Kai Wu

Necrotizing enterocolitis (NEC) is the most devastating gastrointestinal emergency in preterm infants, frequently causing mortality and long-term neurodevelopmental impairment^1,2. Although the gut microbiota has been implicated in NEC pathogenesis^3,4,5, the microbial factors driving NEC and the following brain pathology remain unclear. Here we identify Clostridium perfringens pathobionts containing Delta toxin in the gut microbiome of individuals with severe and fatal NEC. Clinically, infected infants show worse outcomes, including aberrant systemic inflammation, severe neurological damage and coagulopathy. We establish a mouse model in which Delta-toxin-positive C. perfringens infection, or Delta toxin exposure alone, recapitulates clinical features of NEC, including intestinal necrosis and acute cerebral injury. Mechanistically, Delta toxin preferentially targets macrophages, potently activating inflammasome signalling and pyroptosis, triggering a rapid inflammatory cascade that compromises blood–brain barrier integrity and drives neuroinflammation. Through a genome-wide CRISPR screen, we identify the tetraspanin CD9 as the host receptor for Delta toxin. Using systemic, myeloid-specific and intestinal-epithelial-specific Cd9-knockout mice, we show that systemic and myeloid Cd9 ablation protects mice against toxin-induced lethality, blood–brain barrier dysfunction and multiorgan injury, whereas epithelial Cd9 deletion shows partial protection, implicating macrophage targeting as a primary driver of C. perfringens CD9-dependent pathogenesis. Furthermore, a neutralizing anti-Delta toxin monoclonal antibody confers protection against disease progression. These findings suggest that a subset of NEC is a specific-toxin-mediated disorder and validate the Delta toxin–CD9 axis as a therapeutic target for this subset.

Early recurrence of intussusception frequently occurs within 48 h after reduction, yet current assessments lack objective tools for accurate risk prediction. This multicenter study sought to develop and validate a predictive model for early intussusception recurrence using abdominal ultrasound imaging. We retrospectively collected data from 1,314 primary pediatric intussusception cases across three hospitals (2016-2024). Cases from center 1 were split into training (60%) and internal validation (40%) sets, and those from centers 2 and 3 composed the external test sets. An automated segmentation model (DeepLabv3 with ResNet101) was developed for lesion delineation. Radiomics and deep learning features were extracted from the segmented regions. Three classifiers (K-nearest neighbors, random forest, and extreme gradient boosting) were evaluated on the radiomics, deep learning, and fused feature sets. We integrated the fused model with clinical variables to construct a nomogram. The Dice coefficients of the segmentation model were 0.935 (seg-training) and 0.862 (seg-test). The key clinical predictors were age, onset-to-presentation time, and vomiting. After seven radiomics features and nine deep learning features were selected, we integrated the fused model with clinical variables to construct a nomogram. It showed good discrimination (internal validation area under the receiver operating characteristic curve (AUC), 0.892; external test set AUCs, 0.884 and 0.851, respectively). The results of the calibration and decision curve analyses supported its potential as a risk stratification tool. Integrating ultrasound-derived deep learning, radiomics, and clinical data shows promise for predicting recurrence of primary intussusception within 48 h. The model may aid early risk stratification, although prospective validation in diverse populations is needed.

Group 3 innate lymphoid cells (ILC3) are crucial in neonatal necrotizing enterocolitis (NEC); however, the underlying mechanisms remain elusive. Here, we identify NKp46⁻CCR6⁻ (double-negative, DN) ILC3s as the dominant pathogenic subset driving NEC via IL-17 A secretion, which disrupts intestinal barrier integrity. Mechanistically, Atg5 activates autophagy in DN ILC3s during NEC. Atg5 conditional knockout in RORγt⁺ cells mitigates NEC, reduces DN ILC3 accumulation and IL-17 A production. Atg5 deficiency also decreases HIF-1α chromatin accessibility and transcriptional activity, shifting DN ILC3 metabolism from glycolysis to fatty acid oxidation. Lipidomics reveals phosphatidylcholine as a key downstream metabolite of Atg5-mediated autophagy. Phosphatidylcholine supplementation suppresses DN ILC3-driven inflammation, restores metabolic homeostasis, elevates Clostridium abundance, and ameliorates NEC in mice. Importantly, human NEC tissues exhibit increased ILC3 proportions, autophagic activity, and IL-17 A/IL-22 secretion. Thus, we uncover an Atg5–autophagy–glycolipid metabolic axis in DN ILC3s that drives NEC pathogenesis, providing a promising therapeutic target for neonatal NEC.

Neuroblastoma (NB) is the most common extracranial solid tumor in children and is characterized by marked clinical heterogeneity and poor prognosis. MYCN amplification drives NB tumorigenesis through epigenetic reprogramming and is frequently accompanied by a copy-number gain of the long arm of chromosome 17 (17q). Epigenetic dysregulation of enhancer landscapes-particularly large regulatory elements termed super‑enhancers (SEs), which are enriched for H3K27ac and bound by lineage-specific master transcription factors (TFs)-establishes distinct NB cellular identities and states. These SE domains demarcate oncogenes that function as critical regulators of cell proliferation and apoptosis. Therefore, SE-driven genes represent tumor vulnerabilities, offering selective therapeutic opportunities. By integrating ATAC-seq data from 22 NB cell lines, the intratumoral heterogeneity of MYCN-amplified NB was characterized at the level of chromatin accessibility. Subsequently, based on H3K27ac ChIP-seq data from 38 NB cell lines, the ROSE algorithm was employed to identify SE-driven oncogenes. The synergistic mechanism between MYCN amplification and the 17q SE-driven gene MSI2 was investigated through genome‑wide CRISPR/Cas9 loss‑of‑function screens. At single‑cell resolution, we conducted a comprehensive analysis of the characteristics of heterogeneous tumor subpopulations and their immune microenvironment features. This analysis was performed using multiple bioinformatics workflows, including AUCell scoring, SCENIC analysis, copy‑number inference, cell differentiation‑state evaluation, neighborhood abundance tests, and separability tests. Finally, functional validation was performed using NB cell lines (MYCN‑amplified and non‑amplified) to assess gene perturbations. Pronounced epigenetic heterogeneity was observed within MYCN-amplified NB. MSI2 is an SE‑driven oncogene and is highly expressed in MYCN‑amplified NB. MSI2 and MYCN are co-expressed, may be mutually dependent, and are both correlated with cell cycle-related pathways. At the single-cell level, we identified and redefined an NB-MSI2 + MYCN+ subtype characterized by malignant transcriptional features, an immunosuppressive microenvironment, and poor patient prognosis. Building on this, combined targeting of MSI2 and MYCN markedly reduced proliferation and migration in MYCN‑amplified NB cells. The NB‑MSI2 + MYCN+ subtype defines a clinically aggressive, therapy‑refractory state characterized by high proliferation, metabolic reprogramming, and immunosuppression. For patients with MYCN-amplified NB, MSI2 is both a prognostic biomarker and a candidate therapeutic target.

Trophoblast–immune cell communication is crucial during pregnancy, with impairments linked to adverse outcomes. The accumulation of decidual polymorphonuclear myeloid‐derived suppressor cells (dPMN‐MDSCs) in the third trimester is vital for fetal development. This study presents a novel crosstalk mechanism between trophoblasts and dPMN‐MDSCs that improves adverse outcomes associated with advanced maternal age (AMA). A specific dPMN‐MDSC population with high X‐C motif chemokine receptor 1 (XCR1) expression is identified, which interacts with trophoblasts through X‐C motif chemokine ligand 1 (XCL1) during the third trimester. Spontaneous fetal growth restriction observed in AMA and pregnant Xcr1 ^−/− mice is correlated with the disruption of this interaction. Mechanistically, the deficiency in XCL1–XCR1 expression reduces nuclear FOXO1 levels, thereby impairing the transcription of FOXO1‐driven oxidative phosphorylation genes in decidual XCR1 ⁺ PMN‐MDSCs. Restoring the expression of XCL1–XCR1 or FOXO1 in dPMN‐MDSCs mitigates this effect. Crucially, their adoptive transfer or treatment with XCL1/Oltipraz rescues the delayed fetal growth linked to impaired decidual XCR1 ⁺ PMN‐MDSCs and metabolic imbalance. Our findings highlight the importance of trophoblast–dPMN‐MDSC communication via the XCL1–XCR1 axis, proposing metabolic reprogramming of dPMN‐MDSCs as a potential immunotherapeutic strategy for AMA‐related adverse outcomes.

Background: Neonatal necrotizing enterocolitis (NEC) is challenging to diagnose due to its subtle radiological features on abdominal radiographs (ARs) and considerable variability in interpretation among clinicians, especially those with limited experience, which may delay timely intervention. The study aimed to develop a convolutional neural network (CNN) based artificial intelligence (AI) model using ARs to predict NEC and the need for surgical intervention, and to evaluate its ability to assist clinicians in AR interpretation. Methods: We retrospectively collected 738 ARs from 576 preterm infants across three medical centers. The ARs were labeled into three categories: non-NEC, medical-NEC, and surgical-NEC. Data from two centers (set A, n=347; set B, n=130) were randomly split into training and internal testing sets (8:2 ratio), while data from the third center (set C, n=99) served as an external testing set. Six deep learning models [efficientnet, Inception_v3, visual geometry group (VGG), resnet, squeezenet, and densenet] were trained, and the best model was selected. Six clinicians with varying experience interpreted set C ARs, and their accuracy was compared before and after using the best model’s AI-assisted visualizations. Results: The Efficientnet-b0 model, trained on the combined dataset of A and B, performed best. On set C, it achieved area under the receiver operating characteristic curve (AUC) values of 0.883, 0.640, and 0.837 for non-NEC, medical-NEC, and surgical-NEC, respectively, outperforming single-center models (PConclusions: The CNN-based AI model effectively distinguished NEC and identified cases requiring surgery. Clinicians’ AR interpretation accuracy improved with AI assistance.

Background: Neuroblastoma (NB) development is significantly impacted by telomere maintenance where telomerase activation acts as the triggering factor. The study demonstrated the practical use of telomerase activation-related genes in the diagnosis and prognosis of NB. Methods: Weighted Gene Co-expression Network Analysis (WGCNA) and differential analysis were used to screen for prognostic-related genes in NB. Telomerase activation-related genes (TARG) were identified using Genecard and then intersected with the prognostic-related genes to obtain telomerase activation-related NB prognostic genes. The key prognosis genes were pinpointed by random forest and Lasso Cox analysis and confirmed by bioinformatics methods. Then RT-PCR experiments were employed to verify the differences in WNT4 and RPL22L1 gene expressions in tissue samples. The impact of the risk gene on NB cell proliferation, migration, and invasion was confirmed using CCK-8, Wound-healing assay, and Transwell assay. Results: PRPH, RPL22L1, ODC1, ALK, NTRK1, ALCAM, and WNT4 were the seven key telomerase prognostic genes that were found by Lasso cox analysis and confirmed by external datasets in NB. Their expression levels varied in INSS stages, age, and high-risk and low-risk groups, as indicated by the validation results. Subsequently, two previously unverified genes, RPL22L1 and WNT4, were selected to construct a multifactor regression model combined with various clinical features to predict the mortality risk. RPL22L1 was ultimately chosen and confirmed to promote NB cell proliferation, migration, and invasion in vitro . Conclusion: Telomerase activation is strongly linked to the onset and progression of NB. A multivariate regression model using the genes RPL22L1 and WNT4 proved effective in evaluating the prognosis of NB. The new molecular marker RPL22L1 can inhibit the NB cell proliferation, migration, and invasion, making it beneficial for the early diagnosis of high-risk NB.

Objectives This study aimed to develop a diagnostic model utilizing clinical and laboratory data to identify complicated appendicitis in pediatric patients, improving acute appendicitis management. Methods Retrospective analysis of pediatric acute appendicitis cases (2018–2023) at two hospitals collected medical history, clinical criteria, and blood samples. Patients were divided into complicated and uncomplicated appendicitis groups for comparison. Significant variables were identified using the Least Absolute Shrinkage and Selection Operator (LASSO), and incorporated into a logistic regression model to construct a nomogram. The effectiveness of the model was assessed based on sensitivity, specificity, accuracy, and comparison with existing scoring systems. Results Among 323 patients, four variables (neutrophil (NEU), C-reactive protein (CRP), fibrinogen (Fg), and chlorine (Cl)) were identified as significant. The recommended cutoff value of nomogram was 0.730, exceeding that of Alvarado and PAS, with higher sensitivity (81.7%), specificity (82.6%), and accuracy (82.0%), as well as better performance in both internal and external validations. Furthermore, it demonstrated excellent calibration and clinical utility. Conclusions NEU, CRP, Fg, and Cl are effective markers for diagnosing complicated appendicitis in children. The nomogram model can be considered to be incorporated into the diagnosis process of appendicitis as an auxiliary means of surgical intervention decision-making in complex appendicitis cases.

Background: Neuroblastoma (NB) is one of the most common and aggressive pediatric solid tumors, characterized by a highly complex pathogenesis. Within the tumor microenvironment (TME), cancer-associated fibroblasts (CAFs) constitute a major cell population and play a pivotal role in facilitating communication among various stromal cells. However, the specific functions and contributions of CAFs in NB remain incompletely understood. Objective: To investigate the impact of CAFs-related genes on the prognosis of NB, we developed a risk model to facilitate the diagnosis and prognostication of patients. Methods: In this study, a CAFs gene prognostic model for NB was established using single-cell analysis and genomic sequencing data. The effectiveness of this prognostic model was subsequently evaluated through the development of a nomogram, immune infiltration analysis, drug prediction, and gene set enrichment analysis. Ultimately, the expression levels of the identified key genes were experimentally validated in NB tissues. Results: A novel prognostic model for CAFs related to NB prognosis was established through single-cell analysis and transcriptome dataset analysis. The prognosis of the high-risk group was worse than that of the low-risk group. The validity of the model was confirmed by nomogram, drug sensitivity analysis, and immune infiltration methods. Finally, the high expression of the key gene STEAP2 in NB tissues was verified by experiments. Conclusions: The study introduces a new predictive model that uses CAF markers to forecast the prognosis of NB. STEAP2 plays a key role in identifying high-risk neuroblastoma and may become a potential therapeutic target for NB.