MicroRNA-149 Regulates Proliferation, Migration, and Invasion of Pituitary Adenoma Cells by Targeting ADAM12 and MMP14

Pituitary adenomas (PAs) can adapt an aggressive phenotype by invading adjacent brain structures with rapid cellular proliferation. Previous studies demonstrated that excessive expression of metalloproteases ADAM12 and MMP-14 is instrumental for the active proliferation and invasiveness of PA cells in vitro and of tumors in vivo. However, the mechanisms regulating ADAM12 and MMP-14 expression in PAs remain unclear. Target gene prediction and transcriptomic profiling of invasive vs. noninvasive human PA samples were performed to identify miRNA species potentially involved in the regulation of ADAM12 and MMP14. For cellular analyses of miRNA functions, two mouse PA cell lines (AtT20 and TtT/GF) were transfected with miR-149-3p and miR-149-5p, respectively. The effects of miR-149 (3p and 5p) on expression levels of ADAM12 and MMP14 were determined by Western blotting followed by an analysis of proliferation and colony formation assays, scratch migration assays, and invasion assays. A significant downregulation of miRNA-149 was observed in invasive vs. noninvasive PA (0.32 vs. 0.09, P<0.0001). In AtT-20 and TtT/GF mouse PAs cells, transfection of mimic miRNA-149 (3p and 5p) caused a significantly reduced cell proliferation and matrigel invasion, whilst the effect on cell migration was less pronounced. Both strands of miRNA-149 (3p and 5p) markedly reduced protein levels of ADAM12 and MMP-14 by at least 40% in both cell lines. This study proved that the invasiveness of PA cells is, at least partly, regulated by miRNA-149-dependent expression of ADAM12 and MMP-14.

Pituitary adenoma (PA) is the second most common intracranial benign tumor. According to a recent survey, it accounts for about 10%-15% of all intracranial tumors and has even been identified in 27% of non-selected autopsies [1][2][3] . PAs are usually benign and slow-growing. However, about 35% of them are invasive pituitary adenomas (IPAs) [4] , which, like malignant tumors, invade structures adjacent to the pituitary gland, such as the cavernous sinus, thereby wrapping internal carotid arteries. All these complications lead to a failure in total resection of IPAs during surgery, which in turn causes rapid recurrence and unfavorable prognosis. Therefore, understanding the molecular mechanisms of IPAs concomitant with identifying novel biomarkers and therapeutic targets is instrumental to improving the efficacy of IPA treatment. In previous studies comparing the molecular mechanisms underlying PA invasiveness, two major proteases, the matrix metalloprotease MMP-14 and a member of the ADAM family of proteases, ADAM12, were identified as critical proteins that determine invasiveness of PAs [5] . Clinically, these proteases are highly correlated with invasive PAs, and their regulation is dependent on estrogens since anti-estrogens or selective estrogen receptor modulators (SERMs) reduce PA invasiveness and, in particular, the expression levels of MMP-14 and ADAM12 [6,7] .
Despite these findings, the precise mechanism of regulating the invasiveness of PA at the genomic level remains unclear. One such level could be the presence of miRNAs (miR) as short, endogenous, and highly conserved non-coding RNAs that contain approximately 19-25 nucleotides and bind to 3′-untranslated regions (3′-UTRs) of the target mRNA, which could lead to mRNA degradation [8] . MiRNAs can also act as tumor suppressor genes or oncogenes in various tumors, manipulating the cell cycle, cell proliferation, differentiation, and cell apoptosis in various cancers [9] . Various studies have recently confirmed that miRNAs participate in different types of PAs. For instance, miR-205-5p inhibits tumor cell proliferation and migration by directly targeting CBX1 [10] . In addition, miR-26a and miR-106b facilitate the invasiveness of PAs [11,12] , and miR-200c inhibits apoptosis of PA cells by targeting the PTEN/Akt signaling pathway [13] .
Given the differential expression levels of miR-149 in invasive vs. noninvasive PAs, we attempted to investigate whether miRNA-149 could regulate proteases MMP-14 and ADAM12 in PA cells. To address this, we used two mouse pituitary adenoma cell lines: AtT-20 and TtT/GF, to investigate the role of miR-149-3p and miR-149-5p in PA cell proliferation, migration, invasion, and the potential regulation of MMP14 and ADAM12.

Ethical Statement
Before conceiving this study, ethical approval was obtained from local authorities (Ethics Committee, Medical Faculty of Marburg University, file number 185/11 and amendment) following the Helsinki Criteria. Informed written consent was obtained from all patients, and their data were pseudonymized.

Patients and Tumor Characteristics
PAs were collected from patients undergoing surgery in the Department of Neurosurgery, University Hospital Marburg. In total, 7 women and 13 men with a mean age of 51.5 years (range 15-84 years) at the time of surgery were included in miRNA analysis. Based on intraoperative findings as described earlier [5] , 10 invasive and 10 noninvasive cases were selected according to the Knosp classification, which accounts for invasion of the cavernous sinus as a measure of invasiveness (i.e., with Knosp grades 0-2 were considered noninvasive, whereas grades 3 and 4 were considered invasive). Cavernous sinus invasion was not observed to correlate with gender, patient age, tumor size, or hormone secretion type (data not shown).

Proliferation Assay
A total of 2000 cells were seeded in a 96-well plate with 100 μL medium and cultured at 37°C. After 24, 48, and 72 h, 50 μL of CellTiter-Glo® (Promega, Germany) was added to each well after shaking for 10 min and incubated for another 20 min at room temperature in the dark. The absorbance (A) at 450 nm of experimental wells was measured using a FLUOstar OPTIMA Microplate Reader (BMG Labtech, Germany).

Colony Formation Assay
Totally, 1000 cells/well of each cell line were cultured in 6-well plates for two weeks after transfection of respective miRNA mimics. After that, cells were washed with PBS once, fixed with 10% formalin (1 mL/well) for 15 min, then washed with PBS again, followed by staining with 0.1% crystal violet (w/w in water) solution for 30 min at room temperature. The number of colonies with more than 50 cells was counted under the light microscope.

Quantitative Real-Time Polymerase Chain Reaction
Both miRNA and total RNA were extracted by miRNeasy Tissue/Cells Advanced Mini Kit (Qiagen) following the manufacturer's instruction. MicroRNA and RNA quantification was performed using a NanoPhotometer NP80 (Implen, Germany). MicroRNAs were reverse-transcribed by miScript Ⅱ RT kit (Qiagen), and cDNA was obtained using the RNA to cDNA Ecodry TM Premix Kit (Takara Bio Europe, France) according to the manufacturer's protocol. Quantitative polymerase chain reaction (qPCR) analyses were performed in triplicates using the Precision FAST MasterMix with ROX (Primer Design, UK). QuantiTect primers for detecting RNA encoding mouse MMP-2, MMP-9, MMP-14, and ADAM12 were obtained from Qiagen. We used XS13 as a housekeeping gene with primers (fwd-TGGGCAAGAACACCATGATG, rev-AGTTTCTC-CAGCTGGGTTG) purchased from Microsynth (Balgach, Switzerland). MiScript SYBR Green PCR kit (Qiagen) was used to analyze the miRNA expression, and RNU6 (Qiagen) was used as a housekeeping gene. A StepOnePlus TM qPCR system (Thermo Fisher Scientific, USA) was employed for the RT-PCR. The relative gene expression was calculated by 2 -ΔΔCT methods.

Antibodies, Protein Extraction, and Western Blotting
Rabbit polyclonal anti-MMP-14 antibody was purchased from Abcam (Cambridge, UK), and rabbit polyclonal anti-ADAM12 antibody from Proteintech (Rosemont, USA). Anti-β-tubulin (Novus Biologicals) served as a loading control, horseradish peroxidase (HRP)-conjugated donkey anti-rabbit immunoglobulin G from Abcam was used as a secondary antibody in a 1:2000 dilution.
Equal amounts of lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred onto nitrocellulose membranes. To block any unspecific binding, membranes were immersed and gently shaken in 5% non-fat, dried milk in Tris-buffered saline Tween-20 [TBST; 50 mmol/L Tris, pH=7.5; 150 mmol/L NaCl; 0.1% Tween-20 (CARL ROTH, Germany)] for 1 h, after which they were incubated with primary antibodies at 4°C overnight. After three washing steps with TBST, blots were incubated with the secondary antibody for 1 h. After another washing step, signals were detected with Western Bright Chemiluminescence Substrate Sirius (Biozym Scientific, Germany) and quantified by Image J.

Transwell Invasion Assay
The 8 μm pore transwell inserts (Greiner Bio-One, Germany) were used with 24-well plates. The inserts were coated on the upper side with 50 μL Matrigel (Corning® Matrigel® Matrix, Basement Membrane Matrix Growth Factor Reduced, Corning Incorporated, Corning, USA). The Matrigel was solid after 1 h at 37°C under 5% CO 2 . The transwell insert was then turned upside down, and 20 000 TtT/GF and 30 000 AtT-20 cells per well in 50 μL medium containing 0.5% FBS were seeded on the other side. After 4 h of adherence, the transwell insert was turned upside up again. To build an FBS gradient, 250 μL medium containing 20% FBS was applied on top of each Matrigel layer, while 750 μL medium containing 0.5% FBS was placed in each of the 24 wells. After 24 h allowed for invasion, cells were fixed with 4% paraformaldehyde (Sigma), treated with 0.3% octoxynol-9 (Triton TM X-100 buffer; Sigma), and stained with 4',6-diamidin-2-phenylindole (DAPI; Sigma). Cells were counted in five randomly chosen viewing fields. The percentage of cells invading the Matrigel was determined.

Wound-Healing Assay
A total of 60 000 cells/well of AtT-20 and 30 000 cells/well of TtT/GF cells were seeded in a cultureinsert (Ibidi culture-insert 2 well, Ibidi GmbH, Germany). After allowing the cells to attach overnight, we removed the culture insert and washed the cells with PBS to remove non-adherent cells. We then provided a fresh culture medium and photographed the plate every 6 h to capture the two different fields at each time point on each plate. The number of cells that migrated into the wound space was manually counted in three fields per well under a light microscope at 50× magnification. We then quantified the areas using image J analysis.

Statistical Analyses
Continuous variables are presented as the mean ± standard error of mean. One-way ANOVA was carried out for multiple comparisons using GraphPad Prism 5.0 software (GraphPad Software, Inc., La Jolla, USA). Welch's t-test was a statistical test, and P-values ≤0.05 indicated a statistically significant difference.

RESULTS
Expression levels of miR-149-3p and miR-149-5p were initially determined by qPCR in PA tumor tissues. To address this issue, PA specimens collected after surgery were classified into noninvasive vs. invasive PA tumors ( fig. 1). Changes between noninvasive and invasive tumors were highly significant (P<0.0001) for both miRNA 149 strands, although expression levels of miRNA-149-3p were higher than those of miRNA-149-5p. In contrast to tumor tissues, endogenous expression levels of miR-149-3p and miR-149-5p in PA cell lines AtT-20 and TtT/GF were comparable with mean CT values of 25 for miR-149-3p and 26 for miR-149-5p ( fig. S1). To investigate the effect of miR-149 on cellular functions with regard to proliferation, migration, and invasion behaviors, mimic miRNAs for miR-149-3p and miR-149-5p were introduced into both cell lines by transfection ( fig. 2). In comparison to untreated cells (NC) or scramble control (Control), cells transfected with mimic miR-149 showed an at least 100-fold higher expression of the respective miR-149 strands ( fig. 2).

DISCUSSION
This study was initiated to assess the functional consequences of different expression levels of the tumor suppressor miRNA-149 between noninvasive and invasive PAs. As a prerequisite for analyses of miRNA-149, we provided evidence that miRNA-149-3p is expressed in significant amounts in PA tissue and is downregulated in invasive PAs. In contrast, the expression levels of miRNA-149-5p in PA tissues are very low and might not be relevant for controlling invasiveness in vivo, although it forms the guide strand. By using miRNA mimics in PA cell lines, we provided evidence that both strands of miR-149 can regulate MMP-14 and ADAM12 expression, both highly associated with the invasive behavior of PAs [3][4][5] . Based on expression levels, only miRNA-149-3p as the passenger strand might be the relevant strand for pathophysiological regulation of ADAM12 and MMP-14 in PA tumors, which is in accordance with the prediction in miR target prediction database (https://mirdb.org). Silencing of Adam12 and Mmp14 by miRNA-149-3p is likely to be achieved by RNA degradation, as lower mRNA levels for both proteases were found as a consequence of mimic miRNA transfection in PA cell lines. Several lines of evidence verified the regulation of MMP expression by miR-149. For instance, Wang et al demonstrated a negative modulation of MMP2 and MMP9 expression by miR-149 in colon carcinoma [20] . In contrast, miR-149 downregulation caused MMP-9 overexpression [21] , suggesting that this miRNA controls the expression of invasion-related proteases. This relationship seems to be common in diverse pathologies. For instance, in cerebral ischemia, MMP expression is increased, and they are a target of miR-149-5p so that in the middle cerebral artery occlusion (MCAO) model, the use of miR-149-5p mimic can reduce the extent of neurological defects and tissue damage by decreasing the level of MMP2 and MMP9 [22] . Another study demonstrated that in the MCAO model, miR-149-5p significantly decreased the expression of MMP9 [23] . In U251 GBM cells, overexpression of miR-149 inhibits expression of MMP2, thereby suppressing proliferation and invasion of GBM via AKT1 signaling; however, in this work, no distinction is made between both strands of miRNA-149 [24] . Since proliferation is affected by mimic miRNA 149, one of the most important target genes of miRNA-149-3p deduced from the miRNA target database is AKT1, so we can hypothesize that this particular pathway could be affected in PA cells by downregulating AKT1. Concerning MMP-2 and MMP-9, we could not detect their downregulation in PA cells (data not shown), suggesting that gene regulation by miR-149 might be cell-specific.
As a limitation of our study, we cannot wholly explain the biological relevance of miRNA-149 on effects apart from invasion since miRNA-149-3p has 1871 predicted target genes. However, AKT1 is one of the primary targets that could be highly relevant for controlling proliferation. Another limitation is that the mouse PA cell lines do not match the human PA tumor tissue, so the effect of miRNA-149-5p, although detectable in PA cell lines, has no pathophysiological function, as the expression levels are very low. Here, human PA cell lines as tumor models, still lacking in PA research, could be valuable tools to solve this discrepancy.
In our study, we also observed that ADAM12, an active protease involved in growth regulation by shedding of EGF ligands such as HB-EGF [25] , is significantly downregulated by miR-149 as predicted by target sequence analysis with complementary seed sequences in the 3′ region of the mouse Adam12 and the human ADAM12 gene. Interestingly, both strands of miRNA-149, 3p, and 5p, are effective in regulating ADAM12 with a preference for miR-149-5p in both cell lines. For MMP-14, both strands of miR-149 were similarly effective in downregulating its expression levels. Similar results were obtained for other tumor entities, such as renal cell carcinoma [15] . In PA tumors, the main regulatory function of miRNA-149-3p on ADAM12 and MMP-14 expression levels might be orchestrated by higher expression levels of miRNA-149-3p, determining their invasive behavior [5] . Since SERMs were previously shown to regulate MMP-14 and ADAM12 in PA cells [5] , we analyzed a potential regulation of miR-149 by SERMs which could indicate a direct involvement of SERMs in miR regulation. While SERMs could reduce expression levels of ADAM12 and MMP-14, we did not detect any changes in miR expression upon treatment of PA cells with SERMs such as Clomiphene ( fig. S2), suggesting that the SERM effects are not mediated by miR-149. While the functional consequences of miR-149 expression for the invasive behavior of PAs are clearly demonstrated here, it remains to be established whether miR-149 could be a diagnostic marker for the malignancy of pituitary adenomas, i.e., by localization of miR-149 in extracellular vesicles.
In conclusion, here we show that both strands of miR-149 (3p and 5p) regulate the expression levels of ADAM12 and MMP14 in pituitary adenoma tumors and PA cell lines, thereby determining their invasive behavior. Our findings have implications for clinical applications, as these can contribute to the identification of novel biomarkers and therapeutic approaches for PAs.

Open Access
This article is licensed under a Creative Commons Attribution 4.0 International License https://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this