Corrected and Republished from: “A Novel, Multiple-Antigen Pneumococcal Vaccine Protects against Lethal Streptococcus pneumoniae Challenge”

ABSTRACT Current vaccination against Streptococcus pneumoniae uses vaccines based on capsular polysaccharides from selected serotypes and has led to nonvaccine serotype replacement disease. We have investigated an alternative serotype-independent approach, using multiple-antigen vaccines (MAV) prepared from S. pneumoniae TIGR4 lysates enriched for surface proteins by a chromatography step after culture under conditions that induce expression of heat shock proteins (Hsp; thought to be immune adjuvants). Proteomics and immunoblot analyses demonstrated that, compared to standard bacterial lysates, MAV was enriched with Hsps and contained several recognized protective protein antigens, including pneumococcal surface protein A (PspA) and pneumolysin (Ply). Vaccination of rodents with MAV induced robust antibody responses to multiple serotypes, including nonpneumococcal conjugate vaccine serotypes. Homologous and heterologous strains of S. pneumoniae were opsonized after incubation in sera from vaccinated rodents. In mouse models, active vaccination with MAV significantly protected against pneumonia, while passive transfer of rabbit serum from MAV-vaccinated rabbits significantly protected against sepsis caused by both homologous and heterologous S. pneumoniae strains. Direct comparison of MAV preparations made with or without the heat shock step showed no clear differences in protein antigen content and antigenicity, suggesting that the chromatography step rather than Hsp induction improved MAV antigenicity. Overall, these data suggest that the MAV approach may provide serotype-independent protection against S. pneumoniae.


INTRODUCTION
S treptococcus pneumoniae is a common cause of community-acquired pneumonia (CAP), septicemia, and meningitis (1), as well as of noninvasive diseases, such as acute otitis media (AOM) and bronchitis (2). Over 90 different serotypes of S. pneumoniae, determined by the characteristics of the capsular polysaccharide (CPS), have been identified (3). There are currently two vaccines available to prevent S. pneumoniae infections: the pneumococcal polysaccharide vaccine (PPV) and the pneumococcal conjugate vaccine (PCV). Each consists of capsular polysaccharide antigen from a limited panel of S. pneumoniae serotypes. In the United Kingdom, PPV remains the first choice for adult vaccination (4), and PCV is routinely included in childhood immunization schedules worldwide, as it has greater efficacy than PPV in infants. Unfortunately, in developing countries the high cost of PCV restricts its availability, and in addition, serotype coverage is reduced, as PCV was designed to include the most prevalent serotypes in North America (5). Furthermore, serotype replacement in response to PCV vaccination alters the ecology of S. pneumoniae, reducing the efficacy of polysaccharide vaccines over time (6). A vaccine based on protein antigens may provide a low-cost alternative approach capable of inducing cross-serotype protection (7,8).
One vaccine approach dependent on protein antigens is a whole-cell approach, a cost-effective method of immunizing with a large number of potential protein antigens to potentially induce serotype-independent protective immunity. In addition, a wholecell approach could target both humoral and cellular host immunity (9,10), potentially enabling clearance of both disease and colonization. Several groups have therefore studied a whole-cell vaccine approach against S. pneumoniae, including progression to early-phase clinical trials (11)(12)(13). An alternative to maintaining protein antigens as part of the whole S. pneumoniae bacterium is using a bacterial lysate as a vaccine, which could result in a more stable preparation that is better suited to vaccine delivery than a whole bacterium. However, the antigenicity of whole-cell lysates may be weak and require enhancement (14). One method of enhancing immunogenicity is altering the preparation of the lysate to ensure increased representation of immunoprotective proteins. This can be partially achieved using anion-exchange chromatography with a pH 8.0 buffer to preferentially capture several well-known S. pneumoniae antigens, all of which have a pI of 7.5 or lower, including PiuA, PiaA, PsaA, RrgA, RrgB, ClpP, PspA, and Ply. In addition, growth under stress conditions, such as high temperatures, to induce heat shock proteins (Hsps) could increase antigenicity (15), as Hsps facilitate the cross-presentation of peptides (16,17) and act as natural adjuvants by stimulating macrophages and dendritic cells to cause cytokine secretion (18)(19)(20). As a result, Hsps have been studied as vaccines that protect against cancer as well as microbial pathogens (21), with a number of bacterial Hsps showing promise as vaccine candidates (22)(23)(24), including in models of lethal lung infection (25)(26)(27). For example, mice intranasally immunized with the S. pneumoniae Hsp DnaJ (Hsp40) or Hsp caseinolytic protease P (ClpP) were protected from S. pneumoniae infection, including against systemic challenge with a panel of heterologous strains (28). Hence, Hsps are potential vaccine antigens with advantageous immunomodulatory properties that could be used as a component of a broadly protective S. pneumoniae vaccine.
Here we present data on a multiple-antigen approach to a novel S. pneumoniae vaccine based on bacterial lysates that combines the advantages of a whole-cell approach with the potential additional benefit of an increased Hsp and surface antigen content in the vaccine preparation.

RESULTS
Formulation of an S. pneumoniae MAV. A multiple-antigen Hsp-enriched preparation based on a whole-cell preparation was formulated using the ImmunoBiology Ltd. platform technology as previously described (29). Heat shock was used to enrich for Hsps, and anion-exchange chromatography was used to enrich for negatively charged S. pneumoniae antigens (e.g., PspA and Ply) (Fig. 1A). Immunoblots determined which elution fractions contained the highest concentration of the Hsp60 and Hsp70 proteins and demonstrated a marked increase in the expression of both the Hsp60 and Hsp70 content in the multiple-antigen vaccine (heat shock) (MAV hs ) compared to the bacterial heat-killed lysate (HKL) ( Fig. 1B and C). A pooled human IgG preparation known to recognize multiple S. pneumoniae protein antigens (9) was used to probe MAV hs and HKL to determine whether there were differences in their non-Hsp contents. This demonstrated variations in the number, intensity, and molecular weights of bands identified after incubation in sera from MAV hs -or HKL-vaccinated animals (Fig. 1D). Ply activity in the MAV hs , the HKL, and a heat-killed whole-cell (HKWC) preparation formulated with the wild-type TIGR4 strain was assessed using a hemolysis assay. HKWC and HKL did not cause the lysis of red blood cells, probably due to the degradation of Ply during the heat-killing step, whereas MAV hs caused red cell lysis, suggesting that the MAV hs preparation still contained active Ply (Fig. 1E).
Proteomic analysis of MAV hs preparations. Tandem mass spectrometry (MS/MS) sequencing was used to identify proteins in the MAV hs and HKL preparations, with tandem mass tag (TMT) labeling being used to assess relative protein quantities. A total of 627 proteins were identified and compared between the MAV hs and HKL preparations. Of these, 57 were increased .2-fold in MAV hs compared to HKL (see Table S1 in the supplemental material), including several Hsps and important known surface antigens, such as PavB and several lipoproteins, including PsaA, PiaA, and the Th17 antigens SP_0148 and SP_2108 (30) (Table 1). Conversely, 152 proteins were decreased by more than 2-fold by the vaccine preparation process, including multiple proteins required for basic metabolic functions (e.g., ribosomal proteins) and capsule synthesis, but also the virulence factors and protective antigens Ply, PspA, and PspC (Table S2). Previously, we have published data obtained using deltaDOT capillary gel electrophoresis (CGE) demonstrating a consistent protein content between different batches of MAV hs preparations (29).
Vaccination with MAV induces functional antibodies. To assess the immunogenicity of MAV hs , mice were immunized by subcutaneous injection with either MAV hs , HKL, HKWC, or the negative-control buffer using a two-dose schedule 21 days apart. Whole-cell enzyme-linked immunosorbent assays (ELISAs) demonstrated that pooled serum obtained 1 week after the second vaccinaton with MAV hs contained antibody responses to the S. pneumoniae TIGR4 strain markedly higher than those detected in serum from HKL-vaccinated mice. No statistically significant anti-TIGR4 response was identified in sera from mice vaccinated with HKWC ( Fig. 2A). The serum antibody response to MAV hs was dominated by IgG, with no significant IgM response compared to that in buffer-vaccinated mice (Fig. 2B). Significantly increased levels of anti-TIGR4 FIG 2 MAV is immunogenic in a mouse model of subcutaneous vaccination. CD1 mice were vaccinated subcutaneously with 75 mg on day 0 and day 21 and culled at 28 days to obtain serum. (A) Results of a whole-cell IgG ELISA against S. pneumoniae TIGR4 for pooled sera harvested from tail vein bleeds (10 ml per mouse, n = 6). (B and C) Results of whole-cell IgG and IgM ELISAs against S. pneumoniae TIGR4 for pooled sera from MAV-vaccinated mice (n = 5) (B) and against S. pneumoniae TIGR4 for pooled BALF and nasal wash specimens from MAVvaccinated mice (C). Data are presented as the mean and 95% confidence interval. P values were calculated using the Mann-Whitney t test. **, P , 0.01; ns, not significant. IgG were also detected in the bronchoalveolar lavage fluid (BALF) of mice immunized with MAV hs but not in nasal washes (Fig. 2C). Whether the serum IgG induced by the MAV hs , HKL, and HKWC preparations can recognize and bind to the surface of live S. pneumoniae was assessed using a flow cytometry assay that correlates with protection (31). Compatible with the ELISA data, there were higher levels of IgG binding when S. pneumoniae TIGR4 was incubated in serum from mice vaccinated with MAV hs than after incubation in serum from mice vaccinated with HKL or HKWC. Incubation in serum from HKWC-vaccinated mice also resulted in less IgG binding to S. pneumoniae than incubation in sera from HKL-vaccinated mice ( Fig. 3A to C). To investigate IgG binding to heterologous strains, the IgG binding assays were repeated using S. pneumoniae serotype 18C, 23F, 3, and 19F (EF3030) strains. Sera from MAV hs -vaccinated mice significantly increased serum IgG binding to the serotype 18C, 23F, 3, and 19F (EF3030) strains compared to that in a buffer-vaccinated control serum ( Fig. 3D and E). Sera from MAV hs -vaccinated mice bind to multiple protein antigens. Immunoblots against lysates from the S. pneumoniae TIGR4, D39, and 19A strains demonstrated that antibodies in sera from MAV hs -and HKL-vaccinated mice recognized a number of proteins which were largely conserved between the three strains. The antigens recognized after probing with sera from MAV hs -and HKL-vaccinated mice overlapped, although a band at approximately 75 kDa (potentially consisting of multiple proteins) was recognized by serum from MAV hs -vaccinated mice but was not recognized by serum from HKL-vaccinated mice (Fig. 4A). A Meso Scale Discovery (MSD) multiplex assay that measures the levels of IgG to a panel of known S. pneumoniae surface and immunogenic proteins was used to identify some of the protein antigens recognized by sera from vaccinated mice (32). IgG in serum from a MAV hs -vaccinated mouse recognized all the antigens included in the MSD assay panel of proteins, including PspC (CbpA), PspA, PsaA, PiaA, PiuA, and the pilus proteins RrgA and RrgB, all of which have previously been shown to be protective vaccine candidates in mice (33)(34)(35)(36) (Fig. 4B). In contrast, IgG in sera from HKL-vaccinated mice recognized fewer proteins, with no responses to PspC, LytC, PcsB, PiaA, PiuA, family 1 PspA, SP_0609, SP_2027, Spr0057 (StrH), and StkP.
Comparison of MAV preparations made under different conditions using Plydeficient S. pneumoniae. During the development of the bioreactor process, for consistency culture conditions were changed to a temperature of 30°C with a switch to 37°C for the final 30 min. In addition to eliminating potential adverse effects caused by an active Ply, MAV preparations were made using a mutated TIGR4 strain expressing a detoxified Ply (49). To ensure culture at 30°C rather than 37°C did not alter MAV preparation antigen content significantly, two new MAV preparations termed MAV IPS004 and MAV IPS005 were prepared. MAV IPS004 was cultured at 30°C before switching to 37°C for 30 min, and MAV IPS005 was cultured at 37°C throughout. Comparing the MAV IPS004 and MAV IPS005 preparations using immunoblots showed no clear differences in expression of the Hsps Hsp70 and Hsp60 or in the expression of the immunogenic proteins PspA and Ply (Fig. 5A). The capillary gel electrophoresis profiles of both preparations suggested only minor overall differences in their protein constituents (Fig. 5B). Both preparations were used to generate antisera using vaccination experiments in mice conducted at Churchill Applied Biotechnology Ltd. Sera recovered from mice vaccinated with MAV IPS004 and MAV IPS005 showed no major statistically significant differences in flow cytometry assays of IgG binding to live S. pneumoniae bacteria or whole-cell ELISA titers against the TIGR4 S. pneumoniae strain (Fig. 5C). Specific protein antigen ELISAs demonstrated increased antibody titers to Ply in serum from mice vaccinated with heat-shocked MAV IPS004 compared to those in serum from mice vaccinated with MAV IPS005 (Fig. 5D) and reduced titers to PspA (Fig. 5E). When measured using the MSD multiplex assay (32), there were no differences between the MAV IPS004 and MAV IPS005 preparations in responses to other antigens (PiuA, PsaA, RrgA, and RrgB; data not shown). These data indicate that reducing culture temperature for the preparation of MAV to 30°C had limited effects on protein content or overall antigenicity. Protective efficacy of vaccination of mice with MAV TIGR4 preparations. Mouse models were used to determine if vaccination with Hsp-induced MAV resulted in protective immunity against S. pneumoniae. At 24 h after challenge using the pneumonia model, mice vaccinated with MAV hs had at least 1 log 10 fewer numbers of bacterial CFU in both the blood and lungs compared to the buffer-vaccinated controls ( Fig. 6A and B). In contrast, MAV hs vaccination did not reduce the density of the number of bacterial CFU in nasal washes obtained 2 weeks after inoculation in a model of S. pneumoniae nasopharyngeal colonization with TIGR4 (Fig. 6C). In order to eliminate potential adverse effects caused by an active Ply, new MAV preparations, termed MAV IPS004 and MAV IPS014 , denoting different batches, were prepared using a mutated TIGR4 strain expressing a detoxified Ply (37). Both MAV IPS004 and MAV IPS014 contained similar levels of detoxified Ply, as measured by ELISA, that stimulated an antibody response that recognized native Ply (29) and the absence of hemolysis in a red blood cell assay. Sera obtained from rabbits vaccinated subcutaneously on days 0, 21, and 35 with 375 mg of MAV IPS004 , the S. pneumoniae vaccine Prevenar as a positive control, or buffer were used for passive immunization of mice, followed by intraperitoneal challenge with  (Continued on next page) 1 Â 10 4 CFU of TIGR4 S. pneumoniae after 6 h. When culled at 24 h postchallenge, colonies were recovered from the blood of over 65% of the mice given sera from buffervaccinated rabbits, while there no colonies were detected in mice given sera from MAV IPS004 -vaccinated rabbits (Fig. 6D). Two mice (16%) developed septicemia after passive administration of sera from rabbits given the Prevenar vaccine. In an alternative sepsis model, mice were challenged by intravenous inoculation with 5 Â 10 5 CFU of S. pneumoniae TIGR4 or ATCC BAA-1662 (serotype 18C S. pneumoniae) strains after preincubation of the bacteria for 1 h in 100% serum obtained from rabbit immunized with MAV IPS014 , the Prevenar vaccine, or buffer controls. Mice were culled at 4 h to assess the rate of bacterial clearance from the blood by quantifying the number of CFU. Those challenged with the TIGR4 strain were almost completely protected against infection if the bacteria were incubated in sera from either MAV IPS014 -or Prevenar-vaccinated rabbits prechallenge (Fig. 6E). Preincubation of the 18C strain prior to intravenous challenge in sera from either MAV IPS014 -or Prevenar-vaccinated rabbits prechallenge reduced the number of bacterial CFU recovered from the blood by over 1 log 10 compared to the number obtained with preincubation in sera from buffer-vaccinated rabbits (Fig. 6F). To support these data, vaccination with MAV IPS014 was compared to that with a buffer control in a protection study, which demonstrated that mice given a three-dose vaccination schedule with MAV IPS014 were protected against the development of fatal infection after pneumonia challenge with TIGR4 (Fig. 7A). The effects of vaccination on the inflammatory response to pneumonia challenge was assessed using flow cytometry of lung and BALF cell populations 24 h after intranasal infection with TIGR4. Despite clear reductions in the numbers of CFU in the lungs and blood of vaccinated mice, there were no differences in the proportions of neutrophils in BALF between MAV IPS014 -and buffer-vaccinated mice, indicating that MAV IPS014 vaccination resulted in an increased BALF neutrophil response for the level of bacterial infection ( Fig. 7B and C). In addition, in lung homogenates, MAV IPS014 -vaccinated mice had reduced proportions of neutrophils and macrophage lineage cells and a corresponding increase in T cells (Fig. 7D). The increase in the T cell proportion within lung homogenates in vaccinated mice did not alter the CD4/CD8 proportions compared to the data for control mice (data not shown). Lung homogenate cytokine levels were variable between mice but showed increased interleukin-1 (IL-1), IL-6, tumor necrosis factor alpha (TNF-a), and IL-10 responses in vaccinated mice compared to the controls, again suggesting that vaccinated mice were able to mount a more sustained inflammatory response than the controls (Fig. 7E to H). BALF and blood cytokine levels were, in general, too low and variable for consistent patterns to be identified.
Overall, these results indicate that the MAV is able to protect against S. pneumoniae infection with the homologous strain or a heterologous strain at a level of protection similar to that provided by the Prevenar vaccine and is associated with significant changes in the inflammatory response to pneumonic infection.

DISCUSSION
Although the existing childhood conjugated capsular polysaccharide S. pneumoniae vaccines are highly effective, both the lack of protection against nonvaccine serotypes and convincing evidence of serotype replacement in vaccinated populations, as well as the high cost of these vaccines, have stimulated continued interest in alternative vaccine approaches (38,39). Vaccines based on protein antigens could overcome these disadvantages, allowing production of relatively inexpensive vaccines that target cross-protective antigens (40)(41)(42)(43). Presentation of a large number of protein antigens derived from a whole-cell approach has the additional potential advantages of inducing immune responses to multiple antigens, thereby potentially avoiding the development of vaccine escape mutants and inducing stronger cross-protective responses (44,45). The data presented here show that a novel S. pneumoniae multiple-protein-antigen vaccine approach induces antibody that recognizes homologous and heterologous strains and protects against invasive pneumonia and sepsis. The MAV approach uses a whole-cell lysate that has been manipulated to increase the expression of Hsps, and the anion-exchange column and running buffer are optimized for the capture of known S. pneumoniae antigens as well as Hsps. Proteomic analysis confirmed that the MAV approach alters the relative levels of S. pneumoniae proteins within the preparation compared to those in a simple bacterial lysate, with altered expression of a total of 209 proteins of the 627 proteins analyzed. The S. pneumoniae TIGR4 strain actually contains approximately 2,000 genes (46), so the number of proteins with altered content in the MAV may, in fact, be substantially larger. Both immunoblotting and TMT-MS confirmed that the MAV had an increased Hsp content compared to that of a simple lysate. In addition, there was increased expression of multiple surface proteins, including known protective antigens (mainly lipoproteins), although there were also reduced amounts of other protein antigens that are known to induce protective immunity in mice.
Importantly, whole-cell ELISAs, the MSD assay for IgG responses to individual antigens, and a flow cytometry assay of IgG binding to S. pneumoniae all demonstrated enhanced antibody responses in mice vaccinated with the MAV compared to the HKL, demonstrating the benefit of the MAV approach in making a potentially more effective vaccine. Unlike HKL, MAV induced antibody responses to all the protein antigens tested using the MSD system, suggesting that the MAV approach may induce antibody responses to a very high proportion of the proteins in the preparation. Furthermore, although iTRAQ demonstrated a reduced content in the MAV preparation compared to that in HKL for the important antigens PspC and Ply, vaccination with MAV still induced stronger IgG responses to these antigens, when tested using the MSD assay, than vaccination with HKL. These data again suggest that the MAV approach enhances immunogenicity. The improved serological responses to MAV compared to HKL might be predicted to be due to the increased Hsp content leading to the formation of Hsp-peptide complexes and thereby increasing the antigen-presenting ability (27,29). A consistent shift of culture temperature to 42°C was difficult when MAV production was scaled up using bioreactors. This required a change in culture conditions to 30°C with a switch to 37°C for the final 30 min before harvest, conditions which also mimic the transfer of pneumococci from the nasopharynx to the bloodstream during invasive disease. In the studies reported here, MAV prepared under these conditions demonstrated similar antigenicity in comparison to a MAV preparation cultured at 37°C throughout.
Data from a pneumonia model and two separate mouse models of sepsis demonstrated that active or passive vaccination with MAV improved protection against S. pneumoniae. Time course experiments demonstrated that vaccination with the MAV preparation delayed and protected against lethal infection. The total T cell populations in the lung were also increased in MAV-vaccinated mice. This effect in T cell proportions probably reflects the accumulation of antigen-specific cells during pneumonia in infected tissue, but further experiments would be needed to confirm this. Overall, the data demonstrate that vaccination with MAV had significant effects on the inflammatory response during S. pneumoniae pneumonia that are likely to contribute to protective efficacy, with improved neutrophil recruitment into the alveolar space and increased numbers of T cells within the lung parenchyma.
MAV also induced IgG that bound to heterologous strains in the flow cytometry assay, an assay which correlates with improved phagocytosis in vitro and protection in mouse models (31,47,48). Indeed, in a mouse model of sepsis, passive vaccination with sera from MAV-vaccinated rabbits gave a level of protection against both the homologous TIGR4 and heterologous 19C strains similar to that achieved with passive vaccination with sera from Prevenar-vaccinated rabbits. Whether vaccination with MAV can induce stronger protection than vaccination with HKL or HKWC in the mouse models was not tested; due to the relatively low sensitivity of the mouse models, this is likely to require inappropriately large numbers of mice per test group. A lack of sensitivity is also why vaccination followed by disease challenge experiments would be unlikely to show any significant differences between MAV preparations with and without heat shock steps and why these experiments were not performed. The level of protection varied between models, with complete prevention of septicemia being seen in some models, e.g., intravenous TIGR4 infection, but reductions in the number of bacterial CFU only in the blood being seen in others (e.g., the serotype 18C sepsis model). The latter is likely to slow the progression of disease but not prevent fatal infection. Vaccination with MAV failed to protect against colonization, but this is not that surprising, given the lack of detectable antibody in nasal washes and previous data showing that antiprotein antibody is often ineffective at reducing the number of nasopharyngeal S. pneumoniae CFU (in some cases, even when Th17-mediated immunity has also been induced) (9,10,49). Prevention of nasopharyngeal colonization will probably require vaccines that induce strong cell-mediated immune mechanisms, which may require vaccination in combination with specific adjuvants (24,46,50). This would be an important area for further investigation. Future experiments should also assess whether MAV vaccination modulates pulmonary inflammatory responses to S. pneumoniae pneumonia challenge.
The MAV approach described here can induce the cross-protective immunogenicity of protein antigens which stimulate antibody and perhaps Th17 cell responses (30,51) without requiring the identification of specific protective antigens or the production of recombinant proteins for inclusion in subunit vaccines. MAV requires limited downstream processing and rapidly produces a high yield of vaccine product, considerably reducing vaccine costs and making the vaccine more likely to be affordable in lowand middle-income countries. The MAV approach therefore offers a promising opportunity for a novel next-generation S. pneumoniae vaccine and has recently completed a phase I trial in 36 subjects (ClinicalTrials.gov registry number NCT0257635 [52]).
MAV, heat-killed lysate, and heat-killed whole-cell preparation. MAV preparations were cultured in Hoeprichs' media (made in-house) in 1-liter shake flasks to an OD 600 of 1.2. The MAV hs preparation was made from TIGR4 cultured at 37°C before switching to 42°C for 30 min. MAV IPS004 and MAV IPS005 were made from S. pneumoniae TIGR4 B7.1 (PlyD6) in 1-liter bioreactors, with MAV IPS004 cultured at 30°C and then switching to 37°C for 30 min before harvest and MAV IPS005 cultured at 37°C throughout. The bacteria were then centrifuged twice with wash buffer (40 mM Tris, 150 mM NaCl, 1 mM MgCl 2 , pH 8.0) and incubated with lysis buffer (40 mM Tris, 20 mM NaCl, 1 mM MgCl 2 , 0.5% [wt/vol] sodium deoxycholate [NaDOC], pH 8.0) for 1 h at 4°C, before homogenization (EmulsiFlex C5 high-pressure homogenizer; Avestin, Germany) and incubation with 0.1% (wt/vol) octaethylene glycol monododecyl ether (C 12 E 8 ) for 4 h at 4°C. Sample supernatants were harvested using a 5-ml Capto Q column (GE Healthcare, UK). Protein was eluted and collected as 5-ml fractions at 400 mM and 500 mM concentrations of NaCl. MAV IPS014 was made as described above using TIGR4 B7.1 (PlyD6) culturing at 30°C and then switching to 37°C for 30 min. MAVIPS014 was then lysed in NaDOC (0.5%) and Triton X-100 (1%). MAV IPS004 and MAV IPS014 contained similar levels of detoxified Ply when measured by ELISA (32) and the absence of hemolysis in the red blood cell lysis assay. For heat-killed lysates (HKL) of TIGR4 and heat-killed whole-cell (HKWC) preparations, TIGR4 was also cultured in Hoeprichs' medium in 1-liter shake flasks at 37°C to an OD 600 of 1.2, followed by killing by incubation at 65°C for 45 min, which was confirmed by culture on Columbia agar plates. For the HKL, lysis was performed using NaDOC and high-pressure homogenization as described above. C 12 E 8 was added to both the HKL and HKWC before filtration.
Capillary gel electrophoresis. Analysis of the samples by capillary gel electrophoresis (CGE) was conducted by deltaDOT, London BioScience Innovation Centre, using the high-performance capillary electrophoresis (HPCE) platform Peregrine. Peaks were manually selected, and raw data are expressed as the peak area corrected for migration time and then expressed as a percentage of the total corrected peak area or the area under the curve (inclusive of the Triton X-100 peak [peak 1] and interpeak regions [regions A to G]) normalized to an external protein standard. This compensates for day-to-day variation and allows comparison between runs on different sections of the capillary. Peak 1 is the detergent Triton X-100 peak.
In vitro assays. S. pneumoniae whole-cell ELISAs were performed using bacterial cultures at an OD 580 of 0.4 to 5, alkaline phosphatase (AP)-conjugated secondary antibodies, and the substrate para-nitrophenylphosphate (pNPP; Sigma) as previously described (56,57). The absorbance was read at 450 nm, and the readings obtained at 630 nm were subtracted (VersaMax). The ELISA titer represents the theoretical sample dilution that would result in an OD 450 reading minus OD 630 reading of 0.1. For detection of anti-Ply and anti-PspA antibodies, the appropriate antigen was diluted in carbonate buffer to a final concentration of 1 mg/ml, 100 ml was transferred to each well of a 96-well MaxiSorp ELISA plate, and the plate was incubated overnight at 4°C. The plates were then washed 3 times with ELISA wash buffer (1% [vol/ vol] Tween 20-PBS), blocked for 1 h with block buffer (1% [wt/vol] bovine serum albumin [BSA]-PBS) at 37°C, and then washed as described above. Serum samples were diluted to a starting dilution of 1/100 to 1/300. Doubling dilutions of the pooled serum samples were assayed in duplicate. The plates were incubated at 37°C for 1 h and washed, and goat anti-mouse IgG-horseradish peroxidase diluted to 1/ 20,000 was added before incubation at 37°C for 1 h before washing. Tetramethylbenzidine substrate was added before incubation at room temperature in the dark. The plates were read at OD 450 , and endpoint titers were calculated using the linear part of each titration curve. IgG surface binding was assessed using previously described flow cytometry assays (31,47,48) and species-appropriate secondary antibodies: anti-human IgG secondary antibody (1:200) conjugated to phycoerythrin (PE; Sigma-Aldrich) and goat anti-mouse IgG conjugated to fluorescein isothiocyanate (Bio-Rad, USA). Fluorescenceactivated cell sorting (FACS) analysis of bacterial cells was performed on a FACSVerse flow cytometer (Becton, Dickinson, USA) and with FACSuite (Becton, Dickinson) and FlowJo (Becton, Dickinson) software. FACS surface binding comparisons of MAV IPS004 and MAV IPS005 were conducted by ImmBio, as follows: 100 ml of bacterial suspension was placed in each 5-ml FACS tube, and the tubes were incubated overnight at 4°C. Cells were washed with PBS-0.1% Tween 20 (PBS-T). Preadsorbed serum samples were serially diluted (2-fold), starting from 1 in 25, to 1 in 800 in PBS-1% BSA. Cells and sera were incubated together for 2 h at room temperature and then washed with PBS-T. Goat anti-mouse IgG detection antibody in PBS-1% BSA was added, and the mixture was incubated for 2 h. Cells were then fixed with formalin for 30 min at room temperature. Following washing in PBS-T, samples were resuspended in PBS-1% fetal calf serum. The mean fluorescence intensity (MFI) was read by flow cytometry, with 100,000 events being required for each sample. The MFI multiplied by the number of cells in quadrant 1 is shown for each serotype. This represents the degree of antibody binding to each serotype. Multiplexed electroluminescence assays were conducted as previously described (32,58,59) using a Meso Scale Discovery (MSD; MD, USA) platform assay (60) with 5 mg/ml of S. pneumoniae proteins and 10 mg/ml of capsular polysaccharide. After incubation of each antigen-coated plate with blocking agent, washing, and incubation with diluted test serum for 45 min at room temperature, the plates were washed and MSD assay sulfo tag-labeled goat anti-mouse IgG secondary antibody was added for reading using an MSD Sector Imager 2400 or 6000 apparatus.
Quantitative comparison of protein content using TMT and MS. The tandem mass tags (TMT) labeling procedure followed the manufacturer's recommendation (Thermo Fisher). In brief, protein lysates from two replicates of HKL and MAV were reduced with tris(2-carboxyethyl) phosphine and alkylated with iodoacetic acid before an overnight acetone precipitation. Protein pellets were digested overnight at 37°C in 200 mM TEAB solution containing 2.5 mg trypsin (Promega), with the resulting peptides being labeled with different isobaric tags (TMTs 126 to 128). Labeled peptides were mixed and injected onto an XBridge C 18 column (particle size, 5 mm; 4.6 mm [inside diameter] by 25 cm long; Waters) for first-dimension high-pH reversed-phase high-performance liquid chromatography (HPLC) separation under a linear gradient consisting of mobile phase A (10 mM ammonium formate, pH 10.0) and up to 70% mobile phase B (90% acetonitrile in mobile phase A) for 2 h at flow rate of 0.5 ml/min, using a Jasco system consisting of an autosampler, semimicro-HPLC pumps, and a UV detector. The eluted fractions were collected, concatenated into 18 tubes, and vacuum dried.
Nano-liquid chromatography and tandem mass spectrometry (MS/MS) were performed using a U3000 direct nano system coupled with a nano electrospray and LTQ-Orbitrap Discovery mass spectrometer (Thermo). The 12 HPLC fractions containing the mixture of fourplex-labeled peptides were resuspended in 0.1% formic acid, and each was separated on a PepMap C 18 reversed-phase nano column (particle size, 3 mm; 100 Å; 50-cm length; Thermo) under a column flow rate of 0.3 ml/min using a linear gradient of 95% acetonitrile and 0.1% formic acid at 5 to 25% for 180 min, 25 to 32% for 20 min, and 32 to 90% for 10 min. MS scan and MS/MS fragmentation were carried out in the LTQ-Orbitrap Discovery mass spectrometer, using 2 cycles of top 3 data-dependent acquisition with the dynamic exclusion mode enabled and a total cycle time of approximately 30 ms. The first cycle used collision-induced dissociation (CID) fragmentation-generating spectra for peptide sequencing, and the second used high-energy-CID (HCD)-generating spectra both for peptide sequencing and for relative quantitation via report ions.
Mass spectrum processing, database searching, and quantitation against the UniProt S. pneumoniae FASTA database (release 2014.04.03) were performed using Thermo Proteome Discoverer (version 1.4) software with a built-in Sequest program. Spectra from the 12 fractions were added together as one sample during the search. Initial mass tolerances by MS were set to 10 ppm. Up to two missed tryptic cleavages were considered. Methionine oxidation was set as a dynamic modification, whereas carboxymethylation on cysteine and TMT 6-plex labels on the N-terminal amino acid and the lysine side chain were set as static modifications. Peptides at rank 1 with a high confidence were considered to be unambiguously sequenced. Quantification was based on the relative abundances of the TMT tag as the reporter ions for each peptide in the HCD spectra with all TMT channels present. Ratios were calculated from the relative abundances of each labeled peptide in the sample based on reporter ion intensities, and for every protein identified, each was assigned a series of quantification ratios relative to each group.
In vivo methods. All in vivo experiments using mice were performed according to United Kingdom national guidelines for animal use and care. Experiments performed at UCL were approved by the UCL Biological Services Ethical Committee and the UK Home Office (project license PPL70/6510). Experiments used 6-week-old outbred female CD1 mice obtained from Charles River Laboratories. Mice were vaccinated with 75 mg of protein in 100 ml PBS using either intraperitoneal injection at days 0, 14, and 28 or subcutaneous vaccination on days 0 and 21. Tail bleeds (5 ml per mouse) were collected on day 42, and mice were challenged with S. pneumoniae on day 49. For the pneumonia model, mice under isoflurane (4%; MiniRad) anesthesia were inoculated with 5 Â 10 6 CFU of S. pneumoniae in PBS intranasally. After either 24 or 48 h, the mice were euthanized with pentobarbitone, and blood, serum, bronchoalveolar lavage fluid (BALF), the lungs, and the spleen were collected as previously described (9,31,50,61). Lungs and spleens were macerated through a 0.2-mm-pore-size filter. For the colonization model, mice were anesthetized with aerosolized isoflurane (4%) and inoculated with 5 Â 10 6 CFU of S. pneumoniae suspended in 10 ml of PBS. At designated time points postinfection, the mice were culled and nasal washes were obtained by retrograde washing of the nares with 500 ml PBS via the trachea. To assess survival, mice were vaccinated by intraperitoneal inoculation with 75 mg of the MAV IPS014 vaccine together with the adjuvant system (catalog number S6322; Sigma) on days 1, 10, and 22 before intranasal challenge with 1 Â 10 7 CFU of S. pneumoniae TIGR4 on day 50. Disease development was monitored over 6 days, and mice were culled when they exhibited signs of severe disease (61). For the passive transfer model, mice were injected intraperitoneally with 200 ml of serum harvested from rabbits vaccinated by subcutaneous injection of 375 mg MAV on days 0, 21, and 35 or two 200-ml Prevenar13 doses on days 0 and 21 at Envigo, UK. Mice were challenged 6 h later by intraperitoneal inoculation of 1 Â 10 4 CFU of S. pneumoniae and culled at 24 h to obtain blood samples for plating. For the sepsis model, sera from vaccinated mice or rabbits were transferred to mice via intravenous injections into the lateral tail vein. After 4 h, the mice were inoculated intravenously with 5 Â 10 5 CFU of S. pneumoniae and culled 4 h later to collect blood. For the preopsonization clearance model, S. pneumoniae bacteria were opsonized by incubation in 100% rabbit immune serum for 1 h at 37°C, and then 5 Â 10 5 CFU of S. pneumoniae was inoculated intravenously into the mice, which were then culled 4 h later to obtain blood for CFU quantification by plating. To calculate the numbers of target organ CFU, aliquots of blood and lung and spleen tissues were plated at appropriate dilutions on 5% blood Columbia agar plates containing 5 mg/ml gentamicin (Sigma). Additional experiments to raise antisera with different vaccine preparations were performed at a commercial organization, Churchill Applied Biotechnology Ltd., according to institutional guidelines under its UK Home Office project license. For these experiments, six groups of female CD-1 mice (n = 10) were immunized subcutaneously with 0.75 mg of MAV on day 0 and day 21. Mice were culled on day 35 and terminally bled, and serum was prepared for the investigation of antibody responses.
Statistical methods. Statistical analyses were conducted using Prism (version 7) software (GraphPad, USA). Parametric data are presented as means, and error bars represent standard deviations (SD). Comparisons between multiple groups were conducted using analysis of variance (ANOVA) and the Holm-Sidak or Dunnett's posttest to compare experimental groups. Nonparametric data were analyzed using the Mann-Whitney U test. For the disease development model, data were analyzed using the log-rank (Mantel-Cox) test.

ACKNOWLEDGMENTS
W.-Y.C. was supported by an MRC CASE studentship in collaboration with ImmunoBiology Ltd.; G.E. and E.R.-S. are supported by MRC grants MR/R001871/1 and R/N02687X/1, respectively. This work was undertaken at UCLH/UCL, which received a proportion of funding from the Department of Health's NIHR Biomedical Research Centre's funding scheme.
J.S.B. and D.G. have received fees for consultancy work from ImmunoBiology Ltd.; C.E., A.M., P.C., and C.B. are employees of ImmunoBiology Ltd.