Close this search box.

Inter-epitope spacer variation within polytopic L2-based human papillomavirus antigens affects immunogenicity – npj Vaccines

Spacer variant incorporation into the HPV-L2 polytopes does not affect the biochemical properties of the antigens

A GGP tripeptide interposed between individual epitopes as well as at both ends of the display site of Trx (i.e., between the N- and C-terminal portions of the Trx scaffold and the first and the last epitope, respectively; see Fig. 1a) was previously used as a general spacer for the construction of our cutaneous (Trx-L2c12mer-OVX31312) and mucosal (Trx-L2m8mer-OVX31311,15) HPV-L2 vaccine prototypes. To explore the effect of localized spacer variations on the presentation and immunogenicity of specific epitopes, glycine and proline were assembled in different combinations to generate five other distinct spacer antigen variants. The five spacers and ‘GGP’, designated as V1-V6 (see Fig. 1a), were inserted between the HPV3 and the HPV4 L2 epitopes of Trx-L2c12mer-OVX313 (i.e., upstream to the epitope of the sub-optimally neutralized HPV4 type) and between the HPV18 and HPV31 L2 epitopes of Trx-L2m8mer-OVX313 (i.e., upstream to the epitope of the sub-optimally neutralized HPV31 type). In this way, six spacer variants for each antigen, designated as C12merV1 to C12merV6 and M8merV1 to M8merV6, were generated.

Fig. 1: Recombinant expression, purification and stability/oligomerization properties of antigen variants harboring six different peptide spacers inserted into the L2 polytopes of cutaneous and mucosal HPV vaccine candidates.
figure 1

a Representation of the polytope and scaffold proteins compositions of Trx-L2c12mer-OVX313 (containing 12 different cutaneous HPV epitopes) and Trx-L2m8mer-OVX313 (containing 8 mucosal HPV epitopes). Six different spacers, composed of glycine (G) and proline (P) residues (indicated and color-coded at the top), were inserted between HPV3 and HPV4 L2 epitopes to generate antigen variants C12merV1 to V6, and between HPV18 and HPV31 L2 epitopes to generate antigen variants M8merV1 to V6, as indicated. The Trx and OVX313 scaffolds are indicated as green rectangles (shown on both sides of the polytopes) and as blue ovals, respectively. b All the proteins were efficiently purified and migrated as essentially uniform species (indicated by blue arrows) in SDS-PAGE under reducing conditions. Under non-reducing SDS-PAGE conditions, the heptameric, OVX313 disulfide-bonded forms of the antigens were preserved and migrated as larger size bands (marked by pink arrows).

Following expression in Escherichia coli, a comprehensive characterization on antigen variants was conducted, involving solubility and thermal stability tests. All protein variants exhibited high solubility in lysis buffer and shared the same melting temperature, approximately 80 °C for C12mer variants and 75 °C for M8mer variants. The application of thermal treatment to the cleared lysate, followed by cation exchange chromatography, allowed similar purification protocols for all antigen variants. No differences in production yield, and purity level (above 90%) was observed among individual variants of the C12mer and the M8mer antigens, which displayed essentially identical subunit molecular weights upon SDS-PAGE analysis (46 kDa and 36 kDa for the C12mer and M8mer antigens, respectively).

Importantly, when analyzed under non-reducing SDS-PAGE conditions, all antigens migrated as high molecular weight species (marked with pink arrows in Fig. 1b), consistent with the formation of heptameric structures driven by the OVX313 multimerization domain present in all constructs.

Spacer variation affects the neutralization immunogenicity of the corresponding antigens in a HPV type-related manner

The purified antigens were formulated with the AddaVaxTM adjuvant, and four doses were injected intramuscularly into mice at two-week intervals (see Fig. 2a). Following blood collection one month after the last dose, the resulting sera were analyzed by pseudovirion-based neutralization assays (PBNAs). These were applied to a subset of the HPV types represented in the polytopes (seven out of 12 HPV types and four out of eight HPV types for the C12mer and the M8mer polytopes, respectively, including the sub-optimally neutralized HPV4 and HPV31 types; Figs. 2, 3). Our selection of the specific HPV types to be investigated, included those considered hard-to-neutralize (i.e., HPV4 for Trx-L2c12mer-OVX313 and HPV31 for Trx-L2m8mer-OVX313), as well as HPV types located upstream and downstream of the targeted epitope in the polytope string. All antigens induced detectable neutralizing antibody responses, albeit of varying strength, against the examined HPV types (Figs. 2, 3). None of the antigen variants, however, led to a generalized and consistently superior neutralizing antibody response against all the tested HPV types. Still, some statistically significant differences in the strength of the neutralizing responses elicited by some variant antigens against specific HPV types were observed. Most notable was the improvement of HPV1 and HPV2 neutralization associated with the variant antigen C12merV1 compared to C12merV5 and C12merV4 (p-values of 0.0482 and 0.0263, respectively), and the superior neutralization capacity against HPV2 and HPV3 displayed by the C12merV6 antigen compared to C12merV4 and C12merV3 (p-values of 0.0233 and 0.0493) (Fig. 2b–d). The C12merV1 variant also induced an anti-HPV76 neutralization response significantly stronger than that of C12merV4 (p-value of 0.0379) (Fig. 2g). No appreciable difference in neutralization capacity was observed with either variant against the other tested HPV types, including HPV4. Interestingly, the spacers contained in C12merV1 and C12merV6, the two antigens that elicited superior neutralization responses against the above reported HPV types, are four (GPGP) and three (GGP) amino acids in length, respectively, whereas shorter (two amino acids in the case of C12merV5) or longer (six amino acids for C12merV3 and C12merV4) spacers are associated to the more poorly performing antigen variants.

Fig. 2: Antigen spacer variants differently affect the ability of Trx-L2c12merOVX313 to induce neutralizing antibodies against cutaneous HPV types in mice.
figure 2

a Groups of eight mice were immunized with different antigen spacer variants as indicated. bh Neutralizing antibody levels against HPV1, HPV2, HPV3, HPV14, HPV76 and HPV95 were measured by PBNAs. Note that for HPV4 the higher sensitive Fc-PBNA was employed. Each data point represents the neutralizing antibody titer (EC50) measured in one mouse serum. Neutralizing antibody titers lower than 50 or EC50 with R square less than 0.85 were considered as non-neutralizing and set at 0.1. Horizontal bars represent the geometric mean titers (GMT) of each group. P-values ≤ 0.05, as determined by the nonparametric Mann–Whitney test, were considered significant and marked as ‘*’. b C12merV5 proved to be significantly less immunogenic than C12merV1 (p-value = 0.0482). c GMT of sera EC50 titers in the C12merV4 group was significantly lower than that measured in the C12merV1 and C12merV6 groups (p-values of 0.0263 and 0.0233, respectively). d Among the various weakly responding groups, anti-HPV3 neutralizing antibody levels in the C12merV3 was lower than those measured in the C12merV6 group significantly (p-value = 0.0493). g The ability of C12merV4 to induce neutralizing antibodies against HPV76 was significantly lower than that of C12merV1 (p-value = 0.0379). e, f, h No significant difference in anti-HPV4, anti-HPV14 and anti-HPV95 neutralizing antibody induction capacity was observed among the various antigen groups.

Fig. 3: Effect of different spacer variants on the ability of Trx-L2m8mer-OVX313 to induce neutralizing antibodies against mucosal HPV types.
figure 3

ad Neutralizing antibody levels against HPV16, HPV18, HPV31 and HPV33 induced by the indicated M8mer variants; data points represent neutralizing antibody titers (EC50) measured in individual mouse sera. Titers lower than 50 or EC50 with R square less than 0.85 were considered as non-neutralizing and set at 0.1. Horizontal bars represent the geometric mean titers (GMT) of each group. P-values ≤ 0.05, as determined by the nonparametric Mann–Whitney test, were considered significant and are marked as ‘*’. a Strong but not significant reduction of anti-HPV16 neutralization for the M8merV5 variant was observed. b Comparable anti-HPV18 neutralizing responses were induced in all antigen spacer variant groups. c M8merV2 significantly outperformed M8merV1 and M8merV4 in the induction of anti-HPV31 neutralizing antibody titers (p-value of 0.0379 and 0.0368). d M8merV3 induced significantly higher neutralization responses against HPV33 compared to the M8merV1 variant (p-value = 0.0499).

A different situation was observed in the case of the Trx-L2m8mer-OVX313 mucosal vaccine prototype, where a significantly superior HPV31 neutralization performance was found to be associated with the M8merV2 variant, which outperformed both the M8merV1 (p-value of 0.0379) and the M8merV4 (p-value of 0.0368) spacer variants. In the same context, M8merV3 elicited a superior neutralization response compared to M8merV1 against HPV33 (p-value of 0.0499), whose epitope is adjacent to that of HPV31 in the M8mer polytope. Interestingly, both best performing M8mer spacers (V2 and V3) differ from those we identified as best spacers in the case of the C12mer antigens (V1 and V6), as if the epitope spacing effect were somehow influenced by polytope length and/or composition. In addition, M8merV2 contains a highly flexible tetraglycine spacer, whereas a glycine-rich, six-residue spacer is present in M8merV3.

Total and neutralizing anti-L2 antibody levels correlate among antigen spacer variants

We then used peptide ELISA to determine the levels of total anti-L2 antibodies directed against the L2 epitopes of different HPV types. ‘Total anti-L2 antibodies’ refers to all the antibodies capable of binding to the targeted L2 epitope, irrespective of their neutralizing properties. The aim of this analysis was to explore the existence of a correlation between total and neutralizing anti-L2 antibody levels. In particular, we wished to determine whether a reduced neutralization capacity (or lack thereof) may correlate with a deficiency (or lack) of total anti-L2 antibodies against a certain epitope. Provided the effect of linker insertion on immunogenicity is more qualitative than quantitative, i.e. the antibodies induced are functional, then this should lead to a better correlation between peptide ELISA and PBNA. As shown in Fig. 4, ELISA data revealed measurable total anti-L2 antibody titers to each tested HPV-type L2 peptide for most sera, including sera with undetectable neutralizing antibody titers, such as those derived from mice immunized with the poorly performing C12merV3, C12merV4 and C12merV5 antigen variants (see Fig. 2).

Fig. 4: Neutralizing vs. total anti-L2 antibody titers induced by the different antigen spacer variants against distinct HPV types.
figure 4

ak Neutralizing antibody titers (EC50, y-axis) determined by PBNA were correlated with total anti-L2 antibody titers (EC50, x-axis) measured by L2-peptide ELISA in individual immune-sera for cutaneous HPV types (HPV1, HPV2, HPV3, HPV4, HPV14, HPV76 and HPV95) and mucosal HPV types (HPV16, HPV18, HPV31 and HPV33). The shape and color of each data point coding for the different antigen groups are shown in the top right inset. EC50 values from PBNAs lower than 50 or EC50 with R square less than 0.85 were considered as non-neutralizing and set at 0.1. Similarly, EC50 values from ELISAs lower than 10 or EC50 values with R square less than 0.85 were considered as negative data-points and were excluded from analysis. Spearman’s rank correlation coefficients (r) for all mice sera (regardless of the antigen groups) and the corresponding p-values are provided inside each panel; the correlation was considered as statistically significant when the p-value was ≤0.05.

Spearman’s correlation coefficients for total vs. neutralizing anti-L2 antibody titers were then calculated for a subset of cutaneous HPV types (seven out of twelve) and mucosal HPV types (four out of eight, see Fig. 4). For C12mer antigen variants, moderately strong (r > 0.5) positive correlations were found for HPV1, HPV4 and HPV76 (Fig. 4a, d, f, respectively), whereas weaker correlations were obtained for HPV2, HPV3, HPV14 and HPV95 (Fig. 4b, c, e, g, respectively). As to M8mer antigen variants, correlations for HPV18 and HPV33 were significantly strong (Fig. 4i, k, respectively), while no correlations were observed for HPV16 and HPV31 (Fig. 4h, j, respectively). Thus, for some but not all HPV types, total anti-L2 antibody levels strongly correlate with the neutralizing antibody titers induced by the various spacer variant antigens.

A more detailed representation (HPV vs. antigen variant type) of the correlation between L2-peptide ELISA titers and the EC50 values derived from the PBNAs for the corresponding HPV types is reported in Supplementary Table 1 and Supplementary Table 2. No significant correlation between total and neutralizing anti-L2 antibodies could be detected for any of the antigen variant groups for HPV3 and HPV14 for C12mer antigen variants, and HPV31 for M8mer antigen variants. In the case of HPV3, in particular, a large number of ELISA-positive immune-sera failed to display an appreciable neutralization capacity, and a similar lack of correlation was observed, conversely, for PBNA-positive sera. In respect to HPV14 and HPV31, most sera were neutralizing, yet ELISA and PBNA titers did not correlate. Looking at different antigen variants specifically, especially the V4-inserted variants showing comparatively poor performance in regard to immunogenicity (see Figs. 2, 3), we observed a significant correlation between total and neutralizing anti-L2 antibody titers for HPV1, HPV2 and HPV95 in C12merV4, and for HPV33 in M8merC4. For C12mer antigens, a fairly strong correlation was also observed for the antibody titers elicited by C12merV1 against HPV4, by C12merV2 against HPV1 and HPV95, and by C12merV5 against HPV4 as well as HPV76. In contrast, no significant correlation was observed with C12merV3 and the best-performing antigen C12merV6 for all the tested HPV types. In the context of M8mer antigens, a strong correlation was shown for the antibody titers induced by M8merV1 against HPV18, by M8merV2 against HPV18 and HPV33, and by M8merV6 against HPV16. Nevertheless, titers induced by the M8merV3 and M8merV5 antigens showed no significant correlation in respect to the two assays for all tested mucosal HPV types.

Affinity of HPV type-specific monoclonal antibodies to different antigen variants is influenced by the inserted spacers and strongly correlates with immunogenicity

To gain further insight into the influence of inter-epitope spacers on HPV L2 polytope presentation, we set out to use Surface Plasmon Resonance (SPR) for a quantitative analysis of the interaction between all C12mer variants and a subset of neutralizing and non-neutralizing monoclonal antibodies (mAbs) targeting the L2 aa 20-38 epitopes of cutaneous HPV types 1, 2, 3, and 428 (Supplementary Table 3). Since the heptameric, OVX313-containing antigens were stably bound to, and hardly dissociable from the mAbs (data not shown), for this analysis we employed the monomeric forms of the C12mer antigens. Non-reducing SDS-PAGE data showed that these OVX313-lacking, Trx-L2c12mer antigen derivatives were indeed fully monomeric and unable to oligomerize (Supplementary Fig. 1).

The affinity of four neutralizing (1MK2L2, 2TK14L2, 3MK1L2 and 4SA1L2, Fig. 5a) and three non-neutralizing (3SA1Al2, 3SA1BL2 and 3SA2L2, Fig. 5b) mAbs to the monomeric C12mer antigens was then determined based on the equilibrium dissociation constants (KD) provided by SPR measurements. The KD values of non-neutralizing mAbs to all antigens (ranging from 5.3E-09 to 4.3E-08) were consistently higher than those of neutralizing mAbs (KD values ranging from 1.7E-11 to 2.4E-08), indicating an overall lower affinity of non-neutralizing mAbs (Fig. 5).

Fig. 5: Variations in mAb affinity for antigen spacer variants arise from spacer modification.
figure 5

a Affinity of neutralizing mAbs 1MK2L2, 2TK14L2, 3MK1L2 and 4SA1L2 to six monomeric antigen variants (marked with different colors, as indicated) determined by SPR; KD values are shown on the y-axis for different mAbs (x-axis) and antigen variants as indicated. b Same as (a) for the non-neutralizing 3SA1AL2, 3SA1BL2 and 3SA2L2 mAbs (see Supplementary Table 3 for the mAbs properties and target HPV specificities).

The monomeric form of the C12merV6 antigen, one of the best performing antigens revealed by cutaneous HPV neutralization assays (Fig. 2) displayed the highest affinity to all mAbs, with KD values one-two orders of magnitude lower compared to the other variants. In contrast, C12merV4, one of the antigen variants with the weakest ability to induce neutralizing antibodies (Fig. 2), consistently exhibited the highest KD values (i.e., lowest affinity) to all neutralizing mAbs. These results clearly indicate that spacer modification can influence mAbs affinity to the HPV1, HPV2, HPV3 and HPV4 L2 epitopes. In particular, as shown in Fig. 6, we found a consistent correlation between mAbs affinity to different spacer variant antigens and the geometric mean of the PBNA EC50 titers induced by such antigens. In fact, as further documented in Fig. 7, a linear relationship between SPR-derived KD values and neutralizing geometric mean titers (GMT) was seen for HPV2-2TK14L2 (panel b; p = 0.024), HPV3-3MK1L2 (panel c; p = 0.0278) and HPV4-4SA1L2 (panel d; p = 0.0148).

Fig. 6: Correlation between the affinity of HPV type-specific neutralizing mAbs for different spacer variant antigens and the ability of such antigens to induce neutralizing antibody responses.
figure 6

In each graph, referred to different HPV types as indicated, the geometric mean of PBNA EC50 titers (left y-axis) are shown in blue, while the corresponding KD values (right y-axis) determined by SPR for each monomeric antigen spacer variant (x-axis) are shown in red.

Fig. 7: Linear correlation between KD (SPR) and GMT (PBNA) values.
figure 7

ad The dissociation equilibrium constants (KD, y-axis) from SPR tests were correlated with the GMTs of neutralizing antibody titers from the corresponding HPV-type PBNAs (x-axis). Each dot represents one antigen group as indicated. KD (SPR) values for each monomeric antigen variant and mAb (indicated in each panel) are shown on the y-axis; GMT of EC50 (PBNA) values for the neutralization assays of the indicated HPV types are shown on the x-axis, respectively. P-values are reported within each panel under simple linear regression.

Altogether, these data strongly suggests that KD (i.e., mAb affinity) values measured by SPR may represent reliable predictors of neutralizing immunogenicity -much faster and more convenient to determine compared to the measurement of neutralizing antibody titers by PBNAs.