miR-26a plays an important role in cell cycle regulation in ACTH-secreting pituitary adenomas by modulating protein kinase Cδ.

The functional aftermath of microRNA (miRNA) dysregulation in ACTH-secreting pituitary adenomas has not been demonstrated. miRNAs represent diagnostic and prognostic biomarkers as well as putative therapeutic targets; their investigation may shed light on the mechanisms that underpin pituitary adenoma development and progression. Drugs interacting with such pathways may help in achieving disease control also in the settings of ACTH-secreting pituitary adenomas. We investigated the expression of 10 miRNAs among those that were found as most dysregulated in human pituitary adenoma tissues in the settings of a murine ACTH-secreting pituitary adenoma cell line, AtT20/D16v-F2. The selected miRNAs to be submitted to further investigation in AtT20/D16v-F2 cells represent an expression panel including 5 up-regulated and 5 down-regulated miRNAs. Among these, we selected the most dysregulated mouse miRNA and searched for miRNA targets and their biological function. We found that AtT20/D16v-F2 cells have a specific miRNA expression profile and that miR-26a is the most dysregulated miRNA. The latter is overexpressed in human pituitary adenomas and can control viable cell number in the in vitro model without involving caspase 3/7-mediated apoptosis. We demonstrated that protein kinase Cδ (PRKCD) is a direct target of miR-26a and that miR26a inhibition delays the cell cycle in G1 phase. This effect involves down-regulation of cyclin E and cyclin A expression via PRKCD modulation. miR-26a and related pathways, such as PRKCD, play an important role in cell cycle control of ACTH pituitary cells, opening new therapeutic possibilities for the treatment of persistent/recurrent Cushing's disease.

Pituitary adenomas, accounting for ∼10% of intracranial tumors, can be divided in endocrinologically inactive nonfunctioning pituitary adenomas, associated with mass effects, and functioning pituitary tumors, associated with specific clinical syndromes, due to pituitary hormones (prolactin, GH, ACTH, and TSH) excess (1). They are mostly benign, but 5% to 35% of them are locally invasive and are difficult to treat. Among the latter, ACTH-secreting adenomas, which account for 5% to 10% of diagnosed pituitary tumors, frequently persist or recur, and therefore their clinical control is hard to achieve (2–4). A better understanding of the pathogenesis of ACTH-secreting pituitary tumors may identify new pharmacological targets and prompt the development of innovative drugs, with a better chance to control the disease. However, the molecular alterations that may underpin the development and progression of ACTH-secreting pituitary tumors are not completely clear yet. Several hypotheses have been put forward on the basis of molecular studies that led to the identification of many mutations or abnormal mRNA/protein expression patterns (5–8). Recently, microRNA (miRNA) dysregulation has also been referred to as a possible pathogenic mechanism (9).

The miRNAs are highly conserved, noncoding RNAs that powerfully regulate gene expression, reducing mRNA transcription or translation (10, 11), and may play an important role in cell differentiation, apoptosis, and proliferation (12, 13). Recent evidence suggests the potential involvement of miRNAs in human tumor development because they can act as oncogenes or tumor suppressor genes (13–19). We previously demonstrated that miR-15a and miR-16-1 are downregulated in pituitary adenomas (18), suggesting for the first time the role of miRNAs in pituitary tumorigenesis. Moreover, we have identified 29 miRNAs differentially expressed between pituitary adenomas and normal pituitary. Among them, 22 are downregulated (miR-24-1, -144, -009-1, -98, -124a-2, -136, -141, -101-2, 009-2, -015-b, -018, -096, -140, -001,-217, -153-2, -10a, -181b, -144, -21, and -153-1 and let-7g) and 7 are upregulated (miR-212, -026a, -191, -152, -149, -150, and -192) (9, 19).

With regard to ACTH-secreting pituitary adenomas, there is accumulating evidence (9, 19–22) indicating that miRNAs are dysregulated also in these tumors, some being downregulated (let-7a, miR-15a, miR-16, miR-21, miR141, miR-143, miR-145, and miR-150) and others upregulated (miR-122 and miR-493). However, to date, the functional aftermath of miRNA dysregulation in ACTH-secreting pituitary adenomas has not been demonstrated. To address this issue, identification of the miRNA targets is crucial.

Because immortalized human pituitary adenoma cell lines are not available, we investigated the expression of 10 miRNAs among those that were found as the most dysregulated in human pituitary adenoma tissues (19) in the setting of a murine ACTH-secreting pituitary adenoma cell line, the AtT20/D16v-F2 cells. The selected human miRNAs to be submitted to further investigation in the mouse cell line represent an expression panel including 5 upregulated (miR-26a, miR-192, miR-150, miR-212,and miR-191) and 5 down-regulated (miR-124a-2, miR-181b, miR-153-2, miR-98, and miR-024-1) miRNAs. Afterward, among these, we selected the most dysregulated mouse miRNA and set out to find miRNA targets and their biological function.

Materials and Methods

Cell culture

The mouse ACTH-secreting pituitary adenoma cell line, AtT-20/D16v-F2, was obtained from the American Type Culture Collection (Manassas, Virginia). The cell line was maintained in DMEM (Invitrogen, Carlsbad, California) supplemented with 10% horse serum (LGC Standards, Milano, Italy) and antibiotic antimycotic (EuroClone, Milano, Italy). The cells were maintained at 37°C in 5% CO2.

Tissue collection

Tissue samples of human ACTH-secreting pituitary adenomas were collected in accordance with the guidelines of the local committee on human research. Tissue fragments were immediately frozen in liquid nitrogen under ribonuclease-free conditions at the time of surgery and stored at −80°C until protein/RNA isolation was performed. Frozen tissues were disrupted using a dismembrator (B. Braun Biotech International, Milano, Italy).

RNA isolation

Total RNA was extracted with TRIzol reagent (Invitrogen, Milano, Italy) according to the manufacturer's protocol. RNA preparations were treated with ribonuclease-free deoxyribonuclease (Promega, Milano, Italy) to remove DNA contamination. The Experion automated electrophoresis system (Bio-Rad, Milano, Italy) was used to determine the concentration and the integrity of RNA samples. Only samples with RNA quality index >9 were reverse transcribed. An RNA pool from a normal mouse and one from normal human pituitary tissues were purchased from Ambion (Invitrogen).

miRNA reverse transcription and real-time PCR analysis

cDNA was synthesized using the TaqMan MicroRNA reverse transcription kit (Applied Biosystems, Monza, Italy), following the manufacturer's instructions. The miRNA transcripts were reverse transcribed into cDNA using miRNA gene-specific primer sets provided with the TaqMan MicroRNA Assays (Applied Biosystems).

Specific mature miRNA levels were assessed by relative quantitative real-time PCR (RQ-PCR). All the QPCRs were conducted with TaqMan MicroRNA assays (hsa-miR-26a PN4373070, mmu-miR-192 PN4373308, hsa-miR-150 PN4373127, hsa-miR-212, PN4373087, hsa-miR-191 PN4373109, mmu-miR-124a PN4373295, hsa-miR-181b PN4373116, mmu-miR-153 PN4373305, hsa-miR-98 PN4373009, miR-189 ID000488, and hsa-miR-24 PN4373072), run on Applied Biosystems 7700 ABI Prism thermal cycler and analyzed with SDS version 1.9 software (Applied Biosystems). Sno234 was identified as the most stable reference gene (sno234 assay 001234, Applied Biosystems) by geNorm software, version 3.4 (23). Relative expression ratios of miRNAs were calculated by applying the method described by Pfaffl et al (24). Results are expressed as mean value ± SE percent miRNA expression vs normal pituitary from at least 3 independent experiments in 3 replicates.

Bioinformatics analysis

Putative miRNA targets were predicted using the programs PicTar (25), miRanda (26), and Targetscan (27). To minimize the number of false predictions, only miRNA targets predicted using all 3 approaches were considered.

Total RNA reverse transcription and real-time PCR analysis

cDNA was synthesized from total RNA using the first-strand cDNA synthesis kit (Invitrogen), following the manufacturer's instructions.

Gene expression evaluation was performed by RQ-PCR. All the QPCRs were conducted with the TaqMan gene expression assay (Applied Biosystems), run on Applied Biosystems 7700 ABI Prism thermal cycler and analyzed with SDS version 1.9 software (Applied Biosystems). Human glyceraldehyde 3-phosphate dehydrogenase and mouse cyclophilin were identified as the most stable reference genes by geNorm software, version 3.4 (23). Relative expression ratios of SUMO1/sentrin-specific peptidase 5 (SENP5) (Mm00769150_m1 and Hs01090047_m1; Applied Biosystems) and protein kinase Cδ (PRKCD) (Mm00440891_m1 and Hs01090047_m1; Applied Biosystems) mRNAs were calculated by applying the method described by Pfaffl et al (24). Results are expressed as mean value ± SE percent mRNA expression vs normal pituitary from at least 5 independent experiments in 5 replicates.

Construction of 3′-UTR reporter plasmids and luciferase assays

The putative miR-26a recognition sequences from the 3′-untranslated region (UTR) of human PRKCD and SENP5, including XbaI cohesive ends, were obtained by oligo annealing of the following oligonucleotide pairs: 5′-CTAGATTCAAGTAATGAAACACAGGAT-3′ and 5′-CTAGATCCTGTGTTTCATTA-CTTGAA-3′ (PRKCD) and 5′-CTAGATTCAAGTAACAAACCATCTGCT-3′ 5′-CTAGAGCAGATGGTTTGTTACTTGAA-3′ (SENP5). The overhangs created by oligonucleotide annealing were complementary to those generated by XbaI digestion of the pmirGLO vector (Promega), in which the annealed oligonucleotides were cloned 3′ of the firefly luciferase gene (PRKCD-pmirGLO and SENP5-pmirGLO). The inserts from all cloned plasmids were sequenced to verify identities (5′-AGGCGATTAAGTTGGGTA-3′).

AtT-20/D16v-F2 cells were seeded at a cell density of 104 cells per well in 96-well black plates 1 day before transfection. The following day, cells were transfected using TransFast transfection reagent (Promega) according to the manufacturer's instructions, with 0.2 μg PRKCD-pmirGLO or SENP5-pmirGLO in the presence of 30nM pre–miR-26a (product ID PM10249; Ambion, Invitrogen) with or without 30nM anti–miR-26a (product ID AM10249, Ambion, Invitrogen). Pre-miR miRNA precursor negative control (Ambion, Invitrogen) and anti–miR-miRNA inhibitor negative control no. 1 (Ambion) were used as specific control oligos. Luciferase assays were performed 48 hours after transfection using the Dual-Glo luciferase assay (Promega) following the manufacturer's instructions. pmirGLO vector is based on Promega dual-luciferase technology, with firefly luciferase used as the primary reporter to monitor mRNA regulation and renilla luciferase acting as a control reporter for normalization.

Western blot analysis

AtT-20/D16v-F2 cells were seeded at a cell density of 5 × 105 cells in 60-mm plates. After 24 hours, cells were transfected with 30nM pre–miR-26a (Ambion), 30nM anti–miR-26a (Ambion), and 30nM specific control oligos (Ambion). Protein isolation was performed as described previously after 48 hours from transfection (28).

For human ACTH-secreting pituitary adenoma protein evaluation, tissues were lysed as previously reported (29). Total protein extracts were measured by the BCA Protein Assay Reagent Kit (Pierce Biotechnology, Inc, Rockford, Illinois), and 30 μg of proteins were fractionated on SDS-PAGE and transferred by electrophoresis to nitrocellulose membranes (Schleicher & Schuell Italia SRL, Milano, Italy). The following primary antibodies were employed: 1/200 rabbit polyclonal antihuman PRKCD (Santa Cruz Biotechnology, Santa Cruz, California), 1/500 rabbit polyclonal antibody to SENP5 (Histo-Line Laboratories, Milano, Italy), 2 μg/mL mouse monoclonal cyclin A (Abcam, Cambridge, United Kingdom), 1/1000 rabbit polyclonal cyclin E (Abcam), 1/2000 rabbit monoclonal antihuman phospho–c-Jun N-terminal kinase (JNK) (Cell Signaling Technology Inc, Danvers, Massachusetts), 1/1000 rabbit polyclonal antihuman total JNK (Millipore, Billerica, Massachusetts), 1/1000 rabbit p27 Kip1 Xceptional Performance (Cell Signaling Technology Inc), 1/500 anti-p27 Kip1 (phospho-T187) (Abcam), and 1/1000 rabbit β-actin antibody (Cell Signaling Technology). Immunoreactive bands were detected using species-specific horseradish peroxidase-conjugated secondary antibodies (Dako Italia, Milano, Italy) and visualized using the enhanced chemiluminescence Western blotting detection reagents (GE Healthcare Europe GmbH, Milano, Italy). Protein bands were quantified by using ImageJ free software (http://rsbweb.nih.gov/ij/). Protein levels were measured as the ratio between the protein OD and the actin OD of the same sample and expressed as arbitrary units (mean ± SE).

Viable cell number assay

Viable cell number was assessed by the ATPlite assay (PerkinElmer, Waltham, Massachusetts). Briefly, the cells were seeded at 9 × 103 cells per well in 96-well white plates and transfected with indicated oligonucleotides. After 48 hours, an equal volume of reconstituted substrate solution was added at room temperature directly to the cell culture plates. The plates were shaken at 700 rpm for 2 minutes and then measured for luminescent output (relative light units [RLU]) by Victor3 1420 multilabel counter (PerkinElmer). Results are expressed as mean value ± SE percent viable cell number vs mock-transfected cells from 5 independent experiments in 6 replicates.

Caspase activity

Caspase activity was measured 48 hours after transfection using Caspase-Glo 3/7 assay (Promega) following the manufacturer's instruction as described previously (30). Results are expressed as mean value ± SE percent RLU vs mock-transfected cells from 5 independent experiments in 6 replicates.

Cell cycle analysis

Cell cycle phase distribution analysis was performed by flow cytometry after DNA staining as previously reported (31).

Transfection with PRKCD shRNA

Cells were transfected with PRKCD shRNA plasmid (OriGene, Rockville, Maryland), by using Nucleofector II (Amaxa, Gaithersburg, Maryland) according to the manufacturer's protocols. Then, stably transfected clones were selected by incubation in medium containing 2 μg/mL puromycin. Subsequently, PRKCD knockdown cells were transfected with anti–miR-26a, as described above.

Statistical analysis

Results are expressed as ± SE and analyzed statistically using Student's t tests to evaluate individual differences between means. To measure the strength of association between pairs of variables without specifying dependencies, Spearman order correlations were run. P < .05 was considered significant in all tests.

Results

miRNAs are differentially expressed in ACTH-secreting pituitary adenomas

We applied RQ-PCR to examine in a mouse ACTH-secreting pituitary adenoma cell line, AtT20/D16v-F2, the expression of miRNAs that were found as dysregulated in human pituitary adenomas. We found that miR-124a, miR-24, miR-191, and miR-212 were downregulated, whereas miR-181b and miR-26a were upregulated in mouse ACTH-secreting pituitary adenoma cells as compared with mouse normal pituitary, and miR-98, miR-192, miR-153 and miR-150 showed similar expression as compared with mouse normal pituitary. As shown in Figure 1, miR-26a was the most dysregulated miRNA (+300% vs normal pituitary) and was therefore selected for further investigation.

Figure 1.

miRNA expression levels in AtT20/D16v-F2 cells. Mouse miRNA levels were assessed by RQ-PCR. The fold changes (± SE) are expressed as percent mRNA expression vs normal pituitary. At least 3 independent experiments were performed in 3 replicates. **P < .01 vs. normal pituitary.

PRKCD and SENP5 are identified as miR-26a putative target genes

We screened different databases that predict potential miR-26a target genes using diverse computational algorithms. We focused on potential targets that might be involved in pituitary adenoma development and selected PRKCD and SENP5 among the genes identified as putative targets of miR-26a.

PRKCD and SENP5 mRNA expression

PRKCD and SENP5 mRNA expression levels were assessed in AtT20/D16v-F2 cells and in human ACTH-secreting pituitary adenoma tissues. The miR-26a–predicted target gene PRKCD was equally expressed in AtT20/D16v-F2 cells (Figure 2A) as well as in 60% of the investigated human ACTH-secreting pituitary adenomas (Figure 2B) as compared with the mouse and human normal pituitary tissues, respectively. On the other hand, PRKCD was overexpressed (+400%–550% vs normal pituitary) in 40% of the examined tumors (Figure 2B).

Figure 2.

PRKCD and SENP5 mRNA expression. A and B, PRKCD basal mRNA expression levels were assessed in AtT20/D16v-F2 cells (A) and human ACTH-secreting pituitary adenoma tissues (B) by RQ-PCR. C and D, SENP5 basal mRNA expression levels were assessed in AtT20/D16v-F2 cells (C) and human ACTH-secreting pituitary adenoma tissues (D) by RQ-PCR. Results are expressed as mean fold of induction ± SE percent mRNA expression vs normal pituitary from at least 5 independent experiments in 5 replicates. **P < .01 vs normal pituitary.

The miR-26a–predicted target gene SENP5 was downregulated in AtT20/D16v-F2 cells (Figure 2C) as well as in 60% of the investigated human ACTH-secreting pituitary adenomas (Figure 2D) as compared with the mouse and human normal pituitary tissues, respectively. On the other hand, SENP5 was overexpressed (+142%–220% vs normal pituitary) in 40% of the examined tumors (Figure 2D).

Because miRNA may regulate gene expression not only by promoting mRNA degradation but also by inhibiting translation, we assessed whether miR-26a binds the 3′-UTR of PRKCD and SENP5 mRNAs by luciferase assay.

PRKCD is a direct target of miR-26a

The presence of the predicted miR-26a target site in the 3′-UTR of both human and mouse PRKCD and SENP5 mRNAs supports the hypothesis that miR-26a regulates PRKCD and SENP5 expression. To test this hypothesis, we cloned each putative 3′-UTR target site downstream of the luciferase reporter gene of the pmirGLO vector (PRKCD-pmirGLO and SENP5-pmirGLO, respectively). As controls, antisense sequences were used. Each construct was cotransfected in AtT20/D16v-F2 cells together with either the pre–miR-26a, the anti–miR-26a or the specific negative controls.

As shown in Figure 3A, pre–miR-26a transfection did not modify pmirGLO vector luciferase activity. On the other hand, pre–miR-26a transfection significantly reduced luciferase activity by 55% in PRKCD-pmirGLO cotransfectants as compared with pmirGLO vector-only cotransfectants. This effect was completely reversed by anti–miR-26a cotransfection. In addition, pre–miR-26a transfection did not modify luciferase activity in antisense PRKCD-pmirGLO cotransfectants as compared with pmirGLO vector-only cotransfectants, demonstrating the specificity of PRKCD silencing by miR-26a. These data demonstrate that miR-26a binds to the 3′-UTR of the PRKCD gene, regulating its expression.

Figure 3.

PRKCD is a direct target of miR-26a. Panel A, AtT20/D16v-F2 cells were transfected with a luciferase gene linked to the 3′-UTR of PRKCD (PRKCD pmirGLO) or the antisense sequence (antisense PRKCD pmirGLO) and cotransfected with pre–miR-26a, in the presence or absence of anti–miR-26a. Panel B, The same experiments were performed for SENP5 by using SENP5 pmirGLO or antisense SENP5 pmirGLO vectors. Results are expressed as mean value ± SE percent RLU vs vector only (pmirGLO) from 3 independent experiments in 6 replicates. **P < .01 vs pmirGLO vector. Panel C, AtT20/D16v-F2 cells were transfected with pre–miR-26a (P), anti–miR-26a (A), or specific control oligos (cP and cA). Cell extracts were harvested 48 hours after transfection and assayed for PRKCD and SENP5 protein expression by Western blot. Data presented were confirmed in 3 independent experiments. Panel D, PRKCD basal protein expression levels were assessed by Western blot, and miR-26a expression levels were assessed by RQ-PCR in human ACTH-secreting pituitary adenoma tissues. Results are expressed as mean percent expression levels ± SE vs normal pituitary from at least 3 replicates. **P < .01 vs normal pituitary. In the lower panel are representative Western blots showing PRKCD protein levels in human normal pituitary and in human ACTH-secreting pituitary adenomas. Abbreviation: C, control cells.

As shown in Figure 3B, pre–miR-26a transfection did not modify luciferase activity of SENP5-pmirGLO cotransfectants, without or with anti–miR-26a, and of antisense SENP5-pmirGLO cotransfectants, as compared with pmirGLO vector-only cotransfectants. Therefore, the hypothesis that SENP5 is a target of miR-26a is ruled out.

To further verify that miR-26a directly regulates PRKCD, but not SENP5, both protein expression levels were assayed in AtT20/D16v-F2 cells transfected with pre–miR-26a, anti–miR-26a, or specific negative controls. As shown in Figure 3C, pre–miR-26a transfection did not significantly modify basal protein levels of both PRKCD and SENP5. On the other hand PRKCD (+125% ± 25% vs control; P < .01), but not SENP5, protein levels were increased by anti–miR-26a transfection compared with negative controls. Taken together, Western blot and luciferase data demonstrate that PRKCD is a target of miR-26a.

miR-26a inversely correlates with PRKCD protein in human ACTH-secreting pituitary adenomas

To investigate the correlation between miR-26a levels and PRKCD protein expression in human ACTH-secreting pituitary adenomas, we performed RQ-PCR to assess miR-26a levels in a series of human ACTH-secreting pituitary adenomas. In keeping with previous results, we found that in all evaluated tissues, miR-26a was expressed at significantly higher levels as compared with normal pituitary (from +48% to +163%; P < .01). In the same tissues, Western blot analysis showed that PRKCD protein was expressed at significantly lower levels as compared with normal pituitary (from −50% to −65%; P < .01). Moreover, we found a significant negative correlation between miR-26a levels and PRKCD protein levels (r = 0.9573; P < .01).

miR-26a influences ACTH-secreting pituitary adenoma viable cell number

To explore the effect of miR-26a on viable cell number, AtT20/D16v-F2 cells were transiently transfected with pre–miR-26a or anti–miR-26a. Specific control oligos were used as controls. As shown in Figure 4A, pre–miR-26a did not influence viable cell number, whereas anti–miR-26a significantly reduced this parameter (−24% vs control). By contrast, the control oligos did not induce significant effects, indicating that miR-26a specifically influences viable cell number. These data indicate that miR-26a levels directly affect the mechanisms regulating pituitary adenoma cell survival.

Figure 4.

Anti–mir-26a reduces AtT20/D16v-F2 viable cell number without activating caspase-mediated apoptosis. AtT20/D16v-F2 cells were transfected for 48 hours with pre–miR-26a (P), anti–miR-26a (A), pre-miR precursor negative control (cP), and anti–miR-miRNA inhibitor negative control no. 1 (cA). Panel A, Viable cell number was assessed in at least 5 independent experiments with 6 replicates each and is expressed as the mean value ± SE percent viable cell number vs control cells. *P < .05 vs control cells. Panel B, Caspase 3/7 activity was assessed in at least 5 independent experiments with 6 replicates each and is expressed as the mean value ± SE percent caspase 3/7 activity vs control cells. Panel C, Representative Western blot showing caspase 3 expression levels and cleavage. The + sign represents positive control (staurosporine treatment).

miR-26a does not influence caspase-mediated apoptosis

To investigate whether the reduction in viable cell number caused by anti–miR-26a transfection is due to apoptosis activation, caspase 3/7 activity was measured in AtT20/D16v-F2 cells transfected with pre–miR-26a, anti–miR-26a, or control oligos. As shown in Figure 4B, all tested oligonucleotides did not significantly modify caspase activation.

To confirm that miR-26a does not influence caspase 3 cleavage and to exclude possible effects on caspase 3 expression, AtT20/D16v-F2 cells were transfected with pre–miR-26a, anti-miR26a, or specific control oligos, and caspase 3 expression and activation were assessed by Western blot. As shown in Figure 4C, none of the tested oligos modified caspase 3 expression and cleavage.

These results suggest that anti–mir-26a reduces AtT20/D16v-F2 viable cell number independently of caspase-mediated apoptosis.

miR-26a affects G1/S transition via PRKCD

Because anti–miR-26a transfection causes viable cell number reduction but not apoptosis activation, we explored cell cycle progression in AtT20/D16v-F2 cells transfected with pre–miR-26a, anti–miR-26a, or specific control oligos. No significant difference in cell cycle phase distribution was observed in cells transfected with pre–miR-26a as compared with those transfected with pre-miR control. On the other hand, AtT20/D16v-F2 cells transfected with anti–miR-26a displayed an 8% increase in G1 phase compared with those transfected with anti-miR control (P < .05) (Figure 5A). These results suggest that the inhibitory effect of anti–miR-26a on viable cell number could be due to a delay in G1 phase.

Figure 5.

Anti–miR-26a induces an accumulation in G1 cell cycle phase. AtT20/D16v-F2 cells were transfected for 48 hours with pre–miR-26a (P), anti–miR-26a (A), pre-miR precursor negative control (cP), and anti–miR-miRNA inhibitor negative control no. 1 (cA). Panel A, Cell-cycle analysis. The graph shows representative data of 3 independent experiments, which were repeated 3 times. Panel B, Representative Western blot showing the protein levels of cyclin E, cyclin A, total JNK, phospho-JNK, total p27, phospho-p27, and β-actin as loading control. Panel C, Representative Western blot showing the protein levels of cyclin E, cyclin A, total JNK, phospho-JNK, total p27, phospho-p27, and β-actin as loading control in AtT20/D16v-F2 cells treated without or with rottlerin (+). Panel D, Representative Western blot showing the protein levels of cyclin E, cyclin A, total JNK, phospho-JNK, total p27, phospho-p27, and β-actin as loading control, in control mock-transfected (−) and in PRKCD-silenced (+) AtT20/D16v-F2 cells. Abbreviation: C, control cells.

To test this hypothesis, the most important proteins that are known to regulate G1/S transition and to interact with PRKCD (32–36) were analyzed. Protein levels of cyclin E and cyclin A as well as total and phosphorylated JNK and p27 were investigated after transfection with pre–miR-26a or anti–miR-26a. As shown in Figure 5B, no significant difference in protein levels was observed in cells transfected with pre–miR-26a as compared with those transfected with pre-miR control, except for p27. Indeed, phosphorylated p27 levels increased ∼2-fold as compared with mock-transfected control.

On the contrary, cyclin E and cyclin A expression was reduced by anti–miR-26a transfection (∼5-fold and ∼10-fold, respectively). Moreover, the latter reduced JNK phosphorylation by ∼10-fold (T183/Y185) without affecting total protein levels. Similarly, anti–miR-26a transfection reduced p27 phosphorylation (T187) by 75%. Because total p27 levels are not affected, phosphorylation in T187 may not be sufficient for targeting p27 to protein degradation.

Next, we investigated whether the effects of anti–miR-26a on cell cycle proteins are mediated by PRKCD. To this aim, the cells were treated for 24 hours with 3μM rottlerin, a PRKCD inhibitor (37), and were transfected with anti–miR-26a. As shown in Figure 5C, rottlerin pretreatment produced an increase in cyclin A and in phosphorylated p27 protein levels in mock-transfected cells (by ∼10% and by ∼80%, respectively). Similar results were obtained in rottlerin-pretreated cells transfected with anti–miR-26a. Therefore, in the presence of rottlerin, anti–miR-26a fails to downregulate cyclin A and E as well as JNK and p27 phosphorylation. These data support the hypothesis that anti–miR-26a effects are mediated, at least in part, by PRKCD.

To confirm the observed effects, AtT20/D16v-F2 cells were stably transfected with PRKCD shRNA and then transfected with anti–miR-26a. Figure 5D shows that PRKCD silencing abrogates the effects of anti–miR-26a on cyclin E and cyclin A expression, in keeping with the results obtained with rottlerin. On the other hand, p27 and JNK phosphorylation is not detectable in PRKCD-silenced cells, both in the absence and in the presence of anti–miR-26a. Therefore, our data demonstrate that the consequences of miR-26a inhibition are due, at least in part, to PRKCD expression modulation.

Discussion

Several studies investigated the importance of miRNAs in the pathogenesis of pituitary adenomas and identified miRNAs as diagnostic and prognostic biomarkers as well as putative therapeutic targets (38). This issue may be of great importance, especially in the settings of ACTH-secreting pituitary adenomas, that are among the most difficult to treat. Indeed, after surgery, second-line treatments (radical surgery, radiation therapy, medical therapy, and bilateral adrenalectomy) are frequently needed (2) but cannot always control the disease. Cushing's disease is associated with significant morbidity and mortality (39); therefore, finding an effective treatment is crucial to achieve this goal. In these settings, miRNAs can represent a therapeutic target and their investigation may shed light on the mechanisms that underpin pituitary adenoma development and progression. Therapeutic interventions aiming at regulating such newly identified pathways may help in achieving disease control.

The absence of available human pituitary adenoma cell lines makes it extremely difficult to understand the processes in which miRNAs are involved. In this study, using a mouse ACTH-secreting cell line, we profiled the expression levels of 10 miRNAs previously reported as the most dysregulated in human pituitary adenomas (19). Our data confirm the results of Stilling et al (22) that reported miR-212 as downregulated in ACTH-secreting pituitary adenomas versus normal tissue. Moreover, we identified miR-24 and miR-189 as downregulated and miR-26a as overexpressed in both human (19) and mouse pituitary adenomas versus normal tissue. All the other investigated miRNAs show different expression levels between human and mouse. Probably, the poor concordance mirrors the comparison between a human pool of different pituitary adenoma histotypes and a mouse ACTH-secreting adenoma cell line. Nevertheless, we focused our attention on miR-26a, which is the most dysregulated miRNA at the pituitary level and is highly conserved in mouse and humans.

We indeed demonstrate that miR-26a is overexpressed in a murine ACTH-secreting pituitary adenoma cell line, in keeping with previous studies showing miR-26a overexpression in human pituitary adenomas (19) and that PRKCD is a functional target of miR-26a. We found that miR-26a overexpression does not correspond to a reduction in PRKCD mRNA levels in AtT20/D16v-F2 cells and in most human ACTH-secreting pituitary adenomas, suggesting that the regulation of PRKCD mRNA levels does not depend on miR-26a. On the other hand, miR-26a may inhibit PRKCD mRNA translation, as supported by the evidence that miR-26a downregulation by anti–miR-26a causes an increase in PRKCD protein levels in AtT20/D16v-F2 cells. This hypothesis is further strengthened by the evidence that mir-26a levels inversely correlate with PRKCD protein levels in human ACTH-secreting pituitary adenomas.

PRKCD is a serine/threonine kinase involved in different physiological processes, most prominently proliferation, apoptosis, and cell cycle regulation (40, 41). The importance of PRKCD in the control of cell growth has been demonstrated by previous studies showing that loss of PRKCD induces cell proliferation in B cells (42), glial cells (43), and vascular smooth muscle and endothelial cells (44). On the basis of these studies, we investigated whether miR-26a, which targets PRKCD, influences ACTH-secreting pituitary adenoma viable cell number. However, we found that transfection with pre–miR-26a does not influence viable cell number, suggesting that a further increase in miR-26a levels, which are already very high, does not influence cell homeostasis. On the other hand, transfection with anti–miR-26a reduces viable cell number, in agreement with the increased PRKCD levels. These data indicate that miR-26a is important for viable cell number maintenance, because its downregulation causes a reduction in viable cell number.

It is well established that PRKCD regulates apoptosis in various systems (45). Thus, we investigated whether the reduction in viable cell number induced by anti–miR-26a could be due to apoptosis activation. Anti–miR-26a does not influence either caspase 3/7 activity or procaspase 3 expression and cleavage, indicating that miR-26a does not influence basal apoptotic rate in AtT20/D16v-F2 cells. These results are in agreement with the observation that PRKCD activation occurs in response to apoptotic stimuli, such as signals initiated by the death receptor, UV radiation, and etoposide (44). Altogether, these data suggest that, in the absence of proapoptotic stimuli, anti–miR-26a reduces viable cell number without involving apoptosis.

It has been previously demonstrated that PRKCD influences cell cycle progression by inducing cell cycle arrest (39, 46). We here demonstrate that anti–miR-26a, which up-regulates PRKCD protein expression, induces an accumulation in G1 phase, in line with previous reports indicating PRKCD as a mediator of cell cycle arrest in G1 phase (39). Starting from these observations, we investigated the expression of the most important proteins that are known to regulate G1/S transition and to interact with PRKCD (32–36). The G1/S phase transition requires cyclin D complexed with cyclin-dependent kinase (cdk)-4/6, followed by induction of cyclin E complexed with cdk2; each cyclin-activated kinase phosphorylates retinoblastoma, releasing E2F transcription factor 1-3 (E2F1-3) that promote DNA synthesis. The cdk2-cyclin A complexes appear during late S phase and also play a role in the progression toward DNA replication. It has been previously demonstrated that all G1 cyclins are inhibited by PRKCD, which blocks cell cycle progression (47), similar to what was observed in our experimental model. Indeed, miR-26a downregulation, which corresponds to PRKCD increased levels, results in a cell cycle accumulation in G1 phase by reducing both cyclin E and cyclin A expression. This effect is reduced by both a PRKCD inhibitor and PRKCD silencing, confirming that PRKCD activity is important in the control of cell cycle progression and that it mediates anti–miR-26a effects.

The promoting activity of cdks toward cell cycle progression is inhibited by the interaction with small proteins such as p15, p16, p19, p21, and p27. It has been previously demonstrated that p27 overexpression causes G1 arrest due to the inhibition of multiple cdks, including cyclin E-cdk2 (48). In murine fibroblasts, this complex directly phosphorylates p27 (T187), resulting in protein degradation, allowing transit from G1 to S phase (47). In thyroid cancer cells, PRKCD activation induces G1 cell growth arrest, through a MAPK-p27-cyclin E-phosphorylated retinoblastoma pathway (49). Similarly, in our study, PRKCD increased expression induced by anti–miR-26a reduces p27 phosphorylation (T187), which is rescued by rottlerin treatment. On the contrary, phosphorylated p27 protein is undetectable both in control PRKCD silenced cells and in PRKCD silenced cell transfected with anti–miR-26a. Therefore, it is not possible to assess whether the inhibitory action of anti–miR-26a on p27 phosphorylation is mediated by PRKCD or not. Additional studies are necessary to address this issue.

The involvement of the MAPK cascade in the regulation of cell proliferation is widely consolidated. Among MAPK proteins, JNK has been demonstrated to modulate cell cycle progression in the absence of external stimuli (33). Here we demonstrate that anti–miR-26a reduces JNK phosphorylation and that this effect is reversed by a PRKCD inhibitor. The evidence that JNK is a downstream effector of PRKCD (50) further strengthens our data showing that miR-26a affects JNK activity by acting through PRKCD. Indeed, PRKCD inhibition by rottlerin reverses the inhibitory effect of anti–miR-26a on JNK phosphorylation. On the contrary, phosphorylated JNK protein is undetectable both in control PRKCD-silenced cells and in PRKCD-silenced cells transfected with anti–miR-26a. Therefore, it is not possible to assess whether the inhibitory action of anti–miR-26a on JNK phosphorylation is mediated by PRKCD or not. Additional studies are necessary to address this issue.

Taken together, these data demonstrate that miR-26a directly targets PRKCD, which, in turn, mediates miR-26a effects on an ACTH-secreting pituitary adenoma cell line. However, it is well known that each miRNA regulates a large fraction of protein-coding genes. The validated targets of human miR-26a in the database of experimentally supported targets (51) are SMAD1 (mothers against decapentaplegic homolog 1) (52), EZH2 (enhancer of zeste homolog 2) (53), and PLAG1 (pleiomorphic adenoma gene 1) (54), which could influence the observed effects of miR-26a downregulation by anti–miR-26a (ie, AtT20/D16v-F2 viable cell number reduction). On the other hand, the present study demonstrates that PRKCD specifically mediates the effects of miR-26a downregulation by anti–miR-26a in terms of cell cycle regulator expression. Therefore, PRKCD may be considered as a new candidate target for medical therapy in the treatment of ACTH-secreting pituitary adenomas.

We have previously shown that miR-26a is overexpressed at highly variable levels in all human pituitary adenoma subtypes, including ACTH-secreting tumors, which do not significantly differ from other histotypes in miR-26a overexpression (19). This evidence may indicate that targeting the mechanisms regulated by miR-26a might represent a possible new strategy for medical treatment of pituitary adenomas not only of corticotroph origin. However, additional studies are needed to confirm this hypothesis.

In conclusion, our results demonstrate that the mouse ACTH-secreting pituitary adenoma cell line, AtT20/D16v-F2, shows a specific miRNA expression profile that, at least in part, is shared by human pituitary adenomas. miR-26a is overexpressed in human pituitary adenomas and can control viable cell number in the in vitro model without involving caspase 3/7-mediated apoptosis. We validated PRKCD as a direct target of miR-26a and demonstrated that miR-26a inhibition delays cell cycle in G1 phase. The latter effect involves downregulation of cyclin E and cyclin A expression via PRKCD modulation. Therefore, miR-26a and related pathways, such as PRKCD, may play an important role in cell cycle control of ACTH pituitary cells, opening new therapeutic possibilities for the treatment of persistent/recurrent Cushing's disease.