|Home | About | Journals | Submit | Contact Us | Français|
Rapid, high-throughput screening tools are needed to contain the spread of hospital-acquired methicillin (meticillin)-resistant Staphylococcus aureus (MRSA) strains. Most techniques used in current clinical practice still require time-consuming culture for primary isolation of the microbe. We present a new phenotypic assay for MRSA screening. The technique employs a two-photon excited fluorescence (TPX) detection technology with S. aureus-specific antibodies that allows the online monitoring of bacterial growth in a single separation-free process. Different progressions of fluorescence signals are recorded for methicillin-susceptible and -resistant strains when the growth of S. aureus is monitored in the presence of cefoxitin. The performance of the new technique was evaluated with 20 MRSA strains, 6 methicillin-susceptible S. aureus strains, and 7 coagulase-negative staphylococcal strains and two different monoclonal S. aureus-specific antibodies. When either of these antibodies was used, the sensitivity and the specificity of the TPX assay were 100%. All strains were correctly classified within 8 to 12 h, and up to 70 samples were simultaneously analyzed on a single 96-well microtiter plate. As a phenotypic method, the TPX assay is suited for screening purposes. The final definition of methicillin resistance in any S. aureus strain should be based on the presence of the mecA gene. The main benefit afforded by the initial use of the TPX methodology lies in its low cost and applicability to high-throughput analysis.
Methicillin (meticillin)-resistant Staphylococcus aureus (MRSA) is widely accepted to be the most significant multiresistant human pathogen. Strict infection control policies have been introduced, e.g., in Finland to contain the spread of hospital-acquired MRSA strains (15). The adopted policy requires reliable, high-throughput screening tools, as thousands of potential carriers must be tested during outbreak situations.
The traditional screening cultures require at least 48 h until a negative test result for MRSA can be confirmed. These methods have been followed by introduction of faster automated techniques and an agglutination test for the specific detection of a resistance-associated protein, PBP 2a (3). However, most phenotypic techniques still require 18 to 24 h of culturing for primary isolation of the microbe before the identification of Staphylococcus aureus and testing for methicillin resistance can be initiated. As the first example of a rapid nonmolecular MRSA screening test, the BacLite Rapid MRSA assay (Acolyte Biomedica, United Kingdom) is a sensitive (90.4%) and specific (95.7%) test for the detection of ciprofloxacin-resistant MRSA nasal colonization within 5 h (12). The MRSA isolates are first enriched in a selective broth and then extracted by using paramagnetic microparticles coated with monoclonal anti-S. aureus antibodies. Capture and washing are followed by a lysis step to release intracellular adenylate kinase. Finally, the presence of MRSA is confirmed by elevated adenylate kinase activity.
The detection of the mecA gene by PCR is widely recognized as the “gold standard” method for the detection of MRSA (2). For epidemiological screening for MRSA, specimens such as nasal samples often contain coagulase-negative staphylococci (CoNS), which can also carry the mecA gene (1). Therefore, molecular detection of mecA alone cannot be applied for the direct detection of MRSA from such samples, and a traditional culture for the isolation of S. aureus is still needed. A real-time PCR assay (e.g., the IDI-MRSA kit assay [GeneOhm Sciences, San Diego, CA]) allows the detection of MRSA directly from a nasal swab in less than 1 h (10). The assay is based on multiple primers used in combination with several molecular beacon probes. Nevertheless, most diagnostic laboratories are still applying traditional phenotypic methods for high-volume analysis for MRSA due to the high cost of molecular assays and the complexity of sequencing (5, 22).
A two-photon excited fluorescence (TPX) detection technology (ArcDia, Turku, Finland) for use for highly sensitive separation-free bioaffinity assays was originally introduced by Hänninen et al. (8). The instrumentation of an automated TPX plate reader and the physical basis of the detection technique have been presented in detail by Soini et al. (18). The TPX technology is currently applied for the detection of a wide array of analytes, such as the C-reactive protein (21) and oligonucleotides (19). In the present study, we describe the application of a TPX assay for MRSA screening. The assay principle is based on a phenotypic approach: the bacteria in a sample are allowed to grow in the presence of a penicillinase-resistant β-lactam antibiotic while monitoring for an S. aureus-specific fluorescence signal is being performed online by the separation-free detection technique in a single-step procedure. Different progressions of fluorescence signals are recorded for susceptible and resistant S. aureus strains. Descriptions of the earlier stages of development of the assay and a more detailed biochemical basis of the assay have been described elsewhere (14).
In brief, the assay principle is as follows (14). A sample is diluted into a growth medium and incubated with the assay reagents, including polystyrene microparticles coated with capture antibodies, a fluorescently labeled tracer, and a wide-spectrum β-lactam antimicrobial agent. The same S. aureus-specific antibody is used in both the capture and the tracer roles. The temperature of the reaction mixture is held high enough to allow exponential growth of the microbes resistant to the antimicrobial agent used as the selective agent (in this case, cefoxitin at 4 mg/liter). Methicillin-susceptible S. aureus (MSSA) and methicillin-susceptible CoNS are inhibited by the β-lactam while MRSA and methicillin-resistant CoNS are not. Tracer molecules are bound on the microparticle surfaces only via S. aureus antigens; CoNS antigens are not recognized. As a result, the number of fluorescent molecules bound on microparticle surfaces rapidly increases with time if viable MRSA cells are present in the sample. The unique optical setup of a TPX microfluorometer enables the direct measurement of fluorescence from the turbid and strongly scattering reaction mixture without the performance of any separation steps (9). The fluorescence signal from a reaction mixture can be measured repeatedly or can even be monitored continuously, and the growth of the sample bacteria is not affected. A rapid increase in the signal with time confirms a positive MRSA screening test result.
All bacterial strains used in this study are listed in Table Table1.1. They consisted of 11 S. aureus reference strains (6 MSSA and 5 MRSA strains), 7 CoNS reference strains, and 15 defined epidemic MRSA strains. Included were S. aureus strains ATCC 25923 (MSSA), ATCC 29213 (MSSA), and ATCC 43300 (MRSA), as recommended by the Clinical and Laboratory Standards Institute, for use as reference strains in oxacillin susceptibility testing (4). Enterococcus faecalis ATCC 29212 was used as a negative control.
Autoclaved tryptic soy broth (TSB; CM0129; Oxoid, Basingstoke, Hampshire, United Kingdom) was supplemented with polyethylene glycol 6000 (PEG; Fluka, Buchs, Switzerland) to a final concentration of 35 mg/ml to provide TSB-PEG growth medium. PEG facilitates the formation of immunocomplexes and improves the performance of the assay (17). The dry-chemistry assay buffer contained 10 mM Tris (Ultror grade; Calbiochem, La Jolla, CA), 50 mM NaCl (Riedel-de Haën, Seelze, Germany), 0.5% bovine serum albumin (fraction V; Sigma-Aldrich, Steinheim, Germany), and 5% d-sorbitol (Fluka), pH 7.9. The growth medium, water (Rios3 grade; Millipore, Billerica, MA), and other solutions were filtered through 0.2-μm-pore-size filters (GHP Acrodisc; Pall Gelman Laboratory, Ann Arbor MI). Three different anti-S. aureus antibodies (two monoclonal antibodies and one polyclonal antibody) were used. The monoclonal anti-S. aureus antibodies (immunoglobulin G3 isotype [IgG3]; catalog nos. BM3066X and AM01227PU-N) were purchased from Acris Antibodies (Hiddenhausen, Germany). The polyclonal anti-S. aureus antibody (catalog no. ab20920) was obtained from Abcam (Cambridge, United Kingdom). All antibodies were provided in phosphate-buffered saline buffer with sodium azide as a preservative. Carboxyl-modified microparticles (diameter, 3 μm; high acid content, hydrophilic; part 7300-3420) made of cross-linked polystyrene were purchased from Seradyn (Indianapolis, IN). Arctic Diagnostics Ltd. (Turku, Finland) provided the fluorescent labeling reagent dipyrrylmethene-BF2 530 (BF530) (18). Cefoxitin, sodium salt (949 μg/mg; catalog no. C4786), and lysozyme (catalog no. L6876) were purchased from Sigma-Aldrich (Steinheim, Germany). Ninety-six-well microtitration plates with black walls and a μClear bottom (medium binding, half-area, 200 μl/well; catalog no. 675096) were obtained from Greiner Bio-One (Frickenhausen, Germany). Plate sealing films (catalog no. 4ti-0510) were purchased from 4titude (Surrey, United Kingdom), and AeraSeal highly gas-permeable sealing films (catalog no. 229307) was purchased from Porvair Sciences (Middlesex, United Kingdom).
Polystyrene microparticles were coated with antibodies by a procedure described by Waris et al. (21). Solid-phase antibodies were first passively adsorbed on the microparticle surfaces and then bound covalently by the carbodiimide coupling reaction. The tracer antibodies were labeled with a two-photon excitable fluorescent labeling reagent by methods described earlier by Meltola et al. (16). The coated microparticles and labeled antibodies were stored at +6°C.
To evaluate the effect of bacterial lysis on the assay performance, 5 μl of lysozyme solution (10 mg/ml of sterile water) was added to a sample volume of 300 μl, and the mixture was then incubated for 15 min at +37°C.
Dry-chemistry reagents were used to enable simplified assay protocols and a longer reagent shelf-life (13, 14). Cocktails of the assay reagents containing four times the intended final concentrations of coated microparticles, fluorescent tracer, and cefoxitin were prepared in dry-chemistry buffer. The cocktails were dispensed either manually or with an automated dispenser (dispenser automate MiniPrep 60; Tecan Systems Inc.) into the wells of a 96-well plate (25 μl/well). The reagent wells were evaporated to dryness in a desiccator over silica gel (22°C, overnight incubation). The plates were then sealed with gas-impermeable PCR tape and were stored for up to a month at +6°C.
A bacterial strain was cultured aerobically at +37°C on a blood agar plate for approximately 24 h before it was tested. After the incubation, two colonies from the plate were inoculated into 1 ml of TSB-PEG growth medium. Low-density samples for the MRSA screening tests and for assay optimization purposes were prepared by diluting the solution (1 to 1,000) in TSB-PEG medium. Samples for assay performance testing were prepared by serially diluting the solution (1 to 4) in TSB-PEG medium. Plain TSB-PEG medium was used as a zero control for testing.
Prediluted samples were dispensed in the wells of a dry-chemistry 96-well plate (100 μl/well). The plate was then resealed with a plate sealing film and incubated at room temperature for about 30 min to allow dissolution of the dried reagents into the TSB-PEG medium. Fluorescence signals from the resulting reaction mixtures were measured with an automated ArcDia TPX plate reader (PR-6 or PR-10; Arctic Diagnostics). Measurements were done in a fixed preset order and were restarted immediately after the completion of a cycle. As a result, measurements were made for every assay well at regular intervals over a period of up to 20 h. Unless otherwise stated, a measurement time of 3 min per well was used. The plate was automatically shaken (10 s of mechanical pulsing) every 10 min to prevent the settling of microparticles. The temperature of a plate remained relatively constant at +32°C inside an operating reader.
Hundreds of distinct fluorescence bursts (typical duration, 50 ms) were recorded from a single well during the 3-min measurement times; each of these bursts represents the fluorescence from an individual microparticle. The burst data were reduced to a fluorescence signal by use of the robust mean signal estimation algorithm described by Glotsos et al. (7). The R package (version 2.2.1.) was used for statistical computing of these signals. The more bursts that are available for statistical analysis, the more reproducible that the fluorescence signal provided by the algorithm is. The fluorescence signal from each assay well was measured and updated at regular time intervals. The increase in the fluorescence signal rate (ΔF/Δt) was defined as the difference of two consecutive fluorescence signals (ΔF) divided by the time interval (Δt) between the measurements, i.e., an approximation of a tangent to the fluorescence-versus-time curve for an assay well.
The Mann-Whitney U test was used to evaluate the difference between MRSA and non-MRSA samples with respect to the maximal ΔF/Δt values. P values of less than 0.05 were considered statistically significant.
The assay was optimized separately for each microparticle-tracer reagent pair prepared with different antibodies. A lower microparticle concentration theoretically improves the sensitivity of the assay but also reduces the number of fluorescence bursts available for statistical analysis (9). A microparticle concentration that allowed approximately 100 microparticles to be found during a 3-min measurement time was considered optimal. These concentrations were determined experimentally by measuring plain TSB-PEG medium to which various amounts (e.g., 125, 250, and 500 microparticles per μl) of microparticles were added. The predetermined optimal microparticle concentration was used for tracer concentration optimization, in which tracers at eight different concentrations (0.125 to 16 nM) were tested for their ability to detect MSSA ATCC 12600 whole cells. A tracer concentration that provided the highest ratio of the fluorescence signal to the fluorescence signal for the zero control (the signal-to-background ratio [SBR]) was considered optimal. The predetermined optimal microparticle and tracer concentrations were used in the following analyses.
The performance of the assay for the detection of S. aureus whole cells was determined separately for each microparticle-tracer reagent pair. The fluorescence signals from eight sample dilutions containing MSSA ATCC 12600 cells and a negative control were measured once to produce an assay dose-response curve, a fluorescence-bacterial cell count [F(c)] curve, with respect to the cell count (determined by the standard plate count method). It was assumed that no significant growth took place in the assay wells during the measurement. Figure Figure11 presents such a curve when reagents prepared with the monoclonal BM3066X antibody were used. In general, the dose-response curves of all reagent-limited assay methods were characterized by a flat-line response at analyte concentrations below the detection limit, a nearly linear portion within the dynamic range of the assay, and a hook effect with excess analyte concentrations. The SBR, the lowest cell count that was differentiable from the zero control (the limit of detection of the assay), and the cell count that produced the maximal fluorescence signal (the upper limit of the dynamic range of the assay) are given in Table Table22 for both reagent pairs. The best assay performance with the lowest limit of detection, the widest dynamic range, and the highest SBR was clearly achieved with reagents prepared with the monoclonal BM3066X antibody.
The effects of cell lysis on the performance of the TPX assay were evaluated by comparison of the dose-response curves for whole cells of an MSSA reference strain (ATCC 12600) and lysozyme-treated cells (Fig. (Fig.1).1). The procedure described above for determination of the assay dose-response curve, the F(c) curve, with respect to the cell count was used both with and without lysozyme treatment of the samples. A significantly improved SBR and assay limit of detection were seen after the lysozyme treatment. Reagents prepared with the monoclonal BM3066X antibody were used in these experiments.
Repeated on-line measurements allow fluorescence-versus-time [F(t)] curves to be simultaneously recorded from multiple assay wells. As a result of exponential growth, an F(t) curve sharing the same characteristics with the assay F(c) curve was recorded if any viable S. aureus cells were present in the assay well. The lower that the amount of bacteria initially present in a sample was, the longer that it took the assay detection limit to be reached and a fluorescence signal differentiable from that for the zero control was recorded. Figure Figure22 shows F(t) curves recorded for a period of 6 h from four assay wells initially containing various amounts (2,400 to 1.6 × 105 CFU/ml) of cells of an MRSA reference strain (ATCC 43300). Reagents prepared with the monoclonal BM3066X antibody were used in the experiment.
The sensitivities and specificities of the TPX assay for the screening of the 33 study strains (Table (Table1)1) for MRSA were analyzed separately for each microparticle-tracer reagent pair (Table (Table2).2). F(t) curves were recorded for a period of 20 h for assay wells containing samples with low bacterial densities (initial concentration, about 104 to 105 CFU/ml, determined by the plate count method) and cefoxitin at a concentration of 4 mg/liter. Up to 70 samples were simultaneously analyzed in a single test run. The results from the testing of the 20 MRSA strains with reagents prepared with the monoclonal BM3066X antibody are presented as separate F(t) curves in Fig. Fig.3,3, and the corresponding results for the 13 MSSA and CoNS strains are presented in Fig. Fig.4.4. All MRSA strains were clearly differentiated from the non-MRSA strains by a higher maximal increase in the fluorescence signal (ΔF/Δt) rates. Figure Figure55 shows the maximal increase in the fluorescence signal rates for the MRSA strains and the non-MRSA strains obtained with reagents prepared with the monoclonal BM3066X antibody. The average maximal ΔF/Δt rate for the MRSA-negative samples + 3 standard deviations was considered to be a significant increase in the fluorescence signal rate and to confirm a positive screening test result. All samples containing MRSA but none of those not containing MRSA had ΔF/Δt values above the level needed to achieve a significant difference (P < 0.001). Thus, both the sensitivity and the specificity of the TPX assay for screening for MRSA were 100%. Very similar results were obtained when these experiments were duplicated with a highly gas-permeable plate-sealing tape instead of regular tape (data not shown). When reagents were prepared with the monoclonal AM01227PU-N antibody, the sensitivity and the specificity of the screening assay were also 100%. When reagents prepared with polyclonal anti-S. aureus antibody ab20920 were used, the sensitivity of the assay for screening for MRSA was only 70% (results not shown). The measurement time required to confirm that all detectable MRSA strains were positive (ΔF/Δt value > 2.5 × fluorescence background/h) varied from 8 to 12 h (Table (Table22).
Since MSSA and CoNS are commonly present in clinical samples being screened for MRSA, the adequacy of the performance of this method under such conditions was tested with specimens containing different concentrations of MSSA and CoNS, in addition to MRSA. Reagents prepared with the monoclonal BM3066X antibody were used in the experiment. A sample with a low bacterial (MRSA DSM 46320) density (3,000 CFU/ml) was supplemented with various concentrations of MSSA ATCC 29213 (3,200 to 4 × 105 CFU/ml) and CoNS ATCC 14990 (1,900 to 2.4 × 105 CFU/ml). A 133-fold excess of MSSA and an 80-fold excess of CoNS caused an initially high fluorescence signal, which indicated an inconclusive screening test result. The lower MSSA and CoNS concentrations (27-fold or lower excess for MSSA, 16-fold or lower excess for CoNS) had no significant effect on the assay performance.
We evaluated and present here the results of a new culture-based assay for phenotypic MRSA screening. The new method employs the TPX technology to provide a quantitative S. aureus-specific fluorescence signal in a single separation-free process. Different fluorescence signal progressions are recorded for MRSA and MSSA when bacterial growth under conditions of antibiotic pressure is monitored online by continuous or repeated measurements by use of the TPX technology. In the future, this new method might be useful, e.g., for MRSA screening directly from clinical samples without the use of any time-consuming isolation steps. It is of note that as a phenotypic method, the TPX assay is suited for screening purposes. The final definition of methicillin resistance in any S. aureus strain must be based on the presence of the mecA gene. The main benefit afforded by the initial use of the TPX assay lies in its high-throughput capacity and low cost per analysis. It is conceivable that the low costs can be further reduced if smaller amounts of reagents were consumed.
An optimal antibody for use in the TPX screening assay should be highly sensitive for S. aureus but show only minimal cross-reactivity with CoNS. When the two monoclonal antibodies were used, all of the 20 defined MRSA strains were correctly recognized, while none of the MSSA or CoNS strains were falsely identified as MRSA, indicating a 100% sensitivity and a 100% specificity of the methodology. Considerably better assay performance was achieved with reagents prepared with the BM3066X antibody (Table (Table2).2). Although they were carefully selected, our set of 26 S. aureus strains is still limited. Thus, it remains plausible that S. aureus strains not recognized by either of these two monoclonal antibodies might be found in wider-scale testing. A combination of several different monoclonal antibodies could be used to improve the assay performance in case a sufficient sensitivity is not achieved with reagents prepared with any single antibody. Such assays are readily supported by TPX plate readers (18). In the future development of the assay, we would prefer to use better-characterized high-affinity monoclonal antibodies, such as those used in new rapid identification assays, which have proven highly sensitive and specific (e.g., 98.2% and 98.9%, respectively, for the Slidex Staph Plus assay [bioMerieux, Marcy-l'Etoile, France]) for the identification of S. aureus (20). Unfortunately, these antibodies are presently commercially available only as part of diagnostic kits. When reagents prepared with the polyclonal anti-S. aureus antibody were used, the sensitivity of the TPX assay was only 70%, rendering it inadequate for clinical application. The poor sensitivity of the assay was evidently due to the low affinity of the polyclonal antibody. The situation is analogous to the shortcomings observed in association with the first-generation slide agglutination tests for the rapid identification of S. aureus. Many MRSA strains produce agglutination in these tests, which detect only bound coagulase and protein A (20). The agglutination-negative strains possess a polysaccharide capsule which may physically mask cell surface structures.
The decision to use cefoxitin as the antimicrobial agent in this assay was provoked by the results by Felten et al. (6), who showed that cefoxitin might be superior to oxacillin as a selective agent for the phenotypic testing of MRSA. As a wide-spectrum β-lactam agent, cefoxitin inhibits the growth of most microbes potentially present in colonization cultures, but the growth of MRSA or methicillin-resistant CoNS is not affected. β-Lactam antibiotics, such as cefoxitin and oxacillin, induce the autolysis of dividing susceptible bacteria and release a mass of intracellular antigens into the medium. These antigens are presumably recognized by the immunoreagents, and a slow rise of the fluorescence signal is seen with time, even though the live bacterial population is decreasing (Fig. (Fig.4).4). This hypothesis is supported by the effects of lysozyme treatment on the assay performance: lysozyme-induced lysis of bacteria significantly increased the fluorescence SBR and lowered the assay limit of detection (Fig. (Fig.1).1). Both of the monoclonal antibodies used here had a slight cross-reactivity with CoNS, which was shown by a slow rise in the fluorescence signal with time, indicating that some MRSA strains cannot be reliably differentiated from MSSA or CoNS by the absolute fluorescence signals alone. Instead, a steep rise in the fluorescence signal at some point during a measurement appears to characterize MRSA. All of the 20 MRSA strains were clearly differentiated from the non-MRSA strains by higher maximal increases in the fluorescence signal (the ΔF/Δt value) when assay reagents prepared with the monoclonal antibodies were used (Fig. (Fig.55).
An initially high fluorescence signal is seen if the amount of S. aureus present in a sample is already at the start close to or above the upper limit of the assay dynamic range (Fig. (Fig.2).2). In such a case, a low maximal increase in the fluorescence signal rate is recorded whether the bacteria are methicillin resistant or not. A false-positive result can be avoided by keeping in mind that an initially high fluorescence signal is a clear indication of an inconclusive screening test result. Thus, further dilution is required. On the other hand, overdilution of a sample yields an initial amount of MRSA far below the assay detection limit. A longer measurement time is then required before a significant rise in the fluorescence signal is seen and a positive screening test result can be confirmed (Fig. (Fig.2).2). Furthermore, the higher that the assay limit of detection is, the longer that the measurement time is. Alternatively, several duplicates of the sample at different dilutions could be simultaneously run on the same plate to avoid both inconclusive results and extended measurement times due to overdilution.
In this initial state, many factors associated with the potential application of this new concept in the clinical setting remain undefined. For example, we merely relied on overnight cultures, but it is not known whether bacteria just colonizing the host are that metabolically active. Neither is it known whether the inoculum that we used as a starting point fits a clinical situation. Furthermore, since the results reported here rely only on pure cultures, the influence of host flora other than MSSA or CoNS as competing microbes must be evaluated. Under the experimental conditions described here, specimens with comingled bacteria (MRSA and MSSA or CoNS) that closely mimic specimens tested clinically did not jeopardize the assay performance, although the presence of very high excesses of susceptible staphylococci may lead to inconclusive test results.
In conclusion, a new separation-free phenotypic assay for MRSA screening is described. When monoclonal antibodies were used, the assay was 100% sensitive and 100% specific for screening for MRSA from pure cultured samples, and results were available within 8 to 12 h. As a direct continuation of this work, we aim to test the applicability of the TPX methodology for the evaluation of a larger collection of defined staphylococcal strains, including MRSA, MSSA, and CoNS strains. A clinical application of this technique, foreseeable in the future, might be screening for MRSA directly from patient samples without the use of any time-consuming isolation steps. Still, it must be kept in mind that as a phenotypic method, the TPX assay is suited for screening purposes. The final definition of methicillin resistance in any S. aureus strain should be based on the presence of the mecA gene. The main benefit afforded by the initial use of the TPX methodology lies in its applicability to high-volume analysis and its low cost per analysis.
We thank Minna Lamppu for laboratory assistance and Jori Soukka for expert technical support. We also thank Jari Jalava, Harri Härmä, Niko Meltola, and Solja Nyberg for their valuable advice.
Arctic Diagnostics Ltd. provided the TPX plate readers, fluorescent dyes, and several immunoreagents for these experiments. This work was partially supported by an EU FP6 program.
Published ahead of print on 14 September 2009.