Predicting DNA toehold-mediated strand displacement rate constants using a DNA-BERT transformer deep learning model

Dynamic DNA nanotechnology is driving exciting developments in molecular computing, cargo delivery, sensing and detection. Combining this innovative area of research with the progress made in machine learning will aid in the design of sophisticated DNA machinery. Herein, we present a novel framework based on a transformer architecture and a deep learning model which can predict the rate constant of toehold-mediated strand displacement, the underlying process in dynamic DNA nanotechnology. Initially, a dataset of 4450 DNA sequences and corresponding rate constants were generated in-silico using KinDA. Subsequently, a 1D convolution neural network was trained using specific local features and DNA-BERT sequence embedding to produce predicted rate constants. As a result, the newly trained deep learning model predicted toehold-mediated strand displacement rate constants with a root mean square error of 0.76, during testing. These findings demonstrate that DNA-BERT can improve prediction accuracy, negating the need for extensive computational simulations or experimentation. Finally, the impact of various local features during model training is discussed, and a detailed comparison between the One-hot encoder and DNA-BERT sequences representation methods is presented.

An approach for state differentiation in nucleic acid circuits: Application to diagnostic DNA computing

BACKGROUND

Differentiating between different states in nucleic acid circuits is crucial for various biological applications. One approach, there is a requirement for complicated sequential summation, which can be excessive for practical purposes. By selectively labeling biologically significant states, this study tackles the issue and presents a more cost-effective and streamlined solution. The challenge is to efficiently distinguish between different states in a nucleic acid circuit.

RESULTS

An innovative method is introduced in this study to distinguish between states in a nucleic acid circuit, emphasizing the biologically relevant ones. The circuit comprises four DNA logic gates and two detection modules, one for determining fetal gender and the other for diagnosing X-linked genetic disorders. The primary module generates a G-quadruplex DNAzyme when activated by specific biomarkers, which leads to a distinct colorimetric signal. The secondary module responds to hemophilia and choroideremia biomarkers, generating one or two DNAzymes. The absence of female fetus indicators results in no DNAzyme or color change. The circuit can differentiate various fetal states by producing one to four active DNAzymes in response to male fetus biomarkers. A single-color solution for state differentiation is provided by this approach, which promises significant advancements in DNA computing and diagnostic applications.

SIGNIFICANCE

The innovative approach used in this study to distinguish states in nucleic acid circuits holds great significance. By selectively labeling biologically relevant states, circuit design is simplified and complexity is reduced. This advancement enables cost-effective and efficient diagnostic applications and contributes to DNA computing, providing a valuable solution to a fundamental problem.

Heterochiral modifications enhance robustness and function of DNA in living human cells

Oligonucleotide therapeutics are becoming increasingly important as more are approved by the FDA, both for treatment and vaccination. Similarly, dynamic DNA nanotechnology is a promising technique that can be used to sense exogenous input molecules or endogenous biomarkers and integrate the results of multiple sensing reactions in situ via a programmed cascade of reactions. The combination of these two technologies could be highly impactful in biomedicine by enabling smart oligonucleotide therapeutics that can autonomously sense and respond to a disease state. A particular challenge, however, is the limited lifetime of standard nucleic acid components in living cells and organisms due to degradation by endogenous nucleases. In this work, we address this challenge by incorporating mirror-image, ʟ-DNA nucleotides to produce heterochiral "gapmers". We use dynamic DNA nanotechnology to show that these modifications keep the oligonucleotide intact in living human cells for longer than an unmodified strand. To this end, we used a sequential transfection protocol for delivering multiple nucleic acids into living human cells while providing enhanced confidence that subsequent interactions are actually occurring within the cells. Taken together, this work advances the state of the art of ʟ-nucleic acid protection of oligonucleotides and DNA circuitry for applications in vivo.

Engineering protein and DNA tools for creating DNA-dependent protein switches

Switchable proteins are capable of changing conformations from inactive (OFF) to active (ON) forms in response to inputs such as ligand binding, pH or temperature change, or light absorption. A particularly powerful class of protein switches, exemplified by the Cas nucleases of CRISPR systems, are activated by binding of specific DNA or RNA sequences. The mechanism by which oligonucleotide binding regulates biological activity is complex and highly specialized in the case of Cas enzymes, but recent advancements in protein and DNA engineering have made it possible to introduce this mode of control into other enzymes. This chapter highlights recent examples of protein switches that combine these two fields of engineering for the purpose of creating biosensors that detect pathogen and other genomic sequences. One protein engineering method-alternate frame folding-has the potential to convert many proteins into ligand-activated switches by inserting a binding protein (input domain) into an enzyme (output domain). The steps for doing so are illustrated using GCN4 as a DNA recognition domain and nanoluciferase as a luminescent reporter that changes color as a result of DNA binding. DNA engineering protocols are included for creating DNA tools (de novo designed hairpins and modified aptamers), that enable the biosensor to be activated by arbitrary DNA/RNA sequences and small molecules/proteins, respectively. These methodologies can be applied to other proteins to gain control of their functions by DNA binding.

Short DNA-catalyzed formation of quantum dot-DNA hydrogel for enzyme-free femtomolar specific DNA assay

Fast, sensitive, specific, and user-friendly DNA assay is a key technique for the next generation point-of-care molecular diagnosis. However, high-cost, time-consuming, and complicated enzyme-based DNA amplification step is essential to achieve high sensitivity. Herein, a short target DNA-catalyzed formation of quantum dot (QD)-DNA hydrogel is proposed as a new DNA assay platform satisfying the above requirements. A single-stranded target DNA catalyzes the opening cycle of DNA hairpin loops, which are quickly self-assembled with DNA-functionalized QDs to generate QD-DNA hydrogel. The three-dimensional hydrogel network allows efficient resonance energy transfer, dramatically lowering the limit of detection down to ~6 fM without enzymatic DNA amplification. The QD-DNA hydrogel also enables a rapid detection (1 h) with high specificity even for a single-base mismatch. The clinical applicability of the QD-DNA hydrogel is demonstrated for the Klebsiella pneumoniae carbapenemase gene, one of the key targets of drug-resistant pathogenic bacteria.

DNA Strand Displacement Reaction: A Powerful Tool for Discriminating Single Nucleotide Variants

Single-nucleotide variants (SNVs) that are strongly associated with many genetic diseases and tumors are important both biologically and clinically. Detection of SNVs holds great potential for disease diagnosis and prognosis. Recent advances in DNA nanotechnology have offered numerous principles and strategies amenable to the detection and quantification of SNVs with high sensitivity, specificity, and programmability. In this review, we will focus our discussion on emerging techniques making use of DNA strand displacement, a basic building block in dynamic DNA nanotechnology. Based on their operation principles, we classify current SNV detection methods into three main categories, including strategies using toehold-mediated strand displacement reactions, toehold-exchange reactions, and enzyme-mediated strand displacement reactions. These detection methods discriminate SNVs from their wild-type counterparts through subtle differences in thermodynamics, kinetics, or response to enzymatic manipulation. The remarkable programmability of dynamic DNA nanotechnology also allows the predictable design and flexible operation of diverse strand displacement probes and/or primers. Here, we offer a systematic survey of current strategies, with an emphasis on the molecular mechanisms and their applicability to in vitro diagnostics.

Electrochemiluminescence biosensor for miRNA-21 based on toehold-mediated strand displacement amplification with Ru(phen)32+ loaded DNA nanoclews as signal tags

A novel electrochemiluminescence (ECL) biosensor was developed for high sensitive and selective detection of miRNA-21 based on the efficient and specific toehold-mediated strand displacement (TMSD) amplification with Ru(phen)₃²⁺ loaded DNA nanoclews (NCs-Ru(phen)₃²⁺) as signal tags. The stable DNA nanoclews, synthesized by a simple rolling circle amplification reaction, were employed to load with Ru(phen)₃²⁺ efficiently as ECL signal tags to amplify the signals. As for TMSD, the substrate strand (Sub) was initially hybridized with P1 and P2 to form DNA duplex structures with a toehold 1. miRNA-21 could hybridize with the toehold 1 and trigger the TMSD amplification with the help of assist strand, releasing lots of P1 stands from DNA duplex structures. The TMSD technique realized the converting and amplification of the single miRNA-21 input to lots of output DNA (namely P1) with good selectivity, simultaneously. Output P1 were designed to expand the stem-locked region of HP, which were immobilized on the Au electrodes firstly. Subsequently, the opened HP could hybridize with the Ru(phen)₃²⁺, capturing the ECL signal tags closed to the sensing surface. The ECL intensity of the system had a linear relationship with the logarithm of the miRNA-21 concentration in the range of 1.0 fM to 100 pM with a limit of detection of 0.65 fM. The strategy was further applied to detect miRNA-21 in complex samples, and the result was consistent with the qRT-PCR.

Decomposable quantum-dots/DNA nanosphere for rapid and ultrasensitive detection of extracellular respiring bacteria

Extracellular respiring bacteria (ERB) are a group of bacteria capable of transferring electrons to extracellular acceptors and have important application in environmental remediation. In this study, a decomposable quantum-dots (QDs)/DNA nanosphere probe was developed for rapid and ultrasensitive detection of ERB. The QDs/DNA nanosphere was self-assembled from QDs-streptavidin conjugate (QDs-SA) and Y-shaped DNA nanostructure that is constructed based on toehold-mediated strand displacement. It can release numerous fluorescent QDs-SA in immunomagnetic separation (IMS)-based immunoassay via simple biotin displacement, which remarkably amplifies the signal of antigen-antibody recognizing event. This QDs/DNA-nanosphere-based IMS-fluorescent immunoassay is ultrasensitive for model ERB Shewanella oneidensis, showing a wide detection range between 1.0 cfu/mL and 1.0 × 10⁸ cfu/mL with a low detection limit of 1.37 cfu/mL. Moreover, the proposed IMS-fluorescent immunoassay exhibits high specificity, acceptable reproducibility and stability. Furthermore, the proposed method shows acceptable recovery (92.4-101.4%) for detection of S. oneidensis spiked in river water samples. The proposed IMS-fluorescent immunoassay advances an intelligent strategy for rapid and ultrasensitive quantitation of low-abundance analyte and thus holds promising potential in food, medical and environmental applications.