Ex.50.T aptamer impairs tumor-stroma cross-talk in breast cancer by targeting gremlin-1

Abstract

The tumor microenvironment profoundly influences tumor complexity, particularly in breast cancer, where cancer-associated fibroblasts play pivotal roles in tumor progression and therapy resistance. Extracellular vesicles are involved in mediating communication within the TME, specifically highlighting their role in promoting the transformation of normal fibroblasts into cancer-associated fibroblasts. Recently, we identified an RNA aptamer, namely ex.50.T, that binds with remarkable affinity to extracellular vesicles shed from triple-negative breast cancer cells. Here, through in vitro assays and computational analyses, we demonstrate that the binding of ex.50.T to extracellular vesicles and parental breast cancer cells is mediated by recognition of gremlin-1 (GREM1), a bone morphogenic protein antagonist implicated in breast cancer aggressiveness and metastasis. Functionally, we uncover the role of ex.50.T as an innovative therapeutic agent in the process of tumor microenvironment re-modeling, impeding GREM1 signaling, blocking triple-negative breast cancer extracellular vesicles internalization in recipient cells, and counteracting the transformation of normal fibroblasts into cancer-associated fibroblasts. Altogether, our findings highlight ex.50.T as a novel therapeutical avenue for breast cancer and potentially other GREM1-dependent malignancies, offering insights into disrupting TME dynamics and enhancing cancer treatment strategies.

Fig. 1. Ex.50.T specifically targets GREM1.

A Workflow of proteomic analysis and Venn diagrams of the number of proteins for each group. B Western blot of ex.50.T pull-down proteins. PDGFR-β was used as loading control for pulled-down protein; β-actin was used as the loading control of whole lysate; and GREM1 was detected with an anti-GREM1 antibody (Cell Signaling Technology, # 4383S). C ELONA for Ex.50.T on 30 nMol and 60 nMol of recombinant protein. D Increasing concentrations of ex.50.T or CtrlApt were incubated on plates uncoated or coated with GREM1 recombinant protein. Specific binding of ex.50.T was determined by subtracting the values obtained with CtrlApt and reported as Kd ± SE.

Fig. 2. Modeling of ex.50.T on GREM1.

A Schematic representation of ex.50.T in silico modeling from the primary sequence. The minimum free energy 2D structure predicted by 5 programs (RNAfold, RNAstructure, Mfold, pKiss, MXfold2) is shown with line segments between nucleotides representing base pairings. The 3D structure obtained by RNAComposer is shown as a cartoon, with nucleotides highlighted as stubs (color code: A, red; C, yellow; G, green; U, cyan). B Schematic representation of the ex.50.T-GREM1 complex from consensus docking according to the procedure described in the “Materials and methods” section. GREM1 is represented in orange surface and the aptamer as a stub (top panel). Ex.50.T interactions with GREM1 as predicted by MD simulations (bottom panel). Ex.50.T nucleotides and GREM1 residues mainly involved in the aptamer-protein interaction are highlighted as sticks. Hydrogen bonds are depicted as dashed black lines. For the sake of clarity, hydrogen atoms are omitted and only the most conserved interactions along the MD simulations are highlighted.

Fig. 3. Binding analysis of ex.50.T on continuous cell lines and GREM1-modulated cell lines.

A The binding ability of ex.50.T on parental cells of EVs used for exosome SELEX was analyzed by qPCR. Data are mean percentage ± SD over control (fibroadenoma) of two biological replicates. B qPCR analysis of ex.50.T binding on BC continuous cell lines; U87MG cells were used as control cell line. Data are mean percentage ± SD over control of two biological replicates. C Flow cytometry analysis of Alexa-680–ex.50.T binding on PBS-washed MCF7 and BT-549 cells (Total) and high-salt-treated MCF7 and BT-549 cells (Internalized). D Western blot of GREM1 (R&D Systems, # AF956) in MCF7 and BT-549 cell lines. β-actin was used as a loading control. E FACS analysis of ex.50.T in GREM1-overexpressing MCF7 cells (MCF7 24% vs. MCF7 GFP 36%). Data are mean percentage ± SD over control of two independent experiments in three replicates. Western blot analysis was used for overexpression control of GREM1 (Santa Cruz Biotechnology, # sc-515877); β-actin was used as loading control. F FACS analysis of ex.50.T in BT-549 cells after interference with GREM1. Data are mean percentage ± SD over control (BT-549) of all sh-GREM1 interfered points of two independent experiments in three replicates. Western blot analysis was used for down-modulation control of GREM1 (R&D Systems, # AF956); β-actin was used as loading control. G ELISA with GREM1 antibody (sc-515877) (right) and biotinylated ex.50.T ELONA (left) on total protein lysate of U87MG cells overexpressing GREM1. Data are mean percentage ± SD over control U87MG cells of three independent experiments in triplicates. Western blot analysis was used for overexpression control of GREM1 (R&D Systems, # AF956); β-actin was used as loading control. *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 4. GREM1 expression on EVs and Ex.50.T-mediated impairment of BT-549-derived EV uptake and action in NFs.

A Transmission electron micrographs of isolated EVs from BC and glioblastoma cell lines. Each dot is the binding of the primary antibody to GREM1. The GREM1 level appears highest in BT-549 EVs, followed by MCF7 EVs, and finally U87MG EVs. Scale bars = 100 nm. B Transmission electron micrographs of isolated EVs from BC and glioblastoma cell lines. Each dot is the binding of AuNPs functionalized ex.50.T aptamer. The GREM1 level appears higher in BT-549 EVs, compared to MCF7 EVs and U87MG EVs. Scale bars = 100 nm. C Representative images from confocal microscopy of NFs (NFs) from two patients, exposed to PKH26-labeled, BT-549-derived EVs, pre-incubated with either ex.50.T aptamer or CtrlApt. The image shows reduced EV internalization compared to CtrlApt, as evinced by the decreased merged signal (orange). PKH-26 alone was used as negative control (NT). All images were captured at the same settings, enabling direct comparison of staining patterns. NFs were stained using DAPI (blue) and ALEXA488-conjugated anti-β-actin antibody (green), respectively for nuclei and cytoskeleton detection. Magnification 63x. D Western blots on three different NFs for MMP1 and ITGb1 upon treatment of BT-549-derived-EVs, CtrlApt, and ex.50.T, compared to untreated NFs. β-actin was used as loading control. Right panel densitometric analysis of western blot quantification. Data are presented as mean value ± SD over control (NT) of the three biological replicates. E Representative images from confocal microscopy of FAPa and MCT4 (Green, FITC) immunostaining in NFs, exposed to BT-549-derived EVs, pre-incubated with either ex.50.T aptamer or control aptamer (CtrlApt). Nuclei were stained with DAPI (blu).

Fig. 5. Ex.50.T reduces fibroblast-mediated extracellular matrix remodeling.

A Proteome profile Cytokine array analysis of media of NFs treated with EVs alone or in combination with control aptamer or ex.50.T (B). Data are reported as pixel intensity measured with ImageJ. C, D Collagen contraction assay for two patient-derived NFs (top). Histograms (bottom) of collagen plug areas measured with ImageJ. Data are reported as mean + SD of two biological duplicates of two technical replicates. E Collagen contraction assay. Shown are representative pictures of collagen plugs containing NFs treated with EVs from BT-549 cells stably expressing sh-RNA GREM1 (SH-GREM) and a sh-scrambled sequence (NTC). Histograms (bottom) of collagen plug areas, measured with ImageJ. Standard deviations were calculated on replicates from three independent experiments performed with patient-derived NF cells (pt. #72, #75). F Collagen contraction assay. Shown are representative pictures of collagen plugs containing NFs treated with gremlin 1 recombinant protein with or without 500 nMol of ex.50.T (top). Histogram (bottom) of collagen plug areas measured with ImageJ. Standard deviations were calculated on replicates from three independent experiments performed with one patient-derived NF cells (pt. #22). G Serum-starved HUVECs were stimulated with 16 nM of GREM1 for 10 min in the presence or the absence of increasing doses of ex.50.T. Phospho-VEGFR2 (Y1175) and total VEGFR2 (loading control) levels were assessed by Western blot in cell lysates. On the left densitometric quantification of three independent WB experiments. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001.

Fig. 6. BC-derived EVs and ex.50.T aptamer affects NF motility.

A–C Histograms of measured area upon wound healing of NFs treated with EVs alone or pre-incubated with ex.50.T aptamer compared to the CtrlApt. Wound healing areas at 24 h were normalized on the t0 areas and were measured in at least two independent wound sites. Standard deviations were calculated on replicates from three independent experiments performed for each patient. D Histograms of measured area upon wound healing of NFs treated with EVs from BT-549 cells stably expressing sh-RNA GREM1 (SH-GREM) and a sh-scrambled sequence (NTC). Wound healing areas at 24 h were normalized on the t0 areas. Standard deviations were calculated on replicates from three independent experiments performed on NFs. E, F Representative micrographs (top) of NFs migrated through the transwell upon 24 h of treatment, stained with crystal violet. Histograms (bottom) represent the absorbance values of crystal violet (595 nm) eluted from migrated cells, in NFs treated with BT549-derived EVs pre-incubated with ex-50.T or the CtrlApt. Standard deviations were measured in two independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.