Orthogonal Enzyme-Driven Timers for DNA Strand Displacement Reactions

Here, we demonstrate a strategy to rationally program a delayed onset of toehold-mediated DNA strand displacement reactions (SDRs). The approach is based on blocker strands that efficiently inhibit the strand displacement by binding to the toehold domain of the target DNA. Specific enzymatic degradation of the blocker strand subsequently enables SDR. The kinetics of the blocker enzymatic degradation thus controls the time at which the SDR starts. By varying the concentration of the blocker strand and the concentration of the enzyme, we show that we can finely tune and modulate the delayed onset of SDR. Additionally, we show that the strategy is versatile and can be orthogonally controlled by different enzymes each specifically targeting a different blocker strand. We designed and established three different delayed SDRs using RNase H and two DNA repair enzymes (formamidopyrimidine DNA glycosylase and uracil-DNA glycosylase) and corresponding blockers. The achieved temporal delay can be programed with high flexibility without undesired leak and can be conveniently predicted using kinetic modeling. Finally, we show three possible applications of the delayed SDRs to temporally control the ligand release from a DNA nanodevice, the inhibition of a target protein by a DNA aptamer, and the output signal generated by a DNA logic circuit.


1.Experimental Procedures Oligonucleotides
Sequences for RNase H, UDG and Fpg controlled strand displacement systems are listed below. For the ligand-binding device sequence the bold bases represent the portion complementary to the output, the underlined bases represent the ligand-binding site (through Watson-Crick bonds) and italics bases represent the triplex-forming portion.

System 5: RNase H-based temporal-controlled of thrombin activity
Name Sequence Target  5'-CCA ACC ACA CCA ACC TCT CTC CTT TCT CTG ATA CT -3'   Output  5'-GGT TGG TGT GGT TGG -3' RNA-blocker 5'-UCA GAG AAA GGA GAG A -3' The output (thrombin-binding aptamer) and target sequences represent the two  were added. After the stabilization of the signal, the corresponding enzyme was added in the wells at the desired concentration and the fluorescence intensity was recorded over time.

Orthogonal temporal control of strand displacement reactions
All experiments shown in Figure 4 were performed at 30°C in 20 mM Tris-HCl buffer, 10 mM MgCl2, 1 mM EDTA, pH 8.0. The target duplex for each enzyme-controlled strand displacement system were incubated together at 60°C for 2 minutes, using equimolar concentrations of the two relevant strands (target and output) and the desired concentration of blockers. After 30 minutes the preformed duplexes were transferred (100 µL) to 96-well plates where DTT (10 mM) BSA (0.1 mg/mL) and the three input strands (50 nM) were added. After the stabilization of the signal, the three enzymes (at the desired concentration) were added into the wells at the same time and the fluorescence intensity was recorded over time.

UDG-based delay of DNA ligand release from a DNA device
All experiments shown in Figure 5 were performed at 25°C in 10 mM Tris-HCl buffer, 3 mM MgCl2, pH 6.0. The target duplex was prepared as previously described and then transferred (100 µL) to a 96-well plates, where the ligand-binding device (50 nM), the ligand (50 nM) and S7 the input (50 nM) were added. After the stabilization of the signal, the enzyme (at the desired concentration) was added into the well.

Enzyme-based temporally controlled DNA logic circuit
All experiments shown in Figure 7 were performed at 30°C in 20 mM Tris-HCl buffer, 10 mM MgCl2, 1 mM EDTA, pH 8.0. The target duplexes for each enzyme-controlled strand displacement system were incubated together at 60°C for 2 minutes, using equimolar concentrations of the two relevant strands (target and output) and the desired concentration of blockers (150 nM). After 30 minutes the preformed duplexes were transferred (100 µL) to a 96-well plate where DTT (10 mM) and the input strand (100 nM) were added. After the stabilization of the signal, the two enzymes (at the desired concentration) were added into the wells at the same time and the fluorescence intensity for the two outputs was recorded over time.

Emission spectra measurements
Fluorescence spectra measurements were carried out on a Cary Eclipse Fluorimeter

Native-PAGE gel-imaging
All experiments shown in Figures S2,S7 and S10 were performed at 30°C in 20 mM Tris-HCl buffer, 10 mM MgCl2, 1 mM EDTA, pH 8.0. The initial target duplexes (50 nM) were formed using the procedure previously described. After 30 minutes the input strand (50 nM) was added to the preformed complex and the solution was incubated at 30°C for 15 minutes.
Enzyme (at the desired concentration) was then added into the reaction mixture. The reactions were stopped at different times by incubation at 4°C. Samples were diluted to 30 nM in a buffer solution containing 60% glycerol and directly loaded on 15% acrylamide native-PAGE. The gel was allowed to run for 120 minutes at 120 V. The bands were detected by direct gel-imaging using the ChemiDocTM MP imaging system (Bio-Rad).

Light scattering experiments RNase H temporally-controlled thrombin activity
The target duplex (50 nM) was formed by incubating at 60°C for 2 minutes at equimolar concentrations of the two relevant strands (target and output) with the desired concentration of blocker in 20 mM Tris-HCl buffer, 10 mM MgCl2, 1 mM EDTA, 150 mM NaCl, pH 8.0. The input strand (50 nM) was then added to the preformed complex and the solution was incubated at 30°C for 15 minutes. RNase H (at the desired concentration) was then added into the reaction mixture. Aliquots of 117 μL were withdrawn at different times from the above reaction mixture and mixed with 13 μL of fibrinogen (1mg/mL) and then transferred into a quartz cuvette. After the addition of thrombin (0.5 nM) the time-dependent light scattering changes due to the formation and aggregation of insoluble fibrin resulting from thrombin enzymatic activity were monitored at 350 nm at 25°C using a Varian Cary 100 UV-Vis spectrophotometer.

2.Kinetic modelling and curve fitting a. General considerations
To model the experimental results obtained in this work, we developed a minimalistic reaction schemes that included only kinetically-relevant reaction steps by neglecting very fast forward reactions and slow reverse reactions. 1 In particular, strand displacement reactions were described as single-step, irreversible reactions and for enzymatic reactions, protein activity is always assumed to happen under saturation conditions (i.e. concentrations of protein and substrates well above the ). All considered reaction steps were assumed to be of first or second order. Numeric integration of the sets of differential equations resulting from the reaction schemes provided the time courses of the concentrations of the different reaction species.
The conventional strand displacement reactions ( Figure S4) were assumed to follow simple second order kinetics with the displacement rate constant displ : where the target strand (A) and the output strand (B) form an initial target duplex (AB), which reacts with an input strand (C) to release the output. The set of rate equations, that describe strand displacement are: The obtained time-course of the output concentration was related to the measured fluorescence signal using a linear scaling factor S: b. General rate model for DNA strand displacement reactions with enzyme-

based delays
For the delayed strand displacement reactions, first a general/complete reaction scheme is developed that is applicable to the three different enzyme systems ( Figure S5). Adaptation of the model to the particular enzyme system allowed partial exclusion of certain reaction steps.
In the starting configuration, the blocker strands (O) are bound to the toehold region of the initial target duplex (AB) to form the complex ABO. Degradation of the blocker strands on the output-target complex is assumed to occur under pseudo-first order conditions with rate constant deg such that it is described by the reaction: during which unblocked output-target complexes AB as well as degraded blocker fragments R are produced. After unblocking, the target duplex can either rebind another blocker strand with the rate constant bind (as long as intact blockers are in solution): or react with the input strand C to release the output strand B by strand displacement: Furthermore, we allow the parallel degradation of unbound (free) blocker strands in solution with the rate constant deg : where is a dimensionless scaling factor for the degradation rate constant of bound blocker deg . Given pseudo-first order conditions, deg is generally assumed to be proportional to the enzyme concentration [E] but can become reduced by enzyme-inhibition due to accumulation of degraded fragments R. For convenience we took the enzyme concentration S10 given in U/ml as a dimensionless scaling factor for deg . Using the Michaelis-Menten scheme for competitive inhibition allows to express the degradation rate as: where E is the degradation rate at 1 U/ml and R is the dissociation constant of the degraded fragments from the enzyme.
The set of differential equations describing the general The obtained time course for the output strands (B) can again be related to measured fluorescence signal using Equation 4.

c. Curve fitting
All curve fitting was performed using self-written Python (version 3.7) scripts. The scripts use the "odeint" function to numerically integrate a set of ordinary differential equations. The solution was then fitted to the measured kinetics using the nonlinear least-squares function

e. Rate model for the Fpg-based strand displacement reactions with delay
For the Fpg-based reactions (data Figure 3b and Figure S6c,d), free blocker degradation can be neglected ( = 0, see RNAse-H system above), because Fpg shows negligible activity on single stranded compared to double stranded DNA 3 , Fpg inhibition by waste fragments was however considered, since obtained values for were in the order of the blocker concentration.
The general fitting approach was the same as described for the RNase H system. We

f. Rate model for the UDG-based strand displacement reactions with delay
For the UDG-based reactions (Figure 3f and Figure S9c,d), we considered free blocker degradation as well as enzyme inhibition by degradation fragments. UDG has a significantly higher activity on single stranded compared to double stranded DNA 4,5 , such > 0.
The general fitting approach was the same as described for the other two enzymes. We This modifies the previous set of differential equations to: S13 The measured fluorescence decrease was related to concentration of liberated cargo using a linear relation: where 0 is the intensity of the fully bound cargo and the intensity after full release of the cargo, which were free parameters for each trace. Additionally, we included a small fraction of output strands [ ] 0 , which were already present at the start of the reaction due to small errors in pipetting the initial target duplex AB.
Since the reaction conditions for the application of the UDG system differed from that of the characterization (i.e. buffer composition, temperature and pH-value), we employed a new set of rate constants. From a global fit of the cargo release data (Figure 5b (Figure 2). In the absence of the enzyme (Ctrl NO RNase H) the reaction does not proceed. In the absence of the blocker strand (Ctrl NO blocker) the reaction proceeds to completion within 30 minutes. Control using a DNA blocker (Ctrl DNA blocker) shows that no reaction is observed even after the addition of the RNase H enzyme. Experimental conditions used here are the same as in Figure 2.      (Fig. 3). In the absence of the enzyme (Ctrl NO Fpg) the reaction does not proceed. In the absence of the blocker strand (Ctrl NO blocker) the reaction proceeds to completion within 30 minutes. Control using a DNA blocker (Ctrl DNA blocker) shows that no reaction is observed even after the addition of the Fpg enzyme. Experimental conditions used here are the same used in Figure 3. Figure S8. Fluorescence emission spectra of the delayed Fpg-based DNA strand displacement reactions. a) Scheme of the reaction studied. b) Fluorescent spectra at 3 representative times for the same reactions and controls showed in Figure S7.   (Fig. 3). In the absence of the enzyme (Ctrl NO UDG) the reaction does not proceed. In the absence of the blocker strand (Ctrl NO blocker) the reaction proceeds to completion within 30 minutes. Control using a DNA blocker (Ctrl DNA blocker) shows that no reaction is observed even after the addition of the UDG enzyme. Experimental conditions used here are the same used in Figure 3. Figure S11. Fluorescence emission spectra of the delayed UDG-based DNA strand displacement reactions. a) scheme of the reaction studied. b) Fluorescent spectra at 3 representative times for the same reactions and controls showed in Figure S9.