MIEN1, a novel interactor of Annexin A2, promotes tumor cell migration by enhancing AnxA2 cell surface expression

Enhanced expression of MIEN1 is reported in breast cancer compared to normal breast tissues [2]. Analysis of Cancer Genome Atlas (TCGA) data sets identified significantly elevated MIEN1 expression in different subtypes of breast carcinomas (Apocrine, Large Cell Neuroendocrine, Cribiform, Papillary, Ductal, Lobular, Mixed Ductal and Lobular, Mucinous) patients compared to normal tissues (Fig. 1a). In clinical oncology, evaluations of breast tumors are accompanied by an assessment of the molecular status of ER, PR and Her-2 oncogene. To understand the differential expression of MIEN1 in various subtypes of breast cancer, we examined the expression of MIEN1 within the molecular subtypes of breast cancer. Our findings revealed that MIEN1 is predominantly overexpressed in Her-2 positive (85 % cases with elevated MIEN1) and luminal B (63 % cases with elevated MIEN1) subtypes. However, MIEN1 expression in other subtypes basal-like and luminal A were moderate, whereas majority of the normal breast tissues had low MIEN1 (Fig. 1b). Screening of various established breast tumor lines identified increased expression of MIEN1 in majority of the breast tumor lines (MCF10AT, MCF10CA1a, MCF10CA1d, MCF10CA1h, JIMT-1, BT-474, SKBR-3, MDA-MB231, T47D, MCF-7, MDA-MB436 and HCC-70) compared to the immortalized normal mammary epithelial cell line, MCF10A (Fig. 1c). As expected most of the Her-2 amplified cell lines displayed increased expression of MIEN1 protein (CRL-2330, SKBR-3, and BT-474); but MIEN1 expression is not only restricted to Her-2 amplification indicating its distinct transcriptional and post-translational modifications contributing to its elevated expression in breast tumors. Classification of the patient cohort according to high and low MIEN1 expression using TCGA dataset, confirmed a poor survival in breast cancer patients with elevated MIEN1 expression as previously shown [21] (Fig. 1d). Altogether, these findings confirm that MIEN1 is a clinically important oncogene, and its increased expression contributes towards an aggressive disease with poor survival.

In addition to isoprenylation motif [6], MIEN1 harbors several potential phosphorylation sites. Analysis of MIEN1 sequence using computational algorithms (KinasePhos 2.0) identified four potential phosphorylated tyrosine residues (Tyr29, Tyr39, Tyr50 and Tyr85) [22‐25] of which Tyr39 and Tyr50 residues are located in the ITAM domain [21]. The canonical ITAM (immunoreceptor tyrosine based activation motif) is an 18 sequence amino acids (YxxI(6–8)YxxL) where tyrosine is separated from a leucine or isoleucine by any two other amino acids, giving the signature YxxL/I; and these two signatures are typically separated by 6 to 8 amino acids. To demonstrate that the Y39/Y50 in the ITAM domain of MIEN1 is phosphorylated, we investigated the phosphorylation status of MIEN1 (Fig. 2a). Immunoprecipitation of GFP-tagged MIEN1 constructs followed by immunoblot analysis using a generic phospho-tyrosine antibody, demonstrated that MIEN1 wild type (MIEN1^WT) is tyrosine phosphorylated, whereas the Y39F and Y50F phospho-deficient mutants showed lower phosphorylation status. However Y50F mutant showed a greater loss of phosphorylation compared to Y39F (Fig. 2b). Replacement of Y39 and Y50 with phenylalanine (Y39/Y50F) reduced MIEN1-tyrosine phosphorylation by half; indicating that there is some background phosphorylation on the tyrosine residues outside the ITAM domain.

Since, MIEN1 is also modified by isoprenylation at the C-terminal ‘CVIL’ end responsible for its membrane localization [6], we investigated whether phosphorylation of ITAM-tyrosine residues is dependent on isoprenylation. For this, we analyzed the tyrosine-phosphorylation status of MIEN1 in NIH3T3 cells (with low endogenous MIEN1) transfected with GFP vector only, MIEN1^WT or prenylation-deficient MIEN1^C112S constructs. In addition, the MIEN1^WT expressing cells were either treated with a pharmacological inhibitor against GGTase enzyme GGTI (geranylgeranyl transferase I inhibitor) or vehicle control (DMSO). MIEN1^WT expressing cells showed robust tyrosine-phosphorylation whereas GGTI treatment severely abrogated the effect similar to the MIEN1^C112S expressing cells (Fig. 2c). These data indicate that tyrosine-phosphorylation of MIEN1 in the ITAM-domain is dependent on prior isoprenylation. These results demonstrate that correct localization of the protein to the plasma membrane is required for subsequent post-translational modification via tyrosine-phosphorylation at the ITAM-domain.

Next, we evaluated the functional impact of MIEN1-ITAM and isoprenyl mutants compared to the wild-type protein (MIEN1^WT). For this, we stably expressed MIEN1 constructs in NIH3T3-cells with low endogenous protein. Cell migration was significantly enhanced in cells expressing MIEN1^WT relative to the vector control (Fig. 3a, b), whereas MIEN1^C112S failed to migrate into the wound (Fig. 3c), as reported earlier [6]. Similar to the isoprenyl-mutants, NIH3T3 expressing ITAM mutants had a reduced migratory potential (Fig. 3d–f). However, we did not observe significant differences in functional contribution between individual tyrosine mutants (MIEN1^Y39F and MIEN1^Y50F) compared to MIEN1^Y39/50F confirming that the phosphorylation of both tyrosine residues is vital for ITAM induced functions or signaling events to mediate cell migration. The MIEN1^{Y39/50F/C112S} cells did not show any additive effect suggesting both isoprenylation and ITAM-phosphorylation are important for MIEN1 function (Fig. 3g). To determine whether MIEN1 posttranslational modifications are also required for invasive function, we performed matrigel invasive assays with MDA-MB231 cells transfected with MIEN1 constructs. Consistent with the data from the migration assays, MIEN1^WT expression increased the invasive potential of breast cancer cells whereas ITAM and isoprenyl mutants failed to show an invasive phenotype (Additional file 1: Figure S1).

Cell migration is the result of coordination between membrane protrusive structures at a leading edge, adhesion and de-adhesion followed by translocation of the cell body in the direction of migration. In migrating cells, protrusive structures such as filopodia are the pioneers, which probe the environment for cues and guide cell migration. MIEN1 has been shown to induce filopodia formation and mutation of the isoprenyl motif impaired this function [6]. Here we inquired whether the ITAM regulates MIEN1-induced filopodia, aside from the isoprenyl motif. NIH3T3 cells expressing various constructs of MIEN1 were stained with rhodamine-conjugated phalloidin and subjected to immunofluorescence analysis post wound induction. Stable expression of MIEN1 in NIH3T3 cells visibly increased the number and length of filopodia post wound induction compared to MIEN1-mutants expressing cells (Fig. 3i–n) confirming that ITAM and CAAX are important for MIEN1 function. Immunoblotting analysis of cells expressing vector control and MIEN1 constructs confirmed the equal expression of the wild type and mutants MIEN1 protein (Fig. 3o).

We investigated the potential interacting partners of MIEN1 to define the mechanisms associated with tumor cell migration and invasion. In a yeast two-hybrid assay, we identified MIEN1 as potential interactor of AnxA2, a Ca(2+)-dependent phospholipid binding protein which translocates to the cell surface upon cellular signaling. Full-length AnxA2 cDNA cloned into GAL4 DNA-binding domain (GAL4 DNA-BD) of vector pGBKT7 was found to interact with MIEN1 in a yeast two-hybrid screen from a transformed human placental cDNA library as bait. Positive clones were selected on high-stringency medium (synthetic dropout medium) selection markers SD/−Ade/−His/−Leu/−Trp/X-α-gal, and only true interactor-MIEN1 could activate the expression of β-galactosidase (blue color). The left panel shows positive interaction of MIEN1 with AnxA2, but not with p53. The right panel shows the positive control p53 − T-antigen interaction, while the AnxA2 − p53 interaction served as the negative control (Fig. 4a). Following immunoblotting and real-time PCR analysis of breast cell lines expressing both MIEN1 and AnxA2 (Additional file 2: Figure S2A-B), we performed co-immunoprecipitation of endogenous AnxA2 with MIEN1 in BT-474 cells to confirm the yeast two-hybrid data. The reciprocal immunoprecipitation of MIEN1 also pulled down endogenous AnxA2, confirming that these two proteins indeed reside in a complex. The total input used for the immunoprecipitation confirmed equal loading and similar levels of expression of both the proteins (Fig. 4b). Colocalization experiments using confocal microscopy also confirmed interaction of endogenous AnxA2 and MIEN1 primarily in the cytosol, plasma membrane and the perinuclear area of breast cancer cells (Fig. 4c). Finally to confirm that AnxA2 and MIEN1 physically interact intracellularly, we performed FRET detection by fluorescence lifetime imaging microscopy (FLIM) assay to measure the proximity of MIEN1 and AnxA2. The lifetime decays of the donor (MIEN1) and donor-acceptor (MIEN1-AnxA2) pair were measured to be 1.75 and 1.26 ns, respectively. Substituting the lifetime values in the Förster equation, the efficiency of energy transfer was determined to be 28 %, which corresponds to a distance of 50.3 Å between the donor and acceptor pair; indicating that MIEN1 and AnxA2 indeed physically interact and reside in a very close proximity (Fig. 4d). These studies clearly validate MIEN1 as a novel interactor of AnxA2.

We investigated the effects of MIEN1 and AnxA2 interaction on cell migration, given their independent implications in cell migration and invasion (Fig. 5a-c). We silenced MIEN1 and AnxA2 alone or in combination (Fig. 5b) and examined the effects on motility of MDA-MB231 cells. In monolayer migration assays, the wound closure of cells with MIEN1 knockdown was significantly inhibited compared to siGFP (control) transfected MDA-MB231 cells (Fig. 5a, Additional file 3: Figure S3A-B). Silencing of AnxA2 resulted in impairment of cell motility at 6 and 12 h following wound creation (Fig. 5a, Additional file 3: Figure S3A-B). Using similar knockdown strategy, we observed that dual depletion of MIEN1 and AnxA2 led to a two-fold decrease compared to siGFP (Fig. 5c), indicating that MIEN1 and AnxA2 functionally cooperate to promote breast cancer cell migration. AnxA2 is a phospholipid binding protein localizing to the plasma membrane towards the cytosolic side. Previous studies proved that the phosphorylation of AnxA2 at the Y23 residue in the N-terminus of the protein causes its translocation to the extracellular surface, thereby activating extracellular proteolysis [26, 27]. To verify whether MIEN1 expression affects AnxA2 phosphorylation and subsequent translocation to the extracellular surface, we performed immunoblotting analysis followed by total internal reflection microscopy (TIRF-M). As shown in Fig. 5d, over-expression of MIEN1 enhanced AnxA2 phosphorylation at Y23. Conversely, down-regulation of MIEN1 led to a decreased phosphorylation of AnxA2 at Y23 (Fig. 5e). To further confirm this observation, we performed total internal reflection fluorescence microscopy (TIRF-M) in MDA-MB231 cells. Stable MDA-MB231 cells expressing vector control or MIEN1^WT were probed with AnxA2 and phospho-AnxA2^Tyr23 antibodies followed by TIRF microscopy to determine cell surface total-AnxA2 and phospho-AnxA2. MIEN1 overexpression significantly enhanced phosphorylation of AnxA2 at the tyrosine 23 residue and subsequently cell surface localized AnxA2 levels; indicating MIEN1 dependent signaling and interaction activate AnxA2 function (Fig. 5f).

Extracellular surface localized AnxA2 functions as a receptor for tPA and plasminogen; by binding to both enzymes, and catalytically activating plasmin generation [28, 29]. Increased levels of plasmin enhance proteolytic cleavage of ECM thus allowing cancer cells to migrate and invade to distant sites. To confirm that MIEN1 regulates AnxA2 dependent plasmin generation, we investigated the effects of MIEN1 ablation either alone or in combination with AnxA2 plasmin generation in breast cancer cells. Using basal-like HCC70 and luminal MCF7 cells (these two lines express high levels of both AnxA2 and MIEN1), we inquired whether the interaction of MIEN1 and AnxA2 had an effect on plasmin generation. We first confirmed siRNA-mediated knockdown of MIEN1 and AnxA2 in HCC-70 and MCF-7 (Fig. 6a, c) respectively. Western blotting analysis confirmed that the expression of MIEN1 or AnxA2 was dramatically reduced in the cells upon transfection with either MIEN1 or AnxA2 siRNA. Next, we examined the biochemical conversion of plasminogen to plasmin in both cell lines upon siRNA treatment. Silencing of both MIEN1 and AnxA2 led to a significant decrease in plasmin levels in both HCC-70 and MCF-7 (P < 0.05) (Fig. 6b, d). While the knockdown of AnxA2 led to a decrease in plasminogen conversion to plasmin in the initial hours as previously shown [30], the total change in plasmin levels were insignificant compared to the control siRNA treated cells. Interestingly, suppression of MIEN1 in MCF-7 cells led to approximately 1.7 fold decrease in plasmin levels but not in HCC-70; a difference which can be attributed to the expression of various regulators of the plasminogen-plasmin system. Taken together, these data indicate that co-expression of both AnxA2 and MIEN1 enhance plasmin generation and lead to an increase in breast tumor cell migration and invasion which in turn drive the metastatic process.