Secretory Trefoil Factor 1 (TFF1) promotes gemcitabine resistance through chemokine receptor CXCR4 in Pancreatic Ductal Adenocarcinoma

Abstract

Gemcitabine is the first-line treatment option for patients with locally advanced or metastatic pancreatic ductal adenocarcinoma (PDAC). However, the frequent adoption of resistance to gemcitabine by cancer cells poses a significant challenge in treating this aggressive disease. In this study, we focused on analyzing the role of trefoil factor 1 (TFF1) in gemcitabine resistance in PDAC. Analysis of PDAC TCGA and cell line datasets indicated an enrichment of TFF1 in the gemcitabine-resistant classical subtype and suggested an inverse correlation between TFF1 expression and sensitivity to gemcitabine treatment. The genetic ablation of TFF1 in PDAC cells enhanced their sensitivity to gemcitabine treatment in both in vitro and in vivo tumor xenografts. The biochemical studies revealed that TFF1 contributes to gemcitabine resistance through enhanced stemness, increasing migration ability of cancer cells, and induction of anti-apoptotic genes. We further pursued studies to predict possible receptors exerting TFF1-mediated gemcitabine resistance. Protein-protein docking investigations with BioLuminate software revealed that TFF1 binds to the chemokine receptor CXCR4, which was supported by real-time binding analysis of TFF1 and CXCR4 using SPR studies. The exogenous addition of TFF1 increased the proliferation and migration of PDAC cells through the pAkt/pERK axis, which was abrogated by treatment with a CXCR4-specific antagonist AMD3100. Overall, the present study demonstrates the contribution of the TFF1-CXCR4 axis in imparting gemcitabine resistance properties to PDAC cells.

1. Introduction

Pancreatic ductal adenocarcinoma (PDAC), having a low 5-year survival of 12%, is expected to be the second major cause of cancer-related deaths by 2030 [1-3]. Several therapeutic regimens, including gemcitabine, 5-FU, nab-paclitaxel, and FOLFIRINOX, have been developed for PDAC [4]. Although greater overall survivals have been achieved with FOLFIRINOX, severe toxicity issues have limited its utility in the clinical setting, making gemcitabine combined with nab-paclitaxel the standard-of-care chemotherapeutic regimen for PDAC management. However, the adoption of therapeutic resistance to gemcitabine by cancer cells results in poor outcomes in PDAC patients [5, 6]. Many studies have attempted to comprehensively understand the molecular mechanisms mediating gemcitabine resistance in patients, cell lines, and animal models [7-9].

The research focused on utilizing gene signatures for predicting therapeutic response to gemcitabine have stratified PDAC patients into basal, classical, and quasi-mesenchymal subtypes, with the subtypes demonstrating varying responses to drug treatment [10-12]. Interestingly, molecular subtyping studies have indicated the presence of a distinct TFF-based gene signature in PDAC and across multiple tumor types [11, 13]. Trefoil factors (TFFs; TFF1, 2, and 3) are small secretory mucin-associated molecules that play a crucial role in response to gastrointestinal inflammation and injury and have recently gained attention as a part of a gene signature used to distinguish patients into different subtypes [14]. Transcriptomic analysis of human PDAC patient tumors and cell lines by Moffitt et al. and Collison et al. revealed higher expression of TFF1 in the classical subtype that showed poor response to gemcitabine treatment [11, 12]. These studies imply the importance of TFFs in therapy resistance and, thereby, poor survival of cancer patients.

TFF1 is a small secretory protein of 60 amino acids in monomer, dimer, and oligomeric forms [15]. TFF1 plays a crucial role in mucosal defense and repair by forming the mucosal barrier and enhancing cell migration (restitution). Previous studies have shown the tumor-suppressive role of TFF1 in gastric cancer, hepatocellular carcinoma, and retinoblastoma using knockdown (KD) studies and genetically engineered mouse models [15, 16]. Among these, TFF1 depletion has been implicated in inducing the EMT phenotype, suggesting its tumor-suppressive role in PDAC [17]. However, many studies have elaborated that TFF1 function varies context-dependent and may also be associated with metastasis in PDAC [18, 19]. In fact, multiple studies have depicted the oncogenic role of TFFs in various solid malignancies, including breast cancer, colon cancer, and ovarian carcinoma [20, 21]. Our group has recently shown an association between TFF1 and 2 with early-stage PDAC and that the combination of TFFs with CA19.9 stands out as a potential biomarker for discriminating early-stage disease from chronic pancreatitis (CP) and benign controls (BC) [22]. Even so, the functional significance of TFF1 in PDAC progression remains largely unexplored.

Studies from different laboratories have demonstrated an association between TFF1 and stress-related phenomena like senescence [23], DNA damage by chemotherapeutic drugs [24], and apoptosis induction in response to chemoresistance in various cancers [25]. TFF1, as a secretory protein, likely exerts these effects through binding to cell surface receptors. For example, TFF1 plays a crucial role in pancreatic tumorigenesis by suppressing oncogene-induced senescence (OIS) through EGFR [23]. While recent efforts are directed toward identifying cell surface receptors responsible for TFF1-mediated functional attributes in different cancers, no receptors have been established. Conversely, it has been shown that TFF2 binds to transmembrane G protein-coupled receptor CXCR4 on various cancer cell lines and activates Ca²⁺ signaling in Jurkat cells and MAPK signaling in PDAC [26]. The same axis has been linked with the BRAF V600E subtype of colon cancer [27]. Likewise, X-ray structure-based homology modeling studies indicate a high-affinity interaction between TFF3 and the CXCR4/CXCR7 heterodimeric complex, resulting in enhanced cell migration at the ocular surface [28].

Based on the direct correlation between TFF1 expression and the gemcitabine-resistant classical subtype of PDAC and the indication of possible interactions between different TFF proteins and chemokine receptors (CXCRs), we hypothesize that TFF1 is associated with gemcitabine resistance in PDAC and that it elicits these effects through binding to CXCR4. Our analysis of the TCGA dataset suggests that TFF1 expression is inversely correlated with the gemcitabine sensitivity ratio ($\mathsf{GSR}$), an indicator of gemcitabine resistance. Notably, the genetic ablation of TFF1 substantially improves the gemcitabine sensitivity in vitro experiments and pancreatic tumor xenografts. Our biochemical studies suggest that TFF1 binds to CXCR4 and plays a pivotal role in eliciting gemcitabine resistance by increasing stemness, anti-apoptotic genes, and migration propensity of pancreatic cancer cells.

2. Material and Methods

2.1. Cell culture and treatments

SW1990 and COLO357 PDAC cell lines were obtained from ATCC and grown in Dulbecco's Modified Eagle's medium (DMEM) containing high glucose (HyClone, Thermo USA), supplemented with 10% (v/v) fetal bovine serum and 1% penicillin-streptomycin (HyClone, Thermo, USA) at 37°C in a humidified atmosphere containing 5% CO2. The cell lines were authenticated using short tandem repeat profiling. Treatments with recombinant TFF1 (8289-TF-050, R&D), SDF-1α (300-28A, Peprotech), and AMD3100 (R&D) were performed for 24-48 h at 37°C. The samples were evaluated for further analysis using western blotting and cell proliferation experiments. The stable TFF1 knockdown (KD) cell lines SW1990 and COLO357 were generated by transduction with lentivirus shRNA plasmids (Dharmacon).

2.2. Measurement of Gemcitabine Sensitivity Ratio ($\mathbf{GSR}$)

Individual gene expression was downloaded from the cBioPortal database [29]. GraphPad Prism 6 (La Jolla, CA, USA) was used to perform the Pearson/Spearman correlation test (two-tailed) (La Jolla, CA, USA). For the calculation of the $\mathsf{GSR}$, four quantified gemcitabine metabolic genes were taken into account: $\mathsf{hENT}1$, $\mathsf{RRM}1$, $\mathsf{RRM}2$, and $\mathsf{DCK}$. $\mathsf{GSR}$ was first introduced by Nakano et al. [30]. Each gene's expression values were determined and used in the following formula to obtain the ratio.

$$\mathsf{GSR}=[\mathsf{hENT}1]\ast [\mathsf{DCK}]∕[\mathsf{RRM}1]\ast [\mathsf{RRM}2]$$

Because the genes required for metabolism ($\mathsf{hENT}1$ and $\mathsf{DCK}$) are in the numerator and the genes responsible for reducing the effectiveness of the drug ($\mathsf{RRM}1$ and $\mathsf{RRM}2$) are in the denominator, lower values of the ratio denote enhanced gemcitabine resistance.

2.3. Colony Forming Assay

For colony forming assay, 500 cells/well were plated in 2% FBS-containing medium. SW1990Scr and SW1990-TFF1-KD (shTFF1) cells were treated with gemcitabine (1 μM) for 14 days. Then, cells were fixed with 100% methanol, stained with 0.4% crystal violet, and colonies were counted.

2.4. Live cell imaging

The proliferative ability of cells upon TFF1 knockdown or AMD3100 treatments was measured by live cell imaging on Incucyte^R. Briefly, 3000-4000 cells were plated in 96-well plates, and each well was scanned for the phase contrast area or incorporation of nuclear proliferation dye Nuclight NIR (Essen BioScience, 4804) every 4 h for 24-48 h. At the end of the experiment, the data was analyzed for phase area per well and normalized to 0 h using Essen Bioscience software. Likewise, Annexin V dye (Essen BioScience, 4759) was used to measure apoptosis.

2.5. Migration Assay

0.5-0.8 x10⁶ cells were seeded in serum-free medium on the top chamber of polyethylene teraphthalate membranes (six-well insert, pore size 8 μm). 2 mL of 10% serum-containing medium was added to the lower chamber of the well, and the cells were allowed to migrate for 24-48 h depending upon chemotactic drive. Cells that did not migrate through the pores were removed by scraping the membrane with a cotton swab. The migrated cells on the lower side of the membrane were stained with a Diff-Quick Cell Stain Kit (Dade-Behring Inc., Newark, USA). Briefly, migrated cells were subjected to fixation (5 min), then cytoplasmic staining (5 min), and nuclear staining (5 min), and then quickly dipped 3 times in water to remove excess color. The membrane was cut with a scalpel and put on a microscopic slide. The immobilized membrane was imaged in ten random fields of view at 10X magnification. Cell numbers were counted and expressed as the average number of cells/fields of view.

2.6. Side Population and spheroid analysis

We performed a side population (SP) analysis to determine the cancer stem cell population. Briefly, 1X10⁶ cells/mL were used for each experimental group, including control and test. Control cells in each group (parental and gemcitabine-treated) were incubated with 50 μM Verapamil (Sigma, St Louis, MO, USA) for 15 min at 37°C. Then, cells were incubated for an additional 60 minutes at 37°C in a water bath with 5 μg/mL Hoechst 33342 dye. Cells were placed immediately on ice. SP and non-SP fractions were gated separately and analyzed using an LSRII flow cytometer (BD Biosciences, San Jose, CA, USA). For spheroid analysis, an equal number of SW1990Scr and shTFF1SW1990 cells were seeded in stem cell media in a low-attachment plate, and the spheroid forming ability of cells was assessed by real-time imaging with the spheroid module on Incucyte. The data was acquired for 8 days and analyzed using Essence software [31].

2.7. Western blot, immunofluorescence and immunohistochemistry

Immunohistochemistry (IHC) and immunofluorescence (IF) procedures were performed as described previously [32]. PDAC tissue microarray (TMA) was procured from Biomax, USA. Briefly, paraffin-embedded tissues were baked overnight at 58°C, washed with xylene to remove residual paraffin, and subjected to IHC or IF with the following antibodies at the indicated dilutions: TFF1 (polyclonal antibody generated by rabbit immunization with TFF1 peptide) at 3 μg/mL and anti-CXCR4 mAb (R&D, MAB172-SP) at 1:200 dilution, Ki67 (CST, D3B5), Cleaved caspase 3 (Asp175) (CST, 9661). For IHC, universal (mouse/rabbit) polymeric HRP conjugate (Vector Laboratories) was followed by detection with a DAB substrate kit and counterstain with hematoxylin. The stained sections were scored by Jesse L. Cox (Pathologist, UNMC). An intensity score (0–3; 0-negative, 1-weak, 2-moderate, 3-intense staining) and the percentage of positive cells (0–100%) for each tissue were reported. GraphPad Prism 9.0 was used to calculate respective H scores and p values. For IF, Alexa conjugated anti-mouse and anti-rabbit fluoroconjugate antibodies were used, and the images were captured using Zeiss LSM 800 confocal and analyzed using ZEN Imaging Software.

2.8. Apoptosis Array

Protein lysates (200 μg) prepared from gemcitabine-treated scrambled control and shTFF1SW1990 cells were analyzed by a Human Apoptosis Array Kit (Cat. # ARY009, R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s protocol. Appropriate dilutions of protein in the lysates were prepared as per the maximum allowable volume per array recommended by the manufacturer. The recommended quantity of lysates was diluted, pipetted onto the membranes, and incubated overnight at 2–8°C on a rocking platform shaker. Then, the biotinylated secondary antibody cocktail provided by the manufacturer was pipetted onto membranes and incubated for 1 h. After washing, the membranes were incubated with streptavidin-HRP provided by the manufacturer for 30 min. The signals were developed using chemiluminescent reagents and then exposed to X-ray films. The positive signals were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

2.9. Protein-Protein Docking Study

The docking analysis followed a similar methodology described previously [33]. Briefly, the three-dimensional structures of CXCR4 (PDB code: 3ODU) and TFF1 (PDB code: 1HI7) proteins were downloaded from the Protein Data Bank (PDB; http://www.rcsb.org/) and prepared using Protein Preparation Wizard in the Schrödinger suite [34, 35]. The protein-protein docking was performed using the BioLuminate module (Schrödinger Release 2018-2: BioLuminate, Schrödinger, LLC, New York, NY, 2018) present in the Schrödinger suite. BioLuminate searched for the 50 best complexes from 70,0000 possible protein-protein configurations.

2.10. Animal experiments

6–8-week-old male and female BALB/C athymic nude mice were used to study the effects of TFF1 knockdown on gemcitabine sensitivity in PDAC xenograft tumors. The animals used herein were approved by the Institutional Animal Care and Use Committee (IACUC) at UNMC and their care was taken in accordance with institutional guidelines. SW1990Scr and shTFF1SW1990 cells (1x10⁶ cells) in 100 μL of PBS were subcutaneously implanted on the side flanks of each mouse. After the formation of tumors was observed, the mice were randomized (n=4 mice per group) and treated with saline (PBS) or gemcitabine (12.5 mg/kg) every 3 days for 15 days via the intraperitoneal route. Tumor volumes were measured every 3^rd day using a caliper. At the end of the study, the mice were euthanized, and the tumors were excised, weighed, and subjected to endpoint analysis using IHC experiments. Tumor volumes were calculated by using the following formula: length x width² x 0.5.

2.11. Statistical analysis

Correlation analyses were performed using GraphPad Prism 10.0.02 (GraphPad Software Inc.). In all studies, data represent biological replicates (n) and are depicted as mean values ± SD as indicated in the figure legends. Comparison of mean values was conducted with unpaired, two-tailed Student’s t-test as indicated in the figure legends.

3. Results

3.1. Analysis of The Cancer Genome Atlas (TCGA) database identifies an inverse relationship between the $\mathbf{GSR}$ and TFF1

Comparison of TFF1, TFF2, and TFF3 expression in high and low-purity samples showed a significantly higher expression of TFF1 in patients with higher cellular (tumor) content, suggesting their secretion mainly from the epithelial cells (Figure 1A, Supplementary Figure S1A) Further transcript analysis of TFF1 in TCGA and GTEx databases of pancreatic adenocarcinoma (PAAD) (n=179) indicated a significantly higher median expression (~9 transcripts per million) than normal cases (n=171) (Figure 1B). The stagewise expression analysis of PDAC patients suggests an increase in TFF1 levels in stage II and III patients (Figure 1C), validating our previous observations of high TFF1 in PanIN 2 and 3 patients [22], which may explain the poor overall survival (OS) of the high TFF1 expressing patients (Figure 1D). To understand the association of TFF1 with gemcitabine resistance in PDAC, we analyzed the correlation between TFF1 and the $\mathsf{GSR}$, an indicator of gemcitabine resistance [30]. The $\mathsf{GSR}$ was first developed by Nakao et al., which considers a balance of the cellular enzymes of gemcitabine transport and metabolism, including $\mathsf{hENT}1$, $\mathsf{dCK}$, $\mathsf{RRM}1$, and $\mathsf{RRM}2$. We have utilized the publicly available TCGA database (TCGA provisional, Pancreatic Cancer) as our primary source of clinical information and TFF1, $\mathsf{hENT}1$, $\mathsf{dCK}$, $\mathsf{RRM}1$, and $\mathsf{RRM}2$ gene expression values for calculating $\mathsf{GSR}$. Our analysis indicated that TFF1 is negatively correlated with $\mathsf{GSR}$ (Spearman r=−0.2813, p=0.0001) in PDAC samples (n=179) (Figure 1E) and in a panel of PDAC cell lines (Supplementary Figure S1B), suggesting its association with gemcitabine resistance. As discussed previously, the classical subtype has been shown to exhibit gemcitabine resistance. Therefore, to identify the correlation between TFF1 expression and $\mathsf{GSR}$ in the classical subtype, we divided TCGA patient samples into the classical, basal, exocrine, and quasi-mesenchymal (QM) subtypes using genomic characterization [36]. The patient samples, termed classical by one group and basal by another group and vice versa, were excluded from the analysis to minimize the incongruities. We found the lowest $\mathsf{GSR}$ in the classical subtype and the highest $\mathsf{GSR}$ in the quasi-mesenchymal (QM) subtype (Figure 1F, Supplementary Figure 1C). Additionally, we observed that TFF1 is negatively correlated with the classical subtype (Spearman r=−0.3043). In contrast, it positively correlated with basal (Spearman r=0.00602) and quasi mesenchymal (Spearman r=0.04151) subtypes (Supplementary Figure 1C, 1D). We also noted a negative correlation value in the exocrine subtype, but it was insignificant (Spearman r=−0.112) (Supplementary Figure 1E). Extending our correlation analysis between individual molecules of $\mathsf{GSR}$, including $\mathsf{hENT}1$, $\mathsf{dCK}$, $\mathsf{RRM}1$, $\mathsf{RRM}2$, and TFF1, we found a negative correlation between TFF1 and $\mathsf{hENT}1$ (Spearman r=−0.076, p=0.31), TFF1 and $\mathsf{dCK}$ (Spearman r=−0.18, p=0.019) and a positive correlation between TFF1 and $\mathsf{RRM}1$ (Spearman r=0.062, p=0.41) and TFF1 and $\mathsf{RRM}2$ (Spearman r=0.019, p=0.013) (Supplementary Figure S1F). The results suggest a strong association between TFF1 and gemcitabine resistance in PDAC. These results are further substantiated by our individual analysis of the classical subtype, where we observed a very high expression of TFF1 compared to the basal subtype (Figure 1G) and a strong correlation with $\mathsf{GSR}$ (Figure 1H). To further assess these implications, we tested a series of PDAC cell lines for TFF1 expression. Cell lines SW1990, SU86.86, and CD18 (inherent gemcitabine resistance) displayed the highest expression of TFF1, followed by a moderate expression in COLO357, MiaPaca-2, and AsPC1 (Figure 1I).

Figure 1:

TFF1 is associated with poor survival and disease progression of GT patients. A. TFF1 analysis in cellularity corrected TCGA dataset. B. TFF1 expression analysis from GEPIA in PDAC cases (n=179) and normal healthy volunteers (n=171). C. TCGA dataset analysis for determining the stage-wise expression of TFF1 in PDAC patients. D. Overall survival analysis of high and low TFF1 PDAC patients showing TFF1 association with poor overall survival (OS) in TCGA datasets. E. Spearman correlation analysis between TFF1 transcripts and gemcitabine sensitivity ratio ($\mathsf{GSR}$) in the TCGA (Provisional) PDAC clinical dataset (n=179) F. Dot plot showing $\mathsf{GSR}$ correlation with different PDAC subtypes. The radius of the sphere indicates the expression of TFF1. G. The TCGA expression analysis of TFF1 in classical and basal subtypes. H. Spearman correlation between TFF1 and $\mathsf{GSR}$ in the classical subtype of PDAC patients. I. Immunoblot analysis for checking TFF1 expression in a panel of PDAC cell lines (AsPC1, COLO357, CD18, T3M4, MiaPaca-2, SU86.86, and SW1990) and normal human pancreatic epithelial (HPNE) cells. β-actin was used as a loading control. The data are presented as mean±SD. p<0.05; p<0.001.

3.2. TFF1 expression increases upon gemcitabine treatment, and its knockdown sensitizes PDAC cells to gemcitabine.

We observed a significant correlation between TFF1 and gemcitabine resistance in PDAC cell lines and patient datasets. Therefore, we assessed the effect of long-term gemcitabine treatment on TFF1 expression in PDAC cell lines. Treatment with gemcitabine resulted in a 5-fold decrease in $\mathsf{GSR}$ in SW1990 cells (Figure 2A). Accordingly, a ~5-6-fold increase in TFF1 expression was observed in SW1990 and COLO357 PDAC cell lines at the transcript level (Figure 2B and C) and protein level at 48 h (Figure 2D). To identify the role of TFF1 in gemcitabine tolerance in PDAC cells, we knocked down TFF1 in SW1990 and COLO357 cells and confirmed the decrease in TFF1 expression at the transcript level by qPCR (Figure 2E and 2F) and at the protein level by immunoblotting; evaluating the level of TFF expression in shTFF1SW1990 and COLO357 cells compared to scramble controls (Figure 2G). The effect of TFF1 KD (shTFF1) on the proliferative ability was checked on the Incucyte^R live imaging system, and we observed a drastic reduction in the proliferation of shTFF1SW1990 (1.1 to 0.8, p=0.035) and shTFF1COLO357 cells (1.2 to 0.9, p= 0.023) compared to the scrambled controls. These KD cells were also more sensitive to gemcitabine treatment (0.8 to 0.6, p=0.0023; 1 to 0.6, p=0.0052), indicating the importance of TFF1 in maintaining the proliferative ability of cancer cells and their resistance to gemcitabine (Figure 2H and 2I). We observed a similar trend in the colony formation assay where the colony formation ability of shTFF1SW1990 cells was decreased by 36% (p= 0.03) compared to scrambled controls, with further reduction in the colony counts noted in the gemcitabine-treated (GT) shTFF1 cells (50%, p=0.0013) as compared to scrambled controls (Figure 2J).

Figure 2:

TFF1 expression increases upon gemcitabine treatment and is associated with gemcitabine tolerance. A. RT-quantitative PCR (QPCR) analysis of $\mathsf{GSR}$ in long-term GT SW1990 PDAC cells than parental control. B and C. QPCR analysis of long-term GT SW1990 and COLO357 cells indicating an increased expression of TFF1 at the transcript level. D. Immunoblot analysis of TFF1 in gemcitabine-treated PDAC cell lines SW1990 and COLO357. β actin was used as a loading control. E and F. QPCR analysis of TFF1 upon knockdown (KD) using lentiviral vector system in SW1990 and COLO357 cells. TFF1 expression was plotted as fold change. G. Immunoblot analysis of TFF1 KD SW1990 and COLO357 cells indicated a substantial decrease in TFF1 at the protein level. β -actin was used as the loading control. H and I. The proliferative ability of scrambled control, shTFF1SW1990, and shTFF1COLO357 cells in the presence of gemcitabine. The proliferation was measured by the incorporation of nuclear proliferation dye (NIR) into the cells on an incucyte live imaging system. J. The effect of long-term gemcitabine treatment on the survival of SW1990 cells by measuring their colony formation ability in the presence of gemcitabine. Bar diagram showing a decrease in the average number of colonies in shTFF1SW1990 compared to scrambled control cells after treatment with gemcitabine (1 μM). The data are presented as mean±SD. p<0.05; p<0.001.

3.3. TFF1 knockdown decreases stemness and augments the apoptosis of GT PDAC cells.

Multiple studies suggest that gemcitabine resistance in PDAC is directly associated with enhanced stemness [8, 37]. To investigate the effect of gemcitabine on cancer cell stemness, we performed a flow cytometry-based side population (SP) analysis of GT PDAC cells. We observed an increase in the % SP in SW1990 (3%, p<0.05) (Figure 3A) and COLO357 cells (1.5%, p<0.05) (Figure 3B). The knockdown (KD) of TFF1 in SW1990 cells resulted in a decrease in % SP population (0.2%) compared to scrambled control (1.2%) (Figure 3C). The transcript analysis of stemness genes in GT SW1990 cells indicated an upregulated expression of SOX2, SOX9, NANOG, and OCT3/4 compared to control parental cells (Figure 3D). Our bioinformatics analysis of PDAC patients suggested a strong correlation between TFF1 and SOX9 (Supplementary Figure 2A). To substantiate these findings, we performed a QPCR analysis of these stemness genes in TFF1 KD cells. Interestingly, we observed a significant decrease in the relative transcript expressions of SOX9 and OCT3/4 in both shTFF1SW1990 and shTFF1COLO357 cells, thereby validating TFF1 association with stemness in PDAC (Figure 3E). Notably, the spheroid formation ability of SW1990 cells was significantly reduced in TFF1 silenced cells than scrambled controls, as seen by a decrease in total area (p= 0.058) measured by real-time imaging on incucyte (Supplementary Figure 2B). These findings indicate an association of TFF1 expression with PDAC stemness.

Figure 3:

TFF1 knockdown (KD) increases gemcitabine sensitivity in PDAC cell lines by decreasing stemness and augmenting apoptosis. A and B. Effect of long-term gemcitabine treatment in the side population (SP) cells. SW1990 and COLO357 cells were stained with Hoechst 33342 dye in the presence (left) or absence (right) of verapamil and analyzed by flow cytometry. The SP, which disappears in the presence of verapamil, was gated and shown as a percentage of the viable cell population. The corresponding bar diagram represents the percentage of SP populations (p<0.05). C. Effect of TFF1 KD on the stemness of SW1990 cells measured as % side population (SP). D. QPCR analysis of long-term GT SW1990 cells for stem cell markers (SOX2, SOX9, Oct3/4, Nanog, and CD44) at transcript level compared to respective parental cells. E. The impact of TFF1 knockdown on the transcript levels of stemness genes in shTFF1 SW1990 and COLO357 cells. The fold changes are normalized to scrambled control transcripts and, therefore, are shown as relative fold changes. F. The effect of gemcitabine and staurosporine on the apoptosis of SW1990Scr and shTFF1SW1990 cells in vitro. Apoptosis was detected by dual dye staining using annexin V-FITC/PI. The percentage of apoptotic cells in the presence of apoptotic inducer staurosporine (2μM, 8h) and gemcitabine (1 μM, 24h) is presented as the mean ± SD in the corresponding bar diagram. G. Increase in the apoptosis of shTFF1COLO357 cells upon gemcitabine treatment as seen by increased annexin V intensity. H. Immunoblot showing the presence of apoptotic markers cleaved caspase 3 and cleaved PARP in COLO357 cells upon gemcitabine treatment. The data are presented as mean±SD. p<0.05;p<0.001.

To decipher the implications of TFF1 in gemcitabine resistance in PDAC through the anti-apoptotic pathway, we treated both scrambled and shTFF1SW1990 and shTFF1COLO357 cells with apoptosis-inducer staurosporine and gemcitabine. Our result suggested a significant upregulation of apoptotic cells (15%) in shTFF1SW1990 cells as compared to scrambled control (5%) (p<0.05) upon staurosporine treatment (Figure 3F). A similar analysis for apoptosis in shTFF1COLO357 cells was performed using annexin V staining intensity measurement on live cells using the Incucyte system, which revealed an increase in annexin V staining in shTFF1 cells than in the scrambled control and demonstrated that gemcitabine treatment further augmented apoptosis as seen by increased annexin V staining intensity (Figure 3G). These results were corroborated by observing an increase in cleaved caspase-3 (ClC3) and cleaved PARP forms in GT lysates of COLO357 cells (Figure 3H). Next, we investigated the potential mechanisms by which TFF1 modulates apoptosis in SW1990 PDAC cells by screening targets using a human apoptosis antibody array. The array analysis indicated increased cleaved caspase with a reduction in Bcl-X, HSP-70, cIAP in GT shTFF1SW1990 cells (Supplementary Figure 2C and 2D). These results suggest TFF1's contribution to gemcitabine resistance by modulating stemness and anti-apoptotic pathways in PDAC cells.

3.4. TFF1 elicits gemcitabine resistance through binding to CXCR4.

To determine the axis involved in imparting TFF1-mediated gemcitabine resistance, we envisaged the possible involvement of chemokine receptors. TFFs are secretory molecules believed to exert their effect by binding to chemokine or growth factor receptors [15], and chemokine receptor CXCR4 has recently been described as a low-affinity receptor for TFF2 [26]. We have also observed relatively high expression of TFF1 in the classical subtype of PDAC (Figure 1G), which may indicate that TFF1 is available as a ligand for CXCR4 in this patient cohort. In addition, we observed a lower expression of CXCL12 (SDF-1α), a well-known ligand of CXCR4, in the classical subtype compared to the basal subtype (Figure 4A). Therefore, we explored the possibility of CXCR4 as a signaling receptor for TFF1 and sought to understand the involvement of the TFF1-CXCR4 axis in stemness and gemcitabine resistance in the classical subtype of PDAC. Accordingly, we extended our studies to identify possible interactions between TFF1 and the CXCR4 using X-ray structure-based 3D modeling and analysis (Figure 4B). TFF1 interacts with transmembrane helices III and IV present at the N-terminus of CXCR4. TFF1-CXCR4 complex is formed by pi-stacking between amino acid side chains of CXCR4 and TFF1 (CXCR4; Phe172-Chain A: TFF1; Phe60-chain D), hydrogen bonds (CXCR4; Ser-122-Chain A: TFF1; Glu-55-chain D), and (CXCR4; Glu-1-chain C: TFF1; Ile-89-chain A) (Figure 4C). In a nutshell, these modeling results indicate a possible interaction between TFF1 and CXCR4.

Figure 4:

TFF1 interacts with CXCR4 in PDAC. D. TCGA analysis of CXCL12 expression in the classical and basal subtypes of PDAC indicates no differences in the expression in different subtypes. B and C. In silico analysis of TFF1-CXCR4 interaction using BioLuminate. 3D model generated through Bio Luminate indicating the interaction of TFF1 with CXCR4 at Phe-172, Ser-122, and Glu-1 residues through pi-stacking and hydrogen bonding. The docking display of the CXCR4-TFF1 complex is shown in 4B. CXCR4 is represented as a surface figure (cyan color), and TFF1 is depicted in spherical form (green, red, blue, and gray color). D. Real-time binding kinetics of dimeric form of TFF1 to HEK293 cell surface expressed CXCR4 using surface plasmon resonance (SPR). Cu²⁺ maintains the dimeric state of TFF1 and improves the binding to CXCR4. E. Immunohistochemical (IHC) analysis of TFF1 and CXCR4 in stage II and III PDAC patient tumors. The pathologist scored the slides, and the H-scores were calculated using the following formula. H-score= % staining x intensity of staining* 0.5. The Correlation plot showing a direct correlation between TFF1 and CXCR4 in PDAC tumors is shown in parallel. Spearman correlation analysis was performed on GraphPad Prism 10. F. Immunofluorescence analysis of TFF1 (red) and CXCR4 (green) in primary PDAC tumors and the normal pancreas tissue. The graph represents the fluorescence intensity profile calculated on merged images acquired for PDAC tumors stained for TFF1 (red) and CXCR4 (green).

These findings were further validated in real-time kinetic analysis of TFF1 and CXCR4 using surface plasmon resonance (SPR). For this, His tagged CXCR4 was immobilized on a Ni-NTA chip and TFF1 was passed over as an analyte at different concentrations for an association time of 180 s, and dissociation for 300 s. To our surprise, initial experiments of monomeric TFF1 interaction with HEK293T transfected CXCR4 supernatant did not show any binding. Previous studies have indicated the functional activity of TFF1 in the dimeric form in cell-based assays, where Cu²⁺ was essential in maintaining the dimeric form of TFF1 [38, 39]. In line with this, we evaluated the binding between dimeric TFF1 and CXCR4 under similar conditions. Interestingly, the binding was found to be very specific, with no binding on the blank channel (Figure 4D). The addition of Cu²⁺ enhanced the binding of the TFF1 dimer. Of note, the level of binding of TFF1 was lower than CXCL12 (SDF-1α) (Supplementary Figure 3A), a well-known ligand of CXCR4, suggesting TFF1's presence as a low-affinity ligand. The surface competition experiment with TFF1 and SDF-1α in the SPR experiment revealed that TFF1 could not bind to SDF-1α bound CXCR4 (Supplementary Figure 3B), further suggesting that TFF1 binds to CXCR4 and shares a binding pocket with SDF-1α. This intrigued us to investigate further how TFF1, a low-affinity ligand, can compete with SDF-1α for CXCR4 binding and induce gemcitabine resistance. Notably, our bioinformatics analysis suggested a very low expression of SDF-1α in the classical subtype (Figure 4A). It is quite possible that TFF1, owing to its predominant presence in the classical PDAC subtype, interacts with CXCR4 and contributes to gemcitabine resistance. Furthermore, spearman correlation analysis for TFF1 and CXCR4 expression in IHC experiment with stage II and III PDAC tumors suggested a statistically significant positive correlation (R²=0.5724, p=0.0028) between TFF1 and CXCR4 (Figure 4E). The results from immunofluorescence experiments hint towards a significant co-localization and co-expression of TFF1 and CXCR4 in primary PDAC tumors, which in turn indicates a plausible involvement of the TFF1-CXCR4 axis in gemcitabine resistance (Figure 4F).

3.5. TFF1 contributes to gemcitabine resistance through CXCR4 downstream signaling.

To decipher the contribution of TFF1 in imparting gemcitabine resistance properties to cancer cells through CXCR4, we treated TFF1 low-expressing COLO357 cells with TFF1 and SDF-1α in the presence of gemcitabine. We observed an expected increase in the annexin V staining intensity in GT COLO357 cells. However, the cells recovered from apoptosis with the exogenous addition of TFF1 (p=2.16e-9), corroborating our results indicating TFF1’s contribution to gemcitabine resistance (Figure 5A). The annexin V staining intensities were comparable between TFF1 and SDF-1α treatment (p= 7.19e-10), suggesting that both ligands exhibited similar resistance effects. The co-treatment of GT COLO357 cells with TFF1 partially rescued their death through apoptosis and augmented their proliferation (Figure 5B), corroborating the above-mentioned findings of TFF1 association with gemcitabine resistance. Next, we validated TFF1-CXCR4 interactions through inhibitor experiments in which the CXCR4 receptor was blocked with its widely used inhibitor, AMD3100. The co-treatment of TFF1 and GT COLO357 with AMD3100 (1 μM) reverted the anti-apoptotic effects of TFF1 (p=1.37e-6), thereby increasing annexin V staining (Figure 5C). To further understand the mechanistic signaling associated with TFF1-mediated gemcitabine resistance, we treated COLO357 and MiaPaca-2 cells with TFF1 and SDF-1α. TFF1 treatment at 500 nM increased the phosphorylated levels of Akt (pAkt^S473) and ERK (pERK^Y202/204) with maximal increase at 60 min, thereby eliciting the pAkt/pERK signaling axis associated with gemcitabine resistance (Figure 5D). On the other hand, SDF-1α treatment at 100 nM elicited pERK signaling at 30 min, indicating differential affinities of TFF1 and SDF-1α for CXCR4. Furthermore, gemcitabine tolerance with TFF1 was observed at 5-fold higher concentrations (100 nM) than with SDF-1α (10 nM) (Figure 5C), substantiating the findings as mentioned earlier of different affinities of TFF1 and SDF-1α to the CXCR4 receptor. Extending our studies on the functional implications of the TFF1-CXCR4 axis in mediating gemcitabine resistance, we found that the migration (p<0.002) and colony formation ability (p<0.05) of SW1990 cells was significantly increased by TFF1 treatment. Treatment with AMD3100 abrogated these functions of PDAC cells (Figure 5E and 5F). Conversely, treatment with AMD3100 (1 μM) substantially decreased pAkt^S473 and pERK^Y202/204 proteins in TFF1 or SDF-1α treated COLO357 cells, supporting the conjecture that TFF1 exerts gemcitabine resistance in PDAC cells by binding to CXCR4 (Figure 5G).

Figure 5:

TFF1 exerts gemcitabine resistance through CXCR4 in PDAC. A. Apoptosis analysis of gemcitabine-treated COLO357 cells upon the exogenous addition of TFF1 and SDF-1α (a well-known ligand of CXCR4). Apoptosis was measured as annexin V (green) mean intensity in an Incucyte live imaging system. B. The proliferation assay to check the effect of exogenously added TFF1 or SDF-1α on gemcitabine-treated COLO357 cells showing TFF1 mediated rescue from apoptosis and increased proliferation. C. Apoptosis assay with gemcitabine-treated cells in the presence of CXCR4 antagonist AMD3100 (10 nM) that abolished the CXCR4 mediated effects, as seen by an increase in annexin V stained COLO357 cells. D. Immunoblot analysis of TFF1 mediated pAkt/pERK signaling in COLO357 and MiaPaca-2 cells. A parallel analysis with SDF-1α was performed. The cells were treated for 30 and 60 min, and the lysates were collected to measure the Akt and ERK phosphorylated levels. β-actin was used as a loading control. E. Trans well assay to check the migration ability of SW1990 cells in the presence of TFF1 and CXCR4i AMD3100. After preincubation with AMD3100 (1 μM) for 1 h, with indicated drug concentration, and allowed to migrate for 24 h. Error bars mean ± SD (n=5). F. Determining TFF1-mediated colony formation in COLO357 cells. AMD3100 was used to determine TFF1-CXCR4 mediated effects. Crystal violet was dissolved with DMSO, and optical density was measured. Error bars mean ± SD (n=3). G. After preincubation with AMD3100 (1 μM) for 1 h and culture in medium with the indicated drug concentrations (1 μM AMD3100, 100 ng/ml TFF1) for 24 h, cells were subjected to Western blots. Treatment with TFF1 recombinant protein alone and combined with CXCR4 inhibitor (AMD3100) reveals that TFF1 modulates downstream signaling partly through CXCR4. The data are presented as mean±SD. p<0.05; p<0.001.

3.6. Silencing of TFF1 sensitizes PDAC tumors to gemcitabine in vivo.

Next, we evaluated the impact of TFF1 on the tumorigenicity of PDAC cells and how their knockdown impacts the sensitivity of PDAC cells to gemcitabine treatment in vivo. We determined the tumorigenic potential of TFF1 by subcutaneously implanting an equal number of both SW1990Scr and shTFF1SW1990 cells on the side flanks of athymic nude mice in different groups (1x10⁶, 1x10⁵, 1x10⁴, 1x10³). Subsequently, the tumor growth was followed over six weeks, indicating a drastic reduction in the tumor onset in the knockdown group compared to scrambled control cells at cell numbers 1x10⁶ and 1x10⁵ (Supplementary Figure 4). No tumors were observed in low cell numbers (10⁴, 10³). In parallel, a group of mice with subcutaneously implanted cells (1x10⁶) were subjected to gemcitabine treatment through an intraperitoneal (i.p) route at 12.5 mg/kg for two weeks (4 doses) (Figure 6A). At the end of the experiment, we collected tumors for the end-point analysis of proliferation, apoptosis, and TFF1 expression. We observed a statistically significant decrease in tumor volumes (~2-fold, p=0.0390) (Figure 6B) and a drastic reduction in tumor weights in shTFF1 tumors (mean wt =0.65g) than scrambled control cells (1.7g) (Figure 6C). Cumulatively, these results indicated the tumorigenic properties of TFF1 in PDAC cells. Consistent with our in vitro findings, we observed a substantial impact of gemcitabine treatment in TFF1 knockdown tumors than the scrambled control group, as seen by a drastic reduction in both tumor volumes (p=0.0080) and growth (mean wt.= 0.13g) (Figure 6 B and 6 C). To mechanistically investigate the sensitivity of TFF1 silenced tumors to gemcitabine, we performed IHC studies for proliferation index (Ki-67), apoptosis (cleaved caspase 3), and TFF1 expression on tumors from different groups. These studies indicated a high expression of TFF1 in the gemcitabine-treated scrambled control group (Figure 6 D), thereby highlighting a potential role of TFF1 in PDAC tumorigenicity and gemcitabine resistance. In parallel, IHC analysis for Ki67 and ClC3 in these tumors demonstrated a significant decrease in the % of Ki-67 positive cells in the shTFF1SW1990 knockdown tumors (30%) compared to scrambled control (46%) with the lowest expression in gemcitabine treated shTFF1 tumors (23%), suggesting an impact of TFF1 on the tumorigenicity of PDAC cells (Figure 6E). Furthermore, the maximum number of ClC3 positive cells in gemcitabine-treated shTFF1SW1990 tumors hinted towards TFF1 potential in anti-apoptotic gene induction (Figure 6F). Collectively, our in vivo studies corroborated in vitro data and provided compelling evidence that diminishing TFF1 sensitizes PDAC cells to gemcitabine, highlighting its potential as a sensitizer to gemcitabine chemotherapeutic regimens.

Figure 6:

TFF1 confers gemcitabine resistance in subcutaneous xenograft model of PDAC. A. Experimental schema and treatment strategy for evaluating the effects of TFF1 KD on gemcitabine efficacy in subcutaneously implanted SW1990Scr and shTFF1SW1990 cells. Gemcitabine (12.5 mg/kg) or saline was administered through intraperitoneal injection every 3 days for a period of 4 doses. B. Tumor volumes were measured using a caliper every 3 days post-treatment. C. Pictorial and quantitative representation of SW1990Scr and shTFF1SW1990 tumors excised at the end of the experiment. The tumors were weighted and plotted on GraphPad prism 10.0.02. Immunohistochemical (IHC) analysis and quantitative data for D. TFF1 expression in gemcitabine treated vs saline control SW1990 tumors. E. % of intratumoral Ki-67 cells showing change in % proliferation of tumors and F. number of cleaved caspase 3 (ClC3) positive cells indicative of cell apoptosis. The slides were scored by the pathologist for three markers. Error bars indicate standard deviation.

4. Discussion

Chemoresistance has long been considered a foremost impediment in the survival outcomes with existing treatments in PDAC [40, 41], and the result is that the OS for PDAC patients has not been significantly ameliorated in the last two decades [8, 39]. Hence, delineating the molecular mechanisms and pathways behind treatment failure due to chemoresistance demands immediate action to formulate effective personalized therapy for PDAC [6, 42, 43]. Herein, we have reported the contribution of TFF1 in imparting gemcitabine resistance properties to cancer cells. Mechanistically, we described an autocrine impact of TFF1 on PDAC cells, driven partly by its binding to CXCR4, resulting in an enhanced stemness and survival and migration of GT PDAC cells. TFF1 is a significantly upregulated molecule in the classical subtype and demonstrated tumorigenic potential in PDAC [23]. The combination of our analysis of TFF1 and $\mathsf{GSR}$ correlation (negative) and OS from TCGA data in this study demonstrated that TFF1 is strongly associated with gemcitabine resistance and, thereby, poor survival of PDAC patients. Many recent studies, including ours, have shown the involvement of TFFs-associated mucins MUC4 and MUC5AC in imparting gemcitabine resistance to pancreatic cancer cells [44-46].

We analyzed the TCGA-PAAD datasets and the subtype information as defined by previously published reports. A negative correlation between the classical subtype and $\mathsf{GSR}$, a selective and specific enrichment of TFF1 in the classical subtype, and a negative correlation between TFF1 and $\mathsf{GSR}$ cumulatively strengthen the hypothesis that TFF1 is involved in gemcitabine resistance in a subtype-specific manner. It is conceivable that drug tolerance is a subtype-specific attribute in cancer cells. In line with this, a study by Heiser et al. that aimed to evaluate the activity of FDA-approved and investigational compounds on a panel of ~50 different breast cancer cell lines, which are well-defined based on transcriptional and genomic subtype, demonstrated that at least one-third of the drugs showed a subtype-specific response [47]. They also employed a super pathway network to identify potential pathways associated with subtypes exhibiting drug sensitivity, and the analysis indicated the upregulation of FOXA1, a transcription factor regulating TFF1 expression in the luminal subtype of breast cancer, and this subtype showed great sensitivity to the drugs Lapatinib, Trichostatin A, Triciribine among others. Previous literature [11, 12], along with our in-silico analysis, indicates GATA-6 presence in the classical subtype with a negative $\mathsf{GSR}$ correlation and association with TFF1 secretion, thereby suggesting a role of GATA-6 as a master regulator of TFF1-mediated gemcitabine resistance (data not shown). However, extensive research is required to understand their subtype-specific regulation. Nonetheless, our in vitro results of TFF1 enrichment in GT PDAC cells at the transcript and protein levels corroborate the in silico findings of TFF1 association with gemcitabine resistance.

Gemcitabine resistance in PDAC is characterized by the induction of anti-apoptotic mechanisms, increased stemness, EMT, autophagy, and enhanced invasion and migration of cancer cells [31, 48-50]. Accordingly, multiple studies have demonstrated that TFF1 is associated with the induction of EMT and increased migration and invasion of cancer cells [51]. These findings cumulatively persuaded us to mechanistically delineate the potential contribution of TFF1 in gemcitabine resistance in PDAC. Further, an earlier report showed that TFF1 could inhibit doxorubicin-mediated apoptosis in breast cancer cell lines [25]. TFF1 has also been shown to be involved in anti-apoptotic mechanisms in other cancers [15, 52]. The results of our apoptosis array and annexin V staining experiments with GT scrambled and TFF1 KD PDAC cells indicate the role of TFF1 in upregulating anti-apoptotic pathways as a mechanism of gemcitabine resistance. While our studies provided evidence of the involvement of TFF1 in gemcitabine resistance, further exploration is needed to analyze its role in erlotinib tolerance, as the classical subtype also demonstrated higher erlotinib sensitivity.

Though the phenotype of gemcitabine-resistant pancreatic cancer cells and the effect of increased TFF1 are similar, it is the first-ever report on direct exploration of a connection between TFF1 and gemcitabine resistance in PDAC. Extending our studies on the possibility that TFF1 might modulate gemcitabine resistance by reprogramming tumor cells into cancer stem cells (CSCs), we carried out SP analysis and evaluated the spheroid forming ability of TFF1 scrambled and KD cells under GT versus untreated conditions. Interestingly, a significant increase in the percentage of SP cells upon gemcitabine treatment of PDAC cells and a concomitant decrease in the percentage of SP cells upon TFF1 KD support the notion that TFF1 is responsible for the enrichment of stemness in gemcitabine resistance. Our in silico correlation analysis of TFF1 and stemness in the classical subtype and QPCR data highlighted a specific enrichment of stemness genes, including SOX2, SOX9, and NANOG in GT PDAC cells and patient datasets, validating TFF1 association with stemness. These findings are consistent with previous studies showing the role of SOX9 in maintaining stemness and the inhibition of apoptosis responsible for drug resistance in ovarian cancer cells [53]. Indeed, we have recently observed an association of TFF1-associated MUC5AC with stemness in PDAC [54], which stimulates our interest in further investigating if TFF1 and MUC5AC complex together to elicit these resistance mechanisms.

To further investigate how TFF1 induces stemness and anti-apoptotic signaling for gemcitabine resistance, we utilized literature on possible receptors for TFF-mediated functions [28, 55]. Based on recent X-ray 3D structure modeling studies identifying TFF3 interaction with chemokine receptors, including CXCR4 and CXCR7 [28], we employed similar modeling analysis and observed a strong interaction between TFF1 and CXCR4. Another study has established CXCR4 as a low-affinity receptor for TFF2 [26]. TFF1 binding to CXCR4 was proven in real-time kinetic analysis using SPR, and the results established that dimeric TFF1 is a low-affinity ligand for CXCR4. The novel finding of TFF1 binding to CXCR4 in dimeric form suggests a need for TFF1 dimerization for binding to CXCR4 and elicit functional effects. Indeed, a study by Pest et al. in breast cancer cells [56] showed that TFF1 interaction with its receptor is facilitated by its dimerization. Recombinant dimeric TFF1 (linked by a disulfide bond) was 8-fold more potent than monomeric form in stimulating the migration of breast cancer cells. Indeed, our functional assays to decipher the implications of the TFF1-CXCR4 axis in gemcitabine-associated signaling revealed that TFF1 exerted these effects at ~5-10-fold higher concentrations than SDF-1α, further corroborating the starring role of TFF1 as a low-affinity ligand. The potential of CXCR4 in anti-apoptosis, tumorigenesis, stemness, invasion, migration, EMT, and chemoresistance mechanisms in PDAC is well documented [57]. Of note, a previous study has shown the involvement of the SDF-1α-CXCR4 signaling axis in gemcitabine resistance in pancreatic cancer cells, and they reported an elevation in CXCR4 levels in PDAC lines upon treatment with gemcitabine [58], similar to what we observed with TFF1. These results motivated us to explore how TFF1 functions as a ligand for CXCR4 in the presence of its well-known high-affinity ligand SDF-1α. Our findings of a higher expression of TFF1 in the classical subtype compared to the basal subtype, with a low expression of SDF-1α, infers that TFF1 may bind to CXCR4 in the classical subtype and exhibit gemcitabine resistance in a subtype-specific manner. However, a deep mechanistic investigation is warranted to thoroughly understand the subtype-specific role of TFF1 and therapy response in PDAC datasets and patients’ tumors. This analysis will help researchers and clinicians employ TFF1-based subtyping in patient stratification and therapy response [59].

The relevance of TFF1 binding to CXCR4 and its relation to gemcitabine resistance was validated either by exogenous addition of TFF1 or using CXCR4 antagonist AMD3100 in our in vitro experiments. Notably, treatment with dimeric TFF1 decreased gemcitabine-mediated apoptosis of COLO357 cells, which was reverted by AMD3100, indicating that TFF1 binds to CXCR4 and imparts gemcitabine resistance. These studies with PDAC cells clearly emphasize the exclusive role of epithelial cells secreting TFF1 as SDF-1α is mainly secreted from the pancreatic stellate cells (PSCs)/stromal components [60]. Furthermore, the functional effects of TFF1-CXCR4 interactions, such as migration and enhanced cell proliferation, were recapitulated in cell line experiments with TFF1 and AMD3100. Besides, the strong correlation between TFF1 and CXCR4 expression in stage II and III PDAC patients suggests a potential implication of this axis in disease advancement and poor OS [57].

CXCR4 transmembrane helices contain many critical residues that undergo conformational changes upon ligands' activation, thus allowing signaling transmission [61]. In the future, it would be interesting to utilize mutagenesis experiments to study and validate particular interactions of TFF1 with CXCR4 transmembrane helices. Our signaling experiments suggest that TFF1 mediates downstream signaling through CXCR4, as AMD3100 partially blocks it. One of the potential explanations for this partial blocking is that a critical residue Phe-172 (Transmembrane-IV) on CXCR4 is essential for AMD3100 binding, which was also identified as a binding site for TFF1 through our 3D modeling studies. These observations suggest that AMD3100 may block TFF1 binding to CXCR4 through Phe-172 while not interfering with Ser-122 and Glu-1 of CXCR4, and they explain the partial effects observed with AMD3100. It was earlier speculated that this phenylalanine residue is crucial for interactions with the aromatic linker of AMD3100, which may change the overall configuration of the CXCR4 protein. Additionally, it will be interesting to study whether TFF1 has any self-induction/feedback loop mechanism through CXCR4. Overall, our studies highlight the novel role of TFF1 in PDAC gemcitabine resistance. Additionally, we have demonstrated that TFF1 propagates downstream signaling through CXCR4, which needs further investigation using site mutagenesis experiments and tumors from patients with a subtype specificity. Recent enthusiasm in PDAC subtyping at the transcriptional and genetic levels for patient stratification [59] and effective therapy responses impel an in-depth understanding of the molecular mechanisms of TFF1 upregulation and subtype specificity. Moreover, the present study highlights the potential implications of targeting the TFF1-CXCR4 axis for improving gemcitabine efficacy in PDAC patients.

Highlights.

• The de novo expression of Trefoil factor 1 (TFF1) in PDAC is associated with gemcitabine resistance.

• TFF1 imparts gemcitabine resistance properties to pancreatic cancer cells by inducing stemness and anti-apoptotic molecules.

• TFF1 binds to chemokine receptor CXCR4 and elicits gemcitabine resistance in PDAC.

• Increased sensitivity to gemcitabine upon TFF1 silencing underscores its importance in targeting gemcitabine-resistant PDAC.

Abbreviations

PDAC

Pancreatic Ductal Adenocarcinoma

TFF1

Trefoil factor 1

CXCR4

C-X-C chemokine receptor type 4

SPR

Surface plasmon resonance