Accurate top protein variant discovery via low-N pick-and-validate machine learning.

A strategy to obtain the greatest number of best-performing variants with least amount of experimental effort over the vast combinatorial mutational landscape would have enormous utility in boosting resource producibility for protein engineering. Toward this goal, we present a simple and effective machine learning-based strategy that outperforms other state-of-the-art methods. Our strategy integrates zero-shot prediction and multi-round sampling to direct active learning via experimenting with only a few predicted top variants. We find that four rounds of low-N pick-and-validate sampling of 12 variants for machine learning yielded the best accuracy of up to 92.6% in selecting the true top 1% variants in combinatorial mutant libraries, whereas two rounds of 24 variants can also be used. We demonstrate our strategy in successfully discovering high-performance protein variants from diverse families including the CRISPR-based genome editors, supporting its generalizable application for solving protein engineering tasks. A record of this paper's transparent peer review process is included in the supplemental information.

Discovery of regulatory motifs in 5' untranslated regions using interpretable multi-task learning models.

The sequence in the 5' untranslated regions (UTRs) is known to affect mRNA translation rates. However, the underlying regulatory grammar remains elusive. Here, we propose MTtrans, a multi-task translation rate predictor capable of learning common sequence patterns from datasets across various experimental techniques. The core premise is that common motifs are more likely to be genuinely involved in translation control. MTtrans outperforms existing methods in both accuracy and the ability to capture transferable motifs across species, highlighting its strength in identifying evolutionarily conserved sequence motifs. Our independent fluorescence-activated cell sorting coupled with deep sequencing (FACS-seq) experiment validates the impact of most motifs identified by MTtrans. Additionally, we introduce "GRU-rewiring," a technique to interpret the hidden states of the recurrent units. Gated recurrent unit (GRU)-rewiring allows us to identify regulatory element-enriched positions and examine the local effects of 5' UTR mutations. MTtrans is a powerful tool for deciphering the translation regulatory motifs.

Parallel engineering and activity profiling of a base editor system.

Selecting the most suitable existing base editors and engineering new variants for installing specific base conversions with maximal efficiency and minimal undesired edits are pivotal for precise genome editing applications. Here, we present a platform for creating and analyzing a library of engineered base editor variants to enable head-to-head evaluation of their editing performance at scale. Our comprehensive comparison provides quantitative measures on each variant's editing efficiency, purity, motif preference, and bias in generating single and multiple base conversions, while uncovering undesired higher indel generation rate and noncanonical base conversion for some of the existing base editors. In addition to engineering the base editor protein, we further applied this platform to investigate a hitherto underexplored engineering route and created guide RNA scaffold variants that augment the editor's base-editing activity. With the unknown performance and compatibility of the growing number of engineered parts including deaminase, CRISPR-Cas enzyme, and guide RNA scaffold variants for assembling the expanding collection of base editor systems, our platform addresses the unmet need for an unbiased, scalable method to benchmark their editing outcomes and accelerate the engineering of next-generation precise genome editors.

Machine learning-coupled combinatorial mutagenesis enables resource-efficient engineering of CRISPR-Cas9 genome editor activities.

The genome-editing Cas9 protein uses multiple amino-acid residues to bind the target DNA. Considering only the residues in proximity to the target DNA as potential sites to optimise Cas9's activity, the number of combinatorial variants to screen through is too massive for a wet-lab experiment. Here we generate and cross-validate ten in silico and experimental datasets of multi-domain combinatorial mutagenesis libraries for Cas9 engineering, and demonstrate that a machine learning-coupled engineering approach reduces the experimental screening burden by as high as 95% while enriching top-performing variants by ∼7.5-fold in comparison to the null model. Using this approach and followed by structure-guided engineering, we identify the N888R/A889Q variant conferring increased editing activity on the protospacer adjacent motif-relaxed KKH variant of Cas9 nuclease from Staphylococcus aureus (KKH-SaCas9) and its derived base editor in human cells. Our work validates a readily applicable workflow to enable resource-efficient high-throughput engineering of genome editor's activity.

Optochemical Control of mTOR Signaling and mTOR-Dependent Autophagy.

As an important regulator of cell metabolism, proliferation, and survival, mTOR (mammalian target of rapamycin) signaling provides both a potential target for cancer treatment and a research tool for investigation of cell metabolism. One inhibitor for both mTORC1 and mTORC2 pathways, OSI-027, exhibited robust anticancer efficacy but induced side effects. Herein, we designed a photoactivatable OSI-027 prodrug, which allowed the release of OSI-027 after light irradiation to inhibit the mTOR signaling pathway, triggering autophagy and leading to cell death. This photoactivatable prodrug can provide novel strategies for mTOR-targeting cancer therapy and act as a new tool for investigating mTOR signaling and its related biological processes.

A Combinatorial CRISPR-Cas9 Screen Identifies Ifenprodil as an Adjunct to Sorafenib for Liver Cancer Treatment.

Systematic testing of existing drugs and their combinations is an attractive strategy to exploit approved drugs for repurposing and identifying the best actionable treatment options. To expedite the search among many possible drug combinations, we designed a combinatorial CRISPR-Cas9 screen to inhibit druggable targets. Coblockade of the N-methyl-d-aspartate receptor (NMDAR) with targets of first-line kinase inhibitors reduced hepatocellular carcinoma (HCC) cell growth. Clinically, HCC patients with low NMDAR1 expression showed better survival. The clinically approved NMDAR antagonist ifenprodil synergized with sorafenib to induce the unfolded protein response, trigger cell-cycle arrest, downregulate genes associated with WNT signaling and stemness, and reduce self-renewal ability of HCC cells. In multiple HCC patient-derived organoids and human tumor xenograft models, the drug combination, but neither single drug alone, markedly reduced tumor-initiating cancer cell frequency. Because ifenprodil has an established safety history for its use as a vasodilator in humans, our findings support the repurposing of this drug as an adjunct for HCC treatment to improve clinical outcome and reduce tumor recurrence. These results also validate an approach for readily discovering actionable combinations for cancer therapy. SIGNIFICANCE: Combinatorial CRISPR-Cas9 screening identifies actionable targets for HCC therapy, uncovering the potential of combining the clinically approved drugs ifenprodil and sorafenib as a new effective treatment regimen.