INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The genus Staphylococcus currently includes 38 species (13), and coagulase-negative staphylococci (CNS) are isolated frequently in clinical microbiology laboratories (11, 12, 20). CNS are associated with the normal skin flora and mucous membranes and can be isolated from many other sources (including foodstuffs) such as meat, milk and cheese, soil, sand, water, and air (12). CNS may cause bacteremia, endocarditis, catheter-related infections, central nervous system shunt infections, urinary tract infections, endophthalmitis, and infections of prosthetic joints (9). CNS give rise to significant hospital infections often associated with the use of medical devices and immunocompromised patients (13).

With increasing numbers of CNS infections being recognized, it has become necessary to have fast and reliable identification methods (2, 11). Biochemical methods exist (2, 4, 10, 18), but none can reliably identify the important CNS because of the variable expression of some phenotypic characters. Multilocus enzyme electrophoresis or analysis of cellular fatty acid composition have also been used, but identification remains incomplete (21). Genotypic methods for identification have been developed (5, 6,7, 15, 16, 23) but also have limitations. For example, the individual PCRs designed by Gribaldo et al. (6) to identify eight species of CNS are time-consuming and expensive.

Real-time PCR and biprobe (3) assays have recently been used to distinguish related bacterial species of, for example, the genus Campylobacter (14). The aim of the present study was to establish whether real-time PCR was suitable for the identification of CNS involved in human infection.

The LightCycler (Bio/Gene Ltd., Kimbolton, England) is a real-time PCR machine that allows both rapid PCR cycling and continuous monitoring of product formation (24). The formation of double-stranded PCR products is detected by Sybr Green I. Sybr Green I is an intercalating dye which fluoresces strongly when bound to double-stranded DNA; thus, when PCR products are formed an increase in fluorescence is observed (8, 19). After PCR amplification the LightCycler can monitor melting of the DNA with increasing temperature by measuring the decrease in fluorescence as Sybr Green I is released. For convenience, the negative derivative of fluorescence versus temperature is plotted to give a discrete melting peak. When the melting temperature of the PCR products is analyzed in this way, it is not necessary to visualize the PCR products on agarose gels.

Sequence-specific detection of the amplicons can be achieved by the addition of a short-labeled probe. When a probe is included in the reaction, after the melt cycle two peaks can be observed. One corresponds to a decrease in fluorescence due to the melting of the PCR product, and the other is due to the release of the probe. Biprobes are sequence-specific probes labeled with the fluorophore Cy5. When the probe binds to the complementary sequence in the PCR product, the Cy5 label is excited by the energy transfer from Sybr Green I, resulting in an increase of light emitted by Cy5. This fluorescence is measured at a different wavelength from that emitted by Sybr Green I and so can be distinguished from it. Biprobes have another useful feature: as well as hybridizing to a perfectly matched sequence, they will also bind when there is some mismatch (usually up to five mismatches). When the biprobe binds to a mismatched sequence, a melting temperature lower than that of a perfectly matched sequence can be distinguished.

We designed different biprobes to bind to variable regions of the 16S rRNA gene of staphylococci with various degrees of mismatch. We have applied one to four probes to amplicons from 15 species of staphylococci and confirmed melting peaks for a collection of 24 type strains. Further, the identification results of 142 clinical isolates determined by using the biprobes were compared with the results obtained by biochemical and PCR identification.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, culture conditions, phenotypic tests, and PCR. Type strains used in the study are listed in Table 1 and were obtained from the National Collection of Type Cultures (NCTC) (Central Public Health Laboratory, London, United Kingdom). The 142 clinical isolates were selected from those submitted for identification to the Laboratory of Hospital Infection, Central Public Health Laboratory. All strains were grown overnight on blood agar plates at 37°C. Colony suspensions of the strains were made by suspending several colonies from one plate in 1 ml of water, heating to 99°C for 5 min, and centrifuging at 3,000 × g for 2 min. The supernatant was then used in LightCycler assays. Phenotypic identification of strains was made using ID 32 Staph (bioMérieux) according to the manufacturer's instructions (4). PCR identification was made using the primers and conditions of Gribaldo et al. (6).

DNA extraction. DNA was extracted from the type strains using the Wizard Genomic DNA purification kit (Promega, Southampton, United Kingdom) according to the manufacturer's instructions, except that lysis was achieved by resuspending cells in 100 µl of lysostaphin (1 mg/ml in Tris-EDTA [TE] buffer; Sigma) and 50 µl of lysozyme (50 mg/ml in TE buffer; Sigma).

LightCycler primers and biprobes. The sequences of all primers and biprobes are shown in Table 2. All biprobes (3) were labeled with Cy5 at the 5' end and with biotin at the 3' end to prevent them from acting as primers. All primers and probes were synthesized by MWG-Biotech UK Ltd. (Milton Keynes, England). CNS Probe A was used with primers STAR1 (5 pmol/µl) and STAF1 (1 pmol/µl), CNS Probe B was used with STAR1 (1 pmol/µl) and STAF2 (5 pmol/µl), CNS Probe C was used with STBF1 (5 pmol/µl) and STBR1 (1 pmol/µl), and CNS Probe D was used with primers STBF1 (1 pmol/µl) and STBR1 (5 pmol/µl).

LightCycler assays. The LightCycler (Bio/Gene Ltd.) was used for all biprobe assays. PCR mixtures (10 µl) contained 1 µl of colony suspension or 10 ng of DNA, 500 mM Tris-HCl (pH 8.3), MgCl₂ (5 mM), bovine serum albumin (1 mg/ml), deoxynucleoside triphosphates (200 µM concentrations of each; Gibco BRL), forward and reverse primers (one at 5 pmol/µl and the other at 1 pmol/µl), platinum Taq polymerase (0.4 U; Gibco BRL), Sybr Green I (Bio/Gene Ltd.) diluted 0.01%, and a biprobe (5 pmol/µl). The cycling parameters were one denaturation cycle at 94°C for 5 s and 40 amplification cycles (temperature transition rate of 20°C/s) of 94°C for 0 s, 55°C for 1 s, 60°C for 15 s, and 74°C for 10 s. Fluorescence readings were taken after annealing at 55°C for 1 s. PCR cycling was followed by melting curve analysis of 40 to 95°C (temperature transition rate of 0.2°C/s) with continuous fluorescence readings. For ease of interpretation, controls for all of the peaks were run alongside the samples being tested. For example, with CNS Probe A, three controls were run, one for each of the three peaks. This allowed easy scoring of the results and corrected for slight variations in the melting temperatures which were observed from run to run.

DNA sequencing. The primers Epsilon F (AAGAGTTTGATCCTGGCTCAG) and 1510R (GGTTACCTTGTTACGACTT) were used in a PCR to amplify a 1,500-bp amplicon of the 16S rRNA gene. The PCR mixtures (100 µl) contained forward and reverse primers (20 pmol each), deoxynucleoside triphosphates (200 µM concentrations of each), platinum Taq polymerase (0.5 U), MgCl₂ (3 mM), 1× PCR buffer (as supplied with platinum Taq polymerase), and 5 µl of a colony suspension. The PCR was performed in a Perkin-Elmer 9600 machine, and cycling consisted of 1 cycle of 95°C for 5 min, 30 cycles of 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min followed by final extension at 72°C for 5 min. The PCR products were purified using the Wizard PCR Clean up kit (Promega) and quantified by agarose gel electrophoresis. The PCR products were sequenced according to the manufacturer's instructions for the ABI prism sequencing kit (Perkin-Elmer) using the primers Epsilon F and 320R (TTGACCGTGTCTCAGTTCCA).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Biprobe and primer design. The 16S rRNA gene sequences of 35 species of Staphylococcus, Stomatococcus mucilaginosus, and Micrococcus luteus were retrieved from the GenBank database. The sequences were aligned using the Lasergene Dnastar computer package. Two variable regions between bp 1 and 250 and between bp 970 and 1080 were identified as being suitable for the design of biprobes. Four biprobes were designed, CNS Probe A and CNS Probe B from the region between bp 1 and 250 and CNS Probe C and CNS Probe D from the region between bp 970 and 1080. CNS Probe A was designed to be a perfect match to Staphylococcus epidermidis, CNS Probe B was a perfect match to Staphylococcus lugdunensis, CNS Probe C had no mismatches with Staphylococcus warneri, Staphylococcus sciuri, S. lugdunensis, Staphylococcus intermedius, and Staphylococcus schleiferi and CNS Probe D was a perfect match with Staphylococcus capitis. All other species had one or more mismatches with the probes. It was predicted that for S. epidermidis it would be necessary to use only one probe (CNS Probe A) to make an accurate identification. Since S. epidermidis is the most commonly found CNS in nosocomial infections the use of only one probe would allow a fast and inexpensive identification. The probes were all designed to have an annealing temperature at least 5°C higher than that of the primers and were also predicted to have minimum secondary structure that might prevent them from binding to the target sequence.

Biprobe evaluation. The four CNS biprobes were tested on DNA extracted from the 15 type strains of staphylococcal species of clinical interest. During the cycling step the temperature was held at 55°C to allow the primers to anneal. It was also found that for the biprobe to anneal and hence fluoresce it was also necessary to hold the temperature at 60°C for 15 seconds. In addition, it was necessary to reduce the concentration of the primer closest to the probe; otherwise, the primer prevented the probe from binding.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Biprobe peak profiles. (A) Peak profiles obtained with biprobe A. Peak 0 melted at 58°C, peak 1 melted at 52°C, and peak 2 melted at 50°C. (B) Peak profiles obtained with biprobe B. Peak 0 melted at 64°C, peak 1 melted at 60°C, peak 2 melted at 56°C, peak 3 melted at 58°C, peak 4 melted at 53°C, and peak 5 melted at 52°C. (C) Peak profiles obtained with biprobe C. Peak 0 melted at 68°C, peak 1 melted at 60°C, peak 2 melted at 56°C, and peak 3 melted at 54°C. (D) Peak profiles obtained with biprobe D. Peak 0 melted at 60°C, and peak 1 melted at 58°C.

Development of an identification scheme. The four biprobes were tested on 15 Staphylococcus species of interest, namely, S. epidermidis, S. warneri, S. lugdunensis, S. aureus, S. intermedius, S. chromogenes, S. sciuri, S. hominis, S. haemolyticus, S. capitis, S. caprae, S. schleiferi, S. simulans, S. saprophyticus, and S. cohnii. The results are shown in Table 3. Based on these results it was concluded that CNS Probe A identified S. epidermidis and S. warneri, and therefore it was not necessary to use any of the other three probes for identification. CNS Probe B used alone can identify S. lugdunensis and S. haemolyticus, and the combination of CNS Probe A and CNS Probe B allows identification of S. sciuri, S. lugdunensis, S. aureus, S. hominis, S. haemolyticus, and S. cohnii. CNS Probe C in combination with Probes A and B can be used to identify S. intermedius, S. chromogenes, S. schleiferi, and S. saprophyticus. The final probe, CNS Probe D, was designed to distinguish S. capitis and S. caprae.

Clinical isolate testing. A total of 142 clinical isolates of staphylococci were tested with all four CNS probes to determine whether reactions observed with the type strains were consistent. Of the 142, 100 were tested as known isolates and 42 were tested as part of a blind study. Where possible, at least 10 isolates of each species were selected, but for six species only a few clinical isolates were available, and for S. caprae none were obtainable. The results are shown in Table 4. All of the S. hominis, S. sciuri, S. cohnii, S. intermedius, and S. simulans strains showed complete agreement between biochemical identification and LightCycler identification. Four additional S. epidermidis isolates, one additional S. aureus isolate, and one additional S. saprophyticus isolate that were not identified by biochemical tests were identified by real-time PCR.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have described a rapid and accurate test for identification of the main species of CNS of clinical interest, as well as S. aureus, S. intermedius, and S. schleiferi, which are or can be coagulase-positive species. It is becoming increasingly important that reference laboratories identify CNS accurately, particularly S. epidermidis (11).

Of the 142 clinical isolates tested by the biprobe assays, 125 showed concordance between the biprobe assay results and those of the traditional PCR and phenotypic identification method. The 12 strains that did not correlate were further characterized by partial sequencing of the 16S rRNA gene, and these results, except in the case of the S. capitis strains, confirmed the biprobe assay result. There were five strains of S. capitis that were identified by the biprobe assays as S. caprae, but further examination of the 16S rRNA gene sequences indicated that it was not possible to distinguish S. caprae from S. capitis on the basis of 16S rRNA sequence. This has been reported also by Takahashi et al. (22), who compared 16S rRNA gene sequences of 38 species of staphylococci, and by Bannerman and Kloos (1), who carried out DNA-DNA hybridization studies. It may be more convenient to report these species as "S. capitis or S. caprae" or to use an alternative test to identify them.

Several methods are currently being used or have been evaluated to identify CNS. The most common is the use of commercially available biochemical tests that allow phenotypic identification, but these are not always reliable and results are often operator dependent (17). For example, phosphatase-negative variants of S. epidermidis can often be misidentified as S. hominis (7, 13). The introduction of molecular identification of CNS has improved the accuracy of results and allows the identification of the more unusual CNS which occasionally cause infection (6). Species-specific PCRs can be used to replace the phenotypic tests or be used in conjunction with them, but this can require the use of several primer sets, which is time-consuming and expensive. Furthermore, primers are not available for all of the species of CNS that can be implicated in infection. Several genotypic tests are available for the identification of S. epidermidis only (7, 16, 23), but with increasing numbers of the other CNS being implicated in infection it is necessary to have available a fuller range of identification tests.

The real-time PCR and biprobe approach to identifying staphylococci to the species level has several advantages. LightCycler tests can be completed in 30 to 40 min, and it is not necessary to open the tubes after amplification. The risk of contamination is therefore reduced. The assay requires easily prepared colony suspensions, and it is not necessary to perform time-consuming DNA extractions. Only three PCRs need to be performed to characterize up to 15 species of CNS. The more widespread use of these assays is constrained by the initial expense of the LightCycler, and only 32 samples can be included in one run compared to 96 samples in most conventional PCR systems. With future generations of real-time PCR machines having the capacity for additional fluorescence channels, there will be the opportunity to include more than one probe in each reaction, thus reducing the number of PCRs required.

The biprobe assays proved to be accurate and robust, and by using a series of control strains in each experiment the assays are easy to interpret. The rapid identification of S. epidermidis using only one PCR and only one probe will significantly improve the turnaround time for identification to the species level of the most common CNS.