Rapid High-Throughput Whole-Genome Sequencing of SARS-CoV-2 by Using One-Step Reverse Transcription-PCR Amplification with an Integrated Microfluidic System and Next-Generation Sequencing

The long-lasting global COVID-19 pandemic demands timely genomic investigation of SARS-CoV-2 viruses. Here, we report a simple and efficient workflow for whole-genome sequencing utilizing one-step reverse transcription-PCR (RT-PCR) amplification on a microfluidic platform, followed by MiSeq amplicon sequencing.

ABSTRACT

The long-lasting global COVID-19 pandemic demands timely genomic investigation of SARS-CoV-2 viruses. Here, we report a simple and efficient workflow for whole-genome sequencing utilizing one-step reverse transcription-PCR (RT-PCR) amplification on a microfluidic platform, followed by MiSeq amplicon sequencing. The method uses Fluidigm integrated fluidic circuit (IFC) and instruments to amplify 48 samples with 39 pairs of primers, including 35 custom-designed primer pairs and four additional primer pairs from the ARTIC network protocol v3. Application of this method on RNA samples from both viral isolates and clinical specimens demonstrates robustness and efficiency in obtaining the full genome sequence of SARS-CoV-2.

INTRODUCTION

Severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) (family Coronaviridae, genus Betacoronavirus) is responsible for the global pandemic of coronavirus disease 2019 (COVID-19) (1–4). Since its emergence in Wuhan, China in November/December 2019, the disease has rapidly spread worldwide. As of 10 March 2021, there have been over 117 million confirmed cases and over 2.6 million deaths in 192 countries or regions (https://coronavirus.jhu.edu/) (5–8). In addition to the continuously growing number of SARS-CoV-2 infections, increased complexity and diversity of disease symptoms are also observed. Despite the large number of whole-genome sequences of SARS-CoV-2 available in GenBank and other public sources, the unprecedented scale of the viral transmission and the emergence of clinically significant variant strains demand more genomic data to be produced using cost-effective and quality-consistent methodology (9–12).

Targeted, whole-genome amplification and next-generation sequencing (NGS) techniques have been used in sequencing SARS-CoV-2 genomes (8, 10, 13). In contrast to the high-throughput capacity of NGS, using a conventional multiplexing reverse transcription and PCR (RT-PCR) procedures to amplify the large 30-Kb viral genomic RNA of SARS-CoV-2 is technically challenging, requires advanced primer and pooling design, demands experienced hands-on skills, and risks contamination and human error from manual manipulation. A rapid and streamlined approach with fewer manual steps to obtain whole-genome amplicons suitable for NGS is desired.

In this study, we utilize an integrated microfluidic nucleic acid amplification system (14), custom primer design, and a one-step RT-PCR program to amplify whole SARS-CoV-2 genomes. After NGS of the amplicons, an in-house developed bioinformatics pipeline is used to rapidly obtain genome sequences with graphic summaries for data and results visualization.

MATERIALS AND METHODS

SARS-CoV-2 RNA samples and real-time RT-PCR.

SARS-CoV-2 RNA samples used in this study include RNA extracted from viral isolate R4717 in the US Army Medical Research Institute of Infectious Disease (USAMRIID) and RNA extracted from deidentified clinical respiratory specimens. A one-step RT-qPCR method targeting RNA-dependent RNA polymerase (RdRp) was used to quantify SARS-CoV-2 RNA. Serially diluted in vitro transcripts (IVT), corresponding to nucleotide region 15431 to 15530 of NC_045512.2 for strain Wuhan-Hu-1, were prepared and used as real-time RT-PCR standards for quantification of genome equivalent copy number (GE) of SARS-CoV-2 RNA. Real-time RT-PCR was performed using the protocol from Carman et al. (15) with the SuperScript III one-step RT-PCR system and Platinum Taq polymerase. The QuantStudio 7 Flex real-time PCR system and software (Thermo Fisher Scientific, Inc.) were used for data acquisition.

Whole-genome RT-PCR amplification and Illumina sequencing.

For whole-genome RT-PCR amplification, 35 primer pairs covering the 29,903-bp SARS-CoV-2 reference genome (NC_045512.2) were custom designed by Fluidigm Corporation (South San Francisco, CA) (Table 1). The RT-PCR products are approximately 1 kb for all amplicons. One-step RT-PCR amplification was performed using the Fluidigm Access Array (AA) nucleic acids amplification system (Fluidigm Corporation, CA) and the SuperScript III one-step RT-qPCR System with Platinum Taq High Fidelity (Thermo Fisher Scientific, Inc.). Four additional pairs of primers were selected from the ARTIC network protocol v3 (https://www.protocols.io/view/ncov-2019-sequencing-protocol-v3-locost-bh42j8ye) and added to the primer panel to fill the gap/dip regions (Table 1).

TABLE 1.

Primers for RT-PCR amplification of SARS-CoV-2 genome using Fluidigm system^a

Forward primer name	Forward primer sequence	Reverse primer name	Reverse primer sequence	Amplicon length (bp)	From	To
nCoVF1	TTCCCAGGTAACAAACCAACCAA	nCoVR1	AGGTGTCTGCAATTCATAGCTCTT	1011	17	1027
nCoVF2	CCGAACAACTGGACTTTATTGACAC	nCoVR2	GCCTTCTGTAAAACACGCACAGA	1048	912	1959
nCoVF3	ACTGAGTCCTCTTTATGCATTTG	nCoVR3	AGCATCTGCCACAACACA	1048	1852	2899
nCoVF32	AGAAGAAACTGGCCTACTCATGC	nCoVR32	ACATTGGCTGCATTAACAACCAC	1017	2937	3953
nCoVF33	TTTGGAATTTGGTGCCACTTCTG	nCoVR33	CCTCTTGAACAACATCACCCACT	1004	3642	4645
nCoVF4	AGAGTTTGTGTAGATACTGTTCG	nCoVR4	TATAGGAACCAGCAAGTGAGATG	1048	3752	4799
nCoVF5	GTTTCTGTTTCTTCACCTGATGC	nCoVR5	TGGTGCTGACATCATAACAAAAG	1008	4688	5695
nCoVF6	CCTTGTACGTGTGGTAAACAAGC	nCoVR6	GCTAAACCATGAGTAGCAAGGGT	1028	5621	6648
nCoVF7	TACAGAAGAGGTTGGCCACACAG	nCoVR7	TGTACATTCGACTCTTGTTGCTCT	1015	6523	7537
nCoVF8	ATGTGCATGTTGTAGACGGTTGT	nCoVR8	GCAGCACTACGTATTTGTTTTCGT	1009	7455	8463
nCoVF9	GCAGGTAGCAAAAAGTCACAACA	nCoVR9	AGATGCTGATATGTCCAAAGCAC	1036	8368	9403
nCoVF10	GGAGTTTTCTGTGGTGTAGATGC	nCoVR10	AGGTGTCTTAGGATTGGCTGTAT	1041	9311	10351
nCoVF11	TCTAAAGTTGCGTAGTGATGTGCT	nCoVR11	TCCAGTTTGAGCAGAAAGAGGTC	994	9835	10828
nCoVF12	GTTTGTTCGCATTCAACCAGGAC	nCoVR12	ACACTCTCCTAGCACCATCATCA	1031	10360	11390
nCoVF34	ATATGCCTGCTAGTTGGGTGATG	nCoVR34	CTGCATCACGGTCAAATTCAGAT	1026	11726	12751
nCoVF35	GCCTCAGAGTTTAGTTCCCTTCC	nCoVR35	ATTAGTGATTGGTTGTCCCCCAC	1047	12598	13644
nCoVF13	AAGCTGGTAATGCAACAGAAGTG	nCoVR13	TTTCGCATGGCATCACAGAATTG	1011	13023	14033
nCoVF14	TGTAGAAAACCCAGATATATTACGC	nCoVR14	ATTTGTCTAGGTTGTTGACGATG	1007	13935	14941
nCoVF15	TTGATTGTTACGATGGTGGCTGT	nCoVR15	AGGTACACATAATCATCACCCTG	1048	14879	15926
nCoVF16	TGTCTGAAGCAAAATGTTGGACT	nCoVR16	CAACAGCATCACCATAGTCACCT	1044	15821	16864
nCoV-2019_54_LEFT*	TGAGTTAACAGGACACATGTTAGACA	nCoV-2019_56_RIGHT*	ACACTATGCGAGCAGAAGGGTA	1033	16119	17152
nCoV-2019_55_LEFT*	ACTCAACTTTACTTAGGAGGTATGAGCT	nCoV-2019_57_RIGHT*	GTAATTGAGCAGGGTCGCCAAT	1035	16417	17452
nCoVF17	ACCTAGACCACCACTTAACCGAA	nCoVR17	CAGCTTTTCTCCAAGCAGGGTTA	1016	16749	17764
nCoV-2019_56_LEFT*	ACCTAGACCACCACTTAACCGA	nCoV-2019_59_RIGHT*	AAGAGTCCTGTTACATTTTCAGCTTG	1313	16749	18062
nCoV-2019_58_LEFT*	TGATTTGAGTGTTGTCAATGCCAGA	nCoV-2019_60_RIGHT*	GGTACCAACAGCTTCTCTAGTAGC	966	17382	18348
nCoVF18	GCTTAAAGCACATAAAGACAAATCA	nCoVR18	GTGCGCTCAGGTCCTATTTT	1041	17616	18656
nCoVF19	GTCTTATGGGCACATGGCTTTGA	nCoVR19	AGCCACATTTTCTAAACTCTGAAGTC	1050	18589	19638
nCoVF20	ACATGATGATCTCAGCTGGCTTT	nCoVR20	TCACTTTGACAACCTTAGAAACTACA	1047	19535	20581
nCoVF21	TTGGAGAAGCCGTAAAAACACAG	nCoVR21	TTTATAGCCACGGAACCTCCAAG	1045	20123	21167
nCoVF22	TTTAAGACAGTGGTTGCCTACGG	nCoVR22	GGACTGGGTCTTCGAATCTAAAGT	1001	20910	21910
nCoVF23	TAAGGGGTACTGCTGTTATGTCTT	nCoVR23	TCAAGTGCACAGTCTACAGCATC	1025	21419	22443
nCoVF24	GGGTTATCTTCAACCTAGGACTT	nCoVR24	ACATCCTGATAAAGAACAGCAAC	1044	22363	23406
nCoVF25	ACAATTTGGCAGAGACATTGCTG	nCoVR25	AAACCTATAAGCCATTTGCATAGC	1030	23251	24280
nCoVF26	CGGGTACAATCACTTCTGGTTGG	nCoVR26	ACTATGGCAATCAAGCCAGCTAT	1048	24198	25245
nCoVF27	TTCAAAAAGAAATTGACCGCCTC	nCoVR27	CCGTCGATTGTGTGAATTTGGAC	1047	25098	26144
nCoVF28	ACATGTTACCTTCTTCATCTACA	nCoVR28	GACTGTATGCAGCAAAACCTG	1045	26070	27114
nCoVF29	GTGACATCAAGGACCTGCCTAAA	nCoVR29	ATAGGACACGGGTCATCAACTACAT	1012	26998	28009
nCoVF30	CTGTAGCTGCATTTCACCAAGAA	nCoVR30	CAAGCTGGTTCAATCTGTCAAGC	1037	27928	28964
nCoVF31	GAACTTCTCCTGCTAGAATGGC	nCoVR31	TCACATGGGGATAGCACTAC	961	28884	29844

^aThe primers were custom designed by Fluidigm, with four additional pairs (marked with *) from the ARTIC network protocol v3 (https://www.protocols.io/view/ncov-2019-sequencing-protocol-v3-locost-bh42j8ye).

SARS-CoV-2 R4717 RNA was serially diluted to the concentrations of 10⁶, 10⁵, 10⁴, 10³, 10², 10, and 1 GE/μl (genomic equivalence per microliter). For Fluidigm RT-PCR amplification, each sample well in the integrated fluidic circuit (IFC) chip contained 1.45 μl of RNA sample and 3.55 μl of sample mixture solution, consisting of 3 μl of 2× reaction mix, 0.2 μl of dimethyl sulfoxide (DMSO), 0.05 μl of RNaseOUT, 0.05 μl of SuperScript III RT/Platinum Taq High Fidelity enzyme mix, and 0.25 μl of 20× AA loading reagent. Each well of primer mix contained 10 μl of mixture containing one pair of primers at the final concentration of 4 μM for each primer and 1.25 mM MgSO₄ in AA loading reagent. The primed IFC chip was loaded into the Fluidigm FC1 cycler and amplified using the following cycling conditions: 50°C for 30 min (reverse transcription), 94°C for 2 min, 35 cycles of 94°C for 30 s, 53°C for 30 s, 68°C for 90 s, and final extension at 68°C for 7 min, and then kept at 4°C. The RT-PCR products were purified using Agencourt AMPure XP beads (Beckman Coulter, CA, USA) and then analyzed by using the Agilent TapeStation 4200 system and High Sensitivity DNA D5000 kit (Agilent Technologies, CA, USA) to determine quality and quantity of the amplicons.

The NGS libraries were prepared using Nextera DNA Flex library prep (currently Illumina DNA Prep) kit (Illumina, CA, USA). DNA was fragmentized by tagmentation for 15 min followed by indexing and library amplification, with 6 cycles used for 9 to 25 ng of amplicon or 12 cycles used for 1 to 9 ng of amplicon. The libraries were quantified by use of the Agilent TapeStation and DNA D5000 kit and then pooled with equal molar ratio for each library. Pooled libraries were denatured and diluted to a final concentration of 13.5 pM and then sequenced using the Illumina MiSeq system and reagent kit v3 (600 cycles).

Reference genome mapping assembly.

The full SARS-CoV-2 reference genome, NC_045512.2, was downloaded from GenBank (16) and utilized for reference-based genome assembly using the WRAIR viral disease branch in-house bioinformatics pipeline ngs_mapper version 1.5.0 (https://github.com/VDBWRAIR/ngs_mapper). This pipeline incorporates a series of quality control and assembly processes for fastq sequence reads retrieved from the Illumina MiSeq and other platforms. These processes include filtration to drop poorly indexed reads, read trimming based on quality thresholds using Trimommatic (17), read mapping to the reference genome using the Burrows-Wheeler aligner with maximal exact matches (BWA-MEM) (18), read tagging, variant calling file (VCF) generation, read mapping visualizations, fastq statistic generation, and consensus sequence generation from the VCF using an in-house script basecaller.py. Basecaller.py takes into account the depth of coverage, the frequency of a variant, and the quality of bases to make a consensus base call.

A minimum Phred base quality score of 35 and a minimum depth of coverage of 10 were utilized as the configuration parameters for basecaller.py, meaning that any base with a Phred score below 35 was masked and not taken into account for consensus base calling.

Variants present at a frequency of 20% or higher were denoted in the consensus genome as ambiguous positions following the IUPAC code. The assembled genomes were further manually curated utilizing bam, consensus and variant calling files generated from the ngs_mapper pipeline. Geneious R10 software, integrated genome viewer (IGV) (19), and MEGA version 7 (20) were used for quality control of ambiguous calls, insertions, deletions, and primer-induced mutations. The genomes were processed to only include coding sequence regions by clipping the 5′ and 3′ untranslated regions. MEGA7 was used to align genomes using default parameters with multiple sequence comparison by log expectation (MUSCLE) (21).

Data availability.

Next-generation sequencing read data are available under NCBI BioProject no. PRJNA683873 (see also Table S1 in the supplemental material).

RESULTS

Correlation of Fluidigm RT-PCR yield with amount of input SARS-CoV-2 RNA.

Fluidigm RT-PCR whole-genome amplification was done in the Access Array microfluidic chip using only 1.45 μl of RNA input for each sample and with a maximum capacity of 48 samples and 48 pairs of primers. In this study, a set of 35 primer pairs, i.e., nCOVF/R1-35 in Table 1, were designed and tested for genome RT-PCR amplification by using serial dilutions of purified RNA from SARS-CoV-2 isolate R4717. The experiment was done with four replicates processed in parallel. A single band of amplicons with expected sizes of approximately 1 kb was seen for all concentrations, with the band intensities correlated with copy numbers of SARS-CoV-2 in the serial dilutions. The correlation between the concentrations of yielded amplicons and the input SARS-CoV-2 genome copy numbers (Fig. 1) was significant, with a Pearson’s correlation P value of 3.38e−03.

FIG 1.

Agilent TapeStation analysis of SARS-CoV-2 whole-genome RT-PCR amplification using the Fluidigm system. (A) TapeStation profile of RT-PCR products. From left to right, 0, 1, 10, 10², 10³, 10⁴, 10⁵, and 10⁶ copies of SARS-CoV-2 RNA from isolate R4717 were used. (B) SARS-CoV-2 amplicon concentrations for RT-PCR with serial dilutions of R4717 RNA as inputs. The concentration values were the average and standard deviation from four replicates.

Whole-genome coverage and alignment depth of SARS-CoV-2 sequence assembly.

MiSeq data for the quadruplicated R4717 RNA serial dilutions described above were assembled using ngs_mapper pipeline with SARS-CoV-2 complete genome sequence NC_045512.2 as the mapping reference. As expected, genome assembly results correlated well with SARS-CoV-2 copy numbers in each sample. For reactions with 10⁴ or higher SARS-CoV-2 RNA copies, complete genome sequences were readily obtained and, importantly, had uniform coverage depth across the genome with the peaks matching the regions of amplicon alignment (Fig. 2). For reactions with 10³ copies of SARS-CoV-2 RNA, the assembled genome sequences were nearly complete, except for two small dip/gap at positions 1870 to 2500 and 16800 to 17700. The dip/gap regions were successfully filled by adding two extra primer pairs for each region to the panel of 35 primer pairs. These additional primers were selected from the ARTIC v3 protocol, paired to cover the two regions and added into four separate Fluidigm primer wells. In total, 39 pairs of primers were applied for whole-genome amplification in a single RT-PCR using Fluidigm nucleic acids amplification system.

FIG 2.

Genome coverage and sequence mapping alignment depth for MiSeq reads data. SARS-CoV-2 isolate R4717 RNA serial dilutions were subjected to Fluidigm RT-PCR amplification and MiSeq sequencing. Data were mapped using genome NC_045512.2 as the reference. (A) Genome coverage for 10× serial dilutions of R4717 RNA from 1.0E + 06 (top) to 1 (bottom) copies. Mapping depth was indicated in colors as follows: green, normal depth (depth ≥ 10); cyan, dip (depth 1 to 10); red, gap. Results for four replicates are shown. (B) Sequence mapping graphs for 10⁶, 10⁵, 10⁴ and 10³ copies of RNA from one set of serial dilutions.

SARS-CoV-2 genome sequencing of clinical respiratory specimens.

This method was subsequently used in genome sequencing of SARS-CoV-2 in RNA extracts purified from COVID-19-positive nasopharyngeal swabs. The set of 29 samples contained a wide range of titers, with RT-qPCR threshold cycle (C_T) values from 13.5 to 33.4 (379 to 2.72 × 10⁸ GE/μl) of SARS-CoV-2. The RT-PCR cDNA yield has significant correlation with viral titers in the C_T range of 20 to 35, with P value of 1.26e−05. For the samples with C_T values below 20, or exceedingly high concentrations of 1.0 × 10⁷ or greater, RT-PCR yields were substantially lower than projected (Fig. 3). Importantly, this observation suggests that severe suppression of PCR could occur when samples of extremely high titer are used. Nevertheless, full or nearly complete genome coverage was obtained for all clinical specimens with titers above 1.0 × 10⁴ GE/μl, or an approximate C_T value of 29.

FIG 3.

SARS-CoV-2 whole-genome RT-PCR amplification using the Fluidigm system. Twenty nine RNA extracts from COVID-19-positive nasopharyngeal swabs were amplified and sequenced. RT-PCR yield was quantified using the Agilent TapeStation. Reference mapping coverage was determined by mapping MiSeq read data to reference genome sequence NC_045512.2. The represented coverage was for sequences with alignment depth of 10× or greater.

Comparison of assembled consensus sequences and nucleotide variations.

The Fluidigm one-step RT-PCR protocol was applied to SARS-CoV-2 RNA of highly varied titers. For SARS-CoV-2 isolate R4717 RNA 10× dilutions (Fig. 1), a total of 14 full genomes were assembled and curated. All the full genome consensus sequences were identical except for two replicates of 1.0 × 10⁴ GE/μl. One sample had the ambiguous call Y (T or C) in the alignment nucleotide position 2105, while the other three replicates had a C. Another sample had a T in position 23260, while the other three had the ambiguous call Y. These were the only ambiguous positions found in the samples, showing low sample diversity at a variant frequency of 20% or higher. Taken together, this sequencing approach produces high-accuracy results.

DISCUSSION

Even before the disease was named COVID-19, the sequence was swiftly determined using next-generation sequencing technologies and SARS-CoV-2 was identified as the causative pathogen for the emerging acute respiratory disease (5, 22). The first sequence was made publicly available immediately, with a massive number of sequences subsequently generated and shared, which has greatly facilitated research and development (23–25). All the efforts in developing, improving, and sharing materials and/or methods have played an essential role in sequence-based investigations. All the known sequencing protocols use conventional PCR apparatuses with differences in design and selection of primers and reaction parameters. In this study, we applied a one-step RT-PCR protocol on a microfluidic platform (14) to establish a convenient workflow with throughput, speed, simplicity, consistency, and yield suitable for COVID-19 genome sequencing. The Fluidigm Access Array IFC, as well as the latest Fluidigm Juno model, holds 48 RNA samples (inlets) and 48 primer pairs (inlets). Steps for mixing sample with primers are obviated, which not only substantially reduces pipetting manipulation but also effectively mitigates the chance of sample-to-sample cross-contamination. In this report, we selected 39 primer pairs to obtain even genome coverage. Nine more individual pairs of primers can be easily added to the panel of 48 primer inlets to elevate coverage depth in regions that exhibit as dips/gaps and to quickly address emerging SARS-CoV-2 genetic divergence. Moreover, the total number of primer pairs can be further increased without difficulty by pooling together several compatible primer pairs and adding them into one primer inlet. Each sample is mixed with individual primer pairs in IFC microfluidic chambers for nanoliter (nl) RT-PCR amplifications in a simplex-independent reaction manner. In contrast, conventional PCR methods often need optimization of primer pooling and reaction parameters to circumvent primer-to-primer interference and to avoid highly variable yields among amplicons.

Since input RNA samples are partitioned into individual nl reaction chambers to cross-mix with individual primer pairs, microfluidic applications, including Fluidigm IFC, require sufficient genomic copies in order to achieve whole-genome amplification. In consequence, Fluidigm amplification-based whole-genome sequencing has limitations in sequencing low-titer samples. For RNA samples with 1,000 GE/μl or lower concentration, multiplex RT-PCR based methods or SARS-CoV-2 hybridization-based enrichment method might be a more suitable choice for whole-genome sequencing (26–29). The addition of a first-strand cDNA synthesis step prior to Fluidigm amplification, with a change of Fluidigm thermocycling program from RT-PCR to PCR, may help increase genome coverage for low-titer samples. Additionally, concentrating samples using the SpeedVac vacuum concentrator (Thermo Fisher Scientific) or similar devices and loading samples into multiple sample inlets are simple options to generate more amplicons from low-titer RNA samples. Many, if not all, COVID-19 specimens are tested with molecular tests and the C_T values or equivalent titer scores are readily available for deciding whether one-step Fluidigm amplification-based genome sequencing protocol is appropriate.

Using this method, very few sequence assembly errors were observed throughout the tested SARS-CoV-2 sample genomes. These errors might be due to PCR, sequencing or basecalling algorithm errors, as well as due to normal fluctuations in the minor variant frequencies between the sample aliquots. When assembled sequences show potentially significant nucleotide alterations or indels, thorough examination of the data quality processing, primer trimming, curation of sequence assembly, and detailed laboratory records are needed. Whenever possible, running replicate samples, repeating the experiment entirely, or using other methods are important to validate genome variations and minor variants.

In conclusion, our study demonstrates a convenient SARS-CoV-2 whole-genome sequencing protocol by incorporating one-step RT-PCR amplification, microfluidic technology, and next-generation sequencing to achieve a simple and fast workflow with consistent and high-quality data. The performance of the protocol was verified using viral isolate RNA and tested by sequencing clinical respiratory samples of various viral titers.