Serum opacity factor is a major fibronectin-binding protein and a virulence determinant of M type 2 Streptococcus pyogenes

Authors


Harry S. Courtney. E-mail hscourt@ix.netcom.com; Tel. (+1) 901 523 8990 ext. 7548; Fax (+1) 901 577 7273.

Abstract

Serum opacity factor (SOF) is a fibronectin-binding protein of group A streptococci that opacifies mammalian sera and is expressed by some strains that cause impetigo, pharyngitis and acute glomerulonephritis. Although SOF is expressed by ≈35% of known serotypes, its role in the pathogenesis of group A streptococcal infections has not been previously investigated. The sof genes from M types 2, 28 and 49 Streptococcus pyogenes were cloned, sequenced, and their deduced amino acid sequences were compared. The gene for FnBA, a fibronectin-binding protein from Streptococcus dysgalactiae, was also cloned and found to express an opacity factor. The leader sequences, the fibronectin-binding domains, and the membrane anchor regions of these proteins were highly conserved. Short spans of conserved sequences were interspersed throughout the remaining parts of the proteins. The sof2 gene was insertionally inactivated in an M type 2 S. pyogenes strain, T2MR. The resultant SOF-negative mutant (YL3) did not express SOF or opacify serum, and exhibited a 71% reduction in binding fibronectin. Complementation of the SOF-negative defect with sof28 in the recombinant strain YL3(pNZ28) fully restored fibronectin-binding activity and the ability to opacify serum. To determine whether sof plays a role in virulence, mice were challenged intraperitoneally with these strains. None of the 10 mice infected with YL3(pNZ28) survived and only 1 out of 15 mice challenged with T2MR survived, whereas 12 out of 15 mice infected with YL3 survived. These data clearly indicate that SOF is a virulence factor, and they provide the first direct evidence that a fibronectin-binding protein contributes to the pathogenesis of group A streptococcal infections in vivo.

Introduction

Serum opacity factor (SOF) is a large extracellular and surface-bound protein of group A streptococci that causes opalescence of serum (Ward and Rudd, 1938; Krumwiede, 1954). SOF specifically cleaves the apolipoprotein A1 (Apo A1) in high-density lipoproteins (HDL), and it was suggested that the opalescence of serum resulted from the aggregation of HDL particles (Saravani and Martin, 1990). The sof gene from M type 22 Streptococcus pyogenes has been sequenced, and the deduced amino acid sequence was found to contain a fibronectin-binding domain separate and distinct from the domain for enzyme activity (Rakonjac et al., 1995). Another sof gene from an unidentified serotype has also been cloned, sequenced and found to be almost identical to sof22 (Kreikemeyer et al., 1995).

Kreikemeyer et al. (1995) found the sof gene in 43% of isolates from invasive group A streptococcal infections and in 56% of streptococci isolated from wound, throat and skin infections. Strains of group A streptococci that express SOF are also a common cause of impetigo and many are nephritogenic (Wannamaker, 1970; Bisno and Stevens, 1996). The ability to opacify serum is also found in many strains of group C streptococci and staphylococci, but the protein responsible for this opacity reaction has been identified only in group A streptococci. In this report, evidence is provided that the fibronectin-binding protein, FnBA, from Streptococcus dysgalactiae also opacifies serum.

SOF, in addition to M protein, is used to serotype group A streptococci. This typing scheme is based on the observation that SOF contains type-specific determinants that co-vary with the type-specific determinants of M protein (Widdowson et al., 1970; Johnson and Kaplan, 1993). Thus, by determining the SOF type, the M type can also be identified. Currently, there are more than 90 different M protein serotypes and ≈35% of these express SOF (Johnson and Kaplan, 1993).

Expression of sof is controlled by the positive regulator Mga and is part of the Mga regulon, which consists of tandemly linked genes for M-related protein, M protein, M-like protein (or Enn), and C5a peptidase (Caparon and Scott, 1987; Podbielski et al., 1992; Chen et al., 1993). However, sof is not in the same chromosomal locus as other genes in the Mga regulon (McLandsborough and Cleary, 1995), being at least 15 kb away (Rakonjac et al., 1995). Although other gene products under control of Mga have been identified as virulence factors (Scott et al., 1986; Caparon and Scott, 1987; McLandsborough and Cleary, 1995; Podbielski et al., 1996; Courtney et al., 1997), the role of SOF has not been investigated. It is clear that SOF is not an absolute requirement for virulence in all strains of group A streptococci because many serotypes do not contain the sof gene (Kreikemeyer et al., 1995; Rakonjac et al., 1995). However, it is not known whether SOF contributes to the virulence of SOF-positive strains. Purified SOF has been shown to react with fibronectin, but it is not known what role SOF on the surface of streptococci has in binding fibronectin. To investigate the role of SOF in these activities, we inactivated the sof gene in an M type 2 group A streptococcus and complemented this defect with sof28. The parent, mutant and recombinant strains were tested for fibronectin binding and for virulence in a mouse model. The results provide the first direct evidence that SOF is major fibronectin-binding protein and a virulence factor in a SOF-positive group A streptococcus.

Results

Comparison of SOF2, SOF28, SOF49, SOF22 and FnBA

Although it has been shown that sof genes vary among different serotypes (Rakonjac et al., 1995), the degree of variability is unknown. Therefore, we sequenced three additional sof genes and compared their deduced amino acid sequences (Fig. 1). The percentage homology between these proteins ranges from 56% to 65%. The leader sequence and the carboxy-terminal regions containing the fibronectin-binding and the membrane-spanning domains are highly conserved. There are short homologous segments interspersed throughout the molecule between the leader sequence and the fibronectin-binding domain. A comparison of the hydrophobicity plots and the predicted secondary structures revealed remarkable similarity throughout the entire length of these proteins, despite the fact that the overall degree of sequence homology was not very high (data not shown). The amino acid sequence in the region found by Rakonjac et al. (1995) to be necessary for the enzymatic activity of SOF22 appears to be more conserved in SOF2 than in SOF28 and SOF49.

Figure 1.

. Comparison of the amino acid sequences of SOF2, SOF22, SOF28, SOF49 and FnBA. The deduced sequences of sof2, sof 28, sof49 (obtained from this study), sof22 (Rakonjac et al., 1995) and FnBA (Lindgren et al., 1993) were aligned with megalink software using clustal parameters. Gaps were introduced to maximize the degree of homology. White letters on black background indicate identical residues.

The predicted molecular weights of the mature proteins are similar, ranging from 108 806 to 111 289 (assuming cleavage between Ala and Ser within the leader sequence). The calculated isoelectric points of the mature proteins are also similar, ranging from 5.1 to 5.3 and are within the pI range of SOF2 determined experimentally by Martinez et al. (1978).

FnBA, a fibronectin-binding protein from S. dysgalactiae (Lindgren et al., 1993), has a significant degree of homology with SOF (Fig. 1), suggesting that FnBA may be an opacity factor. Therefore, we tested SDS extracts and cell culture medium of S. dysgalactiae for opacification activity. The results indicate that strain S2 of S. dysgalactiae expresses an opacity factor (Fig. 2). In this strain, the opacity factor was found in extracts of cells but, in contrast to type 2 S. pyogenes, was not found in the culture medium. To determine whether this opacity reaction was due to FnBA, its gene was cloned and expressed in E. coli. Lysates of E. coli expressing recombinant FnBA readily opacified horse serum, whereas lysates of control E. coli did not (Fig. 2). These data suggest that FnBA is expressed on the surface of S. dysgalactiae and can opacify serum.

Figure 2.

. Serum opacity reaction of extracts of Streptococcus dysgalactiae and recombinant FnBA. Culture supernatants or SDS extracts of S. dysgalactiae strain S2 were tested for serum opacity reactions. SDS lysates of control E. coli and recombinant E. coli expressing FnBA were also tested for ability to opacify serum.

Analysis of the expression of SOF in T2MR, YL3 and YL3(pNZ28)

As the first step to determining the role of SOF in binding fibronectin and in virulence, the sof2 gene was insertionally inactivated as described in the Experimental procedures. One mutant was selected for further study and labelled YL3. PCR and Southern blot analysis of YL3 indicated that a single cross-over event occurred at the 5′ end of sof2, resulting in the integration of the plasmid containing the omega element and inactivation of sof2. We used a shuttle vector to complement the SOF-negative defect with SOF28 to obtain the recombinant strain YL3(pNZ28). To determine that this insertion event resulted in the loss of expression of functional SOF and that complementation with sof28 restored expression, culture supernatants and cell extracts of T2MR, YL3 and YL3(pNZ28) were tested for their ability to opacify serum (Fig. 3A). Culture supernatants and cell extracts of T2MR and YL3(pNZ28) readily opacified horse serum, whereas neither the culture supernatant nor the cell extract of YL3 was able to opacify serum. In fact, a 100-fold concentrate of the culture supernatant of YL3 did not opacify serum or react in Western blots with anti-SOF2 serum (data not shown). Interestingly, the serum opacity reaction of YL3(pNZ28) was higher than that of T2MR. The basis for this increased opacity reaction could be the result of an increase in the expression of SOF28 because pNZ28 is a multicopy plasmid. It is also possible that SOF28 is more enzymatically active than SOF2.

Figure 3.

. Analysis of the opacifying and fibronectin-binding activity of T2MR, YL3 and YL3(pNZ28). A. Opacifying activity of the growth supernatants (open bars) and SDS extracts of cell pellets (closed bars) of streptococci. B. Fibronectin-binding activity of the streptococci. Fibronectin binding was determined by reacting biotinylated fibronectin with microtitre wells coated with streptococci as described in the Experimental procedures.

Western blot analysis using anti-SOF2 serum confirmed that YL3 did not express SOF, and that both T2MR and YL3(pNZ28) did express SOF (Fig. 4). That anti-SOF2 serum also recognized SOF28 is not surprising because there is ≈55% sequence homology between these proteins. No immunoreactive bands were found in lanes containing YL3 extracts, indicating that there was no detectable expression of a truncated product. SOF was not homogeneous in size and had an Mr that ranged from ≈180 to ≈90. The band with an Mr of 180 is considerably higher than the predicted Mr and may be a dimer of SOF. Horse serum agar overlays of acrylamide gels of recombinant SOF also exhibited bands with a high Mr (data not shown).

Figure 4.

. Western blot analysis of the expression of SOF in T2MR, YL3 and YL3(pNZ28). Cell pellets from 50 ml cultures of T2MR (lane 3), YL3 (lane 2) and YL3(pNZ28) (lane 1) were extracted with SDS, subjected to SDS–PAGE, transferred to nitrocellulose, and probed with a 1:500 dilution of preimmune serum (A) or rabbit anti-SOF2 serum (B). Lane 4 contains prestained molecular weight standards and their Mr are indicated in the right margin.

To verify that YL3(pNZ28) expressed the correct SOF type, we tested culture supernatants with human sera that specifically neutralized the serum opacity reaction of SOF2 but not that of SOF28. The neutralizing serum completely blocked the serum opacity reaction of culture supernatants from T2MR, but had no effect on the serum opacity reaction of culture supernatants of YL3(pNZ28).

Binding of fibronectin to T2MR, YL3 and YL3(pNZ28)

SOF has been previously shown to bind fibronectin (Kreikemeyer et al., 1995; Rakonjac et al., 1995). However, it was not determined whether SOF expressed on the surface of group A streptococci had a role in fibronectin binding. Thus, we compared the binding of fibronectin to T2MR, YL3 and YL3(pNZ28) immobilized on microtitre wells (Fig. 3B). Inactivation of sof2 resulted in a 71% reduction in the binding of fibronectin, indicating that SOF is a major fibronectin-binding protein in this strain. To confirm this finding, fibronectin was also reacted with these streptococci in suspension, and the results indicated that YL3 bound 67% less fibronectin that T2MR (data not shown). The finding that there was not a complete reduction in the binding of fibronectin suggests that there are other receptors for fibronectin in this strain. Complementation of the SOF-negative defect with sof28 fully restored fibronectin binding. In fact, YL3(pNZ28) bound more fibronectin that T2MR, possibly because of multiple copies of pNZ28. SOF2 appears to react similarly with both soluble and immobilized forms of fibronectin because soluble fibronectin blocked the binding of SOF2 to immobilized fibronectin by 75% (data not shown). One surprising result was that YL3 and T2MR bound equally to immobilized fibronectin (data not shown), suggesting that the adhesion to immobilized fibronectin is mediated by another fibronectin-binding moiety(s). One possible explanation for the lack of involvement of SOF in binding to immobilized fibronectin is that the fibronectin-binding domain of SOF is adjacent to the cell wall-anchoring motif, and inaccessible for interactions with immobilized fibronectin. Soluble fibronectin, in contrast, could easily penetrate through surface structures to reach the fibronectin-binding domain of SOF.

To investigate the possibility that fibronectin binding may affect the ability of SOF to opacify serum, tubes were coated with fibronectin and reacted with culture supernatants of T2MR or YL3. SOF in the supernatant of T2MR readily bound to fibronectin and maintained its ability to opacify horse serum (A405 of 1.025 ± 0.026). As expected, the culture supernatant of YL3 did not exhibit any SOF activity in this assay (A405 of 0.003 ± 0.001). Interestingly, SOF2 demonstrated a remarkable stability, remaining bound to fibronectin and retaining enzymatic activity for at least 72 h even after several washes with serum. This showed that the binding of SOF2 to fibronectin was essentially irreversible, and suggested that SOF2 released by streptococci could bind to fibronectin and remain enzymatically active for extended periods.

Localization of the fibronectin-binding and enzymatic domains of SOF2

Rakonjac et al. (1995) found that the C-terminal residues 813–1025 of SOF22 could be deleted without effect on its enzymatic activity. To investigate which regions of the N-terminus of SOF are required for enzymatic activity, we constructed truncated fusion proteins of SOF2 containing a N-terminal histidine tag and tested these proteins for their ability to opacify horse serum (Fig. 5). The fusion proteins SOF2–H(38–1047), SOF2–H(38–843) and SOF2–H(148–843) readily opacified horse serum, whereas the fusion proteins SOF2–H(38–493), SOF2–H(205–843) and SOF2–H(494–1047) did not opacify serum. These data indicate that the enzymatic domain is contained within the amino acid residues 148 and 843 of SOF2. These findings confirm and extend those of Rakonjac et al. (1995). The fusion proteins were also tested for fibronectin-binding activity and only the fusion proteins SOF2–H(38–1047) and SOF2–H(494–1047) reacted with fibronectin. The 494–1047 region of SOF2 contains a sequence of repeating peptides that are highly homologous to the repeats described by Kreikemeyer et al. (1995) and Rakonjac et al. (1995). These investigators found that a peptide containing only the fibonectin-binding repeats was able to bind fibronectin. Together, these findings clearly indicate that the fibronectin-binding and enzymatic functions reside on separate domains of SOF.

Figure 5.

. A. Schematic diagram of the functional domains of SOF. The location of the functional domains of SOF were determined from the data provided in part B of this figure and from the data of Kreikemeyer et al. (1995), Rakonjac et al. (1995) and Schneewind et al. (1993). B. Localization of the fibronectin-binding and enzymatic domains of SOF2. Truncated, histidine fusion proteins were tested for fibronectin-binding and enzymatic activity. The designation of each peptide is indicated above the solid lines and the numbers in parentheses indicate the span of the peptide. Fibronectin-binding activity was determined by reacting biotinylated fibronectin with the peptides immobilized on microtitre wells as described in the Experimental procedures. Opacification activity of the peptides was determined by horse serum agar overlays of SDS–PAGE. NT = not tested.

SOF as a virulence factor

In preliminary experiments, the LD50 of T2MR and YL3 in mice was determined to be ≈1 × 107 and ≈4 × 108, respectively, indicating a 40-fold difference. The role of SOF in virulence was further investigated by injecting ≈4 × 107 cfu of T2MR, YL3 or YL3(pNZ28) into the peritoneal cavities of mice and recording the number of surviving mice (Fig. 6). None of the 10 mice challenged with YL3(pNZ28) survived and only 1 of the 15 mice (7%) challenged with T2MR survived. In contrast, 12 of the 15 mice (80%) challenged with YL3 survived, indicating that SOF contributes significantly to the virulence of SOF-positive group A streptococci.

Figure 6.

. Survival plots demonstrating that SOF is a virulence determinant of group A streptococcal infections in mice. Approximately 4 × 107 cfu of T2MR (closed circles), YL3 (open circles), and YL3(pNZ28) (closed triangles) were injected intraperitoneally into mice. Fifteen mice received injections of T2MR or YL3, and 10 mice received injections of YL3(pNZ28). The number of surviving mice was recorded daily. The number of cfu injected was determined by plating dilutions on blood agar plates. The P-values were determined by Fisher's exact test.

Although there was no significant difference in the overall survival of mice challenged with T2MR and YL3(pNZ28), mice infected with YL3(pNZ28) died more rapidly. Because YL3(pNZ28) appears to express more SOF than T2MR, these data are consistent with the concept that the amount of SOF expressed may also impact on the virulence of streptococci in mice.

Group A streptococci were recovered from the spleens of all mice that died. YL3 recovered from spleens of mice still exhibited the SOF-negative phenotype and no sensitivity to kanamycin was found in over 100 cfu tested, indicating no detectable reversion of YL3 to wild type. YL3(pNZ28) recovered from spleens of mice that died were resistant to both kanamycin and erythromycin in 100 cfu tested. SOF activity from cultures of YL3(pNZ28) was not inhibited by SOF2 neutralizing serum, indicating no reversion to wild type.

To determine whether SOF contributed to the ability of streptococci to resist phagocytosis, the growth of T2MR and YL3 in non-immune human and mouse blood was compared. There was no significant difference in their growth in either human or mouse blood, indicating that SOF does not contribute to the ability of T2MR to resist phagocytosis. The rates of growth of T2MR, YL3 and YL3(pNZ28) in THY broth during a 24 h period were virtually identical, suggesting that the observed difference in virulence was not due to a difference in their rate of growth.

Discussion

The main objectives of this investigation were to determine whether SOF has a role in the pathogenesis of streptococcal infections and in binding fibronectin. That SOF contributes to the virulence of SOF-positive group A streptococci was demonstrated by a significant reduction in the virulence of the SOF-negative mutant in mice. Although it is possible that inactivation of sof2 could have polar effects, the presence of a putative transcription terminator suggests that sof is probably monocistronic and that its inactivation is not likely to affect the expression of adjacent genes. Because we do not yet know what genes are adjacent to sof, it is not possible to provide definitive proof that expression of these genes was not affected. However, the results of complementing the SOF-negative defect with sof28 offers strong evidence that the decrease in virulence was due solely to inactivation of sof2. The recombinant strain expressing SOF28 was as virulent as T2MR, suggesting that at least two different types of SOF can act as a virulence determinant of a SOF-positive group A streptococcus.

The binding of fibronectin to SOF was investigated because of numerous studies indicating that fibronectin may serve as a receptor for bacteria on mucosal surfaces, in wounds and on in-dwelling catheters (for a review, see Hasty et al., 1994). It has been suggested that SOF is not a major fibronectin-binding protein in group A streptococci, but this conclusion was based on the binding of fibronectin to a M49 strain that did not express SOF (Jaffe et al., 1996). Our finding that inactivation of sof2 decreased fibronectin binding by more than 70% indicates that SOF2 is a major fibronectin-binding protein on type 2 group A streptococci. That loss of SOF2 expression was responsible for this decreased binding of fibronectin is supported by the fact that complementation with sof28 fully restored the ability to bind fibronectin. At least five other fibronectin-binding proteins have been identified in group A streptococci (Courtney et al., 1992; Hanski and Caparon, 1992; Pancholi and Fischetti, 1992; Talay et al., 1993; Courtney et al., 1996; Jaffe et al., 1996), but none of these has been shown to be involved in group A streptococcal infections in an animal model. Our data indicating that SOF contributes to the virulence of type 2 S. pyogenes in mice provide the first direct evidence that a fibronectin-binding protein has a role in the pathogenesis of group A streptococcal infections in vivo.

Other streptococci and staphylococci also produce a factor(s) that induces opalescence in mammalian serum (El-Tayeb and Nasr, 1977; Sippel et al., 1995), but the nature of the opacity factor has not been previously identified in any of these bacteria. Our data indicate that a fibronectin-binding protein from Streptococcus dysgalactiae, FnBA, is an opacity factor. The finding that two different streptococcal species express an opacity factor that also binds fibronectin suggests that the linking of these two activities may be important to streptococcal virulence. Interestingly, opacification activity was found only in SDS extracts of cell cultures of S. dysgalactiae and not in the culture medium, whereas SOF2 was found both in cell extracts and in the cell culture medium. One possible explanation for this difference is that FnBA and SOF have a different LPxTG motif. It has been postulated that the LPxTG motif serves as a site for proteolytic cleavage and linkage to the cell wall (Schneewind et al., 1993). This motif in FnBA is LPQTG, whereas SOF has a LPASG motif. It may be that the LPASG variation is more susceptible to proteolytic cleavage and, therefore, is found both extracellularly and as a surface-bound protein. However, as other factors also contribute to the sorting of proteins, the mechanism whereby SOF in S. pyogenes is released into the medium remains to be determined.

SOF is not the only virulence factor in SOF-positive group A streptococci. Both M-related proteins and M proteins are also required for virulence of M type 2 S. pyogenes (Podbielski et al., 1996). Inactivation of mga in a SOF-positive, M type 49 strain resulted in a lack of expression of M49 protein, M-like protein, SOF49 and loss of virulence (McLandsborough and Cleary, 1995), indicating that one or more of these proteins was required for virulence. Taken together, these data suggest that expression of SOF, M proteins and M-related proteins is required for full virulence of SOF-positive streptococci. Although expression of the hyaluronate capsule is required for full virulence in SOF-negative streptococci (Wessels et al., 1991), its role in the pathogenesis of infections due to SOF-positive streptococci has not been addressed.

The mechanism(s) whereby SOF contributes to virulence is not known. SOF apparently has little, if any, role in resistance of streptococci to phagocytosis because the SOF-negative mutant and the parent strain grew equally well in human and mouse blood. The only known functions of SOF are to bind fibronectin and to opacify serum. The ability of SOF to bind irreversibly to immobilized fibronectin and remain enzymatically active for at least 72 h suggests that SOF may accumulate and remain active in the host. The formation of large aggregates of HDL induced by the cleavage of APO A1 by SOF may be detrimental to the host, but this proteolysis may have other effects as well. During the acute-phase response to infections, HDL becomes proinflammatory with a concomitant decrease in the levels of Apo A1 (Van Lenten et al., 1995). Thus, SOF may increase the levels of proinflammatory HDL by proteolysis of Apo A1. Apo A1 has also been shown to prevent cell death of endothelial cells caused by oxidized LDL (Suc et al., 1997), and SOF may inhibit this activity of Apo A1. Although speculative at present, it is possible that SOF contributes to virulence by a combination of these mechanisms.

In summary, SOF is a major fibronectin-binding protein, and at least two different serotypes of SOF can act as virulence determinants in a SOF-positive group A streptococcus. We have demonstrated that SOF is also expressed in group C streptococci. Although it remains to be established whether the factor expressed by other Gram-positive bacteria that opacify serum is similar to SOF, these data raise the possibility that SOF may be a widely distributed virulence determinant among several species of Gram-positive bacteria.

Experimental procedures

All experiments were approved by the Animal Care and Use Committee of the University of Tennessee.

Reagents

Fibronectin was purified by gelatin-affinity chromatography from fresh human serum as described by Engvall and Ruoslahti (1977). The purity of the fibronectin was assessed by SDS–PAGE, and migrated as a typical doublet under reducing conditions with an Mr of ≈220. Although several faint bands were noted in other locations, density measurements indicated that the fibronectin was ≈95% pure. Fibronectin was labelled with biotin as previously described (Courtney et al., 1994). Horse and mouse sera were purchased from Sigma Chemicals.

Organisms and growth conditions

The SOF-positive, M type 2 S. pyogenes strain T2MR was kindly provided by Dr Debra Bessen at Yale University. M type 28 and M type 49 S. pyogenes are from our collection of streptococcal strains. YL3 is an isogenic mutant of T2MR in which sof2 was insertionally inactivated as described in this paper. YL3(pNZ28) is YL3 in which the SOF-negative defect was complemented with sof28 as described in this paper. The group C streptococcus, Streptococcus dysgalactiae strain S2, was generously provided by Dr Magnus Hook at Texas A and M University. The organisms were grown in Todd–Hewitt broth supplemented with 1.5% yeast extract (THY).

DNA techniques

Chromosomal DNA was purified from streptococci as previously described (Courtney et al., 1994). Plasmids were purified from E. coli using Promega Wizard minipreps according to the manufacturer's protocol (Promega).

Cloning and sequencing of sof2, sof28 and sof49

Based on the published sequence of sof22 (Rakonjac et al., 1995), primers 1 and 2 (Table 1) were designed to amplify the entire coding regions of sof2, sof28 and sof49. The PCR product was ligated into the pCRII vector for automated sequencing using M13 forward and reverse primers. Subsequent primers were designed based on the sequences obtained.

Table 1. . List of primers. a. Primers were designed with the aid of oligo software using the published sequence of sof22 (Rakonjac et al., 1995), sequences from the omega element (Trieu-Cuot and Courvalin, 1983; Prentki and Krisch, 1984), and from sequences of fnba (Lindgren et al., 1993).Thumbnail image of

Cloning and expression of fnba from S. dysgalactiae

Primers 5 and 6 (Table 1) were used to amplify fnba and the PCR product was ligated into pBAD TOPO (Invitrogen) and transformed into E. coli Top10. Clones were screened for resistance to ampicillin and for the ability to opacify serum. One recombinant clone was selected for further study.

Construction and purification of SOF2 fusion proteins

Primers were designed to amplify the desired coding sequences of sof2. The amplified products were digested with appropriate restriction enzymes and ligated into pTrcHis. The resulting plasmids were transformed into E. coli Top10. Recombinant colonies were screened for resistance to ampicillin and expression of a product with the correct size. One recombinant clone from each construct was selected for further study. That the correct clones were selected was verified by limited DNA sequencing, by functional activity and by reaction with anti-SOF2 serum. The fusion proteins were purified from SDS lysates of the recombinant E. coli by nickel affinity chromatography using the manufacturer's denaturing protocol (Invitrogen). In some samples, the fusion protein was further purified by preparative SDS–PAGE.

Rabbit antiserum to the fusion protein spanning residues 38–1047 of SOF2, termed SOF2–H(38–1047), was developed by subcutaneous immunization of New Zealand white rabbits with 500 μg of the fusion protein in Freunds complete adjuvant. Booster injections in PBS were given at 4 and at 8 weeks. Serum samples were collected every 2 weeks.

Inactivation of sof2

The Ω-element, which contains a kanamycin resistance marker expressed in both Gram-positive and Gram-negative organisms as well as translational and transcriptional terminators (Prentki and Krisch, 1984), was used for insertional inactivation of sof2. The Sal I and XhoI fragment of sof2 (encompassing bp 307–987 corresponding to amino acid residues 103–329) was ligated into the Sal I site of pUC19 to obtain pSOF2. The HindIII fragment of pBRΩ-KM-2 containing the Ω-element was ligated into the HindIII site of pSOF2. The resultant plasmid, pSOF2-Ω, which cannot replicate in Gram-positive organisms, was electroporated into T2MR. Colonies that grew on Todd–Hewitt agar containing 250 μg ml−1 kanamycin were randomly selected and screened for the lack of expression of sof2. One colony was selected for further study and labelled YL3.

PCR and Southern blot analysis of YL3

PCR analysis of the insertional inactivation of sof2 was performed using chromosomal DNA from T2MR and YL3 as templates. PCR was performed using protocols for Q-solution as described by the manufacturer (Qiagen). Chromosomal DNA from T2MR and YL3 was digested overnight with Pst I, Sal I or Bstz171, subjected to agarose gel electrophoresis, transferred to positively charged nylon membranes, and probed with digoxigenin-labelled Ω-element using procedures recommended by the manufacturer (Boehringer Mannheim). The Ω-element probe was labelled with digoxigenin by PCR using primers 3 and 4 (Table 1), purified Ω-element as template, and the PCR protocol recommended by the manufacturer (Boehringer Mannheim).

Construction and introduction of pNZ28 into YL3

The sof28 gene was amplified by PCR and ligated into the shuttle vector pNZerm, which can replicate in both Gram-positive and Gram-negative bacteria and contains an erythromycin resistance marker (Podbielski et al., 1992). The plasmid was introduced into YL3 by electroporation and colonies were screened for resistance to erythromycin. One colony that expressed SOF28 was selected for further study and labelled YL3(pNZ28).

Assays for opacification of horse serum

Activity of SOF in growth supernatants was tested by centrifugation of overnight cultures of the organisms, sterilization of the media by filtration, and addition of 100 μl of the filtrate to 1 ml of horse serum. After overnight incubation at 37°C, the absorbance at 405 nm was recorded. The growth supernatant was also concentrated 100-fold using a YM50 filter and the concentrate tested as described above. To detect cell-bound SOF activity, the streptococcal pellet from 10 ml of culture was extracted with 0.5 ml of 1% SDS and 100 μl of the extract was added to horse serum. SDS extraction has previously been shown to be an accurate and sensitive method for the detection of SOF activity (Rehder et al., 1995). Variations in the final absorbance values were noted with different lots of horse serum, but the results presented in each figure were obtained with the same lot of horse serum. All assays were carried out in triplicate unless noted otherwise.

For assays detecting opacification activity of SOF2 bound to fibronectin, tubes were coated with fibronectin (20 μg ml−1 in 0.02 M sodium bicarbonate, pH 9.5) for 1 h at 37°C, then overnight at 4°C. The tubes were washed in PBS, pH 7.0, and then blocked with BSA (1 mg ml−1 in PBS). The growth supernatants were mixed with an equal volume of fibronectin (100 μg ml−1) or PBS, and 3 ml of each mixture was added to the fibronectin-coated tubes and rotated for 1 h at 37°C. The tubes were then washed, 3 ml of horse serum was added, the tubes rotated overnight at 37°C, and the absorbance at 405 nm was recorded. Tubes coated with BSA served as a negative control. All assays were carried out in triplicate.

Horse serum agar overlays

Horse serum agar was prepared by mixing equal portions of horse serum and a double concentration of LB agar. The horse serum agar was overlaid on SDS acrylamide gels of extracts containing SOF activity and incubated for 1–2 days. The location of SOF is indicated by the appearance of opaque bands.

SOF neutralization assay

Human sera from donors were screened for SOF2 and SOF28 neutralizing activity. One donor was found whose serum neutralized the serum opacity reaction of SOF2 but had no effect on SOF28. The assay for neutralization consisted of preincubating 100 μl of neutralizing serum and 100 μl of culture supernatant for 30 min at 37°C, then adding 1 ml of horse serum and recording the absorbance at 405 nm after incubating overnight.

Binding of fibronectin to streptococci and to SOF2 fusion proteins

Wells of a microtitre plate were coated for 30 min at 37°C with 100 μl of T2MR, YL3 or YL3(pNZ28) suspended in PBS to an OD530 of 0.4. The wells were rinsed with PBS and blocked with BSA (1 mg ml−1 in PBS). Wells coated with BSA served as a blank. Biotinylated fibronectin (20 μg ml−1 in 0.05 M tris-HCl, 0.15 M NaCl, 1 mg ml−1 BSA, pH 7.4) was added to the wells. After incubating at 37°C for 30 min, the wells were washed and reacted with a 1:2000 dilution of Neutralite avidin peroxidase for 30 min at 37°C. The wells were washed and tetramethylbenzidine was added to the wells. After colour development, stop solution was added and the absorbance at 450 nm was recorded. Non-specific binding of biotinylated fibronectin was determined by adding a 50-fold excess of the unlabelled fibronectin to the wells and was consistently in the same range as the values found for BSA-coated wells. All assays were carried out in quadruplicate. In controls, rabbit antiserum against the group A C-carbohydrate reacted equally with the strains of streptococci immobilized on microtitre wells, indicating that equal numbers of streptococci were bound to the wells. There was also no difference in the binding of fibrinogen or IgA to the immobilized streptococci (data not shown), a further indication that equal numbers of streptococci were bound to the wells.

To measure the binding of fibronectin to the different fusion proteins of SOF2, microtitre wells were coated with the purified proteins (20 μg ml−1 in 0.05 M bicarbonate, pH 9.5) for 30 min at 37°C. The wells were blocked with BSA (1 mg ml−1) and reacted with biotinylated fibronectin and avidin peroxidase as described above. Substrate was added and the absorbance measured after colour development. Wells coated with BSA served as a negative control.

Western blots

Streptococcal pellets from 50 ml of overnight cultures were washed in PBS, extracted with 10% SDS for 1 h, and boiled for 5 min The extracts were subjected to SDS–PAGE under reducing conditions. The proteins were transferred to nitrocellulose and blocked with BSA (2 mg ml−1 in 0.05 M tris-HCl, 0.15 M NaCl, pH 7.4). The blot was then incubated with a 1:500 dilution of rabbit preimmune and anti-SOF2–H(38–1047). After washing, the blot was reacted with a 1:1000 dilution of peroxidase-conjugated goat anti-rabbit IgG. The blot was washed and reacted with 4-chloro-1-napthol.

Bactericidal assays

Streptococci were grown in THY to an OD530 of 0.08. Serial dilutions were prepared in THY and 50 μl of each dilution was added to 450 μl of heparinized human blood from a non-immune donor. The blood was rotated for 3 h at 37°C and the number of cfu was determined by plating an aliquot on blood agar.

Virulence in mice

Mid-log cultures of T2MR, YL3 and YL3(pNZ28) were washed and suspended in PBS. Serial 10-fold dilutions of T2MR or YL3 were injected into the peritoneal cavity of five female NIH Swiss mice and the LD50 determined by the method of Reed and Muench (1938). The number of cfu in the inoculum was determined by plating dilutions on blood agar plates. In an additional experiment, ≈4 × 107 cfu of T2MR, YL3 and YL3(pNZ28) were injected intraperitoneally into female NIH Swiss mice. Fifteen mice were infected with T2MR or YL3, and 10 mice were infected with YL3(pNZ28). The numbers of mice that were dead or moribund were evaluated each morning and evening for 10 days. Moribund mice were sacrificed according to the guidelines of the Animal Use Committee.

Accession numbers

The sequence data for sof2, sof28 and sof49 have been submitted to GenBank under accession numbers AF019890, AF082074 and AF057697 respectively.

Footnotes

  1. *Present address: Washington University, St. Louis, MO, USA

Acknowledgements

We would like to express our appreciation for the excellent technical support of Edna Chiang. These studies were supported by research funds from the US Department of Veterans Affairs (D.L.H. and J.B.D.), by grant AI-10085 (J.B.D.) from the National Institutes of Health, and by grant T35 DK07405-13 (J.L.T.) from the NIH Medical Student Research Fellowship Program.

Ancillary