Development and validation of RT-PCR assays for testing for SARS-CoV-2

Abstract

Background

The recent emergence and rapid global spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) demonstrates the urgent need for laboratory-developed assays for clinical diagnosis and public health interventions in the absence of commercial assays.

Methods

We outline the progression of reverse-transcriptase polymerase chain reaction (RT-PCR) assays that were developed and validated at the Alberta Precision Laboratories, Public Health Laboratory, Alberta, Canada, to respond to this pandemic. Initially, testing was performed using SARS-CoV-2–specific and pan-coronavirus gel-based assays that were soon superseded by real-time RT-PCR assays targeting the envelope and RNA-dependent RNA polymerase genes to accommodate the high anticipated volumes of samples. Throughput was further enhanced by multiplexing the different targets together with the co-detection of an internal extraction control.

Results

These assays are comparable in sensitivity and specificity to the assays recommended by the World Health Organization and the US Centers for Disease Control and Prevention.

Conclusions

The availability of real-time RT-PCR assays early in the pandemic was essential to provide valuable time to local health authorities to contain transmission and prepare for appropriate response strategies.

Background

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was first detected in Wuhan, Hubei, China; it causes coronavirus disease 2019 (COVID-19) (1, 2) and has resulted in a global pandemic (3). Accurate diagnosis using sensitive and specific tests to identify and isolate cases is key to managing the pandemic. Currently, real-time reverse-transcriptase polymerase chain reaction (rtRT-PCR) is the recommended test for the diagnosis of acute cases.

Early in the pandemic, the World Health Organization (WHO) published a list of protocols that included both specific and pan-coronavirus PCR tests for the identification of SARS-CoV-2 (4), followed by widely used rtRT-PCR assays by the Charité Institute of Virology, Universitätsmedizin Berlin (referred to as the E-Sarbeco and RdRp-SARSr assays) (5). The US Centers for Disease Control and Prevention (US CDC) have also developed detection kits that are available to public health laboratories (6). The chronology for the availability of these tests has been summarized by another group (7). The viral genes targeted by the published protocols include the ORF 1ab, nucleoprotein (N), envelope (E), spike (S), and RNA-dependent RNA polymerase (RdRp) genes.

The analytical sensitivities and efficiencies of the primer–probe sets used in the four most common SARS-CoV-2 rtRT-PCR assays developed by the Chinese Center for Disease Control (China CDC) (8), US CDC (9), Charité Institute (4), and Hong Kong University (HKU) (10) have been compared (11). The most sensitive primer–probe sets were the E-Sarbeco, HKU-ORF1, HKU-N, and US CDC-N1; the RdRp-SARSr primer–probe set had the lowest sensitivity.

The public health response to the pandemic by Alberta Precision Laboratories, Public Health Laboratory (ProvLab), Alberta, Canada, is outlined in detail elsewhere (12); it included the development and implementation of two rtRT-PCR assays, including the E-Sarbeco assay and an in-house assay targeting the RdRp gene for the identification of cases. These assays were subsequently multiplexed with the detection of an internal extraction and inhibition control to streamline workflow and increase testing capacity. The validation of these assays and their comparison with the WHO and US CDC recommended assays is presented here.

METHODS

Design of primers and probes

The sequences and source of the oligonucleotides used are provided in Table 1. Different nucleic acid tests were used as the demand for SARS-CoV-2 testing rapidly increased. At the outset, two gel-based assays targeting the RdRp gene, one for the detection of a range of coronaviruses (pan-coronavirus) and one for the specific detection of SARS-CoV-2, were designed, validated, and deployed for patient testing. Sequences representing alpha, beta, and gamma coronaviruses—including transmissible gastroenteritis virus; canine coronavirus; porcine respiratory coronavirus; feline coronavirus; porcine epidemic diarrhea virus; porcine hemagglutinating encephalomyelitis virus; murine hepatitis virus; human enteric coronavirus; human coronaviruses 229E, NL63, OC43, and HKU1; MERS-CoV; avian infectious bronchitis virus; turkey coronavirus; SARS-CoV; and bat coronaviruses—were included in the alignment for primer design (13).

Table 1:

Primers and probes for the detection of SARS-CoV-2

Target	Name	Sequence (5‘–3')	Source
RdRp gene	RdRP_WuCoV-For_qPCR	TTTTAACATTTGTCAAGCTGTCACG	In house
RdRp gene	RdRP_WuCoV-Rev_qPCR	GTTGTAAATTGCGGACATACTTATCG	In house
RdRp gene	RdRP_WuCoV_Prb_qPCR	CACTTTTATCTACTGATGGTAAC (VIC/MGB)	In house
E gene	COVID19_E_For_V2	GAGACAGGTACGTTAATAGTTAATAGCG	In house
E gene	COVID19_E_Rev_V2	CAATATTGCAGCAGTACGCACAC	In house
E gene	COVID19_ E_MGB_FAM	CTAGCCATCCTTACTGCG (FAM/MGB)	In house
MS2	MS2-TM2-F	TGCTCGCGGATACCCG	Dreier et al (14)
MS2	MS2-TM2-R	AACTTGCGTTCTCGAGCGAT	Dreier et al (14)
MS2	MS2-TM2_ATTO647	ACCTCGGGTTTCCGTCTTGCTCGT(ATTO647/Iowa Black)	Dreier et al (14)
E gene	2019nCoV_E_Sarbeco_For	ACAGGTACGTTAATAGTTAATAGCGT	Corman et al (5)
E gene	2019nCoV_E_Sarbeco_Rev	ATATTGCAGCAGTACGCACACA	Corman et al (5)
E gene	2019nCoV_E_Sarbeco_Prb_FAM	ACACTAGCCATCCTTACTGCGCTTCG (FAM/BHQ-1)	Corman et al (5)
RdRp gene	WuhanCov_pol_For	TTATGGGTTGGGATTATCCTAAATGTGA	In house
RdRp gene	WuhanCov_pol_Rev	GTTGTGGCATCTCCTGATGAGGTTCCACC	In house
RdRp gene	PanCoV_Pol_For	TGATGGGTTGGGACTATCCTAARTGTGA	In house
RdRp gene	PanCoV_Pol_Rev	ATTGTATGCTGTGAACAAAATTCATGWGG	In house

SARS-CoV-2 = Severe acute respiratory syndrome coronavirus 2; MGB = Minor groove binding

The gel-based assays were replaced by two rtRT-PCR assays targeting the E (5) and RdRp genes (designed in house). The E gene assay often resulted in weak and late amplification curves and was thus modified with the addition of GC clamps at the 3’ end of the primers and the shortening and addition of a minor groove-binding (MGB) moiety to the probe. The modified E gene and the RdRp assays were multiplexed with the detection of MS2 phage (14) used as a spiked extraction and inhibition control. Probes with MGB proteins were purchased from Applied Biosystems (ABI, Foster City, California, USA); the MS2 probe with an ATTO647 dye was purchased from Integrated DNA Technologies (IDT, Coralville, Iowa, USA), and all other primers and probes were from the University Core DNA services (University of Calgary, Calgary, Alberta).

Gel-based and real-time RT-PCR assays

The gel-based assays were performed using the One-Step RT-PCR kit from Qiagen (Toronto, Ontario, Canada). The 25 µL reaction included 5 µL of template with 5 µL 5x One-Step RT-PCR Buffer, 5 µL of Q-solution, 1 µL of enzyme mix, 5 U of RNaseOUT (Invitrogen, Carlsbad, California), 0.6 µM of sense and antisense primers, and 0.4 mM dNTPs. For both gel-based assays, RT was performed at 50^oC for 30 minutes followed by PCR activation at 95^oC for 15 minutes and amplification for 45 cycles. The SARS-CoV-2–specific assay consisted of denaturation at 94^oC, annealing at 62^oC, and extension at 72^oC for 30 seconds each. The pan-coronavirus assay consisted of denaturation at 94°C, annealing at 55^oC, and extension at 72^oC for 60 seconds each.

Singleplex rtRT-PCR was performed using TaqMan^® Fast Virus One-Step RT-PCR Master Mix (ABI), 0.8 µM each of sense and antisense primers and 0.2 µM of the probes; five microliters of this mix was combined with 5 µL of extracted RNA. Multiplex rtRT-PCR was performed using the same master mix with 0.8 µM each of sense and antisense primers; 0.2 µM of the probes for the SARS-CoV-2 targets; and 0.2 µM and 0.1 µM of the primers and probe, respectively, for the MS2 target. Primer and probe concentrations were optimized for preferential amplification of the SARS-CoV-2 target genes to prevent any competitive inhibition from MS2 amplification. Ten microliters of the RNA was combined with 10 µL of the master mix to improve sensitivity. For all rtRT-PCR procedures, the RT step was performed at 50^oC for 5 minutes followed by incubation at 95^oC for 20 seconds. Amplification included 45 cycles of denaturation at 95^oC for 3 seconds, followed by annealing, extension, and data acquisition at 60^oC for 30 seconds on the 7500 Fast Real-Time PCR system (ABI).

Preparation of RNA transcripts for sensitivity studies

Long oligonucleotide sequences (gblocks), including the detection region with flanking T7 and SP6 RNA polymerase-promoter binding sites, were designed and purchased for all gene targets from IDT. RNA transcription was performed using the RiboMAX™ SP6 RNA Production System (Promega, Madison, Wisconsin, USA). The transcribed RNA was spectrophotometrically quantified for the calculation of copy numbers.

Extraction of viral nucleic acid

Viral RNA from the different specimen types was extracted on one of three platforms using manufacturers’ instructions: easyMAG^® (BioMerieux, Saint-Laurent, Quebec, Canada) with associated reagents; the MagMAX Express 96 or KingFisher Flex automated extraction and purification systems (Thermo Fisher Scientific, Waltham, Massachusetts, USA) with either the MagMAX-96 Viral RNA Isolation Kit (ABI) or the Maxwell HT Viral TNA custom kit (Promega); or the Hamilton STARlet automated extractor (Hamilton, Reno, Nevada, USA) with the Maxwell HT Viral TNA custom Kit. The majority of the specimens tested were nasopharyngeal swabs. Alternate specimen types included throat swabs, nasal swabs, nasopharyngeal aspirates, stool suspensions, bronchoalveolar lavages, endotracheal secretions, blood, and urine. The sample input and output volumes were 200 µL and 110 µL, respectively.

Analytical sensitivity, specificity, reproducibility, and dynamic range of RT-PCR

The analytical sensitivity for the real-time assays was determined by testing 10-fold serial dilutions of quantified in vitro RNA in triplicate on three independent runs. The 95% limits of detection (95% LOD) were calculated by probit analysis. The range of viral loads tested in copies/reaction was 1.01E+07 to 1.01E+00 for the RdRp gene and 1.66E+07 to 1.66E+00 for the E gene. Linear regression fitting of the log RNA copies versus cycle threshold (Ct) allowed for the calculation of PCR efficiency.

The same RdRp in vitro RNA was used for testing by the gel-based assays in triplicate, and the endpoint was determined by gel electrophoresis and Sanger sequencing of the amplified product. Quantitated viral RNA with the pfu/mL for SARS-CoV-2 was obtained from the National Microbiology Laboratory (NML, Winnipeg, Manitoba, Canada), serially diluted, and tested by all the assays for sensitivity comparison.

Specificity of the assay was determined by testing high-viral-load samples of several RNA and DNA viruses and bacteria with clinical symptoms overlapping those of COVID-19. The pathogens tested included coronaviruses (NL63, OC43, 229E, HKU1, MERS-CoV, SARS-CoV-1), influenza A (pdm09 H1N1, H3N2), influenza B, respiratory syncytial virus (A and B), parainfluenza virus (1, 2, 3, 4a, 4b), rhinovirus, enterovirus, adenovirus, bocavirus, human metapneumovirus, cytomegalovirus, herpes simplex viruses (1 and 2), Streptococcus pneumoniae, Mycoplasma pneumoniae, Chlamydia pneumoniae, Legionella pneumophila, Bordetella pertussis, Haemophilus influenzae, and Neisseria meningitidis.

The intra- and inter-assay variability was calculated using two concentrations of in vitro RNA at 1,010 and 1,660 copies for the E and RdRp genes, respectively, to mimic an intermediate viral load with a Ct of 30 and 10.10 and 16.60 copies for the E and RdRp genes, respectively, to mimic a low viral load with a Ct of 36. All dilutions were tested in triplicate on three independent runs.

Specimens tested

At the beginning of the outbreak, all samples tested by ProvLab were also sent to the NML for confirmation using the E-gene assay (5) before the final results were released. A total of 143 samples (136 negatives and 7 positives) tested by the gel-based SARS-CoV-2 assay and singleplex rtRT-PCR assays were sent to NML for parallel testing. Before implementation of the multiplex rtRT-PCR assay, a total of 108 samples (90 negatives and 18 positives) were tested by both the singleplex and the multiplex assays.

During initial validation of the singleplex rtRT-PCR assays, when positive SARS-CoV-2 patient samples were unavailable, a variety of extracted specimen types, including auger suction (n = 8), bronchoalveolar lavage (n = 9), endotracheal tube aspirate (n = 6), nasopharyngeal aspirate (n = 6), nasopharyngeal swab (n = 6), throat swab (n = 5), tracheal secretion (n = 1), and nasal swab (n = 1) were spiked with different concentrations of in vitro RNA. The spiked in vitro RNA concentrations of the RdRp target ranged from 1.01E+03 to 1.01E+05 copies/reaction; for the E gene, they ranged from 1.66E+02 to 1.66E+05 copies/reaction. In addition, five extracts for each of the specimen types (stool, bronchoalveolar lavages, endotracheal secretions, blood, and urine) were validated for the testing of SARS-CoV-2 by spiking with high, intermediate, and low viral loads of SARS-CoV-2 RNA provided by NML.

RESULTS

Assessment of the RT-PCR assay performance: analytical sensitivity, exclusivity, and reproducibility

The results for analytical sensitivity, dynamic range, and assay efficiency are summarized in Table 2. The 95% LOD for the RdRP and E gene targets was 6 and 2 copies/reaction, respectively, for the singleplex assays, and 15 and 4 copies/reaction, respectively, for the multiplex assay. The 95% LOD was confirmed to be the same in the presence of both high and low concentrations of MS2 for the multiplex assay. Preferential amplification of the target with suppression of MS2 amplification was noted in patient samples with high SARS-CoV-2 viral loads. The LOD by the gel-based assays using the pan-coronavirus primers and the SARS-CoV-2–specific primers was 1.01E+04 and 2 copies/reaction, respectively. Log-linear amplification of target was obtained over 7 logs of template concentration for the rtRT-PCR assays. Using these amplification plots, the efficiency of the rtRT-PCR assay was calculated for each of the viral targets, and as indicated in Table 2, these values ranged from 94.08% to 102.21%. All rtRT-PCR assays were able to reproducibly detect 1.20 pfu/mL of quantitated RNA from NML, and some were sensitive enough to detect 1.20E-02 pfu/mL, as shown in Table 3.

Table 2:

Assay characteristics

Assay	95% LOD (copies/reaction)	Dynamic range (copies/reaction)	Slope	Calculated efficiency (%)	R2 value
RdRp-Singleplex	6	1.01E+7 to 1.01E+1	–3.46	94.40	0.9989
E gene-Singleplex	2	1.66E+7 to 1.66E+1	–3.47	94.08	1.0000
RdRp-Multiplex	15	1.01E+7 to 1.01E+1	–3.34	99.20	0.9974
E gene-Multiplex	4	1.66E+7 to 1.66E+1	–3.27	102.21	0.9986
RdRp-SARS-CoV-2	2	1.01E+3 to 1.01E+0	N/A	N/A	N/A
RdRp-pan corona	1.01E+4	1.01E+5 to 1.01E+4	N/A	N/A	N/A

Notes: The singleplex and multiplex assays are real-time assays; the RdRp-SARS-CoV-2 and pan-corona assays are gel based. The 95% LOD for the real-time assays is reported as copies detected per reaction; extraction input and output volumes can be used to calculate the sensitivity per millilitre of patient sample. Linear regression plots of the copy number and Ct values were used to calculate PCR efficiency. Only two dilutions at endpoint were tested for the RdRp-pan corona assay.

LOD = Limits of detection; PCR = Polymerase chain reaction; N/A = Not applicable

Table 3:

Comparison of detection of viral RNA by the different assays

pfu/mL	No. of positive runs/total no. runs
	RdRp_Singleplex	Egene_Singleplex	RdRp_Multiplex	Egene_Multiplex	RdRp-SARS-CoV-2	RdRp-pan corona
1.20E+03	1/1 (25.23)	1/1 (21.12)	1/1 (24.42)	1/1 (20.62)	1/1	1/1
1.20E+02	1/1 (29.15)	1/1 (24.81)	1/1 (28.03)	1/1 (24.27)	1/1	0/1
1.20E+01	1/1 (32.72)	1/1 (28.18)	1/1 (31.01)	1/1 (27.63)	1/1	0/1
1.20E+00	3/3 (35.98)	3/3 (31.78)	3/3 (34.08)	3/3 (31.03)	1/1	0/1
1.20E-01	2/3 (42.04)	2/3 (34.91)	0/3	3/3 (34.27)	1/1	0/1
1.20E-02	0/3	1/3 (35.7)	0/3	2/3 (38.7)	0/1	0/1
1.20E-03	0/3	0/3	0/3	0/3	0/1	0/1

Notes: Average cycle threshhold of the replicates is shown in parentheses. The singleplex and multiplex assays are real-time assays, and the RdRp-SARS-CoV-2 and pan-corona assays are gel based. We used 5 μL of template for the singleplex and gel-based assays and 10µL for the multiplex assay.

The E gene and modified E gene assays detected SARS-CoV-1 as expected but did not amplify any of the other pathogens tested in the specificity panel. The RdRp rtRT-PCR assay did not amplify any of the other pathogens tested, including SARS-CoV-1, thus showing a very high specificity. The gel-based assay designed for the detection of SARS-CoV-2 showed cross-reaction with SARS-CoV-1.

The intra-assay variability (%CV) was calculated using the replicates within the same run and varied from 0.26% to 0.79% for the singleplex assays and 0.19% to 3.19% for the multiplex assay. The inter-assay variability was calculated using values obtained from different runs and ranged from 0.50% to 1.11% for the singleplex assays and 1.01% to 2.74% for the multiplex assay, showing reproducible detection and good precision at different viral loads (Table 4).

Table 4:

Assay variability at different viral loads

Assay	Copies/reaction	Intra-assay variability			Inter-assay variability
		Ct range	SD	%CV	SD	%CV
RdRp_Singleplex	10.1 to 1010	25.69–33.05	0.08–0.18	0.32–0.61	0.16	0.50–0.62
Egene_Singleplex	16.6 to 1660	23.43–31.35	0.06–0.24	0.26–0.79	0.25–0.34	1.07–1.11
RdRp_Multiplex	10.1 to 1010	28.76–37.59	0.07–1.14	0.19–3.19	0.62–1.00	2.31–2.74
Egene_Multiplex	16.6 to 1660	26.69–34.68	0.06–0.88	0.21–2.63	0.27–0.62	1.01–1.87

Ct = Cycle threshhold; %CV = Coefficient of variation

Testing of clinical samples

The gel-based assays were implemented for patient testing on January 21, 2020, followed by the singleplex rtRT-PCR assays on February 14, 2020. Testing was further streamlined with the multiplex assay implemented on March 16, 2020. Of the 143 samples tested with the gel-based RdRp assay, the singleplex rtRT-PCR assays, and NML (which included 136 negative and 7 positive samples), 142 samples gave concordant results by all assays. One in-house positive sample gave a negative result by the NML E gene assay; the in-house results for this sample were E gene rtRT-PCR 2/3 positive with an average Ct of 37.06, RdRp gene rtRT-PCR negative (with curves below threshold), and RdRp gel-based assay positive. Of the 108 samples tested by the singleplex and multiplex assays (90 negatives and 18 positives), all samples gave concordant results, with Ct values from the positive samples ranging from 14.51 to 39.42.

All clinical samples spiked with different concentrations of RdRp and E gene in vitro RNA or viral RNA provided by NML tested positive with expected Ct values. The measured Ct values from five extracts for each specimen type (stool, bronchoalveolar lavage, endotracheal secretions, blood, and urine) spiked at high, intermediate, and low viral loads of SARS-CoV-2 were comparable to the predicted values, thus establishing these specimen types as suitable for testing.

DISCUSSION

For the detection of SARS-CoV-2, the ProvLab initially developed and implemented two gel-based assays targeting the RdRp gene. This was followed by rtRT-PCR assays targeting the E and RdRp genes to increase the testing throughput and improve the turnaround time. The assay design was further streamlined by multiplexing the targets and adding the detection of an internal extraction and inhibition control.

The analytical sensitivity of the commonly used primer–probe sets for the detection of SARS-CoV-2 worldwide have been shown to be comparable (11, 15). The most sensitive primer–probe sets are E-Sarbeco, HKU-ORF1, HKU-N, CCDC-N, US CDC-N1, and US CDC-N3, with LODs of 500 copies/reaction and partial detection of 5 and 50 virus copies per reaction in 25% and 25%–50% of the replicates tested, respectively. The RdRp-SARSr primer–probe set was the least sensitive, being negative at 500 copies/reaction. Another study comparing the E-Sarbeco, RdRp-SARSr, and US CDC-N1/N2 primer–probe sets reports that the US CDC-N2 and E-Sarbeco primer–probe sets are particularly sensitive, with all replicates testing positive at 31.5 viral copies, whereas the RdRp-SARSr and US CDC-N1 are able to detect all replicates at 630 and 63 viral copies, respectively. The sensitivity of the SARS-CoV-2–specific gel-based assay used at ProvLab was two copies/reaction, and the 95% LOD for the assays targeting the E and RdRp genes was less than 15 copies/reaction, with the E gene being slightly more sensitive than the RdRp gene; thus, all the assays, including the RdRp gel-based assay used at the ProvLab, had comparable sensitivity to the assays commonly used worldwide. Comparison of accuracy with the E-Sarbeco assay (5) performed at NML showed 99.3% concordance with the assays performed at ProvLab.

The efficiencies of the different commonly used assays described here were shown to be greater than 90%; the analytical efficiency of the ProvLab assays varied from 94.40% to 102.21%. All rtRT-PCR assays showed 100% analytical specificity because they did not cross-react with any viral or bacterial pathogens that would be included in the differential for COVID-19 (the exception being SARS-CoV-1 for the E gene target, which was expected on the basis of in silico analysis and is not a concern given that this virus is not currently circulating in humans).

The availability of rtRT-PCR assays early in the pandemic was essential to provide valuable time to local health authorities to contain transmission and prepare for appropriate response strategies.

Ethics Approval:

N/A

Informed Consent:

N/A

Registration and Registration No. of the Study/Trial:

N/A

Funding:

No funding was received for this work.

Disclosures:

The authors have nothing to declare.

Peer Review:

This manuscript has been peer reviewed.

Animal Studies:

N/A