Optimization of reaction condition of recombinase polymerase amplification to detect SARS-CoV-2 DNA and RNA using a statistical method.

Recombinase polymerase amplification (RPA) is an isothermal reaction that amplifies a target DNA sequence with a recombinase, a single-stranded DNA-binding protein (SSB), and a strand-displacing DNA polymerase. In this study, we optimized the reaction conditions of RPA to detect SARS-CoV-2 DNA and RNA using a statistical method to enhance the sensitivity. In vitro synthesized SARS-CoV-2 DNA and RNA were used as targets. After evaluating the concentration of each component, the uvsY, gp32, and ATP concentrations appeared to be rate-determining factors. In particular, the balance between the binding and dissociation of uvsX and DNA primer was precisely adjusted. Under the optimized condition, 60 copies of the target DNA were specifically detected. Detection of 60 copies of RNA was also achieved. Our results prove the fabrication flexibility of RPA reagents, leading to an expansion of the use of RPA in various fields.

Introduction

Recombinase polymerase amplification (RPA) exponentially amplifies a target nucleic acid sequence using two opposite primers at a constant temperature near 40 °C [[1], [2], [3]]. In RPA, recombinase binds to the primers, and the primers of the resulting complex bind to the homologous sequences of the DNA template. Then, strand-displacing DNA polymerase extends the primer, and single-stranded DNA-binding protein (SSB) binds to the unwound strand. RPA is more suitable than PCR for use in fields because it does not require a thermal cycler. Indeed, most papers on RPA published to date have been focused on detecting pathogenic organisms with amplicon-detecting technologies in consciousness of field use, such as a lateral flow assay [4], enzyme-linked oligonucleotide assay [5], and electrochemical assay [[6], [7], [8]].

Unlike PCR, the major limitation of RPA is that the RPA kit is sold by only two companies, Twist Dx, which is now owned by Abbott (San Diego, USA), and Jiangsu Qitian Gene Biotechnology (Ningbo, China). As a result, these RPA kits have solely been used in almost all the RPA-related studies. Such limitation makes the researchers have limited flexibility in exploring the effects of the concentration of each component on the reaction efficiency. In order to circumvent it, we previously prepared recombinant recombinase and SSB and used them to examine the effects of pH, temperature, and various additives on the efficiency of RPA [9]. In this study, we established the detection system of SARS-CoV-2 DNA and RNA and attempted to optimize the reaction condition, using one of the well-known statistical methods Taguchi method [10,11]. The results have shown that the sensitivity of RPA increased markedly by optimizing the concentration of each component.

Materials and methods

Preparation of standard DNA and RNA

Standard DNA and RNA corresponding to sequences 28,571–28,970 (System 1 in Fig. 1 A) and 28,171–28,470 (System 2 in Fig. 1B) of the SARS-CoV-2 gene deposited in GenBank NC_045512.2 were prepared as follows. PCR was carried out using the oligonucleotides listed in Tables S1 and S2 and Taq polymerase (Toyobo, Osaka, Japan) for 35 cycles of 30 s at 95 °C, 30 s at 55 °C, and 30 s at 72 °C. The amplified DNA was purified using a MagExtractor (Toyobo). In vitro transcription was carried out using RiboMAX Large Scale RNA Production System (Promega, Madison, WI) at 37 °C for 3 h. The synthesized RNA was purified by NICK Columns (GE Healthcare, Buckinghamshire, UK). The purified DNA and RNA concentrations were determined spectrophotometrically at A 260. The DNA and RNA were stored at −30 °C or −80 °C, respectively.

Fig. 1

Structure of SARS-CoV-2 and a target sequence for RPA amplification. Arrows indicate the sequences to which primers bind. (A) System 1. (B) System 2.

In System 1, the standard DNA was amplified by PCR using 400 nt_F and 400 nt_B as a pair of primers and the mixture of oligonucleotides 400nt-1, 400nt-2, and 400nt-3 as a template. The T7 promoter-bearing standard DNA was amplified by PCR using T7-400 nt_F and 400 nt_B as a pair of primers and the standard DNA as a template (Table S1). Standard RNA was synthesized by in vitro transcription using the T7 promoter-bearing standard DNA as a template. In System 2, the standard DNA was amplified by PCR using 300 nt_F and 300 nt_B as a pair of primes and the mixture of oligonucleotides 300nt-1 and 300nt-2 as a template. The T7 promoter-bearing standard DNA was amplified by PCR using T7-300 nt_F and 300 nt_B as a pair of primers and the standard DNA as a template (Table S2). Standard RNA was synthesized by in vitro transcription using the T7 promoter-bearing standard DNA as a template.

Materials

Recombinant uvsX, uvsY, and gp32 were expressed in Escherichia coli and purified from the cells as described previously [9]. The purified uvsX, uvsY, and gp32 preparations yielded a single band with molecular masses of 43, 22, and 34 kDa, respectively (Fig. S1). Bst DNA polymerase (large fragment) was purchased from New England BioLabs (Ipswich, MA), and creatine kinase was purchased from Roche (Mannheim, Germany). Recombinant thermostable quadruple variant (E286R/E302K/L345R/D524A) of moloney murine leukemia virus (MMLV) reverse transcriptase (RT) was expressed in Escherichia coli and purified from the cells as described previously [12].

RPA reaction and statistical analysis

The reaction mixture (30 μL) for RPA was designed and prepared according to Taguchi's L27 (Table S3) orthogonal array consisting of 13 factors. The reaction was performed in a 0.2 ml PCR tube at 41 °C in PCR Thermal Cycler Dice (Takarabio, Otsu, Japan). The amplified products were separated on 2.0% (w/v) agarose gels and stained with ethidium bromide (1 μg/ml). Each reaction condition for cDNA synthesis was scored as 1, 2, or 3 according to the intensity (no, faint, or clear, respectively) of amplified products. The data were analyzed by the following equation:where S/Nm is signal-to-noise ratio of and sm is the score of each reaction condition (m = 1, 2, -----, 27). The S/Nx,i of level i (= 1, 2, or 3) of factor x (= 1 to 13) was the total of nine out of 27 S/Nm values where the level of factor x is i. For example, the S/N1,1 and S/N2,1 were calculated as follows:S/N S/N

Accordingly, the lowest S/Nx,i value of the three S/Nx,i values (S/Nx,1, S/Nx,2, S/Nx,3) indicates that level i is the most appropriate. Variation (Vx; x = 1 to 13) and percentage contribution (Px) for factor x were calculated using the following equations:

Accordingly, high Vx and Px values indicate that the effect of the difference in the three levels in factor x is high on the reaction efficiency.

Results and discussion

Establishment of the RPA detection systems of SARS-CoV-2 DNA

For use as the assay in the optimization of RPA reaction condition, we established a detection system (System 1) of SARS-CoV-2 DNA (Fig. 1A). In System 1, nucleocapsid phosphoprotein gene was selected as a target according to the previous report [13]. We designed three forward and three reverse primers (Table S1), and selected one combination (1F+4 and 1R+8) that exhibited the best performance in sensitivity. The size of amplified product by the primer combination was 128 bp.

Round 1 of optimization of RPA reaction condition

We designed 13 factors and three concentrations (levels 1–3) for each factor (Table S4). Level 2 was set as the concentration used in the standard condition with which we previously examined the effects of pH, CH3COOK concentration, and temperature on the RPA reaction efficiency [9]. Levels 1 and 3 were set as 25–50% and 200–400%, respectively, of level 2. According to Taguchi's L27 orthogonal array (Table S3), the RPA reaction was carried out with 6 × 108 copies (2 × 107 copies/μL) of standard DNA and primers in System 1. The reaction products at 30, 45, and 60 min was analyzed by agarose gel electrophoresis. As an example, one of the results is shown in Fig. S2. Of the 27 reaction conditions, six (2, 3, 7, 17, 24, and 25) exhibited clear, four (5, 12, 13, and 18) exhibited faint, and the other 17 exhibited no 128-bp band corresponding to the amplified product. The signal-to-noise ratios for each reaction condition (S/Nm; m = 1 to 27), those for each level of each factor (S/Nx,i; x = 1 to 13, i = 1, 2, or 3), and variations (V) and percentage contributions (P) for each factor were calculated (Table S4). The results indicated that the Mg(OCOCH3)2 concentration exhibited the highest V (142) and P (39.7%) values with the optimal concentration of 7 mM (level 1).

It is known that the optimal concentration of Mg(OCOCH3)2 depends on primer and target sequences. Indeed, in the RPA kit sold by Twist Dx, Mg(OCOCH3)2 is not premixed but is added by users. High V and P values in our results suggested that the optimal range of Mg(OCOCH3)2 concentration is relatively narrow.

Rounds 2 and 3 of optimization of RPA reaction condition

The 13 factors and each three levels in Round 2 are shown in Table S5. Based on the results of Round 1 where the Mg(OCOCH3)2 concentrations were set as 7, 14, and 28 mM, for levels 1, 2, and 3, respectively, they were set as 5, 8, and 11 mM in Round 2. The concentrations of PEG35000, dNTPs, ATP, and primers were also altered. Tris-HCl (pH 8.2) was replaced with phosphocreatine. Twenty-seven RPA reactions were carried out with 6 × 108 copies (2 × 107 copies/μL) of standard DNA and primers in System 1. Results are shown in Table S5, indicating that the optimal uvsY, gp32, and ATP concentrations were not level 2 but level 1 (35 ng/μL) for uvsY, level 3 (400 ng/μL) for gp32, and level 1 (3 mM) for ATP concentrations. In addition, the uvsX, uvsY, gp32, and ATP concentrations exhibited relatively high V (21.1, 35.3, 130.4, and 28.6, respectively) and P values (4.2%, 6.9%, 25.7%, and 5.6%, respectively). These results suggested that these concentrations were rate-determining factors.

In the RPA process, the balance of the binding and dissociation between uvsX and DNA primer is important. In the presence of ATP, uvsX binds to DNA primer to form the nucleoprotein with the aid of uvsY. Upon hydrolysis of ATP, uvsX dissociates from DNA primer and is replaced by gp32. Thus, uvsX, uvsY, and ATP shift the balance to the binding, while gp32 shifts it to the dissociation. If this binding affinity is not high enough, the nucleoprotein cannot invade double-stranded DNA, thereby preventing DNA primer from binding to the target sequence. On the other hand, if the binding affinity is too high, uvsX remains occupied even after the elongation starts, preventing another nucleoprotein from binding to the target sequence and starting the elongation. Therefore, it was thought that the binding affinity of the reaction condition consisting of level 2 for all 13 factors’ concentrations was too high.

Based on the results of Round 2, we attempted to lower the binding affinity by increasing the concentration of gp32 and decreasing the concentrations of uvsY and ATP. Thirteen factors and each three levels in Round 3 were determined (Table S6). Twenty-seven RPA reactions were carried out with 6 × 104 copies (2 × 103 copies/μL) of standard DNA and primers using System 1. Results are shown in Table S6. Level 2 was optimal for the uvsX and gp32 concentrations. The uvsY concentration exhibited low V (12.6) and P (3.9%) values. These results suggested that the balance between the binding and dissociation of uvsX and DNA primer was adequately adjusted.

Performance of the optimized reaction condition

For use as the assay in the optimization of RPA reaction condition, we established two detection systems (Systems 1 and 2) of SARS-CoV-2 DNA and RNA (Fig. 1B). In System 2, ORF8 protein gene was used as a target because Centers for Disease Control and Prevention (CDC), USA reported the PCR primers targeting this region, and these primers are widely used in approved diagnostics of SARS-CoV-2 RNA. We designed nine forward and eight reverse primers (Table S2) and selected one combination (2F-15 and 2R-11) that exhibited the best performance in sensitivity. The size of amplified product by the primer combination was 99 bp.

RPA was carried out with 60–6 × 107 copies of standard DNA, and RT-RPA was carried out with 60–6 × 107 copies of standard RNA, both at 41 °C for 1 h. In the analysis of the RPA or RT-RPA products in the subsequent electrophoresis, the optimized conditions detected 60 copies of standard DNA (Fig. 2 A) or RNA (Fig. 2B).

Fig. 2

Comparison of the sensitivities of RPA and RT-RPA. (A) RPA. (B) RT-RPA. The reactions (30 μL) were carried out with 400 ng/μL uvsX, 40 ng/μL uvsY, 400 ng/μL gp32, 0.4 units/μL Bst DNA polymerase, 120 ng/μL creatine kinase, 2 mM DTT, 6% PEG35000, 3.5 mM ATP, 650 μM dNTPs, 3.5 mM ATP, 50 mM Tris-HCl buffer (pH 8.6), 40 mM CH3COOK, 20 mM phosphocreatine, 8 mM Mg(OCOCH3)2, 1 μM 2F-15 primer, and 1 μM 2R-11 primer in the absence (A) or presence (B) of 100 nM MMLV RT, at 41 °C for 30 min. Initial copies of standard DNA (A) and RNA (B): 0 (lane 1), 60 (lane 2), 600 (lane 3), 6 × 103 (lane 4), 6 × 105 (lane 5), and 6 × 107 (lane 6). Black arrow indicates the 99-bp target band.

Finally, we compared the sensitivities of RPA before and after optimization. Using System 2, RPA was carried out with 60–6 × 107 copies of standard DNA (Fig. S3). The condition after optimization detected 60 copies of standard DNA while that before optimization did not detect 600 copies. These results indicated that by optimizing the reaction conditions for three enzymes, 100 to 1000-fold higher sensitivity was achieved.

Generally, the performance of a nucleic acid amplification test depends largely on the performance of the enzymes involved. Indeed, DNA polymerases and RTs whose activity and/or stability have been improved by genetic engineering technique are currently used in PCR and RT-PCR [14,15]. However, such improvement has not been done for recombinase and SSB. It is of note that the optimal concentrations of uvsX, uvsY, and gp32 in the RPA reaction solution are in the range of 1–10 μM, which is 1000-fold higher than reverse transcriptase and thermostable DNA polymerase in cDNA synthesis and PCR. Such high protein concentration makes RPA reagents less flexible in fabrication. To solve this problem, increase in activity and/or binding ability of recombinase and SSB is required. On the other hand, considering the filed use of RPA, it is anticipated that storage of the reagents at room temperature is possible. To address this issue, use of thermostable recombinase and SSB from thermophilic organisms might be useful.

In PCR, various additives that increase the reaction efficiency have been reported: bovine serum albumin, trehalose, sorbitol, glycerol, Triton X-100, and Tween 20 stabilize an enzyme, and dimethyl sulfoxide, formamide, and ammonium sulfate increase specificity [16,17]. Helicase increases specificity by decreasing non-specific binding [18]. Spermidine suppresses reaction inhibition problems encountered while analyzing clinical stool samples [19,20]. In RPA, little is known for such additives except for the recent report that betaine increases specificity [21]. Our results in this study might make the evaluation of the effects of various additives on the RPA reaction efficiency easier.

Insights into the effect of the balance between the binding and dissociation of uvsX and DNA primer on the RPA reaction efficiency

As described above, Round 2 revealed that the reaction efficiency of RPA depends on the balance of the binding and dissociation between uvsX and DNA primer. To further understand this issue, we examined the effects of the concentrations of uvsX, uvsY, gp32, and ATP on RPA efficiency. The optimized condition obtained by Round 4 were used as standard conditions. Fig. 3 shows the analysis of the RPA products at 30 min using agarose gel electrophoresis. An amplified DNA band was observed at 400–4000 ng/μL uvsX, 40–400 ng/μL uvsY, 400 ng/μL gp32, and 0.35–3.5 mM ATP, while it was not observed at 40 and 120 ng/uL uvsX, 4 and 12 ng/μL uvsY, 40, 120, 1200, and 4000 ng/μL gp32, or 10 and 35 mM ATP. These results indicated that the optimal concentration of gp32 is narrower than that of uvsX, uvsY, and ATP, suggesting that the gp32 concentration is critical for the balance of the binding and dissociation between uvsX and DNA primer.

Fig. 3

Effects of the concentrations of uvsX, uvsY, gp32, and ATP on the reaction efficiency of RPA. (A–D) The reactions (30 μL) were carried out with 40–4000 (A) or 400 (B–D) ng/μL uvsX, 4–400 (B) or 40 (A, C, D) ng/μL uvsY, 40–4000 (C) or 400 (A, B, D) ng/μL gp32, 0.4 units/μL Bst DNA polymerase, 120 ng/μL creatine kinase, 2 mM DTT, 6% PEG35000, 0.35–35 (D) or 3.5 (A–C) mM ATP, 650 μM dNTPs, 50 mM Tris-HCl buffer (pH 8.6), 40 mM CH3COOK, 20 mM phosphocreatine, 8 mM Mg(OCOCH3)2, 1 μM 2F-15 primer, and 1 μM 2R-11 primer at 41 °C for 30 min. Initial copies of standard DNA was 6000. The arrow indicates the 99-bp target band.

It is known that uvsX, uvsY, and gp32 form a ternary complex with a single-stranded DNA (ssDNA) [22]. Gajewski et al. performed the crystal structural analysis of the uvsY-ssDNA complex and showed that uvsX exists as a heptamer [23]. They also provided a model showing that uvsY promotes a helical ssDNA conformation that disfavors the binding of gp32 and initiates the assembly of the ssDNA–uvsX filament [23]. We presume that this model might be applicable to the mechanism of RPA reaction.

In conclusion, the sensitivity of RPA and RT-RPA for SARS-CoV-2 DNA and RNA increased by optimizing the concentration of each component using a statistical method. Our results pave the way for use of RPA in various fabrications.

Notes

The authors declare no competing financial interest.