Type strains of C. neoformans (ATCC 32045), Staphylococcus aureus (ATCC 29213, 33592, BAA-44, BAA-41, and BAA-1683), Escherichia coli (ATCC 25922), and Malassezia pachydermatis (ATCC 14522) were purchased from the American Type Culture Collection (ATCC; USA). Field isolates of B. burgdorferi (n = 2), T. gondii (n = 2), E. canis (n = 1), E. ewingii (n = 1), Anaplasma phagocytophilum (n = 2), candidatus Mycoplasma haematoparvum (n = 1), S. pseudintermedius (n = 5), Candida spp. (n = 3), Malassezia spp. (n = 5), Microsporum canis (n = 3), Microsporum gypseum (n = 2), Aspergillus spp. (n = 3), canine parvovirus (n = 3), canine coronavirus (n = 3), and CDV (n = 6) were identified at Chungbuk National University Veterinary Diagnostic Laboratory. Recombinant plasmid vectors for B. dermatitidis, C. neoformans, N. caninum, B. burgdorferi, Bartonella spp., T. gondii, E. canis, and CDV were used to assess the analytical specificity and as positive controls. Nucleic acids (DNA and RNA) extracted from whole blood in two dogs were used as negative controls. All field isolates were identified by culture, conventional PCR, or antigen-capturing ELISA and verified by DNA sequencing. The vectors were constructed by amplifying synthetic oligonucleotides based on sequences obtained from the GenBank database (National Center for Biotechnology Information, USA) using the primers listed in Table 1 and cloned into a T-Blunt PCR Cloning Kit (SolGent, Korea) according to the manufacturer's instructions. Correct insertion into the vector was verified by sequencing. Each recombinant vector was prepared in diethylpyrocarbonate-treated distilled, deionized water as a series of 10-fold serial dilutions and then used to assess the analytic sensitivity of the test.

Total nucleic acid was extracted from the samples using the MagMAX Total Nucleic Acid Isolation Kit (Applied Biosystems, USA) according to the manufacturer's instructions. Briefly, 200 µL of each sample were added to a bead tube containing zirconia beads, followed by 235 µL of lysis/binding solution. The bead tube was then placed on an Ambion Vortex Adapter and beaten with a Vortex-Genie 2 (Scientific Industries, USA) at maximum speed for 15 min. After beating, the tubes were centrifuged at 16,000 × g for 3 min, and the supernatant was then carefully transferred into clean microcentrifuge tubes. Following another centrifugation step at 16,000 × g for 6 min, 115 µL of the supernatant was transferred to a 96-well, deep-well microplate containing 20 µL of paramagnetic beads and 65 µL of 100% isopropanol. The plate was shaken for 1 min on an orbital multi-well-plate shaker at maximum speed without spilling the sample, after which it was placed on a magnetic stand for 5 min and the supernatant was carefully discarded. The bead pellet was then washed twice each with Wash solution I and II (150 µL). Next, the plate was vigorously shaken for 2 min, after which the resultant pellet was resuspended in 50 µL of elution buffer. Finally, the extracted total nucleic acid in elution buffer was stored at -80℃ until use in PCR.

Primer and probe sequences used for ME mqPCR were based on published reports [1011121316202730] and are listed in Table 1. All primers and probes were synthesized and purchased from Bioneer (Korea), except probes for Minor Groove Binder (Applied Biosystems). A 65 bp oligonucleotide was used as an internal control (IC) to monitor false negatives resulting from failure of PCR due to the presence of inhibitory substances. The IC contained a non-specific 16S ribosomal RNA gene sequence flanked by the uvrC gene sequence of Mycobacterium (M.) bovis to minimize cross-reactivity with M. bovis. The IC (0.001 µM) was spiked into each reaction, which resulted in a positive signal with threshold cycle (C_T) values between 36 and 38.

Primers and probes were categorized as mqPCR A, B, C, and D (Table 1). Each mqPCR reaction included two primer/probe sets for the target pathogen, one primer/probe set for the IC, and a reference dye. To minimize cross-interference of the dyes for each probe, a combination of FAM and HEX was used for the target probes. Cy5 was used for the IC, and ROX was used as the reference dye. PCR amplification was conducted using an Eco Real-Time PCR system (Illumina, USA). The Path-ID Multiplex One-Step RT-PCR kit (Applied Biosystems) was used for PCR A, and the QuantiTect Multiplex PCR (Qiagen, USA) kit was used for PCR B, C, and D according to the manufacturer's recommended protocols. Each 20 µL reaction contained 0.4 µM of each primer, 0.2 µM of each probe, and 5 µL (for PCR A) or 2 µL (for PCR B, C, and D) of the template. Cycling conditions were as follows: reverse transcription for 10 min at 45℃ (omitted for PCR B, C, and D), 10 min (for PCR A) or 15 min (for PCR B, C, and D) initial activation at 95℃, and then 35 cycles of 15 sec at 95℃ and 60 sec (for PCR A) or 90 sec (for PCR B, C, and D) at 60℃. A C_T of 35 or lower was considered positive for a given pathogen.

To compare and optimize the PCR conditions, analytical sensitivity was assessed for each pathogen in triplicate as a singleplex assay and the multiplex panel by using the corresponding recombinant vector. The singleplex PCR assay was performed using the same regents and PCR conditions as the multiplex panel. The reaction for E. canis was performed using two mqPCR kits (Path-ID Multiplex One-Step real time-PCR kit; Applied Biosystems; QuantiTect Multiplex PCR; Qiagen) with the recombinant vector as a template to determine whether the two kits had comparable sensitivity.

A total of 47 CSF samples collected from dogs with ME or submitted to Chungbuk National University between January 2010 and August 2013 and six CSF samples collected from healthy dogs were included in this study. All dogs with ME presented with non-specific neurological symptoms, such as seizure, incoordination, or altered consciousness, and had no laboratory abnormalities consistent with a metabolic disorder or radiographic abnormalities (i.e., complete blood count, serum biochemistry profile, urinalysis, X-ray and abdominal ultrasonography) that would suggest non-CNS disease as the cause of symptoms. The criteria for case selection included: (1) evidence of focal or multifocal brain lesions during neurological examination without signs of spinal cord lesions or lesions in the peripheral nervous system; (2) abnormal result of CSF examination indicating inflammation and tissue necrosis (reference interval: < 5 white blood cells/µL, total protein < 0.3 g/L); and (3) evidence of focal or multifocal intracranial lesions upon computed tomography (CT) or magnetic resonance imaging (MRI) [56142129]. After diagnosis, the residual CSF samples were used for the mqPCR assay.

All mqPCR assays detected only the target pathogen and did not produce any non-specific amplification products or products of other pathogens in repeated experiments. In addition, there were no false positives due to cross-talk between dye signals within each reaction. Standard curves for all multiplex assays generated from 10-fold serial dilutions of each template showed correlation coefficients (R²) ranging from 0.970 to 0.995 and slopes of 3.41-3.59 (Fig. 1), indicating good linearity of PCR amplification.

Next, the performance of the mqPCR panel and singleplex PCR for each of the eight pathogens were compared directly to identify any negative effects of multiplexing on PCR detection. The analytical sensitivity (i.e., limit of detection) of the PCR was estimated using serially diluted recombinant vectors with known copy numbers (molecules per µL). The estimated mqPCR detection limits of the mqPCR panel were 3.8 for CDV, 3.7 for E. canis, 3.7 for Bartonella spp., 3.8 for B. burgdorferi, 3.7 for B. dermatitidis, 3.7 for C. neoformans, 38 for N. caninum, and 3.7 for T. gondii (Table 2). The results of singleplex and multiplex PCR were identical except for CDV, and C_T differences between PCR reactions were not significant, demonstrating that multiplexing did not have a significant negative effect on sensitivity of the PCR.

The C_T value obtained for E. canis using the Path-ID Multiplex One-Step RT-PCR kit (Applied Biosystems) was slightly lower than that obtained with the QuantiTect Multiplex PCR kit (Qiagen); however, the difference in C_T was less than 1, suggesting that the kits had comparable performance.

Of the 47 CSF samples obtained from dogs diagnosed with ME, seven (15%) were positive for the target pathogen. The most common pathogens were C. neoformans (3/7), B. dermatitidis (2/7), and B. burgdorferi (2/7). In contrast, conventional CSF analysis and culture was negative for all pathogens. None of the samples were positive for more than one pathogen. For the seven samples with discrepant results between the mqPCR panel and routine test methods, agarose gel electrophoresis, sequencing, and comparison with sequences available in the GenBank database confirmed the presence of the identified pathogen. Among the six CSF samples from healthy dogs, all were negative for the eight pathogens of the mqPCR panel, suggesting that mqPCR did not falsely detect normal CSF constituents as target materials.