|Home | About | Journals | Submit | Contact Us | Français|
We designed primers and probes for the detection and discrimination of Ureaplasma parvum and U. urealyticum and their 14 serovars by real-time PCR. The analytical sensitivity and specificity of the multiplex species-specific PCR were determined by testing corresponding American Type Culture Collection (ATCC) type strains, 47 other microbial species, and human genomic DNA. The limits of the multiplex PCR were 2.8 × 10−2 CFU/μl PCR mixture for detecting U. parvum and 4.1 × 10−2 CFU/μl PCR mixture for detecting U. urealyticum. Clinical specificity and sensitivity were proven by comparison with culture and traditional PCR. For the detection of any Ureaplasma species, the clinical sensitivity and specificity of real-time PCR were 96.9% and 79.0%, respectively, using culture as a reference. Multiplex real-time PCR was also more sensitive than traditional PCR in discriminating the two Ureaplasma species in culture-positive subcultures. Each of the 14 monoplex serovar-specific PCR assays was specific for the corresponding ATCC type strain serovar. This new species identification PCR is specific and sensitive in the detection of Ureaplasma species in clinical specimens, and the serovar-specific PCR assays are the first set of complete genotypic assays to differentiate all 14 known Ureaplasma serovars. These assays provide quick and reliable means for investigating the epidemiology and pathogenicity of ureaplasmas at the serovar level.
Shepard provided the first descriptions of T-strain mycoplasmas, now known as ureaplasmas, in 1954 when he cultivated them in vitro from the urethras of men with nongonococcal urethritis (26), but the genus Ureaplasma was not formally designated until 1974 (28). Ureaplasma urealyticum, eventually shown to consist of two biovars, was considered to be the only species of this genus known to infect humans until 2002, when its two biovars were reclassified as two distinct species, U. parvum and U. urealyticum, based on genome size, 16S rRNA gene sequences, the 16S-23S rRNA intergenic region, enzyme polymorphisms, DNA-DNA hybridization, differential growth responses to manganese, and differences in the multiple-banded antigen (mba) genes (23). At least 14 serovars have been identified in human ureaplasmas; U. parvum contains serovars 1, 3, 6, and 14, while U. urealyticum contains the remaining 10 serovars, 2, 4, 5, 7, 8, 9, 10, 11, 12, and 13 (23).
The two Ureaplasma species can be isolated individually or simultaneously from clinical specimens (35). They are common commensals colonizing the urogenital tract in healthy persons, yet they are also implicated in invasive diseases such as urethritis, postpartum endometritis, chorioamnionitis, spontaneous abortion, and premature birth, as well as low birth weight, pneumonia, bacteremia, meningitis, and chronic lung disease in prematurely born infants (35). The reasons why Ureaplasma spp. are commensals in some instances and produce clinically significant infections in others are still unknown, although an intact functional humoral immune system appears to be very important, as evidenced by the frequent occurrence of systemic Ureaplasma infections such as septic arthritis in adults and children with antibody deficiencies (35). Speculations about the association of particular Ureaplasma species or serovars with certain diseases have provided the rationale for investigations to develop accurate methods for determining the serovars of clinical isolates (35).
Typing methods for Ureaplasma serovar designation using polyclonal or monoclonal antibodies directed against whole cells or purified antigens have included growth inhibition tests (2, 27), metabolic inhibition tests (22), immunofluorescence of colonies on agar (8, 25, 29), immunoperoxidase (20), enzyme-linked immunosorbent assay (33), and immunoblot assays (36, 41). These antibody-based techniques are time-consuming, cumbersome assays that often yield results that are not reproducible, are difficult to interpret, and are inconclusive because of multiple cross-reactions and poor discriminating capacity when used with clinical samples containing two or more serovars. Lack of commercial availability and standardization generally limited serotyping to those laboratories that developed the individual serological reagents.
New rapid molecular methods based on genotypic instead of phenotypic markers are logical choices to replace the conventional antibody-based serotyping methods. PCR-based methods are also becoming an important alternative to conventional culture for the initial detection of ureaplasmas in clinical specimens and have the additional advantage of discrimination between the two Ureaplasma species (24). The gel-based traditional PCR assays for the detection and species identification of human ureaplasmas targeted sequences of the 16S rRNA gene and the 16S-23S rRNA intergenic spacer region (7, 9, 14, 24), the urease gene subunit (3, 19), or the multiple-banded antigen (mba) gene (13-15, 30, 31). Major issues remaining to be resolved for traditional PCR are its specificity and analytic sensitivity, as well as the potential for contamination. Real-time PCR assays targeting the urease gene (5, 17, 37), the mba gene (6), and the 16S rRNA gene (39) have recently been developed to distinguish the two Ureaplasma species. Compared to traditional PCR and culture, real-time PCR species identification assays are quantitative, as well as more rapid, specific, and sensitive and less subject to contamination (5). However, some of the previously published real-time PCR assays available for Ureaplasma species identification require two separate tests (5, 17) or do not have satisfactory sensitivity in one combined test (37).
To characterize the Ureaplasma species at the serovar level by PCR, genotyping primers based on the mba gene and its 5′-end upstream regions have been designed previously and partial serovar identification was achieved. In combination with direct sequencing or restriction enzyme analysis, these assays were capable of distinguishing the 4 serovars of U. parvum and divided the 10 serovars of U. urealyticum into different subgroups (5, 11, 13, 15, 32, 40). Among these PCR typing methods, only one was real-time based (5). Because of limited sequence variation in the mba genes, all of the PCR-based methods reported so far have lacked the capacity for the complete separation of all 14 serovars. Moreover, our personal experience with whole-genome sequencing of all 14 serovars has shown mba to be part of a large gene family present in many variations in different serovars and the gene is phase variable (42). Thus, while the mba gene is still, in some cases, a valid and in fact necessary target for serovar-specific PCRs, attention must be given to the locations of phase-variable forms of these genes so that PCR amplicons do not span recombination sites within these genomes.
We have evaluated the complete genome sequences of all 14 Ureaplasma serovars and designed primers and probes targeting new specific regions for species and serovar identification by real-time PCR assays. One species-specific multiplex PCR and 14 serovar-specific monoplex PCRs were developed and assessed for analytical and clinical specificity and sensitivity.
The 14 type strains of the Ureaplasma serovars obtained from the American Type Culture Collection (ATCC) and used to validate the species and serovar identification assays are described in Table Table1.1. Additional live microorganisms or their DNA used to validate assay specificity are also described in Table Table11.
Two groups of clinical specimens or microorganisms were used in this investigation. They consisted of one group of 132 consecutive neonatal nasal or endotracheal aspirate specimens obtained between 2004 and 2007 for diagnostic purposes and a second group of 194 consecutive Ureaplasma-positive broth subcultures of neonatal nasal or endotracheal aspirates obtained for diagnostic purposes between 1994 and 2003. Bacterial subcultures and clinical specimens were kept frozen at −80°C until processed for species and serovar identification PCR assays. Cultures were performed as described by Waites et al. (34).
Genomic DNA was extracted by the proteinase K method as described previously (4). If inhibition occurred, proteinase K-digested samples were purified using the QIAamp DNA Blood Mini Kit (Qiagen, Valencia, CA). Prepared DNA samples were stored at −80°C.
Primer/probe sets were designed using Roche LightCycler Probe Design Software 2.0 (Table (Table2).2). A multiplex real-time PCR assay was designed to differentiate the two Ureaplasma species simultaneously. The U. parvum primer/probe set anneals to the 477-bp UP063 gene (NP_077893), which encodes a conserved hypothetical protein that is identical in all four U. parvum type strain serovars. The U. urealyticum primer/probe set anneals to a 15,072-bp open reading frame that is almost perfectly (>99.97%) conserved in all 10 U. urealyticum type strain serovars (serovar 10, GenBank ID ACI60066.1). To design the serovar-specific primers and probes, the genomes of the 14 ATCC type strains that had been previously sequenced (Table (Table3)3) were computationally searched to identify 25- to 35-bp candidate regions that are unique to each genome. Each of the candidate region or locus sequences in a genome was then searched against all of the other genomes, and only those that had <80% identity matches were kept. The Roche LightCycler Probe Design Software 2.0 was used to analyze the regions around some of the unique 30-bp regions, and primers and probes were generated. Each possible primer was checked against all 14 Ureaplasma genomes using BLAST, and fewer than 1 in 20 of the unique small regions yielded acceptable serovar-specific primers based on computational analysis. The resulting primers and probes shown in Table Table22 were tested with the LightCycler using the 14 ATCC type stains. The specificity of eight serovars (serovars 2, 3, 6, 7, 8, 9, 13, and 14) came from the two primers, while the other six were from probes, either Simple Probe Chemistry (SPC) probes (serovars 1, 5, 10, and 11) or fluorescence resonance energy transfer (FRET) probes (serovars 4 and 12). Serovar 5 primers may also amplify the other nine U. urealyticum serovars; however, the amplicon is much larger (668 bp versus 121 bp in serovar 5) and the probe has a large unmatched loop (547 bp), so this should not affect specificity. A short extension time that only allows complete amplification of the serovar 5 fragment but not the others enables the distinction. A melting curve analysis will show the differences of the products if unexpected amplification has occurred. Urease-based primers and 32P-labled probes for Ureaplasma species identification by traditional PCR were described previously (10). All primers were synthesized by Invitrogen (Carlsbad, CA), and probes were manufactured by Roche Diagnostics.
Real-time PCR assays were performed on a Roche LightCycler 2.0 instrument. An asymmetric PCR condition was used for the multiplex species identification PCR; the PCR master mix contained less of the forward primers (0.3 μM UP063#1F and 0.2 μM UU127#1F) than of the reverse primers (0.5 μM). The other components were 0.15 μM each UP probe, 0.2 μM each UU probe, 0.5 U of uracil-DNA glycosylase (UNG), 3 mM MgCl2, and 4 μl of 5× Multiplex DNA Master HybProbe buffer (Roche Diagnostics). A volume of 2 μl of the specimen was added to the master mix to reach a total reaction volume of 20 μl. The PCR program was 10 min preincubation at 40°C and 95°C, followed by 45 cycles of amplification at 95°C for 15 s, 55°C for 10 s with a single measurement, and 72°C for 9 s. Melting curves were generated by 95°C for 0 s, 65°C for 30 s, and 95°C with a slope rate of 0.1°C/s in continuous-acquisition mode. Finally, the machine was cooled down to 40°C for 30s. The PCR conditions for 14 monoplex serovar-specific PCRs are summarized in Tables Tables44 and and5.5. All 20-μl reaction volumes contained 0.5 U of UNG and 2 μl of specimen. The serovar 1, 10, and 11 PCRs were asymmetric. Every PCR program included a preincubation procedure and the final cooling procedure described above. Touchdown programs were designed for serovars 3, 5, and 8 to ensure specificity. All PCR runs included a positive control that was the designated species or serovar being evaluated and a negative control (water). For species identification PCR assays, external amplification controls (DNA mixture from serovars 3 and 10) were also included in each run. Conditions for urease gene-based traditional PCR have been described previously (4, 10, 19).
DNA concentrations (ng/μl) of specimens purified by the QIAamp kit were measured by NanoDrop Spectrophotometer ND-1000 (NanoDrop, Wilmington, DE) and transformed into molecules/μl, which is equal to genomes/μl, using the calculated molecular mass of U. parvum serovar 3 (ATCC 27815), which is 4.65 × 108 Da, based on a 0.75-Mb genome length, and that of U. urealyticum serovar 10 (ATCC 33699), which is 5.40 × 108 Da, based on a 0.87-Mb genome length. Tenfold serial dilutions of the QIAamp-purified DNA of U. parvum (serovar 3, ATCC 27815) and U. urealyticum (serovar 10, ATCC 33699) were used to determine the analytical sensitivity of the PCR assay in terms of numbers of molecules/μl. Tenfold serial dilutions of broth cultures of U. parvum (serovar 3, ATCC 27815) and U. urealyticum (serovar 10, ATCC 33699) were purified by the proteinase K method and used to determine the analytical sensitivity of the PCR assay in terms of numbers of CFU/μl.
Using genomic DNA of each of 14 ATCC Ureaplasma serovar type strains as the template, our multiplex real-time PCR differentiated U. parvum and U. urealyticum in two different channels without any cross-reactions. All of the serovars within the two respective species were recognized (Fig. (Fig.1).1). Species-specific amplification was successful in the presence of high copy numbers of the other Ureaplasma species (1:100 ratio; data not shown). No cross-reactivity with any of the other microorganisms listed in Table Table11 was detected. The detection limit of the U. parvum assay was 5.2 DNA molecules/μl (2.8 × 10−2 CFU/μl PCR mixture). For U. urealyticum, the limit was 24 DNA molecules/μl (4.1 × 10−2 CFU/μl PCR mixture). These experiments were repeated three times with similar results.
The species-specific multiplex PCR assay was evaluated in two groups of archived clinical specimens/subcultures in order to include direct comparisons with culture and traditional PCR. Among the 132 clinical specimens in group 1, there were 32 (24.2%) that were culture positive for Ureaplasma spp. and 100 (75.8%) that were culture negative. The multiplex species identification PCR detected 52 Ureaplasma-positive samples and identified 80 as negative. Thus, PCR detected 20 (15.2%) more positive samples than did culture. There was only one culture-positive specimen that was negative by real-time PCR, and the number of organisms in that specimen was <10 CFU/ml. Although PCR is generally considered to be more sensitive than culture, using culture as the reference standard, real-time PCR had a sensitivity of 96.9% and a specificity of 79.0% for the detection of Ureaplasma spp.
Among 194 Ureaplasma-positive subcultures from clinical specimens, traditional PCR detected 176 positives and 18 negatives, whereas the multiplex real-time PCR detected 188 positives and 6 negatives. Thus, the multiplex real-time PCR assay reduced false-negative results among culture-positive specimens from 9.3% to 3.1% compared to traditional PCR. Among the 175 subcultures positive by both PCR methods, 116 (66.3%) contained only U. parvum, 54 (30.9%) contained only U. urealyticum, and 5 (2.9%) contained both species by traditional PCR. By multiplex species identification PCR, 115 (65.7%) contained only U. parvum, 48 (27.4%) contained only U. urealyticum, and 12 (6.9%) contained both species. Overall, real-time PCR recognized six more U. parvum isolates and one more U. urealyticum isolate than traditional PCR. Using traditional PCR as the reference standard, real-time PCR had a diagnostic sensitivity of 98.3% for U. parvum and 96.6% for U. urealyticum; while the specificity of detection was 88.9% for U. parvum versus 99.1% for U. urealyticum. The DNA used for this study was originally isolated in 2004, so it is possible that degradation occurred over a 5-year period, which would account for the fact that some culture-positive specimens were PCR negative when retested for the present evaluation.
Real-time PCR assays specific for 14 Ureaplasma serovars were developed for the ATCC type strains, and the specificity of each PCR was tested against the 13 other serovar type strains. Each individual PCR assay amplified the designated serovar but none of the other 13 serovars (Fig. (Fig.2).2). The primer/probe concentration, MgCl2 concentration, annealing temperature, and annealing time of each PCR program were carefully optimized to make the assay specificity possible. For PCRs with primers only (serovars 2, 3, 6, 7, 8, 9, 13, and 14), the primer concentrations were 0.25 to 0.5 μM; for PCRs that required a combination of primers and probes, asymmetric PCR conditions were used (except for serovars 4 and 12). The annealing time of all programs was kept as short as possible, especially for the eight primer-only programs, which could be as short as 1 s (serovars 7 and 9). The annealing temperatures were kept as high as possible to ensure the specificity of each PCR. Touchdown programs containing a short amplification step under stringent conditions prior to the major amplification program were developed for serovars 3, 5, and 8 to improve specificity. After amplification, a melting curve analysis was performed for each PCR providing additional proof of differentiation power. The detection limits of the serovar-specific PCR assays ranged from 15 (serovar 6) to 3,461 (serovar 9) DNA molecules/μl of PCR mixture. No evaluation was performed directly on broth subcultures to determine the limit of CFU detection as was done for the species-specific multiplex PCR assay, and these assays were not tested directly for specificity against other microbial species since their utility is primarily in the testing of isolates or clinical specimens known to contain U. parvum or U. urealyticum.
Taking advantage of the genomic sequence data now available for all 14 Ureaplasma serovars, our multiplex real-time PCR assay for the simultaneous detection and discrimination of the two Ureaplasma species targets two conserved gene sequences, each present only in U. parvum or U. urealyticum. This differs from previously reported assays for Ureaplasma species identification that targeted small differences within genes conserved in both species. The specificity of this assay was ensured computationally and experimentally verified by BLAST analysis and running the assay against a broad range of other microorganisms.
Compared to culture and traditional PCR, our findings using a single-step multiplex assay format agreed with those of Cao et al. (6), who found the real-time PCR to be the most sensitive in the detection of Ureaplasma species. The detection threshold of our assays is well below the level necessary to detect ureaplasmas in clinical specimens, and our multiplex PCR detected Ureaplasma spp. in 15.2% more clinical specimens than did culture in a direct comparison using optimum cultivation methods developed and validated in our laboratory over several years (34). Even though culture is often considered the reference method for the detection of ureaplasmas in clinical specimens, improved PCR assays may be inherently more sensitive and capable of detecting the organisms in very low numbers. Thus, the PCR-positive and culture-negative specimens we encountered are likely to be true positives rather than false negatives. Whether a PCR assay is more sensitive than culture for the detection of ureaplasmas in clinical specimens depends on the PCR target, the assay conditions, and the method of culture used. Numerous studies have shown clearly that PCR assays can be superior to culture (1, 4, 16, 35, 38). However, even though PCR has the added potential advantages of providing Ureaplasma species identification and same-day turnaround without the necessity of maintaining organism viability, culture is relatively simple and can often provide results in 24 to 48 h. Compared to traditional PCR, the new multiplex real-time PCR greatly reduced the false-negative rate in the detection of Ureaplasma species as defined by culture positivity, making this assay even more attractive than traditional PCR in this setting.
Yi et al. (37) developed a real-time PCR assay to simultaneously detect and discriminate between the two Ureaplasma species in a single test with one pair of common primers for both species and two species-specific TaqMan probes. This assay, however, lacked analytical sensitivity, and its clinical sensitivity was lower than that of traditional PCR. In contrast, our multiplex real-time PCR found a total of six more U. parvum strains alone or in combination than with our traditional PCR and one additional U. urealyticum strain. No other published studies have provided a quantitative comparison of species differentiation using real-time PCR versus traditional PCR.
Molecular genotyping methods based on the mba gene and its 5′ end have been explored as a replacement for the antibody-based phenotyping methods (5, 11-15, 18) that have been in use for almost 3 decades since the original description of the 14-member serotyping scheme by Robertson and Stemke in 1982 (21). However, using the limited sequence data available for previous genotyping methods, only partial serovar identification was possible. The four serovars of U. parvum were readily differentiated from each other by traditional PCR with different sets of primers (13) or by two multiplex real-time PCRs (5). However, to distinguish the 10 serovars of U. urealyticum is still challenging. Using single-stranded conformation polymorphism analysis, the 10 serovars were divided into two groups, A (serovars 2, 5, 8, and 9) and B (serovars 4, 7, 10, 11, 12, and 13) (18). By PCR and sequencing, Kong et al. (14) classified the 10 serovars into three subgroups, 1 (serovars 2, 5, 8, and 9), 2 (serovars 4, 10, 12, and 13), and 3 (serovars 7 and 11). The same investigators then provided better discrimination by dividing the 10 serovars into five MBA genotypes with the added individual separation of serovars 9 and 10, i.e., genotypes A (serovars 2, 5, and 8), B (serovar 10), C (serovars 4, 12, and 13), D (serovar 9), and E (serovars 7 and 11) (13). This approach, using just the sequence of an mba gene specific to each serovar, should be reconsidered in light of the whole genome sequences that show that all Ureaplasma serovars encode multiple members of the phase-variable mba gene family.
In designing our serovar-specific real-time PCR assays, we sought to avoid the mba gene family because of its ubiquity within the ureaplasmas and because of its phase variability. This intent was thwarted by the high interserovar identity among the 14 serovars. The average difference among the 4 U. parvum genomes is 0.56%, and among the 10 U. urealyticum genomes, the average is 0.63%. While it is straightforward to design PCRs for the discrimination of any pair of serovars, the discrimination of 1 serovar from the other 13 is often very challenging. We sought to have all assays employ primers and probes; however, in many cases, this was not possible. We were even forced, in several cases, to target mba genes or mba paralogous gene family genes to obtain serovar identification. In those instances, we made sure our PCR targets did not span sites at which chromosome rearrangements took place during phase variation. Those sites could be identified by analyzing whole-genome shotgun sequencing assemblies. By computationally searching thousands of unique loci in each serovar and testing with 14 ATCC type stains, we created PCRs specific to all 14 serovars without any cross-reactions. Many PCRs predicted to be serovar specific failed when tested and were abandoned. Ultimately, each serovar primer/probe set we used contains at least one serovar-specific primer or probe. In the cases where the set could cross-react, leading to a significantly larger amplicon, the PCR conditions were adjusted to prevent it. The strategy to multiplex the 14 PCR assays, which should be more efficient, was not used because each PCR condition was unique and combination with another assay compromised the specificity and sensitivity of one or both assays (data not shown).
In conclusion, our new multiplex real-time PCR for the detection and discrimination of Ureaplasma species is robust and ready to replace the traditional PCR for clinical diagnostic purposes and is a suitable alternative to culture. We have also shown for the first time that all 14 serovar type strains of Ureaplasma spp. can be clearly differentiated from each other by using our novel real-time PCR technology, which overcomes many of the limitations that hampered the utilization of serologically based typing methods. This should greatly facilitate future investigations of ureaplasmas at the species and serovar levels under clinical conditions to assess differential pathogenicity. Our data regarding the monoplex serovar-specific PCR assays were limited to evaluations using individual type strains representing each serovar. We did not evaluate clinical specimens directly, nor have we applied this technology thus far to clinical isolates. The value of this assay as a diagnostic typing method may ultimately depend on the degree of genetic variability that exists among clinical isolates and the extent of horizontal gene transfer that may influence serovar designation. The analytical sensitivities we obtained suggest that our assay should meet the requirements for direct testing of clinical specimens that may contain serovars alone or in combination. Proof of this will require additional clinical studies, as will the evaluation of how well these assays perform with clinical isolates of unknown serovar status.
This work was supported by federal funds under grants RO1A1072577 and RO1209 and contract NO1-AI-30071 from the National Institute of Allergy and Infectious Diseases and grant AI 28279 from the National Institute of Child Health and Human Development.
Technical support by Donna Crabb and Amy Ratliff is gratefully acknowledged.
Published ahead of print on 16 June 2010.