Could artificial intelligence revolutionize the development of nanovectors for gene therapy and mRNA vaccines?

Abstract

Gene therapy enables the introduction of nucleic acids like DNA and RNA into host cells, and is expected to revolutionize the treatment of a wide range of diseases. This growth has been further accelerated by the discovery of CRISPR/Cas technology, which allows accurate genomic editing in a broad range of cells and organisms in vitro and in vivo. Despite many advances in gene delivery and the development of various viral and non-viral gene delivery vectors, the lack of highly efficient non-viral systems with low cellular toxicity remains a challenge. The application of cutting-edge technologies such as artificial intelligence (AI) has great potential to find new paradigms to solve this issue. Herein, we review AI and its major subfields including machine learning (ML), neural networks (NNs), expert systems, deep learning (DL), computer vision and robotics. We discuss the potential of AI-based models and algorithms in the design of targeted gene delivery vehicles capable of crossing extracellular and intracellular barriers by viral mimicry strategies. We finally discuss the role of AI in improving the function of CRISPR/Cas systems, developing novel nanobots, and mRNA vaccine carriers.

Keywords: AI; CRISPR/Cas; gene delivery vehicles; gene therapy; mRNA vaccine carriers; nanobots.

Figure 2.. Prediction of the in vivo fate of nanoarchitectures using a neural network (NN) workflow.

Proteomic analysis serves as the input and the organ accumulation of nanoparticles and their size forms the outputs. Reprinted with permission from Ref. (71). Copyright 2019, American Chemical Society.

Figure 3.. Schematic illustration of two-layer framework for predicting the cellular uptake efficiency of CPPs.

(A) There are four stages: (I) importing dataset production; (II) extraction of various features from peptide sequences; (III) constructing four different ML-based classifiers; (IIII) comparing the performance of the first-layer prediction model with second-layer prediction model. Reprinted with permission from Ref. (167). Copyright 2018, American Chemical Society. (B) The ANN model for predicting the cell membrane insertion potential of CPPs. Adapted from Ref. (75) with permission. Copyright 2019, Elsevier.

Figure 4.. QNAR analysis of NPs.

(A) Schematic illustration of virtual gold nanoparticle preparation, experimental validation and predictive modelling. Reprinted with permission from Ref. (169). Copyright 2017, American Chemical Society. (B) Quantitative nanostructure-activity relationship (QNAR) model, in which descriptors are obtained from both calculation and experimental determination of manufactured nanoparticles. Reprinted with permission from Ref. (176). Copyright 2017, American Chemical Society.

Figure 5.. Schematic workflow of a ML-based algorithm for detection of intracellular organelles.

(A) The imaging of centrin (centrioles, red) and pericentrin (PCM, green) fluorescence signals, leads to convolutional networks allocating a score for each pixel that helps to determine the likelihood that it is a centrosome. (B) Left panels and right panels indicate the raw images of centrosomes and detected centrioles (white circles), respectively. (C, D and E) show the centriole detection, variation of average precision and a comparison between a noisy image and the original image in terms of precision and recall values in turn. (F) Workflow for segmenting out individual cells. (G) Human-annotated cells are compared to predicted cell segmentation. Adapted from Ref. (245) with permission. Copyright 2020, Cell Science.

Figure 6.. CDs for gene delivery.

A ML-based synthesis technique for construction of highly-fluorescent CDs. Reprinted with permission from Ref. (384). Copyright 2020, American Chemical Society.

Figure 7.. Schematic framework of a ML-based system for determining cytotoxicity, cellular uptake and ribonucleoprotein payload delivery of polymers based on an experimental data set.

(A) Combinatorial polymer synthesis and polyplex assembly. (B) The characterization of polymers was performed based on polyplex size distribution, probe protonation behaviour, charge density and RNP binding affinity. (C) High-throughput biological assays were applied to assess the gene-editing efficiency. (D) The evaluation of cell viability and cellular uptake of polyplexes. (E) The experimental data sets led to the extraction and generation of structure−function maps that correlated polymer properties with cellular toxicity, RNP uptake and RNP delivery using machine learning tools. Reprinted with permission from Ref. (389). Copyright 2020, American Chemical Society.

Figure 8.. Schematic diagram of sensory input-DRLs for efficient navigation of micro/nanobots in an unknown environment based on experimental training information.

(A) Colloidal robots, (B) Micro/nano motors. Reprinted with permission from Ref. (397, 398). Copyright 2020, John Wiley & Sons.

Figure 9.. AI for xenobots and mRNA vaccine development.

(A) Schematic representation of an AI-based algorithm that predicts the kinematic self-replication of synthetic multicellular assemblies (xenobots). Reprinted with permission from Ref. (401). Copyright 2021, Proceedings of the National Academy of Sciences (PNAS). (B) Schematic workflow of a ML-based algorithm for predicting the potential of lipid nanoparticles (LNP) for mRNA vaccine delivery. Reprinted with permission from Ref. (462). Copyright 2022, Elsevier.

Scheme 1.

Schematic showing the use of AI in the development of novel nanovectors that can successfully cross extracellular and intracellular barriers.