Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Jun 4.
Published in final edited form as: Pancreas. 2024 Jun 4;53(10):e808–e817. doi: 10.1097/MPA.0000000000002378

Increased Gremlin1 Expression in Pancreatic Ductal Adenocarcinoma Promotes a Fibrogenic Stromal Microenvironment

Rachel R Tindall 1, Erika Y Faraoni 2, Jiajing Li 1, Yinjie Zhang 1, Shun-Ming Ting 3, Beanna Okeugo 4, Xiurong Zhao 3, Yuying Liu 4, Mamoun Younes 5, Qiang Shen 6, Jennifer M Bailey-Lundberg 2, Yanna Cao 1,*, Tien C Ko 1,*
PMCID: PMC11615151  NIHMSID: NIHMS1996436  PMID: 38829570

Abstract

Objective:

The pancreatic ductal adenocarcinoma (PDAC) microenvironment is primarily composed of cancer-associated fibroblasts (CAFs) and immune cells. Gremlin1 (Grem1) is a profibrogenic factor that promotes tumorigenesis in several cancers. However, the role of Grem1 in the PDAC microenvironment is not adequately defined.

Methods:

We correlated Grem1 levels with activated stroma and immune cells in human PDAC using The Cancer Genome Atlas (TCGA) RNA-sequencing data and characterized the expression of Grem1 transcripts and isoforms in pancreatic cell lines and PDAC tissues. We assessed the role of Grem1 in the microenvironment by in vitro studies.

Results:

Grem1 expression is associated with an activated stroma and increased M1 and M2 macrophages. Only full length Grem1 variant 1 and isoform 1 were detectable in human pancreatic cells, and remarkably high levels of Grem1 were observed in pancreatic fibroblasts (P < 0.05). Immunohistochemistry detected Grem1 protein in PDAC tumor cells and stromal cells, which correlated with infiltrating macrophages in PDAC tumors. Grem1 knockdown in CAFs suppressed transforming growth factor (TGF)-β-induced extracellular matrix proteins (P < 0.05). Grem1 recombinant protein treatment in vitro increased M1 and M2 macrophages (P < 0.05).

Conclusions:

Grem1 acts as a profibrogenic factor in the PDAC microenvironment via modulation of fibroblasts and macrophages. Grem1 may have the potential to be developed as a therapeutic target for PDAC.

Keywords: PDAC, fibrogenic, Grem1, CAFs, macrophages

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) accounts for 85% of pancreatic cancer cases and is currently the third leading cause of cancer-related death1–3. Poor outcomes are due to late diagnosis, which can be attributed to a lack of screening protocols, and a fibrotic tumor microenvironment surrounding the cancer cells creating a chemoresistant barrier4. Understanding the mechanisms involved in development of the fibrotic microenvironment is essential to identifying targets for efficient screening assays and effective therapeutics.

The microenvironment composes over 50% of tumor volume and is inhabited by a variety of cells including cancer-associated fibroblasts (CAFs) and immune cells that play dynamic roles in cancer development and progression3, 5, 6. A recent study demonstrates that a majority of CAFs are derived from resident fibroblasts in the pancreas, which originate from splanchnic mesenchyme7. CAFs can also originate from bone marrow and adipose-derived mesenchymal stem cells8, 9. These CAFs have a heterogeneous role and contribute to PDAC development via a desmoplastic reaction leading to secretion of extracellular matrix (ECM) and soluble factors6, 10, 11. A major portion of the immune cell population includes tumor-associated macrophages (TAMs). TAMs can polarize into pro-inflammatory M1 and anti-inflammatory M2 subtypes and have a dynamic status in PDAC12, 13.

Gremlin1 (Grem1), a bone morphogenetic protein (BMP) antagonist and member of the transforming growth factor (TGF)-β superfamily, is a critical gene expressed during embryogenesis14. Grem1 is overexpressed during certain pathologic conditions including benign pancreatic disease15–20. Furthermore, Grem1 activates fibrogenesis in cases of intestinal fibrosis and breast cancer and directly promotes tumorigenesis in several types of cancer17, 21–28. Grem1 is secreted at high levels by cells of mesenchymal origin in PDAC and is essential in maintaining the heterogeneity of pancreatic cancer via paracrine signaling and a self-inhibitory feedback loop with BMP29. Previously, our group has elucidated the role of Grem1 in pancreatic diseases, showing upregulated Grem1 mRNA and protein play a profibrogenic role in human and animal CP30. Using human PDAC tissue microarray samples, we also showed that Grem1 mRNA expression is primarily localized to CAFs in the stroma and that Grem1 is increased with an order from CP, to pancreatic intraepithelial neoplasm (PanIN), to PDAC cases31.

Three major transcripts and isoforms of Grem1 have previously been identified in normal human tissue32. Expression and function of these transcripts and isoforms have not previously been reported. Understanding the expression and the role of Grem1 in the PDAC tumor microenvironment would deepen our understanding of the molecular events in PDAC and may elucidate potential targets for novel therapeutics.

In this study, we showed Grem1 expression patterns in representative pancreatic cell lines and in PDAC tissue samples and investigated the effect of Grem1 on cells in the tumor microenvironment. We found that Grem1 is secreted by CAFs, mediates TGF-β-induced ECM, and stimulates macrophage activation, suggesting a promoting role of Grem1 in the development of a fibrogenic stromal microenvironment in PDAC.

MATERIALS AND METHODS

The Cancer Genome Atlas (TCGA) RNA-sequencing data analysis

Kaplan-Meier curves were generated using TCGA RNA-sequencing data (FPKM_UQ) for PDAC samples, after excluding non-PDAC samples. Grem1High and Grem1Low expression levels were stratified on the basis of average expression. These groups were determined by first identifying and removing the outliers, which were defined as expression levels less or greater than 1.5 times the interquartile range. The mean and standard error (SE) of the data set was calculated, and the data was filtered into two groups, where Grem1High was the data > mean +SE, and Grem1Low was the data < mean -SE. Statistical analysis was performed using log-rank tests, and hazard ratios were calculated using Prism software (GraphPad Software, Inc.).

The distribution of immune cells in TCGA datasets was extracted from publicly available “The Cancer Immunome Database” (https://tcia.at/home). Data showed immune cell abundance as analyzed by Quantiseq algorithm. Stromal and immune scores were calculated using the MAlignant Tumor tissues using Expression data (ESTIMATE) algorithm, available at https://bioinformatics.mdanderson.org/estimate/index.html.

The mRNA expression as RNA Seq V2 RSEM datasets were downloaded from the TCGA cBioportal. Spearman’s t test was used to evaluate correlation between genes using Prism software. Markers for activated and normal stroma were downloaded, and positively correlated genes were identified as those with a Spearman coefficient > 0.45 in both subtypes described by Moffitt et al.33.

Cancer Cell Line Encyclopedia (CCLE) data analysis

The gene expression data of Grem1 were downloaded from CCLE Expression 23Q4 Public (https://depmap.org/portal/gene/GREM1) and the dataset was filtered to include cell lines isolated from PDAC cases (n=49) and of fibroblastic origin (n=38)34. Outliers were defined as expression levels less or greater than 1.5 times the interquartile range and seven were removed from the PDAC cell lines. The cell lines isolated from PDAC cases were stratified into Grem1High (n=12, > Mean +SE) and Grem1Low (n= 25, < Mean -SE) groups based on average expression. The cell lines of fibroblastic origin were further filtered based on the type of tissue from which they were isolated, and fibroblasts isolated from colorectal carcinomas (n=3) and from melanomas of the skin (n=7) were used for further analysis.

Cell lines and cell culture

Human pancreatic cancer-associated fibroblast (hCAF) and human pancreatic stellate (hPSC) cell lines were obtained from Neuromics Inc. (Edina, MN USA) and ScienCell Research Laboratories Inc. (Carlsbad, CA USA), respectively. Pancreatic tumor cell lines Capan-1, Panc-1, Hs766T, CFPAC, BxPC-3, and Panc10.05 were obtained from American Type Culture Collection (ATCC, Manassas, VA USA). All cells were cultured using manufacturer recommended media under 5% CO2 and at 37°C in humidified incubators.

Reverse transcription-polymerase chain reaction (RT-PCR)

Total RNA was extracted from cells using RNeasy® Plus Micro Kit from Qiagen Sciences Inc. (Germantown, MD, USA) and reverse transcribed to complementary (c)DNA using RETROscript kit from ThermoFischer Scientific (Waltham, MA, USA). The following Grem1 primers were used: forward primer, 5’- ATGAGCCGCACAGCCTACACG-3’, and reverse primer, 5’-GACTGTTGCAGCCTTCCTCG-3’. cDNAs were used for PCR reaction at 95°C for 15 minutes followed by 35 cycles of 95°C for 30 seconds, 55°C for 45 seconds, and 72°C for 1 minute. PCR products were run on agarose gel.

Quantitative PCR

Quantitative (q)PCR was conducted as previously described35, 36 using TaqMan gene expression master mix and specific gene probes for human Grem1 (Hs01879841_s1 and Hs00171951_m) and 18S (Hs99999901_s1) from ThermoFischer Scientific (Waltham, MA, USA). Specific signals acquired were normalized to signals acquired from 18S.

Western blotting (WB)

Protein lysates were prepared from cells in 1× lysis buffer from Cell Signaling Technology (Danvers, MA USA), and the conditioned media were collected. WB was conducted as previously described35–37. Primary antibodies specific for Grem1A (against N-terminal), Grem1B and Grem1C (against C-terminal) were custom antibodies generated by Pacific Immunology Corp. (Ramona, CA USA). An additional primary antibody against Grem1 was from Abgent (San Diego, CA USA). Primary antibodies against fibronectin (FN) were from Santa Cruz Biotechnology (Dallas, TX USA), β-actin from Millipore Sigma (St. Louis, MO USA), ⍺-tubulin from Cell Signaling Technology (Danvers, MA USA), and α-smooth muscle actin (SMA) and collagen1 (Col1) from Abcam (Waltham, MA, USA). IRDye 800CW and 680RD conjugated secondary antibodies for WB were from LI-COR, Inc. (Lincoln, NE USA). Recombinant human and mouse Grem1 proteins were from R&D Systems (Minneapolis, MN USA).

Cell blocks for immunohistochemistry (IHC) controls

Cell paraffin blocks were prepared from hCAFs and Panc10.05 cells using HistoGel (Fisher Scientific, Pittsburgh, PA USA) and used as controls for Grem1A IHC. Briefly, the cells were fixed in 10% Formalin overnight, mixed into the liquid form of HistoGel and allowed the mixture of cells and HistoGel to solidify, which was then fixed in 10% Formalin overnight prior to paraffin embedding for sectioning. Paraffin sections from the cell blocks were used as controls for Grem1A IHC.

PDAC tissue samples and immunohistochemistry

Four PDAC tissue samples were obtained under the protocol approved by Institutional Review Board of George Washington University and used for IHC with Grem1A antibody. Consecutive and freshly cut sections from a pancreatic tumor tissue microarray (TMA) were purchased (PAN040–01A, US Biolab Corporation, Rockville, MD USA), which included 19 PDAC cases and 1 pancreatic acinar cell carcinoma with duplicate cores for each case. IHC was performed using Grem1A antibody (1:800 dilution), Collagen I antibody (1:100 dilution, ab6308, Abcam, Waltham, MA, USA), and CD68 antibody (ready-to-use, GA660961–2, Agilent DAKO, Santa Clara, CA, USA). All IHCs were performed using Dako Omnis auto-stainers.

Grem1 gene knockdown by siRNA and TGF-β treatment

CAFs were seeded in 6-well plates at 5×104/ml overnight and transfected with human Grem1 siRNA (siGrem1, 20 nmol/l, 4392420, ThermoFisher Scientific, Waltham, MA USA) or scrambled siRNA (D-001206–13) as a nonspecific control (siCON). Transfection reagent jetPRIME (114–01, Genesee Scientific, San Diego, CA USA) was used according to the manufacturer’s instructions. After 5 hours, transfection reagents were removed, fresh medium with FBS (0.5%) and TGF-β (=TGF-β1, 1 ng/ml) was added. Cells were cultured for 48 h and cells and conditioned media were collected for WB.

Macrophage isolation and differentiation

Bone marrow-derived mouse macrophages (BM-MΦ) were isolated from femurs of three C57BL/6 mice as previously described38. Isolated BM-MΦ were plated in 6-well plates at 2×106 cells/ml and cultured in macrophage complete medium (RPMI1640, 10% fetal bovine serum, 10 ng/ml mouse colony-stimulating factor (M-CSF, from R&D Systems)) for 7 days. Culture medium was changed every 2 days. Cells were cultured in the M-CSF-free medium for 24h before adding the recombinant mouse Grem1 protein (500 ng/ml). BM-MΦ were harvested 48 hours after treatment for flow cytometry. BM-MΦ were incubated with Fixable Viability Dye (BioLegend, San Diego, CA USA) for 15 min followed by a quick wash with FACS buffer (5% FBS in PBS). Cells were then labeled with surface markers vF450 anti-mouse F4/80 (Tonbo Biosciences, San Diego, CA USA), APC/Fire 750 anti-mouse CD80 (BioLegend), and AF488 anti-mouse CD206 (BioLegend) for 20 min in FACS buffer. In a separate batch, BM-MΦ were intracellularly labeled with APC anti-mouse iNOS (ThermoFisher Scientific) using Fixation/Permeabilization kit (BD Biosciences, San Jose, CA USA). Fluorescence minus one (FMO) controls were used for gating the positive cells.

Naïve CD4+T cell isolation and Treg cell differentiation

Single-cell suspensions were obtained from the spleens of three C57BL/6 mice as previously described39. Naïve CD4+T cells were immediately isolated using EasySep™ Mouse Naïve CD4+ T Cell Isolation Kit (STEMCELL Technologies Inc. Vancouver, Canada) according to the manufacturer’s instruction. Naïve CD4+T cells with > 90% purity were used for Treg differentiation assay. Briefly, the cells (5 × 105 cells/well) were plated in 24-well plates pre-coated with 3 μg/mL anti-CD3 (145–2C11). The differentiation medium contains 3 μg/mL anti-CD28, and 5 ng/mL recombinant mouse IL-2. Cells were treated with 500 ng/mL recombinant mouse Grem1 protein, 5 ng/mL TGF-β1 as a positive control, or vehicle as a negative control. After 5 days of cell culture, cells were harvested for surface and intracellular staining using antibodies including CD4 (GK1.5) conjugated with peridinin-chlorophyll protein/cyanine 5.5 (PerCP/Cy5.5), CD25 (PC61) conjugated with Alexa Fluor 700 (AF700), CD39 (Duha59) conjugated with phycoerythrin (PE), CD73 (TY/11.8) conjugated with allophycocyanin (APC), and Foxp3 (150D) conjugated with Alexa Fluor 488 (AF488). All antibodies were purchased from BioLegend. Intracellular staining of Foxp3 was performed with a Fixation/Permeabilization kit, according to the manufacturer’s protocol (BD Biosciences). Data from the cells were acquired on BD LSRFortessa Flow Cytometer and analyzed using FlowJo (FlowJo, Ashland, OR).

All animal experiments were performed in accordance with the protocol approved by the Animal Welfare Committee of the University of Texas Health Science Center at Houston.

Statistics

Data are expressed as mean ± standard error (SEM). Statistical significance was determined using t test for two group comparison and ANOVA for multiple group comparisons. P < 0.05 were considered statistically significant. The statistics were performed using GraphPad Prism version 9 for Windows (La Jolla, CA USA). Correlation analysis from the pancreatic TMA IHC was performed between antibodies per core using chi-square test.

RESULTS

Increased Grem1 expression is associated with stromal activation

Grem1 expression in 140 tumors from PDAC patients was acquired from TCGA RNA-sequencing data and stratified into Grem1High vs Grem1Low expressing tumors for analysis. Grem1High expressing tumors (n = 50) were associated with a significantly increased stromal score when compared to Grem1Low expressing tumors (n = 90) (P < 0.0001, Fig. 1A). Grem1 levels of Grem1High expressing tumors were positively correlated with 22 out of the 24 genes from the activated stroma subtype previously described by Moffitt et al.33 (Fig. 1B). Contrastingly, Grem1 levels of these tumors positively correlated with only 4 out of a total of 23 genes included in the normal stroma subtype (Fig. 1C). Thus, Grem1High expressing tumors may exhibit characteristics similar to those observed in tumors from the activated stroma subtype such as tumor promotion, an activated fibroblast state, and a higher inflammatory stromal response. Additionally, patients with Grem1High expressing tumors exhibited a marginally decreased, although not statistically significant, probability of survival compared to those with Grem1Low expressing tumors (P = 0.09, Fig. 1D).

Figure 1. Correlation of Grem1High expression with activated stroma in human PDAC cases.

Figure 1.

Open in a new tab

The mRNA expression in 140 PDAC tumors was downloaded from TCGA BioPortal and stratified into Grem1High vs Grem1Low expressing tumors. A. Stromal scores as reported by Malignant Tumor tissues using Expression data (ESTIMATE) algorithm. B. Correlation of mRNA Grem1 expression with activated stroma-related genes. C. Correlation of Grem1 mRNA expression with normal stroma-related genes. Spearman’s t test was used to evaluate correlation between genes. Positively correlated genes were identified with a Spearman coefficient > 0.45. D. Probability of survival by Long-rank test. **** P < 0.0001. n = 50 cases of Grem1High, n = 90 cases of Grem1Low.

Grem1 transcript variant 1 and protein isoform 1 are dominant in pancreatic cell lines

Grem1 expression datasets in cell lines isolated from PDAC tumors and of fibroblastic origin were acquired from the CCLE database. Cell lines isolated from PDAC tumors were stratified into Grem1High (n=12) and Grem1Low (n=25) groups for analysis. Cell lines of fibroblastic origin were filtered based on tissue of isolation from those isolated from tumors of gastrointestinal origin and melanoma of skin for further analysis, as no fibroblastic cell lines isolated from pancreas samples were available. Expression of Grem1 mRNA among cell lines of PDAC tumor origin were variably low compared to cell lines of fibroblastic origin (P < 0.0001, Fig 2A), suggesting that Grem1 expression is largely from fibroblasts.

Figure 2. Characterization of Grem1 transcripts in PDAC.

Figure 2.

Open in a new tab

A. The Grem1 mRNA expression data from 37 PDAC cell lines and 10 fibroblastic cell lines from CCLE. B. Schematic diagram identifies Grem1 transcript variants and location of primers (indicated by arrows) and qPCR probes. C. Representative image of gel electrophoresis from PCR products. D. Grem1 mRNA levels measured by qPCR. Data are presented as mean ± SEM from triplicate wells. * P < 0.05. **** P < 0.0001.

In order to delineate Grem1 expression and distribution in PDAC, we evaluated Grem1 expression in pancreatic cell lines in vitro. First, we evaluated Grem1 transcript variant expression by designing primers that flank the variable region of the Grem1 coding sequence (Fig. 2B). Full-length transcript variant 1 was predominant in all pancreatic cell lines, and the remarkably high levels were observed in pancreatic fibroblasts (Fig. 2C). Truncated transcript variant 2 and 3 were not detectable. In addition, a band around 500bp was visible in several cell lines. We confirmed the specificity of the PCR primers used for the Grem1 gene detection with the Primer-BLAST algorithm40. However, we could not exclude this visible band being a potential spliced variant.

These findings were quantified using qPCR with two Grem1 probe sets, one targeted the variable N-terminus with a probe location at 155 bp that could detect variant 1 and 2 and the other targeted the common C-terminus with a probe location at 3743 bp that could detect all three variants. The qPCR with both probes similarly showed that Grem1 expression was 20–50-fold higher in fibroblastic cell lines than in tumor cell lines (Fig. 2D).

Commercially available Grem1 antibodies have low specificity. In order to address this limitation and to evaluate isoform expression of Grem1 protein, we designed custom antibodies. Three peptides corresponding to 3 regions of Grem1 were generated and used to prepare antibodies (Fig. 3A). WB analysis showed that among three custom Grem1 antibodies and one commercial Grem1 antibody (Abgent), Grem1A antibody detected a major band at 22 kD corresponding to full-length Grem1 isoform 1, indicating a higher specificity (Fig. 3B). The additional custom Grem1B antibody barely detected a band at 22 kD, Grem1C antibody detected a band at 22 kD and also detected other bands below 20 kD. A Grem1 antibody from Abgent detected a weak band at 22 kD. All these 3 antibodies detected multiple bands with relatively higher intensities above 25 kD compared to those with Grem1A antibody. Specificity of the Grem1A antibody was verified by WB following Grem1 knockdown with siGrem1 in CAFs. WB of cell lysates showed the band at 22 kD was diminished with siGrem1 transfection compared to siCON treatment or untreated cells (Fig. 3C).

Figure 3. Characterization of custom Grem1 antibody and Grem1 isoforms.

Figure 3.

Open in a new tab

A. Schematic diagram of Grem1 identifies Grem1 isoforms and location of custom antibodies. B. Images of WB of pancreatic cell lines probed with Grem1 antibodies. C. Images of WB of hCAF cell lysates treated with siGrem1 or siCON, or untreated. Recombinant human Grem1 (rhGrem1) protein was used as a positive control.

Grem1 expression in PDAC correlates with infiltrating macrophages

Based on the characterization experiments, the custom Grem1A antibody was selected over other available antibodies for subsequent experimentation because of its higher specificity. Further analysis with Grem1A antibody was performed with IHC on the cell blocks prepared from hCAFs and Panc10.05 cells for positive controls (Fig 4A). hCAFs stained strong positive and the majority of Panc10.05 cells stained positive. The negative controls were stained with rabbit IgG. IHC with Grem1A was performed on tissue sections from four human PDAC cases (Fig. 4B). The normal tissue away from tumor were essentially negative for Grem1. The tissue adjacent to tumor showed Grem1 positive staining in fibroblasts surrounding acini. The stroma surrounding tumors showed positive Grem1 staining, which is consistent with our previous work31. Fibroblasts immediately surrounding a PanIN3 stained positive, but PanIN cells stained negative. Interestingly, invasive PDAC showed positive staining in some tumor cells, which has not previously been reported.

Figure 4. IHC of Grem1, macrophages, and ECM on PDAC cases and the analysis.

Figure 4.

Open in a new tab

A. Representative images from hCAF and Panc10.05 cell blocks stained with Grem1A antibody (positive control) and rabbit IgG isotype (negative control). B. Representative images from 4 human PDAC cases with Grem1A antibody. C. Representative images from PDAC TMA with Grem1A antibody, CD68 antibody, and Collagen type I antibody from same positive core and same negative core. D. Correlation analysis by Chi-square test between Grem1 epithelial cell staining (Grem1-E) and CD68 or Collagen type I per core on total 34 cores. E. Correlation analysis by Chi-square test between Grem1 stromal cell staining (Grem1-S) and CD68 or Collagen type I per core on total 34 cores. Six cores with technical deficiencies (detached during IHC or no tumor) were excluded for correlation analysis. Scale bar = 100 μm.

Furthermore, we have obtained commercial human pancreatic tissue microarrays and performed IHC with Grem1A antibody, Collagen type I antibody, and CD68 antibody and evaluated the correlation of Grem1 staining with Collagen type I representing ECM and CD68 representing macrophage infiltration (Fig. 4C, 4D, 4E). We excluded 6 cores from 3 PDAC cases with technical deficiencies (detached during IHC or no tumor) and performed correlation analysis between antibodies per core on total 34 cores.

We observed a positive correlation between Grem1 expression in pancreatic cancer epithelial (E) and stromal (S) cells (p<0.001). All 19 cores stained Grem1-E positive also showed Grem1-S positive (100%). Eleven out of 15 cores stained Grem1-E negative were also Grem1-S negative (73%) (Fig. 4D). Grem1-E and Grem1-S staining patterns were positively correlated with stromal infiltration by macrophages as demonstrated by CD68 staining (p<0.001). All 19 cores stained with Grem1-E positive were associated with infiltration by CD68 positive macrophages (100%). Nine of 15 cores stained Grem1-E negative were also negative for CD68 staining (60%). Similarly, all 23 cores stained Grem1-S positive were also positive for CD68 positive macrophages (100%). Nine of 11 cores stained Grem1-S negative were negative for CD68 staining (82%). (Fig. 4E). However, Grem1-E or Grem1-S staining patterns were not significantly correlated with Collagen type I staining (Grem1-E vs. Collagen type I, p=0.218; Grem1-S vs. Collagen type I, p=0.926) (Fig. 4D, 4E).

These results demonstrate that Grem1 is expressed in PDAC in the tumor cells and the stromal cells, and correlates with infiltrating macrophages but not with Collagen type I.

Grem1 mediates TGF-β-induced ECM production from CAFs

In order to elucidate the role of Grem1 in stromal development, pancreatic CAFs were transfected with siGrem1. Whole cell lysates and conditioned media were collected for evaluation of Grem1 levels and ECM production by WB. siGrem1 transfection led to 90 % knockdown of Grem1 expression compared to siCON. Grem1 knockdown alone did not alter cellular or secreted levels of ECM proteins FN or Col1 or α-SMA (Fig. 5A). When treated with TGF-β, a known key profibrogenic factor reported to be upstream of Grem130, the siGrem1 group had reduced cellular levels of FN and α-SMA, and reduced secreted FN and Col1 (Fig. 5B). These data indicate that Grem1 mediates TGF-β-induced ECM production from CAFs. Of note, we observed that commonly used protein loading controls GAPDH and β-actin were also upregulated in TGF-β and siCON treatment group. Currently, we do not know the reason, which deserves future investigation. Since GAPDH and β-actin were not appropriate for the loading controls, therefore, we used tubulin as an WB loading control and observed similar levels of tubulin between siCON and siGrem1 with TGF-β treatment groups.

Figure 5. Effects of Grem1 knockdown on ECM levels in CAFs.

Figure 5.

Open in a new tab

A. Grem1 knockdown effects on baseline cellular (whole cell lysate) and secreted (conditioned medium) FN and Col1 and on cellular α-SMA shown in WB and quantification. B. Grem1 knockdown effects on TGF-β-induced cellular and secreted FN and Col1 and on cellular α-SMA shown in WB and quantification. siCON: siRNA control; siGrem1: Grem1 siRNA; rhGrem: recombinant human Grem1. Data are expressed as mean ± SEM from triplicate wells. * P < 0.05 compared to siCON.

Grem1High expressing PDACs tumors correlate with increased macrophages and Tregs and related genes, and Grem1 increases M1 and M2 macrophages but not Tregs in vitro

To understand whether Grem1 mediates the interaction between immune cells and CAFs within PDAC tumor microenvironment, we first analyzed human PDAC TCGA data. We found that although there was no apparent difference between Grem1 levels and immune score (Fig. 6A), Grem1High expressing tumors showed increased M1 and M2 macrophages (P < 0.0001, Fig. 6B) and increased regulatory T cells (Tregs) (P < 0.0001, Fig. 6C). Furthermore, when compared to Grem1Low expressing tumors, Grem1High expressing tumors presented higher Z score expression values of macrophage-related (Fig. 6D) and Treg-related (Fig. 6E) genes. The expression of immunoexhaustion genes were increased at variable levels in Grem1High expressing tumors (Fig. 6F).

Figure 6. Correlation of Grem1High expressing PDACs tumors with increased macrophages and Tregs and the related genes, and effects of Grem1 on macrophages and Tregs in vitro.

Figure 6.

Open in a new tab

The distribution of immune cells in the TCGA datasets was extracted from publicly available “The Cancer Immunome Database”. Comparison of Grem1High (n=50) vs. Grem1Low (n=90) expressing PDACs tumors on A. Immune score as reported by Malignant Tumor tissues using Expression data (ESTIMATE) algorithm. B. Macrophage fractions. C. Immune cell fractions. D. Macrophage-related genes. E. Treg-related genes. F. Immunoexhaustion genes. G. Macrophages derived from mouse bone marrow were treated with mouse recombinant Grem1 (500 ng/ml) or vehicle control (0.1% bovine serum albumin, 4 mM HCl) for 48 h. H. Naïve CD4+T cells isolated from mouse spleens were treated with mouse recombinant Grem1 (500 ng/ml) or vehicle control in Treg differentiation medium for 5 days. Flow cytometry was performed and analyzed. Data are expressed as mean % or MFI of respective markers ± SEM from triplicate wells. * P < 0.05, ** P < 0.01, **** P < 0.001 compared to vehicle control.

The role of secreted Grem1 on macrophage differentiation was evaluated by treatment with recombinant mouse Grem1 protein. Macrophages were isolated from the bone marrow of mice and treated with Grem1 for 48 hours, followed by flow cytometry assay using macrophage markers. Grem1 treatment increased expression of CD80, iNOS, and CD206, which are markers of total, M1, and M2 macrophages, respectively (Fig. 6G). These results provide functional context to support the human PDAC analysis showing that Grem1High expressing tumors correlate with increased M1 and M2 macrophages (Fig. 6B), suggesting a promoting role of Grem1 in the activation and differentiation of macrophages in the stromal microenvironment.

Furthermore, the role of secreted Grem1 on differentiation of Tregs was evaluated upon treatment with the Grem1 protein. Naïve CD4+T cells isolated from mouse spleens were treated by TGF-β1 (as a positive control), Grem1, or vehicle in the presence of anti-CD3, anti-CD28 and IL-2 for 5 days, followed by staining with T cell markers and analyzed using flow cytometry. Grem1 treatment did not increase the Foxp3+, CD73+, CD39+, or CD25+ cells, all of which are markers of Tregs (Fig. 6H). These results indicate that Grem1 does not directly induce Treg differentiation in vitro.

DISCUSSION

The PDAC microenvironment is a dynamic structure that surrounds cancer cells and provides a chemoresistant barrier. It is highly fibrotic and is largely created by CAFs, which are activated in response to repeated stimuli41. Activation status of fibroblasts can be monitored by measuring α-SMA expression42, which can be used to determine an activated stromal index (ASI). Patients with a greater ASI have been shown to have worse survival outcomes43. In this study, we investigated the relationship of activated stroma with Grem1 expression in PDAC patients and showed that high Grem1 expression is positively correlated with a high stromal score and the expression of activated stromal genes. Thus, Grem1 may be involved in the development and propagation of the activated fibrotic stroma and further investigation into the role of Grem1 in the tumor microenvironment is warranted.

We previously showed that Grem1 mRNA expression was mainly localized to fibroblasts in fibrotic stroma surrounding PDAC using RNA in situ hybridization31. Other studies suggest that Grem1 is expressed in tumor cells21, 23, 44. This study refines the current understanding of Grem1 expression using a highly specific custom Grem1 antibody that overcomes previous limitation of low specificity and provides us with a powerful tool to further investigate the dynamic role of Grem1 in PDAC and other cancers with Grem1 overexpression. Evaluation of pancreatic cell lines with the antibody reveals that Grem1 isoform 1 expression is largely among fibroblastic cell lines with low and variable expression in pancreatic tumor cell lines.

IHC staining with this custom antibody corroborated the cell line findings and showed Grem1 expression location varying among different stages of disease. Grem1 expression is confined to the stroma surrounding the tumor in human PanIN cases but is expressed in tumor cells in invasive PDAC cases, suggesting that Grem1 may be involved in a critical transition to invasive disease45. Further IHC staining on a human PDAC tissue microarray demonstrated that Grem1 protein expression in the tumor cells and the stromal cells correlated with the infiltrating macrophages, consistent with that of the TCGA data. However, Grem1 protein expression in the tumor cells and the stromal cells did not show significant correlation with Collagen type I expression. We speculate that Grem1 expression may correlate with other types of ECM proteins, which deserves future investigation.

We also investigated Grem1 variant expression in PDAC. While cells often express a few alternatively spliced transcripts per gene, expression of a single transcript is generally dominant in normal states. Cancerous cells can defy this trend with increased expression of an alternative transcript, broadening protein activity46, 47. Grem1 has 3 major identified transcripts and corresponding isoforms, with variant 1 and isoform 1 being predominantly expressed in normal human tissue32. Alternative Grem1 isoform expression in cancer has not been previously reported. We evaluated transcript variant and isoform expression among pancreatic cell lines and determined that the full length Grem1 transcript variant 1 and corresponding isoform 1 is predominant in all cell lines, including a normal pancreatic cell line (hPSC) and tumorigenic pancreatic cell lines. This novel insight ensures experiments are clinically relevant in future studies.

Lastly, we investigated the function of Grem1 in the tumor microenvironment. Recent studies have shown that Grem1 maintains epithelial PDAC subpopulations29 and promotes fibrogenesis in intestinal fibrosis and breast cancer22, 28. Lan et al. recently showed that inactivation of Grem1 through gene knockout in an established PDAC in mice resulted in a conversion of epithelial cells to mesenchymal cells and increased metastasis, suggesting Grem1 plays a role in maintaining epithelial heterogeneity in PDAC and may inhibit metastasis29. Conversely, Sung et al. demonstrated that knockdown of Grem1 in breast cancer cells in an orthotopic mouse model resulted in decreased proliferation of cancer cells and decreased EMT27, 48. These contrasting results may be due to different mouse models used and the corresponding tumor stroma. Although these studies suggest a critical and complex role of Grem1 in the development and progression of cancer, further investigation is warranted to determine the specific role of Grem1 in PDAC tumor microenvironment.

We evaluated the specific role of Grem1 in pancreatic CAF fibrogenesis using siRNA knockdown of Grem1. Knockdown of Grem1 in CAFs did not affect activation status but did decrease secretion of fibronectin and collagen1 expression in cells treated with TGF-β, suggesting that the enhancement of the fibrotic tumor microenvironment is a direct result of the molecular mechanisms involving TGF-β/Grem1 in activated fibroblasts. Additionally, we have previously performed IHC of PDAC cases and shown that fibroblasts expressing Grem1 may be associated with increased macrophages and pro-tumor M2 polarization31; however, we did not investigate the presence of M1 macrophages. In this study, we showed treatment of macrophages with Grem1 promoted differentiation of macrophages into both M1 and M2, supporting that fibroblastic secretion of Grem1 promotes a macrophagic immune response. Further investigation into the opposing role of macrophage subtypes in PDAC is warranted.

Bioinformatic analysis of the TCGA PDAC data indicates that Grem1High PDAC is associated with increased Tregs compared to Grem1Low with altered Treg-associated genes. However, our in vitro studies showed Grem1 did not induce Treg. The discrepancy between the TCGA data and the in vitro Treg studies could be a result of the elevated TGF-β level as reported in PDAC49. Although Grem1 acts as a down-stream mediator of TGF-β for the profibrogenic functions30, TGF-β has multiple downstream mediators in addition to Grem1 for its multifunction. Therefore, the upregulation of Treg in TCGA PDAC cases can be mediated by other mediators independent of Grem1. Another potential explanation may be due to the altered immune cells in the tumor microenvironment of PDAC and their communication and interaction50. The microenvironment in the TCGA PDAC cases contains increased macrophages and Tregs51. Tregs can recruit macrophages and can be controlled by macrophages, which can also be Grem1-independent.

Our study addresses the significance of paracrine Grem1 signaling between fibroblasts and macrophages showing that Grem1, which is highly expressed by CAFs, may be a potential therapeutic target in the tumor microenvironment. A recent report on a monoclonal antibody ginisortamab (UCB6114) development shed light on Grem1’s therapeutic potential. This antibody can neutralize Grem1 and restore BMP signaling pathways in human colorectal cancer cell lines and fibroblasts. It is currently being evaluated by a phase 1/2 clinical trial (NCT04393298) in advanced solid tumors including PDAC52. Thus, targeting Grem1 may be an effective adjunctive therapy that supplements other chemotherapies by targeting the PDAC tumor microenvironment. Furthermore, TGF-β superfamily members may also be potential targets for PDAC and other solid tumors. The therapeutic progress and potential regarding TGF-β superfamily members including Grem1 were summarized in our recent review article53.

This study has several limitations. Firstly, the lack of significance in the survival curve in the current study could be due to low power and limited cases. However, the quality IHC for Grem1 detection using the custom antibody in this study was a critical first step for further studying the spatial role of Grem1 in PDAC development. Additionally, a protein not appropriate as a prognostic marker can still be a predictive marker for monitoring the response to targeted treatment54. Secondly, functional studies for cellular interaction were only performed by in vitro studies. As we shown in this study, Grem1 apparently regulates both fibroblasts and immune cells. However, Grem1 may also affect other aspects of the complex tumor microenvironment. Thus, further studies to investigate the effect of Grem1 on the PDAC tumor microenvironment in vivo will provide critical clinical insights.

In conclusion, our present study showed that Grem1 is strongly associated with the activated fibrotic tumor microenvironment and that Grem1 variant 1 and encoded isoform 1 are predominant in all cell lines tested and are remarkably high in pancreatic fibroblasts. Furthermore, we showed that Grem1 has promoting effects on PDAC fibrogenesis and macrophage activation, suggesting that Grem1 may be a potential prognostic maker and a potential therapeutic target for PDAC.

ACKNOWLEDGEMENTS

The authors thank the HistoCore Lab at UTHealth Houston for IHC services.

Disclosure of funding received for this work:

partially supported by the NIH1R21 AA027014-01A1 and Jack H Mayfield M.D. Distinguished Professorship in Surgery (TCK), Dean’s fund for Summer Research Program at UTHealth MMS (RRT).

Footnotes

The authors declare no conflicts of interest to disclose.

References

  • 1.Rahib L, Wehner MR, Matrisian LM, et al. Estimated Projection of US Cancer Incidence and Death to 2040. JAMA Netw Open 2021;4:e214708. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Siegel RL, Miller KD, Fuchs HE, et al. Cancer Statistics, 2021. CA Cancer J Clin 2021;71:7–33. [DOI] [PubMed] [Google Scholar]
  • 3.Ryan DP, Hong TS, Bardeesy N. Pancreatic adenocarcinoma. N Engl J Med 2014;371:2140–1. [DOI] [PubMed] [Google Scholar]
  • 4.Feig C, Gopinathan A, Neesse A, et al. The pancreas cancer microenvironment. Clin Cancer Res 2012;18:4266–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Sunami Y, Haussler J, Kleeff J. Cellular Heterogeneity of Pancreatic Stellate Cells, Mesenchymal Stem Cells, and Cancer-Associated Fibroblasts in Pancreatic Cancer. Cancers (Basel) 2020;12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Moir JA, Mann J, White SA. The role of pancreatic stellate cells in pancreatic cancer. Surg Oncol 2015;24:232–8. [DOI] [PubMed] [Google Scholar]
  • 7.Han L, Wu Y, Fang K, et al. The splanchnic mesenchyme is the tissue of origin for pancreatic fibroblasts during homeostasis and tumorigenesis. Nat Commun 2023;14:1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Miyazaki Y, Oda T, Mori N, et al. Adipose-derived mesenchymal stem cells differentiate into pancreatic cancer-associated fibroblasts in vitro. FEBS Open Bio 2020;10:2268–2281. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Iwamoto C, Ohuchida K, Shinkawa T, et al. Bone marrow-derived macrophages converted into cancer-associated fibroblast-like cells promote pancreatic cancer progression. Cancer Lett 2021;512:15–27. [DOI] [PubMed] [Google Scholar]
  • 10.Ohlund D, Handly-Santana A, Biffi G, et al. Distinct populations of inflammatory fibroblasts and myofibroblasts in pancreatic cancer. J Exp Med 2017;214:579–596. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kalluri R The biology and function of fibroblasts in cancer. Nat Rev Cancer 2016;16:582–98. [DOI] [PubMed] [Google Scholar]
  • 12.Farajzadeh Valilou S, Keshavarz-Fathi M, Silvestris N, et al. The role of inflammatory cytokines and tumor associated macrophages (TAMs) in microenvironment of pancreatic cancer. Cytokine Growth Factor Rev 2018;39:46–61. [DOI] [PubMed] [Google Scholar]
  • 13.Yang S, Liu Q, Liao Q. Tumor-Associated Macrophages in Pancreatic Ductal Adenocarcinoma: Origin, Polarization, Function, and Reprogramming. Front Cell Dev Biol 2020;8:607209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Khokha MK, Hsu D, Brunet LJ, et al. Gremlin is the BMP antagonist required for maintenance of Shh and Fgf signals during limb patterning. Nat Genet 2003;34:303–7. [DOI] [PubMed] [Google Scholar]
  • 15.Marquez-Exposito L, Cantero-Navarro E, R RR-D, et al. Molecular Regulation of Notch Signaling by Gremlin. Adv Exp Med Biol 2020;1227:81–94. [DOI] [PubMed] [Google Scholar]
  • 16.Tsialogiannis E, Polyzois I, Oak Tang Q, et al. Targeting bone morphogenetic protein antagonists: in vitro and in vivo evidence of their role in bone metabolism. Expert Opin Ther Targets 2009;13:123–37. [DOI] [PubMed] [Google Scholar]
  • 17.Sneddon JB, Zhen HH, Montgomery K, et al. Bone morphogenetic protein antagonist gremlin 1 is widely expressed by cancer-associated stromal cells and can promote tumor cell proliferation. Proc Natl Acad Sci U S A 2006;103:14842–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Ma XJ, Dahiya S, Richardson E, et al. Gene expression profiling of the tumor microenvironment during breast cancer progression. Breast Cancer Res 2009;11:R7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Kim HS, Shin MS, Cheon MS, et al. GREM1 is expressed in the cancer-associated myofibroblasts of basal cell carcinomas. PLoS One 2017;12:e0174565. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Dutton LR, Hoare OP, McCorry AMB, et al. Fibroblast-derived Gremlin1 localises to epithelial cells at the base of the intestinal crypt. Oncotarget 2019;10:4630–4639. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Yu Y, Cheng L, Yan B, et al. Overexpression of Gremlin 1 by sonic hedgehog signaling promotes pancreatic cancer progression. Int J Oncol 2018;53:2445–2457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Ren J, Smid M, Iaria J, et al. Cancer-associated fibroblast-derived Gremlin 1 promotes breast cancer progression. Breast Cancer Res 2019;21:109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Karagiannis GS, Musrap N, Saraon P, et al. Bone morphogenetic protein antagonist gremlin-1 regulates colon cancer progression. Biol Chem 2015;396:163–83. [DOI] [PubMed] [Google Scholar]
  • 24.Kim M, Yoon S, Lee S, et al. Gremlin-1 induces BMP-independent tumor cell proliferation, migration, and invasion. PLoS One 2012;7:e35100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Hong D, Liu T, Huang W, et al. Gremlin1 Delivered by Mesenchymal Stromal Cells Promoted Epithelial-Mesenchymal Transition in Human Esophageal Squamous Cell Carcinoma. Cell Physiol Biochem 2018;47:1785–1799. [DOI] [PubMed] [Google Scholar]
  • 26.Sun Z, Cai S, Liu C, et al. Increased Expression of Gremlin1 Promotes Proliferation and Epithelial Mesenchymal Transition in Gastric Cancer Cells and Correlates With Poor Prognosis of Patients With Gastric Cancer. Cancer genomics & proteomics 2020;17:49–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Sung NJ, Kim NH, Surh YJ, et al. Gremlin-1 Promotes Metastasis of Breast Cancer Cells by Activating STAT3-MMP13 Signaling Pathway. Int J Mol Sci 2020;21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Yang Y, Zeng QS, Zou M, et al. Targeting Gremlin 1 Prevents Intestinal Fibrosis Progression by Inhibiting the Fatty Acid Oxidation of Fibroblast Cells. Front Pharmacol 2021;12:663774. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Lan L, Evan T, Li H, et al. GREM1 is required to maintain cellular heterogeneity in pancreatic cancer. Nature 2022;607:163–168. [DOI] [PubMed] [Google Scholar]
  • 30.Staloch D, Gao X, Liu K, et al. Gremlin is a key pro-fibrogenic factor in chronic pancreatitis. J Mol Med (Berl) 2015;93:1085–1093. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Davis JM, Cheng B, Drake MM, et al. Pancreatic stromal Gremlin 1 expression during pancreatic tumorigenesis. Genes & Diseases 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Topol LZ, Bardot B, Zhang Q, et al. Biosynthesis, post-translation modification, and functional characterization of Drm/Gremlin. J Biol Chem 2000;275:8785–93. [DOI] [PubMed] [Google Scholar]
  • 33.Moffitt RA, Marayati R, Flate EL, et al. Virtual microdissection identifies distinct tumor- and stroma-specific subtypes of pancreatic ductal adenocarcinoma. Nat Genet 2015;47:1168–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Ghandi M, Huang FW, Jane-Valbuena J, et al. Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 2019;569:503–508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Gao X, Cao Y, Yang W, et al. BMP2 inhibits TGF-beta-induced pancreatic stellate cell activation and extracellular matrix formation. Am J Physiol Gastrointest Liver Physiol 2013;304:G804–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Cao Y, Chen L, Zhang W, et al. Identification of apoptotic genes mediating TGF-beta/Smad3-induced cell death in intestinal epithelial cells using a genomic approach. Am J Physiol Gastrointest Liver Physiol 2007;292:G28–38. [DOI] [PubMed] [Google Scholar]
  • 37.Gao X, Cao Y, Staloch DA, et al. Bone morphogenetic protein signaling protects against cerulein-induced pancreatic fibrosis. PLoS One 2014;9:e89114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Zhang X, Goncalves R, Mosser DM. The isolation and characterization of murine macrophages. Curr Protoc Immunol 2008;Chapter 14:14 1 1–14 1 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.He B, Hoang TK, Wang T, et al. Resetting microbiota by Lactobacillus reuteri inhibits T reg deficiency-induced autoimmunity via adenosine A2A receptors. J Exp Med 2017;214:107–123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Ye J, Coulouris G, Zaretskaya I, et al. Primer-BLAST: a tool to design target-specific primers for polymerase chain reaction. BMC Bioinformatics 2012;13:134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Huang M, Li Y, Zhang H, et al. Breast cancer stromal fibroblasts promote the generation of CD44+CD24- cells through SDF-1/CXCR4 interaction. J Exp Clin Cancer Res 2010;29:80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Lazard D, Sastre X, Frid MG, et al. Expression of smooth muscle-specific proteins in myoepithelium and stromal myofibroblasts of normal and malignant human breast tissue. Proc Natl Acad Sci U S A 1993;90:999–1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Erkan M, Michalski CW, Rieder S, et al. The activated stroma index is a novel and independent prognostic marker in pancreatic ductal adenocarcinoma. Clin Gastroenterol Hepatol 2008;6:1155–61. [DOI] [PubMed] [Google Scholar]
  • 44.Sato M, Kawana K, Fujimoto A, et al. Clinical significance of Gremlin 1 in cervical cancer and its effects on cancer stem cell maintenance. Oncol Rep 2016;35:391–7. [DOI] [PubMed] [Google Scholar]
  • 45.Chuvin N, Vincent DF, Pommier RM, et al. Acinar-to-Ductal Metaplasia Induced by Transforming Growth Factor Beta Facilitates KRAS(G12D)-driven Pancreatic Tumorigenesis. Cell Mol Gastroenterol Hepatol 2017;4:263–282. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Sebestyen E, Zawisza M, Eyras E. Detection of recurrent alternative splicing switches in tumor samples reveals novel signatures of cancer. Nucleic Acids Res 2015;43:1345–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Kahraman A, Karakulak T, Szklarczyk D, et al. Pathogenic impact of transcript isoform switching in 1,209 cancer samples covering 27 cancer types using an isoform-specific interaction network. Sci Rep 2020;10:14453. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Sung NJ, Kim NH, Bae NY, et al. DHA inhibits Gremlin-1-induced epithelial-to-mesenchymal transition via ERK suppression in human breast cancer cells. Biosci Rep 2020;40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Dardare J, Witz A, Merlin JL, et al. SMAD4 and the TGFbeta Pathway in Patients with Pancreatic Ductal Adenocarcinoma. Int J Mol Sci 2020;21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Zhen Y, Zheng J, Zhao Y. Regulatory CD4+CD25+ T cells and macrophages: communication between two regulators of effector T cells. Inflamm Res 2008;57:564–70. [DOI] [PubMed] [Google Scholar]
  • 51.Tang HD, Wang Y, Xie P, et al. The Crosstalk Between Immune Infiltration, Circulating Tumor Cells, and Metastasis in Pancreatic Cancer: Identification of HMGB3 From a Multiple Omics Analysis. Front Genet 2022;13:892177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Davies GCG, Dedi N, Jones PS, et al. Discovery of ginisortamab, a potent and novel anti-gremlin-1 antibody in clinical development for the treatment of cancer. MAbs 2023;15:2289681. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Tindall RR, Bailey-Lundberg JM, Cao Y, et al. The TGF-β superfamily as potential therapeutic targets in pancreatic cancer. Frontiers in Oncology 2024;14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Mann S, Laucirica R, Carlson N, et al. Estrogen receptor beta expression in invasive breast cancer. Hum Pathol 2001;32:113–8. [DOI] [PubMed] [Google Scholar]

RESOURCES