Table 1, line 4; titers ranged from 640 to 104), although CFA provided a 50-fold boost to autoantibody levels (line 1; titers ranged from 104 to 2 × 105). However, other adjuvants, including Titermax Gold (line 5) and CpG oligonucleotides (line 6), had little or no boosting effect on autoantibody titers.
In contrast to the strong anti–TNF-α response induced when the TNF-α peptide was conjugated to the VLPs, no IgG autoantibodies were detected when mice were inoculated with unfused TNF-α peptide or native TNF-α mixed with unlinked VLPs, even when the potent adjuvant CFA was used (Table 1, lines 7 and 8). Inoculation of SA-TNF-α(3–22) alone without adjuvant also failed to induce autoantibodies, and even SA–TNF-α(3–22) with CFA only induced low IgG autoantibody titers to TNF-α (lines 9 and 10; without CFA anti-TNF-α Abs were never detected, and with CFA titers they ranged from <10 to 103). However, the immunizations that included linked foreign SA antigen induced high-titer IgG Abs to SA, which implies that the low or absent TNF-α Ab production seen with unconjugated SA–TNF-α(3–22) fusion protein was not the result of limiting Th activity. Therefore, when TNF-α was fused to a strong Th epitope(s), but not conjugated to the VLPs, the mouse immune system made a strong distinction between the foreign and self-polypeptides, with titers to SA that were more than 100-fold higher than those against TNF-α. In contrast, conjugation to VLPs resulted in a 1,000-fold increase in autoantibody titers both with and without coadministration of adjuvant. This dramatic increase in autoantibody titer was accompanied by only modest changes in SA Ab titers, resulting in similar Ab titers to TNF-α and SA (the ratio of SA to TNF-α titer ranged from 1.4 to 10, depending on the adjuvant used). Taken together, the findings indicate that conjugation to VLPs was essential for high-titer IgG autoantibody induction and imply that when a self-antigen is displayed at high occupancy on a VLP surface, the humoral immune system loses its ability to distinguish between self and foreign.
To gain insight into the stage of the immune response that is modified by conjugation to VLPs, we measured anti–TNF-α IgM titers and compared them with the IgG titers. Table 2 compares the serum IgG data from 2 weeks after the third vaccination with IgM ELISA titers generated using sera taken 2 weeks after the initial vaccination. In contrast to IgG autoantibody responses, IgM titers to TNF-α were similar regardless of antigen preparation and the IgM titers to self (SA) and foreign (TNF-α) epitopes were comparable for each formulation. Therefore, it does not appear that addition of linked VLPs initially increases the number of B cells that secrete Ig reacting with the self-peptide. Rather, VLP immunization may lead to amplification of the population of autoreactive mature B cells or may prevent these cells from being anergized.
Comparison of IgM and IgG titers in vaccinated C57Bl/6 mice
High-titer autoantibodies persisted in immunized mice for months after vaccination. Three mice injected with VLPs conjugated to SA–TNF-α(3–22) were followed for over 1 year after immunization. Throughout this period of observation, none of the mice exhibited any obvious gross abnormalities. Ab titers against both SA and TNF-α slowly declined to levels that were approximately 5-fold (SA) and 60-fold (TNF-α) lower than peak titers (Figure 2). Thus, exposure to endogenous TNF-α does not appear to restimulate autoreactive B cells or to acutely anergize them. Autoantibody titers could be boosted by subsequent application of conjugated particles. At 58 weeks, the mice were given an additional injection of vaccine, and Ab titers versus SA and TNF-α increased fourfold and tenfold, respectively.
Serum IgG Ab titers in immunized C57Bl/6 mice. Anti–TNF-α (open squares) and anti–SA (filled circles) geometric mean titer (GMT) in sera from three mice immunized with VLPs conjugated to SA–TNF-α(3–22) and followed for over 1 year. Mice were immunized at weeks 0, 2, 4, and 58 (indicated by arrows).
Protective effect of induced autoantibodies. To assess the protective potential of the induced TNF-α autoantibodies, we examined the effects of conjugated VLP vaccination on the development of type II collagen–induced arthritis (CIA) in mice. A total of 15 DBA/1 mice were immunized with conjugated VLPs in two separate experiments. In the first experiment, seven mice were injected three times at 2-week intervals with 5 μg of VLPs maximally conjugated to SA–TNF-α(3–22) in Titermax adjuvant. Five control mice were injected on the same schedule with the same dose of wild-type VLPs. At 1 week and 4 weeks after the final vaccination, mice were inoculated with 200 μg bovine type II collagen. In mice vaccinated with VLP:SA–TNF-α, the incidence of arthritis was apparently reduced to 71% (five of seven), compared with 100% in the control group (five of five). Protection was found to correlate with anti–TNF-α Ab titers (Table 3, experiment 1). The three mice with the highest titers (104) were either fully protected (two of three) or had a delayed onset of disease, whereas all four mice with low Ab titers (103) rapidly developed disease at a rate that was not significantly different from control mice (mean, 15 days after collagen immunization).
Immune responses and disease in immunized DBA/1 CIA mice
These data suggested that protection might be improved by increasing TNF-α Ab titers. To that end, a second experiment was initiated in which mice were given an additional boost of conjugated particles before collagen injection. Eight mice were immunized four times at 2-week intervals with VLPs conjugated to SA–TNF-α(3–22) in Titermax, along with ten VLP-injected control mice. This immunization schedule resulted in six of eight mice with anti–TNF-α titers of 3 × 103 or higher (Table 3). However, there was greater heterogeneity in the avidity of the Ab’s than in experiment 1. As above, type II bovine collagen was administered 1 and 4 weeks after the final vaccination. Vaccination resulted in a reduction of disease incidence and severity, relative to controls (Figure 3). In the vaccinated mice that did develop persistent arthritis, disease was delayed (Figure 3a). While none of the vaccinated mice developed the most severe disease, four of ten control mice displayed complete joint rigidity (clinical score, 3) in at least one limb. Five of the eight TNF-α–vaccinated mice were healthy at the end of the study, although one of the healthy mice experienced transient swelling in one limb. In comparison, 70% (seven of ten) of the control mice developed arthritis (P < 0.05). Histological examination of joints from protected vaccinated mice confirmed the clinical data; these mice displayed a marked reduction in synovitis and joint erosion relative to control mice (Figure 3c). Protection from arthritis induction was associated with both Ab titer and avidity (Table 3). All four mice with high avidity Ab’s (i.e., those with an avidity index score > 50%) remained healthy, whereas the mice with lower avidity Ab’s (< 50%) developed transient disease or became arthritic. Correspondingly, vaccination with conjugated particles gave substantial reduction in the proportion of cells staining positive for TNF-α expression in the joints of protected mice (Figure 3c).
Immunization with conjugated VLPs ameliorates the symptoms of collagen-induced arthritis. DBA/1 mice were immunized four times with either VLPs conjugated to SA–TNF-α(3–22) or wild-type VLPs plus Titermax and then injected intradermally with bovine type II collagen plus Freund’s adjuvant 1 and 3 weeks after the final vaccination. (a) Percentage of arthritic mice; groups immunized with wild-type VLPs (solid line) or VLPs conjugated to SA–TNF-α(3–22) (dashed line). (b) Clinical score: VLP-immunized mice (filled circles; VLP:SA–TNF-α(3–22)–immunized mice (open squares). Clinical score data is for arthritic mice only. Error bars reflect SEM. (c) Representative joints taken from naive or collagen-injected mice immunized with either wild-type VLPs or VLPs conjugated to SA–TNF-α(3–22). Joints were taken 4 weeks after the second collagen injection. The first column of panels shows hematoxylin and eosin (H&E) staining of synovial joints. ×10. B, bone; S, synovial membrane. The second column of panels shows representative immunohistochemical analysis of TNF-α expression in each set of mice. ×40. The percentage of TNF-α–positive cells in joints from protected vaccinated mice was similar to that in naive mice.
We also analyzed the results after combining the data from the two experiments. Within the vaccinated group, the geometric mean autoantibody titer in mice from both experiments that did not develop arthritis was 8,400, and the mean avidity index was 62%, compared with 1,300 and 44%, respectively, in mice that developed arthritis. The conjugated particle vaccine reduced the incidence of arthritis from 80% in the control group (12 of 15) to 53% in the vaccinated mice (8 of 15, P < 0.05). However, substantially better protection was observed in mice with anti–TNF-α Ab titers over 3,000; only two of nine mice (22%, P < 0.0001) in this group developed disease, and both of the sick mice had avidity index values less than 50%.