This study was approved by the ethics committee of Shenzhen Shajing Hospital affiliated to Guangzhou Medical University (Shenzhen, China) prior to commencement. After informed written consent from adult patients or legal representatives of children, throat swab samples, which are determined to be HAdV negative by D3 Ultra DFA Respiratory Virus Screening & ID kit (B&C Biological Technology Co., Shanghai, China), were collected using universal transport media (BD Biosciences, Franklin Lakes, NJ, USA). From February 2016 to June 2017, 3,160 cases of adenovirus-negative throat swab samples were collected from 3 districts (Baoan, Futian, Luohu) of Shenzhen.

For nucleic acid extraction, all throat swab samples were centrifuged at 3,000 × g for 5 min at 4°C to collect virus attached to cells. Nucleic acid was extracted from 200 µl of each sample using the Nucleic Acid Extraction kit according to the manufacturer's protocol (Huiyan Biotechnology Company, Shenzhen, China). Then, 200 µl of the sample was extracted, and nucleic acid was eluted into a volume of 150 µl. Nucleic acid was then processed immediately for qPCR.

Potential primer and probe sequences were selected after being thoroughly analyzed using the Premier Biosoft software and the BLAST (National Centre of Biotechnology Information, National Institutes of Health, Bethesda, MD, USA) program (http://blast.ncbi.nlm.nih.gov/Blast.cgi). BLAST analyses were performed to check the specificity of the primers and the probes against other closely related genome sequences. The Premier Biosoft software allowed the combination of existing proven primer and probe sequences with sequence targets to yield multiplexed assay designs and choose identical cycling parameters. The primer pairs and probes specific to each virus were grouped in a multiplex reaction on the basis of the following criteria: internal primer binding properties for hairpin and primer-dimer potential, length of the desired amplicon, G-C content, and melting temperatures (Tm) of the probes and primers. When the parameter settings were defined, Premier Biosoft software assessed each multiplex component as it was added to the assay pool, avoiding the risk of cross-reactivity issues. The primers and TaqMan probes were purchased from Invitrogen (Thermo Fisher Scientific, Inc., Shanghai, China). TaqMan hydrolysis probes were labeled at the 5′-end with the reporter dyes and quenched with Blackhole Quencher 1or 2 (BHQ 1 or 2) at the 3′-end. For more convenience, the method was somewhat modified and each TaqMan probe was labeled with a different fluorescent reporter dye (FAM, VIC, ROX, or CY5) (Table I).

A 1,000-bp random DNA sequence was generated using the DNAMAN software (Lynnon Biosoft, Vaudreuil, QC, Canada), and four bases (A, T, C, and G) accounted for 25%, respectively. The generated sequence was then synthesized by Takara and introduced into plasmid pBSK.

Three inactivated HAdV isolates were provided as gifts by Professor Rong Zhou (State Key Laboratory of Respiratory Diseases, Guangzhou Medical University, Guangdong, China). These clinical isolates were sent to Hexin Health Technology Company (Guangdong, China) and identified using DHI, immunofluorescence, flow cytometry, and qPCR. The fragments amplified with HAdV primers were sequenced and compared with HAdV sequences in NCBI (National Centre of Biotechnology Information, National Institutes of Health). After comparison, the three clinical isolates were confirmed as HAdV B3, C2, and E4. DNA of three isolates was extracted from 200 µl samples as described earlier.

The concentration of three isolates was detected using the QX100 drop digital PCR system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). The droplet digital PCR (ddPCR) reaction mixture consisted of 10 µl of a 2 × ddPCR Master Mix (Bio-Rad Laboratories, Inc.), 2 µl of HAdV primers/probes, and 5 µl of sample nucleic acid solution in a final volume of 20 µl. The entire reaction mixture was loaded into a disposable plastic cartridge (Bio-Rad Laboratories, Inc.) together with 70 µl of droplet generation oil (Bio-Rad Laboratories, Inc.) and placed in the droplet generator (Bio-Rad Laboratories, Inc.). After processing, the droplets generated from each sample were transferred to a 96-well PCR plate (Eppendorf, Hamburg, Germany). PCR amplification was carried out on a T100 thermal cycler (Bio-Rad Laboratories, Inc.) using a thermal profile beginning at 95°C for 2 min, followed by 40 cycles of 95°C for 10 sec and 60°C for 30 sec, 1 cycle of 95°C for 10 min, and ending at 12°C. After amplification, the plate was loaded on the droplet reader (Bio-Rad Laboratories, Inc.) and the droplets from each well of the plate were read automatically at a rate of 32 wells per hour. ddPCR data were analyzed with Quanta Soft analysis software (Bio-Rad Laboratories, Inc.), and the quantification of the target molecule was presented as the number of copies per ul of PCR mixture.

Then, three HAdV isolates B3, C2, and E4 were equally mixed, and standard curves for adenovirus were drawn by performing qPCR on tenfold serial dilution in DNase- and RNase-free H₂O of the corresponding standard virus templates ranging from 10⁸ to 10² copies (equivalent genomes)/ml. The standard curves for multiplex qPCR were obtained by linear regression analysis of the threshold cycle (C_T) value (y axis) vs. the log of the initial copy number present in each sample dilution (x axis), allowing the determination of the correlation coefficient (R²). The amplification efficiencies (E) of reactions were calculated from the curves using the equation: E=(10(^−1/slope)-1) × 100% (31,32). The resulting curves were used to determine the number of viral genome copy corresponding to each virus found in the experimental samples.

For the multiplex assay, the reaction mixture was prepared, containing 25 µl of iQ Multiplex Powermix (Bio-Rad Laboratories, Inc.), 0.5 µl of each primer (final concentrations, 200 nM), and 0.375 µl of each probe (final concentrations, 75 nM). The final volume of the mixture was adjusted to 30 µl with DNase- and RNase-free H₂O (Nacalai Tesque, Inc., Kyoto, Japan). Finally, 20 µl of sample DNA was added to the reaction mixture on ice. qPCR was performed using a Roche Light Cycler 480 (LC480) (Roche Diagnostics, Basel, Switzerland) with the following amplification protocol: Pre-denaturation for 2 min at 95°C; 40 cycles of denaturation for 10 sec at 95°C, and annealing and primer extension for 30 sec at 60°C. The fluorescence emitted by FAM, VIC, ROX, or CY5 dyes was measured simultaneously and independently at the end of the annealing step. For a background control, no template control (NTC, water) was used.

The results were expressed as the quantification cycle (C_q, cycle at which the fluorescence surpasses the background level). Lower C_q values corresponded to a greater amount of initial template and a negative result was considered to have a C_q value of 40 or more cycles. So the cutoff point was fixed empirically at a C_q value of 40. The status of the sample (positive or negative) was determined first using this chosen cutoff to assess the variability of the observed C_q values. However, the use of such arbitrary cutoffs was not ideal because they might be either too low (eliminating valid results) or too high (increasing false-positive results) (33). Hence, those PCR products, which were initially identified as adenovirus-positive, were sent to the Invitrogen Biotechnology Company (Invitrogen, Shanghai, China) for conventional Sanger sequencing. The nucleotide sequences of HAdV B, C, and E were screened using the BLAST program on the NCBI website to determine the sequence origin.

For a good analysis of the presented results, the performances of the multiplex assays were also assessed using the confusion matrix that contained information about the two assays. The following parameters were defined: True positive (TP): The number of ‘positive’ samples categorized as ‘positive’; false positive (FP): The number of ‘negative’ samples categorized as ‘positive’; false negative (FN): The number of ‘positive’ samples categorized as ‘negative’; true negative (TN): The number of ‘negative’ samples categorized as ‘negative’. Based on these four metrics, the sensitivity [Sn=TP/(TP + FN)] and the specificity [Sp=TN/(TN + FP)] of the multiplex assay could be evaluated (33).

In the study of adenovirus infection in the respiratory tract, most of the literature reported primers and probes based on the hexon and fiber genes (Table I). HAdV serotypes that could be detected by designing primers for the hexon gene included HAdV-C1, C2, B3, E4, C5, C6, B7, B11, B14, B16, B21, B34, and B35. HAdV serotypes that could be detected by designing primers for the fiber gene included HAdV-C2, B3, E4, C5, B7, B11, B16, B21, B34, and B35. As more new adenovirus serotypes (HAdV-B50, B55, and B60) cause respiratory infections, primers for multiplex qPCR should be able to detect adenovirus serotypes as comprehensively as possible. Primers and probes were designed to detect 16 types so as to achieve the coverage of the main HAdV types that cause the respiratory tract infections: 3, 7, 11, 14, 16, 21, 34, 35, 50, 55, and 66 of HAdV species B; 1, 2, 5, and 6 of HAdV species C; 4 of HAdV species E. HAdVs associated with respiratory tract infections were subjected to a conservative analysis. Not only hexon and fiber genes were found to be highly conserved, but E3 genes were also highly conserved. Unique sets of PCR primers and TaqMan probes were designed to target highly conserved sequences in each viral hexon, fiber, and E3 genes. After thorough analysis using the Premier Biosoft software and the BLAST program, sets of primers and TaqMan probes designed for the E3 gene were finally selected for subsequent experiments.

Serial dilutions of mixed adenovirus standards were tested using multiplex qPCR, and individual standard curves were constructed from the C_q values (Fig. 1). The multiplex assay could amplify each plasmid standard between 10² and 10⁸ copies per reaction mixture with a strong linear relationship between the C_q values and the log₁₀ of the input number of copies (HAdV B3, r²=0.99957; HAdV C2, r²=0.99964; HAdV E4, r²=0.99985). The slopes of the standard curves were −3.37007 for HAdV B3, −3.37538 for HAdV C2, and −3.43185 for HAdV E4.

The defined acceptance criteria for accurate and reproducible quantification were as follows: qPCR R² ≥0.98; qPCR E among 90 and 110% (corresponding to the slope of the regression curve that should be between −3.1 and −3.6). For the multiplex assay, the qPCR E ranged from 80 to 120% (corresponding to the slope of the regression curve that should be between −3.9 and −2.9) (31). The regression analysis of the multiplex assay demonstrated that the R² and the PCR amplification E were all within the suitable range. The R² of the standard curves of each multiplex reaction were all greater than 0.99 along with the E values ranging from 95 to 98%, which was considered as acceptable for a multiplex qPCR assay.

A total of 3,160 throat swab samples that tested adenovirus negative by the immunofluorescence assay were retested using multiplex qPCR assay. The sensitivity and specificity of the multiplex qPCR assay were compared with those of the immunofluorescence assay. Further, 2,894 samples were detected as adenovirus negative, which was consistent with the findings of immunofluorescence assay. However, 266 samples were detected as adenovirus positive. The samples were sent for conventional Sanger sequencing to verify whether 266 adenovirus-positive samples were TP. Through the BLAST online software sequence alignment, 254 samples were verified as TP and 12 samples were verified as FP (Fig. 2A). Among the 254 TP samples, HAdV B was detected in 184 samples, HAdV C was detected in 51 samples, and HAdV E was detected in 39 samples. Among the 12 FP samples, HAdV B was detected in 8 samples, HAdV C was detected in 3 samples, and HAdV E was detected in 1 sample (Fig. 2B). Co-infections with two HAdVs B and E were detected in 8 samples and co-infections with HAdVs C and E in 12 samples. No detection of co-infections of more than two HAdVs was reported (Fig. 2C). Briefly, the distribution frequencies of C_q values for a set of 254 analyzed adenovirus-positive samples is displayed in Fig. 2D. Further, 66 positive samples had a C_q value less than 25, 14 a C_q value less than 20, and one with a C_q value less than 15 in HAdV B species. The aforementioned data indicated that the sensitivity and the specificity of multiplex qPCR were much better than those of immunofluorescence assay.

Multiplex qPCR methods in other studies were used to detect not only adenovirus but also other respiratory pathogens (Tables II and III). The multiplex qPCR in this study was used to detect only HAdV B, C and E. 16 HAdV serotypes associated with respiratory infections could be detected using primers and probes in this study. By comparing the number of detected serotypes, sensitivity and specificity, the more comprehensive the serotype detection, the higher the sensitivity of the assay. Therefore, the sensitivity and specificity of this study were still 100% (254/254) and 99.6% (2894/2906), respectively, in the large sample size of the respiratory tract.