Targeting Corticotroph HDAC and PI3-Kinase in Cushing Disease

Abstract

Context

Cushing disease (CD) is a life-threatening disorder. Therapeutic goals include symptom relief, biochemical control, and tumor growth inhibition. Current medical therapies for CD by and large exert no action on tumor growth.

Objective

To identify drugs that inhibit corticotroph tumor adrenocorticotropic hormone (ACTH) secretion and growth.

Design

High throughput screen employing a novel “gain of signal” ACTH AlphaLISA assay.

Setting

Academic medical center.

Patients

Corticotroph tumor tissues from patients with CD.

Interventions

None.

Main outcome measures

Potent inhibitors of corticotroph tumor ACTH secretion and growth.

Results

From a kinase inhibitor library, we identified the dual PI3K/HDAC inhibitor CUDC-907 as a potent inhibitor of murine and human corticotroph tumor ACTH secretion (median effective concentration 1-5 nM), and cell proliferation (median inhibitory concentration 5 nM). In an in vivo murine corticotroph tumor xenograft model, orally administered CUDC-907 (300 mg/kg) reduced corticotroph tumor volume (TV [cm³], control 0.17 ± 0.05 vs CUDC-907 0.07 ± 0.02, P < .05) by 65% and suppressed plasma ACTH (ACTH [pg/mL] control 206 ± 27 vs CUDC-907 47 ± 7, P < .05) and corticosterone (corticosterone [ng/mL] control 180 ± 87 vs CUDC-907 27 ± 5, P < .05) levels by 77% and 85% respectively compared with controls. We also demonstrated that CUDC-907 acts through HDAC1/2 inhibition at the proopiomelanocortin transcriptional level combined with its PI3K-mediated inhibition of corticotroph cell viability to reduce ACTH secretion.

Conclusions

Given its potent efficacy in in vitro and in vivo models of CD, combined with proven safety and tolerance in clinical trials, we propose CUDC-907 may be a promising therapy for CD.

Cushing disease (CD), caused by a pituitary corticotroph tumor, is a debilitating endocrine disorder due to excess adrenal-derived cortisol. Disabling and sometimes fatal symptoms include cardiovascular disease, diabetes, osteoporosis, obesity, and psychiatric disturbances (1). Surgical removal of these typically microadenomas is the current first-line therapy (2). Although initial remission rates are ~80% in expert centers, the disease recurs in up to 50% of cases over 10 to 15 years and repeat pituitary surgery has poorer success rates (3, 4). Other options such as pituitary-directed radiation typically takes years to offer biochemical control and causes hypopituitarism in ~40% of patients, and bilateral adrenalectomy, although resolving hypercortisolism, requires lifelong gluco- and mineralocorticoid replacement. Currently available drugs for CD often do not offer long-term control, are limited by side effects, and, most importantly, do not target the site of disease (5). Pasireotide, 1 of 3 approved agents for CD, inhibits pituitary-derived adrenocorticotropic hormone (ACTH) secretion but causes frequent severe hyperglycemia; whereas the glucocorticoid receptor blocker mifepristone lacks a biomarker of activity limiting its routine clinical use (6-8). Adrenal-directed agents that inhibit cortisol synthesis, such as the approved agents osilodristat, ketoconazole, and metyrapone, do not exhibit tumor-directed actions, and long-term escape from control has been reported (9). A clear unmet need for efficacious and safe therapies for CD that simultaneously inhibit tumor growth and offer biochemical disease control exists and we hypothesize that direct targeting of corticotroph tumors to inhibit ACTH secretion and tumor growth is the optimal way to treat CD.

High throughput screening (HTS) is an automated strategy to identify biologically relevant therapeutic compounds that can be used as chemical probes to study diseases of interest, or can be developed into a series of chemical analogs with improved bioactivities. We developed a novel “gain of signal” homogenous ACTH AlphaLISA assay based on the proximity-dependent chemical energy transfer principle (10) to identify novel inhibitors of murine and human corticotroph tumor ACTH secretion through a HTS. A screen of a kinase inhibitor library led to identification of a potent novel inhibitor (CUDC-907) of corticotroph tumor ACTH secretion and growth in preclinical in vitro and in vivo CD models. Given CUDC-907 has already undergone extensive safety and tolerability studies in lymphomas, we propose that CUDC-907 could be a promising candidate to repurpose as a novel therapy with an acceptable safety and side effect profile for the treatment of CD (11, 12).

Materials and Methods

Cell culture and reagents

AtT-20/D16v-F2 cells (ATCC® CRL-1795™) were purchased from ATCC. This cell line was used only between 10 and 20 passages as it consistently secretes ACTH across this window. Fresh consecutive surgically resected human corticotroph tumor tissues were collected from patients who underwent endoscopic transnasal transsphenoidal surgery in accordance with University of California, Los Angeles Institutional Review Board guidelines (protocol number: 10-001785). A CD diagnosis was confirmed in all patients using standard clinical and biochemical criteria, and a corticotroph tumor was confirmed on pathologic evaluation. Resected corticotroph tumor tissues were washed 3 times in sterile phosphate-buffered saline (PBS), then minced and mechanically dispersed. After centrifugation, cell pellets were resuspended in culture medium supplemented with 10% fetal bovine serum. Cell authentication using short tandem repeat (STR) profiling was performed to confirm some of the surgically derived human corticotroph tumor cultures were human origin and did not match any existing tumor cell line in the ATCC or DSMZ databases. Mycoplasma testing was performed to ensure no contamination.

Antibodies, peptide, and chemicals

Anti-ACTH antibodies (#1 EMD Cat. CBL57; #2 Abcam Cat. Ab20358; #3 Novus Cat. NBP2-34529) were bought from commercial resources. Biotinylated ACTH peptide (1-39 amino acid [aa]) was purchased from AnaSpec (Cat. AS-23968). The antibodies against Acetyl-Histone H3 (Lys9) (Cat. 9649), p27 (Cat. 3686), Phospho-Akt (Ser473) (Cat. 9271), Phospho-4E-BP1 (Thr37/46) (Cat. 2855), Phospho-(Ser/Thr) Phe (Cat. 9631), and Class I histone deacetylase (HDAC) Antibody Sampler Kit (Cat. 65816) were purchased from Cell Signaling Technology (Danvers, MA); the antibodies against Nurr1 (Cat. sc-376984), Actin (Cat. sc-47778), HDAC-10 (Cat. sc-393417), and GR (Cat. sc-1004) were purchased from Santa Cruz Biotechnology Inc. (Dallas, TX); the antibodies against TR4 (Cat. ab109301), LXRα (Cat. ab41902), and Nur77 (Cat. ab109180) were purchased from Abcam (Cambridge, MA). Aliquots of antibodies and peptide were stored at –80°C. CUDC-907, panobinostat, vorinostat, pictilisib, and buparlisib (purity >98%) were purchased from MedChemExpress. The KIL library was purchased from Selleck Chemicals (Houston, TX). Streptavidin-labelled donor beads, and antimouse immunoglobulin (IgG) (Fc specific)-labeled acceptor beads were purchased from PerkinElmer.

Plasmid constructs

Plasmids containing 1000 bp of the human proopiomelanocortin (POMC) promoter fused to a firefly luciferase reporter were used as published previously (13). HDAC1 Flag plasmid (human) was a gift from Eric Verdin (Addgene plasmid #13820; http://n2t.net/addgene:13820; RRID:Addgene_13820) (14). pcDNA3.1-HDAC2 Flag plasmid (mouse) was purchased from Addgene (Addgene plasmid #68117) (15). pD40-His/V5-c-Myc (mouse) was a gift from Rosalie Sears (Addgene plasmid #45597; http://n2t.net/addgene:45597; RRID:Addgene_45597) (16). pcDNA3.1-V5-TR4 plasmid (mouse) was used as published previously (17). pCMV6-Entry-Nurr1 (mouse) was purchased from OriGene Technologies Inc. (Rockville, MD). All constructs were verified by sequencing.

Cell proliferation assay

AtT20 cells, and primary cultures of human pituitary corticotroph ACTH-secreting tumors were suspended in 100 μL of Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum in 96-well plates (2 × 10³ viable cells/well) overnight prior to treatment. Cells were then treated with the indicated compounds for 3 days. Cell viability was determined using a CellTiter-Glo® Luminescent Cell Viability Assay (Promega, Madison, WI) with a luminometer (Wallac 1420 Victor 2 multipliable counter system). Results are presented as relative proliferation rate where the luminescence signal in treated cells was compared with the vehicle control. All experiments were repeated at least 3 times and data are depicted as mean ± standard deviation (SD).

Luciferase reporter assay

AtT20 cells were stably transfected with both a POMC and a Renilla luciferase reporter (the latter as an internal control) and were treated with various compounds as indicated. The luminescence signal was quantitated using the dual-luciferase system (Promega), normalizing the firefly luminescence to the Renilla signals and the fold-change in luciferase activity following treatment calculated relative to the normalized luminescence signal in vehicle control samples. The results shown are representative of 3 independent experiments.

Caspase induction assay

CUDC-907 actions on apoptosis were quantitated by the Caspase-Glo 3/7 assay (Promega). Following CUDC-907 treatment for 24 hours, cells were lysed in 75 μL of chilled cell lysis buffer and incubated on ice for 10 minutes. Luminescence was then measured and expressed relative to vehicle-treated control as an apoptosis index.

Real-time polymerase chain reaction

Total RNA was extracted with RNeasy kit (Qiagen). Ribonucleic acid (RNA) quality and quantity were measured by absorbance at 260 and 280 nm respectively. Total RNA was reverse transcribed into first-strand complementary deoxyribonucleic acid (cDNA) using a cDNA synthesis kit (Invitrogen), and quantitative polymerase chain reaction (PCR) reactions carried out using the CFX real-time PCR Detection System (Bio-Rad Laboratories Inc.). Primer sequences (Invitrogen/Life Technologies) were as follows: mouse β-actin forward primer, 5′-GGC TGT ATT CCC CTC CAT CG-3′; mouse β-actin reverse primer, 5′-CCA GTT GGT AAC AAT GCC ATG T-3′; mouse POMC forward primer, 5′- CCA TAG ATG TGT GGA GCT GGT G-3′; mouse POMC reverse primer, 5′-CAT CTC CGT TGC CAG GAA ACA C-3′; mouse Nurr1 forward primer, 5′-CCG CCG AAA TCG TTG TCA GTA C-3′; mouse Nurr1 reverse primer, 5′-TTC GGC TTC GAG GGT AAA CGA C-3′; human POMC forward primer, 5′-GCC AGT GTC AGG ACC TCA C-3′; human POMC reverse primer, 5′-GGG AAC ATG GGA GTC TCG G-3′; human actin forward primer, 5′-CAC CAT TGG CAA TGA GCG GTT C-3′; human actin reverse primer, 5′-AGG TCT TTG CGG ATG TCC ACG T-3′.

Western blot and coimmunoprecipitation

For Western blot assay, after treatments cells were washed in cold PBS, and proteins extracted in 100 μL of radioimmunoprecipitation assay buffer (Cell Signaling, Danvers, MA) containing a complete protease inhibitor mixture (Roche Molecular Biochemicals, Indianapolis, IN). Protein concentrations were determined by DC protein assay reagent (Bio-Rad, Hercules, CA) and extracts were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and transferred to PVDF membranes (Bio-Rad, Hercules, CA). Membranes were then blocked for 2 hours at room temperature in 0.1% tris buffered saline containing 5% nonfat dried milk and 0.1% Tween-20, washed, and then incubated with the specific primary antibodies over night at 4°C. After washing, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for three hours at 4°C (Santa Cruz Biotechnology Inc., Dallas, TX) and proteins visualized using a Super Signal Chemiluminescence Assay kit (Pierce, Grand Island, NY). For coimmunoprecipitation, cells were harvested and lysed in cell lysis buffer (Cell Signaling Technology) containing complete protease inhibitor (Roche) and 1 mM phenylmethylsulfonylfluoride (Cell Signaling Technology). The majority (90%) of the cell lysates were then incubated with normal mouse IgG (Santa Cruz Biotechnology Inc., Cat. sc-2343) as control, or anti-Nurr1 antibody coupled agarose beads (Santa Cruz Biotechnology, Inc., Cat. sc-76984 AC) at 4°C overnight with agitation. The resin was then washed 5 times with cell lysis buffer. Pulled-down protein complexes were then run on SDS-PAGE gels and immunoblotted using appropriate primary antibodies. The remaining cell lysates (input, 10%) were also analyzed on SDS-PAGE gels and immunoblotted in similar fashion. The results shown were representative of 3 independent experiments. Quantitative densitometric analyses of the protein bands were compared to the individual loading controls using ImageQuant 5.2 software (GE Healthcare, Pittsburgh, PA). Results shown are representative of 3 independent experiments.

Hormone assays

Enzyme-linked immunosorbent assays for mouse and human ACTH, and mouse corticosterone were performed in triplicate using reagents purchased from Biomerica (Irvine, CA) and Calbiotech (Spring Valley, AC).

Tumor xenografts and in vivo treatment studies

The use of mice was approved by the University of California Los Angeles Animal Research Committee and complied with all relevant federal guidelines and institutional policies. AtT20 cells (5 × 10⁵) in 100 μL of Matrigel were injected subcutaneously into 5-week-old Nu/J (JAX) mice to generate corticotroph pituitary tumors. Three days after inoculation, mice were randomly divided into 2 groups (n = 10/group), and treated with either vehicle (30% Captisol) or CUDC-907 300 mg/kg dissolved in vehicle which was administered orally once daily for 21 days. This single dose was chosen as it has been used most commonly in other solid tumor studies and due to cost of the reagent. Tumor sizes were measured every other day in 2 dimensions with Vernier calipers and volumes were calculated using the equation: length × width² × 0.5. Upon completion of treatment, mice were euthanized using CO₂ inhalation, cardiac blood was collected, and tumors were excised and weighed.

Statistics

All in vitro experiments were repeated at least 3 times and in at least 3 individual human corticotroph tumor primary cultures. Results are expressed as mean ± SD. Differences were assessed student t test. P values less than .05 were considered significant.

Results

Development and optimization of an ACTH AlphaLISA assay for high throughput screen

As commercially available ACTH immunoassays require a large sample volume, are time consuming and expensive, they are not ideal for HTS. Therefore, we developed an amplified luminescent proximity homogeneous assay screen (AlphaScreen) (10) for ACTH quantification. This assay comprises streptavidin-labelled donor beads containing a photosensitizer (phthalocyanine), a biotin-labelled ACTH peptide, a mouse anti-ACTH monoclonal antibody, and antimouse IgG (Fc specific) conjugated acceptor beads. When the donor and acceptor beads are brought into close proximity through an antigen–antibody reaction, laser excitation of the donor beads (680 nm) generates a short-lived singlet oxygen molecule that interacts with the acceptor beads within close proximity (200 nm) to generate an amplified chemiluminescent signal (615 nm, Fig. 1A). If a cell culture supernatant (SN) containing secreted ACTH is added to the assay, the free cell SN ACTH analyte competes with the biotinylated ACTH peptide to bind the anti-ACTH antibody, thereby disrupting donor and acceptor proximity to reduce signal intensity (Fig. 1B). In contrast, if cells are exposed to inhibitors of ACTH secretion, the cell SN contains less ACTH, leading to restoration of the chemiluminescent signal.

Figure 1.

Schematic overview of the novel ACTH AlphaLISA assay. (A) Streptavidin-labelled donor beads and antimouse IgG-coated acceptor beads are brought into close proximity by biotinylated ACTH peptide and mouse anti-ACTH antibody. Excitation at 680 nm triggers the release and transfer of a singlet oxygen from donor to acceptor beads within close proximity to produce a measurable chemiluminescence signal at 615 nm. (B) Addition of free ACTH analyte, as in a cell SN, competes with the biotinylated ACTH peptide to displace the donor and acceptor beads and inhibit the Alpha signal.

Available monoclonal anti-ACTH antibodies are exclusively generated against the N-terminus 1 to 24 aa ACTH sequence, which is highly conserved in humans, rats and mice. By comparing 3 commercially available anti-ACTH antibodies (1 nM) with a range of concentrations of biotinylated-ACTH peptide (human 1-39 aa), we selected an antibody that exhibited a robust Alpha signal even at extremely low biotinylated ACTH concentrations (Fig. 2A, αACTH-Ab #1). Further testing of a range of biotinylated ACTH peptide and anti-ACTH antibody concentrations (0.1-0.3 nM) using varying volumes of 3- and 4-day (D) murine corticotroph tumor AtT20 cell–derived SN (ACTH ~10^–10 M) (18) demonstrated robust dose- (SN volume)-dependent reduction in AlphaLISA signals (Fig. 2B). The assay Z′ factor, which is a statistical parameter calculated from the standard deviations of negative and positive controls to assess assay performance and facilitate assay optimization (19), remained consistently >0.7 using 0.3 nM biotinylated ACTH peptide (biotin-ACTH peptide) in combination with 0.1 nM anti-ACTH antibody (αACTH-Ab) with the 3- and 4-D SN (except the lowest volume of 5 μL Fig. 2B). Considering potential compound instability with longer incubation periods, 0.3 nM of biotinylated ACTH peptide, 0.1 nM of anti-ACTH Ab and a 3D AtT20 cell SN were selected as optimal assay conditions (Fig. 2B).

Figure 2.

ACTH AlphaLISA assay development and optimization. (A) Three anti-ACTH antibodies (#1 EMD Cat. CBL57; #2 Abcam Cat. Ab20358; #3 Novus Cat. NBP2-34529) were tested at 1 nM for antibody configuration with biotinylated ACTH peptide (1-20 nM). Antibody #1 (EMD) generated robust AlphaLISA signal at low biotinylated ACTH concentrations and was chosen for further development. (B) Varying volumes of 3- and 4-day murine corticotroph tumor AtT20 cell derived SN in combination with biotinylated-ACTH peptide (0.1 and 0.3 nM) and anti-ACTH antibody (0.1 and 0.3 nM) were compared to assess inhibition of the donor–acceptor bead interaction expressed as raw AlphaLISA signals (upper panel) and as Z′ factor (lower panel). Three-day culture time in combination with 0.3 nM biotinylated ACTH peptide and 0.1 nM anti-ACTH antibody (solid bar and boxed) was chosen. (C) All the assay volumes tested (20, 15, 10, and 5 μL) demonstrated potent inhibition of AlphaLISA signals generating a Z′ factor >0.6, so a 5-μL assay volume was selected (boxed). (D) Using the 5 μL reaction volume with 2 μL of 3D AtT20 cell SN, basal alpha signal reduced in a stepwise fashion with decreasing acceptor bead concentrations (8-4 μg/mL), but Z′ remained >0.7 for all, so a 4 μg/mL acceptor bead was selected (solid bar and boxed). (E) Using this acceptor bead concentration, the Z′ factor dropped below 0.7 when the donor bead concentration was reduced from 10 to 5 μg/mL, so a 10 μg/mL donor bead was chosen (solid bar and boxed). (F) Summary of optimal assay component volumes, concentrations and procedures for the ACTH AlphaLISA.

Having thus far used a 50-μL assay format, we then tested 20, 15, 10, and 5-μL volumes; and varying acceptor and donor bead concentrations to find the lowest effective assay format to optimize cost. Three-day AtT20 cell SN at all assay volumes including 5 μL generated potent inhibition of Alpha signals and a Z′ factor >0.6 (Fig. 2C) with 4 μg/mL of acceptor bead (Fig. 2D) and 10 μg/mL of donor bead concentrations (Fig. 2E). As summarized in Fig. 2F, the final assay comprises a liquid transfer step of 2 μL of SN, followed by addition of 1 μL of biotinylated ACTH peptide and 1 μL of anti-ACTH antibody with 1 hour of incubation, followed by addition of 1 μL of donor and acceptor bead mixture solution with a 2-hour incubation for a total of 3 hours of assay time. The cost of our AlphaLISA is ~$0.1 per reaction.

Identification of CUDC-907 as potent inhibitor of ACTH secretion and corticotroph tumor growth

Given the important role of growth factors and aberrant USP8 mutant-associated EGFR activation in corticotroph tumors (20), we used our ACTH AlphaLISA assay to screen an annotated kinase inhibitor library (KIL, n = 430) at 100 nM, 1 μM, and 10 μM concentrations. The kinase inhibitory library (KIL) contains inhibitors of phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT)/mammalian target of rapamycin (mTOR) (n = 95), protein tyrosine kinases (n = 87), MAPK (n = 45), angiogenesis (n = 44), cell cycle (n = 44), JAK/STAT (n = 26), and others (n = 89, Fig. 3A). By incorporating a permeable nuclear fluorescent dye (Hoechst 33342), we simultaneously quantified compound action on cell proliferation as well as ACTH secretion. Of 430 screened compounds, 6, 20, and 115 compounds exhibited ≥50% inhibition of ACTH secretion (Fig. 3B) and 36, 105, and 263 compounds inhibited proliferation by ≥50% (Fig. 3C) at 100 nM, 1 μM, and 10 μM doses respectively. On combining data from both analyses, we identified 6 compounds that inhibited both murine corticotroph tumor in vitro ACTH secretion and proliferation by ≥50% at 100 nM (Fig. 3D). Correcting ACTH inhibition for cell number as measured by nuclei staining for this set of 6 potent inhibitory compounds, the HDAC and PI3K dual inhibitor CUDC-907 exhibited the highest ACTH inhibitory activity (red circle Fig. 3E). In fact, CUDC-907 potently inhibited both ACTH secretion and proliferation with an EC₅₀ of 1 nM (inhibition of ACTH secretion by 50%, Fig. 4A), and IC₅₀ of 5 nM (inhibition of cell proliferation by 50%, Fig. 4B) in the murine corticotroph tumor AtT20 cells. We next tested actions of CUDC-907 in a series of consecutive human corticotroph tumor–derived primary cultures (n = 8) (details in Fig. 4C). CUDC-907 5 nM resulted in an average of ~51% reduction in ACTH secretion (ACTH secretion fold change relative to vehicle, patient 1# 0.7 ± 0.1, P < .05; patient 2# 0.2 ± 0.04, P < .01; patient 3# 0.5 ± 0.2 n.s., Fig. 4D). CUDC-907 also exhibited a potent antiproliferative effect in the human pituitary corticotroph tumor cultures where it demonstrated an IC₅₀ ranging from 1.2 nM to 10.7 nM with an average of IC₅₀ 5.5 ± 4.8 nM (n = 3 individual corticotroph tumors, Fig. 4E). Using a POMC promoter-driven luciferase assay, we next characterized the effect of CUDC-907 on POMC transcription. As shown in Fig. 4F, CUDC-907 inhibited murine corticotroph tumor POMC transcription with an EC₅₀ of 0.5 nM. Consistent with this, we further demonstrated that CUDC-907 potently inhibited murine corticotroph tumor POMC mRNA (relative POMC mRNA expression, Veh 1.0 ± 0.02; CUDC-907 1.25 nM 0.7 ± 0.02, P < .01; 2.5 nM 0.4 ± 0.03, P < .005; 5 nM 0.5 ± 0.05, P < .01, Fig. 4G). In human pituitary corticotroph tumor primary cultures (n = 3), CUDC-907 also inhibited POMC mRNA expression from 22% to 70% (relative POMC mRNA expression, Veh vs CUDC-907, Patient 3# 1.0 ± 0.06 vs 0.78 ± 0.05, P < .05; Patient 7# 1.0 ± 0.05 vs 0.56 ± 0.07, P < .05; Patient 8# 1.0 ± 0.04 vs 0.29 ± 0.01, P < .01, Fig. 4H).

Figure 3.

Screen of kinase inhibitor library (KIL). (A) Depiction of KIL compound composition (n = 430). (B,C) AtT20 cells were treated with KIL compounds at 100 nM, 1 μM and 10 μM final concentrations. Plotted values corresponding to ACTH secretion and proliferation inhibition rates were calculated from AlphaLISA signals (B) and Hoechst 33342 staining data (C). Compounds exhibiting ≥50% ACTH inhibition (n = 6, 20 and 115, B) and ≥50% proliferation inhibition (n = 36, 105 and 263, C) are highlighted in red at doses of 100 nM, 1 μM and 10 μM respectively. (D) Compounds that exhibited ≥50% inhibition in both ACTH secretion and tumor proliferation are highlighted. (E) ACTH inhibition was corrected for cell number. As depicted (circled in red), CUDC-907 was the most potent hit identified.

Figure 4.

The effects of CUDC-907 on POMC mRNA expression and ACTH secretion in AtT20 cells and human CD cultures. (A,B) AtT20 cells were treated with CUDC-907 at a range of concentrations from 0.4 nM to 40 nM (A, 2-fold dilution) and 1.25-40 nM (B). CUDC-907 EC₅₀ for ACTH secretion inhibition (A) and IC₅₀ for proliferation inhibition (B) were calculated using a sigmoidal dose–response curve (GraphPad Prism). (C) Clinical summary for the human pituitary corticotroph tumor samples (n = 8) used for evaluation of CUDC-907 effects on ACTH secretion, cell proliferation and qPCR. (D,E) Effects of CUDC-907 on human corticotroph tumor primary culture ACTH secretion (D) and cell proliferation (E). (F) Actions of CUDC-907 (0.4 nM-10 µM, 2-fold dilution for 1 day) on murine corticotroph tumor POMC transcription was evaluated in AtT20 cells stably transfected with a POMC promoter luciferase reporter (POMC-Luc) in combination with Renilla luciferase reporter (as an internal control). Luminescence signals were then quantitated using the dual-luciferase system (Promega), and firefly luminescence signals normalized to the Renilla signals and expressed as fold changes, based on which the EC₅₀ for POMC transcription inhibition was calculated using a sigmoidal dose-response curve (GraphPad Prism). (G,H) Effect of CUDC-907 on murine (AtT20 cells, G) and human (H, corticotroph tumor primary cultures; n = 3) POMC mRNA expression detected by real time PCR. The results shown are representative of three independent experiments. *P < .05; **P < .01; ***P < .01.

In an in vivo xenograft model of CD using murine corticotroph tumor AtT20 cells subcutaneously inoculated into athymic nude mice (n = 10 each group), CUDC-907 (300 mg/kg dissolved in 30% Captisol) administered daily by oral gavage was well tolerated. CUDC-907 reduced corticotroph tumor volume (TV [cm³], control 0.2 ± 0.05 vs CUDC-907 0.07 ± 0.02, P < .05, Fig. 5A) and tumor weight (TW [g], control 0.1 ± 0.02 vs CUDC-907 0.04 ± 0.006, P < .05, Fig. 5B) by 65% and 56% respectively compared with vehicle-treated (30% Captisol) controls. Additionally, plasma ACTH (ACTH [pg/mL]; control 206 ± 27 vs CUDC-907 47 ± 7, P < .05, Fig. 5C) and corticosterone (corticosterone [ng/mL] control 180 ± 87 vs CUDC-907 27 ± 5, P < .05, Fig. 5D) levels were reduced by 77% and 85% respectively in CUDC-907-treated mice compared with controls.

Figure 5.

CUDC-907 inhibits in vivo corticotroph tumor growth, plasma ACTH and serum corticosterone secretion. (A,B) CUDC-907 (300 mg/kg dissolved in Captisol) or vehicle (Captisol) was administered daily by oral gavage to mice harboring xenografted corticotroph tumors (n = 10/group). On experiment completion, animals were euthanized and tumor size (A) and weight (B) were measured. (C,D) Blood samples were collected by cardiac puncture and plasma ACTH (C) and serum corticosterone (D) levels compared using enzyme-linked immunosorbent assay between CUDC-907 and vehicle-treated mice. *P < .05.

Comparison of CUDC-907 with single HDAC inhibitors (panobinostat and vorinostat), and PI3K inhibitors (pictilisib and buparlisib)

CUDC-907 was generated by integration of a HDAC inhibitory functional moiety (hydroxamic acid, red dashed circle, Fig. 6A) into a core PI3K inhibitor structure scaffold (morpholinopyridine, green dashed rectangle, Fig. 6A) (21). To better understand the contribution of HDAC versus PI3K inhibitory activities of CUDC-907 in suppressing corticotroph tumor ACTH secretion and proliferation, we compared the actions of CUDC-907 with the single-target HDAC inhibitors (HDACis) panobinostat and vorinostat, and the single-target PI3K inhibitors (PI3Kis) buparlisib and pictilisib. As shown in Fig. 6B and 6C, CUDC-907 and panobinostat potently inhibited ACTH secretion with EC₅₀ of 1 nM and 4 nM respectively (Fig. 6B), compared with another HDACi (vorinostat) or the PI3Kis (buparlisib and pictilisib, Fig. 6C) at the doses tested (0.4-40 nM). Similarly, CUDC-907 (IC₅₀ 5 nM) and panobinostat (IC₅₀ 20 nM) both inhibited AtT20 cell proliferation though CUDC-907 was slightly more potent (Fig. 6D) and both were much more potent than those observed for the HDACi vorinostat (IC₅₀ 2 μM), and the PI3K inhibitors buparilisib (IC₅₀ 0.5 μM) and pictilisib (IC₅₀ 0.8 μM, Fig. 6E). Using a POMC promoter-driven luciferase assay, we next examined for direct actions of these drugs on POMC transcription. As shown in Fig. 6F, CUDC-907 and the HDACis panobinostat and vorinostat inhibited POMC transcription with a range of potencies (CUDC-907 EC₅₀ 0.5 nM to vorinostat 0.5 μM). In contrast the PI3Kis buparlisib and pictilisib increased POMC transcription at higher doses (≥0.1 μM, Fig. 6G). Quantitation of POMC mRNA expression by RT-PCR demonstrated that only CUDC-907 caused a potent reduction in POMC mRNA expression compared with the other HDACis (Fig. 6H), and the 2 PI3Kis did not inhibit POMC mRNA expression (Fig. 6I). To further explore whether the PI3K inhibitory action synergized with our observed HDACi-mediated downregulation of ACTH secretion, we tested effects of combination of single agent nonselective HDACi panobinostat and the PI3Ki buparlisib. As shown in Fig. 6J, buparlisib alone did not inhibit ACTH secretion. In contrast, panobinostat itself (5 and 10 nM, circled) inhibited ACTH secretion by 30% and 66% respectively (ACTH secretion [ng/mL], Veh 39.6 ± 2 vs panobinostat 5 nM 28.7 ± 0.1, P < .05; 10 nM 13.6 ± 1, P < .01, Fig. 6J, circled). Combination treatment of panobinostat (5 and 10 nM) together with buparlisib (for instance, 62.5 nM) further inhibited ACTH secretion (by 51% and 85% respectively, ACTH secretion [ng/mL], buparlisib + panobinostat 5 nM 19.3 ± 0.1, P < .005; buparlisib + panobinostat 10 nM 5.8 ± 1, P < .01, Fig. 6J). Panobinostat alone (5 and 10 nM, circled) marginally inhibited murine corticotroph proliferation compared with vehicle (relative proliferation rate, Veh 1.0 ± 0.06 vs panobinostat 5 nM 1.0 ± 0.01, n.s.; 10 nM 0.8 ± 0.005, n.s., Fig. 6K), and buparlisib alone inhibited murine corticotroph tumor cell proliferation at concentrations >0.1 μM (Fig. 6K). Addition of increasing concentrations of buparlisib to panobinostat (5 and 10 nM) led to increased corticotroph tumor proliferation inhibition compared with panobinostat alone (relative proliferation rate, buparlisib 62.5 nM 1.0 ± 0.06; buparlisib 62.5 nM + panobinostat 5 nM 0.8 ± 0.03, n.s.; buparlisib 62.5 nM + panobinostat 10 nM 0.5 ± 0.01, P < 0.05, Fig. 6K). Although panobinostat (5 nM, circled) itself inhibited POMC-luciferase activity (relative POMC-Luc activity, Veh 1.0 ± 0.06 vs panobinostat 5 nM 0.5 ± 0.07, P < 0.01, Fig. 6L) and mRNA expression (relative POMC mRNA expression, Veh 1.0 ± 0.01 vs panobinostat 5 nM 0.6 ± 0.01, P < .005, Fig. 6M), addition of increasing concentrations of buparlisib did not lead to further reduced POMC-luciferase activity (Fig. 6L) or mRNA levels (Fig. 6M). Taken together, our results suggest that CUDC-907 exerts much of its inhibitory effect on ACTH secretion by its HDAC inhibitory actions to reduce POMC mRNA expression, while PI3K-mediated inhibition of corticotroph tumor cell viability may further contribute to reduced corticotroph tumor ACTH secretion.

Figure 6.

Comparison of CUDC-907 with single-target HDAC and PI3K inhibitors. (A) Chemical structure of CUDC-907 with its core anti-HDAC hydroxamate moiety (circled in red), and PI3K inhibitor skeleton (squared in green). (B,C) AtT20 cells were treated with CUDC-907 and the reference compounds (panobinostat, vorinostat, buparlisib, and pictilisib) at a range of concentrations from 0.4 nM to 40 nM (2-fold dilution) for 3 days after which EC₅₀ for ACTH secretion inhibition was calculated. (D,E) The proliferation inhibition effects of CUDC-907 and single action HDACis and PI3Kis were evaluated by CellTiter Glo assay. (F,G) The effects of CUDC-907 and reference compounds (0.4 nM-10 µM, 2-fold dilution for 1 day) on POMC transcription was evaluated using stable transfectants of murine corticotroph tumor AtT20 cells which expressed a POMC promoter driven firefly luciferase reporter (POMC-Luc) in combination with Renilla luciferase reporter (as an internal control). (H,I) AtT20 cells were treated with CUDC-907 and reference compounds (1.25 nM to 20 nM, 2-fold dilution for 1 day) and the POMC mRNA expression was quantitated by real time PCR. (J-M) ACTH secretion (J), cell proliferation (K), POMC-Luc activity (L), and POMC mRNA expression (M) were quantitated in AtT20 cells treated with either buparlisib alone or in combination with panobinostat (5 & 10 nM). The results shown are representative of three independent experiments. *P < .05; ***P < .01.

CUDC-907 targets Nurr1 expression and activation to inhibit POMC mRNA expression

CUDC-907 potently inhibits HDAC class I (IC₅₀ of 1.7, 5.0, and 1.8 nM for HDAC1, 2 and 3) and II enzymes (IC₅₀ of 2.8 nM for HDAC10) (21). HDAC1 and HDAC2 overexpression in corticotroph tumor AtT20 cells abrogated CUDC-907-mediated inhibition of POMC mRNA (relative POMC mRNA, Veh vs CUDC-907 5 nM, vector 1.0 ± 0.02 vs 0.8 ± 0.01, P < .05; HDAC1 1.1 ± 0.06 vs 1.2 ± 0.03 n.s; HDAC2 1.1 ± 0.03 vs 1.1 ± 0.02 n.s., Fig. 7A). Given HDACs do not contain canonical DNA binding domains, and are recruited to chromatin by protein–protein interactions with other DNA-associated factors (22), we sought to characterize the molecular partners of HDACs involved in CUDC-907’s regulation of POMC mRNA expression. Therefore, we first examined actions of CUDC-907 on expression of several key nuclear receptors (NRs) known to regulate POMC expression, including Nurr1 (NR4A2) (23, 24), LXRα (NR1H3) (25), TR4 (NR2C2) (17), and GR (NR3C1) (26). As shown in Fig. 7B, CUDC-907 treatment (1.25-5 nM) for 24 hours resulted in a striking dose-dependent reduction in corticotroph tumor Nurr1 protein expression (Nurr1/Actin ratio, Veh 1.0 ± 0.02; CUDC-907 1.25 nM 0.5 ± 0.02, P < .05; 2.5 nM 0.4 ± 0.02, P < .05; 5 nM 0.1 ± 0.02, P < .05, top panel-left, Fig. 7B and 7C). In support of this, we also observed that CUDC-907 treatment (24 hours) led to a dose-dependent reduction in Nurr1 mRNA expression (relative Nurr1 mRNA, Veh 1.0 ± 0.02, CUDC-907 1.25 nM 0.7 ± 0.07, P < .05; 2.5 nM 0.6 ± 0.0,3 P < .005; 5 nM 0.6 ± 0.06, P < .001; 10 nM 0.4 ± 0.01, P < .005; Fig. 7D). To further characterize the role of Nurr1 as potential target of CUDC-907 actions, we examined the effect of Nurr1 overexpression on CUDC-907-mediated POMC inhibition. As depicted in Fig. 7E, Nurr1 overexpression potently increased basal POMC mRNA expression (relative POMC mRNA, vector vs Nurr1 1.0 ± 0.02 vs 1.5 ± 0.07, P < .005, Fig. 7E) and blocked the actions of CUDC-907 to inhibit POMC mRNA expression (relative POMC mRNA, Veh vs CUDC-907 5 nM, vector 1.0 ± 0.02 vs 0.4 ± 0.01, P < .01; Nurr1 1.5 ± 0.07 vs 1.4 ± 0.05 n.s., Fig. 7E). We further demonstrated that CUDC-907 treatment reduced serine/threonine phosphorylated Nurr1 levels in line with reduced Nurr1 expression (Fig. 7F). Taken together, these findings suggest that CUDC-907 inhibits POMC mRNA expression by several mechanisms, in part through regulating expression and activation of the nuclear receptor, Nurr1.

Figure 7.

CUDC-907 inhibits POMC transcription via the transcription factor Nurr1. (A) AtT20 cells transiently transfected with HDAC1 and HDAC2 were treated with CUDC-907 and POMC mRNA levels were quantitated by real time PCR. (B,C) The expression changes of several known positive and negative regulators of POMC transcription were detected following CUDC-907 treatment compared with vehicle treatment in murine corticotroph tumor cells. Western blot depiction was shown in B and densitometric quantitation compared to actin loading control was shown in C. (D) Nurr1 mRNA quantitation by real time PCR following CUDC-907 treatment (0.125-10 nM for 24 hours). (E) Quantitation of POMC mRNA expression in AtT20 cells transiently transfected with Nurr1 and treated with CUDC-907 (5 nM for 24 hours) compared with vehicle treatment. (F) CUDC-907 treatment effects on activated (Ser/Thr)-Phe phosphorylated Nurr1 in Nurr1 immunoprecipitates (IPs) compared with IgG control IPs. *P < .05; **P < .01; ***P < .01. All results shown were representative of three independent experiments.

CUDC-907 increased expression of cell cycle inhibitors and induced apoptosis

c-Myc has been reported to be a key target mediating the inhibitory effect of CUDC-907 on cell proliferation in several Myc-dependent cancers (27). Although c-Myc has not been shown to be a key driver in pituitary tumorigenesis, c-Myc overexpression has been observed in pituitary tumors (28, 29) where its expression correlated with increased pituitary tumor growth (30). As shown in Fig. 8A, in murine corticotroph tumor cells, CUDC-907 actually appeared to increase c-Myc protein expression (c-Myc/Actin ratio, Veh 1.0 ± 0.02; CUDC-907 3 nM 2.4 ± 0.2, P < .05; 10 nM 1.7 ± 0.2, P < .05, Fig. 8B). Furthermore, corticotroph tumor c-Myc overexpression did not affect the actions of CUDC-907 to inhibit corticotroph tumor cell proliferation (Fig. 8C). As previously reported, CUDC-907 increased expression of the cell cycle inhibitor p27 (p27/Actin ratio, Veh 1.0 ± 0.02; CUDC-907 10 nM 1.7 ± 0.2, P < .05, Fig. 8A and 8B), in addition to its known histone acetylation target Ac-H3K9 (AcH3K9/Actin ratio, Veh 1.0 ± 0.02; CUDC-907 1 nM 2.7 ± 0.2, P < .05; 3 nM 3.7 ± 0.7, P < .05; 10 nM 3.5 ± 0.2, P < .01, Fig. 8A and 8B). Additionally, given its PI3K inhibitory activity, CUDC-907 treatment as expected, blocked AKT activation (0.1 ± 0.01, P < .05, Fig. 8D and 8E) and the phosphorylation of its downstream target 4E-BP1 (0.4 ± 0.03, P < .05, Fig. 8D and 8E). In addition to its actions to reduce ACTH secretion and reduce corticotroph tumor growth, CUDC-907 led to increased activity of the apoptosis executors, caspase-3 and -7, indicating the drug also induces in vitro corticotroph tumor cell apoptosis (Fig. 8F).

Figure 8.

CUDC-907 actions on c-Myc, p27, AKT pathway and corticotroph tumor apoptosis. (A) c-Myc, acetylated histone H3 at lysine 9 (Ac-H3K9). and p27 expression were detected by Western blot following CUDC-907 (1-10 nM for 24 hours) treatment of murine corticotroph tumor AtT20 cells. (D) AKT pathway activation was detected following CUDC-907 (1-10 nM for 3 days). (B,E) Densitometric quantitation compared to actin loading control was performed using the ImageQuant 5.2 software. (C) Cell proliferation rates were detected in AtT20 cells transiently transfected with pD40-His/V5-c-Myc and treated with either CUDC-907 or vehicle. (F) Murine corticotroph tumor AtT20 Caspase-3/7 activation was determined following CUDC-907 treatment (5-40 nM for 24 hours) by the Caspase-Glo3/7 assay (Promega). *P < .05; **P < .01. All results shown were representative of three independent experiments.

Discussion

CD is a debilitating disorder with high morbidity and mortality (31). Medical therapies for CD require long-term compliance, which is reported to be <30%, either due to escape from control of eucortisolemia or drug side effects (32). Using a novel highly sensitive ACTH AlphaLISA to measure ACTH secretion and proliferation in murine and human corticotroph tumor cells in vitro, we have identified an extremely potent (nM) dual HDAC and PI3K inhibitor named CUDC-907 that suppresses both murine and human corticotroph tumor ACTH secretion and growth in preclinical in vitro and in vivo models of CD. Our findings suggest that CUDC-907 could be a promising candidate as medical therapy for CD.

HDACs remove the acetyl group from histone tails, and decrease the accessibility of transcriptional regulatory proteins to chromatin templates, leading to transcriptional repression (33). There are 18 human HDACs grouped into 4 classes according to their substrate preference. Class I (HDAC1, 2, 3, and 8); II (HDAC4, 5, 6, 7, 9 and 10); and IV (HDAC11) are Zn²⁺-dependent enzymes, whereas Class III (SIRT1-7) use NAD⁺ as a reactant to deacetylate acetyl lysine residues of protein substrates (22). HDACs are typically not mutated in human cancers but aberrant overexpression is seen in a variety of tumors and is associated with poor outcome (34). Prior studies in human pituitary tumors have demonstrated increased HDAC1, 2, and 11 mRNA expression, the latter of which was negatively correlated with low p53 expression, suggesting that HDAC11 may interfere with pituitary tumor p53 expression (35).

To date, HDACis have been used to treat primarily hematologic malignancies as well as HIV infection, Alzheimer’s disease, and Friedreich’s ataxia (36). HDACis bind to the catalytic pocket of HDACs to block substrate accessibility, and are grouped into 4 subtypes, namely short-chain fatty acids, benzamides, cyclin peptides, and hydroxamates (37). Valproic acid, together with sodium phenylacetate and phenylbutyrate, belong to the aliphatic acid compounds which exhibit weak HDAC inhibitory effects, and are approved to treat seizures and urea cycle disorders (37). Interestingly, sodium valproate was previously used to treat 5 patients with Nelson’s syndrome in 1981 by Jones et al. based on the compound’s gamma-aminobutyric acid transaminase inhibitory effect, where it reduced plasma ACTH levels and improved headache symptoms and pigmentation (38).

The hydroxamic acid based HDACi, the antifungal compound trichostatin A, exhibited dramatic antiproliferative effects in cancer cells, but side effects limited further development (37). Vorinostat (also known as suberoylanilide hydroxamic acid) and panobinostat (also known as LBH589) are new-generation hydroxamate HDACis and have been approved for treatment of cutaneous T-cell lymphoma (39) and multiple myeloma (40). Vorinostat (0.5–4 μM) was previously shown to inhibit pituitary tumor cell proliferation and ACTH secretion, and suppress murine and human corticotroph tumor growth in vitro and in vivo (41, 42), in part due to the LXRα and upregulation of a mitochondria-mediated death pathway (42).

CUDC-907 is a bivalent drug that integrates a HDAC inhibitory hydroxamate moiety into a core PI3K inhibitor morpholinopyridine scaffold (21). Therefore the cooperative cytotoxic effect of CUDC-907 due to its synergism of HDACis’ pro-apoptotic effects and PI3Ki-mediated survival inhibition is potentially better than prior HDACis (21). Antitumor growth effects of CUDC-907 have been reported in several animal xenograft models of both hematologic (21, 27, 43) and solid cancers (44). We employed 300 mg/kg in this corticotroph tumor in vivo study, and a similar dose for solid cancer (44). Our studies here comparing CUDC-907 with monofunctional PI3Kis and HDACis demonstrated that the IC₅₀ of CUDC-907 in murine and human corticotroph tumor cells is 4-fold lower than panobinostat, and several hundred-fold lower than both vorinostat and PI3Ki monotherapy. Interestingly, although c-Myc has been demonstrated to be a major target of CUDC-907 in some tumors (27, 43), we did not observe c-Myc inhibition in murine corticotroph tumor AtT20 cells. However, we did observe increased expression of the cell cycle inhibitor p27 in addition to activation of proapoptotic Caspase 3 and 7, underpinning the potent antiproliferative and proapoptotic effects of the combinatory HDACi and PI3Ki actions of CUDC-907.

Based on our studies demonstrating the key role of the HDACi moiety, particularly in reducing the expression of the key POMC transcriptional inducer, Nurr1, we propose that the HDACi actions of CUDC-907 rather than its PI3Ki effects contribute predominantly to its inhibition of POMC mRNA expression. Nurr1 was cloned as a predominantly brain-specific gene whose expression pattern is roughly complementary to that of Nur77 and NOR-1 (45). Unlike Nur77/NOR-1, which requires induction, Nurr1 is constitutively expressed in the hypothalamic paraventricular nucleus, which suggests that it participates in regulation of basal expression of tissue-specific genes (24). Nur factors bind a DNA consensus either as a monomer to a Nur-binding response element (NBRE) containing 5′-AAAGGTCA-3′ (46), or as heterodimer to a Nur-responsive element which contains inverted repeat of NBRE-related octanucleotide separated by 6 nucleotides. A naturally occurring cis-acting element of NBRE is found in the proximal POMC promoter region (-71 GAAGGTCA -63) located within the nGRE (47, 48), and is responsible for GR-regulated repression of the POMC promoter activity (47). In addition to the NBRE, there is a Nur-responsive element (-405 TGATATTTACCGCCAAATGCGA -384) located upstream and close to the tissue restricted E-box_{NeuroD1/Pan1,} T-box_Tpit, and PitxRE, which is responsible for CRH-induced POMC transactivation (49). In response to proinflammatroy cytokines, such as TNFα and IL-1β, Nurr1 expression is stimulated through NFκB-dependent transactivation, to further increase POMC expression and activate the HPA axis as a central response to abrogate infection/inflammation stress (50). These observations highlight the importance of Nurr1 in POMC transcriptional regulation and demonstrate that targeting Nurr1 can potently abrogate POMC expression. In addition, given CUDC-907 reduced global protein serine/threonine phosphorylation, we cannot exclude that these PI3Ki effects are limited to Nurr1, and we suspect multiple mechanisms may be involved. Our demonstration that overexpression of Nurr1 blocked CUDC-907-mediated POMC repression, indicates the key role of this NR in CUDC-907-mediated actions.

Although HDACs are traditionally identified as transcriptional repressors, human genome mapping of histone acetyl transferase and HDAC binding has demonstrated that both localize on the chromatin of active genes with acetylated histones, indicating that chromatin is subject to constant remodeling by a dynamic cycle of transient histone acetyl transferase/HDAC binding (51). Therefore, it is likely that in addition to regulating the expression and action of Nurr1, HDACs may also be recruited to POMC where they may be involved in additional cotranscriptional, post-transcriptional, or elongation events (52, 53). In support of this concept, Bilodeau et al. reported reduced or mislocated HDAC2 in glucocorticoid resistant corticotroph tumors, and demonstrated that HDAC2 was recruited to the POMC promoter only when the GR was activated by dexamethasone where it functioned as a transcriptional corepressor (54). Whether this HDAC action occurs in corticotroph tumors that are not GC resistant is unknown.

CUDC-907 is a well-tolerated drug and in clinical trials resulted in 5 complete and 6 partial responses in patients with relapsed/refractory diffuse large B-cell lymphoma (n = 37) (12). The most common adverse events observed were thrombocytopenia (18%), neutropenia (7%), hyperglycemia (7%), and diarrhea (5%) which all resolved with supportive care (11). It must also be noted that these patients with hematologic malignancies were heavily pretreated prior to CUDC-907 and hematologic adverse effects would be more commonly anticipated. CUDC-907 is typically dosed as 60 mg daily for 5 days followed by a 2-day break (5/2 schedule) (11) and this regimen has been used to successfully treat several solid malignancies, including NUT midline carcinoma where it induced stable disease for 32 months (55). Given the potent efficacy we have demonstrated in in vitro and in vivo models of CD, combined with an acceptable safety profile, we propose that CUDC-907 may be a promising novel therapy for patients with CD and should now be tested in clinical trials.

Glossary

Abbreviations

ACTH

adrenocorticotropic hormone

CD

Cushing disease

EC₅₀

median effective concentration

HDAC

histone deacetylase

HTS

high throughput screening

IC₅₀

median inhibitory concentration

Ig

immunoglobulin

NBRE

Nur-binding response element

NR

nuclear receptor

PBS

phosphate-buffered saline

PI3K

phosphoinositide 3-kinase

POMC

proopiomelanocortin

SN

supernatant

TV

tumor volume

Additional Information

Disclosure Statement: The authors have nothing to disclosure.

Data Availability

Restrictions apply to some or all the availability of data generated or analyzed during this study. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.