Ultrasensitive non-enzymatic protein detection using proximity immunoassay with photonic resonator absorption microscopy

Abstract

Proximity assays have emerged as a leading technology for protein detection, yet broad adoption is currently limited by complex workflows that include enzymatic amplification, thermocycling, and expensive laboratory equipment for signal readout. In this report, we introduce a Proximity Initiated Nucleic Acid Target Amplification (PINATA) assay that is performed at room temperature with a simple two-step, 90-min protocol using a low-cost detection instrument. In contrast to alternative proximity assays, PINATA applies toehold-mediated strand displacement of nucleic acids for enzyme-free reactions that occur at room temperature to uniquely combine linear amplification and digital detection using Photonic Resonator Absorption Microscopy (PRAM). Utilizing human interleukin-6, we demonstrate a detection limit of 37 fg/ml with 6-log dynamic range, high selectivity against non-target cytokines, and assays were performed with complex sample matrices without significant loss of efficacy. We envision that PINATA can address a range of protein quantitation applications for life science research and diagnostics.

Proteins are a well-established class of biomarkers that provide valuable insights into nearly every disease state^1–5. Historically, the sensitivity of protein detection assays has been inferior when compared to assays for detection of nucleic acids, where exponential amplification techniques that include polymerase chain reaction (PCR)⁶, loop-mediated isothermal amplification (LAMP)⁷, and rolling circle amplification (RCA)⁸ enable detection in the attomolar to femtomolar range. In contrast, protein detection using conventional “sandwich” immunoassays (in which the analyte is captured with one antibody and labeled with a second antibody that carries a tag), such as the most widely used commercially available enzyme-linked immunosorbent assays (ELISAs) typically have limits of detection (LOD) near 1–10 pg/mL.

Two main avenues have been previously explored for ultrasensitive protein detection. One approach focuses upon digital detection of individual proteins to increase sensitivity while providing quantitative accuracy. Assays such as single molecule arrays (SiMoAs) have demonstrated detection of individual proteins, achieving detection limits in the pg/mL range⁹. However, SiMoAs require lengthy and complex assay protocols that are difficult to multiplex while relying upon enzymatic amplification and complex instrumentation. Alternative digital resolution detection technologies have been discussed in various reviews, but few meet the combined criteria for simplicity, instrument complexity, detection limit, and cost^10,11. The second avenue to reduce the detection limits of protein detection assays is to utilize nucleic acid amplification techniques by converting a protein biomarker molecule into a nucleic acid sequence. For example, immuno-PCR uses a sandwich assay approach, where a target protein is immobilized on a surface between two antibodies, and the detection antibody is tagged with a DNA sequence that is amplified and detected using qPCR^12–14.

Advances in immuno-PCR led to the development of proximity ligation assays (PLAs)¹⁵ and proximity extension assays (PEAs)¹⁶. Proximity-based affinity assays involve two DNA-labeled antibodies that bind to the same protein on different epitopes, which triggers either ligation or extension. Subsequently, nucleic acid amplification techniques amplify the signal for bulk detection¹⁷. Like immuno-PCR, proximity assays utilize enzymatic nucleic acid amplification with antibody-oligonucleotide conjugates to increase signal detection^18–24. PEAs are now widely used for detection in complex media²⁵ and have been commercialized by Olink^® for high-throughput multiplexing with high sensitivity (pg/mL concentrations) and broad dynamic range²⁶, detecting a variety of biomarkers^27–32. Recently, a proximity-based assay called the NUcleic acid-Linked Immuno-Sandwich Assay (NULISA) was reported that achieves low fg/mL LODs for highly multiplexed detection through the use of bead-based PLA to reduce background before quantification by PCR or NGS³³. Despite the many advantages of proximity assays, all previously reported protocols depend on enzymes or sequencing, therefore requiring complicated storage and stability considerations or expensive readout methods. Enzymatic amplification also requires heating or thermocycling, reducing opportunities for applications that include point-of-care detection. Current proximity assay signal readouts rely upon fluorescence-based detection, so achieving high sensitivity requires expensive lab equipment (with laser excitation and sensitive photodetectors) and large centralized laboratories³⁴. A small number of non-enzymatic proximity approaches have been reported, such as DNA entropy-driven cascade reactions using proximity-driven hybridization chain reactions (HCR) or catalytic hairpin amplification (CHA)^35–42. However, proximity methods without enzymatic amplification have not achieved rapid assay times combined with high sensitivity.

In previous work, we described a nanoparticle-labeled method for digital biomolecule detection called photonic resonator absorption microscopy (PRAM), in which gold nanoparticles (AuNPs) bound to a photonic crystal (PC) biosensor surface are be visualized and individually counted⁴³. As described in prior publications^43–48, the detection principle of PRAM utilizes the resonant PC reflection at a wavelength of λ = 625 nm to provide a high reflected intensity from collimated low intensity LED illumination of the same wavelength into a webcamvariety image sensor. The AuNPs are strategically selected to provide strong absorption by localized surface plasmon resonance at the same wavelength⁴⁹. PRAM imaging offers an attractive alternative to detection of fluorophore tags, as PRAM requires only low power broadband illumination with an LED and can provide time-course data. This makes AuNPs a powerful imaging contrast agent for visualizing the presence of biomolecules without background, photobleaching, blinking, or quenching. We also recently described a portable version of the PRAM instrument with a cost of components totaling $7000⁴⁹, as well as a textbook-size “PRAM Mini” with a single-unit cost of <$1000 for point-of-care environments⁵⁰.

The assay technique called “Activate Capture and Digital Counting” (AC + DC) is a single-step, one-pot assay protocol that is performed simply by placing the test sample and antibody-conjugated AuNPs into a well with the PC biosensor (prepared with the capture antibody) on the bottom surface. Through digital detection with PRAM and AC + DC, we demonstrated rapid detection (1 h) of HIV p24 antigen (LOD of 1 pg/mL)⁵¹ and COVID-19 IgG (LOD of 27 pg/mL)⁴⁶ when detecting a protein in a sandwich assay format, where each protein biomarker is labeled with one AuNP. We also created a sandwich-based digital assay that achieved sub-pg/mL limits of detection for human interleukin 6 (IL-6) through digital counting⁵². In a sandwich assay, the protein target is incorporated within an immunocomplex, where a single antigen can only bind a single nanoparticle tag. While digital detection allows increased sensitivity, the assay time is limited by diffusion, binding rates, field-of-view, and background noise. Recently, we demonstrated the Target Recycling Amplification Process (TRAP)⁵³. The TRAP method reduced assay times and LODs compared to sandwich-based methods, as the target is recycled and linearly amplified to bind AuNPs onto the PC for counting with PRAM. TRAP also showed a 60-fold lower limit of detection for microRNA when compared to PCR, demonstrating that an amplify-then-digitalize approach results in rapid, ultrasensitive, room temperature, and enzyme-free detection that can challenge traditional amplification techniques⁵⁴.

In this work, we present an approach that, for the first time, combines TRAP with a proximity assay for a protein target. Our detection principle is based upon DNA molecular circuits that have been studied for DNA computing applications^55–59. We demonstrate that by creating nucleic acid circuit where an action only occurs by the presence and proximity of two antibody-oligonucleotide (AbO) probes, two simultaneous toehold reactions transduce a protein into a synthetic single-stranded DNA reporter sequence. The reporter is linearly amplified and detected using the TRAP approach with PRAM-based digital readout⁶⁰. By combining non-enzymatic signal amplification with digital readout, the Proximity Initiated Nucleic Acid Target Amplification (PINATA) assay offers a room-temperature, wash-free platform for fast and ultrasensitive protein measurement. We characterize the performance of PINATA by detection of human interleukin-6 (IL-6) as a representative assay that can be bench-marked against other technology platforms and assay methods. The PINATA assay achieves a limit of detection of 37 fg/mL for IL-6, shows excellent selectivity against other cytokines, and a 6-log broad dynamic range from 0.1 pg/mL to 100 ng/mL. The PINATA assay was tested for use for IL-6 spiked into plasma and serum samples, displaying no loss of signal efficiency in complex matrices, and the assay was validated against a commercially available ELISA using IL-6 from HeLa cells. The results highlight the power of entropy-driven digital proximity assays for rapid and ultrasensitive protein detection.

Results

Design of Proximity Initiated Nucleic Acid Target Amplification (PINATA)

The PINATA method (Fig. 1) transduces a target protein into a short single-stranded DNA reporter sequence using two antibody-oligonucleotide conjugates (AbOs) and dual simultaneous toehold-mediated strand displacement reactions. The bridge and reporter sequences (black and red sequences, respectively, in Fig. 1a) are initially annealed to form a stable duplex, with six-nucleotide exposed toehold regions on each end. Proximity probeswere created through antibody-oligonucleotide conjugation to create DNA-reactive affinity targets that can react with the toehold regions, shown by blue and pink complexes in Fig. 1a. Alone, or in the presence of one or both proximity probes in solution, the bridge-reporter duplex is stable, and no reporter will be displaced, so it will not trigger further amplification. If both proximity probes are brought into close vicinity by a target protein, the simultaneous toehold-mediated displacement reactions between the proximity probes and the bridge sequence cause the reporter to be released (Fig. 1b). Next, the ssDNA reporter sequence activates target recycling through additional strand displacement reactions, allowing a single reporter (red) to bind multiple AuNPs to a PC biosensor surface for digital counting (Fig. 1c) by PRAM. The reporter initially interacts with an exposed toehold region on the linker strand (green), displacing the protector strand (blue), which subsequently exposes a second toehold region on the linker that was initially sequestered, allowing the probe sequence conjugated to an AuNP (light purple sequence in Fig. 1c) to begin a second toehold-mediated strand displacement reaction. The final displacement reaction strips off the reporter, allowing a single reporter to enact target recycling to bind many nanoparticles. If no reporteris released into the system, no nanoparticles can bind, indicating the absence of the target protein. The nanoparticles are then imagedusing the PRAM instrument (Fig.1d), which shows an image of each AuNP bound to the PC as a distinctive dark dot. The PRAM images are processed by an AuNP-counting image processing algorithm, as shown in Fig. 1e.

Fig. 1 |. Schematic of the proximity-initiated nucleic acid target amplification (PINATA) assay.

a A dual toehold design on an annealed bridge-reporter duplex has complementary regions to antibody-oligonucleotide proximity probes. b When a target protein is present, the reporter is released through simultaneous toehold-mediated strand displacement reactions, converting the protein into a single-stranded DNA reporter sequence. c Linear amplification of the reporter sequence occurs through toehold-mediated strand displacement reactions with linker-protector duplex and ssDNA probes conjugated to AuNPs. d Photonic resonator adsorption microscopy is used to visualize each bound AuNP. e Image processing is used to quantify and digitally count nanoparticles in PRAM images, with raw, processed, and counted PRAM images shown.

Testing and validation of the reporter toehold-mediated amplification circuit

Amplification relies on multiple toehold-mediated strand displacement exchanges to recycle each reporter sequence to bind multiple nanoparticles. We designed four different reporter lengths that could be amplified using toehold-mediated strand displacement reactions, with amplification sets for bridge-reporter 4, 5, 6, and 7 that included a linker substrate, a protector, and a nanoparticle probe sequence. The number refers to the number of nucleotides on the bridge-reporter sequence in the middle tuning region “c” (Fig. 1a) that are not displaced by the proximity probes. “BR4” refers to a bridge-reporter duplex with a 4-base region that is not displaced, and “BR7” refers to a bridge-reporter sequence with a 7-base region between proximity probes. Each reporter reacts with a 5-nucleotide toehold-1 region that is initially exposed on the pre-hybridized linker-protector duplex (Fig. 2a). The protector is displaced, exposing the second toehold-2 region, which can then react with the probe sequence that will be conjugated to the PC. The linker strand also has a secondary region that can bind to the capture, forming the final capture-linker-probe triplex that captures nanoparticles on the photonic crystal for detection.

Fig. 2 |. Testing and optimization of the nucleic acid recycling method for the reporter sequences in the PINATA assay.

a In-solution testing of the amplification of the reporter using two sequential toehold-mediated strand displacement reactions. b Native PAGE of target recycling amplification process (TRAP) sequences. Lanes 1, 3, 5, and 7 are a negative control, with all TRAP sequences (capture, linker-protector, and NP probe) added, excluding reporter. Lanes 2, 4, 6, and 8 are a positive control and contain all TRAP sequences as well as one-tenth equivalent of the reporter sequence for reporters 4, 5, 6, and 7 to test recycling. The formation of the capture-linker-NP probe (C-L-Pr) triplex indicates a final bound nanoparticle in the PRAM system. c Testing of the pre-annealed bridge-reporter (BR) duplex to determine that duplexes cannot activate recycling. Native PAGE was run after reaction of capture, LP, NP probe, and BR for 4 h at room temperature. Blue arrow indicates region of final bound triplex, with little obvious signal leakage even after 4 h. d Testing of the reporter sequence detection using probe-conjugated AuNPs and capture-functionalized PCs on PRAM for quantification of the reporter 5 (R5) sequence at 10, 20, and 30 min (n = 3, with single images taken in triplicate across three experiments, standard error shown).

Each nucleic acid reporter set went through a series of validation experiments. To test initial amplification of each reporter, the sequences were reacted at stoichiometric equilibrium for 1 hat room temperature, with no reporter added for the negative control, and a one-tenth equivalent of reporter included for the positive control. Samples were then run on a native polyacrylamide gel (PAGE) to examine binding (Fig. 2b). Each sequence set was successfully able to amplify the positive control, showing high formation of the final capture-linker-probe triplex, while the negative lanes showed no triplex formation. Sequence sets were also tested in incubation over a 4-h period at room temperature, then run in native PAGE, to ensure that annealed bridge-reporter would not trigger non-specific amplification without reporter release (Fig. 2c). Finally, the amplification set for reporter 5 was validated using conjugated nanoparticles, functionalized PCs, and PRAM (Fig. 2d). The nanoparticle probe sequence was conjugated to gold 80 nm nanoUrchins (AuNPs) that strongly absorb red light (Fig. S2). The capture sequence was immobilized on red-reflecting photonic crystals (Fig. S1) using silanization to bind the AuNPs to the PC for digital counting. Images were taken by PRAM after 10, 20, and 30 min on the PRAM system. The assay demonstrates rapid detection of the R5 sequence and can detect as low as a 1 aM concentration in 30 min at room temperature with a broad dynamic range.

Design and optimization of the proximity-initiated nucleic acid target amplification (PINATA) assay

As an initial demonstration of PINATA, streptavidin and biotinylated DNA sequences were used as a protein-antibody mimic to assess sets of four different sequence options. When no streptavidin is present, the reporter should not be released from the bridge and cannot amplify. When streptavidin brings the biotinylated proximity probes together, the reporter sequence is displaced and subsequently amplified. To increase the rate of reporter release, the tuning region ‘c’ of the bridge-reporter duplex was shortened, which makes it more thermodynamically favorable for reporter release to occur (Fig. 3a). To decrease the rate of release, the tuning region was lengthened, creating more paired nucleotides that are not displaced by either proximity probe, strengthening the stability of the bridge-reporter duplex (Fig. S5). Results were validated using native electrophoresis PAGE. There was obvious amplification when no streptavidin was present for the BR4 set (Fig. 3b, Lane 1), indicating that the proximity probes were undergoing dual-strand displacement reactions in solution, without requiring proximity of the probe sequences. There was also an extremely low positive signal response to streptavidin for BR7 (Fig. 3b, Lane 8), so BR4 and BR7 were eliminated from further testing. BR5 and BR6 both had a positive signal response to the presence of streptavidin, and no amplification when both proximity probes were present in solution. The BR5 set had much greater positive amplification (Fig. 3b, Lane 4), indicating they were good sequence options for one-pot proximity detection. The bridge-reporter duplexes were purified using magnetic beads to remove any excess unbound bridge sequences and to increase released reporters, confirmed with native PAGE (Fig. S4). Purified BR duplexes and biotinylated proximity probes were additionally tested with the PRAM system and nanoparticles, which was able to detect streptavidin concentrations as low as 10 fg/mL (Fig. S7).

Fig. 3 |. Initial biotin–streptavidin testing of the system to thermodynamically tune the reporter release.

a DNA reporter release was examined using biotinylated sequences with streptavidin to test proximity probe design. BR5 is shown, with five nucleotides in region “c”. b Native PAGE of proximity probes and target recycling amplification process for four different bridge-reporter lengths, reacted for 1 h at room temperature. Odd lanes represent a negative control without streptavidin. Even lanes show positive control with streptavidin added. The blue arrow shows the final amplification product triplex of a bound AuNP (capture-linker-probe).

PINATA for human IL-6 detection using antibody-oligonucleotide conjugates in native PAGE

For further testing, the adapter sequences were conjugated to monoclonal mouse, monoclonal rabbit, and polyclonal goat human IL-6 antibodies using AbO conjugation kits. The AbO conjugates were then hybridized to the proximity probes, with monoclonal mouse to proximity probe 1, monoclonal rabbit to proximity probe 2, and the polyclonal goat AbOs hybridized to either P1 or P2. Monoclonal and polyclonal AbOs were compared for human IL-6 detection using the BR5 and BR6 probe sets through Native PAGE (Fig. S9), along with added nucleic acid target recycling probes. Lanes 1–3 of Fig. S9 used monoclonal AbO proximity probes (mAbOP1 and mAbOP2), while Lanes 4–5 used polyclonal AbOs. The monoclonal AbO proximity probe set demonstrated high signal amplification for BR5 through the presence of the final bound nanoparticle complex (capture-linker-NP probe) when human IL-6 was present, even when tested at a tenfold and hundredfold lower concentration than the initial streptavidin optimization (Lanes 2 and 3) in Fig. S9. It also did not have background amplification in the negative (Lane 1). However, the polyclonal AbOs had very slight triplex formation in the negative control (Lane 4) and lower positive signal (Lane 5) in Fig. S9a, indicating there may be some cross-reactivity with the unpurified probes. BR6 showed comparable results with lower positive amplification, as fewer reporters were released, shown in Fig. S9b.

Initial testing for IL-6 detection initially showed some potential non-specific amplification (Fig. S9a, Lane 1), so the AbO conjugates were purified using magnetic beads to remove any of the excess unbound proximity probes to minimize potential reactions with the bridge-reporter duplex that could trigger amplification without a target present. The purified conjugates were tested using native PAGE for both monoclonal and polyclonal antibody-oligonucleotide conjugates (Fig. 4), using a twofold equivalent of purified bridge-reporter duplex BR5 to capture, linker-protector, and probe. The AbO conjugates were added at aone-fifth equivalent, and the IL-6 target at a one-tenth equivalent. Both monoclonal and polyclonal antibodies demonstrated a good signal-to-noise ratio in native PAGE, although the monoclonal set had a higher positive amplification response. Further optimization using PINATA with PRAM was continued with the purified monoclonal antibody–oligonucleotide conjugates. To balance efficient labeling with preservation of antibody functionality, we targeted a DNA/antibody ratio of approximately 1.5 oligonucleotides per antibody, consistent with commonly adopted practices in proximity-based and amplification-driven immunoassays⁶¹. In this assay architecture, the presence of at least one oligonucleotide per antibody is sufficient to enable signal generation, and modest heterogeneity in conjugation stoichiometry is tolerated without significantly affecting sensitivity. Higher labeling densities can increase the risk of steric effects or unintended cross-linking, which may elevate the background.

Fig. 4 |. Native PAGE testing of PINATA sequences for human IL-6 detection using the purified BR5 sequence set.

Monoclonal and polyclonal antibody-oligonucleotide (AbO) conjugates with adapter sequences were compared for proximity probe use. All target recycling DNA sequences (capture, linker-protector and probe) were reacted at a concentration of 100 nM. The monoclonal and polyclonal AbOs (mAbOP1, mAbOP2, pAbOP1, pAbOP2) were added at a concentration of 25 nM and were preannealed to the two proximity probe sequences at a concentration of 50 nM. Excess unconjugated proximity probe sequences were removed using magnetic beads. Lanes 1 and 2 show a positive and negative control with and without a free reporter for target amplification. For a positive control, a 10 nM concentration of human IL-6 was tested in Lane 4 and 6, for monoclonal and polyclonal antibody oligonucleotide conjugates. Lanes 3 and 5 show a negative control without the IL-6 target. The AbOs, BR, and target were incubated for 1 h at room temperature, then all sequences were added to react for 2 h at room temperature. Sequences were then run on a Native PAGE.

PINATA assay for IL-6 with digital detection using photonic resonator absorption microscopy

The PINATA assay was then adapted for digital detection of IL-6 using nanoparticle labels on photonic crystals with PRAM, where the high absorption of the spiky nanoUrchin AuNPs coupled with the guided resonance of the PC allowed digital detection. We used a two-step assay; the proximity probes were allowed to first react with the target IL-6 and the bridge-reporter duplex for 1 h at room temperature (Fig. S6), and then all sequences were added with the LP and AuNP-Pr onto the capture-functionalized PC surface. The PC was imaged after 30 min at room temperature. The IL-6 assay was evaluated with concentrations ranging from 1 fg/mL to 100 ng/mL (Fig. 5a). Images are shown for the processed and counted images of select concentrations (Fig. 5b). The limit of detection for IL-6 with the PINATA assay was calculated to be 37 fg/mL by considering the concentration at the threshold of three standard deviations of the background above the average of the background (3σ + blank).

Fig. 5 |. IL-6 detection with PINATA using photonic resonator absorption microscopy.

a Standard curve for the detection of human IL-6 on PRAM using PINATA BR5 probe set with monoclonal antibodies. b Representative nanoparticle counts for different concentrations of IL-6 in PRAM images. c Selectivity of the PINATA assay for IL-6 against human IL-2, IL-10, and human serum albumin. IL-6 shown is 10 pg/mL, while other targets were tested at 1000-fold higher concentrations at 10 ng/mL (n = 3, standard error shown). d Detection of a 10 ng/mL concentration of IL-6 in serum and plasma using the PINATA assay (n = 3, standard error shown). e ELISA for testing dilutions of cell media from HeLa cells stimulated with TNF-beta (n = 3, standard error shown). The dotted line shows the background signal with no added cell media. f PINATA for testing of human IL-6 from HeLa cell media (n = 3, standard error shown).

To examine the potential of applying the PINATA assay for clinical samples, we tested the assay use for IL-6 detection in complex matrices, demonstrating detection of IL-6 at 10 ng/mL spiked into commercial plasma and serum (Fig. 5c), which showed slight increases in backgrounds for plasma and serum experiments, but little loss of signal efficiency in positive results. To examine the selectivity of the PINATA assay, we tested the IL-6 AbO proximity probes against two different cytokines, human interleukin-2 and interleukin-10, as well as human serum albumin (Fig. 5d). The non-target cytokines and human serum albumin were added at a concentration of 10 ng/mL, while the IL-6 sample shown was tested at 10 pg/mL. The PINATA assay showed high selectivity against all three, with IL-10 showing the greatest increase in nanoparticle count (62 NPs/image), but still non-significant from the background (42 NPs/image). We also tested human IL-6 from TNF-alpha-stimulated HeLa cell media, which was validated against commercial ELISA results. The human IL-6 ELISA detected a 7680 pg/mL concentration at a 25-fold dilution and a 7350 pg/mL concentration at a 125-fold using comparison to a standard curve (Fig. S10) when considering dilution to find the original concentration of IL-6. The human IL-6 PINATA assay tested the same cell media solution as 7840 pg/mL, 6460 pg/mL, and 8080 pg/mL when tested at dilutions of 4-fold, 200-fold, and 10000-fold, respectively, when considering dilution. This demonstrates that PINATA can achieve accurate quantification of IL-6 even at concentrations down to ~8 pg/mL.

Discussion

Protein biomarkers for cancer, cardiovascular disease, sepsis, traumatic brain injury, neurological disease, and more have been thoroughly validated, providing early screening, predictive monitoring, and personalized therapeutics^2–5. In the broad field of infectious diseases, antigen and antibody detection are effective tools used to identify and control viral and bacterial pathogens, where rapid pathogen testing has demonstrated the utility to improve bedside diagnosis and treatment⁶². The discovery and expansion of non-traditional affinity groups such as aptamers, peptides, single-chain fragment variables, and nanobodies are centering affinity-based protein techniques as powerful, flexible tools for molecular biology and biosensing^63–65. Capability for reduced detection limits for protein assays leads to more rapid assay protocols, utilization of lower volume specimens, and more robust signal-to-noise even when the protein biomarker concentration is well above detection limits. With a simple, inexpensive detection instrument, we envision broader adoption of protein detection that can be performed longitudinally for point-of-care diagnostics that are not possible with existing methods¹⁰.

We present an enzyme-free ultrasensitive method of protein detection using entropy-driven toehold-mediated strand displacement reactions. Here, we focus on digital detection of proteins using the PINATA method with strand displacement nucleic acid amplification. PINATA shares similarities to droplet digital PCR, but without requirements for droplet partitioning, thermal cycles, or enzymes. The same nucleic acid circuit approach used to release the reporterin PINATA could be used with various other standard DNA amplification techniques, both enzymatic and non-enzymatic. The PINATA approach could also be applied to various other reporting systems and labels, digitally and non-digitally detected, such as quantum dots, fluorophores, magnetic nanoparticles or beads, electrochemical sensors, barcoding, or other optical sensing methods. Digital detection methods will achieve the highest sensitivity, but speed or ease of use may be improved by using bulk signal detection.

PINATA could be adapted with a variety of affinity groups for the detection of different targets. Antibodies, aptamers, nanobodies, and other affinity groups could all be applied to protein detection and quantification. Such proximity assays can also be used to study protein-protein interactions and kinetics, or to detect various posttranslational modifications in a point-of-care setting. Using a single DNA sequence-selective proximity probe paired with an affinity probe, the assay could examine protein-RNA or protein-DNA interactions. If using two DNA sequence-specific proximity probes, the assay could be applied to detect nucleic acid targets that could be particularly aided by the requirement of proximity annealing, such as messenger RNA splice variants, circular RNAs, or modified RNA sequences, which are challenging to detect with traditional nucleic acid amplification techniques. Overall, with the sensitivity and rapid assay times demonstrated, we expect the PINATA assay to be a powerful tool for diagnostics once further optimized. The combination of room temperature assay conditions, linearly amplified signal, and the sensitivity of digital detection creates a flexible diagnostic tool for various affinity targets.

Methods

Oligonucleotide–antibody conjugation for human IL-6 detection

Oligonucleotide conjugation kits (Abcam, ab218260) were used to attach 5’ and 3’ single-stranded amine-terminated HPLC-purified adapter oligonucleotides to monoclonal and polyclonal antibodies. Polyclonal antibodies were evaluated at the lower end of the manufacturer-recommended concentration range typical for rapid, kit-based assay deployment, whereas monoclonal antibodies were tested at the upper end of their recommended range to assess maximal performance within this assay architecture. Antibodies were used at a concentration of 1 mg/mL (0.5 mg/mL for polyclonal) in a 100 μL volume of 1× PBS, which was added to the antibody activation reagent vial and incubated for 30 min at room temperature. The ssDNAamine proximity probes were activated at a concentration of 100 μM (dilutedin the provided washbuffer) in the oligo activation reagent vial, then incubated 30 min at room temperature. Then, both the activated oligos and antibodies were desalted using the separating columns provided in the kit: columns were equilibrated with 3 mL of wash buffer, then flow-through was discarded, which was repeated five times total. Then, the 100 μL of activated oligo or antibody was added to the column and allowed to absorb. Totally, 550 μL of wash buffer was added to the column and allowed to completely absorb. Finally, 300 μL of wash buffer was added, and the eluate was collected. To make sure that all antibodies are bound to at least one DNA molecule, but to avoid multiple sequences attaching to multiple locations on the antibodies, we aimed for a DNA/antibody ratio of 1.5. We used a 40 μM concentration of DNA⁶¹, along with 300 μL of the activated protein and 260 μL of wash buffer. The activated oligos and antibodies were mixed and incubated for 1 h at room temperature to conjugate. To purify unbound conjugates, an equal volume of the provided conjugate clean-up reagent was added and incubated on ice for 20 min. The AbO mix was then centrifuged for 5 min at 15k rcf, and the supernatant was removed. The pellet was resuspended in 100 μL of the provided antibody suspension buffer. The adapter oligonucleotides were then hybridized to the chosen proximity probe in saturation, and excess proximity probe sequences were removed using magnetic Dynabeads.

Magnetic bead duplex purification of bridge-reporter duplex and proximity probes

The bridge-reporter duplexes and the antibody-adapter-proximity probe complex were purified using streptavidin-coated magnetic Dynabeads. The bridge-reporter duplexes were initially annealed at a 1:1.2 ratio with bridge in excess. The excess bridge sequences were then purified using Streptavidin MyOne C1 Dynabeads (Thermofischer Scientific, #65001) conjugated to the biotinylated BR purifier for that pair (sequences shown in Table S1), which reacts to the excess bridge without displacing the reporter on properly annealed bridge-reporter duplexes, as it cannot activate either of the toehold regions. The biotinylated DNA purifier sequence was first saturated onto the beads according to the Dynabeads standard procedure using a biotinstreptavidin reaction, then washed four times using a magnetic stand to remove excess unbound purifiers. Totally, 100 μL of beads were reactedwith 300 μL of the BR duplex for 1 h at room temperature. The beads were then removed by magnet, removing any bridge sequences that were not annealed to a reporter. Successful purification was verified by native PAGE results (Fig. S4). Purification for assay bridge-reporter sequences was repeated three times. The excess proximity probes from the antibody–oligonucleotides conjugates that did not bind to the adapter sequence were purified by the same method (Fig. S7), using the proximity probe purifier sequences shown in Table S1.

Gold nanoparticle–oligonucleotide conjugation (AuNP-Pr)

The 80 nm gold nanoUrchins (Cytodiagnostics, #GU-80–20) were initially washed by centrifuging according to recommended times and speeds for 30 min at 600 rcf, then removing the supernatant three times. AuNPs were concentrated to 1 OD, or ~7.8 billion NPs/mL, then were stored in a 0.01× PBS buffer until conjugation. Thiol-modified DNA was reduced using 100 equivalents of TCEP for 2 h at room temperature, then excess TCEP and trityl-SH products were removed using a 0.5 mL 3 K Amicon centrifuge filter (Millipore Sigma, UFC5003) by centrifuging at 14.1 rcf three times for 15 min. The reduced DNA-SH probe sequences were then added to the AuNPs at a final concentration of 100 nM. After 15 min, 1k mPEG-SH (Creative PegWorks) at 40 μg/mL concentration was added as well for surface blocking, then all were incubated for 2 days at room temperature to self-assemble on the gold nanoparticles. Unreacted probe DNA and PEG were then removed through centrifugation, washing three times at 600 rcf for 30 min. Conjugated AuNP-Pr was resuspended at a final OD = 2 in 1×TE, 0.025% tween20, and stored at 4 °C for up to 4 weeks or until use. Properly conjugated AuNPs showed a 1–2 nm increase in radius by DLS (Fig. S3a) and demonstrated a change in the absorbance spectra near 260 nm (Fig. S3b).

Photonic crystal DNA capture surface functionalization

The photonic crystals were first prepared for silanization by sonicating for 2 min each in acetone, isopropanol, and MilliQ^® water to clean. The PCs were dried under nitrogen gas and then heated to 120 °C to completely dry. The PCs were oxygen plasma-treated to create a reactive hydroxylated surface usingthe PicoDiener plasma machine at 0.9 mbar and 100% power. PCs were then immediately added to a 2% (3-Aminopropyl)triethoxysilane (APTES, 99% purity, Sigma Aldrich #440140) solution in anhydrous toluene for 1 h at room temperature, then were removed and washed by sonicating with acetone, ethanol, and then MilliQ^® water for 1 min each. Silanized PCs were dried gently under nitrogen gas, then PCs were cured for an hour at 80 °C and used immediately for capture attachment or stored in a desiccator under nitrogen gas. Gasket wells made from PDMS were attached through light pressure to the PC surface. Disuccinimidyl carbonate (DSC, Millipore Sigma, #225827) was used as a bifunctional linker between APTES and amine-terminated capture DNA and was added to the PC in the wells at 10 mM in 10% DMSO for 20 min at room temperature. The wells were then washed with 1% DMSO, 0.1% DMSO, and water before capture DNA was added at 50 μM for 3 h at room temperature.

IL-6 testing with photonic resonator absorption microscope and PINATA

The target human IL-6 standards (0 pg/mL, 0.1 pg/mL, 1 pg/mL, 10 pg/mL, 100 pg/mL, 1 ng/mL, 10 ng/mL, 100 ng/mL), were added to a reaction tube of 10 μL total volume with 100 pM each of purified monoclonal proximity probe AbO 1 and 2, 25 pM of BR5 in 1×PBS, which were incubated together for 1 h at room temperature to trigger initial reporter release. Capture-functionalized PCs were washed five times with a 1×TE, 0.025% tween20, pH 7.4 washing buffer. PCs were blocked using PBS Superblock for 60 min at room temperature, then washed twice using washing buffer. The proximity probe and standard mixture were then added to the PC surface in separate wells for spatial multiplexing, along with the annealed LP duplex (final concentration of 20 pM) and AuNP-Pr conjugates (final OD = 0.5) in a final buffer of 0.5×TE, 0.5×PBS, 5 mM MgCl₂, and 0.025% tween20 at a 20 μL reaction volume. The PC surfaces were imaged after 30 min, with three images taken per sample well. Testing was repeated in triplicate for each standard, with nine images total.

Oligonucleotide sequence design

Oligonucleotides were designed using the online NUPACK software analysis tool to minimize secondary structures and unwanted cross-reactivity³. The bridge-reporter sequences were chosen by initial testing (not shown) of different toehold and tuning lengths of nucleotides that are not displaced by proximity probes to minimize the background signal when no target was present and to maximize the amount of reporter displaced when the proximity probes are brought into close contact by a target protein. Each proximity probe was designed to be tested with a variety of different bridge-reporter lengths to tune the release of the reporter. After rational probe design through simulation, sequences were tested empirically in native PAGE to determine initial feasibility for reporter displacement and amplification.

Bridge-reporter annealing

Bridge-reporter sequences were initially annealed together at a stoichiometric ratio of 1:1.2 of reporter to bridge, with the bridge sequence in excess. Each bridge sequence hybridizes to its corresponding reporter sequence, i.e., B4 and R4 would be annealed to create the BR4 duplex. To anneal, oligonucleotides were diluted to 5 μM reporter and 7.5 μM bridge in a pH 7.4 1×TE, 12.5 mM MgCl₂ buffer. The oligos were heated to 80°C for 5 min, then slowly allowed to cool to room temperature. BR duplexes were then purified using Dynabeads by the described bead purification method. BR duplexes were stored at 4 °C for short-term use or aliquoted and frozen at −20 °C.

Linker-protector annealing

Linker-protector annealed together at a stoichiometric ratio of 1:1.5 with the protector sequence in excess, at a 500 nM linker concentration in a 1×TE, 12.5 mM MgCl₂ buffer, by heating sequences to 80 °C for 5 min, then slowly allowing to cool to room temperature. LP duplexes were stored at 4 °C or aliquoted and frozen at −20 °C.

Native PAGE

All native PAGE procedures with DNA were loaded with non-SDS 6× purple loading dye (New England Biolabs, B7025S), run on the Biorad Mini Protean Tetra system in a 1×TBE running buffer, stained with Gel Red (Gold Bio, G-720–100), and imaged using the Gel Doc XR imaging system.

Nucleic acid target recycling amplification process feasibility in native PAGE

To verify the target recycling amplification process of the reporter sequence, each reporter sequence was tested with its respective probe, linker, and protector sequences (e.g., R4, L4, P4, Pr4). The same capture sequence was used for all four reporters. The capture, linker-protector, and probe sequences, with reporter added for the positive control, were initially reacted at a 100 nM concentration for 2 h at room temperature in 0.5×PBS, 0.5×TE, 5 mM MgCl₂ buffer, then run on a non-denaturing 15% native PAGE. The final capture-linker-probe triplex, which represents an AuNP binding to the PC, appears in only the positive lanes when the reporter was added, indicating the success of toehold-mediated strand displacement amplification (Fig. S5). The annealed and purified bridge-reporter sequence was also tested to see if it causes non-specific amplification (Fig. S6c), but all showed no amplification after 5-h reaction.

Nucleic acid target amplification testing with PRAM

Capture-functionalized PCs were washed five times with a 1×TE, 0.025% tween20 washing buffer after capture incubation. PCs were then blocked using PBS Superblock (Invitrogen, #37515) for 30 min at room temperature and washed twice using the washing buffer. Annealed LP duplexes (final concentration of 20 pM), nanoparticle AuNP-Pr (final OD = 0.5), and magnesium chloride (final concentration 5 mM), along with the tested reporter concentrations (0 aM, 1 aM, 10 aM, 100 aM, 1 fM, and 10 fM) were added to the 20 μL-containing well on the PC surface in a 0.5×TE, 0.5×PBS buffer. Single PRAM images were captured at 10, 20 or 30 min.

Biotin–streptavidin testing with native polyacrylamide gel electrophoresis

For initial feasibility testing, the DNA sequences were tested using proximity probes terminated with 5′ or 3′ biotin modifications to detect streptavidin (PP1B and PP2B in Table S1). This mimic system of protein-antibody binding allowed more cost-effective initial testing of many potential sequences, without requiring oligonucleotide-antibody conjugation kits. To test each bridge-reporter duplex, the pre-hybridized purified bridge-reporter duplex, and stoichiometric equivalents of PP1B and PP2B were mixed in a 0.5×PBS, 0.5×TE, 5 mM MgCl₂ buffer. For a positive control, one equivalent of streptavidin was added as well. These sequences were reacted for 1 h at room temperature to ensure biotin–streptavidin binding, then the capture, linker-protector and probe sequences for each bridge-reporter length were added. All sequences were then incubated for an additional 1 h at room temperature, and then were run in a 12% native PAGE.

Spiked-in human IL-6 testing in gel electrophoresis

Antibody-oligonucleotide conjugates for monoclonal and polyclonal antibodies for human interleukin 6 were tested in solution with capture, linker-protector duplexes, nanoparticle probes, and annealed and purified bridge-reporter duplexes to examine if the proximity probe AbOs triggered the strand-displacement amplification circuit. A 12% Native PAGE was used to visualize and determine the hybridization and reaction of the various DNA sequences.

IL-6 testing in complex matrices with PINATA

Human IL-6 standard (50 ng/mL) was spiked into commercial human serum (Innovative Research #50-203-6413) or K2 EDTA human plasma (Innovative Research #50-203-6374) samples, which were then diluted to 1/5 by total assay volume for the PINATA reaction to 10 ng/mL. The AbOs and BR5 were reacted with the plasma or serum sample for 1 h at room temperature, then were added to the LP duplex and nanoparticles, and placed on the capture-functionalized PC surface, which was imaged after 30 min. Three images were captured for each test, with tests replicated in triplicate.

HeLa cell culture and treatment with TNF-beta for IL-6 production

The HeLa cell line was cultured in ATCC-formulated Eagle’s Minimum Essential Medium (ATCC 30–2003) with fetal bovine serum (Bio-Techne, S12450H) to a final concentration of 10% and 1× Penicillin–Streptomycin (Gibco 15140122). To induce IL-6 production, 10 ng/mL of Recombinant Human TNF-alpha (Bio-Techne, 10291-TA) was added to the completed medium. The cultured medium was collected at 80% cell confluency and was tested using a commercial ELISA and the PINATA assay with frozen media stored at −80 °C.

Enzyme-linked immunosorbent assay (ELISA) for IL-6 detection from HeLa cell culture medium

IL-6 was quantified from the HeLa cell medium using the DuoSet ELISA (R&D Systems, DY206–05). Ninety-six-well plates were incubated overnight at room temperature with 100 μL of the mouse anti-human IL-6 capture antibody at a concentration of 2 μL/mg. The plate was then aspirated and washed three times with a 1×PBS, 0.05% tween20 washing buffer, which was the washing protocol used throughout the assay. Plates were blocked for 1 h with 300 μL of 0.2 μm syringe-filtered 1×PBS, 1% BSA, then washed. Then, 100 μL of the IL-6 standards or cell culture media were incubated for 2 h at room temperature, then washed. Totally, 100 μL of biotinylated goat anti-human IL-6 detection antibody was added at a concentration of 50 ng/mL and incubated at room temperature for 2 h, then washed, and 100 μL of 1:1 mixture of Color Reagent A (H2O2) and Color Reagent B (Tetramethylbenzidine) (R&D Systems, Catalog # DY999) was added for 20 min, then reaction was halted by adding a stop solution of 50 μL of a 2 N H2SO4 solution (R&D Systems, Catalog # DY994). The plate was imaged using a microplate reader set to a wavelength of 450 nm. Standards for highly purified E. coli-expressed recombinant human IL-6 from R&D Systems were tested from 1200 pg/mL to 9.38 pg/mL by 2-fold serial dilution. Cell culture media was tested across four 5-fold serial dilutions from frozen media.

PRAM image processing and nanoparticle counting

The raw PRAM nanoparticle images were filtered, analyzed, and counted using MATLAB. We first applied a Tophat filter and a Fourier DC component removal algorithm to eliminate uneven background, speckles, and fringes induced by multi-surface reflection or dust diffraction. The noise of the image was suppressed through a simple Wiener filter. An extremum searching algorithm using maximally stable extremal regions (MSER) was applied to recognize individual AuNPs through features such as the threshold of gray scale, size, and circularity. Euler number and equivalent diameters were also considered as criteria to seek the extreme regions generated from AuNPs. To clearly visualize the particle counting process, the detected AuNPs were marked with identical circles on a blank background at their corresponding centroid location (Fig. 1e).

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s44328-026-00090-1.

Data Availability

Additional materials, validation experiments, and data are available in the supporting information. All data are available in the main text or the supplementary materials.