A New Alkaline pH-Adjusted Medium Enhances Detection of β-Hemolytic Streptococci by Minimizing Bacterial Interference Due to Streptococcus salivarius

Abstract

A new selective medium (CNA-P) that reduces or eliminates the inhibitory activity of bacteriocin-producing Streptococcus salivarius against β-hemolytic streptococci has been developed and compared with sheep blood agar (SBA) for the sensitive detection of small numbers of β-hemolytic streptococci in clinical specimens. CNA-P has as its basis a commercial medium (Difco Columbia CNA agar) supplemented with 5% (vol/vol) sheep blood, and the CNA is further modified by addition of 100 mM PIPES buffer [piperazine-N,N′-bis(2-ethanesulfonic acid)] (pH 7.5) to maintain cultures at an alkaline pH during incubation. CNA-P was shown to inhibit the production and/or release of four different types of S. salivarius bacteriocins or bacteriocin-like inhibitory molecules. The efficacies of CNA-P and SBA for detection of β-hemolytic streptococci in 1,352 pharyngeal samples from 376 children were compared. The β-hemolytic streptococcal isolates recovered from the samples included 314 group A (S. pyogenes), 61 group G, 33 group B, and 5 group C streptococci. Of 314 samples that yielded S. pyogenes, 300 were positive on CNA-P (96%) and 264 (86%) were positive on SBA. A significantly greater number of S. pyogenes isolates from these samples were recovered only on CNA-P (50 of 314) compared with the number of isolates recovered only on SBA (14 of 314). In addition, the degree of positivity, a measure of the total numbers of S. pyogenes isolates on the plate, was significantly higher on CNA-P than on SBA (2.40 versus 2.07; P < 0.001). Interestingly, CNA-P was also found to enhance the hemolytic activity of streptolysin O, allowing detection of streptolysin S-deficient S. pyogenes strains which might otherwise go undetected on SBA and other isolation media.

Lancefield group A, B, C, and G β-hemolytic streptococci are major causative agents of a number of illnesses, including acute pharyngitis, skin infections, and sepsis. Group A streptococcus (Streptococcus pyogenes) pharyngitis is of particular importance in pediatric populations because of the ensuing risk of rheumatic fever and glomerulonephritis (3). Recent increases in the incidence of invasive streptococcal disease (15) and a resurgence of rheumatic fever emphasize the need for accurate and sensitive detection of S. pyogenes in cultures (22). In addition, evidence for group C and G streptococcus involvement in outbreaks of pharyngitis, especially in college students and older adults, continues to accumulate (3, 10, 41). While group B streptococci are only rarely implicated in pharyngitis, they pose a significant risk to neonates if the mother is a vaginal carrier of the organism (5). More recently, group B streptococci have been found to be an important cause of invasive infections in adults aged 60 and older and in adults with underlying chronic illnesses such as diabetes mellitus, renal failure, and liver disease (4, 14).

In this laboratory, there is a particular interest in the production of bacteriocin-like inhibitory substances (BLISs) by oral streptococci (34). Bacteriocins are loosely defined as ribosomally synthesized proteinaceous antimicrobial agents produced by members of many bacterial species that inhibit the growth of closely related bacteria. The acronym BLIS followed by the producer strain designation (e.g., BLIS N from Streptococcus salivarius N) has been recommended for use as a temporary designation when putative bacteriocins are first detected. When the corresponding structural gene is identified, a permanent name may then be assigned (19). In this paper, we have for convenience used the acronym BLIS to refer collectively to inhibitory agents that either are known to be bacteriocins or are as yet incompletely characterized.

Previous studies (8, 12, 28, 30–32, 34, 40) have suggested that certain members of the normal oral microbiota may protect against S. pyogenes infection. Our own research has shown that S. salivarius, a predominant oral organism, is frequently a producer of BLISs (36), with at least six different BLIS types being distinguishable (9). A feature of these BLISs is their strong in vitro activity against β-hemolytic streptococci and in particular against S. pyogenes (27, 34, 36). The present focus of this laboratory is to determine the possible role of BLIS-producing S. salivarius in preventing infection by or carriage of β-hemolytic streptococci in the oral cavity.

Knowing that BLIS-producing S. salivarius is able to interfere with the growth of β-hemolytic streptococci in a simultaneous antagonism test on agar medium (9), we explored the possibility that the presence of BLIS-producing S. salivarius in clinical specimens may inhibit recovery of small numbers of β-hemolytic streptococci and lead to false-negative reporting of these bacteria in specimens. With this in mind, we sought to develop a modified blood agar medium that would minimize or abolish the production and/or extracellular release of S. salivarius BLISs.

A well-known feature of many BLISs produced by gram-positive bacteria is their greater antibacterial activity at lower pH values. At alkaline pH they become less stable and their activity decreases. In addition, at alkaline pH, more BLIS molecules remain associated with the producer cell and less BLIS activity is released into the surrounding medium (for a review, see reference 19). By controlling the drop in pH of cultures through the use of buffering agents, we sought to reduce the production, activity, and/or release of BLISs from producer cells. In this paper we report on the development and clinical testing of an alkaline pH-adjusted blood agar medium that enhances detection of β-hemolytic streptococci in the presence of BLIS-producing S. salivarius.

MATERIALS AND METHODS

Strains, media, and growth conditions.

All bacteriological media were obtained from Gibco Laboratories (Madison, Wis.) unless otherwise stated, and all chemicals were from Sigma (St. Louis, Mo.). All test media were supplemented with 5% (vol/vol) defibrinated sheep blood (Life Technologies). Prototype strains were routinely cultured on Columbia agar base supplemented with 5% (vol/vol) human blood and 0.1% CaCO₃ (BA-Ca) or in Todd-Hewitt broth (THB). Mitis-salivarius agar (Difco, Detroit, Mich.) was used for isolation and presumptive identification of S. salivarius (on the basis of colony morphology) (7). Test media included Columbia agar base supplemented with sheep blood (SBA), SBA with 10 μg of colistin sulfate per ml and 10 μg of oxolinic acid per ml (CO agar [26]), SBA with 1 μg of crystal violet per ml (CV-1) or 2 μg of crystal violet per ml (CV-2), and SBA with 25 μg of spectinomycin per ml (SP-25). The new selective medium developed in this study (CNA-P) consists of Difco Columbia CNA agar (which contains 10 μg of colistin sulfate per ml and 15 μg of nalidixic acid per ml) buffered with 100 mM PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)] and adjusted to pH 7.5 prior to autoclaving and then supplemented with 5% sheep blood. Agar media were poured to a depth of 4 mm and were stored at 4°C in plastic bags until required. PIPES (pK_a, 6.76), TES [N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid; pK_a, 7.40], and sodium phosphate buffer (pK_a, 7.20) were added to the appropriate agar media at molarities ranging from 10 to 200 mM, and the pH was adjusted in increments of 0.5 unit from 6.5 to 8.0.

The test strains used for the development of the new medium are listed in Table 1. Clinical specimens included a total of 1,352 paired throat and tongue swab specimens collected from 376 children (ages, 5 to 11 years) between February and May 1997. The number of samples per child ranged from one to nine. Throat swab specimens for culture were taken by trained personnel by using Dacron swabs (Life Technologies) and were plated within 2 h on SBA and CNA-P. The inoculum was streaked into four quadrants for isolation by a standardized procedure (21). The plates were incubated anaerobically at 37°C and were examined after 24 and 48 h. β-Hemolytic colonies were semiquantitated, and the degree of positivity was recorded as 1+ (10 or fewer colonies), 2+ (11 to 100 colonies), 3+ (>100 colonies, with some present in the third quadrant of the plate), and 4+ (β-hemolytic colonies extending into the fourth quadrant).

TABLE 1.

Bacterial strains and BLIS status

Strain	Anti-S. pyogenes activity		P type	Type of BLIS molecule produced	Reference or source
	Simultaneousa	Deferreda
S. pyogenes
M1 (prototype)	−	−	000	BLIS negative	Culture collection, J. R. Tagg
EB1 (M-type 52)	−	−	000	BLIS negative	Tagg and Wannamaker (39)
S. sanguis K11	−	+	257	Putative nonlantibiotic	Skilton and Tagg (33)
S. mitis J	+	+	777	Not known	Culture collection, J. R. Tagg
S. salivarius
Min5	++	++	777	Lantibiotic	Culture collection, J. R. Tagg
20P3	++	++	677	Lantibiotic	Ross et al. (29)
N	−	++	226	Large heat labile	Culture collection, J. R. Tagg
F	++	−	000	Not known	Culture collection, J. R. Tagg
K33	−	−	000	BLIS negative	Culture collection, J. R. Tagg

^aAs described in Materials and Methods. 

Seeding experiments comparing selected media, buffering agents, and pH values.

Seeding experiments were designed to compare the effectiveness of different media, buffering agents, and pH values for the detection of small numbers of β-hemolytic streptococci in salivary specimens containing BLIS-producing S. salivarius. In general, fresh whole saliva was collected from subjects known to be infected with an S. salivarius strain producing a particular type of BLIS and also from subjects consistently found to be colonized with only BLIS-negative S. salivarius strains when representative isolates were examined in both deferred and simultaneous antagonism tests. One milliliter of fresh saliva was seeded with 10⁷ CFU of a β-hemolytic streptococcal culture per ml. A cotton swab that had been dipped into this preparation was used to deliver the primary inoculum to the surface of the test medium, followed by streaking into four quadrants with a wire loop and incubation for 18 h at 37°C in a 5% CO₂-enriched atmosphere. β-Hemolytic colonies were semiquantitated, and the degree of positivity was recorded as 1+ to 4+ as outlined above. The presence or absence of β-hemolytic colonies in each zone and the amount of growth of the nonstreptococcal oral bacteria on each medium was noted. In some experiments 10⁻³ and 10⁻⁴ dilutions of the seeded saliva were also spiral plated (model D; Spiral Systems Inc.) onto each medium and the S. pyogenes isolates were enumerated. To evaluate the abilities of the various test media to interfere with the inhibitory activities of different types of BLISs, a series of BLIS-producing S. salivarius prototype strains (Table 1) were coinoculated at 10⁷ CFU/ml with β-hemolytic streptococci (10⁶ CFU/ml) into either BLIS-negative saliva or saline prior to plating on each test medium.

Detection of bacterial interference.

Deferred and simultaneous antagonism tests were performed essentially as described by Tagg and Bannister (35) on BA-Ca. All incubations were for 18 h at 37°C in a CO₂-enriched atmosphere. Briefly, to detect deferred antagonism, a diametric streak of the test organism was incubated overnight; macroscopic growth was then removed by scraping the growth with the edge of a glass slide, and any remaining bacterial cells were killed by exposure to chloroform vapors for 30 min. After airing, the plates were streaked with samples from overnight THB cultures of nine standard indicator strains at right angles across the line of the original test strain. The pattern of inhibition against the indicator strains (P type) after incubation was recorded as a triplet code. The deferred antagonism test was modified as follows to detect the composite BLIS activities of the streptococcal populations present in clinical samples. Samples swabbed from the dorsal tongue surface were plated onto mitis-salivarius agar. After incubation, a swab was drawn through the confluent growth zone, and this sample was then applied as a diametric streak on the surface of a BA-Ca plate. Deferred antagonism testing of this mixed population was continued as described above. To detect simultaneous antagonism, an indicator lawn culture was first seeded by swabbing an 18-h THB culture of the indicator strain over the surface of a BA-Ca plate, and the colonies to be tested for BLIS production were then stab inoculated into the agar. Inhibitory activity was demonstrated by the interference with growth of the indicator lawn in the vicinity of stab cultures following incubation. The P types and the deferred and simultaneous anti-S. pyogenes activities of the BLIS-producing S. salivarius prototype strains are listed in Table 1.

Identification of cultures.

Initial detection of β-hemolytic streptococci was based on their typical hemolysis and colony morphology (21). Representative β-hemolytic colonies were subcultured onto SBA, and their Lancefield group identities were established serologically by latex agglutination (Oxoid) with group A, B, C, D, F, and G antisera. Lancefield-groupable, small-colony β-hemolytic streptococci (i.e., Streptococcus anginosus) were excluded from consideration in this study.

Statistical analysis.

The sensitivity of each medium for the detection of S. pyogenes was calculated by dividing the number of specimens on each agar exhibiting β-hemolysis due to S. pyogenes (determined by latex agglutination) by the total number of specimens in which S. pyogenes was detected (13). The McNemar chi-square test was used to compare isolation or nonisolation of S. pyogenes on each medium, and the Wilcoxon matched-pairs signed-ranks test was used to compare the degree of positivity of each sample (18).

RESULTS

Comparison of selective agars for reduction of nonstreptococcal oral bacteria.

When throat swab specimens are cultured, normal oral bacteria may overgrow the β-hemolytic streptococci present in the sample and make isolation difficult by masking hemolysis, producing toxic metabolites, or depleting the nutrients needed by the β-hemolytic streptococci. A number of selective media have been developed to circumvent this problem. We compared five selective media (CNA, CO agar, CV-1, CV-2, and SP-25) for their effectiveness at inhibition of normal oral bacterial populations and for their sensitivity of detection of small numbers of S. pyogenes cells when BLIS-producing (inhibitory) S. salivarius is present. For this comparative trial, fresh saliva was obtained from five subjects, two who were colonized with native populations of BLIS-producing S. salivarius and three who apparently lacked BLIS-positive S. salivarius (determined by deferred and simultaneous antagonism tests). These saliva specimens were seeded with an S. pyogenes culture and were swabbed onto each selective medium. A 10⁻³ dilution and a 10⁻⁴ dilution of the seeded saliva was also spiral plated onto each medium, and the S. pyogenes colonies were enumerated.

When cultures from each test medium onto which samples were swabbed were evaluated, CNA and CO agar appeared to be equally effective in suppressing the normal oral bacterial populations in both BLIS-positive and BLIS-negative saliva samples and both media were more effective than CV-1, CV-2, or SP-25. CV-1 and CV-2 were ineffective in inhibiting Neisseria. Although SP-25 effectively inhibited Neisseria, a variety of other oral organisms grew to large numbers, and this medium was also relatively inhibitory to the seeded S. pyogenes isolates. For these reasons, SP-25 was excluded from further consideration.

When the degree of positivity for the S. pyogenes isolates seeded into BLIS-negative saliva was evaluated on each medium, 3+ growth was detected from all samples. However, when S. pyogenes was seeded into BLIS-positive saliva, S. pyogenes was not detected on CO agar, CV-1, or CV-2 and could be detected at only a 1+ level on CNA. On CNA, the β-hemolytic colonies were completely absent from the primary inoculation zone and were visible only in the second quadrant, where they were relatively well separated on the agar surface from other oral bacteria, including BLIS-positive S. salivarius.

When samples of the S. pyogenes-seeded saliva samples were spiral plated onto the various test agars, the counts of S. pyogenes were slightly reduced for BLIS-positive saliva compared with the counts obtained for BLIS-negative saliva. In addition, the β-hemolytic colonies were detected only on the outer edges of the spiral plates. As the colony density increased toward the center of the spiral, the ability to detect β-hemolysis progressively decreased for the BLIS-positive saliva samples, whereas for BLIS-negative saliva samples, β-hemolytic colonies were seen throughout the spirals. The total counts of S. pyogenes in all saliva samples were less than the total count of the original S. pyogenes inoculum, reflecting the antibacterial activities of other salivary components (for a review, see reference 25).

While CO agar and CNA were similarly effective in inhibiting the normal bacterial populations, β-hemolysis by the S. pyogenes isolates seeded into BLIS-positive saliva was detected only when cultures were swabbed onto CNA. Also, the β-hemolysis by S. pyogenes seeded into BLIS-negative saliva was more visible in the confluent growth zone on CNA than on CO agar. For these reasons, CNA was selected as the base medium for the testing of the anti-BLIS activities of various buffer supplements. Since preliminary experiments with group C (Streptococcus zooepidemicus 4881) or group G (Streptococcus subsp. equisimilis strain W2580) streptococci gave results similar to those obtained with S. pyogenes, all further medium development was conducted with only S. pyogenes as the test organism.

Influence of buffering agents at different molarities and pH values.

Many BLIS molecules are relatively unstable in the alkaline pH range, and their production and/or release from the producer cell may also be decreased (for a review, see reference 19). With these two features in mind, PIPES, TES, and sodium phosphate buffers were added to CNA at molarities ranging from 10 to 200 mM, and the CNA was adjusted to pH 7.5. These media were compared for their effectiveness in suppressing S. salivarius BLIS activity. Suppression of BLIS activity was evaluated by recording the sensitivity of each medium (degree of positivity) for the detection of small numbers of S. pyogenes cells seeded in BLIS-positive saliva samples. BLIS-negative saliva samples were used as positive controls.

CNA with either PIPES (200 and 100 mM) or TES (200 mM) and SBA with PIPES (100 mM) showed significantly enhanced abilities to detect S. pyogenes in the presence of BLIS-positive S. salivarius compared with the ability of SBA or CNA without buffering agents (Table 2). On CNA with 200 mM phosphate buffer, no growth of S. salivarius or S. pyogenes was observed, and on CNA with 100 mM phosphate buffer, no β-hemolytic colonies were detected for either BLIS-positive or BLIS-negative saliva, which suggests that 200 mM phosphate buffer is inhibitory to both organisms and that 100 mM phosphate buffer significantly interferes with either the growth or the hemolytic activity of S. pyogenes. CNA with either phosphate, PIPES, or TES buffer at 50 mM was more effective than either unbuffered SBA or unbuffered CNA for detection of S. pyogenes. However, with PIPES, TES, or phosphate buffer at 10 mM, the ability to detect S. pyogenes was similar to that with unbuffered CNA, although detection ability was still slightly better than that with the use of unbuffered SBA. This may be the result of a lower terminal pH of the confluent culture growth on SBA (average agar surface pH, 6.3) than on CNA (pH 6.8) or on CNA with 10 mM buffer (pH 6.8) and/or due to the suppression of other oral bacteria by selective agents present in CNA. At higher molarities of all the buffers tested (100 and 200 mM), some darkening of the media as a result of nonspecific lysis of erythrocytes was observed. However, hemolytic colonies were still much easier to detect on media containing 100 mM buffer than on either unbuffered media or media with 10 to 50 mM buffer. At 200 mM, TES and PIPES buffers gave similar results; however, at 100 mM, PIPES appeared to be more effective, possibly due to a slightly greater buffering capacity (average terminal pH, 7.5 versus 7.4). Taking into account the higher cost of TES than PIPES and the added expense of use of 200 mM buffer than 100 mM buffer, CNA with 100 mM PIPES at pH 7.5 was adopted as the most generally effective medium for minimization of BLIS-mediated interference in the detection of small numbers of S. pyogenes.

TABLE 2.

Influence of buffering agents on suppression of S. salivarius BLIS activity against S. pyogenes

Medium and additiona	Avg pH after growth	Degree of S. pyogenes positivity in saliva with or without S. salivariusb
		BLIS-positive S. salivarius	BLIS-negative S. salivarius
SBA	6.3	0	2+
100 mM PIPES	7.5	3+	3+
CNA	6.8	1+	3+
200 mM PIPES	7.5	3+	3+
100 mM PIPES	7.5	3+	3+
50 mM PIPES	7.3	2+	3+
10 mM PIPES	6.8	1+	3+
200 mM TES	7.5	3+	3+
100 mM TES	7.4	2+	3+
50 mM TES	7.3	2+	3+
10 mM TES	6.8	1+	3+
200 mM PO4	No growth	No growth	No growth
100 mM PO4	NDc	0	0
50 mM PO4	ND	2+	3+
10 mM PO4	ND	1+	3+

^aBuffered media were adjusted to pH 7.5 prior to autoclaving. Unbuffered SBA and CNA had an average pH of 7.3 after autoclaving. 

^bFresh whole saliva from subjects known to have BLIS-positive or BLIS-negative S. salivarius was seeded with S. pyogenes, and the sample was plated on each test medium. After overnight incubation, the degree of positivity was graded from 0 to 4+. 

^cND, not determined. 

Next, a series of CNA media with 100 mM PIPES buffer (pH adjusted in 0.5-unit increments between pH 6.5 and pH 8.0) were tested for their effectiveness in interfering with different types of S. salivarius BLIS activity and for the impact of this on S. pyogenes detection (i.e., degree of positivity for S. pyogenes). This was done by directly evaluating BLIS production by pure cultures on each medium and recording the degree of positivity of small numbers of S. pyogenes when these were present in cocultures with each BLIS producer. Since some S. salivarius isolates are known to produce BLIS activity only in a simultaneous antagonism test (J. R. Tagg, unpublished data), whereas others are inhibitory in both simultaneous and deferred antagonism tests (Table 1), both types of BLIS activities were tested for on each medium (Table 3). S. pyogenes M1 (final concentration, 10⁶ CFU/ml) was inoculated into six aliquots of BLIS-negative saliva. Each aliquot was coinoculated with either one of four prototype BLIS-producing S. salivarius strains (10⁷ CFU/ml), with a BLIS-negative S. salivarius control culture, or with saliva to which S. salivarius was not added. A cotton swab dipped into the sample was used to deliver the primary inoculum, which was then streaked into four quadrants on each test medium. As a further control each strain combination was also inoculated into saline prior to plating on the test media. This was done to demonstrate the degree of positivity for S. pyogenes when non-BLIS-related inhibitory effects from salivary components are removed. A 4+ degree of S. pyogenes positivity was detected from saline suspensions containing the BLIS-negative S. salivarius control strain and from S. pyogenes suspensions to which S. salivarius was not added. The reduction in the degree of S. pyogenes positivity from 4+ in the saline suspension to 3+ in the saliva suspension represents non-BLIS-related activity. In the saliva system, the S. pyogenes culture to which S. salivarius was not added and the S. pyogenes culture coinoculated with K33 (BLIS-negative S. salivarius) showed a 3+ degree of positivity on all media (Table 3). This 3+ degree of positivity represents the greatest sensitivity of detection of S. pyogenes which could be achieved in the saliva system if all BLIS-related interfering activity is eliminated by the new medium. The results from the saliva system experiments are shown in Table 3.

TABLE 3.

Comparison of effectiveness of CNA buffered at a range of pH values in suppressing different types of anti-S. pyogenes BLIS activity

Test medium	Degree of S. pyogenes positivity in BLIS-negative saliva coseeded with prototype BLIS-positive S. salivarius strainsa
	Min5	20P3	N	F	K-33 (BLIS negative)	None
SBA	0	1+	0	0	3+	3+
CNA	0	1+	1+	0	3+	3+
SBA-PIPES (pH 7.5)	2+	3+	1+	2+	3+	3+
CNA-PIPES (pH 6.5)	1+	1+	1+	1+	3+	3+
CNA-PIPES (pH 7.0)	1+	3+	2+	3+	3+	3+
CNA-PIPES (pH 7.5)	2+	3+	3+	3+	3+	3+
CNA-PIPES (pH 8.0)	1+	2+	3+	3+	3+	3+

^aEach BLIS-producing S. salivarius strain or negative control was added to BLIS-negative saliva together with S. pyogenes, and the sample was then plated on each medium. After overnight incubation, the degree of positivity of S. pyogenes was graded from 0 to 4+. 

Interfering activity by S. salivarius against S. pyogenes was observed when samples containing each of the four BLIS-positive S. salivarius strains were plated onto SBA or CNA without added buffering agents. No S. pyogenes was detected in cocultures with BLIS producers Min5, N, or F on unbuffered SBA, and only 1+ S. pyogenes growth was detected in cocultures containing the BLIS producer 20P3. On unbuffered CNA, strains Min5 and F completely blocked detection of S. pyogenes and strains 20P3 and N reduced the degree of positivity to 1+. Of note is the fact that on CNA buffered at pH 6.5, some inhibition of BLIS activity by all four BLIS-producing S. salivarius strains is evident, yet the BLIS activity by strains Min5 and F is not blocked on unbuffered CNA, which has a higher terminal pH (pH 6.8) (Table 2) than CNA buffered at pH 6.5. This suggests that 100 mM PIPES has an effect on BLIS production that is independent of its pH effect. However, as the pH of CNA was increased to 7.5, the sensitivity of detection of S. pyogenes increased to the sensitivity observed in the presence of the BLIS-negative strain (3+) other than in cocultures with strain Min5, which persistently had a somewhat reduced degree of positivity (2+). At pH 8.0, the BLIS activities of strains Min5 and 20P3 were only partially blocked, whereas the ability to detect S. pyogenes in the presence of strain F or N was comparable to that for the BLIS-negative culture. CNA buffered at pH 8.0 was markedly darker in color, and the presence of β-hemolysis was more difficult to interpret than that on CNA at pH 7.5.

On the basis of the results described above, CNA with 100 mM PIPES at pH 7.5 (CNA-P) was selected as the best overall medium for elimination of S. salivarius BLIS-mediated interference with the growth of S. pyogenes. Figure 1 shows a saliva specimen seeded with S. salivarius 20P3 and S. pyogenes M1 plated on SBA (Fig. 1A) or CNA-P (Fig. 1B). BLIS production by strain 20P3 completely inhibits β-hemolysis by S. pyogenes in the primary inoculation zone on SBA, and less than 10 β-hemolytic colonies are visible in the secondary zone. On CNA-P, the BLIS activity of strain 20P3 is inhibited and β-hemolysis can be detected, with numerous β-hemolytic colonies detected within the primary inoculation zone.

FIG. 1.

BLIS production by S. salivarius 20P3 completely inhibits β-hemolysis by S. pyogenes in the primary zone on SBA (A). Less than 10 β-hemolytic colonies are visible in the secondary zone. On CNA-P (B) strain 20P3 BLIS activity is inhibited and β-hemolysis by S. pyogenes is evident in the primary and secondary zones.

CNA-P was also tested for its effectiveness in suppressing the BLIS activity produced by Streptococcus sanguis K11 and Streptococcus mitis J. Surprisingly, we found that S. sanguis K11 acted essentially like a BLIS-negative strain on both SBA and CNA-P in seeding experiments, even though it exhibits strong anti-S. pyogenes activity in a deferred antagonism test (33). This result suggests that the strain K11 BLIS activity may be produced relatively late in its growth cycle. The BLIS activity produced by S. mitis J acted in a fashion similar to that in which the S. salivarius BLIS activities acted (data not shown). Next, CNA-P was compared with SBA for their abilities to detect β-hemolytic streptococci in clinical specimens.

Application of CNA-P and SBA to the detection of β-hemolytic streptococci in clinical samples.

Samples were collected from 376 children. Of a total of 1,352 paired throat and tongue swab specimens, 1,115 were from children asymptomatic for pharyngitis and 237 were from symptomatic children. The throat swab specimens were plated on CNA-P and SBA, and tongue swab specimens were plated on mitis-salivarius agar. By latex agglutination testing of representative colonies, it was demonstrated that 314 of the paired throat swab specimen cultures demonstrating β-hemolysis contained Lancefield group A streptococci, 33 contained group B streptococci, 5 contained group C streptococci, and 61 contained group G streptococci. The composite streptococcal BLIS activity present in the sample at the time of throat swab specimen culture was determined by deferred antagonism testing of the growth from agar. Thirty-three percent (102 of 314) of the samples were BLIS positive (see Table 5).

TABLE 5.

Comparison of S. pyogenes isolation or nonisolation on CNA-P and SBA and average degree of positivity on each medium

Group	Isolation ratea on:			P value	Avg degree of S. pyogenes positivityb		P value
	Both media	CNA-P only	SBA only		CNA-P	SBA
BLIS-positive strainsc	78 (80/102)	19 (19/102)	3 (3/102)	0.001	2.47	2.03	<0.001
BLIS-negative strainsc	80 (170/212)	15 (31/212)	5 (11/212)	0.003	2.35	2.07	<0.001
Asymptomatic childrend	77 (184/240)	18 (44/240)	5 (12/240)	<0.001	2.22	1.87	<0.001
Symptomatic childrend	89 (66/74)	8 (6/74)	3 (2/74)	NSe	2.97	2.70	0.015
Total positive	80 (250/314)	16 (50/314)	4 (14/314)	<0.001	2.40	2.07	<0.001

^aIsolation rate (as a percentage) and the number of samples positive/total number of samples (in parentheses). 

^bAverage degree of positivity of S. pyogenes on each plate (graded from 1+ to 4+). 

^cIsolation of S. pyogenes from samples which had BLIS-positive or BLIS-negative S. salivarius populations by the deferred antagonism test. 

^dIsolation of S. pyogenes from samples taken from children symptomatic for pharyngitis or during routine monthly visits irrespective of BLIS status. 

^eNS, not significant. 

CNA-P demonstrated greater sensitivity than SBA (Table 4). Of the 314 throat swab specimens positive for S. pyogenes, 300 (96%) were positive on CNA-P and 264 (86%) were positive on SBA. The proportion of cultures positive for S. pyogenes was 31% (74 of 237) for symptomatic children and 22% (240 of 1,115) for routine monthly throat swab specimens. The overall prevalence of positive cultures was 23%.

TABLE 4.

Sensitivities of CNA-P and SBA for detection of S. pyogenes

Test medium	No. of specimens			Sensitivity (%)a
	True positive	False negative	True negative
CNA-P	300b	14	1,038	96
SBA	264b	50	1,038	84
Both	314

^aPercentages were based on a total of 314 specimens positive for S. pyogenes as determined by isolation on either medium. 

^bP < 0.001. 

The McNemar chi-square test was used to compare isolation or nonisolation of S. pyogenes on either CNA-P or SBA and on both CNA-P and SBA. The degree of positivity was not considered in this evaluation. The comparisons were made between samples which had BLIS-positive or BLIS-negative deferred antagonism test results, between samples from symptomatic or asymptomatic children, and for the combined total of all positive samples (Table 5). S. pyogenes was isolated on CNA-P but not on SBA from a statistically significant number of samples from all groups, with the exception of the symptomatic group. For the symptomatic group there were insufficient S. pyogenes isolations only on CNA-P (six specimens) or only on SBA (two specimens) to determine whether the increased rate of isolation on CNA-P was significant. Interestingly, the improved isolation of S. pyogenes on CNA-P from samples with BLIS-positive S. salivarius was almost matched by the increased rate of isolation of S. pyogenes on this medium from samples considered to be negative for BLIS-producing S. salivarius by the deferred antagonism test. The improved recovery of S. pyogenes on CNA-P from the presumptive BLIS-negative samples may be due at least in part to the fact that some of these specimens contained S. salivarius strains that produced BLIS activity similar to that produced by strain F (Table 1), which is detected only in a simultaneous antagonism test. The simultaneous antagonism test was not routinely performed with all specimens, as it is labor- and time-intensive. However, the results for four children whose specimens had a significantly better degree of positivity for S. pyogenes on CNA-P on three separate occasions were compared with the results for four children whose specimens had a similar degree of positivity on both media on three occasions. None of these children was colonized with BLIS-positive S. salivarius on the basis of testing by the deferred antagonism method. Twenty individual S. salivarius isolates from each child were tested for simultaneous antagonism activity against S. pyogenes. None of the samples from children with similar degrees of positivity on both media had S. salivarius isolates displaying simultaneous antagonism activity, whereas the S. salivarius isolates from two of the four children giving better S. pyogenes detection on CNA-P demonstrated strong simultaneous antagonism activity (9 of 20 colonies and 5 of 20 colonies, respectively) and isolates from the other two children weakly inhibited the growth of the S. pyogenes indicator lawn (8 of 20 colonies and 4 of 20 colonies, respectively).

All 14 of the S. pyogenes isolates detected only on SBA, and 49 of 50 isolates detected only on CNA-P, produced β-hemolysis on both media when they were grown as pure cultures. The single isolate (isolate 97MH30) that produced hemolysis on CNA-P but not on SBA was investigated further. The unusual hemolytic activity of this isolate led us to suspect that it might be deficient in the major S. pyogenes hemolysin, streptolysin S (SLS). If so, then perhaps one added benefit of CNA-P relates to the sensitization of erythrocytes to subsequent hemolysis by low levels of streptolysin O (SLO), allowing detection of SLS-negative S. pyogenes. Isolate 97MH30 was compared to S. pyogenes C 203U (SLS negative, SLO positive) and Blackmore (SLS positive, SLO negative) (2) for the production of β-hemolysis on SBA and CNA buffered with either PIPES or TES over a pH range of 6.5 to 8.0. Neither 97MH30 nor C 203U was hemolytic on SBA. However, both strains were hemolytic on SBA with PIPES buffer and on CNA with either PIPES or TES buffer at all pH values tested. Both strains displayed slightly stronger hemolysis on the CNA-based media than on the corresponding SBA media. In contrast, strain Blackmore was strongly hemolytic on all media tested.

Isolation frequencies for group B, C, and G streptococci were compared on CNA-P and SBA. Due to the relatively small numbers detected, it was not possible to determine whether their isolation was significantly better on CNA-P than on SBA. However, of the 33 group B streptococci, 9 were found only on CNA-P and 2 were found only on SBA. Of the five specimens from which group C streptococci were recovered, three specimens were positive on both media and two specimens were positive on CNA-P alone. Of the 61 group G streptococci detected, 3 were detected on CNA-P alone and 4 were detected on SBA alone.

The overall degree of positivity for each medium was determined by taking the average of the degree of positivity (1+ to 4+) for all samples positive for S. pyogenes. The Wilcoxon matched-pairs signed-ranks test was used to determine whether the higher average degree of positivity observed on CNA-P than on SBA was statistically significant. When all S. pyogenes-positive samples were compared, a statistically significant higher average degree of positivity was observed on CNA-P than on SBA (2.40 versus 2.07; Table 5). A comparison of samples which had BLIS-positive or BLIS-negative activity associated with a tongue swab specimen taken at the time that the throat swab specimen was taken also demonstrated that similar percentages of both groups had significantly higher degrees of positivity for S. pyogenes on CNA-P than on SBA. Similarly, a comparison of samples from children symptomatic or asymptomatic for pharyngitis revealed that the degree of positivity on CNA-P compared to that on SBA was higher for symptomatic children (2.22) than for asymptomatic children (1.87), probably reflecting the increased numbers of S. pyogenes present in an active infection.

DISCUSSION

Bacterial interference, the use of one bacterium to interfere with the growth of others, predates the use of antibiotics as a therapeutic strategy in the prevention of disease (11). A few researchers have continued to study bacterial interference as a method of preventing rather than treating disease. Interest in bacterial interference is increasing as a consequence of a greater appreciation for the importance of the indigenous microbiota in preventing abnormal colonization by pathogens and increasing recognition that substantial disturbances of the normal microbiota occur as a result of antibiotic administration. Roos et al. (28) demonstrated with a small number of patients that recurrence of S. pyogenes infection was significantly reduced in patients when interfering “α-streptococci” were implanted in their throats following antibiotic therapy than in patients who received a placebo. They propose that disturbances in the normal oral microbiota and/or the absence of α-hemolytic streptococci inhibitory to S. pyogenes is a significant factor in the recurrence of streptococcal pharyngitis. Fujimori et al. (12) found that the levels of inhibitory α-streptococci were significantly lower in children than in adults and that these levels were lowest in pediatric patients with multiple S. pyogenes infections. They found that the inhibitory species with moderate activity against S. pyogenes were predominantly S. sanguis (60%) and S. mitis (20%), whereas the isolates with strong anti-S. pyogenes inhibitory activity were S. salivarius. In a previous longitudinal study of schoolchildren from Dunedin, New Zealand, we found that the presence of a particular S. salivarius BLIS type appeared to correlate with a reduced incidence of streptococcal pharyngitis and that this BLIS type was absent from children who frequently acquired S. pyogenes infections (34). Our interest is in investigating the possible role of BLIS-producing oral organisms, in particular, S. salivarius, in preventing acquisition or carriage of β-hemolytic streptococci. As a prerequisite to this study, a blood agar medium which reduces or abolishes BLIS activity against β-hemolytic streptococci was needed in order to ensure that small numbers of β-hemolytic streptococci are accurately identified. This would allow more precise determination of carriage rates and familial distributions and would provide a more reliable indicator of the success of antibiotic treatment in eliminating the infecting strain.

A number of selective media for the detection of β-hemolytic streptococci have been developed and compared to SBA (for a review, see reference 23). We sought to use a known medium with a selective agent which was not inhibitory to streptococcal species in conjunction with pH control to reduce or eliminate the inhibitory activity of a BLIS. We first compared five selective media without pH adjustment for their abilities to inhibit normal oral flora. Crystal violet formulations were selected because of their activities against staphylococci, CNA was selected because of its inhibition of gram-negative organisms, CO agar was selected because of its activity against staphylococci, gram-negative, and coryneform bacteria, and spectinomycin was selected because of its inhibition of Neisseria. Media containing sulfamethoxazole and trimethoprim were excluded from this study because they are known to interfere with the growth of group C, F, and G β-hemolytic streptococci. Gunn et al. (16) found that the antibiotics in sulfamethoxazole-trimethorpim-containing media suppressed 90% of non-group A and B streptococci as well as α-hemolytic streptococci. Our interest was in the development of a medium that would abolish bacterial interference but that would not preclude detection of group B, C, and G streptococci.

CNA was chosen as the best overall selective medium for use as a basis for specific modification. TES and PIPES buffers were tested in CNA for their effectiveness in suppressing the inhibitory activity from BLIS-producing S. salivarius because they have pK_a values within the desired pH range (pH 7.4 and 6.8, respectively). Sodium phosphate buffer was evaluated as a less expensive alternative to TES and PIPES, but it was found to be less effective in suppressing BLIS activity. In general agreement with our results, tests of P-type 777 BLIS production by Hynes (17) with a number of biological buffers (morpholineethanesulfonic acid [MES], morpholinepropanesulfonic acid [MOPS], PIPES, HEPES, TES, and TRIS) demonstrated that buffered media at pH values above 7.3 inhibited production of BLIS activity. However, at pH values above 7.8 growth of the BLIS producer strain was impaired. Hynes found that with the use of 100 mM buffers the terminal pH could be maintained within 0.1 pH unit of the starting value for all buffers with the exception of Tris buffer, which showed a drop from pH 7.0 to 6.8. Hynes also noted that TRIS and PIPES had an influence on type 777 BLIS production that was independent of the pH effect. He found that PIPES and TRIS blocked BLIS production at pH values that supported BLIS production when MES, MOPS, HEPES, or TES buffer was present. We have also observed that trypan blue dye blocks some BLIS activity. This is thought to be due to interactions between the anionic dye and cationic BLIS molecules (J. R. Tagg, unpublished data). Cayley et al. (6) reported that in the absence of osmoprotectants, MOPS (a PIPES analog) accumulated in the cytoplasm of osmotically stressed Escherichia coli (but not in Salmonella enterica serovar Typhimurium) cells through the activity of an as yet unidentified active transport system and functioned as an anionic osmolyte. Similar observations have not been reported for gram-positive bacteria. The role of PIPES in BLIS suppression other than its function at maintaining the culture pH is as yet unknown and warrants further examination.

We found that media with 100 mM PIPES buffer at pH 7.5 inhibited the widest range of S. salivarius BLIS activity. S. salivarius Min5 and 20P3 are known producers of salivaricin A (29), a bacteriocin of the lantibiotic class of which nisin is the prototype. Nisin is soluble at pH 2 but forms oligomers and is inactivated at alkaline pH (24). The production of the lantibiotic streptococcin A-FF22 is also pH dependent, with optimum production at pH 6.0 to 6.5 and with no detectable production at pH 7.5 or pH 8.0 (37). Strain Min5 also produces a second bacteriocin thought to be a lantibiotic (J. R. Tagg, unpublished data). Strain N produces an incompletely characterized large (approximate molecular mass, 20 to 25 kDa) heat-labile molecule with strong activity against S. pyogenes (J. R. Tagg, unpublished data). The most unusual of the BLISs that we tested is that produced by strain F. Very little is known about the strain F BLIS. It is produced early in the growth cycle and exerts its inhibitory activity only in a simultaneous antagonism test. It has not shown any inhibitory activity by deferred antagonism testing. The activities of all four types of S. salivarius BLISs were suppressed on CNA-P. In addition, the activity of the BLIS produced by S. mitis J was also suppressed. Lantibiotic production by gram-positive species would, in general, be expected to be similarly affected at alkaline pH.

An added benefit of CNA-P is the production of β-hemolysis on this medium by SLS-negative strains of S. pyogenes which are otherwise nonhemolytic on SBA. S. pyogenes produces two hemolysins. SLO is a member of a group of thiol-activated bacterial cytolytic protein toxins. It is reversibly inactivated by oxygen and its activity can be restored by SH compounds and other reducing agents. SLS is an oxygen-stable potent membrane-active toxin. It produces the majority of the hemolysis observed on blood agar medium (1). In most cases, streptococci that are recovered from throat swab specimens and that do not display β-hemolysis when plated on SBA would be regarded as commensal streptococci of the nonhemolytic or viridans group rather than the causative agent of infection. However, nonhemolytic SLS-negative strains have been isolated from human infections, and in one well-documented outbreak of nonhemolytic S. pyogenes infection a number of associated cases of rheumatic fever were reported (20). Previously, Tagg and Vugler (38) developed a staphylococcal beta-lysin-containing medium that enhanced the hemolytic activity of SLS-deficient S. pyogenes. The enhanced hemolysis was due to the interaction of SLO with beta-lysin-sensitized erythrocytes in the medium. It appears that when either PIPES or TES buffer is added at a 100 mM concentration to CNA or SBA, either buffer may act in a manner similar to that of beta-lysin. It may be the increased ionic strength of the medium that sensitizes erythrocytes to SLO hemolysis. This sensitization of erythrocytes correlates with our observation that CNA-P medium at pH values above 7.5 is quite susceptible to nonspecific hemolysis and that preincubation of CNA-P plates at 37°C (to aid with drying) and subsequent storage at 4°C or failure to use freshly collected sheep blood can also result in some nonspecific lysis. Buffering of CNA-P at pH 7.5 may also increase the availability of SLO by inhibiting activation of the streptococcal proteases which are released into the medium when the pH falls below 6.7 (1). The frequency of occurrence of β-hemolytic variant streptococci and their clinical significance need to be addressed, and CNA-P affords an appropriate medium that can aid these studies.

The sensitivity of CNA-P in detecting small numbers of β-hemolytic streptococci was demonstrated by the higher degree of positivity of β-hemolytic streptococci for samples from asymptomatic subjects, which would, in general, have fewer β-hemolytic streptococci than samples from symptomatic subjects. Plating on CNA-P may also enhance recovery of small numbers of β-hemolytic streptococci from symptomatic subjects when samples are obtained during the early or late stages of infection or when delays or deficiencies in specimen transport and processing occur.

CNA-P has been shown to be a sensitive medium for the detection of β-hemolytic streptococci in a clinical setting. Its unique feature is its suppression of the S. salivarius BLIS-mediated inhibition of β-hemolytic streptococci in some cultures. In addition, CNA-P appears to enhance the hemolytic activity of SLO, allowing detection of SLS-deficient S. pyogenes strains that might otherwise go undetected in cultures on SBA or selective media.