RASIP1-positive TECs in pancreatic adenocarcinoma: a potential novel type of endothelial cells correlated with “hot” tumors

Abstract

Pancreatic adenocarcinoma (PDAC) is a highly lethal malignancy characterized by profound resistance to immunotherapy. Converting immunologically “cold” tumors into “hot” tumors by enhancing T cell infiltration represents a promising therapeutic strategy, yet the vascular mechanisms regulating immune recruitment in PDAC remain poorly defined. Here, we integrated single-cell RNA sequencing with GEPIA analysis and spatial transcriptomics to investigate the functional role of Ras Interacting Protein 1 (RASIP1)-positive endothelial cells in PDAC. We identified RASIP1-positive tumor endothelial cells as a distinct endothelial subpopulation enriched in leukocyte transendothelial migration pathways, with upregulated adhesion molecules and spatial co-localization with T-effector and IFN-γ signatures. Multiplex immunohistochemistry of human PDAC tissues revealed prominent perivascular accumulation of CD8⁺/GranzymeB⁺ cytotoxic T lymphocytes surrounding RASIP1-positive vessels. Mechanistically, RASIP1 knockdown reduced ICAM1 expression, whereas RASIP1 overexpression enhanced ICAM1 signaling, and both modulated ERK phosphorylation dynamics, suggesting that RASIP1 regulates endothelial functionality through ERK-related signaling. Collectively, our findings identify a distinct endothelial state that actively shapes the immune microenvironment of PDAC. Targeting RASIP1-positive endothelial cells may represent a potential strategy to enhance tumor immunogenicity and improve responsiveness to immunotherapy.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-40948-y.

Introduction

Pancreatic adenocarcinoma (PDAC) remains one of the leading causes of death among all tumors, which is with poor prognosis and lack of efficient systemic treatments^1,2. Most of pancreatic adenocarcinoma patients are diagnosed at an advanced stage with metastasis^3,4. Moreover, pancreatic adenocarcinoma’s response to immunotherapy is limited due to low immune infiltration⁵.

Immunotherapies, particularly immune checkpoint inhibitors (ICIs), have become a breakthrough treatment modality for several solid tumors, such as melanoma and lung cancer. Pancreatic adenocarcinoma is widely regarded as one of the non-inflamed tumors (“cold” tumors), which is difficult for immune system to recognize and attack⁶. Immune-based therapeutic approaches for treating pancreatic adenocarcinoma face significant challenges, in part due to a shortage of high-quality T cells within the tumor microenvironment⁷. Converting immunogenic “cold” tumors to “hot” tumors by increasing T cells infiltration has become a promising but still challenging strategy in the treatment of pancreatic adenocarcinoma.

Tumor microenvironment is an entity that continuously evolving during tumorigenesis^8,9. It comprises of a complex mixture of cells includes immune cells, stromal cells, and endothelial cells. The formation of freshly developed blood vessels by endothelial cells is a process called angiogenesis, which provides oxygen and nutrition to tumor and is important for tumorigenesis¹⁰. The function of endothelial cells in tumorigenesis is complex and, at times, paradoxical. Previous research has shown that tumor endothelial cells (TECs) are related to immunoevasion, and angiogenesis plays a key role in subverting T-cell-mediated immunosurveillance. Moreover, it is reported that endothelial cells facilitate to recruitment of intra-tumoral CD8 + T cells. However, on the other side, endothelial cells could inhibit T cell activation by expression of the PD-L1 and PD-L2¹¹.

The tumor tissue of pancreatic adenocarcinoma is of hypo vascularity¹². Extensive studies have reported pancreatic adenocarcinoma-driven endothelial ablation¹³. Based on our study, we have identified a subpopulation of tumor endothelial cells (TECs), characterized by Ras Interacting Protein 1 (RASIP1) positivity, that exhibits a latent therapeutic effect in pancreatic adenocarcinoma. It is reported that RASIP1 interacts with H-RAS, K-RAS, as well as RAP1¹⁴. RASIP1 is required for angiogenesis in the early embryo through interacting with GTPase. On the other side, RASIP1 deficiency may lead to instability of blood vessel formation and defect of angiogenesis^15,16. The function of RASIP1 in endothelial cells of pancreatic adenocarcinoma has not been well studied per se. Based on our analysis, RASIP1-positive TECs are likely associated with the conversion of immunogenic “cold” tumors to “hot” tumors by increasing T cells infiltration.

This RASIP1-positive TEC subpopulation represents a novel discovery in PDAC. Its potential clinical value lies in serving as a targeted modulator of the tumor vasculature, which could enhance T-cell trafficking and infiltration into the tumor core. Importantly, this subpopulation’s relevance to immunotherapy stems from its association with pathways that promote leukocyte transendothelial migration, potentially overcoming PDAC’s immunosuppressive barriers and synergizing with ICIs to transform “cold” tumors into immunologically “hot” environments, thereby improving therapeutic outcomes.

Materials and methods

Single cell RNA-seq analysis

The raw expression matrix for single-cell RNA-seq (scRNA-seq) was obtained from the Genome Sequence Archive (GSA) database with the accession number CRA001160¹⁷. The raw expression matrix was analyzed using the R package Seurat¹⁸. Cells with fewer than 200 features were omitted due to low quality. The gene expression matrix of remaining cells was normalized by the global-scaling normalization method, LogNormalize. The FindVariableFeatures function was used to identify genes exhibiting high variability, with the default setting returning 2000 feature genes. The matrix with these variable genes was subjected to Principal Components Analysis (PCA). The first 10 PCA components were selected for further visualization relying on Uniform Manifold Approximation and Projection (UMAP). Cell clustering was performed using the FindClusters function, with the resolution parameter set to 1. The cell types were identified by the expression of marker genes. A dot plot illustrating the expression of marker genes across cell types was generated using the DotPlot function. The top 10 marker genes for tumor and normal endothelial cells were identified and visualized using the DoHeatmap function, while their expression was further illustrated using the FeaturePlot and VlnPlot functions.

Ratio of global unshifted entropy (ROGUE) analysis of endothelial cells

Ratio of Global Unshifted Entropy (ROGUE) analysis is a method to evaluate the purity of a given cluster as a distinct cell population in unsupervised scRNA-seq data analyses. ROGUE analysis of endothelial cells was performed relying on R package ROGUE. The ROGUE values were calculated using the function CalculateRogue and subsequently visualized to compare the values across endothelial cells at various tumor stages.

Pseudo-time analysis for endothelial cells

Cellular trajectory analysis of endothelial cells was performed using R package Monocle 3¹⁹. The trajectory was learned by the function learn_graph with the default setting. The population of normal endothelial cells was set as the root state.

Identification of marker genes of RASIP1-positive TECs

TECs with positive expression of RASIP1 were defined as RASIP1-positive TECs. On the other side, TECs with no expression of RASIP1 were defined as RASIP1-negative TECs. Marker genes of RASIP1-positive TECs were identified by the function of FindMarkers in the R package Seurat, using DESeq2 method. Volcano plot was illustrated by R package EnhancedVolcano.

Gene expression profiling interactive analysis (GEPIA)

Gene Expression Profiling Interactive Analysis (GEPIA) is a widely used online tool for mining and analyzing data from The Cancer Genome Atlas (TCGA) database²⁰. Survival analysis stratified by the proportion of endothelial cells was conducted using the Cell Type-level Survival Analysis tool in GEPIA2021. Additionally, survival analysis of RASIP1, LUM, CX26 and RGS5 expression was performed using the Survival Analysis tool in GEPIA2. RASIP1, LUM, CX26 and RGS5 expression across different stages of pancreatic cancer was visualized using the Stage Plot tool in GEPIA2.

The survival analysis of pan-cancer patients with immunotherapy

The online tool Kaplan Meier plotter was applied in our analysis²¹. The survival analysis was performed based on the data of 1361 cancer patients treated with immune-checkpoint inhibitors, including 797 cancer patients treated with anti-programmed cell death protein 1 (PD1), 486 cancer patients treated with anti-programmed cell death ligand 1 (PDL1), and 131 cancer patients treated with anti-cytotoxic T lymphocyte-associated antigen-4 (CTLA4). In the survival analysis, the data of cancer patients was stratified based on RASIP1 expression, with automatically determined cut-off values using the default settings.

Secreted signaling analysis

The secreted signaling of endothelial cells was analyzed by the R package CellChat (version 1.5.0)²². The secreted signaling information in CellChatDB database was used to analyze the interaction between RASIP positive endothelial cells, RASIP negative endothelial cells, T cells and B cells. The cell communication network was calculated by the function computeCommunProbPathway. The significant interactions were illustrated by the function netVisual_bubble.

Spatial transcriptomics

The public spatial transcriptomics data for pancreatic adenocarcinoma, generated using the 10X Visium platform, was graciously provided by Dr. Jihui Hao’s lab. The data were obtained from four patient-derived xenograft (PDX) mouse models²³. Genes representing the T-effector and interferon-gamma (IFN-γ) gene signature were illustrated, which is considered a hallmark of hot tumors²⁴. The Loup Browser (version 7.0) was used to visualize the spatial expression of RASIP1 and the selected genes, including CXCL10, CXCL9, GBP1, GZMB, IFI16, STAT1, and TAP1. Gene expression clustering was performed using the K-means method. The expression of RASIP1 and the T-effector and IFN-gamma gene signature was visualized using violin plots. The correlation analysis between RASIP1 and the T-effector and IFN-gamma gene signature, immune cell signatures, as well as CD8 and PDCD1, was performed using the online tool TIMER 2.0, based on the TCGA database.

Moreover, we analyzed publicly available human pancreatic cancer spatial transcriptomics data from GSE202740, which contains four samples profiled using 10x Genomics Visium. Raw data were processed using the Seurat package in R. Following standard preprocessing, dimensionality reduction was performed using principal component analysis (PCA), followed by graph-based clustering (Louvain algorithm) and visualization with Uniform Manifold Approximation and Projection (UMAP). Endothelial cell-enriched spots were identified based on high expression of the canonical endothelial marker gene CDH5 (VE-cadherin). Within the spots assigned to the endothelial cell cluster, we evaluated the spatial correlation between RASIP1 expression and selected T cell-related marker genes, including TRAC and TRBC2 (components of the T cell receptor), CD8B (cytotoxic T cell marker), and IFNG (encoding interferon-gamma, a key effector cytokine of activated T cells). Correlation analysis was performed using Spearman’s rank correlation coefficient.

Multiplex immunohistochemistry staining of pancreatic adenocarcinoma tissue

The collection of clinical samples was approved by the Research Ethics Committee of the Second Hospital of Hebei Medical University, under approval number 2021-R127. The research was conducted according to the principles of the World Medical Association Declaration of Helsinki. All participants provided informed consent. After obtaining informed consent, pancreatic adenocarcinoma tissue samples from three patients undergoing pancreatic surgery were collected and subjected to multiplex immunohistochemistry staining. A multiplex immunohistochemistry kit (Pinuofei Biological) was used, with each primary antibody dilution carefully optimized during individual staining. 7-µm-thick slices of pancreatic adenocarcinoma tissue were prepared. The slides were deparaffinized and treated with a citrate-based antigen retrieval solution (Pinuofei Biological, P0026). The tissues were then blocked for 30 min using a blocking buffer containing 10% goat serum (Pinuofei Biological, PN0038). Subsequently, the slides were incubated with the corresponding primary antibody for 2 h at room temperature, followed by incubation with the corresponding secondary HRP-conjugated antibody for 1 h. The TSA dye (Pinuofei Biological) was used for slide staining for 30 min. Subsequently, the slides were washed three times with PBS, each wash lasting for 5 min. The multiplex staining was then continued to the next round of staining. After a total of 5 rounds of staining were completed, DAPI (Pinuofei Biological, PN0015) was applied.

The slides were then scanned using a Pannoramic MIDI scanner (3DHISTECH) and analyzed with CaseViewer software (version 2.4.0.119028 × 64 with CNV module; 3DHISTECH). With guidance from a pathologist and reference to H&E-stained sections, blood vessels (defined as ring structures formed by CD31 + cells) were manually identified. Subsequently, within or adjacent to blood vessels, the staining characteristics of each immunostained protein were manually assessed to identify positively stained cells. Then, the per-vessel proportion of RASIP1-positive endothelial cells was graded (-, +, ++) for subtype categorization. Concurrently, the perivascular density of positively stained immune cells was quantified (-, +, ++).

Tumor-associated vasculature was identified by the presence of ring-like CD31+ (red) endothelial structures with clear lumens. Each vessel cross-section was then classified according to the proportion of endothelial cells showing positive RASIP1 staining as follows: RASIP1++ (high): ≥ 2/3 of the endothelial cells within the vessel are RASIP1-positive. RASIP1+ (intermediate): ≥ 1/3 but < 2/3 of the endothelial cells are RASIP1-positive. RASIP1– (negative/low): < 1/3 of the endothelial cells are RASIP1-positive. Perivascular immune cell infiltration (CD3+, CD8+, and Granzyme B+ cells) was manually quantified in the immediate perivascular region surrounding each vessel cross-section using the following semi-quantitative grading criteria: – (negative/minimal infiltration): Almost no or only sporadic fluorescent signal spots (≤ 5 discrete signals), with scattered and non-clustered distribution. + (moderate infiltration): Fluorescent signals form obvious clustered aggregates, covering approximately 1/3 to 2/3 of the vessel circumference. ++ (strong/dense infiltration): Fluorescent signals form large and dense clusters, covering > 2/3 of the vessel circumference or nearly completely encircling the vessel cross-section area.

Statistical analyses were performed with GraphPad Prism software 9.5.0 (GraphPad Software Inc., San Diego, CA). Composition ratio of categorical variables were compared using Chi-square or Fisher’s exact tests.

Cell culture

Human umbilical vein endothelial cells (HUVECs) were cultured in Endothelial Cell Medium (ECM) supplemented with Endothelial Cell Growth Supplement (ECGS, Cat. #1052), 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, and 1% epidermal growth factor (EGF). Cells were incubated under 5% CO² at 37 ℃.

siRNA transfection

The siRNA targeting RASIP1 was synthesized by GenePharma (Suzhou, China) with the following sequences: sense strand, 5’-CCAGGACUUUGUGGUGUAUTT-3’; antisense strand, 5’-AUACACCACAAAGUCCUGGTT-3’. HUVECs at 60–70% confluence were subjected to siRNA transfection. The day before transfection, the cells were cultured in an antibiotic-free medium. Prior to transfection, HUVECs were digested and resuspended in an antibiotic-free medium. RASIP1 or non-targeting control siRNA complexes were prepared using Lipofectamine 2000 (Invitrogen). The siRNA complex medium was then incubated with HUVECs for 6 h. Next, the normal culture medium was replaced with the siRNA complex medium. After 48 h of incubation, the transfected HUVECs were subjected to further analysis.

Plasmid transfection

The pcDNA3.3-RASIP1-C-3HA was synthesized by BMbios (Shanghai, China). HUVECs with 50–60% confluence was ready for plasmid transfection. 1 h before transfection, normal culture medium was replaced Opti-MEM™ I Reduced Serum Medium (Thermo Fisher). Lipid complexes with a plasmid vector overexpressing RASIP1 or an empty vector were formed using Lipofectamine 3000 (Thermo Fisher) in Opti-MEM ™ Reduced Serum Medium. The DNA-lipid complex was added to the cells. After 6 h, Opti-MEM™ I Reduced Serum Medium was replaced normal culture medium. After 48-hours culture, the transfected HUVECs were collected for Western blotting analysis.

Western blotting

The HUVECs were lysed in RIPA buffer containing protease inhibitor cocktail. Western blotting was performed on the HUVEC lysates using antibodies against RASIP1, extracellular signal-regulated kinase (ERK), phospho-ERK (p-ERK), p38, Ras homolog family member A (RHOA), intercellular cell adhesion molecule-1 (ICAM1), and vascular cell adhesion molecule 1 (VCAM1). The Western blotting procedure followed the protocol described in previous literature²⁵.

Construction of mRNA expression profile of HUVECs

HUVECs with RASIP1 knockdown and control cells were collected in three biological replicates. Total RNA was extracted, and samples meeting the quality criteria were selected for mRNA sequencing. RNA quality was assessed using the Agilent 2100 BioAnalyzer, and accurate quantification of total RNA input was performed with the QUBIT RNA Assay Kit. Library preparation and sequencing were conducted on the Illumina platform by Boao Jingdian (Beijing, China). The raw readings obtained were stored as FASTQ files. Fastp (version 0.23.4) was used to filter the sequencing data, and HISAT2 (version 2.2.1) was applied to align the reads to the reference genome to obtain SAM files. SAMtools (version 1.21) was then used to convert the SAM files into BAM format and perform sorting. Finally, featureCounts in Subread (version 2.0.5) was used for quantitative counting to generate the expression matrix. DESeq2 (version 1.46.0) was then applied for differential expression analysis. The bulk RNA-seq data has been uploaded to the Genome Sequence Archive (GSA) and is available under the accession number HRA008311.

Gene functional enrichment analysis

The top 100 marker genes of normal and tumor endothelial cells were identified using the FindAllMarkers function from the R package Seurat. Gene symbols of these marker genes were analyzed for functional enrichment using the online tool Metascape²⁶. Differentially expressed genes (DEGs) in RASIP1-positive tumor endothelial cells (TECs) and DEGs in siRASIP1-treated HUVECs were also subjected to functional enrichment analysis using Metascape.

TCGA-PAAD bulk RNA-seq data analysis

Bulk RNA-seq gene expression data (TPM values) and corresponding clinical and pathologic information for pancreatic ductal adenocarcinoma (PDAC) cases were downloaded from The Cancer Genome Atlas (TCGA-PAAD) dataset via the Genomic Data Commons (GDC) portal. Samples with available pathologic tumor stage information were included for analysis. Spearman’s rank correlation coefficient was used to evaluate the relationship between RASIP1 expression levels and clinical parameters within the stage I–II subgroup.

Statistical analysis

R (version 4.2.2) was used for data analysis and Figure plotting. Western blotting data were analyzed using ImageJ (version 1.54j) and GraphPad Prism (version 10.0.2). The experimental data were described as means ± standard deviation. All statistical tests were two-sided and p < 0.05 was considered statistically significant.

Results

The hypo vascularity in pancreatic adenocarcinoma

It is well studied that pancreatic adenocarcinoma is of hypo vascularity²⁷. In our analysis, the hypo vascularity was also found in scRNA-seq data of pancreatic adenocarcinoma patients. Clustering of scRNA-seq data was processed (Fig. 1a), and the type of cell populations were identified by marker genes (Fig. 1b). During the progression of pancreatic adenocarcinoma, the abundance of each population was dramatically altered (Fig. 1c). Consistent with previous report, the number of endothelial cells was gradually decreased over the course of tumor development (Fig. 1f). The purity of each endothelial cluster were evaluated by Ratio of Global Unshifted Entropy (ROGUE) analysis, indicating balanced endothelial purity in different stage of PDAC (Fig. 1d). However, fibroblasts and stellate cells, as well as their marker genes, were gradually increased during the progression of PDAC (Fig. 1a, h, i, j, Fig. S1a, b).

Fig. 1.

Single-cell data analysis. (a), UMAP plot showing cell types in single-cell journal samples. (b), Cell types were annotated by the marker genes. (c), UMAP plot showing differences in cell types at different stages during the progression of pancreatic adenocarcinoma. (d), Bar graph evaluating changes in the purity of endothelial cell populations in the progression of pancreatic adenocarcinoma by ROGUE (Ratio of Global Unshifted Entropy). (e), UMAP diagram analyzing the status changes of endothelial cells in different stages of pancreatic adenocarcinoma. (f), Bar graph demonstrating changes in the proportion of endothelial cells in the progression of pancreatic adenocarcinoma. (g), UMAP diagram analyzing the status changes of fibroblasts in different stages of pancreatic adenocarcinoma. (h), Bar graph demonstrating changes in the proportion of fibroblasts in the progression of pancreatic adenocarcinoma. (i), UMAP diagram analyzing the status changes of stellate cells in different stages of pancreatic adenocarcinoma. (j), Bar graph demonstrating changes in the proportion of stellate cells in the progression of pancreatic adenocarcinoma.

The proportion of TECs in pancreatic adenocarcinoma with no relation to overall survival

During the progression of pancreatic adenocarcinoma, the status of endothelial cells underwent dramatically altered (Fig. 2a). However, based on the subsequent investigation utilizing TCGA database, we found that the proportion of endothelial cells was not related to overall survival (OS) of pancreatic adenocarcinoma patients. The p-value for the survival analysis is 0.18, indicating a lack of statistical significance (Fig. S2).

Fig. 2.

The transformation from normal endothelial cells to TECs. (a), Heatmap showing differences in marker genes of expression between pancreatic normal endothelial cells and TECs. (b,c), Pseudochronological analysis revealing transition trajectories from normal endothelial cells to TECs. (d,e), UMAP plots showing the expression of marker genes in normal endothelial cells and TECs. The enrichment analysis of the top 100 differential genes in normal endothelial cells (f) and TECs (g) were performed using the Metascape website.

The transformation from normal endothelial cells to TECs

The new vessel formation is a critical step in tumorigenesis, in which normal endothelial cells are recruited and transformed to TECs¹⁰. Although various origins of TECs may exist, including bone marrow-derived progenitor cells and neoplastic cells, the transformation from normal endothelial cells is still an important source of TECs generation²⁸.

Based on our analysis, the transcriptional profiles of normal endothelial cell and TECs are distinct. Representative marker genes of pancreatic normal endothelial cells and TECs were shown in Fig. 2a. As a dynamic process, the transformation from normal endothelial cells to TECs follows one trajectory which originated from normal endothelial cells (Fig. 2b, c).

During the transformation, the expression of TECs marker genes were gradually elevated (Fig. 2d, e). The marker genes of TECs including Insulin Like Growth Factor Binding Protein 3 (IGFBP3), Secreted Phosphoprotein 1 (SPP1), Complement Factor H (CFH), Immunoglobulin Lambda Like Polypeptide 5 (IGLL5), and TIMP Metallopeptidase Inhibitor 1 (TIMP1). The function of these markers was related to cell proliferation, oxidative stress, complement activation, antigen binding, and degradation of the extracellular matrix, which were critical for the tumor formation.

Instead, the expression of normal endothelial markers was gradually decreased (Fig. 2d, e). The marker genes of normal endothelial cells including Colipase (CLPS), Serine Protease 1 (PRSS1), Chymotrypsinogen B1 (CTRB1), Carbonic Anhydrase 4 (CA4), and Chymotrypsin Like Elastase 3 A (CELA3A). The enzymes CLPS, PRSS1, CTRB1, CA4, and CELA3A expresses in acinar cells, ductal cell type 1 cells, endocrine cells, and endothelial cells (Fig. S3). However, their expression is notably limited in other cell types. The probability of the observed expression in endothelial cells being attributed to “specific contamination” is minimal, because this “contamination” did not happen in other cell types. It indicated that the presence of these enzymes in endothelial cells is likely biologically relevant, not a contamination artifact.

There is growing consensus that loss of normal digestion-related function and emergence of tumor-related function in TECs are the critical events in tumorigenesis of pancreatic adenocarcinoma²⁹. Our analysis is also consistent with these studies. The gene functional enrichment analysis of the top 100 marker genes derived from normal endothelial cells or TECs indicated the functional shift from normal digestion-related signatures to tumor-related signatures (Fig. 2f, g). The most important signatures of normal endothelial cells are pancreatic secretion, digestion and absorption, etc. Instead, the most significant signatures of TECs are blood vessel formation, extracellular matrix degradation, insulin-like growth factor regulation, and wound healing et al..

RASIP1-positive TECs: a population correlated with increased overall survival

The most significant prognosis genes identified using the online bioinformatic analysis tool GEPIA were compared with the top 500 marker genes of TECs (Fig. 3a). The overlapped seven marker genes included SOCS2, RASIP1, EFNB2, MCF2L, S100A16, NCK1, and MPZL2. Based on the scRNA-seq data of pancreatic adenocarcinoma, RASIP1 was chosen for the following analysis due to the highly specificity of its expression in endothelial cells (Fig. 3b, c).

Fig. 3.

Further analysis of RASIP1-positive TECs. (a), Venn diagram for the identification of differential genes. Both violin plots (b), and UMAP plot (c) demonstrating specific expression of RASIP1 in endothelial cells. (d), KM survival curve showing the relationship between RASIP1 expression and overall survival in patients with pancreatic adenocarcinoma. The violin plots respectively showing the expression of RASIP1 in normal pancreas and different stages of pancreatic adenocarcinoma based on the data of scRNA-seq (e) and TCGA database (f). (g), The volcano plot showing the differential expression of specific genes in the up-regulated and down-regulated genes of RASIP1-positive TECs. The enrichment analysis of the down-regulated (h) and up-regulated (i) genes of RASIP1-positive TECs were performed using the Metascape website.

The expression of RASIP1 was related to the OS of pancreatic adenocarcinoma patients (Fig. 3d). However, RASIP1 gene expression in normal pancreas and different stages of pancreatic adenocarcinoma were remained similar (Fig. 3e, f). The p-values in ANOVA analysis based on the data of scRNA-seq and TCGA database were both more than 0.05, indicating no statistically significant.

A potential function of RASIP1-positive TECs: leukocyte transendothelial migration

The up-regulated and down-regulated genes of RASIP1-positive TECs in scRNA-seq data of pancreatic adenocarcinoma were identified (Fig. 3g), using gene functional enrichment analysis. Intriguingly, in the term derived from the up-regulated genes of RASIP1-positive TECs, one of the most significant terms is leukocyte transendothelial migration (Fig. 3h and i and 4a)^30–32. Transendothelial migration of T cells, particular effector T cells, may contribute to the improved responses for immunotherapy by increasing effector T cells infiltration.

Fig. 4.

Exploration of potential functions of RASIP1-positive TECs. (a), KEGG diagram showing the pathway of leukocyte transendothelial migration (Derived from Kanehisa laboratories and used with permission). (b), Violin plots demonstrating the expression of key functional markers of leukocyte transendothelial migration in RASIP1-positive TECs. (c), Scatterplots showing genetic correlation analysis based on TCGA database. (d), Scatterplots showing the expression of RASIP1 associated with different T cells gene signatures in pancreatic adenocarcinoma. * P < 0.05, ** P < 0.01, *** P < 0.001, ns: The difference is not significant.

Moreover, it was found that critical functional markers of leukocyte transendothelial migration were with elevated expression level in RASIP1-positive TECs (Fig. 4b). Based on scRNA-seq analysis of pancreatic adenocarcinoma, ICAM1, intercellular cell adhesion molecule-2 (ICAM2), Cadherin 5 (CDH5), and Junctional Adhesion Molecule 3 (JAM3) were with increased in RASIP1-positive TECs. Furthermore, based on the correlation analysis using TCGA database (Fig. 4c), expression of RASIP1 was significantly correlated with ICAM1, ICAM2, CDH5, and JAM3. Another molecular related to leukocyte transendothelial migration, VCAM1, failed to show elevated expression in RASIP1-positive TECs and no significant correlation with RASIP1 in TCGA database.

RASIP1-positive TECs in pancreatic adenocarcinoma: making “cold” tumors “hot”?

Based on the correlation analysis relying on the public RNA-seq data of pancreatic adenocarcinoma in TCGA database, expression of RASIP1 was significantly correlated with T cells signatures, where included Native T cell, Exhausted T-cell, effector memory T-cell and Th1-like, etc. (Fig. 4d). It may suggest that there were more cellular immune responses in pancreatic adenocarcinoma patients with high RASIP1 expression. Moreover, by analyzing a public spatial transcriptomics data of pancreatic adenocarcinoma, the expression of RASIP1 was co-localized with T-effector and IFN-gamma gene signatures which were the marker of immunogenic “hot” tumors (Fig. 5a, b, c, d, Fig S5, Fig S7). Based on TCGA database, the expression of RASIP1 was also correlated with T-effector and IFN-gamma gene signatures, as well as critical markers for immunotherapy CD8 and Programmed cell death protein 1 (PDCD1) (Fig. 5e, f, g). Intriguingly, we further studied the survival analysis based on patients with pancancer immunotherapy, and we found that the survival time of the RASIP1 high expression group was significantly prolonged, with a p-value of 0.0078 and a HR of 0.73, in patients receiving PD-L1 treatment (Fig. S4).

Fig. 5.

Analysis of immune function of RASIP1-positive TECs. (a,b,c,d), Spatial transcriptome maps revealing expression profiles of RASIP1 and T-effector and IFN-gamma gene signatures. The scatter plots verifying the co-expression of RASIP1 and T-effector and IFN-gamma gene signatures (e), CD8 (f), and PDCD1 (g).

RASIP1-positive TECs exhibited an increased propensity for cell-cell interaction with adaptive immune system

Next, the secreted signaling in scRNA-seq data was analyzed. RASIP1-positive TECs exhibited a more pronounced interaction with T and B cells than RASIP1-negative TECs (Fig. S6). It was shown that the secretion of macrophage migration inhibitory factor (MIF) by RASIP1-positive TECs played a crucial role in mediating their interaction with CD4 + T cells, CD8 + T cells, and B cells. On the other side, the surface receptors CD74, CXCR4 and CD44 were found to bind to MIF on the immune cells, which potentially contributes to the maintenance of the interaction between RASIP1-positive TECs and adaptive immune system.

In terms of the communication from adaptive immune system to endothelial cells, RASIP1-positive TECs were shown to have a closer interaction with the adaptive immune system compared to RASIP1-negative TECs (Fig. S6). CD40 ligand (CD40LG) and secreted phosphoprotein 1 (SPP1) were identified as key mediators of these interactions.

Positive correlation of RASIP1-positive TECs with enhanced tumor immune cell infiltration

To further validate the interaction between RASIP1-positive TECs and immune cells, we collected pancreatic ductal adenocarcinoma tissue samples from three patients undergoing pancreatic surgery and performed multiplex-staining for endothelial cells (CD31+), T lymphocytes (CD3+), cytotoxic T lymphocytes (CD8+), activated cytotoxic T cells/natural killer cells (Granzyme B+), and RASIP1-positive TECs (RASIP1+), among others, on PDAC tissue sections (Fig. 6a).

Fig. 6.

Expression of CD31, RASIP1, CD3, CD8 and Granzyme B in sections of Pancreatic Ductal Adenocarcinoma tissue detected by mIHC. (a), Mono- and pan-chromatic mIHC profile of PDAC tissue.The upper large image shows a merged multispectral fluorescence from CD31, RASIP1, CD3, CD8, Granzyme B and DAPI in three samples. The following six rows of images display representative blood vessels (identified by CD31 + endothelial cells) and immunostaining of RASIP1, CD3, CD8, and Granzyme B in and around blood vessels. RASIP1 immunostaining reflects its expression in vascular endothelial cells, while CD3, CD8, and Granzyme B mark immune cells within or adjacent to vessels. (b), Number of blood vessels stratified by the proportion of RASIP1-positive endothelial cells (-, +, ++) and categorized by the density of CD3+, CD8+, or Granzyme B+ immune cells (-, +, ++) within or adjacent to vessels.

Tumor-associated vasculature was identified by CD31 + labeling of endothelial cell. Strikingly, regions of high-RASIP1 expression (RASIP1++) vasculature demonstrated marked perivascular enrichment of CD3+, CD8+, and Granzyme B+ immune cells, in contrast to low-RASIP1 vasculature (RASIP1−), which exhibited immune desertification phenotypes characterized by minimal lymphocyte infiltration (Fig. 6a). Systematic quantification of 1,381 microvessels across three histopathological sections uncovered a RASIP1 expression-dependent gradient in immune cell aggregation densities. This trend was most prominent for Granzyme B+ immune cells: among vessels formed by low-RASIP1-expressing TECs (−),​those with low Granzyme B immune cell aggregation (−) accounted for the majority. Among vessels with high-RASIP1 expression (++),​those with high Granzyme B immune cell aggregation (++) predominated (Fig. 6b). These findings suggest that RASIP1-positive TECs may regulate the infiltration of CD4+/CD8+/GzmB + T lymphocytes into PDAC tumors through immune cell recruitment or vascular priming.

In vitro model construction to verify the function of RASIP1-positive TECs

To investigate the potential pathway and verify the function of RASIP1-positive TECs, in vitro models using human umbilical vein endothelial cells (HUVECs) were utilized. RASIP1 was knocked down using siRNA, resulting in a significant reduction of the protein levels of RASIP1 in HUVECs (Fig. 7). HUVECs were transfected with either a plasmid vector overexpressing RASIP1 or an empty vector. Increased protein levels of RASIP1 were observed in RASIP1-overexpressed HUVECs (Fig. 8). In order to further verify the function of RASIP1-positive endothelial cells, we performed high-throughput RNA sequencing to identify differentially expressed genes in RASIP1-knockdown HUVECs compared with control samples.

Fig. 7.

In vitro model construction to verify the function of RASIP1-positive TECs. (a), Western blotting analysis assessing the protein levels of RASIP1 and RHOA in HUVECs of negative control (NC) and RASIP1 knockdown samples (SiRASIP1). (b), Western blotting analysis assessing the protein levels of VCAM1 and ICAM1 in HUVECs of NC and SiRASIP1. (c), Western blotting analysis assessing the protein levels of ERK, p-ERK and p38 in HUVECs of NC and SiRASIP1. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, ns: The difference is not.

Fig. 8.

In vitro overexpression model construction to verify the function of RASIP1-positive TECs. (a), Western blotting analysis assessing the protein levels of RASIP1 and RHOA in HUVECs of RASIP1 overexpression(OE_RASIP1) and control(Vector) samples. (b), Western blotting analysis assessing the protein levels of VCAM1 and ICAM1 in HUVECs of OE_RASIP1 and Vector. (c), Western blotting analysis assessing the protein levels of p-ERK, ERK and p38 in HUVECs of OE_RASIP1 and Vector. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, ns: The difference is not significant.

RNA-seq analysis identified 92 upregulated and 55 downregulated genes (Fig. 9a). Metascape analysis revealed that RASIP1 knockdown downregulated cytokine-mediated signaling, T cell activation, and leukocyte migration regulation, while upregulated pathways included negative regulation of immune effector processes and estrogen metabolism (Figs. 9b, c). These findings align with our hypothesis. The top three gene sets of upregulated and downregulated pathways and the genes of our interest were selected for further validation. Consistent with sequencing data, RASIP1 knockdown decreased ICAM1, CDH5, and JAM3 expression but increased RHOA expression, while VCAM1 and ICAM2 remained unchanged (Fig. 9d).

Fig. 9.

High-throughput RNA-sequencing data analysis. (a), The volcano plot showing the differential expression of specific genes in the up-regulated and down-regulated genes in HUVECs with RASIP1 knockdown and samples of control HUVECs. The enrichment analysis of the down-regulated (b) and up-regulated (c) genes of SiRASIP1 and NC were performed using the Metascape website. (d), The heatmap showing differences in the hub genes of expression between HUVECs with RASIP1 knockdown and samples of control HUVECs.

Western blotting showed that RHOA increased when RASIP1 expression was reduced (Fig. 7a). However, overexpressing RASIP1 did not affect RHOA expression levels (Fig. 8a). As previously studied, RASIP1 inhibited RHOA signaling by interacting with RhoGAP, which was involved in endothelial cell morphogenesis and angiogenesis.³³ Additionally previous studies reported that ICAM1 and VCAM1 were important adhesion factors involved in endothelial cell-leukocyte adhesion³⁴, and their increased expression was conducive to T cell infiltration.³⁵ Therefore, we subsequently measured the expression levels of ICAM1 and VCAM1. We found that RASIP1 knockdown led to a decrease in ICAM1 expression, whereas VCAM1 expression remained unchanged (Fig. 7b). Conversely, RASIP1 overexpression resulted in increased ICAM1 expression, with VCAM1 expression again showing no significant changes (Fig. 8b). To further explore potential pathways by which RASIP1-positive TECs maintain their status, we validated the protein levels of ERK, p-ERK, p38. We found that the protein levels of ERK were also reduced upon transfection with RASIP1 siRNA, but the protein levels of p-ERK were the opposite (Fig. 7c), whereas overexpression had the opposite effect. However, the protein levels of p38 remained unchanged following either RASIP1 knockdown or overexpression (Figs. 7c and 8c). These results suggest that RASIP1 could maintain TEC status via ERK pathway regulation, independent of p38 signaling.

Discussion

Ras interacting protein 1 (RASIP1) is a GTPase-binding protein that has been shown to be critical for blood vessel development and angiogenesis³⁶. As an endothelial-specific protein, it not only regulates the movement of endothelial cells, but also participates in the development of cancer cells³⁷. In non-small cell lung cancer, it is widely expressed and promotes the migration of cancer cells²⁵. In diffuse large B-cell lymphoma (DLBCL), overexpression of RASIP1 promotes proliferation, invasion, and tumorigenesis of DLBCL cells³⁸. In our study, we found that pancreatic adenocarcinoma patients with high endothelial expression of RASIP1 exhibited higher expression of adhesion molecules, more T cell infiltrations, and were associated with better survival outcomes.

Although RASIP1 mRNA and protein expression levels remain relatively stable across different pathological stages of pancreatic ductal adenocarcinoma (PDAC), high RASIP1 expression is strongly associated with improved overall survival. This apparent paradox can be explained by the fact that the prognostic impact of RASIP1 is not primarily driven by its absolute abundance, but rather by its role in maintaining the structural and functional integrity of tumor-associated endothelial cells. In other words, high RASIP1 expression per se is not sufficient to promote immune cell infiltration; rather, it is the formation of a specific subset of RASIP1-high, structurally mature, and functionally competent tumor blood vessels that creates an immune-permissive vascular niche, thereby facilitating effective leukocyte adhesion, extravasation, and perivascular accumulation.

Despite the fact that immunotherapies, including ICIs, have proven to be a promising and breakthrough treatment modality for a variety of solid tumors, pancreatic adenocarcinoma has remained a difficult area due to its poor response to immunotherapy. This is mostly due to the tumor’s low immune cell infiltration, especially T cells infiltration, which has been a significant obstacle in achieving successful treatment outcomes. The hypo vascularity in pancreatic adenocarcinoma that we discovered in our study was consistent with previous research. For vascularized tumors, high vascularity is associated with poor prognosis, whereas for avascular tumors, a steady increase in vascularity favors the provision of anticancer immune cells and drug delivery^27,39. Studies have shown that therapies that improve rather than inhibit vascular function may provide a potential therapeutic approach for pancreatic adenocarcinoma⁴⁰. Numerous studies have been reported to address the challenge of poor response to immunotherapy caused by this hypo vascularity. One therapeutic strategy that developed in recent years is the use of cellular immunotherapy. This approach involves the use of modified immune cells, such as chimeric antigen receptor (CAR) T cells, natural killer cells, and tumor-infiltrating lymphocytes, to target and eliminate pancreatic cancer cells⁴¹. In a preclinical study, researchers found that inducing senescence through the retinoblastoma (RB) protein could increase vascularization in pancreatic tumors, leading to improved drug delivery during treatment⁴². Targeting RASIP1-positive TECs could be another effective method to improve the effectiveness of immunotherapy in pancreatic adenocarcinoma. However, despite the potential therapeutic significance of RASIP1-positive TECs, this target has been overlooked previously.

We uncovered a novel cell type, RASIP1-positive TECs, in this study, which provides a potential effective therapeutic mechanism. Intriguingly, although various types of sequencing data were used in our study, including scRNA-seq, bulk RNA-seq in TCGA database, and spatial transcriptomics, the analysis based on these data consistently shown that RASIP1-positive TECs was a factor correlated with “hot” tumors of pancreatic adenocarcinoma. Our study further suggested that RASIP1-positive TECs show closer interactions with the adaptive immune system than RASIP1-negative TECs. As previously reported⁴³, higher levels of CD4 + and CD8 + tumor infiltrating T cells were associated with better prognosis in pancreatic cancer patients after surgery. Therapies that promote T cell priming/activation are considered as one of the strategies to potentiate the efficacy of immunotherapy.⁴⁴.

Moreover, we also established an in vitro model using HUVECs. We further verified the elevated expression of RHOA after knockdown RASIP1. As previously studied, RASIP1 inhibits RHOA, a known inhibitor of lumen formation³³. RASIP1 participates in angiogenesis and morphological maintenance by regulating Rho GTPase signaling⁴⁵, providing new insights for future vascular therapy in pancreatic adenocarcinoma. What’s more, we found that RASIP1 mainly affects the expression of ICAM1 but not VCAM1. ICAM1 is involved in the recruitment and retention of immune cells, the initiation of inflammatory response in endothelial cells⁴⁶, and may be involved in anti-tumor immunity. Finally, we observed that after knocking down RASIP1, the protein level of ERK decreased, which was partly consistent with previous studies⁴⁷. RASIP1 may promote the formation of blood vessels and maintain the state of RASIP1-positive TECs through the ERK pathway. However, the elevated ERK phosphorylation levels may involve more complex regulatory mechanisms. It is still important to acknowledge the limitations of our findings, and further preclinical investigation based on in vitro and in vivo models are needed to confirm the therapeutic value derived from RASIP1-positive TECs. Ideally, to clarify the function of this novel type of cells, the RASIP1-positive TECs should be isolated from clinical tissues and co-cultured with immune cells for several functional assays including cell migration assay based on transwells. The in vitro model base on nude mice with manipulation of tumor tissue with injection of RASIP1-positive TECs is also necessary which could directly illustrate the altered immunoevasion propensity. These findings are expected to be further validated through future research.

In our study, all omics-based analyses are correlative, including scRNA-seq, TCGA bulk RNA-seq, and spatial transcriptomics. Multiplex immunofluorescence (mIHC) and Western blot provide preliminary mechanistic. However, it lacks direct causal validation. Due to technical challenges in isolating stable primary RASIP1 + TECs, advanced co-culture experiments and in vivo functional studies could not be performed. Future work should utilize PDAC mouse models with endothelial-specific RASIP1 manipulation, combined with anti-PD-1/PD-L1 therapy, to establish causality and evaluate its impact on T cell recruitment, tumor growth, and immunotherapy response.

Conclusions

Collectively, our study revealed a novel endothelial cell subtype in pancreatic adenocarcinoma. Through integrative analysis of public single-cell RNA sequencing datasets, TCGA bulk transcriptomic profiles, and spatial transcriptomic data, we demonstrated that RASIP1-positive tumor endothelial cells (TECs) were significantly associated with increased T-cell infiltration and elevated effector T-cell–related and interferon-γ signaling gene signatures. Multiplex immunofluorescence validation further confirmed the spatial colocalization of RASIP1 + TECs with CD8 + cytotoxic T lymphocytes (CTLs) at the tumor–vessel interfaces. Importantly, clinical cohort analyses further suggested ​that RASIP1 + TECs may predict​ improved immunotherapy response and ​potentiated adaptive immune activation. Mechanistically, our findings propose that RASIP1 sustains the functional phenotype of these TECs via ERK pathway modulation. Our investigations on the newly identified RASIP1-positive tumor endothelial cells (TECs) may ultimately enhance immunotherapy efficacy in pancreatic cancer, or at minimum, these findings lay the foundation for developing RASIP1-targeted therapeutic strategies against this malignancy.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

SBZ and WQ designed this study. SBZ, YJH and HC were responsible for data collection. SBZ, YJH and WQ wrote this manuscript. YYT, XW and ZJF reviewed and polished this research. ZJF, WQ, JHL supervised and contributed to the critical reading of manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by Hebei Natural Science Foundation (H2021206314).

Data availability

All data supporting the findings of this study are available within the paper.

Declarations

Competing interests

The authors declare no competing interests.