DNA Hybridization Kinetics Observed at the Single-Molecule Level Using Graphene Near-Field Effects Authors: Maria M Soares, João Freitas, Tiago Queirós, Agnes Purwidyantri, Pedro Alpuim, Jana B Nieder Published: 2024 ABSTRACT AbstractWe present the development of an advanced sensing platform using a monolayer of graphene functionalized with fluorophore-labeled DNA hairpins to detect the kinetics of single hairpins during the hybridization reaction. The near-field photonic effects of graphene induce a distance-dependent quenching effect on the attached fluorescent labels, resulting in distinct optical signals in response to axial displacements resulting from DNA hybridization. Employing a wide-field Total Internal Reflection Fluorescence (TIRF) optical setup coupled with a sensitive Electron-Multiplying Charge-Coupled Device (EM-CCD) camera, we successfully detected fluorescent signals of individual or a low number of individual DNA hairpins within a low-concentration environment DNA target (tDNA). These signals were used to determine the optical setup’s Point Spread Function (PSF) in a novel approach to super-resolution reconstruction. Combining these techniques, the subpixel localization of single hairpin molecules and their respective intensity profiles were extracted, enabling a kinetic assessment of individual DNA hairpins, with estimated unfolding times of approximately 7 s. Observations of kinetic phenomena unveiled intermediate partially hybridized states, extending the time required to unfold the hairpin probes by more than a factor of 2. Furthermore, a developed semiempirical model allowed the conversion of fluorescent signals into fluorophore–graphene distances. At the nanometer scale, we observed a step-like unfolding process characterized by intermittent metastates of unfolding and static periods, which can be attributed to nucleation events in some cases. Our graphene-based sensing platform and optical methodologies can be adopted for further research into the kinetics of different biomolecules under diverse environmental conditions. Introduction Unveiling the mechanisms behind numerous biological processes involves delving into the dynamics of biomolecules, encompassing their movements, conformational changes, and interactions. As information-carrying molecules, nucleic acids1 are a topic where expanding our understanding is paramount. Among the diverse structural configurations that nucleic acids can adopt, DNA hairpin structures deserve significant attention due to their crucial roles in cellular functions, such as the regulation of gene expression2,3 and DNA recombination.4,5 More recently, these structures, characterized by a single-stranded loop and base-paired stem, have been employed as biorecognition units, exploiting the selectivity of the Watson–Crick base pairing rule in biosensing assays.6−8 Therefore, a deeper understanding of the kinetics of DNA hairpins can provide invaluable insights into the various processes in which they are involved. Over the past decades, DNA hairpin kinetics has been the subject of intense research. A spectrum of techniques, such as fluorescence correlation spectroscopy9,10 and temperature-jump spectroscopy,11,12 has been applied to investigate these structures. Earlier studies13 predominantly relied on a simplistic two-state model to describe their kinetics, where hairpins could exist in either a folded or unfolded state. However, recent studies11,14 propose a more intricate kinetic landscape in which numerous intermediate states for the hairpins arise due to competing unfolding/folding pathways. DNA hairpin structures have also been employed to study the dynamics of the DNA hybridization reaction,15 during which DNA strands bind together to form a fully stretched double-stranded molecule. Exploring the sensitivity of the resonance energy transfer rate with the distance16 between a fluorescent molecule, the donor, and a quencher molecule, the acceptor, Chen et al.17 investigated the kinetics of duplexed DNA formation. While the hybridization of two single strands consists of an entropy-driven nucleation process, in DNA hairpin hybridization, there is an additional energy barrier associated with disruption of the hairpin that must be overcome. This extra energy barrier can, consequently, result in the formation of intermediate states that are momentarily stable. Folded, intermediate, and stretched DNA conformations were successfully distinguished by Adão et al.18 during the DNA hybridization reaction. This study of DNA hairpin kinetics was based on graphene near-field effects: the dependence of the fluorescence lifetime on the distance between the fluorescent molecules and graphene. Gaudreau et al.19 were the first to experimentally quantify the z–4 distance dependence of the lifetime for fluorophores positioned in proximity to graphene. As an intrinsic property of a dye, fluorescence lifetime is less susceptible to parameters other than the quenching effect of graphene; consequently, it has become the main parameter of analysis in numerous experimental works when nanometer resolution along the optical axis is required.20,21 In Adão et al.'s work,18 using a custom Fluorescence Lifetime Imaging Microscopy (FLIM) setup, changes in the lifetime of a fluorophore attached to a hairpin immobilized on graphene were tracked as a result of the axial displacement induced by DNA hybridization. By converting these changes into distances from graphene, they were able to monitor in real-time conformational changes in DNA hairpins along the hybridization reaction in a high target concentration regime. However, these and most studies on DNA hairpin kinetics have been conducted at an ensemble level; consequently, they only provide information about the behavior of DNA molecules as a whole rather than providing insights into the dynamics of each molecule. In this work, we propose to detect and resolve the kinetics of individual fluorophore-labeled DNA hairpins during the hybridization reaction exploring the fluorescence quenching principles of graphene.19 On top of a microscope glass coverslip, we developed a sensing platform by functionalizing CVD-grown single-layer graphene sheets with fluorescently labeled DNA hairpins via a linker molecule (Figure 1). Due to the near-field photonic effects of graphene, which produce a distance-dependent quenching effect on fluorescent molecules, we monitored the behavior of each DNA hairpin. Upon the addition of low concentrations of fully complementary (target) DNA strands, the hairpins are expected to unfold, causing the fluorescent dye to move away from the graphene surface to a maximum distance equivalent to the sum of the linker length and the DNA molecular length. Due to this spatial separation between the dye and graphene, triggered by DNA hybridization, the dye’s fluorescence is restored. A wide-field Total Internal Reflection Fluorescence (TIRF) microscope equipped with a powerful Electron Multiplier–Charge-Coupled Device (EM-CCD) was used to detect, with single-molecule sensitivity, these optical changes over time (Figure 2a). Furthermore, we deployed super-resolution reconstruction techniques using the ImageJ plugin ThunderSTORM22 to precisely locate and extract the intensity profiles of individual DNA hairpins. The nanoscale axial motion of the fluorescent label near graphene was also estimated using a semiempirical conversion of Gaudreau et al.19 model into distance-dependent intensity values. The DNA hairpin structures under study were designed to hybridize with a 29-nucleotide complementary single-stranded DNA sequence from Douro wine grapes. Concept of the proposed graphene-based sensing platform developed on top of a microscope glass coverslip. The platform consists of a 2D graphene sheet functionalized with fluorescently labeled DNA hairpins. Initially, the DNA hairpin places the fluorescent dye in the near field of the graphene, leading to significant quenching through resonance energy transfer. Upon the addition of target DNA in a low concentration, hybridization takes place, and the hairpin structure undergoes a conformational change. The hairpin elongates, distancing the dye from the quenching surface, which results in the unquenching of the dye. These optical changes induced by DNA hybridization were detected with advanced optical techniques to investigate the behavior of each DNA hairpin during this reaction. Localization and extraction of single emitters. (a) Visualization of the raw data, a time-dependent wide-field fluorescence image series. (b) Representative extracted time trace of a single emitter with fully revealed kinetics. The gray period (I) corresponds to the period before the addition of tDNA. The black period (II) represents the time of tDNA addition. The red period (III) corresponds to a “lag” phase. The blue interval (IV) shows an increase in fluorescence intensity attributed directly to the unfolding of the DNA hairpin. The green period (V) is characterized by a stable photon count period expected from the unquenched dye at the maximum elongated distance. (c) Theoretical model formulated by Gaudreau et al.19 for the dependency of the fluorescent lifetime of the Atto-488 dye on its distance to a graphene layer. The illustration above the graph depicts the expected conformations before and after DNA hybridization. In contrast to previous studies conducted at an ensemble level, we were able to detect the kinetics of individual DNA hairpins throughout the hybridization reaction. The developed single-molecule-sensitive approach enabled the observation of intermediate partially hybridized states. This approach holds promise for advancing research into the kinetics of DNA molecules under varied environmental conditions and investigating other biomolecules, spanning its relevance in diverse fields such as food safety and healthcare. Experimental Section Graphene Fabrication A graphene monolayer was grown on both sides of a copper foil substrate using Chemical Vapor Deposition (CVD) for subsequent transfer onto #1.5 (0.16–0.19 mm thickness) microscope glass coverslips. A cleaning protocol was implemented to prepare the glass coverslips for graphene transfer. The coverslips were immersed in an acetone bath and sonicated for 10 min, followed by an isopropanol bath and sonication for 10 min. Subsequently, deionized (DI) water was used along with sonication for 5 min, followed by drying with a gentle nitrogen flow. The graphene growth and transfer process to microscope glass coverslips proceeded sequentially as follows: first, the bottom layer of graphene, which exhibited a slower growth rate during CVD, was removed through plasma ashing for 6 min (gas flow rates: O2–200 sccm, Ar—200 sccm; chamber temperature: 50 °C; pressure: 0.60 mbar; voltage: 1500 mV; power: 230 W; frequency: 2.43 GHz). Then, a poly(methyl methacrylate) (PMMA) thin film was spin-coated onto the graphene layer to serve as an intermediate support substrate (16,000 rpm; baking temperature: 80 °C). Following this, the copper foil was dissolved using a 0.5 M iron trichloride (FeCl3) solution for 2 h, and the samples were placed in a DI water bath for 5–10 min. Any remaining iron trichloride was rinsed off through three washing cycles, with each cycle consisting of 10 min in DI water and 30 min in a 2% HCl solution. The manual transfer of the graphene layer onto the glass substrate was followed, and the PMMA layer was dissolved in an acetone bath. Lastly, the samples were left to dry overnight. Bright-field optical microscopy and Raman spectroscopy were used to assess the quality and coverage of the graphene layer (more details are provided in the Supporting Information). Graphene Functionalization with the Florophore-Labeled DNA Hairpin Probe The 2D graphene layer deposited on a microscope glass coverslip was functionalized with fluorophore-labeled DNA hairpin probes. These DNA hairpin probes were immobilized on the graphene surface using the linker molecule 1-Pyrenebutyric Acid N-Hydroxysuccinimide Ester (PBSE). The amino group of the hairpin binds with the reactive succinimide ester of the linker molecule, and the butyric pyrene group of PBSE binds noncovalently to the graphene.23 To carry out this graphene surface functionalization procedure, two sodium citrate (TSC)-based buffers were prepared and utilized: TSC1—0.75 M NaCl + 75 mM Na3C6H5O7 and TSC2—0.3 M NaCl + 30 mM Na3C6H5O7, with the pH adjusted to 7. After transferring the graphene layer onto the microscope glass coverslip, 10 μL droplets of a highly concentrated 10 mM PBSE solution in dimethylformamide (DMF) were drop-casted where the graphene was visible. After reacting for 2 h, the surface was rinsed with dimethylformamide (DMF) and water and dried with a gentle nitrogen flow. Afterward, 10 μL droplets of 10 μM DNA probe were drop-casted to match the expected coverage of the linker molecule and left to react overnight in a fridge at 4 °C, wrapped in aluminum foil to shield it from incoming light. Following this, the sample’s surface was rinsed with TSC1 buffer solution, and 10 μL droplets of a 100 nM solution of ethanolamine (ETA) were added and allowed to react for 30 min to remove excess linker molecules that did not react with the amino group of the DNA hairpin probe. Finally, the samples were rinsed with TSC2 buffer to remove leftover ETA. The functionalized area was left in solution to avoid the collapse of the DNA probe before being stored in the fridge and duly protected from light. This work’s DNA probe (pDNA) and target (tDNA) are 29-mers from flavanone-3-hydroxylase (F3H) gene of Portuguese grapevine varieties. The probe DNA strand has a C6 amino group on its 5′ end. The DNA probe is labeled with the fluorescent dye ATTO-488 on its other end, which is covalently attached to the probe. ATTO-488 has a lower molecular weight compared to the DNA hairpin providing suitable conditions to track hairpin’s dynamics and the DNA hybridization reaction. The target DNA strand was not modified. The sequences of the nucleobases were, for the probe: 5′—C6 Amino—AGC TTC ATA ACC GGC GAA AGG CTG AAG CT—Atto-488—3′ and for the target 5′—AGC TTC AGC CTT TCG CCG GTT ATG AAG CT—3′. The functionalization process was followed by an experimental analysis using the wide-field TIRF microscope on the same day to guarantee the maximum viability of the DNA. Biosensor Activation with Target DNA Solution The functionalized graphene biosensing platform was positioned on the microscope stage. After configuring all the microscope’s optical parameters, the buffer medium was removed using a micropipet, ensuring that the tDNA was not diluted upon addition. A fast time-lapse video lasting 2 min was initiated to capture the entire kinetic active period before adding the tDNA. Following this, a 4 μL droplet of the 100 nM solution of the tDNA was added to the focal region of the graphene sample. Due to the direct intervention with the micropipette, the laser beam scattered on the pipet’s tip during the addition phase, resulting in a temporary period of aberration in the captured fast time-lapse video. Wide-Field TIRF Microscopy Method An optical setup featuring an Inverted Nikon Ti-E microscope (Eclipse Ti series) with a TIRF module was employed to achieve this study’s single-molecule resolution goals. The setup included a Nikon APO TIRF 60x objective (NA: 1.49 Oil, 0.13 WD 0.21) and an incubator chamber housing a motorized X, Y stage coupled with a sensitive motorized piezo Z stage (minimum step of 0.025 μm). A Monolithic Laser Combiner (Agilent Technologies MLC400B) with 4 beam channels was utilized, where only the 488 nm channel was considered with an output power of 90 mW. The optical setup also incorporated an Electron-Multiplying Charge-Coupled Device (EM-CCD) Andor IXon Ultra camera Model U897, characterized by single-photon sensitivity (QE: >95%), a maximum readout speed of 17 MHz, pixel dimensions of 512 × 512, a frame rate of 56 (full frame), 16 μm pixel size, and a 16-bit depth. The continuous beam laser from the MLC was directed to the TIRF module through a fiber-coupled delivery system using a fixed mounting apparatus to maintain permanent alignment of the beam steering optics. This optical configuration was interfaced with a computer running the NIS Elements Advanced Research module, enabling control over the various optical parameters. A cube filter with excitation (482/20 nm) and emission (488/50 nm) band-pass filters was employed. Throughout the study, the laser power was predominantly set at 15% (13.5 mW) to mitigate photobleaching, and the exposure time was fixed at 60 ms to align photon counts with the required time resolution for kinetic assessments. The EM gain function, allowing for photoelectron amplification, was utilized and set at approximately x250. The TIRF angle was carefully adjusted to illuminate only the surface area proximate to the interface, ensuring optimal single-molecule resolution. Single-Molecule Time Trace Extraction The image processing program used in this work was ImageJ, a software developed at the University of Wisconsin, along with its plugin ThunderSTORM. These tools were employed to perform super-resolution reconstruction of the captured video images and to extract subpixel localization and intensity profiles of individual DNA hairpin molecules (more details can be found in the Supporting Information). Results and Discussion In the frames preceding the introduction of DNA (images can be found in the Supporting Information (Figure S3)), although the emitters were expected to be fully quenched, some bright spots remained discernible. These spots might be associated with aggregates of probe DNA or defects or discontinuities in the graphene. However, with the introduction of target DNA, in the fast time-lapse video, a significant number of individual bright spots, which were not evident before the tDNA addition, were observed. These spots demonstrate the successful activation of the DNA hairpin kinetic mechanism. Initially, in the super-resolution reconstruction image of the fast time-lapse video generated using ThunderSTORM’s, over 1500 intensity spots were identified. However, some postprocessing filters were applied to exclude specific spots. First, fluorescent emitters detected in frames corresponding to the addition of the tDNA were removed. This step was taken once the addition of tDNA induced a period of aberration caused by the scattering of the laser on the micropipet, making it uncertain that these particles do not correspond to aberration-induced artifacts. Second, more prominent spots or clusters of molecules were eliminated after analyzing the size distribution of the particles (more details can be found in the Supporting Information (Figure S4a)). In this work’s final normalized Gaussian rendering of the super-resolution conversion, 477 fluorescent intensity spots were detected, in an area of 138.24 μm by 138.24 μm, and subsequently analyzed. Using ImageJ, time traces were extracted for each detected fluorescent emitter, representing the fluorescence intensity values across the video’s time-lapse frames. Afterward, all extracted time traces were analyzed in MATLAB. Single-Molecule Time Traces and Kinetics of Individual DNA Hairpin Probes Out of the detected 477 intensity spots, 183 exhibited minimal to no change in their intensity profile after adding target DNA (Figure S5a, Supporting Information). Before adding the tDNA, the sample was briefly dried once the buffer solution was removed before the DNA addition to avoid uncontrolled and unwanted changes in the target concentration. The addition of tDNA in the hybridization buffer solution induces a transition from a dry medium to the buffer medium, resulting in a change in refractive index that can influence the background fluorescence noise. Attempts were made to determine the background level for graphene samples by capturing graphene images in a buffer solution without tDNA. However, the nature of wide-field analysis complicated this beyond practicality. For any region, the background level is influenced by the contributions of signals and noise from its surroundings, making background levels area-dependent and determining a single absolute value unfeasible. Figure 2b showcases a time trace representative of the remaining 294 emitters, along with the kinetic interpretations of five distinct fluorescent intensity phases, which, in contrast with the 183 low-variance time traces, shows a substantial increase in photon counts after introducing the target DNA. The initial phase I, highlighted in gray, corresponds to the period before the tDNA addition, during which the hairpin is expected to be in its folded state, and the fluorophore is expected to be only a linker length away (1.7 nm) from the graphene layer. At this distance, the fluorescence lifetime is quite short (Figure 2c), being the dye nearly completely quenched. The phase II, highlighted in black, represents the time of tDNA addition with a micropipette, during which laser scattering causes an aberration, resulting in a significant increase in photon counts. This intensity spike occurs in all 477 extracted time traces, and on average, this period takes about 8 s. The small red period, phase III, corresponds to a “lag” phase, where although the tDNA has been added, hybridization with the DNA hairpin has not yet been initiated. In blue, phase IV shows a period in which the fluorescence intensity is increasing before reaching an asymptotic maximum. We attribute this variation to the molecular dynamics of unfolding of the DNA hairpin, whereby the fluorophore attached to the hairpin is pulled away from the graphene, reaching into distances for which the near-field fluorescence quenching is lower. The intensity variation cannot solely be associated with blinking behavior as this typically would not be associated with gradual overall intensity variations. The phase V, highlighted in green, is characterized by a more stable high-intensity level. In this case, we associate this period to the fully hybridized hairpin where the elongated double-strand DNA causes the dye-to-graphene distance to be stable, at a distance of 11.2 nm (linker length + DNA length), resulting in the unquenching of the labeling dye (Figure 2c). Among the 294 time traces that show time-dependent fluorescence intensity changes, we highlight that, after the period of tDNA addition, III presented a “lag phase” (red time interval in Figure 2b) in which the fluorescence intensity did not significantly change and stayed on the background level, as observed in phase I. For these cases, the entire dynamic process associated with the DNA hybridization kinetics (phase IV) can be followed from the outset. Qualitatively assessing the exposure of the kinetic period among the 294 identified spots, in consequence of the aberration period that obscures the relevant signal for 8 s, the time traces were categorized into three distinct cases: “hidden” (Figure S5b, Supporting Information), “partially hidden” (Figure S5c, Supporting Information), and “fully revealed” (Figure 3a). In consequence of the nature of molecular diffusion in an aqueous medium, the hybridization between tDNA and the hairpin probes was initiated at different times. In each case, 41, 89, and 164 fluorescence spots were detected, respectively. Single-molecule time traces and kinetics of individual DNA hairpin probes. (a) Representative time trace showing “fully revealed” kinetics with a single unfolding phase. (b) Representative time trace showing “fully revealed kinetics” with two unfolding phases. (c) Distribution of the unfolding times for time traces with a single unfolding phase. (d) Distribution of total unfolding times in blue for time traces with two unfolding phases. In green is highlighted the distribution of the unfolding times, defined as the sum of the unfolding times for phases I and II. In the case of the “hidden” time traces, the start and end of their unfolding periods are inconclusive, since they are hidden by the laser aberration period. Therefore, their unfolding time was considered to be 8 s, corresponding to the duration of the DNA addition period or less. Regarding the “partially hidden” time traces, some unfolding occurred within the time window of adding tDNA. Consequently, only a portion of the kinetic phase is revealed. For these time traces, their unfolding period was determined to be partially exposed, taking an average of 5 s plus the period of tDNA addition or less. Figure 3a illustrates one of the 164 time traces wherein the unfolding period, which is colored blue, occurs completely beyond the aberration phase. The unfolding phase started at 22 s and lasted approximately 7 s. From the analysis of time traces, whose kinetic period is “fully revealed”, we drew attention to the fact that 14 do not only possess a single unfolding phase but also two unfolding phases, as illustrated and highlighted in green in Figure 3b. The Unfolding Phase I begins at the 22 s mark, just after the addition phase, and lasts approximately 8 s. The Unfolding Phase II starts at 44 s and lasts for around 8 s. These time traces with more complex unfolding periods share exciting characteristics, such as a period between the first and second unfolding phases where the photon counts either slightly decrease or remain the same. For the 14 detected hairpins with two unfolding phases, this intermediate state was estimated to last on average 12 ± 1 s. The observation of time traces exhibiting an intermediate state between two unfolding phases aligns with the findings reported by Chen et al.17 regarding the kinetic properties of DNA duplex formation where an additional energy barrier arises due to the disruption of the DNA hairpin stem, leading to possible intermediate states. Analyzing the time traces of each detected hairpin that unfolds in a single phase, we observed that the unfolding period took on average 7 ± 1 s (Figure 3c). When examining the distribution of the total unfolding time, the time difference between the background level and maximum intensity, for the hairpins with two unfolding phases, it is evident that these hairpins took longer to unfold completely than those with just one unfolding phase (Figure 3d in blue). On average, these hairpins take 29 ± 1 s to unfold. Even when discounting the inactive kinetic period and considering only the time for Unfolding Phase I plus Unfolding Phase II, the unfolding time is longer, averaging 16 ± 3 s (Figure 3d in green). These unfolding times and intermediate steps are linked to rates of hybridization, which are influenced by parameters such as salt concentration, the length, and the structure of the DNA hairpin, interactions between stem and loop nucleotides, and temperature.24,25 Although these variables that influence the kinetics of hairpins can be modeled for certain conditions, in practice, the complex multivariable settings and experiments heavily complicate such determinations, with current reports on unfolding rates ranging from seconds to minutes and hours time scales.18,25−27 Tsourkas et al.25 developed semiempirical models for the kinetics and thermodynamics of DNA hairpin hybridization and unfolding using FRET pairs at an ensemble level. Their findings revealed that for a DNA hairpin probe with a length of 17 base pairs and a stem section of 4 base pairs at 37 °C in an aqueous solution, the unfolding time of the hairpin matches that reported in this work (7 ± 1 s). Furthermore, Tsourkas et al.25 tested the impact of the used fluorophore molecule and did not find an influence on the kinetics of hybridization for a similar DNA molecular beacon system. However, it was also noted that small changes, such as adding one or two extra base pairs in the stem section, could extend the unfolding times beyond several minutes. Due to the sensitivity of hybridization rates and unfolding times to minor alterations in this multivariable landscape, attempts to compare experimental works become highly challenging. The experimental conditions most similar to ours were reported by Adão et al.,18 whose work was developed on a similar graphene-based platform, employing the same DNA hairpin probe design and buffer solutions. However, in contrast to our focus on the single-molecule level, Adão et al.18 examined changes in the fluorescence lifetime of the fluorophore-labeled DNA hairpin at an ensemble level, being the main difference in the concentration of the tDNA used to activate the probe: 10 μM instead of the 100 nM used in this work. Their results suggest that the unfolding of the DNA hairpin is a process spanning several hours, as evidenced by a continuous increase in the fluorescence lifetime over that period. This duration is significantly longer than the single-molecule unfolding times observed in our study, ranging from a few seconds to tens of seconds. These disparities can be explained by the sensitivity of the hybridization rates to external variables. For instance, the room temperature during FLIM characterization (19 °C) was lower than that in the laboratory where our work was conducted (22 °C). Despite similar sample preparation protocols, slight defrosting techniques and sample handling variations may inadvertently contribute to different hybridization rates. Another factor to consider is the potential increase in steric and kinetic hindrance due to higher tDNA concentrations in their study, resulting in closely packed hybridization reactions among neighboring hairpins bound to the graphene substrate. Although fluorescence lifetime is typically the parameter under measurement, several works demonstrated the successful use of intensity values to estimate distances.28,29 Consequently, through a developed semiempirical conversion of Gaudreau et al.’s19 model, which describes quantitatively the near-field interaction between graphene and fluorescent molecules, we were able to achieve highly resolved axial data. The semiempirical conversion was based on determining two critical data points: the background and the maximum intensity. The background level of fluorescence intensity was calculated by averaging the first 10 data points before the addition of tDNA, and the maximum intensity value was determined by averaging 10 data points near the end of the unfolding phase (Figure 4a). These two data points were associated with distances from the fluorophore to the graphene: 1.7 nm, the linker length, which represents the quenched configuration, and 11.2 nm, the linker length plus the molecular length, which corresponds to the expected fully extended duplex strand configuration that is formed as a result of the hybridization reaction. Axially resolved single-molecule hairpin kinetics. (a) A representative time trace in analyses from which two data points (in blue) were extracted to develop the semiempirical model. (b) In black, Gaudreau et al.’s model for the dependency of the fluorescence lifetime of the Atto-488 dye on its distance to a graphene layer. In blue, the Gaudreau et al.’s semiempirical conversion model into intensity as a function of the distance from graphene. (c, d) Axially resolved time traces were obtained using the developed semiempirical conversion model. The left-most black line corresponds to the period of tDNA addition. The unfolding process of the DNA hairpins follows a step-like behavior in that its increase or stability can be discriminated as “steps” (blue) or “plateau” (red). By combining the background and maximum point with Gaudreau et al.’s model19 (Figure 4b in black), a model providing the intensity as a function of the distance from graphene was derived, as illustrated in Figure 4b in blue. Since intensity levels are image area-dependent and vary for each hairpin, this conversion method was performed trace by trace for all analyzed single fluorescent labels that showed fully revealed kinetics. By extracting the corresponding distances from the intensity values based on our developed model, all considered time traces were converted to the distance from the graphene as a function of time to analyze the relevant unfolding phase. Figure 4c,d illustrates examples of this conversion, where it is possible to observe that the hairpins appear to unfold in a step-like fashion, with periods of steady overall increase in distance followed by in-between periods of constant distance plateaus, indicating an overall delay in the unfolding process. We identified these periods as metastable states at distances that were followed in some cases (Figure 4c) by additional unfolding periods where the steady increase in distance is roughly a multiple of nucleation distances (1 base pair length = 0.34 nm). On the other hand, for other cases (Figure 4d), the unfolding process seems more heterogeneous, with higher or smaller plateaus distributed at seemingly random distances. Conclusions Although DNA hairpin kinetics is inherently complex and challenging to predict with certainty due to its multivariable nature, this work demonstrated that kinetic changes are detectable and measurable at a single-molecule level in biologically relevant aqueous solutions by combining sensitive imaging techniques, DNA sensing technology, and the near-field quenching properties of graphene. Our novel approach to super-resolution reconstruction enabled the extraction of subpixel localization of individual DNA hairpin molecules, surpassing the diffraction limit, and their intensity profiles throughout the duration of the assay. The time resolution achieved by the wide-field approach, in conjunction with the optical properties of graphene, allowed us to clearly distinguish between the folded and unfolded states of individual DNA molecules. Furthermore, we determined the relevant kinetic period of the unfolding and hybridization phases, tracking and identifying various phases and trends, with particular emphasis on the observation of intermediate states. In addition, we could estimate the nanoscale axial motion of the fluorescent label away from the graphene by using a trace-specific intensity to distance near-field quenching-based conversion model. The developed wide-field TIRF readout technique, combined with an EM-CCD camera with high quantum efficiency, enables a parallelized read out that can be automated. Future applications of machine learning approaches, coupled with molecular dynamics simulations, may become powerful tools for understanding DNA kinetics in greater detail. This advancement holds potential not only for fundamental research but also for innovative applications such as next-generation unfolding assays and biosensors, with wide-ranging applications in medicine, food authenticity, and safety fields. Additionally, the single-molecule sensitivity approach presented in this work could significantly contribute to deepening our comprehension of the impact of partial DNA mismatches, thereby broadening our understanding of genetic disorders.30 Acknowledgments The authors acknowledge fruitful discussions with Rui Campos and Edite Figueiras. We further thank Rajesh Tamang for supporting the use of ThunderSTORM data analysis software and Ricardo Adão for providing guidance on the use of the near-field conversion model. We acknowledge access and support from the Nanophotonics and Bioimaging (NBI) and Micro and Nano Fabrication (MNF) core facilities of INL. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpca.4c05740. Graphene characterization: Raman spectra and bright-field microscopy images, ThunderSTORM’s method, single-molecule time trace analysis, and the theoretical model of the distance-dependence of the fluorescence lifetime in the near field of graphene (PDF) Graphene characterization: Raman spectra and bright-field microscopy images, ThunderSTORM’s method, single-molecule time trace analysis, and the theoretical model of the distance-dependence of the fluorescence lifetime in the near field of graphene (PDF) Author Present Address † ASML, De Run 6501, 5504 DR, Veldhoven, The Netherlands (J.F.) Author Present Address ‡ School of Chemistry and Chemical Engineering, Queen’s University Belfast BT7 1NN, United Kingdom (A.P.). Author Contributions # M.M.S. and J.F. contributed equally. The research was supported by FEDER, Portugal 2020, and FCT via the project On4SupremeSens with grant no. PTDC/NANOPT/29417/2017 and by the FCT through the Quantum Portugal Initiative (SFRH/BD/150646/2020). The authors declare no competing financial interest. Supplementary Material jp4c05740_si_001.pdf References Associated Data Supplementary Materials jp4c05740_si_001.pdf