2.1 Reagents

Ginkgolic acid C15:1 (#02580585) and crystal violet (#C3886) were purchased from Sigma‒Aldrich (St. Louis, MO). Fibroblast lines were transfected with siRNA using DharmaFECT 2001 (Dharmacon, Lafayette, CO; #T2001). The SUMO1 siRNA (sc-29498) and SUMO2/3 siRNA (#sc-37168) were obtained from Santa Cruz Biotechnology (Heidelberg, Germany).

2.2 Cell culture

CAFs were grown in complete DMEM supplemented with 4.5 g/l glucose and L-glutamine (Sigma Aldrich) in the presence of 10% calf serum. Triple-negative MDA-MB-231 and Luminal A MCF7 cells (ATCC) were grown in complete DMEM supplemented with 4.5 g/l glucose and L-glutamine (Sigma‒Aldrich). T47D (Luminal A) and BT474 (HER2+) cancer cells were grown in RPMI medium. In all the cases, the medium was supplemented with 56 IU/ml penicillin, 56 mg/l streptomycin (Invitrogen, Carlsbad, CA) and 10% fetal bovine serum (FBS; LINUS #16 sV30180.03), and the cells were maintained at 37 °C in a humid atmosphere containing 5% CO2. All cells were routinely tested for mycoplasma contamination. Cell lines were analysed for authentication by the Genomics Core Facility (Instituto de Investigaciones Biomédicas “Alberto Sols” CSIC-UAM. Madrid, Spain) using the STR PROFILE DATA. The STR amplification kits used were as follows: GenePrintR 10 System (Promega); STR profile analysis software, GeneMapper® v3.7 (Life Technologies); and Genomic Analyzer System, ABI 3130 XL (Applied Biosystems).

2.3 Transfection experiments

For the RNAi transfection experiments, a pool of constructs specifically targeting SUMO1 or SUMO2/3 (Santa Cruz Biotechnology) and an RNAi control (sc-37007) were used. In brief, 1 × 106 cells were transfected with DharmaFECT 1 Transfection Reagent (Dharmacon catalog number T-2001) and incubated for up to 72 h before starting the experiment.

2.4 Protein analysis

Proteins were analysed via western blotting as described previously. In brief, proteins were extracted from tumors in TNES buffer [100 mM NaCl, 1% (v/v) NP40, 50 mM Tris-HCl pH 7.6, and 20 mM EDTA] containing protease and phosphatase inhibitor cocktails (Sigma‒Aldrich; #P8340). The proteins recovered were quantified using a BCA Protein Assay Kit (Thermo Fisher Scientific; #23228), resolved by SDS‒PAGE on 10‒12% gradient gels (purchased from Bio-Rad, Berkeley, CA; #456–8085) and transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore, Burlington, MA). The membranes were then probed with the following primary antibodies: anti-SUMO1 (1:1000; #10329-1-AP; Proteintech), anti-SUMO2 (provided by Ron Hay’s laboratory, School of Life Sciences, University of Dundee, Dundee, UK), anti-Rac1 (1:1000; clone 102; BD Biosciences, Franklin Lakes, NJ), anti-Cyclin D1 (1:1000, #sc8396: Santa Cruz Biotechnology), anti-pERK 1/2 (1:1000, #9101: Cell Signaling Technology), anti-ERK total (1:1000, #4696: Cell Signaling Technology), anti-pAKT S473 (1:1000, #32581: Elabscience), anti-AKT total (1:1000, #30471: Elabscience), anti-Tubulin (1:1000; DM1A, Sigma‒Aldrich; #T6199), and anti-GAPDH (1:1000, Sigma‒Aldrich; #G8795). Antibody binding was detected with horseradish peroxidase (HRP)-conjugated anti-mouse, anti-rabbit or anti-sheep secondary antibodies (1:10,000 dilution, Bio-Rad) and visualized by enhanced chemiluminescence (Bio-Rad). The images were obtained with an ImageQuant LAS 500 chemiluminescence CCD camera (GE Healthcare Life Sciences, Chicago, Illinois, USA).

2.5 Rac1 GTPase assay

Endogenous active GTP-bound Rac1 levels were determined using a GST-PAK1 binding domain (PBD) pull-down assay. The cells were lysed in FISH buffer [Glicerol 100X, 1 M Tris pH 7.4, 5 M NaCl, 1% (v/v) NP40 and 1 M MgCl2], and equal volumes were incubated with 20 μg of GST-PBD beads for 15 min at 4 °C. The precipitated complexes were washed three times with excess lysis buffer. After the final wash, the supernatant was discarded, and the beads were resuspended in 10 μl of 5X Laemmli buffer. GTP-bound Rac1 was detected by western blotting. The amount of GTP-bound Rac1 was normalized to the total amount of Rac1 of each sample.

2.6 Cell viability

CAFs and breast cancer cells were plated for 24, 48 or 72 h. After that, they were fixed for 10 min in PBS supplemented with 4% paraformaldehyde at room temperature and stored in PBS supplemented with 0.05% azide at 4 °C prior to staining with crystal violet for 30 min at room temperature. The plates were washed with water, and the crystal violet was removed from the cells with 1% Triton X-100 in PBS, after which the OD was measured at 570 nm.

2.7 Conditioned media experiments

CAFs were grown for two days in the absence of serum, and the culture supernatants were collected and used as conditioned media (CM) in proliferation, migration and invasion experiments. For SUMOylation inhibition, CAFs were treated with distinct concentrations of GA for 24 h or vehicle (ethanol). Then, the culture medium was removed, and the cells were grown for two days in the absence of serum. The resulting supernatant was filtered (0.22 μm filter), to remove unattached cells and cell debris.

2.8 Boyden chamber cell migration assay

Cell migration was assayed in Boyden chambers (8.0 μm pore-size polyethylene terephthalate membrane with a cell culture insert: VWR, Radnor, PA; #VWRI734-2744. The cells were trypsinized and counted, and cell suspensions containing 5 × 104–10 × 104 cells in 200 μl of serum-free medium were added to the upper chamber, with 500 μl of the appropriate medium added to the lower chamber. The transwell inserts were incubated for 24 h for MDA-MB-231 cells or 48 h for T47D cells at 37 °C, after which the cells on the inside of the transwell inserts were removed with a cotton swab, whereas those on the underside of the insert were fixed and stained. Images of five random fields were taken, and the cells were counted to calculate the proportion of transmigrated cells. Boyden chambers with Matrigel (Cell Biolabs, San Diego, CA; #CBA-110) were used to assay cell invasion.

2.9 Wound assays

MCF7 and T47D cells were grown to confluence overnight with 0% FBS. The cell monolayer was scraped in a straight line with a pipette tip to create a linear wound. Debris was removed by washing the cells with PBS, after which the medium was replaced with CM. Images of the wounds (at least 3 fields per condition) were taken with a Nikon TMS microscope with a 10X objective. For each image, the cells that migrated into the original wound were quantified. The migration area was determined by measuring the total area of the wound using ImageJ software.

2.10 Gelatin zymograms

To assess MMP-2 and MMP-9 activity, conditioned media from CAFs were collected. The protein concentration was determined by the Bradford method, and 40 μg of conditioned medium from each condition was assayed for proteolytic activity on gelatin-substrate gels. Briefly, the samples were mixed with nonreducing buffer containing 2.5% SDS and separated on 8% acrylamide gels copolymerized with 1 mg/mL gelatin, as previously described [28]. After electrophoresis at 72 V for 2.5 h, the gels were rinsed twice in 2.5% Triton X-100, and then incubated in 50 mM Tris-HCl (pH 7.4) and 5 mM CaCl2 assay buffer at 37 °C for 24 h. The gels were fixed and stained with 0.25% Coomassie Brilliant Blue G-250 in 10% acetic acid and 30% methanol. Proteolytic activity was detected as clear bands against the background stain of the undigested substrate in the gel. Quantification was performed using ImageJ software (NIH, Bethesda, MD, USA).

2.11 Limiting dilution assays (LDAs)

Limiting dilution assays were performed as previously described [29]. Briefly, cultures of patient samples were plated at a density of 105 cells and incubated with the different treatments for 3 days. Then, the cells were counted and plated in fresh medium in the absence of the drug in 96-well plates at different densities (100, 50, and 10 cells per well). One week later, tumorsphere formation was evaluated; wells in which at least one tumorsphere was present were considered positive. The data in the corresponding representations indicate the fraction of cells with the ability to generate new sphere cultures. Graphs were generated using the ELDA software application [29], which adjusts the data obtained under each experimental condition to the limiting dilution model. In these graphs, the slopes of the depicted solid lines correspond to the fraction of cells with the ability to generate new sphere cultures. A lower slope indicates a lower fraction of cells with the capacity to generate new spheres. Dotted lines represent the 95% confidence intervals.

2.12 SILAC-based quantitative proteomics

CAFs were maintained in DMEM supplemented with 10% dialyzed FBS, 100 units/mL penicillin/streptomycin and either naturally occurring isotopes (“light”) or stable isotope-labelled (heavy”) 13C6 arginine and 13C6 lysine amino acids. The culture media was changed when 100% (1 × 106 cells) confluence was reached. At least 6 doublings were added to the cells to allow full incorporation of the labelled amino acids, which was verified by MS analysis of the protein digest (data not shown). After differential labelling, the control and cells exposed to 10 µM and 30 µM GA were mixed in a 1:1 ratio with the vehicle. Then, the cells were lysed with buffer containing 50 mM Tris, 150 mM NaCl, 0.2 mM EDTA and protease inhibitors. The protein extracts were separated via SDS‒PAGE on a 10% SDS‒polyacrylamide gel, fixed with 7.5% acetic acid and 30% ethanol, and visualized via Coomassie blue staining. The excised gel bands were destained first in 25 mM ammonium bicarbonate followed by 25 mM ammonium bicarbonate/acetonitrile (50:50) and then dried. The gel pieces were rehydrated with 12.5 ng/μL trypsin solution in 25 mM ammonium bicarbonate and incubated overnight at 37 °C. Peptides were extracted using acetonitrile and 5% formic acid, dried by vacuum centrifugation and reconstituted in 12 μg/L 2% acetonitrile and 0.1% formic acid. Peptide mixtures were analysed using nanoflow LC‒MS/MS (Thermo Scientific). Peptides were preconcentrated onto a 20 mm × 75 μgm C18 precolumn and separated using a reversed-phase column (500 mm × 75 μgm fused silica capillary C18 RSLC PepMap, 2 μgm, 100 Å; Thermo Scientific) with a linear gradient of 2–40% acetonitrile in 0.1 formic acid. The samples were delivered over 240 min at a flow rate of 250 nL/min through the analytical column to a stainless steel nanobore emitter (Proxeon). The peptides were scanned with a Q-exactive HF Orbitrap mass spectrometer (Thermo Scientific) operated in full scan MS mode with a resolution of 60,000 in a mass range of 350–2000 Da. MS/MS data were acquired in data-dependent acquisition mode. Up to fifteen precursors were selected in each microscan with charges ranging from +2 to +6 and 27 s of dynamic exclusion, followed by isolation with a ±2 m/z window width and a maximum time of 120 ms for fragmentation by high collision dissociation with a normalized collision energy of 27%. The acquired MS/MS spectra were analysed with Proteome Discoverer 2.4 software (Thermo Scientific) using the MASCOT search engine. The UniProt Homo sapiens database contained 20,416 sequences (19/10/20). Methionine oxidation, protein N-terminus acetylation and 13C6-Arg and 13C6-Lys were specified as variable modifications, and trypsin was used as the specific enzyme, while two missed cleavage points were allowed. The minimum precursor peptide and fragment tolerance were set at 10 ppm and 0.02 Da, respectively. The false positive rate was estimated by means of the percolator algorithm, and a p value < 0.01 was used to filter the identified proteins. The SILAC ratio was defined as the area of the heavy peptide (13C) divided by the area of the light peptide (12C). SILAC ratio variability was assessed, considering exclusively those with a variability less than 30%. Moreover, the quantified proteins were subjected to a t test, and only those with a p value < 0.05 were considered to be significantly different.

2.13 Chicken embryo experiments

In vivo experiments do not require any special allowance as long as the embryos are sacrificed before hatching, as in this study. For the CAM xenografts, we used premium specific pathogen-free (SPF), fertile, 11-day-incubated embryonated chicken eggs. We inoculated 1 × 106 MDA-MB-231 cells in PBS and Matrigel (1:1). Tumors were grown for seven days. Four and six days after inoculation, the tumors were treated with CM from CAFs previously treated with 10 µM or 30 µM GA. After seven days, the tumors were excised and weighed, and the bone marrow was collected for metastasis analysis.

2.14 Quantitative detection of human tumor cell metastasis

The detection of human tumor cells in chick bone marrow (BM) was based on the quantitative detection of human ALU sequences present in chicken lung DNA extracts as a modification of the method developed by Kim et al. [30]. Briefly, the BMs were frozen in liquid nitrogen, and the genomic DNA was extracted using K proteinase. Thirty nanograms of genomic DNA was used per PCR. To detect human cells in chicken tissues, primers specific for the human ALU sequences F-ACGCCTGTAATCCCAGCACTT and R-TCGCCCAGG CTGGAGTGCA were used to amplify the human repeats present in the genomic DNA extracted from chicken BM. Amplification of chicken GAPDH was used as an internal control for the total amount of tissue.

2.15 Human CAF isolation

CAFs from the Luminal-HER2+ and TNBC cell lines were a gift from Paloma Bragado. The nontumor-associated NF cell line and luminal CAFs were a gift from Akira Orimo [31,32,33]. To establish the non-tumour associated NF cell line, we obtained fresh healthy mammary tissue from a woman undergoing reduction mammoplasty [33]. CAFs from the Luminal-HER2+ and TNBC cell lines were established from tumour tissue excised from breast cancer patients: a Luminal B-HER2+ patient and TNBC patients [32]. These fibroblasts (NFs and CAFs) were immortalized by infection with the pMIG (MSCV-IRES-GFP) retroviral vector, expressing both hTERT and GFP, and with a pBabe-puro vector encoding a puromycin resistance gene, before they were co-implanted with breast carcinoma cells into nude mice to generate Lum-CAFs [33].

2.16 Breast cancer patient samples

Ascites effusions were collected by the Manchester Cancer Research Centre (MCRC) Biobank (ethical approval: 12-ROCL-01). The metastatic fluid was centrifuged at 1000×g for 10 min, after which the cell pellet was resuspended in PBS. Centrifugation was repeated, and the cell pellet was resuspended in 5 ml of ACK lysis buffer for 5 min at room temperature to lyse red blood cells. Up to 50 ml of PBS was added, followed by centrifugation, counting and freezing the cells in FBS + 10% DMSO. For the experiments, the samples were defrosted rapidly and seeded as suspension cultures in polyhema-coated plates (P3932-Sigma) in advanced DMEM/F12 supplemented with 5% FBS, 2 mM L-glutamine, 10 mM HEPES, 56 IU/ml penicillin, and 56 mg/l streptomycin. Informed consent was obtained from all patients prior to sampling, and the samples were stored according to the Human Tissue Act (HTA) regulations.

2.17 Statistical analyses

The data are expressed as the mean ± standard error of the mean (s.e.m.) or standard deviation (s.d.) and were evaluated with the Mann–Whitney U test (nonparametric) or t test (parametric) for two groups and with the Kruskal–Wallis test (followed by Dunn’s multiple comparison posttest) or ANOVA (followed by Tukey’s multiple comparisons test) for more than two groups. For all the analyses, p values ≤ 0.05 were considered to indicate statistical significance.