Rapid protein evolution by few-shot learning with a protein language model.

Directed evolution of proteins is critical for applications in basic biological research, therapeutics, diagnostics, and sustainability. However, directed evolution methods are labor intensive, cannot efficiently optimize over multiple protein properties, and are often trapped by local maxima. In silico- directed evolution methods incorporating protein language models (PLMs) have the potential to accelerate this engineering process, but current approaches fail to generalize across diverse protein families. We introduce EVOLVEpro, a few-shot active learning framework to rapidly improve protein activity using a combination of PLMs and protein activity predictors, achieving improved activity with as few as four rounds of evolution. EVOLVEpro substantially enhances the efficiency and effectiveness of in silico protein evolution, surpassing current state-of-the-art methods and yielding proteins with up to 100-fold improvement of desired properties. We showcase EVOLVEpro for five proteins across three applications: T7 RNA polymerase for RNA production, a miniature CRISPR nuclease, a prime editor, and an integrase for genome editing, and a monoclonal antibody for epitope binding. These results demonstrate the advantages of few-shot active learning with small amounts of experimental data over zero-shot predictions. EVOLVEpro paves the way for broader applications of AI-guided protein engineering in biology and medicine.

Rapid, Multiplexed, and Enzyme-Free Nucleic Acid Detection Using Programmable Aptamer-Based RNA Switches.

Rapid, simple, and low-cost diagnostic technologies are crucial tools for combatting infectious disease. We describe a class of aptamer-based RNA switches or aptaswitches that recognize target nucleic acid molecules and initiate folding of a reporter aptamer. Aptaswitches can detect virtually any sequence and provide an intense fluorescent readout without intervening enzymes, generating signals in as little as 5 minutes and enabling detection by eye with minimal equipment. Aptaswitches can be used to regulate folding of seven fluorogenic aptamers, providing a general means of controlling aptamers and an array of multiplexable reporter colors. Coupling isothermal amplification reactions with aptaswitches, we reach sensitivities down to 1 RNA copy/µL in one-pot reactions. Application of multiplexed all-in-one reactions against RNA from clinical saliva samples yields an overall accuracy of 96.67% for detection of SARS-CoV-2 in 30 minutes. Aptaswitches are thus versatile tools for nucleic acid detection that are readily integrated into rapid diagnostic assays.

Rapid and Multiplexed Nucleic Acid Detection using Programmable Aptamer-Based RNA Switches.

Rapid, simple, and low-cost diagnostic technologies are crucial tools for combatting infectious disease. Here, we describe a class of aptamer-based RNA switches called aptaswitches that recognize specific target nucleic acid molecules and respond by initiating folding of a reporter aptamer. Aptaswitches can detect virtually any sequence and provide a fast and intense fluorescent readout, generating signals in as little as 5 minutes and enabling detection by eye with minimal equipment. We demonstrate that aptaswitches can be used to regulate folding of six different fluorescent aptamer/fluorogen pairs, providing a general means of controlling aptamer activity and an array of different reporter colors for multiplexing. By coupling isothermal amplification reactions with aptaswitches, we reach sensitivities down to 1 RNA copy/µL in one-pot reactions. Application of multiplexed one-pot reactions against RNA extracted from clinical saliva samples yields an overall accuracy of 96.67% for detection of SARS-CoV-2 in 30 minutes. Aptaswitches are thus versatile tools for nucleic acid detection that can be readily integrated into rapid diagnostic assays.

Computational Design of RNA Toehold-Mediated Translation Activators.

Translation activators are an important class of riboregulators that respond to nucleic acid signals by activating gene expression. Toehold switches and single-nucleotide-specific programmable riboregulators (SNIPRs) are two types of translation activators that can detect nearly any nucleic acid sequence using interactions initiated by single-stranded domains known as toeholds. Toehold switches operate with high dynamic range, orthogonality, and programmability, making them capable of detecting a variety of pathogens in paper-based cell-free diagnostic assays. SNIPRs are designed to enable the accurate detection of single-nucleotide mutations, making them valuable tools for mutation and drug-resistance assays. Here we describe the computational design process for generating toehold switches and SNIPRs active against different pathogens and mutations of interest. Such riboregulators can be deployed in paper-based diagnostic assays to enable rapid and low-cost disease detection.

Multi-arm RNA junctions encoding molecular logic unconstrained by input sequence for versatile cell-free diagnostics.

Applications of RNA-based molecular logic have been hampered by sequence constraints imposed on the input and output of the circuits. Here we show that the sequence constraints can be substantially reduced by appropriately encoded multi-arm junctions of single-stranded RNA structures. To conditionally activate RNA translation, we integrated multi-arm junctions, self-assembled upstream of a regulated gene and designed to unfold sequentially in response to different RNA inputs, with motifs of loop-initiated RNA activators that function independently of the sequence of the input RNAs and that reduce interference with the output gene. We used the integrated RNA system and sequence-independent input RNAs to execute two-input and three-input OR and AND logic in Escherichia coli, and designed paper-based cell-free colourimetric assays that accurately identified two human immunodeficiency virus (HIV) subtypes (by executing OR logic) in amplified synthetic HIV RNA as well as severe acute respiratory syndrome coronavirus-2 (via two-input AND logic) in amplified RNA from saliva samples. The sequence-independent molecular logic enabled by the integration of multi-arm junction RNAs with motifs for loop-initiated RNA activators may be broadly applicable in biotechnology.

A robust electrochemiluminescence immunoassay for carcinoembryonic antigen detection based on a microtiter plate as a bridge and Au@Pd nanorods as a peroxidase mimic.

The common drawbacks of most traditional electrochemiluminescence (ECL) immunoassays are the strict storage conditions for the ECL electrode and the steric hindrance caused by bovine serum albumin and antigen. The strict storage conditions require that the modified electrode must be stored at 4 °C before measurement, which may cause the degradation of protein molecules and low reproducibility as the time goes by. The steric hindrance can hinder electron transfer between the electrode and the electrochemical active substance unable to transmit proteins on the electrode surface. The current study takes a 96-well microtiter plate (MTP) as a bridge for analyte pre-treatment and Au@Pd nanorods as a peroxidase mimic to assemble a simple and robust ECL immunoassay. Advantages of such assay include not only high sensitivity but also robust detection circumstance. We demonstrated the method by detecting carcinoembryonic antigen from human serum and obtained a good detection limit of 3 fg mL(-1).