Isothermal amplification and fluorescent detection of SARS-CoV-2 and SARS-CoV-2 variant virus in nasopharyngeal swabs

The COVID-19 pandemic caused by the SARS-CoV-2 is a serious health threat causing worldwide morbidity and mortality. Real-time reverse transcription PCR (RT-qPCR) is currently the standard for SARS-CoV-2 detection. Although various nucleic acid-based assays have been developed to aid the detection of SARS-CoV-2 from COVID-19 patient samples, the objective of this study was to develop a diagnostic test that can be completed in 30 minutes without having to isolate RNA from the samples. Here, we present an RNA amplification detection method performed using reverse transcription loop-mediated isothermal amplification (RT-LAMP) reactions to achieve specific, rapid (30 min), and sensitive (<100 copies) fluorescent detection in real-time of SARS-CoV-2 directly from patient nasopharyngeal swab (NP) samples. When compared to RT-qPCR, positive NP swab samples assayed by fluorescent RT-LAMP had 98% (n = 41/42) concordance and negative NP swab samples assayed by fluorescent RT-LAMP had 87% (n = 59/68) concordance for the same samples. Importantly, the fluorescent RT-LAMP results were obtained without purification of RNA from the NP swab samples in contrast to RT-qPCR. We also show that the fluorescent RT-LAMP assay can specifically detect live virus directly from cultures of both SARS-CoV-2 wild type (WA1/2020), and a SARS-CoV-2 B.1.1.7 (alpha) variant strain with equal sensitivity to RT-qPCR. RT-LAMP has several advantages over RT-qPCR including isothermal amplification, speed (<30 min), reduced costs, and similar sensitivity and specificity.


Introduction
COVID-19 is a pandemic disease caused by SARS-CoV-2 [1,2]. The animal reservoir for SARS-CoV-2 is unknown, but it is believed to be of zoonotic origin as SARS-CoV-2 isolates from bats and pangolins are genetically related to human isolates [3,4]. SARS-CoV-2 has initiated a pandemic, and according to the WHO Coronavirus (COVID- 19 used as an endpoint for RT-LAMP there is also improved detection arguing fluorescent RT-LAMP superiority [28]. Further, fluorescent RT-LAMP can be deployed as a point-of-care (POC) test filling a role that is vital for disease management and control.

Optimizing the limit of detection/fluorescent end-point cutoff
The fluorescent LAMP reaction contained an intercalating fluorescent SYBR-like dye with excitation (λ Exci = 497 nm) and emission (λ Emis = 520 nm) when complexed with the doublestranded DNA LAMP reaction product. This allowed for easy monitoring of the accumulation of double-stranded DNA in the RT-LAMP reaction with a real-time thermocycler using a standard SYBR/FAM channel for fluorescence. Using a log dilution series (ranging from 10 8 to 10 copies of the N gene) of purified RNA templates described below (N gene IVT RNA and viral genomic RNA), we measured the signal from the fluorescent RT-LAMP reaction at 1 min intervals analogous to a standard dye-based RT-qPCR reaction utilizing a 30 min total time cut-off. The results were compared to the cycle threshold (Ct) values obtained from our RT-qPCR assay (described below) for the same purified RNA templates. In the fluorescent RT-LAMP reaction, the purified RNA templates representing~100 copies or more of the N gene generated strong fluorescence by 30 min and were readily detectable by RT-qPCR. RNA templates diluted beyond 10 copies of the N gene did not generate fluorescence signal within 30 min in fluorescent RT-LAMP and were negative in RT-qPCR with a cut-off of 40 cycles. Based on this observation, and consistent with other related RT-LAMP publications [29,30], we used fluorescence at 30 min to determine whether a sample was positive or negative.

RT-qPCR
Serial log dilutions of a 2019-nCoV_N_positive control plasmid (IDT) containing the SARS-CoV-2 (Wuhan-Hu-1 strain, GenBank: MN908947) N gene from 10 5 copies/μl to 1 copy/μl, or SARS-CoV-2 genomic viral RNA (vRNA), or SARS-CoV-2 N gene in vitro transcribed RNA (IVT RNA) from 1ng/μl to 0.1fg/μl corresponding to 10 7 copies/μl to 1 copies/μl, or 10 8 copies/μl to 10 copies/μl, respectively, were prepared in nuclease-free water. Nuclease-free water was used as a no-template control for all reactions. The SARS-CoV-2 RNA templates were added to LunaScript RT SuperMix (NEB) at 2 μl in a 20 μl reaction and reverse transcribed for 10 min at 55˚C. The cDNA from this reaction was used directly at 1 μl in a 20 μl reaction with Luna Universal Probe qPCR Master Mix (NEB) and 1 μl of the SARS-CoV-2 CDC EUA N1 probe assay (IDT) and amplified for 40 cycles in a two-step reaction (15s/95˚C, and 30s/60˚C) after an initial 1 min pre-amplification hold at 95˚C on an AriaMx Real-Time PCR System (Agilent). SARS-CoV-2 plasmid DNA was RT-qPCR amplified by adding 1 μl to a 20 μl reaction with the same qPCR master mix and probe assay using the same two-step amplification protocol after an initial 3 min pre-amplification hold at 95˚C. Results from the N gene control plasmid were used to construct a standard curve relating the N gene copy number to RT-qPCR Ct values (S3 Fig). The N gene copy numbers of the SARS-CoV-2 IVT RNA or SARS-CoV-2 genomic vRNA per unit mass were extrapolated from this curve.
For analysis of clinical samples, RNA was extracted from 200 μl of patient nasopharyngeal swab liquid in saline using the RNAdvance Viral Reagent Kit (Beckman Coulter, Indianapolis, IN) and eluted in a final volume of 40 μl of nuclease-free water. RT-qPCR analysis of purified RNA from patient samples as described above was performed using the CDC EUA N1 probe assay and the CDC EUA human RNAseP internal control probe assay to validate sample RNA extraction in separate reactions.

In vitro transcribed (IVT) RNA
The 1260 bp complete coding sequence of SARS-CoV-2 Wuhan-Hu-1 strain N gene (Gen-Bank: MN908947) was amplified from 2019-nCoV_N_Positive Control plasmid (IDT) using primers (IDT) SARS2N_F and SARS2N_R (S1 Table) in standard end-point PCR with Q5 DNA polymerase (NEB). The PCR band representing the N gene coding sequence was subcloned into plasmid pMiniT2.0 (NEB) with the 5' end of the coding sequence proximal to the vector-provided T7 polymerase promoter and confirmed by Sanger sequencing. HiScribe T7 RNA Synthesis Kit (NEB) was used to produce the SARS-CoV-2 N gene single-stranded positive-sense RNA via in-vitro transcription from the pMiniT2.0 construct. The IVT RNA was digested with DNase1 (NEB) and purified using RNAClean XP (Beckman Coulter). Purified IVT RNA was eluted in nuclease-free water, adjusted to a final concentration of 1 μg/μl in nuclease-free water, and stored at -80˚C.

Analysis of NP swab samples
Deidentified samples were collected from consenting adult volunteers by the Georgia National Guard or the University of Georgia (UGA) COVID-19 Task Force following written approval and were tested from frozen -80˚C samples collected by the Georgia Taskforce for COVID-19, the Georgia National Guard, or the University of Georgia Diagnostic Laboratories. NP samples were collected using a sterile swab applicator that was placed in 1 mL of saline. The University of Georgia (UGA) Diagnostic Laboratories determined the SARS-CoV-2 status of the NP samples using an Applied Biosystems TaqPath COVID-19 EUA assay (ThermoFisher) to detect SARS-CoV-2 RNA in a multiplex RT-qPCR format. The assay is specific for three SARS-CoV-2 genome regions, i.e. spike (S) gene, nucleocapsid (N) gene, and orf-1ab. We analyzed the same NP samples using the CDC EUA N1 RT-qPCR assay for the detection of SARS-CoV-2 RNA. The CDC EUA N1 RT-qPCR assay is singleplex and specific for the SARS-CoV-2 N gene. Fifty SARS-CoV-2 positive samples and 84 SARS-CoV-2 negative samples were tested in our laboratory using the CDC EUA N1 RT-qPCR assay to detect the N gene sequence. All NP samples were subsequently tested in a fluorescent RT-LAMP assay for comparative SARS-CoV-2 diagnostic analysis with CDC EUA N1 RT-qPCR assay.

Analytical sensitivity of fluorescent RT-LAMP using IVT RNA and genomic viral RNA
We examined the sensitivity of the fluorescent RT-LAMP assay, by comparison, to the CDC EUA N1 RT-qPCR assay. The limit of detection (LOD) in the RT-qPCR assay using SARS-CoV-2 (Wuhan-Hu-1) IVT RNA was 0.1 fg of RNA corresponding to 10 copies of the N gene sequence (Fig 1). The LOD by RT-qPCR was 100 copies of purified viral genomic RNA from SARS-CoV-2 USA-WA1/2020 (S1A Fig). The fluorescent RT-LAMP assay rapidly and sensitively detected SARS-CoV-2 IVT RNA with a LOD of approximately 10-100 N gene copies with efficient isothermal amplification across seven logs of target concentration (Fig 2). The fluorescent RT-LAMP assay can detect 100 copies of SARS-CoV-2 (USA-WA1/2020 GenBank: MN985325.1) viral genomic RNA equaling the sensitivity of RT-qPCR (S1B Fig). We demonstrated that the fluorescent RT-LAMP assay successfully detected the SARS-CoV-2 B1.1.7 (alpha) variant with equal sensitivity to that of SARS-CoV-2 at a LOD of approximately 100 pfu/ml (Fig 3A and 3B). To confirm RT-LAMP specificity, we tested human coronavirus strains, OC43 (GenBank: AY585228.1), 229E (GenBank: AF304460.1), and NL63 (GenBank: AY567487.2) infected (MOI = 1.0) Vero cells. All human coronavirus strains tested did not produce a detectable fluorescent signal in the fluorescent RT-LAMP assay (Fig 3C). The sensitivity of fluorescent RT-LAMP detection of both the variant and the wild-type SARS-CoV-2 strains compared favorably with the RT-qPCR results for RNA purified from an equivalent amount of SARS-CoV-2 infected Vero cell culture supernatant (Fig 4).

Fluorescent RT-LAMP detects SARS-CoV-2 in nasopharyngeal swab samples
We determined the presence of SARS-CoV-2 directly in clinical nasopharyngeal swab samples collected from patients using fluorescent RT-LAMP. Nasal swabs and NP swabs from suspected COVID-19 patients were collected by personnel from the Georgia COVID-19 Task Force and the University of Georgia Diagnostic Laboratories and confirmed positive or negative for SARS-CoV-2 using an Applied Biosystems TaqPath COVID-19 EUA kit to detect SARS-CoV-2 RNA in a multiplex RT-qPCR format. In comparing fluorescent RT-LAMP to RT-qPCR, we chose to evaluate the singleplex N gene-specific SARS-CoV-2 CDC EUA N1 RT-qPCR assay since it shares close target specificity with our fluorescent RT-LAMP assay (S1 Table). Using the same NP samples, we used the CDC EUA N1 RT-qPCR assay on both SARS-CoV-2 positive (n = 50) and SARS-CoV-2 negative (n = 84) NP samples in triplicate. Positive samples in the CDC EUA N1 RT-qPCR probe assay with Ct values <40 for each of the three repeat samples were considered N gene positives in our comparative analysis. We determined that 84% (n = 42/50) of the previously established SARS-CoV-2 positive NP samples were N gene-positive (Ct<40) by the CDC EUA N1 RT-qPCR assay for each of three repeat measurements ( Table 1). The mean Ct of the SARS-CoV-2 N gene-positive NP samples (n = 42) ranged from 16 to 34 cycles with an overall mean Ct of 25.7, 95% CI [24][25][26][27] using the SARS-CoV-2 CDC EUA N1 RT-qPCR assay. Negative samples with no Ct value in the CDC EUA N1 RT-qPCR assay for any of three repeat measurements were considered true N gene negatives. For the previously confirmed SARS-CoV-2 negative NP samples, we determined that 81% (n = 68/84) gave matching N gene negative results, i.e. no Ct value in any of the three repeat measurements using the SARS-CoV-2 CDC EUA N1 RT-qPCR assay ( Table 2).
The fluorescent RT-LAMP assays were performed in triplicate using the same NP samples that were previously determined to be positive for SARS-CoV-2 (n = 50) by the UGA Diagnostic Laboratories. The fluorescent RT-LAMP assay values for the SARS-CoV-2 N gene-positive NP samples ranged from 18 to 28 minutes with an overall mean of 23 minutes and 95% CI [22][23][24]. The results for N gene-positive NP samples showed 98% (n = 41/42) concordance with RT-qPCR for SARS-CoV-2 positivity (Table 1)   A fluorescent RT-LAMP assay was performed on the same NP samples that were previously shown to be SARS-CoV-2 negative (n = 84) by the UGA Diagnostic Laboratories. Results for the fluorescent RT-LAMP assay on the N gene negative NP samples showed 87% (n = 59/68) concordance with RT-qPCR for SARS-CoV-2 (Table 2), while 13% (n = 9/68) of N gene negative samples gave discordant fluorescent RT-LAMP assay results as determined by a positive result in the 30 minute assay time. When compared to RT-qPCR, N gene-positive samples assayed by fluorescent RT-LAMP had 98% (n = 41/42) concordance, while fluorescent RT-LAMP values were 87% (n = 59/68) concordant with RT-qPCR. Importantly, the Table 1

Mean RT-LAMP (min) [95% CI]
Below assay cut-off (  fluorescent RT-LAMP assay results were obtained without purification of RNA from the NP samples in contrast to the RT-qPCR assay protocol.

Discussion
The benchmark method for the detection of SARS-CoV-2 from patient samples is the RT-qPCR assay requiring RNA extraction, highly trained laboratory personnel, and expensive equipment and reagents. Currently, the majority of clinical diagnosis for SARS-CoV-2 are performed by central testing laboratories that typically takes >24h for return of results. These clinical RT-qPCR assays are used typically with symptomatic patients, are not required to be low-cost, but rely on their analytical sensitivity to produce a definitive diagnosis even if only a single testing opportunity is available. Diagnostic assays used in surveillance are designed to detect the prevalence of SARS-CoV-2. These tests should be cost-effective, rapid, and easy to perform allowing for frequent testing. We sought to improve upon SARS-CoV-2 diagnostic testing by reducing the time for confirmation of a result, lowering costs to acquire the results, and developing a reliable, sensitive, and specific rapid assay. Many emerging studies have shown that RT-LAMP can meet these requirements for screening and testing for SARS-CoV-2 and potentially be a complementary tool to RT-qPCR [18,31,32]. Several RT-LAMP studies have focused on developing point-of-care (POC) diagnostic assays by detection of SARS-CoV-2 RNA in clinical samples [33,34]. Other POC RT-LAMP studies have developed assays to allow multiplexing of more than one genomic target, while others have focused on optimizing RT-LAMP conditions to improve detection of SARS-CoV-2 genomes [35], but direct detection of the virus has been limited.
In this study, we show that fluorescent RT-LAMP has similar performance to RT-qPCR in detecting SARS-CoV-2 from the same clinical samples, and importantly, that fluorescent RT-LAMP can sensitively and reliably detect virus from only minimally pre-processed nasal swabs. Despite differences in SARS-CoV-2 detection in NP samples between the multiplex assay and the RT-LAMP assay, we believe that comparative sensitivity and specificity is important for accelerating diagnostic applications. Likely, the broad target specificity associated with the multiplex RT-qPCR assay used by the UGA Diagnostic Laboratories may in part account for the difference in sensitivity we obtained on the same NP samples using the singleplex CDC EUA N1 RT-qPCR assay. It is also possible that RNA degradation of the NP samples may have occurred by the time we obtained them as they were stored in saline and subsequently underwent several freeze-thaw cycles which may have contributed to differences in sensitivity associated with our RT-qPCR assay.
Of the group of SARS-CoV-2 positive NP samples from the UGA Diagnostic Laboratories, we found 84% (n = 42/50) to be N gene-positive as determined by our RT-qPCR assay. Importantly, fluorescent RT-LAMP assay results were 98% (n = 41/42) concordant as N gene-positive samples. Eight SARS-CoV-2 positive NP samples determined by the UGA Diagnostic Laboratories failed to amplify the N gene target in our RT-qPCR assay. However, the internal control amplicon (human RNase P) amplified in all NP samples tested s that the extracted RNA was intact. We found n = 7/8 positive NP samples by the fluorescent RT-LAMP assay for the failed RT-qPCR NP samples which was discordant with our RT-qPCR result for these NP samples but in agreement with NP samples showing SARS-CoV-2 positive standing by the UGA Diagnostic Laboratories. This may suggest that sample preparation differences (e.g. RNA purification for RT-qPCR or the minimal pre-processing for RT-LAMP) may contribute to the different results. We show that the fluorescent RT-LAMP assay sensitivity is more than sufficient to reliably detect SARS-CoV-2 in NP samples to virus load, as SARS-CoV-2 patients have been reported to have a range from 641 copies/mL to 10 11 copies/mL with a median of 10 4 -10 5 [36]. Of the SARS-CoV-2 negative NP samples from the UGA Diagnostic Laboratories, 81% (n = 68/84) were N gene negatives as determined by our RT-qPCR assay. Fluorescent RT-LAMP assay results were 87% (n = 59/68) concordant confirming N gene negative samples, and 16 NP negative samples were positive in our RT-qPCR assay. These results likely represent RT-qPCR false-positives that were SARS-COV-2 negative as determined by the UGA Diagnostic Laboratories with differences likely owing to the broad-specificity of the RT-qPCR assay. The N gene negative NP specimens by RT-LAMP were 13% (n = 9/68) with positive results discordant with RT-qPCR results.
We show that the fluorescent RT-LAMP assay sensitively and specifically detects wild-type SARS-CoV-2 and the B.1.1.7 (Alpha) variant in virus-infected Vero cell culture supernatants after the minimal pre-processing steps. The fluorescent RT-LAMP assay primers were designed based on the nucleotide sequence of the N gene reported for the SARS-CoV-2 (WA1/2020) strain. A majority of the sequence divergence associated with variant strains is located in the S gene, while the N gene nucleotide sequence is highly conserved. The N gene RT-LAMP primer sequences are 100% conserved among the wild type and variant SARS-CoV-2 strains (S2 Fig). This conservation allows our RT-LAMP assay to detect various SARS-CoV-2 strains. Also, the specificity of our fluorescent RT-LAMP assay is shown by the failure of three closely related human coronavirus strains, i.e. OC43, 229E, and NL63, to give any positive signal over the baseline in this 30 minute assay (S2 Fig).
Overall, the fluorescent RT-LAMP assay was shown to be positive for >98% of NP samples (n = 41/42) that were N gene-positive samples by RT-qPCR, and negative for 87% (n = 59/68) of NP samples that were N gene negative samples by RT-qPCR. Discordant results in RT-qPCR and fluorescent RT-LAMP could represent false positives from spurious amplification from primer-dimer sometimes associated with RT-LAMP assays, sample contamination, variability among different RT-qPCR tests detecting different or multiple SARS-CoV-2 amplicons, or increased sensitivity of RT-LAMP compared to RT-qPCR. RT-LAMP employing six primers has high-specificity for selected amplification targets when compared to standard PCR utilizing only two primers, or RT-qPCR employing an additionally labeled probe oligo. Careful design of the six LAMP primers, particularly the inner primers FIP and BIP to avoid self-complementarity, and optimized formulation can help to mitigate these concerns. Our data shows that the fluorescent RT-LAMP assay is comparable to RT-qPCR and useful for detecting SARS-CoV-2 infected NP samples having a moderate viral load. Importantly, the viral load does not indicate either the course of infection or COVID-19 disease severity because the viral load is not representative of the overall viral burden of an infected individual, and the level of SARS-CoV-2 is not the only attribute affecting disease in an individual [37][38][39].