Epitope-specific immunity against Staphylococcus aureus coproporphyrinogen III oxidase

Staphylococcus aureus represents a serious infectious threat to global public health and a vaccine against S. aureus represents an unmet medical need. We here characterise two S. aureus vaccine candidates, coproporphyrinogen III oxidase (CgoX) and triose phosphate isomerase (TPI), which fulfil essential housekeeping functions in heme synthesis and glycolysis, respectively. Immunisation with rCgoX and rTPI elicited protective immunity against S. aureusbacteremia. Two monoclonal antibodies (mAb), CgoX-D3 and TPI-H8, raised against CgoX and TPI, efficiently provided protection against S. aureus infection. MAb-CgoX-D3 recognised a linear epitope spanning 12 amino acids (aa), whereas TPI-H8 recognised a larger discontinuous epitope. The CgoX-D3 epitope conjugated to BSA elicited a strong, protective immune response against S. aureus infection. The CgoX-D3 epitope is highly conserved in clinical S. aureus isolates, indicating its potential wide usability against S. aureus infection. These data suggest that immunofocusing through epitope-based immunisation constitutes a strategy for the development of a S. aureus vaccine with greater efficacy and better safety profile.

Staphylococcus aureus (S. aureus) is associated with a significant disease burden causing life-threatening diseases, such as deep wound infections, bacteremia, endocarditis, pneumonia, osteomyelitis, and enterotoxin-mediated shock1. Antibiotic resistance, specifically methicillin-resistant Staphylococcus aureus (MRSA), is widespread and of aggravating concern. Although vaccination strategies against S. aureus have attracted much attention in basic and clinical research, no S. aureus vaccine is currently available2,3,4. Specific challenges to development of a S. aureus vaccine include low immunogenicity of pathogen-derived antigens, a lack of natural immunity to S. aureus, multiple virulence and immune evasion factors as well as redundant nutrition acquisition pathways. All of these challenges compromise a straightforward strategy to delineate a correlate of protection.

In general, neutralising antibodies inhibiting pathogen interaction with or entry into host cells or detoxifying virulence factors represent a dominant principle of protection provided by vaccines. For instance, due to nasal colonization most adult humans have high levels of circulating antibodies against many staphylococcal antigens which seem to provide some protection against invasive infection with S. aureus5,6. Classical vaccine approaches, targeting S. aureus toxins for neutralisation or surface antigens for production of opsonising antibodies, have not worked against S. aureus in clinical trials. Similarly, the targeting of S. aureus proteins serving important roles in host-pathogen interactions, including adhesion to host cells, binding to and degradation of extracellular matrix proteins, iron-uptake or intervention with the host fibrinolytic system remained unsuccessful2,3,4. Preclinical and clinical data repetitively indicate that although immunisation with S. aureus antigens usually results in high antibody titers, this does not confer protection against S. aureus infections7.

Induction of a high-titered antibody response by a vaccine is not tantamount to protection and may even be detrimental by causing immune enhancement of disease, which is well known for vaccines against viral pathogens. For example, Song and coworkers recently identified a linear B-cell epitope on the prM protein of dengue virus as a major immunodominant B-cell epitope involved in antibody-dependent enhancement of dengue virus infection8. Although vaccine-mediated immune enhancement has not been an obvious safety concern for S. aureus vaccine development, the knowledge of protective, non-protective and disease enhancing B-cell epitopes represents a strategy for refined vaccine development. In this respect, the use of monoclonal antibodies (mAbs) to design new vaccines has been previously proposed by Burton9. Monoclonal Abs are now an integral part of the ‘reverse vaccinology 2.0’ concept10,11, where mAbs are used to distinguish protective from non-protective epitopes and to support immunofocused antigen design. An epitope-focused vaccine is anticipated to improve its immunogenic precision level, resulting in a vaccine with a greater efficacy and safety profile. Indeed, an epitope-focused strategy has been successfully employed for the development of a vaccine against RSV that has resisted traditional vaccine development in the past12.

We here targeted two non-redundant S. aureus housekeeping proteins, coproporphyrinogen III oxidase (CgoX) and triose phosphate isomerase (TPI), which are essential for heme synthesis and glycolysis, respectively. Staphylococcal CgoX (EC: 1.3.3.15, also known as HemY) catalyses the oxidation of coproporphyrinogen III to coproporphyrin III13,14 but can also oxidise protoporphyrinogen IX to protoporphyrin IX15. Indeed, identical aa sequences are deposited for S. aureus CgoX and protoporphyrinogen oxidase in public data banks like UniprotKB (Uniprot.org, compare Q2FXA5 and A0A0H3K8Y5). The CgoX-mediated generation of protoporphyrin IX, but not coproporphyrin III, is stimulated by heme-bound HemQ, which is mediated by superoxide14. TPI catalyses the reversible interconversion of the triose phosphate isomers dihydroxyacetone and D-glyceraldehyde 3-phosphate. It plays an important role in glycolysis and is essential for efficient energy production. CgoX and TPI were previously identified within a group of anchorless cell wall S. aureus proteins16. The surface location of CgoX and TPI suggests additional functions beyond their role in cellular homeostasis, corresponding to the group of covalently cell wall anchored (CWA) proteins, several of which being multifunctional and involved in S. aureus pathogenesis17. Indeed, TPI has been suggested to have plasminogen binding activity, which might be relevant to staphylococcal invasion18,19. In contrast, whereas the intracellular role of CgoX in heme synthesis is well known, an extracellular function of CgoX has not yet been described. Clearly, the intracellular non-redundant and essential action of CgoX and TPI in S. aureus homeostasis is not accessible for antibodies. Thus, vaccinal targeting of their putative extracellular action is not expected to push directly the development of escape mutants. We here show that immunisation with recombinant (r) CgoX or rTPI protects mice from S. aureus bacteremia. Correspondingly, mAbs raised against CgoX and TPI significantly improved survival in a murine sepsis model. Furthermore, a short, 12 aa linear epitope specifically recognised by a protective CgoX-D3 mAb is demonstrated to provide highly efficient protection against S. aureus infection when used for active immunisation.

Selection of CgoX and TPI as vaccine candidates

We have previously reported on 37 anchorless cell wall associated S. aureus proteins, recognised by naturally occurring antibodies in healthy humans that have potential to serve as new candidates for a protein-based S. aureus vaccine. Indeed, some of these targets induced protective immunity against some laboratory S. aureus strains when tested in a murine sepsis model16,20. Extended testing of this group revealed two further vaccine candidates, coproporphyrinogen III oxidase (CgoX, formerly known as protoporhyrinogen oxidase) and Triose phosphate isomerase (TPI) (Fig. 1). For immunisation studies, recombinant staphylococcal CgoX, and TPI were expressed as His6-tagged proteins in E. coli and purified by affinity chromatography. The purity and integrity of His6-tagged CgoX and TPI were controlled by SDS-PAGE (Fig. 1a, d). Mice were immunised i.p. with 80 µg of recombinant protein and Freund´s adjuvant and boosted with 40 µg antigen in incomplete Freund’s adjuvant s.c. at days 33 and 56. Immunisation with rCgoX and rTPI induced high titers of IgG antibodies recognising the respective recombinant S. aureus protein (Fig. 1b, e). Western blot analysis revealed that the antibody response against CgoX was partially directed against the His6-tag (Fig. 1b). Eight days after the second boost, mice were challenged i.v. with with the methicillin-sensitive Staphylococcus aureus (MSSA) strain ATCC29213 (day 64). Immunisation with rCgoX and rTPI induced significant protection against S. aureus infection in a murine sepsis model (Fig. 1c, f). CgoX and TPI have orthologs in mouse and man. Staphylococcal CgoX shared 23% identity with its human and mouse ortholog, PPOX. Staphylococcal TPI showed 22% identities with its ortholog in mouse and man (Supplementary Fig. 1a, b). Thus, when used as whole antigen for immunisation, these proteins bear a remote risk of eliciting antibodies cross-reacting with human orthologs.

Generation of protective antibodies

To dissect protective from non-protective epitopes of S. aureus antigens, mAbs were raised against CgoX and TPI by standard hybridoma technology20. For each antigen, protective and non- or less-protective mAbs were identified (Fig. 2a). The protective efficacy of mAb CgoX-D3 and mAb TPI-H8 is demonstrated in a murine sepsis model with either the MSSA strain ATCC29213 or MRSA USA300 (Fig. 2b, c). Of note, all mAbs showed protection at doses between 200 µg and 300 µg per mouse (Supplementary Fig. 2c) and worked equally well for MSSA and MRSA strains (Fig. 2b, c and Supplementary Fig. 2a, b). MAb CgoX-D3 and mAb TPI-H8 specifically recognised the respective recombinant protein (Fig. 2d). The heavy chain subtypes isolated for the two S. aureus antigens were predominantly IgG1. The protective mAbs CgoX-D3 and TPI-H8 did not cross-react with their respective human orthologs (Fig. 2e). CgoX and TPI are essential intracellular housekeeping enzymes involved in heme synthesis and glycolysis, respectively. Indeed, the genetic ablation of cgoX attenuated S. aureus proliferation (Fig. 2f), whereas tpiA deletion S. aureus mutants could not be generated suggesting that TPI is essential for S. aureus growth. Given that antibodies do not pass the S. aureus plasma membrane, the protective effects of monoclonal CgoX-D3 and TPI-H8 antibodies should be unrelated to intracellular functions of their respective target antigen. Indeed, none of the mAbs inhibited S. aureus proliferation in vitro (Fig. 2g).

It is noteworthy that humanisation of the murine mAbs CgoX-D3 and TPI-H8 preserved their antigen specificity and function, i.e. recognition of their target antigen and in vivo protectiveness against S. aureus USA300 (Supplementary Fig. 2a–d), suggesting that their mode of action is not dependent on interactions with the murine Fc fragment or distinct IgG subclasses. Like the murine mAbs, humanised mAbs (huMAbs) CgoX-D3 and TPI-H8 did not inhibit in vitro proliferation of S. aureus (Supplementary Fig. 2e, f). Moreover, CgoX-D3 and TPI-H8 provide only marginal opsonizing activity of S. aureus by human neutrophils or murine macrophages, respectively (data not shown).

Epitope mapping of protective and non-protective monoclonal antibodies

For epitope mapping, staggered overlapping peptide fragments spanning the entire respective antigen sequence were spotted on nitrocellulose or glass slides and analysed for binding of the corresponding antibody. A linear epitope could be defined for CgoX-D3, spanning aa 377–388 (Fig. 3a, and Supplementary Fig. 3a). Alanine scanning of the linear epitope revealed Ile384and Arg387 as essential binding determinants for anti-CgoX-D3. The marginal inhibition by the alanine replacement of Leu381, Val385, Arg386 might be functionally irrelevant (Fig. 3b). Indeed, in silico analysis revealed surface exposure of Ile384 and Arg387 (Supplementary Fig. 3b) which is a prerequisite for antibody binding. A blast search revealed strong conservation of the CgoX epitope within 35,361 clinical S. aureus isolates, which predicts that mAb D3 should bind to CgoX in approximately 99% of clinical S. aureus isolates (Fig. 3c). Competitive binding experiments confirmed independent binding of protective and non- or less-protective mAbs to CgoX (Fig. 3e), suggesting distinct epitopes for protective and non-protective mAbs. The mAb CgoX-D3 showed high affinity binding to CgoX with a Kd of 60.38 pM (Fig. 3f).

In contrast to CgoX, the protective TPI-specific mAbs H8 and F3 did not recognise a short epitope comprising less than 15 aa (Supplementary Fig. 4a). Instead, TPI fragment cloning combined with binding specificity analysis revealed that mAb H8 and -F3 bind to a polypeptide of 105 amino acid residues with K114 and N219 representing the N- and C-terminal boundary, respectively (Fig. 4a). Figure 4b shows the binding domain of TPI-H8 in a secondary structure model of TPI. The molecular surface representation of the TPI-H8 defined binding domain revealed many surface exposed aa residues potentially accessible for antibody interactions (Supplementary Fig. 4b). The amino acid residues essential for TPI-H8 binding have not yet been identified. Unlike TPI-H8, two non-protective monoclonal anti-TPI antibodies, C4 and C8 recognised distinct linear epitopes (Fig. 4c). The TPI-C4 epitope spanned aa 93–111, directly adjacent to the N-terminal boundary of the H8 mAb binding site (Fig. 4c). The mAb TPI C8 bound to two cognate epitopes at aa position 21–26 and 244–249, which were distinct from the H8 binding domain (compare Fig. 4b, c). MAb-H8 and -F3 competed with each other for binding to TPI (Fig. 4d) suggesting identical epitopes. No competition was observed between protective and non-protective anti-TPI mAbs (Fig. 4d). The mAb TPI-H8 showed high affinity binding to TPI with a Kd of 15.84 pM (Fig. 4e).

Active immunisation with linear S. aureus CgoX-D3 epitope

Functional monoclonal antibodies have become a valuable tool for immunofocusing in vaccine design, that is, in maximisation of on-target antibody responses to desired epitopes and minimisation of off-target responses9,10,11. We reasoned that the short linear peptide of the 12 aa CgoX-D3 epitope may be suitable for the induction of a more restricted and thus, functionally more precisely targeted antibody response compared to a full-length antigen. We, therefore, tested the protective mAb CgoX-D3 epitope for its potential to elicit a protective immune response when used as an active vaccine. A synthetic peptide representing the protective CgoX-D3 epitope was linked to bovine serum albumin (BSA) as protein carrier. The CgoX-D3-BSA construct specifically competed with CgoX-D3 mAb for binding to the full-length CgoX (Fig. 5a), indicating the preservation of the immunological integrity of the epitope. Mice were immunised s.c. with 80 µg of CgoX-D3-BSA in Freund´s adjuvant and boosted twice with 50 µg CgoX-D3-BSA in Freund´s incomplete adjuvant at day 23 and day 49. Indeed, CgoX-D3-BSA showed a highly significant, peptide-specific IgG response (Fig. 5b). The S. aureus challenge of CgoX-D3-BSA immunised mice resulted in significantly improved survival rates in the murine sepsis model (Fig. 5c). These results indicate that the CgoX-D3 epitope linked to a protein carrier suffices to elicit an effective immune response against S. aureus.

Multiple vaccine candidates for S. aureus infections have shown promise through preclinical development in a range of animal models. However, those that have reached late stage clinical testing have failed to demonstrate efficacy in human trials21,22,23 and occasionally aggravated the course of disease24. Further progress in the design of vaccines against S. aureus with greater precision and efficacy is needed and can be expected from increasing structural and functional characterisation of protective S. aureus antigens and the delineation of protective and non-protective epitopes. Herein we characterise two S. aureus vaccine antigens, CgoX and TPI that elicited a protective immune response against S. aureus infection. Two mAbs CgoX-D3 and TPI-H8 conferred protection against S. aureus bacteremia when used for passive immunisation. MAb CgoX-D3 specifically bound with high affinity to a short and highly conserved epitope. The 12 aa CgoX-D3 epitope conjugated to BSA as a carrier protein elicited excellent protective immunity, providing proof of principle for epitope-based S. aureus vaccine design. Notably, the protective mAb TPI-H8 bound to a discontinuous epitope present on an 11 kD domain within the TPI molecule, suggesting that some, but not all S. aureus antigens may be suitable for epitope-based vaccine design.

CgoX fulfils many criteria of a promising vaccine candidate. CgoX stands out from other vaccine candidates in that a short 12 aa epitope suffices to produce protective immunity, which is advantageous in many ways. Firstly, the peptide sequence of the D3 epitope is highly conserved in more than 35,000 clinical S. aureus isolates which predicts binding of mAb CgoX-D3 to almost 99% of S. aureus isolates (Fig. 3c). Thus, immunisation with the CgoX-D3 epitope warrants great coverage of clinically relevant S. aureus strains. In addition, because CgoX is essential for S. aureus heme biosynthesis, the emergence of deletion mutants is less likely. Secondly, when aligned with its human CgoX ortholog, the D3 epitope shows no identities (Fig. 3d), suggesting a low risk of eliciting cross-reacting antibodies. Indeed, mAb CgoX-D3 did not recognise human CgoX (Fig. 2e). Thirdly, the linear CgoX D3 epitope linked to BSA elicited a statistically greater protective immune response (p = 0.0004) when compared to the full-length CgoX (p = 0.0035). Clearly, an epitope-based vaccine minimises the risk of antibody-mediated enhancement of infection when compared to a whole vaccine protein that likely elicits a broader immune response with non-protective antibodies potentially causing immunopathogenic side effects25. Finally, the observation that mAb CgoX-D3 provided protection when used for passive immunisation demonstrates a great protective potential of anti CgoX-D3 antibodies independent of cellular immune responses.

CgoX operates in the cytosol at the inner leaflet of the S. aureus membrane15 and is thus not accessible for mAbs. However, its expression at the cell surface suggests an additional function of CgoX, possibly at the interface with the host, which has been described for numerous cell wall anchored antigens17. Interestingly, the staphylococcal CgoX (EC 1.3.3.15) catalyses the oxidation of coproporphyrinogen III to coproporphyrin III, whereas in humans, mitochondrial coproporphyrinogen oxidase (CPOX) catalyses the oxidation of coproporphyrinogen to protoporphyrin-IX (EC 1.3.3.3). This divergence between Gram-positive bacteria and humans has been recently exploited for the development of selective antibacterial modalities. Specifically, the activation of staphylococcal CgoX by small molecules resulted in the accumulation of coproporphyrin III, a photoreactive molecule, which sensitised bacteria for light-based antimicrobial therapies26. In humans, accumulation of porphyrins due to hereditary defects of either protoporphyrin oxidase or coproporphyrinogen oxidase are known as human variegate porphyria disease and hereditary coproporphyria27,28,29. As S. aureus invades host cells, it will be interesting to determine whether S. aureus cell surface associated CgoX interferes with host cell porphyrin metabolism. Notably, commercially available human immunoglobulin preparations contain antibodies weakly recognizing the full-sized CgoX16. Thus, a selective bolster of the production of antibodies directed against the D3 epitope might be required to establish immunity against invasive S. aureus infections.

TPI as whole protein produced protective immunity against S. aureus. In addition, passive immunisation with TPI-H8 conferred significant protection. Thus, TPI can be considered as a bona fide vaccine candidate for S. aureus. However, we did not identify linear epitopes of TPI that were suitable for epitope-based immunisation. TPI is a crucial enzyme in the glycolytic pathway catalysing the interconversion of the triose phosphate isomers dihydroxyacetone phosphate and D-glyceraldehyde 3-phosphate. TPI-deficient S. aureus mutants were not obtained, suggesting that TPI is an essential housekeeping gene. In humans, TPI deficiency is a rare autosomal recessive multisystem disorder which is dominated by lifelong hemolytic anemia and severe progressive neuromuscular degeneration30. Like other housekeeping glycolytic proteins such as GAPDH or enolase, TPI is exposed as an anchorless protein on the bacterial cell surface and may interact with extracellular matrix proteins of the host. A single report described an interaction of TPI with plasminogen18. Numerous pathogenic bacterial species intervene with the plasminogen system in vitro, suggesting that pathogenic bacteria use the plasminogen system for migration across tissue barriers or for nutritional demands during infection19. As to S. aureus, fibrinogen binding proteins (FnBPs) are well known plasminogen binding proteins, where the plasminogen binding site is masked and conformationally exposed after previous binding to fibrinogen31,32. Although the putative extracellular function of TPI remains to be resolved, our finding that active immunisation of mice with TPI could provide protection in a S.aureus bacteremia model suggested that TPI acts as S. aureus virulence factor. S. aureus TPI shows significant homology to the human TPI at the protein level (Supplementary Fig. 1). In order to minimise the risk of anti-staphylococcal TPI antibodies cross-reacting with the host TPI, a short S. aureus specific TPI epitope would be best suited for active immunisation. Thus, whereas mAb TPI-H8 does not show cross-reactivity with human TPI and might be considered for passive immunisation, the high overall homology between S. aureus TPI and its human ortholog suggests only a limited advantage of an 11 kDa TPI polypeptide over the whole antigen when used for active immunisation.

The development of human vaccines against S. aureus infections has relied primarily on inducing high titres of opsonising antibodies mediating antibody-dependent phagocytosis and bacterial killing by neutrophils and macrophages. However, all such vaccination attempts have failed eventually in human trials2,7,33,34, suggesting that opsonising antibodies do not correlate with protection. Recently, the critical role of cell-mediated immunity is being appreciated for the resolution of invasive S. aureus infections, and also for detrimental outcomes caused by imbalanced cellular immune responses, both of which have implications for vaccine development4,23. These authors suggest that vaccines targeting staphylococcal toxins and virulence factors are more likely to provide a therapeutic benefit in contrast to attempts aiming at opsonising antibodies that bear the risk of skewing the cellular immune responses, for example, by induction of cytokine production. In any case, focusing of the antibody response to a small protective epitope will avoid the production of a myriad of non-protective antibodies, thereby reducing the risk of adverse immune reactions. The use of mAbs to design new vaccines have been proposed by Burton in 2002 and is now an integral part of the ‘reverse vaccinology 2.0’ concept9,10,11. Indeed, mAbs have been widely and successfully used to understand the mechanistic nature of protection induced by vaccination35,36,37.

A phase IIb/III S. aureus vaccine study investigated the effect of a vaccine targeting S. aureusiron-regulated surface determinant B (IsdB) on the incidence of postoperative S. aureusbacteremia and/or deep sternal wound infection in adult patients undergoing cardiothoracic surgery through postoperative day 90. The trial had to be stopped prematurely after interim analysis showed lack of efficacy as well as a higher mortality rate in the subset of vaccine recipients developing S. aureus infections24. Increased levels of IsdB antibodies in patients with orthopedic infections were found to correlate with increased mortality38, suggesting an antibody-dependent immune enhancement. Notably, immunoprotective as well as non-protective mAbs against the S. aureus iron-regulated surface determinant B (IsdB) were described39. Two noncompeting epitopes were recognised by eight protective IsdB-specific mAbs, whereas two other mAbs also specifically bound to IsdB but were not efficacious in murine infection models. Thus, immunofocusing on a protective epitope of IsdB for active immunisation might result in a better safety profile. Clearly, immunisation with full antigens elicits both protective and non-protective antibodies. In general, non-protective antibodies bear potential risks of cross-reacting with human tissue, of functionally antagonising protective antibodies by steric hindrance of binding to their cognate epitope, or of producing adverse effects including enhancement of severity of infection. Moreover, non-protective epitopes of a given antigen may be immunodominant suppressing the generation of protective antibodies against immune-recessive epitopes. Thus, the use of mAb-defined protective linear epitopes is expected to have a greater efficacy and safety profile compared to a whole antigen. Not least, several short epitopes of different S. aureus vaccine antigens can be easily combined on a single polypeptide for the convenient generation of multicomponent vaccines.