The one-pot RT-asymmetric PCR-combined MCA assay was designed to target the ORF1ab gene and nucleocapsid (N) gene for the detection of SARS-CoV-2. Detection of influenza A and B was achieved by targeting the matrix (M) and hemagglutinin (HA) genes, respectively. The human RPP30 gene was selected as an internal reference and can serve as an indicator of the specimen quality. Primer and probe sequences targeted highly conserved regions of the influenza A, B and SARS-CoV-2 genome and were based on the published literature (van Elden et al., 2001; Duchamp et al., 2010; Dunn et al., 2014; Wei et al., 2020). The SARS-CoV-2 ORF1ab gene and N gene probe, with a carboxy-X-rhodamine (ROX) reporter/QSY quencher, and probes for the influenza A M gene, influenza B HA gene, and human RPP30 gene probe, with a carboxyfluorescein (FAM) reporter/QSY quencher, were purchased from Sangon (Sangon Biotech, Shanghai, China). Five pairs of primers and probes were used to simultaneously detect the three viruses in a single-tube. The optimal sequences of primers and probes are listed in Table 1.

Simulated ssDNAs were designed to simulate the single-stranded amplicons that are generated through asymmetric PCR and were used to simulate the MCA process of the single-stranded amplicon and probe and to evaluate the melting temperature (Tm) value of the detection peak. The simulated ssDNAs were purchased from Sangon (Sangon Biotech) and are listed in Supplementary Table 1.

RNA templates were extracted from MS2-based virus-like particles (MS2-VLPs) corresponding to different viruses as well as the collected clinical samples. The specific virus sequences packaged in MS2-VLPs were quantified by digital PCR, and the results were regarded as the required positive standards. Viral RNA was extracted and purified with the Gene Rotex96 nucleic acid extraction and purification system (Tianlong Science and Technology Co., Ltd., Xi’an, China) using an RNA extraction kit (Tianlong Science and Technology Co., Ltd.) in accordance with the manufacturer’s instructions. The concentration and purity of the extracted RNA samples were assessed using a NanoDrop™ 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States). The qualified RNA samples were subsequently frozen and stored at −80°C.

The assay was performed on the SLAN-96P real-time PCR system (Shanghai Hongshi Medical Technology Co., Ltd). The Abstart One-step RT-PCR Mix (Sangon Biotech, Shanghai, China) was used to reverse transcription and PCR amplification. Each 50-μL PCR mixture contained 2 μL Abstart Taq^® Polymerase with dNTP (0.125 U/μL) supplied in 5 × One-step RT-PCR Buffer (with 3.3 mM Mg²⁺), 5 × Solution I buffer, with 0.01–0.2 μM limiting primers, 0.08–1.6 μM excess primers, 0.005–0.2 μM probes, and 15 μL RNA template. The concentration details for the primers and probes are presented in Supplementary Table 2. The reactions were incubated at 42°C for 30 min and 95°C for 5 min followed by 50 cycles of 94°C for 30 s, 60°C for 60 s, and 72°C for 30 s, and then incubated at 72°C for 10 min, 95°C for 2 min, and 30°C for 2 min. The reaction mixture was subjected to MCA from 30 to 90°C and finally incubated at 40°C for 20 s. To avoid contamination of the PCR products, the whole reaction is performed in a closed PCR amplification tube that is never opened during assay performance.

MS2-VLPs have been widely used as quality control materials for detecting pathogenic RNA viruses (Sun et al., 2013; Wang et al., 2015a; Zhang et al., 2016) due to their similarity in structure to natural viruses, stability, RNase resistance, non-infectivity and inability to replicate by itself both in vivo and in vitro (Sun et al., 2013; Wang et al., 2015a,b). Previous studies showed that MS2-VLPs were used to evaluate the LOD and precision of 7 commercial real-time PCR kits for Zaire ebolavirus (Wang et al., 2015b). Recent studies showed that SARS-CoV-2-MS2-VLPs was used to assess the LOD of rRT-PCR assays when detecting SARS-CoV-2 variants (Chen et al., 2022), as well as be used to evaluate the external quality assessment for molecular detection of SARS-CoV-2 in clinical laboratories (Wang et al., 2021). In our study, the analytical sensitivity was determined quantitatively using MS2-VLPs that contained the SARS-CoV-2 ORF1ab gene and N gene, the influenza A M gene, and the influenza B HA gene, separately. For each type of MS2-VLPs, the concentration was calibrated by digital PCR to 1.0 × 10³ copies/μL. The three types of MS2-VLPs were mixed equally and serially diluted 10-fold with TE buffer to 500, 250, 125, 62.5, 31.3, 15.6, and 7.8 copies/μL. Twenty replicates of each dilution were tested and the lower LoD was defined as the concentration in copies/μL of the lowest dilution that could be detected with 95% probability and determined by probit analysis.

The specificity of each primer and probe oligonucleotide sequence was evaluated by conducting a BLAST analysis¹ using the sequences of SARS-CoV-2, influenza A, and influenza B against the nr/nt database of the National Center for Biotechnology Information and the Global Initiative on Sharing All Influenza Database² to ensure that the primers and probes accurately matched the target gene sequence. Supplementary Figure 1 showed supporting information to better present the blastn analysis.

To determine the specificity of the developed assay, a species-specific sample panel was created that included clinical samples containing other respiratory viruses from patients who had clinical symptoms overlapping with those of influenza infection and COVID-19 (n = 58). The clinical samples were obtained from the Department of Respiratory and Critical Care Medicine at China-Japan Friendship Hospital (Beijing, China) and this study was approved by the Institutional Review Board of the China-Japan Friendship Hospital (No. 2022-KY-248), China. The sample panel contained human rhinovirus (n = 25), human parainfluenza virus (n = 12), adenovirus (n = 4), respiratory syncytial virus (n = 6), and human bocavirus (n = 1), all of which were positively detected via their respective RT-PCR assays. Approximately, 1 ng/μL of each viral RNA (15 μL) was used in the assay.

The precision of the developed assay was assessed in terms of both intra-assay precision (repeatability) and inter-assay precision (reproducibility). The variability of the developed assay was evaluated by detecting three concentrations (500, 125, and 31.3 copies/μL) of equally mixed positive MS2-VLPs. To observe intra-assay variability, each concentration was analyzed five times in one reaction. To observe inter-assay variability, each concentration was analyzed five times in independent reactions performed by different users on separate days. The variability of the Tm value was evaluated using variable analysis.

To evaluate the accuracy of the developed assay, a total of 345 clinical samples were collected and tested, including samples containing influenza A, influenza B, SARS-CoV-2, and co-infection viruses. The pre-determined testing for SARS-CoV-2 was conducted by using the Novel Coronavirus (2019-nCoV) Nucleic Acid Detection Kit [BioGerm Medical Technology Co., Ltd. (Shanghai, China)], while the influenza viruses were using the Influenza A Virus and Influenza B Virus Detection Kit [Coyote Bioscience Co., Ltd. (Beijing, China)]. The clinical samples were obtained from the Department of Respiratory and Critical Care Medicine at China-Japan Friendship Hospital (Beijing, China) and this study was approved by the Institutional Review Board of the China-Japan Friendship Hospital (No. 2022-KY-248). The positive clinical samples selected for this study were identified through pre-determined testing, with Ct values spanning the range of positivity (as indicated in Table 2). Detailed Ct values information for the positive samples were listed in Supplementary Table 3. For influenza A, a total of 111 samples were collected, including 40 positives and 71 negatives, and the specimen types included nasopharyngeal swabs (n = 73), oropharyngeal swabs (n = 38). For influenza B, 105 previously tested samples (29 positives and 76 negatives) were tested, and the specimen types included nasopharyngeal swabs (n = 78) and oropharyngeal swabs (n = 27). For SARS-CoV-2, a total of 129 previously diagnosed samples (42 positives and 87 negatives) were tested, including nasopharyngeal swabs (n = 102), and oropharyngeal swabs (n = 27). In addition, two samples with co-infection of influenza A and SARS-CoV-2 and one sample with co-infection of influenza B and SARS-CoV-2 were tested to assess the ability of the assay to detect co-infection, and the specimen types included nasopharyngeal swab (n = 1) and oropharyngeal swabs (n = 2). The detailed clinical samples information was presented in Supplementary Table 4.

Basic statistical values, including mean, standard deviation, and coefficient of variation for the mean Tm (°C) value, were calculated using Excel (Microsoft Corp., Redmond, WA, USA). The LoD for each virus was calculated using probit regression analysis, which determines the concentration at which the target is successfully detected in 95% of replicates (Miller et al., 2019). Overall percent agreement, positive percent agreement (PPA), negative percent agreement (NPA), and Cohen’s kappa coefficient with associated 95% confidence intervals (95% CI) were calculated using VassarStats online.³ Cohen’s kappa values were interpreted according to Landis and Koch (1977).

We developed a rapid and cost-efficient method for simultaneous detection of the SARS-CoV-2 and influenza A/B viruses. The RNA templates were extracted from both MS2-VLPs and clinical samples and then added into the prefabricated reaction solution for one-pot detection. As shown in Figure 1, the assay consisted of two steps. First, the extracted viral RNA was initially reverse transcribed into cDNA. Due to the insufficient amount of the generated cDNA, we then used cDNA as the template to amplify and generate enough amount of single stranded amplicons through asymmetric PCR. The asymmetric PCR uses unequal concentrations of the primers for amplification. In the early stage of amplification, both primers are available the double-stranded amplicons can be generated exponentially through normal PCR. When the limiting primer is depleted in the reaction mixture, only the still available excess primer is able to continue the amplification of target cDNA and thereby produces single stranded amplicons. Therefore, we generated sufficient amount of distinct detectable viral single-stranded amplicons in one reaction by increasing different viral pairs of primers. Subsequently, the designed fluorescent-labeled probes specifically bind to the corresponding viral single-stranded amplicons to produce detectable peaks by MCA. The feasibility of the method was verified using simulated ssDNAs, and the assay results can be directly visualized as a specific single peak by MCA (shown in Figure 2A). This assay can detect five targeted genes using probes labeled with two different fluorophores, allowing for the simultaneous detection of three viruses, which has many advantages, including short amplification time, low cost, easy accessibility, and high-throughput detection capability.

Considering the differences in operational time caused by various sample preparation and RNA extraction methods, our evaluated turnaround time merely refers to the RNA detection time. In a single experiment, it is feasible to simultaneously detect 96 samples, yielding results in roughly 3 h. Since only the extracted viral RNA needs to be added, the hands-on time is reduced, which in turn reduces exposure risk, particularly given the specificity of COVID-19.

The LoD for each target was determined using probit analysis with dilutions of SARS-CoV-2 and influenza A and B virus MS2-VLPs (Table 3 and Figure 2B). The calculated LoD values were 18.37 copies/μL (95% confidence interval [CI]: 15.395–26.758 copies/μL) for the SARS-CoV-2 ORF1ab gene, 14.96 copies/μL (95% CI: 12.862–19.805 copies/μL) for the SARS-CoV-2 N gene, 15.60 copies/μL (95% CI: 12.516–30.316 copies/μL) for the influenza A M gene, and 26.70 copies/μL (95% CI: 21.304–42.783 copies/μL) for the influenza B HA gene. The SARS-CoV-2 N gene showed the lowest LoD and was the most sensitive among the four targets. Additionally, the analytical sensitivity was slightly and nominally lower for influenza B than for influenza A and SARS-CoV-2 (Figure 2C). These results indicate that the assay has high sensitivity, which supports its evaluation as a potential diagnostic tool.

The specificity of the developed assay was verified in two ways. First, silico analysis and blastn analysis were performed, and the results showed that the primers and probes could match the target gene sequences correctly, with no evidence of non-target matches observed. Second, the assay’s specificity was evaluated by measuring cross-reactivity against 58 human respiratory clinical samples known to contain several different diagnostic respiratory viruses. Table 4 showed that the assay returned negative results for the other respiratory viruses and the blank control, with no amplification of the corresponding nucleic acids. These results showed that a specificity of 100% was achieved, with no cross-reactivity with any of the viruses tested.

All four targeted viral genes and the internal control gene were detected in the tested samples. We conducted a quantitative analysis of the coefficient of variation of the Tm value for each target by calculating the mean and standard deviation Tm values. The percent coefficient of variation (%CV) was calculated as a parameter representing intra-assay variability. Testing mixtures included 500 copies/μL (Figures 2D–F), 125 copies/μL (Figures 2G–I), and 31.3 copies/μL (Figures 2J–L) viral concentrations of the influenza A M gene, influenza B HA gene, and SARS-CoV-2 ORF1ab and N gene. The obtained intra-assay %CV values ranged from 0.12 to 1.16%. The inter-assay variability among three runs ranged from 0.34 to 0.99% for all targets, indicating that the assay achieved reliable detection across different viral loads. Supplementary Table 5 showed the detailed results of intra-assay precision and inter-assay precision.

To assess the accuracy of the developed multiplex assay, a total of 345 clinical samples were tested. The overall percent agreement, as well as positive and negative percent agreement for influenza A were estimated at 99.1% (95% CI: 94.4–99.9%), 97.5% (95% CI: 85.3–99.9%), and 100% (95% CI: 93.6–100%), respectively. Only one specimen showed discordant results between our developed assay and the pre-determined assays. This specimen exhibited a weakly positive result for influenza A in the pre-determined assay (with Ct value of Influenza A gene:34.676, Ct value of internal gene:34.191 and Ct value less than 35 regarded as positive), whereas our assay result was negative. This discrepancy may be attributed to the very low presence of viral RNA in the oropharyngeal swab sample and potential RNA degradation. For SARS-CoV-2 and influenza B, our developed assay yielded results identical to the pre-determined assay. Cohen’s kappa statistic ranged from 0.98 (influenza A) to 1.00 (SARS-CoV-2 and influenza B), indicating nearly perfect agreement between the two assays. Detailed concordance metrics between two assays were shown in Table 5. The results for clinical positive samples were presented in Figures 3A–C. Moreover, this method accurately detected multiple viruses within samples, including cases of co-infection with two different viruses, such as two cases of co-infection with influenza A and SARS-CoV-2 (Figures 3D, E) and one case with influenza B and SARS-CoV-2 (Figure 3F). Additionally, we also observed no false positive results among the 234 samples that were predetermined to be negative for SARS-CoV-2, influenza A, and influenza B.