Introduction

Retinoids are critical for embryonal organogenesis, in particular neural crest and mesoderm-derived organs including the adrenal gland. Several lines of evidence support the role of retinoids in early stage adrenal differentiation1 and zona fasciculata organization2. In adult life, retinoids are known to exert an antiproliferative effect in a variety of cells, including skin, breast and neuronal cells3, and currently play a role in treatment of acute promyelocytic leukemia4 and possibly other tumors, such as, neuroblastoma, breast cancer, melanoma, 3. Retinoic acid has also been tested in adrenal cancer and shown to modulate corticosteroid secretion and cell proliferation5,6.

Most recently, retinoids have been proposed for the treatment of Cushing's disease, a severe endocrine disorder caused by an excess cortisol secretion due to a pituitary corticotropin (ACTH)-secreting tumor7. In vitro studies revealed that all-trans retinoic acid as well as the 9-cis derivate inhibit proliferation in a murine corticotrope tumor cell line and blunt ACTH secretion in human corticotrope adenomas5,8,9,10. Our pilot study in patients with Cushing's disease revealed that all-trans retinoic acid (tretinoin) administration is beneficial in these patients11 and a subsequent study with the 13-cis isomer isotretinoin confirmed these promising results12. In detail, administration of tretinoin or isotretinoin reduced markers of cortisol excess in all patients and normalization of urinary free cortisol levels was achieved in approximately 30% of patients11,12. Amelioration of clinical parameters of Cushing’s disease, e.g., blood pressure, weight, glucose control, hirsutism, was also observed during tretinoin or isotretinoin treatment11,12.

Interestingly, the decrease in cortisol secretion during either retinoid was more pronounced that the change in ACTH levels. While this phenomenon is not uncommon in patients with Cushing’s disease treated with pituitary-acting drugs13,14, it could be due to a direct action on adrenal glands. In our first study on normal adrenal cortex tissue, we observed a dual effect of 9-cis retinoic acid: stimulation of cortisol secretion and STAR expression by roughly 1.5-fold on one side and halving of ACTH receptor synthesis on the other15. We thus hypothesized that the efficacy of retinoids in Cushing’s disease could also be due to a direct effect on the adrenal gland and decided to test this hypothesis in hyperplastic adrenal tissue from these patients.

Aim of the present study was to assess the effects of 9-cis retinoic acid on cortisol secretion and on genes involved in steroidogenesis and retinoid action in adrenal glands from patients with Cushing’s disease. In detail, we evaluated 17hydroxylase, StAR (steroidogenic acute regulatory protein), hormone-sensitive lipase E (LIPE), the ACTH receptor (MC2R), as well as known retinoid target genes, such as retinoic acid receptors alpha and beta, liver X receptor (LXR), peroxisome proliferator activated receptor delta (PPARD), chicken ovoalbumin upstream promoter transcription factor 1 (COUP-TF1), sterol regulatory element binding transcription factor 1 (SREBP1), mitochondrial dehydrogenases mND1 and mND6, and genes involved in both pathways, e.g., dosage-sensitive sex-reversal adrenal hypoplasia critical region in the X chromosome (DAX-1) and steroidogenic factor 1 (SF-1).

Results

9-cis Retinoic acid increased cortisol secretion in adrenal primary cultures from patients with ACTH-dependent Cushing’s syndrome. Cortisol concentrations on average doubled with respect to control wells for both 10 nM and 100 nM 9-cis retinoic acid (Fig. 1a, both p < 0.05); a lesser increase in cortisol was observed with 1 µM 9-cis retinoic acid (Fig. 1a). 9-cis Retinoic acid also increased expression of StAR and CYP17A (Fig. 1b), and decreased NR0B1 (DAX1, Table 1) compared to untreated wells.

Figure 1
figure 1

Effect of 9-cis retinoic acid on baseline cortisol secretion and gene expression in adrenal cultures from patients with Cushing’s disease. (a) Mean cortisol secretion after 24 h incubation with 10 nM–1 µM 9-cis retinoic acid; (b) expression of STAR, CYP17A1 and MC2R after 24 h incubation with 10 nM 9-cis retinoic acid (RA); (c) expression of RARA, RARB and SREBP1 after 24 h incubation with 10 nM 9-cis retinoic acid (RA). Data is expressed relative to untreated, control wells: equal to 100% for cortisol secretion and equal to 1 for gene expression. White bar: control, black bars: 9-cis retinoic acid (RA). *p < 0.05 vs control.

Table 1 Gene expression in adrenal cultures from patients with Cushing’s disease.

As regards the expression of genes related to the retinoic acid pathway, 9-cis retinoic acid increased the expression of both retinoic acid receptor alpha and beta and the transcription factor SREBP1 (Fig. 1c). No significant changes during 9-cis retinoic acid treatment were observed for other factors involved in steroidogenesis, e.g., MC2R, SF-1, LIPE or in retinoic acid action, e.g., liver X receptor, PPARD, COUP-TF1 and the mitochondrial dehydrogenases, i.e., mt-ND1, mt-ND6, (Table 1), compared to control samples.

As expected, incubation with ACTH increased cortisol secretion (335.79 ± 134.64% control, p < 0.05 vs unchallenged wells) and induced the expression of CYP17A1, LIPE , MC2R and StAR. Interestingly, ACTH also induced RARB gene expression; no other retinoid acid-related gene was modified during ACTH incubation (Table 1).

9-cis Retinoic acid blunted ACTH-stimulated MC2R expression by roughly 50% (Fig. 2b) and did not affect the ACTH-induced cortisol response (Fig. 2a) nor ACTH-induced changes in CYP17A and StAR (Fig. 2b). This was replicated at analysis of gene expression normalized to control wells, as the changes in steroidogenic gene expression during retinoic acid-ACTH co-incubated wells were comparable to wells incubated with ACTH alone (Table 1). As regards modulation of retinoic acid receptors, expression of RARA and RARB expression was further enhanced by retinoic acid in ACTH-stimulated wells (Fig. 2c; Table 1). No significant changes compared to ACTH alone were observed for the genes involved in retinoic acid action (NR1H3, PPARD, NR2F1, mt-ND1 and mt-ND6, Table 1).

Figure 2
figure 2

Effect of 9-cis retinoic acid on ACTH-stimulated cortisol secretion and gene expression in adrenal cultures from patients with Cushing’s disease. (a) mean ACTH-stimulated cortisol secretion after 24 h incubation with 10 nM–1 µM 9-cis retinoic acid; (b) expression of STAR, CYP17A1 and MC2R after 24 h incubation with 10 nM ACTH and 10 nM 9-cis retinoic acid (RA); (c) expression of RARA, RARB and SREBP1 after 24 h incubation with 10 nM ACTH and 10 nM 9-cis retinoic acid (RA). Data is expressed relative to ACTH-treated wells: equal to 100% for ACTH-stimulated cortisol secretion and equal to 1 for ACTH-stimulated gene expression; Grey bar: 10 nM ACTH, black bars: 9-cis retinoic acid (RA). *p < 0.05 vs ACTH.

We analysed cortisol secretion during ACTH/retinoid co-incubation by two approaches: the effect of 9-cis retinoic acid on the cortisol response to ACTH was compared to cortisol leves with 10 nM ACTH and the retinoid (10 nM: 92.2 ± 12.8% ACTH; 100 nM: 115.3 ± 16.8% ACTH; 1 µM: 123.6 ± 13.3% ACTH, all comparisons N.S., Fig. 2a) and to cortisol levels with each 9-cis retinoic acid concentration without ACTH (10 nM: 112.9 ± 13.7% RA; 100 nM: 144.6 ± 28.1% RA; 1 µM: 158.6 ± 32.5% RA, all comparisons N.S.); of note, the expected increase in cortisol levels with ACTH is over 300% of unchallenged wells (see above).

Discussion

Retinoids are known modulators of adrenal embryonic development1,16,17 but effects appear to extend beyond adrenal organogenesis. All-trans and 9-cis retinoic acid have been shown to stimulate steroidogenesis in both adrenal and gonadal murine models18,19,20. Conversely, these retinoid agonists proved inhibitors of corticosteroid secretion and adrenal cell proliferation in mouse and human neoplastic adrenal cell lines5,6; on note, retinoic acid signalling pathway stands out in adrenal tumor microarray and miRNA analysis21. Retinoids act mainly via homo- or heterodimerization of ligand-activated RAR and RXR receptors with all-trans retinoic acid and 13-cis retinoic acid acting as RAR agonists and 9-cis retinoic acid binding both RAR and RXR3. These receptors may also heterodimerize with other nuclear receptors, such as PPAR gamma and LXR, or act—as has been shown for 9-cis retinoic acid- on mitochondrial RXRs22, thus adding an additional layer of complexity to retinoid action.

Retinoids has also been shown to affect tumoral corticotrope cells, corticotropes being the main regulators of adrenal cortisol secretion. In fact, studies on both the murine corticotrope cell line AtT-20 and in human adenomatous corticotrope primary cultures5,8,9,10 revealed that both all-trans and 9-cis retinoic acid can inhibit ACTH synthesis and secretion. Both bone morphogenic protein 4 (BMP4), a transcription factor involved in pituitary tumorigenesis, and COUP-TFI, a negative regulator of retinoic acid response pathways, modulated the action of all-trans and 9-cis retinoic acid action in tumoral corticotropes5,8,23. Interestingly, long term (48–192 h) incubation of AtT-20 cells with tambicarotene (Am80), a synthetic RAR alfa and beta agonist, revealed increased ACTH secretion as well as Pomc and Tpit expression24; at 24 h—the time frame evaluated in other experiments5,8,9,10- tambicarotene did not affect ACTH secretion and Pomc secretion, suggesting different involvement of RARs and RXR over time.

This evidence led to clinical trials with tretinoin and isotretinoin in patients bearing an ACTH-secreting pituitary adenoma, i.e., Cushing’s disease11,12, with promising results. So far, over 20 patients with this severe endocrine disorder have been tested and contaiment of cortisol excess could be observed in up to one third of patients, much like it occurs with other pituitary-acting drugs13,14. Interestingly, although the rationale for efficacy of retinoids in these patients rests on evidence collected on the tumoral corticotrope—thus inhibition of ACTH is expected to drive the reduction in adrenal secretion-, the decrease in cortisol levels appeared more pronounced and not strictly parallel to ACTH concentrations. This has been shown to occur with other drugs aimed at the pituitary but, given the known link between retinoids and the adrenal gland, could also be due to a direct action on adrenal cortex cells.

We therefore decided to pursue this avenue of investigation and tested 9-cis retinoic acid first in normal human adrenal tissue. Our study demonstrated that 9-cis retinoic acid stimulates spontaneous cortisol secretion as well as synthesis of STAR15, the rate limiting enzyme for adrenal cholesterol availability, in adrenal cells. At the same time, 9-cis retinoic acid markedly blunted expression of the ACTH receptor, i.e., melanocortin type 2 receptor MC2R, and reduced upregulation of MC2R induced by ACTH itself. The effects of 9-cis retinoic acid on adrenals appeared therefore two-fold: enhancement of spontaneous cortisol secretion and reduction of the ACTH receptor synthesis, suggesting a modulatory role in intraadrenal negative feedback regulation.

In patients with Cushing’s disease, adrenals are continuously exposed to elevated ACTH levels and it stands to reason that the effect of 9-cis retinoic acid on ACTH receptor expression may come to play a major role. Of note, long-standing exposure to excess ACTH usually leads to the development of hyperplastic adrenals in these patients25, attesting to the preeminent role of ACTH on adrenal secretion and trophism. Given the above, we decided to expand upon our previous findings and investigate the effect of 9-cis retinoic acid on adrenals from patients with Cushing’s disease.

Incubation with 9-cis retinoic acid led to doubling of spontaneous cortisol secretion in primary cultures established from adrenal tissues collected patients with Cushing’s disease, indicating that the stimulatory effect observed in normal adrenal tissue15 is mantained in hyperplastic adrenal cortex. Ancillary to increased cortisol secretion, we observed an increase in STAR expression—thereby enhancing cholesterol flux to the mitochondrion26- in keeping with results obtained with all-trans and 9-cis retinoic acid on gonadal and adrenal steroidogenic cell models15,19. 9-cis Retinoic acid also increased CYP17A1 expression, thus potentiating microsomial steroidogenesis; similar results have been reported with all-trans retinoic acid in a murine tumoral Leydig cell line27. As regards the first enzyme of the steroidogenic cascade, cholesterol side-chain cleavage or P450SCC, a primary target of ACTH stimulation, no changes in CYP11A1 expression or protein levels with either retinoid have been reported by us15 and other investigators18,27.

9-cis Retinoic acid also acted upon known targets such as its own receptors, RARA and RARB28, DAX-1, a transrepressor crucial to adrenal development17, and SREBP-1, a sterol regulatory element29. Of note, DAX-1 is a major inhibitor of STAR30 and both STAR31 and CYP17A132 are modulated by SREBP-1. Altogether, this evidence suggests that 9-cis retinoic acid stimulates cortisol secretion via a concerted involvement of DAX-1, SREBP-1, STAR and CYP17A1 in hyperplastic adrenal cells.

In contrast, the most striking effect observed during 9-cis retinoic acid and ACTH co-incubation was the marked reduction in ligand-induced upregulation of the ACTH receptor33,34,35. The ACTH receptor is crucial to cortisol secretion, indeed, mutations in MC2R are associated with severe and often fatal cortisol deficiency36. In this context, our findings on MC2R in hyperplastic adrenal tissues collected from patients with Cushing’s disease confirm and extend results obtained during ACTH stimulation in normal adrenals15,34,35: in addition to MC2R, ACTH proved a strong inducer of STAR, CYP17A1 and LIPE, the hormone-sensitive lipase crucial to adrenal steroidogenesis37.

The stimulatory effect of 9-cis retinoic acid on cortisol secretion and steroidogenic enzyme expression could not be observed in wells co-incubated with ACTH. In fact, cortisol secretion was comparable to levels observed with ACTH alone for 9-cis retinoic acid concentrations up to 1 μM as were DAX-1, STAR and CYP17A1 expression. It is tempting to speculate that the reduction in MC2R induced by 9-cis retinoic acid dampens the adrenal response to ACTH, thus overriding its stimulatory effect on spontaneous cortisol secretion. In our previous study on normal adrenal tissue, the stimulatory effect of 9-cis retinoic acid on ACTH-stimulated cortisol secretion and STAR expression was modest or not significant15. It stands to reason that the stimulatory effect is abolished in ACTH-dependent adrenal hyperplasia, given the preeminent role of ACTH and, thus, MC2R.

Interestingly, ACTH also increased retinoic acid receptor beta (RARB) expression and, further, ligand-induced upregulation of both receptor isoforms was enhanced during ACTH/9-cis retinoic acid co-incubation. Of note, RARB has been shown to be more sensitive than RARA to upregulation by retinoids in some cell models28. Given that the adrenal gland itself produces endogenous retinoid acid16, these changes could come into play in an intraadrenal retinoic acid-ACTH circuit.

In patients with Cushing’s disease, cortisol hypersecretion is driven by ACTH produced by tumoral corticotropes. Although plasma ACTH concentrations are not always markedly elevated, they prove sufficient to determine excess cortisol secretion by the adrenal. In fact, the adrenal MC2R is believed to be upregulated in these patients by virtue of long-standing ACTH stimulation and, further, markedly reduced MC2R expression underlies the absent cortisol response to ACTH testing in patients with Cushing’s disease submitted to surgery38. Along the same line, the reduction of MC2R induced by 9-cis retinoic acid could play a major role in its therapeutic efficacy in Cushing’s disease. As mentioned above, the decrease in cortisol levels observed in patients treated with both tretinoin11 and isotretinoin12 appeared greater and not strictly time-related to changes in ACTH concentrations. The decrease in ACTH is expected by virtue of its action in human tumoral corticotropes5,10 whereas the observed down-regulation of MC2R could contribute to explain a greater decrease in cortisol.

In conclusion, our findings indicate that although 9-cis retinoic acid stimulates unchallenged cortisol secretion, in presence of ACTH the decrease in adrenal ACTH receptor overrides this effect. Thus, an adjunctive, adrenal action could play a role in the efficacy of retinoids in patients with Cushing’s disease.

Material and methods

Adrenal cultures

Adrenals obtained from 6 patients with Cushing’s disease submitted to adrenalectomy were established in culture according to our usual protocols15,39. In brief, the adrenal medulla was carefully removed and adrenal cortex fragments were minced, digested in 0.1% collagenase, plated at approx. 300,000 cells/well, incubated in DMEM supplemented with 10% fetal bovine serum and antibiotics for 3–5 days to allow attachment.

Treatments and assays

Treatment were performed in DMEM containing 0.1% BSA. Cells were incubated with 10 nM, 100 nM and 1 μM 9-cis retinoic acid (Sigma Aldrich, St. Louis, USA) with or without 10 nM ACTH for 24 h. Pregnenolone (10 μM) was included in the test medium in order to promote steroidogenesis40. Both pregnenolone and retinoic acid were dissolved in 100% ethanol and diluted 1000-fold in DMEM; equal volumes of ethanol were added to control wells. Treatments were performed in triplicate for each adrenal specimen. Cortisol in medium was measured using Coat-A-Count radioimmunoassay (Siemens Healthcare Diagnostics, Erlangen, Germany) and normalized to unchallenged or ACTH-stimulated wells, respectively, given the considerable variability in cortisol medium concentrations among specimens (from 360 ng/ml to 990 ng/ml after 24 h incubation).

Quantitative real-time PCR

RNA was extracted from plated cells using TRIzol reagent (Life Technologies, Carlsbad, USA) according to the manufacturer's instruction. The amount and quality of RNA were checked on nanophotometer (Implen GmbH, München, Germany) and 100 ng RNA reverse-transcribed with Superscript Vilo cDNA Synthesis Kit (Life Technologies, Carlsbad, USA). Quantitative Real-Time PCR was performed on a 7900 HT sequence Detection System (Applied Biosystems, Foster City, USA), using the Platinum Quantitative PCR Supermix-UDG with ROX (Life Technologies, Carlsbad, USA) and TaqMan assays. The following genes were evaluated: STAR Hs00264912_m1, MC2R Hs00300820_s1, NR1H3 (LXRa) Hs00172885_m1, NR0B1 (DAX-1) Hs00230864_m1, PPARD Hs00987011_m1, NR2F1 (COUP-TF1) Hs00818842_m1, CYP17A1 Hs01124136_m1, LIPE Hs00943410_m1, RARA Hs00940446_m1, RARB Hs00977140_m1, mt-ND1 Hs02596873_s1, mt-ND6 Hs02596879_g1, NR5A1 (SF-1) Hs00610436_m1, SREBP1 Hs01088679_g1 and normalized to RPLP0 Hs99999902_m1. Expression data was analyzed as 2−ΔΔCt and expressed as fold change vs control or ACTH. Changes in gene expression were evaluated in wells treated with 10 nM 9-cis retinoic acid.

Statistical analyses

Data is expressed as mean ± standard error of the mean (S.E.M.) relative to unchallenged or ACTH-stimulated wells for each adrenal specimen. Comparisons were performed using Wilcoxon's signed-rank test and statistical significance accepted at p ≤ 0.05.

Study approval

The study was approved by the Ethical Committee of the Istituto Auxologico Italiano (project #02C402) and carried out according to guidelines established by the Declaration of Helsinki.

Informed consent

Informed consent for secondary use of surgical tissues obtained from patients by the referring physician prior to surgery.