Variation-Tolerant Capture and Multiplex Detection of Nucleic Acids: Application to Detection of Microbes

Abstract

In contrast to ordinary PCRs, which have a limited multiplex capacity and often return false-negative results due to target variation or inhibition, our new detection strategy, VOCMA (variation-tolerant capture multiplex assay), allows variation-tolerant, target-specific capture and detection of many nucleic acids in one test. Here we demonstrate the use of a single-tube, dual-step amplification strategy that overcomes the usual limitations of PCR multiplexing, allowing at least a 22-plex format with retained sensitivity. Variation tolerance was achieved using long primers and probes designed to withstand variation at known sites and a judicious mix of degeneration and universal bases. We tested VOCMA in situations where enrichment from a large sample volume with high sensitivity and multiplexity is important (sepsis; streptococci, enterococci, and staphylococci, several enterobacteria, candida, and the most important antibiotic resistance genes) and where variation tolerance and high multiplexity is important (gastroenteritis; astrovirus, adenovirus, rotavirus, norovirus genogroups I and II, and sapovirus, as well as enteroviruses, which are not associated with gastroenteritis). Detection sensitivities of 10 to 1,000 copies per reaction were achieved for many targets. VOCMA is a highly multiplex, variation-tolerant, general purpose nucleic acid detection concept. It is a specific and sensitive method for simultaneous detection of nucleic acids from viruses, bacteria, fungi, and protozoa, as well as host nucleic acid, in the same test. It can be run on an ordinary PCR and a Luminex machine and is suitable for both clinical diagnoses and microbial surveillance.

INTRODUCTION

Infections are caused by diverse microbes. They often give similar symptoms. Ideally, in a clinical situation, relevant pathogens, both common and uncommon, plus their antibiotic resistance and virulence, should be detected in one test. Variable nucleic acids (like in RNA viruses), large and expanding pathogen families, and diverse target molecules (antibiotic resistance genes, toxins) are important clinical challenges. They call for broadly targeted and multiplex DNA and RNA detection methods.

Another challenge is the need to encompass both large sample volumes (10 to 100 ml; for the diagnosis of septicemia or the surveillance of food-borne pathogens [11, 18]) and small sample volumes (50 to 200 μl; for the diagnosis of many causes of meningoencephalitis in a small volume of cerebrospinal fluid). Sensitivities down to a few target molecules per reaction and turnaround times of a few hours are often important. PCRs cannot normally be highly multiplex, due to mispriming. However, with conserved targets (16S rRNA and the second internal transcribed spacer [ITS2]), many different pathogens can be detected with one primer pair. Padlock probes are an alternative to PCR. They can be highly multiplex and can be combined with magnetic nanoparticles (17) in a commercial portable read-out device (24), although the short shanks make them especially vulnerable to pathogen variation. There is room for a new principle for the multiplex detection of nucleic acids.

Here we present a new multiplex diagnostic technique encompassing mismatch tolerance, optional capture, and specific, as well as generic, amplification steps. The variation-tolerant capture multiplex assay (VOCMA) (Fig. 1; see also Table S1 in the supplemental material) combines several features in a novel way: (i) a combination of capture probe and primer (i.e., primer-probes), (ii) variation-tolerant primer-probes, (iii) a single-tube, dual-step combination of target-specific preamplification at low primer concentrations and higher annealing temperature, followed by general amplification using generic primers at higher concentrations and lower annealing temperature, (iv) asymmetric amplification for the buildup of excess biotinylated strands, (v) variation-tolerant (VT) detection probe hybridization to target specific sequences, providing a large potential for multiplexing, as well as high sensitivity and specificity, and (vi) a simple and specific readout. Variation tolerance is achieved by using long target-specific primers and probes containing degenerations and the universal base analog deoxyribose-inosine (dInosine) according to variation analysis and principles elaborated in a previous publication (the NucZip algorithm) (15).

Fig 1.

Schematic summary of VOCMA principle. Optional capture is followed by amplification (in tube 1) and a liquid microarray detection step (in tube 2). The first specific VT primer-probe, bound to a magnetic bead (small brown circle), is used to capture the target. Beads are fixed by a magnet (red), and the sample is washed to remove unbound nucleic acid. A one-step PCR mixture containing a low concentration of the long first and second specific VT primer-probes and a high concentration of short generic primers is added. Subsequently, a high annealing temperature of 10 cycles allows the specific VT part of the magnetic bead-bound first specific VT primer-probe to elongate the first copy, which is then copied by the specific VT part of the second specific VT primer-probe. The following 40 to 50 cycles are run at a low annealing temperature, allowing asymmetric amplification by using a high concentration of short generic primers with an excess of the second biotinylated generic primer, generating biotinylated ssDNA. An aliquot is transferred to tube 2 where a Luminex bead-bound (large pink circle) specific VT detection probe hybridizes to the amplified target region. Streptavidin-R-phycoerythrin is added to the reaction to allow streptavidin (orange color) to bind to the biotin. The lasers from the Luminex 200 machine will identify the bead and measure the amount of the fluorescence reporter phycoerythrin (red star). The first specific VT primer-probe, the first generic primer, and the specific VT detection probe (all indicated in blue color) are antisense versus the first strand of the target sequence, whereas the second specific VT primer-probe and the second generic biotinylated primer are sense versus the target (both indicated in brown color).

Here we present the basic conditions for variation-tolerant capture, amplification, and detection, applied to two clinically important situations, diagnosis of sepsis and gastroenteritis (gastro).

MATERIALS AND METHODS

Design of VT primer-probes and detection probes.

The specific VT primer-probes and the specific VT detection probes (sequences and modifications in Tables S2 to S5 in the supplemental material) were designed using BLASTN (NCBI, NIH) (1), ClustalX (8, 19), ConSort (J. Blomberg, described in the supplemental material), Mfold (25), VisualOmp (DNA Software, Ann Arbor, Michigan), and NucZip (15). Target sequences or whole genomes collected in BLASTN were further aligned to map nucleotide conservation and variation with ConSort, which visualizes the frequency of variation and the variation of nucleotide composition in each base position, as well as the number of aligned sequences. The design was evaluated in silico before the in vitro analysis by adjusting and redesigning to avoid hairpin, homodimer, and heterodimer formation, while retaining target specificity. The predicted Gibb's free energy (VisualOmp) and NucZip scores were used as tools to predict hybridization efficiency and to optimize variation tolerance. NucZip simulates nucleation, the initial contact between segments of a perfect match, followed by “zipping” the two strands upstream and downstream (15). The panel was thus built up stepwise, in a recursive loop involving reevaluation and redesign. Variation tolerance was enhanced by using 50 to 70 nucleotide (nt) primer-probes and probes, degenerations and dInosines at sites of target variation, and locked nucleic acid (LNA) residues to increase affinity.

Synthetic targets.

The synthetic targets (Biomers.net GmbH, Ulm, Germany) had a natural primer-probe and detection probe complementary sequence derived from GenBank (see Tables S2 and S4 in the supplemental material). The synthetic target for norovirus genogroup II was artificially designed to perfectly match the primer and probe sequences. Parts of the original sequences of the pathogens were deleted, because of constraints on synthetic oligonucleotide length (138 to 200 nt). Synthetic single-stranded DNA (ssDNA) targets were serially diluted 10-fold from 0.5 × 10⁷ to 0.5 × 10⁰ molecules/μl either with 0.1× Denhardt′s solution (Sigma-Aldrich Sweden AB, Stockholm, Sweden) in the sepsis VOCMA or with 20 ng/μl yeast RNA in diethyl pyrocarbonate (DEPC) water (Ambion, Austin, TX) in the gastro VOCMA.

Nucleic acid extraction from clinical samples.

The patient samples were provided and routinely handled, anonymously, according to Uppsala University Hospital rules. The EasyMag extraction system (bioMérieux AB, Sweden) was used for fecal samples, reference bacteria, and bacteria and fungi from agar plates. The QIAamp DNA blood maxikit (Qiagen GmbH, Hilden, Germany) was used to extract DNA from patient samples (blood culture flasks) that had signaled positivity in the BacT/ALERT blood culture system (bioMérieux). The DNA was extracted from 10 ml of blood and eluted in a total of 2 ml of elution (AE) buffer.

Capture hybridization.

Synthetic ssDNA targets, 10⁵ copies, were diluted in different volumes of buffer ranging from 10 μl to 1,000 μl. A sample of 1 μl was drawn directly to VOCMA PCR, omitting the capture procedure. Capture was then performed on the remaining volumes by adding 10⁵ MyOne magnetic beads (coupled with the first specific VT primer-probe, as described in the supplemental material) per μl with shaking at 600 rpm at 50°C for 30 min in 1× PCR Gold buffer (AmpliTaq Gold, Roche, Branchburg, NJ). After hybridization, the samples were rinsed twice in PCR buffer by fixing the magnetic beads with magnets (MagnaBot 96 magnetic separation device; Promega, Madison, WI). After the final rinse, the beads, with the captured targets, were resuspended in the PCR mix. PCR was performed with the beads present in the reaction. TMAC (3 M tetramethylammonium chloride, 0.1% Sarkosyl, 50 mM Tris-HCl [pH 8.0], 4 mM EDTA [pH 8.0]; Sigma) was used as a PCR inhibitor to a final concentration of 2.7 M in samples containing 0.2 × 10⁶ synthetic target molecules per μl. Five microliters of the sample containing TMAC went directly to VOCMA PCR while 5 μl went through the capture procedure before going to VOCMA PCR (as described above).

Amplification with the 7-plex gastro VOCMA.

Two microliters of 0.5 × 10⁵ copies/μl of synthetic ssDNA target, or 2 μl of extracted nucleic acid was added directly to a 23-μl one-step reverse transcriptase PCR (RT-PCR) master mix or to 5.0 × 10⁵ Dynabeads MyOne microspheres coupled with the first specific VT primer-probe to undergo capture hybridization (see above). The one-step RT-PCR master mix contained a final concentration of 1× iScript buffer (Bio-Rad, Hercules, CA), 0.5 μl iScript reverse transcriptase, 300 nM generic first primer, 500 nM biotinylated generic second primer, and 50 nM each of the 14 first and second specific VT primer-probes to create a 7-plex gastro VOCMA mixture (see Table S3 in the supplemental material). Amplification was carried out on an MJ Research PTC-100 Peltier thermal cycler (sold by Scandinavian Diagnostic Services, Falkenberg, Sweden) as follows: 50°C for 20 min, 95°C for 5 min, followed by 10 cycles of 95°C for 15 s, ramping at 0.1°C/s from 75°C to 65°C, 65°C for 1 min, followed by 40 cycles of 95°C for 15 s, 52°C for 30 s, 60°C for 30 s, followed by 60°C for 5 min, 95°C for 1 min, and 4°C until the next step. A sample of 20 ng/μl yeast tRNA (Sigma) was added as a no-template control (NTC) in the RT-PCR.

Amplification with the 22-plex sepsis VOCMA.

Two μl of 0.5 × 10⁵ copies/μl of synthetic ssDNA target or 2 μl of extracted nucleic acid from positive blood cultures, or other cultures or colonies of bacteria or fungi, was added to a 23-μl master mix containing a final concentration of 2.5 mM MgCl₂, 0.8 mM deoxynucleoside triphosphate (dNTP), 2.5 μl 10× AmpliTaq Gold buffer, 0.2 μl (5 U/μl) of AmpliTaq Gold enzyme (AmpliTaq Gold, Roche, Branchburg, NJ), 300 nM generic first primer (16 nt long), 500 nM biotinylated generic second primer (16 nt long), and 5 nM each of the 42 first and second specific VT primer-probes (see Table S2 in the supplemental material). The fungi Candida albicans and C. dubliniensis used the same VT primer-probes but different specific VT detection probes. Amplification was carried out as follows: 95°C for 9 min, followed by 10 cycles of 94°C for 30 s, 75°C for 1 s, ramping at 0.1°C/s to 62°C, 62°C for 1 min 30 s, 72°C for 30 s, followed by 55 cycles of 94°C for 20 s, 53°C for 30 s, 72°C for 30 s, followed by 72°C for 7 min, 95°C for 2 min, and 4°C until the next step. A sample of 0.1× Denhardt's solution was added as an NTC in the PCR.

Hybridization of biotinylated target DNA to detection probe-linked xMAP beads.

Five microliters of biotin-labeled VOCMA amplified target was mixed with hybridization buffer comprising 33 μl 4.5 M TMAC buffer, 12 μl 1× Tris-EDTA (TE) buffer (pH 8.0), and 0.05 μl (∼2,500 microspheres) of each probe-coupled xMAP bead. The coupling procedure is detailed in the supplemental material. The gastro VOCMA had seven unique beads, each coupled with one of the 7 specific VT detection probes (46 to 68 nt long), and the sepsis VOCMA had 22 unique beads, each coupled with one of the 22 specific VT detection probes (50 nt long). The mixture was heated at 95°C for 2 min, followed by hybridization at 50°C for 30 min with shaking at 600 rpm on a Thermostar (BMG LabTech, Offenburg, Germany) microplate incubator. After centrifugation (5 s, Galaxy minicentrifuge; VWR International), 40 μl was gently aspirated and a mixture of 38 μl 3 M TMAC-TE hybridization buffer with 2 μl (0.05 mg/ml) of streptavidin-R-phycoerythrin (Qiagen, Hilden, Germany) was added. The tubes were further incubated at 50°C for 15 min before analysis for internal bead and R-phycoerythrin reporter fluorescence on the Luminex 200 flow meter (Luminex Corporation, Austin, TX). The quantity of the biotinylated target that hybridized to probe-linked beads was measured as median fluorescence intensity (MFI). The minimum number of beads analyzed per type of bead was set to 100.

Sequencing of VOCMA amplimers.

The VOCMA PCR products of rotavirus and the norovirus-positive samples were purified using the QIAquick PCR purification kit (Qiagen GmbH, Hilden, Germany). The BigDye terminator v3.1 cycle sequencing kit (Applied Biosystems) was used for sequencing 2 μl of purified PCR product, 3.5 μl of 5× BigDye buffer, 0.32 μl of either forward or reverse primer (10 pmol/μl), 1 μl of BigDye mix, and 18 μl of H₂O. The sequencing PCR program was 95°C for 90 s, followed by 35 cycles of 96°C for 10 s, 50°C for 5 s, 60°C for 2 min, and 4°C indefinitely. The sequence readout was made in an Applied Biosystems 3730 XL DNA analyzer machine. Sequencing primers were the generic first and second primers from the gastro VOCMA or sequencing primers for norovirus genogroup II (see Table S3 in the supplemental material).

RESULTS

To illustrate the concepts behind VOCMA (detailed in Table S1 in the supplemental material), two panels were developed: (i) a 22-plex sepsis VOCMA to demonstrate variation tolerance by covering groups of resistance genes (e.g., CTX-M group I) or targeting genes that are normally used for genotyping (e.g., spa gene encoding protein A in Staphylococcus aureus), as well as comprising a high degree of multiplexity, and (ii) a 7-plex gastro VOCMA to demonstrate that the method can detect RNA viruses, i.e., variable targets.

The multiplexity, specificity, and sensitivity of the sepsis VOCMA panel.

A 22-plex sepsis VOCMA panel was created against bacteria and fungi known to be a common cause of sepsis (see Table S2 in the supplemental material). Reduction of the concentration to 10 nM or less of each specific VT primer-probes resulted in a reduction of primer dimers and increased specific signals (see Fig. S1A and B in the supplemental material). All detection probes showed strong specificity, with no or minimal primer dimer formation, using synthetic targets (see Fig. S2A and B). Two unexpected signals were occasionally seen, possibly due to contamination, but otherwise the specific target and probe combination always showed the highest signal. Nucleic acid from different bacterial cultures from clinical samples and reference strains was extracted and analyzed. The 22-plex VOCMA has three target genes for detecting S. aureus. Methicillin-resistant S. aureus (MRSA) harbors the MecA gene, leading to simultaneous VOCMA amplification of four genes and therefore four median fluorescent intensity (MFI) signals from the specific probes (Fig. 2). Further, different Escherichia coli and Klebsiella strains harboring antibiotic resistance genes belonging to CTX-M groups I or IV (14) were accurately detected. Conclusively, the 22-plex VOCMA test showed high specificity (Fig. 2).

Fig 2.

Specificity of sepsis VOCMA against cultured bacteria and fungi. A 22-plex VOCMA experiment on cultured strains of bacteria and fungi, with or without antibiotic resistance, from reference strains or strains from patient samples. The panel of 22 probes is indicated by different colors. The x axis shows the cultures with different colors indicating different detection probes hybridizing. The y axis shows the MFI.

The analytic sensitivity was tested for all 22 synthetic ssDNA targets in a 10-fold dilution series with 10⁷ to 10⁰ molecules per each reaction and resulted in 10¹ to 10³ synthetic targets per reaction (19 out of 22 targets) (see Table S6 in the supplemental material). It is noteworthy that the bead-bound probe displayed highly specific signals (see Fig. S3A and B in the supplemental material) even when the amplimers were very weakly detected in the ethidium bromide (EtBr)-stained agarose gel (see Fig. S3C in the supplemental material). During the expansion of the panel (from 5 sets to 10 sets and 22 sets of primer-probes and detection probes), there were no signs of a decrease in analytic sensitivity (data not shown). Higher multiplexities have not been tested with VOCMA, but it is likely that the limit has not yet been reached.

Blood culture-positive samples were tested and showed that the MFI signal reached a plateau using 2 μl of nucleic acid extract, while using 5 μl tended to decrease the signal more than increase it (Fig. 3). All probes had a reproducible signal maximum, which differed among the specific probes. A small test of the clinical sensitivity, with a total of 11 different blood culture-positive samples and one negative blood control, showed that the negative control became negative and seven of the positive samples were detected correctly (two were false negative). As mentioned previously, two samples occasionally gave a false-positive signal for both Enterococcus faecium and E. coli, probably due to contamination (data not shown).

Fig 3.

Sepsis VOCMA with blood culture-positive samples. A 22-plex VOCMA experiment on blood culture-positive samples from patients with suspected sepsis. The y axis shows the MFI, and the x axis shows the samples as duplicates with different amounts of purified nucleic acid added to the VOCMA PCR (2 μl and 5 μl of extract, respectively). The panel of 22 detection probes is indicated by the different colors in the figure.

Fidelity of detection of unequal copy numbers of simultaneously present targets.

The VOCMA system was tested for its ability to detect the presence of targets of different concentrations (see Fig. S4 in the supplemental material). Synthetic targets of the protein A (Prot A) and coagulase (Coa) genes were coamplified with equimolar or 100-fold difference ratios. The equal ratio resulted in approximately equal detection signals, representing the situation when several genes with the same copy number (from the same chromosome or genome) are amplified. Interestingly, the 100× difference in ratio still allowed detection of the target, with a lower signal for the target present in the smaller amount, roughly in proportion to the target copy number in the reaction.

The multiplexity, specificity, and sensitivity of the gastro VOCMA panel.

The 7-plex gastro VOCMA is designed to detect the most common pathogens causing viral gastroenteritis, including highly variable RNA viruses. Although the method detects most of the intended targets in clinical samples, it would benefit from further optimization and inclusion of an internal control. The MFI signals during optimization of the primer concentration of the first generic primer (200 to 500 nM) and the second generic biotinylated primer (300 to 600 nM) did not vary significantly, while higher concentrations than 50 nM each of the long specific VT primer-probes caused primer dimers and thereby lesser sensitivity (data not shown).

The specificity of the gastro VOCMA using synthetic targets was high, with no cross-hybridization (see Fig. S5 in the supplemental material). Furthermore, the analytic sensitivity was tested in a 10-fold dilution series of 10⁵ to 10⁰ copies of the synthetic targets and resulted in positivity in dilutions with 10⁰ to 10¹ copies/reaction for rotavirus and adenovirus, 10² copies/reaction for astrovirus and poliovirus, 10² to 10³copies/reaction for norovirus genogroup II, and 10⁴ copies/reaction for norovirus genogroup I and sapovirus (data not shown), which is comparable with the analytic sensitivity of the 22-plex sepsis VOCMA. The signal of MFI reached a plateau at ≥10⁵ copies/reaction (see Fig. S6 in the supplemental material).

To better understand the clinical sensitivity, 64 successively collected samples from patients with gastroenteritis (February 2012) were tested (see Table S7 in the supplemental material). The RT-PCR for norovirus genogroup I and genogroup II revealed that 26 samples were positive for norovirus genogroup II. The VOCMA detected 12 of the 15 samples of which contained >20,000 copies per ml, while the 11 samples with <11,000 copies per ml were not detected, correlating with the analytic sensitivity for genogroup II (see Fig. S6 in the supplemental material). However, astrovirus (MFI 539, confirmed by sequencing), rotavirus (MFI 667, confirmed by sequencing), adenovirus (MFI 385), and norovirus genogroup I (MFI 139) were detected in four RT-PCR norovirus genogroup II-negative samples, and a coinfection of genogroup I (MFI 264) and genogroup II (MFI 421) was detected in one sample.

Detection of rotavirus in clinical samples shows the advantages of multiplex testing and tolerance for variation.

Clinical fecal samples from patients with gastroenteritis, previously tested negative with a specific norovirus genogroup I and genogroup II quantitative RT-PCR (qRT-PCR) (4), were analyzed with the gastro VOCMA (see Fig. S7 in the supplemental material). Of the 94 samples, 14 previously unrecognized rotavirus-positive samples (MFI signals 825 to 4477) were found and confirmed by sequencing. A BLASTN search showed a high similarity to human rotavirus A. Two norovirus-positive samples were also detected, but for unknown reasons, sequencing was unsuccessful. The 14-nt overlap of the rotavirus detection probe and first specific VT primer-probe did not create problems, because they were oriented in the same sense. If any remaining free first specific VT primer-probe would bind the biotinylated target the detection probe would still have 53 nt of its total length of 67 nt to hybridize to the target.

The sequences of the rotavirus-positive VOCMA amplicons resulted in six different sequences having 15-nt variant positions, of which 10 positions were within the target region of the detection probe (Fig. 4A). Importantly, the pattern of high variation seen in the 212 aligned sequences from GenBank (Fig. 4B) and the positions of the variations within the sequenced samples (Fig. 4A) matched with the 5 dInosines and 7 degenerations in the detection probe, leaving at most only two uncovered mismatches. The variation-tolerant design obviously sufficed for the detection of rotaviruses.

Fig 4.

Variation tolerance of the rotavirus probe. Amplicons from the rotavirus-positive samples (see Fig. S8 in the supplemental material) were sequenced. (A) The resulting six sequence variants of 13 successfully sequenced clinical samples were aligned in ClustalX 1.83. Red letters in the rota detection probe sequence show the positions of nucleotide degeneration and dInosine. (B) A total of 212 sequences reported to GenBank were aligned utilizing the program ConSort (see the supplemental material). Black bars indicate the average percentage of conservation in each nucleotide position in the GenBank sequences. Gray bars indicate the nucleotide positions that caused mismatching with the probe.

Detection of norovirus in clinical samples shows tolerance for variation.

Different variants of a virus can circulate in a community at the same time, and the virus is also changing over time. Twenty-one norovirus genogroup II samples, known to be positive in RT-PCR (4), were selected from a 2-year period (January 2009 to January 2011). The VOCMA could detect all but one of the norovirus PCR-positive samples (see Fig. S8 in the supplemental material). Sequencing showed that the probe covered the variations in all the eight different variants of the 13 successfully sequenced samples. The adenovirus coinfection detected in one of the samples was confirmed by adenovirus RT-PCR (threshold cycle [C_T], 34.0; 271 copies/ml). The tolerance for variation was also demonstrated with the ability to detect different cultured enteroviruses (see Fig. S9 in the supplemental material).

Capture.

There are two aspects to the concept of “capture” in VOCMA. The first aspect is the enrichment of specific nucleic acid targets from a large volume. In Fig. 5, the trend of decreasing MFIs with increasing sample volumes from samples not subjected to capture is clear (the “no capture” series), while the samples that were subjected to the capture procedure showed higher MFIs over the volume range. The signal with capture was half as strong as that from the noncapture 10-μl solution at a 20-fold higher dilution (200 μl). The second aspect is the rinsing of the nucleic acid of interest from PCR inhibitors (demonstrated with TMAC as inhibitor in Fig. S10 in the supplemental material) and from excess nontarget nucleic acid. A clean sample resulted in a clear MFI signal, while a TMAC-containing sample failed altogether. However, subjecting the TMAC-containing sample to the capture procedure before PCR rescued the sample from inhibition of the PCR.

Fig 5.

Capture in increasing volumes. Copies (10⁵) of synthetic norovirus target were diluted in 10 μl to 1,000 μl. One microliter was drawn from each dilution and processed according to the VOCMA protocol without the capture procedure (no capture series). The remaining sample volumes were processed to include the capture procedure (capture series). The presented data are the means of two separate runs.

DISCUSSION

VOCMA combines several features that are sought after in a diagnostic tool: it is relatively sensitive, specific, tolerant to variations in the target, and multiplex, and it includes removal of inhibitors and enrichment of the target from a relatively large extract volume. Moreover, it is relatively fast (PCR, 3 h; hybridization, 45 min; Luminex, 15 min, excluding hands-on time).

Multiplex testing is gradually becoming important in clinical microbiology. Microarrays (21) and massively parallel (“deep”) sequencing (e.g., see reference 13) have great potential but still have limited sensitivity and are time consuming. Multiplex PCR (6, 7, 20) and multiplex padlock probes with rolling circle amplification (24) are close to routine diagnostic use. The Templex multiplex PCR technique (3) has similarities with VOCMA but handles specific and generic primers differently.

Probes in a suspension array have faster hybridization kinetics (10) than probes on a surface. A prevalent form of suspension array is based on probes bound to Luminex xMAP color-coded beads (e.g., see references 9, 10, and 12). Unlike real-time PCRs, which are performed in closed tubes, a Luminex-based readout requires the opening of tubes after amplification, creating a contamination risk. Although we did not encounter significant amplimer contamination during the development of VOCMA, this step should be performed in a hood in a post-PCR environment, separated from the pre-PCR facility. An advantage of VOCMA over the Luminex xTAG system (12) is that it uses a bead-bound sequence-specific probe, which considerably diminishes the tendency for false-positive signals. VOCMA uses five hybridization events, of which three are target-specific, leading to a high overall specificity (15). Moreover, unlike the xTAG system, VOCMA amplifies and labels in the same tube, without having to open it. It uses a change in the annealing temperature to activate the short biotinylated second generic primer.

Combining a highly multiplex PCR with Luminex technology allows for a multiplex system where the PCR design skill sets the limit for its performance. Standard PCR involves one target sequence and two primer sequences, all in solution. If the concentration of each primer is high (>200 nM), and/or there are many different primers, the likelihood of unwanted interactions increases. In VOCMA, dividing amplification into two steps (target specific and generic) allows for the concentration of free target-specific VT primer-probes to be decreased to as low as 1 nM (see Fig. S1A and B in the supplemental material), simultaneous high multiplexity and specificity, and comparatively high sensitivity (see Table S1 in the supplemental material). The strategy of using a low concentration of the specific primers contributes to the specificity, since the production of primer dimers from the long and numerous specific primers is suppressed (see Table S1). The temperature switch allows all primers to be added as one mix in the tube, which minimizes the risk of contamination (see Table S1) (2). As presented here, the concentration of each specific primer was 5 nM in the 22-plex sepsis VOCMA experiment and 50 nM in the 7-plex gastro VOCMA experiment. To achieve an excess of a single-stranded biotinylated product, the biotinylated generic primer was in excess and the amplimers were denatured before hybridization (see Table S1) (23).

The feature of using three different target-specific oligonucleotides in VOCMA offers freedom of design. A VOCMA panel can include components with different degrees of variation tolerance and different combinations and numbers of specific primer-probes and detection probes targeting highly conserved and more variable targets. This was partially demonstrated in our sepsis panel, which covered both fungal detection using the intervening transcribed sequences and bacterial genomic and antibiotic resistance gene detection components.

In a complex multiplex assay, it can be difficult to control all possible sources of false negativity. By using the same oligonucleotide for capture and primary priming (the first specific VT primer-probe) (Fig. 1), an additional diversity-constricting step is avoided. Target nucleic acid will be greatly enriched versus nontarget nucleic acid, and the first two PCR rounds will be bead bound, which will decrease mispriming (see Table S1 in the supplemental material). The MyOne beads were small enough to stay in suspension during the reactions but could still be collected by a magnetic separator.

The demonstrated capture and enrichment process illustrated that, judging from MFI values (which are not linearly related to the amount of target nucleic acid), 50% of target nucleic acid was recovered from extract volumes that increased up to 20-fold. It is likely that capture from even larger volumes could be achieved by increasing the concentration of the magnetic beads carrying the first specific VT primer-probe. Capture should be especially useful for detection of bacteria in whole blood, of low abundance protozoa and virulence factor genes in feces, and of food-borne pathogens, where sample volumes of over 10 ml may have to be used, giving large extract volumes. The nucleic acid concentration can be prohibitively high in such extracts. Others showed that a MyOne magnetic bead-bound 70-nt long capture probe improves the limit of detection (LOD) 10-fold and 100-fold compared to using a 50-nt or 25-nt long capture probe, respectively (16), corroborating our technique. We also show that long LNA-containing primer-probes are especially suitable for capture.

Target variation, especially in RNA viruses, is a major problem in molecular diagnostics. The recently published NucZip algorithm for long probe hybridization (15), together with the use of the computer program ConSort for mapping the pattern and frequency of variation, was instrumental in the design of VT primer-probes and detection probes. The greatest conservation is placed at the 3′ end for the primer-probes, and both the primer-probes and the detection probes are typically 50 to 70 nt long to provide a high affinity to the target and tolerance to mismatches to the target sequence. The first hybridization event of the specific part of the primer-probe and the pathogen target will often include mismatching nucleotides in several positions, but after the third and fourth PCR cycles, many primer-probes, including the generic part of the primer, will fit the amplimers perfectly. One of the basic rules for VOCMA is that a primer-probe and detection probe should have at least one stretch of nine perfectly matching nucleotides or at least two stretches of five perfectly matching nucleotides (15). Variable positions can be covered by a universal base analog (UBA), e.g., dInosine, and, to a certain extent, the natural base guanosine. However, UBAs tend to decrease the affinity and cannot be used too generously (they should generally be kept below 30% of the length). The choice of target region is dependent on the frequency of variation and, more importantly, on the pattern of variation, since longer segments of perfect matches are preferred and dInosines are especially sensitive to neighboring mismatches (15). Degenerated (redundant) nucleotide positions can also increase variation tolerance and are included in recent versions of NucZip (A. Danielsson, C. Öhrmalm, and J. Blomberg, unpublished data). If the nucleotide positions are neighbored by invariant stretches of at least 3 nucleotides on both sides, there is hybridization cooperativity between degenerate probe positions (15). Thus, the primer-probe and detection probe design algorithm must balance the content of UBAs, guanosine, degenerations, and variable positions against each other to obtain both maximal mismatch tolerance and high affinity and specificity to the target.

Three molar TMAC hybridization buffer increases the melting temperature (T_m) by increasing the binding strength of A·T base pairs to approximately that of G·C base pairs (5, 22). In our sepsis VOCMA panel, all the probes were the same length (50 nt), which means that they will bind with comparable efficiency to a perfect matching target at the same hybridization temperature. The probes in the gastro VOCMA were variable in length (48 to 68 nt) because of the scarcity of conserved target stretches.

The generic first and second primers have artificial sequences which do not have homologous or heterologous complementarity to the other primers and probes. They are only utilized in solution, and their function is to finalize, and to take the brunt of, the amplification. Amplification factors of 10⁵ to 10⁸ are typically achieved with these primers.

At first sight, it appears that the generic primer pair could also amplify falsely. The very high amplification factors and their generic nature could be suspected to cause nonspecificity. However, each target-specific primer-probe and each generic primer must act in succession in an interlocked way to start the second most productive phase of amplification, provided by the generic primers (Fig. 1). The specificity is further enhanced by the placement of the label on the second generic primer. The sequence of the generic primers was tested in silico to avoid hybridization from the human genome and the primers and probes included in the panel. Unspecific amplification from the primer-probes or from the generic biotinylated primer cannot yield unspecific reactions since target-specific detection probes are used in the detection step. However, it will reduce the efficiency of the PCR and thus reduce its sensitivity.

The limits of detection of the 22-plex sepsis VOCMA panel were 10 to 1,000 copies per reaction for 19 of the targets. The differentiation of artificially contaminating staphylococci in sepsis cultures (coagulase-negative bacteria like Staphylococcus epidermidis) from more likely pathogens such as S. aureus relies on the detection of the nuc gene. In this panel, two more genes, the coagulase and protein A genes, were included for the detection of S. aureus. The MecA gene is important for the diagnosis of MRSA. A toxin gene, tss1, was also included to allow rapid identification of highly pathogenic toxin-producing S. aureus strains.

The antibiotic resistance genes from CTX-M groups I and IV and the AmpC genes were also included. To cover the variation in AmpC genes, the primers and probes were divided into three sets: AmpC I, AmpC II, and AmpC III. The CTX-M primers and probes were made to be specific for either CTX-M group I or group IV while still remaining tolerant to variations within the group. All components of the sepsis panel were proven to have enough sensitivity and specificity for rapid identification of sepsis bacteria after overnight culture of patient samples.

The limits of detection of the 7-plex viral gastro VOCMA panel were around 100 to 1,000 copies per reaction. Specificity was proven by detection of the synthetic targets and viruses in clinical samples. Although the norovirus-specific primer-probes and detection probe employed here cannot be considered a “final” design, the system did find noroviruses in samples which had been negative in a conventional real-time PCR. The clearest demonstration of the usefulness of multiplex testing of gastroenteritis was the frequent detection of rotaviruses in samples submitted for norovirus testing and the detection of astrovirus, rotavirus, adenovirus, and norovirus genogroup I in the small clinical sensitivity study. Thus, the gastro VOCMA panel demonstrated the utility of both variation tolerance and multiplex testing.