Disrupting AGR2/IGF1 paracrine and reciprocal signaling for pancreatic cancer therapy

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is highly aggressive and characterized by pronounced desmoplasia. PDAC cells communicate with cancer-associated fibroblasts (CAFs) in a paracrine/reciprocal manner, substantially promoting tumor growth and desmoplastic responses. This study highlights the critical role of anterior gradient 2 (AGR2), an endoplasmic reticulum protein disulfide isomerase, secreted by PDAC cells to activate CAFs via the Wnt signaling pathway. Activated CAFs, in turn, secrete insulin-like growth factor 1 (IGF1), which enhances AGR2 expression and secretion in PDAC cells through the IGF1 receptor (IGF1R)/c-JUN axis. Within PDAC cells, AGR2 acts as a thioredoxin, aiding the folding and cell surface presentation of IGF1R, essential for PDAC's response to CAF-derived IGF1. This reciprocal AGR2/IGF1 signaling loop intensifies desmoplasia, immunosuppression, and tumorigenesis, creating a harmful feedback loop. Targeting both pathways disrupts this interaction, reduces desmoplasia, and restores anti-tumor immunity in preclinical models, offering a promising therapeutic strategy against PDAC.

Keywords: AGR2; IGF1; IGF1R; desmoplastic reaction; immunosuppression; molecular targeting; pancreatic cancer.

Graphical abstract

Figure 1

Secreted AGR2 promotes pancreatic carcinogenesis by activating CAFs (A) Representative immunohistochemical (IHC) images demonstrate AGR2-positive and -negative PDAC tumors (scale bars: 50 μm). (B) Survival curves of patients categorized by AGR2 expression in PDAC tumor samples via IHC (n = 99). (C) Survival curves of patients stratified according to median AGR2 levels, measured by ELISA, in serum from individuals with PDAC (n = 172). (D) Western blot analyses of AGR2 expression in human PDAC cell lines (Capan2 and Panc1) following CRISPR-Cas9-mediated AGR2 knockout (performed in triplicate). (E) Images and volumes of control versus AGR2-knockout subcutaneous xenografts derived from Capan2 and Panc1 cell lines in nude mice (n = 5 per group). (F) Subcutaneous xenografts from AGR2-knockout Capan2 and Panc1 cells (AGR2^KO), following re-expression of wild-type AGR2 (AGR2^WT), AGR2 lacking a nuclear localization signal (AGR2^ΔNLS), and AGR2 lacking a signal peptide (AGR2^ΔSP) (n = 4 per group). (G) ELISA analyses of the supernatant from AGR2-knockout Capan2 and Panc1 cells re-expressing AGR2^WT, AGR2^ΔNLS, and AGR2^ΔSP (performed in triplicate). (H) Representative IHC images and quantification of alpha-smooth muscle actin (α-SMA) staining score and collagen score within xenograft tumors across the four groups (n = 4 per group, scale bars: 50 μm). Statistical significance was determined using a log rank test for (B) and (C) and a one-way ANOVA with multiple comparisons test for (E) through (H). Data are presented as mean ± standard deviation (SD). Significance levels are indicated as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 2

Agr2 secretion correlates with desmoplastic reaction in genetic mouse models of PDAC (A) Schematic illustration of the genotypes for KC mice, KC; Agr2^−/− mice, and KC; Agr2^OE mice. (B) Representative hematoxylin and eosin (H&E), Agr2, Krt19, and α-amylase-stained sections, along with α-SMA/BrdU-stained immunofluorescence images of pancreata from KC, KC; Agr2^−/−, and KC; Agr2^OE mice (scale bars: 50 μm; n = 12 mice per group). The images were scored and thereby quantified (right). (C) ELISA analysis of Agr2 levels in the serum of 8-week-old KC, KC; Agr2^−/− and KC; Agr2^OE mice (n = 12 mice per group). (D) Western blot analysis of α-SMA, α-amylase, and Agr2 expression in the pancreata of KC, KC; Agr2^−/−, and KC; Agr2^OE mice (n = 4 mice per group). (E) Schematic representation of KC mice injected with AAV-EGFP, AAV-Agr2^WT, and AAV-Agr2^ΔSP particles. (F) Representative images of H&E staining, EGFP immunofluorescence, and IHC for Krt19, α-amylase, and α-SMA in the pancreata of KC mice injected with AAV-EGFP, AAV-Agr2^WT, and AAV-Agr2^ΔSP particles (scale bars: 50 μm; n = 3 mice per group). The images were scored and thereby quantified (right). (G) ELISA analysis of Agr2 levels in the serum of KC mice injected with AAV-EGFP, AAV-Agr2^WT, and AAV-Agr2^ΔSP particles. (H) Western blot analysis of α-SMA, α-amylase, Agr2, and EGFP expression in the pancreata of KC mice injected with AAV-EGFP, AAV-Agr2^WT, and AAV-Agr2^ΔSP particles. (I and J) Schematic representation and survival curves for KC and KC; Agr2^−/− mice over a 1.5-year follow-up period. (K) PDAC incidence in KC (11/34, 32.4%) versus KC; Agr2^−/− mice (5/40, 12.5%). (L) ELISA analysis of Agr2 levels in the serum of KC and KC; Agr2^−/− mice with PDAC (n = 5 mice per group). (M) Representative H&E-stained sections showing PDAC tumors in KC and KC; Agr2^−/− mice; Sirius red-stained sections showing collagen distribution in tumors; α-SMA/BrdU-stained immunofluorescence images depicting proliferative α-SMA-positive cells in tumors (scale bars: 50 μm; n = 5 mice per group). (N) Western blot analysis of α-SMA and Agr2 expression in tumors from KC mice and KC; Agr2^−/− mice (n = 5 mice per group). Statistical significance for (B), (C), (G), and (F) was assessed using a one-way ANOVA with multiple comparisons test, (J) with a log rank test, (K) with a chi-squared test, and (L) and (M) with two-tailed, unpaired Student’s t tests. Data are presented as mean ± SD. Significance is denoted as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. “ns” indicates no significance.

Figure 3

IGF1 promotes the secretion of AGR2, which in turn enhances the presentation of the IGF1R on the cell surface (A) The Venn diagram of the upper panel illustrates the count of genes down-regulated in Capan2 and Panc1 cells post AGR2 knockout. The Gene Ontology (GO) analysis of the lower panel identifies enriched biological processes, notably “regulation of IGF receptor signaling pathway” at transcriptional levels, after AGR2 knockout in these cell lines. (B) Western blot and quantitative reverse-transcription PCR (qRT-PCR) analyses assess IGF1R and AGR2 expression in Capan2 and Panc1 cells following AGR2 knockout via the CRISPR-Cas9 system (performed in triplicate). (C) Flow cytometry (FACS) quantifies cell membrane surface expression of IGF1R in Capan2 and Panc1 cells after AGR2 knockout (performed in triplicate). (D) Co-immunoprecipitation assays reveal AGR2’s interaction with pro-IGF1R in Capan2 and Panc1 cells (performed in triplicate). (E) Immunofluorescence imaging displays AGR2 and IGF1R distribution and ER labeling in Panc1 cells (scale bars: 50 μm). (F) Western blot analysis of IGF1R and AGR2 in Panc1 and Capan2 cells with controls (original cell lines), AGR2-knockout (AGR2^KO) post-expression of AGR2^WT, AGR2^ΔNLS, AGR2^ΔSP, and AGR2^C81A mutation (performed in triplicate). (G) Western blot analysis of IGF1R expression in AGR2 in Panc1 and Capan2 cells with AGR2 knockout (KO) treated with 3-methyladenine (3-MA) (15 mM), bafilomycin A1 (30 nM), chloroquine (20 mM), MLN4929 (1 mM), or MG132 (5 mM) for 12 h (performed in triplicate). (H) Western blot analysis of IGF1R, phosphorylated IGF1R, c-JUN, phosphorylated c-JUN, and AGR2 following 12 h of serum starvation and subsequent IGF1 stimulation (50 ng/mL, performed in triplicate). (I) ELISA measures AGR2 secretion after serum starvation and treatment with PPP (1 μM) and IGF1 (50 ng/mL) over time (performed in triplicate). (J) Identification of potential c-JUN-binding sites within the AGR2 promoter region. (K) Western blot analysis of c-JUN, phosphorylated c-JUN, and AGR2 expression following c-JUN knockdown and IGF1 stimulation over time (performed in triplicate). (L) Western blot shows c-JUN, phosphorylated c-JUN, and AGR2 expression post anisomycin treatment over time (performed in triplicate). (M) Chromatin immunoprecipitation followed by quantitative PCR (ChIP-qPCR) demonstrates c-JUN enrichment at AGR2’s transcription start sites (TSSs) before and after IGF1 treatment (performed in triplicate). (N) Integrative Genomics Viewer (IGV) tracks display c-JUN peaks in AGR2’s promoter region post IGF1 treatment. (O) Dual-luciferase reporter assays in Capan2 and Panc1 cells evaluate AGR2 promoter activity under various lengths and site-specific mutations after IGF1 treatment (performed in triplicate). Statistical analyses: (B) and (C) used a one-way ANOVA with multiple comparisons. (I), (M), (N), and (O) were analyzed using two-tailed, unpaired Student’s t tests. Data are presented as mean ± SD, with significance marked as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and “ns” indicates no significance.

Figure 4

Secreted AGR2 promotes IGF1 production from CAFs via the Wnt/β-catenin pathway (A) Western blot analysis evaluates AGR2 and IGF1 levels in three human PDAC-derived CAFs and three PDAC cell lines (Capan2, HPAC, and Panc1) across three independent experiments. (B) qRT-PCR analysis of IGF1 expression and supernatant ELISA analyses of IGF1 secretion in human PDAC-derived CAFs co-cultured with two human PDAC organoids and with or without treatment with Agr2-neutralizing antibody (5 μg/mL) for 48 h (n = 3 independent experiments). (C) qRT-PCR assesses IGF1 expression in PDAC-derived CAFs co-cultured with AGR2-knockout Capan2 and Panc1 cells, following re-expression of AGR2^WT, AGR2^ΔNLS, and AGR2^ΔSP for 48 h (upper); Western blot analysis investigates IGF1R, phosphorylated IGF1R, c-JUN, phosphorylated c-JUN, and AGR2 levels in AGR2-knockout Capan2 and Panc1 cells after co-culture with PDAC-derived CAFs (lower, n = 3 independent experiments). (D) qRT-PCR explores IGF1 expression in two PDAC-derived CAFs after treatment with rAGR2 (500 ng/mL), rTGF-β1 (4 μg/mL), and rIL-1α (200 ng/mL) for 24 h (n = 3 independent experiments). (E) Supernatant analysis quantifies collagen levels in two PDAC-derived CAFs following rAGR2 treatment (500 ng/mL) for 24 h (n = 3 independent experiments). (F) Transwell assays examine cell migration in two PDAC-derived CAFs following rAGR2 treatment (500 ng/mL) for 24 h (n = 3 independent experiments). (G) Left: scRNA-seq identifies iCAFs and myCAFs within 16 PDAC tissues (GEO: GSE155698), showing iCAFs with elevated IGF1 expression (>mean value). Right: volcano plot displays genes differentially expressed between IGF1^high and IGF1^low CAFs (FDR < 0.01; log₂FC > 0.5), accompanied by KEGG pathway analysis of the IGF1-CAF signature. (H) Principal component analysis (PCA) of transcriptomic data from CAFs treated with rAGR2, rTGF-β1, and rIL-1α (n = 3 per group). (I) A heatmap shows genes significantly upregulated in CAFs after treatment with rAGR2, rTGF-β1, and rIL-1α (FDR < 0.01; log2FC > 0.5; left). Bioplant pathway analysis elucidates upregulated gene pathways post rAGR2 treatment in CAFs (right). (J) Identification of potential lymphoid enhancer binding factor 1 (LEF1)-binding sites within the IGF1 promoter region. (K) Western blot analysis shows β-catenin expression in both nuclear and cytoplasmic fractions of PDAC-derived CAFs after AGR2 stimulation (500 ng/mL) for 0.5, 1, 3, and 6 h (n = 3 independent experiments). (L) Western blot and qRT-PCR analyses evaluate β-catenin and IGF1 levels in PDAC-derived CAFs post β-catenin knockdown or following treatment with ICG-001 (Wnt pathway inhibitor) and rAGR2 (500 ng/mL) for 24 h (M) Luciferase reporter assays in three PDAC-derived CAFs transfected with wild-type and site-specific mutagenized IGF1 promoter sequences based on (J) predictions, post rAGR2 treatment (500 ng/mL) for 24 h (n = 3 independent experiments). Statistical analysis: one-way ANOVA with multiple comparisons test was used for (B), (C), (D), and (L); two-tailed, unpaired Student’s t tests were employed for (E), (F), and (M). Data are presented as mean ± SD, with ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 indicating levels of statistical significance.

Figure 5

High serum levels of AGR2 and IGF1 are associated with enhanced desmoplastic reactions and immunosuppression in PDAC (A) ELISA analysis of Igf1 in serum from 8-week-old KC mice, KC; Agr2^−/− mice, and KC; Agr2^OE mice reveals a significant difference (left, n = 12 mice per group). Comparison between KC mice and KC; Agr2^−/− mice with PDAC also shows marked differences (right, n = 5 mice per group). (B) Serum levels of AGR2 and IGF1 exhibit a correlation in 145 human patients with PDAC, analyzed using Pearson’s correlation coefficient. (C) IHC images display α-SMA, podoplanin (PDPN), collagen, and IL-6 positivity in tumor areas, comparing AGR2^high; IGF1^high samples with AGR2^low; IGF1^low samples, demonstrating a difference in desmoplastic reaction. (D) IHC images illustrate the differential presence of CD3, CD8, CD4, FOXP3, CD68, CD206, and CD20-positive cells in tumors between AGR2^high; IGF1^high samples and AGR2^low; IGF1^low samples, indicating variations in immune cell infiltration (scale bars: 50 μm). (E) IHC imaging further reveals the distribution of Cd3, Cd8, Cd4, Foxp3, B220, F4/80, and Cd206-positive cells in tumors from KC mice versus KC; Agr2^−/− mice, emphasizing differences in immunological responses (scale bars: 50 μm). p values in left of (A) was calculated using a one-way ANOVA with a multiple comparisons test, p values in right of (A), (C), (D), and (E) were calculated using two-tailed, unpaired Student’s t tests, and correlation coefficient in (B) was calculated using Pearson’s correlation coefficient. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. ns, no significance.

Figure 6

Combined targeting attenuates desmoplastic reaction and normalizes immunosuppressive microenvironment (A) Western blot analysis reveals Agr2 and Igf1 levels in PSCs isolated from wild-type mice and three mouse PDAC cell lines, highlighting the differential expression patterns. (B) Schematic diagram shows the therapeutic strategy of combining IGF1R inhibitor and AGR2-neutralizing antibody. (C) ELISA and qRT-PCR analyses demonstrate Igf1 levels in PSCs co-cultured with KPC PDAC-derived organoids. The impact of treatments with the IGF1R inhibitor (PPP; 1 μM), Agr2-neutralizing antibody (5 μg/mL) alone, or their combination for 48 h is shown (n = 3 independent experiments). (D) Western blot results display the expression levels of p-Igf1r, Igf1r, c-Jun, p-c-Jun, Akt, p-Akt, Erk, p-Erk, and Agr2 in mouse PDAC-derived organoids after co-culture with PSC cells and subsequent treatments as mentioned in (C) (n = 3 independent experiments). (E) Representative images and quantitative analyses show the growth dynamics of PDAC organoids co-cultured with PSC cells under various treatment conditions over 0, 24, 48, and 96 h (scale bars: 50 μm, n = 3 independent experiments). (F) Tumor volume comparisons in KPC mice post caerulein-induced acute pancreatitis and subsequent treatments with Agr2 antibody (4 mg/kg; intraperitoneally [i.p.], three times per week for 2 weeks), PPP (20 mg/kg; i.p., three times per week for 2 weeks), or their combination (n = 5 for control group, n = 3 for single treatment groups, and n = 5 for combined treatment group). (G) ELISA quantification of Agr2, Igf1, Il-1α, Lif, GM-CSF, and Il-6 in serum samples from the four groups of KPC mice underscores the systemic effects of the treatment modalities on cytokine levels (n = 5 for control group, n = 3 for single treatment groups, and n = 5 for combined treatment group). (H and I) Representative stained sections and quantitative statistics of H&E, Pdpn, α-SMA, collagen, Cd3, Cd4, Foxp3, B220, and Cd206-positive cells within PDAC tumors (scale bars: 50 μm, n = 5 mice per group). (J) Representative IHC highlights CD8-positive cells in lymph nodes adjacent to the tumors (scale bars: 50 μm). p values in (C), (F), and (G) were calculated using a one-way ANOVA with a multiple comparisons test, and p values in (H) and (I) were calculated using two-tailed, unpaired Student’s t tests. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.