CXCL11 recruits immunosuppressive cells to form a barrier at the invasive front of pancreatic cancer by activating the NF-κB/CCL2 axis

Abstract

Background

Pancreatic ductal adenocarcinoma is a malignancy characterized by profound immunosuppression and universal resistance to immunotherapy. The complex tumor microenvironment is a critical determinant of therapeutic failure. The chemokine C-X-C motif chemokine ligand 11 (CXCL11) exerts dual immunomodulatory roles in various tumors; however, its expression pattern, clinical significance, and molecular mechanisms in pancreatic cancer remain unclear.

Methods

This study integrated clinical cohort analysis, tissue microarray, transcriptomic sequencing, and in vitro functional experiments. Immunohistochemistry and multiplex immunofluorescence were used to analyze CXCL11 expression and its spatial relationship with immune cell infiltration. Transcriptomic sequencing was performed to identify differentially expressed genes after CXCL11 knockdown, followed by pathway enrichment analysis. CXCL11 knockdown and overexpression models were established in Panc02 and SW1990 cell lines, and western blotting was used to verify the regulatory effects on the nuclear factor kappa-B (NF-κB) signaling pathway and downstream C—C motif chemokine ligand 2 (CCL2) expression.

Results

CXCL11 expression was higher in paratumor tissues than in tumor tissues of pancreatic cancer, and its high expression was significantly associated with shortened overall survival and more aggressive pathological features. Co-enrichement of CXCL11 with immunosuppressive markers such as CD163, CD11b, and programmed death-ligand 1 in paratumoral regions suggests its involvement in forming a spatially heterogeneous immunosuppressive barrier. Tissue-level analysis revealed a significant positive correlation between CXCL11 expression and the M2 macrophage marker CD163, as well as the myeloid-derived suppressor cell marker CD11b. Transcriptomic and functional experiments confirmed that CXCL11 activates the NF-κB signaling pathway and upregulates CCL2 expression. Tissue microarray further validated their positive correlations at the protein level.

Conclusions

CXCL11 plays a pro-tumorigenic role in pancreatic cancer, with high expression levels predicting poor prognosis. By activating the NF-κB signaling pathway to upregulate CCL2, CXCL11 drives the "CXCL11–NF-κB–CCL2″ signaling axis, which, together with its enrichment in paratumoral regions, promotes the recruitment and activation of immunosuppressive cells and forms a functional immunosuppressive barrier at the invasive front. This provides a mechanistic explanation for immunotherapeutic resistance in pancreatic cancer and suggests CXCL11 and its related pathways as potential therapeutic targets.

Introduction

Pancreatic cancer has the sixth highest incidence and the third highest mortality rate among malignant tumors, representing one of the most prognostically unfavorable malignancies, with approximately 90% of cases being pancreatic ductal adenocarcinoma (PDAC) [1]. Although immunotherapy has achieved remarkable efficacy and improved patient survival in various solid tumors [2,3], the response to immunotherapy in patients with PDAC remains generally poor, and the clinical benefits are far from expected [4,5]. This therapeutic resistance highlights the limitations of current treatment strategies in PDAC and underscores the urgent need to characterize its immune microenvironment and resistance mechanisms, thereby identifying effective approaches to overcome current therapeutic bottlenecks.

The immune microenvironment of PDAC is characterized by profound immunosuppression and is typically described as an immune "cold tumor" [[6], [7], [8]]. Its immune inertness is primarily reflected by the low immunogenicity of tumor cells, insufficient infiltration of functional immune cells into the tumor microenvironment, an abundance of various immunosuppressive cells, and encapsulation within a dense, rigid extracellular matrix [9,10]. Therefore, reversing the "cold tumor" phenotype of PDAC and converting it into an immunologically active "hot tumor" has become a central scientific concern in overcoming immunotherapeutic resistance in PDAC.

CXCL11 is a small interferon‑inducible protein that, by binding to its receptor C-X-C chemokine receptor type 3 (CXCR3), recruits and activates immune cells, such as T and B lymphocytes, thereby playing a key role in tumor immunity [11,12]. Within the tumor microenvironment, CXCL11 expression is upregulated and associated with immunotherapy‑related biomarkers. However, its function is dual: it can mediate antitumor immunity or promote immune escape under specific conditions. High CXCL11 expression enhances the infiltration of CD8+ T cells and other immune cells, activates antitumor immune responses, and correlates with a favorable patient prognosis in colorectal cancer [13], non‑small cell lung cancer [14], glioblastoma [15], and ovarian cancer [16]. In contrast, CXCL11 may promote tumor progression by upregulating the immune checkpoint genes cytotoxic T-lymphocyte-associated protein 4 (CTLA‑4) and programmed death-ligand 1 (PD‑L1) in breast cancer [11], prostate cancer [17], and cutaneous melanoma [18], thereby remodeling the immunosuppressive microenvironment. The function of CXCL11 is subject to complex regulation by CXCR3 splice variants (pro‑tumorigenic CXCR3A and anti‑tumorigenic CXCR3B) and interactions with other chemokines [19], with its net effect depending on cell type and the activation status of downstream signaling pathways.

The expression pattern, clinical significance, and mechanism by which CXCL11 regulates the immune microenvironment in pancreatic cancer remain unclear. This study systematically investigates CXCL11 expression in pancreatic cancer and its relationship with patient prognosis through analysis of pancreatic cancer tissue microarrays, transcriptomic sequencing, and cellular functional experiments. Furthermore, it elucidates the molecular mechanism by which CXCL11 remodels the immunosuppressive microenvironment by activating the NF‑κB signaling pathway and regulating the chemokine network, thereby providing new targets for prognostic assessment and immunotherapeutic strategies in pancreatic cancer.

Materials and methods

Human pancreatic cancer tissues

This study incorporated 129 histopathologically confirmed pancreatic cancer tissue specimens. A tissue microarray (HPanA125Su01) was purchased from Shanghai Outdo Biotech Co., Ltd., containing 81 pancreatic cancer and paratumor tissue samples with their corresponding clinicopathological features and survival data. The array samples originated from pathological tissue specimens of patients diagnosed with pancreatic cancer between September 2004 and December 2008, with patient follow‑up ranging from 0 to 81 months and a median survival of 11 months. The cohort comprised 13 patients with Grade I, 59 with Grade II, and 9 with Grade III disease; TNM staging revealed 32 patients with Stage I, 46 with Stage II, and one with Stage IV disease, and two with unknown stage. Another tissue microarray (ZL‑PanA961) was purchased from Shanghai Zhuoli Biotechnology Co., Ltd., containing 48 pancreatic cancer specimens without prognostic information.

Immunohistochemistry (IHC)

Paraffin‑embedded tissue sections were dewaxed in xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed using ethylenediaminetetraacetic acid buffer (pH 9.0) in a microwave oven to expose epitopes. Sections were then washed with phosphate buffered saline (PBS; pH 7.4) and incubated with 3% hydrogen peroxide solution at 20 °C to block endogenous peroxidase activity. Subsequently, 3% bovine serum albumin (BSA) was applied at room temperature for blocking. Following removal of the blocking solution, diluted primary antibodies against CXCL11 (Proteintech, 10,707‑1‑AP, 1:150) and CCL2 (Huabio, EM1710‑22, 1:200) were added, and sections were incubated overnight at 4 °C in a humidified chamber. The next day, sections were thoroughly washed with PBS, incubated with appropriate horse radish peroxidase (HRP)‑conjugated secondary antibodies for 1 hour at room temperature, and then washed again. Freshly prepared 3,3′-diaminobenzidine (DAB) substrate solution was applied until specific positive signals developed a brown‑yellow color, followed by counterstaining with hematoxylin, differentiation in hydrochloric acid‑ethanol, and bluing in water. Sections were then dehydrated through graded ethanol, cleared in xylene, and mounted with neutral resin. Nuclei appeared blue in the staining results, whereas positive expression of the target protein (localized to the cytoplasm or membrane) appeared brown‑yellow. All section images were acquired and analyzed using KFbio (KF‑FL‑020) and 3D Histech (Pannoramic MIDI) scanners. Images were imported into QuPath software (v0.5.0) for analysis. Tissue core positions on each microarray were automatically identified using the software's built‑in tissue microarray array tool and manually reviewed and corrected. Cell segmentation and feature extraction were performed using the Positive Cell Detection command: the Hematoxylin optical density (OD) channel was selected for detection to ensure that nuclear identification was based on hematoxylin signal; the Intensity features module was used to extract the mean optical density value of the DAB channel for each cell (Cell: DAB OD mean), which directly reflects the expression intensity of the target protein at the cellular level. The mean DAB OD of all cells within each tissue microarray core was used as the final quantitative value for that sample.

Survival analysis

Based on immunohistochemical quantification results, survival analysis for CXCL11 was performed using the "ggsurvplot" package in R software. The optimal cutoff method was used to divide samples into high‑ and low‑expression groups, and Kaplan–Meier survival curves were plotted, with the cutoff value calculated using the "surv_cutpoint" function in the "survminer" package. The optimal DAB cutoff value for CXCL11 was 0.0082, dividing samples into low‑ (24 cases) and high‑expression (57 cases) groups. Univariate Cox regression analysis was performed by integrating CXCL11 expression levels with clinical characteristics of patients with pancreatic cancer to identify prognostically significant variables, and a nomogram was constructed using the "rms" package in R. This nomogram graphically illustrates the influence of these variables on individual patient prognosis and predicts 1-, 2-, and 3-year survival rates, providing a reference model for clinical prognostic assessment.

Multiplex immunofluorescence

A five‑color multiplex immunofluorescence staining method based on tyramide signal amplification (TSA) technology was used. Paraffin sections were dewaxed, rehydrated, and subjected to antigen retrieval, followed by blocking of endogenous peroxidase with 3% hydrogen peroxide and blocking with 10% rabbit serum or 3% BSA at room temperature. Subsequently, five target proteins were stained using a sequential cyclic labeling procedure: rabbit anti‑human CXCL11 primary antibody (Proteintech, 10,707‑1‑AP, 1:50), Panck (Abcam, ab80826, 1:400), CD163 (Youmeng Bio, YM‑B20025, 1:200), PD‑L1 (Youmeng Bio, YM‑B20024, 1:500), and CD11b (Youmeng Bio, YM‑B20023, 1:500). Each labeling cycle comprised overnight incubation with primary antibody at 4 °C, incubation with HRP‑conjugated secondary antibody at room temperature, incubation with the corresponding TSA fluorophore, and subsequent antibody stripping to remove non‑covalently bound antibodies from the previous cycle. After completion of all cycles, nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI), and sections were treated with an autofluorescence quenching reagent before mounting. Finally, multichannel fluorescence images were acquired using a fluorescence microscope and a whole‑slide scanning system. Multiplex immunofluorescence sections were imported into HALO software (Indica Labs, USA) for quantitative analysis. Regions of interest were manually delineated within the target tissues. The positive cell counting module was selected, and fluorescence signals for each channel were manually selected and repeatedly verified to ensure accurate identification of positive signals. Based on DAPI fluorescence signals, the software localized cell nuclei and extended to the cytoplasmic region, calculating parameters including positive cell count, total cell count, and tissue area to derive positive cell number, positive cell ratio (positive cells/total cells × 100%), and positive cell density (positive cells/tissue area, cells/mm²). All analysis parameters were kept constant within the same batch of samples.

Transcriptomics

This study used reference‑based transcriptomic sequencing to analyze gene expression. After quality assessment of total RNA using a Nanodrop and an Agilent 2100 Bioanalyzer, strand‑specific libraries were constructed using the NEBNext Ultra II RNA Library Prep Kit for Illumina. Procedures included enrichment of mRNA using Oligo(dT) magnetic beads, fragmentation, double‑stranded cDNA synthesis, end repair, adapter ligation, and polymerase chain reaction (PCR) amplification. After quality assessment of the constructed libraries using an Agilent 2100 and quantitative PCR, paired‑end 150 bp (PE150) sequencing was performed on an Illumina platform. Raw sequencing data were processed using Cutadapt to remove adapters and filter low‑quality reads (Q < 20), yielding high‑quality data for subsequent analysis. Clean reads were aligned to the reference genome using HISAT2 v2.0.5, and transcript assembly was performed using StringTie to identify novel transcripts. Gene expression levels were quantified using HTSeq for read counts and normalized as fragments per kilobase of transcript per million mapped reads. Differentially expressed genes were screened using DESeq2 with thresholds of |log2FoldChange| > 1 and Q < 0.05. Finally, Gene Ontology (GO) functional and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of differentially expressed genes were performed using the hypergeometric test (P < 0.05).

Cell culture and plasmid transfection

Mouse pancreatic cancer Panc02 cells and human pancreatic cancer SW1990 cells were routinely cultured in RPMI 1640 medium (Panc02) or Dulbecco's Modified Eagle Medium (SW1990) supplemented with 10% fetal bovine serum and 1% penicillin‑streptomycin at 37 °C in a 5% CO₂ incubator. When cells reached 80–90% confluence, they were passaged using trypsin digestion. Three short hairpin RNA (shRNA) oligonucleotide sequences targeting the mouse CXCL11 gene (transcript NM_019494.1), named CXCL11‑i1, CXCL11‑i2, and CXCL11‑i3, were designed and synthesized to knock down the CXCL11 gene in Panc02 cells. The specific sequences were: CXCL11‑i1 (F: gatccGCTGCTCAAGGCTTCCTTATGtcaagagCATAAGGAAGCCTTGAGCAGCtttttt; R: aattaaaaaaGCTGCTCAAGGCTTCCTTATGctcttgaCATAAGGAAGCCTTGAGCAGCg); CXCL11‑i2 (F: gatccGCAGAGATCGAGAAAGCTTCTtcaagagAGAAGCTTTCTCGATCTCTGCtttttt; R: aattaaaaaaGCAGAGATCGAGAAAGCTTCTctcttgaAGAAGCTTTCTCGATCTCTGCg); CXCL11‑i3 (F: gatccGCCTCATAATGCAGGCAATAGtcaagagCTATTGCCTGCATTATGAGGCtttttt; R: aattaaaaaaGCCTCATAATGCAGGCAATAGctcttgaCTATTGCCTGCATT

ATGAGGCg). The shRNA plasmids were transiently transfected into Panc02 cells using Lipo8000 transfection reagent (Beyotime, C0533) to achieve CXCL11 knockdown. For the overexpression of CXCL11 in human SW1990 cells, an overexpression plasmid containing the human CXCL11 coding sequence (NM_005409.4, Miaoling Bio, Cat# P56352) was transfected into SW1990 cells, and CXCL11‑overexpressing stable cell lines were obtained after selection.

Western blotting

Western blotting was performed to detect protein expression levels of CXCL11, P65, and CCL2 in Panc02 and SW1990 cells. Briefly, cells with CXCL11 knockdown (KD) or overexpression (Exp) were collected, and total protein was extracted using radioimmunoprecipitation assay lysis buffer containing protease and phosphatase inhibitors. Protein concentration was quantified using the bicinchoninic acid method. Equal amounts of protein samples were separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto polyvinylidene fluoride membranes. Membranes were blocked with a protein‑free rapid blocking solution (Boster, AR0041) for 1 hour at room temperature and then incubated with primary antibodies against CXCL11 (Proteintech, 10,707‑1‑AP, 1:2000), P65 (Huabio, ET1603‑12, 1:10,000), CCL2 (Huabio, EM1710‑22, 1:1000), and the loading control β‑actin (Huabio, HA722023, 1:10,000) overnight at 4 °C. The next day, after washing with tris-buffered saline with tween, membranes were incubated with secondary antibodies (Servicebio, GB23303, 1:10,000) for 1 hour at room temperature. Finally, signals were developed using enhanced chemiluminescence reagent (Boster, AR1191), and images were captured using a chemiluminescence imaging system.

Results

High CXCL11 expression is associated with poor prognosis in patients with pancreatic cancer

To determine the clinical significance of CXCL11 in pancreatic cancer, we first evaluated the association between its expression level, patient prognosis, and clinicopathological features. Using immunohistochemical detection on pancreatic cancer tissue microarrays, we observed that CXCL11 expression was higher in paratumor tissues than in tumor tissues (Fig. 1A). Survival analysis showed that patients in the high CXCL11-expression group had significantly shorter overall survival than those in the low expression group, indicating a strong association with poor prognosis (Fig. 1B, P < 0.05). Quantitative IHC results demonstrated that CXCL11 expression was significantly higher in paratumor tissues than in tumor tissues (Fig. 1C, P < 0.05). Furthermore, CXCL11 was associated with higher pathological grade (Table 1). A diagram further revealed an association between CXCL11 expression levels and aggressive tumor pathological features (Fig. 1D). Cox regression analysis revealed that N and TNM stages were associated with poor prognosis in patients with pancreatic cancer (Table 2). We integrated CXCL11 expression levels with key clinical parameters such as TNM stage to establish a nomogram and construct a clinical prediction model. This effectively predicted patient survival probabilities at 1, 2, and 3 years (Fig. 1E), and calibration curves showed that the predicted survival probabilities were consistent with actual observations (Figs. 1F–H). Collectively, these results indicate that CXCL11 is differentially expressed in pancreatic cancer tissues, and its high expression is significantly associated with poor patient prognosis and aggressive pathological features. This establishes it as a potential biomarker for prognostic assessment in pancreatic cancer.

Fig. 1.

CXCL11 is associated with poor prognosis in pancreatic cancer.

(A) Representative immunohistochemical images of C-X-C motif chemokine ligand 11 (CXCL11) in pancreatic cancer and paratumor tissues; (B) Kaplan–Meier survival curve based on CXCL11 expression levels; (C) Statistical analysis of CXCL11 expression differences between pancreatic cancer and paratumor tissues; (D) Sankey diagram showing the association between CXCL11 expression and clinicopathological features; (E) Nomogram integrating CXCL11 expression and clinical factors to predict the prognosis of patients with pancreatic cancer. The total score, obtained by summing the points of each clinical indicator, was used to predict patient survival. (F–H) Calibration curves comparing predicted 1-, 2-, and 3-year survival probabilities with observed values. The diagonal dashed line represents ideal prediction, and the blue line indicates the actual values observed at 1, 2, and 3 years.

Table 1.

Relationship between CXCL11 expression level (high vs. low) and clinicopathological characteristics in pancreatic cancer tissues.

Characteristics	CXCL11 low	CXCL11 high	P-value
n	24	57
Grade, n (%)			0.016
Ⅰ	8 (9.9%)	5 (6.2%)
Ⅱ	15 (18.5%)	44 (54.3%)
Ⅲ	1 (1.2%)	8 (9.9%)
Tumor size, median (IQR)	4.5 (3.55, 6.5)	4 (3, 5)	0.437
T, n (%)			0.578
T3	6 (7.5%)	10 (12.5%)
T2	17 (21.2%)	46 (57.5%)
T1	0 (0%)	1 (1.2%)
N, n (%)			0.979
N1	11 (14.5%)	24 (31.6%)
N0	13 (17.1%)	28 (36.8%)
M, n (%)			1.000
M0	24 (29.6%)	56 (69.1%)
M1	0 (0%)	1 (1.2%)
TNM stage, n (%)			1.000
Ⅰ	10 (12.7%)	22 (27.8%)
Ⅱ	14 (17.7%)	32 (40.5%)
IV	0 (0%)	1 (1.3%)

CXCL11. C-X-C motif chemokine ligand 11; IQR, interquartile range; TNM, tumor, lymph node, and metastasis.

Table 2.

Univariate Cox regression analyses in pancreatic cancer tissues.

Characteristics	Total (n)	Univariate analysis
		Hazard ratio (95% CI)	P-value
CXCL11	78
low	24	Reference
high	54	1.589 (0.876–2.880)	0.127
Tumor size	77	0.963 (0.827–1.120)	0.624
T	77
T2	61	Reference
T1	1	2.267 (0.307–16.714)	0.422
T3	15	0.992 (0.499–1.974)	0.982
N	75
N0	40	Reference
N1	35	1.973 (1.143–3.405)	0.015
TNM stage	78
Ⅰ	32	Reference
Ⅱ	46	1.986 (1.134–3.477)	0.016

CI, confidence interval; CXCL11. C-X-C motif chemokine ligand 11; TNM, tumor, lymph node, and metastasis.

Paratumoral regions of pancreatic cancer exhibit pronounced immunosuppressive features

Given the critical importance of spatial heterogeneity in the tumor microenvironment for immune escape mechanisms, we aimed to compare the distribution of immunosuppressive cells and effector molecules between the tumor core and paratumoral regions of pancreatic cancer. IHC results showed that expression levels of the M2 macrophage marker CD163, the myeloid‑derived suppressor cell (MDSC) marker CD11b, and the immune checkpoint molecule PD‑L1 were higher in paratumor tissues than in tumor tissues (Fig. 2A). Quantitative analyses based on three dimensions—positive cell count, positivity rate, and positive density—consistently confirmed the statistically significant enrichment of these immunosuppressive markers in paratumoral regions (all P < 0.05) (Figs. 2B–J). This spatial distribution pattern suggests that the paratumoral region at the invasive front may form an "immune barrier" that restricts effective infiltration of antitumor immune cells. Therefore, the immunosuppressive microenvironment of pancreatic cancer is not confined to the tumor interior but exhibits pronounced spatial heterogeneity in paratumoral regions, which may represent a key mechanism underlying local tumor progression and immune escape.

Fig. 2.

Spatial heterogeneity of the immunosuppressive microenvironment in pancreatic cancer.

(A) Representative immunohistochemical images showing the spatial distribution of CD163, CD11b, and PD‑L1 in pancreatic cancer and paratumor tissues; (B–D) Differential expression analysis of CD163, CD11b, and programmed death-ligand 1 (PD‑L1) based on positive cell count; (E–G) Differential expression analysis of CD163, CD11b, and PD‑L1 based on positivity rate; (H–J) Differential expression analysis of CD163, CD11b, and PD‑L1 based on positive density.

CXCL11 regulates immune-related pathways and exhibits significant statistical correlations with immunosuppressive markers in pancreatic cancer

We successfully established a CXCL11 knockdown (KD) model in pancreatic cancer cells to explore the molecular functions of CXCL11, and verified knockdown efficiency using western blotting (Fig. 3A-B). Principal component analysis revealed a clear separation in gene expression profiles between the CXCL11 KD and control groups (Fig. 3C). The volcano plot showed that 28,191 genes were detected in the CXCL11 KD group versus the negative control group, including 391 differentially expressed genes, of which 166 were upregulated and 225 were downregulated (Fig. 3D). GO enrichment analysis of all differentially expressed genes indicated that CXCL11 was significantly enriched in pathways related to immune cell migration, such as "chemokine‑mediated signaling pathway," "leukocyte chemotaxis," and "neutrophil chemotaxis," as well as key immune biological processes including "major histocompatibility complex class I‑mediated antigen presentation," "T cell receptor binding," and "innate immune response" (Figs. 3E–F). These findings reveal that CXCL11 acts as a chemokine guiding immune cell migration and plays a key role in the immunoregulatory network of pancreatic cancer by influencing antigen presentation and T cell activation. To further validate the expression of CXCL11 and immunosuppressive cells/molecules at the tissue level, we performed multiplex immunofluorescence analysis. Multiplex immunofluorescence staining showed high expression of CXCL11, the M2 macrophage marker CD163, the MDSC marker CD11b, and the immune checkpoint molecule PD‑L1 in pancreatic cancer tissues (Fig. 3G). This spatial co‑expression pattern suggests that CXCL11 may regulate the expression of these immunosuppressive molecules through paracrine or autocrine mechanisms, further supporting its central role in remodeling the immunosuppressive microenvironment. Correlation analyses were performed to quantify the strength of associations between CXCL11 and immunosuppressive markers. Results showed that CXCL11 expression levels were significantly positively correlated with CD163 and CD11b across three dimensions (positive cell count, positivity rate, and positive density); however, no significant correlation was observed with PD‑L1 (Figs. 3H–S). These findings remained largely consistent across multiple quantitative dimensions, indicating a stable synergistic relationship between CXCL11 and immunosuppressive cells/effector molecules and suggesting that high CXCL11 expression is closely associated with the enrichment of M2 macrophages and MDSCs. This potentially contributes to immune escape in pancreatic cancer through the construction of a multifaceted immunosuppressive network.

Fig. 3.

Transcriptomic profiling and immune microenvironment correlation of CXCL11 in pancreatic cancer.

(A-B) Validation of C-X-C motif chemokine ligand 11 (CXCL11) protein knockdown efficiency (western blot); (C) principal component analysis of CXCL11 knockdown and control groups; (D) volcano plot displaying differentially expressed genes between CXCL11‑knockdown and control groups (red: upregulated; blue: downregulated); (E) Gene ontology (GO) enrichment analysis of molecular function (MF) and cellular component (CC); (F) GO enrichment analysis of biological process (BP); (G) Multiplex immunofluorescence staining of CXCL11 in the pancreatic cancer immune microenvironment; (H–K) Correlation analysis of CXCL11 with CD163, CD11b, and programmed death-ligand 1 (PD‑L1) based on positive cell count; (L–O) Correlation analysis of CXCL11 with CD163, CD11b, and PD‑L1 based on positivity rate; (P–S) Correlation analysis of CXCL11 with CD163, CD11b, and PD‑L1 based on positive density.

CXCL11 regulates the chemokine network through the NF‑κB signaling pathway

We further investigated CXCL11’s potential downstream signaling pathways and regulatory networks to elucidate the molecular mechanisms by which it regulates the immune microenvironment. Protein–protein interaction network construction using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database revealed potential interactions between CXCL11 and CXCL10, as well as multiple members of the CCL family (Fig. 4A). Transcriptomic data showed that the expression levels of several chemokines, including CXCL10, CXCL1, CXCL5, and CCL9, were significantly downregulated following CXCL11 knockdown (Fig. 4B). CXCL10, CXCL5, and CXCL1 were downregulated in the CXCL11 KD group (Fig. 4C). KEGG pathway enrichment analysis of key immune‑related genes regulated by CXCL11 suggested that it may be involved in the regulation of the NF‑κB signaling pathway (Fig. 4D). Functional experiments were performed in the mouse pancreatic cancer cell line Panc02 and the human pancreatic cancer cell line SW1990 to validate this finding. Western blot results showed that CXCL11 knockdown significantly inhibited the expression of the NF‑κB core protein P65 and its downstream molecule CCL2, whereas CXCL11 overexpression had the opposite effect (Figs. 4E–L). IHC staining of tissue microarrays further validated the positive correlation between CXCL11 and CCL2 at the tissue level (Figs. 4M–O). In conclusion, CXCL11 may regulate the downstream chemokine network by activating the NF‑κB signaling pathway, thereby contributing to the remodeling of the pancreatic cancer immune microenvironment.

Fig. 4.

Molecular mechanism of CXCL11 regulating the NF-κB signaling pathway.

(A) C-X-C motif chemokine ligand 11 (CXCL11) protein–protein interaction network constructed using the Search tool for the retrieval of interacting genes/proteins (STRING) database; (B) heatmap of key immune‑related genes regulated by CXCL11 based on transcriptomic data (knockdown vs. control); (C) expression differences of CXCL11, CXCL15, CXCL5, and CXCL1; (D) Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis of key regulated genes; (E–H) western blot validation of P65 and C—C motif chemokine ligand 2 (CCL2) following CXCL11 knockdown; (I–L) western blot validation of P65 and CCL2 following CXCL11 overexpression; (M) immunohistochemical staining of CXCL11 and CCL2 in pancreatic cancer tissues; (N) differential expression of CCL2 between pancreatic cancer and paratumor tissues; (O) correlation analysis between CXCL11 and CCL2 at the tissue level.

Discussion

This study systematically investigated the role and molecular mechanism of CXCL11 in the immunosuppressive microenvironment of pancreatic cancer. We confirmed the pro‑tumorigenic role of CXCL11 in pancreatic cancer through clinical sample analysis, transcriptomic sequencing, and in vitro functional experiments. First, CXCL11 expression was higher in paratumor tissues than in tumor tissues, and its high expression was associated with shorter overall survival and more aggressive pathological features. Second, CXCL11 expression was significantly positively correlated at the tissue level with the M2 macrophage marker CD163 and the MDSC marker CD11b, suggesting that CXCL11 primarily participates in the recruitment and enrichment of immunosuppressive cells in pancreatic cancer, thereby promoting immune escape and tumor progression. Finally, mechanistic studies revealed that CXCL11 regulates the downstream chemokine network by activating the NF‑κB signaling pathway, thus contributing to the remodeling of the tumor immune microenvironment.

Our study identified spatial heterogeneity in the pancreatic cancer microenvironment. Multiplex immunofluorescence of human pancreatic cancer specimens revealed that CXCL11 and key immunosuppressive markers (e.g., CD163, CD11b, PD‑L1) were more enriched in paratumoral regions than in the tumor core. The tumor microenvironment of pancreatic cancer is dominated by regulatory T cells, myeloid‑derived suppressor cells, and M2 macrophages, thereby mediating immunosuppression [20]. Multiplex immunofluorescence results showed that PD‑L1 co‑localized with stromal cells and macrophages in pancreatic cancer. This finding is similar to that in colorectal cancer, where PD‑L1 expression is mainly derived from stromal and immune cells, particularly M2 macrophages, and is more abundant at the invasive margin than in the tumor center [21]. High abundance of macrophages and strong PD‑L1 staining observed at the invasive margin of pancreatic cancer often co‑exist with CD8⁺ T cells and tumor cells. However, the cytotoxic effect of immune checkpoint inhibitor (ICI) therapy (e.g., durvalumab or pembrolizumab) on pancreatic cancer cells in vitro 3D co‑culture models was minimal [22]. This encompasses the complex reasons for resistance to ICI therapy in pancreatic cancer. At the invasive margin of pancreatic cancer, the number of M2‑polarized tumor-associated macrophages increases, while the number of cytotoxic T cells decreases, consistent with T cell exclusion in tumor immunity. Moreover, PD‑1⁺ and PD‑L1⁺ CD8⁺ T cells are significantly increased at the invasive margin, indicating that some CD8⁺ T cells progressively acquire an exhausted phenotype [23]. These findings suggest that CXCL11 may recruit antitumor immune cells to the border, but these cells are subsequently "exhausted" or inactivated by the high density of suppressive cells and PD‑L1 signaling in this region, thereby forming an effective immune barrier that physically restricts cytotoxic T cell infiltration and creates favorable conditions for local tumor invasion and metastasis.

Second, our study is the first to preliminarily elucidate the key molecular mechanism in pancreatic cancer by which CXCL11 upregulates CCL2 expression through activation of the NF‑κB signaling pathway, thereby driving the formation of an immunosuppressive microenvironment. Nuclear factor‑κB is a core transcription factor, and persistent activation of the NF‑κB signaling pathway in pancreatic cancer is considered an important driver of cancer development and immune resistance [24,25]. Tumor‑derived extracellular vesicles from pancreatic cancer cells transfer miR‑155‑5p to macrophages, activating the Akt/NF‑κB signaling pathway by targeting ETS homologous factor, thereby promoting macrophage polarization toward the M2 phenotype. Furthermore, as a key upstream regulator, NF‑κB drives CCL2 expression and secretion, thereby amplifying the immunosuppressive effect [26,27]. In pancreatic cancer, CCL2 is primarily secreted by leukocytes, cancer‑associated fibroblasts, and tumor‑associated macrophages, and mediates the recruitment of myeloid‑derived suppressor cells, regulatory T cells, and M2 macrophages through the CCL2 to C—C chemokine receptor type 2 (CCR2) axis, thereby exacerbating immunosuppression [28]. This study preliminarily demonstrates the key role of CXCL11 as an upstream regulator of the NF‑κB‑CCL2 axis in shaping the immune microenvironment of pancreatic cancer. We propose that CXCL11 may activate the NF‑κB pathway within local cells in paratumoral regions of pancreatic cancer through autocrine or paracrine signaling, thereby significantly upregulating the secretion of CCL2 and other chemokines. High levels of CCL2, via the CCL2‑CCR2 axis, recruit MDSCs and M2 macrophages to the tumor margin and upregulate the expression of the immune checkpoint molecule PD‑L1, leading to exhaustion of CD8⁺ T cells. This forms a functional "immunosuppressive barrier" at the tumor‑normal tissue interface, hindering the infiltration and function of cytotoxic T cells into the tumor core.

CXCL11 is generally associated with antitumor immunity and a favorable prognosis in colorectal and ovarian cancer tumors, primarily because it recruits CXCR3⁺ M1 macrophages and cytotoxic T cells [13,16]. Our study suggests that CXCL11 signaling may be reprogrammed in the highly fibrotic and immunosuppressive tumor microenvironment of pancreatic cancer [29,30]. In the context of persistent aberrant activation of the NF‑κB pathway, together with abundant myeloid‑derived suppressor cells and M2 macrophages potentially recruited by factors such as CCL2 [26,31,32], the downstream effects of NF‑κB signaling activated by CXCL11 may be biased toward driving CCL2‑dependent immunosuppression rather than eliciting effective antitumor immune responses.

Nevertheless, this study had certain limitations. First, it was primarily based on retrospective clinical sample analysis and in vitro cellular experiments, lacking in vivo functional validation using immunocompetent animal models to comprehensively confirm the impact of CXCL11 on tumor growth and the immune microenvironment through this mechanism. Second, although we identified a positive correlation between CXCL11 and CCL2 and activation of the NF‑κB pathway, the precise upstream mechanism by which CXCL11 directly or indirectly activates NF‑κB remains to be explored. Future studies are needed to validate this signaling axis in vivo and explore the potential value of targeting this pathway to break the immunosuppressive barrier and enhance immunotherapeutic efficacy.

Conclusion

This study reveals the pro‑tumorigenic role of CXCL11 in PDAC and its unique spatial regulatory pattern. CXCL11 is highly expressed in paratumoral regions, associated with poor patient prognosis, and co‑enriched with immunosuppressive markers (CD163, CD11b, and PD‑L1) at the invasive front, suggesting its involvement in constructing a spatially heterogeneous immunosuppressive barrier. Mechanistically, this study preliminarily demonstrates that CXCL11 upregulates the key chemokine CCL2 by activating the NF‑κB signaling pathway. This "CXCL11–NF‑κB–CCL2″ axis drives the recruitment of CCR2‑expressing MDSCs and M2 tumor‑associated macrophages, forming a PD‑L1‑rich immune‑privileged zone at the tumor–normal tissue interface, thereby hindering the infiltration and function of cytotoxic T cells. This provides a mechanistic explanation for the general resistance of PDAC to ICIs and offers a new theoretical basis for developing novel combination immunotherapy strategies targeting CXCL11 and related pathways.

Ethics approval and informed consent

This study was approved by the Ethics Committee of Shanghai Outdo Biotech Co., Ltd. (Approval No. YBM‑05‑01) and the Ethics Committee of Shanghai Zhuoli Biotechnology Co., Ltd. (Approval No. LLS M‑15‑01). All research procedures involving human participants were in accordance with the Declaration of Helsinki and the International Ethical Guidelines for Biomedical Research Involving Human Subjects, and informed consent was obtained from all patients.

CRediT authorship contribution statement

Fang Wei: Writing – original draft, Investigation, Formal analysis. Lu Yang: Investigation, Data curation. Xianghai Zeng: Investigation, Data curation. Qian Wang: Visualization, Software. Ziqi Wang: Conceptualization. Ben Zhao: Writing – review & editing, Supervision. Xianglin Yuan: Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.