Increased expression of MNK1b, the spliced isoform of MNK1, predicts poor prognosis and is associated with triple-negative breast cancer

MAP kinase interacting kinases (MNKs) modulate the function of oncogene eukaryotic initiation factor 4E (eIF4E) through phosphorylation, which is necessary for oncogenic transformation. MNK1 gives rise to two mRNAs and thus two MNK1 isoforms, named MNK1a and MNK1b. MNK1b, the splice variant of human MNK1a, is constitutively active and independent of upstream MAP kinases. In this study, we have analyzed the expression of both MNK1 isoforms in 69 breast tumor samples and its association with clinicopathologic/prognostic characteristics of breast cancer. MNK1a and MNK1b expression was significantly increased in tumors relative to the corresponding adjacent normal tissue (p < 0.001). In addition, MNK1b overexpression was found in most of the triple-negative tumors and was associated with a shorter overall and disease-free survival time. Overexpression of MNK1b in MDA-MB-231 cells induced an increase in the expression of the MCL1 antiapoptotic protein and promoted proliferation, invasion and colony formation. In conclusion, a high expression level of MNK1b protein could be used as a marker of poor prognosis in breast cancer patients and it could be a therapeutic target in triple-negative tumors.


INTRODUCTION
Breast cancer has an heterogeneous nature [1]. Breast tumors can be classified into subtypes based on their expression profile, and each subtype is associated with distinct histological markers and clinical parameters. The molecular classification is based on the presence of the hormone receptors (estrogen receptor [ER] and/ or progesterone receptor [PR]), the human epidermal growth factor receptor 2 (HER2) and on the epithelial cell of origin (luminal and basal-like). Each subtype has a different prognosis and treatment response: thus, because ER is a therapeutic target, luminal subtypes are amenable to hormone therapy and, in consequence, are associated with less aggressive metastatic disease and longer diseasefree survival [2]. Similarly, the HER2 group is a potential candidate for trastuzumab therapy. However, the absence of expression of a known therapeutic target makes triple-

Research Paper
negative tumors more biologically aggressive, difficult to treat, and often have a poor prognosis.
The MAP kinase interacting kinase 1 (MNK1) has been shown to bind to and to be a substrate for both the mitogen-activated extracellular signal-regulated protein kinases (ERK) 1/2 as well as the stress-activated p38 MAP kinases (p38 MAPKs) both in vitro and in vivo [3]. MNK1b is a truncated isoform lacking the MAP kinasebinding motif present in the C-terminal region of MNK1a, which arises from an alternatively spliced transcript of the MNK1 gene [4]. Both MNK1 isoforms have a nuclear localization signal (NLS) in the N-terminal region that allows the kinase to enter into the nucleus [5]. MNK1a contains a nuclear export signal (NES) that ensures its cytoplasmic localization [5], whereas MNK1b lacks this NES, being cytoplasmic and nuclear [4]. From the different features in human MNK1 isoforms, we have shown that MNK1b has higher basal activity than MNK1a. Moreover, MNK1b activity does not correlate with the phosphorylation of the activation loop residues and seems to be independent of the upstream kinases (ERK1/2 and p38 MAP kinase) [4].
The only well-characterized substrate for MNK1 is the eukaryotic initiation factor 4E (eIF4E) [6]. It has been shown that eIF4E overexpression in a variety of cancers including breast, head and neck, colon, prostate, kidney and lung is related to disease progression [7][8][9]. Elevated levels of phosphorylated eIF4E are found in human cancer tissues obtained from patients with lung, head, colorectal, and gastric cancers and primary pancreatic ductal adenocarcinoma [10,11]. Nuclear eIF4E phosphorylation appears to be important to control the transport of cyclin D1 mRNA and for the transforming properties of eIF4E [12]. Other studies have established that phosphorylation of eIF4E on Ser209 by MNK1/2 is an absolute requirement for the oncogenic action of eIF4E. The inhibition of MNK activity with different MNK1 inhibitors reduces several tumoral characteristics such as proliferation, colony formation or migration in human breast cell lines [13][14][15]. On the other hand, overexpression of the oncogene HMD2 in cancer cells is regulated by eIF4E, so that the overexpression of eIF4E promotes the export of the HDM2 mRNA in a MAP kinase-and Mnk1-dependent manner [16]. In addition, Wendel and colleagues [17] have shown that overexpression of a constitutively active MNK1 diminishes the apoptosis and accelerates the development of tumors in an experimental model of mice while an inactive mutant reduces the development of these tumors. Finally, it has been demonstrated that eIF4E is completely dephosphorylated in MNK1 and MNK2 knock-out mice, which do not exhibit any apparent phenotypic consequences, questioning a vital role of MNK activity [18].
This study aimed to determine the levels of MNK isoforms in breast cancer, whether they are associated with clinicopathological data and whether the constitutively active MNK1b is differentially expressed relative to MNK1a. Interestingly, we observed increased MNK1a and MNK1b levels in breast tumors and that MNK1b expression correlated with some aspects of clinicopathological status, especially with triple-negative tumors. In addition, the role of the overexpression of MNK1a or MNK1b on oncoprotein expression, migration, invasion and proliferation in triple negative breast cell lines has been investigated.

MNK1a/b levels are increased in breast cancer tissues
The expression of MNK1a and MNK1b was analyzed in 69 tumor samples from female patients diagnosed with breast cancer and the corresponding adjacent normal tissue. The patient's characteristics are shown in Table 1. Most of the patients (70%) were older than 50 years, with a median age of 67 (range, 33-96 years). The pathologic tumor stage was predominantly T2-T4 (72.5%) and more of the 50% of the tumors showed positive nodal status. Almost all tumors showed histological grade 2 and 3 (45 and 48%) and only 15% of the cases showed grade 1.
The expression of the two isoforms of MNK1, MNK1a and MNK1b, was analyzed on western blots, and the ratio between tumor (T) and adjacent normal (control, C) tissues was calculated. We used two specific antibodies, anti MNK1(C-20) generated against a C-terminal peptide which only recognized MNK1a isoform, and an antibody raised against a peptide mapping at the N-terminus of Mnk1 of mouse origin (MNK1 (M-20)) that recognizes both isoforms of human MNK1, although experimental results in our laboratory indicated that it recognizes MNK1b with higher affinity than MNK1a in western blot. As shown in Figure 1A, the levels of MNK1a and MNK1b increased significantly in tumor tissue relative to control tissue (about 2.2 and 5.3-fold, respectively). In addition, we analyzed the phosphorylation status and the expression of eIF4E, as a substrate of MNK1a and MNK1b ( Figure 1B). As expected, phosphorylation and levels of eIF4E increased significantly in the tumor samples (about 1.5 and 2.6-fold respectively), what is in agreement with the results shown by other authors [7,10,19]. www.impactjournals.com/oncotarget To check if the changes in the MNK1b levels were due to changes in the mRNA levels, a quantitative analysis of the mRNAs of MNK1a and MNK1b was performed as described in Materials and Methods section. The RNA quality was analyzed using an Experion Station. We used samples with similar RQI factors (RNA quality index) between control and tumor tissues (54 samples). The levels of MNK1a isoform did not change whereas MNK1b mRNA levels increased significantly ( Figure 1C). Interestingly, MNK1b protein and mRNAs levels correlated significantly (Spearman r = 0.403, p = 0.003) and MNK1a/Mnk1b mRNAs correlated inversely (spearman r = -0.417, p = 0.002) indicating that the switch in the ratio of MNK1b/MNK1a mRNAs probably by alternative splicing is responsible of the increased MNK1b expression. Therefore, one can conclude that the increase in MNK1b levels result from an increase in its mRNA expression.

Increased MNK1b levels are associated with clinicopathological status
Next, we analyzed whether the expression of MNK1 isoforms, and the expression and phosphorylation of eIF4E correlate with clinical-pathologic characteristics of the tumors (Supplementary Table 1). No significant correlation was found between MNK1 isoforms expression, eIF4E expression or its phosphorylation status with almost none of the parameters studied. We only found that MNK1a decreased significantly in those tumors positive for Progesterone receptor, MNK1b is higher in pT1 tumors than in pT2-4 tumors and eIF4E levels increased significantly in positive node tumors. Taken into account that the ratio T/C for MNK1b could not be calculated in many cases (>47%) because the protein was undetectable in control tissues by western blot analysis, we decided to perform an additional qualitative analysis. A score of 0 was assigned when no changes in MNK1b expression levels between tumor and control tissue were observed or when the protein was not detected in tumor tissue. A score of 1 was given when MNK1b levels were higher in tumors than in adjacent normal tissue or when MNK1b was only detected in tumor tissue. Higher MNK1b levels were found in 62% of the tumor samples (41/66). As shown in Table 2, there are significant differences in the overexpression of MNK1b between different molecular types, almost all TN tumors (93%) are positive for MNK1b expression. In addition, there are differences although not statistically significant (p = 0.094) in the overexpression of MNK1b between HER2-negative and HER2-positive tumors (67% vs. 43%). It should be noted that, as MNK1a (Supplementary  Table 1), MNK1b is overexpressed in tumors negative for progesterone receptor (77% vs. 52% in positive PR, p = 0.046).
The levels of MNK isoforms, phosphorylated eIF4E and eIF4E were also detected by immunohistochemical staining in 34 paraffin sections (18 Luminal, 7 TN and 9 HER2 positive) as described in Materials and Methods section. The results of the immunohistochemical stainings are shown in Figure 2 and Table 3. MNK1a and MNK1b positive staining (T>N) was observed in the 32% and 44% of the cases respectively, whereas more of the 70% of the cases were positive for eIF4EP and eIF4E. Mnk1a positive staining was associated with the HER2 expression (p < 0.01) ( Table 3). It has been previously reported that activation cascades mediated by HER2 signaling promotes the phosphorylation and translocation to the nucleus of the Y-box binding protein 1 (YB-1) factor where it binds its target genes promoting transcription, including MNK1 as a target gene [20]. MNK1b staining positively correlated with the grade and Ki-67 levels and negatively with the expression of hormone receptors. Interestingly, all the TN cases analyzed were negative for MNK1a staining while 4 of the 7 cases were positive for MNK1b. No association were found between eIF4EP or eIF4E status and the clinicalpathologic characteristics (Table 3). It is interesting to point to that the expression of MNK1b isoform was associated with eIF4EP staining (Chi-square test, p = 0.028).

MNK1b is associated with patient disease-free and overall survival
Finally, the overall and disease-free survival rates were compared among cases low or high MNK1a or MNK1b levels measured by western blot ( Figure 3A-3D) or by immunohistochemistry ( Figure 3E-3H), using the log rank test. MNK1a expression was not associated with overall and disease-free survival rate ( Figure 3A-3B and 3E-3F) whereas the "MNK1b-high group" had lower overall and disease-free survival rates than the "MNK1b low group", being statistically significant for disease-free survival ( Figure 3D and 3H).

The overexpression of MNK1b induces MCL1 overexpression
In order to study the effect of the overexpression of either MNK1a or MNK1b on tumorogenesis, we have analyzed the levels of several oncoproteins whose expression is regulated by eIF4E or by its phosphorylation, in a MNK1 or upstream MAP kinases dependent manner [12,16,17,21,22]. First, we have analyzed the expression of cyclin D1, c-myc, MCL1 and HDM2 by western blot in the stable MDA-MB-231 cells transfected with pCDNA3 (control), MNK1a (MDA-MB-MNK1a) or MNK1b (MDA-MB-MNK1b) previously generated and characterized in our laboratory [14]. As shown in Figure 4A The expression of MCL1 in paraffin tumor samples was analyzed by immunohistochemistry as described in Materials and Methods section ( Figure 4C). MCL1 staining was positive in the 38% of the cases studied, being more frequent in TN (57%) than HER (44%) and Luminal (28%, p = 0.012).

The overexpression of MNK1a or MNK1b induces proliferation, migration, invasion and colony formation in triple negative cell lines
The effect of the overexpression of both isoforms of MNK1 on cell viability, migration, invasion, colony formation and cell cycle was studied in transienly transfected MDA-MB-231 and MDA-MB-468 cell lines as described in Materials and Methods section ( Figure 5A-5E).The effect of the overexpression of MNK1 isoforms was different for the two TN cell lines analyzed. Thereby, the overexpression of MNK1b induced a substantial increase in the proliferation rate, invasion and colony formation in MDA-MB-231 cells ( Figure 5A, 5B, 5D). In addition, MNK1b induced a light but significant increase in the percentage of cells in G2/M phase in this cell line ( Figure 5E). However, the effect of MNK1a overexpression produced a minor effect or did not affect at these oncogenic characteristics ( Figure 5). On the contrary, in MDA-MB-468 cell both MNKs did not have effect, or even induced a decrease, on the colony formation, migration and cell cycle. The effect of MNK1a overexpression on invasion was higher than MNK1b ( Figure 5D) while only MNK1b induced a significant increase in the proliferation rate ( Figure 5A). These results are in agreement with the lack of effect on MCL1 expression in MDA-MB-468 cells.

DISCUSSION
Breast cancer is a complex and heterogeneous disease in terms of its histology, cellular origin, mutations, capacity to metastasize, disease progression and treatment response. Translational deregulation plays an important role in cancer, and a broad number of studies have been undertaken to characterize both the expression of translational regulators and whether these alterations are significant determinants of cancer progression (reviewed in [23]). However, although these studies include eIF4E in breast tumors [8,24,25] and, more recently the phosphorylation status of this protein [10,26], very few studies have addressed the nature of MNK1/2 expression, now considered a potential target for anticancer therapy.
In this study, we determined the expression levels of MNK1a and MNK1b in tumor samples and normal adjacent tissue from breast cancer patients. In the tumor samples, significantly higher levels of both MNK1a and MNK1b protein and higher MNK1b mRNA levels were found compared to healthy tissue. Up to date, only two studies had previously determined the levels of MNK1 in tumors. In the first, the authors showed an increase in both MNK1 mRNA and protein expression in human gliomas [27]. Very recently, Hou et al. [28] have evaluated MNK1 expression in epithelial ovarian cancer. In this paper, the authors demonstrated a significant relationship between MNK1 expression and advanced stage and positive lymph node metastasis. In addition, high MNK1 expression indicated poor clinical outcomes in EOC tissues. However these studies did not discriminate between MNK1a and MNK1b isoforms. In our study, MNK1b levels, but not MNK1a, are associated with the patient outcomes in breast cancer. This finding indicates that MNK1b could be a novel candidate prognostic marker for use in breast cancer patients.
Interestingly, we have found that the MNK1b expression was significantly increased in triple-negative tumors, in which over 92% of the samples expressed more MNK1b in tumor tissue than in control tissue.
Triple-negative tumors are characterized by the lack of expression of ER and PR and the absence of HER2 overexpression, and no targeted therapy is available for this tumor type. The expression of the antiapoptotic protein MCL1 correlates with high tumor grade and lower survival rate in breast cancer [29]. This protein plays a key role for TN cell survival [30,31] and several recent studies indicate that could be considered as a therapy target [32][33][34]. It has been described that activated MNK1 promotes the onset of tumor development which was attributed to the ability of MNK1 to enhance the expression of the   Figure 1A) as cutoff value: cases with Mnk1a levels > 2.2 constitute the high group and cases with Mnk1a levels < 2.2 constitute the low group. This method yields a survival curve that considered both outcomes (death) and censored cases (x). Significance is calculated by log-rank test between two groups. On the top of the curve "n" indicated the number of samples analyzed. www.impactjournals.com/oncotarget  anti-apoptotic protein MCL1 [17]. Our cell experiments, showing that MCL1 expression upregulated by MNK1b correlates with the oncogenic characteristics of the cells, corroborate this result.
The hyperactivation of two major signalling pathways, PI3K/AKT/mTOR and Ras/MAPK/MNK, occurs in the majority of the cancers and eIF4E acts as a convergence point for both signalling pathways to promote tumorigenesis. PI3K activity is often high in tumors due to the loss of the PTEN tumor suppressor, which allows a constitutive activation of AKT kinase, inappropriate mTORC1 signalling, phosphorylation and inactivation of 4EBP1 and, consequenltly, eIF4E available to be phosphorylated by MNK1. This occurs in MDA-MB-468 cell line, where PTEN is mutated, and both signalling pathways are activated (Supplementary Figure 1D). In these cells, overexpressed MNK1a will be quickly activated and, plus endogenous MNK1a, will exercise its oncogenic action. In MDA-MB-231 cell line, where both pathways are less active, the overexpression of the pathways-independent isoform MNK1b promotes an increase in several oncogenic characteristic (mainly invasion and colony formation). We hypothesized that the oncogenic effect of MNK1b could produce worse progression in those tumors with non-hyperactivated AKT/mTOR or/and MAPK pathways, and, in addition, could induce resistence to therapies based on PI3K/AKT inhibitors. Altogether these findings suggest that MNK1b may be used as a therapeutic target in the future in triplenegative tumors ( Figure 6).

Patients
Samples and data from patients included in this study were provided by the Hospital Universitario Ramón y Cajal-IRYCIS Biobank integrated in the Spanish Platform Biobanks Network (RetBioH; www. redbiobancos.es) and were processed following standard procedures with appropriate approval of the Ethical and Scientific Committees. We analyzed 69 breast tumors and normal adjacent tissue included in OCT (optimal cutting compound) frozen via liquid nitrogen and stored at −80º C and 34 paraffin-embebed resected tumor tissues.

Sample preparation
Eight to ten thick tissue sections (20-30 µm) were obtained using a cryostat and used for RNA and protein extraction using TRIZOL ® Reagent (Invitrogen, Barcelona, Spain) following the manufacturer's protocol. In parallel, a 5-µm section was stained with hematoxylin and eosin following a standard procedure and examined under a light microscope to check the percentage of tumor present in the sample. The RNA pellet was resuspended in 50 μL of RNase-free water, treated with 1 μL of DNase I (Fermentas), quantified and stored at −80° C. The protein pellets were dissolved in 50 μL of electrophoresis buffer [180 mM Tris-HCl (pH 6.8), 9% sodium dodecyl sulphate (SDS), 6% 2-Mercaptoethanol, 15% glycerol, 0.05% bromophenol blue], heated at 95° C, cooled and stored at -80° C.

Immunohistochemistry
Tumor sections (5 μm) were deparaffinized and rehydrated described previously [35] and antigen retrieval was achieved by heat treatment in a pressure cooker for 2 min in 10 mM citrate buffer (pH 6.5). Next, endogenous peroxidase was blocked, and the sections were

Western blot
Equal volumes of normal adjacent (control) and tumor samples were resolved in 12% SDS-PAGE gels. MNK1a and MNK1b levels were analyzed using specific anti-MNK1 antibodies from Santa Cruz Biotechnology (both 1:500 dilution); MNK1 (C-20) generated against a C-terminal peptide that only recognizes the MNK1a isoform, and MNK1 (M-20) raised against a peptide mapping to the N-terminal of MNK1. Membranes were incubated with anti-phosphorylated eIF4E antibody (Cell Signaling Technology) and reprobed with an antitotal eIF4E antibody (BD Biosciences). β-actin antibody (Sigma, St Louis, MO, USA) was used to measure housekeeping protein levels as a control to monitor the homogeneity of loading. When enough sample was available, western blot analysis was repeated loading an equal amount of actin. Two or three different western blot analyses were carried out per sample.
Cell lysates (35 μg) were resolved in 12%, 15% or 10% SDS-PAGE gels and western blots were performed as above. Actin was used as a loading control. Antibodies used in this study are listed in Supplementary Table 2.

Transfection and MTT assays
MDA-MB-231 and MDA-MB-468 cells were plated at 6 × 10 3 , 3 × 10 4 or 2.5 × 10 5 cells/well in 96-, 24 or 6-well plates respectively. After 16-24 hours the cells were transfected with the different plasmids Flag-pCDNA3, Flag-MNK1a-pCDNA3 and Flag-MNK1b-pCDNA3 using jetPRIME transfection reagent (Polyplus Transfection) following the manufacturer's instructions. For MTT assays, medium was removed after 72 hours and 100 μl of MTT (1 mg/ml in culture medium) was added to each well, and plates were incubated at 37° C for 3 hour. Next, 100 μl/well of lysis buffer (10% sodium dodecyl sulphate and 10 mmol/l HCl) were added and, after 24 hours of incubation, absorbance was read at 540 nm on a SpectraFluor microplate reader (TECAN). Data were calculated as the percentage relative to the cells transfected with empty vector.

Migration and invasion assays
The migration and invasion assays were carried out using Transwell insert chambers (Corning, USA). For migration assay, cells transfected as described above, were hasterved for 16 hours, collected, counted and 4 × 10 4 cells were plated into the upper chamber in serum-free medium. Complete medium (10% FBS) were added in the lower chamber as chemoattractant. After 24 h, medium was removed from the top and bottom of the wells; cells on the insert were fixed with 4% formaldehyde for 2 min followed by 20 min with 100% methanol. Cells were stained with 30 µM Hoechst 33342 for 15 min, protected from light, washed twice in PBS and cells on the top were removed with a cotton stick. At least five photographs were acquired for each sample using a fluorescent microscope (Olympus IX70) and the number of cells was analysed using Image J software.
For invasion assay, after starvation for 24 hours, 4 × 10 4 cells were seeded as above into upper chambers precoated with matrigel (Corning, USA) in serum-free medium. After incubation for 48 h at 37° C, non-invading cells on the upper surface of filter were removed with cotton swabs and invading cells that migrated to the lower surface of filter were fixed, stained and scored as described above.

Flow cytometric analysis of cell cycle
Twenty-four hours postransfection, cells were collected and fixed in 70% cold ethanol 30 min at -20° C. Afterwards, cells were washed twice with ice-cold phosphate-buffered saline (PBS) and incuabated with 50 µg/mL Propidium Iodide (Sigma) plus 100 µg/mL RNAse A (Sigma) for 1 hour at 37° C at dark. Cell cycle was analyzed by flow cytometry (FACScalibur, BD Biosciences) using selective gating to exclude doublets of cells. Cytometry data was analyzed using Flowing software 2.5.1.

Statistics
The results are expressed as mean values ± standard deviation (SD), unless stated otherwise. Most of the samples were analyzed in triplicate. The significance relative to 1 (ratio T/NT) was analyzed by a one-sample t-test. For each parameter determined, the statistical significance between the two groups was analyzed with the Mann-Whitney U test. Three or more groups were compared by Kruskal-Wallis analysis followed by Dunn's Multiple Comparison Test. Spearman's correlation coefficients (r) were calculated. Pearson's chi-square test was used to compare the difference of MNK1b-positive values among the groups. Univariate survival analyses were carried out according to Kaplan-Meier and logrank tests. The differences were considered statistically significant when P values were < 0.05. The SPSS 15.0 software package was used for the statistical analysis.