Lawrence J. Wangh, Department of Biology, Brandeis University, Waltham, MA 02453-2728, USA.E-mail: email@example.com
The goal of this study was to develop a molecular diagnostic multiplex assay for the quantitative detection of microbial pathogens commonly responsible for ventilator-associated pneumonia (VAP) and their antibiotic resistance using linear-after-the-exponential polymerase chain reaction (LATE-PCR).
Method and Results
This multiplex assay was designed for the quantitative detection and identification of pathogen genomic DNA of methicillin-resistant Staphylococcus aureus (MRSA), Acinetobacter baumannii, Pseudomonas aeruginosa, plus a control target from Lactococcus lactis. After amplification, the single-stranded amplicons were detected simultaneously in the same closed tube by hybridization to low-temperature molecular beacon probes labelled with four differently coloured fluorophores. The resulting hybrids were then analysed by determining the fluorescence intensity of each of the four fluorophores as a function of temperature.
This LATE-PCR single tube multiplex assay generated endpoint fluorescent contours that allowed identification of all microbial pathogens commonly responsible for VAP, including MRSA. The assay was quantitative, identifying the pathogens present in the sample, no matter whether there were as few as 10 or as many 100 000 target genomes.
Significance and Impact of the Study
This assay is rapid, reliable and sensitive and is ready for preclinical testing using samples recovered from patients suffering from ventilator-associated pneumonia.
Nosocomial pneumonia affects 27% of critically ill patients in the United States (Richards et al. 1999). Of those cases, 86% or 300 000 US patients per year are associated with mechanical ventilation (Koenig and Truwit 2006). Ventilator-associated pneumonia (VAP) primarily occurs when bacteria travel into the lower lung either through an endotracheal or tracheostomy tube, or around the cuff holding it in place. Often, bacteria colonize the tube and are embolized into the lungs with each breath. Many of the afflicted individuals have an underlying lung or immune problem. It is therefore not surprising that the mortality rate of VAP patients is up to 50% (Papazian et al. 1996).
Linear-after-the-exponential polymerase chain reaction (LATE-PCR) is an advanced form of nonsymmetric PCR in which amplification of double-stranded DNA is followed by efficient production of single-stranded amplicons. LATE-PCR also allows for a high degree of multiplexing followed by endpoint detection via hybridization of low-temperature fluorescent probes in multiple colours over a wide range of temperatures (Sanchez et al. 2004, 2006; Hartshorn et al. 2005a,b, 2007; Pierce et al. 2005; Salk et al. 2006; Rice et al. 2007). This highly multiplexed single tube approach is not achievable using conventional symmetric PCR which only generates double-stranded amplicons.
We have previously described both relatively simple LATE-PCR assays for foot-and-mouth disease virus (Pierce et al. 2009), African swine fever virus (Ronish et al. 2010), and mitochondrial DNA mutational load (Osborne et al. 2009), and more complex multiplexed assays for TB (Rice et al. 2012a) and Sepsis (Carver-Brown et al. 2012; Gentile et al. 2012; Rice et al. 2012a,b). The single tube, quantitative, multiplex assay described here identifies a set of target genes in several bacterial species that are frequently associated with VAP. These pathogens include: Methicillin-resistant Staphylococcus aureus (MRSA), Acinetobacter baumannii and Pseudomonas aeruginosa. Detection of these types of bacteria is made more difficult by the fact that many strains of MRSA differ in terms of the type of SCCmec cassette that confers methicillin resistance by insertion into the orfX target sequence, as well as by the fact that strains of MSSA, methicillin sensitive Staph. aureus, may have a SCCmec cassette, but no mecA gene. Yet another challenge for construction of a VAP assay is that Acinetobacter bauminii has to be distinguished from Acinetobacter spp. The goal of this study was to consistently distinguish all of these pathogens in a single closed-tube assay when the number of genomes present initially was varied over five orders of magnitude, 101-to-105. No other VAP PCR assays either in the literature or commercially marketed attempt to detect these pathogens, antibiotic resistance genes, or starting genomic DNA copy numbers as low as 10 (Apfalter et al. 2005; Rios-Licea et al. 2010).
Materials and Methods
LATE-PCR multiplex assay design
Linear-after-the-exponential polymerase chain reaction is an advanced form of nonsymmetric PCR. Each LATE-PCR utilizes a limiting primer (L) and an excess primer (X) whose initial melting temperatures, Tm's, fit the formula Tm0L-Tm0X ≥ 0. As a result double-stranded amplicons are reliably generated exponentially until the limiting primer runs out and the reaction switches to linear amplification of a single-stranded amplicon driven by the excess primer. The switch from double strand to single strand is designed to occur soon after the amount of double-stranded amplicon becomes detectable at its threshold cycle, CT. The amount of single-strand target accumulated at any point in the linear phase of amplification is conveniently measured with low-temperature molecular beacon fluorescent probes that hybridize to their target sequences at a chosen temperature below the amplification annealing temperature of the limiting primer. The intensity of the resulting signals is highly reproducible and can give a quantitative measure of the initial copy number of the target genome.
The composition of the VAP assay
The VAP assay was first designed and tested at Brandeis University and then further optimized at Smiths Detection Diagnostics. All reactions carried out at Brandeis University had a final volume of 25 μl and contained a mixture of components comprised of: 1× PCR buffer (cat. no. 84028; Quanta BioSciences, Gaithersburg, MD, USA), 5 mmol l−1 MgCl2 (cat. no. 84030; Quanta BioSciences, Gaithersburg, MD, USA), 400 nmol l−1 dNTPs (cat. no.11874920; Roche, Basel, Switzerland), 30 nmol l−1 PrimeSafe II (Rice et al. 2007), nuclease-free water (cat. no. AM9937; Applied Biosystems, Carlsbad, CA, USA), 100 nmol l−1 of each probe (Biosearch Technologies, Novato, CA, USA), 100 nmol l−1 of the Staph. aureus orfX-SCCmec limiting primers, 50 nmol l−1 of the Ac. baumannii, Ps. aeruginosa, L. lactis and Staph. aureus mecA limiting primers, and 25 nmol l−1 of the Staph. aureus spa limiting primer (Biosearch Technologies, Novato, CA, USA) and 1000 nmol l−1 of each excess primer (Biosearch Technologies, Novato, CA, USA). All reactions contained 2·0 units of AccuStart Taq DNA polymerase (cat no. 95061-01K; Quanta BioSciences, Gaithersburg, MD, USA), 1 μl of mixed genomic DNA target that was prepared from American Type Culture Collection (ATCC) genomic DNA (MRSA cat. no. BA1556; Ps. aeruginosa cat.no. 47085; Ac. baumannii cat. no.19600) as a dry reagent and was suspended in 500 μl of 10 mmol l−1 Tris-HCl pH 8·3.
Each 25 μl test of the VAP assay carried out at Smiths Detection Diagnostics included 1 μl containing 20 ng μl−1 human genomic DNA (cat.no. 11691112001; Roche, Basel, Switzerland), 105 copies of Staphylococcusepidermidis genomic DNA (cat.no.12228D-5, ATCC), 105 copies of Acinetobacter spp. genomic DNA (cat.no.49467D-5, ATCC), 105 copies of Pseudomonas putida genomic DNA (cat.no.47054D-5, ATCC) and 105 copies of L. lactis genomic DNA. (extracted from cheese culture prepared at Smiths Detection Diagnostics using Sigma GenElute Kit- March 2011).
Thermal amplification parameters of the VAP assay
Each reaction carried out at Brandeis University was run using either a Bio-Rad IQ5 Multicolor Real-Time Detection System (Bio-Rad Laboratories, Hercules, California) or a Stratagene Mx3005P (Agilent Technologies, Santa Clara, California). There was no significant difference in the data observed on the two machines. Several no-template-controls (NTC) were also run in parallel in every experiment. An initial denaturation step at 95°C for 3 mins was followed by five amplification cycles of 95°C for 10 s, 58°C for 15 s and 72°C for 30 s. After which, there were 45 amplification cycles of 95°C for 10 s, 62°C for 22 s and 72°C for 30 s. Next, the temperature was dropped from 72°C to 25°C and was increased in 1°C melt steps, 30 s each all the way up to 80°C. Finally, endpoint data were collected during anneal steps from 80°C decreasing the temperature in 1°C, 30 s steps down to 25°C. Fluorescence was measured in the FAM, Cal Orange 560, Cal Red 610, and Quasar 670 channels at each anneal step.
Additional assay optimization at Smiths Detection Diagnostics was carried out on a Stratagene Mx3005P. An initial denaturation step at 95°C for 5 mins was followed by 40 amplification cycles of 95°C for 10 s, 62°C for 22 s and 72°C for 30 s. Next, the temperature was increased from 72°C to 75°C. Finally, endpoint data collection was carried out by decreasing the temperature in 1°C anneal steps, 1 min 6 s each between 75 and 25°C.
Data analysis of the VAP assay
The samples analysed at Brandeis University were processed as follows and then displayed as fluorescent contours, Figs 1 and 2: Each raw fluorescent value in each colour was normalized by dividing it by the value in the same colour at 70°C. This is logical because no probe is bound to any target at this temperature, and the fluorescence in each colour is at the background level for all probes of that colour. Next, each data set was corrected for background temperature-dependant levels of fluorescence as a result of unbound probes in that colour. This was accomplished by subtracting the background temperature-dependent level of fluorescence in that colour in the corresponding NTC.
Data generated at Smiths Detection Diagnostics, Figs 3 and 4, were analysed by normalizing the signal at each anneal step in a particular colour by the mean value of fluorescence in that same colour at a particular temperature. In the FAM dye channel, the mean fluorescence value at 45°C was calculated from a set of replicate reactions containing the orfX-SCCmec. In the Quasar dye channel, the mean fluorescence value at 55°C was calculated from a set of replicate reactions containing spa and mecA; In the Cal Orange dye channel, the mean fluorescence value at 55°C was calculated from a set of replicates containing Ac. baumanni and the L. lactis control; and in the Cal Red dye channel, the mean fluorescence value at 50°C was calculated from a set of replicate reactions containing Ps. aeruginosa. Next, the standard deviation of each set of replicates was calculated for the specific temperature points. Finally, the confidence interval for the standard deviation of each set of replicates was calculated and then displayed graphically to illustrate the statistical significance and linear trending of each target over a wide range of initial target copies, Figs 3 and 4.
The design of the VAP assay
The two-dimensional, temperature vs colour, scheme for the VAP assay, Table 2, builds on our experience constructing a single tube assay for seventeen pathogens responsible for sepsis (Rice et al. 2012ab). Subdivision of the temperature space within a single colour allows for quantitative identification of each target. The gyrB gene of Ac. baumannii, the toxA gene of Ps. aeruginosa and an unnamed control target sequence in L. lactis are each detected using a separate low-temperature molecular beacon having a short loop, a two basepair stem and a terminal fluorophore and black whole quencher. These features render each probe highly specific for its target sequence with a Tm between 63 and 43°C.
Detection and analysis of MRSA in the VAP assay depends on the detection of three genes: (i) the SCCmec-orfX target establishes that the strain of Staph. aureus contains the cassette that could cause the organism to be methicillin resistant; (ii) the mecA target establishes that the SCC cassette is, in fact, methicillin resistance; (iii) the spa target confirms that the organism is, in fact, Staph. aureus. Some current PCR tests for MRSA detect the SCCmec-orfX junction that determines that the SCCmec cassette has been inserted into MSSA, but this strategy can result in false positives and false negatives, either because the cassette loses mecA or because a variant of the SCCmec-orfX cassette goes undetected. Because LATE-PCR is quantitative, such false readings can be distinguished by measuring the ratios between gene concentrations. This is accomplished in the VAP assay by the use of a FAM probe for the detection of the SCCmec-orfX gene target and Quasar 670 probes for the detection of spa and mecA. A 1 : 1 ratio of the spa signal and the mecA signal plus a signal for SCCmec-orfX provides robust evidence of MRSA.
Construction of the VAP assay
Construction of the highly multiplexed assay for VAP began with bioinformatic identification of specific primer pairs that would amplify gene-specific sequences characteristic of each organism. These gene sequences were scrutinized using Basic Local Alignment Search Tool (BLAST) and were then compared with all other sequences in the NCBI GenBank. The chosen sequences were then further analysed to identify relatively conserved organism-specific targets. Each of these sequences, in turn, was downloaded into Visual OMP (VOMP) software (DNASoftware, Inc., Ann Arbor, MI, USA) for primer and probe design.
Linear-after-the-exponential polymerase chain reaction primers and probes were chosen for multiplexing based on the following criteria. All limiting primers were picked to have average concentration-dependent Tm's of 71·1°C (range 73·6–68·0°C), and all excess primers of 68·2°C (range 71·4–66·0°C). In addition, the Tm of each excess primer was designed to have a lower initial Tm than the initial Tm of its paired limiting primer. The design process also circumvented possible primer–primer or primer–probe interactions. Each pair of prospective primers and its corresponding probe sequences was run through another BLAST search to confirm that it had the appropriate homology needed to identify the target sequences and was also sufficiently divergent from related genome targets to prevent amplification or detection of an incorrect target. Collectively, these precautions greatly reduce the chances of false positives. Table 1 shows the primer, probe and target sequences for each of the components of the VAP assay.
The LATE-PCR, multiplexed, VAP assay has the capacity to quantitatively detect all targeted pathogens. It is comprised of fourteen primers and six probes in four fluorescent colours. The first step in testing the assay was to establish that each target pathogen was amplified and detected in a monoplex reaction. Figure 1a illustrates the monoplex amplification of the mecA gene found in MRSA and its detection in the Quasar 670 dye channel for 105 down to 100 starting copies. The next step was to establish that each pathogen has its own unique fluorescent signal in the multiplex. Figure 1b provides an example of the results obtained at this stage of assay construction. In this case, the Ac. baumannii target was quantitatively detected when decreasing numbers of genomes (105–10 starting copies) were added to a set of reactions all of which contained all fourteen primers and six probes.
Construction of the fully optimized assay began upon completion of the above tests. Genomic DNAs of all pathogens were mixed together, and this multitarget sample was diluted in 10-fold steps from 105 to 10 copies. Figure 2 illustrates the quantification of the fluorescent contours that resulted from this multiplex. Figure 2a shows the detection of the SCCmec-orfX cassette in the FAM dye channel, which along with the detection of spa and mecA in the Quasar dye channel (Fig. 2d) identifies MRSA. Figure 2b shows the composite signal in the Cal Orange 560 dye channel that detects the gyrB target of A. baummanii along with the internal control of L. lactis (105 copies). Figure 2c shows detection of toxA of Ps. aeruginosa in the Cal Red 610 dye channel.
To assess the statistical significance of these fluorescent contours, standard deviations and confidence values of particular temperature points of replicate assays were calculated, Fig. 3. Figure 3a shows that the assay is sensitive down to 10 copies of SCCmec-orfX gene target when the FAM signal is read at 45°C. Figure 3b shows that the assay is sensitive down to 10 copies of the gyrB gene target when the Cal Orange 560 signal is read at 55°C in a background of L. lactis. Figure 3c shows that toxA detection in Cal Red 610 at 50°C is sensitive down to 10 copies. Figure 3d shows the fluorescent signal of a 1 : 1 ratio of spa and mecA down to 10 copies of the MRSA genome detected in Quasar 670 at 45°C.
To assess the linear nature of the endpoint fluorescent values at a single temperature point for each fluorescent channel, linear regression analysis was applied to these mean fluorescent values, Fig. 4. Figure 4a,b shows the linear regression of orfX-SCCmec in FAM and A. baummannii with L. lactis in Cal Orange with r-squared values of 0·99. Figure 4c,d for Cal Red and Quasar, respectively, shows slightly lower r-squared values of 0·97.
Results presented here describe the design and construction of a highly multiplexed LATE-PCR single tube assay for detection and identification of the pathogens responsible for ventilator-associated pneumonia. The significance of this assay lies in its quantitative nature at endpoint when all pathogenic targets are present down to 10 initial copies of genomic DNA.
The assay design relied on successful bioinformatic analysis. Primers for specific gene sequences were chosen based on their compatibility with the properties of LATE-PCR. The detection of multiple pathogens in a single tube depended on probe design, dye channel and melting temperature. Initial primers and probe sequences were designed based on our knowledge of gene-specific pathogen detection assays (Rice et al. 2012b).
The primers and probes were arranged in temperature and dye channel space as shown in Table 2. The amplification efficiency of the primers was tested in monoplex reactions using both SYBR green and probes. The multiplex VAP assay was next tested for specificity of pathogen detection in the presence of all primers and probes. This was carried out for each target one at a time. Once the specificity of pathogen detection was established, work on assay optimization began.
Table 2. The VAP LATE-PCR multiplex assay configuration. The fluorescent dye channels are labelled along the top of the table vs temperature showing the detection space for each of the VAP pathogens. The genes used for pathogen and antibiotic resistance detection are shown in parenthesis
Staphylococcus aureus (orf X-SCCmec)
Acinetobacter baumannii (gyrB)
Pseudomonas aeruginosa (toxA)
Staphylococcus aureus (spa)
Staphylococcus aureus (mecA)
During optimization, the assay configuration shifted from single pathogen target amplification to amplification of six pathogen targets in a single tube, and interactions between primers, probes and amplified targets were eliminated. Statistically significant quantitative endpoint analysis was achieved by varying assay components: Taq, magnesium, and dNTP. As a result, the Taq concentration was increased from 1·25 to 2 units, the magnesium concentration was increased from 3 to 5 mmol l−1, and the dNTP concentration was increased from 250 to 400 μmol l−1. Also a mispriming agent, developed in our laboratory, PrimeSafe II, was added at this time. These adjustments enhanced fluorescent signals of all amplicons. Increasing the concentration of these reagents allowed all of the amplicons to be amplified simultaneously.
The quantitative nature of LATE-PCR endpoint fluorescence is dependent on three adjustable factors: number of amplification cycles, probe concentration and primer efficiency. The number of amplification cycles in a thermal profile of a LATE-PCR assay is important for two reasons, one it dictates the amount of single-stranded DNA generated, and two it determines the length of time of the assay. The more cycles added to amplification the closer the assay gets to a terminal concentration. Terminal concentration is reached, when the excess primer is out competed by the single-stranded amplicon. In the initial design, the number of amplification cycles for this assay was 50 cycles. The first five cycles were run with a lower annealing temperature because of the primers inefficiency in the orfX-SCCmec region. After further optimization of the multiplex assay, it was found that the orfX-SCCmec amplification did not significantly benefit from these initial cycles at a lower annealing temperature. In assay optimization, the number of amplification cycles was examined and it was found that 45 amplification cycles with an annealing temperature of 62°C allowed for better quantification.
Quantification is important for this assay because the pathogens associated with VAP are clinically significant at high copy number. Probe concentration at 100 nmol l−1 was found to be optimal for all targets over a range of 10–100 000 starting copies.
The last factor that affects quantification is primer efficiency, which is based on their design and concentration. Some primers such as Staph. aureus (spa) were very efficient because of elements in their design, while others such as Staph. aureus MRSA (orfX-SCCmec) were not efficient because of primer mismatches to the target which delay amplification. To adjust for differences in efficiency, the limiting primer concentrations were titrated. As reported in the Materials and Methods, the Staph. aureus (spa) limiting primer concentration was decreased to 25 nmol l−1. This decreased the number of double strands produced, which then slowed the rate of single-strand production. The Staph. aureus MRSA (orfX-SCCmec) limiting primer concentration was increased to 100 nmol l−1, so more double strands would be produced and would increase the rate of single-strand production.
After incorporating all of the design changes, the optimized LATE-PCR VAP assay achieves quantitative detection and identification of all targeted pathogens in a single closed tube. The presence of each target in its respective dye channel and temperature space is evident by endpoint fluorescence. The signal intensities of these fluorescent contours are statistically significant. Each target has a reliable and reproducible limit of detection down to 10 copies and up to 100 000 copies. The signal intensity is linear over this range. This rapid, reliable and sensitive assay is ready for preclinical testing using samples recovered from patients suffering from ventilator-associated pneumonia.
We are grateful to Dr. Barry Kreiswirth of PHRI, Newark N.J. for genomic DNA from bacterial cultures of twelve strains of Staph. aureus. This research was supported by Smiths Detection Diagnostics, Inc., Watford, UK.