Broadly Protective Vaccine for Staphylococcus aureus Based on an in Vivo-Expressed Antigen

See allHide authors and affiliations

Science  28 May 1999:
Vol. 284, Issue 5419, pp. 1523-1527
DOI: 10.1126/science.284.5419.1523


Vaccines based on preferential expression of bacterial antigens during human infection have not been described. Staphylococcus aureus synthesized poly-N-succinyl β-1-6 glucosamine (PNSG) as a surface polysaccharide during human and animal infection, but few strains expressed PNSG in vitro. All S. aureusstrains examined carried genes for PNSG synthesis. Immunization protected mice against kidney infections and death from strains that produced little PNSG in vitro. Nonimmune infected animals made antibody to PNSG, but serial in vitro cultures of kidney isolates yielded mostly cells that did not produce PNSG. PNSG is a candidate for use in a vaccine to protect against S. aureus infection.

Staphylococcus aureus is the most frequently isolated bacterial pathogen in hospital-acquired infections (1) and is a common cause of community-acquired infections, including endocarditis, osteomyelitis, septic arthritis, pneumonia, and abscesses (1). Staphylococcus aureus is also a significant pathogen in economically important animals (2). Staphylococcal resistance to first-line drugs such as synthetic penicillins has resulted in major problems in treating methicillin-resistant S. aureus (MRSA) strains, which are increasingly common, especially in hospitalized patients. Of greater concern is the recent emergence in several countries of MRSA strains with reduced susceptibility to vancomycin, the antibiotic of last resort (3). The appearance of these vancomycin-intermediate S. aureus (VISA) strains raises the specter of untreatable staphylococcal infections, necessitating a search for alternative therapies (1, 3).

One potential therapeutic target for bacterial infections is products of bacterial genes activated during in vivo infections (4). Presumably these genes encode factors critical for infection and disease progression. Strategies designed to discover in vivo–expressed genes have, of necessity, used animal models for gene identification, and in vivo–expressed bacterial factors have not yet been detected in human infections. In addition, the usefulness in vaccines of any factors encoded by these genes has not been demonstrated. We now provide evidence that antibodies to a distinct surface polysaccharide antigen of S. aureus that is preferentially elaborated in vivo in infected humans and experimental animals are broadly protective against many S. aureusstrains.

Lung tissue removed from two S. aureus–infected cystic fibrosis (CF) patients contained organisms that elaborated a surface polysaccharide that reacted positively with immune rabbit serum specific for poly-N-succinyl β-1-6 glucosamine (PNSG), previously determined to be the Staphylococcus epidermidiscapsular polysaccharide adhesin (5) (Fig. 1A). PNSG is distinct from the serologically and structurally defined S. aureus capsular polysaccharides (CP), including CP5 and CP8, which are made by >70% of all S. aureus strains (6). Only background fluorescence was observed in the lung sections with antibodies to CP antigens (Fig. 1A). In addition, six of nine sputum samples from S. aureus–infected CF patients were positive for PNSG expression by immunofluorescent microscopy (Fig. 1B; 80 to 90% of cells fluorescent) (7), whereas bacteria in these samples expressed little CP antigen (<10% of cells were positive) (8). In vitro cultivation of the S. aureus lung and sputum isolates in trypticase soy broth under aerobic conditions resulted in high expression of CP antigen by all isolates as previously shown (9) and decreased expression of PNSG (8), indicating that PNSG is an in vivo–expressed bacterial factor in human infection.

Figure 1

PNSG is synthesized in infected human tissues, and S. aureus strains carry genes needed for PNSG expression. (A) (Left) Immunofluorescent staining ofS. aureus–infected lung as seen by reaction with antiserum to PNSG (anti-PNSG) and CY3-indocarbocyanine–conjugated antibody to rabbit IgG (yellow fluorescence) (27). (Middle) DAPI staining ofS. aureus cells in a lung section. (Right) Only background immunofluorescence is seen when antibody to the CP microcapsule (anti-CP8) is reacted with S. aureus cells in an infected lung. (B) Immunofluorescent micrographs of a sputum sample from an S. aureus–infected CF patient (19). (Left) Production of PNSG as seen by reaction with antiserum to PNSG (anti-PNSG) and fluorescein-conjugated swine antibody to rabbit IgG. (Right) DAPI staining of S. aureus cells in sputum sample. (C) Immunoelectron microscopic demonstration of expression of surface PNSG by three fresh clinical isolates of S. aureus (5). Only 8 of 43 fresh isolates elaborated detectable PNSG on primary agar plates obtained from the microbiology laboratory. When three of these positive strains were reacted with normal rabbit serum (NRS) and protein A conjugated to 20-nm gold particles, there was little binding of the gold to the extracellular surface, whereas reaction with immune rabbit serum (IRS) to PNSG and protein A–gold resulted in binding of gold particles to the cell surface. Bars, 0.5 μm. (D) Staphylococcus aureus isolates contain the ica locus needed for PNSG synthesis (5, 12). PCR performed on chromosomal DNA from controls and eight isolates of S. aureus(13). Lane 1, molecular weight markers; lane 2, S. carnosus TM300, negative control; and lane 3, positive control DNA from S. epidermidis RP62A. Staphylococcus aureusstrains: lane 4, Reynolds; lane 5, MN8; lane 6, 5827; lane 7, 5836; lane 8, Vas; lane 9, VP; lane 10, 265; and lane 11, Por. Primers were designed to amplify a 2.7–kilobase pair fragment [migrates between 2036–base pair (bp) and 3054-bp markers] containing theicaA, icaD, and icaB genes and a section of icaC genes (12).

A set of animal and human clinical isolates of S. aureus was investigated for PNSG production. With the use of a colony immunoblot assay (10), 29 of 82 (35%) freezer-storedS. aureus isolates from the milk of cows and sheep and 14 of 82 (17%) stored human isolates were PNSG-positive; the latter comprised strains from blood, wounds, and vaginas (associated with toxic shock syndrome). To determine if the low expression of PNSG among these isolates was a result of laboratory storage and passage of strains, we used an enzyme-linked immunosorbent assay (ELISA) inhibition technique (11) to detect PNSG production by 43 fresh clinical bloodstream isolates of S. aureus growing on primary blood agar plates obtained from a hospital microbiology laboratory. The geometric mean percentage of inhibition of antibody binding was only 13% (95% confidence interval of 8 to 20%), and only 8 of 43 (19%) isolates inhibited >50% of the antibody binding to PNSG. However, on some isolates, material on the cell surface bound antibody to PNSG when visualized by immunoelectron microscopy (Fig. 1C). Although in vitro expression by S. aureus of the immunoreactive PNSG antigen was low or absent among most strains, it was detectable as a surface polysaccharide on some strains.

Because immunoreactive PNSG antigen production appeared to undergo an in vivo–in vitro phenotypic variation in S. aureus, we investigated whether S. aureus strains had the intercellular adhesin (ica) locus of S. epidermidis made up of four genes (icaADBC) whose protein products synthesize PNSG (5, 12). The icaADBC genes were detected in eight S. aureus strains by polymerase chain reaction (PCR) (Fig. 1D) (13), whereas a Staphylococcus carnosus strain known to lack the ica locus was negative. In addition, theicaADBC genes are present in the genomes of both S. aureus strains, COL and NCTC 8325-4, that are currently being sequenced (14).

To purify and characterize the antigen from S. aureusreactive with antisera to PNSG, we used strain MN8m, which was recovered from a chemostat culture of strain MN8. Strain MN8m is a constitutive, copious producer of immunoreactive material (Fig. 2). Using the method previously developed for isolating PNSG from S. epidermidis (5), we obtained from strain MN8m an antigen identical to the S. epidermidis PNSG as determined by proton nuclear magnetic resonance (NMR) and chemical analysis (15) (Fig. 2).

Figure 2

PNSG elaboration by S. aureus strain MN8m. (Top) Probing of strain MN8m with IRS to purified PNSG followed by gold-labeled protein A showed binding of the gold particles to a thick extracellular antigen surrounding the bacterial cells in electron micrographs (11). NRS produced only a minimal binding of the gold particles. Bars, 0.5 μm. (Bottom) Proton NMR spectrum of the acid-hydrolyzed PNSG antigen isolated fromS. aureus MN8m and analyzed as described (15). The deduced chemical structure of PNSG is drawn above the spectrum.

A mouse model of renal infection was used for active immunization with PNSG to evaluate protective efficacy against S. aureusinfection (16). Because heavily infected kidneys are found in animals with S. aureus infections of other tissues resulting in endocarditis, arthritis, and bacteremia (17), the colony-forming units (CFU) of S. aureus in the kidneys reflect systemic infection of a variety of tissues. Mice immunized with either PNSG derived from S. aureus MN8m or a control, irrelevant bacterial polysaccharide developed high titers (>500) of immunoglobulin G (IgG) antibodies to the immunizing, but not heterologous, antigen. Five days after intravenous challenge with two different doses of two S. aureus strains (CP5 strain Reynolds and CP8 strain MN8) phenotypically negative for PNSG production in vitro, the PNSG-immunized mice had significant reductions in CFU of S. aureus per gram of kidney compared with the groups immunized with an irrelevant polysaccharide (Fig. 3A).

Figure 3

Protective efficacy of PNSG. (A) Active immunization of mice with PNSG provides protection against challenge with S. aureus (16). Bars indicate mean CFU per gram of kidney in mice immunized with PNSG (solid bars) or with an irrelevant bacterial polysaccharide (speckled bars); challenge doses in CFU per mouse are given below each pair of bars, and Pvalues below the challenge doses were derived by two-tailed, unpairedt tests. N = 5 mice per group. (B) Concentration of S. aureus in infected mouse kidneys 5 days after IV injection of bacteria and IP injection of either antibodies raised to PNSG (solid bars) or antibodies raised to an irrelevant bacterial polysaccharide (speckled bars) (28). Zeros (0) indicate no infected kidneys in any mice (lower limit of detection ∼5 CFU per gram of kidney), bars represent geometric means, and error bars represent the upper standard deviation; challenge doses in CFU per mouse are given below each pair of bars, andP values given below the challenge doses were derived by two-tailed, unpaired t tests. Staphylococcus aureus strains MN8 and Reynolds represent CP8 and CP5 strains, respectively; S. aureus 5827 and 5836 are MRSA-VISA strains;S. aureus 265 is methicillin resistant; S. aureusVas and VP are freezer-stored human isolates; and S. aureusPor is a fresh clinical isolate from a patient with bacteremia.N = 5 mice per group.

Rabbits immunized with PNSG responded by producing high titers (>2500) of PNSG-specific IgG that persisted for at least 8 months. To document that PNSG-specific antibody mediated the protection afforded by active immunization, we passively immunized mice with rabbit antibody before and again 18 hours after intravenous (IV) challenge with eight strains of S. aureus. Controls received comparably titered antisera to an irrelevant polysaccharide fromPseudomonas aeruginosa. Seven of the S. aureuschallenge strains produced little to no PNSG at challenge (0 to 21% ELISA inhibitory activity), whereas the S. aureus strain 5827 inhibited PNSG antibody binding by 48%. The challenge strains had a broad range of minimal doses that resulted in kidney infections in 100% of animals (range of 102 to 106 CFU per mouse), but for all eight strains tested, passive immunization with antibodies to PNSG significantly reduced or completely eliminated bacteria from the kidneys (Fig. 3B) compared with controls. Although it was possible to overwhelm the protective effect with higher challenge doses in some cases (for example, strain Reynolds at a challenge dose of 8 × 105 CFU per mouse), it was also possible to achieve total protection against at least one challenge dose for four of the eight strains (Fig. 3B). The lower numbers of S. aureus in the kidneys of some control animals in the passively immunized groups compared with the numbers in control animals in the actively immunized control groups (Fig. 3A) were likely due to a modest enhancement of resistance to infection from the control antiserum in the passive protection studies.

An additional study was done comparing protection by rabbit IgG antibodies to PNSG with human IgG to CP5 and CP8. The dose of the human IgG that was used (300 μg of CP5-specific and 246 μg of CP8-specific antibody per mouse) was greater than that necessary to protect mice against lethal infection with a virulent, CP5 MRSA strain when both the antibodies and bacteria were injected intraperitoneally (IP) (18). However, in mice injected IP with antibodies and challenged IV with 106 CFU of strain 5836 (a CP5 MRSA-VISA strain), there was no reduction in bacteria in the kidneys by CP-specific antibodies compared with animals given control human antibodies to an irrelevant bacterial polysaccharide [mean log10 CFU per gram of kidney = 3.3 ± 1.1 (SEM) and 3.7 ± 1.1 (SEM), respectively]. In contrast, PNSG-specific rabbit antibodies significantly reduced S. aureus 5836 in infected kidneys [mean log10 CFU per gram of kidney = 0.71 ± 0.3; P ≤ 0.03, one-way analysis of variance (ANOVA) and Fisher probable least significant difference]. CP-specific IgG also did not prevent infection by the representative CP5 and CP8 S. aureus strains Reynolds and MN8 (8).

The protective efficacy of antisera to PNSG against lethal challenge (5 × 107 CFU per mouse) with two MRSA-VISA strains was also tested. None of the 10 infected mice given antibodies to PNSG died (five mice per strain). In contrast, 80% (4/5; P= 0.02, Fisher exact test) of mice infected with strain 5836 and 100% (5/5; P = 0.004, Fisher exact test) of mice infected with strain 5827 died after administration of control antibodies specific to an unrelated bacterial polysaccharide from P. aeruginosa.

Evidence for in vivo production of PNSG after injection of mice with PNSG-low S. aureus challenge strains was investigated. As PNSG is only soluble at pH < 4 (5), kidney homogenates were extracted with 0.1 M HCl to solubilize PNSG, which was then precipitated at neutral pH for probing with PNSG-specific antibodies. Only extracts from kidneys infected with S. aureus reacted with antibodies to PNSG (8). ELISA inhibition assays confirmed that PNSG elaboration was induced during in vivo infection, but expression was reduced or lost after in vitro passage of these bacterial cells on trypticase soy agar (TSA) (Fig. 4A) (19). Increased detection of PNSG expression on the mouse-kidney isolates compared with the fresh human blood isolates was attributed to environmental differences in these tissues that affect induction and stabilization of PNSG expression in vivo. Serum antibody responses to PNSG after S. aureus infection were also analyzed. Between 6 and 14 days after infection, specific IgM antibody responses to PNSG were detected (Fig. 4B), and all of the mice had high concentrations (>105 CFU per gram of kidney) of S. aureus in their kidneys on day 14. In contrast to the vigorous IgG responses elicited in PNSG-immunized mice and rabbits, no IgG responses to PNSG were detected by day 14 in any of the infected animals, possibly explaining the persistence ofS. aureus infection in spite of the immune response to PNSG. Finally, the results of immunoelectron microscopy revealed thatS. aureus cells cultivated directly from infected kidneys produced extracellular PNSG (Fig. 4C).

Figure 4

Detection of PNSG expression during and after infection. (A) ELISA inhibitions for detection of PNSG were performed on S. aureus strains before infection of mice (solid bars; percentage of inhibition value of 0 is shown only with error bar) and on strains freshly isolated from kidney homogenates (diagonal-stripped bars) and after passage on TSA one (cross-hatched bars), two (horizontal-lined bars), three (speckled bars), and four (clear bars) times. Staphylococcus aureus production of PNSG is lost after passage on TSA. Bars depict mean percentage of inhibition of antibody binding and error bars the SE. (B) PNSG elicits specific antibody during S. aureus infection. Mice infected IV with four different strains of S. aureus (Reynolds, solid bars; MN8, speckled bars; 5827, diagonal-stripped bars; and Por, cross-hatched bars) developed PNSG-specific IgM antibodies on the indicated day after infection. Bars depict mean absorbance (A405) and error bars the SE achieved with a serum dilution of 1:100. All postinfection values were significantly different (P < 0.01, ANOVA and Fisher PLSD) from preinfection (day 0) values. (C) Microscopic evidence for PNSG production during animal infection (29). Probing of S. aureus cells growing on primary platings of kidney homogenates from infected mice (postinfection) with antisera to purified PNSG followed by gold-labeled protein A showed binding of the gold particles to an extracellular antigen surrounding the bacterial cells.Staphylococcus aureus cells used to infect the mice (preinfection) showed minimal binding of antibody to PNSG, with small clusters of surface-localized gold particles seen on a minority of cells for most of the strains, as shown in the figure. Bars, 0.5 μm.

Our findings indicate that all S. aureusstrains examined had the ica genes needed for PNSG synthesis but that PNSG was expressed as a surface polysaccharide antigen principally in vivo during human and animal infection where it was a target for protective antibodies. Several in vivo expression strategies have been specifically used with S. aureus (20), although none identified the ica locus in the screens. This may be due to the small number of genes characterized against a background of a large number of candidate genes identified by these strategies. In addition, PNSG may not be made under the conditions used in the various in vivo expression screening approaches. In S. epidermidis, expression of PNSG potentiates infection but PNSG-negative strains can still cause infections if a high enough challenge dose is used (21).

Development of a vaccine for S. aureus is considered a high priority, and current candidates include the CP5 and CP8 microcapsules (18, 22), the recently described RNAIII activating protein that regulates the production of manyS. aureus virulence factors (23), and native and recombinant fragments of S. aureus protein adhesins for host extracellular matrix proteins (24). However, an advantage of a staphylococcal PNSG vaccine would be that PNSG is also elaborated by the majority of clinically important isolates of coagulase-negative staphylococci (CoNS) (25). Together,S. aureus and CoNS account for 40 to 60% of bacterial blood isolates from hospitalized patients (1). PNSG is effective in laboratory animals as a vaccine against CoNS infections (26). Thus, PNSG has potential as a vaccine for protection against hospital-acquired staphylococcal infections, community-acquired S. aureus infections, and infections in farm animals, where staphylococcal diseases have substantial economic impact (2).

  • * To whom correspondence should be addressed. E-mail: gpier{at}


View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article