Metformin inhibits growth and prolactin secretion of pituitary prolactinoma cells and xenografts

Abstract Metformin (MET) is a diabetes drug that activates AMP‐activated protein kinase (AMPK), and is suggested to have anticancer efficacy. Here, we investigated the role of AMPK signalling in prolactinoma (PRLoma), with particular respect to MET and bromocriptine (BC) as a PRLoma treatment. We analysed AMPK phosphorylation, dopamine D2 receptor (D2R), and oestrogen receptor (ER) expression in both BC‐sensitive and ‐resistant PRLoma samples; effects of the AMPK agonist MET (alone or with BC) on in vitro proliferation and apoptosis, xenograft growth and prolactin (PRL) secretion of BC‐sensitive and ‐resistant cells, and ER expression in xenografts. Some BC‐resistant PRLomas showed high D2R expression but extremely low AMPK activation. MET significantly inhibited proliferation of cultured PRLoma cells; MET + BC notably restrained their PRL secretion. MET + BC further decreased tumour growth and serum PRL levels in xenografts than BC treatment alone. ER was down‐regulated after AMPK activation in both cultured cells and xenografts. Together, we propose that the AMPK signalling pathway down‐regulates ERα and ERβ, and suppresses PRLoma growth as well as PRL secretion. Combined MET + BC is a potential treatment for PRLomas.


| INTRODUCTION
Pituitary adenoma comprises 10%-15% of all diagnosed intracranial tumours. 1 PRLoma is the most common type of functional pituitary adenoma, with a prevalence of 100 per 1 million people. 2 Currently, dopamine agonists such as BC are the primary therapy for PRLoma, with a high probability of controlling tumour size and reducing PRL level. Reportedly, BC normalizes PRL levels in 80%-90% of cases, and reduces tumour size in~70% cases. 3 As in previous studies, we herein define BC resistance as the failure to normalize PRL levels or to reduce tumour size by ≥50%, after taking ≥15 mg/day of BC for at least 3 months. 4,5 The mechanisms of BC resistance are unclear, although reduced expression of D2R is believed to be the main factor. 3 The AMPK pathway reportedly mediates proliferation or apoptosis in multiple cancer cell types, including non-small cell lung cancer, 6 glioblastoma, 7 breast cancer, 8 which implies a complex role for AMPK in tumour cell survival. Depending on cancer type, AMPK functions either as an oncoprotein or a tumour suppressor. 9 However, the role of AMPK in PRLoma has not been clear.
Activation of AMPK reportedly inhibits mammalian target of rapamycin (mTOR), and suppression of the mTOR pathway would induce autophagy-dependent cell death in PRLomas. 10 Notably, AMPK can be activated by AMP-mimetic 5-aminoimidazole-4carboxamide ribonucleoside (AICAR), 11 and by MET. 8 The latter is a widely used treatment for type 2 diabetes mellitus. 12 The function of MET is primarily associated with its activity on cellular energy metabolism. By restraining Complex I in the mitochondrial respiratory chain, MET generates cellular energy stress and thus activates AMPK. 13 Accumulating evidence supports an anticancer effect of this drug. 14 To determine whether the AMPK pathway affects PRLoma development, we measured AMPK phosphorylation in human primary PRLoma samples, and investigated the effect and downstream effectors of AMPK agonist MET in PRLoma using BC-sensitive MMQ cells and BC-resistance GH3 cells and their xenografts as models. Our working hypotheses are that the AMPK pathway is a potential therapeutic target for PRLoma, and MET combined with BC is a potential treatment for PRLoma.

| Bromocriptine resistance was associated with down-regulated AMPK activity and high oestrogen receptor expression
To explore whether AMPK is related to drug-sensitivity in PRLomas, we collected samples of BC-sensitive and -resistant PRLomas and examined them for expression of D2R, and AMPK protein and phosphorylation levels. As expected, D2R expression was generally higher in the BC-sensitive group ( Figure 1A), although in the BC-resistant group, Patients 1, 2, 3, and 9 also showed high D2R levels. AMPK activity, as represented by phosphorylated AMPK (p-AMPK), was significantly higher in the BC-sensitive group than in the BC-resistant group. Notably, the four BC-resistant patients with high D2R levels (Patients 1, 2, 3, and 9) all had almost undetectable p-AMPK levels ( Figure 1A). General clinical data of the 16 PRLoma patients were listed in Figure 1B.
As ER expression has been shown to contribute to drug resistance in PRLoma, 15,16 we measured ER levels in the human PRLoma samples.
RT-PCR analysis showed that ERα expression was significantly higher in drug-resistant samples than in the drug-sensitive ones. Median expression of ERβ tended to be higher in the resistant group, although not significantly so (Figure 2A). Immunohistochemistry confirmed higher ERα and ERβ levels in drug-resistant patients ( Figure 2B,C).

| MET activates AMPK in PRLoma cells and inhibits their growth and prolactin secretion
We used MET to activate AMPK in BC-sensitive (D2R-positive) MMQ cells and BC-resistant (D2R-negative) GH3 cells. Obviously, increased AMPK phosphorylation was detected in both cell lines ( Figure 3A). EdU cell cycle analysis revealed that MET markedly suppressed proliferation of both cell lines ( Figure 3B). CCK-8 assay confirmed crippled cell number of GH3 and MMQ upon MET treatment, and showed an additive effect between MET and BC selectively in the D2R-positive MMQ cell ( Figure 3C). AMPK activation also slightly increased proportions of Annexin V-positive cells in both cell lines, but the difference is not significant ( Figure 3D).
Prolactin (PRL) levels in cell supernatants were then measured before and after treatment with different combinations of BC and MET. We found BC used alone reduced PRL in both MMQ and GH3 cells ( Figure 4A,B); as expected, MMQ cells were more sensitive to BC. Importantly, PRL level is greatly decreased after combined BC + MET treatment, indicating that AMPK activation enhances the effect of BC in cellular PRLoma models.

PRLoma xenografts
We next assessed the effect of MET on PRLoma growth in a xeno-

| MET reduces ER expression in PRLoma cells and xenografts
ER is associated with GH3 cell proliferation, 17 and is expressed at higher levels in the BC-resistant human specimens. In the present GAO ET AL.  Treatment with dopamine agonists is currently the first choice for PRLomas because they can normalize PRL levels, reduce the volume of tumour, and restore gonadal function. 18,19 Reportedly, however, about 25% of patients are resistant to BC, and 10% are resistant to cabergoline. 20 While the response of PRLomas is shown closely related to D2R expression level, 21 cases existed that PRLomas of high D2R level are resistant to dopamine agonist, 22 supporting the involvement of other factors in BC resistance. 23 To better characterise BC-resistant PRLomas, we followed strict criteria in patient selection in the present study. All patients had taken BC regularly for at least 3 months and been closely followed up. Eight BCsensitive patients who had severe drug side effects or strongly refused further medication contributed to the precious surgical specimens of drug-sensitive PRLoma. The BC-resistant tumours were from eight age-and sex-matched patients, among them four were shown to express high level of D2R. Interestingly, we found that AMPK was activated in all our BC-sensitive PRLoma specimens, but severely suppressed in the D2R-positive BC-resistant ones. The data from primary PRLoma specimens therefore suggest a role of AMPK activity in regulating PRLoma growth and its BC-resistance. | 6371 administration in animals, and also in men, induced the appearance of pituitary tumours, especially PRLomas. 31,32 Moreover, approximately 30% of patients with PRLomas experience increased tumour volume during pregnancy due to increased oestrogen levels. 33 Clinical studies have confirmed that ER expression is positively correlated with tumour PRL level 34 and is significantly increased in invasive PRLomas. 35 Conversely, ER antagonist fulvestrant showed reduced size and PRL secretion of PRLomas, 16,36 and tamoxifen was successfully used to treat a dopamine agonist-resistant PRLoma. 37 These data, together, suggest that ER may be responsible for the anti- Metformin (4 mmol/L). (One-way ANOVA analysis, ****P < 0.0001) C, F, Plasma PRL levels of MMQ (C) and GH3 (E) xenograft mice in different drug group were examined by ELISA assays (One-way ANOVA analysis, ns, non-significant, **P < 0.01, ****P < 0.0001). Combined effects of MET and BC vs BC treatment alone in MMQ (D) and GH3 (F) xenograft mice were highlighted (post hoc pairwise comparison test, ns: non-significant, *P < 0.05, **P < 0.01 vs blank if not specifically indicated) F I G U R E 3 Metformin activates AMPK, suppressed proliferation, but has no significant effect on apoptosis of both the bromocriptinesensitive MMQ cells and the bromocriptine-resistant GH3 cells. A, MMQ and GH3 cells were treated with metformin of indicated concentrations for 24 hours before subjected to western blotting analysis of AMPK and pAMPK levels. Quantification of the western blotting assay is shown in graphs to the right of each image. (Student's t test, ***P < 0.001 vs blank) B, EdU cell cycle analysis showed markedly repressed proliferation of MMQ and GH3 cells upon the treatment of metformin at indicated concentration. (One-way Anova analysis, ****P < 0.0001) C, CCK-8 assay of GH3 and MMQ cells treated with different combination of MET and BC. (Student's t test, *P < 0.05, **P < 0.01,***P < 0.001 vs "MET 0 mM" group) D, Annexin V-FITC analysis detected no significant changes in apoptosis of MMQ and GH3 cells after treated with Metformin of different dosages GAO ET AL.

| 6373
PRLoma activity of MET and for its reported sexual dimorphism of the anti-PRLoma activity. 29 In addition to MET, we showed that AMPK agonist AICAR also decreased ER expression in both GH3 and MMQ cells. Moreover, we found in the D2R-positive MMQ cells a mutual stimulative relationship between BC/D2R signalling and AMPK activation. BC treatment suppresses ERα and ERβ expression in MMQ and its xenografts, the suppression is much weaker in D2R-negative GH3 cells. Consistently, oral administration of BC was found to significantly decrease ER levels in patients with sensitive PRLomas. 38 The observations support that AMPK activation inhibits of ERα and ERβ expression in PRLomas. A schematic representation of the MET/ pAMPK/ER signalling pathway and its potential crosstalk with BC/ D2R pathway in PRLoma treatment is shown in Figure 8. The physiological relevance and detailed signalling pathway for the interplays of D2R and pAMPK, and for that from pAMPK to ER, are pending further exploration.
Together, our results suggest that the suppressed AMPK signalling pathway contributes to drug resistance of pituitary PRL adenomas and represents an effective target for BC-resistant PRLoma.
We propose that therapeutic effect of BC can be improved by com-

| Cell lines
Rat MMQ cells, GH3 cells, F12 medium, and F10 medium were obtained from the Cell Center, Peking Union Medical College. MMQ cells were cultured in F12 medium supplemented with 2.5% foetal bovine serum (Gibco, USA), 15% horse serum (Gibco), 100 U/mL penicillin (Gibco), and 100 U/mL streptomycin (Gibco) in a humidified incubator at 37°C with 5% CO 2 . GH3 cells were maintained under the same conditions, except that F10 was used as the culture medium.
F I G U R E 6 Both metformin and AICAR reduced ER expression in MMQ and GH3 cells. A, ERα and ERβ expression in in vitro cultured MMQ and GH3 cells were detected by western blotting assays before and after the treatment of metformin at indicated concentrations. Quantification of the western blotting assay is shown in graphs to the right of each image. (Student's t test, ***P < 0.001 vs blank) B-E, MMQ and GH3 xenograft mice were daily treated with saline (blank), 400 mg/kg bromocriptine (BC), 500 mg/kg metformin (MT) or 500 mg/kg metformin in combination with 400 mg/kg bromocriptine (MT + BC), respectively, for five successive weeks. ERα and ERβ mRNA expression in different drug groups of MMQ (B:ERα, C:ERβ) and GH3 (D:ERα, E:ERβ) xenograft tumours were detected by RT-PCR (N = 5 for each group). Metformin inhibits ERα and β expression in both MMQ and GH3 xenografts with or without combined treatment of bromocriptine. (One-way ANOVA analysis, ns, non-significant, *P < 0.05, ***P < 0.001 vs blank if not specifically indicated) F, G, western blotting assays detected F, increased AMPK phosphorylation and G, decreased ERα and ERβ expression in AICAR treated GH3 and MMQ cells. Quantification of the western blotting assay is shown in graphs to the right of the images (Student's t test, ns, non-significant, *P < 0.05, **P < 0.01, ***P < 0.001 vs blank) The mutual stimulative effects of bromocriptine/D2R and AMPK signalling pathways. A, Bromocriptine stimulated AMPK phosphorylation in D2R-positive MMQ cell, but not in the D2R-negative GH3 cell. Quantification of the western blotting assay is shown in graphs below each image. (Student's t test, ***P < 0.001 vs blank) B, C, Both AICAR and Metformin treatment led to D2R up-regulation in MMQ cells at B mRNA and C protein levels. Quantification of the western blotting assay is shown in graphs to the right of each image. (B: One-way ANOVA analysis, C: Student's t test, ns, non-significant, *P < 0.05, **P < 0.01,***P < 0.001, ****P < 0.0001 vs blank if not specifically indicated)  We separated 10 μg of extracted protein on 12% SDS-PAGE gel, which was transferred onto an Immobilon-P membrane. Membranes were blocked in Tris-buffered saline with 5% non-fat milk, then probed with designated primary antibodies overnight at 4°C and incubated with the relevant peroxidase-conjugated secondary antibody (1:5000) for 1 hour at room temperature. Next, membranes were washed and visualised with an enhanced chemiluminescence system. The AMPK, D2R, ERα, and ERβ expression levels were quantified and normalised to the GAPDH control. Group C (500 mg MET/kg only); and Group D (400 mg BC/kg + 500 mg MET/kg). Tumour growth in each drug group (N = 5) were recorded weekly. Five weeks later animals werekilled. Blood was drawn from the retro-orbital sinus, followed by PRL assay. Tumours F I G U R E 8 Schematic representation of metformin/AMPK/ER signalling pathway and its potential crosstalk with bromocriptine/ D2R pathway in prolactinoma treatment. Red arrow/line refers to regulatory relationship proposed in this study. Dashed line with question mark refers to potential AMPK-independent pathway for BC/D2R to regulate ER expression were harvested, followed by photography and western blotting analysis of ERα and ERβ expression.

| Hormone secretion analysis
For MMQ and GH3 cells, supernatants were collected before or after drug treatment dependent on the experiments. For xenografted animals, blood was collected and centrifuged at 3300 rpm for 10 minute at 4°C to separate plasma; PRL levels in supernatant and plasma were determined with an enzyme-linked immunosorbent assay kit from USCN Life Science Inc. (Wuhan, China).

| Statistical analysis
All data are presented as means ± SD. The statistical analyses were performed using one-way analysis of variance (ANOVA), two-way ANOVA, Student's t test or post hoc pairwise comparison test as indicated. Analyses were conducted on GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA).