An Extracellular Protease of Streptococcus gordonii Hydrolyzes Type IV Collagen and Collagen Analogues

Abstract

Streptococcus gordonii is a frequent cause of infective bacterial endocarditis, but its mechanisms of virulence are not well defined. In this study, streptococcal proteases were recovered from spent chemically defined medium (CDM) and fractionated by ammonium sulfate precipitation and by ion-exchange and gel filtration column chromatography. Three proteases were distinguished by their different solubilities in ammonium sulfate and their specificities for synthetic peptides. One of the enzymes cleaved collagen analogs Gly-Pro 4-methoxy-β-naphthylamide, 2-furanacryloyl-Leu-Gly-Pro-Ala (FALGPA), and p-phenylazobenzyloxycarbonyl-Pro-Leu-Gly-Pro-Arg (pZ-peptide) and was released from the streptococci while complexed to peptidoglycan fragments. Treatment of this protease with mutanolysin reduced its 180- to 200-kDa mass to 98 kDa without loss of enzymatic activity. The purified protease cleaved bovine gelatin, human placental type IV collagen, and the Aα chain of fibrinogen but not albumin, fibronectin, laminin, or myosin. Enzyme activity was inhibited by phenylmethylsulfonyl fluoride, indicating that it is a serine-type protease. Maximum production of the 98-kDa protease occurred during growth of S. gordonii CH1 in CDM containing 0.075% total amino acids at pH 7.0 with minimal aeration. Higher initial concentrations of amino acids prevented the release of the protease without reducing cell-associated enzyme levels, and the addition of an amino acid mixture to an actively secreting culture stopped further enzyme release. The purified protease was stored frozen at −20°C for several months or heated at 50°C for 10 min without loss of activity. These data indicate that S. gordonii produces an extracellular gelatinase/type IV collagenase during growth in medium containing minimal concentrations of free amino acids. Thus, the extracellular enzyme is a potential virulence factor in the amino acid-stringent, thrombotic, valvular lesions of bacterial endocarditis.

Streptococcus gordonii is an oral viridans group streptococcus that frequently causes infective endocarditis (6, 9, 44). The bacteria enter the bloodstream as a consequence of trauma to oral tissues (4, 8, 31) and can adhere to heart valves damaged by rheumatic fever, regurgitant blood flow, high-pressure gradients, and stenosis (10, 13, 17), as well as to preexisting thrombotic lesions (8). After colonization of heart valve surfaces, the bacteria become encased in a layer of fibrin and platelets giving rise to macroscopic vegetations in which streptococci grow slowly (17). This reduced rate of growth likely reflects the limited availability of nutrients in this protein-rich but amino acid-poor environment. Nutrient acquisition by streptococci in this environment is most likely enhanced by the production of proteases. Soluble proteases might also contribute directly to pathogenesis by facilitating bacterial erosion of cardiac surfaces (10, 15), degrading host defense proteins such as immunoglobulin and complement components (20, 33), and/or cleaving and activating other streptococcal surface proteins involved in pathogenesis (24).

Species of oral streptococci (S. intermedius, S. mitis, S. mutans, S. oralis, S. salivarius, S. sanguis, and S. sobrinus) have been reported to produce extracellular proteases that degrade host proteins including albumin (25, 38), immunoglobulin A (33), salivary proteins (3), casein (21, 35, 38), gelatin (16, 37, 38), and collagen analogs (16). Of these extracellular proteases, a 52-kDa protease of S. mutans (16) and a 146-kDa protease of S. oralis (25) have been purified and characterized. In this report, we describe the purification of an extracellular 98-kDa serine-type protease of S. gordonii. Production of this enzyme, which cleaves gelatin, collagen IV, and collagen-like peptides, is affected by pH, aeration, and the concentration of free amino acids in the culture medium.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

S. gordonii Challis was grown in tryptic soy broth supplemented with 0.5% yeast extract (TSBY; Difco, Detroit, Mich.) and in a chemically defined medium (CDM) (40) containing 30 mM glucose. In experiments using the CDM, streptococci were conditioned to the medium by being passaged in it twice. Bacterial growth was measured turbidimetrically by using a Klett-Summerson colorimeter equipped with a no. 66 filter. Cultures were harvested at selected times by either centrifugation at 6,000 × g for 30 min at 4°C or tangential-flow filtration through a Filtron Cassette System with 0.45-μm pores (Millipore Corp., Bedford, Mass.). Spent culture medium was sterilized by filtration through membranes with 0.22-μm pores. The CDM was supplemented with 0.05 M HEPES in some experiments to maintain the pH between 6.8 and 7.4 during streptococcal growth. In other experiments, the quantity of the amino acid mixture (20 amino acids) in the CDM (40) was varied from 0.75 to 0.075% (wt/vol).

To determine the ability of streptococci to utilize proteins as the sole source of amino acids, 0.5% gelatin was substituted for the amino acid mixture in the CDM. As a control, an equal amount of gelatin was sterilized in a 8-mm-diameter dialysis sac with a 3,500-molecular-weight exclusion limit (Spectrum Medical Industries, Inc., Laguna, Calif.) and suspended in amino acid-free CDM. After 48 h of equilibration at 37°C, the media were seeded with a 1% inoculum of a logarithmic-growth-phase culture of S. gordonii in unmodified CDM.

Analytical procedures.

Protein concentrations in the culture preparations were determined by the Bradford assay (2) (Bio-Rad Corp., Hercules, Calif.), using bovine serum albumin (BSA) as the standard. Muramic acid was determined by the d-lactate dehydrogenase procedure of Tipper (41), monitoring alkali liberation of d-lactate from previously acid-hydrolyzed peptidoglycan. The presence of histone-like protein in spent culture medium was determined by Western blot assay using a rabbit antihistone serum (36).

SDS-PAGE.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed with a Mini-PROTEAN II gel system (Bio-Rad), using a 1.5-mm-thick 7.5% polyacrylamide gel (22). The proteins were stained with silver nitrate (29). A protein kit containing myosin (205 kDa), β-galactosidase (116 kDa), phosphorylase B (97 kDa), BSA (66 kDa), egg albumin (45 kDa), carbonic anhydrase (29 kDa), and myosin (205 kDa) (Sigma Chemical Company, St. Louis, Mo.) was used as a source of molecular mass standards.

Enzyme isolation.

Extracellular proteins from 15 liters of spent culture medium were concentrated 10-fold by tangential-flow ultrafiltration (Pellicon Cassette System-10,000; Millipore Corp.). The retentate was fractionated by sequential ammonium sulfate precipitation at 30, 55, and 80% saturation points. The protein precipitates were collected by centrifugation at 18,900 × g for 1 h, dissolved in 0.1 M Tris-HCl (pH 7.5), and dialyzed against the same buffer at 4°C for 24 h.

Zymography.

Proteins were separated by SDS-PAGE on 7.5% polyacrylamide–0.1% gelatin gels (14). After electrophoresis, the gel was incubated at room temperature in 2.5% Triton X-100 for 30 min and then in a buffer containing 50 mM Tris-HCl (pH 7.6), 0.2 M NaCl, 5 mM CaCl₂, and 0.02% Tween 20 for 18 h at 37°C. Zones of proteolysis were detected by Coomassie blue staining. In some experiments, samples solubilized in SDS sample buffer at room temperature were resolved by electrophoresis in SDS-polyacrylamide gels; transferred electrophoretically, at 25 V for 24 h at 4°C, into a 0.1% gelatin–7.5% polyacrylamide gel devoid of SDS; and developed for protease activity as described above. Staining of the primary polyacrylamide gels with Coomassie blue indicated that all proteins were transferred from the resolving gel to the substrate-containing gel.

Enzyme assays.

Enzymatic activity was measured by a modification of the methods of Umezawa et al. (42) and Suido et al. (39), using as a substrate one of the following synthetic chromogenic peptides: Nα-benzoyl-dl-Arg β-naphthylamide (BANA), N-carbobenzoxy-Gly-glY-Arg β-naphthylamide (GRNA), gly-Phe β-naphthylamide (GFNA), and Gly-Pro 4-methoxy-β-naphthylamide (GPNA) (all from Sigma). The reaction mixture was prepared by mixing 0.3 ml of a 1 mM substrate solution (100 mM dimethyl sulfoxide), 0.9 ml of 0.1 M Tris-HCl (pH 7.5), and 0.3 ml of protein sample in the same buffer. After incubation at 37°C for 2 h, 0.6 ml of a solution containing 0.5 mg of fast garnet GBC {2-methyl-4-[(2-methylphenyl)-azo]benzene diazonium sulfate; Sigma}/ml of 1 M acetic acid buffer (pH 4.2)–10% Tween 20 was added and the absorbance of the resulting β-naphthylamide–dye complex was measured spectrophotometrically at 525 nm. A reaction mixture without the sample was also incubated as a control to correct for possible spontaneous hydrolysis of the substrate. One unit of enzyme activity was defined as the amount of enzyme catalyzing the release of 1 nmol of β-naphthylamide in 2 h at 37°C. Enzyme preparations were also tested for their abilities to cleave two synthetic peptides related to collagen, 2-furanacryloyl-Leu-Gly-Pro-Ala (FALGPA) and p-phenylazobenzyloxycarboxycarbonyl-Pro-Leu-Gly-Pro-Arg (pZ-peptide), by the procedure of van Wart and Steinbrink (43).

Viable bacteria were also tested for their ability to hydrolyze the GPNA substrate. Cells were harvested by centrifugation from 10-ml aliquots of growing cultures, washed once with 10 ml of phosphate-buffered saline (PBS), and resuspended in 10 ml of PBS. Fifty microliters of the washed-cell suspension was added to the assay mixture described above.

Enzymatic cleavage of native host proteins by the purified streptococcal protease was evaluated by SDS-PAGE. The substrate protein (8 μg) was incubated at 37°C with 0.4 μg of enzyme in a solution containing 50 μl of 0.1 M Tris-HCl buffer (pH 7.5) and 1 μl of toluene preservative for 18 h. As controls, proteins were also incubated under the same conditions without the enzyme. The reaction occurring in the 50-μl incubation mixture was stopped by the addition of an equal volume of 2× SDS-PAGE sample buffer. Human placental type IV collagen, BSA, human fibrinogen, rabbit muscle myosin, and collagenase (type VII grade) of Clostridium histolyticum were purchased from Sigma Chemical Company. Murine laminin and human fibronectin were from Collaborative Biomedical Products (Bedford, Mass.).

Protease activity was also determined in an agar gel diffusion assay. Concentrated culture supernatant fluid, containing up to 40 μg of streptococcal proteins, was added to wells punched in a 2% agar plate containing 0.25% gelatin and 50 mM Tris-HCl buffer (pH 8) and incubated 18 h at 37°C. Zones of proteolysis were detected by flooding the plate with 3 ml of 25% trichloroacetic acid (TCA) to precipitate undigested gelatin. The diameter of the sample well was subtracted from the diameter of the clear zone of proteolyis around the well (27). One unit of enzyme activity was defined as that amount of protein that produced a 1-mm zone of gelatin hydrolysis.

Gelatinase activity of some protein solutions was determined by a dot assay (20 μl) using a developed, single-side coated X-ray film (Cronex MRF34 medical recording film with clear base; E. I. Du Pont de Nemours and Co., Willmington, Del.) (32).

Environmental effects.

The effect of pH on the appearance of protease in the culture supernatant was determined by growing S. gordonii in CDM–0.075% amino acids with and without 50 mM HEPES buffer. Samples were collected at the end of logarithmic growth, centrifuged at 6,000 × g, filter sterilized, and either assayed directly for protease activity by zymography or hydrolysis of GPNA or gelatin or assayed after precipitation with ammonium sulfate at 55 to 80% saturation. To determine whether pH had a direct effect on the activity of the protease, the 55 to 80% ammonium sulfate fraction was dissolved in 0.1 M Tris-HCl buffer at pH values ranging from 6 to 10 and incubated with GPNA dissolved in the corresponding buffer. The effect of aeration on protease production by streptococci was evaluated by comparing GPNA-hydrolyzing activities in samples of static cultures and shaken cultures (200 rpm) obtained at equivalent cell densities (150 Klett units).

Stability and inhibition.

The heat stability of the protease in the 55 to 80% ammonium sulfate fraction was determined by incubation at 30, 40, 50, 60, and 80°C for 10 min in the assay buffer followed by determination of enzymatic activity by the GPNA hydrolysis assay. To determine its susceptibility to known protease inhibitors, the purified S. gordonii protease was preincubated for 2 h at 37°C with 1 mM EDTA, 2 mM p-chloromercuribenzoate (PCMB), 1 mM N-tosyl-l-phenylalanine chloromethyl ketone (TPCK), 1 mM N-α-p-tosyl-l-lysine chloromethyl ketone (TLCK), 10 mM dithiothreitol (DTT), 10 mM iodoacetamide, or 1 mM phenylmethylsulfonyl fluoride (PMSF) and the enzyme activity was measured by the GPNA and gelatin hydrolysis assays.

Fast protein liquid chromatography (FPLC).

The proteins obtained by ammonium sulfate precipitation (55 to 80% saturation) were applied to a Mono Q HR 5/5 ion-exchange column (Pharmacia, Uppsala, Sweden) equilibrated with 20 mM Tris-HCl, pH 8. Proteins were eluted from the column with a linear gradient of NaCl (0 to 1 M) over 55 min and collected in 1-ml fractions. Fractions containing GPNA- and gelatin-hydrolyzing activities were pooled, concentrated by ultrafiltration (CentriCon-10; Amicon Corp., Beverly, Mass.), dialyzed against PBS, and then applied to a Superose 12 HR 10/30 molecular sieve column (Pharmacia) equilibrated with 150 mM NaCl–50 mM Tris-HCl (pH 8). Enzyme-containing fractions from five Superose 12 column runs were pooled, treated with 20 μg (50 U) of PMSF-treated N-acetylmuramidase (mutanolysin; Sigma)/ml for 60 min at 37°C, and then centrifuged at 27,000 × g for 1 h. The soluble fraction was collected and chromatographed on the Superose 12 column.

Preparation of mutanolysin.

Mutanolysin (1 mg/ml) was pretreated with 1 mM PMSF for 1 h at 4°C to inactivate traces of contaminating proteases. The enzyme solution was then dialyzed against PBS for 24 h to inactivate (19) and remove residual PMSF before further use. No protease activity was detected in the PMSF-treated mutanolysin when BSA and rabbit muscle myosin were used as substrates.

RESULTS

Growth conditions and protease release.

S. gordonii was grown aerobically in CDM containing a mixture of free amino acids (0.15 or 0.075% [wt/vol]) and in CDM–0.5% gelatin (Fig. 1); the growth rates, determined from turbidity measurements, were 0.29, 0.14, and 0.09 doubling/h, respectively. At 36 h, the growth rate in gelatin had gradually slowed to 0.008 doubling/h. Further reductions in the amounts of amino acids or gelatin reduced the cell yield significantly, and there was no growth detected in CDM devoid of both amino acids and gelatin. Streptococci did not grow at all in CDM containing an equivalent amount of gelatin confined in dialysis tubing that was equilibrated with the medium for 48 h before inoculation (Fig. 1). Analyses of spent culture media, derived from 18- to 20-h cultures, showed that S. gordonii produced 0.41 gelatin-hydrolyzing units of protease/ml in gelatin-CDM and 0.351 U/ml in CDM–0.075% amino acids but none in CDM–0.15% amino acids. Together, these observations indicated that the amino acids required for streptococcal growth were supplied through the hydrolysis of gelatin by cell-associated or extracellular proteases and was not due to free amino acids or small polypeptides that might have been present as contaminants in the gelatin preparation.

FIG. 1.

Growth of S. gordonii in CDM broth containing free amino acids or gelatin. Fifty milliliters of culture medium in a spectrophotometric side-arm flask was inoculated with bacteria and incubated at 37°C without shaking. Symbols: □, CDM–0.15% amino acids; ▴, CDM–0.075% amino acids; ○, CDM–0.5% gelatin; •, CDM with an equivalent amount of gelatin separated from the streptococci by the use of dialysis tubing.

To determine the effects of free amino acids on the production of extracellular protease, samples of spent culture medium were collected at intervals during growth of streptococci in CDM containing 0.075 or 0.75% amino acids and analyzed for their abilities to cleave GPNA (Fig. 2) and gelatin (data not shown). Protease(s) was released by bacteria throughout the logarithmic growth phase in low-amino-acid medium; however, it was not detected until the end of the logarithmic growth phase in the high-amino-acid medium. Moreover, the amount of aminopeptidase activity produced in the latter medium was only 16% of that produced in the low-amino-acid medium. Production of gelatinase activity followed a similar pattern, reaching a maximum level of 0.48 U/ml in low-amino-acid medium but being undetectable in high-amino-acid medium. In comparison, the GPNA-hydrolyzing activity of viable bacteria was unaffected by the amino acid concentration in the culture medium; activity was directly proportional to cell numbers. Maximum activity, attained at peak turbidity (200 Klett units), was 250 enzyme units (EU)/ml for the cellular fraction and 6.3 EU/ml in the spent culture medium. Comparison of SDS-PAGE profiles of proteins from spent CDM of the two cultures described in Fig. 2 did not reveal any qualitative or quantitative differences (data not shown), indicating that the protease might comigrate with and be masked by another protein in the gel or that the protease comprised a very small percentage of the total streptococcal extracellular protein and was not visibly stained. Spent culture medium collected during the logarithmic phase of S. gordonii growth was devoid (<5 ng/ml) of the cytoplasmic histone-like protein (36), indicating that autolysis did not occur and, thus, the detected protease activity was not of cytoplasmic origin.

FIG. 2.

Effect of amino acids on protease release during growth of S. gordonii. At various time intervals, 10-ml aliquots were collected, filter sterilized, and assayed for aminopeptidase activity, using the GPNA chromogenic substrate. Symbols: □, turbidity of CDM–0.75% amino acids; ○, turbidity of CDM–0.075% amino acids; ■, protease concentration in CDM–0.75% amino acids; •, protease concentration in CDM–0.075% amino acids.

The repressive effect of free amino acids on the release of protease into the culture medium was confirmed when a mixture of 20 amino acids (40) was added at a concentration of 0.3% to a culture of S. gordonii that was actively producing protease. Preliminary experiments had shown that 0.3% amino acids prevented the production of extracellular protease and did not significantly increase the growth rate of the streptococci above that obtained in CDM containing 0.075% amino acids. Figure 3 shows that the appearance of aminopeptidase activity in the culture medium was interrupted for 2 h after amino acid addition and then resumed at the initial rate for the duration of the logarithmic growth phase. At the point of maximum growth, the total amount of aminopeptidase activity was 39% of that detected in the control culture that did not receive supplemental amino acids. Gelatinase production was similarly repressed by the addition of the amino acid mixture (data not shown).

FIG. 3.

Interruption of protease release by the addition of free amino acids to a growing culture. S. gordonii was inoculated into a flask containing 100 ml of CDM–0.075% amino acids, and growth was monitored turbidometrically. After 5.5 h (verticle line), one-half of the culture volume was removed and placed in a separate flask, to which a 3% aqueous solution of amino acids was added to a final concentration of 0.3%; the other half of the culture received water instead of supplementary amino acids. Samples were removed from both cultures at intervals, filter sterilized, and assayed for aminopeptidase activity, using the GPNA chromogenic substrate. Both cultures had essentially identical turbidity values throughout the experiment; thus, only the growth curve for the amino acid-supplemented culture is shown. Symbols: ■, turbidity; •, protease concentration in the control culture; ○, protease concentration in the amino acid-supplemented culture.

The production of extracellular protease by S. gordonii CH1 in CDM was also affected by the pH of the medium and by the amount of aeration. Maximum protease production (10 GPNA EU/ml and 0.47 gelatinase EU/ml) occurred during growth in medium that was maintained between pH 6.5 and 7.0, whereas only 2.5 GPNA EU/ml and negligible gelatinase activity were detected in cultures in which the pH was allowed to reach 5.5 (Table 1). Neutralization of the latter spent culture medium did not increase the enzyme activities, indicating that the enzyme(s) was either produced in smaller amounts at pH 5.5 or irreversibly denatured. Cultures that were grown in flasks on a gyratory shaker (200 rpm) or statically produced 3 and 8 GPNA EU of protease/ml at a turbidity of 150 Klett units, respectively. It is not clear from these observations whether increased oxygenation of the medium inactivated free enzyme or repressed its synthesis by or release from the cells.

TABLE 1.

Effect of pH on release of protease into the culture medium by S. gordonii

Property	Value in culture mediuma:
	CDM + HEPES	CDM
Optical densityb	1.478	1.180
Total protein	1.75 μg/ml	1.09 μg/ml
Zymogram activityc	6–8	<1
Gelatin hydrolysisd	0.44 EU/ml	0.11 EU/ml
GPNA hydrolysise	10 EU/ml	2.5 EU/ml

^aS. gordonii was grown in CDM–0.075% amino acids with and without 50 mM HEPES buffer. The initial pH of both media was 7.2, and the final pH values were 6.7 in CDM + HEPES and 5.5 in CDM. 

^bMaximum turbidity of the culture at 600 nm. 

^cFive micrograms of protein from concentrated spent culture medium was fractionated by SDS-PAGE for zymography. The dimensions of the colorless zone of proteolysis in the Coomassie blue-stained gel are expressed in millimeters. 

^dForty micrograms of concentrated proteins was assayed in a gelatin hydrolysis plate. 

^eThirty-five micrograms of protein was used in the GPNA assay. 

Identification of protease activities.

Fifteen liters of spent culture medium (CDM–0.075% amino acids) was concentrated by ultrafiltration and fractionated by sequential ammonium sulfate precipitation at 30, 55, and 80% saturation points. Analyses of the resulting protein fractions, using synthetic peptide substrates, detected three enzyme specificities (Table 2). The 0 to 30% precipitate contained a trypsin-like protease(s) that cleaved at the arginine residues of BANA and GRNA and a second enzyme that recognized glycine-phenylalanine of the GFNA substrate. The proteins present in the 55 to 80% fraction included an enzyme that cleaved at the proline site of GPNA (Table 2) and the collagen-related peptides FALGPA and pZ-peptide (data not shown). Gelatin was hydrolyzed by enzymes in both the 0 to 30% and 55 to 80% fractions. Subsequent experiments focused on the protease(s) contained in the 55 to 80% ammonium sulfate fraction due to its specificity for collagen-like substrates and its higher solubility. The activity of the collagenase-like protease was inhibited (>90%) by 1 mM PMSF but not (<5%) by 10 mM iodoacetate, 2 mM PCMB, or 1 mM TLCK, TCPK, EDTA, or DTT. When stored at 4°C, the enzyme was stable for at least 2 weeks, and it was stable at −20°C for several months. Also, the 55 to 80% ammonium sulfate fraction could be heated at 50°C for 10 m without any loss of the enzymatic activity. Preliminary experiments also indicated that the protease activity had a pH optimum between 7.0 and 9.0. When assayed above pH 9.5 or below pH 5.0, no proteolytic activity was observed, and the enzyme was not activated when the pH was adjusted to 7 (data not shown).

TABLE 2.

Protease (aminopeptidase) activity in ammonium sulfate fractions of spent culture medium^a

Substrate	A525 value of fraction:
	0–30%	30–55%	55–80%
BANA	0.457	<0.01	<0.01
GRNA	0.720	<0.01	<0.01
GFNA	0.318	<0.01	<0.01
GPNA	<0.01	<0.01	0.385

^aProteins were precipitated sequentially with ammonium sulfate at 30, 55, and 80% saturation points. The precipitates were dissolved in and dialyzed against Tris-HCl buffer, and 100 μg of protein was assayed for enzymatic activity on the indicated substrates. 

SDS-PAGE analyses of concentrated spent culture media obtained from logarithmic-growth-phase cultures revealed 34 distinct bands corresponding to molecular masses ranging from 10 kDa to approximately 400 kDa (Fig. 4, lane 2). Gelatin zymography showed that all of the enzymatic activity was in a diffuse zone at the top of the SDS-polyacrylamide-gelatin matrix, indicating a molecular mass in excess of 400 kDa (Fig. 4, lanes 4 and 5). Silver staining of the same samples after electrophoresis on SDS-polyacrylamide gels devoid of gelatin revealed two weakly staining bands of this size (Fig. 4, lane 3). To determine whether the electrophoretic mobility of the protease was affected by gelatin in the polyacrylamide gel, an enzyme sample was first fractionated in a gelatin-free SDS-PAGE gel and then blotted electrophoretically for 24 h onto a 0.1% gelatin–7.5% polyacrylamide composite gel for subsequent enzymatic analysis. In this blot, the proteolytic activity appeared as a cluster of three bands corresponding to molecular masses of approximately 200, 190, and 180 kDa (Fig. 4, lane 6). Since the migration distances of major streptococcal proteins, seen within the background stain when viewed under bright illumination, were not affected by the presence or absence of gelatin in SDS-PAGE, we concluded that the protease was binding to gelatin during electrophoresis and that 180 to 200 kDa was a more accurate and credible estimate of the molecular mass of the protease in the spent culture medium. It was not clear whether one or all of the gelatinase activities detected on the zymogram could also hydrolyze the GPNA substrate.

FIG. 4.

Electropherograms and zymograms of extracellular proteins of S. gordonii. Lanes 1 to 3, SDS-polyacrylamide gels: lane 1, molecular mass standards; lane 2, 5 μg of proteins from ultrafilter-concentrated spent culture medium (silver stained); lane 3, 5 μg of proteins precipitated by 55 to 80% saturated ammonium sulfate. Lanes 4 to 6, gelatin-polyacrylamide gels: lane 4, 25 μg of concentrated spent culture medium; lane 5, 25 μg of the same material in lane 2; lane 6, 25 μg of the same material as in lane 4 that was separated first by SDS-PAGE and then transferred electrophoretically (blotted) onto the zymogram gel. Lanes 1 and 4 to 6 were stained with Coomassie blue; lanes 2 and 3 were stained with silver nitrate.

Purification of a peptidoglycan-associated protease.

When the proteins in the 55 to 80% ammonium sulfate fraction were subjected to FPLC–anion-exchange chromatography on a Mono Q column (Fig. 5A), the GPNA- and gelatin-cleaving enzyme(s) eluted at 0.3 M NaCl (column fractions 24 and 25). The active fractions were pooled, dialyzed against PBS, and loaded onto a Superose 12 column (Fig. 5B). On this column, the GPNA-hydrolyzing activity eluted in fractions 24 and 25, which corresponded to a molecular mass of 200 kDa as determined by the elution times of standard proteins. Analyses of the chromatography fractions by SDS-PAGE showed that there were 18 to 20 proteins in the Mono Q column pool (Fig. 6, lane 3) and 6 proteins (4 major and two minor) in the Superose 12 column pool (Fig. 6, lane 4). There are three possible explanations for the presence of low-molecular-weight components in proteinase-containing fractions from the Superose 12 column: (i) the protease was released by S. gordonii as a stable complex containing other proteins, (ii) the protease complexed with other proteins during laboratory manipulations, or (iii) some degradation of larger proteins occurred during column chromatography. A second passage of the protease-containing fraction through the Superose 12 column did not change its protein banding pattern on SDS-PAGE, indicating that the companion proteins did not result from inefficient chromatographic resolution or proteolytic degradation of larger proteins. Also, the protein complexes were refractory to dissolution by chromatography on a Superose 12 column equilibrated with 1% SDS (data not shown).

FIG. 5.

Chromatographic purification of the 98-kDa protease. (A) Separation of ammonium sulfate-fractionated (55 to 80% saturation) proteins by FPLC–ion-exchange chromatography on a Mono Q column. The NaCl gradients used were 0 to 0.5 M in 45 min and 0.5 to 1 M in 10 min. (B) Separation of proteins in the protease-containing fractions from the elution profile shown in panel A on a Superose 12 gel filtration column. (C) Superose 12 column chromatography of protease-containing fractions from the elution profile shown in panel B after treatment with mutanolysin. Protease-containing regions of the chromatograms are indicated by bars.

FIG. 6.

Electropherograms of the protease at different stages of its purification. Lane 1, 20 μg of molecular mass standards (Coomassie blue stain); lane 2, 25 μg of protein precipitated by 80% saturated ammonium sulfate; lane 3, 2.5 μg of Mono Q chromatography fractions 24 and 25; lane 4, 2.0 μg of Superose 12 chromatography fractions 24 and 25; lane 5, 0.5 μg of mutanolysin-treated and Superose 12-purified 98-kDa protease. Lanes 2 to 5 are stained with silver nitrate.

To determine whether the protease and its associated proteins were complexed with peptidoglycan fragments released from the cell surfaces (1, 26), the enzymatically active pool from the Superose 12 column was incubated with PMSF-mutanolysin and reapplied to the Superose 12 column for chromatographic separation. The peptidase activity of the mutanolysin-treated pool eluted in fractions 30 and 31 (Fig. 5C) rather than in the previous position of fractions 24 and 25 (Fig. 5B), indicating a significant reduction in molecular mass. There was no detectable loss of enzyme activity after treatment with mutanolysin. Examination of the protease-containing fractions by SDS-PAGE revealed a single band of 98 kDa (Fig. 6, lane 5). The 98-kDa protease was inactive in zymography but hydrolyzed the gelatin-containing emulsion on developed X-ray film in dot assays under nondenaturing conditions, whereas the 200-kDa protein was active in both assays (data not shown), indicating that denaturation of the enzyme by SDS-PAGE was irreversible in the absence of peptidoglycan components. Analyses for muramic acid indicated that the spent culture medium contained 2.26 μg of peptidoglycan/ml and that peptidoglycan comprised 18 to 24% (dry weight) of the extracellular components in the 55 to 80% ammonium sulfate fraction. No muramic acid was detected in mutanolysin-treated protease preparations.

The yields of protease during the purification steps are summarized in Table 3. The final product represented 0.05% of the initial protein in the spent culture medium and 62% of the original GPNA activity. Total purification was 1,350-fold, as indicated by the increase in specific activity. Attempts to verify the purity of the protease by N-terminal amino acid sequencing have not been successful, suggesting that the N terminus may be blocked.

TABLE 3.

Purification of the extracellular 98-kDa protease of S. gordonii

Preparation	Total protein (mg)	Total activity (GPNA EU)	Sp act (U/mg)
Concentrated spent culture mediuma	30	941	31
Ammonium sulfate (55–80%) precipitate	6	297	49
Mono Q chromatography fraction	0.114	689	6,044
Superose 12 chromatography fraction	0.014	586	41,857

^aFifteen liters of spent culture medium was concentrated 15-fold by tangential-flow ultrafiltration. 

Enzymatic activity of the 98-kDa protease.

When the 98-kDa protease (2 μg) was incubated at 37°C for 60 min with the collagen-like substrates (FALGPA and pZ-peptide), it catalyzed absorbance changes of 0.421 and 0.503, respectively. In addition, the protease cleaved human type IV collagen and the Aα chain of fibrinogen but not laminin (Fig. 7), serum albumin, fibronectin, or myosin (data not shown). In comparison, incubation of type IV collagen with 0.2 U of collagenase of C. histolyticum under the same conditions resulted in nearly complete disappearance of all Coomassie blue-staining bands (data not shown).

FIG. 7.

Electropherograms of proteins treated for 18 h with the purified protease of S. gordonii. Lane 1, 8 μg of human placental collagen IV; lane 2, 8 μg of human placental collagen IV after incubation with 400 ng of purified streptococcal protease; lane 3, 8 μg of human fibrinogen; lane 4, 8 μg of human fibrinogen after incubation with 400 ng of purified streptococcal protease; lane 5, 8 μg of murine laminin; lane 6, 8 μg of murine laminin after incubation with 400 ng of purified streptococcal protease. The proteins were stained with Coomassie blue. The component chains of each protein are indicated by arrowheads in the margins of the untreated control lanes. The positions of molecular mass markers are indicated on the left.

DISCUSSION

We have identified and purified an extracellular 98-kDa protease of S. gordonii Challis that hydrolyzes gelatin, native collagen type IV, fibrinogen, and glycine-proline-containing synthetic peptides. Analysis of the extracellular enzyme by gel filtration chromatography and PMSF-mutanolysin treatment indicated that streptococci release the protease as a complex with segments of peptidoglycan chains. The peptidoglycan substituents appear to be responsible for the apparent size heterogeneity of the protease in zymogram blots (Fig. 4) and in column chromatography, since their removal yielded a single protease band (Fig. 6). Production of the extracellular protease was interrupted by the addition of a mixture of free amino acids (>0.3%); however, growth in medium containing gelatin as the sole source of amino acids or in medium with less than 0.075% amino acids resulted in continuous enzyme production during the logarithmic phase of growth. This observation indicates that streptococci might produce significant amounts of extracellular enzymes that can degrade host proteins such as collagen IV and fibrin when the bacteria are located within the amino acid-depleted but protein-rich thrombotic valvular vegetations of infective endocarditis. Streptococci in these lesions grow at a slow but undetermined rate and can reach a density of 10¹⁰ cells per g of tissue (17). We found that S. gordonii Challis had an 11-h generation time in CDM-gelatin and a 71-h generation time in newborn calf serum when incubated aerobically at 37°C (data not shown). This growth rate difference was expected since the concentration of free amino acids in bovine and human blood (0.01%) is far below the optimum amount required for bacterial growth and is below the repression level for protease production. Our findings are consistent with reports by Rogers et al. (35) and Straus (38), who found that the appearance of protease in the culture medium of S. sanguis was maximal under nutritionally limited conditions and that protease was not produced during the logarithmic growth phase in enriched media.

The association of the 98-kDa protease with peptidoglycan fragments in spent culture medium and the presence of GPNA-hydrolyzing activity on intact cells indicate that the enzyme passes through a transition state as a cell wall-associated protein. Release of the protein to the extracellular millieu is likely a consequence of cell wall turnover (1, 26). This conclusion is supported by the observation that mutanolysin treatment of viable logarithmic-phase streptococci liberated a protease whose substrate specificity and molecular mass were identical to those of the extracellular enzyme reported here (data not shown). It is not known whether the protease is covalently linked to the peptidoglycan matrix or is bound through ionic and/or hydrophobic interactions. A covalent linkage is suggested since gel filtration chromatography in 6 M urea, 1 M NaCl, or 1% SDS gels did not resolve the aggregated proteins in the present study. Similar high-molecular-weight complexes have been reported for a fibrillar protein of S. sanguis (28) and for an α-amylase receptor of S. gordonii (7). Comparison of the GPNA-hydrolyzing activities of washed viable bacteria and spent culture medium indicated that the extracellular enzyme represented approximately 3% of the total during peak production. These observations are consistent with a model in which the protease is secreted through the plasma membrane to the internal surface of the cell wall, where it becomes linked to newly synthesized peptidoglycan. The protease-peptidoglycan complex is gradually displaced toward the outer surface of the cell wall as new cell wall synthesis occurs at the plasma membrane interface (34) and as old cell wall matrix is released at the outer surface by the action of other cell surface enzymes (1, 24). Thus, release of the protease to the environment may reflect a natural turnover and shedding of the surface components. However, two other viridans group streptococci, S. mutans and S. sanguis, have been found to not release peptidoglycan fragments into the culture medium during growth (30). This apparent discrepancy from our findings may be explained by the use of TCA precipitation methods by these investigators in their determination of enzymatic degradation of radiolabeled peptidoglycan. The release of TCA-soluble breakdown products into the culture medium might not be detected if streptococcal surface proteins remained closely associated with peptidoglycan fragments; these complexes were readily precipitated by cold 10% TCA in the present study (data not shown).

The observation that the addition of a mixture of amino acids to a growing culture of S. gordonii causes abrupt cessation of extracellular protease production without affecting cell-associated enzyme activity indicates that regulation may affect the synthesis of an autolytic enzyme that mediates the release of peptidoglycan and protease complexes rather than directly repress protease gene transcription. This conclusion, however, must be confirmed by further experimentation.

Teleologically, localization of streptococcal proteases in the outer regions of the cell wall matrix would be advantageous for nutrient acquisition during infection. Host proteins that make contact with the bacterial surface will be digested and their breakdown products will be transported into the bacterium before they diffuse from the site. In infected thrombotic valvular vegetations of endocarditis, in which the levels of soluble protein are low, secreted protease might enable the streptococci to scavenge amino acids from structural proteins of the valve or fibrin clot. Proline-containing peptides are taken up by some oral streptococci and hydrolyzed further by cytoplasmic peptidases (5). Local destruction of tissue proteins would likely have pathologic consequences. Indeed, highly proteolytic strains of Enterococcus faecalis were reported to be more virulent than weakly proteolytic strains in a rabbit model of infective endocarditis (15). A cell-associated collagenase has also been detected in group B streptococci (18), and it hydrolyzes the same collagen analogs as the S. gordonii protease reported here does. In addition, proteases may be important in normal processing of other surface proteins of streptococci; for example, the glycosyltransferase of S. sanguis is converted from 174 to 156 kDa at the cell surface (12). A surface protease of S. sanguis also cleaves an 80-kDa adhesin protein and alters its reactivity with antibodies (23), and the P1 adhesin of S. mutans is released into the culture medium by an endogenous protease designated as a surface protein-releasing enzyme (24). These alterations in cell surface composition might enable the bacterium to escape fouling of its surface by immunoglobulins and other host proteins. An unobstructed cell surface should assure the availability of adhesins for bacterial attachment to host surfaces. S. gordonii adheres avidly to proline-rich salivary proteins present in pellicles adsorbed to tooth surfaces (11) and can degrade the same proteins in saliva (3). The relationship between the proline-rich salivary-protein-degrading enzyme of S. gordonii and the 98-kDa protease reported here remains to be determined. Further studies are also necessary to determine the pathogenic properties of its gelatinase and collagenase-like activity during infective endocarditis.