Budesonide reduces the mesenchymal/invasive features of PDAC cells in 2D cultures without affecting cell proliferation

To investigate whether budesonide may directly affect PDAC growth and development, we first assessed its effect on tumor cell morphology and behavior using two independent patient-derived primary cell lines, PDAC#253 and #354 [15, 16]. To this end, PDAC cells were plated in 2D culture conditions in PDAC growth medium (RPMI +10% FBS) and treated with either budesonide, another glucocorticoid (dexamethasone), or DMSO as a vehicle control (Fig. 1A). After 3 days, the cells were stained with crystal violet and the colony morphology was analyzed. Control colonies showed the expected flat shaped morphology with cells spreading from the colony edges outwards. Colony morphology was largely modified by budesonide, showing a round-shaped/compacted phenotype and higher circularity index compared to controls (Fig. 1B). This phenotypic effect was observed at budesonide concentrations starting from 5 to 10 µM in PDAC#354 and #253 cells, respectively (Fig. 1B). Of note, dexamethasone did not affect PDAC colony morphology even at the highest concentration (20 µM) tested (Fig. 1B), thus suggesting a specific effect of budesonide. We thus assessed whether budesonide may affect cell proliferation and apoptosis in these culture conditions. To this end, PDAC cells ± budesonide were analyzed by flow cytometry using the 5-ethynyl-2-deoxyuridine (EdU) incorporation assay and Annexin V/Propidium iodide staining. The percentage of EdU positive as well as Annexin V/PI positive cells was comparable in control and budesonide-treated cells, suggesting that neither proliferation or apoptosis were affected in the presence of budesonide (Fig. S1A-D). These results were further confirmed with the commercially available pancreatic cancer cell line PANC1. Indeed, budesonide similarly modified the growth behavior of PANC1 cells (Fig. S1E-F), without affecting proliferation and apoptosis (Fig. S1G-H).

Fig. 1

Budesonide promotes cellular adhesion in patient-derived PDAC cells. A Schematic representation of the experimental procedure. PDAC#253 and #354 cells were plated (1.5 × 104 cells/cm2) on gelatin-coated plates at day − 1. On day 0, cells were treated ± budesonide (from 2.5 to 20 µM), dexamethasone (20 µM) or DMSO (control) for 3 days. B Representative pictures of PDAC cells treated with budesonide, dexamethasone or DMSO, at the indicated concentrations and stained with crystal violet (left) and quantification of the circularity index (right). C Representative confocal images of E-CADHERIN (green) staining in PDAC#253 and #354 cells treated ± budesonide (20 µM), or ± dexamethasone (20 µM). Nuclei were stained with DAPI (blue). D Western blot analysis (upper) and densitometric quantification (ADU; bottom) of E-CADHERIN in PDAC#253 and #354 treated from (A). Densitometric analysis (ADU) is shown as fold-change vs. DMSO-treated cells, after normalization to GAPDH. E Representative confocal images of VIMENTIN (green) staining (left) and quantification of VIMENTIN+ cells (right) in PDAC#253 and #354 cells ± budesonide (20 µM). Nuclei were stained with DAPI (blue). F Representative western blot analysis (upper) and densitometric quantification (ADU; bottom) of FIBRONECTIN (FN1) and VIMENTIN in PDAC#253 and #354 cells treated ± budesonide (20 µM). ADU is shown as fold-change vs. DMSO-treated cells after normalization to GAPDH. Data are mean ± SEM (*p ≤ 0.05; **p ≤ 0.005; ***p ≤ 0.001; n = 3, Student’s t-test)

Recent findings showed that budesonide specifically promotes the stabilization of the cell-cell adhesive contacts in embryonic stem cells and triple negative breast cancer cells [11, 13]. To investigate this phenotype in PDAC cells, we analyzed the expression of epithelial and mesenchymal markers by immunofluorescence and western blot. We first assessed the effect of both budesonide and dexamethasone on the expression/localization of the adhesion protein E-CADHERIN in PDAC#354 and #253 cells. Immunofluorescence analysis showed increased accumulation of E-CADHERIN at the cell-cell junctions in budesonide- but not dexamethasone-treated PDAC cells compared to control, which correlates with changes in colony morphology (Fig. 1C). Of note, treatment with budesonide increased expression of E-CADHERIN protein in PDAC#354 but not #253 cells, likely reflecting their different genetic background (Fig. 1D). Conversely, budesonide reduced the expression of the mesenchymal markers VIMENTIN and FIBRONECTIN (FN1) both in PDAC and PANC1 cells (Fig. 1E-F, and Fig. S1I), providing molecular support to the hypothesis that budesonide induces epithelial features in pancreatic cancer cells. To further investigate this phenotype, we analyzed cell migration and invasion by transwell assay and fluorescent gelatin degradation assay in vitro (Fig. 2A). Budesonide significantly reduced PDAC#253 and #354 cell migration (∼10 times) in response to serum gradients in the boyden chamber assay (Fig. 2B). Furthermore, budesonide reduced by more than 80% the capacity of both PDAC and PANC1 cells to degrade and invade Cy3-fluorescent gelatin (Fig. 2C, Fig. S2B-C). In contrast, dexamethasone did not affect the invasive capacity of PDAC and PANC1 cells, further supporting the idea that budesonide exerts a specific activity (Fig. S2A, Fig. S2B-C).

Altogether our findings indicate that budesonide promotes an epithelial phenotype and antagonizes the mesenchymal state in PDAC cells in 2D settings and suggest that it inhibits their ability to migrate and invade the extracellular matrix, without affecting proliferation (Fig. 2D).

Fig. 2

Budesonide reduces mesenchymal markers and PDAC cell migration and invasion. A Schematic representation of experimental procedure. PDAC#253 and #354 cells were plated (1.5 × 104 cells/cm2) on gelatin-coated plates at day − 1. On day 0, cells were treated ± budesonide (20 µM) or DMSO (control). After 3 days, cells were dissociated and plated (2.5 × 104 cells/well) on Boyden chambers or (1 × 105 cells/cm2) on Cy3-conjugated gelatin. B Representative crystal violet images (left) and quantification (right) of PDAC#253 and #354 cells ± budesonide migrating through the transwell. Cell migration was quantified at 6 h (h) after seeding. Data are shown as fold-change vs. control (DMSO) and are mean ± SEM (*p < 0.05; n = 3, Student’s t-test). C Representative confocal images of ACTIN (green) staining (left) and quantification (right) of Cy3-gelatin degraded area in PDAC#253 and #354 cells ± budesonide. Invasion was quantified 6 h after seeding. Nuclei were counterstained with DAPI. Data are mean ± SEM (**p < 0.005, ***p < 0.001; n = 3, Student’s t-test) after normalization vs. the total number of nuclei. D Schematic representation of the effects of budesonide in 2D culture. Budesonide promotes epithelialization and reduces PDAC cell migration and invasion. E Schematic representation of 3D organotypic culture procedure. PDAC (#253 and #354) and PANC1 (1 × 103 cells/cm2) cells were plated on a layer of 100% matrigel in complete medium containing 2% matrigel ± budesonide (20 µM) for 6 days. Medium was refreshed at day 3. F-G Representative phase-contrast images (F, left), frequency of cohesive vs. not cohesive structures (F, right) and quantification of the area (G) of PDAC (#253 and #354) spheroids ± budesonide at day 6 after plating. Data are mean ± SEM (F) or mean ± SD (G) (***p < 0.001; n = 4, Student’s t-test). H Representative images of PKH26-labeled (red) spheroids at day 6 post labeling derived from PDAC#253 and #354 cells treated ± budesonide (20 µM)

Budesonide inhibits PDAC cell proliferation in a three-dimensional environment in vitro and in vivo

While 2D cultures have been largely used to study cancer cell biology, they do not reproduce the three-dimensional cell-cell/cell-matrix interactions proper of the tumor microenvironment [19]. This is particularly relevant when evaluating the effect of a drug on tumor growth. Thus, to investigate the effect of budesonide in a more physiological context, we used 3D organotypic cultures of pancreatic cancer cells. To this end, PDAC cells were seeded as single cells onto a layer of 100% matrigel and incubated in PDAC growth medium (RPMI +10% FBS) with 2% matrigel, either alone or in the presence of budesonide (20 µM), and cultured for 6 days (Fig. 2E). In the presence of budesonide PDAC spheroids were more homogeneous and compacted compared to control (DMSO) (Fig. 2F). Specifically, budesonide increased the fraction (up to 70%) of cohesive spheroids compared to DMSO, which conversely showed a higher fraction of irregular/non-cohesive spheroids with invasive buds (Fig. 2F and Fig. S2D). Comparable results were obtained with 3D organotypic cultures of PANC1 cells (Fig. S2E). Of note, quantification of the spheroid area showed that it was significantly reduced in the presence of budesonide compared to controls (Fig. 2G), raising the hypothesis that budesonide could exert an anti-proliferative effect in 3D culture conditions. To further address this issue, we stained the spheroids with the membrane dye PKH26, which dilutes as the cells divide (Fig. 2E). Quite unexpectedly, while control spheroids lost PKH26 staining at day 6, budesonide-treated spheroids retained the membrane dye (Fig. 2H), indicating that proliferation is significantly reduced in this condition.

To better investigate the effect of budesonide on pancreatic cancer spheroid growth and improve reproducibility, we set up a robust spheroid formation assay by seeding pancreatic cancer cell lines (500 cells/well) into V-shaped ultra-low attachment 96-multiwell plates in PDAC growth medium (RPMI + 10 % FBS), previously used in both 2D and organotypic cultures, without supplementation of growth factors and cytokines. Our initial efforts to generate pancreatic cancer spheroids were unsuccessful due to a reduced propensity of these cells to aggregate, which has already been described [20]. Thus, in the attempt to favor the cell-cell vs. cell-substrate adhesive contacts, cells were passaged at a higher ratio (1:2 vs. 1:6) before seeding (500 cells/well) into V-shaped ultra-low attachment plates (Fig. 3A). This adaptive approach improved the ability of different pancreatic cancer cell lines, including PDAC, PANC1 and the highly aggressive and metastatic L3.6pl cells, to generate floating, round-shaped, compacted cell aggregates that were homogeneous in size, ranging around a mean value of 0.032 ± 0.004 mm3 (Fig. 3B). To assess the effect of budesonide on PDAC, PANC1 and L3.6pl spheroids, cells were allowed to aggregate with either budesonide at 1 or 20 µM or DMSO as a control. Quantification of the spheroid volume revealed that budesonide-treated spheroids were significantly smaller in size compared to controls (Fig. 3B). We thus performed a dose-dependent assay to evaluate the effect of lower concentrations of budesonide (from 10 to 10− 3 µM) on spheroid growth, calculating the change in volume over time (Δvol/Δt). Results showed a clear dose-dependent response to budesonide, which significantly reduced the growth rate of PDAC spheroids (∼3 times) compared to controls, up to a concentration of 10− 2 µM (Fig. 3C, Fig. S3A). Double immunofluorescence staining for the nuclear protein Ki67 and the cytoskeleton protein ACTIN on both whole mount spheroids (Fig. 3D, Fig. S3B) and spheroid cryo-sections (Fig. 3E), showed a significant reduction of Ki67+ cells in budesonide-treated spheroids, which were smaller and highly compacted compared to controls. Complementary to these findings, we showed that both the diameter of the spheroids and the number of cells/spheroid diameter were significantly reduced in the presence of budesonide compared to controls (Fig. S3C). To further investigate this phenotype, we compared the growth rate of PDAC cells ± budesonide in 2D and 3D cultures. PDAC cells seeded in 2D culture plates and treated with budesonide at the highest concentration (20 µM) showed a similar growth rate compared to controls (Fig. 3F, Fig. S3D). Conversely, the doubling time of PDAC spheroids strongly increased in the presence of budesonide (1 µM) from ∼40 h to > 120 h (Fig. 3F, Fig. S3D), further supporting the hypothesis that budesonide exerts an anti-proliferative effect on PDAC cells cultured in 3D conditions.

Fig. 3

Budesonide reduces the growth of pancreatic cancer cell spheroids. A Schematic representation of the experimental design. PDAC (#253 and #354), PANC1 and L3.6pl cells were seeded (5 × 102 cells/well) in ultra-low attachment plates ± budesonide (from 10− 3 µM to 20 µM) or DMSO (control) for 5 days. B Representative pictures (left) of PDAC (#253, #354), PANC1 and L3.6pl spheroids ± budesonide (1 and 20 µM) and quantification (right) of the spheroid volume. Data are mean ± SD (**p < 0.005, ***p < 0.001; n = 3, Student’s t-test). C Representative pictures (left) and growth rate (right) of PDAC#253 spheroids ± budesonide calculated as the mean of the ratio between the Δvolume and the Δtime (48, 72, 96 and 120 h). Data are mean ± SEM (*p < 0.05; **p < 0.005, ***p < 0.001; n = 3, Student’s t-test). D Representative confocal images of ACTIN (red) and Ki67 (grey) staining in PDAC#253 spheroids ± budesonide (1 µM). Nuclei were counterstained with DAPI (blue). E Representative images (left) and quantification (right) of Ki67 (red) staining in cryo-sections of PDAC#253 spheroids. Nuclei were counterstained with DAPI (blue). The number of Ki67+ cells/area is shown as mean ± SD (**p < 0.005; n = 3, Student’s t-test). F Time course quantification of cell number in 3D spheroids (left) and 2D cultures (right) of PDAC#253 cells treated ± budesonide at the indicated concentrations. Data are mean ± SD (*p < 0.05; n = 3, Student’s t-test)

These unexpected findings prompted us to investigate the effect of budesonide in the tumor microenvironment in vivo. To this end, we established a human PDAC xenograft model by injecting PDAC cells subcutaneously into the flanks of CD1 nude mice. Mice were divided into three groups and were injected intraperitoneally (i.p.) with DMSO-vehicle as a control, budesonide (3 mg/Kg; 6 days/week) or gemcitabine (125 mg/Kg; twice/week), which is the gold-standard treatment for PDAC (Fig. 4A). Tumor volume was measured every week and mice were sacrificed at week 4. Budesonide- and gemcitabine- treated mice formed tumors of significantly smaller volumes compared to control mice (control 1439 ± 157 vs. budesonide 718 ± 301 vs. gemcitabine 791 ± 198 mm3; Fig. 4B, Fig. S4A). However, tumors from the 3 groups were similar in their composition, i.e., solid tumor of high grade/poor differentiation, with variable degree of necrosis, ranging from 0 to 40% (Fig. 4C). Immunofluorescence analysis of tumor sections revealed that budesonide strongly reduced the number of proliferating Ki67+ cells compared to control, similarly to gemcitabine (Fig. 4D, Fig. S4B). Furthermore, staining with cleaved CASPASE3 (cCAS3) showed a significant increase of cCAS3+ cells both in budesonide and gemcitabine groups compared to control, suggesting increased apoptosis upon treatment with both budesonide or gemcitabine (Fig. 4E, Fig. S4C).

Fig. 4

Budesonide reduces PDAC tumor growth in vivo. A Schematic representation of the experimental procedure. CD1 athymic nude mice were subcutaneously injected with PDAC#253 cells (3 × 105 cells/flank) and injected intraperitoneally (i.p.) with either budesonide (3 mg/Kg) everyday, vehicle (control) or gemcitabine (125 mg/Kg) 2 times/week. B Quantification of tumor volume (mm3) in mice injected with budesonide (bude), gemcitabine (gem) or vehicle (control). Data are mean ± SEM. n≥ 5 mice/group; n≥ 10 tumors/group. # and * indicate the significance (p < 0.05) of budesonide (bude) and gemcitabine (gem) vs. control (vehicle), respectively. dpi = days post inoculations. C Representative images of H&E-stained tumor sections from control (vehicle), budesonide (bude)- and gemcitabine (gem)- treated mice. Mosaic reconstruction (left) and higher magnifications (middle and right) of tumor sections are shown. Asterisks indicate necrotic areas. D Representative images (left) of Ki67 staining (red) and quantification (right) of Ki67+ cells in tumor sections from budesonide (bude)-, gemcitabine (gem)- and vehicle (control)- treated mice. Nuclei were counterstained with DAPI (blue). E Representative images (left) of cleaved Caspase3 (cCas3) immunohistochemistry and quantification (right) of cCas3+ cells in tumor sections from budesonide (bude)-, gemcitabine (gem)- and vehicle (control)-treated mice. Data are mean ± SD (***p < 0.001; n = 10–12 tumors/group)

All together our findings reveal that budesonide exerts a previously unidentified anti-proliferative effect on PDAC cells, exclusively under 3D growth conditions.

The anti-proliferative effect of budesonide on PDAC spheroids requires an active GC-GR axis

It is well known that glucocorticoids (GCs), like budesonide, may exert both genomic and non-genomic effects, which are either dependent on or independent of the glucocorticoid receptor (GR), respectively. To investigate the underlying mechanism of budesonide activity on PDAC spheroids, we first assessed the effect of other GCs, such as dexamethasone and hydrocortisone. To this end, PDAC spheroids were generated in the presence of either budesonide, dexamethasone, hydrocortisone, or DMSO as a vehicle control. Quantification of the spheroid volume showed that both dexamethasone and hydrocortisone (used at 1 µM) significantly reduced the volume of PDAC#253 and #354 spheroids (Fig. 5A-B). Of note we found a clear dose-dependent response even at lower concentrations of GCs, up to 10− 3 µM in both cell lines (Fig. S5A-B). To directly assess the role of the GR, we silenced its expression in PDAC#253 cells by using an shRNA that targets the NR3C1 gene encoding the GR [18]. We first verified efficient downregulation of NR3C1 at both the RNA and protein level (Fig. S5C), and then tested the effect of budesonide on NR3C1 KD (shNR3C1) and NT (control/ shEmpty) cells in the different culture conditions. In 2D culture, budesonide significantly increased the circularity index of both control and NR3C1 KD PDAC cell colonies (Fig. S5D). Furthermore, budesonide reduced the migration (Fig. 5C, Fig. S5E) and invasion (Fig. 5C-D) of NR3C1 KD PDAC cells as efficiently as control cells, suggesting that the anti-migratory and invasive activities of budesonide are GR-independent and could be ascribed to non-genomic mechanisms.

We then assessed the effect of NR3C1 KD on PDAC tumor spheroids (Fig. 5E). As expected, control (NT) PDAC spheroid volume was significantly reduced in the presence of budesonide as well as in the presence of dexamethasone and hydrocortisone; however, this effect was blunted in NR3C1 KD tumor spheroids (Fig. 5F), suggesting that the anti-proliferative effect of budesonide in 3D spheroids relies on the GR. This hypothesis was further supported by the observation that budesonide reduced the doubling time of control but not of NR3C1 KD PDAC spheroids (Fig. 5G). Conversely, budesonide did not affect the doubling time of both control and NR3C1 KD cells in 2D culture conditions (Fig. 5G) even though the GR is expressed in this condition (Fig. S5C).

All together these data indicate that the anti-proliferative effect of budesonide on PDAC spheroids is GR-dependent, and that it is shared by other GCs (Fig. 5H).

Fig. 5

Budesonide-dependent reduction of PDAC spheroid volume is Glucocorticoid Receptor (GR)-dependent. A-B Representative pictures (left) and volume quantification (right) of spheroids from PDAC#253 A and #354 B cells ± budesonide (bude), dexamethasone (dexa) or hydrocortisone (hydro) at 1 µM. DMSO was used as a control. Data are mean ± SD (***p < 0.001; n = 3, Student’s t-test). C Schematic representation of experimental procedure. NT (control/ShEmpty) and NR3C1 KD PDAC#253 cells were plated (1.5 × 104 cells/cm2) on gelatin-coated plates at day − 1. On day 0, cells were treated ± budesonide (20 µM). After 3 days in culture, cells were dissociated and plated (1 × 105 cells/cm2) on Cy3-gelatin. D Representative confocal images (left) of ACTIN staining (green) in NT and NR3C1 KD PDAC#253 cells ± budesonide (20 µM) and quantification (right) of Cy3-gelatin degraded area. Nuclei were counterstained with DAPI. Data are mean ± SEM (*p < 0.05, n = 3, Student’s t-test), after normalization vs. the total number of nuclei. E Schematic representation of the experimental design. Control (NT) and NR3C1 KD PDAC#253 cells were seeded in ultra-low attachment plates (5 × 102 cells/well) and treated ± budesonide, dexamethasone, hydrocortisone (1 µM) or DMSO for 5 days. F Representative pictures (left) and volume quantification (right) of spheroids generated from control and NR3C1 KD PDAC#253 cells treated with budesonide (bude), dexamethasone (dexa), hydrocortisone (hydro) at 1 µM or DMSO as control. Data are mean ± SD (*p < 0.05; ***p < 0.001; n = 3, Student’s t-test). G Time course analysis of NT and NR3C1 KD PDAC#253 cell proliferation in 3D spheroids ± budesonide (left; bude: 1 µM), and in 2D cultures ± budesonide (right; bude: 20 µM). Data are mean ± SD (*p < 0.05; n = 3, Student’s t-test). H Schematic representation of the GR-independent and -dependent effects of budesonide. In 2D cultures, budesonide (> 2.5 µM) promotes epithelialization and reduces PDAC cell migration independently from the GR. In PDAC spheroids (3D), nanomolar concentrations of budesonide (≤ 10− 2µM) exert a GR-dependent anti-proliferative effect

Genome-wide transcriptome profiling reveals 3D-induced metabolic remodeling of PDAC cells

To gain insight into the mechanism by which budesonide affects PDAC cell behavior in 2D and 3D cultures, we performed RNA-Seq analysis of PDAC cells grown on gelatin-coated plates (2D) and as spheroids (3D) ± budesonide (Fig. 6A).

Principal component analysis (PCA) of RNA-Seq data showed that the cells in the different conditions clustered apart (Fig. 6B). Comparison of the transcriptome profiles of control PDAC cells in 2D and 3D conditions revealed ∼6200 deregulated protein-coding genes (DE; fold change ≥ 1.5; padj ≤ 0.05; Fig. S6A). Gene ontology (GO) analysis (David software; https://david.ncifcrf.gov/#) showed that the up-regulated genes in tumor spheroids were enriched in key biological processes such as cell cycle, cytoskeleton and chromatin organization (Fig. S6B), whereas the down-regulated genes were primarily enriched in metabolic and biosynthetic processes, and OXPHOS (Fig. S6B). Gene set enrichment analysis (GSEA) confirmed a significant enrichment in genes involved in energy metabolism (Fig. 6C-D, Table S3). Among the key upregulated genes in 3D cultures that contributed to the GSEA results were glycolytic enzymes genes, including HK2, PFKL, ALDOA, GAPDH, PGAM1, ENO1, PKM (Fig. 6C). Conversely, several genes of the NDUF, COX, ATP and UQCR gene families that encode enzymatic complexes involved in the OXPHOS/electron transport were down-regulated in 3D growing cells (Fig. 6D, Table S4).

Interestingly, different gene sets involved in lipid metabolism (e.g., Sterol regulatory proteins SREBP, glycerophospholipid and cholesterol metabolism) (Fig. S6C, Table S3) were enriched in 3D tumor spheroids, while gene sets related to protein metabolism (e.g., response to amino acid starvation, ribosome, ribosomal proteins, translation elongation and initiation) were conversely depleted (Fig. S6C, Table S3). Altogether these results suggest that PDAC cells undergo a general metabolic remodeling to meet their needs to grow in a 3D environment. To investigate this hypothesis, we focused on energy metabolism, and evaluated the sensitivity of PDAC cells to inhibition of OXPHOS and glycolysis. Surprisingly, the growth of PDAC spheroids increased in the presence of sublethal concentrations of the OXPHOS inhibitors metformin and rotenone compared to controls (Fig. 6E-F, Fig. S6D). However, as expected, higher concentrations of these inhibitors reduced PDAC spheroid growth and showed a toxic effect (Fig. S6E, F). Conversely, PDAC spheroids were highly sensitive to sublethal concentrations (5 mM) of the glycolysis inhibitor 2-deoxyglucose (2-DG) (Fig. 6E and G, Fig. S6G).

Together, these data support the transcriptome results and suggest that PDAC spheroids mostly rely on glycolysis to sustain their growth.

Fig. 6

Transcriptome profiling of PDAC cells in 2D culture and 3D spheroids and sensitivity to energy metabolism inhibitors. A Schematic representation of the experimental procedure. PDAC#253 were either seeded in ultra-low attachment plates to form 3D spheroids or plated in 2D and treated ± budesonide at the indicated concentrations. DMSO was used as control. After 3 days, spheroids/cells were collected and RNA was extracted for RNA-Seq analysis. B Principal component analysis (PCA) of PDAC ± budesonide in 2D and in 3D culture. C, D GSEA plots related to glycolysis (C, left) and OXPHOS (D, left) positively and negatively enriched in 3D spheroids, respectively. Heatmaps of DEGs related to glycolysis (C, right) and OXPHOS (D, right) in control PDAC cells in 2D vs. 3D culture. E Schematic representation of the experimental design. PDAC cells were seeded (5 × 102 cells/well) in ultra-low attachment plates ± metformin, rotenone, 2-DG or DMSO (control) for 5 days. F Representative pictures (120 h) of PDAC spheroids (upper) ± metformin (0.05 and 0.25 mM) or rotenone (0.05 and 0.5 nM) at the indicated concentrations, and time course analysis (bottom) of spheroid volume at the indicated time points. DMSO was used as control. Data are mean ± SEM (*, # and $ indicate the significance of metformin- or rotenone- treated cells vs. control *p < 0.05, **p < 0.005, ***p < 0.001; n = 3, Student’s t-test). G Representative pictures (120 h) of PDAC spheroids (upper) ± 2-DG (0.5 and 5 mM), and time course analysis (bottom) of spheroid volume at the indicated time points. DMSO was used as control. Data are mean ± SEM (***p < 0.001; n = 3, Student’s t-test)

Budesonide modifies the metabolism of PDAC cells

GCs are well-known regulators of cell metabolism [21]. We thus hypothesized that budesonide could interfere with the observed metabolic remodeling induced in PDAC cells cultured in 3D. To gain insight into the inhibitory effects of budesonide on PDAC spheroids, we compared the transcriptome profiles of PDAC cells (2D) and spheroids (3D) ± budesonide (Fig. 6A-B). Differential expression (DE) analysis revealed 1591 and 3304 DE protein-coding genes (fold change ≥ 1.5; padj ≤ 0.05) in budesonide-treated 2D cells and tumor spheroids (3D), respectively (Fig. S6A). Of note, GO analysis of the down-regulated genes revealed a significant enrichment in terms related to cell migration, locomotion, EMT and ECM organization only in 2D culture conditions (Fig. S7A). Furthermore, GSEA analysis showed that genes involved in ECM organization, including ECM glycoproteins, were negatively enriched in budesonide-treated PDAC cells (Fig. S7B and Table S3). These data provide molecular support to the results that budesonide inhibits PDAC ability to migrate and invade the extracellular matrix in 2D culture (Fig. 2). Interestingly, GSEA analysis of all differentially expressed (DE) genes in budesonide-treated cells revealed a significant enrichment in genes involved in energy metabolism (OXPHOS and respiratory electron transport chain), in both 2D and 3D conditions (Fig. 7A-B, Fig. S7A-B and Table S3-S4) and depletion of those related to protein metabolism (response to starvation, translation initiation and elongation, and ribosome) (Table S3).

These results showing that budesonide induced the expression of genes involved in OXPHOS in 2D and 3D cultures (Fig. 7A-B), and our findings that PDAC cells undergo a metabolic reprogramming towards glycolysis in 3D spheroids (Fig. 6C), led us to hypothesize that budesonide may interfere with the metabolic remodeling that is specifically required for PDAC cells to grow in 3D. This may explain, at least in part, the fact that budesonide inhibits PDAC cell proliferation, exclusively within a 3D environment. To directly investigate this hypothesis, we evaluated the mitochondrial activity of PDAC spheroids ± budesonide by using tetramethylrhodamine, ethyl ester (TMRE), a positively-charged dye that marks mitochondria with membrane potential. Both confocal images and quantification by FACS analysis showed that budesonide significantly increased the fraction of TMRE+ cells compared to controls (Fig. 7C, Fig. S7C), which is in line with the idea that GCs are inducers of mitochondrial activity and gluconeogenesis [21]. Furthermore, the expression of genes involved in OXPHOS metabolism like PDK4 and ATP6V1C2 was strongly induced by budesonide in control (NT/ shEmpty) but not in NR3C1 KD PDAC cells (Fig. S7D), thus, suggesting that budesonide regulates OXPHOS-related genes, at least in part, through the GR. Accordingly, dexamethasone showed a similar effect. In line with this idea and our findings that sublethal concentrations of OXPHOS inhibitors exerted a beneficial effect on PDAC spheroids growth (Fig. 6F), the volume of NR3C1 KD PDAC spheroids was significantly higher compared to that of control spheroids (Fig. S7E-F).

These data support the transcriptome profile and suggest that budesonide induces a GR-dependent metabolic switch toward OXPHOS in PDAC cells, which conversely rely on glycolysis to grow in a 3D environment. We thus hypothesized that a budesonide-induced metabolic imbalance could eventually affect proliferation of PDAC spheroids. GSEA and GO analysis of DE genes in budesonide-treated spheroids (3D) revealed a significant deregulation of genes involved in cell cycle including cell cycle and mitotic spindle checkpoints (Fig. 7D-E, Fig. S7G, Tables S3-S4). Accordingly, cell cycle analysis of 3D spheroids ± budesonide showed that budesonide significantly increased the fraction of PDAC cells in G0/G1 phase, and reduced the fraction of cells in S and G2/M phases (Fig. 7F). Interestingly, among the DEGs we found the cyclin dependent kinase inhibitor 1 C (CDKN1C), a GR target [14] also known as p57Kip2, which was strongly up-regulated (∼100 fold) by budesonide in 3D spheroids. We first validated these data by qPCR analysis. CDKN1C expression was strongly induced by both budesonide and dexamethasone in control 3D spheroids; however, this induction was almost completely abolished in NR3C1 KD spheroids (Fig. S7H), providing unprecedented evidence that CDKN1C is a GR target in PDAC cells. We thus asked whether the anti-proliferative effect of budesonide in 3D tumor spheroids may depend on the induction of CDKN1C. To test this hypothesis, we silenced CDKN1C in PDAC cells by siRNA (Fig. 7F). PDAC spheroids were generated using cells carrying siNT as controls or siCDKN1C (CDKN1C KD) and treated ± budesonide (1 µM) (Fig. 7G and Fig. S7I). As expected, budesonide significantly reduced the volume of control PDAC spheroids, whereas it did not affect spheroid volume in CDKN1C KD cells, suggesting that CDKN1C silencing significantly decreased the susceptibility of PDAC spheroids to budesonide-dependent growth inhibition (Fig. 7G-H).

All together these data suggest that budesonide-induced metabolic reprogramming impairs PDAC growth in 3D at least in part through induction of CDKN1C.

Fig. 7