Real-Time Quantitative PCR Detection of Four Human Bocaviruses^▿

Clinical specimens.

Stool samples were obtained from 250 Finnish patients, 100 with and 150 without a history of travel outside Northern Europe within 1 month before sample acquisition. The median ages were 30.5 years (range, <1 to 74 years) and 52.5 years (range, <1 to 95 years), respectively. The destinations for the 100 travelers were Europe for 54%, Asia for 29%, Africa for 9%, South America for 2%, and unknown for 2% of travelers. Of the 250 subjects, 33 were children aged 0 to 10 years, of whom 12 (36%) had traveled abroad. All samples had been sent to the Helsinki University Central Hospital Laboratory Division for diagnosis of enteric pathogens, as suggested by symptoms of diarrhea. All samples were cultured for enteropathogenic bacteria, with emphasis on Salmonella, Yersinia, Shigella, and Campylobacter. In addition, several samples were studied for enteric parasites and viruses. All samples had further been tested for diarrheagenic Escherichia coli strains by PCR (3).

For PCR, stool swabs were suspended in 200 μl of Tris-EDTA buffer and boiled for 15 min, and the nucleic acids were purified with a NucliSENS kit using the easyMAG platform (bioMérieux, Lyon, France) and eluted in the elution medium to a volume of 25 μl. The median DNA concentration of the fecal extracts was 40 ng/μl (25th percentile, 20 ng/μl; 75th percentile, 77 ng/μl; range, 2 to 200 ng/μl). Potential PCR inhibition had been studied with 80 of the 250 DNA extracts by DNA spiking (17). None of the spiked samples were inhibitory. To further evaluate our qPCR assay, we analyzed three stool DNA extracts shown by qualitative PCR to contain HBoV2, HBoV3, or HBoV4 DNA (12).

Plasmids.

The performances of the qPCR assays were studied with plasmids containing the joint region of the left-hand untranslated region (UTR) and the NS1 gene of the human bocavirus 1 to 4 genomes, corresponding in HBoV1 to nucleotides (nt) 98 to 388 (GenBank accession number EU984245). The corresponding regions of HBoV2 (FJ170279) and HBoV3 (EU918736) were synthesized and cloned into pUC57 by Genscript (NJ), and that of HBoV4 was synthesized and cloned into pJet1.2 (Fermentas, Burlington, Canada) in Helsinki, Finland. The near-full-length HBoV1 clone, pST2, was a kind gift from Tobias Allander (2). Plasmid DNA concentrations were determined by an ND-1000 spectrophotometer (Nanodrop, Wilmington, DE). Each plasmid was diluted serially from 0.5 × 10⁸ to 0.5 × 10¹ copies/μl in 10 mM Tris-EDTA buffer, aliquoted, and stored at −20°C until use for generation of standard curves for DNA quantification.

Of note, the HBoV2 and HBoV4 primer binding sites in the HBoV2 and HBoV4 plasmids are identical except for a single nucleotide mismatch in the sense primer (Fig. 1). This dissimilarity was taken into account by the use of a degenerate nucleotide (C+T; IUPAC code Y). The HBoV2 and HBoV4 plasmids were indistinguishable with respect to assay sensitivity and reproducibility. Moreover, Mfold (22) secondary structure predictions showed no major difference between the amplified regions of the two sequences (data not shown). Consequently, the results reported herein for HBoV2 and HBoV4 are designated HBoV2/4 results and were acquired using only the HBoV2 plasmid unless otherwise stated.

Primers and hydrolysis probe.

Conserved and variable regions of the HBoV1 to -4 genomes were identified by aligning nine HBoV1 sequences, nine HBoV2 sequences, four HBoV3 sequences, and one HBoV4 sequence (Fig. 1). The HBoV1 and -2 strains were selected from a wide variety of geographical locations (10 countries) to take into account intraspecies variation. For HBoV3 and -4, we used all publicly available near-full-length genomic sequences. Primer selection was done with the aid of AlleleId 6.01 software (Premier Biosoft International, CA) for maximum intraspecies similarity and, when appropriate, maximum interspecies heterology. The primers are as follows (with GenBank accession numbers and primer annealing positions in parentheses): HBoV1F, 5′-CCTATATAAGCTGCTGCACTTCCTG-3′ (NC_007455; 152 to 177); HBoV1R, 5′-AAGCCATAGTAGACTCACCACAAG-3′ (NC_007455; 235 to 259); HBoV234F, 5′-GCACTTCCGCATYTCGTCAG-3′ (FJ170279; 50 to 70); HBoV3R, 5′-GTGGATTGAAAGCCATAATTTGA-3′ (EU918736; 205 to 230); and HBoV24R, 5′-AGCAGAAAAGGCCATAGTGTCA-3′ (FJ170279; 128 to 150). The primer region covers the left-hand untranslated region of the human bocavirus genome and the beginning of the NS1 gene. Both of the HBoV1 primers and the antisense HBoV3 primer (HBoV3R) were selected from regions with minimal similarity to the other bocaviruses. The qPCRs for HBoV2, -3, and -4 share a single sense primer (HBoV234F), and the qPCR for HBoV2 and -4 (i.e., the HBoV2/4 qPCR) uses the same reverse primer (HBoV2/4R). The hydrolysis probe sequence 6-carboxyfluorescein (FAM)-5′-CCAGAGATGTTCACTCGCCG-3′-minor groove binder (MGB)-quencher black hole 1 (BHQ1) (FJ170279; 85 to 104) was selected from a region fully conserved between all four known human bocaviruses. In multiplex format, this combination of primers and hydrolysis probe was designed to detect all published types of human bocaviruses. In singleplex format, the primers were designed to distinguish between HBoV1, HBoV3, and HBoV2/4. All oligonucleotides were synthesized by Sigma-Aldrich (Steinheim, Germany). The MGB hydrolysis probe was labeled at the 5′ end with the reporter dye FAM, and the 3′ end was blocked with the nonfluorescent quencher BHQ1. Blasting of primers and the probe against GenBank sequences did not reveal matching sequences that would plausibly result in probe hydrolysis. The amplified template region, together with 100 bp of surrounding sequence in both directions, was also analyzed for secondary structures with Mfold (22). No significant DNA secondary structures were detected (data not shown).

Real-time PCR.

All PCRs were done in a volume of 25 μl using Stratagene Mx3005p and approved clear PCR plasticware (Stratagene, CA). The singleplex reactions consisted of 1× TaqMan Universal Master Mix (Applied Biosystems, CA) with AmpErase uracil-N-glycosylase (UNG), 0.6 μM concentrations of sense and antisense primers, 0.3 μM probe, 2 μl of template, and molecular biology-grade water for a final volume of 25 μl. The multiplex reactions were set up similarly but with all five primers included. UNG was allowed to degrade any potential carryover PCR products for 2 min at 50°C before activation of the AmpliTaq Gold polymerase for 10 min at 95°C. The amplification consisted of 40 cycles of 15 s at 95°C and 1 min at 60°C. Each run included plasmid and no-template controls. To generate baseline-corrected fluorescence (dR) data, baseline fluorescence was automatically determined by the Mx4000 software version 3.01 (Stratagene) baseline algorithm. The cutoff for quantification cycle (C_q) determination was automatically calculated as 20 times the standard deviation of the fluorescence value of the baseline in cycles 5 through 9. Each fluorescent reporter signal from the FAM channel was measured against the internal reference dye (ROX) signal to normalize for non-PCR-related fluorescence fluctuations between samples. Strict laboratory procedures were followed to prevent PCR contamination, including separate spaces for handling of samples, Master Mix ingredients, and plasmid templates. All runs included negative controls, which gave negative results for human bocavirus DNA throughout the study.

Analytical specificity.

Analytical specificities of the real-time qPCR assays were evaluated with 500 ng human DNA per reaction (i.e., 25 μl) from cultured 293T cells and cloned full-length or near-full-length genomes of TT virus (GenBank AY666122), parvovirus B19 genotypes 1 (AY504945), 2 (AY044266) and 3 (AJ249437; a kind gift from Antoine Garbarg-Chenon), and polyomaviruses BK virus (BKV), JC virus (JCV), and simian virus 40 (SV40) (kind gifts from Kristina Dörries). We also tested a partial genome of human parvovirus 4 (PARV4) and the late-region genes of KI polyomavirus (KIPyV) (EF127906; a kind gift from Tobias Allander). We further tested pooled nucleic acid extracts of human stool containing astrovirus, norovirus, and rotavirus. The stool specimens had been found virus positive by diagnostic electron microscopy and/or PCR by the Helsinki University Central Hospital Laboratory Division.

Analytical sensitivity, linearity, and amplification efficiency.

Limits of detection (LOD) of the real-time assays were determined with serial dilutions of control plasmids. At 10 copies/reaction (5,000 copies/ml of eluate), 100% of 20 replicates were positive with the multiplex and all singleplex assays regardless of the presence or absence of human genomic DNA at 500 ng/reaction. This indicates a robust LOD of ≤10 copies/reaction. We further tested 8 replicates of 10⁻² to 10² nominal plasmid copies/reaction diluted in a pool (n = 140) of stool DNA extracts that were qPCR negative for human bocavirus. Testing was repeated with another instrument on a different day. A generalized linear model was fitted to the total observed proportion of positive results and to the nominal numbers of template copies using MATLAB software's (Mathworks, Natick, MA) GLMFIT command with the probit link function. According to the probit analysis, the assay sensitivities in stool extracts remained good, ranging from ∼4 copies/reaction (HBoV2/4) to ∼9 copies/reaction (HBoV1) for both the multiplex and the singleplex assays (data not shown).

The real-time PCRs were linear over the range of 10 to at least 10⁸ copies/reaction (Fig. 2). The average reaction efficiencies for the singleplex and multiplex reactions were high for all assays regardless of the presence or absence of 500 ng of human genomic DNA per reaction, ranging from 97% to 102% (Table 1).

Analytical specificity.

Analytical specificities of the real-time PCR assays were evaluated with purified human DNA, DNA from virus-positive stool samples, and with cloned viral DNA of 9 viruses. All plasmids were tested at 10¹⁰ copies/reaction, corresponding to 5 × 10¹² copies/ml of original sample. Neither the multiplex nor the singleplex qPCR assays showed observable amplification with any of the plasmids or clinical specimens within the 40 PCR cycles.

Cross-reactivity.

Potential cross-amplification of HBoV1 to -4 plasmid templates with the three singleplex assays was studied at high input levels of the plasmids (10⁶ to 10⁹ copies/reaction) with 10 replicates. No cross-amplification was observed with any of the assays at ≤2 × 10⁷ copies/reaction. At 1 × 10⁸ copies, the HBoV3 primers showed cross-amplification of HBoV2 and HBoV4 plasmids, with 9 copies/reaction detected. Similarly, HBoV2 primers showed cross-amplification of the HBoV3 plasmid at 1 × 10⁹ copies/reaction (14 copies/reaction detected); however, no signal was generated at 2 × 10⁸ copies/reaction or less. No cross-amplification with HBoV1 singleplex was observed at 2 × 10⁹ copies/reaction of HBoV2, HBoV3, or HBoV4 templates. Copy numbers higher than 2 × 10⁹ copies/reaction were not tested for cross-reactivity.

All singleplex assays were able to detect the respective bocavirus species also from mixtures of HBoV1 to -4 templates, in which HBoV1 to -4 were simultaneously present in varied quantities. Table 2 shows the calculated and measured amounts of HBoV1, HBoV2/4, and HBoV3 plasmids.

Reproducibility.

Reproducibilities of the multiplex and singleplex assays were studied by replicate analysis of quantification standards in a single run (intraassay variation) and repeated runs (interassay variation) using three replicates, each at concentrations of 10¹ to 10⁸ copies/reaction (Table 3). Of note, the intra- and interassay variations were calculated with measured quantities of target DNA rather than from the corresponding C_q values. Highest intra- and interassay variations were observed at 10¹ copies/reaction, with mean coefficients of variation (CV) of 40% and 33%, respectively. With 10³ copies/reaction, the corresponding figures were reduced to 3% and 16%, respectively.

qPCR of stool extracts.

For clinical evaluation of the qPCR methods presented here, stool samples sent for diagnosis of enteric infections from Finnish travelers and nontravelers were screened for DNA of the four human bocaviruses. Of the 250 samples, 5 (2%) were found reproducibly positive for human bocavirus DNA in the initial multiplex screening. Of these 5 samples, 4 and 1 were positive with the HBoV2 and HBoV3 singleplex assays, respectively, and were further confirmed by sequencing. All four samples with HBoV2 DNA were from children aged ≤2 years, and the sample with HBoV3 DNA was from an 18-year-old female who had recently traveled to Bulgaria. The HBoV2 prevalence among 250 subjects below 95 years of age was only 1.6%; however, it was 10% among the 41 children below 18 years of age, 12.5% among the 32 children aged below 10 years, and 20% among the 20 infants ≤2 years of age. Further patient characteristics, including travel histories, HBoV DNA loads, and other microbial findings, are listed in Table 4. HBoV1 and HBoV4 DNAs were not detected in any of the 250 samples.

With respect to HBoV1 to -3, the qPCR results were validated by rescreening all the samples with a previously described quantitative assay for HBoV1 (1) and qualitative PCR assays for HBoV2 (6) and HBoV3 (19). The results of the confirming HBoV1 qPCR assay were in full accordance with our original results: all 250 samples were negative for HBoV1 DNA. Likewise, the samples that were positive with our HBoV2 or HBoV3 singleplex assays were also positive with the respective qualitative PCR assays, with results confirmed by PCR product sequencing. The sequences from the qualitative PCRs were 98 to 100% identical to previously published HBoV2 and HBoV3 sequences, with no observable regional differences.

The three fecal extracts that had previously been shown to contain HBoV2, HBoV3, or HBoV4 DNA (12) were all reproducibly positive in the respective singleplexes, but not with the heterologous assays. Samples with HBoV2 and HBoV3 DNA were also positive in the multiplex qPCR. The HBoV4 positive sample was not multiplex tested due to the very limited amount of DNA extract available.

qPCR of serum DNA extracts.

To further compare our new HBoV1 singleplex assay with the established qPCR test (1) that we have previously used for HBoV1 quantification (11, 20), we examined with both tests 15 serum samples that had been shown to contain different loads of HBoV1 DNA (1). Correlation between these two methods was high despite the relatively low levels of viral DNA (r = 0.95) (Fig. 3).