Electronic control of DNA-based nanoswitches and nanodevices† †Electronic supplementary information (ESI) available: Experimental procedures. See DOI: 10.1039/c5sc03694a Click here for additional data file.

Here we demonstrate that we can rationally and finely control the functionality of different DNA-based nanodevices and nanoswitches using electronic inputs.


Introduction
The growing and exciting eld of DNA nanotechnology, where synthetic nucleic acids are rationally engineered and designed to build novel responsive nanomachines or functional nanodevices, represents one of the most interesting examples of bioinspired technologies. 1 The majority of these DNA-based nanodevices rely on a common basic mechanism: a target molecular input is recognized by a DNA probe sequence and the binding event is coupled to a mechanical motion or an output signal. 2 Even the most complex functional DNA-based nanostructures (i.e. origami) are thus based on the use of relatively short DNA sequences (or nanoswitches) that in the presence of a specic molecular input undergo binding-induced conformational changes or DNA-based reaction and, by doing so, confer to the nanostructure an useful function. 3,4 Despite the advancements achieved in this eld, the need to have a better control of such nanodevices remains still partially unmet. To fully exploit these platforms it would thus be crucial to nd new strategies to trigger and activate their function in a highly controllable fashion.
Since the revolutionary discoveries of Volta, Faraday and other pioneer electrochemists, 5 the possibility to control redox reactions through an electronic input (applied voltage) has represented one of the major breakthrough in the history of chemistry. It is now more than 200 years that electrochemistry has been applied for a wide range of applications from energy production 6 to industrial manufacturing 7 and sensing. 8 Because of the low cost of instrumentation, possible miniaturization Fig. 1 DNA nanodevices or nanomachines are usually based on responsive DNA sequences (red) that, upon binding to a specific molecular input (green trigger) undergo a conformational-switch or a specific reaction that is converted into a signal. Here we demonstrate that we can electronically activate similar responsive DNA-based nanodevices. In our approach a non-active trigger (grey) is coated on the surface of a chip electrode. The electronic input promotes an electron transfer exchange reaction that activates the trigger and releases it from the chip's surface. The active trigger is then able to bind the DNA nanodevice activating it. We used here four different DNA-based nanodevices: two conformational-change DNA nanoswitches responding to metal ions (Hg(II) and Ag(I)), a stem-loop DNA molecular beacon responding to a specific DNA sequence and a Cu(II)responsive DNAzyme. and high level of control, electrochemistry might also represent an interesting opportunity for novel bio-technological applications. Redox reactions are in fact routinely used in Nature to activate, regulate and control a wide range of biological pathways and reactions (such as photosynthesis and energy storage/ release). 9 In a similar way, electrochemistry could thus be applied to modulate bio-inspired tools and devices. [10][11][12][13] Despite this, the possibility to use electronic inputs to control DNAbased nanodevices has seen very little application to date. 14 Motivated by the above arguments, here we propose an approach to electronically control a wide range of DNA-based nanodevices. We did so by controlling, through an electronic input, electron transfer across an electrode-solution interface thus promoting redox reactions in a highly controllable fashion. More specically, as the electronic input we used here a voltage potential applied at the surface of an electrode chip. Such applied potential promotes an electron-transfer reaction at the electrode-solution interphase leading to the release of a molecular input that ultimately triggers a DNA-based nanodevice in solution ( Fig. 1). To demonstrate the versatility of our approach we have used here four different model DNA-based nanodevices or nanoswitches that can be activated by different molecular inputs.

Results and discussion
As a rst proof-of-principle of our strategy, we demonstrate here that we can electronically trigger the conformational change of a DNA-based nanoswitch. To do this we have used a previously reported DNA-nanoswitch whose binding-induced conformational change can be triggered by Hg(II) ions. 15 The nanoswitch is designed to be in a thermodynamic equilibrium between two low-energy states: a non-binding (OFF) conformation that lacks the Hg(II) binding sites and a binding-competent conformation (ON) that contains multiple T-T Hg(II)-binding mismatches 15 ( Fig. 2A). The sequence is designed so that the non-binding state is more stable and only in the presence of Hg(II) ions this equilibrium is pushed toward the binding-competent conformation, coupling recognition with a large conformational change. Because the nanoswitch is labelled with a uorophore and quencher we can easily follow the conformational change of this nanoswitch by uorescence measurements (Fig. S1 †).
We can trigger the conformational change of such DNAnanoswitch by using an electronic input. To do so we used a gold chip 16 coated with a lm of Hg(0) ( Fig. 2A). By applying on this chip's surface an oxidative potential scan (from 0.2 to 0.65 V vs. Ag/AgCl) it is possible to promote an electron-transfer reaction that will lead to the oxidation of the deposited Hg(0) to Hg(II) ions which, in turn, will diffuse to the bulk of the solution and trigger the conformational change of the DNA nanoswitch ( Fig. 2A). Of note, by varying the density of Hg(0) on the chip surface from 2.4 AE 0.2 Â 10 À6 mol cm À2 to 2.4 AE 0.3 Â 10 À10 mol cm À2 (Fig. S2 †) we can nely regulate the activation of the nanoswitch (Fig. 2B). A control experiment performed under the same experimental conditions but without applying the electronic input leaves the nanoswitch completely inactivated (Fig. 2B, black curve, no electronic input). Similarly, by applying an oxidative potential to a chip without Hg(0) coating results in no activation of the nanoswitch (Fig. 2B, green square). We can also modulate the electronic activation of the nanoswitch by varying the electronic input applied on the chip's surface. More specically by using a chip with a xed Hg(0) density (2.4 AE 0.2 Â 10 À6 mol cm À2 ) we have varied the width of the potential step during the oxidative scan (from 1 mV to 6 mV) thus producing a modulation of the activation of the nanoswitch (from 22 AE 1% to 100 AE 2% respectively) (Fig. 2C). The electronic strategy we propose to control DNA nanodevices is reversible and can achieve regeneration. We demonstrate this by cyclically releasing Hg(II) ions and depositing Hg(0) through the application of oxidative (0.2 V) and reductive potentials (À0.3 V), respectively. By doing so we show that we can cyclically activate and inhibit the DNA nanodevice in a reversible way (Fig. 2D).
By using a similar approach we can also electronically activate a DNA-based switch whose conformational change can be triggered by the formation of multiple C-C Ag(I)-binding mismatches 15 (Fig. S3 †). To do so we employed chips produced  (0) to a gold chip from a Hg(II) solution allows to control the nanoswitch's activation and inhibition respectively. (B) We can modulate the percentage of nanoswitch electronic activation by varying the density of Hg(0) on the chip (from 2.4 AE 0.2 Â 10 À6 to 2.4 AE 0.3 Â 10 À10 mol cm À2 ). Control experiments were performed using (i) Hg(0)-coated chips without applying the oxidation potential (no electronic input, black curve) and (ii) applying the oxidation potential to chips without Hg(0) coating (green square). (C) We can also modulate the Hg(II)-responsive nanoswitch by using chips with fixed Hg(0) density (2.4 AE 0.2 Â 10 À6 mol cm À2 ) and by varying the step potential (E step ) used during the oxidation scan. (D) By cyclically oxidizing Hg(0) (release from chip) and reducing Hg(II) ions (depositing on chip) we can activate and inhibit our Hg(II)-responsive nanoswitch in a reversible way. Here fluorescence measurements were performed in 50 mM sodium phosphate, 150 mM NaCl, pH 7.0 at 25 C containing the Hg(II)-responsive nanoswitch (10 nM). See ESI for experimental details. † by using silver-based conductive inks. By applying a potential scan to this silver-based chip we can control the production and release of Ag(I) ions in solution and we can thus modulate the activation of the DNA-based switch. Of note, in this case we are unable to control the density of Ag(0) on the chip's surface. Despite this, we can gradually modulate the release of Ag(I) ions by varying the potential step used during the oxidative scan observing a behaviour comparable to that observed with Hg(II)activated switches (Fig. S4 †).
To demonstrate the broad implication of our approach we demonstrate here that we can electronically control the activation of other more general DNA-based nanoswitches. More specically, we have employed a classic DNA-based nanoswitch (i.e. a DNA strand adopting a stem-loop structure) whose conformational change can be induced by a specic DNA strand complementary to the loop sequence (Fig. 3A). 17 To electronically induce the opening of this stem-loop nanoswitch we have used a thiol-modied DNA strand (input-strand) and we have deposited it on the surface of a gold chip through spontaneous thiol-gold self-assembly reaction. 16 By applying a constant reductive potential (À1.2 V vs. Ag/AgCl) on the gold electrode surface we can electronically induce the reduction of the thiolgold bond and the release of the input-strand. 18 This allows to control the activation of the DNA-based nanoswitch (Fig. 3B). Of note, using the same input-coated chip but without applying the electronic input, we observe no activation of the nanoswitch (Fig. 3B, black curve). The electronic activation of this DNAnanoswitch is also highly tunable and controllable. For example, by varying the concentration of the input-strand used during the coating step we can modulate the input-strand surface density (from 4.3 AE 0.3 Â 10 À10 mol cm À2 to 1.4 AE 0.4 Â 10 À14 mol cm À2 ). 19 This allows to achieve a gradual electronic activation of the nanoswitch (Fig. 3C). We also note that by depositing an analogue input-strand (non-complementary to the loop sequence) and under the same conditions (i.e. same surface density and electronic input) we do not observe any activation of the DNA nanoswitch (Fig. 3C, mismatch).
We can also modulate the fraction of activated nanoswitches by varying the time of applied potential on the chip's surface (period of electronic input). 18 For example, by gradually changing the period of applied potential from 0.5 to 30 s we can modulate the percentage of activated nanoswitches from 15.7 AE 0.2% to 100 AE 3%, respectively (Fig. 3D). Similarly, because the electron transfer rate efficiency that leads to the input-strand release depends on the electronic input, we can modulate the percentage of activated nanoswitches by varying the applied potential. We demonstrate this by using different applied potentials over the same input-strand surface density (4.3 AE 0.3 Â 10 À10 mol cm À2 ) and maintaining a xed electronic input period (10 s). While applied potentials more negative than À0.8 V vs. Ag/AgCl leads to a complete activation of the nanoswitch, less reductive potentials results in a gradual decrease of the fraction of activated nanoswitches from 84 AE 3% (À0.7 V) to 8 AE 2% (0.0 V) (Fig. 3E). Both the examples shown above demonstrate that the electronic approach proposed here could achieve a gradual and quantitative activation of DNA nanodevices.
As a further demonstration of how electrochemistry can improve the current toolkit of possible inputs in the eld of DNA-based nanotechnology, we also demonstrate here the possibility to regulate, using an electronic input, the activity of nucleic acid enzymes (i.e. DNAzymes). DNAzymes are naturally occurring or in vitro selected RNAs or DNAs that catalyze specic chemical reactions. We employed here a well-known DNAzyme (developed by Breaker and coworkers) displaying a Cu(II)induced nuclease activity. 20 The functionality of this DNAzyme can be easily followed by uorescence measurement by optically Here fluorescence measurements were performed in 50 mM sodium phosphate, 150 mM NaCl, pH 7.0 at 45 C containing the DNA responsive switch (10 nM). We note that 45 C was chosen considering as the best temperature to achieve a good sensitivity and a high signal-to-noise ratio (see Fig. S8 †).
labelling the two DNA strands composing the DNAzyme (Fig. 4A and S9 †). 21 More specically, the two strands are labelled with a uorophore and a quencher and an increase in uorescence signal is observed as a result of the Cu(II)-triggered self-cleaving activity. To electronically control such Cu(II)-dependent DNAzyme we have used here a gold chip coated with a lm of Cu(0). By applying on the surface of this electrode a ramp of potential from À0.65 V to À0.4 V (vs. Ag/AgCl) we can electronically promote the release of Cu(II) ions 22 in a controlled fashion (Fig. S10 †) and thus trigger the DNAzyme activity (Fig. 4B, red curve). Also in this case, under the same experimental conditions but in the absence of the electronic input, we do not observe any signicant DNAzyme activation (Fig. 4B, black curve). Moreover, by controlling the density of Cu(0) coated on the gold chip (from 7.1 AE 0.6 Â 10 À10 mol cm À2 to 1.4 AE 0.3 Â 10 À12 mol cm À2 ) we can nely modulate the percentage of DNAzyme activation (Fig. 4C and D).

Conclusions
A limitation in the recently growing eld of DNA nanotechnology is associated with the fact that DNA-based nanodevices or nanostructures can be controlled and regulated only by a restricted class of molecular cues (i.e. inputs) thus limiting the possibility to achieve an efficient hierarchical control of DNA nanodevices. In response to this, here we have demonstrated the possibility to use electronic inputs to rationally control and regulate DNA-based nanoswitches and nanodevices. We have demonstrated this approach with four model DNA-based nanodevices that are representative of a larger class of conformational-switching and DNA-based enzymes. Of note, each of these model systems is activated by a specic and different input cue that ranges from heavy metal ions (Hg(II), Ag(I) and Cu(II)) to a specic oligonucleotide strand.
To electronically activate these DNA-based nanodevices we have used as electronic input a voltage potential applied at the surface of an electrode chip. Such applied potential promotes an electron-transfer reaction at the electrode-solution interphase leading to the release of a molecular input from the electrode surface that ultimately triggers the DNA-based nanodevice in solution. By varying the electronic input we demonstrate that we can rationally modulate the activation of the nanodevices in a highly controllable fashion.
The possibility to use electronic inputs as a way to control DNA-based nanodevices together with the low-cost and possible miniaturization of electrochemical instruments represents an important advancement that allows to expand the available toolbox to be used in the eld of DNA nanotechnology thus opening the future to new and exciting avenues. Compared to other examples where DNA-based conformational change is triggered solely through the external addition of an input cue, 23,24 we believe our approach could be used to introduce additional control over the formation and functionality of DNA nanostructures with an unprecedented hierarchical control.
Our approach could for example prove useful in cases where the intervention of external operator should be avoided. In such cases, a programmable potential application over an electrode chip might be used to release the triggering input in a completely operator-free fashion. We also note that while we have reported here a limited number of electronically-released inputs, we can easily envision the possibility to use such approach to release or activate other molecular inputs through redox reactions.
HPLC puried oligonucleotides were purchased from Biosearch Technologies (Risskov, Denmark) and IBA GmBH (Göttingen, Germany) and employed without further purication. In this work four different systems were employed. The following oligos modied and non-modied were used for each system: (1) Hg(II)-responsive nanoswitch (see Fig. 2 and S1 †). The DNA-based switch triggered by mercury(II) ions has been optimized and characterized elsewhere. 15 The oligo is internally labeled with FAM (5-carboxyuorescein) and BHQ-1 (black hole quencher 1) and has the following sequence:  (2) Ag(I)-responsive nanoswitch (see Fig. S3 †). The DNAbased switch triggered by Ag(I) ions has been optimized and characterized elsewhere. 15 The oligo is internally labeled with AF680 (Alexa Fluor 680) and BHQ-2 (black hole quencher 2) and has the following sequence: 5 0 -TTTTATTTAATTATA TTATTAAT T (BHQ2) CCTACTT TCATC T (AF680) ATTAATAA CATCAAACTACC-3 0 (3) DNA-responsive nanoswitch (see Fig. 3 and S8 †). The DNA-responsive nanoswitch is a molecular beacon containing a 5-base stem and it is labeled with FAM (5-carboxyuorescein) and a BHQ-1 (black hole quencher 1) and has the following sequence: In the sequence above the underlined bases represent the stem portion, while the italic bases represent the recognition element of the binding-state.
(4) Cu(II)-responsive DNAzyme (see Fig. 4 and S9 †). The Cu(II)-responsive DNAzyme has been also characterized elsewhere. 21 The system is composed of two different strands. The rst strand is labeled with AF680 (Alexa Fluor 680) and BHQ-2 (black hole quencher 2) and the second strand is labeled with BHQ-2 (black hole quencher 2). The sequences of the two strands are given below: Strand 1: 5 0 -(BHQ2) AGC TTC TTT CTA ATA CGG CTT ACC (AF680)-3 0 Strand 2: 5 0 -(BHQ2) GGT AAG CCT GGG CCT CTT TCT TTT TAA GAA AGA AC-3 0 See the cartoon in Fig. 4 to clarify the copper-binding site and mechanism for obtaining information on the activity of the DNAzyme.

Electronic activation of DNA-based nanodevices
In this work we report the use of electronic inputs to activate all the four different DNA-based nanodevices described above. All experiments were performed using a portable PalmSens potentiostat instrument connected to a laptop. Briey, the molecular input of each DNA-based nanodevice has been deposited onto the surface of a screen printed disposable electrode. Through an electrochemical input (applied potential) we have released the molecular input in a controlled way and we have thus triggered the DNA-based nanodevice. The occurred activation of the DNA-based nanodevice has been followed through uorescent measurement. The detailed experimental procedure employed for each system is described in the ESI document. †

Standard binding curves
Standard binding curves were obtained for each system employed in this work by adding at increasing concentrations the specic molecular input and measuring the uorescence signal. For each system the same buffer solution used for the electronic activation experiment (see ESI †) and the same uorometer parameters were used. The targets used for each system are the following: HgCl 2 (as a source of Hg(II) ions) for Hg(II)responsive nanoswitch, AgNO 3 (as a source of Ag(I) ions) for Agresponsive nanoswitch, the thiol-labeled strand target for DNAresponsive nanoswitch, CuNO 3 (as a source of Cu(II) ions) for the Cu(II)-responsive DNAzyme.
The observed uorescence, F [target] , was tted using the following four parameter logistic equation 25 where, F min and F max are the minimum and maximum uorescence values, K 1/2 is the equilibrium target concentration at half-maximum signal, n H is the Hill coefficient, and [target] is the concentration of the target added. This model is not necessarily physically relevant, but it does a good (empirical) job of tting effectively binding curves such as those we obtain for most of our systems employed in this work.