Reagents and materials
Detailed information about reagents, including the commercial vendors and stock concentrations, is provided in Supplementary Table 3.
Clinical samples and ethics statement
Clinical samples were de-identified and acquired from clinical studies evaluated and approved by the Institutional Review Board/Ethics Review Committee of the Massachusetts General Hospital and Massachusetts Institute of Technology (MIT) or Redeemer’s University Ethical Review Committee. De-identified clinical samples from Boca Biolistics were obtained commercially under their ethical approvals. The Office of Research Subject Protection at the Broad Institute of MIT and Harvard University approved the use of samples included in this study.
Viral and extracted sample preparation and RT-qPCR testing
For side-by-side comparisons of the two-step SHERLOCK assay and RT-qPCR on viral seedstocks, the 2019-nCoV/USA-WA1-A12/2020 isolate of SARS-CoV-2 was provided by the US CDC. The virus was passaged at the Integrated Research Facility-Frederick in high containment (BSL-3) by inoculating grivet kidney epithelial Vero cells (American Type Culture Collection (ATCC) #CCL-81) at a multiplicity of infection of 0.01. Infected cells were incubated for 48 or 72 h in Dulbecco’s Modified Eagle Medium with 4.5 g/L D-glucose, L-glutamine, and 110 mg/L sodium pyruvate (Gibco) containing 2% heat-inactivated fetal bovine serum (SAFC Biosciences) in a humidified atmosphere at 37 °C with 5% CO2. The resulting viral stock was harvested and quantified by plaque assay using Vero E6 cells (ATCC #CRL-1586) with a 2.5% Avicel overlay and stained after 48 h with a 0.2% crystal violet stain.
For side-by-side comparisons of the two-step SHERLOCK assay and RT-qPCR on patient samples, nasal swab or combined nasal and saliva samples were collected from symptomatic patients in whom COVID-19 was suspected. Nasal swabs were collected and stored in viral transport medium (VTM)42. All nucleic acid extractions were performed using the QIAamp Viral RNA Mini Kit (Qiagen). For a subset of patients, saliva samples were combined with nasal samples during extraction. The starting volume for extraction was 70 μL and extracted nucleic acid was eluted into 60 μL of nuclease-free water. RT-qPCR was performed using either the RT-PCR Reagent Set for COVID-19 Real-time detection (DaAn-GENE) or the GeneFinder™ COVID-19 Plus RealAmp Kit (OSANG Healthcare) using the N target (primer and probe sequences not publically available). RT-qPCR cycling conditions for the DnAn-GENE Kit were as follows: RT at 50 °C for 15 min, heat activation at 95 °C for 15 min and 45 cycles of a denaturing step at 94 °C for 15 s followed by annealing and elongation steps at 55 °C for 45 s. RT-qPCR cycling conditions for the OSANG Healthcare’s Kit were as follows: RT at 50 °C for 20 min, heat activation at 95 °C for 5 min, and 45 cycles with a denaturing step at 95 °C for 15 s followed by annealing and elongation steps at 58 °C for 60 s.
Nasal swabs were collected and stored in UTM (BD) or VTM and stored at −80 °C prior to nucleic acid extraction. For the initial set of 50 NP patient samples, nucleic acid extraction was performed using MagMAXTMmirVanaTM Total RNA Isolation Kit. The starting volume for the extraction was 250 μL and extracted nucleic acid was eluted into 60 μL of nuclease-free water. Extracted nucleic acid was then immediately Turbo DNase-treated (Thermo Fisher Scientific), purified twice with RNACleanXP SPRI beads (Beckman Coulter), and eluted into 15 μL of Ambion Linear Acrylamide (Thermo Fisher Scientific) water (0.8%).
Turbo DNase-treated extracted RNA was then tested for the presence of SARS-CoV-2 RNA using a laboratory-developed, probe-based RT-qPCR assay based on the N1 target of the CDC assay30. RT-qPCR was performed on a 1:3 dilution of the extracted RNA using TaqPathTM 1-Step RT-qPCR Master Mix (Thermo Fisher Scientific) with the following forward and reverse primer sequences, respectively: forward GACCCCAAAATCAGCGAAAT, reverse TCTGGTTACTGCCAGTTGAATCTG. The RT-PCR assay was run with a double-quenched FAM probe with the following sequence: 5′-FAM-ACCCCGCATTACGTTTGGTGGACC-BHQ1-3′. RT-qPCR was run on a QuantStudio 6 (Applied Biosystems) with RT at 48 °C for 30 min and 45 cycles with a denaturing step at 95 °C for 10 s followed by annealing and elongation steps at 60 °C for 45 s. The data were analyzed using the Standard Curve module of the Applied Biosystems Analysis Software.
Patient samples for side-by-side SHINE and RT-qPCR testing (from Boca Biolistics) were extracted using the QIAamp Viral RNA Mini Kit (Qiagen). The starting volume for the extraction was 100 μL and extracted nucleic acid was eluted into 40 μL of nuclease-free water. Extracted RNA was then tested for the presence of SARS-CoV-2 RNA using the laboratory-developed, probe-based RT-qPCR assay mentioned above (based on the N1 target of the CDC assay). Primers, probes, and conditions are the same as mentioned above.
SARS-CoV-2 assay design and synthetic template information
SARS-CoV-2-specific forward and reverse RPA primers and Cas13-crRNAs were designed as previously described19. In short, the designs were algorithmically selected, targeting 100% of all 20 publicly available SARS-CoV-2 genomes at the time, and predicted by a machine learning model to be highly active (Metsky et al., in preparation). Moreover, the crRNA was selected for its high predicted specificity toward detection of SARS-CoV-2, versus related viruses, including other bat and mammalian coronaviruses and other human respiratory viruses (https://adapt.sabetilab.org/covid-19/).
Specificity target sequences were generated using the same design software noted above by providing the amplicon coordinates of the designed assay within the viral species of interest and an alignment of the selected viral species. The specificity targets tested represent the overall medoid of sequence clusters at the provided amplicon for each selected viral species within the designed SARS-CoV-2 SHERLOCK assay.
Synthetic DNA targets with appended upstream T7 promoter sequences (5′-GAAATTAATACGACTCACTATAGGG-3′) were ordered as double-stranded DNA (dsDNA) gene fragments from IDT and were in vitro transcribed to generate synthetic RNA targets. In vitro transcription was conducted using the HiScribe T7 High Yield RNA Synthesis Kit (New England Biolabs (NEB)) as previously described24. In brief, a T7 promoter ssDNA primer (5′-GAAATTAATACGACTCACTATAGGG-3′) was annealed to the dsDNA template and the template was transcribed at 37 °C overnight. Transcribed RNA was then treated with RNase-free DNase I (QIAGEN) to remove any remaining DNA according to the manufacturer’s instructions. Finally, purification occurred using RNAClean SPRI XP beads at 2× transcript volumes in 37.5% isopropanol.
Sequence information for the synthetic targets, RPA primers, and Cas13-crRNA is listed in Supplementary Table 4.
Two-step SARS-CoV-2 assay
The two-step SHERLOCK assay was performed as previously described19,24,26. Briefly, the assay was performed in two steps: (1) isothermal amplification via RPA and (2) LwaCas13a-based detection using a single-stranded RNA (ssRNA) fluorescent reporter. For RPA, the TwistAmp Basic Kit (TwistDx) was used as previously described (i.e., with RPA forward and reverse primer concentrations of 400 nM and a magnesium acetate concentration of 14 mM)26 with the following modifications: RevertAid reverse transcriptase (Thermo Fisher Scientific) and murine RNase inhibitor (NEB) were added at final concentrations of 4 U/µL each, and synthetic RNAs or viral seedstocks were added at known input concentrations making up 10% of the total reaction volume. The RPA reaction was then incubated on the thermocycler for 20 min at 41 °C. For the detection step, 1 µL of RPA product was added to 19 µL detection master mix. The detection master mix consisted of the following reagents (final concentrations in master mix listed), with magnesium chloride added last: 45 nM LwaCas13a protein resuspended in 1× storage buffer (SB: 50 mM Tris pH 7.5, 600 mM NaCl, 5% glycerol, and 2 mM dithiothreitol (DTT); such that the resuspended protein is at 473.7 nM), 22.5 nM crRNA, 125 nM RNaseAlert substrate v2 (Thermo Fisher Scientific), 1× cleavage buffer (CB; 400 mM Tris pH 7.5 and 10 mM DTT), 2 U/µL murine RNase inhibitor (NEB), 1.5 U/µL NextGen T7 RNA polymerase (Lucigen), 1 mM of each rNTP (NEB), and 9 mM magnesium chloride. Reporter fluorescence kinetics were measured at 37 °C on a Biotek Cytation 5 plate reader using a monochromator (excitation: 485 nm, emission: 520 nm) every 5 min for up to 3 h.
Single-step SARS-CoV-2 assay optimization
The starting point for optimization of the single-step SHERLOCK assay was generated by combining the essential reaction components of both the RPA and the detection steps in the two-step assay, described above24,26. Briefly, a master mix was created with final concentrations of 1× original reaction buffer (20 mM HEPES pH 6.8 with 60 mM NaCl, 5% PEG, and 5 µM DTT), 45 nM LwaCas13a protein resuspended in 1× SB (such that the resuspended protein is at 2.26 µM), 136 nM RNaseAlert substrate v2, 1 U/µL murine RNase inhibitor, 2 mM of each rNTP, 1 U/µL NextGen T7 RNA polymerase, 4 U/µL RevertAid reverse transcriptase, 0.32 µM forward and reverse RPA primers, and 22.5 nM crRNA. The TwistAmp Basic Kit lyophilized reaction components (1 lyophilized pellet per 102 µL final master mix volume) were resuspended using the master mix. After pellet resuspension, cofactors magnesium chloride and magnesium acetate were added at final concentrations of 5 and 17 mM, respectively, to complete the reaction.
Master mix and synthetic RNA template were mixed and aliquoted into a 384-well plate in triplicate, with 20 µL per replicate at a ratio of 19:1 master mix:sample. Fluorescence kinetics were measured at 37 °C on a Biotek Cytation 5 or Biotek Synergy H1 plate reader every 5 min for 3 h, as described above. We observed no significant difference in performance between the two plate reader models.
Optimization occurred iteratively, with a single reagent modified in each experiment. The reagent condition (e.g., concentration, vendor, or sequence) that produced the most optimal results—defined as either a lower LOD or improved reaction kinetics (i.e., reaction saturates faster)—was incorporated into our protocol. Thus the protocol used for every future reagent optimization consisted of the most optimal reagent conditions for every reagent tested previously.
For all optimization experiments, the modulated reaction component is described in the figures, associated captions, or associated legends. Across all experiments, the following components of the master mix were held constant: 45 nM LwaCas13a protein resuspended in 1× SB (such that the resuspended protein is at 2.26 µM), 1 U/µL murine RNase inhibitor, 2 mM of each rNTP, 1 U/µL NextGen T7 RNA polymerase, and 22.5 nM crRNA, and TwistDx RPA TwistAmp Basic Kit lyophilized reaction components (1 lyophilized pellet per 102 µL final master mix volume). In all experiments, the master mix components except for the magnesium cofactor(s) were used to resuspend the lyophilized reaction components, and the magnesium cofactor(s) were added last. All other experimental conditions, which differ among the experiments due to real-time optimization, are detailed in Supplementary Table 5.
Single-step SARS-CoV-2 optimized reaction
The optimized reaction (see Supplementary Protocol for exemplary implementation) consists of a master mix with final concentrations of 1× optimized reaction buffer (20 mM HEPES pH 8.0 with 60 mM KCl and 5% PEG), 45 nM LwaCas13a protein resuspended in 1× SB (such that the resuspended protein is at 2.26 µM), 125 nM polyU [i.e., 6 uracils (6U) or 7 uracils (7U) in length, unless otherwise stated] FAM quenched reporter, 1 U/µL murine RNase inhibitor, 2 mM of each rNTP, 1 U/µL NextGen T7 RNA polymerase, 2 U/µL Invitrogen SuperScript IV (SSIV) reverse transcriptase (Thermo Fisher Scientific), 0.1 U/µL RNase H (NEB), 120 nM forward and reverse RPA primers, and 22.5 nM crRNA. Once the master mix is created, it is used to resuspend the TwistAmp Basic Kit lyophilized reaction components (1 lyophilized pellet per 102 µL final master mix volume). Finally, magnesium acetate is the sole magnesium cofactor and is added at a final concentration of 14 mM to generate the final master mix.
The sample is added to the complete master mix at a ratio of 1:19, and the fluorescence kinetics are measured at 37 °C using a Biotek Cytation 5 or Biotek Synergy H1 plate reader as described above.
For the specificity data, fluorescence kinetics were measured at 37 °C using a Molecular Devices SpectraMax M2 plate reader using the same excitation and emission parameters described above; notably, this plate reader model required twice the reporter concentration (250 nM polyU FAM) to achieve a comparable LOD to the Biotek models.
Detection via in-tube fluorescence and lateral flow strip
Minor modifications were made to the optimized single-step and the two-step SARS-CoV-2 reaction to visualize the readout via in-tube fluorescence or lateral flow strip.
For in-tube fluorescence with the optimized single-step reaction, we generated the master mix as described above, except the 7U FAM quenched reporter was used at a concentration of 62.5 nM. The sample was added to the complete master mix at a ratio of 1:19. Samples were incubated at 37 °C, and images were collected after 30, 45, 60, 90, 120 and 180 min of incubation, with image collection terminating once experimental results were clear. A dark reader transilluminator (DR196 model, Clare Chemical Research) or Gel DocTM EZ Imager (BioRad) with the blue tray was used to illuminate the tubes.
For lateral flow readout with the two-step SHERLOCK method, we generated the Cas13-based detection mix as described above, except we used a biotinylated FAM reporter at a final concentration of 1 µM rather than RNase Alert v2. For lateral flow readout using the optimized single-step SHERLOCK assay, we generated the single-step master mix as described above, except we used a biotinylated FAM reporter at a final concentration of 1 µM rather than the quenched polyU FAM reporters. For both two-step and single-step SHERLOCK, the sample was added to the complete master mix at a ratio of 1:19. After 1–3 h of incubation at 37 °C, the detection reaction was diluted 1:4 in Milenia HybriDetect Assay Buffer, and the Milenia HybriDetect 1 (TwistDx) lateral flow strip was added. Sample images were collected 5 min following incubation of the strip. Lateral flow results were assessed either by the user or in an automated fashion by measuring the pixel intensity of the test band using ImageJ.
In-tube fluorescence reader mobile phone application
To enable smartphone-based fluorescence analysis, we designed a companion mobile application pipeline. Using the application, the user captures an image of a set of strip tubes illuminated by a transilluminator. The user then identifies regions of interest in the captured image by overlaying a set of pre-drawn boxes onto experimental and control tubes. Image and sample information is then transmitted to a server for analysis. Within each of the user-selected squares, the server models the bottom of each tube as a trapezoid and uses a convolutional kernel to determine the location of maximal signal within each tube, using data from the green channel of the RGB image. The server then identifies the background signal proximal to each tube and fits a Gaussian distribution around the background signal and around the in-tube signal. The difference between the mean pixel intensity of the background signal and the mean pixel intensity of the in-tube signal is then calculated as the background-subtracted fluorescence signal for each tube. To identify experimentally significant fluorescent signals, a score is computed for each experimental tube; this score is equal to the distance between the experimental and control background-subtracted fluorescence divided by the standard deviation of pixel intensities in the control signal. Finally, positive or negative samples are determined based on whether the score is above (positive, +) or below (negative, −) 1.5, a threshold identified empirically.
HUDSON nuclease and viral inactivation were performed on viral seedstock as previously described with minor modifications to the temperatures and incubation times25. In short, 100 mM TCEP (Thermo Fisher Scientific) and 1 mM EDTA (Thermo Fisher Scientific) were added to non-extracted viral seedstock and incubated for 20 min at 50 °C, followed by 10 min at 95 °C. The resulting product was then used as input into the two-step SHERLOCK assay.
The improved HUDSON nuclease and viral inactivation protocol was performed as previously described, with minor modifications26. Briefly, 100 mM TCEP, 1 mM EDTA, and 0.8 U/µL murine RNase inhibitor were added to clinical samples in UTM, VTM, or human saliva (Lee Biosolutions). These samples were incubated for 5 min at 40 °C, followed by 5 min at 70 °C (or 5 min at 95 °C, if saliva). The resulting product was used in the single-step detection assay. In cases where synthetic RNA targets were used, rather than clinical samples (e.g., during reaction optimization), targets were added after the initial heating step (40 °C at 5 min). This is meant to recapitulate patient samples, as RNA release occurs after the initial heating step when the temperature is increased and viral particles lyse.
For optimization of nuclease inactivation using HUDSON, only the initial heating step was performed. The products were then mixed 1:1 with 400 mM RNaseAlert substrate v2 in nuclease-free water and incubated at room temperature for 30 min before imaging on a transilluminator or measuring reporter fluorescence on a Biotek Synergy H1 [at room temperature using a monochromator (excitation: 485 nm, emission: 520 nm) every 5 min for up to 30 min]. The specific HUDSON protocol parameters modified are indicated in the figure captions.
Data analysis and schematic generation
Conservation of SARS-CoV-2 sequences across our SHERLOCK assay was determined using publicly available genome sequences via GISAID. Analysis was based on an alignment of 5376 SARS-CoV-2 genomic sequences. Percent conservation was measured at each nucleotide within the RPA primer and Cas13-crRNA-binding sites and represents the percentage of genomes that have the consensus base at each nucleotide position.
As described above, fluorescence values are reported as background-subtracted, with the fluorescence value collected before reaction progression (i.e., the latest time at which no change in fluorescence is observed, usually time 0, 5, or 10 min) subtracted from the final fluorescence value (3 h, unless otherwise indicated).
Normalized fluorescence values are calculated using data aggregated from multiple experiments with at least one condition in common and for the specificity testing where all conditions were performed in the same experiment on a SpectraMax M2 (Molecular Devices). The maximal fluorescence value across all experiments is set to 1, with fluorescence values from the same experiment set as ratios of the maximal fluorescence value. Common conditions across experiments are set to the same normalized value, and that value is propagated to determine the normalized values within an experiment.
The Wilcoxon rank-sum test was conducted in MATLAB (MathWorks). Schematics shown in Figs. 1a and 3a were created using www.biorender.com. All other schematics were generated in Adobe Illustrator (v24.1.2). Data panels were primarily generated via Prism 8 (GraphPad), except Fig. 3e that was generated using Python (version 3.7.2), seaborn (version 0.10.1), and matplotlib (version 3.2.1)43,44.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.