Exosomes define a local and systemic communication network in healthy pancreas and pancreatic ductal adenocarcinoma

Abstract

Pancreatic ductal adenocarcinoma (PDAC), a lethal disease, requires a grasp of its biology for effective therapies. Exosomes, implicated in cancer, are poorly understood in living systems. Here we use the genetically engineered mouse model (ExoBow) to map the spatiotemporal distribution of exosomes from healthy and PDAC pancreas in vivo to determine their biological significance. We show that, within the PDAC microenvironment, cancer cells establish preferential communication routes through exosomes with cancer associated fibroblasts and endothelial cells. The latter being a conserved event in the healthy pancreas. Inhibiting exosomes secretion in both scenarios enhances angiogenesis, underscoring their contribution to vascularization and to cancer. Inter-organ communication is significantly increased in PDAC with specific organs as most frequent targets of exosomes communication occurring in health with the thymus, bone-marrow, brain, and intestines, and in PDAC with the kidneys, lungs and thymus. In sum, we find that exosomes mediate an organized intra- and inter- pancreas communication network with modulatory effects in vivo.

Fig. 1. The ExoBow transgene efficiently labels pancreas cells and its derived exosomes.

a The ExoBow construct is inserted in the intron one of ROSA 26 (R26) locus and is under the action of a strong synthetic promotor (CAG). Upstream of the exosomal marker CD63 mouse gene there is a neomycin resistance cassette with a stop-codon flanked by Frt sites that prevents further transcription. Following CD63 there are 4 fluorescent reporters: mCherry, phiYFP, eGFP and mTFP, each with a polyA sequence. b Schematic of the monoreporter mouse model Panc-CD63-mCherry in which Pdx1 drives the expression of the Flp recombinase. Pancreas imaging using IVIS Lumina System illustrating CD63-mCherry expression (535 excitation laser and DsRed emission filter). Pancreas of control (R26^CD63-XFP/+, no recombinases, left) and Panc-CD63-mCherry mice (right; experiment repeated with a total of 6 mice). c Confocal microscopy images of a maximum projection of a Panc-CD63-mCherry pancreas section depicting exocrine and endocrine CD63-mCherry positive cells. Immunofluorescence against mCherry (red; experiment repeated with a total of 3 mice). d Schematic representation of the isolation of interstitial EVs from the pancreas tissue according to Crescitelli et al.. Anti-mCherry western-blot in small and large EVs fractions isolated from pancreas tissue of wild-type (WT, control) or Panc-CD63-mCherry mice (experiment repeated with a total of 3 mice). e Schematic representation of the multireporter mouse model, Panc-ExoBow, in which both Flp and Cre recombinases are under the control of Pdx1 promoter. Pancreas imaging using IVIS Lumina System illustrating CD63-eGFP, CD63-phiYFP, and CD63-mTFP (465 excitation laser and GFP emission filter). Pancreas of control (R26^CD63-XFP/+, left) and Panc-CD63-mCherry mice (right; experiment repeated with a total of 5 mice). f Confocal images of a maximum projection of a Panc-ExoBow pancreas section depicting CD63-mTFP, CD63-phiYFP and CD63-eGFP positive cells. Immunofluorescence for mTFP (cyan), phiYFP (yellow) and eGFP (green; experiment repeated with a total of 3 mice). g Schematic representation of the isolation of interstitial EVs from the pancreas tissue according to Crescitelli et al.. Anti-mTFP, anti-phiYFP and anti-eGFP western-blot in small and large EVs fractions isolated from pancreas tissue of WT or Panc-ExoBow mice (experiment repeated with a total of 2 mice). h Schematic representation of the PDAC monoreporter mouse model, KPF CD63-mCherry. Pancreas images of control (no ExoBow transgene, left) and KPF CD63-mCherry mice (right; experimented repeated with a total of 4 mice). i Confocal microscopy images depicting CD63-mCherry positive cancer cells in the pancreas of a KPF CD63-mCherry mouse. Immunofluorescence against mCherry (red; experiment repeated with a total of 3 mice). j Schematic representation of the PDAC multireporter mouse model, KPC-ExoBow. Pancreas images of control (no ExoBow transgene, left) and KPC-ExoBow mice (right; experiment repeated with a total of 4 mice). k Confocal microscopy images depicting CD63-mTFP and CD63-eGFP positive cells in the pancreas of a KPC-ExoBow mouse. Immunofluorescence against mTFP (cyan), and eGFP (green; experiment repeated with a total of 2 mice). Mice age in b, c, e, f is of 8 weeks, d and g between 8–1 weeks, h i 16.3 weeks, and j k 17 weeks. In western-blots Ponceau S as loading control and 25 μg of protein samples was used. In all images nuclei are counterstained with hoechst (blue) and scale bars are 20 μm. Source data are provided as a Source Data file. Schemes created with BioRender.com.

Fig. 2. Exosomes mediate intra-pancreas communication in PDAC and CAFs spatial distribution.

a Dot plot representing the percentage of cancer-associated fibroblasts (CAFs; CD140A⁺), endothelial cells (CD31⁺) and immune cells (CD45⁺) that received PDAC CD63⁺ Exos in tumors PDAC reporter mice analyzed by flow cytometry (n = 5 biologically independent animals). b Representative confocal microscopy images of PDAC Exos (green) accumulation in CAFs (alpha-smooth muscle actin –αSMA—in red). Nuclei were counterstained with hoechst (blue). Scale bar 5 μm. Experiment repeated with a total of 3 mice. c Dot plot representing the fluorescence intensity of Rab27a in different lesions (n = 43) of a KPC tumor, with representative confocal microscopy images of regions with Rab27a low or Rab27a high PDAC lesions. EpCAM (cancer cells) in magenta, Rab27a in green and nuclei were counterstained with hoechst (blue). Scale bar 20 μm. Dashed line represents the median of Rab27a intensity. d Linear regression of αSMA and Rab27a per PDAC lesion (left) and circular radial profile of the αSMA fluorescence intensity over centered Rab27a high or low PDAC lesions in two KPC tumors (upper (n = 43 PDAC lesions) and lower (n = 31 PDAC lesions) left) with representative confocal microscopy images of a Rab27a^Low/αSMA^High and Rab27aHigh/αSMA^Low. Rab27a in green, αSMA in red and EpCAM in magenta, nuclei were counterstained with hoechst (blue). Scale bar 10μm. Dashed lines represent the median radius fitted to the manual PDAC lesions’ ROI. Kolmogorov-Smirnov, ****p < 0.0001. Data are Mean ± SEM. Source data are provided as a Source Data file.

Fig. 3. Pancreas exosomes mediate local communication and constrain angiogenesis.

a Dot plot representing the percentage of endothelial cells (CD31⁺) that received pancreas-derived CD63⁺ Exos in health (Panc-CD63-mCherry, n = 8) and in PDAC (KPF CD63-mCherry n = 2 and KPC-ExoBow n = 3) analyzed by flow cytometry. b Representative confocal microscopy images of pancreas-derived Exos (green) accumulation in endothelial cells (CD31⁺ in red) in healthy pancreas of Panc-CD63-mCherry mice (upper panel) or tumors of KPF CD63-mCherry mice. Nuclei were counterstained with hoechst (blue). Scale bar 5 μm. c Dot plot representing the percentage of endothelial cells (CD31⁺) in the pancreas microenvironment in health (Panc-CD63-mCherry, n = 8) and in PDAC (KPF CD63-mCherry n = 2 and KPC-ExoBow n = 3) analyzed by flow cytometry. d Schematic representation of the healthy Rab27a KO GEMM in which pancreas cells have impaired secretion of exosomes. e Representative CD31 IHC images (10 x, left) and respective quantification (right) in the pancreas of wild-type (Rab27aWT, n = 6) and Pdx1 Rab27a^Frt/Frt (Rab27aKO, n = 6) mice. Two-tailed unpaired t-test, p = 0.0261. f Schematic representation of the PDAC Rab27a KO GEMM in which pancreas cells have impaired secretion of exosomes upon tamoxifen administration. The PKT Rab27a model is the control group which lacks the R26^{LSL-Flpo-ERT2} allele, hence upon tamoxifen administration expresses Rab27a and has proficient exosomes secretion. g Representative CD31 IHC images (10 x, left) and respective quantification (right) in the pancreas of control PKT Rab27a (n = 8) and PKT iRab27a (n = 6) mice. Two-tailed unpaired t-test, p = 0.0485. Data are Mean ± SEM. Source data are provided as a Source Data file. Schemes created with BioRender.com.

Fig. 4. PDAC exosomes mediate local communication with cells of the immune system.

a Dot plot representing the percentage of cells of the immune system (CD45⁺) that received pancreas-derived CD63⁺ Exos in health (Panc-CD63-mCherry, n = 8) and in PDAC (KPF CD63-mCherry n = 2 and KPC-ExoBow n = 3) analyzed by flow cytometry. b Dot plot representing the percentage of cells of the immune system (CD45⁺) in the pancreas microenvironment in health (Panc-CD63-mCherry, n = 8) and in PDAC (KPF CD63-mCherry n = 2 and KPC-ExoBow n = 3) analyzed by flow cytometry. Two-tailed Mann–Whitney test, p = 0.0186. c Dot plot representing the percentage of T cells (TCRβ⁺), cells of the monocyte lineage (CD11b⁺Ly6G/C⁻) and natural killer (CD11b⁺NK1.1⁺) cells that received PDAC CD63⁺ Exos (KPF CD63-mCherry n = 2 and KPC-ExoBow n = 3, except for monocyte-lineage n = 3 KPC-ExoBow) analyzed by flow cytometry. d Dot plot representing the percentage of T helper cells (Th, TCRβ⁺CD4⁺), cytotoxic T cells (Tc, TCRβ⁺CD4^-), and regulatory T cells (Treg, CD4⁺Foxp3⁺) that received PDAC CD63⁺ Exos (KPF CD63-mCherry n = 2 and KPC-ExoBow n = 3) analyzed by flow cytometry. e Representative confocal microscopy images of PDAC CD63⁺ Exos (green) accumulation in different subpopulations of the tumor microenvironment (red) including T helper cells, cytotoxic T cells, and regulatory T cells, cells of the monocyte lineage and natural killer cells. Nuclei were counterstained with hoechst (blue). Scale bar 5μm. Data are Mean ± SEM. Source data are provided as a Source Data file.

Fig. 5. Pancreas exosomes-mediated inter-organ communication increases throughout PDAC progression.

a Dot plot representing the average radiant efficiency fluorescence levels of CD63-mCherry⁺ Exos in the different organs in health (Panc-CD63-mCherry, n = 6) and in PDAC (KPF CD63-mCherry, n = 4). Two-tailed Mann Whitney, p = 0.0095. b Fold change of the average radiant efficiency fluorescence levels of CD63-mCherry⁺ Exos in the different organs in PDAC (KPF CD63-mCherry, n = 4) in relation to the healthy context (Panc-CD63-mCherry, n = 6). Two-tailed Mann-Whitney test, *p = 0.0119, **p = 0.0095. c Dot plot representing the average radiant efficiency fluorescence levels of PDAC CD63-mCherry⁺ Exos across different organs in KPF CD63-mCherry mice (n = 4) at time of euthanasia (left), with representative IVIS images (535 excitation laser and DsRed emission filter; right). d Dot plot representing the average radiant efficiency fluorescence levels of pancreas CD63⁺ Exos across all organs in different disease stages, healthy (Panc-CD63-mCherry, n = 6), early PDAC (KPF CD63-mCherry, n = 1 and KPC-ExoBow, n = 2) and late PDAC (KPF CD63-mCherry, n = 4). Two-tailed Mann-Whitney test, in healthy vs. early PDAC: lung p = 0.0119, kidneys p = 0.0238, heart p = 0.0238 and healthy vs. late PDAC: lung p = 0.0119, kidneys p = 0.0095. e Dot plot representing the average radiant efficiency fluorescence levels of CD63-XFP⁺ Exos present in the different organs in health (Panc-CD63-mCherry, n = 6) and in early PDAC (KPF CD63-mCherry, n = 1 and KPC-ExoBow, n = 2). Two-tailed Mann–Whitney test, p = 0.0476. f Representative confocal microscopy images of PDAC CD63⁺ Exos (green) accumulation in the lungs and kidneys of KPF CD63-mCherry mice at early (upper panel) or late (lower panel) PDAC stages. Nuclei were counterstained with hoechst (blue). Scale bar 20 μm. Data are Mean ± SEM. ING LN, inguinal lymph nodes, AXL LN axillary lymph nodes, MST LN mesenteric lymph nodes. Source data are provided as a Source Data file.

Fig. 6. PDAC exosomes are enriched in circulation and are taken up by specific cell types in the kidneys and lungs.

Representative confocal microscopy images of PDAC CD63⁺ Exos (green) accumulation in the kidneys (Megalin, Aquaporin-2 and Podoplanin positive cells in red) or lungs (Uteroglobin, Podoplanin and TTF1 positive cells in red) of KPF CD63-mCherry mice at a early PDAC stages or b late PDAC stages. Nuclei were counterstained with hoechst (blue). Scale bar 20 μm. Experiments performed in a total of 3 mice. c Nanoparticle tracking analysis of the small EVs population isolated from the pancreas of healthy (n = 9, wild-type n = 3, Panc-CD63-mCherry n = 3 and Panc-ExoBow n = 3) or PDAC mice (PKT, n = 5) according to Crescitelli et al.. Two-tailed Mann–Whitney test, p = 0.0005. Data are Mean ± SEM. d Nanoparticle tracking analysis of the exosomes found in serum of healthy (wild-type, n = 8) or PDAC mice (PKT, n = 5). Two-tailed Mann–Whitney test, p = 0.0016. Data are Mean ± SEM. e Image stream analysis of CD63-XFP in exosomes isolated from serum of mice at an early PDAC stage (CD63-mTFP, left) and a late PDAC stage (CD63-mCherry, right). Experiment repeated in a total of 2 and 3 mice, respectively. Scale bar 10 μm. Source data are provided as a Source Data file.

Fig. 7. EVs content reflect the local intratumor communication of PDAC.

a Schematic representation of the experimental approach for proteomics analysis (upper panel) and heatmap depicting the deferentially expressed proteins common across the 3 samples for wild-type (WT) or KPC small EVs (lower panel). Unsupervised hierarchical clustering showing separation of the two protein clusters WT and KPC small EVs. b Venn-diagram of total proteins detected in EVs from WT or KPC small EVs. Gene set enrichment analysis of the c Angiogenesis pathway that does not separate the WT from KPC small EVs, and the d Cell activation pathway which distinguishes WT from KPC small EVs. GSEA (Gene set enrichment analysis, in c and d) uses a ranked gene list, in our case, sign(fold change gene)⋅−log10(P), encompassing the differential expression between two conditions (KPC vs WT), and the Kolmogorov-Smirnov statistic to score the enrichment of a priori defined set of genes that share common biological function. Significance of the score is evaluated using an empirical permutation test correcting for multiple hypothesis testing. e Top 15 enriched reactome pathways in KPC small EVs in comparison to WT small EVs. MBP—Macromolecule Biosynthetic process. f Schematic representation of the experimental approach for proteomics analysis (upper panel) and heatmap depicting the deferentially expressed genes of WT or KPC small EVs following RNA Seq analysis (lower panel). g Volcano plot representing the downregulated and upregulated genes in KPC small EVs in comparison to WT small EVs using DESeq2. DESeq2 uses negative binomial generalized linear models for the differential analysis of count data and uses the Wald test with multiple correction (Benjamini–Hochberg method) for significance testing. Shrinkage of log2FC estimates to control for small sample sizes and low read counts was done by the apeglm method. h Top 15 enriched reactome pathways in the upregulated genes of KPC small EVs, common to KPC cell line exosomes in comparison to WT small EVs. Cellular response to oxygen-containing compound (O₂-CC); Cellular protein-containing complex assembly (CC). i Heatmap depicting the Log2FoldChange > 1 in RNA Seq analysis of cancer associated fibroblasts (CAFs) or endothelial cells (bEnd.3) exposed to cancer exosomes or not (control). j Top 20 enriched reactome pathways in the upregulated genes of CAFs upon cancer exosomes exposure. k Top 20 enriched reactome pathways in the downregulated genes of bEnd.3 upon cancer exosomes exposure. MOP Multicellular Organism Process, ST Signal Transduction, SP Signaling Pathway, SF Structure Formation. Over-Representation (ORA) analysis in (e, h, j and k) employs a hypergeometric test, corrected for multitple testing using Benjamini–Hochberg method, to determine the statistical significance of the up or down-regulated DEGs in each GO term. Circos plot depicting the interaction between the KPC small EVs RNA (red) and protein (blue) cargo with the altered genes upon cancer exosomes exposure in l CAFs and m endothelial cells (bEnd.3). In all cases are represented the upregulated differentially expressed genes or proteins. n Number of entries for RNA or protein identified in KPC small EVs that are present in the upregulated differentially expressed genes of CAFs or endothelial cells exposed to cancer exosomes. Source data are provided as a Source Data file. Schemes created with BioRender.com.

Fig. 8. Schematic of the intra- and inter- pancreas communication established by CD63 positive exosomes in PDAC.

The ExoBow mouse model can be crossed with specific Flp and Cre lines driven by the Pdx1-pancreas promoter which will render the conditional expression of CD63-XFP reporter proteins by pancreas cells. Crossing this model with well-stablished PDAC genetically engineered mouse models, enables the assessment of CD63 exosomes biodistribution locally and systemically in both healthy and PDAC contexts. Our work demonstrates an intra-pancreas connectome mediated by pancreas-derived exosomes mainly with CAFs (in a PDAC-specific context), followed by endothelial cells and, in lower amounts, with cells of the immune system. We were also able to demonstrate that the inter-pancreas connectome mediated by pancreas-derived exosomes varies from physiological conditions to a cancer context. Furthermore, we show that these routes of communication also vary along PDAC progression. PDAC, pancreatic ductal adenocarcinoma; CAFs, cancer-associated fibroblasts. Schemes created with BioRender.com.