Evaluation of two automated low-cost RNA extraction protocols for SARS-CoV-2 detection

Background Two automatable in-house protocols for high-troughput RNA extraction from nasopharyngeal swabs for SARS-CoV-2 detection have been evaluated. Methods One hundred forty one SARS-CoV-2 positive samples were collected during a period of 10-days. In-house protocols were based on extraction with magnetic beads and designed to be used with either the Opentrons OT-2 (OT-2in-house) liquid handling robot or the MagMAXTM Express-96 system (MMin-house). Both protocols were tested in parallel with a commercial kit that uses the MagMAXTM system (MMkit). Nucleic acid extraction efficiencies were calculated from a SARS-CoV-2 DNA positive control. Results No significant differences were found between both in-house protocols and the commercial kit in their performance to detect positive samples. The MMkit was the most efficient although the MMin-house presented, in average, lower Cts than the other two. In-house protocols allowed to save between 350€ and 400€ for every 96 extracted samples compared to the commercial kit. Conclusion The protocols described harness the use of easily available reagents and an open-source liquid handling system and are suitable for SARS-CoV-2 detection in high throughput facilities.


Introduction
The SARS-CoV-2 pandemic has called for the use of massive qPCR tests in order to detect positive cases and to trace contacts to stop community transmission. Prior to qPCR testing, clinical samples are usually subject to RNA extraction [1][2][3][4]. Given the number of samples tested every day, manual RNA extraction methods are unfeasible for most facilities and thus, automatic systems are widely used for this task [5][6][7]. As a drawback, automatic systems significantly increase the final costs, which can hinder massive testing in some areas. Moreover, due to the increased demand, stock shortage of extraction reagents has caused major delays in diagnostics.
In this work, two low-cost alternatives for automatic RNA extraction are described. The first one using the OT-2 system (Opentrons, New York, NY, USA), an open-source liquid handler robot capable of automating self-designed protocols, and the second using the rapid and easy-to-use nucleic acid extractor MagMAX TM Express-96 system (Thermo Fisher Scientific, Waltham, MA, USA). The later extracts up to 96 samples in 30 min, though it requires a previous manual dispensation of the reagents, magnetic beads and samples in 96-well plates adding 30 min. As an alternative, the OT-2 in-house protocol can process up to 48 samples in 104 min in a fully automated manner.

Sample collection
During ten days 141 consecutive SARS-CoV-2 positive nasopharyngeal swabs with viral transport medium (Deltalab, Barcelona, Spain) that had been stored at 4˚C were collected. Before processing, the samples were inactivated by mixing 500 μL of the viral medium and 500 μL of 4M guanidine Isothiocyanate (GTC) (Qiagen, Hilden, Germany) with 5 μg/mL carrier RNA, followed by heating the samples 2 min at 80˚C and a short vortex mixing.

Equipment and reagents
Automatic extraction of nucleic acids was done using two systems: the MagMAX TM Express-96 Deep Well Magnetic Particle Processor (King Fisher Instrument, Thermo Fisher Scientific, Waltham, MA, USA) and the open system OT-2 (Opentrons, New York, NY, USA) with a GEN1 magnetic module (Opentrons, New York, NY, USA) and an in-house protocol. Mag-MAX™ Express 96 plates and Deep Well plates (Thermo Fisher Scientific, Waltham, MA, USA) were used with the two systems.
Nucleic acid extraction was done with three methods: 1) Using the MagMAX TM with the commercial MagMAX CORE Nucleic Acid Purification kit (MM kit ) (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer instructions; 2) The OT-2 system with generic reagents (OT-2 in-house ) such as ethanol (Emsure1, Merck KGaA, Darmstadt, Germany), 2-Propanol (Emsure1, Merck KGaA, Darmstadt, Germany), Elution Buffer (Omega BIO-TEK, Norcross, GA, USA), Nuclease free water (Ambion TM , Thermo Fisher Scientific, Waltham, MA, USA) and magnetic beads (Mag-Bind1 TotalPure NGS, Omega Bio-Tek, Norcross, GA, USA); 3) MagMAX TM with the same protocol as the commercial kit but the reagents used in the OT-2 method (MM in-house ). In-house protocols are a modification of the procedure described by Hui He et al. [8]. Briefly, inactivated respiratory samples are mixed in a 1:1 proportion with isopropanol to a final volume of 500 μl for the OT-2 in-house and 1000 μL for the MM in-house , 40 μL of magnetic beads are added and the mixtures are incubated for 5 min at room temperature. Next, magnetic beads are pulled to one side of the tubes with a magnet and supernatant is discarded. Magnetic beads are then washed two times with 500 μL of freshly prepared ethanol 70%. After the second wash, the ethanol 70% is discarded and the magnetic beads are air dried at room temperature. Finally, the beads are resuspended in 100 μL of the elution buffer and separated with the magnet again to recover the eluted viral RNA (Table 1). 3. Put all the plates in the system and run the script MagMAX_CORE_KF-96.

Protocol design and validation
OT-2 in-house protocol: 1. Dispense in a Deep well plate 40μL of magnetic beads (Mag-Bind1 TotalPure NGS, Omega Bio-Tek, Norcross, GA, USA), 250μL of Isopropanol and 250μL of sample per well. Mix by pipetting five times and incubate 5 min at room temperature.
3. Collect the supernatant and discard.
7. Turn off the GEN1 magnetic module and add 100μL of Elution Buffer (Omega BIO-TEK, Norcross, GA, USA).
8. After 30 sec turn on the GEN1 magnetic module.
9. After 90 sec collect the supernatant and transfer to a 96-well microtiter plate.

The sample input was the maximum volume allowed by the automatic pipetting systems and the volumes of the other reagents and therefore was different for the three methods
OT-2 in-house protocol was written in Python according to the Opentrons instructions. The scripts have been deposited in the GitHub repository (https://github.com/HULPopentrons/ RNA_extraction_OT2opentrons).
To validate the performance of the in-house protocols for nucleic acid extraction, simulated samples were made with standard, low and very low viral loads using the DNA positive control TaqMan 2019-nCoV Control Kit v1 (Thermo Fisher Scientific, Waltham, MA, USA). The Control Kit comes at a concentration of 1 x 10 4 copies/μL. The standard viral load sample was prepared by mixing 10 μL of the positive control with 490 μL of viral transport medium and 500 μL of GTC. The low and very-low viral loads were prepared in the same manner but using ten-fold serial dilutions of the positive control. All mock samples were prepared in triplicate and processed in parallel with the OT-2 in-house , the MM kit and the MM in-house and then tested by qPCR with the TaqMan 2019-nCoV Assay Kit v1. Final concentration of the positive control in the simulated samples with the standard, low and very-low viral loads was 1 x 10 6 copies/mL, 1 x 10 5 copies/mL and 1 x 10 4 copies/mL respectively. Negative controls were included in all the runs.
Nucleic acid extraction efficiency was calculated by comparing the Ct values of the positive controls in the simulated samples (Ct ss ) with the Ct values of the positive controls prepared directly from the stock correcting the amounts to match the dilution factor and the amount of initial sample used in each protocol (Ct pc ) (R = 2 -ΔCt = 2 -(Ctss-Ctpc) ).

qPCR
Nucleic acid amplification of the SARS-CoV-2 viral RNA was done using the TaqMan 2019-nCoV Assay Kit v1 that targets the orf1ab, spike (S), nucleocapsid (N) and human RNa-seP genes and the TaqMan 2019-nCoV Control Kit v1 as positive control with the qPCR conditions recommended by the manufacturer (https://assets.thermofisher.com/TFS-Assets/LSG/ manuals/MAN0019096_TaqMan2019nCoVAssayKit_PI.pdf). All qPCR assays were performed in a CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). To reduce inter-assay variability, extracted nucleic acid samples were dispensed automatically in qPCR strips with another OT-2 module.

Statistical analyses
The three protocols were compared pairwise. McNemar's test was used to compare their performance in assigning samples as positive or negative. Ct values are not distributed normally, therefore the Wilcoxon Signed-Rank test was used for comparing the Ct values of each target.
For each target only the samples that were amplified by the three methods were considered. Median confidence intervals were calculated with a bootstrap method.

Efficiencies of the extraction protocols
Nucleic acid extraction efficiencies of the two in-house protocols were compared with the MM kit protocol by extracting the mock samples prepared with the SARS-CoV-2 positive control that simulate different viral loads. The MM kit and OT-2 in-house protocol successfully amplified the orf1ab, S and N targets in all mock samples with the standard viral load. For these samples, the mean extraction efficiency of all replicates and genes was 33.9% (SD: 13.8) for the MM kit , 19.6% (SD: 2.2) for the OT-2 in-house protocol and 16.0% (SD: 5.03) for the MM in-house .
In the low viral load mock samples, the MM kit failed to amplify the S target in all samples and the N target in one of the replicates while the OT-2 in-house and MM in-house detected all the genes. Finally, the MM kit and OT-2 in-house protocols failed to amplify the very-low viral load samples, except for one replicate in which the N gene could be amplified with the OT-2 in-house protocol but the MM in-house could detect the positive control in all samples, one of them with the three targets, one with them with the orf1ab and N targets and the last one with the orf1ab target (S1 Table).

Nucleic acid extraction from clinical respiratory samples
The 141 positive clinical samples collected were subjected to nucleic acid extraction in parallel with the MM kit , the OT-2 in-house , and the MM in-house methods and the eluted SARS-CoV-2 RNA was tested by qPCR. Positive/negative results and amplification cycle threshold values (Ct) for each target were registered. According to the manufacturer's instructions samples were considered positive when at least one of the three-targeted regions amplified with a Ct<40. Negative controls included in all the runs yielded in all cases negative PCR results.
Of the 141 samples, 123 were positive for SARS-CoV-2 by at least one extraction method while 18 were negative by all three methods. This was probably due to sample degradation during the collection period, because they had been stored for about 48 h at 4˚C [9]. By method, 114 samples tested positive using the MM kit , 111 with the OT-2 in-house and 118 with the MM inhouse . Pairwise comparisons found no significant differences in their performances for detecting SARS-CoV-2 (MM kit Vs OT-2 in-house , P = 0.5465; MM kit Vs MM in-house , P = 0.3865; OT-2 in-house Vs MM in-house , P = 0.0961). Eighteen samples were negative by some of the three methods. These had high Ct values, with median values of the orf1ab and N targets of 38.68 and 38.19 respectively (in those samples the S target was not amplified by any method). Pairwise linear regression and correlation analyses comparing all methods and targets showed R 2 values between 0.85 and 0.95 and Bland-Altman analyses showed the agreement to be uniform through the whole Ct range (S1 Fig). The three targets (orf1ab, N and S) were detected in 43% of the samples extracted with MM kit , 49% of those extracted with OT-2 in-house and 49% with MM in-house (Fig 1). In 34%, 39% and 30% the orf1ab and N targets were detected, and in 19%, 10% and 17% only the N target was amplified. Few samples were considered positive by the amplification of either the orf1ab target alone (6), the orf1ab + S (1), or the N + S (5) targets, and no sample had the S gene as the only positive marker. In fact, the S target in this assay had an evident lack of sensitivity and would be irrelevant to take diagnostic decisions. Samples with two or three targets detected were 87% (97 samples) with the OT-2 in-house protocol, 82% with the MM in-house (95 samples) and 79% with the MM kit (90 samples).
When considering paired samples, no significant differences were found in the Cts between the MM kit and the OT-2 in-house protocol in the Ct values for the orf1ab (P = 0.437), N (P = 0.686) and S (P = 0.794) targets (Fig 2). The MM in-house protocol did present significantly lower Cts in all targets compared to the MM kit (orf1ab; P< 0.00001, N; P< 0.00001, S; P = 0.00148) and the OT-2 in-house (orf1ab; P< 0.00001, N; P = 0.00008, S; P = 0.00252).
The median amplification cycle (CI:95%) of the orf1ab target using the MM kit , the OT-2 inhouse and the MM in-house were, respectively, 35.

Extraction costs and hands-on time
Nucleic acid extraction of 96 samples using the OT-2 in-house protocol had an overall cost of 107 € (37€ reagents and 70€ labware) with a hands-on time of 10 min and an extraction time of 3hours and a half (the script handles 48 samples in each run so it has to be completed two times). The MM in-house cost was 66€ (37€ reagents and 29€ labware) and the MM kit, 472€ (443 € reagents and 29€ labware). Both protocols had a hands-on time of 40 min and an extraction time of 30 min. Notably, although the OT-2 protocol for 96 samples was 40€ more expensive than the MM in-house , the required initial investment is significantly lower, being approximately 10,000€ for the OT-2 system (including modules and pipettes) and 50,000€ for the Mag-MAX TM system.

Discussion
In this work, two methods are presented for low-cost viral RNA extraction and detection of SARS-CoV-2 in clinical samples with a performance comparable to that of a commercial kit. Although in-house protocols extracted the DNA control less efficiently than the commercial kit, in clinical samples the overall sensitivity was not affected, probably because in-house protocols use larger sample starting volumes that compensate the lower efficiency. Setting up the OT-2 system for RNA extraction required intensive programming for a nonexperienced team but provided a cost-effective alternative capable of extracting 48 samples in a single run. The scripts used in this work were uploaded to the open access software repository GitHub, where many protocols for these and similar applications can be found. This should help other workers to avoid or reduce the programming steps. The protocol demands little hands-on time and uses easily available reagents. In addition, the equipment requires a lower investment compared to other extraction systems, making it suitable for middle to low resource facilities. The MagMAX TM Express-96 is a fast, semi-automatic equipment capable of extracting 96 samples per run. The MM kit provides reagents for inactivation, lysis, washing and elution and can be used for different matrices and samples types at a competitive cost. On the other hand, the MM in-house protocol takes advantage of the semi-open approach of this system by substituting the commercial solutions by other chemicals making it more cost effective. In both cases, preparing the deepwell plates with buffers and samples must be done manually, increasing the total extraction time and the manipulation error risk. The main drawback of the two in-house protocols comes when handling viscous samples, the presence of mucus, highly viscous polysaccharides, leukocytes, erythrocytes, hemoglobin, proteases and cell detritus with high amounts of cellular nucleic acids can preclude the RNA extraction or inhibit the PCR

PLOS ONE
Automated low-cost SARS-CoV-2 RNA extraction protocols reaction [10][11][12]. To partially overcome this problem, samples can be heated and vortexed or subjected to a centrifugation prior to extraction, though this would increase the hands-on and response times. When using the MM in-house , the eluted samples may have in some cases traces of the magnetic beads though this did not affect the performance of the RT-PCR reactions.
With the three systems negative controls were in all cases negative, indicating that crosscontamination is not an issue when working with primary samples in these systems. Nevertheless, same as with non-automated systems, precautions should be taken to keep pre-PCR and post-PCR operations separated. If an OT-2 module is used to set-up post-PCR reactions (eg. sequencing) it should not be used to extract nucleic acids from samples unless it is thoroughly decontaminated.

Study strength and limitations
The major strength of the work is that the methods were tested with real samples in a clinical microbiology laboratory. The comparison between methods was not systematic, and this is an important limitation. In fact, the design was intended to optimize the results for each system, and so the inputs were different. A systematic exploration of the input sample volumes, as well as other reagent volumes would be helpful and informative, but it was beyond the aim of this work that was to compare the performance of the systems in a clinical laboratory environment with real samples.

Conclusion
In summary, the two in-house nucleic acid extraction methods presented here are efficient for the high throughput diagnosis of SARS-CoV-2 at fraction of the costs of other commercial kits without losing sensitivity.
Supporting information S1 Table. Cts obtained with the simulated samples at three different concentrations (Ctss) and Cts of the diluted controls used to calculate the efficiency (Ctpc). (n.d. not detected). (TIF) S1 Fig. Bland-Altman plot showing the agreement between the three methods with three marker genes. The same set of samples was extracted with the three methods and analyzed with a commercial PCR targeting three marker genes. The horizontal axes show the average Cts, and the vertical axes show the difference between Cts for each sample with the two methods indicated. The horizontal continuous line marks the average difference, and the discontinuous lines indicate the 95% limits of agreement (average difference ± 1.96 standard deviation of the difference). (TIF) S1 Data. (XLSX)