key: cord-0908417-xza6erwf
authors: Cherkaoui, Dounia; Huang, Da; Miller, Benjamin S.; Turbé, Valérian; McKendry, Rachel A.
title: Harnessing recombinase polymerase amplification for rapid multi-gene detection of SARS-CoV-2 in resource-limited settings
date: 2021-05-14
journal: Biosens Bioelectron
DOI: 10.1016/j.bios.2021.113328
sha: b2a0398ad0fc4cb8c104404a0c8f2ea0e73c15e7
doc_id: 908417
cord_uid: xza6erwf

The COVID-19 pandemic is challenging diagnostic testing capacity worldwide. The mass testing needed to limit the spread of the virus requires new molecular diagnostic tests to dramatically widen access at the point-of-care in resource-limited settings. Isothermal molecular assays have emerged as a promising technology, given the faster turn-around time and minimal equipment compared to gold standard laboratory PCR methods. However, unlike PCR, they do not typically target multiple SARS-CoV-2 genes, risking sensitivity and specificity. Moreover, they often require multiple steps thus adding complexity and delays. Here we develop a multiplexed, 1-2 step, fast (20-30 minutes) SARS-CoV-2 molecular test using reverse transcription recombinase polymerase amplification to simultaneously detect two conserved targets - the E and RdRP genes. The agile multi-gene platform offers two complementary detection methods: real-time fluorescence or dipstick. The analytical sensitivity of the fluorescence test was 9.5 (95% CI: 7.0-18) RNA copies per reaction for the E gene and 17 (95% CI: 11-93) RNA copies per reaction for the RdRP gene. The analytical sensitivity for the dipstick was 130 (95% CI: 82-500) RNA copies per reaction. High specificity was found against common seasonal coronaviruses, SARS-CoV and MERS-CoV model samples. The dipstick readout demonstrated potential for point-of-care testing in decentralised settings, with minimal or equipment-free incubation methods and a user-friendly prototype smartphone application. This rapid, simple, ultrasensitive and multiplexed molecular test offers valuable advantages over gold standard tests and in future could be configurated to detect emerging variants of concern.

antigen and molecular biomarkers. Antigen testing, based on the detection of viral proteins, 92 has some advantages over PCR, as these tests are often low-cost, fast and can be performed 93 outside the laboratory (Guglielmi, 2020). However, their sensitivity is often significantly 94 lower than nucleic acid amplification-based tests, missing up to 60% of PCR-positive 95 asymptomatic patients (University of Liverpool, 2020). of-care application. By taking advantage of their processing speeds, display, storage capacity, 142 their high resolution camera and connectivity, smartphones are useful devices to store 143 information relative to the test and the patient, analyse test results with an enhanced readout 144 compared to the naked eye, to communicate the result to a local hospital and send alerts in 145 case a new outbreak or cluster is detected (Brangel et al., 2018; Budd et al., 2020) . 146

Here we report the development of a rapid molecular diagnostic for the detection of SARS-148

CoV-2 by RT-RPA (Fig. 1a ) simultaneously detecting two gene targets (Fig. 1b) . We aimed 149 to design two alternative readouts, which are both multiplexed: real-time fluorescence (Fig.  150 1c) and dipsticks (Fig. 1d) . Offering two detection methods makes the assay more accessible 151 to different settings, depending on their resources. To the best of our knowledge, this is the 152 first one-pot multiplexed RPA-based assay for SARS-CoV-2. We also explore the use of 153 low-cost handwarmers to achieve the required temperature, and also a smartphone app to 154 capture and interpret test results. 155 156 Fig. 1 Schematic representation of the rapid and multiplex RT-RPA assay with real-cDNA template and 12.2 µL of nuclease-free water. Finally, 2.5 µL of magnesium acetate (at 191 concentration 280 mM) was added to start the reaction. Three cDNA concentrations (50, 500, 192 5000 copies) were tried along with a non-template control (NTC). The reactions were 193 incubated at 39˚C for 30 minutes and real-time fluorescence was recorded using a microplate 194 reader (excitation wavelength 495 nm and emission 520 nm). The screen was done in 195 technical replicates (N=2). Background correction was done to remove potential variation 196 due to initial mixing and normalised the data to compare relative fluorescence increase. The 197 measurement at 60 seconds was used to set the fluorescence to zero, as initial mixing of 198 reactions can lead to variations in fluorescence. This method of background subtraction 199 allowed to set all first measurement values to zero, remove potential variation due to initial 200 mixing and enabled to compare relative fluorescence increase. Then, the average values of 201 duplicates were plotted on GraphPad Prism along with error bars, corresponding to the 202 standard deviation. The fluorescence threshold value for the RPA screen of the four genes 203 with cDNA was set to 25,000. This threshold value was calculated by averaging fluorescence 204 signals from several NTC reactions and adding 3 times the associated standard deviation. The 205 average time to threshold, defined as the time corresponding to the intersection of the 206 amplification curve with the threshold value, was determined for each gene. 207 208

The plasmid cDNA encoding for the E and RdRP genes were digested using a pair of 210 restriction sites of the plasmid. Double digestion allowed to isolate the sequence of interest 211 and get linear DNA. The product of digestion was run on a 1% agarose gel with a DNA 212 ladder. The band of interest was excised from the gel and the DNA was purified. To generate 213 positive-sense RNA transcripts, a T7 promoter sequence was added via PCR amplification done with 2.5 hours incubation, with several rounds of DNase I treatment to remove the DNA 217 template, and the RNA was purified. The RNA was tested by PCR using the RPA primers 218 (also suitable for PCR) to check for traces of DNA impurities ( Supplementary Fig. 2c ). The 219 concentration of the RNA transcripts was measured using the Qubit, then the RNA was 220 diluted in DEPC-treated water and stored at -80˚C with RNase inhibitor. A dilution series 221 was used to measure the analytical sensitivity of the molecular test. Table 1 ) for duplex 237 detection on the dipsticks, which incorporate carbon nanoparticles conjugated to neutravidin. 238

The E gene primers were modified with biotin and digoxigenin for detection on test line (1), 239 whereas the RdRP gene primers were modified with FAM and biotin for detection on test line 240

(2). To eliminate non-specific binding due to dimers forming, the assay was tested without 241 any template (negative controls) with modified primers at concentration 10 µM, 2 µM, 1 µM 242 and 0. to strengthen the specificity of the assay. 266

The analytical sensitivity of the RT-RPA with real-time fluorescence readout was done using 267 these thresholds. Repeats were run five times for a range of RNA inputs: 1, 2.5, 5, 7.5, 10, 268 10 2 and 10 3 (only for the RdRP gene). The fraction of positive reactions (reactions which 269 reached the threshold in less than 20 minutes) was calculated separately for both genes and 270 probit analysis was done on Matlab (R2020b). 271

The EC 95 was calculated from the probit analysis with its 95% CI. The EC 95 was defined as 272 the analytical sensitivity of the test. The "CovidApp" smartphone application was developed in Android Studio using java 284 libraries. Screenshots were taken from the emulator using a Galaxy Nexus API 28. The 285 complete code for the Android application is available on request. 286

The application opens onto a homepage where the users can choose between three activities 287 "Test", "Alerts" or "Map outbreak". The "Test" activity includes first recording of patient 288 information (patient ID, date of birth, GPS and symptoms). The GPS coordinates are 289 J o u r n a l P r e -p r o o f captured in real-time by clicking the button and time and date are also automatically 290 captured. Then, the user can click on the "Take Test Picture" button to get access to the 291 smartphone camera and take a photograph of the lateral flow test. Manual cropping is 292 required to crop onto the result area of the test (where the lines are). Another activity enables 293 image analysis of the cropped image for enhanced visualisation of the test lines and plotting 294 of the test line intensity. Finally, the user can select between the options "three lines", "two 295 lines", "one line", "no line" which records the test result as "positive", "presumptive 296 positive", "negative" or "invalid". If the user test result is "positive" or "presumptive 297 positive", the user is taken to the "Contacts" page when clicking on the "Next step" button. 298

This will enable to record the contacts of the positive case so then can be later reached by the 299 local contact tracing system. Finally, the activity "Map outbreak" opens to visualise the 300 location of the tested case on the map. In the "Alerts" activity, the information of the patient, 301 with the test result, can be seen. 302 reading frame 1a/b (Orf1ab). The RPA assay design was optimised for amplicon size of ~200 319 bp and long primers of ~30 bp. BLAST analysis indicated that these pairs and probes 320 specifically detect SARS-CoV-2 (100% identity). In addition, primers and probes sequences 321 were also screened through BLAST against seasonal coronaviruses, SARS-CoV and MERS-322

CoV which revealed low identity score and high E value. 323

A preliminary gene screening aimed to identify the best two primers/probe sets among these 324 four targets, able to achieve rapid and sensitive detection of SARS-CoV-2 in a real-time RPA 325 assay. The gene screening was conducted with cDNA controls rapidly made available by 326 suppliers (Supplementary Fig. 3a) . A single fluorescence threshold was used to compare the 327 four targets. All reactions using template, except one (50 copies for the N gene), showed 328 successful amplification of 50, 500 and 5,000 copies with fluorescent signals reaching the 329 threshold in less than 30 minutes ( Supplementary Fig. 3b) . Then, the average time to 330 threshold was determined for each gene and it was used to compare them (Supplementary 331 Fig. 3c ). The two genes with the shortest average time to threshold with 50 copies of cDNA 332

were the E gene, in 14 minutes, and the RdRP gene, in 19 minutes. The Orf1ab gene was 333 slightly slower than RdRP gene, while the N gene showed particularly low sensitivity in the 334 RPA protocol and did not reach the threshold with 50 copies. Eventually, the E and RdRP 335 genes were selected and multiplexed to make an in-house duplex RT-RPA protocol to detect

The RT-RPA assay was developed with two complementary detection systems (Fig. 1a) . The dipstick-based platform was developed to be as low-cost and minimalist as possible. The 357 primer sequences used were the same as for the real-time fluorescence readout above, but 358 these primers were modified with small molecules to mediate capture of the amplicons on the 359 test lines of the dipstick (Supplementary Table 1 ). Optimisation of the primer concentration 360 was needed to eliminate non-specific binding on the test lines ( Supplementary Fig. 5 ) 361 attributed to binding of dimerised primers when used in excess (> 1 µM). After the 362 amplification was performed, detection of the two amplicons was possible on two distinct test that the test had worked properly (Fig. 1d) . 365 366

The analytical sensitivity was measured for the real-time RT-RPA assay and defined as the 368 concentration of analyte, here synthetic SARS-CoV-2 viral RNA copies per reaction, that can 369 be detected ≥ 95% of the time (< 5% false negative rate). 370

To determine the analytical sensitivity of the RT-RPA fluorescence readout two thresholds 371 were calculated, to account for the different background fluorescence of the FAM and HEX 372 fluorophores (see Material and methods section). The resulting thresholds were 112 for the E 373 gene (FAM) and 13 for the RdRP gene (HEX). RT-RPA reactions were run for different 374 RNA inputs ranging from 1 copy to 10 5 copies and real-time fluorescence was recorded. The 375 time to threshold was determined for reactions reaching threshold in 20 minutes of 376 amplification (Fig. 2a) . The amplification time was fixed at 20 minutes, as the assay was able 377 to detect as little as 1 RNA. To measure the analytical sensitivity of both genes, we calculated 378 the fraction positive to find and plot the EC 95 (see "Methods" section). The analytical 379 sensitivity was 9.5 RNA copies per reaction (95% CI: 7.0-18) for the E gene and 17 RNA 380 copies per reaction (95% CI: 11-93) for the RdRP gene (Fig. 2b) . 381

The specificity of the RT-RPA assay was tested with model samples against common 382 seasonal coronaviruses, namely HCoV-NL63, HCoV-OC63 and HCoV-229E, as their 383 symptoms could be easily confused with COVID-19, and we also tested cross-reactivity with 

The analytical sensitivity of the dipstick detection method was approximated by running a 396 range of RNA inputs, from 1 to 10 5 copies. Six replicates were performed (Supplementary 397 Fig. 6 ) of which one representative dipstick per RNA concentration is shown in Fig. 3a . The 398 test line intensity analysis was used to quantify test line intensity. Single-copy detection was 399 possible for 2/6 repeats (33%), giving in a positive result, defined as both test lines visible by 400 eye or with image analysis. The probit analysis was performed to determine the analytical 401 sensitivity of the assay using the fraction positive (Fig. 3b) . The analytical sensitivity of the 402 dipstick method was 130 (95% CI: 82-500) RNA copies per reaction. 403

The specificity of the dipstick detection method was assessed against the common seasonal 404 coronaviruses, SARS-CoV and MERS-CoV (Fig. 3c) . The dipstick showed high specificity 405 for only SARS-CoV-2 viral RNA and no cross-reactivity was seen with the other 406 coronaviruses. 407 The tests could be read visually by eye. In addition, we developed a smartphone application 416 as a prototype towards a connected-diagnostic dipstick test. The architecture and screenshots 417 of the prototype application are presented in Fig. 4a . The smartphone application proposed 418 allowed input and storage of patient's information, symptoms, capture of geo-location, test 419 lines intensity analysis of the dipstick and a record of the test results. If the test is "Positive" 420 or "Presumptive Positive" the user could insert the names of close contacts for contact tracing 421 purposes. The application also included geographic visualisation of the tested patients to map 422 'hotspots'. 423 424

The major advantage of RPA, compared to other approaches such as PCR and LAMP, is its 425 isothermal amplification at ~37˚C. We investigated the potential of different incubation 426 methods which could be more suitable for point-of-care settings. RT-RPA was performed to 427 detect 100 copies of RNA using four incubation approaches: an incubator, a water bath, a 428 disposable hand warmer bag and simply holding the tube in our hands. Incubators and water 429 baths are often found in well-equipped laboratories, but we also tried using a low-cost hand 430 warmer bag (based on an exothermic reaction shown to deliver a constant temperature of 431 ~36-37˚C for several hours (Wang, 2010) ) and holding the tube in one hand (using body 432 temperature ~37˚C) to show inexpensive and equipment-free alternatives. The results are 433 shown in Fig. 4b . While amplification in the incubator seemed to show the best results with 434 two test lines visible on the dipstick, two test lines were also visible for the reaction incubated 435 in the water bath, although slightly fainter. The reactions incubated on a hand warmer bag 436 and handheld appeared less sensitive, showing only a signal on test line (1). However, we 437 proved that very simple methods could be successfully used to amplify SARS-CoV-2 RNA 438 via RT-RPA and visual dipstick detection. 439 440 We investigated further the hand warmer bag as an affordable incubation method for RPA 441 reactions. We recorded the temperature on the bag surface and in the solution contained in 442 the PCR tube ( Supplementary Fig. 7) . We observed that the temperature on the bag surface 443 falls within the RPA range (grey shaded area) after 15 minutes of air-activation and remained 444 in the right range for hours (at least 2 hours). Moreover, we showed the solution temperature 445 inside the tube (incubated on the flat hand warmer bag) reached the RPA temperature range. 446

To assess the cost-effectiveness of the assay, we estimated the cost of the reagents for the 447 RT-RPA assay (for both readouts) and compared them to two commercial kits for qRT-PCR 448 protocols (Supplementary Table S2 ). The RPA reagents cost between ~£4-5.7 which from 449 estimation was half the price of qRT-PCR reagents. 450 451 Finally, preliminary analysis was performed to assess the potential of the dipstick test to be 452 compatible with mock clinical samples, using human saliva with spiked RNA transcripts to 453 mimic mouth swabs (Fig. 4c) . Saliva is an easy specimen for self-collection that has been 454 FDA-approved for molecular testing of COVID-19 (U.S. Food and Drug Administration, 455 2020b). The E gene was clearly detectable on test line (1) with ≥ 1 RNA copy per reaction, 456 and a faint signal was seen on test line (2) for the RdRP gene with 1 and 100 RNA copies per 457 reaction. Two strong test lines were visible for 10 5 copies per reaction. Therefore, the 458 findings of this small study suggest that conducting the assay in saliva compared to buffer did 459 not have a substantial impact on the assay sensitivity. We also showed that the assay was lower cost than qRT-PCR protocols, since it does not 496 require a sophisticated thermocycler and reagents cost less than some common qRT-PCR 497 commercial kits. 498

The amplification time for the RT-RPA assay was set to run for 20 minutes since it was 500 sufficient to achieve single-copy detection of the E gene with real-time fluorescence and 501 visual dipstick readouts, showing the ultrasensitive potential of the test. High sensitivity is 502 necessary to detect viral loads that are clinically relevant for COVID-19. The World Health 503

Organisation considers as acceptable an analytical sensitivity for confirmation of acute 504 SARS-CoV-2 infection when equivalent to 10 3 genomic copies/mL (~50 copies per reaction) 505 (World Health Organization, 2020a). 506

The analytical sensitivities for the E and RdRP genes comparable with those reported by 507

Charité-Berlin for its qRT-PCR assay, which were 3.9 copy per reaction (95% CI: 2.8-9.8) 508

for the E gene and 3.6 copy per reaction (95% CI: 2.7-11.2) for the RdRP gene; in 509 comparison to 9.5 RNA copies per reaction (95% CI: 7.0-18) for the E gene and 17 RNA 510 copies per reaction (95% CI: 11-93) for the RdRP gene reported herein. Notably, the 511 difference for the E gene is non-significant, with a 95% CI overlapping our reported mean. 512 some of the inherent delays associated with shipping samples to centralised laboratories for 514 gold standard tests and waiting for test results. 515

Our RT-RPA assay was shown to be highly specific to SARS-CoV-2, with no observed 516 cross-reactivity with the closely related coronaviruses tested, such as SARS-CoV, MERS-517

CoV, HCoV-NL63, HCoV-OC43 and HCoV-229E. This high specificity was demonstrated 518 for both detection methods and reduces the risk of false positives with closely related viruses. 519 520

The prototype smartphone application was proposed as a powerful tool for data capture, 521 analysis and visualisation when testing in decentralised settings. Smartphones are widely 522 accessible, easy-to-use and can act as a substitute to sophisticated laboratory equipment as 523 they integrate a high-resolution camera, large data storage space, real-time location and 524 connectivity. 525

Moreover, the use of inexpensive methods for incubation at ~37˚C of the RT-RPA reaction 526 for detection on dipsticks, especially with a hand warmer bag, emphasises the simplicity of 527 the assay for resource limited settings. 528 529

To close, we have developed an ultrasensitive and specific multi-gene diagnostic for SARS-531

CoV-2 viral RNA using isothermal RPA technology, and proposed two different detection 532 methods, both showing high accuracy. While real-time fluorescence detection developed here 533 offers more sensitivity and faster results (10 minutes faster than dipstick method), the 534 proposed detection on dipsticks appeared as the preferred method for decentralised testing. 535

We showed this method has the potential to meet the ASSURED and REASSURED criteria; 536 it is affordable, rapid, has high analytical sensitivity and specificity, it is user-friendly and can 537 be performed with minimal equipment. We also proposed the addition of real-time 538 connectivity through a smartphone application and the potential use of saliva as a non-539 invasive specimen. Having an alternative to qRT-PCR that has comparable analytical 540 performance, but with a shorter time to result, using different supply chains, requiring less 541 equipment and non-extensive laboratory experience, could help to alleviate the pressure on 542 healthcare systems and curb the COVID-19 pandemic worldwide. 

How ID NOW Tackles COVID-19

Streamlined inactivation, amplification, and Cas13-based 593 detection of SARS-CoV-2

Rapid detection of SARS-CoV-2 by low volume real-time 597 single tube reverse transcription recombinase polymerase amplification using an exo 598 probe with an internally linked quencher (exo-IQ)

A Serological Point-of-Care Test for the Detection of IgG Antibodies against

Ebola Virus in Human Survivors

Beyond the lateral flow assay: A review of paper-based microfluidics

2019-Novel Coronavirus

Real-time rRT-PCR Panel Primers and Probes

Rapid electrochemical detection of coronavirus

Development of Point-of-Care Biosensors for COVID-19

An integrated 627 paper-based sample-to-answer biosensor for nucleic acid testing at the point of care

Suitcase Lab for Rapid Detection of SARS-CoV-2 Based on

Testing is our first line of defence

Evidence supporting transmission of 647 severe acute respiratory syndrome coronavirus 2 while presymptomatic or 648 asymptomatic

The species Severe acute respiratory syndrome-related 652 coronavirus: classifying 2019-nCoV and naming it SARS-CoV-2

Syndrome Coronavirus 2 (SARS-CoV-2)

Diagnostics for the 669 developing world

Spin-enhanced nanodiamond biosensing for ultrasensitive diagnostics

Loop-mediated isothermal amplification (LAMP): a rapid, 675 accurate, and cost-effective diagnostic method for infectious diseases

Specific primers and 678 probes for detection of 2019 novel coronavirus

Investigation of novel SARS-COV-2 variant Variant of 690 Concern 202012/01

An enhanced isothermal amplification assay for viral detection

A fully 697 integrated paperfluidic molecular diagnostic chip for the extraction, amplification, and 698 detection of nucleic acids from clinical samples

COVID-19 spurs wave of innovative diagnostics

Electrochemical sensors for rapid diagnosis of 703 pathogens in real time

Coronavirus (COVID-19) Update: FDA

Authorizes First COVID-19 Test for Self-Testing at Home

Coronavirus (COVID-19) Update: FDA

Authorizes First Diagnostic Test Using At-Home Collection of Saliva Specimens. announcements/coronavirus-covid-19-update-fda-authorizes-first-diagnostic-test-using-713 home-collection-saliva

Liverpool COVID-19 Community Testing Pilot

Considerations for diagnostic COVID-19 tests

What's inside disposable hand warmers?

Laboratory testing for coronavirus disease 2019 (COVID-19) in suspected 724 human cases

Taking Mobile Health Connected Infectious Disease 729

Diagnostics to the Field

WHO Director-General's opening remarks at the media 741 briefing on COVID-19 -13

WHO Director-General's remarks at the media briefing 745 on 2019-nCoV on 11

WHO Coronavirus Disease (COVID-19) Dashboard

A 751 new coronavirus associated with human respiratory disease in China

Single-copy sensitive, field-deployable, and simultaneous dual-754 gene detection of SARS-CoV-2 RNA via modified RT-RPA

Transcription Recombinase-Aided Amplification Assay for Rapid Detection of the 2019

Novel Coronavirus (SARS-CoV-2)