A label-free and enzyme-free platform with a visible output for constructing versatile logic gates using caged G-quadruplex as the signal transducer† †Electronic supplementary information (ESI) available: Experimental details and supplementary figures. See DOI: 10.1039/c7sc04007e

A complete set of elementary logic gates and two multilevel circuits have been constructed on a label-free and enzyme-free biocomputing platform using caged G-quadruplex as the signal transducer.


General procedures for logic gate operation
All assays were performed in 20 mM Tris-HCl buffer (pH = 7.4, 200 mM NaCl, 20 mM MgCl 2 ) at a final DNA concentration of 1 M. The samples, without the inputs, were heated to 90 o C for 10 min, and then cooled down to room temperature for 90 min. The input was added to the above solution, and the mixture was incubated for 120 min at room temperature. The resulting solution was used for subsequent colorimetric detection.

Colorimetric analysis
10 L of 20 mM Tris-HCl buffer solution (pH = 7.4, containing 2 M hemin, 15 mM KCl, 5 mM MgCl 2 , 0.03% Triton X-100, and 1% DMSO) was added to 40 L of the above solution, and the mixture was incubated at room temperature for 45 min to form the active hemin/G-quadruplex

Circular dichroism (CD) experiment
CD spectra were measured on a Chirascan Circular Dichroism Spectrometer (Applied Photophysics Ltd, England, UK) at room temperature. The DNA solution (3 M) was prepared in 20 mM Tris-HAc buffer (pH 7.4) containing 50 mM KAc and 50 mM Mg(NO 3 ) 2 . The solution was heated to 90 o C for 5 min, gradually cooled to room, and incubated at RT for 3h. After incubation with 3 M hemin for 1 h, CD spectra from 320 to 220 nm were recorded in 1 mm path length cuvettes and averaged from three scans with the buffer background subtracted.

Effects of pH and ion strength
To investigate the effects of pH and ion strength on the response of the logic system, the (1,1) state of the AND logic gate was tested under different pH and ion strength. As shown in Fig. S1, a wide range of pH (from 5 to 9) and ion strength (with NaCl concentrations from 50 to 400 mM) did not change the logic functions of the computation states, indicating the robustness of this logic system which is beneficial for practical applications.

Circular dichroism results
Taking the AND logic gate as an example, the formation of the G-quadruplex structure was verified by circular dichroism (CD) spectroscopy. As shown in Fig. S2, there was no obvious peak in the (0,0) state of the AND gate. In the (1,0) or (0,1) state, the intermediate DNA structure showed a positive peak near 275 nm and a negative peak near 245 nm, which referred to the typical characteristic of double helix structure. 3,4 In the presence of both inputs (1,1), the final DNA structure had two characteristic peaks at 264 and 240 nm. The positions of the positive and negative peaks match the reported characteristic peaks of the parallel G-quadruplex, 5,6 indicating the successful formation of the G-quadruplex structure in the (1,1) state of the AND logic gate.

XNOR logic gate
The construction of the XNOR logic gate is outlined in Fig. S3. An additional DNA template (DNA1) is introduced to bind with the DNAzyme subunits (DNA2 and DNA3) to form an active DNAzyme. This original state (0,0) results in the cleavage of the substrate (DNA6), releases the caged G-quadruplex sequence, and gives rise to the output of 1. The addition of either input could disassemble the active DNAzyme structure and the output is 0. The overhung segments C and D in DNA1 serve as the toehold domains and proceed to invade and displace the previously hybridized DNAzyme subunits vis the toehold-mediated strand-displacement reaction in presence of input1 (1,0) or input2 (0,1). If both inputs are present (1,1), the partial duplex between input1 and input2 is formed (domain E is complementary with domain E' ). The cooperative interactions of the DNAzyme subunits (DNA4 and DNA5) with the input1-input2 yields the assembly of another active DANzyme, consequently leading to the cleavage of DNA6 and resulting in blue solution. Fig. S4A presents typical photo images of the XNOR gate. Fig. S4B depicts absorption spectra from 500 to 800 nm. Fig.   S4C shows the corresponding absorption intensity at  = 650 nm. This gate has an output of 1 (blue solution) only if the two inputs are present or absent. The truth table and circuitry are given in Fig.   S4D and S4E, respectively.  Electronic equivalent circuitry.

NAND logic gate
A NAND gate is the inverted result of an AND gate, which causes all 0 output to switch to 1 and vice versa. As shown in Fig. S5, in the absence of any input, the template DNA1 bridges the DNAzyme subunits (DNA2-DNA4) to assemble two different active DNAzymes, leading to the cleavage of the mutual substrate DNA6 and to the generation of a true output. In the presence of either input1 or input2, the input-initiated toehold-mediated strand-displacement reaction could disassemble one of the formed two active DNAzyme structure using segments C and F as the toehold binding domains. The other DNAzyme keeps intact and active that leads to the cleavage of the substrate and also yields a true output. In the presence of both inputs, the hybridization of the inputs to the template DNA1 can eliminate the two active DNAzymes, leading to a false output. Fig. S6A presents typical photo images of the NAND gate. Fig. S6B depicts absorption spectra from 500 to 800 nm. Fig. S6C shows the corresponding absorption intensity at  = 650 nm. The NAND logic gate has an output of 0 only if both inputs are 1. The truth table and circuitry are given in Fig. S6D and S6E, respectively.

NOR logic gate
A NOR gate is the result of the negation of an OR gate. As illustrated in Fig. S7, in the (0,0) state, the template DNA1 is hybridized with the DNAzyme subunits (DNA2 and DNA3) to construct the active DNAzyme structure that cleaves the substrate (DNA4) and yields an output of 1. In the presence of either input1 or input2, the input could disassemble the active DNAzyme structure from the template vis the toehold-mediated strand-displacement reaction using segments C and D as the toehold binding domains. This leads to the false output of the gate. Similarly, in the presence of both inputs, the template is more favorably hybridized with the inputs to form duplexes by displacement mechanism. This process results in the dissociation of the active DNAzyme, consequently leading to an output of 0. Fig. S8A presents typical photo images of the NOR gate. Fig. S8B depicts absorption spectra from 500 to 800 nm. Fig. S8C shows the corresponding absorption intensity at  = 650 nm.
The NOR logic gate has an output of 1 only if both inputs are 0. The truth table and circuitry are given in Fig. S8D and S8E, respectively.  Electronic equivalent circuitry.

INHIBIT logic gate
An INHIBIT logic gate, which consists of two DNAzyme subunits (DNA1 and DNA2), the substrate DNA3, and the input DNA, is also designed. As illustrated in Fig. S9, in the presence of input1 (1,0), an active DANzyme is formed, leading to the cleavage of the substrate and to the generation of a blue color signal. In the presence of another input (0,1), the system fails to yield any active DANzyme for readout because of the identical DNA sequences between the recognition arms (segments A' and B') of the DNAzyme subunits and the input2. Triggering the system with both inputs (1,1) results in the formation of the duplex structure between input1 and input2. This prohibits the formation of an active DANzyme and the generation of a true output. Fig. S10A presents typical photo images of the INHIBIT gate. Fig. S10B depicts absorption spectra from 500 to 800 nm. and only if one input is 1; otherwise, the ouput is 0. The truth table and circuitry are given in Fig. S10D and S10E, respectively.

IMPLICATION logic gate
An IMPLICATION logic gate is the reverse output result of an INHIBIT logic gate. As shown in Fig. S11, an additional DNA template (DNA1) can hybridize with the DNAzyme subunits (DNA2 and DNA3) to form an active DNAzyme, giving rise to the output of 1 in the absence of input. To trigger the disassembly of the active DNAzyme, the input1 (1,0) is introduced to initiate the toehold-mediated strand-displacement reaction to displace one DNAzyme subunit (DNA2) from the DNA template. As expected, the disassembly of the active DNAzyme results in a false output. In the (0,1) and (1,1) combinations, the strand displacement reactions are prohibited for the disassembly processes, because there is no toehold binding between the input and the template DNA. Thus, the formed active DNAzyme keeps the intact structure for the cleavage of the substrate. Fig. S12A presents typical photo images of the IMPLICATION gate. Fig. S12B depicts absorption spectra from S16 500 to 800 nm. Fig. S12C shows the corresponding absorption intensity at  = 650 nm. The IMPLICATION logic gate provides the true output of 1 in all circumstances, except for the situation that one particular input is 1. The truth table and circuitry are given in Fig. S12D and S12E, respectively.

Practical applications of the logic gates
To demonstrate the feasibility of the logic system for practical applications, the XOR, OR, AND, and XOR+AND logic gates were validated in human serum samples. Serum samples were prepared by diluting to 10% in dilution buffer (20 mM Tris-HCl, 200 mM NaCl, 20 mM MgCl 2 , pH 7.4). The input1 and input2 spiked in 10% human serum were used as inputs to perform the logic operations.
As shown in Fig. S13A, in the presence of either spiked input1 (1,0) or input2 (0,1), the solution turned blue (output = 1). In the absence of input (0,0) or in the presence of both spiked inputs (1,1), the logic gate failed to generate a blue solution (output = 0). Those behaviors corresponded to the XOR logic gate. Fig. S13B depicts absorption spectra from 500 to 800 nm. Fig. S13C shows the corresponding absorption intensity at  = 650 nm. The truth table and circuitry are given in Fig.   S13D and S13E, respectively. As shown in Fig. S14A, without the spiked input1 and input2 (0,0), the solution is colorless in the OR gate, but when serum samples were spiked with either input1 (1,0) or input2 (0,1) or both inputs (1,1), the blue color solution can be observed. Fig. S14B depicts absorption spectra from 500 to 800 nm. Fig. S14C shows the corresponding absorption intensity at  = 650 nm. The truth table and circuitry are given in Fig. S14D and S14E, respectively. As shown in Fig. S15A, the blue color solution was generated only when the AND gate was subjected to the two spiked inputs together (1,1). The other three input combinations all fail to produce an obvious colorimetric response. Fig. S15B depicts absorption spectra from 500 to 800 nm. Fig. S15C shows the corresponding absorption intensity at  = 650 nm. The truth table and circuitry are given in Fig.   S15D and S15E, respectively. As shown in Fig. S16A, in the presence of either spiked input1 (1,0) or input2 (0,1), the solution turned blue (output = 1), which corresponded to the XOR logic gate. When the integrated circuit was subjected to the two spiked inputs together (1,1), the solution also turned blue, which corresponded to the AND logic gate. Thus, an integrated circuit established form the XOR and AND gates could be performed parallelly in human serum samples. Fig. S16B depicts absorption spectra from 500 to 800 nm. Fig. S16C shows the corresponding absorption intensity at  = 650 nm. The truth table and circuitry are given in Fig. S16D and S16E, respectively. These results indicate that the proposed logic system performs well even in relatively complex sample matrixes and is not affected when the inputs existed in human serum samples, which holds great promise for POC diagnostics and in-field applications.   (E) Electronic equivalent circuitry.