HGT is widespread in insects and contributes to male courtship in lepidopterans.

Horizontal gene transfer (HGT) is an important evolutionary force shaping prokaryotic and eukaryotic genomes. HGT-acquired genes have been sporadically reported in insects, a lineage containing >50% of animals. We systematically examined HGT in 218 high-quality genomes of diverse insects and found that they acquired 1,410 genes exhibiting diverse functions, including many not previously reported, via 741 distinct transfers from non-metazoan donors. Lepidopterans had the highest average number of HGT-acquired genes. HGT-acquired genes containing introns exhibited substantially higher expression levels than genes lacking introns, suggesting that intron gains were likely involved in HGT adaptation. Lastly, we used the CRISPR-Cas9 system to edit the prevalent unreported gene LOC105383139, which was transferred into the last common ancestor of moths and butterflies. In diamondback moths, males lacking LOC105383139 courted females significantly less. We conclude that HGT has been a major contributor to insect adaptation.

Eukaryotic Acquisition of a Bacterial Operon.

Operons are a hallmark of bacterial genomes, where they allow concerted expression of functionally related genes as single polycistronic transcripts. They are rare in eukaryotes, where each gene usually drives expression of its own independent messenger RNAs. Here, we report the horizontal operon transfer of a siderophore biosynthesis pathway from relatives of Escherichia coli into a group of budding yeast taxa. We further show that the co-linearly arranged secondary metabolism genes are expressed, exhibit eukaryotic transcriptional features, and enable the sequestration and uptake of iron. After transfer, several genetic changes occurred during subsequent evolution, including the gain of new transcription start sites that were sometimes within protein-coding sequences, acquisition of polyadenylation sites, structural rearrangements, and integration of eukaryotic genes into the cluster. We conclude that the genes were likely acquired as a unit, modified for eukaryotic gene expression, and maintained by selection to adapt to the highly competitive, iron-limited environment.

Tempo and Mode of Genome Evolution in the Budding Yeast Subphylum.

Budding yeasts (subphylum Saccharomycotina) are found in every biome and are as genetically diverse as plants or animals. To understand budding yeast evolution, we analyzed the genomes of 332 yeast species, including 220 newly sequenced ones, which represent nearly one-third of all known budding yeast diversity. Here, we establish a robust genus-level phylogeny comprising 12 major clades, infer the timescale of diversification from the Devonian period to the present, quantify horizontal gene transfer (HGT), and reconstruct the evolution of 45 metabolic traits and the metabolic toolkit of the budding yeast common ancestor (BYCA). We infer that BYCA was metabolically complex and chronicle the tempo and mode of genomic and phenotypic evolution across the subphylum, which is characterized by very low HGT levels and widespread losses of traits and the genes that control them. More generally, our results argue that reductive evolution is a major mode of evolutionary diversification.

Incongruence in the phylogenomics era.

Genome-scale data and the development of novel statistical phylogenetic approaches have greatly aided the reconstruction of a broad sketch of the tree of life and resolved many of its branches. However, incongruence - the inference of conflicting evolutionary histories - remains pervasive in phylogenomic data, hampering our ability to reconstruct and interpret the tree of life. Biological factors, such as incomplete lineage sorting, horizontal gene transfer, hybridization, introgression, recombination and convergent molecular evolution, can lead to gene phylogenies that differ from the species tree. In addition, analytical factors, including stochastic, systematic and treatment errors, can drive incongruence. Here, we review these factors, discuss methodological advances to identify and handle incongruence, and highlight avenues for future research.

Fragment-based drug discovery supports drugging 'undruggable' protein-protein interactions.

Protein-protein interactions (PPIs) have important roles in various cellular processes, but are commonly described as 'undruggable' therapeutic targets due to their large, flat, featureless interfaces. Fragment-based drug discovery (FBDD) has achieved great success in modulating PPIs, with more than ten compounds in clinical trials. Here, we highlight the progress of FBDD in modulating PPIs for therapeutic development. Targeting hot spots that have essential roles in both fragment binding and PPIs provides a shortcut for the development of PPI modulators via FBDD. We highlight successful cases of cracking the 'undruggable' problems of PPIs using fragment-based approaches. We also introduce new technologies and future trends. Thus, we hope that this review will provide useful guidance for drug discovery targeting PPIs.

Identification of miRNA858 long-loop precursors in seed plants.

MicroRNAs (miRNAs) are a class of nonprotein-coding short transcripts that provide a layer of post-transcriptional regulation essential to many plant biological processes. MiR858, which targets the transcripts of MYB transcription factors, can affect a range of secondary metabolic processes. Although miR858 and its 187-nt precursor have been well studied in Arabidopsis (Arabidopsis thaliana), a systematic investigation of miR858 precursors and their functions across plant species is lacking due to a problem in identifying the transcripts that generate this subclass. By re-evaluating the transcript of miR858 and relaxing the length cut-off for identifying hairpins, we found in kiwifruit (Actinidia chinensis) that miR858 has long-loop hairpins (1,100 to 2,100 nt), whose intervening sequences between miRNA generating complementary sites were longer than all previously reported miRNA hairpins. Importantly, these precursors of miR858 containing long-loop hairpins (termed MIR858L) are widespread in seed plants including Arabidopsis, varying between 350 and 5,500 nt. Moreover, we showed that MIR858L has a greater impact on proanthocyanidin and flavonol levels in both Arabidopsis and kiwifruit. We suggest that an active MIR858L-MYB regulatory module appeared in the transition of early land plants to large upright flowering plants, making a key contribution to plant secondary metabolism.

Machine learning enables identification of an alternative yeast galactose utilization pathway.

How genomic differences contribute to phenotypic differences is a major question in biology. The recently characterized genomes, isolation environments, and qualitative patterns of growth on 122 sources and conditions of 1,154 strains from 1,049 fungal species (nearly all known) in the yeast subphylum Saccharomycotina provide a powerful, yet complex, dataset for addressing this question. We used a random forest algorithm trained on these genomic, metabolic, and environmental data to predict growth on several carbon sources with high accuracy. Known structural genes involved in assimilation of these sources and presence/absence patterns of growth in other sources were important features contributing to prediction accuracy. By further examining growth on galactose, we found that it can be predicted with high accuracy from either genomic (92.2%) or growth data (82.6%) but not from isolation environment data (65.6%). Prediction accuracy was even higher (93.3%) when we combined genomic and growth data. After the GALactose utilization genes, the most important feature for predicting growth on galactose was growth on galactitol, raising the hypothesis that several species in two orders, Serinales and Pichiales (containing the emerging pathogen Candida auris and the genus Ogataea, respectively), have an alternative galactose utilization pathway because they lack the GAL genes. Growth and biochemical assays confirmed that several of these species utilize galactose through an alternative oxidoreductive D-galactose pathway, rather than the canonical GAL pathway. Machine learning approaches are powerful for investigating the evolution of the yeast genotype-phenotype map, and their application will uncover novel biology, even in well-studied traits.

Saccharomycotina yeasts defy long-standing macroecological patterns.

The Saccharomycotina yeasts ("yeasts" hereafter) are a fungal clade of scientific, economic, and medical significance. Yeasts are highly ecologically diverse, found across a broad range of environments in every biome and continent on earth; however, little is known about what rules govern the macroecology of yeast species and their range limits in the wild. Here, we trained machine learning models on 12,816 terrestrial occurrence records and 96 environmental variables to infer global distribution maps at ~1 km2 resolution for 186 yeast species (~15% of described species from 75% of orders) and to test environmental drivers of yeast biogeography and macroecology. We found that predicted yeast diversity hotspots occur in mixed montane forests in temperate climates. Diversity in vegetation type and topography were some of the greatest predictors of yeast species richness, suggesting that microhabitats and environmental clines are key to yeast diversity. We further found that range limits in yeasts are significantly influenced by carbon niche breadth and range overlap with other yeast species, with carbon specialists and species in high-diversity environments exhibiting reduced geographic ranges. Finally, yeasts contravene many long-standing macroecological principles, including the latitudinal diversity gradient, temperature-dependent species richness, and a positive relationship between latitude and range size (Rapoport's rule). These results unveil how the environment governs the global diversity and distribution of species in the yeast subphylum. These high-resolution models of yeast species distributions will facilitate the prediction of economically relevant and emerging pathogenic species under current and future climate scenarios.

A disordered rock salt anode for fast-charging lithium-ion batteries.

Rechargeable lithium-ion batteries with high energy density that can be safely charged and discharged at high rates are desirable for electrified transportation and other applications1-3. However, the sub-optimal intercalation potentials of current anodes result in a trade-off between energy density, power and safety. Here we report that disordered rock salt4,5 Li3+xV2O5 can be used as a fast-charging anode that can reversibly cycle two lithium ions at an average voltage of about 0.6 volts versus a Li/Li+ reference electrode. The increased potential compared to graphite6,7 reduces the likelihood of lithium metal plating if proper charging controls are used, alleviating a major safety concern (short-circuiting related to Li dendrite growth). In addition, a lithium-ion battery with a disordered rock salt Li3V2O5 anode yields a cell voltage much higher than does a battery using a commercial fast-charging lithium titanate anode or other intercalation anode candidates (Li3VO4 and LiV0.5Ti0.5S2)8,9. Further, disordered rock salt Li3V2O5 can perform over 1,000 charge-discharge cycles with negligible capacity decay and exhibits exceptional rate capability, delivering over 40 per cent of its capacity in 20 seconds. We attribute the low voltage and high rate capability of disordered rock salt Li3V2O5 to a redistributive lithium intercalation mechanism with low energy barriers revealed via ab initio calculations. This low-potential, high-rate intercalation reaction can be used to identify other metal oxide anodes for fast-charging, long-life lithium-ion batteries.

Shuffling the yeast genome using CRISPR/Cas9-generated DSBs that target the transposable Ty1 elements.

Although homologous recombination between transposable elements can drive genomic evolution in yeast by facilitating chromosomal rearrangements, the details of the underlying mechanisms are not fully clarified. In the genome of the yeast Saccharomyces cerevisiae, the most common class of transposon is the retrotransposon Ty1. Here, we explored how Cas9-induced double-strand breaks (DSBs) directed to Ty1 elements produce genomic alterations in this yeast species. Following Cas9 induction, we observed a significant elevation of chromosome rearrangements such as deletions, duplications and translocations. In addition, we found elevated rates of mitotic recombination, resulting in loss of heterozygosity. Using Southern analysis coupled with short- and long-read DNA sequencing, we revealed important features of recombination induced in retrotransposons. Almost all of the chromosomal rearrangements reflect the repair of DSBs at Ty1 elements by non-allelic homologous recombination; clustered Ty elements were hotspots for chromosome rearrangements. In contrast, a large proportion (about three-fourths) of the allelic mitotic recombination events have breakpoints in unique sequences. Our analysis suggests that some of the latter events reflect extensive processing of the broken ends produced in the Ty element that extend into unique sequences resulting in break-induced replication. Finally, we found that haploid and diploid strain have different preferences for the pathways used to repair double-stranded DNA breaks. Our findings demonstrate the importance of DNA lesions in retrotransposons in driving genome evolution.

Functional and Evolutionary Integration of a Fungal Gene With a Bacterial Operon.

Siderophores are crucial for iron-scavenging in microorganisms. While many yeasts can uptake siderophores produced by other organisms, they are typically unable to synthesize siderophores themselves. In contrast, Wickerhamiella/Starmerella (W/S) clade yeasts gained the capacity to make the siderophore enterobactin following the remarkable horizontal acquisition of a bacterial operon enabling enterobactin synthesis. Yet, how these yeasts absorb the iron bound by enterobactin remains unresolved. Here, we demonstrate that Enb1 is the key enterobactin importer in the W/S-clade species Starmerella bombicola. Through phylogenomic analyses, we show that ENB1 is present in all W/S clade yeast species that retained the enterobactin biosynthetic genes. Conversely, it is absent in species that lost the ent genes, except for Starmerella stellata, making this species the only cheater in the W/S clade that can utilize enterobactin without producing it. Through phylogenetic analyses, we infer that ENB1 is a fungal gene that likely existed in the W/S clade prior to the acquisition of the ent genes and subsequently experienced multiple gene losses and duplications. Through phylogenetic topology tests, we show that ENB1 likely underwent horizontal gene transfer from an ancient W/S clade yeast to the order Saccharomycetales, which includes the model yeast Saccharomyces cerevisiae, followed by extensive secondary losses. Taken together, these results suggest that the fungal ENB1 and bacterial ent genes were cooperatively integrated into a functional unit within the W/S clade that enabled adaptation to iron-limited environments. This integrated fungal-bacterial circuit and its dynamic evolution determine the extant distribution of yeast enterobactin producers and cheaters.

ACC SYNTHASE4 inhibits gibberellin biosynthesis and FLOWERING LOCUS T expression during citrus flowering.

Flowering is an essential process in fruit trees. Flower number and timing have a substantial impact on the yield and maturity of fruit. Ethylene and gibberellin (GA) play vital roles in flowering, but the mechanism of coordinated regulation of flowering in woody plants by GA and ethylene is still unclear. In this study, a lemon (Citrus limon L. Burm) 1-aminocyclopropane-1-carboxylic acid synthase gene (CiACS4) was overexpressed in Nicotiana tabacum and resulted in late flowering and increased flower number. Further transformation of citrus revealed that ethylene and starch content increased, and soluble sugar content decreased in 35S:CiACS4 lemon. Inhibition of CiACS4 in lemon resulted in effects opposite to that of 35S:CiACS4 in transgenic plants. Overexpression of the CiACS4-interacting protein ETHYLENE RESPONSE FACTOR3 (CiERF3) in N. tabacum resulted in delayed flowering and more flowers. Further experiments revealed that the CiACS4-CiERF3 complex can bind the promoters of FLOWERING LOCUS T (CiFT) and GOLDEN2-LIKE (CiFE) and suppress their expression. Moreover, overexpression of CiFE in N. tabacum led to early flowering and decreased flowers, and ethylene, starch, and soluble sugar contents were opposite to those in 35S:CiACS4 transgenic plants. Interestingly, CiFE also bound the promoter of CiFT. Additionally, GA3 and 1-aminocyclopropanecarboxylic acid (ACC) treatments delayed flowering in adult citrus, and treatment with GA and ethylene inhibitors increased flower number. ACC treatment also inhibited the expression of CiFT and CiFE. This study provides a theoretical basis for the application of ethylene to regulate flower number and mitigate the impacts of extreme weather on citrus yield due to delayed flowering.

The influence of the number of tree searches on maximum likelihood inference in phylogenomics.

Maximum likelihood (ML) phylogenetic inference is widely used in phylogenomics. As heuristic searches most likely find suboptimal trees, it is recommended to conduct multiple (e.g., ten) tree searches in phylogenetic analyses. However, beyond its positive role, how and to what extent multiple tree searches aid ML phylogenetic inference remains poorly explored. Here, we found that a random starting tree was not as effective as the BioNJ and parsimony starting trees in inferring ML gene tree and that RAxML-NG and PhyML were less sensitive to different starting trees than IQ-TREE. We then examined the effect of the number of tree searches on ML tree inference with IQ-TREE and RAxML-NG, by running 100 tree searches on 19,414 gene alignments from 15 animal, plant, and fungal phylogenomic datasets. We found that the number of tree searches substantially impacted the recovery of the best-of-100 ML gene tree topology among 100 searches for a given ML program. In addition, all of the concatenation-based trees were topologically identical if the number of tree searches was ≥ 10. Quartet-based ASTRAL trees inferred from 1 to 80 tree searches differed topologically from those inferred from 100 tree searches for 6 /15 phylogenomic datasets. Lastly, our simulations showed that gene alignments with lower difficulty scores had a higher chance of finding the best-of-100 gene tree topology and were more likely to yield the correct trees.

OrthoSNAP: A tree splitting and pruning algorithm for retrieving single-copy orthologs from gene family trees.

Molecular evolution studies, such as phylogenomic studies and genome-wide surveys of selection, often rely on gene families of single-copy orthologs (SC-OGs). Large gene families with multiple homologs in 1 or more species-a phenomenon observed among several important families of genes such as transporters and transcription factors-are often ignored because identifying and retrieving SC-OGs nested within them is challenging. To address this issue and increase the number of markers used in molecular evolution studies, we developed OrthoSNAP, a software that uses a phylogenetic framework to simultaneously split gene families into SC-OGs and prune species-specific inparalogs. We term SC-OGs identified by OrthoSNAP as SNAP-OGs because they are identified using a splitting and pruning procedure analogous to snapping branches on a tree. From 415,129 orthologous groups of genes inferred across 7 eukaryotic phylogenomic datasets, we identified 9,821 SC-OGs; using OrthoSNAP on the remaining 405,308 orthologous groups of genes, we identified an additional 10,704 SNAP-OGs. Comparison of SNAP-OGs and SC-OGs revealed that their phylogenetic information content was similar, even in complex datasets that contain a whole-genome duplication, complex patterns of duplication and loss, transcriptome data where each gene typically has multiple transcripts, and contentious branches in the tree of life. OrthoSNAP is useful for increasing the number of markers used in molecular evolution data matrices, a critical step for robustly inferring and exploring the tree of life.

Zbtb16 increases susceptibility of atrial fibrillation in type 2 diabetic mice via Txnip-Trx2 signaling.

Atrial fibrillation (AF) is the most prevalent sustained cardiac arrhythmia, and recent epidemiological studies suggested type 2 diabetes mellitus (T2DM) is an independent risk factor for the development of AF. Zinc finger and BTB (broad-complex, tram-track and bric-a-brac) domain containing 16 (Zbtb16) serve as transcriptional factors to regulate many biological processes. However, the potential effects of Zbtb16 in AF under T2DM condition remain unclear. Here, we reported that db/db mice displayed higher AF vulnerability and Zbtb16 was identified as the most significantly enriched gene by RNA sequencing (RNA-seq) analysis in atrium. In addition, thioredoxin interacting protein (Txnip) was distinguished as the key downstream gene of Zbtb16 by Cleavage Under Targets and Tagmentation (CUT&Tag) assay. Mechanistically, increased Txnip combined with thioredoxin 2 (Trx2) in mitochondrion induced excess reactive oxygen species (ROS) release, calcium/calmodulin-dependent protein kinase II (CaMKII) overactivation, and spontaneous Ca2+ waves (SCWs) occurrence, which could be inhibited through atrial-specific knockdown (KD) of Zbtb16 or Txnip by adeno-associated virus 9 (AAV9) or Mito-TEMPO treatment. High glucose (HG)-treated HL-1 cells were used to mimic the setting of diabetic in vitro. Zbtb16-Txnip-Trx2 signaling-induced excess ROS release and CaMKII activation were also verified in HL-1 cells under HG condition. Furthermore, atrial-specific Zbtb16 or Txnip-KD reduced incidence and duration of AF in db/db mice. Altogether, we demonstrated that interrupting Zbtb16-Txnip-Trx2 signaling in atrium could decrease AF susceptibility via reducing ROS release and CaMKII activation in the setting of T2DM.

ACFIS 2.0: an improved web-server for fragment-based drug discovery via a dynamic screening strategy.

Drug discovery, which plays a vital role in maintaining human health, is a persistent challenge. Fragment-based drug discovery (FBDD) is one of the strategies for the discovery of novel candidate compounds. Computational tools in FBDD could help to identify potential drug leads in a cost-efficient and time-saving manner. The Auto Core Fragment in silico Screening (ACFIS) server is a well-established and effective online tool for FBDD. However, the accurate prediction of protein-fragment binding mode and affinity is still a major challenge for FBDD due to weak binding affinity. Here, we present an updated version (ACFIS 2.0), that incorporates a dynamic fragment growing strategy to consider protein flexibility. The major improvements of ACFIS 2.0 include (i) increased accuracy of hit compound identification (from 75.4% to 88.5% using the same test set), (ii) improved rationality of the protein-fragment binding mode, (iii) increased structural diversity due to expanded fragment libraries and (iv) inclusion of more comprehensive functionality for predicting molecular properties. Three successful cases of drug lead discovery using ACFIS 2.0 are described, including drugs leads to treat Parkinson's disease, cancer, and major depressive disorder. These cases demonstrate the utility of this web-based server. ACFIS 2.0 is freely available at http://chemyang.ccnu.edu.cn/ccb/server/ACFIS2/.

Holistic immunomodulation for small cell lung cancer.

Small cell lung cancer (SCLC) is characterized by a high mortality rate, rapid growth, and early metastasis, which lead to a poor prognosis. Moreover, limited clinical treatment options further lower the survival rate of patients. Therefore, novel technology and agents are urgently required to enhance clinical efficacy. In this review, from a holistic perspective, we summarized the therapeutic targets, agents and strategies with the most potential for treating SCLC, including chimeric antigen receptor (CAR) T therapy, immunomodulating antibodies, traditional Chinese medicines (TCMs), and the microbiota, which have been found recently to improve the clinical outcomes and prognosis of SCLC. Multiomics technologies can be integrated to develop effective diagnostic methods and identify new targets for new drug discovery in SCLC. We discussed in depth the feasibility, potential, and challenges of these new strategies, as well as their combinational treatments, which may provide promising alternatives for enhancing the clinical efficacy of SCLC in the future.

Clinical efficacy of tumor organoid-guided cancer therapy for locally advanced unresectable or metastatic breast cancer.

Patient-derived organoids (PDOs) may facilitate treatment selection. This retrospective cohort study evaluated the feasibility and clinical benefit of using PDOs to guide personalized treatment in metastatic breast cancer (MBC). Patients diagnosed with MBC were recruited between January 2019 and August 2022. PDOs were established and the efficacy of customized drug panels was determined by measuring cell mortality after drug exposure. Patients receiving organoid-guided treatment (OGT) were matched 1:2 by nearest neighbor propensity scores with patients receiving treatment of physician's choice (TPC). The primary outcome was progression-free survival. Secondary outcomes included objective response rate and disease control rate. Targeted gene sequencing and pathway enrichment analysis were performed. Forty-six PDOs (46 of 51, 90.2%) were generated from 45 MBC patients. PDO drug screening showed an accuracy of 78.4% (95% CI 64.9%-91.9%) in predicting clinical responses. Thirty-six OGT patients were matched to 69 TPC patients. OGT was associated with prolonged median progression-free survival (11.0 months vs. 5.0 months; hazard ratio 0.53 [95% CI 0.33-0.85]; p = .01) and improved disease control (88.9% vs. 63.8%; odd ratio 4.26 [1.44-18.62]) compared with TPC. The objective response rate of both groups was similar. Pathway enrichment analysis in hormone receptor-positive, human epidermal growth factor receptor 2-negative patients demonstrated differentially modulated pathways implicated in DNA repair and transcriptional regulation in those with reduced response to capecitabine/gemcitabine, and pathways associated with cell cycle regulation in those with reduced response to palbociclib. Our study shows that PDO-based functional precision medicine is a feasible and effective strategy for MBC treatment optimization and customization.

A non-invasive artificial intelligence model for identifying axillary pathological complete response to neoadjuvant chemotherapy in breast cancer: a secondary analysis to multicenter clinical trial.

BACKGROUND: This study aims to develop a stacking model for accurately predicting axillary lymph node (ALN) response to neoadjuvant chemotherapy (NAC) using longitudinal MRI in breast cancer. METHODS: We included patients with node-positive breast cancer who received NAC following surgery from January 2012 to June 2022. We collected MRIs before and after NAC, and extracted radiomics features from the tumour, peritumour, and ALN regions. The Mann-Whitney U test, least absolute shrinkage and selection operator, and Boruta algorithm were used to select features. We utilised machine learning techniques to develop three single-modality models and a stacking model for predicting ALN response to NAC. RESULTS: This study consisted of a training cohort (n = 277), three external validation cohorts (n = 313, 164, and 318), and a prospective cohort (n = 81). Among the 1153 patients, 60.62% achieved ypN0. The stacking model achieved excellent AUCs of 0.926, 0.874, and 0.862 in the training, external validation, and prospective cohort, respectively. It also showed lower false-negative rates (FNRs) compared to radiologists, with rates of 14.40%, 20.85%, and 18.18% (radiologists: 40.80%, 50.49%, and 63.64%) in three cohorts. Additionally, there was a significant difference in disease-free survival between high-risk and low-risk groups (p < 0.05). CONCLUSIONS: The stacking model can accurately predict ALN status after NAC in breast cancer, showing a lower false-negative rate than radiologists. TRIAL REGISTRATION NUMBER: The clinical trial numbers were NCT03154749 and NCT04858529.

Noninvasive Artificial Intelligence System for Early Predicting Residual Cancer Burden during Neoadjuvant Chemotherapy in Breast Cancer.

OBJECTIVE: To develop an artificial intelligence (AI) system for the early prediction of residual cancer burden (RCB) scores during neoadjuvant chemotherapy (NAC) in breast cancer. SUMMARY BACKGROUND DATA: RCB III indicates drug resistance in breast cancer, and early detection methods are lacking. METHODS: This study enrolled 1048 patients with breast cancer from four institutions, who were all receiving NAC. Magnetic resonance images were collected at the pre- and mid-NAC stages, and radiomics and deep learning features were extracted. A multitask AI system was developed to classify patients into three groups (RCB 0-I, II, and III ) in the primary cohort (PC, n=335). Feature selection was conducted using the Mann-Whitney U- test, Spearman analysis, least absolute shrinkage and selection operator regression, and the Boruta algorithm. Single-modality models were developed followed by model integration. The AI system was validated in three external validation cohorts. (EVCs, n=713). RESULTS: Among the patients, 442 (42.18%) were RCB 0-I, 462 (44.08%) were RCB II and 144 (13.74%) were RCB III. Model-I achieved an area under the curve (AUC) of 0.975 in the PC and 0.923 in the EVCs for differentiating RCB III from RCB 0-II. Model-II distinguished RCB 0-I from RCB II-III, with an AUC of 0.976 in the PC and 0.910 in the EVCs. Subgroup analysis confirmed that the AI system was consistent across different clinical T stages and molecular subtypes. CONCLUSIONS: The multitask AI system offers a noninvasive tool for the early prediction of RCB scores in breast cancer, supporting clinical decision-making during NAC.