LRP-1 Matricellular Receptor Involvement in Triple Negative Breast Cancer Tumor Angiogenesis

Background: LRP-1 is a multifunctional scavenger receptor belonging to the LDLR family. Due to its capacity to control pericellular levels of various growth factors and proteases, LRP-1 plays a crucial role in membrane proteome dynamics, which appears decisive for tumor progression. Methods: LRP-1 involvement in a TNBC model was assessed using an RNA interference strategy in MDA-MB-231 cells. In vivo, tumorigenic and angiogenic effects of LRP-1-repressed cells were evaluated using an orthotopic xenograft model and two angiogenic assays (Matrigel® plugs, CAM). DCE-MRI, FMT, and IHC were used to complete a tumor longitudinal follow-up and obtain morphological and functional vascular information. In vitro, HUVECs’ angiogenic potential was evaluated using a tumor secretome, subjected to a proteomic analysis to highlight LRP-1-dependant signaling pathways. Results: LRP-1 repression in MDA-MB-231 tumors led to a 60% growth delay because of, inter alia, morphological and functional vascular differences, confirmed by angiogenic models. In vitro, the LRP-1-repressed cells secretome restrained HUVECs’ angiogenic capabilities. A proteomics analysis revealed that LRP-1 supports tumor growth and angiogenesis by regulating TGF-β signaling and plasminogen/plasmin system. Conclusions: LRP-1, by its wide spectrum of interactions, emerges as an important matricellular player in the control of cancer-signaling events such as angiogenesis, by supporting tumor vascular morphology and functionality.


Introduction
Breast cancer (BC) is the most diagnosed cancer in women worldwide and the leading cause of cancer-related death. It is an heterogenous disease characterized by diverse phenotypes and a considerable heterogeneity in molecular and histopathological features [1]. Based on transcriptomics analysis, five BC subtypes have been identified: luminal A, luminal B and human epidermal growth factor 2 receptor (HER2)-enriched, basal-like, proteomics analysis of TCM showed that LRP-1 supports angiogenesis and tumor growth through the TGF-β signaling and plasminogen/plasmin system modulation, among others. Thus, the matricellular receptor LRP-1, by its wide spectrum of interactions within the microenvironment, appears as a key factor in the control of BC signaling events such as angiogenesis.
For all procedures, the cells were harvested using a 1X Trypsin-EDTA solution (Sigma Aldrich, St. Louis, MI, USA) or Accutase ® (C-41310; Promocell, Handschuhsheimer, Germany) and were maintained at 37 • C in a humidified atmosphere of 5% CO 2 .

Tumor-and HUVEC-Conditioned Media Preparation
48-h tumor-conditioned media (TCM): shLRP-1 or shCtrl MDA-MB-231 cells were seeded at 3.2 × 10 6 in T150 culture flasks. Forty-eight hours after seeding, the media were replaced by 8 mL of DMEM containing 1% FBS. After 48 h of incubation, the supernatant was centrifuged at 10,000× g for 10 min. In parallel, cells were harvested and counted using Scepter TM 2.0 (Merck Millipore, Molsheim, France). Cell equivalents between shLRP1 and shCtrl TCM were made by diluting the most concentrated TCM with DMEM. The resulting TCM, equivalent in pairs at a cell concentration from 0.8 to 1.2 million cells/mL, were stored in aliquots at 20 • C to avoid multiple freeze-thaws. 24-h TCM: shLRP-1 or shCtrl MDA-MB-231 cells were seeded at 1.2 × 10 6 in a 35-mm culture dish. Forty-eight hours after seeding, the media were replaced by 3.5 mL of FBS-free, phenol-red-free DMEM after washing the cells with PBS twice. After 24 h of incubation, the supernatant was centrifuged at 10,000× g for 10 min. In parallel, cells were detached and counted using Scepter TM 2.0 (Merck Millipore, Molsheim, France). Cell equivalents between shLRP-1 and shCtrl TCM were made by diluting the most concentrated TCM in DMEM. The resulting TCM, equivalent in pairs at a cell concentration from 0.8 to 1.2 million cells/mL, were stored in aliquots at 20 • C to avoid multiple freeze-thaws. 24-h TCM-stimulated HUVEC-conditioned medium (CM): HUVECs were seeded at 1.2 × 10 6 in a 35-mm culture dish. Twenty-four hours after seeding, the media were replaced by 24 h of shLRP-1 or shCtrl MDA-MB-231 TCM as a pre-treatment for 24 h after washing the cells with PBS twice. After treatment incubation, the media were replaced by 3.5 mL of FBS-free, phenol-red-free DMEM after washing the cells twice with PBS. After 24 h of incubation, the supernatant was centrifuged at 10,000× g for 10 min. The resulting CMs were stored in aliquots at 20 • C to avoid multiple freeze-thaws.

In Vivo Studies
Mice (5)(6) week-old female Balb/c nu) purchased from Janvier (Janvier labs, Le Gnest-Saint-Isle, France) were housed in ventilated cages under filtered air and acclimatized for one week prior to manipulation. The experiments with animals were approved and carried out in compliance with ethics rules under the authorization number APAFIS#4373-2016030410575189 vI, "Study of LRP-1 receptor involvement in TNBC models in mice", distributed by the higher education and research administration attached to the French National Education Ministry. All procedures were conducted under general anesthesia induced by the inhalation of 3% isoflurane and maintained with 1.5% during imaging.

Orthotopic Xenograft Model
shLRP-1 or shCtrl MDA-MB-231 cells were harvested using Accutase ® , washed and resuspended into a 5 × 10 7 /mL cell solution before inoculation. Twelve mice were injected with 100 µL into the mammary fat pad. Tumor growth was assessed by measuring the length (A) and width (B) with a digital caliper every week. The volumes were calculated using 1/2(A × B 2 ). The mice were sacrificed 28 days after inoculation. After excision, the tumor tissues were immersed in liquid nitrogen, transferred to a vial, and stocked at −80 • C or fixed in 4% paraformaldehyde (Sigma Aldrich, Saint-Louis, MI, USA) for 24 h and embedded in paraffin.

Matrigel ® Plug
A total of 2 × 10 5 of shLRP-1 or shCtrl MDA-MB-231 cells were resuspended in 0.1 mL of growth medium, mixed with 0.4 mL of growth factor-reduced Matrigel ® (Corning ® , BD Biosciences, Franklin Lakes, NJ, USA) at 8.6 mg/mL, and implanted subcutaneously into the flank of each 7-week-old female BALB/c-nu mouse (Janvier labs, Le Genest-Saint-Isle, France) (n = 12/group). Twenty-one days after the injection, the animals were sacrificed, and the Matrigel ® plugs were excised, photographed, and fixed in 4% paraformaldehyde (Sigma Aldrich, Saint-Louis, NJ, USA) for histological analysis.

Optical Imaging
Fluorescent molecular tomography (FMT) was conducted using an FMT-4000 scanner (PerkinElmer, Waltham, MA, USA) calibrated beforehand with fluorophores according to the supplier's instructions. Fluorescence quantification was achieved with the TrueQuant 3.0 software (PerkinElmer, Waltham, MA, USA). The AngioSense TM -750/AngioSense TM -680 or HypoxiSense TM -680 contrast agent was used, with an excitation filter of 750 ± 3 nm or 670 ± 3 nm, respectively, and an emission filter of 690-740. 3D and 2D trans-illumination acquisitions were carried out 24 h after the injection of a 100 µL intravenous contrast agent.

MRI Imaging
Images were acquired using a three-dimensional (3D) coronal T2-weighted fast spin echo sequence (FSE) with echo time (TE)/repetition time (TR) = 68/5000 ms, a 60 mm × 60 mm field of view, 1 mm slice thickness, and a 512 × 240 matrix (with 2× oversampling in the read direction) zero-padded to 512 × 256 during reconstruction (resolution = 0.234 mm × 0.25 mm), echo train length = 8, number of averages = 1, and an effective receiver bandwidth (BW) = 20 kHz); a transverse T2-weighted fast spin echo sequence (FSE) with echo time (TE)/repetition time (TR) = 68/5000 ms, a 40 mm × 40 mm field of view, 1 mm slice thickness, and a 256 × 240 matrix zero-padded to 256 × 256 during reconstruction (resolution = 0.156 mm × 0.167 mm), echo train length = 8, number of averages = 2, receiver bandwidth (BW) = 20 kHz); with echo time (TE)/repetition time (TR) = 11/720 ms, a 16 mm × 16 mm × 16 mm field of view, a 128 × 128 × 80 matrix (resolution = 0.125 mm × 0.125 mm × 0. time was~40 min. Intensity analyses were performed after a 4D reconstruction with the OsiriX software (Pixmeo, Swiss) using the ROI enhancement plugin. The intensity of the ROI in the plugs was quantified with the OsiriX software after the 4D reconstruction. A mouse body coil was used. The temperature was maintained by a heating air flow built into the animal bed system. Respiration was monitored throughout the entire scan.

Chick Chorioallantoic Membrane (CAM) Assay
Fertilized chicken eggs (EARL Les Bruyères, FR) (less sentient model according to 3R rule) were incubated in egg-racks at 37 • C and a 65% humidified atmosphere in a rotatory incubator allowing rotations of 25 degrees every 6 h. On day 3 of embryonic development, a window was made in the eggshell and sealed with adhesive film (Durapore tape). On embryonic day 10, once the CAM was fully formed, a gentle laceration of its surface was performed using a scalpel, and plastic rings (made from Nunc Thermanox coverslips) were put on the surface of the CAM [21]. Then, 4 × 106 of shLRP-1 or shCtrl MDA MB-231 cells were deposited as a thin layer on the surface of the fertilized chicken eggs. On day 17, CAMs were fixed in vivo (4% paraformaldehyde, RT, 30 min) and included. Digital photographs were taken under a Nikon SMZ800 stereomicroscope. A quantitative analysis was performed using a MATLAB routine developed by Dr El-Hadi Djermoune [22] in which the vascular structures' segmentation leaned on the vesselness probability map of the 2D images [23].

Histology
A histological analysis of formaldehyde-fixed and paraffin-embedded tumors and Matrigel ® plugs was performed on 4 µm hematein, eosin, and saffron (HES)-stained sections prepared using routine methods. The necrosis area was calculated as the ratio of (necrosis surface/tumor surface) × 100 and the mitotic index in 10 consecutive high-power fields (HPF = 0.31 mm 2 ) in the highest mitotic activity area.

Immunohistochemistry and Immunofluorescence
For immunohistochemistry, tumor and Matrigel ® plug sections were deparaffinized in xylene and rehydrated in solutions of graded ethanol. The antigen retrieval was performed at 95 • C for 40 min in 0.01 M sodium citrate buffer sections, which were then immersed in 3% hydrogen peroxide for 30 min to block the endogenous peroxidase activity. For vessel labeling, the sections were incubated with an anti-CD31 [EPR17259] (1/100, rabbit monoclonal, ab18298; Abcam, Cambridge, UK) primary antibody (1/100) at 4 • C overnight. Following primary antibody incubation and washing, the sections were treated using the labeled polymer peroxidase AEC method (Dako EnVisionThermo ScientificTM HRP Lab VisionTM RTU AEC Substrate System Kit, DakoFisher Scientific, CarpinteriaPittsburgh, CAPA) for 60 min. The proteins were visualized using a liquid diaminobenzidine substrate kit (Zymed Laboratories, San Francisco, CA, USA). The sections were counterstained with hematoxylin before mounting. Appropriate positive and negative controls were used throughout the experiment. CD31-immunostained vessels, exhibiting a lumen, were numerated on 5 consecutive HPF in the area with the vessels' greatest density. The interpretation was fulfilled blindly by an external anatomopathologist.
For immunofluorescence, an anti-vimentin (V6) (1/100, mouse monoclonal, sc-6260; Santa Cruz, CA, USA) primary antibody was used. For the dilution of the antibodies, the preincubation and washing of the cryosections, 0.4% Triton X-100, and 1% bovine serum albumin diluted in a potassium phosphate buffer (pH 7.4) were performed. All processing of cryosections was done in humidified chambers. The primary antibody was applied overnight at 4 • C. After four washes within 20 min, the secondary antibody was applied for 60 min. After repeated washing, the sections were mounted under coverslips with ProLong TM Gold Antifade Mountant (Invitrogen, Waltham, MA, USA). Appropriate positive and negative controls were used throughout experiments.

RNA Isolation and Real-Time PCR
Total mRNA was extracted using the TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA). PCR primers were synthesized by Eurogentec (Liege, Belgium) as follows (5 -3 ) for LRP1: GCTATCGACGCCCCTAAGAC and CGCCAGCCCTTTGAGATACA (Table S1). mRNA analyses were performed on breast cancer cell lines. The total RNA was isolated using Extract-All (Eurobio, Les Ulis, France) and DNase treated (RQ1 RNase-Free DNase, Promega) as described in the manufacturer's instructions. The RNA quality was checked by 1% agarose gel electrophoresis, and the total RNA concentration (ng/µL) was measured at 260 nm by a NanoDropTM One (Thermo Fisher Scientific, Waltham, MA, USA) for each sample. Two hundred and fifty nanograms of total RNA were used for reverse-transcription using the VERSO cDNA kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions, using a mix of random hexamer primers and anchored oligo dT. The transcript levels were determined by a real-time quantitative analysis using an Absolute SYBR Green Rox mix (Fisher Scientific) on a CFX 96 touch real time PCR detection system (Bio-Rad). PCR reactions were carried out in duplicates in 96-well plates (15 µL per well) in a buffer containing 1× SYBR Green mix (including Taq polymerase, dNTPs, SYBR Green dye), 280 nM forward and reverse primers, and a 1:10 dilution of reverse transcript RNA. After denaturation at 95 • C for 15 min, the amplification occurred in a two-step procedure: 10 s of denaturation at 95 • C and 45 s of annealing/extension at 60 • C, with a total of 40 cycles. Identical thermal cycling conditions were used for all targets. The specificity of PCR amplification was checked using a heat dissociation curve from 65 • C to 95 • C following the final cycle. The cycle threshold (Ct) values were recorded with the Bio-Rad CFX ManagerTM 3.1 software (Bio-Rad). Specific primers were designed using the Primer3 and BLAST softwares (National Center for Biotechnology Information) and are presented in the Supplementary Table S1. The PCR efficiency of the primer sets was calculated by performing a real-time PCR on serial dilutions and was 90% to 110%. For each experiment, PCR reactions were performed in duplicate and 3 independent experiments were analyzed. The results correspond to the means ± standard deviation (SD) of the duplicate reactions of three independent experiments. The relative gene expression was determined with the formula fold induction: 2 −∆∆Ct , where ∆∆Ct = (Ct GI [unknown sample] − Ct GI [reference sample]) − (Ct reference genes [unknown sample] − Ct reference genes [reference sample]). GI is the gene of interest. RS18 and RPL32 were used as internal controls. The reference sample is the MDA-MB-231 WT or shCtrl sample, chosen to represent 100% of the GI expression. The means ± SEM originated from 3 independent experiments realized in duplicates.

Tubule Formation
A growth-factor-reduced (GFR) Matrigel ® (Corning ® , BD Biosciences, Franklin Lakes, NJ, USA) at 8.6 mg/mL was thawed on ice at 4 • C overnight before use. Ten microliters of GFR Matrigel ® were loaded into each well of a pre-cooled µ-Slide Angiogenesis plate, ibiTreat (ibidi TM , Martinsried, DE, USA), and the plate was incubated at 37 • C for 30 min. As mentioned in the Materials and Methods section, 1.5 × 10 4 GFP-HUVECs cells were seeded in 50 µL of TCM to be tested and for controls, EGM-2, EBM-2, and 0.8% FBS DMEM. The plate was then incubated at 37 • C in a humid atmosphere in the presence of 5% CO 2 for 8 h. A photography of each well was taken using a fluorescence microscope (X4) coupled to a camera. After 8 h at 37 • C, the cells were imaged at ×4 magnification on a Nikon eclipse 300 inverted microscope. The total network length and branching number were assessed using AutoTube [24]. The results are the means of random fields in 3 replicates and were repeated three times.

Endothelial Proliferation and Migration
An MTT assay was realized as described in [25]. Briefly, HUVECs were seeded in 96-well plates at a density of 1 × 10 4 cells/mL in 100 µL of growth medium. Twenty-four hours later, the medium was replaced by 100 µL of TCM to be tested or control conditions (EGM-2, EBM-2 and 1% FBS DMEM) after rinsing the cells with PBS. Then, 20 µL of MTT (5 mg/mL) were added into each well after 0, 24, 48, and 72 h of treatment. Four hours later, 150 µL of dimethyl sulfoxide were added to each well. The absorbance (optical density, OD) at 560 nm was measured using a microplate reader (Tecan Infinite ® , Mannedorf, Switzerland). The experiments were performed in triplicate. Migration experiments were carried out using ThinCert TM cell culture inserts (BD Biosciences, Franklin Lakes, NJ, USA) in 8-µm-pore, fibronectin-coated membranes in a 24-well plate, as described in [26]. Briefly, HUVECs were seeded at a density of 0.15 × 10 6 cells/cm 2 on ThinCert TM pre-coated with fibronectin from bovine plasma (Sigma-Aldrich, Saint-Louis, MI, USA) at 7 µg/mL overnight. The surplus was eliminated. After 30 min of hood drying, the lower well was filled with 800 µL of EGM-2, EBM-2, 0.8% FBS DMEM, and 48 h TCM to be tested containing 182 µL of fresh DMEM 3.5% FBS (for a final FBS concentration of 0.8%). Two hundred microliters of the HUVEC cell solution adjusted to 5 × 10 4 cells/mL in EBM-2 were added to the upper well of each insert. The 24 well-plates were incubated at 37 • C in a humid atmosphere in the presence of 5% CO 2 . After 8 h, the medium was removed and replaced with cold methanol for 15 min at RT to fix the cells. The inserts were then rinsed by successive baths in distilled water. The cells that did not migrate on the upper well of the insert were eliminated using a cotton swab. The membranes were excised from inserts and mounted on microscopic observation slides with a ProLong ® Gold Antifade Reagent mounting medium (with DAPI (4 6-diamidino-2-phenvlindole)) (Invitrogen, Waltham, MA, USA). The cells were counted on 9 random microscopic fields per membrane using a fluorescence microscope (X20) (Evos, Thermo Fisher Scientific, Waltham, MA, USA) coupled to a camera. The experiments were carried out in triplicate and repeated with three independent TCM.

Proteomics
For label-free quantitative proteomics, three independent biological replicates on secretome extracts for shLRP-1 and shRNA-control cell lines have been performed. Ten micrograms of proteins were loaded on a 10% acrylamide SDS-PAGE gel, and the proteins were visualized by Colloidal Blue staining. The migration was stopped when the samples had just entered the resolving gel, and the unresolved region of the gel was cut into only one segment. The steps of sample preparation and protein digestion by trypsin were performed as previously described [27]. A nanoLC-MS/MS analysis was performed using an Ultimate 3000 RSLC Nano-UPHLC system (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a nanospray Orbitrap Fusion TM Lumos TM Tribrid TM Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Each peptide extract was loaded on a 300-µm ID × 5 mm PepMap C18 precolumn (Thermo Fisher Scientific, Waltham, MA, USA) at a flow rate of 10 µL/min. After a 3-min desalting step, the peptides were separated on a 50-cm EasySpray column (75 µm ID, 2 µm C18 beads, 100 Å pore size, ES803A rev.2, Thermo Fisher Scientific, Waltham, MA, USA) with a 4-40% linear gradient of solvent B (0.1% formic acid in 80% ACN) in 115 min. The separation flow rate was set at 300 nL/min. The mass spectrometer operated in positive ion mode at a 2.0 kV needle voltage. The data were acquired using the Xcalibur 4.1 software in a data-dependent mode. MS scans (m/z 375-1500) were recorded at a resolution of R = 120,000 (@ m/z 200) and an AGC target of 4 × 10 5 ions collected within 50 ms, followed by a top speed duty cycle of up to 3 s for MS/MS acquisition. Precursor ions (2 to 7 charge states) were isolated in the quadrupole with a mass window of 1.6 Th and fragmented with HCD@30% normalized collision energy. MS/MS data were acquired in the ion trap with the rapid scan mode, an AGC target of 3 × 10 3 ions, and a maximum injection time of 300 ms. The selected precursors were excluded for 60 s. Protein identification and Label-Free Quantification (LFQ) were done Biomedicines 2021, 9, 1430 8 of 22 in Proteome Discoverer 2.4. The MS Amanda 2.0, Sequest HT, and Mascot 2.4 algorithms were used for protein identification in batch mode by searching against a Uniprot Homo sapiens database (75,093 entries, release 20 May 2020). Two missed enzyme cleavages were allowed for trypsin. Mass tolerances in MS and MS/MS were set to 10 ppm and 0.6 Da. Oxidation (M), acetylation (K), and deamidation (N, Q) were searched as dynamic modifications, and carbamidomethylation (C) as a static modification. Peptide validation was performed using the Percolator algorithm [28], and only "high confidence" peptides were retained, corresponding to a 1% false discovery rate at the peptide level. A Minora feature detector node (LFQ) was used along with the feature mapper and precursor ion quantifier. The normalization parameters were selected as follows: (1) Unique peptides, (2) Precursor abundance based on intensity, (3) Normalization mode: total peptide amount, (4) Protein abundance calculation: summed abundances, (5) Protein ratio calculation: pairwise ratio-based, and (6) Hypothesis test: t-test (background-based). Quantitative data were considered for master proteins, quantified by a minimum of 2 unique peptides, fold changes above 2, and a statistical p-value lower than 0.05. The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE [29] partner repository with the dataset identifier PXD022978.

S-2251 TM Assay
A chromogenic substrate selective for plasmin, S-2251 (Chromogenix, Diapharma, Westchester, NY, USA), was used to follow the initial rate of plasminogen activation by measuring p-nitroaniline generation. TCM were mixed with a buffer containing 0.1 M Tris-HCl pH 7.8 and 20 µL of plasminogen EACA reconstituted at 10 U/mL (Calbiochem, Darmstadt, Germany). To initiate the reaction, 20 µL of 3.5 mM S-2251 TM were added to each well. The generation of plasmin was detected by measuring the absorbance of the p-nitroaniline release every 30 min at 405 nm during 10 h (Tecan Infinite ® , Mannedorf, Switzerland).

Patient Tumor-Derived Breast Cancer Xenografts (PDX)
Tumor fragments used to generate PDXs were collected from patients upon signing an informed consent, and PDX models have been generated according to previous studies [30].

RNA Seq
The RNA seq analysis was outsourced either to Integragen (France) or Novogene (China) by using poly-T oligo enriched RNA, a strand-specific library, and paired-end sequencing (Illumina).

Statistical Analysis
The data are expressed as means +/− SEM, as indicated in the figure legends. For the statistical analysis, an independent t-test in vitro and a non-parametric Mann-Whitney in vivo were used to assess the significance of the mean differences. The differences were considered significant at a p value of 0.05 or less. * p < 0.05; ** p < 0.01; *** p < 0.005; **** p < 0.001.

LRP-1 Is Preferentially Expressed in TNBC Cell Lines
Using RNA-sequencing (Xentech biotechnology), we analyzed the quantity and the sequences of LRP-1 RNA in xenograft (PDX) derived from 20 breast cancer patients, including 12 TNBC and 8 non-TNBC (seven luminal and one HER2+). We highlighted no significant differences between the two groups but a trend of higher LRP-1 RNA expression in the TNBC group. However, LRP-1 RNA expression was found to be higher in 8/12 of TNBC PDXs compared to the average expression of the non-TNBC PDXs (with a mean of 67.86 vs. 23.07) ( Figure 1A). We also evaluated the LRP-1 expression level in TNBC cell lines, MDA-MB-231, Hs-578T, BT-20, and 4T1, and in non-TNBC cell lines, MCF-7, SKBR3, and T47D. LRP-1 was found to be more expressed at the transcriptional and translational levels in TNBC cell lines (MDA-MB-231 > 4T1 > Hs578T >  in comparison to non-TNBC cell lines (T47D > MCF-7 > SK-BR3) ( Figure 1B,C). Therefore, to investigate LRP-1's role in TNBC progression, we used the stably transfected MDA-MB-231 cell line to allow for a constitutive expression of LRP-1-targeting shRNA (shLRP-1) or a scrambled shRNA (shCtrl). RT-qPCR and the immunoblot showed a significant decrease in LRP-1 mRNA (by 60%) and protein (by 67%) expression, respectively, in shLRP-1 MDA-MB-231 cells compared with shCtrl ( Figure 1D-F). These results validated our LRP-1 study model in MDA-MB-231 cells. As shown in Figure S1, the LRP-1 expression in MDA-MB-231 without antibiotic selection pression showed no significant difference up to 35 days, indicating that LRP-1-targeting shRNA was stable over time and compatible with in vivo experiments ( Figure S1). sequences of LRP-1 RNA in xenograft (PDX) derived from 20 breast cancer patients, including 12 TNBC and 8 non-TNBC (seven luminal and one HER2+). We highlighted no significant differences between the two groups but a trend of higher LRP-1 RNA expression in the TNBC group. However, LRP-1 RNA expression was found to be higher in 8/12 of TNBC PDXs compared to the average expression of the non-TNBC PDXs (with a mean of 67.86 vs. 23.07) ( Figure 1A). We also evaluated the LRP-1 expression level in TNBC cell lines, MDA-MB-231, Hs-578T, BT-20, and 4T1, and in non-TNBC cell lines, MCF-7, SKBR3, and T47D. LRP-1 was found to be more expressed at the transcriptional and translational levels in TNBC cell lines (MDA-MB-231 > 4T1 > Hs578T >  in comparison to non-TNBC cell lines (T47D > MCF-7 > SK-BR3) ( Figure 1B,C). Therefore, to investigate LRP-1′s role in TNBC progression, we used the stably transfected MDA-MB-231 cell line to allow for a constitutive expression of LRP-1-targeting shRNA (shLRP-1) or a scrambled shRNA (shCtrl). RT-qPCR and the immunoblot showed a significant decrease in LRP-1 mRNA (by 60%) and protein (by 67%) expression, respectively, in shLRP-1 MDA-MB-231 cells compared with shCtrl ( Figure 1D-F). These results validated our LRP-1 study model in MDA-MB-231 cells. As shown in Figure S1, the LRP-1 expression in MDA-MB-231 without antibiotic selection pression showed no significant difference up to 35 days, indicating that LRP-1-targeting shRNA was stable over time and compatible with in vivo experiments ( Figure S1).   (Figure 2A). To examine the in vivo functional aspects of neo-formed vascular networks within tumors, we used the Dynamic Contrast Enhancement (DCE)-MRI and Fluorescent Molecular Tomography (FMT) imaging methods. As shown in Figure 2B, the temporal changes in contrast enhancement due to the gadolinium (Clariscan ® ) concentration within tumors after an intravenous bolus injection allowed us to observe fully perfused shCtrl tumors, while shLRP-1 tumors appeared only superficially perfused for a quarter of their circumference. To keep exploring the functional aspect of the vascular network, we used a long-circulating near-infrared fluorescent blood-pool agent (AngioSense TM -750). We observed a clear heterogeneity within tumor groups that did not allow us to conclude significantly on the slighter AngioSense TM -750 signal trend in shLRP-1 tumors compared to shCtrl ( Figure 2C). However, the major population of shCtrl tumors [1] with an AngioSense TM -750 signal from 180 to 260 pmol presented a comparable signal on the tumors' edges ( Figure 2C, right panel). One of shCtrl tumors [2] stood out with a different profile and half the signal recovered (87 pmol) compared with the others. Concerning shLRP-1 tumors, we observed different profiles. From one low vascularized tumor with 38 pmol [5] to what seems to be a hyperpermeable marked profile with 269 pmol of AngioSense TM -750 signal [4]. Nevertheless, we found a major shLRP-1 population [3] with the same profile, characterized by an accumulation of AngioSense TM -750 from 111 pmol to 251 pmol in the heart of the tumor ( Figure 2C, right panel). The results obtained using a HypoxiSense TM -680 fluorescent imaging agent that detects carbonic anhydrase 9 (CA IX) tumor cell surface expression revealed a rise of hypoxia in shLRP-1 tumors compared to shCtrl (0.079 ± 0.020 vs. 0.010 ± 0.04 pmol/mm 3 ) ( Figure 2D). Both LRP-1 immunoblots and RT-qPCR realized from tumors samples confirmed a LRP-1 protein repression of more than 50% ( Figure 2E) and more than 70% at the transcriptional level ( Figure 2F) at the end of the protocol in shLRP-1 tumors compared to shCtrl. CD31 labeling followed by a microvascular density (MVD) analysis were performed and highlighted differences of vascularization, revealed by a decrease of the vessel number in the shLRP-1 tumor section (−50% ± 7%, ** p < 0.01) ( Figure 2G). In line with these observations, HES staining showed the largest necrosis areas in shLRP-1 tumors compared to shCtrl (52 ± 6% vs. 20 ± 4% of tumor area, ** p < 0.01) ( Figure 2H). The count of mitoses did not reveal any difference between the two groups ( Figure 2I). Thus, the vascular networks formed within shLRP-1 tumors presented morphological and functional differences compared to shCtrl, which were decisive for the primary tumor progression. This seems to be explained by the microenvironment's physicochemical properties modulation, especially hypoxia.

MDA-MB-231 Secretome Analysis Reveals That LRP-1 Angiogenic Effects Involved TGF-β and Plasminogen/Plasmin Pathways
To decipher the mechanisms by which LRP-1 can influence tumor progression and angiogenesis, 24 h shLRP-1 and shCtrl cells secretomes were investigated using mass spectrometry-based proteomics. Intracellular proteins, most certainly coming from exosomes, were excluded. When LRP-1 is stably repressed in the cells, many factors (whether pro-or anti-angiogenics) are modulated, as shown on the representative heatmap ( Figure 5A). Based on an in-depth analysis via the Proline software and using the GSEA and Ingenuity Pathways for pathway representation, we highlighted a preferential modulation scheme of certain pathways, such as the transforming growth factor-β (TGF-β) signaling (notably TGFβ-1, TGFβ-2, TGFβI) and the plasminogen/plasmin (PP) system (including PLG, PLAT, and a batch of SERPIN) ( Figure 5B). In addition, TIMP-1, TIMP-2, and TIMP-3 with ratios of 35.37, 3.79, and 98.13, respectively, were enriched in a shLRP-1 secretome compared to shCtrl, as well as THBS1 with a ratio of 39.17 ( Figure S3), suggesting a strong regulation of proteinase activity and anti-angiogenic effects. Pro-angiogenic molecules such as ECM1, GRN, and FST were also enriched with ratios of 77.49, 12.04, and 15.31, respectively ( Figure S3). The modulation of the PP system was confirmed by measuring plasmin activity using S-2251 TM (HD-Val-Leu-Lys (pNA)) ( Figure 6). The photometric measurements of plasmin activity demonstrated an exponential increase in plasmin activity in shCtrl MDA-MB-231 TCM, reaching an optical density at 405 nm (OD405) of 2.70 ± 0.1 after 630 min. In contrast, a slower conversion of plasminogen into plasmin was measured in shLRP-1 MDA-MB-231 TCM with an OD405 of 1.70 ± 0.02 after 630 min ( Figure 6A). The data obtained from 24 h HUVEC-conditioned media by shLRP-1 or shCtrl TCM showed more pronounced effects ( Figure 6B). Thus, HUVECs stimulated by shLRP-1 TCM exhibit a decreased plasmin activity compared to HUVECs stimulated by shCtrl, leading to a lesser propensity to migrate and invade.    Among selective known genes linked to cancer progression and/or angiogenesis, protein-protein interactions were mapped using Ingenuity Pathways Analysis. TGF-β signaling (governed by TGF-β1, on the left) and the plasminogen/plasmin system (represented by PLG/PLAT, on the right) stand out for their privileged place within these multiple interactions organized around LRP-1.
Colored heatmap generated from proteomics analysis data using the ggplot2 R package reflecting LRP-1′s influence in 24 h shLRP-1 and shCtrl MDA-MB-231 TCM. Comparison of proteomics profiles between shLRP-1 and shCtrl triplicate. Logarithmic scale of fold change from 1.5 to -1.5. (B) Representative pathway of LRP-1 modulations in MDA-MB-231 secretome. Among selective known genes linked to cancer progression and/or angiogenesis, protein-protein interactions were mapped using Ingenuity Pathways Analysis. TGF-β signaling (governed by TGF-β1, on the left) and the plasminogen/plasmin system (represented by PLG/PLAT, on the right) stand out for their privileged place within these multiple interactions organized around LRP-1.

Discussion
This study aimed to clarify LRP-1′s role in TNBC tumor growth, and more precisely its involvement in tumor angiogenesis using an MDA-MB-231 cell line-based model. Previous data from our group and others have shown that the expression of LRP-1 and LDLR was higher in mammary tumor tissues [31,32], contributing to LDL-C uptake from the blood [33] and a poor prognosis [34]. In particular, LRP-1 was shown to be involved in the invasiveness of luminal and TNBC subtypes of BC [18,35,36]. In recent years, LRP-1 was shown to be involved in angiogenesis, notably by regulating LRP1-dependent signaling pathways in different endothelial processes, such as proliferation, migration, permeability, and tube formation [37,38]. LRP-1 also plays an essential role in vascular homeostasis, by having a protective role in atherosclerosis pathogenesis and aneurysm formation [39]. Furthermore, some studies have revealed the angio-modulatory capacities of LRP family members in various solid tumors, including BC [40][41][42]. A PDX analysis comparing the LRP-1 RNA expression of TNBC versus non-TNBC showed no significant results, in line with the searched databases. This could be a consequence of the inherent heterogeneity of this aggressive subtype [43]. However, 3/4 of TNBC PDXs we had access to have a higher expression than the average non-TNBC PDXs. Therefore, the study of the role of LRP-1 appears to be relevant for a majority of TNBC. Moreover, a more accurate TNBC subtyping of the PDXs-such as a basal-like or non-basal-like distinguo-could show potential

Discussion
This study aimed to clarify LRP-1's role in TNBC tumor growth, and more precisely its involvement in tumor angiogenesis using an MDA-MB-231 cell line-based model. Previous data from our group and others have shown that the expression of LRP-1 and LDLR was higher in mammary tumor tissues [31,32], contributing to LDL-C uptake from the blood [33] and a poor prognosis [34]. In particular, LRP-1 was shown to be involved in the invasiveness of luminal and TNBC subtypes of BC [18,35,36]. In recent years, LRP-1 was shown to be involved in angiogenesis, notably by regulating LRP1-dependent signaling pathways in different endothelial processes, such as proliferation, migration, permeability, and tube formation [37,38]. LRP-1 also plays an essential role in vascular homeostasis, by having a protective role in atherosclerosis pathogenesis and aneurysm formation [39]. Furthermore, some studies have revealed the angio-modulatory capacities of LRP family members in various solid tumors, including BC [40][41][42]. A PDX analysis comparing the LRP-1 RNA expression of TNBC versus non-TNBC showed no significant results, in line with the searched databases. This could be a consequence of the inherent heterogeneity of this aggressive subtype [43]. However, 3/4 of TNBC PDXs we had access to have a higher expression than the average non-TNBC PDXs. Therefore, the study of the role of LRP-1 appears to be relevant for a majority of TNBC. Moreover, a more accurate TNBC subtyping of the PDXs-such as a basal-like or non-basal-like distinguo-could show potential correlations with LRP-1 expression.
Here, we showed that LRP-1 plays a more decisive role, not only by contributing to cell survival and proliferation [44]; it modulates (directly or indirectly) the angiogenic balance through its pivotal roles within the tumor microenvironment. We showed that LRP-1 repression in MDA-MB-231 tumors led to a significant tumor growth decrease (64%) compared to the control group. The lower proliferative capacities of shLRP-1 cells observed in vitro (15-20%, data not shown) are not sufficient to explain such a difference in tumor volume. Otherwise, no significant difference in the mitotic index in the viable parts of the tumors was found.
As angiogenesis is required for tumor progression and growth [11], DCE-MRI experiments were conducted to assess tumor perfusion and enable the depiction of physiological alterations as well as morphological changes [45]. shLRP-1 tumors characterized by a decreased tumor perfusion in vivo exhibited numerous unsuccessful structures, displaying a CD31 signal but without lumen, suggesting that the stimulation of angiogenesis was present and sustained but unable to reach shCtrl vascular achievement. The in vivo vascular density evaluation in FMT confronted us with intra-tumor heterogeneity. Two major distinct populations were found according to the signal distribution-either peripheral tumors, in shCtrl, or central, in shLRP-1 tumors. An accumulation of fluorochrome in the peritumoral tissue is thought to be due to highly leaky vessels or a potential hemorrhage within tumors [46]. Certain CD31-stained shLRP-1 tumor sections exhibited large structures resembling hemorrhagic lakes rather than vessels, but anastomoses were also observed, highlighting a marked vascular anarchy when LRP-1 is repressed in MDA-MB-231. shLRP-1 tumors showed a significant increase in necrosis compared to shCtrl, as a direct result of the increased hypoxia. As LRP-1 is known to be upregulated by hypoxia [47], we ascertained that its expression was still low enough in our in vivo tumor model at the protocol end. As a common phenomenon in most malignant tumors, hypoxia leads to an advanced but dysfunctional vascularization, by inducing an imbalance between proand anti-angiogenic factor production, thus leading to a rapid and chaotic blood vessel formation increase [48].
By focusing on in vivo and in ovo angiogenic assays, we highlighted the LRP-1silenced cells' difficulties in supporting angiogenesis. The in vivo and ex vivo vascular densities found in MPs were indeed lower when LRP-1 expression was repressed, which is consistent with the decreased vessel numbers in CD31-stained MPs sections. In line with these data, blood perfusion appeared less efficient in shLRP-1 MPs. As for the CAMs assay, it demonstrates that the vascular networks generated by shLRP-1 cells exhibited a lesser overall length and a lower number of branchings. These results corroborate that LRP-1 plays a significant role in the outcome of angiogenic processes in MDA-MB-231 tumor cells.
In vitro, LRP-1 influences the tumor cells' secretome which shapes EC behaviors among the microenvironment cells. We showed that the shLRP-1 MDA-MB-231 cells' secretome decreases the angiogenic potential of HUVECs by impacting their ability to form a 3D-tubular network on Matrigel ® and, unsurprisingly, their migratory capacities. However, we found that shLRP-1 TCM led to a higher EC proliferative rate over time than shCtrl. The overall growth of a vasculature is the result of both proliferation and migration controlled by a myriad of factors in the tumor microenvironment, including many pro-and anti-angiogenic factors [49]. In a computational model, the authors have modulated proliferation and migration rates separately. They have demonstrated that an EC proliferation increase at the expense of migration leads to an increase in sprouts, which then mostly exhibit anastomoses preventing vessel functionality [50].
Through a proteomic approach, we demonstrated the extensive LRP-1's influence on the MDA-MB-231 tumor cells' secretome, where 962 proteins were identified. When it comes to identifying by which precise molecular pathways LRP-1 plays its part on tumor progression and angiogenesis, the task is intricate. We highlighted a solid modulation of TGF-β signaling as well as a modulation of the plasminogen/plasmin (PP) system. Under physiological conditions and in early stages of carcinogenesis, TGF-β acts as a tumor suppressor by restricting cell growth and stimulating apoptosis to maintain homeostasis in the tissues. However, in advanced tumors, cancer cells escape TGF-β's initial suppressive effects and use its regulatory functions to promote their progression with clear roles in processes supporting cancer cell invasion, epithelial-mesenchymal transition (EMT), immune response suppression, angiogenesis, and metastasis [51]. In addition, TGF-β contributes to matrix remodeling by increasing the expression of MMPs [52] and plasmin, creating a permissive environment allowing cancer cells to metastasize [53]. Through endogenous TGF-β1 activation, it has been shown that thrombospondin-1 (THBS1) up-regulates the PP system and promotes tumor cell invasion in MDA-MB-231 cells [54]. THBS1, overexpressed in LRP-1-repressed MDA-MB-231, is established as an anti-angiogenic and anti-tumoral protein [55]. Notably, THBS1 binds to free and cell-associated VEGF [56], and THBS1/VEGF complexes are internalized via LRP-1 [57], suggesting that LRP-1 contributes to VEGF bioavailability during neovascularization. No clear modulation of VEGF by LRP-1 could be demonstrated, as no significant difference in VEGF-immunostained tumor sections was measured. However, an increase in VEGF transcriptional expression in tumors has been shown, certainly in response to hypoxia, because this increase was not measured in vitro (data not shown). In THBS1 up-regulated cells, the secreted VEGF could be sequestrated, and is thus not sufficient for the cells to ensure a proper VEGF-stimulated angiogenesis. As THBS1 regulates vessel stabilization, its overexpression has been shown to suppress vascular growth and expand vessel diameter [58], suggesting that it could be associated with dysfunctional angiogenesis, like in Fabry disease [59]. Despite an increased plasminogen expression and one of its activators in shLRP-1 TCM, a decreased plasmin activity was measured. The explanation appears more sophisticated than the unavailability of plasminogen or its activators, suggesting the involvement of system inhibitors such as SERPINE1/2 (PAI-1/2) or SERPINC1 (antithrombin-III), able to thwart the enzymatic cascade [60]. Angiogenesis is associated with an important extracellular remodeling involving different proteolytic systems, among which the PP system plays an essential role. EC migration is associated with significant proteolysis upregulation, and, conversely, PP system inhibition reduces angiogenesis in vitro [61]. Thus, the prevention of in vitro HUVECs' tubular structure formation in shLRP-1 TCM is consistent with the decreased plasmin activity in HUVECs CM after shLRP-1 TCM stimulation, given that pseudotube formation is based on ECs' proteolytic activity and migratory capacities generated in response to their environment. However, genetically altered mice for the PP system developed without overt vascular anomalies, indicating a possible compensation by other proteases in vivo [61]. Furthermore, SERPINF1, expressed five times more in shLRP-1 TCM, has been described as an inhibitor of hypoxia-induced angiogenesis by either directly targeting HIF-1 or regulating HIF-1's target genes signaling cascades, thus blocking EC survival, proliferation, and migration or leading to their apoptosis [62].
Although we have previously shown that shLRP-1 cells revealed an increased cell rigidity in vitro, with the drop in membrane extension dynamics directly reflecting their altered migratory capacities [19], these results could be divergent in vivo. When we set an experimental configuration that mimics the in vivo environment or approaches it, whether it is a CAMs assay or the formation of 3D spheroids, shLRP-1 cells grafts or spheroids exhibit a more invasive profile than expected compared to shCtrl ( Figure S4). As hypoxia contributes to TGB-β up-regulation and EMT phenotype acquisition, resulting in cell mobility and metastasis, it could be the trigger of invasiveness in vivo. Moreover, a long exposure to hypoxia is associated with DNA breaks and a high frequency of replication errors, potentially leading to genetic instability and mutagenesis [63], and increasing the metastatic potential. A hypoxic environment, unfavorable to cell proliferation and survival, participates in the selection of cell clones that have acquired insensitivity to oxygen and nutrient deprivation [48]. In particular, MDA-MB-231 cells have been shown to secrete heat shock protein 90 alpha (eHsp90α) to mediate their survival under hypoxia [64]. The integration of such survival signals, leading to the epithelial-to-mesenchymal transition and migration in breast cancer cells, is dependent on the LRP-1 receptor [65]. Although the expression of Hsp90α was not identified in our analysis, it should nonetheless be excluded from future investigations, given its direct link with LRP-1 and the potential for its inhibition in TNBC [66]. As a mecanosensor of the tumor microenvironment, LRP-1 temporal expression during tumorigenesis could modulate the sensitivity of cells in response to stresses such as hypoxia. Thus, the question of whether LRP-1-repressed cells, less proliferative, with lower migratory properties in vitro, and forming primary tumors of smaller sizes in vivo, could surpass shCtrl MDA-MB-231 cells' aggressiveness in the late tumorigenesis stages due to the hypoxia rise and a permissive signaling such as TGF-β is more than relevant and will be addressed later.

Conclusions
In the present study, we showed that LRP-1 emerges as an important matricellular player in the control of cancer-signaling events such as angiogenesis, by supporting tumor vascular organization in a way that appears dispensable but that is ultimately essential for the vascular effectiveness for tumor growth.

Informed Consent Statement:
Tumor fragments used to generate PDXs were collected from patient upon Informed consent signature from all subjects involved in the study.

Data Availability Statement:
The proteomics analysis revealed that LRP-1 supports tumor growth and angiogenesis through TGF-β signaling and the plasminogen/plasmin system. Concerning these last analyses, mass spectrometry proteomics data have been deposited into the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org (accessed on 15 August 2021)) via the PRIDE partner repository with the dataset identifier PXD022978.