Gasdermin D permeabilization of mitochondrial inner and outer membranes accelerates and enhances pyroptosis.

Gasdermin D (GSDMD)-activated inflammatory cell death (pyroptosis) causes mitochondrial damage, but its underlying mechanism and functional consequences are largely unknown. Here, we show that the N-terminal pore-forming GSDMD fragment (GSDMD-NT) rapidly damaged both inner and outer mitochondrial membranes (OMMs) leading to reduced mitochondrial numbers, mitophagy, ROS, loss of transmembrane potential, attenuated oxidative phosphorylation (OXPHOS), and release of mitochondrial proteins and DNA from the matrix and intermembrane space. Mitochondrial damage occurred as soon as GSDMD was cleaved prior to plasma membrane damage. Mitochondrial damage was independent of the B-cell lymphoma 2 family and depended on GSDMD-NT binding to cardiolipin. Canonical and noncanonical inflammasome activation of mitochondrial damage, pyroptosis, and inflammatory cytokine release were suppressed by genetic ablation of cardiolipin synthase (Crls1) or the scramblase (Plscr3) that transfers cardiolipin to the OMM. Phospholipid scramblase-3 (PLSCR3) deficiency in a tumor compromised pyroptosis-triggered anti-tumor immunity. Thus, mitochondrial damage plays a critical role in pyroptosis.

Ring domains are essential for GATOR2-dependent mTORC1 activation.

The GATOR2-GATOR1 signaling axis is essential for amino-acid-dependent mTORC1 activation. However, the molecular function of the GATOR2 complex remains unknown. Here, we report that disruption of the Ring domains of Mios, WDR24, or WDR59 completely impedes amino-acid-mediated mTORC1 activation. Mechanistically, via interacting with Ring domains of WDR59 and WDR24, the Ring domain of Mios acts as a hub to maintain GATOR2 integrity, disruption of which leads to self-ubiquitination of WDR24. Physiologically, leucine stimulation dissociates Sestrin2 from the Ring domain of WDR24 and confers its availability to UBE2D3 and subsequent ubiquitination of NPRL2, contributing to GATOR2-mediated GATOR1 inactivation. As such, WDR24 ablation or Ring deletion prevents mTORC1 activation, leading to severe growth defects and embryonic lethality at E10.5 in mice. Hence, our findings demonstrate that Ring domains are essential for GATOR2 to transmit amino acid availability to mTORC1 and further reveal the essentiality of nutrient sensing during embryonic development.

Energy status dictates PD-L1 protein abundance and anti-tumor immunity to enable checkpoint blockade.

Aberrant energy status contributes to multiple metabolic diseases, including obesity, diabetes, and cancer, but the underlying mechanism remains elusive. Here, we report that ketogenic-diet-induced changes in energy status enhance the efficacy of anti-CTLA-4 immunotherapy by decreasing PD-L1 protein levels and increasing expression of type-I interferon (IFN) and antigen presentation genes. Mechanistically, energy deprivation activates AMP-activated protein kinase (AMPK), which in turn, phosphorylates PD-L1 on Ser283, thereby disrupting its interaction with CMTM4 and subsequently triggering PD-L1 degradation. In addition, AMPK phosphorylates EZH2, which disrupts PRC2 function, leading to enhanced IFNs and antigen presentation gene expression. Through these mechanisms, AMPK agonists or ketogenic diets enhance the efficacy of anti-CTLA-4 immunotherapy and improve the overall survival rate in syngeneic mouse tumor models. Our findings reveal a pivotal role for AMPK in regulating the immune response to immune-checkpoint blockade and advocate for combining ketogenic diets or AMPK agonists with anti-CTLA4 immunotherapy to combat cancer.

Experimental evidence for the functional importance and adaptive advantage of A-to-I RNA editing in fungi.

Adenosine-to-inosine (A-to-I) editing is the most prevalent type of RNA editing in animals, and it occurs in fungi specifically during sexual reproduction. However, it is debatable whether A-to-I RNA editing is adaptive. Deciphering the functional importance of individual editing sites is essential for the mechanistic understanding of the adaptive advantages of RNA editing. Here, by performing gene deletion for 17 genes with conserved missense editing (CME) sites and engineering underedited (ue) and overedited (oe) mutants for 10 CME sites using site-specific mutagenesis at the native locus in Fusarium graminearum, we demonstrated that two CME sites in CME5 and CME11 genes are functionally important for sexual reproduction. Although the overedited mutant was normal in sexual reproduction, the underedited mutant of CME5 had severe defects in ascus and ascospore formation like the deletion mutant, suggesting that the CME site of CME5 is co-opted for sexual development. The preediting residue of Cme5 is evolutionarily conserved across diverse classes of Ascomycota, while the postediting one is rarely hardwired into the genome, implying that editing at this site leads to higher fitness than a genomic A-to-G mutation. More importantly, mutants expressing only the underedited or the overedited allele of CME11 are defective in ascosporogenesis, while those expressing both alleles displayed normal phenotypes, indicating that concurrently expressing edited and unedited versions of Cme11 is more advantageous than either. Our study provides convincing experimental evidence for the long-suspected adaptive advantages of RNA editing in fungi and likely in animals.

Disulfiram blocks inflammatory TLR4 signaling by targeting MD-2.

Toll-like receptor 4 (TLR4) sensing of lipopolysaccharide (LPS), the most potent pathogen-associated molecular pattern of gram-negative bacteria, activates NF-κB and Irf3, which induces inflammatory cytokines and interferons that trigger an intense inflammatory response, which is critical for host defense but can also cause serious inflammatory pathology, including sepsis. Although TLR4 inhibition is an attractive therapeutic approach for suppressing overexuberant inflammatory signaling, previously identified TLR4 antagonists have not shown any clinical benefit. Here, we identify disulfiram (DSF), an FDA-approved drug for alcoholism, as a specific inhibitor of TLR4-mediated inflammatory signaling. TLR4 cell surface expression, LPS sensing, dimerization and signaling depend on TLR4 binding to MD-2. DSF and other cysteine-reactive drugs, previously shown to block LPS-triggered inflammatory cell death (pyroptosis), inhibit TLR4 signaling by covalently modifying Cys133 of MD-2, a key conserved residue that mediates TLR4 sensing and signaling. DSF blocks LPS-triggered inflammatory cytokine, chemokine, and interferon production by macrophages in vitro. In the aggressive N-methyl-4-phenyl-1,2,3,6-tetrahydropyridine mouse model of Parkinson's disease (PD) in which TLR4 plays an important role, DSF markedly suppresses neuroinflammation and dopaminergic neuron loss, and restores motor function. Our findings identify a role for DSF in curbing TLR4-mediated inflammation and suggest that DSF and other drugs that target MD-2 might be useful for treating PD and other diseases in which inflammation contributes importantly to pathogenesis.

A non-pheromone GPCR is essential for meiosis and ascosporogenesis in the wheat scab fungus.

Meiosis is essential for generating genetic diversity and sexual spores, but the regulation of meiosis and ascosporogenesis is not clear in filamentous fungi, in which dikaryotic and diploid cells formed inside fruiting bodies are not free living and independent of pheromones or pheromone receptors. In this study, Gia1, a non-pheromone GPCR (G protein-coupled receptor) with sexual-specific expression in Fusarium graminearum, is found to be essential for ascosporogenesis. The gia1 mutant was normal in perithecium development, crozier formation, and karyogamy but failed to undergo meiosis, which could be partially rescued by a dominant active mutation in GPA1 and activation of the Gpmk1 pathway. GIA1 orthologs have conserved functions in regulating meiosis and ascosporogenesis in Sordariomycetes. GIA1 has a paralog, GIP1, in F. graminearum and other Hypocreales species which is essential for perithecium formation. GIP1 differed from GIA1 in expression profiles and downstream signaling during sexual reproduction. Whereas the C-terminal tail and IR3 were important for intracellular signaling, the N-terminal region and EL3 of Gia1 were responsible for recognizing its ligand, which is likely a protein enriched in developing perithecia, particularly in the gia1 mutant. Taken together, these results showed that GIA1 encodes a non-pheromone GPCR that regulates the entry into meiosis and ascosporogenesis via the downstream Gpmk1 MAP kinase pathway in F. graminearum and other filamentous ascomycetes.

Valsa mali PR1-like protein modulates an apple valine-glutamine protein to suppress JA signaling-mediated immunity.

Apple Valsa canker (AVC) is a devastating disease of apple (Malus × domestica), caused by Valsa mali (Vm). The Cysteine-rich secretory protein, Antigen 5, and Pathogenesis-related protein 1 (CAP) superfamily protein PATHOGENESIS-RELATED PROTEIN 1-LIKE PROTEIN c (VmPR1c) plays an important role in the pathogenicity of Vm. However, the mechanisms through which it exerts its virulence function in Vm-apple interactions remain unclear. In this study, we identified an apple valine-glutamine (VQ)-motif-containing protein, MdVQ29, as a VmPR1c target protein. MdVQ29-overexpressing transgenic apple plants showed substantially enhanced AVC resistance as compared with the wild type. MdVQ29 interacted with the transcription factor MdWRKY23, which was further shown to bind to the promoter of the jasmonic acid (JA) signaling-related gene CORONATINE INSENSITIVE 1 (MdCOI1) and activate its expression to activate the JA signaling pathway. Disease evaluation in lesion areas on infected leaves showed that MdVQ29 positively modulated apple resistance in a MdWRKY23-dependent manner. Furthermore, MdVQ29 promoted the transcriptional activity of MdWRKY23 toward MdCOI1. In addition, VmPR1c suppressed the MdVQ29-enhanced transcriptional activation activity of MdWRKY23 by promoting the degradation of MdVQ29 and inhibiting MdVQ29 expression and the MdVQ29-MdWRKY23 interaction, thereby interfering with the JA signaling pathway and facilitating Vm infection. Overall, our results demonstrate that VmPR1c targets MdVQ29 to manipulate the JA signaling pathway to regulate immunity. Thus, this study provides an important theoretical basis and guidance for mining and utilizing disease-resistance genetic resources for genetically improving apples.

Clinicopathological Characteristics of Breast Cancer Patients with HER-2 Low Expression Receiving Neoadjuvant Therapy.

INTRODUCTION: Human epidermal growth factor receptor-2 (HER-2) low expression breast malignant tumors have become a research hotspot in recent years, but it is still unclear whether HER-2 low expression represents a special subtype of breast cancer. However, this molecular type requires more effective treatment regimens in the neoadjuvant therapy stage. METHODS: This study enrolled breast cancer patients who were treated at Harbin Medical University Cancer Hospital with neoadjuvant treatment between October 2011 and May 2019 and was a single-center retrospective study. RESULTS: A total of 1,053 breast cancer patients who received preoperative therapy, including 279 (26%) HER-2 low expression patients, were included in this retrospective study. The HER-2 low expression group had a higher proportion of patients under 50 years old than the other two molecular subtype groups (p = 0.047, 62.0% vs. 57.2% and 52.5%), and the percentage of patients with Ki67 index above 15% was lower than that in HER-2-negative and HER-2-positive patients (p < 0.001, 50.2% vs. 63.6% and 71.5%). Most of the patients with HER-2 low expression were hormone receptor (HR) positive (p < 0.001, 85.7% vs. 60.4% and 36.0%), and their pathologic complete response (pCR) rate after neoadjuvant therapy was significantly lower than that of HER-2-negative and HER-2-positive patients (p < 0.001, 5.7% vs. 11.8% and 20.5%). The results of the subgroup analysis showed HR-positive patients with HER-2 low expression had a lower pCR rate (p < 0.001, 4.6% vs. 14.6%) and objective response rate (p = 0.001, 77.8% vs. 91.0%) than HER-2-positive patients and had no significant difference in these rates compared to HER-2-negative patients. There were no significant differences in overall survival (OS) and disease-free survival (DFS) up to 67 months (the median follow-up time) among HER-2 low, HER-2-negative, and HER-2-positive patients. The results of Cox hazard proportional showed that the Ki67 index and T stage (T3) were independent influencing factors for DFS. In terms of OS, Ki67 index, P53, T stage, and objective response were independent influencing factors for OS in HER-2 low expression patients. CONCLUSIONS: In general, further studies are needed to confirm that HER-2 low expression is a special breast cancer molecular subtype. The efficacy of neoadjuvant therapy in patients with HER-2 low expression is relatively poor, and the efficacy of neoadjuvant therapy can predict the prognosis of patients with HER-2 low expression.

Twin-Field Quantum Key Distribution with Local Frequency Reference.

Twin-field quantum key distribution (TFQKD) overcomes the linear rate-loss limit, which promises a boost of secure key rate over long distance. However, the complexity of eliminating the frequency differences between the independent laser sources hinders its practical application. We analyzed and determined the frequency stability requirements for implementing TFQKD using frequency-stabilized lasers. Based on this analysis, we proposed and demonstrated a simple and practical approach that utilizes the saturated absorption spectroscopy of acetylene as an absolute reference, eliminating the need for fast frequency locking to achieve TFQKD. Adopting the 4-intensity sending-or-not-sending TFQKD protocol, we experimentally demonstrated the TFQKD over 502, 301, and 201 km ultralow-loss optical fiber, respectively. We expect this high-performance scheme will find widespread usage in future intercity and free-space quantum communication networks.

Self-assembly of CuAuTA nanozymes for intelligent detection of ginkgolic acids.

Toxic ginkgolic acids (GAs) are a challenge for Ginkgo biloba-related food. Although a detection method for GAs is available, bulky instruments limit the field testing of GAs. Herein, by assembling gold nanoclusters with copper tannic acid (CuTA), CuAuTA nanocomposites were designed as peroxidase mimics for the colorimetric determination of GAs. Compared with single CuTA, the obtained CuAuTA nanocomposites possessed enhanced peroxidase-like properties. Based on the inhibitory effect of GAs for the catalytic activity of CuAuTA nanozymes, CuAuTA could be utilized for the colorimetric sensing of GAs with a low limit of quantitation of 0.17 µg mL-1. Using a smartphone and the ImageJ software in conjunction, a nanozyme-based intelligent detection platform was developed with a detection limit of 0.86 µg mL-1. This sensing system exhibited good selectivity against other potential interferents. Experimental data demonstrated that GAs might bind to the surface of CuAuTA, blocking the catalytically active sites and resulting in decreased catalytic activity. Our CuAuTA nanozyme-based system could also be applied to detect real ginkgo nut and ginkgo powder samples with recoveries of 93.12-111.6% and relative standard deviations less than 0.3%. Our work may offer a feasible strategy for the determination of GAs and expand the application of nanozymes in food safety detection.

Wavelength-versatile deep-red laser source by intracavity frequency converted Raman laser.

We demonstrate an efficient wavelength-selectable output in the attractive deep-red spectral region from an intracavity frequency converted Nd:YLF/KGW Raman laser. Driven by an acousto-optic Q-switched 1314 nm Nd:YLF laser, two first-Stokes waves at 1461 and 1490 nm were generated owing to the bi-axial properties of KGW crystal. By incorporating intracavity sum-frequency generation and second-harmonic generation with an angle-tuned bismuth borate (BIBO) crystal, four discrete deep-red laser emission lines were yielded at the wavelengths of 692, 698, 731, and 745 nm. Under the incident pump power of 50 W and the repetition rate of 4 kHz, the maximum average output powers of 2.4, 2.7, 3.3, and 3.6 W were attained with the pulse durations of 3.4, 3.2, 4.3, and 3.7 ns, respectively, corresponding to the peak powers up to 177, 209, 190, and 245 kW. The results indicate that the Nd:YLF/KGW Raman laser combined with an angle-adjusted BIBO crystal provides a reliable and convenient approach to achieve the selectable multi-wavelength deep-red laser with short pulse duration and high peak power.

High-power diode-end-pumped 1314†nm laser based on the multi-segmented Nd:YLF crystal.

We demonstrate the first multi-segmented Nd:YLF laser, to the best of our knowledge. The multi-segmented crystal was designed to straightforwardly aim for the minimum thermal stress without sacrificing the overall laser efficiency, with the influence of the pump beam waist position considered in particular. Integrating the enhanced thermo-mechanical resistance of multi-segmented crystal and the alleviated heat load of low quantum defect pumping, this end-pumped 1314 nm Nd:YLF laser system delivered a maximum continuous-wave output power of up to 35.5 W under a pump power of 105 W, corresponding to an optical-to-optical efficiency of 33.8%. Furthermore, by incorporating an acousto-optic modulator, an active Q-switching oscillator was accomplished, yielding a maximum average output power of 22.9 W at a pulse repetition frequency (PRF) of 20 kHz and a largest pulse energy of 13.6 mJ at a PRF of 1 kHz.

Twin-Field Quantum Key Distribution without Phase Locking.

Twin-field quantum key distribution (TF-QKD) has emerged as a promising solution for practical quantum communication over long-haul fiber. However, previous demonstrations on TF-QKD require the phase locking technique to coherently control the twin light fields, inevitably complicating the system with extra fiber channels and peripheral hardware. Here, we propose and demonstrate an approach to recover the single-photon interference pattern and realize TF-QKD without phase locking. Our approach separates the communication time into reference frames and quantum frames, where the reference frames serve as a flexible scheme for establishing the global phase reference. To do so, we develop a tailored algorithm based on fast Fourier transform to efficiently reconcile the phase reference via data postprocessing. We demonstrate no-phase-locking TF-QKD from short to long distances over standard optical fibers. At 50-km standard fiber, we produce a high secret key rate (SKR) of 1.27 Mbit/s, while at 504-km standard fiber, we obtain the repeaterlike key rate scaling with a SKR of 34 times higher than the repeaterless secret key capacity. Our work provides a scalable and practical solution to TF-QKD, thus representing an important step towards its wide applications.

Experimental Twin-Field Quantum Key Distribution over 1000 km Fiber Distance.

Quantum key distribution (QKD) aims to generate secure private keys shared by two remote parties. With its security being protected by principles of quantum mechanics, some technology challenges remain towards practical application of QKD. The major one is the distance limit, which is caused by the fact that a quantum signal cannot be amplified while the channel loss is exponential with the distance for photon transmission in optical fiber. Here using the 3-intensity sending-or-not-sending protocol with the actively-odd-parity-pairing method, we demonstrate a fiber-based twin-field QKD over 1002 km. In our experiment, we developed a dual-band phase estimation and ultra-low noise superconducting nanowire single-photon detectors to suppress the system noise to around 0.02 Hz. The secure key rate is 9.53×10^{-12} per pulse through 1002 km fiber in the asymptotic regime, and 8.75×10^{-12} per pulse at 952 km considering the finite size effect. Our work constitutes a critical step towards the future large-scale quantum network.

Free-Space and Fiber-Integrated Measurement-Device-Independent Quantum Key Distribution under High Background Noise.

Measurement-device-independent quantum key distribution (MDI QKD) provides immunity against all attacks targeting measurement devices. It is essential to implement MDI QKD in the future global-scale quantum communication network. Toward this goal, we demonstrate a robust MDI QKD fully covering daytime, overcoming the high background noise that prevents BB84 protocol even when using a perfect single-photon source. Based on this, we establish a hybrid quantum communication network that integrates free-space and fiber channels through Hong-Ou-Mandle (HOM) interference. Additionally, we investigate the feasibility of implementing HOM interference with moving satellites. Our results serve as a significant cornerstone for future integrated space-ground quantum communication networks that incorporate measurement-device-independent security.

CeO2-Based Porous Carbonaceous Frameworks as Antioxidant Nanozymes for Scavenging Reactive Oxygen Species and Adsorbing Benzo[a]pyrene.

Barbecue smoke, car exhaust, cigarette smoke, and other waste gases contain toxic reactive oxygen species (ROS) and polycyclic aromatic hydrocarbons (PAHs). Herein, CeO2-based porous carbonaceous frameworks (CeO2 PCFs) were explored as antioxidant nanozymes to scavenge ROS and absorb benzo[a]pyrene (B[a]P). Using cerium-based frameworks as the precursors, CeO2 PCFs were constructed by high-temperature calcination. Due to excellent superoxide dismutase-like and catalase-like activity, CeO2 PCFs could effectively eliminate superoxide radical, hydroxyl radical, and hydrogen peroxide. The 2,2'-azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) free radical scavenging assay had substantiated free radical scavenging ability of CeO2 PCFs. In addition, with a large surface area and porous structure, CeO2 PCFs could adsorb B[a]P efficiently. The designed CeO2 PCFs may provide a new opportunity as scavengers of ROS and absorbents of PAHs in some harmful gases.

The ubiquitination of rag A GTPase by RNF152 negatively regulates mTORC1 activation.

mTORC1 is essential for regulating cell growth and metabolism in response to various environmental stimuli. Heterodimeric Rag GTPases are required for amino-acid-mediated mTORC1 activation at the lysosome. However, the mechanism by which amino acids regulate Rag activation remains not fully understood. Here, we identified the lysosome-anchored E3 ubiquitin ligase RNF152 as an essential negative regulator of the mTORC1 pathway by targeting RagA for K63-linked ubiquitination. RNF152 interacts with and ubiquitinates RagA in an amino-acid-sensitive manner. The mutation of RagA ubiquitination sites abolishes this effect of RNF152 and enhances the RagA-mediated activation of mTORC1. Ubiquitination by RNF152 generates an anchor on RagA to recruit its inhibitor GATOR1, a GAP complex for Rag GTPases. RNF152 knockout results in the hyperactivation of mTORC1 and protects cells from amino-acid-starvation-induced autophagy. Thus, this study reveals a mechanism for regulation of mTORC1 signaling by RNF152-mediated K63-linked polyubiquitination of RagA.

Prediction of nonsentinel lymph node metastasis in breast cancer patients based on machine learning.

BACKGROUND: Develop the best machine learning (ML) model to predict nonsentinel lymph node metastases (NSLNM) in breast cancer patients. METHODS: From June 2016 to August 2022, 1005 breast cancer patients were included in this retrospective study. Univariate and multivariate analyses were performed using logistic regression. Six ML models were introduced, and their performance was compared. RESULTS: NSLNM occurred in 338 (33.6%) of 1005 patients. The best ML model was XGBoost, whose average area under the curve (AUC) based on 10-fold cross-verification was 0.722. It performed better than the nomogram, which was based on logistic regression (AUC: 0.764 vs. 0.706). CONCLUSIONS: The ML model XGBoost can well predict NSLNM in breast cancer patients.

Opposing functions of Fng1 and the Rpd3 HDAC complex in H4 acetylation in Fusarium graminearum.

Histone acetylation, balanced by histone acetyltransferase (HAT) and histone deacetylase (HDAC) complexes, affects dynamic transitions of chromatin structure to regulate transcriptional accessibility. However, little is known about the interplay between HAT and HDAC complexes in Fusarium graminearum, a causal agent of Fusarium Head Blight (FHB) that uniquely contains chromosomal regions enriched for house-keeping or infection-related genes. In this study, we identified the ortholog of the human inhibitor of growth (ING1) gene in F. graminearum (FNG1) and found that it specifically interacts with the FgEsa1 HAT of the NuA4 complex. Deletion of FNG1 led to severe growth defects and blocked conidiation, sexual reproduction, DON production, and plant infection. The fng1 mutant was normal in H3 acetylation but significantly reduced in H4 acetylation. A total of 34 spontaneous suppressors of fng1 with faster growth rate were isolated. Most of them were still defective in sexual reproduction and plant infection. Thirty two of them had mutations in orthologs of yeast RPD3, SIN3, and SDS3, three key components of the yeast Rpd3L HDAC complex. Four mutations in these three genes were verified to suppress the defects of fng1 mutant in growth and H4 acetylation. The rest two suppressor strains had a frameshift or nonsense mutation in a glutamine-rich hypothetical protein that may be a novel component of the FgRpd3 HDAC complex in filamentous fungi. FgRpd3, like Fng1, localized in euchromatin. Deletion of FgRPD3 resulted in severe growth defects and elevated H4 acetylation. In contract, the Fgsds3 deletion mutant had only a minor reduction in growth rate but FgSIN3 appeared to be an essential gene. RNA-seq analysis revealed that 48.1% and 54.2% of the genes with altered expression levels in the fng1 mutant were recovered to normal expression levels in two suppressor strains with mutations in FgRPD3 and FgSDS3, respectively. Taken together, our data showed that Fng1 is important for H4 acetylation as a component of the NuA4 complex and functionally related to the FgRpd3 HDAC complex for transcriptional regulation of genes important for growth, conidiation, sexual reproduction, and plant infection in F. graminearum.

The prediction of distant metastasis risk for male breast cancer patients based on an interpretable machine learning model.

OBJECTIVES: This research was designed to compare the ability of different machine learning (ML) models and nomogram to predict distant metastasis in male breast cancer (MBC) patients and to interpret the optimal ML model by SHapley Additive exPlanations (SHAP) framework. METHODS: Four powerful ML models were developed using data from male breast cancer (MBC) patients in the SEER database between 2010 and 2015 and MBC patients from our hospital between 2010 and 2020. The area under curve (AUC) and Brier score were used to assess the capacity of different models. The Delong test was applied to compare the performance of the models. Univariable and multivariable analysis were conducted using logistic regression. RESULTS: Of 2351 patients were analyzed; 168 (7.1%) had distant metastasis (M1); 117 (5.0%) had bone metastasis, and 71 (3.0%) had lung metastasis. The median age at diagnosis is 68.0 years old. Most patients did not receive radiotherapy (1723, 73.3%) or chemotherapy (1447, 61.5%). The XGB model was the best ML model for predicting M1 in MBC patients. It showed the largest AUC value in the tenfold cross validation (AUC:0.884; SD:0.02), training (AUC:0.907; 95% CI: 0.899-0.917), testing (AUC:0.827; 95% CI: 0.802-0.857) and external validation (AUC:0.754; 95% CI: 0.739-0.771) sets. It also showed powerful ability in the prediction of bone metastasis (AUC: 0.880, 95% CI: 0.856-0.903 in the training set; AUC: 0.823, 95% CI:0.790-0.848 in the test set; AUC: 0.747, 95% CI: 0.727-0.764 in the external validation set) and lung metastasis (AUC: 0.906, 95% CI: 0.877-0.928 in training set; AUC: 0.859, 95% CI: 0.816-0.891 in the test set; AUC: 0.756, 95% CI: 0.732-0.777 in the external validation set). The AUC value of the XGB model was larger than that of nomogram in the training (0.907 vs 0.802) and external validation (0.754 vs 0.706) sets. CONCLUSIONS: The XGB model is a better predictor of distant metastasis among MBC patients than other ML models and nomogram; furthermore, the XGB model is a powerful model for predicting bone and lung metastasis. Combining with SHAP values, it could help doctors intuitively understand the impact of each variable on outcome.