Development and Clinical Evaluation of an Internally Controlled, Single-Tube Multiplex Real-Time PCR Assay for Detection of Legionella pneumophila and Other Legionella Species

Abstract

A multiplex real-time PCR assay for detection of Legionella pneumophila and Legionella spp. and including an internal control was designed. Legionella species, L. pneumophila, and the internal control were detected simultaneously by probes labeled with 6-carboxy-fluorescein, hexachlorofluorescein, and indodicarbocyanine, respectively. Therefore, no postamplification analysis was required in order to distinguish the targets. The sensitivity of both assays was 2.5 CFU/ml, and from analysis of 10 culture-positive and 74 culture-negative samples from patients investigated for legionellosis, 100% agreement was observed by both assays in comparison to culture. Four additional positives were found by the multiplex real-time PCR assay in the Legionella culture-negative samples.

Legionella pneumophila and some Legionella species have been shown to cause community-, travel-, and nosocomially acquired pneumonia. Infection of humans is usually via inhalation of aerosols from a variety of natural or man-made water systems. Although there are currently 48 species of Legionella (7), only 19 Legionella species are documented as human pathogens on the basis of their isolation from clinical material (18). However, 91.5% of the clinical cases are caused by Legionella pneumophila (27). Potentially, all legionellae may cause human disease. Mortality of patients with pneumonia may exceed 30% for elderly or immunocompromised patients (2). Microbiological diagnosis is warranted, as clinically legionellosis is not distinguishable from other pneumonias. Moreover, rapid diagnosis improves the prognosis for the patient by timely and appropriate treatment (10).

Culture on buffered charcoal yeast extract (BCYE) plates is considered the gold standard for the laboratory diagnosis of Legionella infections. However, legionellae are slow-growing and fastidious bacteria, and successful culture requires selective medium and prolonged incubation periods. In addition, culture is less optimal for the non-L. pneumophila legionellae, as these often require additional supplements. Many laboratories use urine antigen detection tests, which are much more rapid for routine diagnostics, but these assays only detect a limited number of serogroups of L. pneumophila, especially the Binax Now, which detects non-L. pneumophila serogroup 1 (6). These rapid urine antigen tests have sensitivities between 60% and 85% (5, 6, 11, 14). Serological diagnosis is also commonly used, and a sensitivity of 80% has been reported (8). However, a diagnosis by a fourfold immunoglobulin G (IgG) titer increase can only be made retrospectively, and in 25% of patients no IgG titer increase is observed (8). An IgM assay can make a rapid diagnosis, but the drawback is that IgM may persist for 2 years or more (13).

Therefore, nucleic acid amplification techniques are attractive tools for detection of legionellae in clinical samples, as they are able to detect all legionellae and provide rapid results. It is, however, still important to be able to distinguish between L. pneumophila and non-L. pneumophila species, as currently more severe morbidity and mortality has been observed with L. pneumophila. Therefore, clinically the identification of the actual Legionella species is required. Diagnostic PCR assays have principally targeted specific regions within 16S rRNA genes (rDNA) (3, 9, 12, 21, 23, 26), 5S rDNA (9), or the macrophage inhibitor potentiator (mip) gene (1, 9, 15, 16, 17, 20, 22, 26). Real-time PCR has added benefits to routine diagnosis, as it minimizes manual time for the PCR and negates the use of post-PCR analysis. A few real-time protocols have been reported with the Lightcycler (1, 9, 21, 23, 26) but the dual target assays require melt-curve analysis to distinguish L. pneumophila from non-L. pneumophila (21, 23).

Here we describe the development of a sensitive, specific multiplex PCR assay for the simultaneous detection and discrimination of Legionella spp. and Legionella pneumophila with an internal control in clinical specimens.

MATERIALS AND METHODS

Bacteria.

A range of bacterial strains were used to test the specificity of the real-time PCR, including Bordetella pertussis, Bordetella parapertussis, Mycobacterium pneumoniae, Streptococcus pneumoniae, Haemophilus influenzae, Streptococcus pyogenes, Mycobacterium catarrhalis, Streptococcus viridans, S. enterica serovar Enteritidis, S. enterica serovar Typhimurium Enterococcus faecalis, Staphylococcus aureus, Acinetobacter baumannii, Neisseria meningitidis, Pseudomonas aeruginosa, Enterobacter aerogenes, Enterobacter faecium, Klebsiella pneumoniae, Chlamydophila pneumoniae, Chlamydophila psittaci, and Escherichia coli. Each was cultured with appropriate agar, cells, and medium.

Twelve different serogroups of L. pneumophila (serogroups 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, and 14) and 24 non-L. pneumophila species were used establish the specificity and sensitivity of the PCR assays. The 24 non-L. pneumophila Legionella species were L. anisa L. birminghamensis, L. bozemanii group 1, L. brunensis, L. cherrii, L. dumoffii, L. feeleii −1, L. gormanii, L. israelensis, L. jamestowniensis, L. jordanis, L. longbeachae 1, L longbeachae 2, L. maceachernii, L. micadadei, L. moravica, L. oakridgens, L. parisensis, L. rubrilucens, L. sainthelensis, L. santicrusis L. spiritensis, L. steigerwaltii, and L. tucsonensis. The Legionella strains were grown at 37°C on buffered charcoal yeast extract (BCYE) agar for 48 to 72 h. The incubation period for the diagnostic culture was 10 days.

To evaluate the sensitivity of the PCR test, all the Legionella strains (36 in total) were suspended in sterile physiological saline at a concentration equivalent to 10⁸ cells per ml, based on McFarland turbidimetric standards. Serial dilutions were made from the suspension and aliquots were extracted and tested in the PCR to determine sensitivity.

Clinical material.

To evaluate the sensitivity of the PCR test for legionellae on clinical samples, sputum, throat swab, and bronchoalveolar lavage samples were all spiked with L. pneumophila (ATCC 33152) based on McFarland turbidimetric standards. Serial dilutions were made from the suspension and aliquots were extracted and tested in the PCR to determine sensitivity.

Retrospectively, 10 clinical samples which were known to be culture positive for Legionella species were tested. Five were obtained from the Kapellen outbreak (28), four from patients presenting with community-acquired pneumonia, and one from a patient presenting with nosocomial pneumonia. All the samples from the Kapellen outbreak were grouped as L. pneumophila serogroup 1, the other patient samples were L. pneumophila serogroup 1 (three samples) and serogroup 6 (one sample), and the nosocomial patient sample was L. bozemanii. Prospectively 64 bronchoalveolar lavage samples and 10 throat swab samples from hospitalized patients with pneumonia were also tested; all were cultured for Legionella species. A summary of the clinical samples is shown in Table 3.

TABLE 3.

Culture and real-time PCR results for clinical samples^a

Clinical samples	No. of positive samples
	Culture	Real-time PCR	Total
Retrospective
L. pneumophila	9	9	9
Legionella spp.	1	1	1
Prospective
Bronchalvelor lavage
L. pneumophila	0	0	64
Legionella spp.	0	2	64
Throat swabs
L. pneumophila	0	2b	10
Legionella spp.	0	0	10

^aCulture positive samples were tested retrospectively, and 74 samples were prospectively analyzed.

^bShowed recent Legionella infection by serology.

Proficiency panel.

A proficiency panel for the quality control of L. pneumophila PCR was subjected to the real-time PCR as well. This panel had been used in the first national external quality assessment for Belgian Laboratories (see Report, First External Quality Assessment in Belgian Laboratories Performing Molecular Microbiology, organized by the working group of molecular microbiology of the National Committee for Molecular Diagnostics) and consisted of 20 spiked samples, 10 bronchoalveolar lavage samples, and 10 sputum samples. Twelve samples were positive (eight weakly positive and four strongly positive) and the other eight samples were negative for all Legionella species. Six of the positive samples were L. pneumophila serogroup 1, and the other six were L. pneumophila serogroup 6.

DNA extraction.

Nucleic acids were extracted from the suspensions of all bacterial colonies and the clinical samples with the High Pure PCR Template preparation kit (Roche, Mannheim, Germany). Sputum samples were treated with sputolysin reagent prior to DNA extraction. (Calbiochem, Darmstadt, Germany). The sputum was mixed with three parts sputolysin, vortexed, and incubated for 10 min at room temperature. The mixture was centrifuged for 10 min at 1,000 × g. The sediment was resuspended in Tris-EDTA buffer and stored at −80°C prior to extraction. All other nucleic acid extraction from samples was performed without any preceding steps according to the manufacturer's instructions.

Briefly, 200 μl of the samples was lysed by adding 200 μl of binding buffer and proteinase K and incubation at 72°C for 10 min. After adding 100 μl of isopropanol, the solution was passed through a silica column by centrifugation at 6,000 × g for 1 min. The columns were washed three times and centrifuged at 6,000 × g for 1 min. Purified nucleic acids were eluted in 200 μl of elution buffer, which was warmed to 70°C and stored at −20°C. DNA of L. pneumophila serogroup 1 (ATCC 33152) was used as a positive control. The controls included in each run encompassed a no-template control after every 10 samples, a control for the extraction, and any buffers to ensure no contamination with Legionella species. In the negative control, molecular grade water was added instead of specimen.

Primers and probes for real-time PCR.

Primer and molecular beacon sequences were selected from an alignment of nucleotide sequences of the 16s RNA gene of Legionella species from GenBank The aim of this assay was to select primers and probes that would detect all Legionella species but no other organisms. Primer and molecular beacon sequences were selected from the mip gene of L. pneumophila with criteria required for design of molecular beacon assays (accession no. AF095229). The aim of this assay was to amplify only L. pneumophila serogroups and no other Legionella species. The PCR primers were optimized with the primer 3 program (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) to ensure no secondary structures.

The molecular beacon was designed with the Mfold Zuker program (http://www.bioinfo.rpi.edu/applications/mfold/old/dna/). Criteria for a good molecular beacon included a melting temperature of 5 to 10°C over the melting temperature of the primers and limited secondary structure in the target sequence. The stem sequence was selected to have a compatible melting temperature to the molecular beacon. The beacon formed a stable structure at 50°C to 55°C, the proposed annealing temperature, with no secondary structures. For the Legionella species assay, the fluorescent reporter on the 5′ end of the probe was 6-carboxy-fluorescein (FAM), and the quencher on the 3′ end was Dabcyl. For the L. pneumophila assay, the fluorescent reporter on the 5′ end of the molecular beacon was hexachlorofluorescein (HEX) and the quencher on the 3′ end was Dabcyl. The phocine herpesvirus (PhHV) internal control probe was labeled with indodicarbocyanine (Cy-5) and Black Hole Quencher 2. Biolegio prepared the molecular beacons and primers (Biolegio, Malden, The Netherlands). Selected primers and probes are shown in Table 1.

TABLE 1.

Primers and probes for PCR amplification of Legionella spp. (16S rRNA gene) and L. pneumophila (mip gene)

Target	Sequencea	Function	Product size (bp)
16S rRNA	AGGCTAATCTTAAAGCGCC	Forward primer	212
	CCTGGCTCAGATTGAACG	Reverse primer	212
	FAM-CCGAGCGGTGAGTAACGCGTAGGAATATGGCTCGG-Dabcyl	Molecular beacon
mip	TGGTGACTGCAGCTGTTATG	Forward primer	124
	CATTGCTTCCGGATTAACAT	Reverse primer	124
	HEX-GCTGCCGCATTGGTGCCGATTTGGGAGGCAGC-Dabcyl	Molecular beacon

^aThe stem structure of the molecular beacon is underlined. FAM, 6-carboxy-fluorescein; HEX, hexachlorofluorescein; Dabcyl, quencher.

Real-time PCR conditions for individual assays.

Real-time PCR was performed in 50 μl of reaction mixture consisting of 25 μl of HotStart Taq master mix kit (Qiagen, Hilden, Germany), 4.5 mM MgCl₂, 0.3 μM each primer 0.44 μM molecular beacon, and 10 μl of template. The PCR thermal profile consisted of an initial incubation of 15 min at 95°C, followed by 50 cycles of 30 s at 95°C, 30 s at 55°C, and 30 s at 72°C. Amplification, detection, and data analysis were performed with the iCycler IQ real-time detection system (Bio-Rad, Veenendaal, The Netherlands).

Multiplex PCR including internal control.

The multiplex PCR was designed with a PhHV internal control as described by Templeton et al. (24). The real-time PCR for Legionella spp. was performed in a reaction mixture consisting of 25 μl of HotStary Taq master mix kit (Qiagen, Hilden, Germany) 4.5 mM MgCl₂, 0.3 μM each of the four Legionella primers, 0.2 μM phocine herpesvirus primers, 0.44 μM each of the two Legionella molecular beacons, 0.24 μM phocine herpesvirus molecular beacon, and 10 μl of template. The PCR thermal profile was identical to that of the individual assays. Amplification, detection, and data analysis were performed with an iCycler IQ real-time detection system (Bio-Rad, Veenendaal, The Netherlands).

Determination of intra-assay and interassay variation.

DNA was extracted from L. pneumophila serogroup 1 (ATCC 33152) and stored in AE (Tris-EDTA) buffer (Qiagen, Hilden, Germany). The DNA was diluted to a concentration equivalent to 5,000 CFU/ml and stored in small aliquots at −20°C. In order to determine inter- and intra-assay variation, an aliquot was thawed and run in quintuplicate in five consecutive runs of the multiplex real-time PCR assay.

DNA sequencing and Legionella species identification.

Products of the 16S rRNA gene PCR were sequenced with a 310 Genetic Analyser (Applied Biosystems, Warrington, England) with a Big Dye terminator cycle sequencing ready reaction (Applied Biosystems, Warrington, England. For the sequencing, 0.1 μM primer was used. Identification of Legionella species was done by comparison of the gene sequence obtained with the sequences in GenBank.

RESULTS

Specificity.

The Legionella real-time PCR assay detected DNA from 12 L. pneumophila serogroups and 24 non-L. pneumophila Legionella species. The L. pneumophila real-time PCR assay amplified DNA from 12 L. pneumophila serogroups but none of the 24 Legionella species tested. Both assays generated negative results with a selection of other bacteria, including Bordetella pertussis, Bordetella parapertussis, Mycobacterium pneumoniae, Streptococcus pneumoniae, Haemophilus influenzae, Streptococcus pyogenes, Mycobacterium catarrhalis, Streptococcus viridans, S. enterica serovar Enteritidis, S. enterica serovar Typhimurium Enterococcus faecalis, Staphylococcus aureus, Acinetobacter baumannii, Neisseria meningitidis, Pseudomonas aeruginosa, Enterobacter aerogenes, Enterobacter faecium, Klebsiella pneumoniae, Chlamydophila pneumoniae, Chlamydophila psittaci, and Escherichia coli.

Sensitivity.

Sensitivity was determined by analyzing dilutions of the extracted DNA of Legionella spp. and L. pneumophila. The number of cells per milliliter was based on McFarland turbimetric standards. The dilutions obtained were cultured on BCYE to determine CFU; 200 μl of the dilutions obtained which related to the CFU were extracted and resuspended in 200 μl of buffer. The sensitivity was 2.5 CFU/ml for all the L. pneumophila serogroups. The Legionella. assay also had a sensitivity of 2.5 CFU/ml for the 24 non-L. pneumophila strains tested.

Optimization of multiplex assay.

Both Legionella real-time PCR assays were first optimized as a monoassay. A 10-fold dilution series of Legionella DNA was used to make a standard curve, and this was analyzed by means of the iCycler IQ real-time detection apparatus software to determine assay efficiency and correlation coefficient. Theoretically, no loss in sensitivity is observed in a multiplex assay if two 100% efficient assays are combined. The efficiency obtained was 100%, 103%, and 97% for the Legionella spp. assay, L. pneumophila assay, and PhHV assay, respectively (Fig. 1). The multiplex assay was optimized, and thereafter, the cycle threshold (Ct) values obtained from testing a dilution series of L. pneumophila in the individual assay and the multiplex assay were similar, indicating the same level of sensitivity. (Table 2). The specificity of the multiplex assay was no different from that of the individual assay. The different fluorescent labels on the beacons had no cross-reaction in the iCycler real-time detection system, and therefore the different probes could be distinguished easily (Fig. 2).

FIG. 1.

(A) Relative fluorescence units obtained for a 10-fold dilution series of L. pneumophila ATCC 33152 from 10⁷ to 10² CFU/ml for the Legionella spp. assay. Each dilution gives a Ct value which can be plotted against the CFU to obtain a standard curve. The value obtained for the slope of the line gives a value for PCR efficiency; in this case it is 100%. (B) Relative fluorescence units obtained for a 10-fold dilution series of L. pneumophila ATCC 33152 from 10⁷ to 10² CFU/ml for the L. pneumophila assay. Each dilution gives a Ct value which can be plotted against the CFU to obtain a standard curve. The value obtained for the slope of the line gives a value for PCR efficiency; in this case it is 102.7%.

TABLE 2.

Comparison of Ct values in monoassay and multiplex assay after real-time amplification of a dilution series of an L. pneumophila serogroup 1 standard

CFU/ml	Ct
	Monoassay		Multiplex assay
	L. pneumophila	Legionella spp.	L. pneumophila	Legionella spp.
100,000	27.2	27.7	27.4	27.8
10,000	31.2	30.6	31.6	32.8
1,000	34.3	34.5	35.2	35.5
100	38.5	38.3	38.6	39.4
10	43.4	42.4	43.2	44.7

FIG. 2.

Determination of results. Each panel displays the readout from the multiplex PCR with the fluorophore indicated. (A) Sample positive for L. pneumophila, showing fluorescence with FAM (Legionella spp.), HEX (L. pneumophila), and Cy-5 (internal control). (B) Sample positive for Legionella spp. but not L. pneumophila, showing fluorescence with FAM (Legionella spp.) and Cy-5 (internal control). (C) Negative samples, showing fluorescence only with Cy-5 (internal control).

Reproducibility of multiplex real-time PCR assay.

Samples containing 5000 CFU/ml were assayed in quintuplicate in the multiplex assay to determine the interassay and intra-assay variability. As determined from the Ct values obtained from five consecutive runs, the interassay mean Ct value was 32.2 (standard deviation, 0.4) and 29.3 (standard deviation, 0.6) for L. pneumophila and Legionella species, respectively. The mean of the intra-assay variability was 1.1 (range, 0.4 to 1.8) and 0.6 (range, 0.4 to 0.7) for the L. pneumophila and Legionella species assay, respectively.

Evaluation of clinical material in multiplex assay.

Sputum, throat swab, and bronchoalveolar lavage samples were all spiked with L. pneumophila (ATCC 33152) at 5,000 CFU/ml, 500 CFU/ml, and 50 CFU/ml, and serial dilutions of the extracted DNA were performed. The sensitivity of both assays in the multiplex was 10 CFU/ml of spiked sputum, bronchoalveolar lavage, and throat swab samples. The amount of sample spiked into the samples did not affect the final sensitivity.

The results from the clinical samples tested as part of the proficiency panel were all as expected. The 12 positive samples were detected by both assays, and the eight negative samples were negative by both assays, resulting in a 100% score. In this assay, the weakly positive samples had a mean Ct value of 27 (range, 23 to 29).

Clinical samples shown to be culture positive were analyzed retrospectively and were all detected by the multiplex PCR. Nine of the samples were shown to contain L. pneumophila and were detected by both assays. One sample was L. bozemanii, and this was only detected by the Legionella species assay. Of the clinical samples analyzed prospectively, all were culture negative, but 2 (3%) of the 64 bronchoalveolar lavage samples were positive for Legionella species by PCR. All 64 bronchoalveolar lavage samples were negative for the L. pneumophila assay. One of the two positive specimens was found to contain L. bozemanii and the other L. anisa. The L. bozemanii bronchoalveolar lavage sample was from a patient who had other culture-positive samples for L. bozemanii. Two (3%) of the 64 samples were found to contain inhibitory compounds. Two of the 10 throat swabs were positive for L. pneumophila, and all were culture negative; all 10 were negative in a Legionella spp. assay. Acute and convalescent serology samples from these patients demonstrated the presence of recent Legionella infection. A summary of the results obtained with clinical samples is shown in Table 3.

DISCUSSION

The real-time PCR assay described provides an effective way to detect and discriminate infections with L. pneumophila and Legionella species and check for appropriate DNA extraction and inhibition in a single tube. Other methods that discriminate between L. pneumophila and other Legionella species either rely on final discrimination with melting curve analysis (21, 23) or have to be performed in separate tubes (9, 26). The multiplex real-time assay described here is as sensitive as the individual assays. Multiplexing can be affected by the addition of several primer sets. Resichl et al. showed that in order to multiplex with the Lightcycler, more optimal results were obtained with one set of primers as opposed to two (23). The software analysis possible with the iCycler IQ real-time detection system resulted in complete optimization of the monoassays and facilitated their use in a multiplex reaction.

In the multiplex reaction, an internal control reaction has been included to check the validity of the PCR result by monitoring the nucleic acid extraction as well as presence of inhibitors. Of the 64 samples tested, 2 were shown to inhibit the PCR. The samples that are likely to be used for clinical diagnosis include sputum, bronchoalveolar lavage, and lung tissue samples, all of which may have inhibitors to PCR. Others have performed the internal control step separately (21, 23) or in duplex with either one of the Legionella assays (26), but not in the same tube.

The sensitive detection of Legionella spp. at the genus level as well as the differentiation between L. pneumophila and non-L pneumophila species is important in the case of clinically suspected pneumonia. As Legionella infection does not present with a distinctive clinical syndrome, it cannot be reliably differentiated from pneumonia due to other respiratory pathogens on the basis of signs and symptoms (19). The differentiation between L. pneumophila and Legionella species is also required by clinicians because L. pneumophila is associated with more significant morbidity and mortality. However, the importance of Legionella spp. other than L. pneumophila may be underrecognized, as routine diagnostics are more directed to L. pneumophila (18).

Several studies have found that PCR methods have a higher rate of detection than culture-based methods (9, 16), Although in this study the numbers are small, four more Legionella samples were identified by PCR. Two of the additional positives were for L. pneumophila, and those were both in throat swabs; no sputum sample was available in these cases. It is known that throat swabs are not as good a sample for culture, so this may explain why the culture was negative, as the serology was also positive. However, these results show that sputum production is not always possible in patients with Legionella infection and that PCR is a better alternative. In addition, the use of real-time PCR not only has good sensitivity but also decreases the turnaround time for results and shows a marked reduction in the risk of carryover contamination.

In order to perform this assay, additional controls to monitor for environmental contamination with Legionella spp. is also very important. Diagnostic results obtained should relate to Legionella in the clinical sample only. Use of kit-form DNA isolation systems does not diminish the problem of reagent-derived Legionella contamination (25). In addition, Taq polymerase has also been shown to be contaminated with bacterial DNA (4), and hence Legionella contamination is also possible in PCR master mixes.

The characteristics of this Legionella multiplex real-time PCR assay make it a useful addition for Legionella diagnosis. The assay is performed in a single tube with no post-PCR analysis required, and discrimination between L. pneumophila and other Legionella species is obtained in a single reaction.