Annexin A1 attenuates EMT and metastatic potential in breast cancer

Metastasis is the major cause of carcinoma-induced death, but mechanisms involved are poorly understood. Metastasis crucially involves epithelial-to-mesenchymal transition (EMT), causing loss of epithelial polarity. Here we identify Annexin A1 (AnxA1), a protein with important functions in intracellular vesicle trafficking, as an efficient suppressor of EMT and metastasis in breast cancer. AnxA1 levels were strongly reduced in EMT of mammary epithelial cells, in metastatic murine and human cell lines and in metastatic mouse and human carcinomas. RNAi-mediated AnxA1 knockdown cooperated with oncogenic Ras to induce TGFβ-independent EMT and metastasis in non-metastatic cells. Strikingly, forced AnxA1 expression in metastatic mouse and human mammary carcinoma cells reversed EMT and abolished metastasis. AnxA1 knockdown stimulated multiple signalling pathways but only Tyk2/Stat3 and Erk1/2 signalling were essential for EMT.


2
Eger et al, 2000; Reichmann et al, 1992). Stable knockdown of AnxA1 in EpRas and EpC40 cells yielded EpRas-siAnxA1 and EpC40-siAnxA1 cells, which showed a similar, fibroblastoid EMTphenotype as EpRasXT cells (Figs. S10B-1,B-2). All EpH4-derived cells were cultivated in DMEM/F12 medium (GIBCO) plus 10% fetal bovine serum (FCS), 100 U/ml penicillin and 50µg/ml streptomycin, 1mM HEPES, 2mM L-glutamine. However, for cultivation of EpRas cells FCS was reduced to 4%, while EpRasXT cells were kept in medium with 15%FCS. To guarantee similar culture conditions of all cells used to produce culture supernatants for analysis of secreted factors, cells were cultivated overnight in media without serum and supernatants concentrated by ultrafiltration.
The MDCK-DDRafER cell line, originally obtained from J. Downward (Lehmann et al, 2000) was optimized by cloning to resemble the original, fully polarized MDCK cells in absence of ligands (Tanos & Rodriguez-Boulan, 2008). Cells were grown in DMEM/F12 plus 10% FCS. For EMT-induction, the cells were seeded and maintained on plastic at semi-confluent density and treated or not treated with 3x10 -6 M estradiol (E2) for 2 days. Cells were then trypsinized, re-seeded at near confluency on porous supports (translucent filters with 1.5 µm mean pore size, 1.7 or 4.5 cm in diameter) and treated or not treated with E2 plus 3ng/ml TGFβ1 for 4-5 days, during which time untreated cells got dense enough to fully polarize (as shown with apical markers and Z-stack analysis, not shown). Treatment with drugs or inhibitors was performed while the cells were expanding for 4-5 days on filters.

Tissue collections and databases used for analysis of endogenous AnxA1 levels:
Human breast cancer tissue samples  were obtained from the Biobank of the Medical University Graz, Department of Pathology, as a kind gift of Dr. Kurt Zatloukal. All tissues were snap-frozen in methylbutane pre-cooled with liquid nitrogen within 20min after surgery. The study was approved by the ethical committee of the Medical University Graz. RNA was prepared from these tissue samples according to standard procedures and processed for quantitative RT-PCR (qRT-PCR) as described in the main text and below. The samples consisted of 4 benign mammary tissue samples and 20 samples from invasive breast tumors.
Analysis of the Gene Logic database for AnxA1 expression in normal and cancerous breast, colon and lung tissues (Fig. S10A-5) was performed by Boehringer Ingelheim Austria (BI Austria) personnel (Ferby et al, 2006), who provided us with the graph in Fig. S1A, which depicts data from evaluating Affymetrics chips containing probes for essentially all human genes and hybridized with probes from >10 different human carcinoma types prepared from 20 to >100 tissue samples from cancer patients as well as normal tissues. The data shown in Fig. S1A  All procedures were performed according to Gateway protocol's instructions (Invitrogen).
The following two different oligonucleotides were selected for RNAi of murine AnxA1: 5´AGGGTAGGGGCTCACTGCTGACCCAGGA3` 5´GGAGACCATAAT CCTGATCAATGCCTTA3`, annealed with respective, second oligonucleotides and the resulting, double-stranded shRNA fragments were inserted into the gateway-compatible pSHAG vector (Paddison et al, 2002) according to the Gateway protocol's instructions (Invitrogen). The resulting U6-promoter-driven shRNAs were then transferred to the pMSCV-pgk-puromycin retroviral vector (Clontech) by recombination, according to Gateway protocols instructions. A basic BLAST alignment search of all target sequences did not reveal significant sequence homology with other genes. All shRNAexpressing vectors were confirmed by sequence analysis of the target insert. Vectors containing 28nt target sequence for human AnxA1 were generated as described above for the mouse constructs, selecting three oligonucleotide sequences: To generate retroviral vectors expressing human AnxA1, we PCR-amplified full length AnxA1 from a cDNA library. This DNA was inserted into the p201 gateway donor vector and then subcloned into two different target vectors by recombination according to the gateway protocol (Invitrogen): the pMSCV-pgk-puromycin vector to infect MDA-MB-231 cells and the pMSCV-GFP target vector to infect the already puro-resistant EpC40-siAnxA1 cells. The insert was confirmed by sequencing using specific primers in both constructs.

Generation of stable cell lines expressing the above retroviral vector:
Retrovirus producer cell lines (NIH-3T3gp+86 cells) were transfected with the above retroviral constructs using Lipofectamine2000 reagent (Invitrogen). These producer cells were seeded into 6-well plates (4.5 ml medium) at densities allowing generation of semiconfluent cell layers at three consecutive days. Twice daily, supernatants were harvested from producer cells having the correct density, filtered through a 0.45 µm filter and used to infect target cells in the presence of polybrene (2µg/ml) by respective medium changes. Cells were then selected for 2 weeks with 2.5µg/ml Puromycin or were FACS sorted for GFP-positive cells and the resulting pools tested for reduction (AnxA1-RNAi) or elevation (exogenous expression of AnxA1) by Western Blot analysis. Pools with significant AnxA1 knockdown (obtained from vectors harboring different AnxA1-specific siRNAs) or AnxA1 overexpression were cloned via single-cell FACSsorting into 96-well plates, expanded clones were verified by Western Blot analysis and frozen in liquid nitrogen. Stable knock down clones of Jak1, Jak2 and Tyk2 were generated using MISSION Lentiviral Transduction particles (Sigma-Aldrich; Jak1: NM_146145; Jak2: NM_008413; Tyk2: NM_018793) according to the manufacturer's instruction. Briefly, 1,6 x 10 4 cells/well were seeded into 96 well plates in growth medium and grown to 50% confluency. 1-3 x 10 5 transduction units (TU) in presence of polybrene were added to the cells. Following 24hr incubation, the medium was changed and the cells were selected with puromycine. Unfortunately, normally proliferating knockdown clones obtained showed no strong reduction of the respective target mRNAs.
Most pharmacological inhibitors were used as 5-50 mM stock solutions in DMSO, stored at -80°C and added daily to cells in collagen gels -or porous supports in case of MDCK-DDRafER cells -starting 24 hours after cell seeding and avoiding > 1/1000 DMSO dilutions. Salinomycin was dissolved at a concentration of 50mM in Methanol. Powder and stock solutions of the Ras (farnesyl-transferase) inhibitor L739.749 were stored at -70ºC, dilutions were prepared immediately before addition to cells, kept on ice and made fresh every day. Inhibitor working concentrations leading to a complete response were determined earlier on cell types known to respond to the above inhibitors against p38MAPK (Maschler et al, 2004), PI3K, MEK, and RAS (Janda et al, 2002a;Janda et al, 2002b); TGFβRI (Waerner et al, 2006) and Jak inhibitor I (Grebien et al, 2008).

Detailed procedure for Western Blot analysis:
For Western Blot (WB) analysis, cells were lysed on ice for 20 min in ice-cold RIPA buffer, containing 150mM NaCl, 50mM Tris pH 7.4, 1% NP40, 1% Na-desoxycholate, 1mM Na 3 VO 4 , 25mM NaF, 1mM PMSF, 5mM β-glycerol-phosphate, 10mM DTT, protease inhibitor cocktail tablets (Complete mini; Roche, according to manufacturer`s instructions). Lysates were cleared for 5 min at 12500xg and 4°C and protein concentrations determined using the Bradford assay (Biorad). 40µg of protein were denatured in SDS-sample buffer, followed by SDS-PAGE and transferred to PVDF membranes. After treatment with blocking solution (5% bovine serum albumin (BSA) in TBST; i.e. Tris-buffered saline + Tween 20), membranes were incubated with the respective antibody in blocking solution for 1hr at room temperature or over night at 4°C, washed with plain TBST and incubated with the respective secondary antibody in blocking solution for 1hr at room temperature. After 5 washes with TBST for 30 min, the blots were developed using an enhanced chemiluminescence kit (ECL, Amersham). Quantitation of WB data from human breast carcinoma cells analysed for AnxA1 protein expression was performed by densitometry, using ImageJ software according to the manufacturers instructions, using loading control data (β-actin) for normalization.

Immunoprecipitation of cell surface-and metabolically labelled proteins: detailed procedure:.
For exclusive labelling of cell surface proteins, cell surface biotinylation was performed as described previously (Maschler et al, 2005) with minor modifications. Briefly, cells were detached by trypsinisation, washed in Hank`s balanced salt solution (HBSS) and biotinylated at 2x10 6 cells per ml for 30 min, using 1mg/ml Sulfo-NHS-Biotin (Sigma) in HBSS. After stopping the biotinylation reaction with DMEM containing 0.6 % bovine serum albumin (BSA) for 10 min, cells were washed extensively in HBSS and lysed in 50mM Tris (pH 7.5), 0.1% Triton X100, 50mM NaCl. Equal amounts of protein were immunoprecipitated with the respective antibody for 1 hour and immune complexes collected using Protein A/G plus beads (Santa Cruz). The immunoprecipitates (removed from beads with SDS sample buffer) were subjected to SDS-PAGE, transferred to nitrocellulose membranes (Amersham) and incubated with blocking solution (see Western Blot) at 50°C for 1 hour. Biotinylated proteins were detected by treating the membrane with streptavidine-horse radish peroxidase (Sigma, 1:5000 in 1%BSA for 1hr at room temperature) and visualized using ECL-detection reagent.
For metabolic labelling, equal cell numbers were plated on 10 cm plastic dishes in 10 ml medium at 1-3 x10 6 cells/dish and pre-incubated with methionine and cysteine-free MEM (GIBCO) for 1hr, then labelled with 100µCi of [ 35 S]-L-methionine/L-cysteine (MP Biomedicals) in 10 ml of the above medium for 5hr and lysed as described for Western blot analysis. Aliquots of cell lysates were cleared by centrifugation and aliquots containing 2 x 10 6 cpm of incorporated [ 35 S] were subjected to immunoprecipitation with appropriate antibodies. Precipitated proteins were subjected to SDS-PAGE. Gels were fixed in 7% acetic acid, 25% methanol in aqua dest. for 30 min, treated with ENHANCE amplifier solution (NEN), dried and exposed for autoradiography on Kodak X-ray film in enhancer screen cassettes.

Collagen gel culture methods and immunofluorescence analysis:
For successful 3D culture of MDA-MB-231 cells in collagen gels, we optimized the cell numbers seeded (between 1000 and 3000 cells per gel). TGFβ was added 2 days after seeding where indicated, while pharmacological inhibitors were added 1 day after seeding. Depending on 8 the growth rate of the collagen structures, gels were maintained for 7-10 days, after which they were photographed, fixed and processed for antibody staining.

Non-Permeant cellular ELISA: details and minor modifications:
Cell surface EGFR was determined using a non permeant cellular ELISA assay described in (Lee et al, 2002) with minor modifications. Briefly, 1x10 4 cells were seeded into 12-well plates in 1.5 ml medium and cultivated for 24 hours. Cells were then fixed with 4% paraformaldehyde in 50mM phosphate buffer (pH 7.2) for 20min at room temperature. Cells were washed three times with ice cold PBS containing 0.1M glycine and incubated with 3% H 2 O 2 in PBS for 5 minutes to minimize endogenous peroxidase activity. After blocking with 5%FCS+1% BSA in PBS for 30 minutes, cells were incubated for 1 hour at room temperature with the primary anti-EGFR-antibody (Santa Cruz, sc-03, 1µg/ml in blocking solution), subsequently washed three times with ice cold PBS and treated with the peroxidase conjugated secondary antibody. Then cells were washed and the colour reaction was carried out with the substrate solution of the IL6 ELISA kit. The coloured reaction end product was transferred to a microtiter plate and stop solution (from the IL6 ELISA kit) was added. After 30min, the colour reaction was measured with an ELISA reader at 450nm, using 570nm as the control wavelength.

Analysis of AnxA1 expression in human tissue tumor arrays by immunohistochemistry:
Slides bearing 3 µm sections from two human breast cancer tissue tumor arrays (TA2, TA3) using primary tumor tissue from 141 patients with known disease history were dried at 65°C for 45 minutes, deparaffinated by washing with xylene three times followed by incubation in 100% and 50% ethanol. Antigen retrieval was performed by incubating the slides in antigen retrieval buffer (DAKO S1699) for 30minutes at 95°C. After the slides were allowed to cool down they were washed once in 1xPBS and incubated in blocking solution (15% goat serum / PBS) for 15 minutes.
Primary antibody (anti-AnxA1; Becton Dickinson; BD 610067) was added at a dilution of 1:1500 in 1%BSA / PBS and slides were incubated for 60 minutes. Endogenous peroxidase was blocked by treating the samples with 0.3% H 2 O 2 / PBS for 20minutes. After washing 3 times with 1xPBS the secondary antibody (biotinylated, DAKO E0413) was added at a dilution of 1:500 in 1%BSA / PBS and the samples were incubated for 30minutes followed by three washing steps with PBS. The streptavidine-biotin complex (DAKO K0377) was added for 30 minutes. Then the slides were washed three times in PBS and colour development was carried out in AEC (1% w/v 3,3-amino-9ethyl-carbazole; Sigma 5754 in N, N-dimethylformamide). Counter-stain was performed with Hematoxylin (Sigma MHS-16). The slides were then washed in tap water, dried for 15minutes and mounted in mounting medium (Aquatex, Merck 1.08562) and analysed by high resolution imaging (MIRAX-software) of the stained arrays. After normalization of staining intensity using Adobe Photoshop CS4 (Waerner et al, 2006), using blue nuclear and cytoplasmic staining for normalization, intensity and subcellular localization of the brown AnxA1 signal was recorded and correlated in Kaplan-Meier plots to metastasis free survival and overall survival (Waerner et al, 2006), using the available patient disease history information.

RNA Isolation and cDNA synthesis:
Total RNA was isolated using the TRIzol Reagent (Invitrogen) according to the manufacturer's instruction. cDNA was synthesized from 2µg of total RNA using the "Ready to go you prime First Strand" beads kit (Amersham Biosciences) and random oligonucleotides according to the manufacturer`s instruction.

qRT-PCR of AnxA1 levels in cell lines and tumor tissues:
RNA was prepared from cell pellets containing 1-2x10 8 cells or from tumor tissues according to standard procedures. qRT-PCR of the expression levels of AnxA1 in these RNA samples was performed using the SYBR green 2x PCR kit (Applied Biosciences) and a primer mix from Qiagen and were normalized to GAPDH levels.

qRT-PCR of E-cadherin repressor in cell lines:
1µl of cDNA was used per reaction, which was performed using SYBR Green (Molecular Probes) and an Opticon2 Monitor Fluorescence Thermocycler in a 25µl reaction volume for 40 cycles. Comparable Ct (cycle threshold) method was used to quantify the amplified fragments and RNA expression levels were normalized to GAPDH.

Tumor formation and metastasis assay: minor modifications and details:
Tumor formation ability was determined by mammary gland fat pad or subcutanous injection as described earlier (Jechlinger et al, 2002;Maschler et al, 2005;Zhao et al, 2003), with minor modifications. Briefly, cells to be injected were trypsinized, washed in PBS and 1x10 5 cells were 10 injected into the mammary gland fat pad of anesthetized MF1 nu/nu mice (2 injection sites per mouse, 3 mice each). For subcutaneous injection of cells from the HMEC cell series described above, trypsinized cells were washed with PBS and 4x10 6 cells were injected in 0.1 ml PBS (4 injection sites per mouse, 2 mice per cell line). In the subcutaneously injected mice receiving the HMEC-series, tumors occurred after approximately 4 weeks. Tumor weight was measured by dissecting out the tumor and determining tumor weight, by immersing tumors into PBS-filled, preweighed dishes and weighing them again.
Metastasis formation was measured after injection of the respective cell lines into the tail vein of nude mice (Janda et al, 2002a;Waerner et al, 2006). 1x10 5 cells in 0.1ml of PBS were injected into the tail vein of MF1 nu/nu mice, using 3 mice per cell type with MDA-MB231-control, MDA-MB231+AnxA1, EpRas-sicontrol, and EpRas-siAnxA1 cells and 7 mice per cell type for EpC40sicontrol and EpC40-siAnxA1 cells. Mice were inspected every few days and sacrificed when becoming moribund, which occurred between 25 and 45 days in the mice injected with EpC40-siAnxA1 and MDAMB231-control vector cells, while the lungs remained metastasis-free for >100 days with EpC40-sicontrol cells and for >55 days in case of the MDA-MB-231+AnxA1 cells. Two experiments were done with three and four mice injected with EpC40-sicontrol, and EpC40-siAnxA1 cells (using the AnxA1/2 and AnxA1/6 cell clones) and pooling the data in Fig. 3f, the observation times of the control mice in these two experiments were 16 and 12 weeks. In case of the EpRas-sicontrol and EpRas-siAnxA1 cells, the latter formed already large metastases at a time where the control EpRas cells had just started to form tiny metastases. Due to these "black and white" differences in the first two models, metastases were only counted on the surface of the dissected lungs immersed in PBS under a dissection microscope, rather than counting metastases in serial sections.

Supplemental Material: AnxA1 expression in various human tumors (summary of published data; focus on metastasis)
AnxA1 was downregulated in breast cancer (Shen et al, 2005;Shen et al, 2006), thyroid cancer (Petrella et al, 2006), head and neck cancer (Garcia Pedrero et al, 2004), larynx tumors astrocytomas (Ruano et al, 2008) and renal cell carcinomas (Zimmermann et al, 2007). Although not discussed by the respective authors, high AnxA1-expression was typical for early stage cancers like adenomas in most reports, while loss of AnxA1 was mainly seen in more strongly progressed tumors (Lim & Pervaiz, 2007;Mussunoor & Murray, 2008). Importantly, however, very little has been published on AnxA1-expression during metastasis, particularly with respect to tumor cells at the invasion front, which (often transiently) lose epithelial polarity and gain EMT. In addition, the subcellular localization of AnxA1 -changing from cytoplasmic-to vesicle-localization during butyrate-or inosine-induced colon carcinoma epithelial differentiation (Guzman-Aranguez et al, 2005), and between plasma-membrane-and cytoplasmic localization in renal cell carcinomas Neu) were maintained separated from males until tumors developed (6-9 months). Tumors were snap frozen in liquid nitrogen and kept at -80°C. For histology and expression of EMT-markers in these tumors, see (Jechlinger et al, 2006). For Western Blot analysis, small tumor pieces were lysed in RIPA buffer using a homogenizer and lysates were then subjected to SDS-PAGE (see main text, Fig. 1E).
To test, whether or not loss of AnxA1 would enhance metastasis in an in vivo metastasis mouse model, we used 129/Bl6-Annexin A1-/-mice (kindly provided by R.Flowers, London, GB) 12 and wt mice of the same mixed background. These mice were crossbred to the abovementioned MMTV-neu-or MMTV-neu x MMTV-TGFβ transgenic mice to generate these strains on a AnxA1-/-background. During an observation time of >1.2 years, these mice developed tumors and metastases at similar ages as the two respective FVB/N-transgenic strains not bred to AnxA1-/mice, indicating that neu was not sufficient to enhance metastasis in AnxA1-/-mice which show early compensation by other family members. In line with this, EpH4 cell expressing activated HER2* underwent reversible rather than metastable EMT upon exposure to TGFβ (A. Pacher and HB, unpublished), suggesting that HER2* in contrast to oncogenic Ras cannot sustain a complete EMT phenotype.

EpRasXT cells by forced expression of human AnxA1
A) Western blot (WB) analysis of EpH4, EpRas and EpRasXT cells (Fig. S10A-1  days. WB analysis for activated (pErk1/2) and total Erk1/2 shows that Erk1/2 phosphorylation is not affected by H14 (due to hyperactivation of the Erk/MAPK pathway; Janda et al, 2002a) but completely inhibited in EpC40-sicontrol cells (Erk/MAPK signalling not hyperactive), while pErk1/2 is not reduced by H14 in EpC40-siAnxA1 cells, suggesting that knockdown of AnxA1 activates Erk1/2 in an EGFR-independent fashion. D), E) EpC40-sicontrol and EpC40-siAnxA1 cells were treated with the EGFR kinase inhibitor H14 in collagen gels. Bright field images (D) reveal no obvious effect of EGFR-tyrosine kinase inhibition. Similarly, no clear effects of H14 were observed after staining the collagen gels for EMT markers (E). This was also true for EpRasXT cells (E), suggesting that EGFR-signaling is not involved in Ras/TGFβ and Ras/AnxA1-RNAi induced EMT.

Ras/AnxA1-RNAi-induced EMT, but reverses Ras/TGFβ-induced EMT
A) EpC40-sicontrol and EpC40-siAnxA1 cells were seeded on porous supports and treated after 24 hours with 5 µM salinomycin for 5 days. Confocal immunofluorescence images revealed, however, that Salinomycin induced re-expression of mostly cytoplasmic, but also plasma membrane E-cadherin, partially delocalized ZO-1 and caused partially cortical β-actin localization, but did not affect high Vimentin and Fibronectin expression (bar 20 µm). B) MDCK-DDRafER cells induced to undergo complete EMT upon treatment with Estradiol (E2) and TGFβ after seeding on porous supports were treated or not treated with 6 µM Salinomycin for 5 days, starting 2 days after seeding. Salinomycin completely prevented EMT induction seen without inhibitor (see Fig. S9, bottom), the cells expressing plasma membrane E-cadherin, cortical β-actin, no Fibronectin and little Vimentin, similar as untreated control cells (bars 20 µm).

Supplemental Figure S9: EMT induction in MDCK-DDRafER cells by estradiol and TGFβ
MDCK-DDRafER cells are indistinguishable from MDCK control cells with respect to their polarized epithelial phenotype. They show a typical, epithelial morphology in phase contrast (inset, green frame). Confocal immunofluorescence images (Bar 20 µm, in insets 10 µm) show that the cells express strictly plasma membrane localized E-cadherin and ZO-1, which fully colocalize with cortical β-actin. In contrast, these cells essentially lack Vimentin and Fibronectin. When induced to undergo EMT in response to estradiol (E2) and TGFβ, they form multilayers of fibroblastoid cells, which lack E-cadherin and ZO-1, show β-actin localizing to stress fibers or the cytoplasm, and strongly express Vimentin and Fibronectin (for details of EMT-induction see Suppl. Exp. Proc.).

Supplemental Figure S10: Explanatory schemes of the cell-and tissue systems employed
This scheme depicts the various cellular models used to analyze A) endogenous AnxA1 expression in murine and human tumor cells and tissues, and B) cellular models to study phenotypic changes induced by knockdown (RNAi) or exogenous expression of AnxA1. The different models and test systems are shown in numbered subpanels (panels A1-A7; panels B1-B4) and are referred to in the text as e.g. Fig. S10A-2. The scheme should be used together with the detailed description of the various cell models in Suppl. Exp. Proc., where the parental normal epithelial cell line EpH4 (used to derive systems A1, A2, B1, B2) and the various EpH4 derivatives and other cell systems used are described. The names of the various cell types shown within a given subpanel have not been inserted into the schemes, but are explained in Suppl. Exp. Proc., e.g. EpRas cells after TGFβ treatment (EpRasXT cells; see subpanel S10A-1). The other cell systems used (see A3, A4, B3, B4) are also explained in detail in Suppl. Exp. Proc., in which we also describe the sources and processing of the tumor samples and databases referred to in panels A5, A6, A7. The symbols for epithelial and mesenchymal markers used in the cartoons and the circular arrow depicting autocrine 22 TGFβ are explained at the bottom of the figure. EMT, epithelial-mesenchymal transition; red cross, AnxA1 knockdown via RNAi. Differently colored nuclei depict exogenous expression of various, defined genes of interest.