Aptamer selection
Aptamer selection was performed using GO-SELEX38. Briefly, the SELEX selection process begins with a random library of single-stranded DNA (ssDNA). Each 76 base pair ssDNA is comprised of a central 30 base pair sequence, which is randomly distributed, with two ends (23 bp each) of known sequences to serve as primers (SI Table 8). In the first round of SELEX, the ssDNA from the random pool is diluted to 20 ng/µl in water. Peanut protein (or gluten) is mixed in extraction/assay buffer (20 mM EPPS, 0.2% Brij-58, 2% PEG 8000, 2% Pluronic F-127, pH 8.4), 15 nM AF647-P1-16, and 30 mM MgCl2) and diluted to the desired concentration, which is dependent on the round. 100 µl of diluted ssDNA is added to 300 µl of diluted peanut protein, and the resulting mixture incubated at room temperature with shaking for a set amount of time dictated by the round (SI Tables 5, 6). This ssDNA/protein mixture is added to 600µL of graphene oxide (GO), incubated for 20 min at room temperature with shaking to remove ssDNA that is not bound to protein, and centrifuged at 10,000g for 3 min. The supernatant, containing ssDNA bound to the target protein, is cleaned by adding 10% Strataclean resin and heating to 80 °C for 3 min, followed by centrifugation at 10,000g for 3 min. The supernatant is collected, and the Strataclean step repeated. The concentration of ssDNA in the final supernatant is measured and compared to the initial concentration prior to addition of protein and GO, to define percent recovery. The cleaned ssDNA pool is then amplified by PCR using a biotinylated reverse primer using standard cycling procedures and the product cleaned using the ChargeSwitch Pro PCR Clean Up Kit. To regenerate ssDNA for the next round, clean PCR product is added to streptavidin coated magnetic beads (Dynabeads MyOne Streptavidin C1) and base is added to denature the dsDNA. Resulting ssDNA is collected using the magnetic beads and concentrated using an Amicon Ultra 3 k Spin Column. Subsequent round of SELEX are performed on the resulting ssDNA. SELEX rounds performed for peanut protein and gluten are presented in SI Tables 5 and 6, respectively. After SELEX selection, ssDNAs were cleaned using a ChargeSwitch Pro PCR Clean Up Kit (Thermo Fisher Scientific) and sent to deep sequencing (Illumina Miseq, UMass Deep Sequencing Core). A heatmap of the sequences was created using Galaxy (open-source bioinformatics tools), which indicated that the top 40 sequences could be divided into 5 families based on a consistent core sequence (SI Table 7). The top sequence from each of the families was synthesized and tested for binding affinity to peanut protein as shown in Fig. 1.
Aptamer modifications
To prepare aptamers for incorporation into a chip-based assay we added either C-Rich or G-rich supplemental sequences to the 5′ and 3′ ends. The sequences chosen were based on the 5’ and 3’ ends of the selection library and modified to be G/C rich to improve binding to the anchors. These sequences were folded (using IDT oligo-analyzer and M-fold) and analyzed for similarities and significant 3-dimensional structure. From those data, 10–15 sequences were synthesized and screened for Kd based on FP (Fig. 1—data shown for top 5 sequences representing the 5 families). 5′ and 3′ sequences of the selected aptamers are shown in SI Table 8.
Aptamers were modified with fluorophores chosen for compatibility with assays performed throughout the study. All three fluorophores have similar excitation and emission spectrums; Texas Red (625 g/mol, Em.640 nm, Ex. 580 nm), Alexa647 (1155 g/mol, Em.665, Ex 650), and Cy5 (792 g/mol, em. 670 nm Ex. 651 nm). Texas Red was used in FP assays, since it is the smallest molecule with the least interference to Brownian motion, with a shorter relaxation time, and is commonly used in FP studies39. To transition the assay to chip format, we moved towards CY5 which is commonly used in microarray assays. As our development progressed and we aimed to create a consumer product that is stable at room temperature for long periods of time we decided to shift toward Alexa64740, which is more resistant to photobleaching over time.
Affinity measurements
Fluorescence polarization (FP) was used to determine dissociation constants (Kds) for PT-31, P1-10, P1-16, P2-8, and P2-18 interaction with potential targets. The aptamers were synthesized from Integrated DNA Technologies with a Texas Red fluorophore attached to its 5’ end in order to measure changes in fluorescence polarization. Each experiment was performed on a TECAN Spark 10 M plate reader (excitation 570 nm/emission 625 nm) set to 5 kinetic cycles. Samples were prepared in 50 µL with FP buffer (50 mM Tris–HCl, 0.1% Tween-20, pH 9) containing 5 nM aptamer and increasing concentration of purified AraH1 (Indoor Biotechnologies) or peanut matrix (Teddie brand unsalted peanut butter or Protein Plus brand roasted natural peanut flour) ranging from 0 to 50 µM and incubated for 10 min prior to reading on the spectrofluorometer. Peanut matrices were prepared by homogenizing samples at a stock concentration of 100,000 parts per million in FP buffer and clarifying by centrifugation at 5000×g for three minutes. Nonlinear regression analyses were used to determine Kds (Prism 8, GraphPad).
Anchor screening
CY5-labeled aptamers and 40 short DNA anchors with a ten-oligonucleotide sequence complementary to the aptamers were synthesized at Integrated DNA Technologies. Half of the anchor sequences contained a poly-A tail and all anchor sequences contained an amine linker at the 5’ end of the oligonucleotide. Each anchor was spotted on epoxy silane-treated slides at various concentrations (1–40 µM) at Applied Microarrays (Tempe, Arizona). Each slide was pre-blocked with 1% bovine serum albumin in HEPES for two minutes prior to incubation of CY5-aptamer/peanut flour mixture. Aptamer was mixed with peanut flour at different concentrations and allowed to incubate in binding buffer prior to loading onto the well. After a two-minute incubation with mild shaking on an orbital rotator, the slides were washed with binding buffer and scanned for fluorescent CY5 signal. After selection of the 5′-(Amine-6C)-anchor, the linker was extended by six additional carbon atoms and printed at a concentration of 5 µM on epoxy silane-treated slides for confirmation.
Assay details
COP chips were placed in an air-tight chamber with fluidic channels connected to reservoirs from which wash solution or filtered food homogenate were drawn via a pump system. First, 100 µL of wash solution (20 mM Trizma base, 0.2% Brij-L4, 0.2% Capstone FS-31, 0.25 mM MgCl2) was delivered to the chamber, followed by a short air purge. Then, 100 µL of test sample (containing homogenization buffer (20 mM EPPS, 0.2% Brij-58, 2% PEG 8000, 2% Pluronic F-127, pH 8.4), 15 nM AF647-P1-16, and 30 mM MgCl2) was delivered to the chamber at a rate of 1000 µL/min. The chamber was cleared, an image captured, and the intensity of the control spots assessed. If less than 30 rfu, another aliquot of 100 µL of test sample was delivered. The process was repeated until the intensity of the control spots is greater than 30 rfu. Then the chamber was washed with 200 µL of wash solution and imaged.
Guard band study
For the matrix interference studies, 15 nM AF647-P1-16 was incubated briefly with the listed additives and components in homogenization buffer (20 mM EPPS, 2% Pluronic F-127, 2% PEG 8000, 0.2% Brij-58, pH 8.4). The percentage represents the amount in a food sample, meaning for a value of 100%, 0.1 g of component was added to 3 mL of assay buffer. Peanut flour was then added to the 50 ppm samples, and the assay was run as described above.
Specificity studies
AF647-P1-16 aptamer (20 nM) was incubated with increasing concentrations of purified AraH proteins (Indoor Biotechnologies, AraH1, #NA-AH1-1; AraH2, #NA-AH2-1; AraH3, #NA-AH3-1; AraH6, #NA-AH6-1; AraH8, #RP-AH8-1) in assay buffer and 30 mM MgCl2. Commercially available nut flours (pecan, walnut, pistachio, hazelnut, almond, sunflower, and cashew) were homogenized with assay buffer and clarified by centrifugation at 5000×g for three minutes. To study the specificity of P1-16 aptamer when tree nut is present, 0 or 50 ppm peanut flour was spiked into clarified 50 ppm tree nut flour in assay buffer or 0.1% non-fat milk (dry powder, American Bio). Samples were assayed as described for the guard band study.
Matrix testing to validate assay
Chips were printed with both 12.5 µM P1-16 anchor and 7 µM of the control anchor. Foods were sampled at 0.1 g and homogenized for 45 s in 3 mL of assay buffer with 7.5 nM P1-16. For peanut containing samples, 30 µL of 5000 ppm peanut homogenate was also added. Finally, 23 mM MgCl2 was added to all samples. The assay ran as described above. The foods tested were: vanilla ice cream, sugar-free vanilla wafer, gelato, milk chocolate, mint chocolate chip ice cream, nacho cheese, pasta sauce, mushroom soup, sweetened cereal, white chocolate, applesauce, oat cereal, chicken gravy, hoisin sauce, packaged cupcakes, rice noodles, vanilla crispy squares, blue cheese dressing, alfredo sauce, frosting, pink meringue cookie, fluff, marshmallow cereal, sauerkraut, fruit flavored chewy candy, Asian dressing, fruit punch, coconut milk, coffee creamer, flavored tortilla chips, French dressing, corn chips, sweetened cereal, electrolyte beverage, granola, honey, chocolate-covered wafer, mashed potatoes, olive oil, pear baby food, rainbow sherbet, ranch dressing, white rice, shortbread cookie, sweet red chili sauce, tomato soup, chocolate-covered cookie bar, and yogurt.
Gluten assay
GN5 aptamer was incubated in gluten assay buffer (GAB, 15.4 mM MES buffer, 0.08% Tween-20, 30% ethanol, and 1 mM MgCl2, pH 5) with increasing concentrations of gluten. Gluten (wheat source, Sigma Life Science) was extracted in GAB and diluted in GAB with 20 nM GN5. For the food testing, commercially available foods were paired with the closest match for the gluten-free counterpart: wheat round crackers versus gluten-free round crackers (corn starch and rice flour); wheat frosted blueberry toaster pastry versus gluten-free frosted blueberry toaster pastry (rice flour); wheat pretzel sticks versus gluten free pretzel stick (corn and potato starches); country white bread versus gluten-free white bread (pea, tapioca, and rice starches); animal crackers (wheat) versus gluten-free animal crackers (pea and potato starches) were tested with GN5. Each food was prepared as described in the matrix testing description; after filtration the food filtrate was diluted by an additional 1:10 with GAB. For both assays, 250 µL of GAB is delivered to the chamber, followed by a short air purge. Then, 500 µL of test sample (GAB and 20 nM GN5, with or without gluten) is delivered to the chamber at a rate of 1000 µL/min. Then the chamber was washed with 525 µL of GAB with 10 mM MgCl2 with the same flow rate. The chips were then air dried and imaged.
Long-term stability experiments
AF647-P1-16 (10 nM) was formulated in autoclaved homogenization buffer (20 mM EPPS, 0.2% Brij-58, 2% PEG-8000, 2% Pluronic F-127, pH 8.4) under aseptic, sterile environmental conditions within a clean room facility. Aliquots of such samples were subjected to accelerated aging at 37 °C. At each time point, samples were compared with respect to age-matched fresh P1-16.
