All-in-one Biocomputing Nanoagents with Multilayered Transformable Architecture based on DNA Interfaces

Abstract

The pathogen diversity to infiltrate the host organism highlights the demand for equally sophisticated mechanisms for their prevention. The development of "intelligent" agents with molecular logic capabilities are of great hope, but their full theranostic potential has yet to be realized. Methods: The original concept of nanoagents based on "Biocomputing based on particle disassembly" technology has been extended to nucleic acids (NAs) interfaces and inputs. By exploiting the unique properties of NAs, we designed nanostructures that can implement all basic single- and dual-input logic gates on a unified nanoparticle platform through DNA strand displacement triggered by oligonucleotide inputs. Performance of nanostructures was investigated across various output signal detection formats including specific interaction with nanosized objects and targeting cells. Results: Here, we demonstrate autonomous theranostic biocomputing agents based on nanoparticles and DNA interfaces ("DNA-transformers") capable of executing a functionally complete set of Boolean logic gates (YES, NOT, AND, and OR) within a single all-in-one particle structure. Each DNA-transformer is constructed through a multi-layered self-assembly of nanoparticles via DNA-interfaces. The route of the agent's disassembly induced by the particular combination of the specific ssDNA inputs determines the agents' ability to produce the programmed outputs compatible with theranostic applications such as specific targeting of HER2/neu-positive cancer cells. Conclusions: The developed all-in-one DNA-based nanoagents represent a significant advancement in molecular logic devices, establishing a versatile platform for smart nanoagents equally suitable for diagnostic and therapeutic applications.

Keywords: DNA interfaces; HER2/neu cell targeting; biocomputing; smart materials; strand displacement.

Figure 1

General scheme for implementing the simplest logic YES gate based on DTs. (A) Self-assembly of DTs to implement the YES gate; (B) DT affinity switching scheme by free NA fragments (INPUT); (C) options for generating the output signal as a result of executing the YES function. Strep/HRP denotes a conjugate of streptavidin and horseradish peroxidase.

Figure 2

Possible schemes for implementing one-, two- and four-input logic functions using toehold-mediated strand displacement on the DT surface. The detailed scheme of the composite two-input logic gate operating is depicted in Figure S1.

Figure 3

Options for constructing a single-input logic YES gate based on DTs. (A) Input and output receptors, which are certain nucleotide sequences and biotin, respectively, are located on the same DNA molecule (i/o-rec, blue) immobilized on a core magnetic polymer particle (Core Particle). Together with i/o-rec ligand (partially complementary to i/o-rec, green) immobilized on SN, they form “DNA-transformer” (DTs) and biotin in shielded (not active) state. Such a construction can be further disassembled both by attacking with Input1 and/or Input2 (fully complementary oligonucleotides to i/o-rec and to i/o-rec ligand, respectively), forming the Core Particles with non-shielded (activated) biotin. System switching is recorded by a signal generation system (SGS, based on ferrihydrite nanoparticle-streptavidin-HRP conjugate) followed by colorimetric registration of the interaction of HRP with the substrate (TMB/H₂O₂). (B) A variant of the separate location of the input and output receptors on two different oligonucleotides, which are co-immobilized on Core Particle. (C) A variant of the separate arrangement of receptors with multiplexing of the output signal, in which the ligand for the output receptor (o-rec1 ligand) is located on the auxiliary SN conjugate. The specific nucleotide sequences used in the scheme are shown in Figure S2.

Figure 4

Optimization of the DT assembly conditions for the implementation of the YES gate according to the scheme in Figure 3A. (A, B) Comparison of the shielding efficiency of magnetic microspheres of different nature: (A) SEM images (BSE mode, in which heavy metal compounds, primarily gold nanoparticles, are visible). Scale bars correspond to 1 µm; (B) The result of a quantitative comparison of optical signals after interaction with the signal generation system (SGS, conjugate of ferrihydrite particles with streptavidin and horseradish peroxidase) followed by spectrophotometric detection. The numbers indicate the ratio of the signals before (CP) and after (CP+SN) screening of the output ligands on the CPs. One and two asterisks indicate the significance level of the difference (p < 0.05 and 0.01) between shielded and unshielded states of DTs, respectively. The original names of the used commercial particles are used. (C) Dependence of the output optical signal of the reaction of DTs with SGS on the concentration of shielding nanoparticles (black color, incubation time 30 min, CP conc. 3 g/L) and the time of their incubation with CP (red color, GNP and CP conc. 0.15 nM and 3 g/L, respectively). The selected optimal conditions (where the signal reaches a plateau) are indicated by a dotted line.

Figure 5

Options for building CP/SN interfaces for implementing a YES gate and their impact on the DT functionality. (A) Tested combinations of randomly generated oligonucleotides immobilized on CP (blue) and SN (pink) and “staples” connecting them (green). The central column shows the Shield Ratio (Ratio of output signal intensity in unshielded and shielded states of DTs) versus Adjusted Minimum Free Energy (AMFE, the ratio of the minimum internal energy of the complex to the total number of nucleotide pairs in the complementary region). The data were obtained using NUPACK software (http://www.nupack.org/), and specific sequences are shown in Figure S7. (B-D) The dependence of the optical signal on unshielded (Core) and shielded (Shielding) DTs, as well as in the presence of specific (Specific) and non-specific (Non-specific) oligonucleotides, on the type of DNA interface for several variants of its implementation. The signals were obtained after interaction with the signal generation system and normalized to the value of the signal from the CP in the unshielded state. The sequences used are shown in Figure S7.

Figure 6

Demonstration of the operation of YES and NOT gates. Implementation scheme for YES (A,C) and NOT (E,G) gates relative to two INPUT DNA sequences (InA and InB). (B), (D), (F), (H) - Dependences of DT optical signals in YES (B,D) and NOT (F,H) gates in response to the presence of specific (InA and InB) and nonspecific (InN) oligonucleotides (10^‑6 M). The signals are normalized by the corresponding 'No input' signal, and the trigger threshold (geometric average of the maximum and minimum output signals) is indicated by a red dotted line. The sequences of the oligonucleotides used are shown in Figure S8.

Figure 7

Demonstration of two-input OR and AND gates. Implementation scheme for OR (A) and AND (C) gates relative to the combination of INPUT DNA (InA and InB). Dependences of optical signals of DTs realizing OR (B) and AND (D) gates in response to the presence of specific (InA and InB) and nonspecific (InN) oligonucleotides (10^-6 M). Absorbance signals are normalized to the corresponding 'No input' signal. The trigger threshold (geometric average of the maximum and minimum output signals) is indicated by a red dotted line. The sequences of the oligonucleotides used are shown in Figure S9 and S10.

Figure 8

Cell targeting as result of the YES gate implementation. Flow cytometry (A,B) and imaging flow cytometry (C-E) data on registration of the interaction of HER2-positive (SK‑BR‑3) and HER2-negative (CHO) cells with DTs in the presence of specific and non-specific input oligonucleotide. Binding to cells (output = 1) occurred through the interaction of output DT ligands (DNA-bound biotin) with surface HER2 receptors preliminary labeled with anti-HER2/streptavidin conjugate. Histograms of signal distribution in side scatter (SSC) channels (A, C) were generated, and the operation of the YES gate was demonstrated by comparing the medians of the output signals (B). The thresholds for the YES gate (geometric mean of the maximum and minimum output signals) were indicated by a dotted line for both types of the cells. Representative images in bright-field (BF) and SSC channels (D) were recorded, and the specificity of DTs-cell interactions was analyzed using convolutional neural network analysis with the number of bound particles as a quantitative criterion (E). Designations: Cells only - background signal (cells in the absence of particles); Core w/o Mabs - signal from cell interaction with DTs in the absence of preincubation with anti-HER2-Str conjugate (specificity control); Core, Shield - signals from cell interaction with CP and CP+SN components of DTs, respectively; +InA, +InB, +InA,InB, +InN - signals from the interaction of cells with fully assembled DTs in the presence of the first (specific) and second (nonspecific) inputs (10^-6 M), their joint combination and a nonspecific oligonucleotide of arbitrary composition, respectively. The sequences of the oligonucleotides used are shown in Figure S9 and S10.

Figure 9

Cell targeting as result of the AND gate implementation. Flow cytometry (A,B) and imaging flow cytometry (C-E) data on registration of the interaction of HER2-positive (SK‑BR‑3, BT-474, CT-26) and HER2-negative (CHO) cells with DTs in the presence of specific and non-specific input oligonucleotide. Histograms of signal distribution in side scatter (SSC) channels (A, C) were generated, and the operation of the AND gate was demonstrated by comparing the medians of the output signals (B). AND-gates switching thresholds are indicated by a dotted line for all types of cells in compliance with color coding; (D) Representative images in BF and SSC channels; (E) Results of convolutional neural network analysis of the specificity of DT-cell interactions. Further details and designations can be found in the caption of Figure 8.