Nature biotechnology最新文献

Rewriting RNA information to alter function requires controllable tools to edit RNA sequences within a user-defined region. Here we report a single-strand deaminase-assisted platform for adjustable RNA information manipulation (AIM). AIM is composed of a loop-forming guide RNA bound to an RNA-targeting Cas protein and an evolved TadA. AIM induces a loop, flanked by paired regions, in the target RNA; the loop size can be adjusted to allow conversions of single and multiple bases. We evolve TadA to achieve A-to-I, C-to-U or simultaneous A+C editing in coding and noncoding regions. We apply AIM to suppress the ochre nonsense codon (UAA) in disease-relevant cell and animal models, in which the two As must be simultaneously edited to rewrite the coding information. Moreover, we use AIM to manipulate adjacent phosphorylation sites important for protein function. Collectively, AIM is a versatile platform for manipulating RNA information within user-defined regions, opening additional avenues for functional RNA modulation.

Assessing the accuracy of long-read assemblies, especially from complex environmental metagenomes that include underrepresented organisms, is challenging. Here we benchmark four state-of-the-art long-read assembly software programs, HiCanu, hifiasm-meta, metaFlye and metaMDBG, on 21 PacBio HiFi metagenomes spanning mock communities, gut microbiomes and ocean samples. By quantifying read clipping events, in which long reads are systematically split during mapping to maximize the agreement with assembled contigs, we identify where assemblies diverge from their source reads. Our analyses reveal that long-read metagenome assemblies can include >40 errors per 100 million base pairs of assembled contigs, including multi-domain chimeras, prematurely circularized sequences, haplotyping errors, excessive repeats and phantom sequences. We provide an open-source tool and a reproducible workflow for rigorous evaluation of assembly errors, charting a path toward more reliable genome recovery from long-read metagenomes.

Detection of the off-target effects of base editors is important for identifying their safety risks but current methods for understanding their global activities have limitations in terms of sensitivity or bias by computationally selecting a subset of sites for experimental analysis. We present CHANGE-seq-BE, a method to assess the guide RNA-dependent off-target profile of both adenine and cytosine base editors that is simultaneously sensitive and unbiased. CHANGE-seq-BE relies on selective sequencing of base-editor-modified genomic DNA in vitro and provides comprehensive identification of genome-wide off-target mutations. We found that 98.8% of validated off-target sites were unique to ABE8e adenine base editors compared to Cas9 nuclease, suggesting substantially higher off-target activity of the former. We further applied CHANGE-seq-BE to support genotoxicity studies in an emergency investigational new drug application for customized adenine base editor treatment for a person with CD40L-deficient X-linked hyper IgM syndrome. Our results emphasize the importance of using a base-editor-specific method for identifying off-target activity.

The synergistic combination of two antimicrobial drugs is a promising therapeutic modality for many infectious diseases. However, systemic fungal infections still have a high mortality rate because of distinct in vivo distributions of the two drugs. Here we address this challenge by designing an antifungal polymer that forms micelles suitable for delivering a second antifungal agent to achieve temporal and spatial consistency of delivery. We show that the polymer, which mimics host defense peptides, exerts a synergistic effect with the antifungal amphotericin B (AmB). The AmB-encapsulated micelles (AmBmicelles) greatly reduce the toxicity of AmB through slow release and expand its therapeutic window in vivo. AmBmicelles can selectively target fungal pathogens through charge interactions with the fungal membrane. In mouse models of systemic candidiasis and cryptococcal meningitis, AmBmicelles increase the survival rate by 67-100% compared to the state-of-the-art drug AmBisome or AmBisome and 5-flucytosine combination, suggesting that the strategy may be effective in combating drug-resistant fungal infections including meningitis.

Small molecules can bind RNAs to regulate their fate and functions, providing promising opportunities for treating human diseases. However, current tools for predicting small molecule-RNA interactions (SRIs) require prior knowledge of RNA tertiary structures. Here we present SMRTnet, a deep learning method that uses multimodal data fusion to integrate two large language models with convolutional and graph attention networks to predict SRIs on the basis of RNA secondary structure. SMRTnet achieves high performance across multiple experimental benchmarks, substantially outperforming existing tools. SMRTnet predictions for ten disease-associated RNA targets identified 40 hits of RNA-targeting small molecules with nanomolar-to-micromolar dissociation constants. Focusing on the MYC internal ribosome entry site, SMRTnet-predicted small molecules showed binding scores correlated closely with observed validation rates. One predicted small molecule downregulated MYC expression, inhibited proliferation and promoted apoptosis in three cancer cell lines. Thus, by eliminating the need for RNA tertiary structures, SMRTnet expands the scope of feasible RNA targets and accelerates the discovery of RNA-targeting therapeutics.