DNA microarray-based detection of Coxiella burnetii, the causative agent of Q fever

Background An easy-to-handle microarray assay based on the cost-effective ArrayTube™ platform has been designed for the rapid and unequivocal identification of Coxiella burnetii, the causative agent of Q fever. The gene targets include the chromosomally coded markers icd, omp/com1, and IS1111 as well as the plasmid coded markers cbbE and cbhE. Results A representative panel comprising 50 German C. burnetii isolates and 10 clinical samples was examined to validate the test. All tested isolates harboured plasmid QpH1 and were correctly identified, corresponding to 100% sensitivity. The assay’s limit of detection was 100 genome equivalents (GE) for icd, omp/com1, cbbE and cbhE and 10 GE for IS1111. Assay specificity was 100% as determined by analysing a panel of 37 non-Coxiella strains. Conclusions The present array is a rational assembly of established and evaluated targets for the rapid and unequivocal detection of C. burnetii. This array could be applied to the screening of vaginal swabs from small ruminants; screening of environmental samples e.g. on farms or screening of human samples.


Background
The causative agent of the zoonosis Q fever, Coxiella burnetii is listed by the United States Centers for Disease Control and Prevention (CDC) as a Category B pathogen and potential bioterrorism agent [1]. This small, Gramnegative, nonmotile, obligate intracellular bacterium is highly infectious and experimental data suggest that less than 10 organisms can cause infection. In animals, infection takes mostly a subclinical or inapparent course but abortions or birth of weak offspring, especially in small ruminants, can occur [2,3]. In humans, the clinical picture can range from asymptomatic to severe, usually presenting with fever, severe headache, myalgia and fatigue, frequently accompanied by atypical pneumonia and/or hepatitis. Chronic Q fever, i.e. persistence of infection exceeding a period of six months duration, may lead to endocarditis, which can be fatal. Additionally, chronic hepatitis, osteomyelitis, and septic arthritis are known sequelae [4].
Clinical diagnosis of C. burnetii infections in man and animal usually relies on serology, despite molecular methods such as PCR-based assays being more suitable in terms of speed and specificity, especially within the first couple of weeks after onset of disease [5][6][7]. However, although PCR assays are generally very fast and sensitive, their multiplexing capacity is limited. Moreover, due to their high specificity, they are incapable of detecting e.g. novel species or variants of a known species [8]. Microarrays on the other hand can be designed with a multitude of different probes either suitable for species identification by using highly specific probes, or for the detection of related or novel species by using probes lying within conserved regions [8]. A further benefit of the multiple targets on an array is that they can partly mitigate the weakness of diagnostic PCR assays when the PCR primer target contains point mutations. These mutations can be present in variants within a species and can lead to false negative results.
In this study, we describe a microarray-based method adapted to the ArrayTube™ (AT) platform, using three chromosomal (icd, omp/com1, and IS1111) and two plasmid (cbbE, cbhE) coded targets for the detection of C. burnetii. This platform has repeatedly been shown to be suitable for detecting other biological agents such as Burkholderia mallei/pseudomallei, Brucella spp., Bacillus anthracis, and Chlamydia spp. [9][10][11][12], for bacterial species differentiation and genotyping [13] and for automation-based applications.

Cell culture and bacterial isolates
The heat inactivated preparations of C. burnetii isolates and clinical samples used in this study were obtained from the National Reference Laboratory of Q Fever at the Federal Research Institute for Animal Health (Friedrich-Loeffler-Institut (FLI), Jena, Germany) [14] (Table 1).
All non-Coxiella bacteria used in this study were obtained from the German Collection of Microorganisms and Cultures (DSMZ, Braunschweig, Germany), and from the strain collection of the Institute of Bacterial Infections and Zoonoses at the Federal Research Institute for Animal Health (FLI, Jena, Germany). Bacteria were grown on standard media under conditions recommended by the respective bacterial strain collections. DNA from Bartonella spp, Chlamydia spp, Francisella spp, Salmonella typhimurium, Haemophilus influenzae, and Legionella pneumophila was obtained from the Institute for Medical Microbiology and Infection Control, Goethe University, Frankfurt/M., from the National Reference Laboratory of Psittacosis, from the National Reference Laboratory of  (Table 2).

DNA extraction and quantification
Genomic DNA from inactivated preparations of C. burnetii isolates and from non-Coxiella bacteria was isolated using the High Pure PCR Template Preparation Kit™ (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions. Quality and purity of the DNA were determined using a Nanodrop ND-1000 spectrophotometer (PEQLAB Biotechnologie GmbH, Erlangen, Germany). DNA quantification was performed with a TaqMan based real-time PCR assay targeting the transposase element IS1111 or the isocitrate dehydrogenase gene (icd) as described by Klee et al. [15] and Hildebrandt et al. [16]. DNA quantification for NineMile RSA493 was performed with the IS1111 real-time PCR assay and genome equivalents (GE) were calculated with 20 IS1111 copies per genome. Priscilla Q177 DNA was quantified with an icd real-time PCR assay and GE calculation was done with one icd copy per genome. Reaction conditions have been described previously [16] with the exception of the master mix (Maxima™ Probe qPCR Master Mix, Fermentas, St. Leon-Rot, Germany) and thermocycler (Mx3000P Thermocycler, Agilent Technologies, Santa Clara, CA, USA).

Primer and probe design
Gene specific primers and probes were designed and optimised using the Array Design software package (Alere Technologies GmbH, Jena, Germany) and the published target sequences from the reference strains NineMile RSA493 (GenBank: AE016828), Henzerling RSA331 and Priscilla Q177. After design, all primers and probes were blasted against the complete GenBank™ database (http:// www.ncbi.nlm.nih.gov/BLAST/).

Microarray layout and fabrication
Oligonucleotides were purchased as 3'-amino-modified oligonucleotides from metabion (Martinsried, Germany) and diluted in Spotting Buffer 1 (Quantifoil Micro Tools, Jena, Germany) to a final concentration of 10 μM. Arrays were spotted with six-fold redundancy on surfacecoated glass using Alere Technologies spotting machines and assembled into array tubes as described previously [17,18]. Internal staining controls (3'-amino and 5'-biotin-modified oligonucleotides) were included for staining control and orientation, as well as negative controls (0.1 M Sodium Pyrophosphate (NaPP) Standard pH 9). Probe sequences and array layout are shown in Table 3, and Figure 1, respectively.  Table 4). Amplifications were performed using a Mastercycler ep® (Eppendorf, Germany). The reaction was started with a 1 min denaturation step at 96°C, followed by 35 cycles with 15 s of denaturation at 96°C, 20 s for annealing at 55°C, and 30 s for extension at 70°C. After a final 5 min extension step at 70°C the reaction was stopped and the PCR products submitted to electrophoresis. Hybridisation control DNA was generated from 10 5 GE of the C. burnetii type strain (NineMile RSA493) using the PCR conditions above and the primer pair HybCtr-F (biotinylated) and HybCtr.

Hybridisation and data analysis
Hybridisation was carried out at 55°C for 60 min [10]. The hybridisation reaction was monitored using the ATR-01™ array tube reader (Alere Technologies) at 25°C for 10 min, recording one image at the end of incubation time. Signal intensity data with local background correction were obtained using the Iconoclust™ software, version 3.0 and Pionir/PARTISAN™ arrayLIMS software (Alere Technologies GmbH). The normalised intensity

Determination of assay specificity and detection limit
In order to assess assay specificity, DNA from a panel of 37 type or in-house reference strains of different bacterial origin was submitted to the above described PCR procedure and subsequent hybridisation ( Table 2). The detection limit was the smallest amount of Coxiella DNA detected by the test system for each individual target, and was determined in duplicate using tenfold serial dilutions containing 10 5 to 10 0 genome equivalents (GE) of NineMile RSA493 and Priscilla Q177, respectively.

PCR amplification of target genes
Using the reference strains NineMile RSA493 (GenBank: AE016828), Henzerling, and Priscilla Q177, five separate PCR assays where established and optimised for each target (Table 3). All gene targets were successfully amplified and the obtained fragment sizes corresponded to the theoretical values (Table 4) calculated by in silico analysis of the respective target sequences (data not shown).

Oligonucleotide probes and microarray layout
The selection of the microarray probe panel was based on five well characterised genetic markers of C. burnetii.
The chromosomal target regions (icd, omp/com1, IS1111) were chosen due to their genetic stability. The plasmid coded target regions (cbbE, cbhE) were selected to identify the Coxiella-specific plasmids QpRS and QpH1 [19,20]. After initial BLAST analysis, all available sequences for the specific targets were used for sequence alignments. The final microarray layout consisted of 16 oligonucleotide probes of 26-38 bp in length with six-fold redundancy ( Figure 1).

Hybridisation results
Initially, hybridisation experiments were done with three reference strains (Nine Mile RSA493, Henzerling RSA331 and Priscilla Q177). Biotin-labelled amplification products of every individual target were pooled and applied to the array. The hybridisation patterns of all three reference strains complied with two patterns; these had been predicted from the in silico analysis of target region sequences derived from the GenBank™ sequence database: pattern 1 for RSA493 and RSA331, and pattern 2 for Q177, respectively ( Figure 1). In order to validate the new diagnostic assay, German C. burnetii isolates obtained from the German Q-fever network (http://www.fli.bund.de/de/startseite/institute/ institut-fuer-bakterielle-infektionen-und-zoonosen/projekte/bmbf-verbundprojekt-q-fieber.html), and from the German National Reference Laboratory for Q fever were tested; a sample panel of 50 different isolates and 10 clinical samples (vaginal swabs) collected in Germany between 1997 and 2013 from sheep, goats, cattle, fallow deer, ticks, and humans were applied to the array. All analysed isolates revealed hybridisation signals corresponding to pattern 1.

Determination of detection limits and assay specificity
The assay detection limit of every target was tested with RSA493 and Q177 in serial dilutions of DNA. Every marker was reliably detected down to DNA amounts of 100 genome equivalents (GE), whereby IS1111 detection was more sensitive with a detection limit of 10 GE (data not shown). The assay specificity was evaluated with 37 type or in-house reference strains of various bacterial

Discussion
Diagnosis of Q fever in animal and man usually relies on serological methods, such as indirect immunofluorescence, complement fixation or enzyme-linked immunosorbent assays [5,21]. However, since C. burnetii-specific antibodies only begin to appear several weeks after infection, diagnosis is delayed. DNA based diagnostic assays are therefore more appropriate in terms of speed and specificity, and a plethora of PCR assays suitable for the detection of C. burnetii have been developed [15,22]. Although these assays are usually fast and sensitive and therefore more than adequate for most diagnostic applications, their capacity for multiplexing is limited [8]. This shortcoming has been resolved by using different types of DNA microarrays which have proved to be suitable for a broad range of applications in the microbial research and diagnostics of Q fever [7,23,24]. The microarray used in the present study was designed to target the chromosomally coded single copy genes isocitrate-dehydrogenase icd, the outer membrane proteincoding gene omp/com1 and the transpoase gene in the multi copy insertion element IS1111 as well as the plasmid coded markers cbbE on QpRS and cbhE on QpH1. All these markers have been characterised extensively and their suitability as targets for detecting C. burnetii has been confirmed in numerous studies [19,[25][26][27][28][29].
The analytical sensitivity of the assay used in the present study was 100 GE for the single copy gene targets icd and omp/com1 as well as for the plasmid coded targets cbbE and cbhE and 10 GE for the IS1111. Comparable values have been reported by Janse et al. [7], for icd and IS1111 using suspension microarrays and for IS1111 in a multiplex PCR targeting icd, omp/com1 and IS1111 by de Bruin et al., [29]. The specificity of the array was 100% with the tested non-Coxiella strains, including the phylogenetic neighbours Legionella spp. and Francisella spp. as well as with Bartonella spp. which have been reported to cross react with Coxiella in serological assays [30]. We did not observe any cross reactions between probes or primers and sample DNA as reported by Janse et al. [7], which could be due to the different oligonucleotide design and different array chemistry.
C. burnetii can harbour four different plasmids of different size and composition, namely QpH1, QpRS, QpDG, QpDV and the chromosomally integrated plasmid-like sequences IPS [25,31,32]. Their role in the biology of C. burnetii is still not clear, but some plasmid genes appear to be essential for conserved functions [33]. Moreover, a potential influence of the different C. burnetii plasmids on animal and human disease has been discussed [34]. Early studies suggested that the plasmids QpH1 and QpRS are markers for acute or chronic disease, respectively [32], but this assumption could not be verified in a later study [35]. A small, recent study testing the hypothesis that obstetric complications in C. burnetii infections were associated with a particular genotype and the presence of QpDV, found that this plasmid type was detected more frequently in isolates associated with abortions [36]. All samples tested in the present study originated from Germany and harboured the plasmid QpH1. This plasmid type seems to be the predominant type in the isolates circulating in Germany and allegedly the Netherlands [14,36]. Since the present array only includes probes for the plasmids QpH1 and QpRS, which currently appear to be the most frequently occurring plasmid types in C. burnetii [14,35], the existing array could be improved by implementing additional probes for the other known plasmids of C. burnetii.
Several arrays have been developed for the detection of C. burnetii [7,23,36,37]. These arrays range from easyto-handle, cheap but inflexible in terms of further probe implementation, to highly sophisticated, expensive-onpurchase but flexible regarding assay design. The Array-Tube™ platform is an easy-to-handle, middle price range (approx. 15 Euro/Array and PCR), molecular test for high-throughput and parallel analysis of samples. Array designs can be readily expanded by adding further gene targets. Moreover, the modular composition of the platform allows the rapid assembly of custom-made assays, targeting different biological agents, e. g. as part of a Bioweapon-agent-array detecting C. burnetii, Brucella spp., B. mallei/pseudomallei, B. anthracis and Chlamydia spp. [9][10][11][12].

Conclusions
The present array is a rational assembly of established and evaluated targets for the rapid and unequivocal detection of C. burnetii. This array could be applied to the screening of vaginal swabs from small ruminants, screening of environmental samples e.g. on farms, for screening patients with infective endocarditis [38], or of blood donors in regions of high endemicity for Q fever, e.g. the German-Dutch border area, or in a diagnostic assay screening for atypical pneumonias [23]. Competing interests GS and LDS have no financial or other competing interests. RE is an employee of Clondiag/Alere, the producer of the ArrayTube™ platform.