Low-cost genotyping method based on allele-specific recombinase polymerase amplification and colorimetric microarray detection
The costs of current genotyping methods limit their application to personalized therapy. The authors describe an alternative approach for the detection of single-point-polymorphisms using recombinant polymerase amplification as an allele-specific technique. The use of short and chemically modified primers and locked nucleic acids allowed for a selective isothermal amplification of wild-type or mutant variants at 37 °C within 40 min. An amplification chip platform containing 100 wells was manufactured with a 3D printer and using thermoplastic polylactic acid. The platform reduces reagent consumption and allows parallelization. As a proof of concept, the method was applied to the genotyping of four SNPs that are related to the treatment of tobacco addiction. The target polymorphisms included rs4680 (COMT gene), rs1799971 (OPRM1 gene), rs1800497 (ANKK1 gene), and rs16969968 (CHRNA5 gene). The genotype populations can be well discriminated.
KeywordsPharmacogenomics SNP genotyping Isothermal amplification Micro-well plate Microchip Tobacco addiction 3D–printer COMT gene OPRM1 gene ANKK1 gene CHRNA5 gene
Pharmacogenomics is currently considered one of the most active areas of the personalized medicine paradigm. However, numerous barriers have been encountered to launch DNA variation analyses, such as single-nucleotide polymorphisms (SNPs), in health systems . Among other factors, available platforms have a marked technological limitation. Most genomic findings have been obtained from high-throughput technologies, such as Illumina and Affimetrix platforms. However, the incorporation of these methods into primary care centers is limited, and cost is the main drawback [2, 3, 4]. Dramatically cutting the turnaround times of these platforms is an important goal for implementing SNP testing into clinical scenarios. Therefore, the novel generation of simple diagnostic tools is absolutely necessary for the real adoption of personalized medicine [5, 6].
A large family of high-potential methods to be developed in simple systems is that based on allele-specific (AS) amplification . Polymerase chain reaction (PCR) uses primer pairs, deliberately designed at SNP sites. Primers have a single-base variation at the 3′ end (allele-specific primers), so extension and amplification reactions take place with only perfectly-matched sequences of target regions. This approach has been successfully used as a pharmacogenomic tool combined with several detection systems [8, 9, 10]. Nevertheless, these techniques require particular thermal cycling, consequently there are several limitations for their future integration as point-of-care devices. PCR demands an accurate temperature control system to quickly heat/cool reaction solutions. The high temperature reached (up to 95 °C) leads to variations in the volume reaction due to water evaporation and gas bubble formation, which renders accurate process control necessary.
Several new technologies have emerged to improve DNA-based analyses . Many efforts have been made focusing on developing low-cost systems to be used for point-of-care applications or in small laboratories located at the physician’s office or in primary health centers.
A revolution in the development of new methods is currently being witnessed, and is associated with the application of isothermal solutions for microanalyses . These approaches are based on using proteins that separate DNA strands instead of thermal approaches, so that target nucleic acids are synthesized at constant temperature. Nevertheless, the application of isothermal amplification for SNP genotyping is still minimum [13, 14]. Among isothermal reactions, recombinase polymerase amplification (RPA) has many interesting properties; e.g., short reaction times (20–60 min), robustness and low energy requirements (close to room temperature) . Recently, our research group has demonstrated that polycarbonate-based substrates can be used to support RPA assays in chip , micro-reactors , or dynamic formats . A well array chip has been described to process miniaturized RPA assays, applied for pathogen detection in under 30 min . The chip was manufactured from a silicon substrate by a complex fabrication technique, including photolithography and chemical treatment, and is only available in specialized laboratories.
The capability of RPA technology for SNP genotyping remains an unsolved issue. A recent study evaluated the influence of sequence mismatches on the amplification specificity of closely-related pathogens . A proof of concept assay describes the discrimination of a single-point mutation of the HRAS gene . To this end, DNA extracted from cell lines was selectively amplified and genetic variants were distinguished by measuring the wavelength shift on silicon micro-ring resonators. However, this technology is far from being adopted generally in health centers.
In the present study, the capability of allele-specific RPA (AS-RPA) is evaluated and a low-cost method is developed for the SNP genotyping of clinically relevant polymorphisms. In a first approach, amplification is performed in conventional polypropylene vials in a single format (detection of a single polymorphism). The advantages of this disposable format are its low cost and compatibility with a huge number of laboratory equipment. However, lab-on-a-chip or μ-total analysis systems offer important advantages for diagnostic devices, such as high-throughput and miniaturization, among others .
There are many ways to microfabricate plastic-based materials; e.g. laminate, embossing or injection molding. Additive manufacturing is being examined given its growing interest in the microfluidics field . It is capable of producing customized structures that range from a few microns to several centimeters in a single step. The main limitations of 3D printers are related to spatial resolution, dimensional fidelity, surface quality, biocompatibility, optical transparency, among others . Advantages include low infrastructure costs and easy manufacturing compared to photolithography or soft lithography approaches. In order to evaluate the potential of this technology, a well array chip for performing AS-RPA was designed and developed with a commercial 3D printer. The objective was to demonstrate the capability of RPA as a genotyping method and to compare its performance in an advanced platform compared to the standard format (vials).
As proof of concept, the solution was applied as a pharmacogenomic tool to treat smoking cessation and the highly addictive properties of nicotine [25, 26]. In such diseases, the benefits of personalized medicine based on genotype populations has been demonstrated. However, the high cost of current genotyping technologies, compared to the cost of ineffective or erroneous treatment, compromises the application of the test. Therefore, the approach was designed by following the analytical quality and health system sustainability goals.
Material and methods
The target polymorphisms for the tobacco use disorder were rs4680 (COMT gene), rs1799971 (OPRM1 gene), rs1800497 (ANKK1 gene) and rs16969968 (CHRNA5 gene). The wild-type variants are G, A, G, and G, and the mutant variants are A, G, A, and A, respectively. The pharmacogenomic information about these variants is included as Supplementary Material (Tables S1, S2 and S3). The human beta actin (ACTB) gene was selected as an endogenous control. The list of oligonucleotides for the genotyping of each SNP is found in Table S4.
Patient samples and reference discrimination method
Subjects (n = 17) were recruited for the present study according to ethics and with informed consents. Buccal smear samples were collected using sterile swabs. They were submitted to digestion and purification steps with a PureLink Genomic DNA Mini Kit (Invitrogen, Thermo Fisher Scientific, USA, www.thermofisher.com). The isolated DNA extracts were eluted from the spin-columns of the kit with Tris-HCl buffer (10 mM Tris, pH 8.6) and stored at −20 °C until analyzed.
Allele-specific PCR in a single format was applied as the reference genotyping method. Briefly, the extracted genomic DNA (4 ng) was amplified using two PCR master mixes (Biotools, Spain, www.biotools.eu) and employing 300 nM of each variant primer pair (reverse and allele-specific forward). To confirm amplification, products were diluted in 0.5× SyBR Safe (Invitrogen, Thermo Fisher Scientific, USA) and submitted to fluorescence measurements in a plate reader (Victor 3TM V1420, PerkinElmer, Finland, www.perkinelmer.com). Genotypes were also verified by 3% agarose gel electrophoresis, followed by dying with an intercalating agent (Real Safe, Durviz, Spain, www.durviz.com) and visualization under UV light.
Fabrication of structured platforms
The 100-well array chip was modeled with the Autodesk Inventor Professional 2015 software (Autodesk, USA, www.autodesk.com) and fabricated with a 3D printer (Ultimaker 2 Extended, UltimakerB.V., the Netherlands, www.ultimaker.com). Polylactic acid (PLA) filament (RS Pro, Spain; 2.85 mm diameter, www.rs-online.com) was employed as the printing material. Fabrication was carried out using a 0.4 mm diameter nozzle at 210 °C and a bed operation temperature of 60 °C. In order to evaluate the best printing conditions, prototypes were fabricated with different layer thicknesses (up to 0.2 mm) and at various printing speeds (up to 300 mm⋅s−1). Subsequently, printed structures were cleaned with a 30-min ultrasonic bath and dried with compressed air.
The fabrication quality of the PLA-chips was monitored by optical microscopy imaging. Surface pictures were captured (1.2× magnification) by an Olympus SZ61 stereo microscope (Olympus Co., Japan, www.olympus.com). Images were analyzed with the Image J software to provide an estimated roughness for each sample. Surface hydrophobicity was estimated from the contact angle data. Measures of the deionized water droplets (4 μL) were taken using a Dino-Lite Digital Microscope (AnMo Electronics Co., Taiwan, www.dino-lite.com) at the 1.3 megapixel resolution.
A mass loss study was also done to evaluate possible sample evaporation on the platform. The chip was loaded with 10 RPA samples (4 μL each) and placed in an oven at 37 °C. Mass measurements of the set were periodically taken during 4 h and compared with those of an unloaded reference chip.
Assay protocol: amplification
The amplification step was performed using a TwistAmp Basic RPA kit (TwistDx, UK, www.twistdx.co.uk). Eight allele-specific mixes (2 per SNP) were prepared with rehydration buffer, 14 mM of magnesium acetate, 480 nM of allele-specific forward primer and reverse digoxigenin-labeled primer, and the enzyme pellet. Mineral oil (8%) was also added to minimize sample evaporation. Solutions were loaded onto the 100-well array chip and the DNA template (2.56 ng) was added to allow the simultaneous amplification of eight different allelic variants for 10 patient samples and controls (human ACTB gene). The chip was then covered with a polyester plate sealer (Corning, USA, www.corning.com) and gently vortexed to mix reagents and samples. Amplification was carried out in a heating oven (Memmert UF30, Germany, www.memmert.com) at 37 °C for 40 min.
The AS-RPA reactions were also performed in 0.2 mL-polypropylene vials (Labbox, Spain, www.labbox.com) and polycarbonate home-made array chips. These chips were fabricated using a computer numerical control drilling machine (Bungard CCD, Karo 5410, Germany, www.bungard.de). The feed speed and rotational rate of the drill were respectively 2000 mm·s−1 and 48,000 rpm. The diameter of each well was 5 mm and their depth was 1.1 mm. The composition of the RPA mixtures was the same as that previously described, but volumes were 25 μL and 4 μL per reaction for vials and chips, respectively. The heating system used was a thermocycler (TC-4000, Techne, UK, www.techne.com) and an oven, respectively.
Assay protocol: detection and data analysis
The AS-RPA products were detected by a hybridization assay on polycarbonate chips, adapted from reference . Briefly, the mixtures of the wild-type or mutant products for all four SNPs were prepared from the respective single RPA solutions. For this purpose, 2 μL of each amplification product were diluted in 16 μL of hybridization buffer composed of NaCl 225 mM, sodium citrate 22.5 mM, 10% formamide and 2.5× Denhardt’s solution, pH 7. Subsequently, mixtures were heated at 95 °C for 10 min for denaturation and transferred to the chips with the immobilized probes in the microarray format. After 60 min of incubation at 37 °C, chips were washed with diluted hybridization buffer. The immunoreaction protocol with enzymatic labelling was followed to develop the duplex of the probe-RPA product, as described in reference . The oxidized form of 3,3′,5,5′-tetramethylbenzidine (substrate of horse-radish-peroxidase) produced a blue precipitate over the positive or control spots. Chips were then read with a desktop scanner (Epson Perfection 1640SU Office, Epson, Japan, www.epson.com).
The resulting gray-scale images (Tagged Image File Format, color depth 16 bit) were processed by an in-house software for the microarray analysis. The optical intensity signals of each spot and local background were quantified by generating a data matrix of the signal-to-noise ratios. The genotype determination rule was constructed according to the replicated responses of the specific probes for each polymorphism. A discrimination index was calculated from the signal of the wild-type (WT) and mutant (MUT) variants according to this equation: (WT − MUT)/(WT + MUT). The Statgraphics Centurion statistical package for Windows v.16 was used for the data analysis.
RPA capability as a genotyping tool
The use of RPA as an allele-specific amplification technique was analyzed by considering the role of each element in the process. A recombinase (T4 uvsX) recognizes targeted DNA templates and specific primers at a high affinity and catalyzes subsequent homologous pairing and strand exchange . Polymerase produces the correct elongation of the perfect-annealed primer/template, and is the key reaction in the DNA duplication process . Furthermore, the Pol I large fragment (Bsu polymerase) lacks exonuclease activity (3′ → 5′) that may modify the target nucleotide. Therefore, we expected the presence of mismatches on their 3′-extreme to hamper the nonspecific reaction due to the combined action of two enzymes, even at a low working temperature.
Complementarily, oligonucleotide sets were carefully selected to satisfactorily amplify/detect the given template region. The in silico design restrictions were primer length, absence of secondary structures, and primer/template duplex stability. Both these last parameters were estimated from the thermodynamic models available for DNA duplexes . Although the recommended length for RPA primers should be 30–35 bases long according to the manufacturer’s instructions, shorter primers (19–21 mer) were chosen to improve selectivity. The free energy (ΔG) values for the self-annealing and hairpin structures were restricted to 1.0 kcal⋅mol−1 (the equivalent to melting temperatures <50 °C). The selected oligonucleotides produced primer/template duplexes with changes in free energy (ΔG) of −25.6 ± 0.2 kcal⋅mol−1 (the equivalent to a melting temperature of 75.4 ± 0.1 °C) for totally complementary primers. The duplexes between the template and mismatched primers were less stable (−24.0 ± 0.6 kcal⋅mol−1, 72.7 ± 0.6 °C).
Conventional and chemical-modified primers, locked nucleic acids (LNA), were compared for AS-RPA. Figure 1b shows that nonspecific amplification took place for the mismatched duplexes between the primer and templates. Nevertheless, the amplification yield was significantly higher for the totally complementary duplexes (ANOVA, p-value < 0.001). With the LNA primers, differences were more marked, and even nonspecific amplification was similar to the negative controls. These experiments demonstrated that the presence of this nucleoside at the 3′ terminal base improved allelic discrimination.
A multiplex reaction was studied for the simultaneous amplification of more than one target in a single reaction. However, reaction yields were not satisfactory and there were sensitivity losses. One system displayed dominating and/or inhibiting activity over other primers and amplicons, and even genotyping capability was lost. These results agreed with conventional RPA behavior and can be associated with their high sensitivity to the total primer concentration .
Design and fabrication of well array chip
The amplification assays, described in the previous section, were performed in 0.2-mL polypropylene vials. The next step was to reduce the reaction volume by performing the assay in a well array chip. This kind of platforms improves amplification capabilities, particularly high-throughput . Fused filament fabrication was chosen as the additive manufacturing technology, with a biocompatible polymer, e.g., PLA, as the thermoplastic material. This technique was selected because it produces innovative bioanalytical platforms that can be customer-designed and fast prototyped by a 3D printer.
The hydrophilic/hydrophobic nature of the reaction vessel can affect the easy-filling of wells. Hence the wettability of the raw PLA-chips was estimated and the contact angle of well surfaces was measured. The results were 97 ± 4 ° and 77 ± 2 ° for the wall and the bottom surface, respectively. These values indicated how the patterned topography modified the interfacial tension between the liquid and thermoplastic compared to the base material (PLA, contact angle of about 80 °). Two chemically modified chips (UV/ozone irradiation and PEG passivation) were tested. The surface topographies, estimated by microscope image analyses, were comparable to those from the raw chip (p-value >0.05). After applying soft UV/ozone irradiation, the contact angle was 71 ± 2 °. This lower value indicated that some photo-oxidized polar groups on the surface were formed, consequently fiber adhesion increased (higher surface energy). Addition of PEG also produced a coating over all the active surfaces, and hydrophilicity increased (62 ± 3 °). Regardless of the surface treatment, reagent solutions were easily loaded in wells.
The effect of the unpolished surfaces and chemical treatment on the amplification yield was studied. Replicate RPA reactions were performed using genomic DNA for wild-type/mutant patients (order of magnitude: 103 pg of gDNA). Effective amplification was achieved in all the wells of the raw and chemically modified PLA-chips. Nevertheless, the PEG coating was chosen because this treatment can help block chip surfaces, and prevent nonspecific signals and sample losses through protein and amplification product adsorption .
Comparison of amplification platforms
Characteristics of the tested amplification platforms. a.u.: arbitrary units. Data from three replicated assays
Our approach based on reaction vessels was compared with microfluidic chips in virtue of their high applicability as point-of-care systems . The microfluidic platforms, generally based on poly(methyl-methacrylate) (PMMA) or similar polymers, allow a higher degree of assay integration and lower reaction volumes (nanoliter scale). In exchange, the PLA multi-well chip presents easier manipulation, no fluidic control equipment requirements and a simpler fabrication process with a 16-fold lower cost.
This study can open up ways to test PLA-microfluidic devices, e.g., integration of RPA amplification and real-time detection, prior to their mass production in other thermoplastics, such as PC (e.g., injection molding).
Analytical performance of the genotyping assay
Having demonstrated that AS-RPA can be used for SNP genotyping in a single format, the capabilities of a multiplex detection method were studied. Among the techniques currently available (i.e. optical, electrochemical, etc.), AS-RPA on PLA chips combined to a hybridization assay on PC chips was tested. This detection approach showed excellent performance to simultaneously identify several PCR products .
Our detection method of AS-RPA products displayed comparable performance to others previously described for AS-PCR, such as capillary electrophoresis , real-time fluorescence , the fluorescent-based droplet technique  and hybridization to covalently immobilized probes in fluorescent magnetic beads .
Analysis of patient samples
The applicability of chip-based AS-RPA for genotyping screening in tobacco cessation treatment, or for drugs used in nicotine dependence, was investigated. As a biological sample, buccal swab extracts were selected. In a clinical scenario, the use of buccal swabs is noninvasive, less stressful and a much easier technique to collect DNA samples. Moreover, sample storage does not require refrigeration and DNA extraction is a much simpler process than blood samples. As a detection strategy, a desktop scanner was used for chip reading. Then the procedure was performed using low-resource laboratory materials and equipment (i.e. a primary health center). The analysis time was 210 min (DNA extraction: 50 min, amplification: 60 min, hybridization-detection: 100 min).
The absorption measurements indicated that a sufficient amount of high quality amplifiable human DNA was isolated from all the tested samples. The reference (ACTB gene) and targeted regions were amplified and submitted to on-chip hybridization. Fig. S3 illustrates some examples of the microarray images. Presence of mutated variants in rs4680 and rs1799971 was detected. A subsequent comparison of the acquired chip signals with patient stratification based on the reference method clearly demonstrated perfect matching. The clinical implications of the provided genotyping results are a review of drug treatment, including anti-depressives or nicotine replacement products (e.g. patch). Functional polymorphisms in dopamine pathways (rs4680) are associated with the use of bupropion to mitigate lapsing to smoking following a quit attempt . Better prolonged abstinence rates have been reported after the nicotine replacement therapy tailored to each smoker, and based on either genotype in the opioid receptor (rs1799971) .
In the last few decades, patient genome information has been proposed to select individual clinical care, particularly drug treatment decisions. However, the impact of personalized medicine is low compared to the research advances made. The results reported herein study demonstrate how pharmacogenomics knowledge combined with emerging analytical methodologies can benefit clinical practice more broadly. Although more in-depth research must be conducted, the combination of two innovative solutions was a success. Firstly, the advantages of the isothermal amplification reaction were incorporated to acquire the demanded copy number for sensitive SNP loci detection. The best features were their fast-response (4-fold compared to the PCR), low temperature (37 °C) and few design restrictions. However, the reduced multiplexing capabilities forced single parallelized reactions. Secondly, we employed additive manufacturing based on using a 3D printer as the chosen technology to create a customer-tailored platform for high-throughput analyses. The 100-well PLA-chip design considerably reduced reagent consumption and avoided expensive manufacturing processes or complex pumping systems associated with some DNA detection instruments. In fact the assay can be performed with standard materials (i.e. pipettes, oven) found in clinical laboratories. The present work demonstrates that PLA is an adequate material for performing enzymatic reactions in a static format. The following step is to achieve better point-of-care performance and the next challenge is to develop microfluidic devices fabricated with this material that integrate all DNA assay steps, from extraction to detection. The advantages include the method’s flexibility and accessibility compared to other micro-prototyping or micro-fabrication techniques.
Regarding the clinical impact, today pharmacogenomics is applied mainly to certain treatments in psychiatry, oncology and cardiology. One main reason is the cost-effectiveness of genotyping methods. In addition, only some diseases apply to this approach because their treatment generally involves expensive pharmaceutical products or drugs with highly probable adverse effects. With our approach, personalized therapies based on incorporating genetics into patient stratification can be offered, and even for relatively less-impact treatments. The methodology’s cost-effectiveness, flexibility and portability will support the well-known genetic marker for predicting drug responses. In the particular case of tobacco addiction, genotyping information will help predict the degree of success in smoking cessation.
The authors acknowledge the financial support received from the Generalitat Valenciana (GVA-PROMETEOII/2014/040 project and GRISOLIA/2014/024 PhD grant) and the Spanish Ministry of Economy and Competitiveness (MINECO CTQ2013-45875-R project).
Compliance with ethical standards
The author(s) declare that they have no competing interests.