Nicole L. Gentile, UC Davis Pathology and Laboratory Medicine, 3455 Tupper Hall, Davis, CA 95616, USA. E-mail: firstname.lastname@example.org
To verify monoplex and multiplex gene-specific linear-after-the-exponential polymerase chain reaction (LATE-PCR) assays for identifying 17 microbial pathogens (i.e., Klebsiella sp., Acinetobacter baumannii, Staphylococcus aureus, Enterobacter sp., Pseudomonas aeruginosa, coagulase negative staphylococci, Enterococcus sp., Candida sp.) commonly associated with septicaemia using clinical isolates.
Methods and Results
Clinical isolates of each target pathogen were collected from the University of California, Davis Medical Center (UCDMC) microbiology laboratory. Five microlitres (μl) of each culture suspension (1 × 108 CFU ml−1) were added to 20 μl of monoplex mastermix. DNA extracted from clinical isolates was tested in multiplex. Monoplex assays demonstrated 100% sensitivity at this input level, except Enterobacter cloacae (2·7%), Ac. baumannii (57%) and Ps. aeruginosa (97·8%). All clinical isolates were positive in multiplex, with the exception of two Ac. baumannii, two Klebsiella oxytoca and two Candida parapsilosis isolates.
Sixteen pathogens can be identified by monoplex LATE-PCR assays with sensitivities ≥97·8%. The multiplex assay demonstrated 91·4% sensitivity when tested with DNA extracted from 70 different target strains.
Significance and Impact of the Study
This study demonstrates the potential of LATE-PCR to serve as an adjunct to culture if the reagents are optimized for sensitivity. Results warrant further testing through analytical and clinical validation of the multiplex assay.
The goal of the study was to test and verify monoplex and multiplex formats for a suite of gene-specific assays developed for the detection of 17 microbial pathogens (Table 1) commonly associated with septicaemia using clinical isolates collected from the University of California, Davis Medical Center (UCDMC) clinical microbiology laboratory. The assays were an asymmetric format of polymerase chain reaction (PCR), called linear-after-the-exponential (LATE)-PCR (Sanchez et al. 2004).
Table 1. The multiplex gene-specific sepsis assay characteristics
Pathogen (probe name)
ATCC, American Type Culture Collection; gDNA, genomic DNA; MRSA, methicillin-resistant Staphylococcus aureus; Tm, melting temperature.
Candida dubliniensis was not a target originally considered for this assay. Detection was unexpected, yet consistent and, therefore, has been added to the sepsis panel. However, purified ATCC gDNA positive controls were not available through ATCC for testing and clinical isolates were not collected. Data presented for this organism is strictly from DNA extracted from the ATCC whole pathogen.
Traditional asymmetric PCR is able to produce an amplicon product yielding a fluorescent signal 6–57% brighter than those of conventional symmetric PCR (Poddar 2000). LATE-PCR is an advanced form of asymmetric PCR that utilizes ‘limiting primers’ (LP) with higher melting temperatures (Tm) and lower concentrations relative to ‘excess primers’ (XP), which maintain reaction efficiency (Sanchez et al. 2004). A multitude of single stranded (ss) DNA products are produced once the LPs are exhausted and reaction product is predominantly driven by extension from the XPs. The ssDNA products ultimately allow for molecular beacon (probe) hybridization across a broad range of temperature without incurring competition from a second DNA strand which enables multiplexing using both temperature and fluorescence as unique differentiators. Such ability to simultaneously identify a list of pathogens in a single reaction would increase the clinical impact of an already fast and highly sensitive method.
With regards to pathogen detection, LATE-PCR has demonstrated feasibility for acting as a point-of-care test for identification of foot and mouth disease virus (Pierce et al. 2010) and providing species detection in African Swine Flu Virus down to 1 copy/reaction (Ronish et al. 2011). The LATE-PCR assay proposed in the current study is the first of its kind designed to detect both bacterial and fungal pathogens associated with human disease in a single tube by utilizing amplification and end-point detection of gene-specific sequences.
Materials and methods
Clinical isolate collection
A UC Davis Institutional Review Board (IRB) approved this study for clinical isolate collection and testing (protocol #221517-4). The number of clinical isolates collected for each target species is reflected in Table 2. All isolates were identified by the BD Phoenix™ Automated Microbiology System (Becton Dickinson, Franklin Lakes, NJ, USA) and collected from the UCDMC clinical microbiology laboratory. Isolates were stored in tryptic soy broth with 15% glycerol at −80°C. Ten microlitres (10 μl) of each archived organism was streaked on 5% sheep blood and nutrient agar and YM agar for bacterial and fungal isolates respectively. Plates were incubated at 37°C for 24 h.
Table 2. Monoplex assay sensitivities and melting temperature shifts
ATCC, American Type Culture Collection (whole pathogen); CoNS, coagulase-negative Staphylococcus; gDNA, genomic DNA (purified from ATCC whole pathogen); MRSA, methicillin-resistant Staphylococcus aureus; MSSA, methicillin sensitive SA; P, positive isolates; T, total number of isolates; Tm, melting temperature.
Shifts in Tms are relative to ATCC purified gDNA.
−2°C; one isolate
−3°C; seven isolates
−2°C; eight isolates
No Tm shifts identified
No Tm shifts identified
No Tm shifts identified
No Tm shifts identified
No Tm shifts identified
No Tm shifts identified
−6°C; two isolates
+11°C; two isolates
−6°C; two isolates
−7°C; one isolate
No Tm shifts identified
No Tm shift identified
No Tm shifts identified
No Tm shifts identified
No Tm shift identified
Primers and probes were designed by the Wangh Laboratory at Brandeis University (Waltham, MA, USA) (Rice et al. 2012). Table 1 lists the pathogens by dye channel and identifies the melting temperatures, target gene and acquisition numbers for American Type Culture Collection (ATCC, Manassas, VA, USA) purified genomic DNA (gDNA) and the whole organisms used as positive controls.
PCR mastermix was comprised of: 1× PCR buffer (cat. no. 10966-034; Invitrogen, Carlsbad, CA, USA), 3 mmol l−1 MgCl2 (cat. no. 10966-034; Invitrogen), 250 nmol l−1 dNTPs (Invitrogen), water (cat. no. BP2819-1; Fisher Scientific), 100 nmol l−1 of each probe (Biosearch Technologies, Novato, CA, USA), 50 nmol l−1 of each LP (Sigma-Aldrich, St Louis, MO, USA) and 1000 nmol l−1 of each XP (Sigma-Aldrich). All reactions contained 1·25 units of Platinum Taq DNA polymerase (cat. no. 10966-034; Invitrogen).
Monoplex assay sensitivity testing
All clinical isolates were tested against their specific LATE-PCR monoplex assay using spike controls. Spike control suspensions were created by inoculating molecular grade water with colonies from pure culture until suspensions were approximately 1·5 × 108 colony forming units (CFU) ml−1 for bacteria and 1 × 106 CFU ml−1 for fungi relative to a 0·5 McFarland turbidity standard. Clinical isolates were tested in parallel with ATCC whole pathogen and purified gDNA standards as positive controls.
Five microlitres of spike control or purified ATCC gDNA was added to each PCR tube containing 20 μl of mastermix, which included LP, XP and probe for only the particular organism in question. No template controls (NTCs) were tested using 5 μl of molecular grade water. Reactions were run in triplicate. Sensitivity was defined as (True positives)/(True positives + False negatives). A true positive was defined as a positive PCR result across all replicates, while a false negative was defined as a negative result in one or more replicates. Shifts in Tm relative to each pathogen's gDNA standard were also reported. Such deviations indicate differences in the target gene sequence relative to the consensus sequence used for assay design.
Multiplex assay verification
Suspensions were created as described above for three clinical isolates of each target species (five isolates of Acinetobacter baumannii) and for the corresponding ATCC pathogen reference strains as positive controls. Three separate ATCC strains having different staphylococcal cassette chromosome (SCC) mec types were tested for methicillin-resistant Staphylococcus aureus (MRSA). DNA was extracted from all suspensions using a commercial sample chemistry extraction method developed at Smiths Detection Diagnostics (SDD, Edgewood, MD, USA).
The SDD proprietary sample preparation chemistry is similar to many commercially available kits, such as Dynal, and utilizes nucleic acid immobilization and washing on silica-type paramagnetic particles (PMP). It is unique in the initial steps for liberation of gDNA from bacterial and fungal cells, but does not introduce any exotic components that could affect PCR. Capture, retention and recovery were all optimized for gDNA and PCR performance, and designed to be both scalable and compatible with automation. These methods will be reported elsewhere.
Extracted DNA concentrations were estimated spectrophotometrically at 260 nm. All isolates were tested in triplicate in the full multiplex assay at approximate concentrations of 106 copies per reaction by adding 5 μl of DNA to 20 μl of mastermix, which included all 11 LPs, 11 XPs and 11 probes for all target organisms in each PCR tube. NTCs and ATCC preparations of purified gDNA for each organism were also tested in triplicate as controls.
Data analysis and fluorescence interpretation
All amplification reactions were run using the Stratagene Mx3005P thermocycler (Agilent Technologies, Santa Clara, CA, USA). The raw data for each temperature step and each dye channel was exported to Microsoft Excel 2007. Data were normalized by dividing all values by the value of the signal at 70°C and subtracting the average background temperature-dependent level of fluorescence, using the signal profiles of the three NTC samples to derive the average backgrounds for each sample/fluor channel combination (Rice et al. 2012). All fluorescent signals greater than the assigned threshold value of 0·2 for each pathogen in Cal Red, Cal Orange and Quasar channels were considered positive. To account for the higher background noise in the FAM channel relative to the other three channels, fluorescent signals >0·15 were considered positive. Fluorescence was divided by the highest value to normalize the raw data. All negative isolates were retested to demonstrate reproducibility of results.
Preliminary multiplex analysis of purified DNA samples
In accordance with results from Rice et al. (2012), Figure 1 illustrates that all 17 pathogens are detectable and identifiable in their respective fluorescent dye channels with purified ATCC gDNA as template.
Monoplex sensitivity analysis using clinical isolate spike controls
Monoplex assay sensitivities are listed in Table 2. Each monoplex assay demonstrated 100% sensitivity for the input copy number used, with the exception of Ac. baumannii (57%), Enterobacter cloacae (2·7%) and Pseudomonas aeruginosa (98·7%). Several isolates displayed probe Tm shifts (Table 2). Most notably, for two isolates of Klebsiella oxytoca the probe Tm was shifted up by 11°C (Fig. 2a), resulting in melting temperatures and normalized fluorescent contours (Fig. 2b) similar to that of Klebsiella pneumoniae (Fig. 1a1). Both of these Kl. oxytoca isolates were negative in multiplex (Fig. 3a).
Cultures of two PCR positive and five PCR-negative isolates, identified by the UCDMC microbiology laboratory as Ac. baumannii, were sent to Midi Labs (Newark, DE, USA) for 16S sequencing. Sequence data from these seven clinical isolates identified the five PCR monoplex negative isolates as Acinetobacter genomospecies 3. Of these five negative isolates, two were tested in multiplex and were confirmed negative (Fig. 3a).
No fluorescence signal was observed for 35 of the Ent. cloacae isolates, indicating false negative results for 97·3% of isolates. However, postamplification reaction mixtures from selected reaction wells of these Ent. cloacae assays were further analysed on a DNA electrophoretic chip (Agilent Bioanalyzer) and verified presence of an amplification product with the expected length (99 base pairs) of the Ent. cloacae amplicon.
Multiplex analysis of DNA extracted from clinical isolates
Multiplex assay normalized anneal curves (fluorescence) are shown in Figs 3 and 4. All clinical isolates and positive controls for each organism were positive in multiplex, with the exception of two Ac. baumannii (Fig. 3a), two Kl. oxytoca (Fig. 3a) and two Candida parapsilosis (Fig. 4) isolates. Negative isolates were retested and in each case yielded the same result. Additionally, the multiplex assay detected DNA extracted from the ATCC whole organism, Candida dubliniensis. Further testing indicated that this organism was detected by the reagents designed for Candida albicans, Candida tropicalis and C. parapsilosis. All NTCs were negative.
Rapid administration of antibiotics is one of the most pivotal benefactors for septic patient survival. It has been shown that for every hour for which antibiotics are administered earlier, survival rate increases by 7–10% (Kumar et al. 2006). Unfortunately, only a handful of diagnostic methods are currently available to provide accurate and efficient identification of pathogens associated with sepsis. Two common forms of identification include mass spectroscopy (MS) and PCR. MS has been narrowed down to electrospray ionization (ESI) MS and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS (Ho and Reddy 2011). MALDI-TOF MS is able to give positive results in 20 min postincubation of 7–14 h through the use of BacT/ALERT blood culturing (Loonen et al. 2012). However, reduced incubation followed by MALDI-TOF MS does not result in faster identification and pathogens detected by this method are only correctly identified at the species level ≤78% and approximately 90% of the time from positive BacT/ALERT and BACTEC cultures respectively (Christner et al. 2010; Moussaoui et al. 2010; Loonen et al. 2012).
PCR has shown feasibility in providing early, species level pathogen differentiation in a clinical setting (Lehmann et al. 2008; Louie et al. 2008; Tran et al. 2012) and can decrease inadequate antibiotic treatment days by 22·8 days per 100 tests, thereby reducing mortality by 2·6% (Lehmann et al. 2009, 2010). When compared with conventional blood cultures, PCR has shown a 1·5 to 2-fold higher positivity rate, but does not provide antibiotic susceptibility data for most organisms that have acquired multi-drug resistance (Wallet et al. 2011). Newer PCR assay strategies, such as LATE-PCR, can allow strain differentiation with increased multiplexing capability, in turn providing more information in a single tube (Sanchez et al. 2004). However, results accrued from the current LATE-PCR sepsis assay under study demonstrate the need for further optimization of the assay reagents before clinical testing can occur.
Through assay development and verification testing, characteristics of each individual assay were examined under several conditions, including the type and purity of target input, for monoplex and varying configurations of multiplex. Without the influence of a clinical sample matrix like blood (i.e., bacterial targets were suspended in water only), assay results were similar or identical for whole bacterial cell spikes (see Fig. 2 as one example, other data not shown), DNA purified using the SDD chemistry (Figs 3 and 4), and purified ATCC gDNA reference materials (Fig. 1). Additionally, estimation of input copy number (in CFU per reaction), was identical for both raw spikes and extracted DNA using the SDD chemistry demonstrated by plating dilutions of the sample suspensions. For simplicity, characterization of monoplex assays was done using whole bacteria. These suboptimal test samples would, if anything, challenge each assay with a better likelihood of revealing performance issues.
Some of the assays when in more complex multiplex reaction mixtures, however, were very sensitive to test sample characteristics, with loss of sensitivity and in some cases assay failure caused by using whole bacterial cells as the test sample. For these particular assays, robustness was compromised by the presence of the additional assay components (primers, probes) and any additional organic molecules spiked into the assay caused failure, hence the use of extracted DNA for multiplex verification.
Rapid detection of Ac. baumannii is valuable due to its tendency to spread and its innate and acquired antimicrobial resistance (Higgins et al. 2007). Many species in the Acinetobacter genus, such as Acinetobacter genomospecies 3 and Acinetobacter genomospecies 13, are more recently recognized as important but distinct nosocomial pathogens (Higgins et al. 2007). However, detection of these particular organisms is complicated because they are closely related genetically (gyrB) and difficult to differentiate phenotypically (Gerner-Smidt et al. 1991; Gerner-Smidt 1992; Yamamoto et al. 1998; Higgins et al. 2007). Therefore, these unnamed species tend to be grouped with Acinetobacter calcoaceticus in the Ac. calcoaceticus-Ac. baumannii complex.
Unfortunately, substrates used by current bacterial species identification systems have not been tailored specifically to identify the majority of Acinetobacter sp. Particularly, the three clinically relevant members of the Ac. calcoaceticus-Ac. baumannii complex (i.e. Ac. baumannii, Acinetobacter genomic species 3 and Acinetobacter genomic species 13TU) cannot be separated by the majority of the widely available methods and are therefore uniformly identified as Ac. baumannii (Peleg et al. 2008). These known complications combined with the low sensitivity (57%) of Ac. baumannii in the monoplex LATE-PCR assay prompted further investigation.
PCR-negative isolates were grown on blood agar to test for haemolysis, indicative of Acinetobacter haemolyticus isolates mislabelled as Ac. baumannii; however, no obvious haemolysis was seen. Additionally, it is expected that the current hospital automated antimicrobial sensitivity identification system (BD Phoenix™ Automated Microbiology System) would have correctly identified Ac. haemolyticus, as it is among the five Acinetobacter species specified on the panel (Salomon et al. 2005).
Seven of the 42 isolates collected were 16S sequenced for species identification. Sequencing identified five PCR-negative isolates as an unnamed Acinetobacter species, called Acinetobacter genomospecies 3, of which the BD Phoenix™ Automated Microbiology System cannot specifically classify. These results show that the primers and/or probes designed for Ac. baumannii are highly specific. However, the high Ac. baumannii specificity observed in this assay may or may not prove useful for clinical diagnostics, as treatment may be the same regardless of the Acinetobacter infection.
Some researchers have questioned whether there is a need for species identification amongst the Acinetobacter genus (Gerner-Smidt et al. 1991). Others argue that it is necessary to distinguish between the Ac. baumannii group and other Acinetobacter sp. since the latter organisms rarely have infection control implications (Peleg et al. 2008). The lack of knowledge regarding epidemiology, pathogenicity and clinical impact of a virulent Acinetobacter species (e.g. Acinetobacter genomospecies 3) combined with the absence of capable detection methods may delay the physician from administering the proper therapy for an infection. Development of primers and probes designed to target additional Acinetobacter species may be relevant for clinical diagnostics; however, this issue warrants further investigation.
The Tm for two Kl. oxytoca isolates was shifted up (increased) by 11°C (Fig. 2a), making their assay profile similar to that of Kl. pneumoniae (Fig. 1a). Possible reasons for the temperature shift are as follows: the probe target sequence in that particular strain is different from the reference to which that probe had been designed, the molecular beacon redundant end sequences (stem structure of the probe) may match the target for these isolates, primer and/or probe specificity may be inherently low, or mislabelling of isolates by the BD Phoenix™ Automated Microbiology System may have occurred. If the isolates were indeed mislabelled as Kl. oxytoca, but were actually Kl. pneumoniae, it is expected that the specific Kl. oxytoca probes would not have picked up the two organisms in monoplex. However, recent data demonstrated cross-reactivity of Kl. oxytoca reagents in monoplex in the presence of high concentrations of Kl. pneumoniae DNA extracted from pure culture (data not shown).
The negative multiplex results for these two Kl. oxytoca isolates further support the interpretation that the isolates may either be a Klebsiella sp. closely related to Kl. oxytoca, or a closely related organism of a different genus, such as Escherichia coli. In a documented case study, the phoE gene from an E. coli contaminant was unexpectedly amplified in a PCR assay specific to Klebsiella, leading to false positive results (Bastian and Bowden 1996). The possibility of amplification occurring from an organism other than a Klebsiella species cannot be ruled out until analytical specificity experiments can be done to demonstrate that related organisms do not cause false positive results. Regardless of whether the organism was of a different genus, or among Klebsiella species, it is likely that the sequence variations in the phoE gene may have been significant enough to impair the Kl. oxytoca primers and/or probe from hybridizing to its target in the presence of all other reagents in multiplex.
The lack of fluorescence signal and presence of the 99 base pair Ent. cloacae product (amplicon) indicates that the gene sequence was amplified by the primers and the issue most likely resides with the probe hybridizing to the amplification product. The low sensitivity (2·7%) of the Ent. cloacae assay further indicates that this organism should be removed from the test panel until the molecular beacon, and perhaps overall design, can be improved to allow detection of all subspecies of Ent. cloacae.
Conversely, the positive result seen with the C. dubliniensis ATCC organism indicates that the reagents designed for Candida sp. can, indeed, pick up additional Candida targets not previously validated with gDNA. This finding warrants further testing with clinical isolates to verify reagent sensitivity.
In conclusion, 16 pathogens commonly associated with septicaemia can be identified by monoplex LATE-PCR sepsis assays with sensitivities >97·8%. The multiplex assay demonstrated 91·4% sensitivity when tested with DNA extracted from 70 different target strains in pure culture. This study demonstrates the potential of the assay to provide a molecular diagnostic screening method for septic patients and may prove useful as an adjunct to blood culture provided that the reagents can be optimized to improve the sensitivity of the multiplex assay.
Within the scope of the present study, the optimal Mg2+ and dNTP conditions for the full multiplex assay were prescribed as a design input, based on the nonclinical testing (Rice et al. 2012). Optimization would require titrating Mg2+ and dNTPs to determine concentrations that will provide the most efficient amplification and detection in addition to varying extension time and temperature for determining the optimal conditions for the reaction. However, based on recent findings (to be published elsewhere), varying these reaction mixture conditions and/or cycling conditions would not improve performance characteristics for the problem assay elements, especially with the variables introduced with complex test samples (i.e. blood). Therefore, the next step towards improving the current assay would be to remove or redesign specific parts of the assay. Furthermore, results warrant further testing through analytical and clinical validation of the multiplex assay.
We would like to thank Kristen Hale-Katz, Lucy LaMar, Dr Chris Polage and the staff at the UCDMC Clinical Microbiology Laboratory for their help with clinical isolate collection. Also, special thanks to Samaan Mahmoudzadeh and Isaac Sergei Horowitz of the Kost Lab at UCD for reviewing this manuscript. This study was completed at the University of California, Davis and supported by an ARRA grant (award number RC1EB010643) from the National Institute of Biomedical Imaging and Bioengineering. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Biomedical Imaging and Bioengineering or the National Institutes of Health.