Multiplex PCR Assays for Identifying All Major SARS-CoV-2 Variants

Synthetic RNA targets

RNA standards of the SARS-CoV-2 whole genome and variant spike proteins were provided by Exact Diagnostics through Bio-Rad Laboratories, Hercules, CA (catalog numbers COV019, COV019CE, COVA, COVB, COVE, COVG). Additional spike protein RNA sequences, such as those containing the E484Q and T478K mutations and the omicron spike RNA, were generated from a double-stranded synthetic DNA (gBlock) from Integrated DNA Technologies (Coralville, IA). A 5’-T7 promoter was appended to it, which enabled transcription by T7 RNA polymerase, which was performed using the HiScribe™ T7 High Yield RNA Synthesis Kit (New England Biolabs, Ipswich, MA). The RNA produced was purified with a Monarch RNA Cleanup Kit, 50 µg, (New England Biolabs) and by denaturing polyacrylamide gel electrophoresis, eluting the band of correct size in gel elution buffer (400 mM NaCl, 20 mM Tris-HCl (pH 8.0), and 1 mM EDTA). The concentration of the eluted RNAs was determined by a NanoDrop spectrophotometer (Thermo Fisher Scientific).

Primer Design

Primers were designed by using the primer3 server¹³. 200 nucleotides surrounding the mutation of interest on either side of the SARS-CoV-2 genome was used as input, the target-binding region of the molecular beacon was used as the hybridization probe, and the amplicon size was set between 50 to 300 nucleotides. The sequences of all the primers that were used in the assay are shown in Table 1.

Design and characterization of single nucleotide discriminating molecular beacons

A key consideration in the design of single-nucleotide discriminating molecular beacons is that the melting temperature (T_m) of the probe-target hybrid with the perfect target (in the present case, the mutant sequence) is minimally above, only 3 to 7 °C, the annealing temperature of the PCR (which is the temperature at which fluorescence will be monitored). This makes it so that the mismatched hybrid (in the present case, the wild-type sequence) will not appreciably bind to the molecular beacon at the fluorescence monitoring temperature. Furthermore, the melting temperature of the stem of the molecular beacon should be above the detection temperature.

We designed our primers so that they will function at the PCR annealing temperature of 58 °C. We selected a region of sequence around the site of each mutation such that the T_m of the perfect probe-target hybrid was above the annealing temperature, but that the mismatched single-nucleotide polymorphism had a T_m lower than the annealing temperature. These T ‘s were predicted by using the DINAMelt¹⁴ two-state melting hybridization server with the following parameters: 0.25 µM oligonucleotide concentration, [Na⁺] = 60 mM, [Mg²⁺] = 3 mM, for DNA at 58 °C.

After deciding upon the target binding region, a hairpin stem was created by appending 5 or 6 nucleotides on either side (with 4 to 5 G:C pairs) such that they are complementary to each other, but not complementary to the target. In addition, we ensured that the 5’ nucleotide is not a guanosine, because that could lead to fluorescence quenching due to the 5’ position of the fluorophore. These sequences were then ordered from either LGC Biosearch Technologies (Petaluma, CA) or from Eurofins (Luxembourg City, Luxembourg). They were purified by HPLC through a PRP-3 Reversed Phase HPLC Column (Hamilton Company, Reno, NV) in a Beckman System Gold HPLC System (Fullerton, CA). The desired fractions were then ethanol precipitated and resuspended in water, and their concentration was determined by NanoDrop spectrophotometer. The sequences of all molecular beacons that were used in the assay are shown in Table 1.

To determine if the designed sequences could sufficiently discriminate the alleles at the annealing temperature of the PCR assay, we measured the fluorescence of the molecular beacons in the presence of either a mutant target, a wild-type target or, no target, under the buffer conditions that are expected to be present during PCR. Each 25 µL thermal denaturation reaction contained 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl_2, 1 µM oligonucleotide target, and 0.4 µM molecular beacon. The fluorescence intensity of each reaction was measured at every degree in the appropriate fluorescence channel as the temperature was lowered from 95 °C to 30 °C (decreasing 1 °C/10 seconds) in a Bio-Rad CFX96 Touch Real-Time PCR Detection System.

Melt curve signals were normalized, for each molecular beacon. The difference between a normalization point and the other melt curve fluorescent signals was subtracted from the raw fluorescent intensity values and then all values in that channel were divided by the maximum intensity for that molecular beacon, to derive relative fluorescence units, which range between 0 and 1. These data are shown in Figure S1.

Optimizing the concentrations of the qRT-PCR multiplex primers and molecular beacons

qRT-PCR was performed with TaqMan Fast Virus 1-Step Master Mix (No ROX) to optimize conditions, and TaqPath™ 1-Step Multiplex Master Mix (No ROX) (both from Applied Biosystems, Waltham, MA) for final experiments and patient samples.

Asymmetric PCR was performed for each amplicon, where the primer generating the strand complementary to the molecular beacon (labeled reverse) was in 5-fold excess over the other primer (labeled forward). All molecular beacons were designed in the same forward strand direction. G:T mismatches can be discriminated from each other by switching the strand being interrogated to the complementary strand¹⁵. The molecular beacon and primer concentrations were adjusted by trial and error, for sufficient amplification and signal. The final working concentrations of each of the components for each tube are listed in Table 1 along with their sequences.

Each reaction had a final volume of 20 µL comprised of TaqPath or TaqMan Master 1-Step Mix (No ROX), water, and primers and probes as previously described. 15 µL of this complete master mix was pipetted into each PCR tube, and then 5 µL of sample was added and mixed, before running on the thermal cycler.

Assay sensitivity and specificity

The sensitivity of the assay was determined by preparing dilution series of in vitro transcribed targets, which were diluted 10-fold from 1,000 copies/µL to 0.1 copies/µL. 5 µL of each of these dilutions were combined with 15 µL of completed master mix per reaction. The tubes were sealed, briefly shaken, spun down at 800 g for 1 min. and then run on a Bio-Rad CFX96 Touch Real-Time PCR Detection System with the TaqPath or TaqMan Fast specified thermal cycler conditions, and with an annealing temperature of 58 °C. Specificity of the assay was determined by comparing the known sequences of the in vitro transcribed targets with the signals in these dilution series. For patient samples, sensitivity and specificity were determined by comparing this assay’s results to sequencing results, which are the gold standard.

Collection of patient samples and sequencing

Patient mid-nasal swabs were collected for clinical testing in 1 ml of DNA/RNA Shield Reagent (Zymo, Irvine, CA). 400 µL aliquots were used for RNA extraction using the Kingfisher Flex magnetic bead extraction system (Thermo Fisher Scientific) and the Zymo Quick RNA/DNA Magbead kit (Zymo) per manufacturer’s instructions. qRT-PCR screening was performed using 2 µL of purified RNA, TaqPath 1-step Multiplex Master Mix (No ROX) and an in-house validated multiplex screening assay. This screening assay, outlined in Figure S2, detects conserved sequences in the N and RdRp genes of SARS-CoV-2, along with human ACTB (β-actin) mRNA, with high sensitivity. The primers and molecular beacons targeting regions in the RdRp, N, and human ACTB genes were modified and adapted from previously validated assays¹⁶ (https://www.cdc.gov/coronavirus/2019-ncov/lab/rt-pcr-panel-primer-probes.html, date of last access: 12/7/2021), and are listed in Table 2. Samples were run on a Quantstudio 7 (Thermo Fisher Scientific). Amplicon sequencing was performed using the Ion AmpliSeq SARS-CoV-2 Insight Research Panel (Ion Torrent, Guilford, CT). Each library contained six samples, one positive control sample and one no template control. Barcoded libraries were prepared using the Ampliseq Kit for Chef DL8 and Ion 530 Kit-Chef (Ion Torrent). Sixteen libraries were combined to run on a single Ion 530 chip. Templates were prepared using the Ion 530 Chef Reagents template kit. Sequencing was performed on the Ion S5XL instrument with Ion S5 Sequencing Solutions and Reagents (Ion Torrent). Sequence data was analyzed for on-target and mean depth of coverage using the SARS_COV_2_Coverage Analysis plug-in v1.3.0.2 (Ion Torrent). FASTA files were created using the IRMA report plug-in v1.3.0.2 (Ion Torrent).

Sequence data and corresponding metadata were uploaded to GISAID and are included as supplementary data. Aliquots of purified RNA samples no longer needed for diagnosis were deidentified and transferred to Rutgers’s laboratory for initial testing of the multiplex variant molecular beacon assay.

RNA Purification from inactivated SARS-CoV-2 Virus

10,000 copies of each inactivated Variant of Concern SARS-CoV-2 virus from ZeptoMetrix (Buffalo, NY) (catalog# NATSARS(COV2)-VP) was treated with proteinase K in a 210 μL reaction (1% SDS, 25 mM EDTA, 10 mM Tris pH8.3 and 20 μg proteinase K (Thermo Fisher, catalog #25530049)) at 45°C for 30min, to decrosslink the protein matrix entrapping the viral RNA. After the addition of 1-2µg of yeast carrier tRNA (Roche, Basel, Switzerland, catalog #10109495001), purification of the RNA was performed using the QIAgen RNeasy Mini kit (catalog #74104).

Multiplex qRT-PCR with samples

RNA purified patient and inactivated virus samples were tested in a similar manner to how we tested the assay’s sensitivity and specificity. 5 µL of each sample was combined with 15 µL each completed master mix using TaqPath and run as mentioned before.

qRT-PCR analysis and plotting

Data generated from the CFX96 Touch Real-Time PCR Detection System was analyzed with Bio-Rad’s CFX Maestro Software Version 2.0, which corrects for fluorescence drift, subtracts the baseline, and fits curves to each qRT-PCR signal. Threshold cycles (C_t) were determined via a manually specified single threshold of our analyzed data, which was clearly above all negative control signals. For the specificity plots, fluorescent signals were normalized by dividing all values by the highest fluorescent value in each channel for the respective tubes.

Design of a two-tube, eight mutation, multiplex qRT-PCR assay for SARS-CoV-2 Variants

In our assay, SARS-CoV-2 variants were identified by probing the coding region of the spike protein transcript for the presence of eight different point mutations, as detailed in Figure 1. Since only the mutants are expected to yield a positive signal, we included an additional set of primers and molecular beacons for a conserved region of the viral N gene in Tube one as a positive control for the presence of the virus. Since the multiplexing capacity of common thermal cyclers is 4 or 5 colors, we divided the assay into two tubes. The colors of the molecular beacons and the locations of the primers are indicated in Figure 1A. In Tube one, there were three amplicons, and in Tube two, all four targets were present in the same amplicon.

• Download figure
• Open in new tab

Figure 1.

A. A schematic representation of the primers and molecular beacons used in the assay. A small black x on the molecular beacons target binding regions indicates that the molecular beacon is single nucleotide discriminant. B. Identifying key mutations present in each variant. The mutations detected in the assay are shown in the rows and variants are shown in the columns. Each of the variants of concern can be identified by these combinations of mutations as well as the variants of interest, see supplemental chart. Abbreviations used: CFG = Cal Fluor Gold 540, CFR =Cal Fluor Red 610, Q670 = Quasar 670, Q705 = Quasar 705.

Even though individual isolates of each variant may possess other mutations, we focused on sets of mutations that are functionally the most important. By using in vitro binding and epidemiological data of how these mutations affect immune evasion and binding to the ACE2 receptor^2–5,17,18, we chose to identify the mutations shown in Figure 1: d69-70, K417N/T, L452R, T478K, E484K/Q, and N501Y. We also chose these mutations such that a single mutation in Tube one is sufficient to identify the VOCs, which of course is subject to change as the virus evolves. The mutations targeted in Tube two complete the identification of the VOCs and identify other common mutations, which are useful for identifying other variants.

The variant molecular beacon assay is specific for the identification of common SARS-CoV-2 variant mutations

Before assembling our multiplex assay, we designed molecular beacons for each targeted mutation, determined their thermal denaturation profiles, and tested them in monoplex PCRs. To determine whether the molecular beacons were allele discriminating at the annealing temperature, we determined the thermal denaturation profiles of the molecular beacons by themselves, and together with mutant or wild-type targets. In these experiments, the targets were synthetic oligonucleotides, and they were used in a molar excess above the molecular beacon concentration. A representative profile is shown in Figure 2, and the profiles of all the molecular beacons are shown in Supplementary Figure 2. Although, the window of discrimination varies somewhat from molecular beacon to molecular beacon, all of our molecular beacons permitted single nucleotide discrimination (see Figure 3A). After qualifying the molecular beacons in this manner, we performed a set of monoplex PCR assays for each mutation, using wild-type targets and mutant targets in alternate reactions.

• Download figure
• Open in new tab

Figure 2.

The florescence of the T478K discriminating molecular beacon, as a function of temperature, in the presence and absence of targets. When no target is present, dotted line, the fluorescence increases as temperature rises due to the helical order of the stem giving way to a random-coil configuration, separating the fluorophore from the quencher, and resulting in fluorescence. When the molecular beacon binds to its target its stem dissociates which turns on its fluorescence. Due to the higher binding strength of the perfect probe-target hybrid, solid line, compared to the imperfect probe-target hybrid, dashed line, the imperfect-probe target hybrid dissociates at a lower temperature. The PCR monitoring temperature (58°C) is indicated by the vertical dashed line and the window of good discrimination by the shaded area.

• Download figure
• Open in new tab

Figure 3.

Demonstration of specificity and sensitivity of the variant molecular beacon assay. A. Each panel shows the amplification curves of a single multiplex reaction using either the Tube one or Tube two specified primers and probes (columns), and the indicated in vitro transcribed RNA targets (rows). The ‘wildtype’ template is RNA of the original strain of SARS-CoV-2’s entire genome. The labeled strains contain the RNA sequences of the S gene from the respective SARS-CoV-2 variants and the sample labeled Kappa/Delta uses an RNA corresponding to (positions 1251-1630nt in the Spike protein coding sequence) and containing the E484Q and T478K mutations, which are from the B.1.617 (Kappa) and B.1.617.2 (Delta) variants. This Kappa/Delta Spike RNA fragment was generated and used due to the lack of commercially available in vitro-transcribed RNA’s at the time. B. Sensitivity analysis was run by diluting templates containing the specified mutations from 5,000 copies down to 0.5 copies per reaction. The C_t values are plotted against the template copy number and the correlation coefficients of the natural log of the copy number to the C_t are greater than 0.98 for all amplification products, with around a 3 C_t delay between 10-fold dilutions, both as expected.

When all the monoplex reactions indicated successful amplification and allele discrimination, we assembled two multiplex reactions that included all the primers and molecular beacons, as detailed in Figure 1. To demonstrate the specificity of the multiplex assays, six pairs of reactions were carried out, in which each reaction received 5,000 copies of either the wild type, B.1.1.7 (Alpha), B.1.351 (Beta), P.1 (Gamma), B.1.427 (Epsilon), or E484Q and T478K (Kappa/Delta)-containing spike RNA sequences. The targets were in vitro-transcribed RNAs corresponding to a portion of the spike protein, or commercially available RNA standards, as detailed in methods and in the legend to Figure 3. The results of these reactions are presented in Figure 3A, where the targets are indicated inside the panels, and the color response of each molecular beacon is indicated by the color of the line. This data shows that when the wild-type SARS-CoV-2 genome is present, only the N gene positive control signal occurs. When testing spike RNAs containing the variant mutations, the correct mutations were identified. For example, the Beta variant (B.1.351) contains not only the uniquely identifying K417N mutation, but also the N501Y and E484K mutation.

The variant molecular beacon assay can detect at as few as five copies of viral RNA

To demonstrate the sensitivity of the variant molecular beacon assay, PCR assays were initiated with serial dilutions of each target. The results are shown in Figure 3B. Each dilution series tested from 5,000 copies to 0.5 copies. Where a clear signal was present, the threshold cycle was determined by a manually identified single threshold for each channel. Reactions starting with as little as five copies yielded a detectable signal in each case. The correlation coefficients between the logarithm of the copy number and the threshold cycles in each case are each greater than 0.98, showing a very strong exponential relationship between the two. These results also indicate that in Tube one, which contains four pairs of primers, the amplification of multiple different amplicons does not interfere with each other, which would diminish sensitivity.

Inactivated virus and patient sample strains of SARS-CoV-2 are accurately identified

After demonstrating the specificity and sensitivity of the variant molecular beacon assay with in vitro-transcribed RNAs, we performed this assay on RNA purified from inactivated virus obtained from ZeptoMetrix (Buffalo, NY). As can be seen in Figure 4A, our assay identified each of the mutations as listed in Figure 1B for the respective VOC. We next tested a set of patient samples (RNA extracted from mid-nasal swabs) that were SARS-CoV-2 positive, as identified by a multiplex screening assay, which we previously developed (Figure S2). After being identified as positive for SARS-CoV-2 by this assay, the samples were sequenced to identify all mutations present in the strains’ genomes, this sequencing data is available online at https://data.mendeley.com/datasets/v3n2dzt8st/1. Afterwards, the RNA extracted from the samples were then provided to Rutgers in a blinded manner for variant identification.

• Download figure
• Open in new tab

Figure 4.

Threshold cycle (C_t) data for virus and patient samples using the variant molecular beacon assay. A. Virus samples. All VOC virus samples tested were accurately identified as shown in Figure 1B. B. Patient samples. All samples test positive for the N gene control. For the other targets, a C_t value less than 40 indicates the presence of the specified mutation, which can be used to identify the variants of the virus. For example, the presence of the T478K and L452R in pt19 indicates that this virus is the Delta variant. All the specified mutations shown in this figure have been sequence confirmed by deep sequencing.

These RNA samples were tested using the two-tube variant molecular beacon assay, as described above. Figure 4B shows the threshold cycles for each molecular beacon target in each sample. All samples tested positive for the N gene positive control signal. Two samples were identified as the Alpha variant (B.1.1.7 with d69-70 and N501Y), two samples were identified as the Epsilon variant (B.1.427/429 with only L452R), eight samples were identified as the Delta variant (B.1.617.2 with L452R and T478K), ten samples had the E484K mutation, although not associated with any VOCs, and four samples had none of the targeted mutations and therefore are not any of the currently identified VOCs.

After strain identification was performed with the variant molecular beacon assay, the results were compared with sequencing data, which was in complete agreement. Our results indicate 100% sensitivity and specificity of the assay as tested so far. It is also notable that one of the samples possessed an L452Q mutation, which is similar to the L452R mutation, but the L452R molecular beacon did not show a fluorescent signal due to the L452Q mutation, further confirming the specificity of the results of our single nucleotide discriminating molecular beacons.

Design and implementation of an Omicron identifying molecular beacon assay

The recent emergence of the omicron variant, which possesses over 30 mutations in the spike protein, necessitates the design of a new identifying single nucleotide discriminating molecular beacon. Since a number of Omicron mutations occur in the target regions of the two-tube assay, it is not feasible to incorporate uniquely Omicron specific molecular beacons into the two-tube assay discussed above in a straightforward manner. Therefore, a new two-color Omicron specific assay was designed by targeting the E484A mutation, which is unique in the Omicron variant (represented in Fig. 5A), along with N gene detection. Previous studies have also shown that this mutation is important for immune evasion¹⁹. The melt curve for the designed molecular beacon is shown in Figure 5B, which shows high specificity for its intended target at the annealing temperature of the assay of 58°C. Next the sensitivity was tested using in vitro-transcribed Omicron spike fragment RNA, similar to what was done previously. The results in Figure 5C show the sensitivity to detect 50 copies of the RNA target. The sequences of probes and primers are listed in Table 3 and the amplicon schematic is shown in Figure 5A. This Omicron assay was tested against the other VOC sequences and the results in Figure 5D show clear identification of the Omicron variant, with no signals from any other target sequences. We then tested thirty-three recent patient samples identifying seventeen of them as Omicron (Figure 5E). These samples were also analyzed using the TaqPath COVID-19 Multiplex Diagnostic Solution assay (ThermoFisher, Waltham, MA) in which Omicron variant fails to produce a S gene specific signal because one of the primers binds in the Omicron variant’s d69-70 mutation. This S gene dropout data was provided to Rutgers in a blinded manner and upon comparison it was found to be in complete agreement, indicating high accuracy of our assay. However, in contrast with TaqPath COVID-19 Multiplex Diagnostic Solution assay our assay yields a positive signal for a unique mutation in the Omicron variant.

• Download figure
• Open in new tab

Figure 5.

Duplex assay for the Omicron variant. A. Schematic showing the locations of primers and molecular beacons. B. Thermal denaturation profile of the molecular beacon with various target sequences. C. Linearity of detection of in vitro-transcribed RNA corresponding to a portion of Omicron S gene. D. Specificity of the Omicron assay using synthetic targets (described in Figure 3). E. Analysis of 33 recently collected SARS-CoV-2 samples by the duplex omicron assay.