Structural Design of Chimeric Antigens for Multivalent Protein Vaccines

The development of prophylactic vaccines against pathogenic bacteria is a major objective of the World Health Organisation. However, vaccine development is often hindered by antigenic diversity and the difficulties encountered manufacturing immunogenic membrane proteins. Here, we employed structure-based design as a strategy to develop Chimeric Antigens (ChAs) for subunit vaccines. ChAs were generated against serogroup B Neisseria meningitidis (MenB), the predominant cause of meningococcal disease in the Western hemisphere. MenB ChAs exploit the lipoprotein factor H binding protein (fHbp) as a molecular scaffold to display the immunogenic VR2 epitope from the integral membrane protein PorA. Structural analyses demonstrate fHbp is correctly folded and that PorA VR2 epitope adopts an immunogenic conformation. In mice, ChAs elicit antibodies directed against fHbp and PorA, with antibody responses correlating to protection against meningococcal disease. ChAs offer a novel approach for generating multivalent subunit vaccines, containing of epitopes from integral membrane proteins, whose composition can be selected to circumvent pathogen diversity.

However, significant challenges remain in developing vaccines against pathogens for which toxoid and capsule-based vaccines are not viable. These pathogens include non-typeable strains of H. influenza and S. pneumonia (Keller et al., 2016, Murphy, 2015, unencapsuated pathogens such as Neisseria gonorrhoeae and Moraella catarrhalis (HPA, 2013, HPA, 2016, Jerse, Bash et al., 2014, Schaller, Troller et al., 2006 and encapsulated serogroup B N. meningitidis, for which a capsule based vaccine is not feasible (Finne, Leinonen et al., 1983). Given the exponential rise in the emergence of multi-drug resistant bacteria (WHO, 2014, Wi, Lahra et al., 2017, new approaches for vaccine development are paramount. However, strategies for generating successful vaccines are hampered by pathogen diversity (Telford, 2008) and the difficulties associated with presenting epitopes from membrane-embedded surface proteins to the immune system (Carpenter, Beis et al., 2008).
Two main approaches have been used to develop vaccines against serogroup B N.
We employed a structure-based approach to generate Chimeric Antigens (ChAs) against serogroup B N. meningitidis. ChAs exploit fHbp as a molecular scaffold to present the surface exposed PorA VR2 loop, which is achieved by inserting the VR2 loop into a β-turn region in fHbp. ChAs retain epitopes from both fHbp and PorA, and can elicit functional immune responses against both antigens. We demonstrate integration of a VR2 loop does not alter the overall architecture of fHbp and that the VR2 loop folds into a conformation recognized by a bactericidal mAb. We provide proof-in-principle that ChAs can be used to display selected epitopes from integral membrane proteins, such as PorA. ChAs incorporate epitopes from multiple antigens into a single vaccine antigen, which can be selected to circumvent pathogen antigenic diversity. Furthermore, ChAs contain epitopes from integral membrane proteins, which have previously hindered vaccine development, owing to the difficulties encountered during manufacture.

Design and construction of chimeric fHbp:PorAs
Immunisation with N. meningitidis proteins fHbp and PorA elicits bactericidal antibody responses, which provide a correlate of protection against meningococcal disease (Green et al., 2016, Serruto et al., 2012 (Figure 1A). fHbp is a lipoprotein that expresses as a soluble protein in Escherichia coli following removal of the N-terminal lipobox motif (Masignani et al., 2003).  (Table S1). The ChAs all express to high levels in E. coli and were purified by nickel affinity chromatography. Western blot analyses confirm all ChAs retain epitopes recognised by an α-P1.16 mAb and α-fHbp pAbs ( Figure 1C).

fHbp:PorAs are stable and can bind CFH
Stability of an antigen is an important property of a vaccine, and insertion of PorA epitopes might disrupt the overall structure of the ChA scaffold. Therefore, we determined the thermal stability of ChAs by differential scanning calorimetry (DSC, Table 1). Insertion of a PorA loop into the N-or C-terminal β-barrel of fHbp decreased the thermal stability of that β-barrel by 1.0°C to 15.5°C, with little or no effect on the other β-barrel. Overall, the lowest measured melting temperature (T m ) of any β-barrel was 60.5°C, which is considerably higher than the N-terminal T m of V3.45 (41°C), one of the fHbps in Trumenba® (Table 1).
A key property of fHbp is its ability to bind CFH (Madico et al., 2006, Schneider et al., 2006, Schneider et al., 2009 (Figure S1A). Therefore, surface plasmon resonance (SPR) was used to determine the affinity of each ChA for domains 6 and 7 of CFH (  Figure   S1B), potentially inhibiting CFH binding.

fHbp:PorA ChAs elicit protective immune responses
To examine the ability of ChAs to elicit immune responses, groups of CD1 mice were immunized with ChAs using alum or monophospholipid A (MPLA) as the adjuvant (Figure   2A  Significant binding (p≤0.05), is observed with fHbp V1.1 :PorA 267/P1.16 /MPLA antisera to the H44/76∆fHbp∆porA negative control strain, which is due to non-specific binding ( Figure   S2).
The serum bactericidal assay (SBA) assesses the ability of antibodies to initiate complement-mediated lysis of N. meningitidis. When using baby rabbit complement, an SBA titre of ≥8 is an accepted correlate of protective immunity against N. meningitidis (Andrews, Borrow et al., 2003). SBAs conducted with each set of pooled ChA/adjuvant antisera and wild-type N. meningitidis H44/76 all had titres of ≥128 ( Figure 3A).  (Figure 2C and 2E), these antisera did not have PorA-dependent SBA titres.
To activate the classical pathway, bound immunoglobulin (Ig) must recruit the C1q subunit of C1 (Frank, Joiner et al., 1987). The ability of Ig classes to bind C1q varies; a single IgM can be sufficient for C1q recruitment (Poon, Phillips et al., 1985), while several IgGs must be bound in close proximity and in a particular conformation (Burton, 1990, Hughes-Jones & Gardner, 1979, Sledge & Bing, 1973. Therefore, we examined which Ig isotypes are elicited by ChAs. Flow cytometry demonstrates that IgG1 is the main Ig in immune sera that binds the surface of N. meningitidis ( Figure 3D-3G). When compared with sera from mice immunised with PBS/adjuvant alone, all ChAs elicit significant α-fHbp IgG1 responses ( Figure 3D and 3F Figure 3G).

ChAs retain the architecture of the fHbp scaffold and PorA loop
To further characterise the fHbp:PorA ChAs, we determined the atomic structures of the  loops in the ChAs adopt a conformation that can induce bactericidal antibody responses (Oomen, Hoogerhout et al., 2003, Oomen, Hoogerhout et al., 2005.

ChAs containing an expanded range of PorA VR2 loops generate immune responses
To test the adaptability of our fHbp:PorA ChAs, we generated several ChAs composed from  Table 2).

Discussion
During infection pathogens present our immune system with an assortment of surface exposed lipid anchored and integral membrane proteins, both of which can be used as components in subunit vaccines. Whilst lipoproteins can be simply engineered for recombinant expression (by removal of their lipid anchor), integral membrane proteins present several challenges for vaccine development. Recombinant forms of integral membrane proteins are often poorly expressed and their native conformations may be compromised during purification, potentially reducing their ability to elicit immune responses against conformational epitopes, such as those found in surface loops (Bagal, Brown et al., 2013, Carpenter et al., 2008. Furthermore, immunisation with integral membrane proteins can generate irrelevant immune responses, which are directed towards epitopes masked by the outer membrane (Zhu, Thomas et al., 2005). To circumvent these issues, we used structure-based design to develop ChAs. We selected a key surface exposed epitope (VR2) from the integral membrane protein PorA and inserted it into the immunogenic scaffold of the lipoprotein fHbp. Multivalent ChAs generate immune responses against two key surface antigens that can elicit protective immunity (Claassen et al., 1996, Green et al., 2016, Kaaijk, van Straaten et al., 2013, Serruto et al., 2012, providing proof in principle that immunogenic epitopes from integral membrane proteins can be introduced into soluble molecular scaffolds to create ChAs.
Combining two antigens within a single recombinant ChA could diminish the immunogenicity of one or both antigens. We found that fHbp in all ChAs is highly immunogenic, inducing Linear PorA VR2 P1.16 peptides elicit antibodies that fail to recognise the native protein and are non-bactericidal, while cyclic PorA VR2 peptides, with identical residues but fixed into a β-turn, can elicit antibodies that recognise native PorA and are bactericidal (Christodoulides, McGuinness et al., 1993, Christodoulides, Rattue et al., 1999, Hoogerhout, Donders et al., 1995, Oomen et al., 2003, Oomen et al., 2005. Structural data shows that when the structure of cyclic VR2 peptide mirrors that of a linear peptide bound by a bactericidal mAb, and thus locked into an immunogenic conformation, the cyclic peptide induces bactericidal responses (Oomen et al., 2003, Oomen et al., 2005. In ChAs, the N-and C-termini of the VR2 loop are bound by neighbouring fHbp β-strands, fixing the VR2 epitope into a β-turn, as observed in cyclic peptides. This was confirmed by the atomic structures of ChAs, in which the PorA adopts the same conformation as when bound by a bactericidal Fab fragment (van den Elsen et al., 1997). In previous work, cyclic peptides were coupled to carrier proteins and used with adjuvants not licensed for human use (Christodoulides et al., 1993, Christodoulides et al., 1999, Oomen et al., 2003, Oomen et al., 2005. In the resulting SBAs, titres were only observed with antisera from some mice (Oomen et al., 2005), similar to our findings with ChAs and licensed adjuvants.  (Green et al., 2016, Serruto et al., 2012.

Bacterial strains and growth
The bacterial strains used in this study are shown in Table S2 and Table S3. N.

Western blot analyses
Western blots of purified proteins ( All sera were stored at -80°C until required and once defrosted sera were stored at 4°C.
Baby rabbit complement (Cedar lane, lot #15027680) was diluted with DPBS-G to a final dilution of 1 in 10. Serum, pooled or from individual mice, was heat inactivated for one hour at 56°C and added to the wells in a serial two-fold dilution, starting with a dilution of 1 in 5 or higher. Control wells contained no serum or no complement. Following static incubation for one hour at 37°C in the presence of 5% CO 2 , 10 µl from each well was plated onto BHI plates in triplicate and colonies from surviving bacteria counted. The bactericidal activity is expressed as the dilution of serum required to kill ≥ 50% of bacteria in assays containing both complement and serum in comparison with control assays containing serum or complement alone. SBAs using pooled sera were repeated three times, and assays using sera from individual mice were repeated twice. SBA titres were input into GraphPad Prism and statistical analyses comparing titres obtained from alum immunisations with titres obtained from MPLA immunisations were performed using two-way ANOVA (statistical significance of p ≤ 0.05) and Dunnett's method of multiple comparisons.

Surface Plasmon Resonance
SPR was performed using a Biacore 3000 (GE Healthcare       and ChA/adjuvant antisera. SBAs with H44/76 and H44/76∆porA were conducted using pooled ChA/adjuvant antisera, and SBAs with H44/76∆fHbp were conducted with ChA/adjuvant antisera from individual mice. Geometric mean and SD of independent assays (n>2) are indicated. Flow cytometry was used to detect binding of mouse isotypes IgG1, IgG2a, IgG2b, IgG3 and IgM in ChA antisera to N. meningitidis strains H44/76∆porA (D and F) and H44/76∆fHbp (E and G). Mouse isotypes binding to N. meningitidis were detected with isotype specific secondary antibodies. SD of independent assays (n=3) is indicated. Two-way ANOVA and Dunnett's method of multiple comparison were used to compare SBA titres from pooled antisera (A and B) and ChA antisera to PBS control sera (D-G) (* p≤0.05, ** p≤0.01, *** p≤0.001, **** p≤0.0001).       Comparison of the geometric means from these histograms, and the histograms from two other independent repeats, is shown in Figures 2C and 2D