key: cord-0889214-lwvvnlmv
authors: Cannon, Gemma A.; Carr, Michael J.; Yandle, Zoe; Schaffer, Kirsten; Kidney, Rachel; Hosny, Ger; Doyle, Allen; Ryan, John; Gunson, Rory; Collins, Terry; Carman, William F.; Connell, Jeff; Hall, William W.
title: A low density oligonucleotide microarray for the detection of viral and atypical bacterial respiratory pathogens
date: 2009-07-26
journal: J Virol Methods
DOI: 10.1016/j.jviromet.2009.07.005
sha: f372a72948ea2a6ed9713e677f78e365992132ad
doc_id: 889214
cord_uid: lwvvnlmv

Acute respiratory tract infections are a major cause of morbidity and mortality worldwide and exert a considerable economic burden on healthcare systems. Acute respiratory tract infections of the upper and lower respiratory tract are caused by a wide variety of viral and bacterial pathogens, which require comprehensive laboratory investigations. Conventional serological and immunofluorescence-based diagnostic methods for acute respiratory tract infections lack sensitivity when compared to polymerase chain reaction (PCR)-based approaches and the development of new diagnostic methodologies is required, to provide accurate, sensitive and rapid diagnoses. In the present study, a PCR-based low density oligonucleotide microarray was developed for the detection of 16 viral and two atypical bacterial pathogens. The performance of this DNA microarray-based analysis exhibited comparable sensitivities and specificities to multiplex real-time reverse transcription polymerase chain reactions (rtPCRs) confirming the potential diagnostic utility of the method. In contrast to routine multiplex PCR, the microarray incorporates an intrinsic redundancy as multiple and non-identical probes per target on the array allow direct intra-assay confirmation of positives. This study demonstrates that microarray technology provides a viable alternative to conventional serological-based approaches and multiplex PCR for pathogen identification in acute respiratory tract infections.

Acute respiratory tract infections are the most common reason for general practitioner visits and acute admission to hospital during the winter season irrespective of age or gender (Elliot et al., 2008) . In 2002, lower respiratory tract infections accounted for 3.9 million deaths worldwide; 6.9% of all deaths that year (WHO, 2004) . Viral and bacterial co-infections are frequently observed using PCR-based investigation, which in many cases are not detected by current serological-based and direct antigen detection methods. Therefore, the development of new diagnostic methods is required to enable the detection of a broad range of pathogens. Indeed an accurate, sensitive and rapid differential diagnosis can influence patient management, reduce potentially inappropriate antibiotic use and aid infection control measures in institutional settings.

The current serological-based methods for the diagnosis of acute respiratory tract infections are limited in assay specificity and sensitivity, often resulting in underdiagnosis, e.g. antigen detection of the Adenoviridae (Arnold et al., 2008) . However, viral antigen detection tests are used increasingly because of low demand on equipment and cost effectiveness (Grandien, 1996) . In addition, virus isolation is slow, expensive and labour intensive; often requiring 1-5 days incubation to detect or confirm the absence of cytopathic effects. In addition, with decreasing resources fewer clinical laboratories have the appropriate facilities and expertise to undertake culturebased investigations (Koenig et al., 2001) . Culture-based methods are also used for bacterial detection and are also hampered by slow turnaround times for reporting of results and insufficient sensitivity (Peters et al., 2004) . Serological profiling is also used often, however, this can require more than two weeks for antibodies to develop and sequential samples, which are often not collected and are usually required to confirm the diagnosis (Nilsson et al., 2008; Wong et al., 2008) .

Advances in molecular diagnostic methods have enabled rapid diagnosis of infectious disease (Murata, 2008) . Molecular techniques offer several advantages over conventional serology-based methods, including speed, ease of standardisation, automation and high assay sensitivities and specificities (Dong et al., 2008) . To date, DNA microarray technology has been used predominantly for gene expression profiling studies in basic research applications (Hegde et al., 2000) . Clinical presentations of acute respiratory tract infections are diverse and non-specific; therefore an efficient laboratory assay that can detect a panel of common respiratory pathogens would be advantageous. In addition, large-scale screening may identify co-infections, which would not have been detected previously. Microarray-based analysis is a cost-effective approach that yields reproducible results and can allow replicate analyses in a single assay run. This is achieved by spotting of solid phase arrays with multiple copies of the same probe, alternative probes within the same target amplicon and/or mismatch probes allowing intra-assay confirmation of results that is not readily possible with multiplex rtPCR-based approaches.

A novel, non-fluorescent, low-cost, low density oligonucleotide microarray format has been developed which facilitates the use of DNA microarrays for clinical diagnostic purposes. The ArrayTube TM (AT) platform offers a fully automated system for clinical diagnosis. Several studies have used this platform for a variety of applications (Borel et al., 2008; Sachse et al., 2006 Sachse et al., , 2005 . The AT system represents a cost-effective platform involving 2.4 mm × 2.4 mm glass biochips, integrated into the bottom of standard 1.5 ml plastic microtubes. The chip may contain a number of different probe types depending on the specific application and allows all steps of the hybridisation reaction to be conducted within the AT vessel, obviating the requirement for a separate hybridisation chamber or other additional laboratory equipment. The hybridised target is visualised by enzyme-catalysed precipitation and the microtubes are read using a simple low cost, transmission scanner. A previous study revealed that specific hybridisation to virus-specific oligonucleotide probes can be obtained from a single PCR amplifiable target copy .

The development of an oligonucleotide microarray-based system for the detection, differentiation and subtyping of 18 viral and bacterial respiratory pathogens simultaneously, comprising 16 viruses and two atypical bacteria is described. PCR-derived amplicons from external quality assessment panels and known positive samples were used as targets to establish probe sensitivity. A validation panel was assembled, comprised of throat swab specimens, collected from adults who presented with symptoms of respiratory disease to a tertiary care hospital during the 2007-2008 winter season, to evaluate the potential role of the microarray assay compared to multiplex rtPCRs.

Oligonucleotide probe genomic target regions are listed in Table 1 . Oligonucleotide probes were evaluated using Primer3 (http://frodo.wi.mit.edu/). Probes on the array were between 18 and 35 nucleotide bases in length and were modifications of existing TaqMan assays. The most important parameter for modification of existing published TaqMan probes was that they would have similar T m values so that hybridisation would be uniform for all denatured amplicons on the array. All probes have a T m value close to the average of all probes and T m values were calculated using Primer3 which uses the SantaLucia method for T m calcula-tion (SantaLucia, 1998 

Oligonucleotide probes (Metabion, Martinsried, Germany), ∼18-35 bp in length, were spotted in fourfold redundancy on glass arrays using a proprietary technology (ArrayTube, Clondiag Technologies, Jena, Germany).

Biotinylation end-point PCR primer sequences and multiplex set-up are shown in Table 2 . The OneStep RT-PCR kit (Qiagen, Crawley, UK) was used according to the manufacturing instructions with the modified addition of 0.15 mM dTTP and 0.2 mM dATP, dGTP, dCTP and 0.05 mM biotin-21-dUTP per reaction (Annecis Ltd., Lancaster, UK) for target labelling. PCR conditions were 1 cycle of 50 • C for 30 min, 1 cycle of 95 • C for 15 min, 50 cycles of 94 • C for 30 s, 55 • C for 1 min, 72 • C for 1 min and final extension for 1 cycle of 72 • C for 10 min on a PTC-200 thermocycler (MJ Research, Waltham, MA, USA).

The microarrays were washed in nuclease-free water (Promega, Madison, WI, USA) and hybridisation buffer [250 mM NaPOi, pH 7.2 (Camida Ltd., Tipperary, Ireland); 4.5% SDS; 1 mM EDTA, pH 8.0; 1× SSC] (Sigma, Dublin, Ireland) for 5 min each at 55 • C at 550 rpm in the Thermomixer comfort (Eppendorf AG, Hamburg, Germany). A 0.5 l volume of each target solution in a final volume of 100 l of hybridisation buffer was heated at 95 • C for 5 min and then chilled on ice. The biotinylated target reactions were hybridised to the microarrays at 55 • C for 1 h at 550 rpm. DNA arrays were washed in solutions of decreasing stringency [2× SSC, 0.01% Triton X; 2× SSC; 0.2× SSC] (Sigma, Dublin, Ireland) for 5 min each at 30 • C at 550 rpm. Blocking was performed using 2% BSA in 6XSSPE, 0.005% Triton X (Sigma, Dublin, Ireland) for 15 min at 30 • C at 550 rpm. A 1 in 5000 dilution of streptavidin-horseradish peroxidise (SA-HRP) conjugate (Pierce, Dublin, Ireland) was added for 15 min at 30 • C at 550 rpm. Washing was repeated in solutions of decreasing stringency [2× SSC, 0.01% Triton X; 2× SSC; 0.2× SSC] (Sigma, Dublin, Ireland) for 5 min each at 30 • C at 550 rpm to remove unbound SA-HRP conjugate. Visualisation of the hybridised targets was achieved by incubation with 100 l tetramethyl bendzidine (TMB) (KPL Ltd., MD, USA) at 20 • C for 10 min.

Hybridisation signals were measured at 20 • C after 10 min using the ATR03 AT DNA microarray transmission scanner (Clondiag Technologies, Jena, Germany). Quantitative staining values (QSVs) with local background correction were obtained for each probe spot via the Iconoclust software, version 2.3 (Clondiag Technologies, Jena, Germany). The criteria for assignment of hybridisation patterns were as follows:

Background-corrected signal intensities were given as NI = 1-M/BG, with NI being Normalised Intensity, M is Mean spot intensity, and BG local background intensity. Spot intensities were measured as light transmission, with M values ranging from 1 for complete transmission (background, weak spots) to 0 for complete absorption (dark spots). Normalised intensities ranged Corless et al. (2002) . d Kuypers et al. (2005) . e Dare et al. (2007) . f Gunson et al. (2005) . g Lu et al. (2008) . h Gullsby et al. (2008) . between 0 and 1. QSVs were reported as the median of all four replicate signals for each probe.

Real-time PCR primer sequences and multiplex set-up are shown in Table 2 and carried out using the Invitrogen Superscript III one-step qPCR system according to the manufacturer's instructions (Invitrogen, Paisley, UK). For an internal control for the extraction efficiency, sample addition and the absence of PCR inhibitors in each specimen we amplified a human gene (k-ras) from all throat swabs.

Biotinylated PCR products were purified using the High Pure PCR product Purification Kit according to manufacturers instructions and eluted in 50 l (Roche Applied Science, Mannheim, Germany). DNA concentration was determined using the NanoDrop ND-1000 spectrophotometer (Thermo Scientific, DE, USA) and all DNA target solutions normalised to 50 ng/l. A series of 10-fold dilutions (5000-5 pg/l) were prepared and subjected to DNA microarray analysis.

Throat swabs were collected from adult patients with respiratory symptoms presenting at an emergency department in a tertiary care hospital in Dublin during the 2007-2008 winter season. Swabs were stored in viral transport medium at −80 • C before processing. DNA and RNA were extracted from respiratory specimens using the QIAamp Virus BioRobot MDx kit on the BioRobot MDx workstation, according to the manufacturer's instructions (Qiagen, Crawley, UK).

We have developed and validated a diagnostic PCR-based low density oligonucleotide microarray for the detection of 16 viral and 2 bacterial respiratory pathogens. In initial studies the oligonucleotide microarray probe specificity was assessed using known positive specimens (i.e. either external quality assessment samples or clinical material confirmed by rtPCR and/or sequencing) and then analytical sensitivity was evaluated by serial dilution of biotinylated target amplicons and hybridisation to the array. Table 3 lists the QSVs for each of the individual pathogen probes, the mean internal staining controls, the background signal for the assay, the background signal for an equivalent 'negative' assay as well as the factor by which the signal is greater than the background level. Excluding Coronavirus OC43 which had a QSV of 0.55, the remaining probes exhibited QSVs between 0.62 and 0.95, with the median QSV being 0.77. All internal staining controls had a QSV of ≥0.84. The minimum factor by which the probe signal was greater than the background was a factor of 80 (coronavirus OC43) and the maximum was 529 (RSV subtype B). Certain pathogens that had multiple probes of differing sequences showed a greater specificity for a single probe sequence, e.g. Parainfluenza-3; 0.85, 0.81 and 0.73 for all 3 probes respectively. The analytical sensitivity of the microarray was evaluated using a serial dilution of individual targets of known concentration. The majority of the probes exhibited a reduced sensitivity below 50 pg of target DNA/cDNA and a complete loss of signal below 5 pg of target DNA for all probes (data not shown). Some signal was observed at 5 pg of target DNA for the probes RSV subtype A, Rhinovirus-03 and -04, with Rhinovirus-03 still yielding a QSV of approximately 0.63.

A cohort of clinical specimens were tested by both multiplex rtPCR and DNA microarray analysis (n = 50) to assess the performance of the DNA microarray for clinical investigation. The results are summarised in Table 4 Table 4 illustrates the clinical sensitivity of the DNA microarraybased assay. Real-time PCR C t values are listed for all positive specimens alongside DNA microarray results. All six targets not detected by DNA microarray analysis were weakly positive via rtPCR analysis, and had C t values >40 with 4/6 having C t values of ≥45. For all of the clinical samples tested, the QSVs were ≥0.60 and had a background of ≤0.10. We next attempted to determine the clinical relevance of the findings from the rtPCR and microarray analysis. Only the specimens positive by DNA microarray analysis and with C t values <40 in rtPCR were included in this analysis (n = 17) ( Table 5 ). Among these 17 positive test results two patients had a concomitant bacterial respiratory tract infections (M. pneumoniae). Antibiotic treatment had been implemented in 13 of the remaining 15 patients and 8 patients had been admitted to hospital. Two patients with an admitting diagnosis of pneumonia and an infiltrate on chest X-ray were positive for M. pneumoniae and their antibiotic treatment was therefore con-sidered appropriate. As the majority of other patients neither had an elevated peripheral white cell count nor significant chest X-ray infiltrates, it seems plausible that if a positive viral test result would have been available for clinicians less antibiotics would have been prescribed.

Current diagnostic methodologies for the detection of respiratory pathogens are cell culture isolation, antigen/antibody detection and immunofluorescence and conventional/rtPCR. However, not all of these methodologies have the sensitivity, specificity and turnaround time required to have an impact on clinical management. Variations in sample quality may impact on the viability of virus for culture and also the number of available cells for immunofluorescence investigation, whereas PCR is less impacted by poor sample quality. Therefore, molecular techniques are required for an accurate and rapid diagnosis. PCR tests are considered very sensitive, and our data has shown that the combined PCR and DNA microarray-based assay was capable of detecting and identifying 18 pathogens from throat swab specimens in a single run. Our study also shows that the DNA microarray assay for respiratory pathogens has: (i) the sensitivity required for the testing of clinical samples and (ii) the high specificity and capacity to achieve identification and differentiation of a pathogen(s) in a single test in a single working day.

Importantly, the redundant spotting of probes and multiple nonidentical probes per target on the array allows a direct intra-assay confirmation of positives. This ability to perform replicate analyses in a single assay run provides a major advantage over real-time technologies that are limited by the number of available dyes and the current detection technology platforms. In addition, precipitation staining utilised in the microarray approach described herein does not have problems with respect to signal stability, quenching effects and cross-talk between detectors which can be encountered with rtPCR. Most importantly, microarrays have a significantly greater multiplexing capability than rtPCR as presently a maximum of five targets can be analysed per well which in best practice would employ one filter for the internal control leaving only four filters to detect pathogens per reaction. In addition, there is increasing concern regarding the accuracy of the multiple formats compared to the amplification in a simplex reaction.

The PCR-based DNA microarray assay failed to detect a number of samples that were reported as weak positives by rtPCR; however these positives with C t values >40 are of dubious clinical significance. This technology allows for differential diagnoses and an advantage of the microarray is that it provides more clinically relevant information. This technology will inevitably impact on the management strategy of respiratory infections and potentially could be employed to limit outbreaks in institutions as well as to decrease empiric antibiotic usage where the pathogen responsible can be shown to be viral. The ability to attach large numbers of probes on each solid phase and the comparatively low cost per pathogen tested of such an oligonucleotide microarray-based assay compared to multiplex rtPCR suggests it may be a potential alternative for clinical testing. Furthermore, the increased capacity of such testing will broaden the approach to diagnosis and patient management. Timely molecular test results can contribute to good antimicrobial stewardship and help prevent secondary complications associated with antibiotic treatment, such as intravenous line infections or potential toxic/allergic side effects. Of the 7 patients with an Influenza A or B positive test result, two had been admitted to general medical wards. The admission of influenza patients into open wards poses a significant infection control hazard as this could potentially result in influenza outbreaks. Molecular testing of respiratory specimens could help initiate appropriate isolation and infection control measures to protect other patients.

To date, high-density spotted DNA microarrays have almost exclusively been used for research purposes for a broad spectrum of applications (Dankbar et al., 2007; Gruden et al., 2008; Scaria et al., 2008; Sidders et al., 2007) . However, this technology is also being employed in the clinical laboratory for diagnosis of viral pathogens in clinical specimens and can potentially become a frontline screening assay supplanting quantitative PCR in the management of herpesvirus and adenoviral infections in the immunocompromised patient population (Muller et al., 2009) . While high-density in situ synthesised DNA microarrays can offer a greater analytical sensitivity and can accommodate a greater number of probes for a more advanced approach to analysis, e.g. resequencing microarrays (Lin et al., 2007) and genotyping (Lodes et al., 2007) , low density DNA microarray-based analysis is a cost-effective assay concept. In the future, combining microarray hybridisation with high-throughput, automated DNA extraction and amplification will have a significant impact on high-throughout analysis. Regarding the role of spotted DNA microarrays in diagnostics, we have shown that a suitable low density platform is available for the development of diagnostic assays. Though less sensitive, this may paradoxically be of benefit in the identification of clinically significant levels of pathogen. The cost and ease of automation incorporating liquid handling and multiplexing capability makes this platform an attractive alternative technique for the standardisation purposes required in clinical diagnostics. This is specifically relevant when current commercial assays do not meet the test algorithms for a particular clinical laboratory and this may have an impact on the approach to specimen testing in the clinical laboratory and hospital setting. This will inevitably aid the detection, monitoring and therapeutic treatment of disease. DNA microarray-based analysis may thus be a potential alternative for clinical testing, providing more information on pathogens allowing better assessment of the viral aetiopathogenic agent(s) responsible for respiratory illness in the case viral and bacterial of co-infections.

Undiagnosed respiratory viruses in children

Direct identification of chlamydiae from clinical samples using a DNA microarray assay: a validation study

Development and evaluation of a 'real-time' RT-PCR for the detection of enterovirus and parechovirus RNA in CSF and throat swab samples

Diagnostic microarray for influenza B viruses

Human coronavirus infections in rural Thailand: a comprehensive study using real-time reverse-transcription polymerase chain reaction assays

Emerging pathogens: challenges and successes of molecular diagnostics

Optimized DNA microarray assay allows detection and genotyping of single PCR-amplifiable target copies

Acute respiratory infections and winter pressures on hospital admissions in England and Wales 1990-2005

Viral diagnosis by antigen detection techniques

Expression microarrays in plant-virus interaction

Simultaneous detection of Chlamydophila pneumoniae and Mycoplasma pneumoniae by use of molecular beacons in a duplex real-time PCR

Real-time RT-PCR detection of 12 respiratory viral infections in four triplex reactions

A concise guide to cDNA microarray analysis

Rapid and quantitative detection of human adenovirus DNA by real-time PCR

Comparison of Light-Cycler PCR, enzyme immunoassay, and tissue culture for detection of herpes simplex virus

Detection and quantification of human metapneumovirus in pediatric specimens by real-time RT-PCR

Using a resequencing microarray as a multiple respiratory pathogen detection assay

Identification of upper respiratory tract pathogens using electrochemical detection on an oligonucleotide microarray

Real-time reverse transcription-PCR assay for comprehensive detection of human rhinoviruses

Detection of herpesvirus and adenovirus co-infections with diagnostic DNA-microarrays

Respiratory syncytial virus infection in adults

Real-time PCR for diagnosis of human bocavirus infections and phylogenetic analysis

Polymerase chain reaction is superior to serology for the diagnosis of acute Mycoplasma pneumoniae infection and reveals a high rate of persistent infection

New developments in the diagnosis of bloodstream infections

The use of DNA microarray technology for detection and genetic characterisation of chlamydiae

DNA microarraybased detection and identification of Chlamydia and Chlamydophila spp

A unified view of polymer, dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics

Microarray for molecular typing of Salmonella enterica serovars

Quantification of global transcription patterns in prokaryotes using spotted microarrays

The World Health Report 2004-Changing History

Detection of a broad range of human adenoviruses in respiratory tract samples using a sensitive multiplex real-time PCR assay

The authors would like to thank Ralf Ehricht and Peter Slickers from Clondiag (Jena, Germany) for assistance in the design of oligonucleotide probes for the microarray. We are grateful to col-leagues in the Department of Virology, Barts and the London NHS trust for providing the sequences for the k-ras assay.