Denaturing and dNTPs reagents improve SARS-CoV-2 detection via single and multiplex RT-qPCR

Ethics statement

This study was approved by the Research Ethics Committees of the Industrial University of Santander, the Chicamocha Clinic, and the Chicamocha Clinical Laboratory (L3C) (Bucaramanga, Santander, Colombia). The study participants were patients either hospitalized-diagnosed with SARS-CoV-2 or persons that presented to the emergency room or volunteers. All participants provided written informed consent and voluntarily participated in our study. Further, all participants were kept anonymous and were informed that the present study was conducted strictly for research purposes and that its outcomes were not intended to serve as treatment or diagnosis.

SARS-CoV-2 sample collection and storage

Nasopharyngeal swab samples were acquired by L3C personnel. Two samples were obtained per study participant, one for clinical diagnosis, as authorized by the Ministry of Health and Social Protection of Colombia, and the other for our study. The samples were placed in Universal Transport Medium (UTM) and stored at −80°C until required for downstream analyses (Rogers et al., 2020). The samples were collected over the course of one week and then appropriately packaged according to the sanitary legislation of Colombia (Ministry of Health and Social Protection, 2020) prior to their transportation and processing. All samples were shipped to the Central Research Laboratory of the Industrial University of Santander Health Faculty (LCI-FS-UIS) and maintained at 4°C while the informed consents were processed and digitalized.

Afterwards, the samples were aliquoted in a Class II, Type A2 Biosafety Cabinet (Thermo Fisher Scientific). A total of three aliquots were obtained from each sample, including a 200-μL aliquot for total RNA extraction, which was used immediately, and two 650-μL aliquots used as backup samples, which were stored at −80°C.

SARS-CoV-2 genome database consolidation

Due to the appearance of mutations in several genes of the SARS-CoV-2 reference genome that could generate false negatives with the primers and probes authorized for identification by RT-qPCR. A region coding for the accessory protein ORF8 was selected, hypothesizing that it did not have a high mutation frequency. Therefore, monthly consensus sequences were generated to determine its identity pattern with respect to new genomes. Publicly available SARS-CoV-2 genome sequences were obtained from the GenBank and GISAID databases (RRID:SCR_002760; RRID:SCR_018251) (Sayers et al., 2021; Shu & McCauley, 2017). To gather data from the GenBank database, a Python script (RRID:SCR_008394) was executed in the GUANE-1 High Performance and Scientific Computing Center of the Industrial University of Santander (SC3UIS) through the Entrez Programming Utilities interface using the Biopython package (RRID:SCR_013249; RRID:SCR_007173) (Cock et al., 2009; Geer et al., 2010), whereas all GISAID data were manually downloaded. Genomes with uncharacterized regions/nucleotides were discarded using another Python script (RRID:SCR_008394), which was used to conduct monthly FASTA sequence alignments using the MAFFT software (RRID:SCR_011811) (Katoh & Standley, 2013) coupled with previously described DNA loss model parameters (Martínez-Pérez et al., 2002, 2007). The consensus sequences were generated using BioEdit version 7.2 (RRID:SCR_007361) with a 100% threshold frequency (Hall, 1999).

Design and synthesis of ORF8-specific primers, probes, and substrate oligos

Using the SARS-CoV-2 (NC_045512) reference genome and consensus sequences from January to April 2020 (Zhu et al., 2020), an approximately 150-base pair (bp) region of the ORF8 gene was selected based on the secondary structure of its RNA transcript, which in turn was predicted using the algorithms proposed by Zuker (Zuker & Jacobson, 1998) via the Mflod software (RRID:SCR_001360) (Zuker, 2003). Thus, the last set containing the codons of the central region of the ORF8 gene, which encodes the secretion protein ORF8, was chosen because it allows proper viral adhesion to the host cell (Chan et al., 2020). The obtained sequence was used as a template to create primers, a TaqMan FAM-BBQ probe, and substrate oligos for RT-qPCR. The specificity of the aforementioned molecules was confirmed via GenBank BLAST analyses (RRID:SCR_004870) (Ye et al., 2006). Genes E and N from the Berlin and CDC protocols were used as controls (Biotek, 2020; Corman et al., 2020).

All molecules were synthesized by Bioneer (Korea). The primers were purified via separation on a reverse-phase cartridge, whereas the probe and substrate oligos were purified via high-performance liquid chromatography (HPLC) and polyacrylamide gel electrophoresis (PAGE), respectively. The ORF8 RT-PCR conditions were implemented as described by the Berlin protocol (Corman et al., 2020) using the aforementioned molecules coupled with the 2019-nCoV TaqMan RT-PCR Kit (Norgen Biotek Corp) developed by the CDC (Biotek, 2020).

Preparation of denaturing solutions and adjustment of dNTP concentrations

A:T and G:C ratios were calculated based on the monthly SARS-CoV-2 consensus sequences to obtain an average for each nucleotide. These averages were then used to determine the minimum concentrations of TEA (ABCAM-USA), DMSO (Scharlab-Spain), and dNTPs (100 mM each, Promega-USA) in molecular-grade ultrapure water (Promega-USA), in addition to the MgSO₄ concentration recommended by the Berlin protocol.

RNA extraction

Total RNA extraction was conducted using the MagMAX™ Viral/Pathogen II (MVP II) Nucleic Acid Isolation Kit (2000 RXNs) (Applied Biosystems-USA) using a KingFisher Duo Prime (5400110) DNA/RNA extraction system according to the manufacturer’s instructions (Thermo Fisher Scientific-USA) (Fang et al., 2007).

SARS-CoV-2 single and multiplex RT-qPCR

RT-qPCR was conducted using the ORF8-specific primers and probe designed herein, in addition to the E (Berlin protocol) and N genes (N1-N2; CDC protocol). The RNase P gene was used as an external control, as proposed by both of the aforementioned protocols. The reactions were conducted using the SuperScript III One-Step RT-PCR System with Platinum Taq DNA Polymerase (Ref. 11732088; Invitrogen-USA) and the 2019-nCoV TaqMan RT-PCR Kit (Ref. TM67100; Norgen-Biotek-Canada). Each reaction for each diagnostic system was conducted in either 15- or 25-μL reaction volumes consisting of 2 μL of patient-derived purified RNA, 2× One-Step RT-PCR Master Mix, 2× nuclease-free buffer, and the respective primers/probe at the concentrations recommended by the CDC. Moreover, a denaturing stock solution was added to obtain a final concentration of 0.7% TEA, 0.2% DMSO, and 0.8 mMol MgSO₄. The dNTP reagent had a final concentration of 12 mMol dATP-dTTP, 10 mMol dCTP-dGTP, and 0.8 mMol MgSO₄. The reaction conditions were the following: 55°C for 15 minutes, 95°C for 3 minutes followed by 45 cycles of 95°C for 15 s and 58°C for 30 s for the SuperScript™ III One-Step RT-PCR kit; and 95°C for 3 s followed by 55°C for 20 s for the 2019-nCoV TaqMan RT-PCR kit. Fluorescence signals were quantified using a QuantStudio 1 Real-Time PCR System (No. A40427) in a 96-well 0.2 μL block (Thermo Fisher Scientific-USA).

The above-described procedure was conducted using two different Multiplex One-Step RT-qPCR protocols. In the first instance, the primers and probes for the E, ORF8, and N (N1) genes were mixed, whereas the other reaction was performed by mixing the N1 and N2 sets of the N gene. The RNase P gene was independently assessed in both cases. All reactions were performed as described above.

Calculation of sensitivity, specificity and predictive values of SARS-CoV-2

To calculate the sensitivity, specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) of each of the reactions with the single and multiplex primers used, a script was implemented in R using the Caret package to calculate the values generated by each RT-qPCR with their respective primers/probes (https://github.com/GenomicUIS/Sensitivity-specificity-PPV-and-NPV-for-SARS-CoV-2.git) (Kuhn, 2008; R Core Team 2020). Cycle threshold (Ct) was defined for each of the primers in the single reactions of each of the processed samples, where, the value of 18-35 Ct was considered true positive, Ct 35 > false positive, Ct 18 < true negative. Subsequently, the Cts values obtained were pooled and evaluated per sample. Therefore, if the result generated by a set of primers/probes was at the above-mentioned threshold, it was considered a true positive identification. Whereas, results outside the threshold were considered false negative and indeterminate. Cts were identical for multiplex RT-qPCR reactions, but the results were evaluated as a whole.

SARS-CoV-2 GenBank and GISAID databases

A total of 19,317 genomes were retrieved from the GenBank database from January to October 2020 based on our search criteria, whereas the GISAID database rendered 107,259 full genomes from January to December 2020. In both cases, the number of available sequences increased substantially each month. However, the frequency of base substitutions in the monthly consensus sequences for the E, ORF8, and N genes was higher in the GISAID database compared to the GenBank database (Figure 1).

Figure 1. Primers and probes to RT-qPCR to SARS-CoV-2.

The nucleotides correspond to: forward primer in green box and lower primer blue box. The Probes is in yellow box. The conceptual translation is indicated in the top of each alignment. The number indicate the nucleotide combination. The nucleotides values combination by position corresponds to international nomenclature. The nucleotide position according to SARS-CoV-2 reference genome is indicated in the lower consensus alignment.

Primer, probe, and substrate oligo design for SARS-CoV-2 detection based on RNA secondary structure

The region of the E gene employed herein exhibited a 113 bp length, whereas the amplicons of the N1-N2 system of the N gene were 72 and 67 bp in length, respectively, all of which exhibited loop-bubble structures. Interestingly, 154 bp of the first half of the ORF8 gene presented a loop-bubble structure, which was similar to that of the other two genes. Nevertheless, the third loop of the N1 system and the second loop of the ORF8 gene were formed via four and seven canonical pairings, respectively. These structures can be considered scorpion-type primers or probes, which are often used in RT-qPCR; however, the loop-bubble structures used for these purposes are typically formed via 2–6 canonical pairings (Figure 2). Table 1 summarizes the primer sequences used in this study.

Figure 2. Secondary structures of the primers and probes used for genes E, ORF8 and N (N1-N2).

Stem and stem-bubble structures are observed in the 5' and 3' direction. The numbers represent the amount of nucleic acids used for each set and the parentheses individually delimit each primer and probe within each segment.

SARS-CoV-2 detection via E, ORF8, and N RT-qPCR

A total of 49 samples were collected from October 25, 2020, to January 21, 2021, of which 22 tested positive and 27 were negative. The October samples were used to confirm that the E, ORF8, and N genes could be used as effective indicators to detect SARS-CoV-2 via RT-qPCR (Table 2, Figure 3). However, given that our secondary structure and nucleotide ratio analyses indicated that the A:T ratio was always between 63% and 70% regardless of the consensus sequence, denaturing and dNTP solutions were also incorporated in these reactions (see the Methods for more details). These conditions generally improved the Ct values of the reactions compared to those of the commercial procedures; however, some exceptions were observed (Table 2, Figure 3).

Figure 3. RT-qPCR curves for the detection of SARS-CoV-2 from patient-derived samples.

RNA obtained from patients and primers and probes corresponding to the E gene (WHO), ORF8 (this study), and the N gene (N1-N2; CDC). The green, yellow, and red arrows indicate the curves generated by denaturalization (Den), the dNTP reagent, and commercial kits, respectively. The RNase P gene was used as a control.

Given that all samples were freshly acquired and never stored, we concluded that the gradual loss of amplicon systems and adjuvant solutions began in December. This was confirmed first for the E gene system, then for the N2 system, and finally for ORF8. This trend was determined using ten samples collected from December 14 to 21, 2020. The results of our experiments could be generally classified into three outcomes depending on the viral detection system: (1) the system worked following the manufacturer’s instructions with the addition of the denaturing solution (2); the system rendered positive results with the denaturing solution but negative without it; and (3) the opposite of the second outcome (Table 2).

In silico estimation of E, ORF8, and N RT-PCR primer and probe combinations

Our analysis of the global SARS-CoV-2 mutation patterns based on the GenBank and GISAID databases indicated that most mutations began in March 2020, and subsequently increased significantly, relative to the SARS-CoV-2 reference genome (GenBank access: NC_045512.2). This trend was particularly noticeable for the quencher and forward primer sequences of the ORF8 and N genes. Moreover, the potential primer and probe combinations for each gene increased substantially starting in November according to the GISAID database and were significantly higher in December. For example, the ORF8 system and the N1 set exhibited 3.4×10¹¹ and 1.7×10¹² combinations, respectively, whereas the N2 set and E had only 1.4×10⁵ and 32 combinations (Figure 1).

Multiplex RT-qPCR

The multiplex RT-qPCR controls of systems E, ORF8, and N (N1) with the positive samples 02 and 03 rendered the expected results, which were thereafter confirmed using samples 05 and 06 obtained in October 2020. Nevertheless, using the same reaction conditions, the denaturing solution decreased the Ct values for the positive samples obtained in October but enhanced those of the samples obtained in November. In contrast, the dNTP solution had the opposite effect (Table 3, Figure 4).

Figure 4. Multiplex RT-qPCR (Mp) curves for the detection of SARS-CoV-2 using E-, ORF8-, and N (N1)-specific.

The RNA samples from 6 different patients are indicated with the respective numbers. For each curve, the first number in the nomenclature indicated the top result and the second the lower result. The RNase P gene was used as a control. The arrows and nomenclature are the same as in Figure 3.

The influence of the denaturing and dNTP solutions was further confirmed using nine samples obtained from December 4 to 7, 2020, and from January 14 to 15, 2021, using the N1–N2 primers and probe as described by the CDC. A total of six samples exhibited the originally documented pattern and the three remaining samples tested negative, both in terms of curve trends and Ct values (Table 3, Figure 5). Therefore, the initial multiplex reaction effectively detected the SARS-CoV-2 virus in the nine samples obtained from January 19 to 29, 2021 (Table 3).

Figure 5. RT-qPCR Multiplex (Mp) curves for the detection of SARS-CoV-2 using from N gene (N1-N2).

Primers and probes, as indicated by the CDC coupled with the denaturalization reagent. The RNA samples from 9 patients are indicated with the respective numbers. The multiplex curves corresponding to N1-N2 are indicated at the top, whereas the results corresponding to the RNase P primers and probe used as controls are indicated below. The abbreviators and nomenclature are the same as in Figure 1.

Sensitivity, specificity and predictive values of SARS-CoV-2

The performance of single and multiplex RT-qPCRs with the solutions used showed that single denaturing was more sensitive than the commercial solution, but the latter is 33% more specific than the former. As for the multiplex reactions, the commercial solution showed non-significant difference in sensitivity with respect to the denaturing solution. Similarly, the specificity of the single reactions was better for the commercial solution with respect to the denaturing solution, but in the multiplex reactions this difference was not evident in the detection of true negatives as both solutions yielded results of 0. dNTPs were not evaluated due to the number of positive results (Table 4).