Recombinase polymerase amplification (RPA) with lateral flow detection for three Anaplasma species of importance to livestock health

Anaplasma marginale, A. ovis, and A. phagocytophilum are the causative agents of bovine anaplasmosis, ovine anaplasmosis, and granulocytic anaplasmosis, respectively. The gold standard for diagnosis of post-acute and long-term persistent infections is the serological cELISA, which does not discriminate between Anaplasma species and requires highly equipped laboratories and trained personnel. This study addresses the development of a rapid, isothermal, sensitive, species-specific RPA assays to detect three Anaplasma species in blood and cELISA A. marginale-positive serum samples. Three RPA primer and probe sets were designed targeting msp4 genes of each Anaplasma species and the internal control (GAPDH gene) for each assay. The limit of detection of gel-based or RPA-basic assays is 8.99 × 104 copies/µl = A. marginale, 5.04 × 106 copies/µl = A. ovis, and 4.58 × 103 copies/µl = A. phagocytophilum, and for each multiplex lateral flow or RPA-nfo assays is 8.99 × 103 copies/µl of A. marginale, 5.04 × 103 copies/µl of A. ovis, 4.58 × 103 copies/µl of A. phagocytophilum, and 5.51 × 103 copies/µl of internal control (GAPDH). Although none of the 80 blood samples collected from Oklahoma cattle were positive, the RPA-nfo assays detected all A. marginale cattle blood samples with varying prevalence rates of infection, 83% of the 24 cELISA A. marginale-positive serum samples, and all A. phagocytophilum cell culture samples. Overall, although early detection of three Anaplasma species was not specifically addressed, the described RPA technique represents an improvement for detection of three Anaplasma in regions where access to laboratory equipment is limited.

Artificial positive control (APC). Endpoint PCR was used to confirm the size of the two complete APC sequences. The PCR product size of the RPA-basic APC was 250 bp (from A. phagocytophilum forward (Ap_L) to GAPDH reverse (GAPDH2_R) primers, Supplemental Fig. 1A) and the RPA-nfo for lateral flow APC, which was 442 bp (from A. phagocytophilum forward (Ap_L) primer to A. ovis reverse (Ao2_R) primer, Supplemental Fig. 1B). Furthermore, all targets of both APCs were amplified individually using RPA-basic reactions. All the resulting amplicons were detected by agarose gel electrophoresis. The product size between APCs and actual Anaplasma reference controls were slightly different as expected. The RPA amplicons amplified using the RPA-basic_APC with primers for A. marginale (Am3R/L), A. ovis (Ao2R/L), A. phagocytophilum (ApR/L), and GAPDH (GAPDH2R/L) were 93 bp, 193 bp, 188 bp, and 152 bp, respectively; while the products amplified using reference A. marginale, A. ovis, A. phagocytophilum, and GAPDH DNA were 103 bp, 184 bp, 202 bp, and 168 bp, respectively. The expected RPA product size of the RPA-nfo_APC using this synthetic control were 137 bp, 189 bp, 220 bp, and 160 bp which were amplified using A. marginale primers (Am3R/L), A. ovis primers (Ao2R/L), A. phagocytophilum primers (ApR/L), and GAPDH primers (GAPDH2R/L), respectively.
Optimization of RPA-basic and RPA-nfo conditions. Betaine, temperature, and incubation time were optimized using RPA-basic reactions based on agarose gel results. Betaine (10 µl) was added to the RPA reaction to minimize false-positives and to reduce mis-priming. Agarose gel results confirmed that non-template Table 1. Anaplasma marginale, A. ovis, A. phagocytophilum, and GADPH gene (internal control) RPA primers and probes*. *Primers, probes, and the amplification kit are protected by a patent application. www.nature.com/scientificreports/ controls did not amplify non-specific targets when betaine was added to the reaction nor did it interfere with the amplification of RPA targets. The three primer sets amplified the three expected diagnostic products from the predicted msp4 gene of A. marginale (103 bp), A. ovis (184 bp), and A. phagocytophilum (202 bp) within a range of six temperatures (35-40 °C). The reaction times were 20 or 40 min, indicating adequate performance in a broad range of temperatures and reaction times. The band intensity was the same for each assay so 37 °C with a reaction time of 20 min were selected for routine assay. None of the Anaplasma species primer sets amplified non-specific products from the non-template control (water). The amplified RPA-nfo products of A. marginale, A. ovis, and A. phagocytophilum were detected in lateral flow assay (LFA). The presence of control line C confirmed the lateral flow assay was working properly, test line 1 showed the internal control (GADPH gene), and test line 2 verified the presence of three Anaplasma species in the samples. All tests were consistent for each repeated assay demonstrating consistency in the amplification of all bacterial targets tested, only the control line C developed in the negative non-template control (water).
Faint test lines were observed when the incubation time was 5 min using lateral flow assay. However, as the incubation time was increased to 10 min and 15 min, stronger positive signals were detected. Based on these results, the RPA-nfo reaction time consisted of two-steps: first, RPA amplification in dry bath incubator at 37 °C for 20 min, and second, lateral flow assay at room temperature for 10 min as recommended for a total incubation time of 30 min.
RPA-basic reactions were performed using total DNA extracted from A. marginale-infected cattle blood and sheep blood spiked with A. ovis DNA. The RPA primer set GAPDH2R/L (internal control) amplified a product from the predicted target of the GAPDH gene at 37 °C for 20 min. The RPA products were obtained within the expected amplification size of 168 bp. The reverse and forward RPA primers did not amplify products from the negative control tick DNA and the non-template control (water).
Multiplex RPA-nfo reactions targeted each Anaplasma species with the GAPDH housekeeping gene (internal control). The results of sixteen multiplex RPA-nfo assays using an Artificial Positive Control (APC: 1 ng/µl) as a template demonstrated A. marginale and internal control test lines were clear and intense when primers were loaded at a volume of 1.8 µl (0.36 µM final concentration) and the probes 0.2 µl (0.04 µM final concentration). The best combination for A. ovis and A. phagocytophilum was 1.05 µl of primers (0.21 µM final concentration) and 0.6 µl of probes (0.12 µM final concentration). In these assays, the two test bands (line 1 and 2) were equally intense and clear. No signal was observed in non-template control (water).
Specificity of RPA-basic and RPA-nfo assays. Each species-specific primer set for A. marginale, A. ovis, and A. phagocytophilum amplified the specific sample DNA and no amplification was observed when primers were tested against the other two Anaplasma species (Fig. 1A,C,E). No cross-amplification was observed in any reaction. Lab-reared tick DNA was used as negative control and NTC (no template control) was also included in each of the tests. These negative controls did not produce any reaction using the RPA-basic reactions.
The multiplex RPA-nfo tested positive only in each of the Anaplasma species target; a solid positive test internal control band developed in each of the Anaplasma species DNA, whereas no signals (lines 1 and 2) were observed in the negative dipsticks (tick and NTC) (Fig. 1B,D,F). The results indicated that the primer-probe combinations designed for RPA-nfo reactions were specific to each of the corresponding Anaplasma species targets. The species-specific isothermal reactions consistently detected and discriminated the three Anaplasma species.
A quantitative PCR standard curve was used to determine the concentration of A. marginale, A. ovis, and A. phagocytophilum in total extracted DNA. The concentration of bacteria in the samples was 9.21 × 10 7 copies/µl for A. marginale:, 6.3 × 10 6 copies/µl for A. ovis:, and 6 × 10 10 copies/µl for A. phagocytophilum: Therefore, these three samples were used in RPA-basic and RPA-nfo limit of detection assays. Additionally, the three quantitative PCR showed a similar significant tendency (r = 0.99).
A. marginale quantified sample was diluted from 8.99 × 10 9 to 8.99 × 10 3 copies/µl, the limit of detection of RPA-basic using total DNA was 8.99 × 10 3 copies/µl (Fig. 2B). The limit of detection of RPA-basic to detect A. marginale from plasmid DNA was 10 times more sensitive than from an infected blood sample. The A. ovis quantified sample was diluted from 5.04 × 10 6 to 5.04 × 10 3 copies/µl, the limit of detection of RPA-basic using total DNA was 5.04 × 10 6 copies/µl (Fig. 3B). The A. phagocytophilum quantified sample was diluted from 4.58 × 10 9 to 4.58 × 10 3 copies/µl, the limit of detection of RPA-basic using total DNA was 4.58 × 10 3 copies/µl (Fig. 4B). The limit of detection of RPA-basic to detect A. phagocytophilum and A. ovis from plasmid DNA and an extracted total DNA sample was equivalent. No amplification was observed with non-template control in each.
The limit of detection of RPA-nfo assays was measured using a ten-fold serial dilution of the Artificial Positive Control (APC). The results of A. marginale, A. ovis, and A. phagocytophilum RPA reactions using primers and probes shown that method allows detecting as low as was 8.99 × 10 3 copies/µl of A. marginale, 5.04 × 10 3 copies/µl of A. ovis, 4.59 × 10 3 copies/µl of A. phagocytophilum, and 5.51 × 10 3 copies/µl of internal control (GAPDH) with APC and bacterial measured total DNA (Figs. 2C,D, 3C,D, 4C,D). Clear test (1,2) and control lines appeared on each strip; however, 1 and 2 test lines were faint when DNA concentration was decreasing. Only the control line band was observed with non-template control (water) in each assay. Therefore, the limit of detection of RPA-nfo was higher than the RPA-basic amplification detected by agarose gel electrophoresis.  Twenty-five A. marginale positive blood samples and three A. phagocytophilum positive cell culture samples were simultaneously detected by qPCR using optimized RPA primers and multiplex RPA-nfo (Supplemental Table 2 Table 2 and Supplemental Fig. 2). While the amount of quantified Anaplasma DNA in total extracted blood DNA may have varied between blood samples and infection levels due to unspecified methodologies in the source lab, the RPA assay still detected A. marginale DNA in all of them (Supplemental Table 1). All cell culture samples tested positive for A. phagocytophilum by qPCR and RPA-nfo reactions (Supplemental Table 2 and Supplemental Fig. 3). No amplification was observed with non-template control (water) of both assays. Two test lines (1,2) were observed with APC, A. marginale, and A. phagocytophilum reference positive control in PCRD cassette. www.nature.com/scientificreports/ All 80 field-collected blood samples tested negative to A. marginale by endpoint PCR. To preserve the use of lateral flow assay, two randomly selected DNA blood samples of each sale barn were assayed using RPA-basic and RPA-nfo. All eight samples tested negative for A. marginale. Expected PCR and RPA product sizes of 344 bp and 103 bp were visible with A. marginale reference positive control, respectively. Two test lines (1, 2) were observed with APC and A. marginale reference positive control in lateral flow assay and no amplification was observed with non-template control (water) of the three assays.

Discussion
Rapid diagnostic tests are useful for screening livestock at point-of-care must not only be able to detect low parasitemias of Anaplasma infections in chronically-infected animals, but must also be simple to use by untrained personnel. This feature involves the use of no expensive equipment and easy to interpretation 22 . This study describes the development of three RPA primer and probe sets for rapid, sensitive, and species-specific detection of A. marginale, A. ovis, and A. phagocytophilum by RPA-basic and multiplex RPA-nfo using GAPDH gene as internal control. Coupling RPA-nfo assays in a lateral flow assay creates the opportunity to develop rapid pointof-care diagnostic tests for three Anaplasma species which affect cattle, sheep, and goats. When put together with an Elution Independent Collection Device (EICD) prototype 23 , the whole process from blood extraction to accurate detection of species-specific Anaplasma species at the point of sampling is 60-70 min.
Although 2019 was a low year for Anaplasma infections in Oklahoma cattle (Justin Talley, personal communication), the usefulness of the rapid Anaplasma detection (RAD) assay was demonstrated in the detection of 83% of the cELISA A. marginale-positive serum samples from cattle compared with only 4% by endpoint PCR. These results are due to the dramatic differences in limit of detection between the multiplex RPA-nfo assay  Sensitive diagnostics are used to detect Anaplasma infections in livestock in diagnostic labs worldwide 10, 11 , but there is a need for point-of-care diagnostics for the detection of Anaplasma that will screen livestock in fieldconditions. The species focused on in this study are important globally. Anaplasma marginale is a problem in the US, not only in the states where most livestock is raised but also in other states as well [32][33][34][35][36][37] , in addition to cattle and goats worldwide [38][39][40][41][42] . While A. ovis has not been reported to adversely affect sheep and goats, it continues to be identified in ovine and caprine species, domestic and wild, worldwide 43,44 . The strains of A. phagocytophilum in the United States do not appear to infect bovine populations 23, 45 but they do cause death and morbidity in European cattle 46,47 . One of the issues that is often reported for the standard test used in professional diagnostic labs is that the cELISA does not discriminate between Anaplasma species as it is based on the msp5 gene which www.nature.com/scientificreports/ is cross-reactive among Anaplasma species 2, 48 . In the current assay, each RPA primer/probe set was based on the msp4 gene which specifically differentiates between the species and ensures that each assay will only detect one species of Anaplasma. In addition to being based on a species-specific gene, we recognized the many strains of A. phagocytophilum worldwide, some affecting humans while some affect cattle 7, 49 . Our A. phagocytophilum RPA primer/probe set was developed from over 20 different strains to ensure the detection of the majority of strains currently reported. While our assay is built for consensus detection, it would be possible to build strain specific primer/probe sets as well. Altogether, there is still much to learn about the different strains of these pathogens and how they affect livestock species in different countries. As these studies continue to reveal the extent to which A. marginale, A. ovis, and A. phagocytophilum strains are impacting livestock development globally, the need for point-of-care detection of these pathogens at the local, community level becomes ever more important. RPA is a relatively new technology, so we identified some key aspects that are imperative when developing sensitive and specific RPA primers and probes. First, development of accurate RPA reactions requires primers (30-35 nucleotides) and probes (46-52 nucleotides) which are longer than conventional PCR primers, however there are no optimal software packages from which to develop the RPA primers. We designed the primers using the web interface application Primer3 50 and manually created the RPA probes according to the selection parameters for the optimal RPA primers and probes described in TwistAmp Design Manual 51 . As RPA product size influences the quality of bands, the best RPA primers were A. ovis and A. phagocytophilum, which generated intense and clear bands (more than 150 bp) compared with bands generated by RPA primers were A. marginale and GAPDH internal control. As the variation was not clear, further studies focused on primer and probe parameters are needed to improve band intensity of A. marginale. Non-specific amplification was eliminated by  www.nature.com/scientificreports/ using betaine which is used in PCR, LAMP, and RPA to prevent secondary structures due to high GC content (varying between 40 and 60%) in target sequences, primers, and probes which may favor hairpin formation and create false-positive results 52 . The addition of betaine allowed positive results at a wider range of temperatures (35-40 °C). Finally, while commonly reported for LAMP assays 53,54 , inactivation of RPA reactions at 80 °C for 5 min prevented possible cross-contamination among reactions.
To develop multiplex RPA-nfo assays, we optimized the multiplex RPA primers and probes to avoid crossinteraction between dyes and primer/probe secondary structure or hairpins formation which cause lower signal intensity in the test lines or false-positive results, respectively 55,56 . In this study, two targets, A. marginale msp4 and GAPDH genes, A. ovis msp4 and GAPDH genes, and A. phagocytophilum msp4 and GAPDH genes, were tested in single multiplex RPA-nfo reaction with optimized primer and probe combinations. In total, the amplifications obtained two clear, intense lines on each reaction unit which could be easily read in any light. In the future, it would be possible to place all three reactions with their controls into one unit as some commercial serological tests are currently packaged.
One of the challenging aspects of developing molecular diagnostics is the need for target pathogen DNA on which to validate that the assays are detecting the pathogen and reducing the changes of false-negative results. Positive controls are often difficult to obtain because not all proteins or pathogens are available, or some pathogens are exotic and not commercially available 57 . At times, samples containing infectious material are also used as positive controls, however, shipping and handling of these samples is risky and require permits 58 . The development of an artificial positive control (APC) involves the use of customized synthetic DNA inserts based on linear arrays of primer sequences designed from a variety of organisms or targets important in detection, diagnostics or research 59 . Used previously to mimic multiple pathogens 22 , APCs can reduce risks associated with in vivo positive controls and improve accuracy of molecular detection techniques. In the current study, we used two constructed synthetic DNA positive controls targeting A. marginale, A. ovis, A. phagocytophilum, and GAPDH housekeeping genes using species-specific RPA primers and nfo probes synthesized in a pUC57 vector system. When these two APCs were compared with Anaplasma-infected blood samples as reference positive controls in the RPA-basic and RPA-nfo assay systems, we found they both demonstrated accurate amplification. The amplicons generated by the RPA assay varied slightly in size from those generated by the Anaplasma reference positive controls. This was most likely due to differing annealing sites of target sequences in vivo and the distribution of primers and probe sequences in APC. However, these variations were not an issue because the Anaplasma species and internal control targets amplified as well demonstrating that APCs worked correctly.

Conclusion
The Anaplasma-specific RPA assays developed in this study are part of a wider project to develop rapid diagnostic assays that can detect the extremely low parasitemia levels of Anaplasma infections in chronically infected animals. To augment this goal, the RPA assays were developed in conjunction with a lateral flow assay to make the technique simple to use at a point-of-sampling site, the results easy to interpret, and able to be used by untrained personnel. When put together with an Elution Independent Collection Device (EICD) prototype 23 , the whole process from blood extraction to accurate detection of species-specific Anaplasma took 60-70 min. While this process continues to be streamlined to reduce the testing period, this marks an important step in the development of point-of-care diagnostics for Anaplasma species which can be used by field-based veterinarians as well as APHIS agents to monitor livestock at ports of entry into the United States. This becomes even more important as co-infections of two or more Anaplasma species have been reported in ticks, deer, and cattle in different parts of the world 2, 60 . In this study, specificity assays of each set of RPA primers and probes as well as the multiplex www.nature.com/scientificreports/ RPA-nfo reactions detected only the specific Anaplasma sample targets. Both RPA-basic and multiplex RPA-nfo identified and discriminated among three Anaplasma species and detected Anaplasma marginale DNA in the serum of 83% of cELISA A. marginale-positive cattle and 96% of A. marginale positive blood samples. By combining these species-specific RPA assays for three Anaplasma species with appropriate controls in a lateral-flow delivery system, we have demonstrated the flexibility and utility of this molecular technique in the development of many types of field-diagnostics.

Materials and methods
Source of samples. Reference  RPA primers were designed using Primer3 50 while the thermodynamics and tendency to form self-dimers was analyzed using mFold 62 . The selected parameters for optimal RPA primers were as described in the TwistAmp Design Manual 51 . The specificity in-silico assay of primer sets were performed using BLASTn 63 . RPA primers and probes were synthesized by Integrated DNA Technologies (IDT) and Biosearch Technologies Inc., respectively. The nfo probes were designed to be located between forward and reverse primers. To adapt the RPA reaction for lateral flow assay, three modifications were added to the probes: 6-carboxyfluorescein or digoxigenin tag at 5′, tetrahydrofuran located around 30 bp of the 5′-end and a polymerase blocking group (C3 spacer) at the 3′-end. A. marginale, A. ovis, and A. phagocytophilum probes were labeled at 5′ position with fluorescent dye FAM, the GAPDH probe was labeled at 5′ position with DIG (Digoxigenin), and the reverse primers with biotin.
Artificial positive control (APC). Two artificial positive controls (APC) were designed, one based on tandem of forward and reverse complement sequences of RPA primers, and the second based on tandem of forward and reverse complement sequences of RPA primers and nfo probes targeting A. marginale, A. ovis, A. phagocytophilum and GAPDH gene as reported 59 . An APC is a cloneable, synthetic, multi-target, and non-infectious control used for routine application in detection and diagnostics assays 59 . Each sequence was designed and made synthetically then inserted into a pUC57 restriction site (GenScript Inc, USA) (Supplemental Fig. 3A,B). The multiplex RPA-nfo assay was performed using RPA TwistAmp® nfo kit (TwistDx, UK) according to the manufacture's protocol with modifications (10 μl betaine). A factorial combination assay to include four volumes of primers (2.1 µl, 1.575 µl, 1.05 µl, and 0.525 µl) and probes (0.6 µl, 0.45 µl, 0.3 µl, and 0.15 µl) was evaluated in order to obtain the optimal combination between internal control and Anaplasma species. The RPA incubation followed the same protocol as described above. After amplification, 6 μl of RPA products were mixed with 84 μl of buffer (Abingdon Health, UK), 75 μl of the diluted sample was added to a PCRD Nucleic Acid Detector cassette (Abingdon Health, UK). The results were registered after 15 min.
Cloning of diagnostic Anaplasma species fragments. The three RPA primer sets were adapted for end-point PCR to clone the amplified diagnostic fragments. The PCR amplified products from A. marginale, A. ovis, and A. phagocytophilum were purified from excised agarose gel bands using QIAquick Gel Extraction Kit (Qiagen, USA). The TOPO TA cloning kit (Invitrogen, USA) was used to clone the three amplified segments of Anaplasma according to the manufacturer's instructions. PCR products were inserted into the pCR'4-TOPO plasmids which were incubated in Escherichia coli competent cells following the manufacturer's protocol. The amplified PCR products carried into transformed bacterial colonies were sequenced to verify whether the Anaplasma fragments corresponded to each of the expected Anaplasma species in the plasmids. The bacterial plasmids were purified using Plasmid Mini Kit (Qiagen, USA) following the manufacturer's instructions.
LC green qPCR of Anaplasma DNA. Total DNA extracted from cattle blood infected with A. marginale, sheep blood spiked with A. ovis DNA, and cattle blood spiked with A. phagocytophilum DNA was quantified using LC green quantitative PCR (qPCR). Ten-fold dilutions of previously described plasmid DNA A. marginale (8.99 × 10 9 -8.99 × 10 3 copies/µl), A. ovis (5.04 × 10 9 -5.04 × 10 3 copies/µl), and A. phagocytophilum (4.58 × 10 9 -4.58 × 10 3 copies/µl), were used to plot standard curves for each of the three Anaplasma species DNA quantification. Plasmid DNA concentration was measured by Qubit 4 Fluorometer (Thermo Fisher, USA). The qPCR amplification was carried out in 20 µl final volume containing 10 µl of One Taq Hot Start DNA Polymerase (Biolabs, USA), 0.5 μl of each RPA sense and antisense primer (10 μM), 2 μl of LC green (BioChem, USA), 1 μl of plasmid or DNA sample, 6 μl nuclease-free water. Each reaction was tested in triplicate. The PCR cycling parameters were initial start of 50 °C for 3 min, initial denaturation of 94 °C for 4 min, 40 cycles of denaturation at 95 °C for 20 s, annealing at 62 °C for 20 s, extension at 72 °C for 20 s and final extension final at 72 °C for 4 min. The assays were performed in a Rotor Gene 6000 series (Corbett Research, Qiagen, USA) and the mean of each set of replicates was calculated. The quantity of DNA in each sample was determined by converting the copy number using the formula (amount of DNA (ng) × 6.022 × 10 23 ) / (length of DNA (bp) x 10 9 × 650).
Limit of detection and specificity of the RPA-basic and RPA-nfo assays. The limit of detection of the RPA-basic and RPA-nfo assays was assessed using the three cloned plasmids. The limit of detection for the three primer sets was determined using a ten-fold serial dilution of A. marginale (8.99 × 10 9 -8.99 × 10 3 copies/µl), A. ovis (5.04 × 10 9 -5.04 × 10 3 copies/µl), and A. phagocytophilum (4.58 × 10 9 -4.58 × 10 3 copies/µl). One microliter of each dilution was used as template for RPA-basic and RPA-nfo assays.
The specificity of the three RPA primer pairs was tested against the A. marginale, A. ovis, and A. phagocytophilum DNA reference positive controls. Non-infected tick DNA was used as negative control and NTC (no template control) was also included in all assays. All results were observed by electrophoresis (1X TAE) on agarose gel and lateral flow assay (PCRD).

Screening of serum and blood samples.
Twenty-four cELISA A. marginale-positive serum samples, twenty-five blood samples infected with A. marginale, and three cell culture preparations infected with A. phagocytophilum were screened using published endpoint PCR 44 , qPCR with the optimized RPA primers and the multiplex RPA-nfo. All molecular assays were performed using 3 µl of each sample. Twenty microliters of amplified PCR product were electrophoresed and 6 μl of amplified RPA product was mixed with 84 μl of lateral flow buffer, and 75 μl of the diluted sample was loaded to a PCRD.