TREM2 depletion in pancreatic cancer elicits pathogenic inflammation and accelerates tumor progression via enriching IL-1β⁺ macrophages

Abstract

BACKGROUND & AIMS:

Pancreatic ductal adenocarcinoma (PDAC) has a complex tumor microenvironment enriched with tumor-associated macrophages. Triggering receptor expressed on myeloid cells 2 (TREM2) is highly expressed by a subset of macrophages in PDAC. However, the functional role of TREM2 in PDAC progression remains elusive.

METHODS:

We generated a novel transgenic mouse model (KPPC;Trem2^−/−) that enables the genetic depletion of TREM2 in the context of spontaneous PDAC development. Single-cell RNA-sequencing analysis was utilized to identify changes in the tumor immune microenvironment upon TREM2 depletion. We evaluated the impacts of TREM2 depletion on the tumor immune microenvironment to elucidate the functions of TREM2 in macrophages and PDAC development.

RESULTS:

Unexpectedly, genetic depletion of TREM2 significantly accelerated spontaneous PDAC progression and shortened the survival of KPPC;Trem2^−/− mice. Single-cell analysis revealed that TREM2 depletion enhanced pro-inflammatory macrophages and exacerbated pathogenic inflammation in PDAC. Specifically, TREM2 functions as a key braking mechanism for the NLRP3/NF-κB/IL-1β inflammasome pathway, opposing to microbial lipopolysaccharide (LPS) as the key activator of this pathway. TREM2 deficiency orchestrated with microbial LPS to trigger IL-1β upregulation and pathogenic inflammation, thereby fueling PDAC development. Notably, IL-1β inhibition or microbiome ablation not only reversed the accelerated PDAC progression caused by TREM2 depletion, but also further inhibited PDAC progression in the TREM2-depleted context.

CONCLUSIONS:

TREM2 depletion accelerates tumor progression by enhancing pro-inflammatory macrophages and IL-1β-mediated pathogenic inflammation in PDAC. The accelerated tumor progression by TREM2 depletion can be reversed by blocking IL-1β-associated pathogenic inflammation.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies, characterized by its highly desmoplastic tumor microenvironment¹. Despite extensive efforts, multiple therapeutic interventions targeting this tumor microenvironment have yielded disappointing results in clinical trials. Our recent studies have highlighted the functional heterogeneity of tumor stromal components, revealing their dual roles with both tumor-promoting and tumor-restraining functions, depending on the context^2–5. These findings underscore the critical need to selectively target the tumor-promoting functions of PDAC stromal components without abrogating their tumor-restraining properties. Moreover, our previous findings have elucidated the immune-regulatory roles of key stromal components^2–4, which can significantly impact the efficacy of immunotherapy in PDAC. Further insights into the immune-regulatory functions of stromal components using clinically relevant transgenic mouse models may facilitate the development of new immunotherapeutic strategies for PDAC.

Triggering receptor expressed on myeloid cells 2 (TREM2), highly expressed by tumor-infiltrating myeloid cells/macrophages, has garnered attention in cancer research due to its upregulation in various cancer types^6–11. These studies have elucidated the tumor-promoting functions of TREM2 in models of sarcoma, colorectal, and lung cancers, suggesting potential therapeutic strategies targeting TREM2 in these contexts^{7, 8, 11}. Specifically, TREM2 depletion using Trem2^™/ transgenic mouse model impeded tumor progression and enhanced anti-tumor immune responses⁸. However, in hepatocellular carcinoma, head and neck squamous cell carcinoma, and glioblastoma, TREM2 has been shown to exert tumor-restraining functions^{12, 13}. These conflicting findings underscore that TREM2 can have either tumor-promoting or tumor-restraining functions, depending on the distinct microenvironmental cues in different cancer types. Therefore, it is pivotal to re-evaluate the roles of TREM2 and potential outcomes of TREM2-targeted therapeutic strategies in a cancer-type-specific manner. In the case of pancreatic cancer, it remains unclear whether TREM2 has tumor-promoting or tumor-restraining functions, necessitating further investigations using preclinical genetically engineered mouse models (GEMMs) with autochthonous pancreatic tumor formation. Even less is known regarding the functional role of TREM2 in shaping macrophage phenotypes and the tumor immune microenvironment of pancreatic cancer.

To unravel the functional role and therapeutic relevance of TREM2 in pancreatic cancer, we developed a novel transgenic mouse model, LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre;Trem2^−/− (KPPC;Trem2^−/−). The KPPC;Trem2^−/− mouse model allows the genetic depletion of TREM2 (encoded by the Trem2 gene) in the context of spontaneous pancreatic tumor development. Subsequently, we utilized single-cell RNA-sequencing (scRNA-seq) analysis to comprehensively investigate how TREM2 depletion in KPPC;Trem2^−/− tumors impacts macrophage phenotypes and the tumor immune microenvironment in PDAC from a systemic perspective. Taken together, this study leverages novel transgenic mouse models and single-cell analysis to investigate the functional roles of TREM2 in pancreatic tumor development and immune response. By elucidating these important roles of TREM2, we aim to provide new insights into potential therapeutic opportunities associated with TREM2 and related pathways in pancreatic cancer.

Methods

Please refer to the online Supplementary Materials for additional information.

Mice

The Trem2^−/− mice (harboring a stop codon at amino acid 17) were purchased from the Jackson Laboratory (Strain No: 027197). LSL-Kras^G12D/+, Trp53^loxP/loxP, and Pdx1-Cre mice were purchased from the Jackson laboratory (Strain No: 032429 and 014647). LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre (KPPC) mice were crossed with Trem2^−/− mice to get the LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre;Trem2^−/− (KPPC;Trem2^−/−) mice. LSL-Kras^G12D/+;Pdx1-Cre (KC) mice were crossed with Trem2^−/− mice to get the LSL-Kras^G12D/+;Pdx1-Cre;Trem2^−/− (KC;Trem2^−/−) mice. All mouse strains were maintained with the C57BL/6 background. Both female and male mice with KPPC, KPPC;Trem2^−/−, KC, and KC;Trem2^−/− genotypes were used as experimental mice with spontaneous pancreatic intraepithelial neoplasia (PanIN) and/or pancreatic ductal adenocarcinoma (PDAC) lesions. The C57BL/6J mice were obtained from the Jackson Laboratory. All mouse strains were housed in standard housing conditions at MD Anderson Cancer Center (MDACC) north campus animal facility. All animal procedures were reviewed and approved by MDACC Institutional Animal Care and Use Committee (IACUC; protocol number 00002328-RN00).

Single-cell RNA-sequencing (scRNA-seq) analysis

The KPPC pancreatic tumor group contains a total of 7432 cells from KPPC mice (n = 3; at 70-day age), and the KPPC;Trem2^−/− pancreatic tumor group contains a total of 6813 cells from stage-matched KPPC;Trem2^−/− mice (n = 3; at 50-day age). To ensure that the pancreatic tumors from KPPC and KPPC;Trem2^−/− mice were stage-matched, each tumor sample was divided into two mirror sections: one for single-cell analysis and the other for histological evaluation by a pathologist after H&E staining. All samples were processed following the same protocol as described in our recent studies ^{2, 4, 14} and analyzed by the Advanced Technology Genomics Core Facility of MDACC. Unfractionated live cells from KPPC tumors or KPPC;Trem2^−/− tumors were captured with the 10X Genomics’ Chromium controller and Single Cell 3’ Reagent Kits v3. cDNA was synthesized and amplified to construct Illumina sequencing libraries. The libraries were sequenced by Illumina NovaSeq 6000 system. The run format was 26 cycles for read 1, 8 cycles for index 1, and 124 cycles for read 2. Library Seurat_5.0.2, dplyr_1.1.4 and cowplot_1.1.3 were installed into the R package (version 4.3.2) to set quality control threshold, normalize data, cluster cells, and identify maker genes. To filter out the cells with poor sequencing quality, a threshold was set at least 200 genes and a maximum of 7500 genes per cell. Cells with more than 10% of the mitochondrial gene and 30% of the ribosome gene were excluded. “RunUMAP”, “JackStrawPlot”, “ElbowPlot”, “FindAllMarkers”, “DoHeatmap”, “VlnPlot”, and “DotPlot” functions were utilized to plot the expression profiles of indicated genes in various cell populations, as previously described. The expression profiles of indicated genes were also visualized by library matrixStats_1.2.0, Nebulosa_1.12.0, and scCustomize_2.0 packages. The ‘CellChat’ algorithm ¹⁵ was utilized to visualize the cell-cell interactions between different cell populations based on the ligand-receptor expression profiles. The ‘Monocle 3’ algorithm was used to visualize the macrophage pseudo-temporal trajectory using learn_graph and order_cells ¹⁶. Differentially expressed genes were analyzed using the FindMarkers function and then visualized as volcano plot using ggplot2_3.5.0. KEGG and/or GSEA analyses were conducted using enrichplot_1.22.0, clusterProfiler_4.10.0, and GSEABase_1.64.0 packages to identify the pathways of differentially expressed genes.

Animal studies

KPPC and KPPC;Trem2^−/− mice (stage-matched) were treated with indicated agents upon developing palpable pancreatic tumors. CSF1R inhibitor Pexidartinib hydrochloride (HY-16749A, MedChemExpress) was formulated in 10% DMSO and 90% corn oil and was administered daily by gavage (50 mg/kg). Diacerein (HY-N0283, MedChemExpress) was formulated in 10% DMSO and 90% corn oil and was administered daily by gavage (35 mg/kg). Pexidartinib, Diacerein, and vehicle treatments started in stage-matched KPPC mice (around 5-week age) and KPPC;Trem2^−/− mice (around 4-week age) with palpable tumors. Cerulein-induced pancreatitis was performed as described previously ¹⁷. KC and KC;Trem2^−/− mice (5 mice per group) were treated with Cerulein (CCKS-001, CPC scientific) by two intraperitoneal injections (50 μg/kg) at a 4-hour interval every other day. After 20 days of the last injection, the mice were euthanized for tissue collection and analysis. Wild-type (WT) and Trem2^−/− mice were treated with L-Arginine to induce pancreatitis as previously described ^{18, 19}, with two rounds of intraperitoneal injections on Monday and Thursday of each week, for a total of eight weeks. For each round of injections, L-Arginine (A5131, Sigma) was intraperitoneally injected (4 g/kg) twice at a 1-hour interval. Then these mice were euthanized after the last injection for further analyses.

To deplete the microbiota, KPPC and KPPC;Trem2^−/− mice were administrated with broad-spectrum antibiotics (ABX) via oral gavage as previously described ³, including vancomycin (50 mg/mL), neomycin (10 mg/mL), metronidazole (100 mg/mL), and amphotericin (1 mg/mL). During the experiments, mouse drinking water was supplied with ampicillin (1 mg/mL), vancomycin (0.5 mg/mL), neomycin (0.5 mg/mL), metronidazole (1 mg/mL), and amphotericin (0.5 μg/mL). ABX and vehicle treatments started in stage-matched KPPC mice (around 5-week age) and KPPC;Trem2^−/− mice (around 4-week age) with palpable tumors. At the endpoint of this experiment, mouse tumor and stool samples were collected for further analyses.

KPPC-Luc cells²⁰ (0.3×10⁶ cells in 50 μL PBS) were orthotopically injected into the tail of the pancreas of WT and Trem2^−/− mice using a 27-gauge Hamilton syringe. At three weeks following orthotopic injection, mice were sacrificed, then tumors were dissected, weighed, and taken images. To detect the tumor burden, luciferase expression was measured by injecting the mice i.p. with D-Luciferin (LUCK-1G, Goldbio) following the manufacturer’s instructions and imaged via IVIS (Xenogen Spectrum) at the MDACC Small Animal Imaging Facility (SAIF).

Results

Triggering receptor expressed on myeloid cells 2 (TREM2) is upregulated in the APOE^HighIL1B^Low subset of tumor-associated macrophages in pancreatic cancer

To identify the cell types that express TREM2 in pancreatic ductal adenocarcinoma (PDAC), we firstly analyzed the single-cell RNA-sequencing (scRNA-seq) datasets from tumor samples from both genetically engineered mouse models (GEMMs) and human patients. The scRNA-seq data analysis revealed that Trem2 gene was highly expressed in macrophage/monocyte population but not other cell populations (Figure 1A) in the autochthonous pancreatic tumors from LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre (KPPC) mouse model, as used in our recent studies^{3, 14}. Furthermore, the high expression of Trem2 in a major subset of macrophages was positively correlated with that of Apoe (Apolipoprotein E, ApoE), C1qa, C1qb, and C1qc (Figure 1B, Supplementary Figure 1A). Specifically, the Trem2^High macrophage subtype revealed high expression levels of Apoe, C1q genes, and Mrc1 (Figure 1C and 1D). In contrast, Trem2^Low macrophage subtype exhibited high expression levels of Il1b (encoding interleukin-1β/IL-1β) and major histocompatibility complex class II (MHCII) genes (Figure 1C and 1D). Consistent results were observed on tumors of LSL-Kras^G12D/+;Trp53^R172H/+;Ptf1a-Cre (KPC) mouse model (Figure 1E and 1F) from a recent dataset²¹. We also analyzed the scRNA-seq data of sarcoma mouse model⁸ and identified that Trem2 expression in macrophages exhibited a positive correlation with Apoe and C1q family genes, in contrast to a reverse correlation with Il1b (Supplementary Figure 1B and 1C), consistent with our observations on pancreatic tumor mouse models.

Figure 1. Expression profile of TREM2 in macrophages of human and mouse pancreatic tumors.

(A-D) Analyses on the single-cell RNA-sequencing (scRNA-seq) data of mouse primary pancreatic tumors from LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre (KPPC) mice. The expression profile of Trem2 gene in various cell populations was shown in violin plot (A). The expression profiles of Trem2, Apoe, C1qa, and Il1b in macrophages were shown in UMAP plot (B). Trem2^Low and Trem2^High macrophage subpopulations were defined in UMAP plot (C), with signature genes shown in dot plot (D).

(E and F) Analyses of mouse primary pancreatic tumors from LSL-Kras^G12D/+;Trp53^R172H/loxP;Ptf1a-Cre (KPC) mice based on the scRNA-seq data (GEO: GSE202651). Trem2^Low and Trem2^High macrophage subpopulations were defined in UMAP plot (E), with signature genes shown in dot plot (F).

(G-J) Analyses on the scRNA-seq data of human pancreatic tumors (GEO: GSE229413). The expression profile of TREM2 gene in various cell populations from was shown in violin plot (G). The expression profiles of TREM2, APOE, C1QA, and IL1B in macrophages were shown in UMAP plot (H). Indicated macrophage subpopulations were defined in UMAP plot (I). Violin plot for the expression profiles of indicated genes in macrophage subpopulations was shown in (J).

(K) Spatial transcriptomic profiles of TREM2 and APOE based on a recent dataset (GEO: GSE233254) of human primary pancreatic tumor samples.

(L) Scatterplot illustrating the correlation between the expression levels of TREM2 and macrophage-related genes, including APOE and C1QA, based on the bulk RNA-sequencing (RNA-seq) data from pancreatic cancer patient cohort of The Cancer Genome Atlas (TCGA) database. Pearson’s correlation test was used. ****P < .0001.

Consistently, scRNA-seq dataset of human pancreatic tumor samples²² also confirmed that TREM2 is highly expressed by a major subset of macrophages, which was positively correlated with APOE, C1QA, C1QB, and C1QC expression, but no clear correlation with FOLR2 expression (Figure 1G and 1H, Supplementary Figure 1D). Further analyses on macrophage subclusters in human pancreatic tumors revealed a TREM2^HighAPOE^High subcluster with low IL1B expression, in contrast to a TREM2^Low subcluster with high IL1B expression (Figure 1I and 1J).

Next, we analyzed a recently published spatial transcriptomic dataset of human pancreatic tumor sections²³. TREM2 expression was mostly localized within APOE^High tissue areas, but less frequent in FOLR2^High regions (Figure 1K, Supplementary Figure 1E). Consistently, bulk RNA-seq data of The Cancer Genome Atlas (TCGA) pancreatic cancer cohort also revealed that the expression of TREM2 is positively correlated with that of APOE or C1Q genes (Figure 1L, Supplementary Figure 1F). These results suggested that TREM2^High macrophages in PDAC represent a subcluster with higher expression levels of APOE and C1Q, while TREM2^Low subcluster has higher expression levels of IL-1β and MHCII. This was supported by immunofluorescence staining on human pancreatic tumor tissue sections showing the colocalization of TREM2 with ApoE, but not with IL-1β or MHCII (Supplementary Figure 1G).

Genetic depletion of TREM2 accelerates pancreatic tumor development

The functional roles of TREM2 in PDAC development and immune response remain unknown. To achieve genetic depletion of TREM2 (Trem2 gene) in the context of autochthonous pancreatic tumorigenesis, we developed a new transgenic mouse model, LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre;Trem2^−/− (KPPC;Trem2^−/−) (Figure 2A). Genetic depletion of Trem2 in KPPC;Trem2^−/− mice was confirmed by genotyping PCR and Western blot assays (Supplementary Figure 2A and 2B). Interestingly, KPPC;Trem2^−/− mice exhibited significantly accelerated tumor development and shortened overall survival, as compared to background-matched KPPC control mice (Figure 2B, Supplementary Figure 2C–E). KPPC;Trem2^−/− mice of 4- or 6-week-age developed significantly larger areas of acinar-to-ductal metaplasia (ADM) and pancreatic intraepithelial neoplasia (PanIN) lesions than age-matched KPPC mice (Figure 2C and 2D, Supplementary Figure 2F). At 8 weeks of age, KPPC;Trem2^−/− mice reached moribund status and exhibited larger areas of late-stage PDAC with poorly differentiated tumor histology, as compared with KPPC control mice (Figure 2E and Supplementary Figure 2G and 2H). The histology of other major organs from TREM2-depleted mice did not exhibit any abnormality as compared to TREM2-intact control mice (Supplementary Figure 3). In addition, KPPC;Trem2^−/−, but not KPPC;Trem2^−/+ tumors revealed significantly larger areas of cytokeratin-19 (CK19) and phosphorylated-ERK positivity than KPPC tumors (Figure 2F and Supplementary Figure 4). The levels of alpha-smooth muscle actin (αSMA) and collagen deposition were not significantly altered by TREM2 depletion (Figure 2F). KPPC;Trem2^−/+ mice with heterozygous deletion of Trem2 did not exhibit shortened survival (Figure 2B) or accelerated tumor progression (Figure 2E) as compared with KPPC mice.

Figure 2. TREM2 depletion accelerates spontaneous PDAC development in a novel transgenic mouse model and shortens the survival.

(A) Genetic strategy to delete Trem2 in the context of autochthonous PDAC development using the LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre;Trem2^−/− (KPPC;Trem2^−/−) mouse model.

(B) Survival of KPPC;Trem2^−/− mice (n = 13), as compared to that of background-matched KPPC mice (n = 13). Survival of KPPC;Trem2^−/+ mice harboring heterozygous Trem2 deletion was also shown (n = 12). Log-rank (Mantel-Cox) test was used.

(C-E) Histology of pancreatic tissues from KPPC and KPPC;Trem2^−/− mice at the same age of four weeks (C), six weeks (D), and eight weeks (E). Quantitative histology evaluation of tumors from KPPC (n = 5) and KPPC;Trem2^−/− (n = 5) mice was shown. Pancreatic tissues from tumor-free wild-type (WT) mice, pancreatic tissues from tumor-free Trem2^−/− mice, and pancreatic tumors from KPPC;Trem2^−/+ mice (with heterozygous Trem2 deletion) were also examined as additional control samples in (E). Scale bar: 100 μm.

(F) Representative images of immunohistochemistry staining for cytokeratin-19 (CK19), phosphorylated ERK (P-ERK), αSMA, type I collagen (Col1), and Picrosirius Red on KPPC and KPPC;Trem2^−/− tumors (n = 5/group). Staining positivity quantification was shown with Student’s t test. Scale bar: 100 μm.

*P < .05, **P < .01, ****P < .0001; ns: not significant.

Single-cell analysis identifies the significant changes of macrophages in TREM2-depleted pancreatic tumors

Next, we conducted single-cell analysis (scRNA-seq) to compare the cellular compositions and phenotypes of various cell populations between KPPC;Trem2^−/− tumors and stage-matched KPPC control tumors (Figure 3A, Supplementary Figure 5A–5C). KPPC;Trem2^−/− tumors exhibited dramatically enriched macrophage/monocyte population (Figure 3B and 3C). Particularly, the percentage of tumor-associated macrophage subcluster-1 (TAM-1) among total macrophages significantly increased upon TREM2 depletion, becoming the dominant macrophage subtype in KPPC;Trem2^−/− tumors (Figure 3D). In contrast, the percentage of TAM-2 subcluster significantly decreased in TREM2-depleted tumors (Figure 3D). These results indicated that KPPC;Trem2^−/− tumors exhibited altered macrophage polarization switching from anti-inflammatory TAM-2 (with high expression of Arg1) to pro-inflammatory TAM-1 (with high expression of MHCII genes) (Figure 3E, Supplementary Figure 6A). Cell-cell interaction analysis using the CellChat algorithm further revealed the significantly enhanced signaling transduction associated with TAM-1 subcluster in KPPC;Trem2^−/− tumors (Figure 3F, Supplementary Figure 6B). Specifically, the TAM-1 and TAM-2 subtypes in our mouse models resemble the human macrophage TREM2^LowIL1B^High and TREM2^HighAPOE^High subtypes, respectively (Supplementary Figure 6C). In addition to the TAM-1/2 subcluster ratio changes, we observed significantly elevated total numbers of macrophages (TAMs) in scRNA-seq analysis and immunostaining of Adgre1 (F4/80) and Csf1r (CSF1R) (Figure 3G–J). We further performed additional immunostaining of F4/80, IL-1β, Arg1, CD19, MPO, CD3, and CD11c to characterize the immune profiles (Supplementary Figure 7A and 7B). Due to the direct correlation between TREM2 and ApoE, TREM2 depletion led to decreased ApoE expression KPPC;Trem2^−/− tumors (Supplementary Figure 8A–8B). The decreased levels of ApoE and TREM2 were also associated with downregulation of cholesterol efflux signaling and phagocytosis pathways in macrophages of KPPC;Trem2^−/− tumors (Supplementary Figure 8C).

Figure 3. scRNA-seq analysis identifies altered cell composition in TREM2-depleted pancreatic tumors.

(A-C) scRNA-seq analysis of unfractionated live cell mixture from autochthonous pancreatic tumors from KPPC mice (with 7432 total cells from 3 mice) and stage-matched KPPC;Trem2^−/− mice (with 6813 total cells from 3 mice). The major cell clusters were defined in UMAP plot (A). Split view of cell composition profiles comparing KPPC tumors and KPPC;Trem2^−/− tumors was shown in UMAP plot (B). cDC, classical dendritic cell; pDC, plasmacytoid dendritic cell. The relative abundance (%) of indicated cell populations in KPPC tumors and KPPC;Trem2^−/− tumors was compared in (C).

(D) UMAP showing cell compositions of indicated macrophage subpopulations (left). Split view of macrophage subtype compositions comparing KPPC tumors and KPPC;Trem2^−/− tumors was shown in UMAP plot (right). The relative abundance of various macrophage subtypes was shown in pie chart plot.

(E) The signature genes of defined macrophage subpopulations were shown in dot plot.

(F) The cell-cell interaction networks among indicated macrophage subpopulations in KPPC and KPPC;Trem2^−/− tumors, shown as intensity of interactions based on the “CellChat” algorithm. Dot size represented the cell number of indicated cell subpopulation.

(G) UMAP showing Adgre1 (encoding F4/80) expression in macrophages from KPPC and KPPC;Trem2^−/− tumors.

(H) Representative images of immunohistochemistry staining for F4/80 in KPPC and KPPC;Trem2^−/− tumors (n = 5/group). Staining positivity quantification was shown with Student’s t test. Scale bar: 100 μm.

(I) UMAP showing Csf1r (encoding CSF1R) expression in macrophages from KPPC and KPPC;Trem2^−/− tumors.

(J) Representative images of immunohistochemistry staining for CSF1R in KPPC and KPPC;Trem2^−/− tumors (n = 5/group). Staining positivity quantification was also shown with Student’s t test. Scale bar: 100 μm.

**P < .01.

In order to evaluate whether the enriched macrophages in KPPC;Trem2^−/− tumors functionally contributed to the accelerated PDAC progression in KPPC;Trem2^−/− mice, we treated the mice with Pexidartinib (PLX-3397), an inhibitor of CSF1R (Figure 4A). Pexidartinib treatment reversed the accelerated tumor progression caused by TREM2 depletion, resulting in prolonged overall survival of KPPC;Trem2^−/− mice (Figure 4B) with improved tumor histology (Figure 4C, Supplementary Figure 8D). The effect of Pexidartinib treatment was associated with decreased macrophage accumulation in KPPC and KPPC;Trem2^−/− tumors (Figure 4D). The tumor progression and survival of KPPC control mice were not improved by Pexidartinib treatment (Figure 4B–4D).

Figure 4. TREM2 depletion enhances the accumulation of macrophages with pro-inflammatory phenotypes, which can be reversed by CSF1R inhibition.

(A) Schematic of CSF1R inhibitor (CSF1Ri) Pexidartinib treatment in KPPC and KPPC;Trem2^−/− mice. Indicated treatments started in stage-matched KPPC and KPPC;Trem2^−/− mice with palpable tumors (n = 5/group).

(B) Survival of KPPC and KPPC;Trem2^−/− mice with indicated treatments in (A). Log-rank (Mantel-Cox) test was used.

(C and D) Representative images of H&E staining (C) and F4/80 immunohistochemistry (D) staining on endpoint pancreatic tumors from vehicle- or Pexidartinib-treated KPPC and KPPC;Trem2^−/− mice (n = 5/group). Staining positivity quantification was also shown with one-way ANOVA with Tukey’s multiple comparisons test (D). Scale bar: 100 μm.

(E) Volcano plot showing differentially expressed genes in macrophages between KPPC and KPPC;Trem2^−/− tumors, based on scRNA-seq data.

(F) KEGG analysis of top enriched gene pathways in macrophages of KPPC;Trem2^−/− tumors, as compared to macrophages of KPPC tumors.

(G-I) Pseudo-time trajectory of macrophages from KPPC and KPPC;Trem2^−/− tumors, as plotted by the Monocle 3 inference analysis. The UMAP distributions of macrophages from KPPC and KPPC;Trem2^−/− groups were compared in (G). The pseudo-time trajectory (H) and distribution of indicated macrophage subpopulations (I) were also shown.

**P < .01, ***P < .001, ****P < .0001.

To further elucidate the macrophage phenotype alterations induced by TREM2 depletion, we conducted gene enrichment analysis to compare the differentially expressed genes in macrophages between KPPC;Trem2^−/− tumors and KPPC tumors. Macrophages in KPPC;Trem2^−/− tumors exhibited significant upregulation of pro-inflammatory genes such as Il1b (Figure 4E), which was associated with enhanced NOD-like receptor (NLR) and NF-κB pathways (Figure 4F). These results are consistent with previous studies that identify TREM2 as a key anti-inflammatory factor in macrophages^{24, 25}. Pseudo-time trajectory analysis by Monocle 3 algorithm further revealed the significantly altered phenotype trajectory of macrophages in KPPC;Trem2^−/− tumors (Figure 4G and 4H), featured by enhanced TAM-1 differentiation (Figure 4I) with elevated Il1b expression (Supplementary Figure 8E). These results are consistent with our previous results showing the reverse correlation between TREM2 and IL-1β (Figure 1).

TREM2 depletion elicits IL-1β upregulation and pathogenic inflammation, which can be reversed by IL-1β inhibition

Next, we validated that the accelerated progression in KPPC;Trem2^−/− tumors was associated with elevated IL-1β expression in enriched pro-inflammatory macrophages (Figure 5A and 5B, Supplementary Figure 9A). IL-1β upregulation and macrophage enrichment were observed at various stages in KPPC;Trem2^−/− tumors (Supplementary Figure 9B–E). IL-1β upregulation was also accompanied by elevated levels of Casp1 expression (Figure 5C) and cleaved Caspase-1 (Figure 5D), a key enzyme for IL-1β release²⁶. ELISA assay further validated the elevated IL-1β levels in the serum samples of KPPC;Trem2^−/− mice (Supplementary Figure 9F). Additionally, the elevated IL-1β expression and IL-1β⁺ macrophage by TREM2 depletion could be mitigated by CSF1R inhibitor Pexidartinib treatment (Supplementary Figure 10A–10C). Consistent with recent studies identifying IL-1β as an essential pro-inflammatory factor that can promote PDAC progression^27–29, our in vitro results also demonstrated that IL-1β could promote pancreatic cancer cell proliferation (Supplementary Figure 10D).

Figure 5. TREM2 depletion upregulates IL-1β pro-inflammatory pathway in macrophages, which can be reversed by IL-1β inhibitor treatment.

(A) UMAP showing Il1b (encoding IL-1β) expression in macrophages of KPPC and KPPC;Trem2^−/− tumors.

(B) Representative images of immunohistochemistry staining for IL-1β in KPPC and KPPC;Trem2^−/− tumors (n = 5/group). Staining positivity quantification was shown with Student’s t test. Scale bar: 100 μm.

(C) UMAP showing Casp1 (encoding Caspase-1) expression in macrophages of KPPC and KPPC;Trem2^−/− tumors.

(D) Representative images of immunohistochemistry staining for Cleaved Caspase-1 in KPPC and KPPC;Trem2^−/− tumors (n = 5/group). Staining positivity quantification was shown with Student’s t test. Scale bar: 100 μm.

(E) Schematic of IL-1β inhibitor (IL-1βi) Diacerein treatment in KPPC and KPPC;Trem2^−/− mice. Indicated treatments started in stage-matched KPPC and KPPC;Trem2^−/− mice with palpable tumors (n = 7/group).

(F) Survival of KPPC and KPPC;Trem2^−/− mice treated with or without IL-1β inhibitor Diacerein (n = 7/group). Log-rank (Mantel-Cox) test was used.

(G and H) Representative images of H&E staining (G) and F4/80 immunohistochemistry staining (H) on tumor sections from vehicle- or Diacerein-treated KPPC and KPPC;Trem2^−/− mice at endpoint. Staining positivity quantification was shown with one-way ANOVA with Tukey’s multiple comparisons test (H). Scale bar: 100 μm.

(I) Gene set enrichment analysis (GSEA) analysis showing upregulation of NF-κB pathway genes in macrophages of KPPC;Trem2^−/− tumors.

(J and K) qRT-PCR analysis of NOD-like receptor (NLR) pathway genes in KPPC and KPPC;Trem2^−/− tumors, as shown in heatmap plot (J) or normalized bar graph (K). Student’s t test was used.

**P < .01, ***P < .001, ****P < .0001.

The above findings suggested the key role of IL-1β in TREM2-depletion-accelerated tumorigenesis, which prompted us to examine the efficacy of IL-1β inhibition on KPPC;Trem2^−/− mice (Figure 5E). We tested the IL-1β inhibitor Diacerein, an anti-inflammatory drug for the treatment of joint diseases such as osteoarthritis. Intriguingly, IL-1β inhibition using Diacerein not only reversed the tumor-promoting effect of TREM2 depletion, but also significantly inhibited tumor progression in the TREM2-depleted context and prolonged the overall survival of KPPC;Trem2^−/− mice (Figure 5F) with improved tumor histology (Figure 5G, Supplementary Figure 10E). The enhanced therapeutic efficacy of IL-1β inhibition in KPPC;Trem2^−/− mice was associated with significantly decreased macrophage infiltration (Figure 5H), consistent with previous studies showing that IL-1β upregulates the expression of monocyte chemoattractant proteins (MCPs)³⁰ while IL-1β ablation inhibits macrophage accumulation²⁷. Consistently, we observed that several MCP genes were downregulated after IL-1β inhibitor Diacerein treatment (Supplementary Figure 10F). Diacerein treatment significantly downregulated the IL-1β pathway downstream genes in KPPC;Trem2^−/− tumors (Supplementary Figure 10G), which was associated with decreased numbers of IL-1β⁺ macrophages (Supplementary Figure 10H). Taken together, these results indicated the accelerated PDAC development by TREM2 depletion is associated with IL-1β upregulation, which can be reversed by IL-1β inhibition.

TREM2 depletion enhances IL-1β⁺ pro-inflammatory macrophages via NLRP3/NF-κB pathway activation

To elucidate the mechanism by which TREM2 depletion upregulates IL-1β in PDAC, we conducted Gene Set Enrichment Analysis (GSEA) and identified the significant upregulation of NF-κB pathway genes in macrophages of KPPC;Trem2^−/− tumors (Figure 5I). Upregulation of Il1b, Nfkb1, Nlrp3, and other NLR-related genes in KPPC;Trem2^−/− tumors were also identified in total mRNAs from KPPC;Trem2^−/− tumors (Figure 5J and 5K). Our findings indicated that TREM2-depletion-induced IL-1β upregulation is regulated through the NLRP3/NF-κB pathway. To validate the clinical relevance of our observations on transgenic mouse models, we continued the analysis on the single-cell datasets of the macrophages from human PDAC samples²². Our previous results revealed the TREM2^HighAPOE^High macrophage subcluster with low IL1B expression and a TREM2^Low macrophage subcluster with high IL1B expression (Figure 1J). We further identified the TREM2^LowIL1B^High macrophage subcluster exhibited high expression levels of NLRP3, NFKB1, TLR2, and RIPK2, all of which are associated with the NLRP3/NF-κB pathway (Figure 6A and 6B). Pathway analysis also revealed upregulation of pro-inflammatory pathway genes in TREM2^LowIL1B^High macrophage subcluster as compared to TREM2^HighAPOE^High subcluster (Figure 6C and Supplementary Figure 10I). Consistently, macrophages of KPPC;Trem2^−/− tumors exhibited upregulation of similar pro-inflammatory pathway genes, as compared to macrophages of KPPC tumors (Figure 6D, Supplementary Figure 10J).

Figure 6. TREM2 depletion orchestrates with microbial lipopolysaccharide (LPS) to ignite NLRP3/IL-1β inflammatory pathway, which can be reversed by antibiotics treatment.

(A-C) Analysis on scRNA-seq data of macrophages from human pancreatic tumors (GEO: GSE229413). Cell compositions of indicated macrophage subpopulations were shown in UMAP plot (A). The expression profiles of indicated genes were shown in violin plot (B). GSEA analysis revealed the upregulation of inflammatory response pathway genes in TREM2^-IL1B⁺ macrophage subpopulation as compared with TREM2⁺APOE⁺ subpopulation (C).

(D) GSEA analysis identified the upregulation of inflammatory response pathway genes in macrophages of KPPC;Trem2^−/− tumors, as compared to macrophages of KPPC tumors.

(E) Enhanced activation of IL-1β⁺CD11b⁺ bone marrow-derived macrophages (BMDMs) from Trem2^−/− mice as compared to BMDMs from WT mice, as measured by the percentage of emerging IL-1β⁺CD11b⁺ macrophages after stimulation with LPS at 20 ng/mL for three hours (n = 4 biological replicates). Significance was determined by one-way ANOVA with Tukey’s multiple comparisons test.

(F) qRT-PCR analysis of Il1b in the total mRNA samples from WT and Trem2^−/− BMDMs treated with or without LPS at 20 ng/mL for four hours (n = 3 biological replicates). Significance was determined by one-way ANOVA with Tukey’s multiple comparisons test.

(G) Schematic representation of broad-spectrum antibiotics (ABX) treatment in KPPC and KPPC;Trem2^−/− mice. Indicated treatments started in stage-matched KPPC and KPPC;Trem2^−/− mice with palpable tumors (n = 5/group).

(H) Survival of KPPC and KPPC;Trem2^−/− mice treated with or without ABX (n = 5/group) from (G). Log-rank (Mantel-Cox) test was used.

(I) Representative images of H&E staining on endpoint tumors from KPPC and KPPC;Trem2^−/− mice treated with or without ABX. Quantitative histology analysis of each group was plotted in bar graph (n = 5/group). Scale bar: 100 μm.

(J) Representative images of immunohistochemistry staining for IL-1β in KPPC and KPPC;Trem2^−/− tumors treated with or without ABX from (G). Staining positivity quantification was also shown with Student’s t test (n = 5/group). Scale bar: 100 μm.

(K) qRT-PCR analysis of Il1b in the total mRNA samples from KPPC and KPPC;Trem2^−/− tumors treated with or without ABX from (G) (n = 5/group). Student’s t test was used.

(L) qRT-PCR analysis of total bacterial 16S DNA gene expression in fecal samples collected from KPPC and KPPC;Trem2^−/− mice treated with vehicle control or ABX (n = 5/group). The number of 16S DNA copies was normalized to the number of mouse genomic DNA copies. The values of vehicle-treated groups were used to normalize the value of ABX-treated groups. Student’s t test was used.

*P < .05, **P < .01, ***P < .001, ****P < .0001; ns: not significant.

Next, we isolated the primary bone marrow-derived macrophages (BMDMs) and conducted in vitro activation of BMDMs following the standard protocol of bacterial lipopolysaccharide (LPS) treatment. We identified that BMDMs from Trem2^−/− mice exhibited significant IL-1β upregulation than BMDMs from wild-type mice under the same condition of LPS induction, as examined by flow cytometry and qRT-PCR assays (Figure 6E and 6F). Consistently, Trem2^−/− BMDMs also exhibited significantly higher IL-1β upregulation upon treatment of NLRP3 inflammasome pathway activator Nigericin, which could be diminished by NLRP3 inhibitor MCC950 treatment (Supplementary Figure 10K). In addition, we identified that hypoxia pathway and HIF-1α expression were upregulated in TREM2-depleted macrophages based on our scRNA-seq, qRT-PCR, and immunostaining results (Supplementary Figure 11A–11D). Consistent with previous studies³¹, the upregulation of IL-1β in LPS-activated Trem2^−/− BMDMs could be suppressed by HIF-1α inhibitor LW6 treatment (Supplementary Figure 11E). Taken together, these results indicate that the elevated IL-1β expression in Trem2^−/− macrophages is mediated through NLRP3/NF-κB/HIF-1α related pathway, especially in the presence of LPS stimulation.

TREM2-depletion-induced IL-1β upregulation and pathogenic inflammation can be reversed by microbiota ablation with antibiotics

In order to investigate the functional contribution of tumor microbiome (key source of LPS) to macrophage activation and inflammation in PDAC, we treated the KPPC;Trem2^−/− mice and KPPC mice with broad-spectrum antibiotics (ABX) (Figure 6G). By ablating the gut and tumor microbiota, ABX treatment not only reversed the TREM2-depletion-accelerated tumor progression, but also significantly prolonged the overall survival of KPPC;Trem2^−/− mice as compared to KPPC mice (Figure 6H and 6I). Additionally, IL-1β upregulation caused by TREM2 depletion was abolished by ABX treatment (Figure 6J and 6K). In contrast, the survival of KPPC control mice was not significantly improved by ABX treatment, consistent with our previous observations³. Efficient microbiota depletion by ABX treatment was confirmed by quantitative PCR analysis of 16S DNA levels (Figure 6L). Additionally, the prolonged overall survival and suppressed PDAC progression in ABX-treated KPPC;Trem2^−/− mice were accompanied by increased infiltration of CD8⁺ T cells and reduced infiltration of IL-1β⁺ macrophages into the tumors (Supplementary Figure 11F and 11G).

To explore the possibility whether TREM2 depletion may affect cancer cells, we established primary cancer cell lines from pancreatic tumors of KPPC and KPPC;Trem2^−/− mice. We observed no significant difference in cell morphology, proliferation rate, or basal/classical subtype (Supplementary Figure 12A–12D). Also, KPPC and KPPC;Trem2^−/− cancer cells exhibited similar impact on co-cultured macrophages (Supplementary Figure 12E). In comparison, orthotopically implanted LSL-Kras^G12D/+;Trp53^loxP/loxP;Pdx1-Cre;Luciferase (KPPC-Luc) cancer cells formed pancreatic tumors in a more aggressive manner in Trem2^−/− mice than in WT control mice (Supplementary Figure 12F–12L).

TREM2 depletion elicits pathogenic inflammation in both pancreatitis and PDAC

Inflammation and pancreatitis intricately associate with PDAC initiation and progression^{28, 32–34}. However, the role of TREM2 in pancreatitis remains unclear. To examine the clinical relevance of TREM2 in human pancreatitis, we firstly analyzed a recently published scRNA-seq dataset³⁵ of human chronic pancreatitis as compared to normal control pancreatic tissues (Figure 7A). Interestingly, TREM2 and APOE expression levels significantly decreased in macrophages of chronic pancreatitis tissues, accompanied by profound upregulation of IL1B (Figure 7B and 7C). Moreover, normal control pancreatic tissues predominantly harbored the TREM2^HighIL1B^Low subtype of macrophages, in contrast to chronic pancreatitis tissues harboring mostly TREM2^LowIL1B^High pro-inflammatory macrophage subtype (Figure 7B). In addition, chronic pancreatitis tissues displayed elevated expression levels of HIF1A, NLRP3, and NFKB1 (Figure 7C and 7D), analogous to our prior observations on TREM2-depletion pancreatic tumors from KPPC;Trem2^−/− mice. These results demonstrated that TREM2 deficiency is associated with exacerbated pathogenic inflammation in human pancreatitis, featured by IL-1β upregulation, which further supports the essential anti-inflammatory function of TREM2 in both pancreatitis and PDAC.

Figure 7. TREM2 deficiency is associated with aggravated pancreatitis and inflammation-induced PDAC development.

(A-D) Analysis of a recently published scRNA-seq dataset of human chronic pancreatitis and non-diseased pancreatic tissues (GEO: GSE165045). Violin plot for the expression profiles of TREM2 and APOE in various cell populations was shown in (A). Expression profiles of TREM2 and IL1B in macrophages, compared across non-diseased control, hereditary chronic pancreatitis, and idiopathic chronic pancreatitis groups, were shown in UMAP plot (B) and violin plot (C). TREM2^High and TREM2^Low macrophage subpopulations were defined in UMAP plot, with indicated signature gene expression profiles shown in dot plot (D).

(E) Genetic strategy to genetically deplete Trem2 in the context of Kras^G12D-driven autochthonous disease development using the LSL-Kras^G12D/+;Pdx1-Cre;Trem2^−/− (KC;Trem2^−/−) mouse model.

(F) Schematic representation of pancreatitis induction in KC and KC;Trem2^−/− mice at the age of 10 weeks (n = 5/group). Each arrow indicates two intraperitoneal injections of Cerulein at a 4-hour interval (50 μg/kg) on each indicated day.

(G-I) Quantitative histology evaluation (G), representative H&E staining images (H), and representative CK19 immunohistochemistry staining images (I) on the pancreatic tissues of KC and KC;Trem2^−/− mice treated with or without Cerulein as shown in (F). CK19 staining positivity quantification was shown with Student’s t test. Scale bar: 100 μm.

***P < .001; ns: not significant.

To investigate the functional role of TREM2 in pancreatitis-associated PDAC development, we utilized an additional transgenic mouse model, LSL-Kras^G12D/+;Pdx1-Cre;Trem2^−/− (KC;Trem2^−/−), with TREM2 depletion in slow-progressing KC background (Figure 7E). KC;Trem2^−/− mice and KC control mice were injected with Cerulein to induce pancreatitis (Figure 7F), which could accelerate PanIN/PDAC development due to the combined effects of both inflammation and Kras^G12D mutation. KC;Trem2^−/− mice upon Cerulein injections exhibited significantly larger PanIN and PDAC areas than KC control mice with identical Cerulein injections (Figure 7G–7I). Furthermore, the accelerated PDAC development in Cerulein-treated KC;Trem2^−/− mice was associated with enhanced recruitment of macrophages with IL-1β upregulation, as compared to Cerulein-treated KC mice (Supplementary Figure 13A–13C). These results indicated that the elevated inflammation following TREM2 depletion functionally contributes to the pancreatitis-accelerated PDAC development.

In addition, we examined Trem2^−/− mice and wild-type mice with L-Arginine-induced pancreatitis, in the absence of oncogenic Kras^G12D mutation or PDAC initiation (Supplementary Figure 13D). TREM2 depletion significantly aggravated L-Arginine-induced pancreatitis in Trem2^−/− mice, as compared to wild-type mice with identical L-Arginine injections (Supplementary Figure 13E). Consistently, the aggravated pancreatitis in L-Arginine-treated Trem2^−/− mice was associated with enhanced recruitment of macrophages and IL-1β upregulation, as compared with L-Arginine-treated wild-type mice (Supplementary Figure 13F and 13G). Our observations on the accelerated inflammation and elevated IL-1β levels in TREM2-depleted mouse models are consistent with previous observations showing that IL-1β promotes pancreatitis, which contributes to PDAC development²⁹.

Taken together, our findings demonstrate that TREM2 depletion significantly increases IL-1β⁺ macrophages and exacerbates pathogenic inflammation, thereby fueling PDAC development (Supplementary Figure 13H).

Discussion

The tumor immune microenvironment of PDAC is characterized by a significant enrichment of myeloid cells or tumor-associated macrophages³⁶. Over recent years, TREM2, prominently expressed by myeloid cells or macrophages in various cancer types, has garnered considerable attention as a potential therapeutic target^6–11. However, conflicting results have emerged across different cancer types. In models of sarcoma, colorectal, and lung cancers, depleting TREM2 has shown promise in restoring the anti-tumor immunity and enhancing the efficacy of immunotherapy^{7, 8, 11}. Conversely, in the context of hepatocellular carcinoma and glioblastoma, TREM2 depletion has been shown to promote tumor progression^{12, 13}. These observations are in concordance with our recent studies supporting the functional heterogeneity of various stromal components with either tumor-promoting or tumor-restraining functions^2–5.

Our findings revealed that TREM2 depletion elicits pathogenic inflammation and enrichment of proinflammatory IL-1β⁺ macrophages, resulting in accelerated tumor development. The IL-1β-driven pathogenic inflammation fuels PDAC progression but can be reversed by the treatment of either IL-1β inhibition or antibiotics-mediated microbiome ablation. More importantly, inhibiting IL-1β or ablating the microbiome not only reversed the accelerated PDAC progression by TREM2 depletion, but also further suppressed PDAC progression in the TREM2-depleted context. Nevertheless, additional pre-clinical and clinical studies are still required to carefully examine the outcomes of therapeutic strategies targeting TREM2, IL-1β, and/or microbiome.

Our results are consistent with previous studies showing the elevated activation level of NLRP3 inflammasome pathway in Trem2^−/− macrophages following microbial infection or LPS stimulation^{37, 38}. Furthermore, our findings highlight the significant connection between pathogenic inflammation upon TREM2 depletion and intra-tumoral microbiome-derived LPS in PDAC. Recent studies by our group³ and others^{39, 40} have underscored the functional role of intra-tumoral microbiome in PDAC, which is also the predominant source of LPS to activate the pro-inflammatory macrophages.

The contributions of tumor microbiome^{39, 41} and IL-1β pro-inflammatory pathway^27–29 to PDAC development have been reported by previous studies. This study suggests that tumor microbiome may activate IL-1β⁺ macrophages to promote pathogenic inflammation, which is negatively regulated by TREM2. Our results support the potential therapeutic implications of either microbiome ablation with antibiotics, or anti-inflammatory drugs targeting the IL-1β pathway with Diacerein/Canakinumab/Rilonacept, to mitigate inflammation in PDAC. Nevertheless, additional pre-clinical investigations are still needed.

PDAC is characterized by a complex tumor microenvironment. While immunosuppression fosters a permissive environment for PDAC progression, inflammation can exacerbate tissue damage and orchestrate with oncogenic driver mutations to propel PDAC development^{28, 29, 32, 33, 42}. This present study suggests that TREM2 depletion in PDAC may induce a pro-inflammatory state featured by enriched IL-1β⁺ macrophages, which can promote PDAC development. Thus, the pathogenesis of PDAC involves dysregulation of co-existing inflammatory signals and immunosuppressive signals, necessitating further investigation into the intricate mechanisms and therapeutic implications. These IL-1β⁺ macrophages are not equivalent to conventionally defined ‘M1-like’ anti-tumor macrophages. Instead, these IL-1β⁺ macrophages promote pathogenic inflammation and tumorigenesis, consistent with recent findings²⁸. As a master regulator of inflammation, IL-1β can profoundly impact the immune microenvironment even with a marginal increase in IL-1β levels^{28, 29}.

Our findings support the notion that TREM2 in macrophages may have distinct roles with either tumor-promoting or tumor-restraining functions on a cancer type-specific basis, necessitating the need of discerning the heterogeneous functions of TREM2 in different cancer types. Future pre-clinical and clinical studies on targeting TREM2 in PDAC and other cancer types may need to take into consideration the potential elevation of inflammation, particularly induced by IL-1β or other pro-inflammatory factors. In addition to the alterations in macrophages, we also observed a noticeable decrease of B cells in TREM2-depleted tumors. The functional influence of B cell alterations is still unclear, especially considering the distinct roles of B cells in PDAC shown in recent studies^{43, 44}. Further studies are needed to clarify the impact of B cell alterations in our model system.

In summary, this study utilizes TREM2-deficient transgenic mouse models of PDAC, combined with scRNA-seq analysis, to unravel the pivotal role of TREM2 in modulating the immune microenvironment and restraining PDAC development. These findings provide insights into the complex tumor immune microenvironment of PDAC, with implications for potential therapeutic interventions aimed at more efficiently targeting this disease.

Data availability

The data that support the findings of this study are available from the corresponding author on request.