Recurrent somatic mutations underlie corticotropin-independent Cushing’s syndrome

See allHide authors and affiliations

Science  23 May 2014:
Vol. 344, Issue 6186, pp. 917-920
DOI: 10.1126/science.1252328


Cushing’s syndrome is caused by excess cortisol production from the adrenocortical gland. In corticotropin-independent Cushing’s syndrome, the excess cortisol production is primarily attributed to an adrenocortical adenoma, in which the underlying molecular pathogenesis has been poorly understood. We report a hotspot mutation (L206R) in PRKACA, which encodes the catalytic subunit of cyclic adenosine monophosphate (cAMP)–dependent protein kinase (PKA), in more than 50% of cases with adrenocortical adenomas associated with corticotropin-independent Cushing’s syndrome. The L206R PRKACA mutant abolished its binding to the regulatory subunit of PKA (PRKAR1A) that inhibits catalytic activity of PRKACA, leading to constitutive, cAMP-independent PKA activation. These results highlight the major role of cAMP-independent activation of cAMP/PKA signaling by somatic mutations in corticotropin-independent Cushing’s syndrome, providing insights into the diagnosis and therapeutics of this syndrome.

Candidate Cushing's culprit identified

Cushing's syndrome is a rare condition resulting from the excess production of cortisol. About 15% of Cushing's syndrome cases are associated with an adrenocortical tumor. However, the genetic etiology of these adrenocortical tumors is ill defined (see the Perspective by Kirschner). Cao et al. and Sato et al. both performed whole-exome sequencing of tumors from individuals with adrenal Cushing's syndrome and compared it with the patient's own matched non-tumor DNA and identified recurrent mutations in the protein kinase A catalytic subunit alpha (PRKACA) gene, as well as less frequent mutations in other putative pathological genes. The most common recurrent mutation activated the kinase, which may suggest a potential therapeutic target.

Science, this issue p. 913, p. 917; see also p. 804

Cushing’s syndrome is a systemic disorder associated with various constitutive symptoms—such as hypertension, impaired glucose tolerance, central obesity, osteoporosis, and depression—that are ascribed to cortisol overproduction (13). Cortisol biosynthesis is primarily regulated by corticotropin secreted from the anterior pituitary gland. Corticotropin acts by binding to the melanocortin-2 receptor, increases cyclic adenosine monophosphate (cAMP) production, and activates cAMP-dependent protein kinase [protein kinase A (PKA)] (4). While aberrant corticotropin secretion from pituitary adenoma or other ectopic sites is the leading cause of Cushing’s syndrome (corticotropin-dependent Cushing’s syndrome), excess cortisol is autonomously produced by adrenocortical tumors in patients with corticotropin-independent Cushing’s syndrome (2, 5, 6). Although mutations in the regulatory subunit type I alpha of PKA (PRKAR1A) (7, 8) guanine nucleotide-binding protein subunit alpha (GNAS) (9), phosphodiesterase-8B (PDE8B) (10, 11) and -11A (PDE11A) (12), and armadillo repeat containing 5 (ARMC5) (13) are responsible for rare syndromic or hereditary disorders with bilateral adrenocortical hyperplasia, molecular pathogenesis of cortisol-producing adrenocortical adenomas, which account for a large portion of corticotropin-independent Cushing’s syndrome (6), are less studied.

To investigate genetic lesions in corticotropin-independent Cushing’s syndrome, we performed whole-exome sequencing (WES) of eight adrenocortical tumors and matched normal specimens (fig. S1 and table S1). We identified a total of 45 validated nonsynonymous and 59 putative synonymous somatic mutations (tables S2 and S3) (see the supplementary materials). Remarkably, the gene encoding the catalytic subunit (C subunit) of PKA (PRKACA) was recurrently mutated in four out of the eight cases, resulting in an identical c.T617G mutation predicted to cause a conversion of leucine to arginine at amino acid position 206 (p.L206R). Together with a GNAS mutation (p.R201C), five of the eight cases had somatic mutations in genes involved in the cAMP/PKA signaling pathway. No allelic imbalances were observed at both gene loci in single-nucleotide polymorphism (SNP) array analysis (fig. S2), indicating that these mutations were heterozygous, which was consistent with the observation that, in exome sequencing, variant allele frequencies (VAF) of PRKACA/GNAS mutations were comparable to those of other somatic mutations (fig. S3). Mutations were not detected in any other known causative genes, including PRKAR1A, PDE11A, PDE8B, and ARMC5.

We performed follow-up sequencing of PRKACA and GNAS as well as previously reported genes (PRKAR1A, PDE11A, PDE8B, and ARMC5) in an additional 57 cases (see the supplementary materials). The L206R mutation in PRKACA was found in 30 out of the 57 follow-up cases, of which 24 cases were confirmed as being somatic, whereas GNAS mutations were found in 10 cases, with somatic origin being confirmed in six cases (table S1, Fig. 1, and fig. S4). No mutations were found in the previously reported genes. The eight samples double-negative for PRKACA and GNAS mutations were tested for mutations in seven additional genes that were mutated and expressed in three double-negative exome cases, but no more recurrent mutations were identified. Combined with the four PRKACA and one GNAS mutations in the discovery cases, PRKACA and GNAS were mutated in 34 (52.3%) and 11 (16.9%) out of the 65 cases with corticotropin-independent Cushing’s syndrome, respectively, where both mutations were completely mutually exclusive (Fisher’s exact test, P = 9.46 × 10−5) (Fig. 1). The somatic origin was confirmed for 28 out of 28 PRKACA and 6 out of 6 GNAS mutations thus far tested. In addition, VAFs of PRKACA and GNAS mutations in deep sequencing were distributed between 0.08 and 0.35, whereas those of most heterozygous SNPs (83%) were between 0.4 and 0.6 (fig. S5), indicating that most of these mutations were somatic in origin.

Fig. 1 Recurrent mutations in PRKACA and GNAS.

(A) Mutations in PRKACA (top) and GNAS (bottom) identified in 65 patients with corticotropin-independent Cushing’s syndrome (arrowheads). Confirmed somatic mutations are indicated in red. (B) Mutually exclusive distribution of PRKACA and GNAS mutations.

Patients with mutated PRKACA showed significantly higher cortisol levels on the 1-mg dexamethasone suppression test (DST) compared with wild-type PRKACA and GNAS (t test, P = 2.60 × 10−3) (Fig. 2A). PRKACA-mutated adenomas had a significantly smaller tumor diameter than those with no known mutations (t test, P = 4.93 × 10−5) (Fig. 2B), suggesting that PRKACA-mutated adenomas may have higher cortisol production. GNAS-mutated adenomas also showed higher cortisol levels on 1-mg DST and have a smaller tumor size. Seventy-six percent of the patients with clinical Cushing’s syndrome had a mutation of either gene, whereas only two of nine patients with subclinical Cushing’s syndrome had mutated PRKACA or GNAS genes (Fisher’s exact test, P = 2.87 × 10−3) (Fig. 2C and table S4), indicating that these mutations were enriched for clinical Cushing’s syndrome.

Fig. 2 Relationship between mutation status and clinical features.

(A and B) Serum cortisol level of 1-mg dexamethasone (A) and the diameter of adrenocortical adenoma (B) according to the mutation status of PRKACA/GNAS genes. (C) Clinical and subclinical Cushing’s syndrome as attributed to mutated PRKACA, GNAS, or other, currently undetected, cause.

PRKACA is the catalytic (C) subunit of the tetrameric PKA holoenzyme, which binds tissue-specific dimeric regulatory (R) subunits (1416). In the native state, the C subunit is kept inactivated through the binding of the R subunit, which masks the catalytic site of the C subunit (Fig. 3A). However, when intracellular cAMP is up-regulated by external stimuli, each R subunit binds two cAMP molecules, which causes a conformational change in the R subunit to promote dissociation of the C subunit from the PKA complex (1416), allowing for the translocation of the dissociated, and thereby activated, free C subunit to the nucleus and phosphorylation of its target substrates therein (17, 18). The highly conserved L206 residue of PRKACA, which resides within the P+1 loop of the C subunit, lies on the surface of the large lobe of PRKACA and is thought to be essential for the catalytic activity of the kinase (Fig. 3A, and fig. S6) (1416). In the absence of cAMP, the inhibitory region of the R subunit docks to the active site cleft of the C subunit, including the P+1 loop. The L206 residue is located at the interface between the C subunit and the inhibitory region in the R subunit to form a hydrophobic interaction with the I99 residue of the R subunit (Fig. 3B). Thus, the substitution from the small hydrophobic leucine to a large hydrophobic arginine is predicted to cause steric hindrance and abolish the binding of the C and R subunits (Fig. 3C), resulting in constitutive, cAMP-independent activation of PKA.

Fig. 3 Effect of L206R mutation on three-dimensional structures of PKA.

(A) Three-dimensional structure of the PKA complex, composed of C (PRKACA) (pink) and R (PRKAR1A) (cyan) subunits, is depicted using the University of California–San Francisco Chimera program, based on the Research Collaboratory for Structural Bioinformatics Protein Data Bank (PDB ID: 2QCS). L206 is shown in red. (B and C) A predicted effect of the L206R mutation within the P+1 loop on the interaction with the R subunit (C) in comparison with wild-type PRKACA (B).

In fact, when the PKA complex was reconstituted in vitro using purified proteins (fig. S7), PRKAR1A binds the wild-type C subunit and suppresses its PKA activity in the absence of cAMP, and the suppression is recovered in the presence of cAMP with substantially reduced interaction with the wild-type C subunit (Fig. 4, A and B). In contrast, the R subunit can no longer bind the L206R PRKACA mutant with constitutive PKA activation, regardless of the presence or absence of cAMP (Fig. 4, A and B). The loss of binding to the R subunit and consequent cAMP-independent PKA activation for the mutant PRKACA was also demonstrated in vivo. When expressed in human embryonic kidney 293T (HEK293T) cells (fig. S8), wild-type PRKACA, but not the L206R PRKACA mutant, coimmunoprecipitated with PRKAR1A (Fig. 4C). Both mock- and wild-type PRKACA-transduced cells showed increased PKA activity accompanied by an elevated phosphorylation of cAMP response element–binding protein (pCREB), one of the major downstream targets of PKA activation (Fig. 4, D and E, and fig. S9), on cAMP induction by forskolin treatment (see the supplementary materials), whereas mutant PRKACA-transduced cells demonstrated a higher basal level of PKA activity and CREB phosphorylation, regardless of forskolin treatment (Fig. 4, D and E, and fig. S9). To confirm the cAMP-independent activation of the mutant PRKACA, we examined the effect of two PRKACA inhibitors (H89 and KT5720) and a competitive inhibitor of cAMP binding for the R subunit (Rp-cAMPS) on the activity of the L206R mutant (see the supplementary materials). PKA activation in the L206R mutant–transduced cells in the absence of forskolin was suppressed by H89 and KT5720 (t test, P = 1.60 × 10−3 and P = 1.86 × 10−3, respectively) but not in Rp-cAMPS–treated cells, supporting further that the consequence of the L206R mutation is constitutive, cAMP-independent activation of PKA (Fig. 4F and fig. S10).

Fig. 4 Functional characterization of PRKACA mutant.

(A) Immunoblot analysis of antibody to FLAG immunoprecipitates of the PKA reconstructed in vitro with large-scale–purified proteins, in which PRKACA and PRKAR1A subunits are tagged with hemagglutinin (HA) and FLAG, respectively. (B) In vitro PKA activities of wild-type and mutant PRKACA in the presence or absence of PRKAR1A and cAMP with purified proteins. (C) Immunoblot analysis of the total cell lysates (input) and antibody to HA immunoprecipitates from HEK293T cells stably transduced with either mock, wild-type, or mutant PRKACA tagged with HA with indicated antibodies. (D) PKA activities in HEK293T cells stably transduced with either mock, wild-type, or mutant PRKACA in the presence or absence of forskolin stimulation. (E) Relative expression of pCREB to total CREB in HEK293T cells stably transduced with either mock, wild-type, or mutant PRKACA as determined by densitometory of the immunoblots. (F) The effect of inhibition of PKA (H89 and KT5720) and cAMP (Rp-cAMPS) in HEK293T cells transduced with mock or mutant PRKACA. (G) PRKACA expression standardized for mRNA expression in PRKACA-mutated and unmutated adenomas and matched normal adrenocortical tissues. (H) Western blot analysis of indicated fractions of cell lysates from wild-type and mutant PRKACA-transduced HEK293T cells using indicated antibodies. Standard errors and significant differences are indicated [(C) to (G)].

Finally, as predicted from low stability of free C subunits, primary PRKACA-mutated tumors showed significantly lower PRKACA protein expression compared with unmutated tumors (t test, P = 5.70 × 10−3) and normal adrenocortical tissues (t test, P = 2.09 × 10−2) (Fig. 4G and fig. S11), although no significant difference was observed for downstream signaling, such as pCREB or steroidogenic acute regulatory protein (StAR). The reduced intracellular expression of the mutant PRKACA protein was also observed for exogenously introduced PRKACA in different cell types (Fig. 4H and fig. S12). However, compared with the wild-type protein, mutant PRKACA was more enriched in the nuclear than in the cytoplasmic fraction (Fig. 4H).

In conclusion, frequent somatic mutations in PRKACA and GNAS genes underlie corticotropin-independent Cushing’s syndrome (fig. S13). Strikingly, in accordance with a recent report (19) PRKACA mutations were found in more than 50% of the current cohort. All the mutations were the identical L206R substitution, which prevents binding to the inhibitory R subunits and results in constitutive, cAMP-independent activation of PKA. GNAS mutations were also detected in a substantial fraction of the present cohort (11/65 or 16.9%), in which all were confined to the R201 residue. Steroid hormone biosynthesis in steroidogenic cells is primarily regulated through activation of the cAMP/PKA signaling pathway (20). PRKACA and GNAS mutations in adrenocortical cells affecting this pathway are thus a likely factor responsible for the excessive production of cortisol and together account for as many as 70% of the current cohort of the patients. In contrast, the remaining 30% of the patients with no mutations in either gene tended to show lower cortisol levels on 1-mg DST and had a larger tumor size compared with PRKACA-mutated patients. In addition, the identified mutations were confined almost exclusively to patients with clinical Cushing’s syndrome, suggesting that double-negative cases have distinct pathogenesis, with driver mutations still to be identified. It should be warranted to identify the genetic basis of double-negative cases in future studies.

Supplementary Materials

Materials and Methods

Figs. S1 to S13

Tables S1 to S8

References (2126)

References and Notes

  1. Acknowledgments: We thank Y. Mori, M. Nakamura, N. Mizota, S. Ichimura, and M. Yamakawa for their technical assistance. The retroviral vector, pGCDNsamIRES-EGFP, was kindly provided by M. Onodera (National Research Institute for Child Health and Development). This work was supported by the Japan Society for the Promotion of Science (JSPS) through Grants-in-Aid for Scientific Research (KAKENHI) grant number 22134006 and the Funding Program for World-Leading Innovative Research and Development on Science and Technology (FIRST Program). Data for exome sequencing, as well as SNP array analysis, are found in the European Genome-phenome Archive (EGA) under accession EGAS00001000661. Author contributions: Y.Sa., S.Ma., K.Y., Y.N., T.Y., H.S., and A.K. performed library preparation and DNA sequencing. Y.Shira., Y.Shio., K.C., H.T., and S.Mi. were committed to bioinformatics analyses of resequencing data. Y.Sa. and A.S.-O. performed SNP array analysis. Y.Sa., S.Ma., M.S., R.I., K.K., and O.N. performed the functional analyses of PRKACA mutant. T.M., H.K., M.F., and Y.H. provided specimens and were also involved in planning the project. Y.Sa., S.Ma., M.S., R.I., and S.O. generated figures and tables and wrote the manuscript. S.O. led the entire project. All authors participated in the discussion and interpretation of data and results. Y.Sa., S.Ma., M.S., Y.H., and S.O. are inventors on a patent applied for by Kyoto University that covers the inspection method for Cushing’s syndrome and biomarker and theraputic agent thereof (2014-37189).

Stay Connected to Science

Navigate This Article