Vaccine composition formulated with a novelTLR7-dependent adjuvant induces high and broadprotection against Staphylococcus aureusFabio Bagnolia, Maria Rita Fontanaa, Elisabetta Soldainia, Ravi P. N. Mishraa, Luigi Fiaschia, Elena Cartoccia,Vincenzo Nardi-Deia, Paolo Ruggieroa, Sarah Nosaria, Maria Grazia De Falcoa, Giuseppe Lofanoa, Sara Marchia,Bruno Gallettia, Paolo Mariottia, Marta Bacconia, Antonina Torrea, Silvia Maccaria, Maria Scarsellia, C. Daniela Rinaudoa,Naoko Inoshimab, Silvana Savinoa, Elena Moria, Silvia Rossi-Paccania, Barbara Baudnera, Michele Pallaoroa,Erwin Swennena, Roberto Petraccaa, Cecilia Brettonia, Sabrina Liberatoria, Nathalie Noraisa, Elisabetta Monacia,Juliane Bubeck Wardenburgb, Olaf Schneewindc, Derek T. O’Hagana, Nicholas M. Valiantea, Giuliano Bensia,Sylvie Bertholeta, Ennio De Gregorioa, Rino Rappuolia,1, and Guido Grandia,1

aNovartis Vaccines Research Center, 53100 Siena, Italy; bDepartments of Pediatrics and Microbiology, University of Chicago, Chicago, IL 60637;and cDepartment of Microbiology, University of Chicago, Chicago, IL 60637

Contributed by Rino Rappuoli, January 25, 2015 (sent for review September 1, 2014)

Both active and passive immunization strategies against Staphylococ-cus aureus have thus far failed to show efficacy in humans. With theattempt to develop an effective S. aureus vaccine, we selected fiveconserved antigens known to have different roles in S. aureus path-ogenesis. They include the secreted factors α-hemolysin (Hla), essextracellular A (EsxA), and ess extracellular B (EsxB) and the twosurface proteins ferric hydroxamate uptake D2 and conserved staph-ylococcal antigen 1A. The combined vaccine antigens formulatedwithaluminum hydroxide induced antibodies with opsonophagocytic andfunctional activities and provided consistent protection in four mousemodels when challengedwith a panel of epidemiologically relevant S.aureus strains. The importance of antibodies in protection was dem-onstrated by passive transfer experiments. Furthermore, when for-mulated with a toll-like receptor 7-dependent (TLR7) agonist recentlydesigned and developed in our laboratories (SMIP.7–10) adsorbed toalum, the five antigens provided close to 100% protection againstfour different staphylococcal strains. The new formulation inducednot only high antibody titers but also a Th1 skewed immune responseas judged by antibody isotype and cytokine profiles. In addition, lowfrequencies of IL-17–secreting T cells were also observed. Altogether,our data demonstrate that the rational selection of mixtures of con-served antigens combinedwith Th1/Th17 adjuvants can lead to prom-ising vaccine formulations against S. aureus.

Staphylococcus aureus | vaccine | TLR7 | adjuvant | Hla

Current antibiotics are not efficacious against emergingmultidrug-resistant strains of Staphylococcus aureus, a major

human pathogen. Therefore, there is an urgent need to developvaccines to target this pathogen. Two prophylactic vaccines havebeen tested recently for efficacy in humans: StaphVAX, whichcontained capsular polysaccharides type 5 and 8 (CP5 and CP8),and V710, based on a single protein antigen (IsdB) (1, 2). Bothvaccines failed in phase III efficacy trials (3, 4). On the basis ofthese disappointing results and taking into account that S. aureusproduces a plethora of virulence and immune evasion factors,different vaccine candidates, constituted by multiple compo-nents, are currently in phase I/II trials, but efficacy data are notavailable yet (5). In line with the multicomponent strategy, ourlaboratory has undertaken a vaccine discovery project aiming atthe identification of conserved antigens, which play importantroles in S. aureus virulence and pathogenicity. The main objec-tive of the study was to combine the selected antigens in thepresence of appropriate adjuvants and to demonstrate protectiveefficacy against a panel of genetically different S. aureus clinicalisolates in different mouse models.

ResultsAntigen Selection. The antigens included in our candidate combi-nation vaccine were selected among surface and secreted factorspreviously shown to be protective and involved in S. aureus viru-lence. Two of them, the ferric hydroxamate-binding lipoproteinFhuD2 and the putative lipoprotein named conserved staphylo-coccal antigen 1A (Csa1A), are surface-exposed antigens that wereidentified in our laboratories using MS-based surfome analyses andbioinformatics (6, 7). FhuD2 is a lipoprotein involved in iron uptakeand in early stages of invasive S. aureus infection (6, 8, 9). Csa1A ishighly conserved across different S. aureus isolates (Fig. 1A) andbelongs to a family of proteins encoded in at least four distinct locisharing from 54% to 91% sequence identity (7). The other threeselected antigens are secreted virulence factors and includeα-hemolysin (Hla), ess extracellular A (EsxA), and ess extracellularB (EsxB). Hla is one of the best characterized toxins of S. aureusand has been shown to play a prominent role in early stages of

Significance

Staphylococcus aureus is a human pathogen causing life-threatening infections. The high incidence of methicillin-resistant S. aureus isolates resistant to all antibiotics makes thedevelopment of anti-S. aureus vaccines an urgent medicalneed. However, the unique ability of S. aureus to producevirulent factors, which counteract virtually all pathways of in-nate and adaptive immunity, has hampered all vaccine dis-covery efforts. Starting from the assumption that to be effec-tive a vaccine should induce highly functional antibodies andpotentiate the killing capacity of phagocytic cells, we selecteda cocktail of five conserved antigens involved in differentmechanisms of pathogenesis, and we formulated them witha potent adjuvant. This vaccine provides an unprecedented pro-tective efficacy against S. aureus infection in animal models.

Author contributions: F.B., R.R., and G.G. designed research; F.B., R.P.N.M., L.F., S.N.,M.G.D.F., G.L., B.G., M.B., A.T., S. Maccari, C.D.R., N.I., S.R.-P., E. Swennen, R.P., C.B.,S.L., and E. Monaci performed research; F.B., M.R.F., E. Soldaini, P.R., M.S., S.S., E. Mori,B.B., M.P., N.N., J.B.W., O.S., G.B., S.B., E.D.G., R.R., and G.G. analyzed data; E.C., V.N.-D.,S. Marchi, P.M., D.T.O., and N.M.V. contributed new reagents/analytic tools; F.B., R.R.,and G.G. wrote the paper; and F.B., P.R., O.S., G.B., S.B., E.D.G., R.R., and G.G.study supervision.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. Email: rino.rappuoli@novartis.com orgrandiguido9@gmail.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424924112/-/DCSupplemental.

3680–3685 | PNAS | March 24, 2015 | vol. 112 | no. 12 www.pnas.org/cgi/doi/10.1073/pnas.1424924112

invasive and skin infections in animal models (10, 11). EsxA andEsxB are two factors secreted through the ESAT-6 secretion system(ESS) of S. aureus associated with abscess formation and may fa-cilitate persistence and spread of the pathogen in the infected host(12, 13). To be exploited as vaccine components, Hla, EsxA, andEsxB were modified as follows. Hla was detoxified by a histidine toleucine substitution at position 35 (HlaH35L), resulting in a mu-tated protein unable to form pores and lyse host cells (see Fig.4C) and to cause cell junction dissolution (14). EsxA and EsxBwere fused together, creating a recombinant 24-kDa EsxABchimera, which was stable and well expressed in Escherichia coliunlike the individual proteins. Therefore, we named the combi-nation vaccine 4C-Staph (four-component S. aureus vaccine).

Selected Vaccine Antigens Are Conserved and Expressed in Vivo. Thepresence and conservation of fhuD2, csa1A, hla, esxA, and esxBgenes in S. aureus were investigated by analyzing 60 genomesequences available in public databases, as well as by sequencingthe genes from 49 isolates of our internal strain collection(Tables S2 and S3). As reported in Fig. 1A, fhuD2, csa1A, andesxA are present in all strains and highly conserved (amino acididentity ranging from 83% to 99%). In general, hla is also wellconserved, but the sequence contains premature stop codons in11 strains belonging to clonal complex 30 (CC30) of the 26 usedin this study. Finally, esxB is missing in 37% of the analyzedstrains, although most of these strains belong to the same clonalcomplex (CC30). However, when present, EsxB protein is highlyconserved. Next, we investigated the expression of the antigensin six epidemiologically relevant S. aureus strains that were alsoused in mouse models of S. aureus infection (Fig. 1B and TableS2). Expression analysis was carried out by Western blot on cellextracts and culture supernatants. As shown in Fig. 1B, FhuD2and EsxA/EsxB were expressed in all six strains. However, be-cause EsxA and EsxB comigrated in the gel, we could not un-ambiguously confirm that the strains simultaneously expressedboth proteins. Furthermore, all strains but Mu50 expressed Hla.This result was unexpected because the sequence analysis of the

hla gene in the Mu50 strain did not reveal frameshifts or pre-mature stop codons. Finally, Csa1A was expressed in Newman,Los Angeles County clone (LAC), Mu50, and NRS216. InStaph19, in addition to the immune reactive band correspondingto Csa1A, a second band migrating at a slightly lower molecu-lar mass was present. A similar immune reactive protein wasdetected in MW2 where Csa1A was poorly or not expressedunder the conditions used. As already pointed out, Csa1Abelongs to a family of homologous proteins encoded by differentloci in S. aureus genome and the number of homologs varies inthe different strains. Therefore, it is not surprising that differentimmune reactive protein profiles are visible in the total extractsof different S. aureus isolates. To understand if the antigens in-cluded in the vaccine are expressed and immunogenic duringinvasive infection, we analyzed sera of mice infected with theS. aureus LAC strain for their reactivity against the purifiedvaccine proteins. Of the four vaccine components, only Hla hada detectable immunoreactivity in a dot-blot assay (Fig. 1C). Wehypothesized that the lack of reactivity against the other vaccinecomponents was due to the expression of the staphylococcalprotein A (SpA) by S. aureus during infection. Indeed, SpA isa B-cell superantigen: interaction with B-cell receptors can in-duce clonal expansion and subsequent apoptosis of B1 and MZ Bcells dampening the humoral response against staphylococcalantigens (15). Therefore, the elicitation of vaccine antigen-specific antibodies could be prevented by SpA through deletion ofantigen-specific B cells. To test this hypothesis, we decided to usea nontoxigenic form of SpA (SpAKKAA), developed previously byKim et al. (16), to immunize animals before their infection. Bydoing so, we inhibited the activity of SpA as previously demon-strated (16). As expected, sera of mice immunized with SpAKKAAand then infected with S. aureus recognized all of the vaccinecomponents (Fig. 1C).

Combination of HlaH35L, EsxAB, FhuD2, and Csa1A Induces ConsistentProtective Immunity Against Epidemiologically Relevant StaphylococcalStrains in Different Mouse Models. One of the criteria for selectingthe four antigens was their ability to induce partial protection inone or more mouse models of S. aureus infection. We then askedthe question of whether, when combined in a single formulation,the antigens could work synergistically and induce broad pro-tection. Mice were immunized with each antigen alone or withthe combination of the four antigens formulated with aluminumhydroxide (4C-Staph). After immunization, mice were chal-lenged i.v. with a sublethal dose of one of the five strains,Newman, LAC, Mu50, Staph19, and MW2, and 4 d later, bac-teria recovered from kidney homogenates were counted. Asshown in Table 1, immunization with 4C-Staph resulted in astatistically significant reduction of bacterial load regardless ofthe challenge strain used compared with control mice immunizedwith alum alone (Alum). Reduction of colony forming units(CFUs) varied from a minimum of 1.37 to a maximum of 2.44logs. Immunization with single vaccine components and IsdB didnot provide such a consistent protection, and it was generallyinferior to that provided by 4C-Staph (Table S4). IsdB, anS. aureus-conserved surface antigen shown to be protective inmice and tested in clinical trials (1, 2), was used as comparatorthroughout these experiments. The most effective single antigenin this model was FhuD2, which protected against four of fivestrains with a reduction in CFUs ranging from 0.94 to 2.13 logs.CFU counts measured following immunization with 4C-Staphand single antigens were closely paralleled by reduction of thearea and number of abscesses (Fig. S1 and Table S5). We alsocompared the performance of 4C-Staph with respect to singleantigens in the peritonitis infection model by challenging micewith a lethal dose of either Newman or MW2 strain. None ofthe single antigens significantly increased survival rate of micechallenged with the strain Newman (Table 2), and only HlaH35L,

Fig. 1. Gene conservation and expression of EsxA/B, Hla, FhuD2, and Csa1A.(A) Conservation of esxA/B, hla, fhuD2, and csa1A genes among represen-tative S. aureus strains. Conservation data of csa1A are limited to 60staphylococcal strains with genomes available in databases. (B) In vitro ex-pression of 4C-Staph antigens was assessed by Western blotting in the in-dicated strains. See Tables S1–S3 for further information about the strains.Sera raised against the vaccine antigens recognized the proteins expressedby S. aureus at the expected molecular weight. (C) S. aureus-infected micehave antibodies against the vaccine antigens. Sera from BALB/c mice (n = 15)immunized with SpAKKAA (closed bars) or with PBS/adjuvant control (openbars) and then challenged by i.v. inoculation with 5 × 106 CFU S. aureusUSA300 LAC were analyzed for antibody responses to staphylococcal anti-gens by dot-blot analysis. Data represent the mean ± SEM and are repre-sentative of three independent experiments.

Bagnoli et al. PNAS | March 24, 2015 | vol. 112 | no. 12 | 3681

APP

LIED

BIOLO

GICAL

SCIENCE

S

Csa1A, and FhuD2 significantly increased the median survivaltime of the mice (Table S6). On the other hand, vaccination withCsa1A, EsxAB, and FhuD2, but not with HlaH35L and IsdB,significantly increased the number of surviving mice challengedwith strain MW2 (Table 2). Csa1A and FhuD2 also significantlyincreased the median survival time of mice infected with thestrain MW2 (Table S6). Overall, 4C-Staph induced the greatestlevel of protection against both strains (Table 2 and Table S6).The protective efficacy of 4C-Staph was further investigatedwith three additional types of experiments. First, we expandedthe peritonitis protection data by challenging the vaccinatedmice with four additional strains (LAC, Staph19, Mu50, andNRS216), and we followed mouse survival over a period of 15 d.As shown in Fig. 2A, the 15-d protection elicited by 4C-Staphranged from 53% to 78% and was always significantly superior tothat observed in mock (for all strains) and IsdB-immunizedanimals (for strains Newman, LAC, MW2, and Staph19). IsdBimmunization was also tested in independent experimentsagainst strains NRS216 and Mu50 (Fig. S2). As observed againstthe other staphylococcal strains, IsdB immunization did notconfer significant protection against these latter two strains.Second, we tested the protective efficacy of 4C-Staph in thepneumonia model, challenging mice with five different strains:Newman, LAC, Staph19, Mu50, and NRS216. In this model,survival rate of mice immunized with 4C-Staph ranged between87% and 47%, with the exception of strain Mu50 (Fig. 2B).However, this was due to the low mortality rate associated withthis strain in the pneumonia model (23% mortality only in thecontrol group). For the same reason strain MW2 was not in-cluded in these experiments. This is consistent with the fact thatHla has been shown to play a predominant role in the pneu-monia model (11) and that its expression is low or nondetectablein strains MW2 and Mu50, respectively (Fig. 1B). IsdB neverconferred significant protection in this model. Finally, to furtherassess the protective efficacy of 4C-Staph against differentstaphylococcal disease outcomes, we used a skin infection model.In this model, mice were inoculated by s.c. injection in theshaved right flank with S. aureus LAC strain. Mouse skin andabscesses were harvested on day 4 after inoculation for CFUenumeration and histopathology. In a different set of animals,abscess mass and dermonecrotic area were monitored at 24-hintervals for 14 d. Immunization with 4C-Staph significantly re-duced abscess formation and CFU counts (Fig. 3 A and B).Furthermore, as shown in the gross histology pictures and theH&E-stained tissue sections, dermonecrosis was substantiallyabsent in the vaccinated mice (Fig. 3 C–F).

Protection Induced by 4C-Staph Was Largely Mediated by theElicitation of Functional Antibodies. To dissect the possible mech-anisms of protection induced by the vaccine, we focused ourattention on the analysis of functional antibodies. To this aim,rabbits were immunized with 4C-Staph, and subsequently therabbit serum was used in passive protection experiments. Inparticular, 150 μL of serum were administered i.v. to mice, and24 h later animals were challenged with S. aureus Newman strainaccording to the abscess and peritonitis models. As shown in Fig.4A, CFU counts in the kidneys of animals that received the hy-perimmune serum were significantly lower than in control ani-mals. Likewise, survival of passively immunized mice to the i.p.S. aureus challenge was significantly greater than that of the controlgroup (58% vs. 19%; Fig. 4B). On the other hand, rabbit serumagainst IsdB did not provide significant protection against nei-ther abscess formation nor in the peritonitis model (Fig. 4 A andB). We next investigated the mechanisms by which 4C-Staphantibodies protected mice against S. aureus challenge. First, wetested anti-Hla neutralizing activity of the rabbit serum used inthe passive protection experiment. To this end, rabbit eryth-rocytes were incubated for 30 min with 50 nM recombinant WTHla together with different concentrations of rabbit serum. Asshown in Fig. 4C, Hla-mediated hemolysis was inhibited in adose-dependent manner, with 0.5% serum concentration suffi-cient to mediate close to complete inhibition. No inhibition wasmediated by high concentrations (8%) of control serum. Fur-thermore, we analyzed the ability of the 4C-Staph mouse anti-serum to promote opsonophagocytosis of S. aureus Newman bydifferentiated HL60 cells in the presence of rabbit complement.As shown in Fig. 4D, the serum from animals immunized with4C-Staph mediated ∼40% killing of bacteria within 1 h of in-cubation. Preimmune serum and serum from mock-immunizedanimals did not promote bacterial killing. The same was truewhen 4C-Staph rabbit antiserum was tested in the absence ofphagocytes or active complement.

The Protective Efficacy of 4C-Staph Was Further Improved whenFormulated with a Novel Small Molecule Immune Potentiator. It islikely that an efficacious S. aureus vaccine would induce bothfunctional antibodies and CD4+ T cell-mediated immunityskewed toward a T helper (Th) type 1/Th17 response (4, 17–20).We therefore hypothesized that if 4C-Staph was formulated with

Table 1. Summary of 4C-Staph efficacy in reducing bacterialload in the renal abscess model

ImmunizationChallenge strain anddose (CFU/mouse) N

Mean logCFU ± SE

Log CFUreduction* P

Alum ST254 (Newman) 16 7.53 ± 0.304C-Staph 2.0 × 107 16 5.96 ± 0.36 1.57 0.002Alum USA300 (LAC) 20 7.01 ± 0.184C-Staph 3.8 × 107 19 4.57 ± 0.49 2.44 <0.0001Alum USA400 (MW2) 16 7.11 ± 0.244C-Staph 2.9 × 107 20 4.80 ± 0.50 2.31 0.001Alum USA100 (Mu50) 19 7.49 ± 0.174C-Staph 4.2 × 107 19 5.98 ± 0.39 1.51 0.003Alum ST80 (Staph19) 18 7.43 ± 0.194C-Staph 4.9 × 107 20 5.20 ± 0.33 2.23 <0.0001

Mice were immunized with 4C-Staph or with alum alone and thenchallenged as indicated (at least two separate experiments). One-tailedMann-Whitney u test. Values of P < 0.01 are highlighted in bold. Completeresults including single antigens and IsdB are reported in Table S2.*Log CFU reduction = mean log CFU alum ctrl − mean log CFU vaccinated.

Table 2. Summary of 4C-Staph efficacy in increasing survivalrates in the peritonitis model

ImmunizationChallenge strain

and dose NPercentsurvival

P vs.alum*

P vs.4C-Staph*

Alum Newman 32 22 0.001Csa1A 5.5 × 108 32 31 0.29 0.01EsxAB 32 38 0.14 0.04HlaH35L 32 31 0.29 0.01FhuD2 32 38 0.14 0.044C-Staph 32 63 0.001IsdB 32 38 0.14 0.04Alum MW2 60 30 0.0001Csa1A 8 × 108 60 60 0.0008 0.087EsxAB 60 48 0.03 0.004HlaH35L 60 43 0.09 0.0008FhuD2 60 52 0.01 0.014C-Staph 60 61 0.0001IsdB 59 44 0.08 0.001

Mice were immunized with the indicated antigens or with alum aloneand then challenged with the indicated strains (at least three separateexperiments).

*Fisher’s exact test. Values of P < 0.01 are highlighted in bold. Completeresults are reported in Table S6.

3682 | www.pnas.org/cgi/doi/10.1073/pnas.1424924112 Bagnoli et al.

an adjuvant stimulating this kind of immune response, its pro-tective properties might be further enhanced. In the course ofour internal adjuvant discovery program, an engineered smallmolecule immune potentiator (SMIP.7–10) targeting Toll-likereceptor 7 (TLR7) was generated (21). Such an immune po-tentiator has been chemically functionalized to be efficientlyadsorbed to aluminum hydroxide, and the resulting adjuvant wasnamed SMIP.7–10-Al(OH)3. We tested the vaccine combinationformulated with SMIP.7–10-Al(OH)3 (4CT7-Staph) in the peri-tonitis model, and protection was compared with that achievedby 4C-Staph (formulated with alum). Ten days after vaccination,mice were challenged with a lethal dose of Mu50, Newman,LAC, and MW2 strains, and survival was followed over a periodof 15 d. As shown in Fig. 5A, 4CT7-Staph was highly efficaciousagainst all strains, with protection levels ranging from 80% to90% and significantly higher than that obtained when the antigencombination was formulated in alum alone. Protection corre-lated with an increase in total IgG titers against all antigensexcept Hla (Fig. 5B). We next analyzed CD4 T-cell responsesagainst the combined antigens induced by 4CT7-Staph or4C-Staph immunization. To this aim, we performed two types ofexperiments. First, we looked at the isotype profile of antigen-

specific IgGs induced by 4C-Staph and 4CT7-Staph. As shown inFig. 5B, a significant increase in antigen-specific IgG2a was ob-served in animals vaccinated in the presence of SMIP.7–10. Anincrease was also observed for IgG2b and IgG1, albeit not forall antigens. Second, we analyzed antigen-specific CD4 T-cellresponses induced by the vaccine formulated with the twoadjuvants. For this purpose, we measured intracellular cytokinesproduced by CD4+CD44high T cells from spleens of mice im-munized with the two formulations in response to in vitro stim-ulation with the combined vaccine proteins or OVA as a negativecontrol. As shown in Fig. 5C, both formulations induced statis-tically higher frequencies of IL-2–, TNF-, and IFN-γ–producingCD4+CD44high T cells than those induced by the respective ad-juvant alone, whereas only 4C-Staph induced also IL-4/IL-13–producing cells. Greater frequencies of IL-17+ cells were observedin mice vaccinated with 4CT7-Staph, although differences werenot statistically significant.

DiscussionAmong all bacterial vaccines still missing, S. aureus is probablyone of the most difficult to develop for a number of reasons.First, there is no clear evidence of natural protective immunityagainst S. aureus, and there are no correlates of protectionestablished yet. Second, S. aureus expresses a plethora of toxinsand immune evasion factors (22, 23). Third, a large proportion ofat-risk population includes immune-compromised subjects (e.g.,HIV, cancer and hemodialysis patients, as well as the elderly).Taking into account these considerations, our strategy to developa S. aureus vaccine was (i) to select antigens with different rolesin pathogenesis, (ii) to increase the reliability and predictivevalue of this study by using different mouse models and func-tional assays, (iii) to include an adjuvant that could elicit highantibody titers and potentiate vaccine efficacy.The antigens of 4C-Staph were selected based on the hy-

pothesis that they could work synergistically by eliciting anti-bodies with different mechanisms of action. On one hand, we

Fig. 2. 4C-Staph generates greater protection against clinically relevantS. aureus strains than IsdB in the peritonitis and pneumonia models. In allplots, lines with triangles indicate 4C-Staph; lines with closed circles indicateIsdB; and lines with open circles indicate mice treated with aluminum hy-droxide alone. (A) In the peritonitis model, mice were challenged with thestrains Newman, LAC, MW2, Mu50, Staph-19, and NRS216, respectively (n =32–88 per group, at least two separate experiments). (B) In the pneumoniamodel, mice were challenged with the strains Newman, LAC, Staph-19, andNRS216, respectively (n = 30 per group, two separate experiments). Statis-tical analysis was performed by log-rank (Mantel-Cox) test.

Fig. 3. Reduction of abscess formation and dermonecrosis following im-munization with 4C-Staph. Active immunization with 4C-Staph decreasessize of abscesses (A) and CFU counts (B) associated with S. aureus LAC in-fection. Abscess formation was monitored once per day for 2 wk (A) or 4 dafter s.c. infection with 1 × 107 of LAC 10 d after secondary immunization(B). Results are the mean ± SEM and N = 10 mice per group. (C) Represen-tative mouse skin lesions (day 5). (D and E) Representative histological sec-tions showing mouse LAC abscess with dermonecrosis (at 3 d after infection)following immunization with either alum alone (D) or 4C-Staph (E). Mag-nification of images is ×200. (F) Dermonecrotic area monitored once per dayfor 2 wk. Results are the mean ± SEM and N = 10 mice per group. Data wereanalyzed using a one-way ANOVA and Dunnett’s test. *P < 0.05.

Bagnoli et al. PNAS | March 24, 2015 | vol. 112 | no. 12 | 3683

APP

LIED

BIOLO

GICAL

SCIENCE

S

showed that our vaccine induces antibodies that target threesecreted virulence factors: the pore-forming toxin Hla, known toplay key roles in pneumonia and skin infection (10, 11), andEsxA/EsxB, involved in abscess formation (12). In addition, be-cause of the presence of the surface-associated lipoproteinsFhuD2 and Csa1A, our vaccine elicits antibodies that are notonly opsonophagocytic but also interfere with important bi-ological functions. In the case of anti-FhuD2 antibodies, they canimpair iron uptake (6, 9), whereas as far as anti-Csa1A anti-bodies are concerned, we are accumulating evidence that theycould inhibit cell wall biosynthesis, a process that Csa1A is likelyto participate in. Our sequence analysis revealed that the fiveselected antigens are conserved in most S. aureus isolates. Inaddition of being conserved, the antigens were found expressedunder laboratory conditions and in experimentally infected mice.In vivo expression was deduced by detecting antigen-specificantibodies in mice infected with sublethal doses of S. aureus.Interestingly, as previously reported (16), these antibodies werefound only if infected mice had preexisting SpA antibodies,which could neutralize the capacity of SpA to induce B-cell ap-optosis. By analyzing sera from healthy adult volunteers, wefound that with the exception of Hla, antibodies against thevaccine antigens were present in few individuals (Fig. S3).Considering the high frequency with which humans are exposedto S. aureus, it is plausible to believe that the low immunogenicityof the vaccine antigens is not due to their poor expression but

rather to the same antibody inhibitory activity exerted by SpAin mice. This observation might have important implications forthe strategy to be used for vaccine candidate identification. Afrequently used approach to vaccine antigen selection is the analysisof sera from convalescent patients. Although this approach hasbeen successful for a number of pathogens, S. aureus mightrepresent an important exception. When formulated in alum, thevaccine antigens induced robust protection in four differentmouse models of infection. Importantly, protection was achievedwhen mice were challenged with six genetically different S. aureusstrains, isolated from human patients and belonging to important

Fig. 4. 4C-Staph generates functional antibodies. (A and B) Mice receivedi.v. sera of rabbits immunized with alum alone, 4C-Staph, or IsdB, and 24 hlater animals were infected i.v. (abscess model) or i.p. (peritonitis model)with S. aureus Newman. In the abscess model, the dashed line indicates thelower limit of CFU detection (n > 20 per group, at least two separateexperiments). Statistical analysis was performed by log-rank (Mantel-Cox)test for the peritonitis model and Mann–Whitney U test for the abscessmodel. (C) Rabbit antisera against the 4C-Staph vaccine neutralize Hla he-molytic activity. The black and gray columns represent hemolysis obtainedincubating rabbit erythrocytes with 50 nM Hla without rabbit serum or withserial dilutions of a serum from a rabbit immunized with 4C-Staph or Alumalone, as indicated in the figure. The white column corresponds to the he-molysis observed with 50 nM HlaH35L. Columns represent mean value ± SD offour independent rabbit sera. (D) Sera against 4C-Staph mediate S. aureusopsonophagocytosis. Mouse vaccine antisera, rabbit complement, HL-60,and the S. aureus strain Newman were incubated for 1 h and plated ontryptic soy agar for CFU counting. Percent killing was calculated as the ratiobetween the percent killing achieved with and without sera. No bacterialkilling was observed in the presence of sera from alum-treated mice (Alum)or using heat-inactivated complement (HI) and in absence of HL-60 cells [(−)HL-60]. Error bars represent SD. Statistical analysis was performed by pairedt test.

Fig. 5. Protective efficacy, isotype profile of vaccine-specific IgGs, and CD4+

T-cell responses induced by 4C-Staph and 4CT7-Staph. (A) Protective efficacyof 4C-Staph and 4CT7-Staph in the peritonitis model against S. aureus strainsMu50, MW2, LAC, and Newman. In all graphs, white columns indicate theadjuvants alone [SMIP.7–10-Al(OH)3 = SMIP-Alum]. Gray and black columns,one of the two vaccine formulations as indicated (n = 32–64 per group, atleast two separate experiments). (B) Antigen-specific IgGs in mice immu-nized with 4C-Staph or 4CT7-Staph. (C) CD4+CD44high T-cell responses inimmunized mice. Graphs show results from mice immunized with adjuvantalone (white bars, Alum or SMIP-Alum) or with vaccine antigens formulatedwith the two adjuvants (black bars, 4C-Staph or 4CT7-Staph). Splenocytescollected from each mouse and stimulated in vitro with the vaccine antigensor OVA, used as a control. The graphs show the frequencies of CD3+CD4+

CD44high T cells expressing IL-2, TNF, IFN-γ, IL-4, and/or IL-13, and IL-17A.Data are expressed as mean ± SE from individual mice (n = 7) and are rep-resentative of two independent experiments. Statistical significance of miceimmunized with vaccine + adjuvant vs. adjuvant only (***P ≤ 0.001).

3684 | www.pnas.org/cgi/doi/10.1073/pnas.1424924112 Bagnoli et al.

epidemic lineages. None of the single antigens performed simi-larly well, suggesting that each antigen contributed at least tosome extent to protection induced by 4C-Staph. Immunizationwith IsdB, which was used as comparator throughout this study,elicited nonconsistent protection against the S. aureus strainstested in the kidney abscess model. In addition, in the peritonitisand pneumonia models, IsdB did not confer significant pro-tection and was significantly inferior to the 4C-Staph. Protectiveimmunity was largely mediated by antibodies, as indicated by thefact that similar levels of protection were obtained when micewere passively immunized with sera from rabbits vaccinated with4C-Staph. Although antibodies are expected to have an impor-tant role in preventing S. aureus human infections, the need toefficiently kill the pathogen once it gets internalized intophagocytic cells suggests that the addition of a Th1/Th17 adju-vant could be beneficial in eliciting optimal anti-S. aureus im-mune responses. Indeed, a Th1/Th17 polarized immune responsehas been demonstrated to potentiate the opsonophagocytic ac-tivity of neutrophils and macrophages (17–19). We tested thishypothesis by formulating the vaccine antigens with SMIP.7–10-Al(OH)3, which is a small molecule TLR7 agonist chemically func-tionalized to allow stable adsorption to aluminum hydroxide. Thisfeature prevents side effects due to systemic exposure of the SMIPand allows the codelivery of the antigen-SMIP complex to antigenpresenting cells, a prerequisite for optimal activation of antigen-specific B and T cells (21). 4CT7-Staph was found highly pro-tective in the mouse models used, with almost 100% of micesurviving lethal doses of different S. aureus isolates. The presenceof the SMIP not only enhanced the antibody titers against vaccineantigens but also skewed the immune response toward a Th1/Th17profile, as judged by antibody isotype and cytokine profiles.

Therefore, the increased efficacy associated with SMIP.7–10-Al(OH)3 compared with the alum formulation may likely be due toits combined effects on antibody and T-cell responses.Two prophylactic vaccine trials have recently failed, and in

both cases, the vaccines did not include adjuvants. In particular,the most recent phase III trial (3) was carried out with a single doseof IsdB with no adjuvants. Our data seem to indicate that part of thefailure might be due to an inefficient stimulation of cell-mediatedimmunity and prompt to test the use of new vaccines able to acti-vate this arm of the immune system, which appears to be particu-larly important to combat S. aureus infections.

Materials and MethodsVaccine antigens were cloned from the S. aureus NCTC8325 strain, and dif-ferent purification protocols were used for each protein. Highly purifiedantigens formulated with aluminum hydroxide or adsorbed to SMIP.7–10-Al(OH)3 were used to immunize mice before infection with S. aureusstrains. 4C-Staph antisera generated in mice or rabbits were used to per-form opsonophagocytosis and Hla assays, as well as passive transferexperiments. Splenocytes isolated 10 d after the second immunization ofmice were stimulated with HlaH35L, EsxAB, FhuD2, or Csa1A in combinationin the presence of Brefeldin A. Experimental methods are described indetail in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank our colleagues A. Covacci, F. Berti, G. DelGiudice, P. Costantino, M. Singh, U. D’Oro, J. Telford, M. Nissum, M. Giuliani,F. Legay, M. Soriani, D. Maione, M. Barocchi, D. Serruto, C. Gianfaldoni,M. Giraldi, D. Laera, P. Boucher, S. Nuti, S. Tavarini, C. Sammicheli, F. Mancini,C. Facciotti, F. Falugi, M. Tortoli, G. Corsi, T. Wu, and M. Cooke for scientificand technical support to the study. In addition, we thank A. DeDent andH. K. Kim at the University of Chicago; G. Mancuso, G. Teti, and A. Midiri atthe University of Messina; and G. Rossi at the University of Camerino forscientific support to the study.

1. Kim HK, et al. (2010) IsdA and IsdB antibodies protect mice against Staphylococcusaureus abscess formation and lethal challenge. Vaccine 28(38):6382–6392.

2. Kuklin NA, et al. (2006) A novel Staphylococcus aureus vaccine: Iron surface de-terminant B induces rapid antibody responses in rhesus macaques and specific in-creased survival in a murine S. aureus sepsis model. Infect Immun 74(4):2215–2223.

3. Fowler VG, et al. (2013) Effect of an investigational vaccine for preventing Staphy-lococcus aureus infections after cardiothoracic surgery: A randomized trial. JAMA309(13):1368–1378.

4. Bagnoli F, Bertholet S, Grandi G (2012) Inferring reasons for the failure of Staphylo-coccus aureus vaccines in clinical trials. Front Cell Infect Microbiol 2:16.

5. Proctor RA (2012) Challenges for a universal Staphylococcus aureus vaccine. Clin InfectDis 54(8):1179–1186.

6. Mishra RP, et al. (2012) Staphylococcus aureus FhuD2 is involved in the early phase ofstaphylococcal dissemination and generates protective immunity in mice. J Infect Dis206(7):1041–1049.

7. Schluepen C, et al. (2013) Mining the bacterial unknown proteome: Identification andcharacterization of a novel family of highly conserved protective antigens in Staphy-lococcus aureus. Biochem J 455(3):273–284.

8. Mariotti P, et al. (2013) Structural and functional characterization of the Staphylo-coccus aureus virulence factor and vaccine candidate FhuD2. Biochem J 449(3):683–693.

9. Sebulsky MT, Heinrichs DE (2001) Identification and characterization of fhuD1 andfhuD2, two genes involved in iron-hydroxamate uptake in Staphylococcus aureus.J Bacteriol 183(17):4994–5000.

10. Kennedy AD, et al. (2010) Targeting of alpha-hemolysin by active or passive immu-nization decreases severity of USA300 skin infection in a mouse model. J Infect Dis202(7):1050–1058.

11. Bubeck Wardenburg J, Schneewind O (2008) Vaccine protection against Staphylo-coccus aureus pneumonia. J Exp Med 205(2):287–294.

12. Burts ML, Williams WA, DeBord K, Missiakas DM (2005) EsxA and EsxB are secreted byan ESAT-6-like system that is required for the pathogenesis of Staphylococcus aureusinfections. Proc Natl Acad Sci USA 102(4):1169–1174.

13. Korea CG, et al. (2014) Staphylococcal Esx proteins modulate apoptosis and release ofintracellular Staphylococcus aureus during infection in epithelial cells. Infect Immun82(10):4144–4153.

14. Wilke GA, Bubeck Wardenburg J (2010) Role of a disintegrin and metalloprotease 10in Staphylococcus aureus alpha-hemolysin-mediated cellular injury. Proc Natl Acad SciUSA 107(30):13473–13478.

15. Silverman GJ, Goodyear CS (2006) Confounding B-cell defences: Lessons froma staphylococcal superantigen. Nat Rev Immunol 6(6):465–475.

16. Kim HK, Cheng AG, Kim HY, Missiakas DM, Schneewind O (2010) Nontoxigenic pro-tein A vaccine for methicillin-resistant Staphylococcus aureus infections in mice. J ExpMed 207(9):1863–1870.

17. Stockinger B, Veldhoen M (2007) Differentiation and function of Th17 T cells. CurrOpin Immunol 19(3):281–286.

18. Ye P, et al. (2001) Requirement of interleukin 17 receptor signaling for lung CXCchemokine and granulocyte colony-stimulating factor expression, neutrophil re-cruitment, and host defense. J Exp Med 194(4):519–527.

19. Ellis TN, Beaman BL (2004) Interferon-gamma activation of polymorphonuclearneutrophil function. Immunology 112(1):2–12.

20. Miller LS, Cho JS (2011) Immunity against Staphylococcus aureus cutaneous infections.Nat Rev Immunol 11(8):505–518.

21. Wu TY, et al. (2014) Rational design of small molecules as vaccine adjuvants. Sci TranslMed 6(263):263ra160.

22. Foster TJ, Geoghegan JA, Ganesh VK, Höök M (2014) Adhesion, invasion and evasion:The many functions of the surface proteins of Staphylococcus aureus. Nat RevMicrobiol 12(1):49–62.

23. Foster TJ (2005) Immune evasion by staphylococci. Nat Rev Microbiol 3(12):948–958.

Bagnoli et al. PNAS | March 24, 2015 | vol. 112 | no. 12 | 3685

APP

LIED

BIOLO

GICAL

SCIENCE

S

Supporting InformationBagnoli et al. 10.1073/pnas.1424924112SI Materials and MethodsBacterial Strains, Media, and Growth Conditions. S. aureus clinicalisolates were collected from patients with infective endocarditis(IE) by University of Pavia (Italy) (kindly provided by P. Spezialeand S. Rindi, Dipartimento di Medicina Molecolare, Universitàdi Pavia, Pavia, Italy) and with necrotizing pneumonia by IstitutoSuperiore di Sanità (Rome, Italy). The strain collection includedalso representative hospital-acquired and community-acquiredmethicillin-resistant clones (HA-MRSA and CA-MRSA) re-ceived by Geneva University Hospital (Switzerland) and Uni-versity of Chicago, as well as laboratory strains routinely used inanimal studies. The strains used in this study are listed in TablesS2–S3. S. aureus strains were grown at 37 °C in Tryptic soy broth(TSB; Difco Laboratories) or in trypticase soy agar. For prepa-ration of bacterial challenge inocula for infection studies in an-imals, S. aureus strains were grown at 37 °C in TSB and grown toan OD660 0.5–2. For each strain, bacterial pellets from liquidcultures were collected by centrifugation, washed once with PBS,and resuspended in PBS, providing the desired amount of CFUdepending on the model.For expression of recombinant proteins, E. coli was grown in

Luria Bertani broth containing 100 μg/mL ampicillin (His-taggedproteins) or 30 μg/mL kanamycin (tagless proteins) up to OD = 0.4and then induced with 1 mM isopropyl-β-D1-thiogalactopyranosideand grown for 3 h at 25 °C.

Vaccine Antigen Gene Analysis. Genomic DNA was isolated by astandard protocol for gram-positive bacteria by lysostaphintreatment of bacterial cells using a Gene Elute Bacterial GenomicDNA kit (Sigma-Aldrich) according to the manufacturer’s in-structions. Genes were amplified using the primers, listed inTable S1, which were selected on the basis of the available ge-nome sequences and specifically annealed to conserved genomicregions external to the coding sequences. PCR products werepurified on a Biomek FX liquid handling robotic workstationusing AMPure magnetic beads (Agencourt) and sequenced inboth directions using the BigDye Terminator v3.1 Cycle Se-quencing kit (Applied Biosystems) on a 3730xl DNA Analyzer(Applied Biosystems). Sequence identity was measured by pair-wise BLAST with Vector NTI Suite 11 (Invitrogen), with gapsincluded. Sequence alignments were performed using ClustalW(1.83, GCG Wisconsin Package, version 11.1).

Multilocus Sequence Typing. Multilocus sequence typing (MLST)was performed by sequencing the internal fragments of the arc,aro, glp, gmk, pta, tpi, and yqi genes. The seven housekeepinggenes were PCR amplified by genomic DNA using the primerpairs indicated at saureus.mlst.net/misc/info.asp#experimental.Alleles from theMLSTWeb site (saureus.mlst.net) were downloadedfor alignment analyses and sequence type (ST) determination.

Vaccine Antigen Expression Analysis. The expression of vaccineantigens was evaluated in vitro by Western blot analysis. S. aureuscultures at exponential growth phase were centrifuged at 1,940 × grpm for 20 min. Proteins in 1.5 mL of supernatants were pre-cipitated with 15% (wt/vol) trichloroacetic acid (TCA) overnightat 4 °C and then pelleted by centrifugation. All TCA precipitateswere washed with ice-cold acetone, solubilized in 2% (wt/vol)SDS, and used for detecting the expression of Hla. Bacterialpellets were washed once with PBS, suspended in Tris·HCl(50 mM, pH 6.8), containing complete protease inhibitors (Ro-che) and 100 μg/mL lysostaphin (Sigma-Aldrich), and incubated

for 1 h at 37 °C. After three cycles of freeze and thaw, solubleproteins were separated from insoluble materials and cellulardebris by centrifugation and used for assessing the expression ofEsxA/EsxB and FhuD2 antigens. The remaining pellet (mem-brane-enriched fraction) was solubilized using 2% (wt/vol) SDSbuffer and used for Csa1A protein detection. Equal amountsof bacterial proteins from each strain were separated on SDS/PAGE and transferred to nitrocellulose membranes using iBlottransfer (Dry blot system; Invitrogen). The membranes were probedwith rabbit antisera followed by an anti-rabbit HRP-conjugatedsecondary antibody (Dako). Immunoreactive signals were finallyrevealed by chemiluminescence using SuperSignalWest Picochemiluminescent substrate (Pierce).The presence of specific antibodies against vaccine antigens in

human sera collected from healthy people was assessed by im-munoblot analysis using purified recombinant proteins expressedin, and purified from, E. coli. Proteins (100 ng) were resolved bySDS/PAGE and transferred to nitrocellulose membranes thatwere probed with human sera (1:2,500 dilutions). Immunoreactivebands were detected by chemiluminescence as described above.

Cloning and Purification of Vaccine Antigens.Vaccine antigens wereamplified by PCR from the S. aureus NCTC8325 strain andcloned as N-terminal 6-histidine-tagged (His-tag) or taglessconstructs. The his-tagged PCR products were cloned into thepET-15b+ vector using the polymerase incomplete primer ex-tension (PIPE) technique (1). The tagless PCR products werecloned into the pET-24b+ vector. Both kinds of constructs weretransformed in E. coli BL21 (DE3). His-tag proteins were puri-fied by immobilized metal ion affinity chromatography purifi-cation using a nickel-activated chelating Sepharose column (GEHealthcare His-trap HP). Different purification protocols wereused for each tagless protein as described below.FhuD2-expressing cells were then harvested by centrifugation

and suspended in 50 mM sodium phosphate buffer (pH 7.2),followed by mechanical disruption. After centrifugation at12,000 × g for 30 min at 4 °C, the E. coli extract supernatantswere filtered through a 0.22-μm membrane. The soluble cellextract was diluted with bidistilled water to adjust conductivity to1.8–1.9 mS/cm and was loaded onto an SP Sepharose FF HiTrapresin (5-mL column; GE Healthcare). The bound protein waseluted with a gradient from 0 to 1 M NaCl in 50 mM sodiumphosphate buffer (pH 7.0), with a flow rate of 2 mL/min. Frac-tions containing FhuD2 were identified by SDS/PAGE [12%(wt/vol) polyacrylamide gel] analysis and were loaded on toa hydroxyapatite ceramic type I 40-μm column (Bio-Rad Labo-ratories) equilibrated in 10 mM sodium phosphate (pH 7.0). Thebound protein was eluted with a 15-CV (column volumes) sodiumphosphate gradient from 10 mM to 1 M (pH 7.0) at a flow of2.5 mL/min. The pooled fractions containing FhuD2 were dialyzedin PBS, filtered through a 0.22-μm membrane, and analyzed forpurity and integrity by SDS/PAGE, size exclusion (SE)-HPLC,and reverse phase (RP)-HPLC.Csa1A-expressing cells were harvested by centrifugation and

suspended in 10 mM Tris (pH 8.5) followed by mechanical dis-ruption, addition of 5% (wt/vol) polyethyleneimine (PEI) in50 mM Tris, pH 7.5, to a final concentration of 0.25% and 3 h ofmagnetic stirring at room temperature. After centrifugation at12,000 × g for 30 min at 4 °C, 0.6 volumes of 100 mM Tris,pH 9.0, were added to the E. coli extract and filtered througha 0.22-μm membrane. The soluble extract was then loaded on toCapto Q anion exchange resin (GE Healthcare) equilibrated in

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 1 of 10

10 mM Tris, pH 8.5. Bound protein was eluted with a gradient0–250 mM NaCl in equilibration buffer. Fractions containingCsa1A were identified by SDS/PAGE [12% (wt/vol) poly-acrylamide gel] analysis and were loaded onto a hydroxyapatiteceramic type I 40 μm column (Bio-Rad Laboratories) equili-brated in 10 mM sodium phosphate (pH 7.2); protein was elutedwith a 20-CV sodium phosphate gradient from 10 to 600 mM(pH 7.2). The pooled fractions containing Csa1A were dialyzedin PBS, filtered through a 0.22-μm membrane, and analyzed forpurity and integrity by SDS PAGE, SE-HPLC, and RP-HPLC.EsxAB-expressing cells were harvested by centrifugation and

suspended in 30 mM Tris (pH 8.0) followed by mechanical dis-ruption. After centrifugation at 16,000 × g for 60 min at 4 °C, theE. coli extract supernatants were filtered through a 0.22-μmmembrane. The soluble cell extract was loaded on to CaptoQ Sepharose (GE Healthcare) equilibrated in 30 mM Tris, pH8.0. Bound protein was eluted with a gradient from 0 to 1 MNaCl in equilibration buffer. Fractions containing EsxAB wereidentified by SDS/PAGE [12% (wt/vol) polyacrylamide gel]analysis, were salted to 3 M NaCl, and loaded on to a PhenylSepharose 6 fast flow HS (GE Healthcare) previously equili-brated in 15 mM Tris·HCl (pH 8.0) and 1 M NaCl; bound pro-tein was eluted with a gradient from 15 mM Tris·HCl (pH 8.0)and 1 M NaCl to 10 mM sodium phosphate, pH 7.0. Fractionscontaining EsxAB were identified by SDS/PAGE [12% (wt/vol)polyacrylamide gel] analysis and were loaded onto a hydroxyap-atite ceramic type I 40 μm column (Bio-Rad Laboratories) equili-brated in 10 mM sodium phosphate (pH 7.0); protein bound to thecolumn was washed with 40 mM sodium phosphate (pH 7.2) andeluted with a 40- to 250-mM gradient of sodium phosphate (pH7.2). The pooled fractions containing EsxAB were dialyzed in PBS,filtered through a 0.22-μm membrane, and analyzed for purity andintegrity by SDS PAGE, SE-HPLC, and RP-HPLC.HlaH35L-expressing cells were harvested by centrifugation

and suspended in 20 mM Tris, pH 7.0, followed by mechanicaldisruption. After centrifugation at 16,000 × g for 60 min at 4 °C, theE. coli extract supernatant was filtered through a 0.22-μm mem-brane and salted to 1.0 M (NH4)2SO4. The soluble cell extract wasloaded onto phenyl sepharose 6 Fast Flow (FF) High Sub (HS)(GE Healthcare) equilibrated in 10 mM Tris and 1 M (NH4)2SO4,pH 7.0. Bound proteins were washed with equilibration buffer andeluted with a gradient of 100% equilibration buffer → 100 H2O.Fractions containing HlaH35L were identified by SDS/PAGE[12% (wt/vol) polyacrylamide gel] analysis, dialyzed overnightagainst 10 mM NaH2PO4, pH 7.0, and subsequently filtered througha 0.22-μm membrane. The protein solution was then loaded ontoa hydroxyapatite ceramic type I 40-μm HS (Bio-Rad) previouslyequilibrated in 10 mM NaH2PO4, pH 7.0; bound proteins werewashed with equilibration buffer and eluted with a gradient from 10to 500 mM NaH2PO4, pH 7.0. Fractions containing HlaH35Lwere identified by SDS/PAGE [12% (wt/vol) polyacrylamidegel] analysis and were loaded onto a hydroxyapatite ceramic type I40-μm column (Bio-Rad Laboratories) equilibrated in 10 mMsodium phosphate (pH 7.0); protein bound to the column waswashed with equilibration buffer and eluted with a gradient from10 to 500 mM NaH2PO4, pH 7.0. The pooled fractions containingHlaH35L were concentrated on an Amicon Cell 10-kDa mem-brane to one-sixth of the starting volume and loaded onto a Su-perdex 75 size exclusion column (GE Healthcare), with PBS asthe mobile phase. Fractions containing HlaH35L were identifiedby SDS/PAGE analysis [12% (wt/vol) polyacrylamide gel], col-lected, filtered through a 0.22-μm membrane, and analyzed forpurity and integrity by SDS/PAGE, SE-HPLC, and RP-HPLC.IsdB-expressing cells were harvested by centrifugation and

suspended in 50 mM Na phosphate and 300 mM NaCl, pH 8.0,followed bymechanical disruption. After centrifugation at 18,000× gfor 30 min at 4 °C, the E. coli extract supernatant was filteredthrough a 0.22-μm membrane and loaded onto chelating Sepharose

FF (GE Healthcare) equilibrated in 50 mM sodium phosphate and300 mM NaCl, pH 8.0. Bound proteins were washed with equil-ibration buffer and eluted with a gradient 0–250 mM imid-azole in equilibration buffer. Fractions containing IsdB wereidentified by SDS/PAGE analysis [10% (wt/vol) polyacrylamidegel] and dialyzed overnight against 50 mM NaH2PO4, pH 6.5.Precipitated material was removed by centrifugation for 30 min at18,000 × g, and the supernatant was filtered through a 0.22-μmmembrane. The protein solution was then loaded onto an SPSepharose FF (GE Healthcare) previously equilibrated in 50 mMNaH2PO4, pH 6.5; bound proteins were washed with equilibrationbuffer and eluted with a gradient of 0–1 M NaCl in 20 CVequilibration buffer. Fractions containing IsdB were identified bySDS/PAGE analysis [10% (wt/vol) polyacrylamide gel]. Thepooled fractions containing IsdB were concentrated on AmiconCell 10-kDa membranes to one-fifth of the starting volume andloaded onto a Superdex 200 size exclusion column (GE Health-care), with PBS as the mobile phase. Fractions were collected, andthose containing the protein of interest were pooled, filteredthrough a 0.22-μm membrane, and analyzed for purity and in-tegrity by SDS/PAGE, SE-HPLC, and RP-HPLC.

Ethics Statement. Mice and rabbits were monitored twice per dayto evaluate early signs of pain and distress such as respiration rate,posture, and loss of weight (more than 20%) according to humaneend points defined for each model. Animals showing such con-ditions were euthanized in accordance with experimental pro-tocols, which were reviewed and approved by the Novartis AnimalWelfare Body and the Italian Ministry of Health (Protocols 136/2010-B for mouse studies and 201103 for rabbit studies) or by theInstitutional Biosafety Committee (IBC) and Institutional Ani-mal Care and Use Committee (IACUC) at The University ofChicago. The University of Chicago Animal Resource Center isaccredited by the American Association for Accreditation ofLaboratory Animal Care and the Department of Health andHuman Services (DHHS No. A3523-01).

Active Immunization. Four- or 5-wk-old mice (CD1, BALB/c, orC57BL/6) were immunized i.p. or i.m. with prime-booster injectionswith a 14-d interval. Vaccine formulations of 200 (i.p.) or 100 μL(i.m.) containing either 20 μg of each purified protein (when micewere immunized with single antigens) or 10 μg (when mice wereimmunized with the combined vaccine) adsorbed to aluminum hy-droxide adjuvant (alum, 2 mg/mL) were used to immunize animals.For formulations containing the TLR7 agonist, protein anti-

gens were adsorbed to SMIP.7–10, ensuring a final SMIP dose of50 μg per animal, thus corresponding to a SMIP concentration ofeither 500 μg/mL for i.m. or 250 μg/mL for i.p. immunizations.All formulations were adjusted by the addition of 10 mM histi-dine buffer and 9 g/L NaCl to the recommended physiologicalranges of pH (6.5 ± 0.5) and isotonicity (osmolality, 300 ± 60mOsm/kg). Control mice received equal amounts of PBS and alumadjuvant. Serum was collected from mice both before and aftervaccination to document serum antibody titers to each proteincomponent in the combination vaccine.Control mice received equal amounts of PBS and alum adju-

vant. Serum was collected from mice both before and aftervaccination to document serum antibody titers to each proteincomponent in the combination vaccine.

Antibody Quantification. Antibody titers present in sera of im-munized mice were measured by Luminex technology (Luminex200 TM). HlaH35L, EsxAB, FhuD2, Csa1A, and IsdB purifiedproteins were covalently conjugated to the free carboxyl groupsof microspheres using an N-hydroxysulfosuccinimide–enhancedcarbodiimide-mediated conjugation chemistry. Antigen-specificantibodies were revealed by phycoerythrin-labeled secondary

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 2 of 10

antibodies. The assay readout is a measure of fluorescence in-tensity at fixed serum dilution.Isotype characterization based on Luminex technology was

used to classify IgG1, IgG2a, and IgG2b in mouse sera specific forthe antigens. Biotin-SP–conjugated detection antibody was addedto each well and incubated for 30 min. Plates were washed threetimes with PBS, and streptavidin PE-conjugated was added to eachwell. Plates were incubated for a further 15 min, washed threetimes with PBS, and then resuspended in 200 μL of PBS. At thispoint, plates were ready for analysis through the Luminex system.Mice that had been immunized with SpAKKAA or received

alum alone and then were infected with S. aureus LAC wereexamined by quantitative immunoblot. Nitrocellulose membranewas blotted with 2 μg of purified vaccine antigens. IRDye700DX-conjugated anti-mouse IgG (Rockland Immunochemicals)was used to quantify signal intensities from mice using the Odysseyinfrared imaging system (LI-COR Biosciences). Signal intensities inmouse sera were quantified and normalized using anti-His6 anti-body with the Odyssey.

Murine Pneumonia Model. On documentation of appropriate an-esthesia with ketamine (100 mg/kg) and xylazine (5 mg/kg),C57BL/6 mice were challenged intranasally with S. aureusNewman, LAC, NRS216, Mu50, or Staph 19. For each strain,bacterial pellets from 50-mL culture aliquots were collectedby centrifugation, washed once with PBS, and resuspended in750 μL of PBS providing 2–3 × 108 CFU per 30-μL volume ofsuspension. For Mu50 studies, inocula were increased to 6–8 × 108

CFUs. A 30-μL volume was administered intranasally (15 μL pernares) to each mouse as previously described. Following bacte-rial challenge, disease severity was monitored at 8- to 12-h in-tervals, with morbidity determined based on signs specified bythe University of Chicago IACUC. Survival curves were determinedvia the Kaplan–Meier method, and statistical significance was as-sessed with the log-rank test using GraphPad Prism software.

Murine Skin Infection Model. On documentation of appropriateanesthesia with ketamine (100 mg/kg) and xylazine (5 mg/kg),BALB/c mice were challenged s.c. with 1–3 × 107 CFU S. aureusLAC in 50 μL PBS, essentially as previously described (2–5).Bacterial inocula were prepared as described for intranasal in-fection and adjusted to the appropriate CFUs accordingly. Massand abscess formation (size and dermonecrosis) were monitoredat 24-h intervals over a course of 14 d. The size of an abscess andassociated overlying dermonecrotic lesion was determined usinga standard formula for area [A = (π/2) × l × w]. For histologicalanalysis of mouse skin, tissue was harvested on day 4 after in-oculation and fixed in 10% (vol/vol) neutral-buffered formalinfor 48 h. Paraffin-embedded tissue blocks were sectioned at 5 μmand stained with H&E to facilitate visualization. Analysis ofbacterial burden in skin abscess lesions was determined 4 d afterinfection. Following CO2 inhalation for euthanasia, skin ab-scesses were excised and homogenized in 0.01% Triton X-100.Tissue homogenates were serially diluted and plated, and theresulting CFUs were enumerated following overnight growth at37 °C on tryptic soy agar. Aggregate analysis of abscess size anddermonecrosis for statistical significance was performed withGraphPad Prism software using the unpaired two-tailed Studentt test at each time point over the 14-d observation period. Re-sults are expressed as mean ± SEM.

Passive Immunization. New Zealand rabbits were immunized byintradermal injection with the 4C-Staph vaccine or IsdB (50 μgof each purified protein) adsorbed to aluminum hydroxide(2 mg/mL) on days 0, 21, and 35. Rabbit immune sera (150 μL)were injected into the tail vein of 8-wk-old CD1 mice 24 h beforechallenge with S. aureus infection according to the renal abscessor peritonitis models. Control mice were injected with the same

volume of sera from rabbits that had been immunized with PBSand aluminum hydroxide.

Renal Abscess Model. Experiments using the renal abscess modelwere performed as previously described (6, 7). Briefly, immu-nized CD1 mice were challenged on day 24 (10 d after secondimmunization) by i.v. injection of a sublethal dose of S. aureus(∼2–6 × 107 CFUs, specific inoculum varied depending on thechallenge strain). On day 28, mice were euthanized, and kidneyswere removed, homogenized in 2 mL PBS, and plated on agarmedia in duplicate for determination of CFUs. Kidneys werealso processed for histopathology as described below.

Peritonitis Model. Experiments using the peritonitis model wereperformed as previously described (6, 8, 9). Immunized CD1mice were challenged on day 24 (10 d after second immuni-zation) by i.p. injection of S. aureus. Mice were infected with∼2–8 × 108 CFUs of S. aureus (specific inoculum varied de-pending on the challenge strain) and monitored daily for 14 d.

Histopathology.Mouse kidneys were cut in two equal halves alongthe sagittal plane and then fixed in 4% (wt/vol) buffered form-aldehyde and processed for paraffin embedding. Four-micrometersections were cut from each kidney and stained with H&E).Other sections were stained with periodic acid–Schiff (PAS),Sirius red (SR), and acid fuchsin–orange G (AFOG) for betterevaluation of tissue damage. Before AFOG staining, deparaffi-nized sections were postfixed in Bouin’s fixative. Images of eachstained kidney section were acquired by a Mirax Scan 150 slidescanner equipped with a 40× lens. Two sections for each kidneywere examined, and the one with the most abundant abscesses wasused for quantification. Abscesses were enumerated, and the totalarea of abscessualization, including suppurative infiltrate areas, wasquantified in each kidney using the Mirax viewer software. Thearea of abscesses for a given section was calculated as the ratio ofthe area covered by the abscesses over the total area of the sec-tion. The sections were examined in blind by an experienced an-imal pathologist, and nephritis and nephrosis were scored asfollows: 0 = normal; 1 = mild; 2 = moderate; 3 = severe.

Opsonophagocytosis Killing Assay. The opsonophagocytosis killingassay (OPKA) assay was performed as previously described(6, 10). Briefly, human promyelocytic leukemia cells HL-60 (ATCCCCL240) were maintained in enriched medium and differ-entiated into phagocytes using 0.8% N,N-dimethylformamide(DMF; Sigma). Following heat inactivation (30 min, 56 °C),mouse antiserum against 4C-Staph was diluted 1:50 in HBSSbuffer (with Ca2+/Mg2+). S. aureus Newman grown overnight inTSB were washed once in PBS and then incubated with sera(75,000 CFUs/well) at 4 °C for 20 min. Differentiated HL-60cells were distributed at 3.7 × 106/well (HL-60:bacteria ratio,50:1), and rabbit complement was added at a 10% (vol/vol) finalconcentration. Plates were then incubated at 37 °C for 1 h, underagitation at 600 rpm, and samples were plated onto TSA platesfor CFU count determination.

Hla Neutralization Assay. Ability of 4C-Staph antibodies to inhibitHla-induced hemolysis was evaluated in an in vitro assay. Briefly,serial twofold dilutions of antisera against the vaccine, or antiseraagainst Alum adjuvant alone, were incubated with 50 nM Hla for30 min at 37 °C and agitated at 350 rpm. Then, erythrocytesderived from defibrinated rabbit blood were added, and in-cubation was prolonged for a further 30 min at 37 °C. Incubationwith water + 1% Triton X-100 was used as a positive control(maximal hemolysis). Plates were then centrifuged for 5 min at1,000 × g, and the supernatant was analyzed spectrophotomet-rically by a SpectraMax 340PC384 absorbance microplate readerat 540 nm.

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 3 of 10

Intracellular Flow Cytometry. Splenocytes were isolated 10 d afterthe second immunization, plated at 2 × 106 cells per well in 96-wellplates in complete medium [RPMI 1640 (Gibco) supplementedwith 25 mMHepes (Gibco), heat-inactivated 10% (vol/vol) FBS, lowendotoxin (HyClone), and penicillin/streptomycin (Gibco)], con-taining anti-CD28 and anti-CD49d (2 μg/mL each; BD Biosciences)only or HlaH35L, EsxAB, FhuD2, or Csa1A in combination(10 μg/mL each) at 37 °C for 16–18 h in the presence of BrefeldinA (5 μg/mL) for the last 4 h. OVA (30 μg/mL; Hyglos) was usedas a negative control. The cells were then stained with Live/Deadyellow (Invitrogen), fixed, permeabilized with Cytofix/Cytoperm(BD Biosciences), washed in Perm/Wash buffer (BD Biosciences),incubated with anti-CD16/CD32 Fc block (BD Biosciences)for 20 min at room temperature, stained with fluorochrome-conjugated mAbs (BD Pharmingen, unless specified) [anti–CD3-PerCP-Cy5.5, anti–CD8-PE Texas Red (Invitrogen), anti–CD4-V500, anti–CD44-V450, anti–IFN-γ-PE, anti–IL-2-APC,anti–TNF-AlexaFluor700, anti–IL-4-AlexaFluor488 (eBioscience),anti–IL-13-AlexaFluor488 (eBioscience), and IL-17A-PE-Cy7

(eBioscience)] in Perm/Wash buffer (BD Biosciences) for 20 minat room temperature, washed twice in Perm/Wash buffer, andsuspended in PBS. Samples were acquired on a LSRII flow cy-tometer (BD Biosciences) and analyzed using FlowJo software(TreeStar).

Statistical Analysis. At least two independent experiments, rununder the same conditions, were performed for all studies. For theperitonitis and pneumonia models, statistical significance wasassessed with the log-rank test using GraphPad Prism software.The significance of the survival rate and median survival in theperitonitis model was assessed by Fisher’s exact test and Mann–Whitney U test. The Student paired t test was used to analyze thestatistical significance of opsonophagocytosis experiments. TheMann–Whitney U test was used to analyze the statistical signif-icance of experiments performed with the renal abscess model,as well as of the experiment of Hla inhibition mediated by vac-cine antibodies. T-cell responses to vaccination were analyzedusing ANOVA and Tukey’s posttest correction.

1. Klock HE, Lesley SA (2009) The Polymerase Incomplete Primer Extension (PIPE)method applied to high-throughput cloning and site-directed mutagenesis. MethodsMol Biol 498:91–103.

2. Kennedy AD, et al. (2010) Targeting of alpha-hemolysin by active or passive immu-nization decreases severity of USA300 skin infection in a mouse model. J Infect Dis202(7):1050–1058.

3. Bunce C, Wheeler L, Reed G, Musser J, Barg N (1992) Murine model of cutaneousinfection with gram-positive cocci. Infect Immun 60(7):2636–2640.

4. Inoshima N, Wang Y, Bubeck Wardenburg J (2012) Genetic requirement for ADAM10in severe Staphylococcus aureus skin infection. J Invest Dermatol 132(5):1513–1516.

5. Bubeck Wardenburg J, Palazzolo-Ballance AM, Otto M, Schneewind O, DeLeo FR(2008) Panton-Valentine leukocidin is not a virulence determinant in murine modelsof community-associated methicillin-resistant Staphylococcus aureus disease. J InfectDis 198(8):1166–1170.

6. Mishra RP, et al. (2012) Staphylococcus aureus FhuD2 is involved in the early phase ofstaphylococcal dissemination and generates protective immunity in mice. J Infect Dis206(7):1041–1049.

7. Schluepen C, et al. (2013) Mining the bacterial unknown proteome: Identification andcharacterization of a novel family of highly conserved protective antigens in Staph-ylococcus aureus. Biochem J 455(3):273–284.

8. Mariotti P, et al. (2013) Structural and functional characterization of the Staphylococcusaureus virulence factor and vaccine candidate FhuD2. Biochem J 449(3):683–693.

9. Kuhn ML, et al. (2014) Structure and protective efficacy of the Staphylococcus aureusautocleaving protease EpiP. FASEB J 28(4):1780–1793.

10. Becherelli M, et al. (2013) Protective activity of the CnaBE3 domain conserved amongStaphylococcus aureus Sdr proteins. PLoS ONE 8(9):e74718.

Fig. S1. AFOG staining of kidney sections of S. aureus-infected mice. Mice immunized with alum alone (Alum) (D–F) or with the combined vaccine (4C-Staph)(A–C) and then challenged with S. aureus LAC. Representative images, i.e., having abscess area and number close to the geometric mean for the group, areshown. (A and D) Whole kidney section. (B and E) Cartoon indicating in black the abscess areas for A and D, respectively. (C and F) High magnification of theareas surrounded by white lines in A and D, respectively, showing abscesses with Staphylococcal abscess communities (SACs) in the core, indicated by white arrows,and surrounded by infiltrating polymorphonuclear leukocytes (PMNs).

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 4 of 10

Fig. S2. Protective efficacy elicited by IsdB immunization against S. aureus strains NRS216 and Mu50 in the peritonitis model. Mice were challenged with thestrains NRS216 and Mu50 (n = 8–16 per group). In all plots, lines with solid circles indicate IsdB, whereas lines with open circles are those mice treated withaluminum hydroxide alone. Statistical analysis was performed by log-rank (Mantel–Cox) test.

Fig. S3. Healthy humans have antibodies against the vaccine antigens. (A and B) Immunoreactivity of sera from healthy donors toward vaccine antigensassessed by Western blot analysis. (A) Representative blots of three different sera. The first panel on the left shows a serum, which recognized all of the fourvaccine components, whereas in the second and third panels, sera reacted, respectively, only against HlaH35L, EsxAB, and FhuD2 or HlaH35L and EsxAB. (B)Number of sera reactive against at least one of the four vaccine components among the 100 sera tested.

Table S1. Primers used in this study

Primer Sequence (5′ to 3′)Expected

fragment (bp) Description

FhuD2 for CACTAGCTATTGTAAATGTATATGC 1,028 Primers for PCR amplification of fhuD2 geneFhuD2 rev GATATGTTTCAGACTCTCATTTCAC

Hla for ATAAATATTTGATATGTCTCAACTGC 1,056 Primers for PCR amplification of hla geneHla rev CATCATTTCTGAAGTTATCGGC

esxA for CACTTTTTCAAAAAATAGTGTCCC 416 Primers for PCR amplification of esxA geneesxA rev AAAATGATAAGCAAAGCCACATTAG

esxB for GAATCCCTCCAAAAAGCTAAGG 495 Primers for PCR amplification of esxB geneesxB rev CAAACAATTCAAAATTTACGAGAGG

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 5 of 10

Table S2. S. aureus strains used in this study

Strain Description Reference/source

Newman Clinical strain, MSSA, CC8, ST254, CP5 (1)LAC Clinical strain, MRSA, USA300, SCCmec IV, CC8, ST8, CP5 (2)MW2 Clinical strain, MRSA, USA400, SCCmec IV, CC1, ST1, CP8 (3)Mu50 Clinical strain, HA-VR-MRSA, USA100, SCCmec II, CC5, ST5, CP5 (4)Staph 19 Clinical strain, MRSA, SCCmec IV, ST80, CP8 This studyNRS216 Clinical Strain, MSSA, ST30, CC30, CP8 This studyReynolds Laboratory strain, MSSA, CC25, ST25, CP5 (5)Wright Laboratory strain, MSSA, CP8 This studyBecker Laboratory strain, MSSA, CP8 (6)ATCC6538 Laboratory strain, MSSA, CC97, ST467, CP5 This studyLowenstein Laboratory strain, MSSA, CC25, ST25, CP5 (7)BD1686 Clinical strain, HA-MRSA, USA100, CC5, ST5, CP5 This studyBD1534 Clinical strain, HA-MRSA, USA200, CC30, ST36, CP8 This studyBD1449 Clinical strain, CA-MRSA, USA1000, CC59, ST59, CP8 This studyNRS382 Clinical strain, HA-MRSA, USA100, CC5, ST5, CP5 This studyNRS248 Clinical strain, MRSA, CC1, ST1, CP8 This studyNRS252 Clinical strain, MSSA, CC30, ST30, CP8 This studyStaph 015 Clinical strain, CA-MRSA, ST30, CC30, CP8 This studyStaph 017 Clinical strain, CA-MRSA, ST8, CC8, CP5 This studyStaph 018 Clinical strain, CA-MRSA, ST8, CC8, CP5 This studyStaph 021 Clinical strain, CA-MRSA, SCCmec IV, ST80, CP8 This studyMSSA 94 ISS Clinical strain, MSSA, CC22, ST22, CP5 This studyIT-SA1 Clinical strain from infective endocarditis, CC30, ST30, CP8 (8)IT-SA2 Clinical strain from infective endocarditis, CP8 (8)IT-SA3 Clinical strain from infective endocarditis, CC101, ST101, CP8 (8)IT-SA4 Clinical strain from infective endocarditis, CC30, ST34, CP8 (8)IT-SA5 Clinical strain from infective endocarditis, CC8, ST8, CP5 (8)IT-SA6 Clinical strain from infective endocarditis, CC5, ST5, CP5 (8)IT-SA7 Clinical strain from infective endocarditis, CC20, ST20, CP5 (8)IT-SA8 Clinical strain from infective endocarditis, CC15, ST15, CP8 (8)IT-SA9 Clinical strain from infective endocarditis, CC121, ST120, CP8 (8)IT-SA10 Clinical strain from infective endocarditis, CC15, ST15, CP8 (8)IT-SA11 Clinical strain from infective endocarditis, CC5, ST5, CP8 (8)IT-SA12 Clinical strain from infective endocarditis, CC72, ST72, CP5 (8)IT-SA14 Clinical strain from infective endocarditis, CC5, ST5, CP5 (8)IT-SA15 Clinical strain from infective endocarditis, CC45, ST45, CP8 (8)IT-SA16 Clinical strain from infective endocarditis, CC72, ST72, CP8 (8)IT-SA17 Clinical strain from infective endocarditis, CC121, ST120, CP8 (8)IT-SA18 Clinical strain from infective endocarditis, CC5, ST5, CP5 (8)IT-SA19 Clinical strain from infective endocarditis, CC5, ST5, CP5 (8)SW-ST239-III Clinical strain, MRSA, SCCmec III, CC8, ST239, CP8 This studySW-ST80-IV-PVL Clinical strain, MRSA, SCCmec IV, pvl+, ST80, CP8 This studySW-ST5-IV-PVL Clinical strain, MRSA, SCCmec IV, pvl+, CC5, ST5, CP5 This studySW-ST30-PVL Clinical strain, MRSA, pvl+, CC30, ST30, CP8 This studySW-ST398 Clinical strain, MRSA, CC398, ST398, CP5 This studySW-ST228-I Clinical strain, MRSA, CC5, ST288, CP5 This studySW-ST8-IV Clinical strain, MRSA, CC8, ST8, CP5 This studySW-ST88 Clinical strain, MRSA, CC88, ST88, CP8 This studySW-ST45 Clinical strain, MRSA, CC45, ST45, CP8 This studySW-ST42 Clinical strain, MSSA, Singleton, ST42, CP8 This studySW-ST152 Clinical strain, MRSA, CC152, ST152, CP5 This studySW-ST59 Clinical strain, MRSA, CC59, ST59, CP8 This study

CC, clonal complex; CP, capsule type; LAC, Los Angeles County clone; MRSA, methicillin-resistant S. aureus;MSSA, methicillin-sensitive S. aureus; ST, sequence type.

1. Baba T, Bae T, Schneewind O, Takeuchi F, Hiramatsu K (2008) Genome sequence of Staphylococcus aureus strain Newman and comparative analysis of staphylococcal genomes:polymorphism and evolution of two major pathogenicity islands. J Bacteriol 190(1):300–310.

2. Miller LG, et al. (2005) Necrotizing fasciitis caused by community-associated methicillin-resistant Staphylococcus aureus in Los Angeles. N Engl J Med 352(14):1445–1453.3. Baba T, et al. (2002) Genome and virulence determinants of high virulence community-acquired MRSA. Lancet 359(9320):1819–1827.4. Kuroda M, et al. (2001) Whole genome sequencing of meticillin-resistant Staphylococcus aureus. Lancet 357(9264):1225–1240.5. Karakawa WW, et al. (1985) Method for the serological typing of the capsular polysaccharides of Staphylococcus aureus. J Clin Microbiol 22(3):445–447.6. Cook J, et al. (2009) Staphylococcus aureus capsule type 8 antibodies provide inconsistent efficacy in murine models of staphylococcal infection. Hum Vaccin 5(4):254–263.7. Fattom A, et al. (1990) Synthesis and immunologic properties in mice of vaccines composed of Staphylococcus aureus type 5 and type 8 capsular polysaccharides conjugated to

Pseudomonas aeruginosa exotoxin A. Infect Immun 58(7):2367–2374.8. Rindi S, et al. (2006) Antibody response in patients with endocarditis caused by Staphylococcus aureus. Eur J Clin Invest 36(8):536–543.

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 6 of 10

Table S3. S. aureus genomes available in National Center forBiotechnology Information database used to estimate the levelof conservation of vaccine antigen genes

Strain name Clonal complex, sequence type

NCTC8325 CC8, ST8Mu3 CC5, ST5N315 CC5, ST5MRSA252 (EMRA16) CC30, ST36MSSA476 CC1, ST1TW20 (0582 Sanger) CC8, ST239COL CC8, ST250NRS112 (MN8) CC30, ST30EMRSA15 CC22, ST22FRP3757_ USA300 CC8, ST8USA300_TCH1516 CC8, ST8JH1 CC5, ST105JH9 CC5, ST105RF122 CC705, ST151ED98 CC5, ST5402981 CC5, ST225A5937 CC5, ST5A5948 CC8, ST8A6224 CC5, ST5A8115 CC5, ST5A9299 CC5, ST5A9635 CC45, ST278A9719 CC5, ST5A9763 CC5, ST5A9781 CC5, ST5132 CC8, ST855/2053 CC30, ST3065–1322 CC30, ST3068–397 CC30, ST30E1410 CC30, ST30M876 CC30, ST30Mu50-omega CC5, ST5TCH130 CC8, ST72TCH60 CC30, ST30TCH70 CC1, ST1USA300TCH959 CC7, ST1159CF-Marseille CC5, ST5JKD6008 CC8, ST239JKD6009 CC8, ST239A017934 CC30, ST30A10102 CC5, ST5A8117 CC5, ST5Btn1260 CC30, ST30C101 CC30, ST30C160 CC30, ST34C427 Singleton, ST42D139 CC10, ST145M899 CC30, ST30WW2703 CC30, ST30930918 CC8, ST8D30 CC8, ST858424 CC30, ST30M1015 CC30, ST30M809 ST431A9765 CC8, ST8MR1 CC5, ST5A9754 CC8, ST8

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 7 of 10

Table S4. Summary of 4C-Staph efficacy in reducing CFU counts in the renal abscess model

Immunization Challenge strain and dose N Mean log CFU ± SE Log CFU reduction* P vs. Alum† P vs. 4C-Staph‡

Alum ST254(Newman)2.0 × 107 16 7.53 ± 0.30 — — 0.002Csa1A 14 6.66 ± 0.31 0.87 0.042 0.868EsxAB 16 7.24 ± 0.26 0.29 0.403 0.028HlaH35L 16 6.55 ± 0.39 0.98 0.043 0.792FhuD2 16 6.29 ± 0.33 1.24 0.011 0.8654C-Staph 16 5.96 ± 0.36 1.57 0.002 —

IsdB 16 6.47 ± 0.30 1.06 0.011 0.283

Alum USA300(LAC)3.8 × 107 20 7.01 ± 0.18 — — <0.0001Csa1A 20 5.37 ± 0.46 1.64 0.007 0.172EsxAB 20 5.25 ± 0.47 1.76 0.003 0.257HlaH35L 19 6.27 ± 0.39 0.74 0.166 0.006FhuD2 20 4.95 ± 0.47 2.06 0.001 0.5804C-Staph 19 4.57 ± 0.49 2.44 <0.0001 —

IsdB 19 4.93 ± 0.48 2.08 0.0004 0.575

Alum USA400(MW2)2.9 × 107 16 7.11 ± 0.24 — — 0.001Csa1A 17 6.06 ± 0.40 1.05 0.027 0.086EsxAB 20 5.84 ± 0.42 1.27 0.027 0.112HlaH35L 19 5.89 ± 0.29 1.22 0.003 0.196FhuD2 20 6.17 ± 0.36 0.94 0.054 0.0184C-Staph 20 4.80 ± 0.50 2.31 0.001 —

IsdB 20 5.92 ± 0.39 1.19 0.019 0.100

Alum USA100(Mu50)4.2 × 107 19 7.49 ± 0.17 — — <0.0001Csa1A 18 6.29 ± 0.43 1.20 0.021 0.466EsxAB 20 6.89 ± 0.25 0.60 0.026 0.084HlaH35L 20 7.10 ± 0.22 0.39 0.091 0.051FhuD2 20 6.19 ± 0.24 1.30 <0.0001 0.8884C-Staph 19 5.98 ± 0.39 1.51 0.003 —

Alum USA100§4.1 × 107 39 7.00 ± 0.16 — — 0.00034C-Staph 39 5.62 ± 0.27 1.38 0.0003 —

IsdB 20 5.82 ± 0.35 1.18 0.002 0.648

Alum ST80(Staph 19)4.9 × 107 18 7.43 ± 0.19 — — <0.0001Csa1A 20 6.28 ± 0.33 1.15 0.002 0.011EsxAB 17 6.38 ± 0.37 1.05 0.014 0.011HlaH35L 17 6.49 ± 0.31 0.94 0.012 0.008FhuD2 18 5.30 ± 0.46 2.13 <0.0001 0.5894C-Staph 20 5.20 ± 0.33 2.23 <0.0001 —

Alum ST80§3.7 × 107 38 6.89 ± 0.20 — — <0.00014C-Staph 40 5.48 ± 0.25 1.41 <0.0001 —

IsdB 20 5.77 ± 0.37 1.12 0.004 0.490

Mice were immunized with the indicated single or combined antigens, or with alum alone, and then challenged with the indicated strains (at least twoseparate experiments).*Log CFU reduction = mean log CFU alum ctrl − mean log CFU vaccinated.†Significance vs. alum group, one-tailed Mann–Whitney U test.‡Significance vs. 4C-Staph group, two-tailed Mann–Whitney U test.§This experiment was conducted independently from the one above. Values of P < 0.01 and 4C-Staph rows are highlighted in bold.

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 8 of 10

Table

S5.

Summaryofefficacy

of4C

-Staphva

ccinein

reducingab

scessesan

dtissuedam

agein

therenal

abscessmodel

Immunization

Challengestrain

anddose

N

Abcess

area

*Abscessnumber

Nep

hritis

Nep

hrosis

Geo

mea

nVaccinated

/co

ntrol†

Pvs.

alum

‡

Pvs.

4C-Staph§

Geo

mea

nVaccinated

/co

ntrol{

Pvs.

alum

‡

Pvs.

4C-Staph§

Mea

nscore

±SE

Pvs.

alum

‡

Mea

nscore

±SE

Pvs.

alum

‡

Alum

ST25

4(New

man

)2.0

×10

716

1.37

——

0.03

711

.14

——

0.06

31.7±

0.2

—1.6±

0.1

—

Csa1A

141.02

74.67

0.49

20.11

82.77

24.89

0.11

00.98

3n.d.

n.d.

EsxA

B16

5.16

377.56

0.00

6<0.00

0113

.39

120.21

0.30

50.03

0n.d.

n.d.

HlaH35

L16

1.32

96.77

0.43

30.11

72.78

24.94

0.04

60.97

0n.d.

n.d.

FhuD2

160.54

39.73

0.06

80.73

41.52

13.67

0.01

40.46

0n.d.

n.d.

4C-Staph

160.29

21.28

0.03

7—

2.12

19.02

0.06

3—

1.2±

0.2

0.10

10.5±

0.1

0.02

0IsdB

160.90

65.90

0.15

90.30

81.98

17.74

0.00

10.18

4n.d.

n.d.

Alum

USA

300(LA

C)3.8

×10

720

0.96

——

0.00

77.67

——

0.01

82.3±

0.3

—1.3±

0.2

—

Csa1A

200.40

42.12

0.17

90.15

01.90

24.71

0.15

50.23

8n.d.

n.d.

EsxA

B20

0.21

22.07

0.06

90.40

30.92

11.99

0.21

10.37

1n.d.

n.d.

HlaH35

L19

0.98

102.28

0.26

40.02

99.13

119.05

0.40

60.01

9n.d.

n.d.

FhuD2

200.74

76.94

0.32

80.04

24.44

57.84

0.19

70.08

5n.d.

n.d.

4C-Staph

190.12

12.72

0.00

7—

0.40

5.21

0.01

8—

1.3±

0.3

0.01

00.9±

0.2

0.05

8IsdB

190.94

98.48

0.15

50.03

22.45

31.94

0.46

20.09

6n.d.

n.d.

Alum

USA

400(MW

2)2.9×10

714

1.20

——

0.41

79.33

——

0.15

52.0±

0.3

—1.1±

0.2

—

Csa1A

160.52

43.05

0.24

60.57

61.92

20.55

0.04

60.30

9n.d.

n.d.

EsxA

B20

1.65

137.79

0.26

40.35

86.62

71.04

0.12

00.82

9n.d.

n.d.

HlaH35

L19

1.33

110.68

0.26

20.43

93.04

32.59

0.02

60.31

7n.d.

n.d.

FhuD2

200.63

52.43

0.35

00.93

51.54

16.51

0.02

10.29

6n.d.

n.d.

4C-Staph

200.78

65.27

0.41

7—

3.76

40.33

0.15

5—

1.4±

0.3

0.08

00.7±

0.1

0.05

1IsdB

201.73

144.43

0.16

20.16

65.41

58.00

0.10

90.88

6n.d.

n.d.

Alum

USA

100(Mu50

)4.2

×10

719

1.77

——

0.03

44.77

——

0.00

81.10.3

—0.9±

0.1

—

Csa1A

180.55

30.95

0.18

90.90

31.38

28.90

0.21

40.39

9n.d.

n.d.

EsxA

B20

0.96

54.40

0.34

20.13

22.12

44.41

0.40

00.13

8n.d.

n.d.

HlaH35

L20

2.09

118.27

0.13

70.02

85.20

109.15

0.32

60.01

1n.d.

n.d.

FhuD2

200.44

24.85

0.03

40.71

41.09

22.87

0.02

60.74

4n.d.

n.d.

4C-Staph

190.67

37.82

0.03

4—

1.32

27.67

0.00

8—

0.6±

0.2

0.06

20.6±

0.1

0.06

4Alum

ST80

(Staph19

)4.9

×10

718

0.87

——

0.00

73.17

——

0.02

41.7±

0.3

—0.9±

0.2

—

Csa1A

200.83

95.71

0.12

70.03

84.13

130.35

0.34

60.06

3n.d.

n.d.

EsxA

B17

0.95

109.54

0.29

90.02

94.95

156.07

0.36

40.06

0n.d.

n.d.

HlaH35

L17

0.46

53.28

0.24

90.19

11.37

43.10

0.14

80.53

8n.d.

n.d.

FhuD2

180.11

13.04

0.00

30.44

70.33

10.54

0.00

40.26

9n.d.

n.d.

4C-Staph

200.22

25.33

0.00

7—

0.77

24.13

0.02

4—

0.9±

0.3

0.01

50.5±

0.1

0.05

2

Micewereim

munized

withalum

alone,

thesingle

antigen

s,orthe4C

-Staphva

ccine,

andthen

challenged

withtheindicated

strainsofS.

aureus(atleasttw

oseparateex

perim

ents).From

left

toright,thetable

showsthegeo

metricmea

nva

lues

ofCFU

counts,ab

scessarea

andnumber,an

dnep

hritisan

dnep

hrosismea

suredin

micetrea

tedwiththeva

riousva

ccines.Statisticalsignifican

cetowardalum-treated

miceor

4C-Staph

isindicated

.Morphometrican

alysis

oftheab

scessesreve

aled

that

immunization

with

theco

mbined

vaccinewas

able

tosignifican

tlyreduce

thesize

and/orthenumber

oftheab

scessesin

mice

challenged

withallstrainstested

(P<0.05

),in

agreem

entwiththeCFU

resultsshownin

Table

1,withtheex

ceptionofstrain

MW2,

forwhichthereductionwas

notsignifican

t.Protectiveefficacy

inducedby

4C-Staphwas

significan

tlygreater

than

those

obtained

withsingle

antigen

sin

seve

ralinstan

ces.Rep

resentative

imag

esofAFO

G-stained

kidney

sectionsfor4C

-Staphan

dalum

controlg

roupsareshownin

Fig.S

1.Im

munizationwith4C

-Staphwas

also

foundto

significan

tlyreduce

nep

hritisan

dnep

hrosisin

somecasesan

dnea

rlysignifican

tlyin

most

oftheothers.Values

ofP<0.05

and4C

-Staphrowsarehighlig

htedin

bold.n.d.,notdetermined

.*A

reaev

aluated

asthepercentofthearea

ofthesectionco

veredbytheab

scess.

†Vaccinated

/controlratio=

(geo

metricmea

nab

scessarea

vaccinated

/geo

metric

mea

nab

scessarea

control)×10

0.‡One-taile

dMan

n–W

hitney

Utest.

§Tw

o-tailedMan

n–W

hitney

Utest.

{ Vaccinated

/controlratio=

(geo

metricmea

nnumber

ofab

scessesva

ccinated

/geo

metricmea

nnumber

ofab

scessesco

ntrol)×10

0.

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 9 of 10

Table S6. Summary of 4C-Staph efficacy in increasing survival in the peritonitis model

ImmunizationChallenge strain

and dose N

Survival rate Survival kinetics

Percentsurvival P vs. alum* P vs. 4C-Staph*

Mediansurvival (d) P vs. alum† P vs. 4C-Staph†

Alum ST254(Newman)5.5 × 108 32 22 — 0.001 1.0 — <0.0001Csa1A 32 31 0.29 0.01 8.5 0.048 0.0009EsxAB 32 38 0.14 0.04 1.5 0.08 0.0009HlaH35L 32 31 0.29 0.01 9.0 0.0001 0.008FhuD2 32 38 0.14 0.04 7.5 0.004 0.0054C-Staph 32 63 0.001 — 15.0 <0.0001 —

IsdB 32 38 0.14 0.04 2.5 0.06 0.0008Alum USA400(MW2)8 × 108 60 30 — 0.0001 6.0 — <0.0001Csa1A 60 60 0.0008 0.087 15.0 <0.0001 0.1EsxAB 60 48 0.03 0.004 10.0 0.08 0.0008HlaH35L 60 43 0.09 0.0008 8.0 0.07 0.0002FhuD2 60 52 0.01 0.01 15.0 0.008 0.0074C-Staph 60 61 0.0001 — 15.0 <0.0001 —

IsdB 59 44 0.08 0.001 8.0 0.06 0.0005

Mice were immunized with the indicated single or combined antigens, or with alum alone, and then challenged with the indicated strains (at least threeseparate experiments). Values of P < 0.01 and 4C-Staph rows are highlighted in bold.*Fisher’s exact test.†Mann–Whitney U test.

Bagnoli et al. www.pnas.org/cgi/content/short/1424924112 10 of 10