This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lindahl, G. Articles by Areschoug, T. Search for Related Content PubMed PubMed Citation Articles by Lindahl, G. Articles by Areschoug, T.

Surface Proteins of Streptococcus agalactiae and Related Proteins in Other Bacterial Pathogens

Department of Medical Microbiology, Dermatology and Infection, Lund University, Sölvegatan 23, Lund, Sweden

SUMMARY

INTRODUCTION

PATHOGENESIS OF S. AGALACTIAE INFECTIONS

Neonatal and Other Infections

Early-onset and late-onset disease.

Incidence of neonatal disease.

Animal models.

Known virulence factors.

THE POLYSACCHARIDE CAPSULE AND SEROLOGICAL TYPES

GENOMICS

Genome Sequences and Comparative Genomics

Predicted Surface Proteins

A Putative High-Virulence Clone

SURFACE PROTEINS

THE Alp FAMILY OF PROTEINS

A Family of Highly Repetitive Proteins That Elicit Protective Immunity

Four Members of the Family: the {alpha}, Rib, R28, and Alp2 Proteins

The {alpha} protein.

The Rib protein.

The R28 protein (Alp3).

The Alp2 protein.

Proteins in the Alp family have chimeric but stable sequences.

Immunological and Immunochemical Properties

Variation in repeat number and its effect on immunological properties.

Immunological relationship between Alp family members.

Laddering pattern in gels.

Biological Function

Virulence of bacterial mutants.

Interactions with human epithelial cells.

Related Proteins in Other Bacterial Species

Esp in E. faecalis and a closely related protein in E. faecium.

Other related proteins.

THE C5a PEPTIDASE

General Properties

Molecular biology of S. pyogenes C5a peptidase (ScpA).

Molecular biology of S. agalactiae C5a peptidase (ScpB).

Role of enzymatic activity.

Role in fibronectin binding and cellular invasion.

Role in virulence.

Immunization studies.

THE ß PROTEIN

General Properties and Sequence

Binding of human IgA-Fc and factor H.

Immunization studies.

Related proteins in other bacterial species.

THE Lmb PROTEIN

The Lmb Proteins of S. agalactiae and S. pyogenes

Genetics of the region encoding Lmb and ScpB: linked genes for surface proteins.

OTHER SURFACE PROTEINS OF S. AGALACTIAE

The Fibrinogen-Binding FbsA Protein

The Sip Protein

CspA and Other Surface-Localized Enzymes

The Spb1 Protein

Additional Surface Proteins, Including R and X Antigens

REGULATION OF SURFACE PROTEIN EXPRESSION

S. AGALACTIAE SURFACE PROTEINS AS VACCINE COMPONENTS

Protein Vaccines and Conjugate Vaccines

CONCLUDING REMARKS AND PERSPECTIVES

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
Gram-positive cocci cause some of the most common infections in humans (69, 96, 166, 237). Major efforts are therefore under way to find new ways to prevent and treat infections caused by these pathogens, to analyze pathogenetic mechanisms, and to develop efficient vaccines. While much work is devoted to studies of polysaccharide capsules (7, 186, 266), surface proteins are attracting increasing interest for studies of pathogenesis and as potential vaccine components.

Here, we review current knowledge about surface proteins of Streptococcus agalactiae (group B streptococcus [GBS]), the most important cause of invasive bacterial disease in newborns (60). This bacterium is part of the normal vaginal flora of many women and is therefore strategically located to cause serious infections in neonates, whose immune response is depressed compared to older children and adults (60, 63). In addition to its significance as a pathogen, S. agalactiae is a valuable model system for analysis of infections caused by encapsulated pathogens and for analysis of infections in the perinatal period.

Surface proteins of S. agalactiae are likely to play important roles during different stages of an infection and also hold promise as vaccine components (33, 140). Moreover, S. agalactiae surface proteins are of interest for use in the analysis of some major problems in medical microbiology, such as adhesion to epithelial cells, interactions with human extracellular matrix or plasma proteins, and escape from host immunity. This review focuses on the S. agalactiae surface proteins that have so far been purified and characterized at the molecular level, with emphasis on proteins that are known to elicit protective immunity and therefore are of interest for vaccine development. To provide the proper background, the first part of the review briefly describes the pathogenesis of S. agalactiae infections, the polysaccharide capsule, and current knowledge about S. agalactiae genomics. A detailed description of S. agalactiae infections is beyond the scope of this article, but excellent reviews are available (60, 179).

By tradition, S. agalactiae has most commonly been referred to as group B streptococcus or GBS, but in recent years there has been a tendency to use the Latin name for this pathogen, as for other pathogens, and this practice is followed here.

PATHOGENESIS OF S. AGALACTIAE INFECTIONS

Top Previous Next References

 

S. agalactiae is found in the vaginal and/or rectal flora of 15 to 40% of adult women, and children born to these women may develop disease due to exposure to the bacteria before birth or during the neonatal period (60, 220). These neonatal S. agalactiae infections are divided into early-onset and late-onset infections. Early-onset infection, which is the most common type of neonatal S. agalactiae disease, occurs within the first week of life, while late-onset infection occurs in infants between 1 week and 3 months of age (60, 220).

Early-onset and late-onset disease. In early-onset disease, the neonate is infected by exposure to S. agalactiae before or during birth (60). Recent mortality rates of 4 to 6% have been reported in the United States (218). Some early-onset infections may occur when the neonate is exposed to S. agalactiae during passage through the birth canal, but most early-onset infections are probably caused by ascending spread of the organism from the maternal genital tract through ruptured membranes into the amniotic fluid, in which the bacteria multiply, allowing them to colonize the respiratory tract of the fetus. In some cases, transmission of S. agalactiae into the amniotic fluid may even occur through intact membranes. After bacterial entry into the respiratory tract, pneumonia may develop and the bacteria may further spread to the bloodstream and cause septicemia. Bloodstream dissemination allows the bacteria to reach multiple body sites, where subsequent tissue penetration may result in manifestations such as meningitis and osteomyelitis (60, 181, 211). Thus, S. agalactiae probably has to adhere to, invade, and transcytose several epithelial/endothelial cell barriers to cause disease.

In an important experimental study mimicking early-onset disease, S. agalactiae was used for intra-amniotic inoculation of pregnant Macaca nemestrina primates; this was followed by analysis of infected tissues (76, 211). Interestingly, bacteria were observed in vacuoles within fetal lung epithelial, endothelial, and fibroblast cells. In agreement with this observation, S. agalactiae has been demonstrated to invade human epithelial and endothelial cells (76, 118, 180, 212, 260, 271) and even macrophages (259) in tissue culture, and bacteria may also survive intracellularly (76, 212, 259). Thus, invasion of human cells may play an important role in pathogenesis. However, all available evidence indicates that S. agalactiae should be viewed as an extracellular pathogen, although it might survive intracellularly at certain stages of an infection. In contrast to a typical intracellular pathogen such as Listeria monocytogenes, there is little evidence that S. agalactiae can replicate intracellularly (76, 212, 271). Moreover, it should be noted that the ability to invade human cells in vitro may not always reflect the in vivo situation. For example, Staphylococcus aureus readily invades human cells in tissue culture but was not found intracellularly in an animal model (161). However, the observations in the monkey model studied by Rubens et al. (211) indicate that the ability of S. agalactiae to invade cells in culture is indeed of relevance to human infections.

Late-onset neonatal disease is less common than early-onset disease but is becoming relatively more important because its incidence has not declined as a result of the prophylactic measures that have caused a decrease in the incidence of early-onset infections (see below). Little is known about the pathogenesis of late-onset infections, but vertical transmission of the organism from the mother to the infant probably explains most infections during this period. In some cases, late-onset disease occurs as nosocomial epidemics that primarily affect preterm neonates, who probably acquire the bacterium by horizontal transfer from nursery personnel. The two most common clinical manifestations of late-onset disease are meningitis and bacteremia without a focus of infection. Fatality ratios for late-onset disease are generally lower than those for early-onset infections and have been reported to range from 2 to 6% (60, 218, 220).

Importantly, a substantial proportion of neonates who survive S. agalactiae infection suffer from sequelae. After cases of neonatal meningitis, neurological sequelae occur in up to 50% of the survivors and include mental retardation, cortical blindness, deafness, uncontrolled seizures, hydrocephalus, hearing loss, and speech and language delay (60).

Because illness develops in only a small fraction of neonates colonized by S. agalactiae during birth, many studies have tried to identify factors that may increase the risk for development of disease. Rupture of membranes before labour onset and increased interval between membrane rupture and delivery are considered to be major risk factors. Other predisposing factors include premature delivery, low birth weight, dense vaginorectal colonization, and intrapartum fever (60). The infant's susceptibility to S. agalactiae infection may also be enhanced by low levels of maternal antibodies against the polysaccharide capsule (8) or against surface proteins (C. Larsson and G. Lindahl, unpublished data).

Incidence of neonatal disease. The overall incidence of invasive neonatal S. agalactiae disease has recently been estimated to be 0.6 to 1.8 cases per 1,000 live births in the United States (60, 218) and 0.7 cases per 1,000 live-births in the United Kingdom and Ireland (90). Mortality rates were reported to be between 4 and 6% in the United States (218) and 10% in the United Kingdom (90). Interestingly, it has been suggested that the incidence of neonatal disease is considerably greater than reported, because the requirement for positive cultures from blood or cerebrospinal fluid may underestimate the true burden of disease (151). Of note, neonatal S. agalactiae disease is more common than some other well-known diseases affecting newborns, such as rubella and spina bifida.

The incidence of neonatal S. agalactiae disease declined considerably in the United States during the 1990s, probably because of the introduction of surveillance programs and intrapartum antibiotic prophylaxis (169, 217, 218). However, the widespread use of antibiotic prophylaxis to inhibit S. agalactiae disease may be problematic, because it has been accompanied by an increase in the incidence of early-onset sepsis caused by Escherichia coli (239), a finding that stresses the need for an S. agalactiae vaccine (63).

Animal models. Like many other pathogens, strains of S. agalactiae isolated from human infections appear to be adapted to their human host, and there is no good animal model. Most S. agalactiae strains isolated from cases of bovine mastitis even have properties distinct from human isolates (21, 60). For example, molecular analysis has shown that most bovine strains lack genes for the surface proteins ScpB and Lmb, which are found in all human isolates, and lack the gene for the immunoglobulin A (IgA)-binding ß protein, which is common among human isolates (55, 73). Moreover, human isolates may express components that specifically interact with the human host. For example, the surface-localized C5a peptidase of S. agalactiae (ScpB) may degrade human C5a but not mouse C5a (23, 28) and the IgA-binding ß protein binds human IgA but probably not rat IgA (146, 214). Nevertheless, the mouse and the rat are valuable model systems for analysis of S. agalactiae infections, and several bacterial components are known to act as virulence factors in these models. However, results obtained with animal models must be interpreted with caution, and lack of evidence that a bacterial component contributes to virulence in an animal model does not exclude the possibility that it is of major importance in human infections. In the future, the use of transgenic mice may facilitate studies of some aspects of the infectious process, as described in other systems (100, 143, 207).

Of note, infection of mice (or rats) is usually performed by the intraperitoneal (i.p.) route, a situation thought to at least partially reflect invasive infection in humans. Moreover, several investigators have used mouse or rat pups for i.p. infection, because S. agalactiae is of particular importance as a pathogen in the neonatal period (209). However, all i.p. infections bypass the adhesion step that probably is essential in natural infections, so a bacterial mutant affected in adhesion may have normal virulence in an i.p. infection model.

Known virulence factors. Several S. agalactiae virulence factors have been identified in the mouse and rat models. In particular, the polysaccharide capsule and the secreted hemolysin are of major importance for virulence (57, 178, 213, 263). Moreover, superoxide dismutase and D-alanylated lipoteichoic acid play significant roles (199, 200). In contrast, the use of animal models has provided little evidence that surface proteins contribute to virulence, although a recent report indicates that the surface-localized protease CspA makes a contribution (87). This situation may at least partially reflect species specificity in the interactions between surface proteins and the infected host.

In an interesting study of a strain of capsular type Ia, Jones et al. (113) used signature-tagged mutagenesis (STM) and a neonatal rat model of i.p. infection to identify novel S. agalactiae genes implicated in virulence. Most of the genes identified were predicted to affect transport, regulation, intermediate metabolism, and cell wall metabolism, stressing the importance of such functions for virulence. In contrast, only one typical surface protein was identified, a result that may be due to the inherent limitations of the animal model used. Moreover, many surface proteins may contribute to adherence and colonization, steps that were bypassed in the STM model (113). The surface protein identified was the laminin-binding Lmb protein or a closely related protein recently identified in the S. agalactiae genome (113, 242). In addition, the STM screening identified a mutant with strongly decreased virulence that lacks the penicillin-binding protein PBP1a (113), which may be surface exposed and contribute to phagocytosis resistance (114).

THE POLYSACCHARIDE CAPSULE AND SEROLOGICAL TYPES

Top Previous Next References

 
All clinical isolates of S. agalactiae express a polysaccharide capsule; nine different capsular types have been identified so far (123). From the early days of S. agalactiae research, much interest has been focused on this capsule, because anticapsular antibodies confer protective immunity in an animal model and because capsular typing is epidemiologically important (60, 133, 134). The capsule is a major virulence factor with antiphagocytic function (158, 213, 263), and attempts are in progress to develop a multivalent capsular conjugate vaccine (9, 192).

The four "classical" capsular serotypes, identified by Lancefield, are types Ia, Ib, II, and III. When S. agalactiae emerged as an important neonatal pathogen, strains of serotype III predominated as the cause of serious infections, in particular as the cause of meningitis (60). However, in some (but not all) recent surveys, type Ia strains have been equally common among isolates from invasive infections, although type III remains the most important cause of late-onset infection and meningitis (88, 117, 145). In addition, strains of type V have recently emerged as a significant cause of S. agalactiae infection (22) and strains of types VI and VIII are the most common strains isolated from healthy women in Japan (131).

With regard to surface proteins, the subject of this review, the capsule is of particular relevance for at least two reasons. First, expression of a given surface protein is often correlated with the capsular type (6, 89, 112, 124, 130, 163, 236). Such associations may represent evolutionary lineages and could lack functional significance, but an interesting hypothesis predicts that immune selection has favored strains with certain combinations of surface structures because of their greater fitness (83). Second, the capsule might have been expected to interfere with the function of antibodies directed against surface proteins, either by blocking the access of antibodies or by interfering with antibody effector functions. However, this does not appear to be the case, because antibodies to surface proteins do protect against infection, at least in an animal model, a situation that encourages efforts to develop a protein-based vaccine. It is possible that the capsule does not block opsonization because it is thinner in vivo than in vitro, because it is expressed only in certain growth phases, and/or because it is very loose and flexible.

GENOMICS

Top Previous Next References

 

Comparison of the S. agalactiae genome with the genomes of two other pathogenic streptococci, Streptococcus pyogenes and Streptococcus pneumoniae (66, 253), showed that most of the predicted S. agalactiae proteins have homologs in at least one of the two other species (78, 252). As summarized in Fig. 1, the genome of the sequenced type V strain is predicted to comprise 2,144 protein-encoding genes, of which 683 are unique to S. agalactiae while 1,060 are shared with both of the other genomes. In addition, 401 putative proteins of this S. agalactiae strain are shared with one, but not both, of the other species (not indicated in Fig. 1). Moreover, the chromosomal order of genes is highly conserved between S. agalactiae and S. pyogenes, stressing the relatedness between these two species, while the gene order showed low conservation between S. agalactiae and S. pneumoniae, possibly reflecting the importance of transformation and recombination in the evolution of S. pneumoniae (78).

View larger version (36K): [in this window] [in a new window]   FIG. 1. Comparison of protein-encoding genes in the genomes of three streptococcal species: S. agalactiae, S. pyogenes, and S. pneumoniae. The total number of predicted protein-encoding genes in a bacterial species is given below the name of that species. The numbers of proteins unique to each species are given in the corresponding shaded areas, while the number of proteins shared by all three species is indicated in the innermost part of the figure. For example, S. agalactiae was predicted to have 2,144 protein-encoding genes, of which 683 are unique to this species, while 1,060 protein-encoding genes have homologs in both of the two other streptococcal species. In addition, 401 S. agalactiae proteins have homologs in only one of the two other species (not shown). The comparison is based on three published genome sequences (66, 252, 253). Modified from reference 252 with permission of the publisher.

The two S. agalactiae genomes were predicted to encode 21 or 24 LPXTG proteins, respectively (78, 252), and may also encode some proteins with LPXTG-related motifs (78). The type III genome (78) was predicted to encode 36 lipoproteins, while as many as 51 putative lipoproteins were identified in the type V genome (252). A separate bioinformatic analysis of the two genomes identified a core of 39 common putative lipoproteins, along with 2 proteins unique to the type III strain and 5 proteins unique to the type V strain (242). For many of these predicted surface proteins, hypothetical functions can be predicted from comparisons with similar proteins in other bacteria. The predicted LPXTG proteins include putative adhesins and enzymes, but most of the LPXTG proteins identified from genome sequences have no obvious function (78, 252). The predicted lipoproteins include components of ABC transport systems, putative adhesins, cation-binding proteins, enzymes, and components involved in protein localization and folding (78, 242, 252).

Concerning surface proteins that have been purified and characterized, proteins with an LPXTG-like motif have so far been most extensively studied, as described in this review. Among these proteins are putative or known adhesins and several enzymes. Of note, many of the LPXTG proteins of S. agalactiae are not present in S. pyogenes or S. pneumoniae, implying that each pathogen possesses a specific repertoire of LPXTG proteins (78).

In addition to the predicted LPXTG-containing proteins and lipoproteins mentioned above many proteins might be located on the surface, as suggested by immunological analysis and by comparison with known surface proteins identified in other bacterial species (78, 252). Indeed, an immunological analysis reported by Tettelin et al. (252) indicated that the type V strain studied by them expressed more than 55 different surface proteins, of which only a minority were LPXTG proteins or lipoproteins. Some of these putative surface proteins may be anchored to the bacterial cell wall by novel mechanisms (41, 46). For example, analysis of the type V genome identified a putative secreted metalloprotease containing so-called GW repeats, which have been implicated in the binding of some Listeria monocytogenes surface proteins to lipoteichoic acid (116). Moreover, it has been reported that several enzymes usually considered to be intracellular may also be expressed on the S. agalactiae surface (98). It will be of obvious interest to analyze all of these predicted surface proteins to determine their role in pathogenesis and evaluate their suitability as possible vaccine components.

Several reports indicate that serotype III strains in one clone are strongly enriched among invasive-disease isolates, implying that strains in this clone have increased pathogenic potential and are of particular relevance for studies of pathogenesis (20, 115, 172, 245). This putative high-virulence clone was first described by Musser et al. (172), and the available evidence indicates that their clone ET-1 is identical to clones designated III-3 (245), ST-17 (115), or GIII (20, 70). Interestingly, a recent analysis employing MLST indicated that this putative high-virulence clone (ST-17) corresponds to a bovine lineage introduced among humans (21). However, analysis of two strain collections did not support the conclusion that this type III clone has increased virulence (53, 89, 93). Thus, the situation is unclear, possibly reflecting different epidemiological settings in the different geographical areas where the strains were collected (93). Of note, neither of the two S. agalactiae genome sequences published so far represents the putative high-virulence type III clone (78, 252).

There is some evidence that strains in the putative high-virulence clone have unique surface properties. These strains have significantly increased sialic acid content, a property that might contribute to increased virulence (245, 246). Moreover, subtractive hybridization of DNA from a strain in the III-3 clone has allowed the identification of a novel surface protein, Spbl, found only in strains of that clone (1, 27). With regard to surface properties, it may also be of relevance that most strains in the putative high-virulence clone carry the group II intron GBSi1 inserted between the genes encoding the two surface proteins ScpB and Lmb, suggesting that expression of these proteins could be altered (79, 247).

SURFACE PROTEINS

Top Previous Next References

 
Historically, the first surface protein antigen identified in S. agalactiae was the c antigen, which was detected with antisera raised against whole bacteria (268). The c antigen (also designated Ibc) has been found in many strains but not in the clinically important type III strains (67, 163). Characterization of this antigen subsequently showed that it is composed of two unrelated protein components, the trypsin-resistant protein and the trypsin-sensitive ß protein (18), both of which have now been extensively characterized, as described below. Of note, a strain that is reported to express c antigen may express either or both of the and ß proteins. Because and ß together constitute the c (or C) antigen, the designations alpha C and beta C, and other related designations, have also been used for these proteins. Moreover, the designation ACP (alpha C protein) was recently used for the protein (11). It is not known why the and ß proteins are commonly expressed by the same S. agalactiae strain (16, 112, 163).

Lancefield et al. (136) made the important observation that antibodies to the c antigen confer protection against S. agalactiae infection in a mouse model, implying that not only the capsule but also surface proteins may elicit protective immunity. As pointed out by Ferrieri (67), this finding remained relatively ignored, possibly because the c antigen is expressed by only some strains and not by the clinically important type III strains. However, early work by Bevanger and Naess (19) demonstrated that each of the and ß proteins, the two components of the c antigen, elicit protective immunity. Subsequent work on type III strains identified Rib, a novel surface protein that elicits protective immunity and is expressed by most strains not expressing (236). This finding suggested that an S. agalactiae vaccine based solely on surface proteins might be developed and focused interest on surface proteins for analysis of S. agalactiae pathogenesis.

Below, we summarize work on those S. agalactiae surface proteins that have so far been studied most extensively. Of note, the genes encoding these different surface proteins are widely distributed over the S. agalactiae chromosome, except for the closely linked scpB and lmb genes, which are located on a putative composite transposon (73).

THE Alp FAMILY OF PROTEINS

Top Previous Next References

 

Four members of the Alp family have so far been identified: the , Rib, R28, and Alp2 proteins (Fig. 2A). In addition to their relevance for vaccine development and for analysis of S. agalactiae pathogenesis, these proteins are of interest as models for other repetitive proteins, which are common in gram-positive bacteria (120, 177) and in higher organisms (162, 240, 262). Because Alp-related proteins have been identified in several bacterial pathogens, studies of Alp family proteins are also of general interest for the analysis of pathogenetic mechanisms.

View larger version (34K): [in this window] [in a new window]   FIG. 2. The Alp family of proteins. (A) Comparison of the four known members of the Alp protein family: the , Rib, R28, and Alp2 proteins. S, signal peptide; N, nonrepeated N-terminal region; R, repeat region; C, C-terminal region. The number of amino acid residues in each region and the percent residue identity (bold numbers) between corresponding regions are indicated. Note that the designation Alp3 has been used as an alternative name (130) for the R28 protein (235). Each of the three proteins , Rib, and R28 has 9 to 12 repeats with a length of 79 or 82 residues, as indicated. These repeats are identical within one protein but not between proteins, and the number of repeats varies among clinical isolates. The nonrepeated region of R28 contains a 195-residue subregion, designated RN, which does not fit into the alignment with Rib. This region is 99% identical to a type of repeat found in the N-terminal half of the Alp2 protein. Thus, the Alp2 protein contains two types of tandem repeat (designated RN and R), unlike the other members of the family. This figure is based on published sequences for the four proteins (78, 130, 165, 235, 261). The data for Alp2 are based on the sequence reported by Glaser et al. (78). (B) Schematic representation of the R28 protein. As indicated, R28 can be viewed as a chimera, derived from the two Alp family proteins and Rib and the unrelated ß protein (235). Predictions about the three-dimensional structure are given below the protein. The region that shows similarity to ß corresponds to a region in ß predicted to have an IgSF fold (12), and our analysis of the R28 region suggests that it has a similar structure. The repeats in the C-terminal part of R28 have been proposed to have a fold related to the IgSF fold (35). Modified from reference 235 with permission of the publisher. (C) Immunological relationship among the four known members of the Alp protein family. Solid arrows indicate immunological cross-reactivity, and broken arrows indicate lack of cross-reactivity. For example, the protein cross-reacts with Alp2 but not with Rib or R28. Most of the data summarized in the figure were obtained with rabbit antisera raised against purified proteins (132, 234-236), but the relationship between R28 and Alp2 was analyzed with an extract of strain D136C (NEM316), which expresses Alp2 (78, 234). (D) Laddering pattern in the , Rib, and R28 proteins analyzed by SDS-PAGE. Purified preparations of the three proteins, and the control protein ß (which is not a member of the Alp family), were subjected to SDS-PAGE under ordinary conditions (with boiling of samples at pH = 7) or after boiling of samples for 15 min at pH = 4. Under the latter conditions, the three Alp family proteins form a regularly laddered pattern due to hydrolysis of acid-sensitive Asp-Pro bonds in the repeats during sample preparation. Modified from reference 235 with permission of the publisher.

Except for the Alp2 protein (which appears to be rare, as described below) there is only one type of repeat in each protein. This type of repeat, designated R in Fig. 2A, is present in all four proteins. Within a given protein, these repeats are identical or virtually identical, but they vary in sequence among different Alp proteins. The residue identity between different repeats varies from 95% for the pair Rib-R28 to 40% for the pair -Alp2. These repeats have a length of 76 to 82 amino acid residues, and the number of repeats in a protein varies among strains (see below). The Alp2 protein differs from other members in the family in having a second type of tandem repeat (here designated RN) in the N-terminal half of the protein (Fig. 2A).

Interestingly, it has been predicted that the repeats in Alp family proteins have a structure related to the Ig-like fold (35), a structure identified also in some other repetitive prokaryotic proteins (86, 152, 160, 198). In agreement with this sequence analysis, computer modeling of the Rib repeat predicts that it contains at least four ß-sheets in a compact structure (E. Lindahl and G. Lindahl, unpublished data). Because domains with the Ig-like fold are often implicated in molecular recognition, these data suggest that Alp family proteins may have such a function (35). Alternatively, the repeat region may act as a rod needed for exposure of a unique ligand-binding region, by analogy to the situation in intimin from E. coli and invasin from Yersinia, two proteins in which the repeat regions have Ig-like fold (86, 152).

Four Members of the Family: the , Rib, R28, and Alp2 Proteins

The protein. The protein is unrelated to the ß protein, the other component of the c antigen (92, 107), and is commonly found in strains of serotypes Ia, Ib, and II but almost never in type III strains and only rarely in type V strains (6, 16, 67, 112, 124, 163, 236, 243) (Table 1). The protein is resistant to trypsin, a property that contributed to its identification and has been used to analyze strains for expression of (67, 163, 268). Bevanger and Naess (19) reported that purified from streptococci elicits antibodies that protect mice against infection, a result that was confirmed in experiments with recombinant protein (164).

The gene for was cloned from strain A909, a commonly used reference strain of serotype Ia. Sequence analysis of the -encoding gene, named bca (not to be confused with bac, the gene encoding the S. agalactiae ß protein [92]) revealed that a large part (75%) of the protein was composed of nine 82-residue tandem repeats, for which the sequence was also identical at the nucleotide level (165). The calculated molecular mass for the mature protein (i.e., the protein without the 56-residue signal sequence [236]) is 103 kDa. Of note, the repetitive structure of the bca gene made the sequencing difficult (165), a problem also encountered during sequencing of genes encoding other members of the Alp family (235, 261). Because these genes cloned in E. coli are unstable, sequencing of such cloned genes may underestimate the number of repeats (261). However, the true number of repeats may be estimated from ladder patterns in PCR or sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (see below).

The Rib protein. The Rib protein was identified as a unique high-molecular-weight protein in cell wall extracts of type III strains (236). Characterization of the purified protein indicated that it was related to , although the two proteins did not cross-react immunologically. In particular, the N-terminal sequences of the two purified proteins were found to be related but not identical. Moreover, Rib was resistant to trypsin, varied in size among clinical isolates, and gave rise to a ladder pattern in Western blots, like the protein. Most importantly, antibodies to Rib conferred protective immunity against lethal infection with Rib-expressing strains (236). The available evidence indicates that Rib is expressed by the large majority of serotype III strains, by many type II strains, and by a few type V strains (6, 139, 236, 252).

The sequence for Rib was determined in strain BM110, a member of the putative high-virulence type III clone (261). The processed form of Rib in this strain has a 174-residue N-terminal region and 12 identical 79-residue repeats and has a calculated molecular mass of 123 kDa (Fig. 2A). As previously reported for , the repeats of Rib are identical even at the nucleotide level (165, 261). The amino acid residue identity to is 61% in the N-terminal region and 47% in the repeat region. Thus, the sequencing of Rib showed that Rib and are members of a family of streptococcal surface proteins with extremely repetitive sequence (261). Of note, the genome of the sequenced type V strain (252) has the gene (rib) encoding Rib. The sequence deduced for Rib in that strain is completely identical to the sequence previously determined in a type III strain (261), except that the type V protein has 14 repeats. The paper on the type V sequence (252) reports that the Rib protein of that strain has repeats with a length of only 67 amino acid residues, but that conclusion is due to a mistake during analysis of the nucleotide sequence reported in the same paper.

The R28 protein (Alp3). The R28 protein was first identified in strains of S. pyogenes (group A streptococcus) (137), and molecular characterization demonstrated that it is a member of the same family as Rib and (235). The nonrepeated N-terminal region in R28 is considerably longer than the corresponding regions in and Rib, while the repeat region is closely related (95% identity) to that in Rib. This similarity between R28 and Rib may explain an immunological cross-reactivity observed in early studies (137). The R28 protein is expressed not only by some strains of S. pyogenes but also by many S. agalactiae strains of serotypes V and VIII (6, 130, 131), and these S. agalactiae strains express R28 proteins that are identical except for different number of repeats (124, 130). Lachenauer et al. (130) used the novel name Alp3 for the protein expressed by S. agalactiae, but because the proteins expressed by S. pyogenes and S. agalactiae are 98% identical, we prefer to retain the original name R28 for both proteins. Like other members of the Alp protein family, R28 elicits protective immunity (234, 235).

The R28 protein may be viewed as a chimera derived from the three S. agalactiae proteins , ß, and Rib (Fig. 2B; also see below). This finding suggests that the gene encoding R28 may have arisen in S. agalactiae, followed by horizontal gene transfer to S. pyogenes (235).

Interestingly, there is some evidence that the R28 protein may have played a pathogenetic role in the epidemics of puerperal fever (childbed fever) that caused the death of numerous parturient women during earlier centuries and represents the classical example of a nosocomial infection (3, 235). These epidemics were due to infections with S. pyogenes (135) and should not be confused with neonatal infections caused by S. agalactiae, but in both cases the infection was most probably initiated by bacteria initially present in the vagina. Discussion of puerperal fever is beyond the scope of this review, but the putative association between R28 and nosocomial infections may be of relevance for studies of R28-related proteins such as Esp of the nosocomial pathogens Enterococcus faecalis and E. faecium (see below).

The Alp2 protein. The Alp2 protein was identified in a serotype V strain (132) and has a sequence virtually identical to that of R28 in the N-terminal two-thirds (130) (Fig. 2A). This protein is more rare than the other members of the Alp family and was found in only a few strains of serotypes Ia, III, and V (78, 124, 130). Unlike the other Alp family proteins, Alp2 has two types of tandem repeat. One type of repeat, here designated RN, corresponds to a nonrepeated part of R28 and includes the region in R28 that is related to the ß protein (235) (Fig. 2B). The sequenced variants of Alp2 have one or two such repeats (78, 130). The second type of tandem repeat, located in the C-terminal half of Alp2 and here designated R, corresponds to the repeats identified in the other members of the Alp family. For Alp2, the number of such repeats varies between two and four in different S. agalactiae strains (78, 130). The most C-terminal part of Alp2 includes a single region most similar to repeats in the protein. Thus, the sequence organization in the Alp2 protein is more complex than in the other members of the Alp family. Of note, the genome of the sequenced type III strain (78) contains the gene encoding Alp2.

Proteins in the Alp family have chimeric but stable sequences. The known proteins in the Alp family appear to represent stable and distinct entities. For example, the Rib protein expressed by two S. agalactiae strains of serotypes III and V have identical sequences except for different number of repeats (252, 261) and the R28 protein sequenced in S. pyogenes is 98% identical to that sequenced in several type V and type VIII strains of S. agalactiae (130, 235). Moreover, partial sequencing of the four known Alp family members in many S. agalactiae strains indicated that these proteins have very stable sequences (124). Thus, different members of the Alp family vary only a little or not at all in sequence between strains, except for variation in repeat number.

In spite of their stability, members of the Alp family appear to have a chimeric structure, as first observed for the R28 protein, which can be viewed as a chimera derived from the three S. agalactiae proteins , ß, and Rib (235) (Fig. 2B). Similarly, the Alp2 protein can be viewed as a chimera derived from R28 and other members of the Alp family (130). The most surprising result of this analysis is that one region in R28 shows sequence homology to a region in the S. agalactiae ß protein, which is not a member of the Alp family and has very different sequence organization (see below). Interestingly, the corresponding region in ß is predicted to be evolutionarily related to eukaryotic proteins and to have the Ig superfamily (IgSF) fold (12). The ß-like region in R28 may also have this structure (our unpublished data), suggesting that it corresponds to a mobile domain with the IgSF fold.

Although the chimeric structure of Alp proteins suggests that the corresponding genes can recombine with each other, and possibly also with other genes, the emergence of a new family member appears to be a rare event, because only four members of the family have been identified so far, and these proteins are genetically stable. The family members identified so far may have been selected because of their superior fitness. Nevertheless, it is conceivable that novel Alp family members will arise and become common in other epidemiological settings.

Variation in repeat number and its effect on immunological properties. For all members of the Alp protein family, the protein varies in size among different clinical isolates and the size variability can readily be explained by variation in the number of repeats (130, 153, 154, 235, 236). For example, the size of the intact protein varies between 65 and 165 kDa in different strains (153), corresponding to proteins with 8 to 19 repeat regions, and the size of the Rib protein varies between 65 and 125 kDa (236), corresponding to proteins with 8 to 15 repeat regions. Such size variation may arise through recA-independent slipped-strand mispairing during DNA replication (201) and is common in repetitive DNA sequences (150).

In spite of the extremely repetitive sequence, the size of proteins in the Alp family remains remarkably stable during growth in vitro, suggesting that the molecular events giving rise to size variability represent rare events (201, 236). However, rare size variants can be demonstrated to arise in vitro, as described in the system (201). In contrast, considerable size variability has been found in vivo, when bacteria infect a host pretreated with protective anti- serum (154). Interestingly, variants of with fewer repeats are selected under these conditions, suggesting that reduction in repeat number allows bacteria to escape immunity. In agreement with this hypothesis, an S. agalactiae strain expressing an protein with only one repeat was more virulent than a strain expressing the wild-type variant containing nine repeats when the virulence of the two strains was compared in mice pretreated with antibodies to the wild-type protein (82). However, no difference in virulence was observed in untreated mice, so these findings do not provide evidence that the repeats of affect virulence under nonimmune conditions. Importantly, naturally occuring variants of and other Alp proteins contain multiple repeats, suggesting that a long repeat region enhances virulence in humans. The appearance of variants with few repeats may therefore be interpreted as a side effect of host immunity, and it seems possible that it represents a dead end for the bacterium rather than an immune evasion mechanism. Nevertheless, this phenomenon may contribute to the size variation encountered among naturally occurring variants. Possibly, the average number of repeats in an Alp family protein represents the net result of a selective force (bacterial virulence) favoring proteins with many repeats and another selective force (host immunity) favoring variants with few repeats. While immune selection would result in rapid "contraction" of the repeat region, escape from such selection would allow expansion of the repeat region, which may be required for spread to a new host. In vivo, such dynamic selections might cause frequent size variability, resulting in the appearance of repeats that are identical even at the nucleotide level.

In addition to the possible role of the repeat region for immune evasion in an immune host, this region has been suggested to affect immune escape in another way, by reducing the immune response to the whole protein, and in particular to the N-terminal region, in a nonimmune host (80). This conclusion was based on the observation that the antibody response in mice to the N-terminal region of was drastically reduced when the number of repeats was increased (80).

Immunological relationship between Alp family members. The immunological relationship among different Alp proteins has implications for the development of an S. agalactiae vaccine and for the use of antisera for serological typing of strains. Cross-reactivities observed among the four Alp family members are summarized in Fig. 2C (130, 132, 139, 234-236). Most of these results were obtained with hyperimmune rabbit antisera raised against purified proteins, and they do not necessarily reflect the situation during a natural infection. In general, the cross-reactivities are surprisingly limited, given the extensive residue identity between different members of the Alp family. For example, the protein showed no cross-reactivity with any of the Rib and R28 proteins, in spite of 61% residue identity in the N-terminal regions and 45% identity in the repeat regions (235, 261). Similarly, the Alp2 protein did not cross-react with Rib, in spite of the 60 to 67% residue identity in different regions (132). Even more surprisingly, the cross-reactivity between Rib and R28 was limited, in spite of the 59% residue identity in the most N-terminal region and the 95% identity in the repeat region (234). These data indicate that the sequence variability could represent antigenic variation, allowing bacteria expressing one protein in the family to evade protective immunity directed against some of the other proteins.

From a vaccine development point of view, it is of relevance that even a limited cross-reactivity may be sufficient to confer cross-protection, as shown by work with the Rib and R28 proteins (234). In that study, vaccination of mice with purified Rib from S. agalactiae protected against infection with an R28-expressing strain of S. pyogenes and vice versa; i.e., the cross-reacting Rib and R28 proteins even elicited cross-protection between strains of two different species (234). Moreover, a very weak cross-reactivity that is hardly detectable by immunochemical techniques may allow some cross-protection in vivo, as suggested by work with the Rib and proteins in the mouse model (139). Of note, the cross-reactivities summarized in Fig. 2C indicate that a vaccine composed of the two proteins Rib and might protect against S. agalactiae strains expressing any of the four Alp proteins so far identified (140).

Laddering pattern in gels. A remarkable property of proteins in the Alp family is their ability to form a regular ladder pattern in SDS-PAGE and Western blots, as first described for the protein (153, 165) (Fig. 2D). Because of this property, the proteins in the family have often been referred to as laddering proteins. The distance between the steps in the ladder corresponds to the size of one repeat, and it was initially suggested that a novel mechanism may allow -protein fragments of different sizes to be synthesized (165). However, immunochemical analysis subsequently indicated that the laddering phenomenon most probably is due to hydrolysis of acid-sensitive Asp-Pro bonds in the repeats during sample preparation (261). Indeed, purified preparations of the and Rib proteins do not form a ladder in SDS-PAGE when samples are boiled under standard conditions at pH 7, while samples boiled at pH 4 give rise to the characteristic ladder pattern (Fig. 2D). Although these observations imply that the ladder pattern is an artifact, it is an interesting phenomenon that may aid in the identification of novel Alp family members. However, members of the Alp family may exist that lack acid-sensitive bonds in the repeats and therefore do not form ladders. Of note, the number of steps in a ladder observed on SDS-PAGE may be used to estimate the number of repeats in a protein (261). Laddering has also been observed during analysis of PCR products generated from genes encoding Alp proteins, and the distance between the steps in that ladder was estimated to correspond to one repeat (261). Such ladders, which may arise through slippage of the DNA polymerase during replication, can also be used to estimate the number of repeats in the gene.

Virulence of bacterial mutants. Use of an -negative mutant suggested that might have some effect on mouse virulence in the early parts of an infection, but it was not excluded that this result was due to a polar effect (144). Moreover, experiments with bacterial mutants lacking Rib or R28 have not provided evidence that these proteins act as virulence factors in the mouse (235; T. Areschoug and G. Lindahl, unpublished data). Together, these data suggest that proteins in the Alp family have little or no effect on virulence in the mouse model of invasive infection. Nevertheless, each of the different Alp proteins is a target for protective antibodies in the mouse system (19, 132, 235, 236). These findings may appear to be paradoxical but can be readily explained by assuming that Alp proteins act only as targets for antibodies in the mouse, allowing complement deposition and/or Fc-mediated phagocytosis.

Interactions with human epithelial cells. The R28 protein, which was first identified in S. pyogenes, promotes binding of that pathogen to the human cervical epithelial cell line ME-180, indicating that R28 may act as an epithelial cell adhesin (235). However, attempts to demonstrate that other members of the Alp family also act as adhesins have been negative. Thus, a mutant of S. agalactiae lacking the protein was not affected in binding to ME-180 cells (30), and similar results were obtained with a Rib-negative mutant (J. Waldemarsson, G. Lindahl, and T. Areschoug, unpublished data). However, the protein may promote invasion of ME-180 cells by S. agalactiae (30), possibly by interaction with host cell glycosaminoglycans (11).

View larger version (29K): [in this window] [in a new window]   FIG. 3. Proteins in other bacterial species related to members of the Alp protein family. (A) Schematic representations of six Alp-related bacterial proteins. The bacterial species and the name of the protein are indicated to the left. For all proteins, the C-terminal part includes a region with long tandem repeats showing sequence similarity to the C-terminal repeat region of Alp family members. The lengths of the different repeats are indicated below each protein. The percentages indicate residue identity to the repeats of the Rib protein, the Alp family member to which these proteins show the greatest similarity. Note that the Esp protein of E. faecalis has a more complex structure than the other proteins and has three types of repeats, the tandem A and C repeats and the nontandem B repeats, of which the B and C repeats show similarity to the repeats of Rib and other Alp proteins. The complete sequence is not available for the Rib-like protein (Rlp) of L. fermentus. With one exception (BipA), these different proteins have an LPXTG motif in the C-terminal region, which may allow covalent anchoring to the cell wall. The BipA protein of the gram-negative bacterium B. pertussis is probably bound to the bacterium via an N-terminal hydrophobic region. The black boxes to the left represent signal peptides. See the text for details and references. (B) Alignment of sequences in a conserved region present within the repeats. Because these sequences are most similar to the corresponding sequence in Rib among different Alp proteins, the corresponding sequence in Rib is also included. Residues identical to those in Rib are shown in bold type, and the number of identities is indicated to the right. The consensus sequence represents residues present in at least five of the eight sequences.

Esp in E. faecalis and a closely related protein in E. faecium. The 202-kDa Esp protein of E. faecalis was identified through the presence of localized sequence identity to the Rib and proteins of S. agalactiae (225). Because the esp gene was found in many E. faecalis strains isolated from patients with invasive infections but only rarely in commensal isolates, it was suggesed that Esp contributes to the virulence of this important cause of nosocomial infections (225). Other investigators have reported that the esp gene may also be common among commensal isolates, a result that might reflect geographical differences (258). The function of Esp is not known, but it enhances the ability of E. faecalis to cause urinary tract infection in an animal model (224) and contributes to the formation of biofilms on abiotic surfaces (255), although other surface structures also are important for biofilm formation (127, 167, 255).

Esp differs from the Alp family proteins in having three different types of repeat (A, B, and C), as well as longer N- and C-terminal regions (Fig. 3A). As a result, Esp is about twice the size of the S. agalactiae proteins. The A and C repeats of Esp are tandemly arranged, while the two B repeats are located in different parts of the sequence. Within the A and C repeat regions, the sequences of different repeat units are almost identical, but the number of repeats varies among different clinical isolates. However, all esp genes include A and C repeats, suggesting that both regions are important for the function of the protein (225, 255). In contrast to the variability in repeat number between different clinical isolates, the number of repeats within one strain remained stable after passages on laboratory medium, as described also for the Rib and proteins of S. agalactiae (201, 236). The 69-residue B repeats and the 82-residue C repeats are related and show extensive sequence similarity to members of the Alp protein family (225). Of note, these repeats do not contain the acid-labile Asp-Pro bonds that give rise to a ladder pattern in Alp family proteins analyzed by SDS-PAGE (261).

The esp gene is located on a large PAI, supporting the notion that the Esp protein contributes to virulence (223). Within this PAI, a well-defined short deletion occurs at considerable frequency during in vitro growth, causing the elimination of several genes in the PAI, including the entire esp gene. The biological relevance of this irreversible deletion is unknown, but it is reminiscent of the appearance under certain conditions of -protein variants with few repeat regions (154). In vivo, such deletions might represent a dead end for the bacterium, because the deleted fragment would be irreversibly lost. It has been suggested that loss of Esp may cause detachment of bacteria from a biofilm, thereby allowing dissemination, but the problem with irreversible gene loss remains (257).

The gene for a protein closely related to Esp of E. faecalis is found in many strains of E. faecium, an increasingly important cause of nosocomial infections (59, 269), and epidemiological analysis indicates that this esp-related gene is associated with virulence (142, 269). As described for the Esp protein of E. faecalis, the number of A and C repeats varies among different clinical isolates of E. faecium, and some isolates may even lack A repeats (59, 142). Interestingly, the esp gene of E. faecium may be located on a pathogenicity island, like the esp gene of E. faecalis, but, except for esp and an araC-like gene, these two putative islands contain different genes (142).

Other related proteins. A predicted "R28-like" protein was recently identifi