A Melting Curve-Based Multiplex RT-qPCR Assay for Simultaneous Detection of Four Human Coronaviruses

Human coronaviruses HCoV-OC43, HCoV-229E, HCoV-NL63 and HCoV-HKU1 are common respiratory viruses associated with acute respiratory infection. They have a global distribution. Rapid and accurate diagnosis of HCoV infection is important for the management and treatment of hospitalized patients with HCoV infection. Here, we developed a melting curve-based multiplex RT-qPCR assay for simultaneous detection of the four HCoVs. In the assay, SYTO 9 was used to replace SYBR Green I as the fluorescent dye, and GC-modified primers were designed to improve the melting temperature (Tm) of the specific amplicon. The four HCoVs were clearly distinguished by characteristic melting peaks in melting curve analysis. The detection sensitivity of the assay was 3 × 102 copies for HCoV-OC43, and 3 × 101 copies for HCoV-NL63, HCoV-229E and HCoV-HKU1 per 30 μL reaction. Clinical evaluation and sequencing confirmation demonstrated that the assay was specific and reliable. The assay represents a sensitive and reliable method for diagnosis of HCoV infection in clinical samples.


Selection of Primer Sets, and Modification of OC43 Primers
To develop a melting curve-based multiplex RT-qPCR assay for simultaneous detection of four human coronaviruses (HCoV-OC43, -229E, -NL63 and -HKU1), we designed 10 sets of primers for amplification of the four HCoVs. The theoretical Tm values for 10 amplicons were from 81.44 to 84.97 • C (Table 1). In order to select the optimal combination of the primer sets for the four HCoVs, the actual Tm values of 10 specific amplicons were measured by a melting curve analysis. The actual Tm values were from 80.00 to 83.96 • C ( Table 1). The difference between theoretical and actual Tm values varied from 0.01 to 1.01 • C. To correctly distinguish the four HCoVs, the primer set 1 of each HCoV was selected to combine into a detection group ( Figure 1A). The corresponding Tm values were 83.39, 83.96, 81.10, and 80.00 for HCoV-229E, HCoV-OC43, HCoV-NL63, and HCoV-HKU1, respectively ( Figure 1A). Although HCoV-229E, -NL63, and -HKU1 can be clearly distinguished from each other by Tm value, the small difference in Tm values (0.57 • C) between amplicons of HCoV-229E and HCoV-OC43 may result in weak discrimination of both HCoVs.  To improve the discrimination between HCoV-229E and HCoV-OC43 amplicons, we developed a strategy to increase the Tm value of the HCoV-OC43-specfic amplicon by modifying the specific primers. First, we generated a series of GC-rich 9-mer oligonucleotides using the shuffle program in the online software package SMS (http://www.bioinformatics.org/SMS/), and added the GC-rich 9-mer oligos into the 5′ end of the HCoV-OC43 primers. Then, we evaluated these GC-modified HCoV-OC43 primers using Vector NTI (Version 11.5.4; Thermo Fisher Scientific Inc., Bedford, MA, USA; http://www.thermofisher.com/cn/zh/home/life-science/cloning/vector-nti-software/vectornti-advance-software/vector-nti-advance-downloads.html) and melting curve analysis (data not shown). One set of GC-modified HCoV-OC43 primers was selected because the Tm value of the corresponding amplicon was improved from 83.96 to 84.88 °C ( Figure 1B). The GC-modified HCoV-OC43 primers are GC-HCoV-OC43-F: ACGTGCGCGATGTCAATACCCCGGCTGAC and To improve the discrimination between HCoV-229E and HCoV-OC43 amplicons, we developed a strategy to increase the Tm value of the HCoV-OC43-specfic amplicon by modifying the specific primers. First, we generated a series of GC-rich 9-mer oligonucleotides using the shuffle program in the online software package SMS (http://www.bioinformatics.org/SMS/), and added the GC-rich 9-mer oligos into the 5 end of the HCoV-OC43 primers. Then, we evaluated these GC-modified HCoV-OC43 primers using Vector NTI (Version 11.5.4; Thermo Fisher Scientific Inc., Bedford, MA, USA; http://www.thermofisher.com/cn/zh/home/life-science/cloning/vector-nti-software/vectornti-advance-software/vector-nti-advance-downloads.html) and melting curve analysis (data not shown). One set of GC-modified HCoV-OC43 primers was selected because the Tm value of the corresponding amplicon was improved from 83.96 to 84.88 • C ( Figure 1B). The GC-modified HCoV-OC43 primers are GC-HCoV-OC43-F: ACGTGCGCGATGTCAATACCCCGGCTGAC and GC-HCoV-OC43-R: CCAGGCGGTGGCTCTACTACGCGATCCTG ( Figure 1B). Compared with the original OC43-F/R primers, the GC-HCoV-OC43-F/R primers also appeared to improve the amplification reactions ( Figure S1). When the GC-HCoV-OC43-F/R primers were added into the detection group by replacing the original OC43-F/R primers, four amplicons of HCoVs were clearly distinguished from each other by melting peaks ( Figure 1C). The corresponding Tm values were 84.88, 83.39, 81.10, and 80.00 for HCoV-OC43, -229E, -NL63, and -HKU1, respectively. These imply the establishment of a melting curve-based quadruplex RT-qPCR for detection of the four HCoVs. To obtain well performance, the primer concentration of the assay was optimized as 0.8 µM, 0.8 µM, 0.8 µM and 0.2 µM of HCoV-229E, -NL63, -HKU1, and -OC43 specific primers, respectively, by orthogonal experiments (Supplemental result and Figure S2).

Specificity of the Melting Curve-Based Multiplex RT-qPCR
The specificity of the melting curve-based multiplex RT-qPCR was assessed using 13 common respiratory viruses. Except the four HCoVs, no amplification was observed for other 9 respiratory viruses, as well as the negative control ( Figure 2A). Melting analysis showed that except the four HCoVs specific melting peaks, there was no melting peak for other 9 respiratory viruses and negative control ( Figure 2B). Furthermore, there was also no cross-amplification among these four HCoVs. These indicated that the assay had good specificity for detection of the four HCoVs. GC-HCoV-OC43-R: CCAGGCGGTGGCTCTACTACGCGATCCTG ( Figure 1B). Compared with the original OC43-F/R primers, the GC-HCoV-OC43-F/R primers also appeared to improve the amplification reactions ( Figure S1). When the GC-HCoV-OC43-F/R primers were added into the detection group by replacing the original OC43-F/R primers, four amplicons of HCoVs were clearly distinguished from each other by melting peaks ( Figure 1C). The corresponding Tm values were 84.88, 83.39, 81.10, and 80.00 for HCoV-OC43, -229E, -NL63, and -HKU1, respectively. These imply the establishment of a melting curve-based quadruplex RT-qPCR for detection of the four HCoVs.

Specificity of the Melting Curve-Based Multiplex RT-qPCR
The specificity of the melting curve-based multiplex RT-qPCR was assessed using 13 common respiratory viruses. Except the four HCoVs, no amplification was observed for other 9 respiratory viruses, as well as the negative control ( Figure 2A). Melting analysis showed that except the four HCoVs specific melting peaks, there was no melting peak for other 9 respiratory viruses and negative control ( Figure 2B). Furthermore, there was also no cross-amplification among these four HCoVs. These indicated that the assay had good specificity for detection of the four HCoVs.
From 61 HCoVs positive clinical samples identified by the melting curve-based multiplex RT-qPCR, we selected 21 samples having relatively strong melting peak for sequencing confirmation. Blast using the obtained sequences as query sequence revealed there were 2 HCoV-229E, 9 HCoV-OC43, 9 HCoV-NL63, and 1 HCoV-HKU1 sequences ( Figure S4). Each obtained sequence corresponded to the right HCoV as judged by the melting curve-based multiplex RT-qPCR assay (Figure S5), indicating a 100% specificity at least for these sequenced samples.

Discussion
Because HCoVs are one of the most commonly observed pathogens causing respiratory illnesses [2,10,11], the development of reliable and rapid detection method is greatly needed. The multiplex RT-qPCR assay has high sensitivity and specificity, and was often used in the detection of respiratory viruses [27][28][29][30]. A hydrolysis probe-based multiplex RT-qPCR assay was previously developed for HCoV detection [12]. However, difficulty in probe design for highly divergent viruses, as well as high cost in probe synthesis, limits the use of probe-based multiplex RT-qPCR assay.
As an alternative to hydrolysis probes, fluorescent dyes (e.g., SYBR geen I) that bind preferentially to double-stranded DNA (dsDNA) have been widely used in the qPCR because they are cheaper than hydrolysis probe. However, because the binding of fluorescent dyes to dsDNA lacks sequence specificity, they are rarely used in the multiplex qPCR. Recently, several melting curve-based multiplex RT-qPCR assays using fluorescent dye were developed for simultaneous detection of bacterial pathogens [33][34][35]. Up to our knowledge, no similar method was used for detection of human viruses. In this study, we developed a melting curve-based multiplex RT-qPCR assay for the detection of four HCoVs (HCoV-229E, -OC43, -NL63, and -HKU1). Because SYBR Green I inhibits PCR reaction and preferentially binds to GC rich DNA sequences, which can affect the melting curve analysis [36], we used fluorescent dye SYTO 9 to replace SYBR Green I in the assay. In order to correctly distinguish the four HCoVs in a single RT-qPCR reaction, we designed four sets of primers to generate four specific PCR products with obviously different Tm values ( Table 1). The melting curve analysis showed that the Tm values of the specific products for HCoV-229E, -OC43, -NL63, and -HKU1 were 83.39 • C, 83.96 • C, 81.10 • C, and 80.00 • C, respectively ( Figure 1A). To further increase the discrimination between HCoV-229E and HCoV-OC43, we developed a strategy to improve the Tm value of HCoV-OC43 amplicon by introducing a 9-mer GC-rich oligonucleotide into the 5 end of the specific primers. The GC-modified OC43 primers generated an amplicon with Tm value of 84.88 • C, 0.92 • C higher than that obtained with the original OC43 primers ( Figure 1B). Under the new primer sets, the melting curve-based multiplex RT-qPCR assay can well detect and distinguish the four HCoVs ( Figure 1C).
The detection sensitivity of the new assay was determined using the standard strains or in vitro transcribed RNA of the four HCoVs. The detection sensitivity was 3 × 10 2 copies/µL for HCoV-OC43, and 3 × 10 1 copies/µL for HCoV-NL63, HCoV-229E and HCoV-HKU1 (Figure 3). The specificity of the assay was assessed using a panel of respiratory viruses commonly found in children with acute respiratory infection. No amplification signal was observed for any of the strains except the four targeted HCoVs (Figure 2). Using the new assay, we detected 88 clinical samples. Among them, 61 were identified as HCoV-positive, including 9 dual infections with two of the four HCoVs. To confirm the results, we selected 21 samples (including 9 HCoV-OC43, 9 HCoV-NL63, 2 HCoV-229E and 1 HCoV-HKU1) from these positive samples for sequencing confirmation. The sequence and blast results of the 21 samples were completely consistent with that obtained by the melting curve-based multiplex RT-qPCR assay (Figures S4 and S5). Furthermore, non-specific amplification was not observed when pooled samples were used (data not shown). These results indicated that the new assay was reliable.
In summary, we developed a melting curve-based multiplex RT-qPCR assay for simultaneous detection of HCoV-229E, HCoV-OC43, HCoV-NL63 and HCoV-HKU1. The assay is a sensitive and reliable method for screening of HCoV infection in clinical samples (e.g., nasopharyngeal swabs). In addition, we developed a strategy to improve the Tm value of amplicon by adding GC-rich oligonucleotides to the 5 end of the related primers. A total of 88 nasopharyngeal swabs from outpatient children with fever and respiratory symptoms in Shanghai Nanxiang Hospital were used in this study for clinical evaluation. The samples were obtained from two previous studies with 59 being identified as HCoV-positive using a multiplex RT-PCR [37,38]. During this study, these samples were re-detected using the same method. Only 29 were identified as HCoV-positive, which may be ascribed to degradation of viral RNA among these samples caused by multiple freezing and thawing. The study was approved by the Ethics Committee of Taizhou

Primer Design
All available genomic sequences of HCoV-229E, HCoV-OC43, HCoV-NL63, and HCoV-HKU1 were downloaded from GenBank, and aligned using MEGA 5.05 (http://www.megasoftware.net/) to find the most conserved region. The M gene of HCoV-229E, and N genes of HCoV-OC43, HCoV-NL63, and HCoV-HKU1 were selected as target region for primer design. The primers were designed using Primer Express 3.0 software (Applied Biosystems, Foster City, CA, USA), and their specificity was confirmed by a Nucleotide BLAST search. In order to develop a multiplex RT-qPCR assay based on melting analysis for simultaneous detection of the four HCoVs, the primers were designed to generate amplicons with different melting temperature (Tm) for different coronavirus. Three sets of primers were designed for HCoV-229E, -OC43, and -HKU1, and one set of primers was for HCoV-NL63 ( Table 1). The theoretical Tm of each amplicon was predicted using the online tool Oligo Calc: Oligonucleotide Properties Calculator (Northwestern University, Chicago, IL, USA), (http://biotools. nubic.northwestern.edu/OligoCalc.html) based on the amplicon sequence [39]. Actual Tm of the amplification product was determined by the melting curve analysis.

RT-qPCR and Melting Curve Analysis
The RT-qPCR was performed using QIAGEN OneStep RT-PCR Kit (QIAGEN, Hilden, Germany) with SYTO 9 (Life technologies, Carlsbad, CA, USA) as the fluorescent dye. Thirty µL reactions including 3 µL template input were run on a Light Cycler 96 RT-qPCR System (Roche Diagnostics, Mannheim, Germany). Reaction conditions were: 30 min RT at 50 • C, 15 min at 94 • C for inactivation of reverse transcriptase (RT), followed by 40 cycles of 94 • C for 30 s, 50 • C for 30 s and 72 • C for 1 min. Melting curve analysis was performed under the condition of 95 • C for 60 s, 40 • C for 60 s, 65 • C for 1 s, then followed by a slow increase from 65 • C to 95 • C with a speed of 0.07 • C per second.

Specificity and Sensitivity of the Melting Curve-Based Multiplex RT-qPCR
The specificity of the melting curve-based multiplex RT-qPCR was assessed using 13 common respiratory viruses including 9 standard strains and 4 clinical isolates. The 9 standard strains were described above and the 4 clinical isolates were PIV1, PIV2, PIV4, and HCoV-NL63. Because there was no standard strain and clinical isolate available for HCoV-HKU1, we synthesized plasmid containing HCoV-HKU1 N gene downstream of the T7 promoter. HCoV-HKU1 RNA was obtained via in vitro transcription. Using the same strategy, we also obtained the RNA stocks of HCoV-229E, HCoV-OC43, and HCoV-NL63 for sensitivity experiments. Serially diluted RNA stocks for four HCoVs from 1 × 10 6 to 1 × 10 1 copies/µL were used to determine the detection limit of the melting curve-based multiplex RT-qPCR.

Evaluation of the Melting Curve-Based Multiplex RT-qPCR Using Clinical Samples
A total of 88 clinical samples were used to evaluate the performance of the melting curve-based multiplex RT-qPCR for four HCoVs detection. To avoid the possibility of the non-specific PCR product with same or similar Tm value to specific products, we confirmed the detection results by sequencing the amplicons. Because the weak melting peak implies few amplification products that are not enough for direct DNA sequencing, we selected the amplification products with strong melting peaks for sequencing confirmation. Of HCoV-positive clinical samples identified by the multiplex RT-qPCR, 21 were subjected to sequencing confirmation, including 2 HCoV-229E, 9 HCoV-OC43, 9 HCoV-NL63, and 1 HCoV-HKU1. The sequencing reactions were performed by Shanghai Biosune Biotechnology Co., Ltd. (Shanghai, China).