mTOR Repression in Response to Amino Acid Starvation Promotes ECM Degradation Through MT1‐MMP Endocytosis Arrest

Abstract Under conditions of starvation, normal and tumor epithelial cells can rewire their metabolism toward the consumption of extracellular proteins, including extracellular matrix‐derived components as nutrient sources. The mechanism of pericellular matrix degradation by starved cells has been largely overlooked. Here it is shown that matrix degradation by breast and pancreatic tumor cells and patient‐derived xenograft explants increases by one order of magnitude upon amino acid and growth factor deprivation. In addition, it is found that collagenolysis requires the invadopodia components, TKS5, and the transmembrane metalloproteinase, MT1‐MMP, which are key to the tumor invasion program. Increased collagenolysis is controlled by mTOR repression upon nutrient depletion or pharmacological inhibition by rapamycin. The results reveal that starvation hampers clathrin‐mediated endocytosis, resulting in MT1‐MMP accumulation in arrested clathrin‐coated pits. The study uncovers a new mechanism whereby mTOR repression in starved cells leads to the repurposing of abundant plasma membrane clathrin‐coated pits into robust ECM‐degradative assemblies.


Supplementary Experimental Section
Cell culture, transfection and siRNA treatment. Human MDA-MB-231 breast adenocarcinoma cells obtained from ATCC (ATCC HTB-26) were grown in L-15 medium (Sigma-Aldrich) supplemented with 15% fetal calf serum (FCS) and 2 mM Gln (ThermoFisher Scientific) at 37°C in 1% CO 2 . The human pancreas adenocarcinoma cell line Bx-PC3 obtained from ATCC (ATCC CRL-1687) was grown in RPMI-1640 medium (ThermoFisher Scientific) supplemented with 10% fetal calf serum at 37°C in 5% CO 2 . Both cell lines were routinely tested for mycoplasma contamination. MDA-MB-231 cells stably expressing TKS5 GFP or MT1-MMP pHLuorin were generated by lentiviral transduction. [1] For transient expression, MDA-MB-231 cells were transfected with the plasmid constructs using AMAXA nucleofection (Lonza) and analyzed by live cell imaging 48 h after transfection. For starvation experiments, cells were cultured in EBSS medium (ThermoFisher Scientific) supplemented with MEM Vitamins (Gibco, composition in Table S1, Supporting Information) at 37°C in 5% CO 2 . For siRNA treatment, except for α-adaptin and clathrin heavy chain (CHC), we used SMARTpool reagents consisting of a mix of four individual siRNAs used at lower concentration in order to reduce potential off-target effects (see Table S2, Supporting Information, for a list of siRNAs used for this study).
Cells were treated with the indicated SMARTpool mix (50 nM final concentration) using Lullaby (OZ Biosciences) according to manufacturer instructions and analyzed after 72 hrs of treatment. For silencing of α-adaptin and CHC, cells were treated twice with the siRNA (50 nM final concentration) at 48 hrs interval and analyzed 120 hrs after initial treatment.
Antibodies and drugs. The source of commercial antibodies used for this study are listed in Table S3 (Supporting Information). The source and working concentration of drugs used in this study are listed in Table S4 (Supporting Information).
Polymerization of type I collagen gel. A type I collagen polymerization mix was prepared on ice by adding 25 µM HEPES (final concentration) to a 2.2 mg mL -1 acidic-extracted type I collagen solution (Corning) and pH was adjusted to 7.5 with 0.34 N NaOH. When required for microscopic visualization of the collagen network, 2 to 5% of a ~2 mg mL -1 solution of AlexaFluor 647-conjugated type I collagen was added to unlabeled collagen in the polymerization mix. When required, drugs were added to the appropriate final concentration in the polymerization mix (see Table S4, Supporting Information). Polymerization was started by incubation at 37°C in a humidified chamber (CO2 cell incubator).
Antibodies were detected using the Enhanced Chemiluminescence reagent (ECL, Amersham RPN2232) on the ChemiDoc MP Imaging System (Bio-Rad).
Quantification of pericellular collagenolysis. To measure pericellular collagenolysis on a thin layer of type I collagen gel, a 18-mm diameter glass coverslip was layered with 200 µl of the ice-cold 2.2 mg mL -1 AlexaFluor 647(AF 647 )-labeled type I collagen polymerization mix as described above. Excess collagen solution was removed by pipette aspiration to leave a thin smear of collagen solution on the glass coverslip. After 3 min of polymerization at 37°C, the collagen gel was gently washed in PBS and 7x10 4 cells were added and incubated for 1 at 37°C in CM or EBSS 6 medium in the presence or in the absence of AA supplements or drugs as indicated.
Cells were pre-extracted with 0.1% Triton X-100 in 4% PFA in PBS for 90 sec at 37°C and fixed in 4% PFA in PBS for 20 min at 37°C. Coverslips were treated with 1% BSA in PBS for 30 min at room temperature then incubated with Col1-¾C and anti-cortactin antibodies diluted in 1% BSA in PBS for 2 hrs at 4°C. After three washes with PBS at 4°C, samples were counterstained with Cy3-conjugated antirabbit IgG and A488-conjugated anti-mouse IgG antibodies for 60 min at 4°C, extensively washed in PBS and mounted in Prolong-DAPI mounting medium (Invitrogen). Images were acquired with a wide-field microscope (Eclipse 90i Upright; Nikon) using a 100x Plan Apo VC 1.4 oil objective and a cooled interlined chargecoupled device (CCD) camera (CoolSnap HQ 2 ; Roper Scientific). A z-dimension series of images was taken every 0.2 µm by means of a piezoelectric motor (Physik Instrumente). The system was steered by Metamorph software. Deconvolution was processed by Nikon NIS-Elements software (3D-deconvolution module; Lucy-Richardson algorithm).
For quantification of pericellular collagenolysis in a 3D collagen network, 40 µl of a 6x10 4 cells/mL cell suspension in the 2.2 mg mL -1 type I collagen polymerization mix was added on top of a 12-mm diameter glass coverslip and polymerization was performed for 30 minutes at 37°C. The indicated culture medium was added and samples were incubated for 6 hrs at 37°C. Samples were fixed, permeabilized and stained with Col1-¾C antibody as described above except that samples were counterstained with Phalloidin-Alexa488 to visualize cell shape. Image acquisition was performed with an A1R Nikon confocal microscope with a 40x NA 1.3 oil objective using high 455 sensitivity GaASP PMT detector and a 595 +/-50 nm bandpass filter. Quantification of Col1-¾C signal (cleaved collagen) was performed with a homemade ImageJ macro. Acquired z-planes were projected using maximal intensity projection in Fiji and Col1-¾C signal was determined using the thresholding command excluding regions <50-px to avoid non-specific signal. Col1-¾C signal area was normalized to the total cell surface (thin layer) or to the number of nuclei in field (3D network) and values normalized to control cells.
Fluorescent gelatin degradation assay. MDA-MB-231 cells were plated for 1 to 5 hrs on Oregon Green 488 (OR 488 ) or AF 594 -conjugated cross-linked gelatin (Invitrogen) in EBSS or CM medium in the presence or absence of rapamycin as previously described. [2] Cells were pre-extracted with 0.1% Triton X-100 in 4% PFA in PBS for 90 sec at 37°C and fixed in 4% PFA in PBS for 20 min at 37°C and then stained with the indicated antibodies or with fluorescently-labeled phalloidin to stain F-actin. Cells were imaged with a 100x objective on a wide-field microscope equipped with a piezoelectric motor as above. For quantification of degradation, the area of degraded matrix (black pixels) measured with the threshold command of ImageJ was divided by the total cell surface and values were normalized to control cells. The regions of interest delimiting the gelatin degradation were saved for further analysis, such as the assessment of AP-2 association (see below). Linescans were performed using Fiji software. Deconvolution was processed by Nikon NIS-Elements software (3D-deconvolution module; Lucy-Richardson algorithm). After fixation and permeabilization, cells were stained with α-adaptin as described above. CCPs in the entire cell were detected using the Find Maxima command of ImageJ and the number of detected CCPs was divided by the area of the cell. CCPs positions were saved for further analysis (see below).

Randomization of AP-2 distribution over gelatin degradation spots.
To measure the association of α-adaptin positive CCPs with gelatin degradation spots, CCPs and degradation spots were detected as described above and their positions as well as the position of all pixels inside the cell (total pixels) defined by their X and Y coordinates were saved. For each CCP, (X, Y) positions were randomly drawn from all pixels of the cell, effectively changing the position of CCPs inside the cell in a random fashion (see Figure S3B, Supporting Information). This randomization procedure was performed 5,000 times per cell and the number of CCPs associated with gelatin degradation was measured each time. The true value of CCP association with gelatin degradation was calculated and compared to the randomized values.
Synthetic images displaying cell contour (white line), degradation spots (black) and associated CCPs (red crosses) were generated with ImageJ. This procedure was repeated for ten independent cells with p-values ranging from 0 to 0.0142 (mean pvalue = 0.001).
Tfn uptake assay. 7x10 4 MDA-MB-231 cells plated on a 18-mm diameter glass coverslip were incubated overnight at 37°C in CM. Cells were washed twice with PBS before incubation in EBSS or CM medium for 1 hr at 37°C, then transferred on ice and washed twice with ice-cold EBSS or L15 medium supplemented with 1% BSA and 20 mM HEPES pH 7.5. Coverslips were incubated with 20 µg mL -1 of AF 546conjugated Tfn (ThermoFisher) in the same medium for 1 hr at 4°C. Cells were fixed with 4% PFA in PBS or incubated in pre-heated CM or EBSS for 2, 5 or 10 min at 37°C before fixation. After permeabilization with 0.1% Triton X-100 in PBS for 15 min, samples were incubated with anti-α-adaptin (overnight at 4°C) or with anti-EEA1 antibodies (1 hr at room temperature), and then counterstained with AF488conjugated anti-mouse antibodies (1 hr at room temperature). Stacks of images were acquired with a wide-field microscope (Eclipse 90i Upright; Nikon) steered by Metamorph software as described above. For analysis, the plane corresponding to the plasma membrane was selected. CCPs positive for α-adaptin in a selected region were detected and segmented using the manual threshold command of ImageJ. The regions of interest (ROI) were saved and copied on the Tfn image. The mean intensity of Tfn inside each ROI was measured and a frequency histogram was generated with a normalization to T0.
Quantification of LC3-positive puncta. 7x10 4 MDA-MB-231cells were plated on collagen-coated or on non-coated 18-mm diameter glass coverslips as previously described and incubated for 4 at 37°C in CM or in EBSS medium. Cells were fixed with 4% PFA in PBS for 10 min and permeabilized with 0.05% saponin (Sigma-Aldrich) in PBS for 10 min. Samples were blocked in PBS with 0.05% saponin and 5% FCS for 30 min at room temperature and stained with anti-LC3 and anti-p4E-BP1 antibodies for 2 hrs at room temperature. After three washes, samples were counterstained with Cy3-conjugated anti-mouse IgG and Alexa488-conjugated antirabbit IgG antibodies and mounted in Prolong-DAPI medium. Image acquisition was performed by wide-field microscopy as previously described. Quantification of LC3positive vesicles was performed by maximal orthogonal projection of the series of optical sections (the distance between two sections is 0.2 µm). Cells were manually delimited using the p4E-BP1 signal while LC3 signal was denoised and thresholded to detect LC3-positive autophagic vesicles. Detected spots were counted and saved for visual verification. No manual correction was done. The average number of LC3positive puncta per cell was normalized to the value in CM-treated cells set to 1.

Dynamics of TKS5-and µ-adaptin-positive structures by live cell total internal reflection fluorescence microscopy (TIRF-M). MDA-MB-231 cells transfected with
GFP-tagged TKS5 and mCherry-tagged µ-adaptin were plated in CM or EBSS on glass bottom dishes (Ibidi Corporation) layered with unlabeled cross-linked gelatin as previously described. Simultaneous dual color TIRF-M sequences were acquired with an inverted microscope (Eclipse-Ti-E, Nikon) equipped with a 100x PlanApo TIRF objective (1.47 NA), a TIRF arm, an image splitter (DV; Roper Scientific) installed in front of the EMCCD camera (Photometrics) and a temperature controller. GFP and m-Cherry were excited with 491-and 561-nm lasers, respectively (50 mW, Gataca Systems) and fluorescent emissions were selected with bandpass and longpass filters (Chroma Technology Corp). The system was driven by Metamorph. For quantification of CCP dynamics, CCP lifetime was measured using the TrackMate plugin of FIJI. [3] At least 300 CCPs from at least 6 cells per condition and per experiment were tracked from three independent experiments. Data are expressed as mean lifetime ± sem.
Ex-vivo culture of TNBC patient-derived xenografts. Breast cancer patient derived xenografts were obtained from triple-negative breast tumors and generated as described. [4] After surgical excision of the tumor xenograft, tumor cells were dissociated in DMEM/F12 medium supplemented with collagenase and hyaluronidase (SIGMA-Aldrich, 1X final) in 10 mM HEPES, 7.5% BSA Fraction V (Gibco), 5 µg mL -1 insulin (SIGMA-Aldrich) and 50 µg mL -1 gentamycin (GIBCO) for 1 hr at 37°C on a rotating wheel at 180 rpm as previously described. [5] Samples were     Bound proteins were analyzed with MT1-MMP and -adaptin antibodies Equal loading was controlled using GAPDH antibody (not shown).

Supplementary Movie Legends
Movie S1: TKS5 localizes to highly dynamic matrix fiber-remodeling elongated invadopodia in cells grown in nutrient-replete conditions. Shown is a MDA-MB-231 cell expressing TKS5 GFP (green). The cell is plated on a layer of fibrillar collagen (magenta) and is grown in nutrient-replete conditions. Images were acquired every 1 min for 60 min. The last sequence of the movie are still images of the time projection