Introduction

Following the report of a case in Wuhan on 31 December 2019, the rapid spread and highly infectious nature of the newly emerged coronavirus has resulted in a worldwide pandemic, as declared by the World Health Organization (WHO) on 11 March 20201. The causative agent of Coronavirus Disease 2019 (COVID-19) has been classified as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and is closely related to the severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) coronaviruses, which were responsible for outbreaks in 2003 and 2012, respectively2. As of 8 July, there have been 11,850,000 SARS-CoV-2-confirmed cases worldwide, with 544,000 deaths in 213 countries and territories3. The fast rate of SARS-CoV-2 human-to-human transmission has resulted in an unprecedented need for diagnostic testing, placing a great strain on public health departments in every country. Diagnostic testing is essential not only for the identification of infection in patients but also for tracking and containment of viral spread within communities, testing of unresolved cases, and daily screening of medical frontline workers.

Automated workflows are highly preferable over manual protocols to achieve meaningful throughput, diagnostic precision, and to exclude human error from the sample processing pipeline. Typical automated systems such as the Roche cobas® unit can process hundreds of samples per day with minimal staff support, while ensuring uniform processing and sample tracking. As with other similar automated diagnostic testing platforms, they are costly, not available in the numbers needed to process hundreds of thousands of samples per day in the United Kingdom and currently suffer from reagent supply shortages. Thus, an urgent need has arisen for the adaptation of alternative automated liquid-handling platforms and diagnostic test approaches and workflows, ideally designed in an open and modular way to allow for diversification of reagent supply away from mainstream and overstretched reagent sources.

Many research institutions around the world have established non-commercial Biofoundries, which offer integrated infrastructure including state-of-the-art automated high-throughput (HT) equipment to enable the design-build-test cycle for large-scale experimental designs in synthetic biology4. This infrastructure, in combination with technical expertise in molecular biology, analytics, automation, engineering, and software development, provides an excellent, self-sufficient, and agile capability to quickly establish platforms for prototy** biological testing standards and develo** liquid-handling workflows, such as those needed for automated diagnostic testing of SARS-CoV-2. In the London Biofoundry, we rapidly re-configured existing liquid-handling infrastructure to establish an automated HT SARS-CoV-2 diagnostic workflow with reverse-transcriptase quantitative PCR (RT-qPCR), CRISPR-Cas13a, and RT-loop-mediated isothermal amplification (RT-LAMP)-mediated outputs.

Armoured RNA particles are non-infectious RNA virus surrogates consisting of MS2 bacteriophage capsids containing an RNA template of choice5. Previously, they have been employed as diagnostic reference tools for the detection of respiratory viruses such as Influenza A and B, as well as SARS-CoV6,23,24. This toolkit increases the resilience of the SARS-CoV-2 NAT in case of shortages in extraction materials, RT-qPCR master mix, and laboratory equipment availability. This modularity is created not only by generating custom protocols for several commercially available kits, but also by adapting the CRISPR-Cas13a detection and colorimetric LAMP systems to HT SARS-CoV-2 diagnostic testing. CRISPR-based detection technologies are also currently being developed by Sherlock Biosciences and Mammoth Biosciences to provide at-home point-of-care testing kits10,19,25, as well as in CARMEN-Cas13, a microwell array that multiplexes virus detection26.

Our workflow is easy to scale up, cost-effective, and can provide similar output capacity to that offered by the gold standard of commercial automated systems. For example, a single FeliX liquid handler and qPCR thermocycler can match the largest state-of-the-art Roche cobas® 8800 platform, which can process 960 samples in eight hours. In addition, excess viral RNA remaining from the FeliX patient sample RNA extraction can be diverted to alternative analysis workflows such as next-generation sequencing, which is not possible for some commercial platforms. Finally, our automated RNA extraction and qPCR workflow requires minimal specialist training and can be launched within one day. It is currently installed—and used—in NHS diagnostic labs, where patient sample testing has been validated against large commercially available platforms, matching their precision and throughput.

Although NHS labs currently rely on qPCR workflows for all SARS-CoV-2 diagnostic testing, the potential of alternative detection technologies would allow for HT testing for population screening and in low-resource settings. Miniaturizing LAMP and CRISPR reactions results in a slight loss of sensitivity and therefore may not be suitable for making diagnostic decisions where qPCR capacity is available; however, their isothermal incubation allows for thousands of samples to be tested simultaneously (Comparison of Methodologies in Supplementary Table 2). LAMP is a particularly attractive technique, because it has also been shown to be sensitive with heat-inactivated samples, removing the bottleneck of RNA extraction27. Furthermore, these solutions can be deployed for community testing in low-resource settings or at the point-of-care without expensive equipment requirements.

Engineered VLPs have been widely reported and commercially used as controls and standards in nucleic acid-based diagnostic tests (Asuragen), and have been developed as antigen epitopes in serological assays, where they are used to detect patient antibodies (Native Antigen)28,29. MS2 VLPs carrying RNA payload, such as those used in this study for the detection of SARS-CoV-2 N-gene RNA, provide a quick and reproducible system for generating extremely stable NAT controls. As such, we have purified and quantified large batches that are ready to be shared with and employed by others for diagnostic test development that relies on viral RNA detection. Furthermore, our VLP production and characterization workflow can be modified to rapidly generate new controls mimicking emerging viral threats, thus enhancing preparedness for the development of new diagnostics in future epidemic or pandemic scenarios. In addition, automation equipment available in biofoundries can be used for large-scale testing of antigen-presenting VLPs in develo** antibody-based enzyme-linked immunosorbent assay diagnostics and for performing HT antiviral drug screens. The London Biofoundry is a founding member of the Global Biofoundry Alliance, which currently encompasses 26 such entities worldwide30. This network allows for easy sharing of reagents, protocols, and technical know-how. Therefore, automated diagnostic workflows developed by one partner can be quickly replicated around the world and increase capacity for testing and drug development to help counteract and prevent the global spread of emerging pathogens.

Methods

Primers and probes

Primers and probes were ordered from IDT or Biolegio and can be found in the Supplementary Information in Supplementary Tables 3, 4, and 5.

VLP preparation

The nucleic acid sequence of the N-gene of SARS-CoV-2 (accession number: NC_045512) was ordered from GeneArt (Thermo Fisher Scientific). The N-gene was cloned into a MS2 VLP expression plasmid backbone (Addgene #128233) using Type IIs assembly. The sequence-verified (Eurofins Genomics) plasmid (Addgene #155039) was then transformed into Rosetta 2 (DE3) pLysS cells (Merck Millipore). An overnight culture was used to inoculated 200 mL of Terrific Broth (Merck) supplemented with 35 µg/mL of Chloramphenicol (Merck) and 50 µg/mL of Kanamycin (Merck), and grown at 37 °C, 200 r.p.m. until an OD of 0.8. The culture was induced by supplementing with 0.5 mM IPTG (Merck) and grown at 30 °C for a further 16 h. Cells were collected at 3220 × g at 4 °C and stored at −20 °C for later purification.

All protein purification steps were performed at 4 °C. The cell pellet was resuspended in 4 mL Sonication Buffer (50 mM Tris-HCl pH 8.0, 5 mM MgCl2, 5 mM CaCl2, and 100 mM NaCl) with 700 U RNase A (Qiagen), 2500 U BaseMuncher (Expedeon), and 200 U Turbo DNase (Thermo Fisher Scientific). The cells were sonicated for a total of 2 min (50% amplitude, 30 s on, 30 s off) on wet ice. The lysate was then incubated for 3 h at 37 °C. The lysate was centrifuged at 10,000 × g for 10 min at room temperature in a microcentrifuge. The supernatant was then filtered with a 5 µm cellulose acetate (CA) filter before being mixed 1 : 1 with 2× Binding Buffer (100 mM monosodium phosphate monohydrate pH 8.0, 30 mM Imidazole, 600 mM NaCl).

Supernatant was applied to a 5 mL HiTrap® TALON® Crude column (Cytiva) with a HiTrap® Heparin HP column (Cytiva) in series on an ÄKTA pure (Cytiva) primed with Binding Buffer (50 mM monosodium phosphate monohydrate pH 8.0, 15 mM Imidazole, 300 mM NaCl). The protein was eluted with a linear gradient of elution buffer (50 mM monosodium phosphate monohydrate pH 8.0, 200 mM Imidazole, 300 mM NaCl) and then desalted and buffer exchanged into STE buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA, 100 mM NaCl) using an Amicon Ultra-15 10 K Centrifuge Filter (Merck). The protein concentration was measured using the Qubit Protein Assay Kit and Qubit 3 Fluorometer (Thermo Fisher Scientific). The protein was then diluted in STE buffer, aliquoted, and stored at −80 °C.

Reverse-transcriptase droplet digital PCR

Droplet digital PCR was performed using the Bio-Rad QX200 Droplet Digital PCR system. Reactions were set up using the One-Step RT-ddPCR Advanced Kit for Probes (Bio-Rad) with primer and probe concentrations of 500 nM and 125 nM, respectively. Data were exported in CSV format and analysed using a custom Python implementation (https://github.com/mcrone/plotlydefinerain) of an online tool (http://definetherain.org.uk). The online tool uses a positive control to define positive and negative droplets using K-means clustering, with rain being determined as anything outside three standard deviations from the mean of the positive and negative clusters. It then calculates final concentration based on Eq. 1.

$$c = - {\it{ln}}\frac{{N_{\mathrm{{neg}}}}}{N}/V_{\mathrm{{droplet}}}$$
(1)

c = calculated concentration (copies/µL)

Nneg = number of negative droplets

N = total number of droplets

Vdroplet = average volume of each droplet (0.91 × 10−3 µL).

Dynamic light scattering

DLS was performed using a Zetasizer Nano (Malvern Panalytical) according to the manufacturer’s instructions.

Quantitative PCR

qPCR experiments were designed using the combination of SAS JMP and Riffyn. Primers, probes, and their relative concentrations were those recommended by the CDC and were ordered from IDT. TaqPath 1-Step RT-qPCR Master Mix (Thermo Fisher Scientific), TaqMan Fast Virus 1-Step Master Mix (Thermo Fisher Scientific), or Luna Universal Probe One-Step RT-qPCR (NEB) were used as the relevant master mixes. qPCR reactions were otherwise set up according to the manufacturer’s instructions and thermocycling settings (according to the CDC protocol). Liquid transfers were performed using an Echo 525 (Labcyte). Plates were sealed with MicroAmp Optical Adhesive Films (Thermo Fisher Scientific) and spun at 500 × g in a centrifuge. An Analytik Jena qTower3 auto was used for thermocycling and measurements were taken in the FAM channel.

LwCas13a purification

A plasmid expressing LwCas13 [pC013-Twinstrep-SUMO-huLwCas13a was a gift from Feng Zhang (Addgene plasmid # 90097)] was transformed into Rosetta 2 (DE3) pLysS cells (Merck Millipore). An overnight culture was inoculated into 1 L of Terrific Broth (Merck) supplemented with 35 µg/mL of Chloramphenicol (Merck) and 50 µg/mL of Kanamycin (Merck), and was grown at 37 °C, 160 r.p.m. to an OD of 0.6. The culture was then induced with 0.5 mM IPTG (Merck), cooled to 18 °C, and grown for a further 16 h. Cells were collected at 3220 × g at 4 °C and stored at −20 °C for later purification.

All protein purification steps were performed at 4 °C. The cell pellet was resuspended in lysis buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 1 mM dithiothreitol (DTT)) supplemented with protease inhibitors (cOmplete Ultra EDTA-free tablets, Merck) and BaseMuncher (Expedeon), and sonicated for a total of 90 s (amplitude 100% for 1 s on, 2 s off). Lysate was cleared by centrifugation for 45 min at 38,758 × g at 4 °C and the supernatant was filtered through a 5 µm CA filter.

Supernatant was applied to a 5 mL StrepTrap™ HP column (Cytiva) on an ÄKTA pure (Cytiva). The buffer of the system was changed to SUMO digest buffer (30 mM Tris-HCL pH 8, 500 mM NaCl, 1 mM DTT, 0.15% Igepal CA-630). SUMO digest buffer (5 mL) supplemented with SUMO enzyme (prepared in-house) was then loaded directly onto the column and left to incubate overnight. The cleaved protein was then eluted with 5 mL of SUMO digest buffer. The elution fraction was diluted 1 : 1 with Ion Exchange low salt buffer (20 mM HEPES pH 7, 1 mM DTT, 5% Glycerol), applied to a Hitrap® SP HP column (Cytiva), and eluted using a gradient of the ion exchange high-salt buffer (20 mM HEPES pH 7, 2000 mM NaCl, 1 mM DTT, 5% Glycerol). The eluted protein was then pooled, concentrated, and buffer exchanged into Storage buffer (50 mM Tris-HCl pH 7.5, 600 mM NaCl, 2 mM DTT, 5% Glycerol) using an Amicon Ultra-15 30 K Centrifuge Filter (Merck). The protein concentration was measured using the Qubit Protein Assay Kit and Qubit 3 Fluorometer (Thermo Fisher Scientific). The protein was then diluted, aliquoted, and stored at −80 °C.

crRNA transcription and quantification

DNA was ordered as ssDNA oligonucleotides from IDT and resuspended at 100 µM in Nuclease Free Duplex Buffer (IDT). Oligos contained a full-length reverse strand and a partial forward strand that contained only the T7 promoter sequence. Oligos were annealed by combining forward and reverse strands in equimolar concentrations of 50 µM and heating to 94 °C for 5 min and slow cooling (0.1 °C/s) to 25 °C in a thermocycler.

RNA was then in vitro transcribed using the TranscriptAid T7 High Yield Transcription Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions with a DNA template of 100 nM. Reactions were incubated for 16 h at 37 °C. DNAse I was then added and incubated for 15 min at 37 °C.

Automated purification was performed using the CyBio FeliX liquid-handling robot (Analytik Jena) using RNAClean XP beads (Beckman Coulter) according to the manufacturer’s instructions.

For automated quantification, samples were loaded into a 384 PP Echo plate (Labcyte). Qubit RNA BR Dye and Qubit RNA BR Buffer (Thermo Fisher Scientific) were premixed at a ratio of 1 : 200 and loaded into a 6-well reservoir (Labcyte). Experimental design was performed using a custom Python script and Riffyn with each sample having four technical replicates that were randomly distributed in a Greiner 384 PS Plate (Greiner Bio-One). A standard curve of 9 concentrations (0, 5, 10, 15, 20, 40, 60, 80, 100 ng/µL) was prepared using the standards provided with the Qubit RNA BR Kit (Thermo Fisher Scientific).

A volume of 9.95 µL of the mix of Qubit Dye and Qubit buffer was added to each well using an Echo 525 (Labcyte). A volume of 0.05 µL of sample was then added to each well using the Echo 525 (Labcyte) and the plate was sealed with a Polystyrene Foil Heat Seal (4titude) using a PlateLoc Thermal Microplate Sealer (Agilent). Plates were centrifuged at 500 × g for 1 min before being kept in the dark for 3 min.

Plates were read using a CLARIOstar Plus (BMG Labtech) plate reader, using the following settings: excitation wavelength of 625–15 nm, dichroic of 645 nm, and emission of 665–15 nm and the Enhanced Dynamic Range (EDR) function. RNA molar concentration values were calculated, and the concentration was then normalized, RNA aliquoted and subsequently stored at −80 °C.

CRISPR-Cas13a assays with PCR amplification

Experiments were designed and randomized using SAS JMP and Riffyn. Targets were pre-amplified using the Luna Universal One-Step RT-qPCR kit (NEB) with a primer concentration of 500 nM for 45 cycles. All concentrations are final CRISPR reaction concentrations and the final CRISPR reaction volumes were 5 µL. An Echo 525 (Labcyte) was used to transfer CRISPR Master Mix (50 nM LwCas13a, 1 U/mL murine RNAse inhibitor (NEB), 4 mM Ribonucleotide Solution Mix (NEB), 1.5 U/µl T7 RNA Polymerase (Thermo Fisher Scientific) and 1.25 ng/µL HEK293F background RNA) in Nuclease Reaction Buffer (20 mM HEPES pH 6.8, 60 mM NaCl, 9 mM MgCl2) to a 384-well Small Volume LoBase Microplate (Greiner Bio-One). crRNA (25 nM) and 200 nM poly-U fluorescent probe (5′-/56-FAM/rUrUrUrUrU/3IABkFQ/-3′) were then added separately. An Echo 550 (Labcyte) was used to transfer pre-amplified products from a 384 LDV Plus Echo plate (Labcyte) to initiate the reaction, the plate was sealed, spun at 500 × g for 1 min and read using a CLARIOstar Plus (BMG Labtech) plate reader with an excitation wavelength of 483-14 nm, emission of 530-30 nm, dichroic filter of 502.5 nm, and EDR enabled. Double orbital shaking of 600 r.p.m. for 30 s was performed before the first cycle. The reactions were incubated at 37 °C with readings taken every 2 min. Each reaction was normalized between a water input (background fluorescence) as 0 and an RNase I (Thermo Fisher Scientific) input (0.25 U) as 1 (RNase I cleaves all of the fluorescent probe and thus serves as a positive relative control).

Colorimetric LAMP reactions with VLPs

Experiments were designed and randomized using SAS JMP and Riffyn. Colorimetric LAMP reactions (NEB WarmStart® Colorimetric LAMP 2× Master Mix) were performed with a lower final reaction volume of 5 µL. Master Mix, primers, and template were transferred to a 384-well small volume LoBase plate (Greiner Bio-One) using an Echo 525 (Labcyte). The plate was then sealed with a MicroAmp Optical Adhesive Film (Thermo Fisher Scientific) and centrifuged for 1 min at 500 g. The plate was incubated at 65 °C in a CLARIOstar Plus (BMG Labtech) plate reader and absorbance measurements were taken at 415 nm every minute for 60 min. Double orbital shaking of 600 r.p.m. for 30 s was performed before the first, sixth, and eleventh cycles.

RNA extraction

RNA extraction was performed using a custom Analytik Jena CyBio FeliX script (available on reasonable request) for the Analytik Jena InnuPREP Virus DNA/RNA Kit-FX or the Promega Maxwell HT Viral TNA Kit. Samples of 200 µL were run and eluted in 50 µL of RNase Free Water.

qPCR patient validation

Clinical material (viral transport medium from throat/nose swabs), provided for validation by NWLP, included samples left over after clinical diagnosis as per standard practice for the validation of new assays and platforms. Patient samples were stored at room temperature for no more than 48 h after the initial analysis by NWLP before they were purified and analysed on our platform. Results (Ct values) were compared directly with those obtained by NWLP. As NWLP uses a nested PCR method, Ct values were reported as being the summation of the first and second PCR steps.

qPCR reactions were set up using the TaqPath 1-Step RT-qPCR Master Mix, CG kit, and the CDC N1 Primers according to the manufacturer’s instructions and thermocycling settings (according to the CDC protocol). Final reaction volumes were 10 µL with 5 µL of extracted RNA template. Liquid transfer of the qPCR master mix was performed using an Echo 525 (Labcyte) from a 6-well reservoir (Labcyte). Extracted RNA templates were transferred using a multichannel pipette. Plates were sealed with MicroAmp Optical Adhesive Films (Thermo Fisher Scientific) and spun at 500 × g in a centrifuge. An Analytik Jena qTower3 auto was used for thermocyling and measurements were taken in the FAM channel.

CRISPR-Cas13a assays with RT-RPA amplification

Experiments were designed and randomized using SAS JMP and Riffyn. Targets were pre-amplified using the TwistAmp Liquid Basic Kit (TwistDx) supplemented with 0.5 U/µL Murine RNase Inhibitor (NEB) and 0.08 U/µL Omniscript (Qiagen). Final reactions had a final volume of 14 µL and were set up in Echo 384 LDV Plus plates (final primer concentration of 0.45 µM and 2 µL of purified patient RNA template) and incubated at 42 °C for 30 min in a CLARIOstar Plus (BMG Labtech) plate reader with double orbital shaking of 600 r.p.m. for 30 s at 5 min. All concentrations are final CRISPR reaction concentrations and the final CRISPR reaction volumes were 5 µL. An Echo 525 (Labcyte) was used to transfer CRISPR Master Mix (50 µM LwCas13a, 1 U/µL Murine RNase inhibitor (NEB), 4 mM Ribonucleotide Solution Mix (NEB), 1.5 U/µL T7 RNA Polymerase (Thermo Fisher Scientific), and 1.25 ng/µL HEK293F background RNA) in Nuclease Reaction Buffer (20 mM HEPES pH 6.8, 60 mM NaCl, 9 mM MgCl2) to a 384-well Small Volume LoBase Microplate (Greiner Bio-One). crRNA (25 nM) and 200 nM poly-U fluorescent probe (5′-/56-FAM/rUrUrUrUrU/3IABkFQ/-3′) were then added separately. An Echo 550 (Labcyte) was used to transfer pre-amplified products (0.25 µL) from the 384 LDV Plus Echo plate (Labcyte) to initiate the reaction, the plate was sealed, centrifuged at 500 × g for 1 min and read using a CLARIOstar Plus (BMG Labtech) plate reader with an excitation wavelength of 483-14 nm, emission of 530-30 nm, dichroic filter of 502.5 nm and EDR enabled. Double orbital shaking of 600 r.p.m. for 30 s was performed before the first cycle. The reactions were incubated at 37 °C with readings taken every 2 min. Each reaction was normalized between a water input as 0 (background fluorescence) and an RNase I (Thermo Fisher Scientific) input (0.25 U) as 1 (RNase I cleaves all of the fluorescent probe and thus serves as a positive relative control).

Colorimetric LAMP reactions with patient samples

Experiments were designed and randomized using SAS JMP and Riffyn. Colorimetric LAMP reactions (NEB WarmStart® Colorimetric LAMP 2× Master Mix) were performed as previously described11 but with a lower final reaction volume of 5 µL and template of 2 µL. Master Mix, primers, and template were transferred to a 384-well Small Volume LoBase plate (Greiner Bio-One) using an Echo 525 and Echo 550 (Labcyte). The plate was then sealed with a MicroAmp Optical Adhesive Film (Thermo Fisher Scientific) and centrifuged for 1 min at 500 x g. The plate was incubated at 65 °C in a CLARIOstar Plus (BMG Labtech) plate reader and absorbance measurements were taken at 415 nm every minute for 60 min. Double orbital shaking of 600 rpm for 30 seconds was performed before the 1st, 6th, and 11th cycles.

Ethics statement

Surplus clinical material was used to validate the assay as per normal practice and does not require ethical review.

Reporting summary

Further information on research design is available in the Nature Research  Reporting Summary linked to this article.