Supplementary Table S3. Antibodies used in the study.

Narrow urban rivers remain challenging for Sentinel-2-based chlorophyll-a (Chl-a) retrieval because mixed pixels, adjacency effects, and bank shadows can contaminate spectral signals, while in situ observations are often limited. This study develops an integrated workflow that integrates a high-confidence urban-river water mask with explicit shadow suppression to constrain feature extraction and sample selection, cross-resolution pseudolabel transfer by mapping a GF-1 Chl-a field to the Sentinel-2 10 m grid and stratifying it to form pseudolabeled samples for augmentation, and a leakage-controlled out-of-fold stacking regressor with an XGBoost metalearner to improve stability under data scarcity. On the in situ test set (n = 6), the GF-1 retrieval achieves R-2 = 0.86 and RMSE = 3.32 mu g/L. Using mapped pseudolabels for Sentinel-2 training, the model attains R-2 = 0.84 and RMSE = 3.88 mu g/L on a held-out pseudolabel set (n = 40), while the in situ only baseline yields R-2 = 0.70 and RMSE = 4.47 mu g/L on the separate in situ test set. These results indicate that cross-resolution pseudolabel augmentation and leakage-controlled ensembling can be effective under limited field supervision; the pseudolabel evaluation reflects internal consistency rather than independent ground truth.

Supplementary Table S1. Clinicopathological characteristics of BLCA patients in in-house scRNA-seq cohort.

Supplementary Figure S3. CCL5 hi-CD4 + T cells enhance M1 macrophage infiltration. (A) Dot plot showing CCL5 expression across cell subsets within tumor microenvironment. (B) Representative flow cytometry plots showing the proportions of non-T cells (CD4 -CD8 -), CD4 + cells, and CD8 + cells across different CCL5 expression levels. (C) Quantification of CD8 +CD4 + , CD4 + , CD8 + , and CD4 -CD8 - cell proportions within CCL5-high-expression cells (n = 3/group). (D) CellChat analysis predicting the interaction likelihood between CCL5 hi-CD4+ T cells (source) and cell subsets within tumor microenvironment (target). Dot size, color intensity, and line width represent the number of interactions. (E) UMAP plots showing distinct myeloid subsets in human BLCA. (F) Dot plot of top differentially expressed marker genes for each myeloid subset. (G) Difference in proportion of CD68+CD86+ cells among total CD68+ cells between CR and non-CR groups (left, n = 12 CR samples; right, n = 28 non-CR samples). (H) Correlation between CCL5 hi-CD4 + T cells and M1 macrophages across TCGA pan-cancer datasets. (I) Representative mIHC images from each treatment group showing F4/80+CD86+ dual-positive cells (F4/80: magenta; CD86: cyan; nuclei: DAPI/blue); scale bars: 50 μm. (J) Quantification of F4/80+CD86+ cells within total F4/80+ populations (n = 7 mice/group). (K) Representative flow cytometry plots showing M1 macrophages (CD206 -CD86 + F4/80 + ) and M2 macrophages (CD206 +CD86 -F4/80 + ) subsets across treatment groups. (L) Differences in M2 macrophage infiltration among treatment groups (n = 5–7 mice/group). (M) Representative mIHC images from each treatment group showing CD8+ cells (CD8: yellow; nuclei: DAPI/blue); scale bars: 100 μm. (N) Quantification of CD8+ cells as a percentage of total cells across treatment groups (n = 7 mice/group). (O) Expression of canonical CCL5 receptors in human myeloid cell subsets. (P) Representative mIHC images showing spatial proximity between CCL5 hi-CD4 + T cells (CCL5 +CD4 + , yellow arrows) and CCR1 + M1-like macrophages (CCR1 +CD86 + , white arrows); scale bars: 50 μm. (Q) Expression of phenotypical markers (M2) on RAW264.7 cells co-cultured with CCL5 hi-CD4+ T cells or CCL5 hi-CD4+ T cells plus BX471. P values were calculated as follows: two-sided paired t-test (C), two-sided unpaired Wilcoxon rank-sum test (G), one-way ANOVA (J, N and Q), and Kruskal–Wallis tests (L).

Supplementary Table S4. Primers used in the study.

Conventional chemotherapy achieves clinical efficacy not only through its cytotoxic effects but also by reactivating immune surveillance. However, whether chemotherapy can inversely suppress antitumor immunity remains largely unexplored. Here, we integrate cross-species single-cell and spatial transcriptomics to investigate how chemotherapy programs immune cell plasticity. Our findings reveal that chemotherapy-educated liver-resident Kupffer cells (KCs) promote immune evasion and chemoresistance in liver metastases. These reprogrammed KCs, characterized by leptin receptor expression (LEPR+), originate from preexisting KCs and differentiate via STING (Stimulator of interferon genes)-ID1 signaling, driven by paracrine cGAMP (cyclic GMP-AMP) released from chemotherapy-treated tumor cells. Unlike conventional KCs at the tumor periphery, LEPR+ KCs infiltrate tumors and suppress antitumor immunity through MerTK-dependent efferocytosis that eliminates chemotherapy-induced immunogenic cell death (ICD) signals. Targeting LEPR+ KCs enhances tumor immunogenicity and promotes antitumor T cell responses. Together, our study highlights the potential of combining efferocytosis inhibitors with immunotherapy to overcome chemoresistance.

Immune checkpoint inhibitors (ICI) have transformed cancer therapy, yet their efficacy remains limited to a subset of patients, underscoring the need for robust predictive biomarkers and deeper mechanistic insights into treatment resistance. In this study, we identified a population of CCL5hiCD4+ T cells that were characterized by memory-like activation markers and strongly correlated with ICI therapeutic responses in bladder cancer. Functionally, these T cells enhanced antitumor immunity by promoting M1-like macrophage polarization through CCL5/CCR1 signaling. Importantly, tumor-derived prostaglandin E2 (PGE2) acted as a critical microenvironmental factor that suppressed the differentiation of CCR6hiCD4+ T cells into immunostimulatory CCL5hiCD4+ T cells, thereby driving resistance to ICI therapy. These findings extend the understanding of CD4+ T-cell heterogeneity and its role in shaping immune responses to ICI. By elucidating that CCL5hiCD4+ T cells enhance myeloid-mediated tumor control and that tumor-derived PGE2 disrupts CCL5hiCD4+ T-cell differentiation to promote immune evasion, this work highlights potential therapeutic strategies to improve ICI efficacy in bladder cancer.Significance: CCL5 hiCD4+ T cells with memory-like activated characteristics enhance antitumor immunity in bladder cancer by reprogramming macrophages, supporting the potential of these cells as biomarkers and targets to enhance immunotherapy efficacy.

Supplementary Figure S4. Characteristics of different subsets along the differentiation trajectory towards CCL5 hi-CD4 + T cells. (A) Non-negative least-squares regression (NNLS) analysis showing the similarity of CD4+ T cell subsets in tumor-only scRNA-seq data and scRNA-seq data from the tumor–lymph-node–peripheral-blood mixture. (B) Heatmap showing effector-marker and memory-activated CD4 + T-cell signatures across CD4 + T cell subsets. (C) Validation of OVA protein levels in the MB49 cells. (D) Schematic illustrating the generation of OT-II-CCL5 hi-CD4 + T cells via adoptive transfer. (E) Relative tumor-cell viability after co-culture with CCL5 hi-CD4+ T cells versus OT-II-CCL5 hi-CD4+ T cells (n = 5/group). (F) Change in tumor volumes at multiple time points among different treatment groups (n = 5 mice/group). (G) Change in tumor volumes at multiple time points among different treatment groups (n = 5 mice/group). (H) Differential abundance of TXK hi-CD4+ T cells and CCR7 hi-CD4+ T cells between CR/PR and SD/PD groups in the IMvigor210 cohort. (I) Kaplan-Meier survival analysis stratified by CCR6 hi-CD4 T cells infiltration levels. (J) Correlation of CCR6 hi-CD4+ T cell infiltration abundance with M1 TAMs in the TCGA dataset. (K) Flow cytometry plots and quantification of CCL5 + cells within CD25 -CCR6 -CD4 + versus CD25 -CCR6 +CD4 + T cells (n = 3/group). P values were calculated as follows: two-sided unpaired t-test (E), repeated measures two-way ANOVA (F, G), two-sided unpaired Wilcoxon rank-sum test (H), log-rank test (I), Spearman correlation (J) and two-sided paired t-test (K).

Supplementary Figure S2. Single-cell analysis of bladder cancer identifies diverse cell types including distinct CD4 + T cell subsets. (A) Heatmap of top differentially expressed marker genes for each cell type in BLCA (left); dot plot comparing transcriptional profiles between CD4+ and CD8+ T cell subsets (right). (B) Bar graph showing the relative abundance of indicated cell subsets in BLCA. (C) Heatmap displaying expression patterns of CD4-3 subset-specific marker genes across CD4+ T cell subsets. (D) UMAP plots showing distinct CD4+ T cell subsets in human BLCA derived from the PRJNA662018 dataset (n = 11 samples). (E) Dot plot depicting the memory activated CD4+ T cell signature and the expression of selected CD4-3 marker genes across CD4 + T cell subsets in PRJNA662018 dataset. (F) Bar graph showing the relative abundance of indicated cell subsets in BLCA. (G) UMAP plots showing memory-activated CD4+ T cell signature scores across CD4+ T cell subsets. (H) Comparison of CD4 + T cell subsets in this study with published human CD4 + T cell states. (I) Dot plots depict the memory coinhibitory scores across CD4 + T cell subsets. (J) Violin plot showing expression of immune-checkpoint genes in tumor versus peritumoral tissues. (K) Volcano plot of differentially expressed genes in CCL5 hi-CD4 + T cells (tumor vs peritumoral tissues) with immune-checkpoint genes highlighted in red. (L) Differential proportion of CCL5 hi-CD4+ T cells between CR/PR and SD/PD groups in the IMvigor210 cohort. (M) Representative mIHC images and quantification of the proportion of CD4+CXCL13+ cells among total CD4+ T cells in CR (n = 12) versus non-CR (n = 28) patients. (CD4: green; CXCL13: red; nuclei: DAPI/blue). Scale bars: 50 μm. (N) Representative mIHC images and quantification of the proportion of CD4+FOXP3+ cells among total CD4+ T cells in CR (n = 12) versus non-CR (n = 28) patients. (CD4: green; FOXP3: red; nuclei: DAPI/blue). Scale bars: 50 μm. (O) Representative mIHC images and quantification of the proportion of CD8+ cells among total cells in CR (n = 12) versus non-CR (n = 28) patients. (CD8: white; DAPI/blue). Scale bars: 3 mm. (P) UMAP plots showing distinct CD4+ T cell subsets in merged Daping Cohort 1 and Cohort 2. (Q) UMAP plots showing distinct CD4+ T cell subsets in Daping Cohort 2. (R) PCA projection showing the similarity of CD4+ T cell subset between Daping Cohort 2 and Cohort 1. (S) The proportion of each subset within CD4 + T cells across samples in Daping Cohort 2. (T) Differential abundance of memory activated CD4 + T cells between CR/PR and SD/PD groups in SKCM (melanoma) and STAD (gastric cancer). (U) Violin plot depicting the expression of selected CCL5 hi-CD4 marker genes and Cxcr6 across CD4 + T cell subsets in mouse BLCA. (V) Dot plot depicting the expression of CXCR6 across CD4 + T cell subsets in human BLCA. (W) Flow cytometry plot showing the sorting strategy for CCL5 hi-CD4 cells. (X) Schematic workflow for isolation of CD45.1 +CD25 -CXCR6 +CD4 + T cells, adoptive transfer into CD45.2 recipients and analyzed at various time points. (Y) Representative flow cytometry plots showing CD45.1 vs CD45.2 progression from days 1 to 5, and quantification of the CD45.1 + :CD45.2 + ratio within CD4 +FOXP3 -CCL5 + T cells. P values were calculated as follows: two-sided unpaired t-test (B, F, M, N, and O), two-sided unpaired Wilcoxon rank-sum test (L, T).

Supplementary Figure S5. The relationship between tumor-derived PGE2, CCL5 hi-CD4 + T cell infiltration in bladder cancer. (A) Patient stratification by CCR6 hi-CD4 + T cell and CCL5 hi-CD4 + T cell abundance (group A-D). (B) Abundance of memory activated CD4 + T cells (left) and memory resting CD4 + T cells (right) in group B and C. (C) Shared upregulated genes (n = 45 genes) in group C versus group B across TCGA (n = 323 genes) and IMvigor210 (n = 91 genes) cohorts. (D) Expression of the 45 upregulated genes across cell subsets within tumor microenvironment. (E) UMAP plots showing distinct tumor cell subsets in human BLCA. (F) Dot plot of selected top differentially expressed marker genes for each tumor subset. (G) UMAP plots showing selected gene signature scores across tumor cell subsets. (H) Differential cell-cell communication networks between tumor cells of distinct scoring groups and CCR6 hi-CD4 + T cells. (I) Kaplan-Meier survival analysis stratified by PGE2 score levels. (J) ELISA showing the difference in PGE2 levels between CR (n = 4) and non-CR (n = 6) patients. (K) Correlation of PGE2 scores with CCR6 hi-CD4 differentiation-related tumor genes score. (L) Heatmap showing the enrichment of PGE2 score across tumor subsets and cell populations within tumor microenvironment. (M) Representative mIHC images showing COX2 localization to the tumor epithelium (COX2: red; CK8/18: yellow); scale bars: 100 μm. (N) Representative mIHC images showing the spatial relationship between CCL5 hi-CD4 + T cells (CCL5 +CD4 + , white arrows) and tumor with different COX2 expression levels; scale bars: 800 μm. (O) Quantification of the proportion of CCL5 +CD4 + cells among CD4 + T cells in tumor regions with high versus low PGE2 levels. (P) Validation of COX2 protein levels in MB49 cells after COX2 knockdown. (Q) Representative flow cytometry plots comparing the proportion of CCL5 + or CCR6 + cells within CD25 -CD4 + T cells and of M1-macrophage (CD86 +CD206 -F4/80 + ) across treatment groups. (R) Representative flow cytometry plots showing the proportion of CCL5 + cells within CD45.1 +FOXP3 -CD4 + T cells across treatment groups at the indicated days after adoptive transfer of CD45.1 +CD25 -CCR6 +CD4 + T cells. (S) Representative flow cytometry plots comparing the proportion of CCL5 + cells among all CD4 + T cells across different treatment groups. P values were calculated as follows: two-sided unpaired Wilcoxon rank-sum test (B), log-rank test (I), two-sided unpaired t-test (J, O) and Spearman correlation (K)

Supplementary Table S2. Clinicopathological characteristics of BLCA patients receiving neoadjuvant chemo-immunotherapy in the Daping cohort

Fructose consumption increases the risk of obesity-related metabolic diseases and some cancers, but its role in hepatocellular carcinogenesis (HCC) remains controversial. Animal studies suggest that high fructose promotes HCC, whereas human data fail to support the positive link between fructose intake and elevated risk of liver cancer. Moreover, fructose metabolism is progressively attenuated in HCC with the loss of key fructolytic enzymes, including fructose-1,6-bisphosphate aldolase B (ALDOB). Here, we report that fructose suppresses HCC through fructose 1-phosphate (F1P)-mediated inhibition of mannose phosphate isomerase (MPI) in the context of ALDOB deficiency. Transcriptomic and metabolic flux analyses using human HCC cells and tissues revealed that liver cancer cells retain a significant ability to metabolize fructose despite the downregulation of fructolytic genes, with ALDOB showing the earliest and most pronounced suppression compared with GLUT2 and KHK. Dietary supplementation with 10% fructose suppressed HCC in liver-specific Aldob knockout mice. Further spatial and single-cell transcriptomic analyses of clinical HCC samples revealed the spatiotemporal dynamics of fructolytic gene expression and identified subsets of cancer cells that retain fructose uptake and phosphorylation capacity (SLC2A2⁺/KHK⁺) but lack ALDOB expression. Upon fructose exposure, accumulated F1P binds to and inhibits MPI, reducing protein N-glycosylation and triggering apoptosis due to maladaptive ER stress. We further performed virtual high-throughput screening of FDA-approved and clinical-trial drugs and identified ebselen as a potent MPI inhibitor. Taken together, the results of our study reveal a novel mechanism by which dietary fructose inhibits HCC through the F1P-MPI axis, suggesting a therapeutic strategy targeting metabolic vulnerabilities in cancer.

Abstract Background Colorectal cancer (CRC) is a leading cause of cancer‐related death and is associated with high recurrence rates. Solute carrier family 7 member 5 (SLC7A5), a core transporter that facilitates the transmembrane movement of tryptophan, plays a role in various cancers. However, whether and how SLC7A5 promotes colorectal liver metastasis (CRLM) through tryptophan metabolism reprogramming and immune remodelling remain unexplored. Methods We integrated public datasets and clinical specimens to analyse SLC7A5 expression and prognosis, and validated its role in proliferation and metastasis using in vitro assays and in vivo models. Targeted metabolomics and isotope tracing identified kynurenine (Kyn) and xanthurenic acid (XANA) as downstream metabolites of SLC7A5. Single‐cell RNA sequencing (scRNA‐seq) and conditioned medium experiments were used to assess the impact of SLC7A5 on the tumour immune microenvironment (TIME). Results SLC7A5 expression increases sequentially in normal tissue, primary tumours and liver metastases, and higher SLC7A5 levels are associated with worse prognosis. SLC7A5 facilitates CRC cell growth, metastasis and epithelial‒mesenchymal transition (EMT) by promoting the production of Kyn and XANA and subsequent activation of the aryl hydrocarbon receptor (AhR). scRNA‐seq analysis and conditioned medium experiments demonstrated that SLC7A5 knockdown reprograms the TIME by driving macrophages towards an antigen‐presenting phenotype, alleviating CD8+ T‐cell exhaustion, polarising CD4+ T cells towards Th1/Th17 subsets and triggering antigen‐driven immunoglobulin G (IgG)‐secreting B‐cell clonal expansion, effects that were reversed by exogenous Kyn/XANA supplementation. Moreover, combination therapy with the SLC7A5 inhibitor JPH203 and anti‐programmed cell death protein 1 (PD‐1) antibody produced synergistic tumour growth inhibition and heightened antitumour immune responses. Conclusions Collectively, our findings reveal that SLC7A5 drives CRLM through tryptophan/Kyn/XANA–AhR signalling and concomitant remodelling of the TIME, positioning SLC7A5 as a promising target for combination therapy with anti‐PD‐1 in CRLM.

BACKGROUND:N-acetyltransferase 10 (NAT10) is an RNA acetyltransferase that catalyzes N4-acetylcytidine (ac⁴C) modification and regulates mRNA stability. However, its biological function and mechanistic role in colorectal cancer (CRC) remain poorly defined. METHODS:NAT10 expression was analyzed across multiple GEO cohorts and paired CRC clinical specimens. Gain- and loss-of-function experiments were performed to assess the effects of NAT10 on CRC cell proliferation, migration, colony formation, and tumor growth in vivo. Transcriptomic correlation and enrichment analyses were used to identify NAT10-associated pathways. RIP-seq mining, NAT10-RIP-qPCR, and ac⁴C-RIP-qPCR were applied to verify downstream targets. Actinomycin D chase assays were used to evaluate mRNA stability. The functional relevance of CDK4 was examined using genetic NAT10 perturbation, Remodelin-based pharmacologic treatment, and CDK4 rescue experiments. RESULTS:NAT10 was markedly up-regulated in CRC tissues compared with normal mucosa and was maintained at high levels in malignant CRC lesions. NAT10 overexpression enhanced CRC cell proliferation, migration, and colony formation, whereas NAT10 knockout suppressed these phenotypes. In vivo, NAT10-deficient cells formed significantly smaller and slower-growing xenograft tumors, with markedly reduced tumor volume and weight compared with controls. Pathway analyses indicated strong enrichment of cell-cycle programs, particularly the G1/S transition. CDK4 was identified as a NAT10-associated ac⁴C-modified target. NAT10 depletion destabilized CDK4 mRNA, reduced CDK4-associated cell-cycle protein expression, and induced G1/S accumulation, while NAT10 overexpression produced the opposite effects. Remodelin treatment, used as a pharmacologic perturbation of NAT10-associated signaling, suppressed CDK4 expression and CRC cell growth, and CDK4 overexpression partially rescued these inhibitory effects. CONCLUSIONS:This study identifies a mechanistic NAT10-ac⁴C-CDK4 regulatory axis that stabilizes CDK4 mRNA, promotes G1/S transition, and drives CRC progression. Targeting NAT10 or its downstream CDK4 pathway represents a potential therapeutic strategy for CRC.

Accurate assessment of bifacial photovoltaic (PV) potential is essential for global energy planning. However, optimizing installation parameters remains challenging due to the complex geometry of ground-reflected radiation. This study addresses this issue by presenting a framework for mapping global bifacial PV potential, which involves determining the optimal tilt and azimuth angles. We integrate a height-corrected view factor model into a hybrid optimization algorithm that combines particle swarm optimization with gradient-based local search. This overcomes the limitations of traditional cross-strings methods by explicitly accounting for module mounting height. We simulate the optimal configuration and energy yields of bifacial PV worldwide using multi-source global datasets, including CERES radiation, ERA5 meteorological reanalysis and MODIS albedo products. The simulation is validated against in-situ irradiance and energy yield measurements from bifacial experimental sites (NREL and Sandia). Our model outperforms standard models (pvlib, PVRT, and RT) in irradiance simulation. For the energy conversion step, we incorporate a bifacial thermal model and a mismatch loss factor, which further improves the accuracy. The optimized installation angles reveal distinct geographic patterns driven by latitude and ground albedo. The optimal tilt angles for bifacial systems are generally 0-10 degrees lower than those for monofacial counterparts, particularly in high-latitude and high-albedo regions. Furthermore, our performance analysis shows that optimized bifacial systems achieve a global average energy gain of 10.92% (ranging from 0% to 46.17%) compared to monofacial systems. Angular optimization can provide an additional yield increase of up to 10% in high-latitude regions. This work reveals the global patterns in bifacial PV performance, providing valuable insights for PV system design, resource assessment, and energy yield forecasting.

Patients with myelodysplastic syndrome (MDS) harboring SRSF2 (serine and arginine rich splicing factor 2) mutations exhibit poor prognosis and aberrant inflammatory activation, underscoring an urgent need for therapies. Here, we reveal that low messenger RNA expression of SETD2 (SET domain containing 2) in hematopoietic stem and progenitor cells (HSPCs) from patients with MDS carrying SRSF2P95 mutations (SRSF2P95-Mut MDS) correlates with adverse outcomes and increased inflammation. Multivariate analysis confirmed the correlation between low SETD2 expression and poor prognosis in patients with SRSF2P95-Mut MDS. Furthermore, Setd2 loss in the Srsf2P95H/+ mouse model resulted in lethal MDS with hyperinflammation and expansion of myeloid-derived suppressor cells (MDSCs). Mechanistically, SETD2 methylates SRSF2P95H at lysine-17 and lysine-65 to inhibit aberrant splicing of CEACAM1-4 (isoforms of carcinoembryonic antigen cell adhesion molecule), which enhances interleukin-1β (IL-1β) signaling through Slc7a11 (solute carrier family 7 member 11)-mediated cystine uptake, thereby promoting HSPC differentiation into MDSCs, establishing an IL-1β-driven immunosuppressive microenvironment. These findings identify the SRSF2P95HK17me1K65me2-CEACAM1-4 signaling axis as a promising therapeutic target in SRSF2P95-Mut MDS.

Supplementary Table S5. Abbreviations and Full Names of Cancer Types.