Abstract
Metastasis, a fatal complication of breast cancer, does not fully benefit from available therapies. In this study, we investigated whether ATIP3, the major product of 8p22 MTUS1 gene, may be a novel biomarker and therapeutic target for metastatic breast tumors. We show that ATIP3 is a prognostic marker for overall survival among patients with breast cancer. Notably, among metastatic tumors, low ATIP3 levels associate with decreased survival of the patients. By using a well-defined experimental mouse model of cancer metastasis, we show that ATIP3 expression delays the time-course of metastatic progression and limits the number and size of metastases in vivo. In functional studies, ATIP3 silencing increases breast cancer cell migration, whereas ATIP3 expression significantly reduces cell motility and directionality. We report here that ATIP3 is a potent microtubule-stabilizing protein whose depletion increases microtubule dynamics. Our data support the notion that by decreasing microtubule dynamics, ATIP3 controls the ability of microtubule tips to reach the cell cortex during migration, a mechanism that may account for reduced cancer cell motility and metastasis. Of interest, we identify a functional ATIP3 domain that associates with microtubules and recapitulates the effects of ATIP3 on microtubule dynamics, cell proliferation, and migration. Our study is a major step toward the development of new personalized treatments against metastatic breast tumors that have lost ATIP3 expression. Cancer Res; 73(9); 2905–15. ©2013 AACR.
Introduction
The occurrence of distant metastasis is a dreadful complication of breast cancer and a leading cause of death by malignancy in women worldwide. Metastasis is a multistep process that involves cancer cell migration and invasion across the extracellular matrix to reach the blood flow, followed by extravasation and colonization of secondary organs (1). Among millions of invasive cancer cells that reach the blood circulation, only few will establish at distant sites and grow as metastases (2–5). Breast cancer metastases can remain latent for several years following primary tumor removal, and the identification of molecular markers that may predict the risk of metastasis occurrence, and/or progression is of invaluable help for the follow-up of the patients and choice of therapeutic options (5, 6). Over the past decade, extensive studies have led to the classification of breast tumors into distinct molecular subtypes, allowing subsequent development of efficient targeted treatments for a majority of primary tumors (7–9). However, available therapies have limited effect on cancer metastasis and new genetic determinants contributing to essential steps of the metastatic process need to be characterized.
Microtubule-targeting drugs such as taxanes are used for standard first-line treatment of breast cancer metastasis, and new microtubule-targeting agents, such as epothilones and eribulin, are under clinical evaluation (10). Microtubules are polarized and highly dynamic structures that rapidly switch between periods of polymerization (growth) and depolymerization (shrinkage) at the plus ends, a process termed dynamic instability (11–13). The extent and rate of microtubule growth, as well as transitions between growth and shrinkage, are parameters of dynamic instability that can be measured by tracking end-binding proteins at the microtubule plus ends (13–15). Dynamic instability is essential for the microtubule plus ends to explore the cytosol and ensure cytoskeleton reorganization during cell division and migration. Targeting the expression or activity of metastasis genes that regulate microtubule dynamics represents a promising option for cancer therapy.
ATIP3 is a microtubule-associated protein encoded by 8p22 candidate tumor suppressor gene MTUS1 (16–18). We have previously shown that ATIP3/MTUS1 levels are significantly downregulated in 47.7% of invasive breast carcinomas and 62.4% of metastatic tumors (19). Restoring ATIP3 expression at normal levels in breast cancer cells significantly reduces cancer cell proliferation in vitro and tumor growth in vivo (19). However, effects of ATIP3 on breast cancer metastasis have not yet been evaluated.
In this study, we investigated whether ATIP3 may represent a new biomarker and therapeutic target for breast cancer metastasis. We present evidence that low ATIP3 levels correlate with the decreased probability of survival among patients with breast cancer metastasis, and that ATIP3 expression into ATIP3-deficient cancer cells markedly impairs the establishment of metastatic foci in vivo. Loss of ATIP3 increases breast cancer cell migration and alters microtubule dynamics. We show that ATIP3 associates with microtubules through a central basic domain that retains the functional properties of the full-length protein. Our study thus identifies ATIP3 as a new promising therapeutic target against metastatic breast tumors of poor prognosis.
Materials and Methods
Breast tumor samples and gene arrays
Microarray data for a series of 150 infiltrating ductal primary breast carcinomas and 11 normal breast tissues from the Institut Curie (Paris, France) and clinical data for the patients were described elsewhere (19, 20). Gene expression profiles from an independent cohort of 162 invasive breast carcinomas were obtained from patients included in the prospective database of the Institut Gustave Roussy (IGR; Villejuif, France) between 1984 and 1994. This study was approved by the Institutional Review Boards of the IGR. Data have been submitted to the Array Express data repository at the European Bioinformatics Institute (Saffron Walden, United Kingdon; http://www.ebi.ac.uk/arrayexpress/) under accession number E-MTAB-1389. MTUS1 gene expression in a meta-analysis of 2,898 patients with breast cancer with known clinical outcome was retrieved from Kaplan–Meier plotter database (21, 22).
Cell lines, plasmid constructs, and transfections
Human breast cancer cell lines MDA-MB-468 and MCF7 and stable clones were described previously (19). MDA-MB-231-Luc-D3H2LN breast cancer cells (designated D3H2LN) obtained from Caliper Life Science (Xenogen) were derived from an in vivo-selected metastatic subclone of MDA-MB-231 cells expressing luciferase and grown as described (23). HeLa cells were provided by Dr. Mounira Amor-Gueret (Institut Curie, Orsay, France). RPE-1 [human telomerase reverse transcriptase (h-TERT)-immortalized, retinal pigment epithelial] cells were from Dr. Franck Perez (Institut Curie, Paris). MRC5-SV lung fibroblasts were grown in Dr. A. Akhmanova's laboratory as described (24). All cells were used at passages 2 to 20 after thawing and grown as described by the provider. Cells were routinely authenticated by morphologic observation and tested for absence of mycoplasma contamination using MycoAlert Assay detection kit (Lonza).
Plasmid constructs are described in the Supplementary Methods. Transfections using ATIP3-specific siRNAs (si#1 and si#2) were conducted as described (19) and verified by immunoblotting using anti-MTUS1 polyclonal antibodies (ARP-44419, Aviva Systems).
Intracardiac experimental mouse model of metastasis
Experimental metastasis was conducted as described (23, 25, 26) following intracardiac injection of stable ATIP3-negative [WT (wild-type), GFP] or positive (Cl3, Cl6) D3H2LN cell clones. All injected cells showed similar viability as measured by Annexin V apoptosis kit (Beckman Coulter). The experiment was carried out with the approval of the Département d'Expérimentation Animale, Institut d'Hématologie, Hopital St-Louis ethical committee, and was conducted twice (9 mice per group).
Clonogenicity, cell migration, and adhesion assays
Analyses of colony formation, Boyden chambers chemotaxis, transendothelial migration, wound healing, and cell adhesion were conducted as described (23). Time-lapse videomicroscopy analyses of cell motility are described in the Supplementary Methods. For cell polarity measurements, transiently transfected D3H2LN were allowed to migrate for 90 minutes and analyzed using bright field microscopy. Polarized cells were identified on the basis of nucleus position and cytoplasm extension at the leading edge.
Immunostaining, fluorescence microscopy, analysis of microtubule dynamics
Cells were plated on glass coverslips and transfected for 24 hours (plasmids) or 72 hours (siRNA), fixed in ice-cold methanol for 5 minutes, and incubated for 1 hour at room temperature with anti-α-tubulin clone F2C (27), monoclonal anti-γ-tubulin (Sigma), anti-end-binding 1 (EB1; clone 5; BD Biosciences), or anti-acetylated tubulin (clone 6-11B-1; Sigma). Secondary antibodies and fluorescence images capture are described in the Supplementary Methods.
Linescan analyses of α-tubulin and EB1 fluorescence intensity were done (ImageJ) on a 6 μm line along the length of microtubule tip. At least 10 microtubules per cell in 4 separate cells were measured. EB1-comet maximal intensity was obtained by subtracting the intensity value of the EB1-dot (100 a.u.) to the maximal staining intensity.
Analyses of microtubule stability, regrowth, and microtubule dynamic instability are described in the Supplementary Methods.
Statistical analysis
Statistical analyses were done using JMP-7 and GraphPad Prism softwares. Overall survival (OS) curves were plotted according to the method of Kaplan–Meier and compared by the log-rank test. Data in bar graphs (mean +/− SD) were analyzed using 2-tail unpaired Student t test. Dot plot analyses were done using Mann–Whitney test. P < 0.05 was considered statistically significant.
Results
ATIP3 is a prognostic marker of poor outcome in metastatic breast cancer
The prognostic value of ATIP3 as a marker for metastatic progression and OS was evaluated in 3 independent cohorts of patients with breast cancer. Comparison of MTUS1 Affymetrix probeset intensities with clinicopathologic data of the patients in a panel of 150 invasive breast carcinomas (Supplementary Table S1) showed that the overall probability of survival is strongly reduced in patients with tumors expressing low as compared with normal ATIP3 transcript levels (Fig. 1A and Supplementary Fig. S1A). Relapse-free survival (RFS) of the patients was also significantly reduced in low ATIP3-expressing tumors (Supplementary Fig. S1B). Similar results were obtained by analyzing MTUS1 levels in an independent cohort of 162 patients with breast cancer (Fig. 1B and Supplementary Table S2) and in a meta-analysis of 2,898 patients with breast cancer (Fig. 1C and Supplementary Fig. S1C and S1D). Of note, correlation between ATIP3 expression and OS of the patients was independent of the estrogen receptor (ER) status of the tumor (Fig. 1D).
Low levels of ATIP3 predict poor outcome among metastatic tumors. A, OS curves for patients from the Institut Curie cohort, with tumors expressing normal (> 0.5) or low (< 0.3) ATIP3 levels, relative to the median value of MTUS1 probeset (212096_s_at) in normal tissues. B, OS curves for patients from the IGR cohort, with tumors expressing normal or low ATIP3 levels [inferior or superior to the median value of MTUS1 probeset (A_23_P347169) intensities in the 162 tumors analyzed]. C, OS curves for patients with tumors expressing normal to high (gray) or low (black) MTUS1 (212096_s_at) using Kaplan–Meier plotter. The best performing threshold was used as a cutoff. D, OS curves from patients expressing normal or low ATIP3 as in A, among ER negative (ER−, left) and positive (ER+, right) tumors. E, percentage of patients remaining alive after 5 years with nonmetastatic (−) and metastastic (+) tumors expressing ATIP3 levels as in A. F, OS curves for patients with nonmetastatic (left) or metastatic (right) tumors expressing ATIP3 levels as in A. G, OS time (in months) for patients with metastatic tumors expressing ATIP3 levels as in A. Median values are on the right. *, P = 0.0119. H, OS curves as in F for patients from the IGR cohort. Nonmetastatic (left) and metastatic (right) tumors were classified according to ATIP3 levels as in B. I, OS time (in months) for patients as in G with metastatic tumors classified as in H. Number of patients is under brackets. *, P = 0.0172.
Low levels of ATIP3 predict poor outcome among metastatic tumors. A, OS curves for patients from the Institut Curie cohort, with tumors expressing normal (> 0.5) or low (< 0.3) ATIP3 levels, relative to the median value of MTUS1 probeset (212096_s_at) in normal tissues. B, OS curves for patients from the IGR cohort, with tumors expressing normal or low ATIP3 levels [inferior or superior to the median value of MTUS1 probeset (A_23_P347169) intensities in the 162 tumors analyzed]. C, OS curves for patients with tumors expressing normal to high (gray) or low (black) MTUS1 (212096_s_at) using Kaplan–Meier plotter. The best performing threshold was used as a cutoff. D, OS curves from patients expressing normal or low ATIP3 as in A, among ER negative (ER−, left) and positive (ER+, right) tumors. E, percentage of patients remaining alive after 5 years with nonmetastatic (−) and metastastic (+) tumors expressing ATIP3 levels as in A. F, OS curves for patients with nonmetastatic (left) or metastatic (right) tumors expressing ATIP3 levels as in A. G, OS time (in months) for patients with metastatic tumors expressing ATIP3 levels as in A. Median values are on the right. *, P = 0.0119. H, OS curves as in F for patients from the IGR cohort. Nonmetastatic (left) and metastatic (right) tumors were classified according to ATIP3 levels as in B. I, OS time (in months) for patients as in G with metastatic tumors classified as in H. Number of patients is under brackets. *, P = 0.0172.
Tumors were then classified according to their metastatic properties and MTUS1 probeset intensities were compared with the probability of patient survival. As shown in Fig. 1E, the percentage of patients with metastatic disease surviving after 5 years was markedly reduced when tumors expressed low ATIP3 (6.25%) compared with normal ATIP3 levels (31.6%), whereas in patients with nonmetastatic tumors, 5-year survival (Fig. 1E and Supplementary Fig. S1E) and OS rates (Fig. 1F and H) were independent of the levels of ATIP3. Within patients with metastatic disease, OS rates (Fig. 1F and H and Supplementary Fig. S1E) and survival time (Fig. 1G and I) were also reduced when tumors expressed low levels of ATIP3. Thus, ATIP3 expression is an important indicator of clinical outcome for patients with metastatic breast tumors. Correlation between low ATIP3 levels and reduced survival rates among patients with advanced breast cancer suggests major effects of ATIP3 on metastatic progression.
ATIP3 limits breast cancer metastatic colonization in vivo
In vivo effects of ATIP3 on the metastatic potential of breast cancer cells were evaluated using a well-defined experimental mouse model of metastasis monitored by intravital bioluminescence imaging (23, 25, 26). Highly metastatic, ATIP3-negative, D3H2LN breast cancer cells were transfected with either GFP or GFP-ATIP3, and independent stable cell clones (Cl3 and Cl6) expressing moderate levels of ATIP3 were selected (Fig. 2A, left). All cell clones exhibited similar levels of luciferase activity (Fig. 2A, right). Metastatic cancer cells were injected intracardiacally into the bloodstream of nude mice to recapitulate the late, rate-limiting, steps of the metastatic process, and examine metastatic dissemination to various organs while avoiding any effect of ATIP3 on primary tumor growth. Four groups of 18 mice were analyzed in two independent experiments. For each animal, the total number of metastatic foci and the number of photons/s were quantified every 2 days for 24 days (Supplementary Table S3). As shown in Fig. 2B, the time-course of metastasis formation was markedly delayed in mice injected with ATIP3-positive as compared with ATIP3-negative cell clones. The number of cancer cells growing at secondary sites increased exponentially from day 17 after injection of ATIP3-positive clones, as compared with day 10 for mice injected with control cells (Fig. 2B). As shown in Fig. 2C, the number of mice developing metastasis was strongly diminished upon ATIP3 expression. Importantly, the number of detectable metastases per mouse was also significantly reduced at all times in the presence of ATIP3 (Fig. 2D). At day 24, the number of mice invaded with large metastases reached 13 of 18 (72.2%) in the control group as compared with 2 of 18 (11.1%) following injection of ATIP3-positive cells (Fig. 2E and F), indicating a prominent effect of ATIP3 on cancer cell growth and colonization at secondary sites. Accordingly, on day 24, the total number of photons/s per mouse was 50- and 25-fold lower following injection with Cl3 and Cl6 clones, respectively, compared with WT (Supplementary Fig. S2A). For ethical reasons, mice had to be sacrificed at day 24, therefore OS of the two groups of mice could not be quantified. Furthermore, ex-vivo and histologic analysis of metastatic nodules (Supplementary Fig. S2B) confirmed that bioluminescent signals indeed correspond to metastases of human tumor cells having infiltrated mouse tissues. Metastases were mainly detected in the bones, the lungs, and the brain, which are the most frequent sites of metastatic dissemination of human breast tumors. No preferential location of metastatic nodules in ATIP3-positive versus ATIP3-negative cell types could be observed. Altogether, these results identify ATIP3 as a potent antimetastatic molecule, and support a role for ATIP3 in metastatic growth and colonization in vivo.
ATIP3 expression slows metastatic progression in vivo. A, characterization of stably transfected D3H2LN cell clones. Left, immunoblots of nontransfected (WT) and GFP-ATIP3-expressing D3H2LN clones (Cl3, Cl6) using anti-GFP antibodies, reprobed with anti-α-tubulin (α-tub) antibodies. Right,measurement of luciferase activity per cell (n = 3). B, number of photons/s per mouse (n = 9) at days 6 to 24 following tumor cell inoculation. C, left, representative pictures of bioluminescence (5/9 mice) at day 17 following intracardiac injection. Scale is on the right. Right, number of mice with at least one detectable metastasis at day 17. D, total number of metastastic sites per mouse at indicated days after inoculation of control (Ctrl, WT and GFP, n = 18) and ATIP3 positive (ATIP3, Cl3 and Cl6, n = 18) cells. **, P < 0.01, ***, P < 0.001. E, number of mice with large metastases at different times after inoculation as in D. F, representative pictures (day 24) as in C.
ATIP3 expression slows metastatic progression in vivo. A, characterization of stably transfected D3H2LN cell clones. Left, immunoblots of nontransfected (WT) and GFP-ATIP3-expressing D3H2LN clones (Cl3, Cl6) using anti-GFP antibodies, reprobed with anti-α-tubulin (α-tub) antibodies. Right,measurement of luciferase activity per cell (n = 3). B, number of photons/s per mouse (n = 9) at days 6 to 24 following tumor cell inoculation. C, left, representative pictures of bioluminescence (5/9 mice) at day 17 following intracardiac injection. Scale is on the right. Right, number of mice with at least one detectable metastasis at day 17. D, total number of metastastic sites per mouse at indicated days after inoculation of control (Ctrl, WT and GFP, n = 18) and ATIP3 positive (ATIP3, Cl3 and Cl6, n = 18) cells. **, P < 0.01, ***, P < 0.001. E, number of mice with large metastases at different times after inoculation as in D. F, representative pictures (day 24) as in C.
ATIP3 impairs breast cancer cell proliferation and migration
Metastatic colonization involves cancer cell migration, invasion through the extracellular matrix, and proliferation at the secondary site. As expected from our previous studies (19), cell proliferation was significantly reduced in ATIP3-positive clones Cl3 and Cl6 as compared with control (Supplementary Fig. S3A). In addition, Boyden chambers assays of chemotaxis and invasion revealed more than 90% reduction in the promigratory properties of Cl3 compared with GFP (Fig. 3A). Similar effects were observed using stably transfected MDA-MB-231 cells (Supplementary Fig. S3B). Conversely, ATIP3 silencing in metastatic MDA-MB-468 breast cancer cells expressing endogenous ATIP3 induced a 2- to 2.5-fold increased chemotactic migration (Fig. 3B), suggesting that cancer cells having lost ATIP3 may acquire a promigratory phenotype and may be more prone to develop distant metastasis.
ATIP3 reduces breast cancer cell migration. A, Boyden chamber migration of stable D3H2LN cell clones across filters coated (coll) or not (no coll) with collagen I. Results (percent) are mean ± SEM (n = 3). Right, representative picture of cells migrating to the bottom of the well. B, Boyden chamber assay using ATIP3-positive (WT and siCtrl) and ATIP3-negative (si#1 and si#2) MDA-MB-468 cells. Results are shown as in A. Top, immunoblots of MDA-MB-468 cells after siRNA silencing [anti-MTUS1, reprobed with anti-α-tubulin (α-tub) antibodies]. Right, representative pictures of the lower face of the filter. C, transendothelial migration, mean ± SEM (n = 3). D and E, cancer cell adhesion (mean ± SEM, n = 3) to endothelial cells (D) and collagen (E). F, migration of ATIP3-positive (siCtrl) and ATIP3-silenced (si#1, si#2) HeLa cells (n = 3). Left, representative pictures of wound at times T0 and T22. Right, quantification (percent) of wound closure at T22. Results are mean ± SEM (n = 2). Top, immunoblots of siRNA-transfected HeLa cells as in B. G, directional migration of stably transfected D3H2LN clones (n = 4). Left, representative pictures of wound at times T0 and T7. Right, quantification of wound closure at T4 and T7. Results are mean ± SD (n = 3). H and I, cell tracking of D3H2LN stable clones during wound closure. H, cell velocity scattered dot plot. I, diagrams of migration trajectories (12 hours). Number of cells is under brackets. Directionality coefficient (Dir) is inside the graph. A, B, F, and G, magnification, ×100. *, P < 0.05; **, P < 0.001; ***, P < 0.0001.
ATIP3 reduces breast cancer cell migration. A, Boyden chamber migration of stable D3H2LN cell clones across filters coated (coll) or not (no coll) with collagen I. Results (percent) are mean ± SEM (n = 3). Right, representative picture of cells migrating to the bottom of the well. B, Boyden chamber assay using ATIP3-positive (WT and siCtrl) and ATIP3-negative (si#1 and si#2) MDA-MB-468 cells. Results are shown as in A. Top, immunoblots of MDA-MB-468 cells after siRNA silencing [anti-MTUS1, reprobed with anti-α-tubulin (α-tub) antibodies]. Right, representative pictures of the lower face of the filter. C, transendothelial migration, mean ± SEM (n = 3). D and E, cancer cell adhesion (mean ± SEM, n = 3) to endothelial cells (D) and collagen (E). F, migration of ATIP3-positive (siCtrl) and ATIP3-silenced (si#1, si#2) HeLa cells (n = 3). Left, representative pictures of wound at times T0 and T22. Right, quantification (percent) of wound closure at T22. Results are mean ± SEM (n = 2). Top, immunoblots of siRNA-transfected HeLa cells as in B. G, directional migration of stably transfected D3H2LN clones (n = 4). Left, representative pictures of wound at times T0 and T7. Right, quantification of wound closure at T4 and T7. Results are mean ± SD (n = 3). H and I, cell tracking of D3H2LN stable clones during wound closure. H, cell velocity scattered dot plot. I, diagrams of migration trajectories (12 hours). Number of cells is under brackets. Directionality coefficient (Dir) is inside the graph. A, B, F, and G, magnification, ×100. *, P < 0.05; **, P < 0.001; ***, P < 0.0001.
The ability to migrate through a monolayer of endothelial cells (transendothelial migration) was significantly reduced (58 ± 16%) in Cl3 compared with control (Fig. 3C). Adhesion of clones Cl3 and Cl6 to endothelial cells was significantly elevated (3-fold and 2.8-fold, respectively) compared with WT (Fig. 3D), suggesting that increased tumor-endothelial cell adhesion may account for reduced transendothelial migration. Cell adhesion to collagen I was also increased in Cl3 (1.85-fold) and Cl6 (1.93-fold) compared with WT (Fig. 3E). Altogether, these data indicate that ATIP3 concomitantly increases cell adhesion and limits cell migration.
The consequences of ATIP3-silencing on cancer cell motility were analyzed in HeLa cells that express endogenous ATIP3 and are well suited for analyses of wound closure. As shown in Fig. 3F, ATIP3 silencing in HeLa cells increased (1.84- to 2.6-fold) directional migration. Conversely, stable ATIP3 expression into D3H2LN (Cl3 and Cl6, Fig. 3G) and MCF7 cells (Supplementary Fig. S3C) significantly reduced wound closure. Time-lapse microscopy (Supplementary Movies S1 and S2) and tracking of D3H2LN-migrating cells further indicated that stable ATIP3 expression impairs both cancer cell velocity (0.34 μm/s and 0.55 μm/s for Cl3 and GFP clones, respectively; ref. Fig. 3H) and directionality (Fig. 3I). Similar results were obtained (Supplementary Fig. S3D and S3E) by analyzing cell tracking following transient transfection of GFP or GFP-ATIP3 into D3H2LN cells (Supplementary Movies S3 and S4). Of note, the number of GFP-ATIP3–positive cells reaching the wound edge was reduced compared with GFP-expressing cells. GFP-ATIP3 expressing cells were overtaken by nontransfected cells reaching the border of the wound (Supplementary Fig. S3F), further confirming the inhibitory effect of ATIP3 on cancer cell migration.
ATIP3 alters microtubule dynamics
We hypothesized that ATIP3, being closely associated with microtubules (19), may limit cell proliferation and migration by regulating microtubule dynamics. We first analyzed the consequences of ATIP3 depletion on the sensitivity of microtubules to nocodazole that prevents repolymerization of dynamic microtubules. Stable microtubules that are not affected by nocodazole treatment are typically stained by anti-acetylated tubulin. As shown in Fig. 4A, ATIP3-silenced HeLa cells were highly sensitive to nocodazole. The number of cells retaining stable microtubules was decreased by 51% ± 10 and 53% ± 14 following transfection of siRNA#1 and siRNA#2 compared with control. Conversely, stable transfection of GFP-ATIP3 into MCF7 cells significantly increased the number of cells retaining stable, nocodazole-resistant, microtubules as assessed by antiacetylated tubulin labeling (Fig. 4B) and immunoblotting (Fig. 4C). ATIP3 expression also significantly delayed microtubule regrowth following nocodazole washout (Fig. 4D). At 5 minutes, microtubule length around the centrosome was reduced by 57 ± 20% in GFP-ATIP3 compared with GFP-transfected clones, supporting the notion that ATIP3 may impair microtubule dynamics.
ATIP3 reduces nocodazole sensitivity and microtubule outgrowth. A, immunostaining (anti-α-tubulin (α-tub) and anti-acetylated tubulin [Ac-tub antibodies] of ATIP3-positive (siCtrl) and -negative (si#1) HeLa cells incubated without (0) or with 1 μmol/L nocodazole (Nz). Right, immunoblots of HeLa cells after siRNA silencing (anti-MTUS1, reprobed with antitubulin antibodies). Bottom, quantification (%) of cells retaining stable microtubules, mean ± SEM (n = 3). B, immunostaining of stable MCF7 clones incubated with or without 10 μmol/L nocodazole, as in A. Right, quantification as in A, mean ± SEM (n = 3). C, immunoblot analysis of acetylated-tubulin (Ac-tub) and ezrin content in stably transfected MCF7 clones, either nontreated (−) or treated with DMSO (D) or increasing concentrations of nocodazole. Right, quantification of the ratio between Ac-tub and ezrin intensity. D, microtubule regrowth in transiently transfected RPE-1 cells (n = 4). Shown is α-tubulin staining at indicated times after nocodazole (10 μmol/L) washout. Right, quantification of microtubule density at 4 μm around the centrosome, mean ± SD (n = 4 to 10 cells). *, P < 0.05; **, P < 0.001. Scale bar, 10 μm.
ATIP3 reduces nocodazole sensitivity and microtubule outgrowth. A, immunostaining (anti-α-tubulin (α-tub) and anti-acetylated tubulin [Ac-tub antibodies] of ATIP3-positive (siCtrl) and -negative (si#1) HeLa cells incubated without (0) or with 1 μmol/L nocodazole (Nz). Right, immunoblots of HeLa cells after siRNA silencing (anti-MTUS1, reprobed with antitubulin antibodies). Bottom, quantification (%) of cells retaining stable microtubules, mean ± SEM (n = 3). B, immunostaining of stable MCF7 clones incubated with or without 10 μmol/L nocodazole, as in A. Right, quantification as in A, mean ± SEM (n = 3). C, immunoblot analysis of acetylated-tubulin (Ac-tub) and ezrin content in stably transfected MCF7 clones, either nontreated (−) or treated with DMSO (D) or increasing concentrations of nocodazole. Right, quantification of the ratio between Ac-tub and ezrin intensity. D, microtubule regrowth in transiently transfected RPE-1 cells (n = 4). Shown is α-tubulin staining at indicated times after nocodazole (10 μmol/L) washout. Right, quantification of microtubule density at 4 μm around the centrosome, mean ± SD (n = 4 to 10 cells). *, P < 0.05; **, P < 0.001. Scale bar, 10 μm.
The effects of ATIP3 on microtubule dynamic instability parameters were further analyzed by measuring EB1 protein accumulation at the microtubule plus tips (13–15, 28, 29) in RPE-1 epithelial cells and lung fibroblasts (MRC5-SV), which have a sparse microtubule array and are well suited for distinguishing individual microtubule tips. As shown in Fig. 5A, ATIP3 expression in RPE-1 cells led to a significant reduction in the number and size of EB1 comets that rather appeared as dots. Decreased accumulation of EB1 at microtubule plus ends was not associated with decreased EB1 expression (Supplementary Fig. S4A). In ATIP3-depleted HeLa cells, significantly more EB1 comets of increased length and intensity were detected compared with control cells (Fig. 5B), suggesting that ATIP3 silencing increases microtubule dynamics. Time-lapse total internal reflection fluorescence (TIRF) videomicroscopy analysis of EB1-GFP comets (Supplementary Movie S5) and subsequent microtubule-tips tracking indicated that microtubule growth episodes were significantly longer in ATIP3-silenced HeLa cells compared with control (Fig. 5C). ATIP3 depletion increased microtubule growth rate and decreased the time spent in pause as well as the frequency of catastrophes (Fig. 5C), accounting for increased microtubule dynamics. Conversely, videomicroscopy of EB3-GFP comets following expression of mCherry-ATIP3 in MRC5-SV cells (Supplementary Movie S6), and corresponding kymographs (Supplementary Fig. S4B), indicated that ATIP3 expression decreases microtubule dynamics and reduces the rate of microtubule growth.
ATIP3 regulates microtubule dynamics. A, immunostaining (anti-EB1, anti-mCherry antibodies) of RPE-1 cells transiently transfected with mCherry-ATIP3 (Ch-ATIP3). Insets, EB1 comet-like structures in ATIP3-negative (1) and positive (2) cells. Distribution of EB1 (black), α-tubulin (dashed) and ATIP3 (gray) at the microtubule tip (linescans) and quantification of comets intensity (scattered dot plot). Number of comets analyzed is under brackets. Shown is 1 experiment out of 5. B, EB1 localization in siRNA-silenced HeLa cells. Insets, EB1 comet-like structures in ATIP3-positive (1) and ATIP3-negative (2) cells. Distribution of EB1 (black) and α-tubulin (dashed) at the microtubule tip (linescans). Quantification of comets intensity as in A. Shown is 1 experiment out of 3. C, time-lapse images of siRNA-silenced HeLa cells expressing EB1-GFP. Arrowhead indicates the position of EB1 comet over time (in seconds). Parameters of microtubule dynamics (EB1-GFP comets) in siRNA-transfected HeLa cells (n = 100 comets) are shown in scattered dot plot and histograms. D, immunostaining [anti-α-tubulin (α-tub) and anti-GFP] of transfected D3H2LN in migration. Arrows indicate the direction of migration. Cell margin (black line) is visualized by bright field microscopy. Insets show microtubule array at the cell border of ATIP3-negative (1, 2) and GFP-ATIP3-positive (3, 4) cells. Right, immunoblots (anti-GFP, antitubulin) of transiently transfected D3H2LN cells. Bottom left, quantification of microtubules reaching given distance from the cell cortex in GFP- and GFP-ATIP3- expressing cells. Bottom right, mean distance between microtubules and cell cortex. Number of microtubules analyzed is under brackets. E, quantification (percent) of polarized D3H2LN cells during migration. Number of cells is under brackets. *, P < 0.05; **, P < 0.001, ***, P < 0.0001. A, B, C, and D, scale bar, 10 μm.
ATIP3 regulates microtubule dynamics. A, immunostaining (anti-EB1, anti-mCherry antibodies) of RPE-1 cells transiently transfected with mCherry-ATIP3 (Ch-ATIP3). Insets, EB1 comet-like structures in ATIP3-negative (1) and positive (2) cells. Distribution of EB1 (black), α-tubulin (dashed) and ATIP3 (gray) at the microtubule tip (linescans) and quantification of comets intensity (scattered dot plot). Number of comets analyzed is under brackets. Shown is 1 experiment out of 5. B, EB1 localization in siRNA-silenced HeLa cells. Insets, EB1 comet-like structures in ATIP3-positive (1) and ATIP3-negative (2) cells. Distribution of EB1 (black) and α-tubulin (dashed) at the microtubule tip (linescans). Quantification of comets intensity as in A. Shown is 1 experiment out of 3. C, time-lapse images of siRNA-silenced HeLa cells expressing EB1-GFP. Arrowhead indicates the position of EB1 comet over time (in seconds). Parameters of microtubule dynamics (EB1-GFP comets) in siRNA-transfected HeLa cells (n = 100 comets) are shown in scattered dot plot and histograms. D, immunostaining [anti-α-tubulin (α-tub) and anti-GFP] of transfected D3H2LN in migration. Arrows indicate the direction of migration. Cell margin (black line) is visualized by bright field microscopy. Insets show microtubule array at the cell border of ATIP3-negative (1, 2) and GFP-ATIP3-positive (3, 4) cells. Right, immunoblots (anti-GFP, antitubulin) of transiently transfected D3H2LN cells. Bottom left, quantification of microtubules reaching given distance from the cell cortex in GFP- and GFP-ATIP3- expressing cells. Bottom right, mean distance between microtubules and cell cortex. Number of microtubules analyzed is under brackets. E, quantification (percent) of polarized D3H2LN cells during migration. Number of cells is under brackets. *, P < 0.05; **, P < 0.001, ***, P < 0.0001. A, B, C, and D, scale bar, 10 μm.
Microtubule stabilization and decreased growth rate at the cell periphery should be responsible for an inhibition of microtubule targeting and capture at the cell cortex (30). As shown in Fig. 5D, in migrating D3H2LN cells, microtubules projected radially toward the cell periphery and microtubule plus ends were close to the cell edge (mean distance 1.43 ± 0.7 μm), whereas in the presence of ATIP3, microtubules were bended and more than 50% of microtubule tips did not reach the cell margin (mean distance 2.31 ± 1.2 μm). Of note, reduced ability of microtubules to reach the cell cortex in migrating ATIP3-positive cells was accompanied by a 34% decrease in cell polarity (Fig. 5E). Taken together, these results suggest that ATIP3-dependent regulation of microtubule dynamics results in decreased ability of microtubules to reach the cell cortex, which contributes to reduced cell polarity and migration.
Microtubule-binding domain D2 recapitulates the functional effects of ATIP3
The ATIP3 polypeptide was cleaved into 3 fragments designated D1, D2, and D3 (Fig. 6A), which were fused to GFP and expressed in RPE-1 cells (Fig. 6B). As shown in Fig. 6C, the GFP-D1 fusion protein did not associate with microtubules and was rather diffuse in the cytosol. Accordingly, GFP-D1 expression had no significant effect on the number, size, or intensity of EB1 comets (Fig. 6D). In contrast, GFP-D2 clearly colocalized with the microtubule cytoskeleton and centrosome in living cells (Fig. 6C). As for GFP-ATIP3, GFP-D2 was entirely retrieved in the pellet fraction in microtubule cosedimentation assays (Supplementary Fig. S5A). Of interest, upon expression of GFP-D2, accumulation of EB1 as comet-like structures at the microtubule plus ends was strongly impaired (Fig. 6E), indicating that expression of the D2 domain is sufficient to stabilize microtubules. Expression of GFP-D3 (Fig. 6C) led to the formation of large aggregates containing tubulin, probably due to oligomerization of coiled-coil motifs present in the C-terminal region of ATIP3 (31). Because of these aggregates, functional properties of GFP-D3 could not be evaluated further.
The D2 region of ATIP3 decorates and stabilizes microtubules. A, scheme of ATIP3 regions D1, D2, and D3. Amino acid numbering is according to accession number NP_001001924. B, immunoblots (anti-GFP, antitubulin) of RPE-1 cells transfected (24 hours) with GFP-D1, GFP-D2, and GFP-D3. C, immunostaining (anti-GFP, antitubulin) of RPE-1 cells transiently transfected as in A. D, anti-EB1 immunostaining of GFP-D1–transfected RPE-1 cells. Insets show EB1 comets in GFP-D1-negative (1) and GFP-D1-positive (2) cells. Distribution of EB1 (blue), α-tubulin (pink), and GFP-D1 (green) at the microtubule tip (linescans) and quantification of comets intensity (scattered dot plot). Number of comets is under brackets. E, GFP-D2-transfected RPE-1 cells stained with anti-EB1 antibodies and analyzed as in D. ***, P < 0.0001. B, C, and D, scale bar, 10 μm.
The D2 region of ATIP3 decorates and stabilizes microtubules. A, scheme of ATIP3 regions D1, D2, and D3. Amino acid numbering is according to accession number NP_001001924. B, immunoblots (anti-GFP, antitubulin) of RPE-1 cells transfected (24 hours) with GFP-D1, GFP-D2, and GFP-D3. C, immunostaining (anti-GFP, antitubulin) of RPE-1 cells transiently transfected as in A. D, anti-EB1 immunostaining of GFP-D1–transfected RPE-1 cells. Insets show EB1 comets in GFP-D1-negative (1) and GFP-D1-positive (2) cells. Distribution of EB1 (blue), α-tubulin (pink), and GFP-D1 (green) at the microtubule tip (linescans) and quantification of comets intensity (scattered dot plot). Number of comets is under brackets. E, GFP-D2-transfected RPE-1 cells stained with anti-EB1 antibodies and analyzed as in D. ***, P < 0.0001. B, C, and D, scale bar, 10 μm.
Altogether our results identify D2 as the ATIP3 domain able to associate with microtubules and suppress their dynamics. Of importance, the D2 domain also retained the ability of ATIP3 to inhibit cell proliferation (91.6% inhibition for GFP-D2 and GFP-ATIP3 compared with GFP; ref. Fig. 7A). In wound healing assays, cells expressing GFP-D2 showed reduced cell migration and directionality (Fig. 7B). Cell tracking of transient transfectants (Supplementary Movie S7) indicated that similar to GFP-ATIP3, GFP-D2–positive cells mostly remained at the back of the wound and were overtaken by untransfected cells (Fig. 7C and Supplementary Fig. S5B). Thus, the microtubule-binding domain D2 is sufficient to recapitulate the functional features of ATIP3.
Microtubule-binding domain D2 is the functional domain of ATIP3. A, colony formation of GFP-, GFP-ATIP3-, GFP-D2- transfected MCF7 cells, and quantification (mean ± SD, n = 4). Shown is 1 representative experiment out of 3. B, migration trajectories (17 hours) covered by GFP- (n = 30), GFP-ATIP3- (n = 17), and GFP-D2- (n = 24) expressing D3H2LN cells. Directionality coefficient (Dir) is inside the graph. C, aligned dot plots show Euclidean distance covered by untransfected (−) and transfected cells (+) cells as indicated. *, P < 0.05; **, P < 0.001; ***, P < 0.0001.
Microtubule-binding domain D2 is the functional domain of ATIP3. A, colony formation of GFP-, GFP-ATIP3-, GFP-D2- transfected MCF7 cells, and quantification (mean ± SD, n = 4). Shown is 1 representative experiment out of 3. B, migration trajectories (17 hours) covered by GFP- (n = 30), GFP-ATIP3- (n = 17), and GFP-D2- (n = 24) expressing D3H2LN cells. Directionality coefficient (Dir) is inside the graph. C, aligned dot plots show Euclidean distance covered by untransfected (−) and transfected cells (+) cells as indicated. *, P < 0.05; **, P < 0.001; ***, P < 0.0001.
Discussion
We report here that ATIP3 is an important prognostic marker for survival of the patients with breast cancer, independently of the ER status of the tumor. Using 3 different patient cohorts, we show that among metastatic breast tumors, low ATIP3 levels correlate with reduced probability for overall survival of the patients, suggesting that ATIP3 may be an important indicator of metastatic progression. Examination of ATIP3 levels in breast tumors may contribute to identify a population of patients at high risk of fatal complication, who should be the subject of careful medical follow-up.
Using a bioluminescence-based experimental mouse model for cancer metastasis (23, 25, 26), we showed that restoring ATIP3 expression into highly metastatic ATIP3-deficient D3H2LN breast tumor cells significantly delays the time-course of metastasis and reduces the number of detectable metastases per mouse at all times examined. ATIP3 expression in cancer cells also strongly reduces the size of metastatic foci as well as the number of mice fully invaded with large metastases. These observations, together with above mentioned results on human patients, suggest that ATIP3 may have a prominent effect on metastatic colonization.
Essential steps of metastatic progression include the ability of cancer cells to reach a secondary organ and grow in the new microenvironmental context (3–5). This requires active cell migration and proliferation, two important biologic processes that are significantly increased in breast cancer cells following ATIP3 silencing. By promoting dual effects on cancer cell proliferation and migration, ATIP3 likely regulates both early (tumorigenic) and late (metastatic) stages of cancer development. Beneficial actions of ATIP3 on a wide range of cancer-related processes, including invasion, transendothelial migration, cell migration, and proliferation may explain its potent anti-metastatic effects in preclinical studies.
Other studies have shown that the MTUS1 gene encoding ATIP3 is significantly downregulated in various types of cancers including from the pancreas (32), ovary (33), head-and-neck (34, 35), colon (36), and bladder (37). Low MTUS1 levels were also correlated with reduced overall survival of the patients with bladder cancer (37) and oral tongue squamous cell carcinomas (35), highlighting the potential importance of ATIP3 as a new prognostic marker in a variety of solid tumors.
At the molecular level, we show that ATIP3 is a microtubule-associated protein with potent microtubule-stabilizing effects. We propose that by stabilizing microtubules, ATIP3 decreases their dynamics therefore leading to impaired ability of microtubule tips to reach the cell cortex during migration. Microtubule dynamics at the cell cortex is essential for generating a polarized microtubule array, required for cell polarity and migration (30). Reduced microtubule dynamics may thus represent a major mechanism accounting for anti-migratory and anti-metastatic effects of ATIP3 in breast cancer. Accordingly, loss of ATIP3 leads to increased microtubule growth rate, less time spent in pause, and decreased frequency of catastrophes. Alteration of microtubule dynamics parameters in ATIP3-depleted cells may explain uncontrolled cancer cell motility that is associated with increased metastasis and poor prognosis in patients with ATIP3-negative breast cancer. The association of ATIP3 with the microtubule lattice involves an internal basic region designated D2 whose expression is sufficient to recapitulate all effects of the full-length protein on microtubule stabilization, as well as cell proliferation, motility, and directional migration. The microtubule-binding D2 region thus represents the functional domain of ATIP3. Further characterization of this domain and identification of intracellular interacting partners may help deciphering the molecular mechanisms by which ATIP3 limits breast cancer cell migration and hence, metastasis. Our study paves the way to the design of peptides or small molecules able to mimic the effects of ATIP3, which is a prerequisite for the development of targeted therapy. These may be particularly beneficial to the subset of breast tumors that have lost ATIP3 expression and are prone to metastasize.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: S. Honoré, M. Di Benedetto, C. Nahmias, S. Rodrigues-Ferreira
Development of methodology: A. Molina, L. Ghouinem, M. Abdelkarim, M. Di Benedetto
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Velot, B.P. Bouchet, A.-C. Luissint, I. Bouhlel, A. Vincent-Salomon, O. Delattre, B. Sigal-Zafrani, F. André, B. Terris, M. Di Benedetto, S. Rodrigues-Ferreira
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Molina, L. Velot, S. Honoré, D. Braguer, N. Gruel, O. Delattre, F. André, B. Terris, A. Akhmanova, M. Di Benedetto, C. Nahmias, S. Rodrigues-Ferreira
Writing, review, and/or revision of the manuscript: A. Molina, L. Velot, A.-C. Luissint, A. Di Tommaso, S. Honoré, B. Sigal-Zafrani, F. André, A. Akhmanova, C. Nahmias, S. Rodrigues-Ferreira
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Morel, E. Sapharikas
Study supervision: C. Nahmias, S. Rodrigues-Ferreira
Acknowledgments
The authors thank the Cochin Imaging facility, the Hospital St Louis Animal facility, Nicolas Cagnard (Genomics facility of the University Paris Descartes), Muriel Andrieu, Karine Labroquère (Cybio platform), Maryline Favier, and Corinne Lesaffre (Histology facility) of the Cochin Institute.
Grant Support
This work was supported by the University Paris Descartes, Inserm, CNRS, the Ligue Contre le Cancer-IDF, Inserm-Transfert, the association “Le Cancer du Sein, Parlons-en!”, Roche laboratories, the fondation RAJA-Danièle-Marcovici, Odyssea, and Prolific.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.