All primers, FQ reporter (quenched fluorescent DNA reporter FAM-TTATT-BHQ1) and LF reporter (lateral flow strip reporter, FAM-TTATT-Biotin, FAM-T₂₀-Biotin) were synthesized by Sangon Biotech (Shanghai, China). The detailed sequences are listed in Supplementary Table 1. EnGen^ⓇLbaCas12a and NEBuffer 3.1 (10×) (100 mM NaCl, 50 mM Tris-HCl, 10 mM MgCl₂, 100 μg/ml) were purchased from New England Biolabs (MA, USA). The TwistAmp^ⓇBasic kit for RPA and portable real-time fluorometer TwistDX T8 (TwistDX Co. UK) were purchased from TwistDx (Cambridge, United Kingdom). The lateral flow strips (Cat. No. JY0301) and fast nucleic acid release regents (Cat. No. BT0068) were from Tiosbio Biotechnology Co, Ltd. (Beijing, China). Plant DNA extraction kit with magnetic beads (DP342), RNase inhibitor (NG209), and DNase/RNase-free ddH₂O were purchased from Tiangen (Beijing, China). Dithiothreitol (DTT) (Cat. No. 43816) was purchased from Sigma Aldrich (St. Louis, MO, USA). The Qubit^Ⓡ2.0 Fluorometer and the Qubit^ⓇdsDNA HS Assay kit were purchased from Invitrogen (Life technologies, Carlsbad, CA). PDMS prepolymer was purchased from Dow Corning (Midland, USA). The fluorescence quantifications were measured with a Roche LightCycler 480 (Roche, USA).

The hyphae or spores of cultured fungi were collected into a 1.5 ml tube, ground with quartz sand, and the genomic DNA was extracted using a plant DNA extraction kit. For plant materials, about 500 mg samples were ground into powder with a mortar and pestle after adding liquid nitrogen, transferred into a 1.5 ml tube, and the genomic DNA extraction was performed according to the manufacturer's instructions. The absorbed DNA on the magnetic beads was eluted using appropriate autoclaved distilled water, and stored at −20°C. The concentration of DNA was quantified using the Qubit^Ⓡ2.0 Fluorometer and the Qubit^ⓇdsDNA HS Assay kit following the manufacturer's instructions.

For the crude DNA extraction, fungal hyphae or plant tissues in a 2 ml Eppendorf tube were crushed with quartz sands and plastic pestle, then 100 μl of fast nucleic acid release reagents was added, and mixed by pipetting up and down 20 times. After 2 min of standing at room temperature, the supernatant was directly used as the template for RPA pre-amplification.

The primers for RPA reaction were designed according to the ITS sequence of L. maculans (GenBank No.: MG569595.1), L. biglobosa “brassicae” (GenBank No.: MH861367.1), and L. biglobosa “Canadensis” (GenBank No. FJ172238.1) using DNAMAN 8.0. The ITS genes of L. maculans and L. biglobosa were searched against nr/nt NCBI database in order to select the most conserved and specific molecular markers to discriminate L. maculans from L. biglobosa and other sequenced fungi. The primers were synthesized by Sangon Biotech (Shanghai, China).

A master mix containing 29.5 μl supplied rehydration buffer, 2.4 μl forward primer (10 μM), 2.4 μl reverse primer (10 μM), and 8.6 μl nuclease free water was added to an RPA TwistAmp^TMbasic pellet to dissolve the enzyme, followed by the addition of 2.5 μl magnesium acetate (280 mM) (Supplementary Table 2). After a brief vortex and spin down the reaction mixture, 10 μl of the reconstituted RPA reaction was added to new PCR tubes, then 1 μl purified DNA was added to each reaction and mixed by carefully pipetting up and down. One pellet could yield approximately four individual RPA reactions. The reactions were incubated at 37°C for 15–30 min.

The crRNA for the CRISPR detection system was designed according to the amplified ITS gene from L. maculans, and aligned to determine the highly conserved regions. The crRNAs for the CRISPR detection system was designed specifically for the amplification region and synthesized commercially by Sangon Biotech.

Genomic DNA extracted from fungi hyphae or diseased plant tissue was used as input for RPA reaction. The system of Cas12a-mediated fluorescent assay contained 0.125 μM of Cas12a, 0.25 μM of crRNA, 0.1 μM of FQ reporter, 10 U of recombinant RNase inhibitor, 2.5 mM DTT, 1×NEBuffer 3.1, and 1 μl of RPA pre-amplification buffer in 20 μl reaction volume (Supplementary Table 3). The reaction was performed at 37°C in a TwistDX T8 or LightCycler 480 instrument for 30 min. The fluorescence intensity was detected with λ_Ex/Em of 488/520 nm.

The Cas12a-mediated lateral flow assay was performed in a total of 20 μl reaction mixture containing 0.125 μM Cas12a, 0.25 μM crRNA, 0.5 μM LF reporter, 10 U recombinant RNase inhibitor, 2.5 mM DTT, 1×NEBuffer 3.1, and 2μl of RPA pre-amplification buffer (Supplementary Table 4). LF reporter was labeled with carboxyfluorescein (6-FAM) at the 5' and biotin at the 3' end of ssDNA, respectively. The 20 μl reaction mixture was incubated at 37°C with a portable metal incubator for 20 min, followed by the addition of 80 μl of ddH₂O. Then a lateral flow strip was put into this solution for about 3 min. If the sample is positive, both the C line and the T line or only the T line (strong positive sample) will appear, while only the C line appears for the negative sample. The strip images were converted to 8-bit grayscale using ImageJ (https://imagej.nih.gov/ij/), to determine the gray value of T line and C line intensity.

The RPA pre-amplification reagents (Table 1) were added to the TwistAmp^TM basic enzyme pellet, and 10 μl of the reconstituted RPA reaction buffer was distributed to a new PCR tube. One pellet yielded approximately four RPA reactions. If a larger number of reactions are needed, the master mix volume and pellets are scaled up accordingly. The CRISPR reaction reagents were prepared according to the components list in Table 1, then 4 μl of the mixture was added to the PCR tube lid. One microliter DNA sample was added to the bottom of the PCR tube and mixed by carefully pipetting up and down, then the PCR tubes were incubated at 37°C for 20 min. Subsequently, the CRISPR reagents pre-placed on the PCR tube lid were mixed with the RPA reaction buffer by inverting and centrifuging the tube, and incubated at 37°C for another 20 min. When the FQ reporter was used, the fluorescence signals were monitored via a portable fluorometer or real-time PCR fluorometer. When the LF reporter was used, a lateral flow strip was directly inserted into the reaction buffer after 85 μl water was added, and stand for about 3 min.

The mold for the microfluidic chip was first designed using AutoCAD (Supplementary Figure 1) and fabricated by a 3D printer using high-temperature resin. Next, the PDMS prepolymer, a two-component mixture of base and the crosslinking agent was put on the resin mold, and cured at a moderately elevated temperature (52°C for 12 h) to replicate the desired feature. An inlet (Φ 0.5 mm) was drilled, and covered with a Parafilm. To make this RPA/CRISPR method applicable for on-site field assay, the 13 μl of RPA reagents and 4.0 μl of CRISPR reagents (Table 1) were pre-stored in the corresponding reaction chamber, and lyophilized in a benchtop freeze dry system. Finally, the chip was quickly attached with another PDMS layer with double-sided tape (DSMS, Deer Brand, Taiwan) on a super clean bench.

The all-in-one chip system included a self-contained microfluidic chip and a portable heater pad or box. First, 15 μl extracted DNA was introduced to the RPA reaction chamber of the chip, which was put in a portable heater box set at about 37°C. After a 20 min incubation, the RPA amplicons in the RPA reaction chamber were pushed to the CRISPR reaction chamber by air with the pipette. After a 20 min-incubation, 85 μl deionized water was added to drive the CRISPR reaction buffer into the lateral flow strip chamber, where a lateral flow strip was inserted. The visual test result could be directly observed with naked eyes from the lateral flow strip in the detection chamber.

To avoid the interference by other fungi which infect Brassica napus/B. oleracea, or have the similar morphology of fungal hyphae and spores, we selected three Leptosphaeria fungi and three Diaporthe fungi having similar fungi colony shapes as the specificity evaluation. Seven pathogens including target pathogen L. maculans, approximate species L. biglobosa “brassicae”, L. biglobosa canadensis”, L. lindquistii, Diaporthe helianthin, Diaporthe phaseolorum var. caulivora, and Diaporthe phaseolorum var. meridionalis were detected to determine the specificity of RPA/Cas12a-based fluorescent assay and lateral flow assay, respectively.

The genomic DNA of L. maculans hyphae and spores were 10-fold diluted from 2.21 ng/μl (4.7×10⁴ copies/μl) to 0.221 pg/μl (4.7 copies/μl), which were detected with RPA/Cas12a-based assay. In the two-step method, template DNA was amplified according to the section “Recombinase Polymerase Amplification,” then the RPA products were detected according to the section “RPA/Cas12a-based Fluorescent Assay” and “RPA/Cas12a-based Lateral Flow Assay,” respectively. In one-pot detection, template DNA was detected with FQ reporter or LF reporter according to Section “One-pot detection.” For all-in-one chip detection, 15 μl diluted DNA solution was added and detected as explained in Section “RPA/Cas12a-based All-in-on Chip Detection.”

To check the applicability of the RPA/Cas12a-based assay for field samples, 200 mg L. maculans hypha was ground in liquid nitrogen and dispersed in 400 μl water, then 50 μl L. maculans hypha in water was added to 0.1 g healthy oilseed rape seeds, stems and cabbage leaves, respectively. The extracted DNA from healthy plant tissues and L. maculans hypha were compared using the RPA/Cas12a-based fluorescent assay and the AOCLFA. To evaluate the efficiency of fast nucleic acid release reagents for the plant tissues, 20, 40, and 60 μl L. maculans hypha in water were added to healthy cabbage leaves, and the DNA were extracted using plant DNA extraction kit and fast nucleic acid release reagents, respectively. Both extracted nucleic acids were detected using an RPA/Cas12a-based fluorescent assay. Furthermore, DNA extracted from sixteen unknown fungi hyphae were detected with the AOCLFA. These fungi DNA were also amplified with primer ITS1/ITS4, sequenced, and aligned using NCBI Blastn.

Statistical analyses and Analysis of Variance (ANOVA) were performed with Microsoft Excel. RPA/Cas12a-based fluorescent experiments were repeated at least 3 times for each sample. All experimental results were shown as Mean ± SD unless otherwise stated.

Lateral flow assay is one of the most convenient point-of-care testing (POCT) techniques and has been widely used for pathogen detection (Corstjens et al., 2001; Mao et al., 2009; Wang et al., 2020a). In this study, we adopted a paper-based lateral flow strip to develop two portable platforms, i.e., one-pot detection and an all-in-one chip lateral flow assay (AOCLFA) for the visual detection of L. maculans, which could reduce the reliance on equipment, simplify the operations, and avoid contamination. As illustrated in Figure 1, this portable platform integrated (i) RPA pre-amplification of sample DNA, (ii) sequence-specific recognition and non-specific cleavage by Cas12a/crRNA, and (iii) colorimetric readout of cleaved product from LF reporter. Both RPA pre-amplification reagents and Cas12a/crRNA reagents were pre-stored in one tube or different reaction chamber in one chip. For the negative result, only the control band would appear on the lateral flow strip, while both the test band and control band or only the test banc appeared in the presence of the positive sample. A portable incubator pad or thermal box with a mobile battery that can provide a temperature between 37 and 42°C was found to be suitable to meet the needs of both RPA and Cas12a reactions. By employing the fast plant DNA extraction procedure, the total time from sample preparation to results readout was <1 h. The minimum equipment required to operate the protocol included only pipettes, reagent tubes, a portable thermal box or pad, and lateral flow strips. All the equipment could be integrated into a portable suitcase. Therefore, the proposed method had great potential to enable on-site field assay of plant pathogen fungi outside of the laboratory.

The internally transcribed spacer (ITS) of L. maculans has limited sequence variance compared with its approximate species, i.e., L. biglobosa “brassicae” (Lbb) and L. biglobosa “canadensis” (Lbc) (Figure 2A). There are two regions (from 1 to 182 and from 365 to 546) having obvious variation between L. maculans and its approximate species. Since the region from 1 to 182 has more variance (56 bp) than the region from 365 to 546 (30 bp), primers Lm-F/Lm-R are located in the regions (1–182) were designed to produce amplicons of 154 bp. The protospacer adjacent motif (PAM) sequence of TTTN (Gootenberg et al., 2017; Chen et al., 2018) is required to recognize the target dsDNA by CRISPR/Cas12a. Among the five PAMs sites (TTTN) in the RPA target region of L. maculans, only the 20 bp after PAM1 site showed the maximum 9 nt difference with Lbc and 12nt difference with Lbb (Figure 2A). For the specificity of this crRNA designed from PAM1, the fluorescence signals using FL reporter as the ssDNA reporter indicated that only L. maculans genomic DNA generated high fluorescence signals (Figure 2B, line 1), but DNAs extracted from approximate species, such as L. biglobosa “brassicae,” L. biglobosa “canadensis,” L. lindquistii, Diaporthe helianthi, Diaporthe phaseolorum var. caulivora and Diaporthe phaseolorum var. meridionalis produced no obvious signals (Figure 2B, line 2–7). The lateral flow strip results also indicated that only L. maculans showed positive test lines using this Cas12a/crRNA detection system with LF reporter, and other allied species showed no positive lines (Figure 2C). The gray value ratio of the T line between the samples and the negative control indicated that L. maculans could produce detectable signals much stronger than other fungi (Figure 2D). Therefore, the designed primers and crRNA had good specificity for L. maculans.

For RPA pre-amplification time, results (Figures 3A,B) indicated that shorter RPA pre-amplification time produced less amplicons, thus yielding lower CRISPR/Cas12a signals. There was no obvious signal difference when RPA time was more than 20 min, indicating that 20 min was enough for RPA pre-amplification (Figure 3C). Thus, 20 min was used as the optimized RPA pre-amplification time. For CRISPR/Cas12a reaction time, about 15 min was enough to produce a detectable test band for the RPA pre-amplification products of L. maculans (Figures 3D,E). However, prolonged incubation (e.g., 50 and 60 min) may increase non-specific products (Figures 3F,G). Therefore, 20–30 min was the optimized Cas12a/crRNA reaction time throughout the experiment.

The ssDNA reporter in CRISPR/Cas12a system was first evaluated using the FQ reporter involved in fluorescence detection. Four concentrations of FQ reporter, i.e., 25, 50, 100, and 200 nM were adopted in the Cas12a/crRNA detection system with the RPA pre-amplification products of L. maculans or water as the template. Results showed that FQ reporters with higher concentrations provided higher fluorescence signals (Figures 3H–J). Taking the cost into account, 100 nM of FQ reporters were used in the subsequent experiment. For lateral flow assay, it has been reported that 1 μM LF reporter (FAM-TTATT-Biotin) can avoid the false positive band (Lu et al., 2020). However, we found that both concentration and length of LF reporter affected the generation of the false-positive band. The results showed that a false-positive band appeared when a 5 bp LF reporter with a lower concentration than 100 nM was used; In contrast, when a 20 bp LF reporter (FAM-T₂₀-Biotin) was used, false-positive bands could be avoided at lower concentrations (Figures 3K,L). Thus, the 20 bp ssDNA LF reporter with 100 nM was adopted in the following study.

To evaluate the possibility of all-in-one chip detection, we first performed the integration of RPA and Cas12a cleavage in a single tube. NEBuffer 3.1 was not included in RPA pre-amplification buffer, but on the tube lid to avoid the negative effect of Mg²⁺ in NEBuffer 3.1 on the RPA pre-amplification. Results showed that the addition of all the components except Cas12a enzyme and NEBuffer 3.1 to the RPA reaction produced no fluorescence signal (Figure 4A, line 4). To investigate which component affects the fluorescence signal, crRNA and FQ reporter were added separately to the RPA reaction, and the results showed that the inclusion of crRNA/DTT/RNase inhibitor in the RPA buffer produced no fluorescence signal (Figure 4A, line 5), while FQ reporter did not affect the generation of the fluorescence signal (Figure 4A, line 3). DTT and RNase inhibitors were jointly used to protect crRNA from degradation. Therefore, in one-pot detection, NEBuffer 3.1, Cas12a enzyme, DTT, RNase inhibitor, and crRNA were added to the tube lid or tube wall. After the RPA pre-amplification finished, the CRISPR/Cas12a reaction components were shaken into the RPA reaction buffer, and then the tube was incubated at 37°C for another 20 min.

For the RPA pre-amplification time involved in one-pot detection, four time points, i.e., 15, 20, 25, and 30 min were tested, and the results showed that although longer RPA pre-amplification produced higher fluorescence signals for the same sample (4.7 × 10³ copies DNA), the difference is not obvious, indicating 15 min was enough to produce RPA amplicons for one-pot detection (Figure 4B). For the lateral flow assay, 4 μl FQ reporter (2 μM) was replaced with 4 μl LF reporter (10 μM), and the cleaved LF reporter was detected by the lateral flow strip. To optimize the Cas12a reaction time, the products of four time points, i.e., 15, 20, 25, and 30 min were tested with lateral flow strips. The results showed that 15 min was enough to produce a detectable T line (Figures 4C,D), which was consistent with the two-step RPA/Cas12a experimental results (Figure 3D). The total time of RPA pre-amplification and Cas12a reaction was about 30 min.

The sensitivity of RPA/CRISPR/Cas12a detection was evaluated using the genomic DNA of L. maculans from 2.21 ng/μl (4.7 × 10⁴ copies/μl) to 0.221 pg/μl (4.7 copies/μl). For the two-step detection, the genomic DNA was pre-amplified with RPA for 20 min, then 2 μl pre-amplification products were added into the 18 μl of Cas12a/crRNA reaction buffer for another 20 min. Every reaction was repeated 3 times. The results showed that 0.221 pg (4.7 copies) of L. maculans could be detected with the developed Cas12a-mediated fluorescence detection (Figure 5A). This sensitivity is 100 times higher than that of fluorescent RPA and 10 times higher than that of real-time PCR (Lei et al., 2019), indicating that Cas12a-mediated detection combined with RPA had high sensitivity. For the two-step detection with lateral flow strips, the results showed that 2.21 pg (47 copies) of L. maculans was obviously detected, and 0.221 pg (4.7 copies) was weakly detected (Figure 5B). The gray value of the T line of 0.221 pg L. maculans DNA was higher than that of the negative control (Figure 5C). Thus, this sensitivity is 100 times higher than that of fluorescence RPA (Lei et al., 2019), indicating that RPA/CRISPR/Cas12a with lateral flow strips also had good sensitivity.

For one-pot detection, the same L. maculans genomic DNA samples as those in two-step detection were used to evaluate the sensitivity. The results showed that the sensitivity of the developed one-pot detection method using fluorescence reporter (Figure 5D) was the same as that of two-step detection (Figure 5A). Although the T lines of 2.21 ng and 0.221 ng L. maculans DNA were lighter than those in two-step detection, the T line of 0.221 pg L. maculans DNA was obviously stronger than that of negative control (Figure 5E). The gray value of the T line of 0.221 pg L. maculans DNA was higher than that of negative control (Figure 5F) indicating that one-pot lateral flow assay has the same sensitivity as that of two-step detection.

The sensitivity of AOCLFP was discounted compared with the two-step and one-pot detection. As shown in Figures 5G,H, the T line of 15 μL DNA (2.21 pg/μl) can be weakly detected, and the LOD was calculated as 705 copies. The lower sensitivity in AOCLFP may result from the inadequate mixing of injected DNA samples with the pre-stored lyophilized RPA reagents or CRISPR/Cas12a reagents in the chip chamber.

The efficiency of crude DNA extraction is 75.2, 61.4, and 52.0% for 60, 40, and 20 μl L. maculans hyphae spiked plant tissues (Figures 6A,B). The extracted DNA samples from healthy plant tissues and L. maculans hypha spiked plant tissues were first determined with the RPA/Cas12a-based fluorescent assay. Results showed that there were no fluorescence signals for the healthy plant tissues (Figure 6C, line a–c), but L. maculans hypha spiked plant tissues had strong signals (Figure 6C, line a'–c'). The AOCLFP results showed that the healthy plant tissues showed no positive results, but the L. maculans hypha spiked plant tissues showed positive T lines (Figure 6D). To meet the field detection requirement, we developed a portable suitcase, which integrated pipettes, tubes, chips with lyophilized reagents, a portable thermal box, and lateral flow strips altogether (Figure 6E). This AOCLFA was applied to detect sixteen fungi samples, which were meanwhile identified with ITS1/ITS4 amplification and sequencing. The results showed that only L. maculans presented positive signals, and other fungi DNA presented negative signals (Figure 6F). The AOCLFA detection results of these 16 fungi were consistent with the NCBI Blastn results of the sequencing (Supplementary Table 5).