Rapid Identification of Penicillin and Macrolide Resistance Genes and Simultaneous Quantification of Streptococcus pneumoniae in Purulent Sputum Samples by Use of a Novel Real-Time Multiplex PCR Assay

Abstract

We evaluated a real-time quantitative PCR combined with a multiplex PCR assay for the quantification of Streptococcus pneumoniae and the simultaneous detection of drug-resistant genes by gel-based PCR, using purulent sputum samples. This assay correctly quantified S. pneumoniae and identified their penicillin and erythromycin susceptibilities directly from samples within 3 h.

Streptococcus pneumoniae is a crucial pathogen that causes community-acquired pneumonia (CAP). CAP in adults is often treated with a combination of β-lactam antibiotics and macrolides (18, 22). However, alarmingly high frequencies of penicillin- and macrolide-resistant pneumococci have been reported, especially in several Asian countries, including Japan (4, 24, 30). The resistance of S. pneumoniae to penicillin has been shown to be closely associated with mosaic mutations in the pbp1a, pbp2b, and pbp2x genes (12, 35). Macrolide resistance is generally mediated by two mechanisms: 23S rRNA methylation encoded by the erm(B) gene or macrolide efflux via the mef(A) gene (23, 32). Detection of drug-resistant S. pneumoniae by classical techniques usually takes several days (17, 27). Recently, the utility of the real-time PCR method in the detection and quantification of S. pneumoniae has been examined using lower respiratory tract samples (2, 5, 15, 20, 36). The results of these studies suggest that clinical infection correlates with increased pneumococcal load, and these studies also mentioned the capacity of quantitative real-time PCR to distinguish between colonization and true infection (15, 36). Many investigators have evaluated the accuracy of multiplex PCR methods, which are used in the screening of S. pneumoniae strains demonstrating penicillin and macrolide resistance (13, 14, 21, 33, 34). Most of these assays, however, require separate tubes, are only able to detect two or three gene fragments, and have not been evaluated for clinical respiratory tract samples. In the present study, we developed a simultaneous single-tube real-time quantitative PCR combined with a multiplex PCR (RQ-mPCR) assay that rapidly quantifies S. pneumoniae; identifies alterations in pbp1a, pbp2b, and pbp2x genes; and detects the presence of erm(B) and mef(A) genes. We first verified this method by using clinical S. pneumoniae strains and then evaluated the effectiveness of the method using purulent sputum samples.

We used 200 clinical isolates of S. pneumoniae screened by optochin susceptibility (susceptible) and bile solubility (soluble) that were collected from April 2004 to March 2006 by a laboratory at the Nagasaki University Hospital. Strains were propagated on 5% sheep blood agar (Nissui Co., Ltd., Tokyo, Japan) at 37°C with 5% CO₂. A mixture of 24 bacterial species from the American Type Culture Collection (ATCC; see Table S1 in the supplemental material) were selected from species commonly isolated from respiratory tract and from species that are genetically related to S. pneumoniae (3). In addition, 17 clinical strains of Streptococcus mitis and 12 clinical strains of Streptococcus oralis, which had been isolated from respiratory tract samples, were collected for cross-reactivity studies.

The MICs were determined by using broth microdilution techniques as described by the Clinical and Laboratory Standards Institute (CLSI) guidelines (7, 8). S. pneumoniae ATCC 49619 was used for quality control.

A total of 200 purulent sputum samples, which were collected from April 2004 to March 2005 and from June 2007 to August 2007 by a laboratory at the Nagasaki University Hospital, were used. Only good-quality sputum samples (P2 and P3 according to the classification of Miller and Jones (19) were used. Sputum samples were diluted 1:100 and 1:10,000 with 0.45% sodium chloride and treated with Sputazyme solution (Kyokuto Pharmaceutical Industries Co., Ltd., Tokyo, Japan). The diluted samples were spread on 5% sheep blood agar plates with a DS500 spiral plater (InterScience, Inc., Markham, Ontario, Canada) and then incubated at 37°C in 5% CO₂. Optochin sensitivity and bile solubility were used to identify S. pneumoniae. Nucleic acids were isolated from clinical strains and sputum samples by using a QIAamp DNA blood minikit (Qiagen, Hilden, Germany).

The lytA, pbp1a, pbp2b, pbp2x, ermB, and mefA genes were amplified by PCR. Primers LytA-F and LytA-R and probes LytA-DCR and LytA-ACR were designed to target a 173-bp fragment of the single copy autolysin (lytA) gene of S. pneumoniae and were gleaned from published sequence (26). The primers for amplification of the pbp1a, pbp2b, and pbp2x genes were newly designed as follows: pbp1a (353 bp), 5′-₁₇₀₉AGTATATCAAGAACACTGGCTACG₁₇₃₂ and 5′-₂₀₆₁GCTTGGAGTGGTTGAGCTA₂₀₇₉-3′; pbp2b (442 bp), 5′-₁₂₉₁AAATTGGCATATGGATCTTTTC₁₃₁₂-3′ and 5′-₁₇₃₂TATTCATCTCTGTCGGTTGC₁₇₅₁-3′; and pbp2x (339 bp), 5′-₉₉₀AAGTAACTATGAACCAGGATCAG₁₀₁₂-3′ and 5′-₁₃₈₈CGAAGCATTTGTGTTTGTGT₁₄₀₇-3′. The resistance pbp1a primers were designed to target four amino acid substitutions (Thr-574→Asn, Ser-575→Thr, Gln-576→Gly, and Phe-577→Thr) that are common to all penicillin G (PCG)-intermediate and -resistant isolates (29). The resistance pbp2x primers were designed to target amino acid substitutions in the 337STMK motif, and the resistance pbp2b primers were designed to target amino acid substitutions close to the 448SSN motif (25, 28). The primers for amplification of the ermB (224-bp) and mefA (294-bp) genes were gleaned from a published sequence (21). All of the primers used for RQ-mPCR had almost identical annealing temperatures (range, from 59.0 to 62.5°C), which reduces the occurrence of unwanted bands originating from nonspecific amplification. The PCR product amplified from S. pneumoniae ATCC 49619 using the LytA-F and LytA-R primer set was ligated into the pTAC-1 plasmid vector (BioDynamics, Tokyo, Japan) by using the TA PCR cloning technique. Plasmid standards containing 2.9 × 10⁶ to 2.9 × 10⁰ copies/μl were prepared by diluting the plasmid extracts in water. The standard curve was generated and exported by using LightCycler software (v3.5). PCR was performed on a LightCycler instrument. The final 20-μl single-tube reaction mixture contained 2× LightCycler FastStart DNA Master HybProbe (Roche Diagnostics, Basal, Switzerland), 5 mM MgCl₂, 0.5 μM concentrations of each primer (LytA-F, LytA-R, pbp1a-F, pbp1a-R, pbp2b-F, pbp2b-R, pbp2x-F, pbp2x-R, mef-F, mef-R, erm-F, and erm-R), 0.2 μM concentrations of each hybridization probe (LytA-DCR and LytA-ACR), and 2 μl of DNA template. The cycling conditions were as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 5 s, 60°C for 10 s, and 72°C for 15 s. All runs included a negative control of water and a calibrator/positive control of 2.9 × 10⁶ copies of the plasmid/μl used for the standard curve. The data were analyzed by using LightCycler Software (v3.5) in the F2/F1 mode with a fit point calculation method. After amplification, 10 μl of the PCR product was separated by electrophoresis on a 3% agarose gel (Cambrex BioScience/Rockland, Inc., Rockland, ME) for 30 min at 100 V. The positions of DNA fragments are shown in Fig. 1A, lane M (100-bp ladder; Amersham Biosciences). Figure 1B shows the presence of the amplified products after agarose gel electrophoresis when DNA was extracted from representative PCG- and erythromycin (EM)-resistant S. pneumoniae strains (lane 1, ATCC 49619 (MICs of PCG = 0.25 μg/ml and EM < 0.5 μg/ml); lane 2, clinical strain 4808/S (MICs of PCG < 0.015 μg/ml and EM < 0.5 μg/ml); lane 3, clinical strain 1512/F (MICs of PCG = 0.03 μg/ml and EM = 2 μg/ml); lane 4, clinical strain 1565/F (MICs of PCG = 0.03 μg/ml and EM = 32 μg/ml); lane 5, clinical strain 8315/F (MICs of PCG = 0.12 μg/ml and EM = 0.5 μg/ml); lane 6, clinical strain 4827/F (MIC of PCG = 0.12 μg/ml and EM = 32 μg/ml); lane 7, clinical strain 8605/Z (MICs of PCG = 8 μg/ml and EM < 0.5 μg/ml); lane 8, clinical strain 8729/Z (MICs of PCG = 4 μg/ml and EM = 8 μg/ml); lane 9, clinical strain 1824/F (MICs of PCG = 2 μg/ml and EM = 16 μg/ml); lane M, 100-bp ladder).

FIG. 1.

New multiplex PCR assay for simultaneous detection of lytA, penicillin resistance genes (altered pbp1a, pbp2x, and pbp2b), and macrolide resistance genes [erm(B) and mef(A)]. (A) Comparison of single-target versus multiplex PCRs using “resistant” control strain 1824/F. Lane M, 100-bp ladder (Amersham Biosciences). (B) Agarose gel electrophoresis of PCR products amplified with RQ-mPCR using control strains. Lane 1, ATCC 49619 (MICs of 0.25 μg/ml for PCG and <0.5 μg/ml for EM); lane 2, clinical strain 4808/S (MICs of <0.015 μg/ml for PCG and <0.5 μg/ml for EM); lane 3, clinical strain 1512/F (MICs of 0.03 μg/ml for PCG and 2 μg/ml for EM); lane 4, clinical strain 1565/F (MICs of 0.03 μg/ml for PCG and 32 μg/ml for EM); lane 5, clinical strain 8315/F (MICs of 0.12 μg/ml for PCG and 0.5 μg/ml for EM); lane 6, clinical strain 4827/F (MICs of 0.12 μg/ml for PCG and 32 μg/ml for EM); lane 7, clinical strain 8605/Z (MICs of 8 μg/ml for PCG and <0.5 μg/ml for EM); lane 8, clinical strain 8729/Z (MICs of 4 μg/ml for PCG and 8 μg/ml for EM); lane 9, clinical strain 1824/F (MICs of 2 μg/ml for PCG and 16 μg/ml for EM); lane M, 100-bp ladder (Amersham Biosciences).

Six, S. pneumoniae ATCC strains were all positive for the lytA gene. None of the DNA extracts (≥10 ng/μl) from 47 nonpneumococcal organisms (including 17 clinical isolates of S. mitis and 12 clinical isolates of S. oralis) cross-react with the primer-probe set, showing that the lytA primer-probe set was 100% specific for detecting S. pneumoniae strains. We used S. pneumoniae ATCC 49619 (positive for lytA and pbp1a), S. pneumoniae clinical strain 4808/S (positive for lytA only) and S. pneumoniae clinical strain 1824/F [positive for lytA, erm(B), mef(A), pbp1a, pbp2x, and pbp2b] to examine the analytical sensitivity of S. pneumoniae quantification by RQ-mPCR (see Fig. S1A in the supplemental material). The detection limit of lytA quantification was 20 copies/assay (10 copies/μl), which corresponds to 5 × 10² CFU/ml. To verify the analytical sensitivities of drug resistance genes in RQ-mPCR, we used S. pneumoniae clinical strain 1824/F and assessed the appearance of PCR products by gel electrophoresis (see Fig. S1B in the supplemental material). The detection limits of the five drug-resistant genes were between 5.8 × 10¹ and 5.8 × 10² copies/assay (between 10³ and 10⁴ CFU/ml). To validate the RQ-mPCR technique, all of the 200 S. pneumoniae strains that were tested by RQ-mPCR were also screened for the presence of individual resistance genes by a single PCR using the PCR conditions described above. The results of the two methods were in full agreement (data not shown), suggesting that the multiplex PCR primer sets are reliable. The RQ-mPCR results of 200 S. pneumoniae strains and the MIC distribution of PCG and EM are shown in Tables 1 and 2. All of the 200 strains were positive for the lytA gene. The multiplex PCR correctly identified penicillin susceptibility (MIC ≤ 0.06 μg/ml) or nonsusceptibility (MIC ≥ 0.12 μg/ml) in 189 (94.5%) of 200 isolates evaluated. The sensitivity, specificity, positive predictive values, and negative predictive values of our assay were 98.1% (103/105), 90.5% (86/95), 91.9% (103/112), and 97.7% (86/88), respectively. Our assay also correctly identified EM susceptibility (MIC ≤ 0.5 μg/ml) or resistant (MIC ≥ 1 μg/ml) in 200 (100%) out of 200 isolates evaluated. All of these isolates yielded 100% sensitivity, 100% specificity, 100% positive predictive values, and 100% negative predictive values.

TABLE 1.

RQ-mPCR results and PCG MICs in 200 pneumococcal isolates^a

RQ-mPCR result	PCG MIC distribution (no. of isolates)
	≤0.015 μg/ml	0.03 μg/ml	0.06 μg/ml	0.12 μg/ml	0.25 μg/ml	0.5 μg/ml	1 μg/ml	2 μg/ml	4 μg/ml	8 μg/ml
None	50	26	10	2
pbp2x only		1	6	9	4
pbp1a only			2	4	7	4
pbp1a + pbp2x				4	1	3	2
pbp1a + pbp2x + pbp2b					4	6	16	20	18	1

^aThe CLSI PCG MIC breakpoints for S. pneumoniae are as follows: susceptible, ≤0.06 μg/ml; intermediate, 0.12 to 1 μg/ml; and resistant, ≥2 μg/ml (7).

TABLE 2.

RQ-mPCR results and EM MICs in 200 pneumococcal isolates^a

RQ-mPCR result	EM MIC distribution (no. of isolates)
	<0.5 μg/ml	1 μg/ml	2 μg/ml	4 μg/ml	8 μg/ml	16 μg/ml	≥32 μg/ml
None	56
mef(A) only		4	15	17	9	1
erm(B) only				3	7	10	67
mef(A) + erm(B)					1	2	8

^aThe CLSI EM MIC breakpoints for S. pneumoniae are as follows: susceptible, ≤0.25 μg/ml; intermediate, 0.5 μg/ml; and resistant, ≥1 μg/ml (7).

Of 200 purulent sputum samples, 56 samples were S. pneumoniae positive, and the remaining 144 samples were S. pneumoniae negative as determined by the conventional culture method. All 56 S. pneumoniae-positive samples were also positive for the lytA gene, and 143 of the 144 S. pneumoniae-negative samples were negative for the lytA gene. Therefore, the sensitivity and specificity for the identification of S. pneumoniae using purulent sputum samples in RQ-mPCR compared to the conventional culture method were 100% (56/56) and 99.3% (143/144), respectively. The correlation between the conventional culture counts and the level of lytA gene expression by RQ-mPCR using the 56 pneumococcal culture-positive sputum samples is shown in Fig. S2 in the supplemental material.

The penicillin- and macrolide-resistant genes detected by RQ-mPCR in purulent sputum samples are shown in Table 3. We compared these results to those in isolated S. pneumoniae from the same sputum samples. Of 56 pneumococcal culture-positive sputum samples, all were positive for lytA gene, and for 51 samples the detected genes were in complete agreement with the isolated S. pneumoniae. For the remaining five samples, the genes were not in complete agreement with the isolated S. pneumoniae: two samples were false positives for erm(B); one sample was false positive for mef(A), pbp1a, pbp2x, and pbp2b; one sample was false positive for pbp2x and pbp2b; and one sample was false positive for mef(A) and false negative for pbp1a. The sensitivities and specificities of this assay for detecting genes directly from sputum samples relative to isolated S. pneumoniae were 100 and 93.9% for erm(B), 100 and 94.8% for mef(A), 94.4 and 97.5% for pbp1a, 100 and 94.1% for pbp2x, and 100 and 95.6% for pbp2b, respectively. With regard to the 144 pneumococcal culture-negative sputum samples, the detection rates of drug resistant genes were 0% for pbp1a, 2.7% (4/144) for pbp2x, 1.4% (2/144) for pbp2b, 20.8% (30/144) for erm(B), and 11.1% (16/144) for mef(A).

TABLE 3.

Comparison of RQ-mPCR results between in pneumococcal positive sputum samples and in S. pneumoniae isolates

Sample group and gene	No. of samples	Positive sputum samples by RQ-mPCR among samples with disagreement (n = 5)b					No. false positive/no. false negative	% Sensitivity	% Specificity
		102/5F01	107/5F23	113/5F28	101/6S11	101/6D13
Samples with complete agreement	51/56a
Samples with disagreement
lytA	51	+	+	+	+	+	0/0	100	99.3
erm(B)	27	−	−	+	FP	FP	2/0	100	93.9
mef(A)	19	FP	+	FP	FP	FP	2/0	100	94.8
pbp1a	15	FN	−	FP	+	+	1/1	94.4	97.5
pbp2x	24	+	FP	FP	+	+	2/0	100	94.1
pbp2b	12	+	FP	FP	+	+	2/0	100	95.6

^aThat is, the number of samples in complete agreement with RQ-mPCR results in S. pneumoniae isolates/number of all samples.

^bFP, false-positive; FN, false-negative.

Microorganisms closely related to S. mitis and harboring lytA gene, which are classically associated with S. pneumoniae, have been reported previously (37). However, positive results were not obtained from the mitis group of streptococci (including our collected clinical strains of 17 S. mitis and 12 S. oralis) that were tested for cross-reactivity in the present study. Compared to past reports (13, 14, 21, 34, 35), we obtained lower pbp2b detection rates in PCG-intermediate S. pneumoniae strains. This may have been due to a difference in the regions targeted by the pbp2b primers. The PCR results of macrolide-resistant genes in the present study matched previous report using the same primers (21), and they were also consistent with reports using other primers (11, 31). Although the analytical sensitivities of resistance genes on gel electrophoresis were lower than those seen with lytA quantification, the detection limit of 10³ to 10⁴ CFU/ml in resistant strains is within the permissible range for use with clinical samples, including bronchoalveolar fluid, which requires a diagnostic sensitivity for pathogens in excess of 10⁴ CFU/ml (1, 9). With regard to one sputum sample which showed a false positive for the lytA gene, the RQ-mPCR assay (confirmed repeatedly) showed the presence of 10⁴ CFU of S. pneumoniae/ml, and this specimen grew to 2 × 10⁷ CFU of Staphylococcus aureus/ml by the culture method. This discrepancy may have resulted from a failure to detect S. pneumoniae that was surrounded by S. aureus and/or have resulted from the detection of atypical S. mitis or S. oralis, both of which harbor the lytA gene (37).

A discrepancy in RQ-mPCR specificity for samples 102/5F01 [false positive for mef(A)], 107/5F23 (false positive for pbp2x and pbp2b), 113/5F28 [false positive for mef(A), pbp1a, pbp2x, and pbp2b], and 101/6S11 and 101/6D13 [false positive for erm(B)] was confirmed by single PCR. This discrepancy indicates the presence of similar resistance genes in other microorganisms in the sputum samples. Avoiding these problems with cross-reactivity is difficult because the mosaic genes that encode altered, low-affinity pbp genes are considered products of recombination events involving horizontal transfer from closely related species (10, 16) and also because erm(B) and mef(A) macrolide-resistant genes in S. pneumoniae are highly homologous to genes in other Streptococcus-related species (6). In sample 102/5F01, pbp1a was detected by a single PCR, suggesting that the discrepancy in sensitivity (false negative) was due to the presence of PCR inhibitors or to a decline of sensitivity in this sputum sample. Although the sensitivity and specificity rates of our assay for sputum samples were in general satisfactory, further investigations are needed to remove PCR inhibitors from samples and to increase the sensitivity of the assay. Although our RQ-mPCR assay showed high correlation between the conventional culture counts and the level of lytA gene expression, further data from patients with pneumonia are needed to evaluate and interpret the results of our assay.

In summary, the RQ-mPCR method developed here had high sensitivity and specificity for pneumococci and could detect drug resistance in both clinical S. pneumoniae strains and sputum samples. Furthermore, the results can be obtained directly from clinical samples within 3 h (2 h for DNA extraction and preparation of PCR mixture and 1 h for PCR assay and electrophoresis), and this assay requires only a single tube. This method may be helpful for the rapid screening of resistance in pneumococcal isolates and should allow the administration of earlier, more focused and effective treatment of drug-resistant S. pneumoniae.