Directing Nanoparticle Organization in Response to Diverse Chemical Inputs

Abstract

Signaling cascades are crucial for transducing stimuli in biological systems, enabling multiple stimuli to regulate a downstream target with precisely controlled timing and amplifying signals through a series of intermediary reactions. Developing a robust signaling system with such capabilities would be pivotal for programming complex behaviors in synthetic DNA-based molecular devices. However, although “software” such as nucleic acid circuits could potentially be harnessed to relay signals to DNA-based nanostructure hardware, such explorations have been limited. Here, we develop a platform for transducing a variety of stimuli via messenger-mediated reactions to regulate the release and reloading of gold nanoparticles (AuNPs) in a 3D DNA framework. In the first step, an in vitro transcription circuit is engineered to sense and amplify chemical stimuli, including arbitrary DNA sequences and proteins, producing RNA. In the second step, the RNA releases the DNA-coated AuNPs from the DNA framework via a strand displacement reaction. AuNP reloading is controlled by a separate step driven by degradation of the RNA. Our platform holds promise for applications requiring dynamic multiagent control over DNA-based devices, offering a versatile tool for advanced molecular device engineering.

Graphical Abstract

INTRODUCTION

In biological systems, stimuli are transduced by signaling cascades wherein a series of intermediary reactions culminates in downstream target activation. This relay mechanism, mediated by messenger molecules produced along the pathway, enables signals from diverse independently timed stimuli to converge on the target; enzymatically controls signal amplification; and provides independent control of response termination by reactions that degrade the messenger.¹ Such reactions enable, for example, the olfactory system to translate tiny amounts of odor molecules into the patterned electrical response of neurons.² The dynamic multiagent control and precise timing, scalability, and duration made possible by such signaling systems suggest a similar strategy could be transformative for deploying synthetic molecular devices in biosensing, targeted therapy applications, and metamaterials.^3–6

Recently, diverse DNA-based materials and devices responsive to stimuli,^7,8 including light-responsive machineries,^9–14 systemic stimulated (pH and temperature) devices,^15–22 and nanomotors triggered by DNA or RNA strands^23–28 have been developed. Such devices primarily rely on the direct interaction of the device with the stimulus. As a result, the device needs to be redesigned when a new stimulus is desired, and the timing, scale, and duration of the response directly reflect stimulus presence and magnitude. On the other hand, chemical reactions involving biomolecules have been used to design software-like functionalities for information processing and regulation of chemical outputs, such as bistable switches,^29,30 DNA logic gates,^31–33 pattern recognition networks,^34,35 and digital circuit computation.^36,37 Their highly specific, tunable input-output mechanisms provide a rich resource for engineering signal processors, but there have been limited explorations^38–42 into harnessing them to regulate hardware. A critical challenge has been integrating dissipative DNA software reactions with DNA hardware. In this work we develop a protocol that enables the use of enzymatic DNA software over at least 24 h without significant nanostructure deformation, opening the door for a range of new materials. We sought to extend these explorations by combining essential elements of control, termed “software”, such as strand-displacement reactions and nucleic acid synthesis, to transduce various stimuli into downstream responses, termed “hardware”.

Here, we present a versatile signaling strategy for transducing stimuli through multistep reactions to regulate the release of gold nanoparticles (AuNPs) from a 3D DNA framework (Figure 1). The framework^43,44 is a spatially well-defined 3D microscale assembly of DNA voxels in which each voxel can encapsulate a single DNA-coated AuNP through DNA hybridization between the single-stranded (ss) DNA grafted onto AuNP surface and the complementary strands within the voxels.^43,45 Our strategy comprises two primary signaling steps and an additional step to return the hardware to its original state by reloading AuNPs to the framework (Figure 1a). In the first step, a chemical stimulus activates an in vitro transcription circuit, generating RNA strands via RNA polymerase. In the second step, the RNA strands displace the AuNPs from the DNA voxel via a strand displacement reaction (Figure 1b). In an additional step to reset the system, the RNA is degraded by adding of ribonucleases (RNases), which catalyze cleavage of RNA molecules, allowing AuNPs to rebind to the voxel.

Figure 1.

Schematic of a signaling system to control the removal and reloading of gold nanoparticles (AuNP) from the DNA framework. (a) In vitro transcription circuits transduce stimuli (DNA sequences or proteins) into the production of RNA molecules that direct AuNP (diameter: 10 nm) release and reloading. Scale bars: 5 μm. (b) Schematic of the mechanism whereby RNA molecules disrupt the connection between AuNPs and DNA voxels. Each voxel typically contains eight AuNP-voxel bonds; for simplicity, only four bonds are depicted in the illustration. (c) Structural analysis using Small-Angle X-ray Scattering (SAXS) reveals structure factors, S(q) indicative of an AuNP-loaded DNA framework (red) and of a partially vacant DNA framework following AuNP release by RNA (purple), a model of a simple cubic (SC) lattice with 100% AuNP filling of the DNA framework with an interparticle center-to-center distance of 59 nm (black dashed line) is shown for comparison.

We demonstrate how this system can transduce diverse chemical signals into material transformation output (Figure 2). Specifically, signals such as DNA species unrelated to the framework or nanoparticles, and proteins, control the release and reloading of AuNP within the 3D DNA voxel framework. The strategy allows the response to be driven by dissipative reactions in both directions—the hydrolysis of nucleoside triphosphates (NTPs) in the polymerase reaction, and the breakdown of RNA by RNases—providing separate control over response onset and duration. Using RNA as a messenger also amplifies the signal to levels required to mediate AuNP release at the nano- and microscale, enabling the system to process a range of stimulus concentrations, including levels lower than that of DNA directly interacting with the voxels. The timing and rate of the response can also be tuned by RNA concentration. This versatility highlights the modularity of our signaling system and provides a foundation for coupling DNA-based hardware and software in the design and programming of molecular devices.

Figure 2.

Molecular mechanism of the signaling system. (a) The stimuli, DNA or protein, alter the conformation of a designed DNA template to regulate transcription of RNA in the presence of NTPs. (b) The produced RNA replaces the DNA strand grafted to the AuNP at the voxel interior strand, triggering AuNP release from the DNA framework via a strand displacement reaction. (c) In the reset step, the input of RNases breaks down the RNA. (d) RNA degradation makes the voxel sites accessible to AuNPs again, resulting in AuNP reloading. The two-state switch is driven by dissipative reactions in both the forward and reverse directions (cleaving NTPs and RNA species, respectively). Scale bars: 10 μm.

RESULTS AND DISCUSSION

An AuNP-Hosting DNA Framework.

The 3D DNA framework studied here is built on a previously developed 3D framework⁴³ of octahedral DNA voxels⁴⁶ each containing 12 six-helix bundle edges (29 nm) (Note S1). Each voxel is a scaffolded DNA origami nanostructure, formed by folding the M13mp18 single-stranded (ss) DNA using 144 staple strands. The resulting voxels self-assemble into a simple cubic lattice⁴³ via vertex-to-vertex hybridization. Each voxel presents eight ssDNA sticky ends (S1)—domains extended from eight staples—projecting toward the interior. The S1 can hybridize with ssDNA (N1) covalently bonded (Au-thiol bond) on target guest AuNPs (Note S2), such that each voxel encages one AuNP (Figure 3a). This design enables AuNP release and reloading within the framework to be regulated by dehybridization and rehybridization between S1 and N1 (Note S3).

Figure 3.

RNA-directed AuNP release and reloading in DNA frameworks. (a) Mechanism of AuNP release and reloading triggered by input and degradation, respectively, of rD1 RNA. (b) Optical micrographs of one framework domain during AuNP release (purple arrow) and of a second framework domain during AuNP reloading at 37 °C (red arrow). Scale bars: 10 μm. (c) Absorbance/area of individual frameworks (dots) at different times (labels, upper right) after the addition of rD1 (purple) or RNases (red). Absorbance range: 400 to 700 nm. Data are analyzed based on Note S5.

To verify the degree of AuNP filling, we used Small Angle X-ray Scattering (SAXS) technique to compare the AuNP loaded and released 3D DNA frameworks. As shown in Figure 1c, the measured structure factor (red curve), S(q), reveals a simple cubic lattice (SC) configuration with an interparticle center-to-center distance of about 59 nm for AuNPs loaded in the DNA framework, indicating that the AuNPs are placed within the predetermined positions of each DNA voxel. A computed S(q) for AuNP arranged in the simple cubic lattice (100% AuNP filling, black dashed line) agrees well with the experimental S(q), indicating a substantial amount of AuNPs homogeneously loaded within the DNA framework.

Since AuNPs exhibit significantly higher optical absorbance than DNA materials, particularly at the AuNP plasmonic resonance regime (wavelengths λ ~ 520 nm), the absorbance of the voxel framework serves as a quantitative indicator of AuNP loading. Therefore, the extent of AuNP loading and release was monitored by measuring the light transmission of individual frameworks, namely the cubic crystals observed in microscopic images at λ = 400–700 nm (Figure 3b and Note S5).

AuNPs were initially loaded by mixing and incubating AuNPs with the framework at a molar ratio of AuNP: voxel = 2:1 ([AuNP] = 8 nM, [voxel] = 4 nM) at 37 °C. The majority of the DNA frames were filled with AuNPs after 48 h incubation, as evidenced by significantly larger absorbance per DNA framework area (absorbance/area) observed for these frameworks compared to either vacant frameworks or the vacant framework mixed with AuNPs coated with control DNA strands that could not hybridize with S1 (Figure S5b).

The AuNPs were released (Figure S2) by the addition of a displacing ssDNA strand (D1), which can displace N1 because D1 forms 12 more base pairs with S1 than N1 does (Note S8). Displacing N1 releases AuNPs from their S1–N1 origami bonds, allowing AuNPs to diffuse out of the DNA framework. Transmission electron microscopy (TEM) experiments (Figure S4) were conducted to probe the AuNP loading and release within individual voxels. 75.4% ± 2.7% (95% CI) of voxels contained AuNPs (N = 252) in the initial preloading sample, while after the addition of 10× D1 for 2 h incubation, 91.9% ± 4.6% (95% CI) of voxels contained no AuNPs (N = 135). To measure the concentration dependence of AuNP release within DNA frameworks, we added 1× (D1: S1 = 1:1, [S1] = [D1] = 32 nM), 2× ([D1] = 64 nM), and 10× ([D1] = 320 nM) excess of D1 to solutions containing DNA frameworks. We observed that higher D1 concentrations resulted in a faster rate of absorbance decrease (Figures S8 and 9): absorbance/area decreased from 3000 (loaded-AuNP framework) to 1540, 990, 292 AU for 1×, 2×, and 10× [D1], respectively. These results indicate that the rate of AuNP release can be modulated by varying the concentration of molecules that displace N1 from S1.

Release and Reloading of AuNPs via Direct RNA Input and Degradation.

To investigate whether RNA molecules can regulate AuNP release by displacing N1 from S1 (Figure 3a), we synthesized RNA strand equivalents of D1 (rD1) with domains A and B (Note S6, the same sequence as D1 except with T in place of U) and used RNases to enzymatically degrade rD1. Within the DNA framework, the S1 sticky end composed of A* and B* (X* refers to the complementary sequence of X) domains binds to the N1 strand with the A domain coating the AuNP. Adding rD1 to the AuNP-loaded framework led to strand displacement of N1 from S1 and triggered AuNP release from the DNA framework. Subsequent addition of RNases A/T1 and H, which catalyze endonucleolytic cleavage of ssRNA⁴⁷ and of RNA hybridized to DNA,⁴⁸ respectively, led to the degradation of rD1, enabling rehybridization of S1 and N1 and reversion to the AuNP-filled framework.

We first studied the AuNP release and reloading processes within individual voxels by rD1 and RNases using TEM (Figure S4). AuNP release was triggered by the addition of 10× rD1 and AuNP reloading was facilitated by 50 U/mL RNase A/T1 and H. After a 2 h incubation with 10× rD1, 96.9% ± 3.0% (95% CI) of voxels contained no AuNPs (N = 130). Subsequently, we added 50 U/mL RNases A/T1 and H to the rD1 release sample and incubated for 24 h. The corresponding TEM images showed that 45.5% ± 4.2% (95% CI) of voxels contained AuNPs (N = 143).

We then monitored the AuNP release and reloading processes within DNA frameworks using optical microscopy; the high light absorption by particles means that the transparency of a frameworks is lower when it contains AuNPs (Figure 3b). We found that a substantial release of AuNPs occurred after 24-h incubation with rD1, as indicated by a decrease in the absorbances of individual frameworks–the average absorbance/area of frameworks decreased from 2150 to 180 AU (Figure 3c). After 24-h incubation with RNases, the average absorbance/area recovered to 2150 AU, indicating that the reloading of AuNPs can be directed by RNase-catalyzed reactions that degrade the ssRNA and the RNA hybridized with AuNP recognition strand within a voxel (Figures 2a and 3a). We also used SAXS to compare the AuNP loading, release and reloading within DNA frameworks. Structural analysis shows that upon rD1 addition, the SC lattice pattern disappeared with the reduction of the peaks during the process (Figure 1c), indicating AuNP release, while RNases input restored SC lattice pattern (Figure S14), demonstrating the reloading of AuNPs. These findings support the notion that rD1 RNA can induce AuNP-DNA voxel unbinding, and that AuNP release and reloading can be induced by separate steps of RNA input and degradation. Note that there is a modest proportional relationship between absorbance/area of individual framework and the measured framework area. This phenomenon can likely be attributed to the varied thickness of the framework structure, as thicker frameworks generally contain more AuNPs leading to a higher absorbance per pixel.

Ligated DNA Framework Can Resist Disassembly Caused by Adding RNA Polymerase (RNAP).

In vitro transcription (IVT), involving T7 RNAP and simple chemical reagents–double-stranded DNA templates, Mg²⁺ ions and NTPs as substrates for the synthesis of RNA molecules, is an essential reaction of transcriptional circuits and a key element of our strategy for linking the AuNP response to upstream signals. However, T7 RNAP has been reported to nonspecifically bind and transcribe a variety of DNA sequences, which could cause undesired deformation of DNA nanostructures.^30,49,50

We hypothesized that DNA ligation could mitigate the nonspecific transcription within DNA nanostructures by reducing the number of 3′ ends where promoter-independent transcription is initiated readily, and therefore provide protection against disassembly. We first used T4 DNA ligase to ligate the staple strands in the DNA voxel in between the 5′-phosphate and the 3′ hydroxyl groups of adjacent nucleotides. We observed that, without ligation of the DNA voxels, adding T7 RNAP led to disassembly of the nanostructures. It was presumably because of promoter-independent transcription occurring on the DNA voxels, leading to undesired strand displacement reactions and subsequent unfolding of the DNA voxel structures (Figure 4a). Conversely, ligated DNA voxels retained their octahedral shape after incubation with T7 RNAP for 4 h (Figure 4b). We further applied the ligation protection to the DNA framework and observed that the ligated DNA framework remained largely intact while the nonligated framework was corroded after incubation with T7 RNAP for 4 h (Figure S7). We also found that ligation did not affect AuNP loading (Note S5). We, therefore, chose to ligate DNA frameworks to ensure their structural stability when exposed to T7 RNAP in subsequent experiments (Note S7).

Figure 4.

Protection of DNA voxel nanostructures against nonspecific transcription by T7 RNAP. (a) DNA voxels can disassemble when incubated with T7 RNAP. (b) DNA ligation can protect DNA nanostructures from disassembly caused by nonspecific transcripts. Scale bars: 100 nm.

In Vitro Transcription (IVT) Directed AuNP Release.

As the next step in constructing the signaling system, we investigated whether rD1 can be synthesized by transcription in situ and then direct sequence-specific AuNP release. As shown in Figure 5a, we designed an rD1 template with promoter (orange domain) and downstream rD1 sequence (blue and purple domains) to produce the target RNA rD1. To measure the production of RNA independently of its interactions with AuNPs and the DNA framework, we developed a real-time fluorescent rD1 reporter pair serving as an indicator of rD1 synthesis, wherein rD1 displaces a Cy3 fluorophore-modified DNA strand and enhances the measured fluorescence (Figure 5b). Using this reporter system (320 nM rD1 reporter) (Tables S4 and S8) in the absence of the DNA framework, we found that rD1 can be synthesized in situ and that the rate of synthesis (41 to 259 nM activated reporter after a 2 h reaction) can be modulated by varying the rD1 template concentration between 1 and 10 nM (Figure 5b). To determine whether DNA nanostructures interfere with rD1 transcription, we compared the reporter fluorescence when IVT was performed with 10 nM rD1 template in the presence and absence of DNA voxels (Figure S11). The rate of fluorescence increase was similar in the two cases, implying that rD1 synthesis is not significantly impacted by the presence of DNA voxels.

Figure 5.

In vitro transcription (IVT) directed AuNP release from the frameworks. (a) IVT reaction (yellow box: promoter domain in orange, rD1 in blue (B–B*), purple (A-A*) and black where A-A* and B–B* domains correspond to domains involved in the strand displacement reaction within the DNA frameworks) and AuNP release and reloading using rD1 and RNases. (b) Quantitative measurement of transcriptional rate using a reporter complex (dashed-line box: Cy3 fluorophore and quencher shown as green and black dot, respectively), and a graph of measured transcriptional rates for different template concentrations. (c) Optical micrographs of AuNP-loaded framework following IVT reaction for 10 min and 2 h. Scale bars: 10 μm. (d) Plots of absorbance/area of individual frameworks at different times (labels, upper right) during in vitro transcription (IVT, yellow) or after the addition of RNases (red).

To ask whether RNA transcription could drive AuNP release, we conducted IVT with 10 nM rD1 template in the presence of the reporter system (320 nM rD1 reporter, Figures 5b), AuNP-loaded individual voxels (Figure S4) and the AuNP-loaded framework (Figure 5c,d). IVT activated 260 nM rD1 reporters within 2 h of reaction, which resulted in 78.4% ± 3.8% (95% CI) of individual voxels containing no AuNPs (N = 116), and a corresponding reduction of the absorbance/area of individual frameworks from 3000 to 1690 AU. After 24 h of reaction, the absorbance/area decreased to 160 AU, confirming that RNA produced by IVT can drive AuNP release. To assess the impact of nonspecific transcription on the measured framework absorbance change, we combined T7 RNAP with the AuNP-loaded frameworks in the absence of rD1 template. We observed a ~18% reduction in the average absorbance of individual frameworks over a 24-h reaction period (Figure S12), indicating that nonspecific transcription can contribute to AuNP release but at a very low rate compared to the rD1-induced response.

We then examined whether AuNP release rate can be controlled by varying the IVT rate. Conducting IVT with 2 nM rD1 template (Figure S13) activated 4 nM and 104 nM rD1 reporter within 10 min and 2 h, respectively, and led to a reduction in absorbance/area to 2200, 1860, and 260 AU within 10 min, 2 and 24 h, respectively. These results suggest that a lower rD1 template concentration leads to a lower IVT rate, resulting in slower AuNP release.

To determine whether degradation of the in situ synthesized rD1 in the presence of the frameworks can direct the reloading of the AuNPs in the frameworks, we added 50 U/mL each of RNases A/T1 and H to the framework solutions. After 24 h of incubation, the absorbance/area of the frameworks increased to 1500 AU, suggesting successful reloading of AuNPs. Completion of AuNP loading takes approximately 72 h (Note S3).

Transduction of DNA Stimuli via Genelet-Directed rD1 Synthesis.

We then explored whether a transcriptional circuit can direct AuNP release within the DNA frameworks in response to a stimulus input. One well-established unit within the transcriptional circuit repertoire is the genelet,^30,51 a short partially dsDNA sequence containing an incomplete promoter site for T7 RNAP. Transcription of an RNA output only occurs when a ssDNA activator strand binds to and completes the promoter site. We used a genelet (termed G) IrD1 template (which couples Input sensing and rD1 generation), a partial DNA duplex with an incomplete promoter site, and A1 activator that can hybridize to the IrD1 template and complete the promoter site (Figure 6a). In the absence of the A1 activator, the genelet is in an OFF state, so that rD1 transcription is halted. The addition of the A1 activator turns the genelet device to an ON state, triggering rD1 transcription. Importantly, A1 has no direct interaction with the DNA framework or AuNPs, suggesting that nanoparticle release can only occur because of IVT of the genelet IrD1.

Figure 6.

Directing AuNP release using DNA sequence stimuli that control rD1 transcription rate. (a) A genelet device, termed G, can serve as a transcription template when bound to the input DNA signal A1. (b) Quantification of transcription rates from G ON (green line) and G OFF (gray line). (c) Optical micrographs of AuNP-loaded framework 10 min and 2 h following the addition of 40 nM A1, which activates transcription of rD1. Scale bars: 10 μm. (d) Plots of absorbance/area of individual frameworks in the G ON state (top row) and in the G OFF state (bottom row) 10 min, 2 h, and 24 h after adding DNA signal A1.

We then studied the AuNP release response to stimulus A1. As shown in Figure 6b–d, when the genelet was in the OFF state (gray curve and spots, 8 nM IrD1 template, no A1), less than 10 nM reporter was active and only a minor reduction in absorbance/area from 2650 (at 10 min) to 2300 AU was observed after 2 h. When the genelet was in the ON state (green curve and spots) (i.e., 40 nM A1 input was present), 190 nM reporter was activated, and the average absorbance/area was reduced to 1190 AU within the same time frame. After 24 h of reaction time, there was still no significant reduction in absorbance/area for an OFF genelet that had not received an input, while frameworks where input had been added had a further decrease in average absorbance/area to 600. Note that these tests of genelet-controlled AuNP release were conducted using buffers where the genelet behavior has been studied previously (Note S4). These results indicate that a genelet can transduce DNA input signals to drive AuNP removal from frameworks.

To examine whether IrD1 template concentration affects AuNP release speed, we added 40 nM A1 to the AuNP-loaded framework but reduced the IrD1 template concentration to 4 nM. The absorbance/area decreased (Figure S14b) from 2560 at 10 min to 1779 after 2 h and to 1147 after 24 h. These results indicate that the release rate of AuNPs at 4 nM IrD1 is slower than at 8 nM IrD1 under identical conditions of A1 input, suggesting that the AuNP release rate can be modulated by altering IrD1 template concentration. We also investigated the influence of A1 activator concentration in AuNP release kinetics by adding 1.6 nM A1 to the IrD1 solution with 8 nM template. The absorbance/area decreased (Figure S18) from 2426 at 10 min to 1390 after 2 h and to 777 after 24 h, indicating that we can manipulate AuNP release rate by varying A1 input.

Transduction of Protein Stimuli via Aptamer-Based Circuit Directed rD1 Synthesis.

To demonstrate the capacity of a processor to transduce diverse stimuli other than nucleic acid sequences, we adapted an aptamer-based transcriptional circuit⁵² that responds to specific concentrations of input proteins to drive AuNP removal. The circuit uses aptamer-protein binding to regulate the transcription of a DNA template (Figure 7a). The transcription template contained a DNA aptamer for an example protein IFN-γ; binding of the protein to the aptamer represses transcription of the template. The resulting DNA template for Aptamer-Regulated Transcription (dART) for IFN-γ was coupled with a reference template to create a comparator circuit. The comparator includes two transcriptional units: (1) ref-rD1-dART with an aptamer domain that does not bind IFN-γ, designed to produce rD1; (2) IFN-rD1′-dART with the IFN-γ aptamer domain, designed to produce an output rD1’ with partial complementarity to rD1 in the absence of IFN-γ. When [IFN-γ] is low, the concentration of IFN-γ-unbound IFN-rD1′-dART is higher than the ref-rD1-dART, leading to [rD1’] > [rD1]. In this case, more rD1’ would bind to and sequester rD1, making rD1 unable to interact with other molecules and thus resulting in a low reporter signal or weak AuNP response. As [IFN-γ] increases, more IFN-rD1′-dART binds to IFN-γ, suppressing the rD1’ transcription rate and thereby making more rD1 available to interact with either a reporter or with the DNA frameworks (Figures7b,c).

Figure 7.

Protein-directed AuNP release using a circuit that processes a protein input using aptamer-regulated transcription. (a) A comparator circuit, composed of transcriptional units IFN-rD1′-dART and ref-D1-dART, produces high [rD1] in response to the presence of the protein IFN-γ. ref-D1-dART produces rD1 which can be deactivated by rD1’ generated from IFN-rD1′-dART. The IFN- γ is designed to suppress IFN-rD1′-dART, leading to lower [rD1’] and consequently higher [rD1]. This comparator circuit is referred to as A; A with an up arrow indicates a circuit is active because its input is present, A with a down arrow is an inactive circuit. (b) Expected (digital) response of the comparator circuit to [IFN-γ]. (c) Quantification of transcription rates with 100 nM IFN- γ (navy line) and without IFN- γ (light blue line). (d) Optical micrographs of AuNP-loaded framework following aptamer-based circuit with 100 nM IFN- γ for 10 min and 2 h. Scale bars: 10 μm. (e) Plots of absorbance/area of observed individual frameworks within reaction periods of 10 min, 2 h, and 24 h after adding input. Top, after IFN-γ is added as input, bottom, no IFN-γ input.

Combining 25 nM of ref-rD1-dART and 50 nM of IFN-rD1′-dART in the presence of the 3D DNA framework in the buffers where the dART was previously studied (Note S4), we observed significantly more AuNP release (Figure 7d,e) in the presence of 100 nM (navy spots, Figure 7e) and 50 nM (Figure S19) IFN-γ than 5 nM IFN-γ (Figure S19) and in the absence of IFN-γ (light blue spots, Figure 7e). We observed a reduction in absorbance/area to 990 and 300 AU in the presence of 100 nM IFN-γ within 2 and 24 h, respectively. Conversely, in the absence of IFN-γ, the absorbance/area decreased from 3000 AU (DNA frameworks without stimuli) to 2000 AU after a 2-h reaction, with little change observed over a 24-h reaction period. We also observed that the frameworks remained intact in the solution used for these experiments, which had a high potassium ion concentration (Note S4). These findings underscore the aptamer-based circuit’s ability to transduce specific protein signals into a prescribed response within the DNA frameworks.

Moreover, we adapted an amplified comparator⁵³ to produce more rD1, which we hypothesized would increase the AuNP release rate (Note S11). We added a genelet downstream of the comparator circuit, so that the output (RNA strand C1) of the comparator activated transcription of the genelet IrD1 (Note S11). This circuit rapidly activated 1000 nM reporter in 2 h after adding 200 nM IFN-γ (Figure S16a). Integrating this system with the 3D DNA frameworks, we observed a notably faster AuNP release rate compared to a nonamplified comparator: addition of 200 nM IFN-γ caused a reduction of the absorbance/area of individual frameworks to 430 and 130 AU within 2 and 24 h, respectively (Figure S16b).

CONCLUSION

Here we have developed a multistep signal processing platform that transduces a variety of stimuli to modulate AuNP release from a downstream DNA framework. Transcription circuits process a range of ssDNA and protein inputs presented at different concentrations to synthesize varying amounts of RNA. The RNA then relays the signals to the framework and directs AuNP release with a wide range of controllable speeds. By separating the upstream stimuli from the downstream response, the system can be reset—AuNPs reloaded into the framework—by a separate, stimulus-independent dissipative reaction, RNA degradation. This allows response onset and timing to be controlled independently.

More broadly, this study shows how the modularity of in vitro transcriptional circuits can be used for sensing, amplifying, and relaying signals to DNA nanoscale materials and devices. This processor design serves as a foundation for integrating additional circuitry, such as logic gates³⁴ and probabilistic computation⁵⁴ without a structural overhaul of the material/device. This signaling system offers the potential to develop reconfigurable optical metamaterials, as many advanced optical functionalities stem from the intricate 3D structural arrangement of nanomaterials which are often difficult to fabricate and even more so to reconfigure postfabrication. Within our signaling system, we can potentially program nanoparticle manipulations by controlling steps of addressable loading and unloading independently driven by multiple stimuli. For example, the presented approach might allow for switching a photonic crystal’s symmetry by controlling the positioning of minor perturbations within its unit cells, altering the photonic band structure, and yielding novel spectral features with controllable lifetimes and polarization states.^55,56 In addition, this signaling concept can extend to dynamically controlling the organization of catalytic objects, such as enzymes,^57–59 where factors such as colocalization, proximity, and local environment profoundly influence their catalytic performance.