ABSTRACT
Tuberculosis (TB) remains a major health problem worldwide. Control of TB requires rapid, accurate diagnosis of active disease. However, extrapulmonary TB is very difficult to diagnose because the clinical specimens have very low bacterial loads. Several molecular methods involving direct detection of the Mycobacterium tuberculosis complex (MTBC) have emerged in recent years. Real-time PCR amplification simultaneously combines the amplification and detection of candidate sequences by using fluorescent probes (mainly TaqMan or Molecular Beacons) in automated systems. The multiplex real-time PCR-short assay is performed using locked nucleic acid (LNA) probes (length, 8 to 9 nucleotides) in combination with CodUNG to detect multiple pathogens in clinical samples. In this study, we evaluated the performance of this novel multiplex assay for detecting the MTBC in comparison with that of the conventional culture-based method. The multiplex real-time PCR-shortTUB assay targets two genes, whiB3 (redox-responsive transcriptional regulator) and pstS1 (phosphate-specific transporter), yielding limits of detection (LOD) of 10 copies and 100 copies, respectively, and amplification efficiencies of 92% and 99.7%, respectively. A total of 94 extrapulmonary samples and pulmonary samples with low mycobacterial loads (all smear negative; 75 MTBC culture positive) were analyzed using the test, yielding an overall sensitivity of 88% and a specificity of 95%. For pleural fluid and tissues/biopsy specimens, the sensitivity was 83% and 85%, respectively. In summary, this technique could be implemented in routine clinical microbiology testing to reduce the overall turnaround time for MTBC detection and may therefore be a useful tool for the diagnosis of extrapulmonary tuberculosis and diagnosis using pulmonary samples with low mycobacterial loads.
INTRODUCTION
Tuberculosis (TB) is an ancient infectious disease caused by the Mycobacterium tuberculosis complex (MTBC), and it remains one of the most serious health problems worldwide, causing high morbidity and mortality rates (1). It is the ninth leading cause of death and the leading cause from a single infectious agent, ranking above HIV/AIDS. In 2017, an estimated 10 million new TB cases occurred, with 1.3 million deaths among HIV-negative people and an additional 300,000 deaths among HIV-positive people. Although TB typically affects the lungs (pulmonary TB), the extrapulmonary form of the disease represented, on average, 14% of notified cases (range, 8% to 24%) in 2017, and the frequency of presentation increased markedly in HIV-infected patients, according to the WHO Global Tuberculosis Report 2018 (1).
The most important and representative bacterial species causing TB is Mycobacterium tuberculosis, which belongs to the Mycobacterium tuberculosis complex (MTBC), although other members of the MTBC, such as Mycobacterium bovis strains (including BCG), Mycobacterium africanum, Mycobacterium canettii, and Mycobacterium microti, may also be of clinical relevance (2, 3).
One of the principles of tuberculosis control is the rapid and accurate diagnosis of the disease, in order to allow prompt initiation of antimicrobial therapy and prevent transmission. Unfortunately, the conventional microbiological diagnosis of TB has important limitations, such as the poor sensitivity of acid-fast bacillus detection by microscopy and the slow growth of the tubercle bacillus in culture media (4). Screening tests to detect nucleic acids, which reduce diagnostic times and improve sensitivity, have been widely implemented over the past two decades, thus greatly contributing to resolving the shortcomings of conventional methods (3, 5, 6).
The efficacy of these new molecular methods of detecting MTBC in extrapulmonary and some lung samples (such as biopsy samples) remains a challenge due to the low bacillary loads in these types of samples (7, 8). However, thanks to sequencing techniques (whole-generation sequencing [WGS]), it is possible to study new targets conserved in bacterial genomes (9), which could be used to design new diagnostic real-time PCR-based tools. A new real-time PCR platform that allows simple, fast, and economical modification of the detection targets would therefore be useful.
In the present study, we decided to evaluate the use of TaqMan LNA probes (length, 8 to 9 nucleotides) (10) as “short platform” probes to detect MTBC. These nucleotides are analogous to strong bonds with the cDNA strand, and they enable shorter targets to be detected with higher sensitivity and specificity. Furthermore, we used the uracil-N-glycosylase (UNG) enzyme isolated from the Atlantic cod, Gadus morhua (CodUNG), which has been shown to be 100 times more efficient at eliminating contaminating amplicons than Escherichia coli UNG (10). Although Champlot et al. (11) used a 103-bp-long PCR amplicon containing either deoxyuridine (dU) or deoxyribosylthymine (dT), these researchers found that the CodUNG enzyme specifically degraded 99.9% ± 0.04% of the dU-containing amplicon, whereas the bacterial UNG degraded only 92.8% ± 0.53% (11).
Finally, we used a combination of the LNA probes (length, 8 to 9 nucleotides) and the CodUNG enzyme to demonstrate the presence of the MTBC by detection of the whiB3 (redox-responsive transcriptional regulator) and pstS1 (phosphate-specific transporter) genes (which are single-copy genes in the DNA genome) (12–16) in samples with low mycobacterial loads. We selected these conserved genes (whiB3 and pstS1) after considering the data from the Integrated Microbial Genomes & Microbiomes system (IMG/M; https://img.jgi.doe.gov/m/) (Table 1).
Bioinformatic analysis of conservation of the whiB3 and pstS1 genes in the MTBC
MATERIALS AND METHODS
DNA extraction.(i) Isolates. DNA was automatically extracted from the MTBC, from nontuberculous Mycobacterium spp. (Mycobacterium avium, Mycobacterium intracellulare, Mycobacterium kansasii, Mycobacterium malmoense, Mycobacterium abscessus, Mycobacterium chelonae, Mycobacterium xenopi, and Mycobacterium gordonae), and from Nocardia spp. (Nocardia asteroides and Nocardia farcinica) using the MagNA Pure system (Roche, Germany) and following the manufacturer’s instructions. The DNA was identified with the rpoB housekeeping gene sequencing method (17). The specificity and sensitivity of these primers and probes in the multiplex real-time PCR-shortTUB assay were analyzed with DNA isolated from nontuberculous mycobacteria and from Nocardia spp.
(ii) Clinical samples. A total of 94 samples (75 MTBC culture positive and 19 culture negative) from 94 patients were assayed. All samples were cultured on solid medium (Coletsos medium; bioMérieux Clinical Diagnostics, Spain). The different types of sample included sterile fluids (pleural, cerebrospinal, and ascitic fluids), urine, abscess aspirates and swabs, tissues (joint, colon, lung, skin, bone, liver, kidney, and intervertebral disc biopsy specimen), and stool.
The samples were obtained and stored frozen (1-ml aliquots) between 2007 and 2016, after smear microscopy for acid-fast bacilli (AFB) and subsequent culture. The samples were retrospectively analyzed (between July and September 2017) using the multiplex real-time PCR-shortTUB platform in the Microbiology Department, Bellvitge Hospital (Barcelona, Spain). All samples were automatically extracted using the MagNA Pure compact instrument (Roche, Germany). For sterile liquids, including cerebrospinal fluids, 1 or 2 ml of sample was used. Moreover, the sterile liquids were pretreated with 50 μl of lysozyme according to the manufacturer’s indications (Sigma-Aldrich, Merck, Spain). For liquid aspirates and swabs, 200 or 400 μl of the liquid obtained from a lymph nodes or abscesses was used. Finally, the DNA was extracted from tissues (25-g samples) and biopsy specimens (20 μl) by using the QIAamp DNA minikit (Qiagen, Germany).
Multiplex real-time PCR-shortTUB.The primers and probes used in the study and listed in Table 2 were designed from conserved DNA regions from the whiB3 gene (GenBank accession number AXA88767.1) and pstS1 gene (GenBank accession number AWY85876.1). Moreover, the multiplex assay also includes an internal control (β-globin gene) to ensure that the extraction and amplification processes are functioning correctly (Table 2).
Primers and probes used in this studya
The multiplex real-time PCR-shortTUB assay (m2000rt real-time instrument; Abbott) was performed using TaqMan locked nucleic acid (LNA) probes 8 to 9 nucleotides in length (Roche; Exiqon). LNA nucleotides are analogous to strong bonds (2′-O-4′-C methylene bridges), which maintain the melting temperature and are highly discriminatory (single mismatch). The multiplex real-time PCR-shortTUB assay was then conducted with DNA extracted from the clinical samples. The total reaction volume (20 μl) included 10 μl of LightCycler 480 DNA Probes Master, 5 μl of DNA sample, 0.75 μl of each primer (concentration 100 μM), 0.5 μl of each probe (10 μM TaqMan LNA probe), and 3 μl of water. The samples were incubated with 0.2 μl of CodUNG (ArcticZymes) for 15 min at 25°C (to induce enzyme activation) in order to remove contaminating amplicons of length 65 to 120 bp and containing either dU or dT. CodUNG proved more effective at eliminating amplicons of around 100 bp long than uracil-N-glycosylases from bacteria (11). The following program was applied (duration, 50 min): (i) preincubation for 10 min at 95°C, (ii) amplification for 10 s at 95°C, 20 s at 60°C, and 1 s at 72°C (45 cycles), and (iii) cooling for 10 s at 40°C. The water for the reaction mixture of the multiplex real-time PCR-shortTUB assay was used as a negative control. Finally, in order to determine the limit of detection of the whiB3, pstS1, and β-glob genes in the multiplex real-time PCR-shortTUB assay, serial 10-fold dilutions of a mixture of plasmid DNA from E. coli strains (TOPO TA cloning kit, Thermo Fisher Scientific) of the PCR-targets were tested at concentrations ranging between 1 and 107 copies per well, with 3 replications per dilution. In addition, the calibration curve parameters were analyzed in order to determine the amplification efficiency of the assay.
Evaluation of the multiplex real-time PCR-shortTUB assay for clinical samples with low mycobacterial loads.The results of the multiplex real-time PCR-shortTUB assay in clinical samples were classified as follows: (i) positive, when amplification of either of the target genes occurred and the crossing point was lower than 45; (ii) negative, when neither of the target genes was amplified, with a crossing point lower than 45; (iii) invalid, amplification of the negative control (water from the PCR mix); and (iv) inhibited, amplification failure of the internal control. The sensitivity and specificity of the multiplex real-time PCR-shortTUB for detection of the MTBC were also calculated relative to the conventional culture technique.
Statistical analysis.Qualitative data analysis was performed using SPSS, with either the chi-square test or Fisher’s exact test, as appropriate. P values less than 0.05 were considered statistically significant. A normal distribution was assumed, and both P values and 95% confidence intervals were calculated for each item.
RESULTS
The limit of detection (LOD) and amplification efficiency (%) of each gene in the singleplex and multiplex formats of the assay are shown in Tables S1, S2, and S3. In the singleplex assay, the LOD of the three genes (whiB3, pstS1, and β-glob) was 10 copies, and the amplification efficiency was 95 to 100%. In the multiplex assay, the LOD of the three above-mentioned genes was between 10 and 100 copies, and the amplification efficiency was 99.7% for the whiB3 gene, 92% for the pstS1 gene, and 97.2% for the β-glob gene. Neither whiB3 or pstS1 yielded positive results in the multiplex real-time PCR-shortTUB assay for nontuberculous Mycobacterium DNA or actinomycetal DNA (Table S4).
For analysis of clinical samples with low mycobacterial loads (all negative by smear microscopy), we included a total of 94 samples from 94 patients attended in our center. Of these, 79.8% had tested positive for MTBC by traditional culture method. The multiplex real-time PCR-shortTUB assay yielded a global sensitivity of 88% (66 out of 75) and a specificity of 95% (18 out of 19). Of the 66 positive results obtained with the assay, 47 detected both whiB3 and pstS1 genes (71.2%). Ten samples were whiB3 positive and pstS1 negative, which can be attributed to the fact that the detection limit of the multiplex assay was lower for the whiB3 gene than that for the pstS1 gene (10 copies versus 100 copies). On the other hand, nine samples only showed a positive signal for the pstS1 gene. In 4 of these 9 clinical samples, species-level identification revealed the presence of Mycobacterium bovis strains (including BCG), probably indicating the higher ability of pstS1 gene amplification for this strain.
This finding is also supported by the results obtained in the bioinformatic analysis, as mentioned in the Discussion, which shows higher conservation of the pstS1 gene than of the whiB3 gene in M. bovis (94% versus 92%, respectively).
Finally, eight of the positive culture samples tested negative for both targets in the assay. In two of these cases, only the solid culture (Coletsos medium; bioMérieux Clinical Diagnostics, Spain) was positive, after incubation for more than 30 days, suggesting the presence of very low bacterial loads in the samples. One sample was inhibited, as shown by a negative result in the internal control. The samples were classified as follows according to the source: (i) sterile fluids, (ii) nonsterile fluids (urine), (iii) aspirates and swabs, (iv) tissues, and (v) stool specimens (Table 3). The sensitivity did not differ significantly in relation to the source of the sample. Among the 19 negative-culture samples analyzed, 18 yielded a negative result in the multiplex real-time PCR-shortTUB assay, and one stool specimen yielded a late positive result (threshold cycle [CT] > 39) for both whiB3 and pstS1 genes. This PCR-positive result could not be retested because of an insufficient amount of sample. The overall specificity was 95% (see Table 3).
Results obtained with the multiplex real-time PCR-shortTUB assay according to the source and result of the MTBC culture
DISCUSSION
Numerous studies have assessed the use of real-time PCR to detect the MTBC, with probes of around 15 to 20 nucleotides in length (7, 18–24) (Table 4). In this study, we evaluated for the first time the ability of the multiplex real-time PCR-shortTUB assay, which uses LNA probes of around 8 to 9 nucleotides in length (25) in combination with uracil-N-glycosylase (CodUNG) to detect MTBC.
Sensitivity and specificity of real-time PCR tests
We used bioinformatic tools (IMG/M; https://img.jgi.doe.gov/m/) to analyze two targets (whiB3 and pstS1) for inclusion in the multiplex real-time PCR-shortTUB assay, as these genes are present in 99% of the 1,000 genomes in the MTBC (see Table 1). Both genes are fully conserved in 31 genomes of M. africanum and in only one genome of M. microti. On the other hand, this bioinformatic tool reveals greater conservation of the pstS1 gene than of the whiB3 gene in the 63 genomes of M. bovis (94% versus 92%, respectively). The efficacy of the assay was evaluated relative to that of conventional methods.
The whiB3 gene (of which a single copy occurs in the DNA genome) is conserved throughout the MTBC (see Table 1), and we therefore chose it as the target gene for detecting the pathogen, with the aim of improving the detection of all species in the complex (12, 15). Together the pstS1 and pstS11 genes constitute the Pst system in the MTBC. The Pst system is responsible for transporting phosphate and is essential for survival of the bacteria under certain types of stress, such as low phosphate concentration or acidic conditions. The system has been conserved (single copy in the DNA genome) throughout evolution of the bacterium and is present in all species of the tuberculosis complex. The pstS1 gene is a potentially valuable target for detecting different species in the MTBC (13, 14). The experimental findings indicated a greater efficiency of the assay to identify M. bovis. However, as only 4 samples were positive for M. bovis, we carried out a bioinformatic analysis with the 63 M. bovis genomes deposited in the database to compensate for the lack of samples (see Table 1). The results showed that the pstS1 gene could potentially be used to differentiate species within the MTBC. Nonetheless, further studies with clinical samples are required to confirm this hypothesis.
The real-time PCR technique has been used to detect MTBC with different DNA targets and effectiveness (Table 4). The DNA targets most frequently used to detect MTBC are rpoB and IS6110 (7, 18, 21, 23, 24), and other DNA targets used include rpnB (22), 16S rRNA and rRNA (rDNA) (20), and the intergenic region SenX3-RegX3 (19).
The highest overall sensitivity and specificity of the real-time PCR technique in these studies, which analyzed a large number of clinical samples with low mycobacterium loads, were obtained by Hinic et al. (100% and 100%, respectively) (23) and Queipo-Ortuño et al. (19) (93.3% and 100%, respectively) (Table 4). We also highlight our results with the multiplex real-time PCR-shortTUB assay (88% and 95%, respectively) and those of Tortoli et al. (21) (81.3% and 99.8%, respectively) (Table 4).
In relation to the sensitivity of the real-time PCR technique with sterile fluids, the highest value was reached with the multiplex real-time PCR-shortTUB assay (85%), followed by the Xpert MTB/RIF Ultra assay (60.5%) (24). However, the Xpert MTB/RIF (GX) technique showed low sensitivity with sterile fluids (26.9%) (7) and cavitary fluids (<50%) (21).
One of the most outstanding features of the GeneXpert MTB/RIF platform is its ability to detect rifampin-resistant strains, which is of paramount importance in regions where the burden of the disease is mainly due to multidrug-resistant strains of the MTBC (26). Nevertheless, in industrialized countries, most tuberculous infections affect aging populations and immunocompromised individuals (both of which are increasing in numbers). This has led to a situation in which the greatest concern is disseminated TB, mainly caused by susceptible strains in patients who require hospitalization (27, 28). The availability of diagnostic tools that can detect the presence of microorganisms in a variety of sample types is therefore essential for integral management of patients, and resistance to rifampin will therefore be of secondary importance, although the nondetection of these strains is a limitation of our assay. Moreover, the multiplex real-time PCR-shortTUB assay could be developed with RNA from biological samples in which the MTBC is still viable and indicating active infection. This would represent a great advantage in the clinical use of this technique for diagnosing tuberculosis (29). Garcia and colleagues used the real-time PCR-shortTUB technique to study the M. tuberculosis RNA transcriptional profile in sputum samples, showing it to be a powerful tool for evaluating treatment responses in vivo (30).
In conclusion, the multiplex real-time PCR assay short platform was used to detect the Mycobacterium tuberculosis complex (multiplex real-time PCR-shortTUB assay) by targeting the whiB3 and pstS1 genes, and it showed excellent sensitivity and specificity in clinical samples with low mycobacterial loads, including smear-negative samples. This technique may therefore be a valuable tool for the rapid diagnosis of low-mycobacterial-load extrapulmonary and pulmonary TB when clinical suspicion is high.
ACKNOWLEDGMENTS
This study was funded by grant PI16/01163 awarded to M. Tomás within the State Plan for R+D+I 2013–2016 (National Plan for Scientific Research, Technological Development and Innovation 2008–2011) and cofinanced by the ISCIII-Deputy General Directorate for Evaluation and Promotion of Research–European Regional Development Fund “A way of Making Europe” and Instituto de Salud Carlos III FEDER, the Spanish Network for Research in Infectious Diseases (REIPI; grants REIPI RD16/0016/0006 and RD16/0016/0011), the Study Group on Mechanisms of Action and Resistance to Antimicrobials-GEMARA and Mycobacterial Infections Study Group–GEIM (SEIMC; http://www.seimc.org/). M. Tomás was financially supported by the Miguel Servet Research Program (SERGAS and ISCIII).
This work is involved in two patents: (i) ES2449665 (Spain), “Procedimiento para el diagnóstico clínico de una enfermedad infecciosa basada en el empleo de la PCR cuantitativa, oligonucleótidos marcados de 8 a 9 nucleótidos de longitud y la UNG del Bacalao del Atlántico (Gadus morhua),” whose holders are Servizo Galego de Saúde (SERGAS) and Fundación Profesor Novoa Santos, and (ii) PCT/EP2017/069849 (International), “Use of short probes between 8 and 9 nucleotides in multiplex assays,” whose holders are Servizo Galego de Saúde (SERGAS) and Fundación Profesor Novoa Santos.
FOOTNOTES
- Received 7 May 2019.
- Returned for modification 28 May 2019.
- Accepted 9 June 2019.
- Accepted manuscript posted online 12 June 2019.
Supplemental material for this article may be found at https://doi.org/10.1128/JCM.00733-19.
- Copyright © 2019 American Society for Microbiology.
