Cadmium (Cd) is a toxic metal which is released into the environment due to anthropogenic activities (Industrial and agricultural activities). Therefore, Cd is frequently found as a persistent pollutant in soil, water and air (Jomova et al. 2025; Wang et al. 2023). With the development of industrialization and intensive agriculture, the population is increasingly exposed to Cd levels which are very high in industrial areas, but also in certain foods such as cereals and bread (Wang et al. 2021). Cd accumulates durably in living organisms (plants and animals) due to its exceptionally long biological half-life of around 10–30 years. In the human body, Cd has the capacity to accumulate preferentially in the liver and in the kidneys.

The cytotoxicity associated with Cd has been demonstrated for the first time by the appearance of bone lesions in Japan following the discharge of Cd-rich waste from mining operations into rivers, and the consumption of high doses of Cd by the population (Genchi et al. 2020). Cd is biologically active in the form of divalent cations Cd²⁺ which substitute for essential divalent cations (Ca²⁺, Mg²⁺, Zn²⁺, ...) to deregulate their physiological functions (Đukić-Ćosić et al. 2020; Peana et al. 2022; Rani et al. 2014). Cd²⁺ cations enter into the cells through the same membrane transporters than divalent cations (Thevenod et al. 2019). Interestingly, studies on cellular models have shown that Cd chronic exposure induces transformation of non-cancer epithelial cells into a cancer-like phenotype (Benbrahim-Tallaa et al. 2009; Person et al. 2013; Qu et al. 2012).

Cd has been classified as a type I carcinogen by the International Agency for Research on Cancer due to its impact on lung cancer incidence in industrial workers (Boffetta 1993; Waalkes 2003). However, a growing body of research suggests a link between Cd contamination and an increased risk of chronic disease in the general population (Jomova et al. 2025). For example, it was shown that Cd can accumulate in the exocrine pancreas and particularly in tumoral tissues (Djordjevic et al. 2019; Pallagi et al. 2024; Soleimani et al. 2025).

The pancreatic ductal adenocarcinoma (PDAC) represents the most frequent and the deadliest form of pancreatic cancer, it ranks 12th in terms of incidence and 6th in terms of mortality worldwide (Bray et al. 2024). The incidence of PDAC is rising at an alarming and partly unexplained rate, making it the 2nd leading cause of cancer death by 2040 according to recent epidemiological projections (Rahib et al. 2021). The earliest event of pancreatic carcinogenesis is the KRAS mutation which is present in > 90% of PDAC, leading to the development of pancreatic intraepithelial neoplasia (PanINs) which progress toward the development of invasive PDAC (Luo 2021).

PDAC is characterized by an intense and fibrous remodelling of their stroma called desmoplastic reaction, which promotes cancer progression and resistance to therapies (Whittle and Hingorani 2019). The desmoplastic reaction is initiated by the activation of pancreatic stellate cells (PSCs) (Thomas and Radhakrishnan 2020). Quiescent PSCs represent 4–7% of the healthy pancreas. They store vitamin A in lipid droplets and are involved in extracellular matrix (ECM) regulation by secreting ECM components. PSCs activate in response to various stresses including ethanol, fatty acids, hypoxia, inflammation or oxidative stress. Activated PSCs loss their lipid droplets and acquire a myofibroblast-like phenotype characterized by enhanced migratory and secretory properties (Sarkar et al. 2023). Moreover, activated PSCs are present in the PDAC microenvironment where they facilitate tumour progression and metastasis by interacting with cancer cells through secreting soluble factors.

We recently highlighted the role of ion channels in the activation of PSCs (Auwercx et al. 2025). Ion channels are transmembrane protein which regulate numerous cellular mechanisms. Among these ion channels, we have identified the Transient Receptor Potential Cation Channel Subfamily M Member 7 (TRPM7) as a potential biomarker of activated PSCs (Auwercx et al. 2022). TRPM7 has been shown to allow the entry of metal cations including Cd²⁺ (Li et al. 2006; Monteilh-Zoller et al. 2003). Interestingly, we have shown that chronic exposure to Cd increases TRPM7 expression in pancreatic non-cancer cells leading to enhanced migratory and invasive properties (Vanlaeys et al. 2020).

The aim of this study was to assess the effect of Cd chronic exposure on human PSC activation and TRPM7 expression, as well as in mice genetically modified with KRAS^G12D mutation inducing spontaneous PanIN precancerous lesions.

As previously conducted in epithelial pancreatic cells, we cultivated human PSCs with culture media supplemented with 1 µM CdCl₂ for more than 30 weeks to mimic Cd chronic exposure (Qu et al. 2012; Vanlaeys et al. 2020). By using western-blot and immunofluorescence staining, we studied α-Smooth Muscle Actin (α-SMA), desmin, Glial Fibrillary Acidic Protein (GFAP), and vimentin expressions as defining criteria of PSC (Froeling et al. 2009) (Fig. 1). High α-SMA and vimentin expression are found in activated PSCs while desmin and GFAP have been identified as quiescent PSC markers (Erkan et al. 2012). PSCs exposed to Cd have a higher expression of α-SMA compared to the control cells (Figs. 1A, E) but no significant difference was found for vimentin expression (Figs. 1B, F). On the other hand, a lower expression of GFAP (Figs. 1D, H) was detected in Cd-exposed PSCs compared to the control cells. No difference was detected for desmin expression between control and Cd exposed PSCs (Figs. 1C, G).

As activated PSCs synthetize excessive extracellular matrix (ECM) proteins leading to a dense fibrotic stroma, we studied the expression of type I collagen, elastin and fibronectin by western blot and immunofluorescence (Fig. 2).

Figure 1 Expression of pancreatic stellate cell markers in PS-1 cells chronically exposed to Cd. A Protein level expression of α-SMA studied by western-blot (n = 6). B Protein level expression of vimentin studied by western-blot (n = 7). C Protein level expression of desmin studied by western-blot (n = 7). D Protein level expression of GFAP studied by western-blot (n = 5). E Protein level expression of α-SMA studied by immunofluorescence (n = 3). F Protein level expression of vimentin studied by immunofluorescence (n = 3). G Protein level expression of desmin studied by immunofluorescence (n = 3). H. Protein level expression of GFAP studied by immunofluorescence (n = 3). Bar = 100 µm. *p < 0.05.

An increase of both type I collagen and elastin expression were observed in Cd exposed PSCs compared to the control ones (Figs. 2A, D, B, E). However, we did not observe a statistically significant change of fibronectin expression in Cd exposed PSCs (Figs. 2C, F). Moreover, we studied the expression of IL-8 and IL-10 in the PSC conditioned media by ELISA. As indicated in the Figs. 2G and 2H, an increase of both IL-8 and IL-10 is observed in Cd exposed PSCs compared to the control ones. In Cd exposed cells, we also observed a decrease of lipid droplets detected by Nile Red staining (Figure S1). Collectively, our data showed that Cd exposure enhanced the secretory activity of PSCs.

Figure 2 Expression of matrix extracellular proteins and inflammatory cytokines in PS-1 cells chronically exposed to Cd. A Protein level expression of type I collagen studied by western-blot (n = 8). B Protein level expression of elastin studied by western-blot (n = 11). C Protein level expression of fibronectin studied by western-blot (n = 8). D Protein level expression of type I collagen studied by immunofluorescence (n = 3). E Protein level expression of elastin studied by immunofluorescence (n = 3). F Protein level expression of fibronectin studied by immunofluorescence (n = 3). G Expression of secreted IL-8 (n = 3). H Expression of secreted IL-10 by ELISA assays (n = 3). **p < 0.01; ***p < 0.001.

It has been described that high α-SMA expression is linked to transdifferentiation of PSCs into a myofibroblast-like phenotype (Erkan et al. 2012). In pancreatic cancer, it has been shown that stromal fibroblasts secrete Matrix Metalloproteinase-2 (MMP-2) and urokinase Plasminogen Activator (uPA) leading to cancer progression and metastasis (He et al. 2007). Thus, we measured the effect of Cd exposure on MMP-2 and uPA secretion by PSCs (Fig. 3). Both MMP-2 and uPA secretions are increased in Cd-exposed PSCs compared to the control ones (Figs. 3A, B). Moreover, Cd exposure stimulated PSC migratory properties evaluated by wound healing (Fig. 3C) and by Boyden chamber assays (Fig. 3D). Taken together our results strongly suggest that Cd exposure induced a cell reprogramming into a myofibroblast-like phenotype in human PSCs.

We have previously shown that TRPM7 expression was increased following Cd exposure in pancreatic epithelial cells leading to enhanced cell migration (Vanlaeys et al. 2020). Moreover, TRPM7 expression is also related to PS-1 activated state (Auwercx et al. 2022). Here we showed that TRPM7 protein expression was increased in Cd-exposed PSCs (Fig. 4A). We also showed an increase of Mg²⁺-inhibited cation (MIC) current recorded by whole-cell patch-clamp (Fig. 4B). We used siRNA strategy to specifically silence TRPM7 expression in PSCs and to study the impact of TRPM7 silencing on MIC current and PSC migration. As shown in Fig. 4C, TRPM7 silencing was validated by the TRPM7 mRNA decrease in both control and Cd-exposed PSCs. It is interesting to note that Cd-exposure increased TRPM7 protein expression but not at transcriptomic level (Figs. 4A and C), suggesting that Cd affects translation but not transcriptional mechanisms in PSCs. TRPM7 silencing strongly inhibited MIC currents (Fig. 4D), confirming that MIC currents are generated by TRPM7 activity in PSCs as previously shown (Auwercx et al. 2022). Finally, TRPM7 silencing abolished PSC migration in both control and Cd-exposed cells (Figs. 4E, F). This indicated that Cd-induced PSC migration was TRPM7-dependent. These results were confirmed by using NS8593 (25 µM) as TRPM7 pharmacological blocker (Figs. 4E, G).

Activated PSCs are present in the desmoplastic stroma of pancreatic tumour where they communicate with cancer cells and stimulate their aggressivity. By using indirect co-culture model, we placed PSCs in the bottom of wells and we evaluated the MIA PaCa-2 cell migration through the transwell membrane pores (Figs. 5A, C, E). We showed that the presence of PSCs in bottom wells stimulated the MIA PaCa-2 cell migration (Figs. 5A, B). Cd-exposed PSCs have a greater effect on MIA PaCa-2 pancreatic cancer cell migration than control ones (Figs. 5A, B). Next, we assessed the role of TRPM7 on the PSC properties to stimulate pancreatic cancer cell migration by using pharmacological or siRNA strategies. NS8593 (25µM) abolished pancreatic cancer cell migration stimulated by control PSCs or by Cd-exposed PSCs in a similar manner (Figs. 5C, D). TRPM7 silencing also inhibited pancreatic cancer cell migration stimulated by control PSCs but we observed a greater inhibition when stimulated by Cd-exposed PSCs (Figs. 5E, F).

We evaluated the effects of Cd exposure (1 mg/kg, 13 weeks) on the Pdx1-Cre; LStopL-K-ras^G12D mouse model, which develop PanINs (Vasseur et al. 2015). We observed an increase of the pancreas masses for the Cd exposed mice compared to the control ones (Fig. 6A). This was not associated with a statistically significant increase of PanIN percentage in Cd exposed mice, despite a greater heterogeneity compared to control mice (Fig. 6B). By using second harmonic generation (SHG) imaging, we evaluated collagen deposition in mice pancreas. As shown in the Fig. 6C, a more intense staining was observed in Cd exposed mice, mainly at the periphery of tissue (Fig. 6C). This was confirmed by Masson’s trichrome staining protocol for collagen fibre detection (Figs. 6D, E). Thus, a more intense blue staining was detected in pancreatic tissues from Cd-exposed mice compared to the control ones (Figs. 6D, E). Moreover, a strong immunostaining of α-SMA detected by IHC was found in fibrotic areas (Fig. 6F). A moderate TRPM7 staining was detected in the same areas suggesting that TRPM7 is overexpressed in fibrotic stroma (Fig. 6G). Taken together, our results strongly suggest that Cd exposure exacerbated the pancreatic fibrosis that accompanied PanIN formation and that TRPM7 is preferentially found in fibrotic pancreatic stroma.

Cadmium (Cd) is a trace metal that accumulates in human tissue, and its levels are rising alarmingly in the general population. In this study, we aimed to describe the effect of chronic Cd exposure on pancreatic stellate cell (PSC) activation which is the initial step for extracellular matrix (ECM) deposition and stiffening occurring during pancreatitis and pancreatic ductal adenocarcinoma (PDAC) development.

We showed that chronic Cd exposure transforms human pancreatic stellate cells (PSCs) into a myofibroblast-like phenotype and worsens fibrosis in the Pdx1-Cre; LStopL-K-ras^G12D mice model of pancreatic intraepithelial neoplasia (PanINs) formation. Moreover, Cd exposure increased TRPM7 expression in PSCs, and TRPM7 was preferentially expressed in fibrotic pancreatic stroma. In this work, we used the immortalized PS-1 cells, a model of non-pathological PSCs (Froeling et al. 2009). PS-1 were cultivated with 1 µM CdCl₂ for 30 weeks as previously described to induce pancreatic epithelial cell transformation (Qu et al. 2012; Vanlaeys et al. 2020). Cd is a bioaccumulative metal and it has been shown that the concentrations used here (1µM for cell culture and 1 mg/kg CdCl₂, by intraperitoneal injection in mice) are significantly below that reported by Djordjevic et al. in human pancreatic cancer tissues (between 11 and 166 µM) and for animal studies (15 and 30 mg/kg) (Djordjevic et al. 2019).

PSCs exist under two states, quiescent in the normal pancreas or activated in the pathological one. PSC activation can be induced by various stimuli including ethanol, hypoxia, oxidative stress, pro-inflammatory cytokines and growth factors leading to a myofibroblast-like phenotype mainly characterized by the overexpression of α-SMA (Bynigeri et al. 2017; Ferdek and Jakubowska 2017; Fu et al. 2018). In accordance with previous studies, cells that have been exposed to Cd exhibited the hallmark features of activated cells including α-SMA overexpression, lipid droplet disappearance (stained by Nile Red, figure S1), downregulation of GFAP, ECM protein (type I collagen, elastin, MMP-2, uPA) and cytokine (IL-8, IL-10) production (Bynigeri et al. 2017; Ferdek and Jakubowska 2017; Fu et al. 2018). In addition, Cd-exposed PSCs have a greater migratory capacity than non-exposed cells. This confirmed that Cd exposure activated human PSCs because activation of PSCs has been shown to be a prerequisite for migration (Phillips et al. 2003). Pancreatic stellate cells are key players in the pancreatic tumour microenvironment, in which they communicate with cancer cells and stimulate their metastatic and chemoresistant properties. Indirect co-culture models using Boyden chamber assays have been developed to simulate the interaction between PSCs and pancreatic cancer cells (Kikuta et al. 2010; Lu et al. 2014). By using similar approach, we confirmed that PSCs stimulated pancreatic cancer cell migration and we highlighted that Cd exposure promoted these interactions. Taken together our results showed that Cd exposure activates PSC activation into a myofibroblast-like phenotype but also enhances their secretory properties.

We have shown that the expression level of TRPM7 can be correlated with the activation state of PSCs (Auwercx et al. 2022). Here we showed that chronic Cd exposure results in an increase of TRPM7 expression and membrane currents recorded by patch-clamp technique. Similar results have been previously shown by our group in mammary and pancreatic non-cancer epithelial cells exposed to Cd (Vanlaeys et al. 2020). In Cd-exposed PSCs, TRPM7 inhibition (by using TRPM7 silencing or NS8593) prevented enhanced cell migration as well as the interaction with the pancreatic cancer cells. The effect of NS8593 on PSC and pancreatic cancer cell interaction was more pronounced than the effect of TRPM7 silencing. This could be explained by the fact that NS8593 will also inhibit TRPM7 channels expressed at the MIA PaCa-2 plasma membrane, inducing a direct inhibitory effect on MIA PaCa-2 cell migration. Thus, we have already shown that TRPM7 regulates MIA PaCa-2 cell migration (Lefebvre et al. 2020; Rybarczyk et al. 2017). These results reinforced the concept that TRPM7 is an important regulator of activated PSCs. Initial studies showed that the TRPM7 pore is permeant to divalent metal cations including cadmium (Li et al. 2006; Monteilh-Zoller et al. 2003). However, the ability of TRPM7 to pass Cd is still debated (Thevenod et al. 2019) because Cd influxes through TRPM7 have been recorded using supraphysiological [Cd²⁺]_o from 10 to 30 mM (Li et al. 2006; Monteilh-Zoller et al. 2003). Moreover, Mellott et al. showed that TRPM7 blockade by NS8593 fails to reverse Cd-induced cytotoxicity in Jurkat human leukemic T lymphocytes (Mellott et al. 2020). On the other hand, Martineau et al. showed that TRPM7 contributes to Cd uptake in human osteoblasts (Martineau et al. 2010). Based on the property of Cd²⁺ to bind Fura-2 probe and induce a positive quench, we have shown that TRPM7 promotes Cd²⁺ entry into non-cancer mammary epithelial MCF10A cells (Vanlaeys et al. 2020). More recently, Correia et al. showed that TRPM7 knockout by CRISPR/Cas9 mitigates Cd-induced cytotoxicity in human lung cancer A549 cells (Correia et al. 2025). They showed similar effects on both A549 and primary alveolar type 2 (ATII) cells by using NS8593 or VER155008 as potent inhibitors of the TRPM7 channel (Correia et al. 2025; Nadezhdin et al. 2023). Although we showed that Cd exposure upregulates TRPM7 expression in PSCs, our study does not allow us to determine whether TRPM7 is involved in Cd accumulation in these cells.

We previously identified TRPM7 as a biomarker of PDAC (Auwercx et al. 2021; Rybarczyk et al. 2012). In the present study, we used the Pdx1-Cre; LStopL-K-ras^G12D preclinical mouse model which develop spontaneously pancreatic intraepithelial neoplasia (PanINs) (Vasseur et al. 2015). Interestingly, the Cd-treated mice have a bigger pancreas but we did not detect a larger proportion of PanIN lesions which is probably due to the important heterogeneity of the Cd group. However, a larger deposition of collagen was detected through Masson’s trichrome staining and SHG in the Cd group. Collagen is a major component of the desmoplastic reaction which occurs during exocrine pancreatic disease such as chronic pancreatitis and PDAC (Ferdek et al. 2022; Liot et al. 2021). α-SMA staining by IHC revealed that activated PSCs were localized mainly in collagen rich areas confirming that activated PSCs are the major source of collagen in pancreatic stroma. Moreover, TRPM7 staining also showed that it was more abundantly expressed in these fibrotic areas enriched in activated PSCs. By analysing the TCGA databases, we have already shown that TRPM7 expression was enriched in pancreatic cancer associated fibroblasts (CAFs) compared to fibroblasts from normal pancreas (Auwercx et al. 2022). The results from our in vivo study in mice harbouring the oncogenic KRAS^G12D mutation showed that Cd exposure stimulates the formation of fibrotic areas enriched with activated PSCs expressing TRPM7.

To the best of our knowledge, we show for the first time that Cd activates PSCs and worsens pancreatic oncogenic remodelling by promoting fibrosis (Fig. 7). The present study strongly suggests that Cd pollution is a risk for exocrine pancreatic disease progression and confirms that TRPM7 is a major mediator of PSC activation and Cd-induced cell transformation.

PSC migratory capacities were assessed by Boyden chamber and by wound healing assays.

Cell migration was performed using 8 µm pore size Boyden chambers. 4x10⁴ PS-1 cells were seeded in the upper compartment in growth medium with 1% FCS and the lower compartment was filled with growth medium supplemented with 10% FCS as chemoattractant. After 24 hours of incubation at 37°C, the remaining cells in the upper compartment were removed by scrubbing. Migrating cells were washed in PBS, fixed with methanol, and stained with haematoxylin solution. Cell migration was then quantified by counting 20 different fields at x400 magnification under an inverted microscope (Nikon Eclipse TS100). An MTT test was carried out at each experiment to avoid seeding or counting errors.

For wound healing assay, 2.5x10⁵ cells were seeded in 35-mm Petri dish containing 2 mL of culture media supplemented with 10% FCS. After 48h, a physical gap was created within cell monolayer using a 20–200µL pipette tip (P200). The process of cell migration into the gap was monitored by taking photos immediately after wounding (T0), after 18h (T18), and finally after 24h (T24). The gap closure rate was analyzed by using ImageJ software.

PS-1 cells were seeded in 24-wells plate at a confluence of 7.5x10⁴ cells per well in DMEM/F-12 medium supplemented with 1% FCS. Cells were incubated at 37°C for 24h prior to the coculture experiments. Then, 4x10⁴ MIA PaCa-2 cells were seeded in the upper compartment of the Boyden chamber in DMEM/F-12 medium with 1% FCS. The Boyden chambers containing MIA PaCa-2 cells were placed in the wells containing the PS-1 cells, then incubated for 24h at 37°C. Migrating MIA PaCa-2 cells were counted as previously described for the migration assay.

PS-1 cell total RNA was extracted using TRIzol reagent protocol. RNA concentration and purity were assessed using NanoDrop spectrophotometer and 2 µg of total RNA was then reverse transcribed into cDNA using the High-Capacity cDNA RT-kit following the manufacturer's instructions. Quantitative real-time PCR (RT-qPCR) was performed using PowerUp SYBR Green Master Mix on QuantStudio™ 7. Relative gene expression was calculated Pfaffl method (Pfaffl 2001). ZNF384 and DNAJC14 have been used as housekeeping genes using Genorm method (Vandesompele et al. 2002).

Cells were lysed in RIPA buffer supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). Cellular debris was pelleted by centrifugation of lysates at 10000 g for 10 minutes at 4°C. Protein concentration of the supernatant was quantified using Biorad Protein Assay (BioRad, Marnes-La-Coquette, France) according to the manufacturer’s instructions. Samples were diluted in Laemmli buffer, reduced by 10 mM dithiothreitol and denaturated 5 minutes at 95°C. Samples were submitted to SDS-PAGE (50 µg of total protein per lane) and then transferred onto polyvinylidene difluoride membranes (GE Healthcare Life Sciences). Membranes were blocked with 5% BSA or milk in TBS-T (0.1% Tween 20, 50 mM Tris HCl buffer, 150 mM NaCl, pH 7.5) for 2h at room temperature. They were then incubated overnight at 4°C with primary antibodies diluted in TBS-T supplemented with 1% BSA (the references of primary antibodies are indicated in the Table 1 of the supplemental data). After washing with TBS-T, membranes were incubated for 1h at room temperature with Horseradish peroxidase (HRP)-conjugated secondary antibodies diluted 1/10000 in TBS-T-1% BSA. After washing with TBS-T and TBS, immune complexes were revealed using ECL prime chemiluminescence detection kit (GE Healthcare, Orsay, France) according to the manufacturer’s instructions.