Neutrophil profiling illuminates anti-tumor antigen-presenting potency.

Neutrophils, the most abundant and efficient defenders against pathogens, exert opposing functions across cancer types. However, given their short half-life, it remains challenging to explore how neutrophils adopt specific fates in cancer. Here, we generated and integrated single-cell neutrophil transcriptomes from 17 cancer types (225 samples from 143 patients). Neutrophils exhibited extraordinary complexity, with 10 distinct states including inflammation, angiogenesis, and antigen presentation. Notably, the antigen-presenting program was associated with favorable survival in most cancers and could be evoked by leucine metabolism and subsequent histone H3K27ac modification. These neutrophils could further invoke both (neo)antigen-specific and antigen-independent T cell responses. Neutrophil delivery or a leucine diet fine-tuned the immune balance to enhance anti-PD-1 therapy in various murine cancer models. In summary, these data not only indicate the neutrophil divergence across cancers but also suggest therapeutic opportunities such as antigen-presenting neutrophil delivery.

Deep Learning Prediction Boosts Phosphoproteomics-Based Discoveries Through Improved Phosphopeptide Identification.

Shotgun phosphoproteomics enables high-throughput analysis of phosphopeptides in biological samples. One of the primary challenges associated with this technology is the relatively low rate of phosphopeptide identification during data analysis. This limitation hampers the full realization of the potential offered by shotgun phosphoproteomics. Here we present DeepRescore2, a computational workflow that leverages deep learning-based retention time and fragment ion intensity predictions to improve phosphopeptide identification and phosphosite localization. Using a state-of-the-art computational workflow as a benchmark, DeepRescore2 increases the number of correctly identified peptide-spectrum matches by 17% in a synthetic dataset and identifies 19% to 46% more phosphopeptides in biological datasets. In a liver cancer dataset, 30% of the significantly altered phosphosites between tumor and normal tissues and 60% of the prognosis-associated phosphosites identified from DeepRescore2-processed data could not be identified based on the state-of-the-art workflow. Notably, DeepRescore2-processed data uniquely identifies EGFR hyperactivation as a new target in poor-prognosis liver cancer, which is validated experimentally. Integration of deep learning prediction in DeepRescore2 improves phosphopeptide identification and facilitates biological discoveries.

Dual Inhibition of CDK4/6 and XPO1 Induces Senescence With Acquired Vulnerability to CRBN-Based PROTAC Drugs.

BACKGROUND & AIMS: Despite the increasing number of treatment options available for liver cancer, only a small proportion of patients achieve long-term clinical benefits. Here, we aim to develop new therapeutic approaches for liver cancer. METHODS: A compound screen was conducted to identify inhibitors that could synergistically induce senescence when combined with cyclin-dependent kinase (CDK) 4/6 inhibitor. The combination effects of CDK4/6 inhibitor and exportin 1 (XPO1) inhibitor on cellular senescence were investigated in a panel of human liver cancer cell lines and multiple liver cancer models. A senolytic drug screen was performed to identify drugs that selectively killed senescent liver cancer cells. RESULTS: The combination of CDK4/6 inhibitor and XPO1 inhibitor synergistically induces senescence of liver cancer cells in vitro and in vivo. The XPO1 inhibitor acts by causing accumulation of RB1 in the nucleus, leading to decreased E2F signaling and promoting senescence induction by the CDK4/6 inhibitor. Through a senolytic drug screen, cereblon (CRBN)-based proteolysis targeting chimera (PROTAC) ARV-825 was identified as an agent that can selectively kill senescent liver cancer cells. Up-regulation of CRBN was a vulnerability of senescent liver cancer cells, making them sensitive to CRBN-based PROTAC drugs. Mechanistically, we find that ubiquitin specific peptidase 2 (USP2) directly interacts with CRBN, leading to the deubiquitination and stabilization of CRBN in senescent liver cancer cells. CONCLUSIONS: Our study demonstrates a striking synergy in senescence induction of liver cancer cells through the combination of CDK4/6 inhibitor and XPO1 inhibitor. These findings also shed light on the molecular processes underlying the vulnerability of senescent liver cancer cells to CRBN-based PROTAC therapy.

Impact of respiratory aerosol size and number distribution on the relative importance of different routes in SARS-CoV-2 transmission.

Exploring the relative importance of different routes in SARS-CoV-2 transmission is crucial in infection prevention. However, even in the same environmental setting, the relative importance of different routes has varied in different studies. We hypothesize that respiratory aerosol size and number distribution might play a key role. In this study, size and number distribution of respiratory droplets emitted from breathing, talking, and coughing were identified from PubMed and Web of Science. The infection risk of SARS-CoV-2 via airborne, droplet, and fomite transmission routes was modeled in a household and a healthcare setting. The relative importance of three routes varied with different size distributions in both settings. Generally, the contribution of the airborne route increased with the volume percentage of respirable droplets emitted. And the increase of the total number of emitted droplets leads to an increase in the contribution of tdroplet route. In the healthcare setting, as the total number of emitted droplets increased from 110 to 4,973, the contribution of droplet route increased from 62.24% to 98.11%. Next, by considering the combination of breathing, coughing, and talking when the infected person was asymptomatic, the airborne route predominated over the droplet and contact routes. When the infected person had developed symptoms, that is, cough, the droplet route played a dominant role in SARS-CoV-2 transmission. In conclusion, risk analyses will be improved with improved sampling methods that enable characterization of viruses within respiratory droplets of different sizes.

An algorithm-assisted automated identification and enumeration system for sensitive hydrogen sulfide sensing under dark field microscopy.

Hydrogen sulfide (H2S) is an active physiological molecule, and its intracellular level has great significance to life functions. In this study, an effective and sensitive method was developed for H2S sensing with dark field microscopy (DFM). The proposed method employed AuNPs as the signal source, DFM as the readout system, and an intelligence algorithm as the image processing and output systems, respectively. The AuNP surface was modified with azido and alkynyl in advance, and then added into a tube cap. As the H2S evaporated from the solution and selectively reduced azido to amino, the click chemistry reaction was inhibited, which resulted in the AuNPs being well dispersed in the solution; otherwise, AuNP aggregation occurred. The scattering colour of single AuNPs could be easily distinguished from that of AuNP aggregations with DFM, and the number or ratio of single AuNPs could also be easily obtained by the custom algorithm. The results showed that the H2S content could be linearly analyzed in a range from 2-80 µM. Furthermore, the proposed sensing strategy has been applied for H2S detection in cell lysate. Compared with the traditional colorimetric method, the results showed no significant difference, indicating the good prospects of the algorithm and proposed H2S sensing method.

Increasing contributions of airborne route in SARS-CoV-2 omicron variant transmission compared with the ancestral strain.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron variant has become the dominant lineage worldwide. Experimental studies have shown that SARS-CoV-2 Omicron variant is more stable on various environmental surfaces than the ancestral strains of SARS-CoV-2. However, the influences on the role of the contact route in SARS-CoV-2 transmission are still unknown. In this study, we built a Markov chain model to simulate the transmission of the Omicron and ancestral strains of SARS-CoV-2 within a household over a 1-day period from multiple pathways; that is, airborne, droplet, and contact routes. We assumed that there were two adults and one child in the household, and that one of the adults was infected with SARS-CoV-2. We assumed two scenarios. (1) Asymptomatic/presymptomatic infection, and (2) symptomatic infection. During asymptomatic/presymptomatic infection, the contact route contributing the most (37%-45%), followed by the airborne (34%-38%) and droplet routes (21%-28%). During symptomatic infection, the droplet route was the dominant pathway (48%-71%), followed by the contact route (25%-42%), with the airborne route playing a negligible role (<10%). Compared to the ancestral strain, although the contribution of the contact route increased in Omicron variant transmission, the increase was slight, from 25%-41% to 30%-45%. With the growing concern about the increase in the proportion of asymptomatic/presymptomatic infection in Omicron strain transmissions, the airborne route, rather than the fomite route, should be of focus. Our findings suggest the importance of ventilation in the SARS-CoV-2 Omicron variant prevention in building environment.

Induction of functional hepatocyte-like cells from mouse fibroblasts by defined factors.

The generation of functional hepatocytes independent of donor liver organs is of great therapeutic interest with regard to regenerative medicine and possible cures for liver disease. Induced hepatic differentiation has been achieved previously using embryonic stem cells or induced pluripotent stem cells. Particularly, hepatocytes generated from a patient's own induced pluripotent stem cells could theoretically avoid immunological rejection. However, the induction of hepatocytes from induced pluripotent stem cells is a complicated process that would probably be replaced with the arrival of improved technology. Overexpression of lineage-specific transcription factors directly converts terminally differentiated cells into some other lineages, including neurons, cardiomyocytes and blood progenitors; however, it remains unclear whether these lineage-converted cells could repair damaged tissues in vivo. Here we demonstrate the direct induction of functional hepatocyte-like (iHep) cells from mouse tail-tip fibroblasts by transduction of Gata4, Hnf1α and Foxa3, and inactivation of p19(Arf). iHep cells show typical epithelial morphology, express hepatic genes and acquire hepatocyte functions. Notably, transplanted iHep cells repopulate the livers of fumarylacetoacetate-hydrolase-deficient (Fah(-/-)) mice and rescue almost half of recipients from death by restoring liver functions. Our study provides a novel strategy to generate functional hepatocyte-like cells for the purpose of liver engineering and regenerative medicine.

Optimized Hepatocyte-Like Cells with Functional Drug Transporters Directly-Reprogrammed from Mouse Fibroblasts and their Potential in Drug Disposition and Toxicology.

BACKGROUND/AIMS: To develop a suitable hepatocyte-like cell model that could be a substitute for primary hepatocytes with essential transporter expression and functions. Induced hepatocyte-like (iHep) cells directly reprogrammed from mice fibroblast cells were fully characterized. METHODS: Naïve iHep cells were transfected with nuclear hepatocyte factor 4 alpha (Hnf4α) and treated with selected small molecules. Sandwich cultured configuration was applied. The mRNA and protein expression of transporters were determined by Real Time PCR and confocal. The functional transporters were estimated by drug biliary excretion measurement. The inhibition of bile acid efflux transporters by cholestatic drugs were assessed. RESULTS: The expression and function of p-glycoprotein (P-gp), bile salt efflux pump (Bsep), multidrug resistance-associated protein 2 (Mrp2), Na+-dependent taurocholate cotransporting polypeptide (Ntcp), and organic anion transporter polypedtides (Oatps) in iHep cells were significantly improved after transfection of hepatocyte nuclear factor 4 alpha (Hnf4α) and treatment with selected inducers. In vitro intrinsic biliary clearances (CLb,int) of optimized iHep cells for rosuvastatin, methotrexate, d8-TCA (deuterium-labeled sodium taurocholate acid) and DPDPE ([D-Pen2,5] enkephalin hydrate) correlated well with that of sandwich-cultured primary mouse hepatocytes (SCMHs) (r2 = 0.984). Cholestatic drugs were evaluated and the results were compared well with primary mice hepatocytes. CONCLUSION: The optimized iHep cells expressed functional drug transporters and were comparable to primary mice hepatocytes. This study suggested direct reprogramming could provide a potential alternative to primary hepatocytes for drug candidate hepatobiliary disposition and hepatotoxicity screening.

Hypergraph-Based Multi-Modal Representation for Open-Set 3D Object Retrieval.

The traditional 3D object retrieval (3DOR) task is under the close-set setting, which assumes the categories of objects in the retrieval stage are all seen in the training stage. Existing methods under this setting may tend to only lazily discriminate their categories, while not learning a generalized 3D object embedding. Under such circumstances, it is still a challenging and open problem in real-world applications due to the existence of various unseen categories. In this paper, we first introduce the open-set 3DOR task to expand the applications of the traditional 3DOR task. Then, we propose the Hypergraph-Based Multi-Modal Representation (HGM 2 R) framework to learn 3D object embeddings from multi-modal representations under the open-set setting. The proposed framework is composed of two modules, i.e., the Multi-Modal 3D Object Embedding (MM3DOE) module and the Structure-Aware and Invariant Knowledge Learning (SAIKL) module. By utilizing the collaborative information of modalities derived from the same 3D object, the MM3DOE module is able to overcome the distinction across different modality representations and generate unified 3D object embeddings. Then, the SAIKL module utilizes the constructed hypergraph structure to model the high-order correlation among 3D objects from both seen and unseen categories. The SAIKL module also includes a memory bank that stores typical representations of 3D objects. By aligning with those memory anchors in the memory bank, the aligned embeddings can integrate the invariant knowledge to exhibit a powerful generalized capacity toward unseen categories. We formally prove that hypergraph modeling has better representative capability on data correlation than graph modeling. We generate four multi-modal datasets for the open-set 3DOR task, i.e., OS-ESB-core, OS-NTU-core, OS-MN40-core, and OS-ABO-core, in which each 3D object contains three modality representations: multi-view, point clouds, and voxel. Experiments on these four datasets show that the proposed method can significantly outperform existing methods. In particular, the proposed method outperforms the state-of-the-art by 12.12%/12.88% in terms of mAP on the OS-MN40-core/OS-ABO-core dataset, respectively. Results and visualizations demonstrate that the proposed method can effectively extract the generalized 3D object embeddings on the open-set 3DOR task and achieve satisfactory performance.

Multi-model analysis of gallbladder cancer reveals the role of OxLDL-absorbing neutrophils in promoting liver invasion.

BACKGROUND: Gallbladder cancer (GBC) is the most common and lethal malignancy of the biliary tract that lacks effective therapy. In many GBC cases, infiltration into adjacent organs or distant metastasis happened long before the diagnosis, especially the direct liver invasion, which is the most common and unfavorable way of spreading. METHODS: Single-cell RNA sequencing (scRNA-seq), spatial transcriptomics (ST), proteomics, and multiplexed immunohistochemistry (mIHC) were performed on GBC across multiple tumor stages to characterize the tumor microenvironment (TME), focusing specifically on the preferential enrichment of neutrophils in GBC liver invasion (GBC-LI). RESULTS: Multi-model Analysis reveals the immunosuppressive TME of GBC-LI that was characterized by the enrichment of neutrophils at the invasive front. We identified the context-dependent transcriptional states of neutrophils, with the Tumor-Modifying state being associated with oxidized low-density lipoprotein (oxLDL) metabolism. In vitro assays showed that the direct cell-cell contact between GBC cells and neutrophils led to the drastic increase in oxLDL uptake of neutrophils, which was primarily mediated by the elevated OLR1 on neutrophils. The oxLDL-absorbing neutrophils displayed a higher potential to promote tumor invasion while demonstrating lower cancer cytotoxicity. Finally, we identified a neutrophil-promoting niche at the invasive front of GBC-LI that constituted of KRT17+ GBC cells, neutrophils, and surrounding fibroblasts, which may help cultivate the oxLDL-absorbing neutrophils. CONCLUSIONS: Our study reveals the existence of a subset of pro-tumoral neutrophils with a unique ability to absorb oxLDL via OLR1, a phenomenon induced through cell-cell contact with KRT17+ GBC cells in GBC-LI.

Cell fate conversion: direct induction of hepatocyte-like cells from fibroblasts.

One of the essential features of stem cells is their cellular plasticity to differentiate into daughter cells with defined functions. Recently, induction of pluripotent stem cells from somatic cells by defined transcription factors led to the focus on cellular plasticity of terminally differentiated cells. This approach is adopted by other studies to demonstrate the cell fate conversion between different lineages of terminally differentiated cells. We and others showed that induced hepatocyte-like (iHep) cells are directly converted from mouse fibroblasts by overexpression of liver-enriched transcription factors. iHep cells as well as pluripotent stem cell- or mesenchymal stem cell-derived hepatocyte-like cells provide potential cell sources for disease modeling, transplantation, and tissue engineering independent of donor organs. Here, we review the latest advances in generating hepatocyte-like cells and summarize general criteria for evaluating these cells. In addition, we propose a possible role of the p19(Arf) /p53 pathway in cell fate maintenance, which apparently limits the formation of induced pluripotent stem (iPS) cells and iHep cells.

HGNN+: General Hypergraph Neural Networks.

Graph Neural Networks have attracted increasing attention in recent years. However, existing GNN frameworks are deployed based upon simple graphs, which limits their applications in dealing with complex data correlation of multi-modal/multi-type data in practice. A few hypergraph-based methods have recently been proposed to address the problem of multi-modal/multi-type data correlation by directly concatenating the hypergraphs constructed from each single individual modality/type, which is difficult to learn an adaptive weight for each modality/type. In this paper, we extend the original conference version HGNN, and introduce a general high-order multi-modal/multi-type data correlation modeling framework called HGNN + to learn an optimal representation in a single hypergraph based framework. It is achieved by bridging multi-modal/multi-type data and hyperedge with hyperedge groups. Specifically, in our method, hyperedge groups are first constructed to represent latent high-order correlations in each specific modality/type with explicit or implicit graph structures. An adaptive hyperedge group fusion strategy is then used to effectively fuse the correlations from different modalities/types in a unified hypergraph. After that a new hypergraph convolution scheme performed in spatial domain is used to learn a general data representation for various tasks. We have evaluated this framework on several popular datasets and compared it with recent state-of-the-art methods. The comprehensive evaluations indicate that the proposed HGNN + framework can consistently outperform existing methods with a significant margin, especially when modeling implicit data correlations. We also release a toolbox called THU-DeepHypergraph for the proposed framework, which can be used for various of applications, such as data classification, retrieval and recommendation.

Timing and Magnitude of the Second Wave of the COVID-19 Omicron Variant - 189 Countries and Territories, November 2021 to February 2023.

What is already known about this topic?: The first nationwide wave of coronavirus disease 2019 (COVID-19), driven by the Omicron variant, has largely subsided. However, subsequent epidemic waves are inevitable due to waning immunity and the ongoing evolution of the severe acute respiratory syndrome coronavirus 2. What is added by this report?: Insights gleaned from other nations offer guidance regarding the timing and scale of potential subsequent waves of COVID-19 in China. What are the implications for public health practice?: Understanding the timing and magnitude of subsequent waves of COVID-19 in China is crucial for forecasting and mitigating the spread of the infection.

Graph Learning on Millions of Data in Seconds: Label Propagation Acceleration on Graph Using Data Distribution.

Graph-based semi-supervised learning methods have been used in a wide range of real-world applications, e.g., from social relationship mining to multimedia classification and retrieval. However, existing methods are limited along with high computational complexity or not facilitating incremental learning, which may not be powerful to deal with large-scale data, whose scale may continuously increase, in real world. This paper proposes a new method called Data Distribution Based Graph Learning (DDGL) for semi-supervised learning on large-scale data. This method can achieve a fast and effective label propagation and supports incremental learning. The key motivation is to propagate the labels along smaller-scale data distribution model parameters, rather than directly dealing with the raw data as previous methods, which accelerate the data propagation significantly. It also improves the prediction accuracy since the loss of structure information can be alleviated in this way. To enable incremental learning, we propose an adaptive graph updating strategy which can update the model when there is distribution bias between new data and the already seen data. We have conducted comprehensive experiments on multiple datasets with sample sizes increasing from seven thousand to five million. Experimental results on the classification task on large-scale data demonstrate that our proposed DDGL method improves the classification accuracy by a large margin while consuming much less time compared to state-of-the-art methods.

Deep learning prediction boosts phosphoproteomics-based discoveries through improved phosphopeptide identification.

Shotgun phosphoproteomics enables high-throughput analysis of phosphopeptides in biological samples, but low phosphopeptide identification rate in data analysis limits the potential of this technology. Here we present DeepRescore2, a computational workflow that leverages deep learning-based retention time and fragment ion intensity predictions to improve phosphopeptide identification and phosphosite localization. Using a state-of-the-art computational workflow as a benchmark, DeepRescore2 increases the number of correctly identified peptide-spectrum matches by 17% in a synthetic dataset and identifies 19%-46% more phosphopeptides in biological datasets. In a liver cancer dataset, 30% of the significantly altered phosphosites between tumor and normal tissues and 60% of the prognosis-associated phosphosites identified from DeepRescore2-processed data could not be identified based on the state-of-the-art workflow. Notably, DeepRescore2-processed data uniquely identifies EGFR hyperactivation as a new target in poor-prognosis liver cancer, which is validated experimentally. Integration of deep learning prediction in DeepRescore2 improves phosphopeptide identification and facilitates biological discoveries.

Genome-Scale CRISPR screen identifies LAPTM5 driving lenvatinib resistance in hepatocellular carcinoma.

ABBREVIATIONS: cld-CASP3: cleaved caspase 3; cld-PARP: cleaved PARP; DTP: drug tolerant persister; GO: Gene Ontology; GTEx: The Genotype-Tissue Expression; HCC: hepatocellular carcinoma; HCQ: hydroxychloroquine; IC50: half maximal inhibitory concentration value; KEGG: Kyoto Encyclopedia of Genes and Genomes; LAPTM5: lysosomal protein transmembrane 5; NT: non-targeting; PDC: patient-derived primary cell lines; PDO: patient-derived primary organoid; TCGA: The Cancer Genome Atlas.

An Inflammatory Checkpoint Generated by IL1RN Splicing Offers Therapeutic Opportunity for KRAS-Mutant Intrahepatic Cholangiocarcinoma.

KRAS mutations are causally linked to protumor inflammation and are identified as driving factors in tumorigenesis. Here, using multiomics data gathered from a large set of patients, we showed that KRAS mutation was associated with a specific landscape of alternative mRNA splicing that connected to myeloid inflammation in intrahepatic cholangiocarcinoma (iCCA). Then, we identified a negative feedback mechanism in which the upregulation of interleukin 1 receptor antagonist (IL1RN)-201/203 due to alternative splicing confers vital anti-inflammatory effects in KRAS-mutant iCCA. In KRAS-mutant iCCA mice, both IL1RN-201/203 upregulation and anakinra treatment ignited a significant antitumor immune response by altering neutrophil recruitment and phenotypes. Furthermore, anakinra treatment synergistically enhanced anti-PD-1 therapy to activate intratumoral GZMB+ CD8+ T cells in KRAS-mutant iCCA mice. Clinically, we found that high IL1RN-201/203 levels in patients with KRAS-mutant iCCA were significantly associated with superior response to anti-PD-1 immunotherapy. SIGNIFICANCE: This work describes a novel inflammatory checkpoint mediated by IL1RN alternative splicing variants that may serve as a promising basis to develop therapeutic options for KRAS-mutant iCCA and other cancers. This article is featured in Selected Articles from This Issue, p. 2109.

Pharmaco-proteogenomic characterization of liver cancer organoids for precision oncology.

Organoid models have the potential to recapitulate the biological and pharmacotypic features of parental tumors. Nevertheless, integrative pharmaco-proteogenomics analysis for drug response features and biomarker investigation for precision therapy of patients with liver cancer are still lacking. We established a patient-derived liver cancer organoid biobank (LICOB) that comprehensively represents the histological and molecular characteristics of various liver cancer types as determined by multiomics profiling, including genomic, epigenomic, transcriptomic, and proteomic analysis. Proteogenomic profiling of LICOB identified proliferative and metabolic organoid subtypes linked to patient prognosis. High-throughput drug screening revealed distinct response patterns of each subtype that were associated with specific multiomics signatures. Through integrative analyses of LICOB pharmaco-proteogenomics data, we identified the molecular features associated with drug responses and predicted potential drug combinations for personalized patient treatment. The synergistic inhibition effect of mTOR inhibitor temsirolimus and the multitargeted tyrosine kinase inhibitor lenvatinib was validated in organoids and patient-derived xenografts models. We also provide a user-friendly web portal to help serve the biomedical research community. Our study is a rich resource for investigation of liver cancer biology and pharmacological dependencies and may help enable functional precision medicine.

Kullback-Leibler Divergence Metric Learning.

The Kullback-Leibler divergence (KLD), which is widely used to measure the similarity between two distributions, plays an important role in many applications. In this article, we address the KLD metric-learning task, which aims at learning the best KLD-type metric from the distributions of datasets. Concretely, first, we extend the conventional KLD by introducing a linear mapping and obtain the best KLD to well express the similarity of data distributions by optimizing such a linear mapping. It improves the expressivity of data distribution, which means it makes the distributions in the same class close and those in different classes far away. Then, the KLD metric learning is modeled by a minimization problem on the manifold of all positive-definite matrices. To deal with this optimization task, we develop an intrinsic steepest descent method, which preserves the manifold structure of the metric in the iteration. Finally, we apply the proposed method along with ten popular metric-learning approaches on the tasks of 3-D object classification and document classification. The experimental results illustrate that our proposed method outperforms all other methods.

Multimodule characterization of immune subgroups in intrahepatic cholangiocarcinoma reveals distinct therapeutic vulnerabilities.

BACKGROUND: Immune microenvironment is well recognized as a critical regulator across cancer types, despite its complex roles in different disease conditions. Intrahepatic cholangiocarcinoma (iCCA) is characterized by a tumor-reactive milieu, emphasizing a deep insight into its immunogenomic profile to provide prognostic and therapeutic implications. METHODS: We performed genomic, transcriptomic, and proteomic characterization of 255 paired iCCA and adjacent liver tissues. We validated our findings through H&E staining (n=177), multiplex immunostaining (n=188), single-cell RNA sequencing (scRNA-seq) (n=10), in vitro functional studies, and in vivo transposon-based mouse models. RESULTS: Integrated multimodule data identified three immune subgroups with distinct clinical, genetic, and molecular features, designated as IG1 (immune-suppressive, 25.1%), IG2 (immune-exclusion, 42.7%), and IG3 (immune-activated, 32.2%). IG1 was characterized by excessive infiltration of neutrophils and immature dendritic cells (DCs). The hallmark of IG2 was the relatively higher tumor-proliferative activity and tumor purity. IG3 exhibited an enrichment of adaptive immune cells, natural killer cells, and activated DCs. These immune subgroups were significantly associated with prognosis and validated in two independent cohorts. Tumors with KRAS mutations were enriched in IG1 and associated with myeloid inflammation-dominated immunosuppression. Although tumor mutation burden was relatively higher in IG2, loss of heterozygosity in human leucocyte antigen and defects in antigen presentation undermined the recognition of neoantigens, contributing to immune-exclusion behavior. Pathological analysis confirmed that tumor-infiltrating lymphocytes and tertiary lymphoid structures were both predominant in IG3. Hepatitis B virus (HBV)-related samples tended to be under-represented in IG1, and scRNA-seq analyses implied that HBV infection indeed alleviated myeloid inflammation and reinvigorated antitumor immunity. CONCLUSIONS: Our study elucidates that the immunogenomic traits of iCCA are intrinsically heterogeneous among patients, posing great challenge and opportunity for the application of personalized immunotherapy.