Gαs–Protein Kinase A (PKA) Pathway Signalopathies: The Emerging Genetic Landscape and Therapeutic Potential of Human Diseases Driven by Aberrant Gαs-PKA Signaling

1. Cholera

Perhaps one of the best examples of Gαs-PKA pathway–mediated pathophysiology is the severe diarrhea caused by infection with Vibrio cholera, or cholera. Cholera continues to be a global health concern, contributing to hundreds of thousands of deaths each year (Ali et al., 2015). Cholera toxin has a unique ability to ADP-ribosylate Gαs at arginine 201. The addition of an ADP-ribose group inhibits the GTPase activity of Gαs and renders it constitutively active in a manner similar to the disease-associated R201 mutations (discussed in section A. Mutations in GNAS) (Landis et al., 1989; Kaper et al., 1995). Overactivation of Gαs by cholera toxin leads to cAMP production and PKA activation in the intestinal epithelium (Fig. 8A). In crypt cells, PKA activity enhances secretion of Cl⁻ into the intestinal lumen due to direct regulation of the cystic fibrosis transmembrane conductance regulator (CFTR) channel. Under normal physiologic conditions, the degree of phosphorylation of four PKA phosphosites controls the degree of CFTR channel opening. Thus, overactive PKA in response to cholera leads to maximal channel opening. In villous cells, PKA also functions to decrease Cl⁻ absorption by inhibiting Na⁺/Cl⁻ cotransporters and Na⁺/H⁺ exchangers (Goodman and Percy, 2005). As a result of osmotic imbalance, water rapidly moves out of cells into the intestinal lumen, overwhelming reabsorption mechanisms and producing severe, watery diarrhea and dehydration that can prove deadly if left untreated (Fig. 8A). Interestingly, patients with cystic fibrosis are resistant to the effects of cholera toxin as a result of mutations in the CFTR channel. Notably, the majority of patients harbor the F508del mutation in the regulatory region of CFTR. This mutation causes PKA phosphorylation defects that alter trafficking through the endoplasmic reticulum and Golgi to the cell surface as well as disrupt the conformational cues induced by PKA phosphorylation that are critical to channel opening (Kaper et al., 1995; Goodman and Percy, 2005; Bharati and Ganguly, 2011; Chin et al., 2017).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8

Gαs-PKA pathway signalopathy pathophysiology. (A) Pathophysiology of cholera. Cholera is an intestinal parasite that enters the digestive tract when consumed via contaminated water. In the intestinal epithelium, cholera toxin ADP-ribosylates and activates Gαs, leading to overactivation of PKA. PKA directly phosphorylates the CFTR to facilitate channel opening. Efflux of chloride ions disrupts normal ionic gradients, and water passes into the intestinal lumen to compensate. Consequently, the clinical manifestations of cholera include watery diarrhea and dehydration. (B) Cushing syndrome pathophysiology. ACTH is secreted by the pituitary gland in the brain and travels through the bloodstream to the adrenal gland located on top of the kidney. ACTH binds to the melanocortin receptor (MC₂R) on the surface of adrenocortical cells to activate PKA and stimulate cortisol secretion. In Cushing syndrome, loss-of-function mutation in RIα (or gain-of-function mutation in Cα) leads to persistent PKA activation and excess cortisol secretion. Clinical manifestations of the disease exacerbate the effects of cortisol and include hypertension, hyperglycemia, and obesity. (C) Fibrous dysplasia pathophysiology. Fibrous dysplasia is a postzygotic disease caused by activating mutation in GNAS. Persistent activation of PKA in mesenchymal stem cells impairs proper differentiation to adipocyte, chondrocyte, and osteogenic lineages. In particular, accumulation of osteogenic precursors shifts the balance of osteoblasts and osteoclasts to favor bone resorption by osteoclasts. Resulting clinical manifestation of the disease includes brittle bone and frequent fracture or deformity.

1. Carney Complex

Carney complex is a rare disease that is characterized by multiple neoplasms of both endocrine (commonly adrenal, pituitary, or thyroid glands and gonadal tissues) and nonendocrine tissues (commonly heart, skin, or eye). First described in 1985, only about 750 individuals have been diagnosed worldwide (Correa et al., 2015). Interestingly, 70% of the cases are familial, following autosomal dominant inheritance patterns, with the majority of patients having inactivating mutations in PRKAR1A (Kirschner et al., 2000b; Bertherat et al., 2009). Additionally, 35% of sporadic cases are also caused by these same mutations (Kirschner et al., 2000b). In fact, Carney complex was the first disease to be associated with mutations in the PKA holoenzyme (Kamilaris et al., 2019). As mentioned previously, the vast majority of mutations are not actually expressed due to NMD, creating PRKAR1A haploinsufficiency, ultimately resulting in catalytic subunit hyperactivity (Bertherat et al., 2009). Aligned with this concept, a patient with Carney complex with copy number gains in PRKACB has also been documented (Forlino et al., 2014).

Carney complex is a heterogeneous disease with typical onset around age 20, but some patients have even been diagnosed as children (Correa et al., 2015). Interestingly, patients with PRKAR1A mutations tend to present at a younger age with specific phenotypes (Bertherat et al., 2009). Most patients present with Cushing syndrome (see section 2. Cushing Syndrome and Adrenocortical Adenomas) and endocrine phenotypes. One of the most common physical characteristics is the presence of pigmented skin lesions, like café-au-lait spots, caused by the hyperproliferation of melanocytes (also seen in McCune-Albright syndrome; see section 4. Fibrous Dysplasia and McCune-Albright Syndrome). Another common characteristic is cardiac myxoma, a neoplasm of the heart. Cardiac myxoma represents a major cause of mortality in Carney complex because of its rapid growth and recurrence, resulting in obstruction of blood flow in the heart (see section 3. Cardiac Myxoma) (Wang et al., 2018b). Finally, the most common endocrine phenotype is primary pigmented nodular adrenocortical disease, affecting up to 60% of patients with Carney complex. As the name suggests, it manifests as pigmented nodules on the adrenal gland (Bertherat et al., 2009). This results in adrenocorticotropic hormone (ACTH)-independent Cushing syndrome, which is discussed in the next section. Interestingly, primary pigmented nodular adrenocortical disease can occur outside of Carney complex and not only is caused by mutations in PRKAR1A but can also be caused PDE8B or PDE11A mutations (Bertherat et al., 2009; Kamilaris et al., 2019) (Fig. 7A; Supplemental Table 2). This highlights that overactive PKA is a driver of this disease, regardless of how it is achieved.

Similarly, the physical manifestations of the disease are in line with the importance of PKA signaling to the cell types affected by Carney complex. In these tissues, normal programs such as growth and development and energy metabolism are driven by the hormone-GPCR-Gαs-PKA signal transduction axis (see section F. Metabolic Regulation for more information on energy metabolism). Acting through cAMP second messengers, PKA mediates systemic responses to hormones of the pituitary, adrenal gland, thyroid, parathyroid, and hypothalamus as well as more local responses in tissue such as the pancreas, kidney, liver, and gonads (Tilley and Fry, 2015). Of note, individual mutations in their cognate GPCRs can also cause endocrine phenotypes related to Carney complex or other Gαs-PKA pathway signalopathies (Lania et al., 2006). However, when dysregulation of this signaling occurs through loss of RIα function, it typically results in neoplastic growth and tumorigenesis across these tissues. In fact, as evidence to the importance of PKA in global growth and development, Prkaca knockout mice weigh 65% less than control littermates and exhibit a significant growth delay (Skålhegg et al., 2002).

2. Cushing Syndrome and Adrenocortical Adenomas

Cushing syndrome is a rare disease that affects around two individuals per million per year across the world (Steffensen et al., 2010). It can present with very broad symptoms, including hypertension, hyperglycemia, obesity, skin changes, mood disorders, and other hormonal changes. Although these symptoms can have multiple etiologies, Cushing syndrome is specifically characterized by exposure to excess cortisol (Sharma et al., 2015). Cortisol is a hormone that helps control the stress response by regulating blood pressure and blood sugar as well as dampening the immune response. The release of cortisol is regulated by ACTH, which is secreted by the pituitary glands at the base of the brain. Once in the bloodstream, ACTH travels to the adrenal gland, located on top of the kidneys, where it binds the melanocortin receptor (MC₂R). MC₂R is a Gαs-linked GPCR located on the surface of the adrenocortical cells, which when stimulated activates PKA to trigger cortisol secretion (Fig. 8B).

Cushing syndrome has many etiologies, including overuse of glucocorticoid medication, ACTH-secreting pituitary tumors (termed Cushing disease), or cortisol-secreting adrenocortical adenomas (Sharma et al., 2015). Although rare, Cushing syndrome can also have genetic causes that converge on overactivation of the PKA pathway. One of the most common genetic causes of Cushing syndrome is the PRKACA L206R mutation. As mentioned previously, L206R disrupts regulatory subunit contacts, leading to constitutive PKA activity. This mutation, along with loss-of-function mutations in PRKAR1A, underlie ACTH-independent Cushing syndrome (Fig. 8B). Similarly, germline PRKACA copy number gains (Beuschlein et al., 2014; Lodish et al., 2015) and somatic PRKACB S54L mutations can also cause cortisol-producing adrenocortical adenomas/hyperplasias and Cushing syndrome (Espiard et al., 2018). Somatic mutations in GNAS and PDE8D have also been identified (Espiard et al., 2018). In general, patients with PKA gene mutations have earlier onset of disease with more comorbidities. There is some evidence, at least for germline PRKACA amplifications, that this is a dose-dependent effect, with patients harboring PRKACA triplication having the most severe symptoms and earliest onset (Lodish et al., 2015). Interestingly, patients with GNAS and PRKACA mutations have smaller tumor sizes, which is a sign that the tumor is capable of efficient cortisol production and secretion (Goh et al., 2014). This finding is also in line with the role of cAMP in controlling regulated exocytosis, which contributes to hormone secretion in endocrine cells. For instance, in the pituitary, cAMP increases the size of secretory granules (Seino and Shibasaki, 2005), and in the adrenal gland, basal PKA signaling is required to maintain the vesicle pools that are primed and ready to be exocytosed (Nagy et al., 2004). In general, increase in intracellular Ca²⁺ is the main driver of exocytosis, but cAMP can also modulate the response at several different levels through mechanisms involving both PKA and EPAC.

Although Cushing syndrome is the most prominent diagnosis, primary macronodular adrenal hyperplasia is a related disorder that reflects a spectrum of disease ranging from subclinical hypercortisolism all the way to overt Cushing syndrome. Of note, it can also be part of the manifestations of McCune-Albright syndrome (see section 4. Fibrous Dysplasia and McCune-Albright Syndrome) (De Venanzi et al., 2014). It is characterized by large functional nodules on the adrenal gland that alter cortisol secretion. Although rare, primary macronodular adrenal hyperplasia can be caused by activating mutations in MC2R (encoding MC₂R) (Hiroi et al., 1998; Swords et al., 2004) or GNAS (Fragoso et al., 2003; Hsiao et al., 2009).

3. Cardiac Myxoma

Cardiac myxomas (CMs) can occur in the context of Carney complex, and this accounts for about 7% of all CM cases (Milunsky et al., 1998). The vast majority of the patients with Carney complex have loss-of-function mutations in PRKAR1A (70%) (Bertherat et al., 2009; Wang et al., 2018b). For these patients, CMs typically present earlier in life (with frequent reoccurrence) and can affect any chamber of the heart with multiple lesions. Conversely, isolated sporadic CMs typically occur as a single lesion in middle-aged women (mean age 51 years) and preferentially in the left atria (Carney, 1985; Reynen, 1995; Stratakis et al., 2001). Interestingly, it is estimated that anywhere from 31% (Maleszewski et al., 2014) to 64% (He et al., 2017) of isolated sporadic CMs are also caused by loss-of-function mutations in PRKAR1A. Although the vast majority of CMs are sporadic, there are also a few reports of familial CMs not associated with Carney complex. Typically, these familial mutations follow autosomal dominant inheritance. For instance, in one family, both the father (44 years of age) and daughter (20 years of age) developed CM as a result of the V164D frameshift deletion (c.491_492delTG) in PRKAR1A. The woman’s uncle and brother did not harbor the mutation and had no signs of CM to date (Ma et al., 2019). CMs are the most common primary tumor in the heart, and although they are benign, they can cause significant morbidity and mortality because of their location (Reynen, 1995). The mechanism of tumorigenesis for CM is not fully understood, but it is thought that mesenchymal stem cells (MSCs) from the endocardium and epicardium are the cell of origin (Di Vito et al., 2015). Effects on this MSC population may also account for GNAS mutations found in intramuscular and cellular myxomas (>90% GNAS mutants) (Sunitsch et al., 2018). Of note, MSCs are also the cell of origin for fibrous dysplasia, which is discussed in the next section.

4. Fibrous Dysplasia and McCune-Albright Syndrome

Fibrous dysplasia (FD) is a rare skeletal disorder that is characterized by painful and brittle bones that are prone to fracture and deformity. The clinical presentations can be very heterogeneous, affecting one bone (monostotic) or multiple bones (polyostotic) with variable severity. FD can also present with additional manifestation of café-au-lait spots or endocrine hyperfunction, which is termed McCune-Albright syndrome (MAS) (Feller et al., 2009; Riminucci et al., 2010). Additionally, if FD presents with intramuscular myxomas, tumors of musculoskeletal soft tissue, it is termed Mazabraud syndrome. FD/MAS is caused by postzygotic somatic activating mutations in GNAS (GNAS R201C/H) (Fig. 8C); thus, the disease is not inherited. The heterogeneity of FD/MAS results from somatic mosaicism, wherein some cells inherit the defect, whereas others do not. The tissues involved in FD/MAS arise from all three embryonic germ layers (ectoderm, endoderm, mesoderm), suggesting that in most cases the mutation may be acquired prior to gastrulation, before cell lineage decisions are made (Riminucci et al., 2006; Feller et al., 2009).

Recent studies by our groups and others have demonstrated that expression of GNAS activating mutations in mesenchymal/skeletal stem cells is necessary and sufficient to drive FD development in mouse models (Zhao et al., 2018). Interestingly, germline expression of the FD mutation is embryonic lethal (Khan et al., 2018), but when expression is induced during embryogenesis or postnatally, FD lesions develop rapidly (Zhao et al., 2018). The severity of the disease, however, is not linked to stage of development in which the mutation is acquired but, rather, the degree to which mutated cells contribute to critical functions within the tissues (Riminucci et al., 2006; Feller et al., 2009). For instance, patients with a higher ratio of mutated cells to normal cells in the osteogenic progenitor pool will develop more severe FD, whereas patients with a higher ratio of normal cells to mutant cells will display milder phenotypes. In fact, isolation of bone marrow stroma progenitors from patients with FD revealed that the stroma is a mosaic of mutant and normal cells. Mosaic stromal marrow engrafts into immunocompromised mice, whereas purified mutant marrow fails to engraft (Bianco et al., 1998). Therefore, it has been proposed that there is a “critical mass” of mutated cells that are necessary to drive symptomatic disease (Riminucci et al., 2006; Feller et al., 2009).

Under normal physiologic conditions, bone is constantly being remodeled, which is a balance between bone production by osteoblasts and bone resorption by osteoclasts. Overactivation of Gαs signaling through PKA induces proliferation of osteogenic precursors but impairs proper differentiation of osteoblasts and mineralization while enhancing osteoclast differentiation (Riminucci et al., 1997; Zhao et al., 2018) (Fig. 8C). Ultimately, this shifts the balance toward bone resorption, which is a histologic marker of FD in patients.

5. Acromegaly, Gigantism, and Pituitary Tumors

Acromegaly and gigantism are rare diseases characterized by overproduction of growth hormone (GH). GH is normally secreted by the pituitary gland into the bloodstream, where it travels to the liver to stimulate insulin-like growth factor-1 production and growth of bones and body tissues. Gigantism occurs early in childhood before growth plate fusion, resulting in dramatic vertical growth, whereas acromegaly occurs in adulthood and is characterized by growth and swelling of many body tissues, including hands, feet, nose, lips, jaw, and brow (Hannah-Shmouni et al., 2016). In most cases, acromegaly and gigantism are caused by somatotropinoma or GH-secreting pituitary tumors. The majority of GH-secreting pituitary tumors occur sporadically, but there are a few examples of familial cases. The most common sporadic alteration in acromegaly is GNAS activating mutations (40%–60%) (Freda et al., 2007; Hage et al., 2018). Typically, these patients have smaller tumors but very high GH secretion, highlighting again the physiologic role of the cAMP in secretion. Of note, no mutations have been identified in the PRKACA or PRKACB (Larkin et al., 2014), and GNAS mutations specifically enrich in GH-secreting pituitary tumors over other subtypes of pituitary tumors (Bi et al., 2017). In about 10% of gigantism, patients have very-early-onset disease (before the age of 4), known as X-linked acrogigantism (XLAG). In addition to overproduction of GH, patients with XLAG also overproduce the hormone prolactin. XLAG is caused by duplications in GPR101, an orphan GPCR on the X chromosome. XLAG predominates in females, but some males also acquire sporadic mutations (Iacovazzo and Korbonits, 2016; Gadelha et al., 2017). Additionally, there have been two independent families that display GPR101 duplications. GPR101 is predicted to couple to Gαs and has been show to stimulate cAMP production in vitro; however, there is some evidence it could couple to Gαi as well (Bates et al., 2006; Martin et al., 2015; Iacovazzo and Korbonits, 2016).

Acromegaly and gigantism are also associated with Carney complex and McCune-Albright syndrome, but in these cases, it is generally caused by hyperplasia of the somatotrophs, GH-secreting cells in the pituitary, instead of overt tumors. In Carney complex, most patients have PRKAR1A loss-of-function mutations, leading to PKA activation and GH and prolactin excess, but only about 10% of patients actually present with acromegaly. For McCune-Albright syndrome, a smaller percentage of patients have pituitary involvement, but of those, 36% develop gigantism, whereas the other 64% develop acromegaly (Boikos and Stratakis, 2007; Gadelha et al., 2017).

6. Hyperthyroidism

Hyperthyroidism is a disease in which the thyroid gland is overactive, producing too much of the hormones that control metabolism, triiodothyronine and tetraiodothyronine. This leads to increased appetite and unintentional weight loss, rapid and irregular heartbeat, restlessness, and potentially goiter (enlargement of the thyroid gland) (De Leo et al., 2016). Hyperthyroidism can have many causes, but as previously mentioned, it can be a component of Carney complex and McCune-Albright syndrome. Whether patients present as part of a broader syndrome, these nonautoimmune hyperthyroidisms can be caused by activating mutations in the thyroid-stimulating hormone receptor (TSHR, encoded by TSHR) or GNAS. As a GPCR, TSHR couples to Gαs to control secretion of triiodothyronine and tetraiodothyronine, but activating mutations in this pathway can cause thyroid adenomas that autonomously secrete hormones (Hébrant et al., 2011; Lacka and Maciejewski, 2015). Of these thyroid adenomas, 5%–10% are caused by GNAS mutations, and 70%–80% are caused by TSHR mutations (Palos-Paz et al., 2008; Nishihara et al., 2009). A recent report suggested that for hot thyroid nodules (nodules that preferential take up radioactive iodine, generally with excess thyroid-stimulating hormone secretion), GNAS and TSHR are the only driver mutations, with a clear preference for TSHR mutations (Stephenson et al., 2020). Over 30 different mutations in TSHR have been documented. Some mutations have been identified in adenomas as well as sporadic and familial cases, whereas others have preference for specific subsets (Hébrant et al., 2011). The reason for this preference is a balance between mutation expression and strength of activation. Strong clonal mutations are likely to cause adenomas and sporadic hyperthyroidism, whereas weaker germline mutations expressed in all cells are likely to cause familial cases. Although there is no defined syndrome, it is probable that particularly strong germline TSHR mutations are embryonic lethal since thyroid hormones are critical to fetal development (Lacka and Maciejewski, 2015).

7. Inactivating Parathyroid Hormone/Parathyroid Hormone–Related Peptide Signaling Disorder

Unlike the other diseases discussed so far, inactivating parathyroid hormone/parathyroid hormone–related peptide signaling disorder (iPPSD), represents a heterogeneous group of disorders that is characterized by inactivating defects in the Gαs-PKA signaling pathway. Clinical features of this disease are diverse and overlapping among subtypes. Common features include skeletal deformities (brachydactyly, short stature), obesity, cognitive impairment, and hormone insensitivity, leading to improper mineral metabolism and delayed reproductive development, among other manifestations (Mantovani and Elli, 2018, 2019). The current iPPSD nomenclature encompasses diseases such as Blomstrand chondrodysplasia/Eiken syndrome, pseudohypoparathyroidism, acrodysostosis, Albright hereditary osteodystrophy, and progressive osseous heteroplasia, but the specific distinctions are beyond the scope of this review (Mantovani and Elli, 2019). Here, we will focus on the molecular underpinnings of the iPPSD subtypes.

The clinical features of iPPSD highlight the physiologic roles of PTH signaling in a wide variety of developmental and homeostatic mechanisms. PTH is secreted from the parathyroid glands located in the neck to regulate calcium and phosphate homeostasis by signaling through the parathyroid hormone receptor (PTHR). PTHR is a Gαs-coupled GPCR that is expressed at particularly high levels in the bone and kidney. Not surprisingly, inactivating mutations in PTHR (PTH1R) cause iPPSD1 with predominately skeletal defects. Gαs itself is also subject to heterozygous loss-of-function mutations or, more commonly, genomic imprinting that reduces Gαs mRNA and protein levels by around 50% (iPPSD2/3) (Turan and Bastepe, 2015; Mantovani and Elli, 2019). Clinical phenotypes, particularly heterotopic ossification, are recapitulated in mice with Gnas knockout in mesenchymal progenitor cells (Regard et al., 2013). GNAS is also subject to tissue-specific maternal imprinting or loss of paternally imprinted methylation patterns in particular regions on the GNAS locus. Patients with loss of function in Gαs display variable resistance to hormones, including PTH, thyroid-stimulating hormone, gonadotropin, and GHRH, which determine their clinical manifestations (Mantovani and Elli, 2018, 2019). For instance, all patients of these subtypes display bone and adipose phenotypes due to biallelic expression of Gαs in these tissues, whereas individuals with maternally inherited loss of function will present with additional cognitive and endocrine phenotype due to paternal imprinting of Gαs in these tissues (Mantovani et al., 2004; Long et al., 2007; Mouallem et al., 2008; Turan and Bastepe, 2015). In line with the importance of the Gαs-PKA signaling pathway, mutations in RIα, PDE4D, and PDE3A characterize the remainder of the molecularly defined iPPSD subtypes (iPPSD4/5/6) (Mantovani and Elli, 2019). Of particular note, mutations in PDE3A further highlight the importance of cAMP in driving the pathophysiology of iPPSD. As mentioned previously, PDE3 family members can hydrolyze both cAMP and cGMP. Interestingly, mutations in PDE3A have been shown to enhance the cAMP-hydrolyzing activity without altering enzymatic activity toward cGMP, ultimately resulting in reduced cellular cAMP levels (Maass et al., 2015; Ercu et al., 2020).

1. GNAS–Protein Kinase A as Oncogenes: Beyond Endocrine Tumors

A real shock to the field came with the discovery of a PKA fusion protein that drives a rare form of liver cancer (<1% of cases), known as FL-HCC (Honeyman et al., 2014). Affecting children and young adults with no underlying pathology, FL-HCC could not be more different from the majority of liver cancers, which affect adults with liver damage commonly due to viral infection or alcoholism. As mentioned previously, patients with FL-HCC were found to express an in-frame fusion of DNAJB1 with PKA Cα (DNAJB1-PRKACA) that resulted in increased PKA activity due to relative overexpression of the catalytic subunit (Riggle et al., 2016), but importantly, overexpression of PRKACA does not completely recapitulate the oncogenicity of the fusion protein (Kastenhuber et al., 2017) (see section D. Fusion Proteins: An Emerging Mutational Theme) (Fig. 6A). To date, across multiple studies, DNAJB1-PRKACA has been identified in nearly 80% of patients with FL-HCC (Cornella et al., 2015). Of note, several patients with FL-HCC lacking the DNAJB1-PRKACA fusion protein, but with a history of Carney complex and other tumors, exhibited a complete loss of RIα protein instead (Graham et al., 2018). Recent studies have pointed to an even broader role of PKA fusion proteins, including additional fusions with PRKACB and ATP1B1, suggesting that they may also be driver oncogenes in extrahepatic cholangiocarcinoma, intraductal oncocytic papillary neoplasms, and intraductal papillary mucinous neoplasms (IPMNs) of the pancreas and bile duct (Nakamura et al., 2015; Singhi et al., 2020; Vyas et al., 2020) (Fig. 6A).

Although DNAJB1-PRKACA in FL-HCC clearly establishes PKA as an oncogenic driver, broader analysis of cancer genomes by our group revealed that GNAS is the most highly mutated G protein, harboring mutations in over 4% of all sequenced tumors to date, with the majority representing hotspot mutations (O'Hayre et al., 2013; Wu et al., 2019; Arang and Gutkind, 2020). Surprisingly, we and others have noted that among GNAS mutated cancers, there is a clear enrichment of gastrointestinal cancers, including colorectal adenocarcinoma (4%–10%), stomach adenocarcinoma (6%–10%), and pancreatic adenocarcinoma (5%–12%), a finding which extends to GPCRs and other G protein subunits (O'Hayre et al., 2013; Innamorati et al., 2018; Wu et al., 2019; Arang and Gutkind, 2020). GNAS and PKA also seem to be particularly important to neuroendocrine cancers of the pancreas, prostate, liver, and lung (Deeble et al., 2007; Boora et al., 2015; Kastenhuber et al., 2017; Innamorati et al., 2018; Coles et al., 2020). Expanding on these observations, we find that GNAS mutation frequency is even more significant in less-studied cancers, such as those of the bone (40%) and the peritoneum (53%) (Fig. 9A; Supplemental Table 5). Although GNAS mutation is recognized for its importance in cancer and is routinely included in clinical sequencing panels, such as FoundationOne (https://www.foundationmedicine.com/), analysis of the broader pathway reveals that mutations occur at every node. There are particularly good examples of each, such as ADCY2 mutations in liver (20%), PDE4D mutations in prostate (25%), and SPHK1-interactor and AKAP domain-containing protein (SPHKAP) in skin (26%) (Fig. 9A; Supplemental Table 5). Given that there are many genes representing each node of the pathway, when we consider the mutation frequency of each gene family, it becomes clear that some gene families are preferentially mutated in certain tissues; for instance, GNAS mutations predominate in hormone-sensitive tissues (Fig. 9B; Supplemental Table 5). Somewhat strikingly, we find that adenylyl cyclase mutations constitute the bulk of the mutations across many tissue types. Intriguingly, AKAPs are mainly mutated in the stomach and pancreas, whereas PKA catalytic subunits have a consistent low level of mutation across most tissues (Fig. 9B; Supplemental Table 5). Of important note, most patient samples harbor only one or two pathway mutations (57%), with the majority of those (41%) being single pathway mutations (Supplemental Table 5). As we discussed previously, there is limited knowledge on the functional importance of mutations within these other nodes of the pathway (see section E. Expanding the Mutational Themes), but given the emergence of genomic medicine and the success of targeted therapies, the role of the Gαs-PKA pathway in cancer certainly warrants further study. For the remainder of this review, we will highlight examples of the clinical and biologic function of the Gαs-PKA pathway in cancer.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9

Protein kinase A pathway mutations in cancer. (A) Frequency of specific pathway gene mutation across several tumor and cancer types. Heatmap is colored by mutation frequency (0% to >50%), with darker purple representing higher mutational frequency. All gene mutations from whole genome sequencing data sets are included (COSMIC database) (Tate et al., 2019). (B) Frequency of pathway mutation grouped by gene family across tumor and cancer types. A sample is considered to have a pathway mutation if it harbors at least one mutation in a family gene member.

2. Mucin Production Drives Clinical Phenotypes

One of the most striking and clinically relevant features of GNAS mutant cancers is their high level of mucin production across several tissue types (lung, stomach, bile duct, pancreas, appendix, colorectum, and gonads) (Innamorati et al., 2018). Mucins are large glycoproteins, either secreted or membrane-bound, with important physiologic and homeostatic roles. In the intestine, mucin provides the first line of defense against microbes and is critical to preserving epithelial barrier integrity. Mucins also have important structural roles to help physically maintain the microvilli architecture that is so important to intestinal function (Pelaseyed and Hansson, 2020). Consequently, the dysregulation of mucin can have profound impacts on disease. For instance, mucin 2 (encoded by Muc2) knockout mice have defects in goblet cell differentiation. This results in increased epithelial cell proliferation and migration coupled with decreased apoptosis and lack of acidic mucin production. Ultimately, these Muc2 knockout mice spontaneously develop tumors in the small and large intestine that progress to invasive carcinoma (Velcich et al., 2002). Interestingly, the Gαs-PKA pathway is known to directly regulate MUC2 expression through the G protein–coupled E-type prostanoid receptor 4 in the intestine. PKA-mediated activation of CREB triggers binding to the CRE in the MUC2 promoter and transcriptional upregulation (Nishikawa et al., 2013; Dilly et al., 2017). In pancreatic ductal cells, GNAS mutation is known to dramatically increase the expression of another mucin, MUC5AC. MUC5AC is one of the predominant mucins overexpressed in IPMNs of the pancreas, which commonly harbor GNAS hotspot mutations (discussed below) (Ideno et al., 2013; Komatsu et al., 2014). Transcriptional upregulation of mucin production is also augmented by the role of cAMP and PKA in vesicular transport. PKA is involved in constitutive transport of vesicles through the trans-Golgi network to the cell surface (Muñiz et al., 1996). Specifically, AKAPs anchor PKA to the cytoplasmic surface of the endoplasmic reticulum (AKAP1) and Golgi (AKAP1/9), where it can be activated in response to extracellular stimulation (Rios et al., 1992; Huang et al., 1999; Ma and Taylor, 2008; Mavillard et al., 2010).

At a molecular level, mucin overexpression in cancer has been implicated in dysregulation of cell polarity and disruption of proper cell-cell contacts. Further, mucin can facilitate aberrant oncogenic signaling, such as β-catenin activation, and receptor tyrosine kinase oligomerization and activation (Kaur et al., 2013; Pelaseyed and Hansson, 2020; Pothuraju et al., 2020). Mucin is also thought to play an important role in modulating the tumor microenvironment, serving as a bridge to nutrient-rich stroma through neoangiogenesis as well as by providing immunosuppressive mechanisms to evade immune surveillance. In addition to biologic effects on the tumor microenvironment, mucin can also serve as a physical barrier, sequestering local growth factors and protecting neoplastic cells from cytotoxic agents (Hollingsworth and Swanson, 2004; Kaur et al., 2013). Consequently, mucinous adenocarcinoma (in which >50% of the tumor mass is mucin) and tumors with a mucinous component (<50% of tumor mass is mucin) are implicated, with poor prognosis and chemoresistance across many tissue types (Schiavone et al., 2011; Lee et al., 2013; Kajiyama et al., 2014; Asare et al., 2016; Xie et al., 2018). Of note, pseudomyxoma peritonei (PMP) is one of the most devastating examples of mucin dictating clinical outcomes, for which the 5-year survival rate of high-grade disease is only 23% (Nummela et al., 2015). PMP is an extremely rare subtype of mucinous adenocarcinoma (typically originating from the appendix) in which the peritoneal cavity is colonized by mucin-secreting neoplastic cells. The excess mucin (>90% of tumor volume, dominated by MUC2 and a lesser extent MUC5AC) (O'Connell et al., 2002) overtakes the peritoneum, obstructing normal intestinal function and ultimately killing the patient. GNAS hotspot mutations are found in 63% of all PMPs, including both low- and high-grade disease (56% and 70%, respectively). Currently, the only therapeutic options for these patients are reductive surgery and intraperitoneal chemotherapy, which have significant treatment-associated morbidity. Thus, targeting the Gαs-PKA pathway as a means to limit mucin production has been proposed for patients with PMP (Nummela et al., 2015). Interestingly, in recurrent PMP, patients with GNAS mutations have poorer outcomes after chemotherapy, but it is uncertain whether this is because of the biology of GNAS mutants or whether GNAS is a biomarker of therapeutic resistance (discussed in section 5. Gαs–Protein Kinase A Induced Therapeutic Resistance in Cancer) (Pietrantonio et al., 2016).

When considering the prevalence of GNAS mutations in PMP, among other cancer subtypes, another trend that becomes rapidly apparent is a co-occurrence with mutations in the KRAS proto-oncogene (encoded by KRAS) Interestingly, 63%–72% of GNAS mutant PMPs also harbor KRAS mutations (Nummela et al., 2015; Ang et al., 2018). Furthermore, in mucinous neoplasms of the appendix, 69% of patients with GNAS mutations actually harbor GNAS and KRAS comutations. Nearly all of these patients had low-grade histology (Alakus et al., 2014). Another study corroborated this, finding that 50% of patients with low-grade appendiceal mucinous neoplasm were positive for both GNAS and KRAS mutations (Nishikawa et al., 2013). Interestingly, 38%–43% of patients with IPMNs of the pancreas, which are analogous low-grade lesions of the pancreas, harbor both GNAS and KRAS mutations (Molin et al., 2013; Amato et al., 2014). Furthermore, 58% of villous adenocarcinomas of the colorectum, which are characterized by noninvasive tissue architecture (similar to low-grade appendiceal mucinous neoplasm and IPMN) and profound mucin production are also GNAS and KRAS comutants (Yamada et al., 2012). Together these co-occurrence patterns highlight that GNAS and KRAS mutation give rise to unique biology in neoplastic diseases that cannot be achieved be either gene alone.

3. GNAS and Protein Kinase A Link Inflammation to Cancer Initiation

Consistent with clinical evidence that GNAS mutations are predominantly found in benign, noninvasive lesions, mouse models reveal that GNAS mutation alone is insufficient to induce epithelial tumorigenesis (Wilson et al., 2010; Patra et al., 2018). Our team showed that, in the context of KRAS mutations in the pancreas, GNAS drives lesions toward the cystic lineage; together, these comutants form well differentiated, mucinous cysts that resemble IPMNs, instead of noncystic pancreatic intraepithelial neoplasias. Somewhat counterintuitively, GNAS R201C expression does not accelerate KRAS-driven progression to pancreatic adenocarcinoma (PDAC). Instead, inactivation of tumor suppressors, like p53 (TP53), cyclin dependent kinase kinhibitor 2a (CDKN2A), or SMAD family member 4 (SMAD4), are needed to facilitate efficient progression to PDAC (Ideno et al., 2018; Patra et al., 2018). Interestingly, in the context of PDAC, GNAS R201C expression through activation of PKA actually attenuates aggressiveness and invasiveness due to epithelial differentiation (Pattabiraman et al., 2016; Ideno et al., 2018). This is supported by clinical evidence that patients with GNAS mutant have a better overall survival in appendix cancer (Ang et al., 2018). However, in small cell lung cancer, a neuroendocrine disease, GNAS and PKA activity is critical to cancer stem cell maintenance and increases rate of initiation and progression (Coles et al., 2020). This suggests that GNAS and PKA can play disparate roles within the various stages from neoplastic initiation to carcinogenic progression. Analysis of colorectal tissues on this spectrum from adenoma to carcinoma revealed that the frequency of GNAS mutation drops with progression. For instance, adenomas had the highest frequency of mutation, followed by carcinomas with residual benign adenoma, carcinomas with adenoma, and regions of invasion, and finally, no mutants were detected in pure carcinomas (Zauber et al., 2016). This suggests that in epithelial tissues, GNAS is most important in early initiation events. Indeed, several studies have highlighted that GNAS mutation can accelerate tumorigenesis (Wilson et al., 2010; Ideno et al., 2018; Patra et al., 2018; Coles et al., 2020). Given that tumors are heterogeneous, GNAS may confer a selective advantage initially, in which context additional mutational insults, like KRAS and subsequently TP53, can drive malignant growth ultimately independent of GNAS mutation. To this end, sequencing of normal human colon crypts unsurprisingly shows that KRAS and TP53 mutations are rare, suggesting that they are more important in intermediate and late events. However, reanalysis of available data highlights 55% of normal crypts harbored GNAS mutations (five of nine subjects), supporting the notion that GNAS may be important in neoplastic initiation and tumorigenesis (Lee-Six et al., 2019).

This idea of GNAS mutations participating in neoplastic initiation tracks well given the established Vogelgram of colorectal cancer (CRC) mutation accumulation. In the original model, KRAS mutations participated in intermediate events, facilitating the progression of adenomas, whereas TP53 loss served as the final barrier to carcinogenesis (Fearon and Vogelstein, 1990). As we have gained more understanding of the molecular events involved in carcinogenesis, COX-2–mediated inflammation has been defined as one of the earliest events in initiation (Markowitz and Bertagnolli, 2009). COX-2 is not expressed under normal conditions but is rapidly upregulated in response to stress and inflammatory stimuli. Naturally, COX-2 has become a prominent biomarker in colorectal cancer and many others, including lung (Hida et al., 1998), pancreas (Tucker et al., 1999), breast (Ristimäki et al., 2002), liver (Shiota et al., 1999), esophagus (Zimmermann et al., 1999), cervix (Ryu et al., 2000), and skin cancer (Buckman et al., 1998). COX -2 is the inducible form of the COX enzymes, which converts arachidonic acid to lipid signaling molecules, including prostaglandins and thromboxanes. These inflammatory mediators are ligands for a number of GPCRs in the prostanoid family (Hata and Breyer, 2004). Most notably, PGE₂ is the ligand for two Gαs-coupled GPCRs, E-type prostanoid receptors 2 and 4 (encoded by the PTGER2 and PTGER4 genes, respectively). PGE₂ has been shown to increase proliferation in colon cancer cells and mediate activation of β-catenin (through Gαs) and other mitogenic signaling molecules, like phosphoinositide 3-kinase (PI3K) and protein kinase B (Akt) (through Gβγ effects) (Castellone et al., 2005).

Frequent and early genomic alteration in the adenomatous polyposis coli (APC) gene are often concurrent with COX-2 overexpression in early initiation events of CRC (Fearon and Vogelstein, 1990; Markowitz and Bertagnolli, 2009), thus highlighting the interplay between their regulated pathways. APC acts a major tumor suppressor in CRC, inhibiting the Wnt–β-catenin signaling route (Kolligs et al., 2002). The Wnt pathway is a major determinant of cell fate decisions, helping to promote stem cell maintenance and tissue renewal from embryogenesis to adulthood. Consequently, these normal programs are frequently co-opted by disease. Wnt signaling controls β-catenin, a coactivator that drives transcription through binding of nuclear transcription factors (i.e., T-cell factor or TCF). When the pathway is inactive, β-catenin is sequestered in the cytoplasm by a protein complex termed the destruction complex and ultimately targeted for degradation (Fig. 10A). This destruction complex consists of key molecules like GSK3, casein kinase 1α (CK1α), Axin, and APC. CK1α and GSK3 provide the phosphorylation signals that target β-catenin for ubiquitination and degradation. Canonically, the pathway becomes activated by extracellular Wnts or changes in adherens junctions (Angers and Moon, 2009; Valenta et al., 2012). However, the Gαs-PKA pathway can modulate β-catenin activity at several levels (Fig. 10A). When activated by receptors, Gαs has been shown bind to Axin, leading to the stabilization and activation of β-catenin (Castellone et al., 2005). Many components of the destruction complex are also phosphorylated by PKA. The predominant mechanisms highlight the ability of PKA to phosphorylate and inhibit GSK3, releasing β-catenin to enter the nucleus (Fang et al., 2000). This, coupled with direct PKA phosphorylation of β-catenin to inhibit ubiquitination and degradation, helps drive β-catenin–mediated transcription (Hino et al., 2005). These mechanisms have important biologic consequences, including stem cell maintenance and tissue regeneration and repair (Goessling et al., 2009; Wang et al., 2016). Crosstalk with the Gαs-PKA pathway is also particularly important for the endocrine Gαs-PKA pathway signalopathies (Walczak and Hammer, 2015). For instance, β-catenin expression is very strong in adrenal tumors and Carney complex caused by genetic defects in the Gαs-PKA pathway (Almeida et al., 2012). This contributes to dysregulated Wnt signaling and loss of cell cycle control (Almeida et al., 2010). In CRC, activation of the Wnt–β-catenin pathway by Gαs-PKA may represent a key event in CRC initiation and progression, whether it is achieved by mutations in GNAS or perhaps more often by PGE₂ and COX-2–initiated, Gαs-linked GPCR signaling (Castellone et al., 2005; Wu et al., 2019).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 10

Aberrant protein kinase A pathway activity leads to dysregulation of signaling and transcriptional programs. (A) Wnt and PKA activity drive β-catenin–mediated gene transcription. Canonically, Wnt binds to Frizzled receptors and coreceptors like LPL receptor related protein 6 (LRP 6) on the surface of the cell to inhibit the activity of the destruction complex. Destruction complex members include APC, GSK3, CK1α, and Axin. Inhibition of this complex releases β-catenin to drive target gene transcription through the transcription factor TCF. Production of PGE₂ through COX-2 leads to activation of Gαs-coupled GPCRs, E-type prostanoid receptor 2 (EP2), and E-type prostanoid receptor 4 (EP4). Activation of Gαs leads to direct phosphorylation and inhibition of GSK3 as well as stabilizing phosphorylation of β-catenin. These effects coupled with the direct binding of Gαs to Axin lead to accumulation of β-catenin and activation of target gene transcription. (B) PKA inhibits Hippo pathway and YAP/TEAD-mediated transcription. The Hippo pathway is regulated by a kinase cascade whereby the upstream kinase MST phosphorylates and activates LATS kinase. Phosphorylation of YAP by LATS inactivates YAP through cytoplasmic sequestration and degradation. PKA-mediated phosphorylation of LATS, among other mechanisms, also inhibits YAP activity and consequently blocks target gene transcription through TEAD. (C) PKA regulates hedgehog (HH) signaling in the cilium to inhibit GLI transcriptional activity. When HH ligand is present, it binds to and inhibits the receptor Patched (PTCH1), allowing the Gαi-like GPCR SMO to traffic to the ciliary membrane. SMO inhibits cAMP production and PKA activity, allowing GLI-mediated transcription to proceed. When HH ligand is absent, PTCH1 constitutively inhibits SMO and allows the Gαs-coupled GPCR GPR161 to traffic to the ciliary membrane. When present at the membrane, GPR161 stimulates cAMP production and PKA activity. PKA in turn phosphorylates and inhibits GLI, eventually leading to its degradation.

Aligned with this perspective, PGE₂ dramatically increases intestinal tumor burden in CRC mouse models, and the inhibition of PGE₂ production with COX-2 inhibitors, such as by nonsteroidal anti-inflammatory drugs (NSAIDs), reduces tumor burden (Hansen-Petrik et al., 2002; Kawamori et al., 2003; Wang and DuBois, 2010). In humans, retrospective studies have revealed a reduced incidence of colorectal cancer with prolonged NSAID use, and NSAIDs can directly reduce polyp size and number in patients with familial CRC. Unfortunately, the clinical response to NSAIDs is incomplete, and long-term use can have limiting toxicities (Giardiello et al., 1993; Brown and DuBois, 2005). Of available NSAIDs, aspirin has been used successfully long-term in cardiovascular disease. In these patient populations, aspirin has also been shown to reduce CRC incidence and mortality. Interestingly, the benefit of aspirin in chemoprevention was most pronounced after 10 years (Chan et al., 2005; Chubak et al., 2015; Drew et al., 2016). One mechanism by which aspirin is thought to reduce mortality is by preventing metastasis, particularly in the progression of local adenoma to metastatic disease (Rothwell et al., 2012). Given the consistent efficacy of aspirin and other NSAIDs in chemoprevention, numerous clinical trials have tested their efficacy in other settings. Notably, NSAIDs have shown efficacy in some adjuvant settings but failed when operating as single-agent chemotherapeutics (Brown and DuBois, 2005; Wang and DuBois, 2010). The clinical efficacy of NSAIDs as chemopreventive agents, but failure as chemotherapeutics, highlights the true complexity of prostaglandin signaling. It is likely that PGE₂ and others participate in autocrine and paracrine signaling loops that involve both tumor, stroma, and immune components. To this end, Gαs-PKA activation, downstream of the proton-sensing GPCR GPR68, has been shown to drive the secretion of interleukin 6 (IL- 6) from cancer-associated fibroblasts and subsequent proliferation of PDAC in trans (Wiley et al., 2018). Further, Gαs-linked GPCRs, like prostanoid (Zelenay et al., 2015; Böttcher et al., 2018; Pelly et al., 2021) and adenosine receptors (Visser et al., 2000; Novitskiy et al., 2008; Young et al., 2014; Young et al., 2018), contribute to tumor immune evasion and drive immune suppression by dampening T-cell responses, as well as interfering with immune cell migration and maturation. For example, these mechanisms can include direct PKA-mediated phosphorylation of C-terminal Src kinase (CSK) and other components involved in T-cell receptor signaling and activation, as well as PGE₂-mediated suppression of chemokine production and dendritic cell recruitment (Wehbi and Taskén, 2016; Böttcher et al., 2018). Recent evidence also points to the specific role of PKA Cβ2 (an immune-specific spliceform) in regulating immune responses in inflammatory disease (Moen et al., 2017). Together, this highlights that the Gαs-PKA pathway can participate in tumor initiation and progression through autocrine and paracrine (oncocrine) mechanisms (Wu et al., 2019). Even in the absence of overt mutations, these oncocrine signals can have important effects throughout the tumor microenvironment, including contributions to a cancer immune evasion and therapeutic resistance (see section 5. Gαs–Protein Kinase A Induced Therapeutic Resistance in Cancer).

4. GNAS–Protein Kinase A as Tumor Suppressors

Thus far, our discussions of the Gαs-PKA pathway in cancer have focused on the role of GNAS and DNAJB1-PRKACA as oncogenes. Paradoxically, however, there are several examples in which the Gαs-PKA pathway functions as a tumor suppressor. A study by our group unexpectedly found that genetic ablation of Gnas or inhibition of PKA in the epidermis was sufficient to drive basal cell carcinoma, with dramatic expansion of the stem cell compartment residing in the hair follicle. Conversely, overactivation of the pathway with the GNAS R201C mutation drove the same stem cell population to terminal differentiation and exhaustion. Mechanistically, stem cell expansion in the hair follicle is controlled by PKA-mediated repression of yes-associated protein (YAP) and glioma-associated oncogene (GLI) transcriptional activity, with no effect on other stem cell programs like Wnt (Iglesias-Bartolome et al., 2015). Of note, PKA has been shown to repress YAP activity in pancreatic cancer (in which PKA functions as an oncogene) but still induce a differentiation phenotype (Ideno et al., 2018). Much like GNAS and PKA, YAP has also been shown to behave as either an oncogene or a tumor suppressor depending on the cellular context.

The Hippo pathway controls growth, differentiation, and cell death, balancing these processes to ensure proper organ development and size. In mammals, YAP and tafazzin (TAZ) are the main effectors that regulate transcriptional output through binding to transcription factors like TEA domain transcription factor (TEAD) in the nucleus. YAP/TAZ are regulated by phosphorylation from upstream kinases large tumor suppressor kinase 1 and 2 (LATS 1/2), whereby phosphorylation induces YAP/TAZ cytoplasmic sequestration and subsequent degradation (Fig. 10B). LATS1/2 in turn can be regulated by many upstream signals, including GPCRs. Gαs-coupled GPCRs activate LATS1/2 to repress YAP/TAZ (Yu et al., 2012). PKA directly phosphorylates LATS1/2 to enhance its kinase activity, and mutation of the PKA phosphosites abrogates PKA regulation of LATS1/2 while other regulatory mechanisms remain intact (Kim et al., 2013) (Fig. 10B). Physiologically, this is important because PKA is known to induce adipogenesis and neurogenesis through suppression of YAP (Kim et al., 2013; Yu et al., 2013). In general, YAP phosphorylation and inactivation is critical for cell cycle exit and terminal differentiation, and it is thought that PKA contributes to this regulation (Lee et al., 2008; Kim et al., 2013). This can explain in part why many neoplasms and cancers characterized by Gαs-PKA pathway activation are of well differentiated histology and typically less proliferative or low-grade (as discussed previously, see section 3. GNAS and Protein Kinase A Link Inflammation to Cancer Initiation).

In line with the additional effects of Gαs on GLI in basal cell carcinoma, low GNAS expression is also a feature of the sonic hedgehog (SHH) subtype of medulloblastoma (SHH-MB). Medulloblastoma is the most common pediatric brain cancer, with SHH-MB representing 30% of patients (Kijima and Kanemura, 2016). Within this subtype, activation of the SHH pathway (through multiple mechanisms) is thought to drive tumor initiation. Interestingly, patients with SHH-MB with low GNAS expression have significantly worse prognosis compared with patients with high GNAS expression (50% 5-month survival versus 100% 5-month survival). Similar to the hair follicle model, knockout of Gnas in neural progenitor cells induced expansion of this stem cell population in neonatal mice that progressively developed into a tumor resembling medulloblastoma by adulthood. The tumors were marked by upregulation of GLI and SHH signaling with no effect on the Wnt pathway, a pattern that matches the signature of patients with SHH-MB (He et al., 2014). Around 6% of patients with SHH-MB actually have GNAS mutations, including frameshift and nonsense inactivating mutations (He et al., 2014; Huh et al., 2014; Kool et al., 2014). Perhaps more surprisingly through, around 80% of SHH-MBs overexpress C-X-C motif chemokine receptor 4 (CXC R4), which is a Gαi-coupled GPCR (Sengupta et al., 2012). These patients are typically younger (∼50% were infants) with desmoplastic histology (He et al., 2014). Although CXCR4 is not often mutated, CXCR4 and its ligand, C-X-C motif chemokine ligand 12 (CXCL 12), are markers of poor prognosis and earlier onset in other brain tumors, like gliomas (Calatozzolo et al., 2006; Bian et al., 2007). For these patients, cAMP elevating agents, such as PDE inhibitors, have been proposed as potential therapeutic options (Rao et al., 2016).

The importance of Gαs in the SHH-MB subtype of pediatric brain cancer reflects the fundamental importance of Gαs-PKA in brain development. As a testament to its importance, Gnas homozygous knockout mice are embryonic lethal (Yu et al., 1998). Similarly, only 27% of Prkaca homozygous knockout mice survive past weaning (Skålhegg et al., 2002). As mentioned previously, both Cα1 and Cβ1 are ubiquitously expressed and capable of some degree of compensation. Therefore, it is not surprising that Cα and Cβ1 double knockout mice are embryonic lethal. Restoration of one allele in either gene (Cα or Cβ1) confers survival, but mice die from severe neural tube defects. Histologically, these mice have an expansion of cell types that are dependent on hedgehog (HH) signaling (Huang et al., 2002).

In a more pathway-specific fashion, PKA is known to regulate HH signaling, both SHH and Indian hedgehog, within the context of cilia. Interestingly, the ciliary structure is essential to proper signaling and development controlled by the HH pathway, a feature that is not shared by other developmental programs. The GLI family of transcription factors are the main effectors that respond to upstream stimulus from HH ligands. In the absence of pathway stimulation, GLI is sequestered and eventually degraded (Carballo et al., 2018). AKAPs position PKA at the base of the cilium, where the catalytic subunit phosphorylates GLI to facilitate GLI’s proteolytic processing and degradation, ultimately preventing transcriptional activation (Fig. 10C). Recent work has demonstrated that the Gαs-coupled GPCR, GPR161, contains an AKAP domain enabling it to directly recruit PKA to cilia (Bachmann et al., 2016). Of note, GPR161 is regulated by trafficking and only capable of signaling when it is present on the ciliary membrane (Bangs and Anderson, 2017). Activation of GPR161, among other Gαs-coupled GPCRs, is important to trigger production of cAMP and subsequent PKA activation. Generally, PKA activity is quite high when HH ligand is absent (Tschaikner et al., 2020). However, when HH is present, the Gαi-like-coupled GPCR Smoothened (SMO) traffics to the cilium to trigger a reduction in cAMP levels and inhibition of PKA activity (Ogden et al., 2008), allowing full-length GLI to activate transcription. This trafficking is regulated by the binding of HH to its receptor patched homolog 1 (PTCH1) at the membrane, thereby relieving the inhibition on SMO (Bangs and Anderson, 2017). Recent evidence has also demonstrated that SMO can directly inhibit PKA through binding to the free catalytic subunits at the membrane (Arveseth et al., 2021) (Fig. 10C). Numerous other GPCRs, such as CXCR4 (Gαi-coupled) and PAC1 (Gαs-coupled), can also contribute to the modulation of ciliary cAMP levels and PKA activity, although some of these roles are complex and cell type–dependent (Niewiadomski et al., 2013; Mukhopadhyay and Rohatgi, 2014; Schou et al., 2015; Mykytyn and Askwith, 2017; Amarante et al., 2018; Tschaikner et al., 2020). Ultimately, the degree of GLI transcriptional output is dependent on the level of PKA activity as a balance of these various inputs (Tschaikner et al., 2020). Consequently, the overexpression of PKA Cα is sufficient to inhibit SHH-stimulated proliferation and induce differentiation (Barzi et al., 2010). Recently, several mutations in Cα and Cβ, which display increased sensitivity to cAMP, show reduced HH pathway activation (Palencia-Campos et al., 2020). Conversely, deletion of Gαs in the mouse augments SHH signaling with developmental defects that mirror PKA deletion, or deletion of other negative regulators of the SHH pathway (Regard et al., 2013). SHH signaling is particularly important in guiding development of the nervous system and limb patterns, whereas Indian hedgehog is important in skeletal development (Bangs and Anderson, 2017). This explains why patients with loss-of-function mutation in the Gαs-PKA pathway can develop SHH-MB or severe skeletal deformities as part of iPPSD (discussed previously, see section 7. Inactivating Parathyroid Hormone/Parathyroid Hormone–Related Peptide Signaling Disorder). Furthermore, recent reports have described mutations in PRKACA that cause skeletal ciliopathies (Palencia-Campos et al., 2020; Hammarsjö et al., 2021).

5. Gαs–Protein Kinase A Induced Therapeutic Resistance in Cancer

Our discussions have already highlighted some features of the Gαs-PKA pathway that contribute to therapeutic resistance, including supporting an immune suppressive tumor microenvironment, and clinical evidence of poor outcomes and chemoresistance due to mucinous disease. Here, we will focus on additional evidence of the therapeutic resistance potential of the Gαs-PKA pathway in cancer.

Building on the evidence of GNAS and KRAS functioning as codrivers of carcinogenesis, several unbiased studies have identified Gαs and PKA as key drivers of resistance to MAPK pathway inhibition. In metastatic melanoma, about half of all patients have BRAF mutations and are primarily treated with BRAF inhibitors. Although most patients have clinical responses, approximately 20% of patients with BRAF mutation have intrinsic resistance to BRAF inhibitors (Sanchez et al., 2018). Unfortunately, many initial responders later develop acquired resistance from genetic (60%) or epigenetic and transcriptomic (40%) changes, primarily through reactivation of MAPK signaling outputs (Kakadia et al., 2018). Several studies have aimed at understanding these mechanisms of resistance and reactivation. Gain-of-function open reading frame and CRISPR activation screens in BRAF V600E melanomas have been used to identify programs that confer resistance to multiple BRAF and MAPK inhibitors. Surprisingly, GPCRs were consistently among the top hits, many of them being Gαs-coupled (Johannessen et al., 2013; Konermann et al., 2015). Downstream, ADCY9 and PKA Cα also confer resistance to MAPK inhibitors, with PKA Cα having a higher score than even RAF1. Further analysis revealed that PKA via CREB was able to activate transcriptional programs that MAPK normally activates (Johannessen et al., 2013). In melanocytes, there is a fine balance between MAPK control of proliferation and cAMP control of differentiation (Dumaz et al., 2006). This balance is achieved in part because PKA can phosphorylate and inhibit RAF1, while BRAF continues signaling downstream to ERK (Cook and McCormick, 1993; Dhillon et al., 2002). Interestingly , when RAS is mutated, RAF1 predominantly signals to ERK, a program that BRAF controls when it is mutated. This type of compensatory crosstalk is the basis for PKA-mediated resistance to MAPK pathway inhibition. Of note, this crosstalk is not present in all cell types, e.g., fibroblasts (Dumaz et al., 2006).

As we discussed previously, inflammatory signaling through COX2-PGE₂-Gαs contributes to the pathogenesis of many cancers. Recently, this pathway has also been implicated as a mechanism of resistance to combination BRAF and MAPK pathway inhibition in BRAF V600E colorectal cancer. Using a high-throughput kinase activity screen, the SRC proto-oncogene(SRC) was identified as having increased activity after inhibitor treatment. SRC in particular was shown to initiate a proinflammatory autocrine loop mediated by PGE₂ and Gαs that was sensitive to COX2 inhibition. Dramatically, the addition of a COX2 inhibitor to two or three drug combinations targeting the MAPK pathway led to greater rates of tumor regression in patient-derived xenograft resistance models (Ruiz-Saenz et al., submitted manuscript). The mechanisms of resistance through COX2-PGE₂-Gαs and PKA include survival of cancer stems cells as well as immune suppression (Tong et al., 2018). In BRAF V600E mutant melanoma, for instance, COX-2 was shown to drive tumor immune escape, a response that underlines the preclinical synergy of COX-2 inhibitors in combination with immune checkpoint blockade (Zelenay et al., 2015). Similarly, the ability of PKA to drive tumor immune evasion has also limited the efficacy of other immune-based therapies such as chimeric antigen receptor T cells (CAR-Ts) (Newick et al., 2016). This suppression of CAR-Ts and T cells in general is mediated by PKA/AKAP associations that negatively regulate T-cell function (Ruppelt et al., 2007). Interestingly, disruption of this PKA/AKAP interaction can improve CAR-T efficacy and enhance tumor killing (Newick et al., 2016). Building on the understanding of these immune suppressive mechanisms (see section 3. GNAS and Protein Kinase A Link Inflammation to Cancer Initiation), NSAIDs as well as prostanoid and adenosine receptor antagonists are being investigated as agents to combat tumor immune evasion and enhance the clinical efficacy of immune therapies (Leone et al., 2015; Hamada et al., 2017; Take et al., 2020). Finally, the Gαs-PKA pathway has effects on migration and metastasis. This role is somewhat controversial, as PKA has been shown to drive epithelial differentiation, instead of the epithelial-to-mesenchymal transition phenotypes generally recognized as metastatic (Pattabiraman et al., 2016). However, PKA is also known to play a role in cytoskeletal changes through direct AKAP interactions that are required for many of the hallmarks of cell migration (Howe, 2004). Importantly, it seems that these effects are context-dependent, since Gαs and PKA serve as a central regulatory hub integrating many signaling pathways and biologic functions.

PKA can also contribute to therapeutic resistance by co-opting other normal mechanisms, including energy adaptation. The mitochondria are the main producers of energy in the cell; thus, maintaining mitochondrial homeostasis is critically important to cell health. Mitochondrial homeostasis represents a dynamic balance between fusion (joining) and fission (division) events. PKA is particularly well studied in its ability to inhibit mitochondrial fission through phosphorylation of dynamin-related protein 1 (DRP 1), a dynamin-like GTPase. DRP1 functions to bring mitochondrial membranes close to each other to facilitate fission events. PKA phosphorylation at serine 637 inhibits DRP1 GTPase activity and recruitment to the mitochondria (Chang and Blackstone, 2007). By inhibiting fission, fusion is allowed to proceed, resulting in elongated mitochondria and increased respiration. Increased cAMP and PKA activity has also been linked to decreased mitophagy and ultimately control of mitochondrial recycling; however, it remains unclear whether this is primarily due to increased fusion or additional effects of cAMP and PKA. Together, the actions of the cAMP-PKA pathway on the mitochondria provide a prosurvival signal (Di Benedetto et al., 2018; Ould Amer and Hebert-Chatelain, 2018). Under physiologic conditions of low nutrients, cells elongate mitochondria to compensate. Interestingly, this physiologic adaptation can be exploited by cancer cells, which, although somewhat counterintuitive, rely heavily on glycolysis for energy (Vander Heiden et al., 2009). For instance, KRAS transformed cells die in low-glucose conditions, but activation of cAMP/PKA rescues their survival under these conditions. PKA-mediated activation of mitochondrial respiration ramps up oxidative phosphorylation and ATP levels (Acin-Perez et al., 2009; Palorini et al., 2013; Ould Amer and Hebert-Chatelain, 2018). Coupled with reduction in reactive oxidative species and increased autophagy, cAMP and PKA metabolically rewire cells to promote survival (Palorini et al., 2013; Palorini et al., 2016; Ould Amer and Hebert-Chatelain, 2018). Under physiologic conditions of low nutrients, PKA also liberates energy from glycogen and lipid stores through direct phosphorylation, as well as transcriptional regulation, of the enzymes involved in these processes (Rogne and Taskén, 2014; Yang and Yang, 2016) (see section F. Metabolic Regulation). However, it remains unclear to what extent cancer cells exploit these energy sources. Together, energy adaptation mechanisms and prosurvival signals provide some insight as to why GNAS and PKA serve as biomarkers of therapeutic resistance in many cancer types and further why GNAS and KRAS often comutate in cancer.

Finally, the role of Gαs and PKA in resistance can be seen clinically in breast cancer, a tissue type in which GNAS mutations are rarely found. One study profiled circulating free DNA before and after treatment with targeted therapy in metastatic, human epidermal growth factor receptor 2–positive (HER2+) breast cancer. Surprisingly, they found that GNAS mutations were only present in patients that were resistant to targeted therapy (Ye et al., 2017). Similarly, PRKACA transcripts were elevated in HER2+ patients that were resistant to trastuzumab (HER2 inhibitor) (Moody et al., 2015). In vitro models of resistance have also demonstrated that knockdown of PRKAR2A, to activate PKA, confers partial resistance to trastuzimab (Gu et al., 2009). Unlike in melanoma, this resistance could not be explained by MAPK pathway reactivation but, rather, by restoration of antiapoptotic signaling (Moody et al., 2015). In another subtype of breast cancer, estrogen receptor–expressing, patients receive antiestrogen therapies such as tamoxifen. Tamoxifen binds to estrogen receptor α to induce a conformation that prevents its activation and signaling. Interestingly, PKA has been found to phosphorylate the estrogen receptor α, an interaction coordinated by AKAP13. This phosphorylation prevents the inhibitory conformational change induced by tamoxifen and renders tamoxifen ineffective (Michalides et al., 2004; Bentin Toaldo et al., 2015). GNAS amplifications have been identified in 20% of HER2+ breast cancers and 13% of hormone receptor–positive breast cancers (Kan et al., 2010). Although further studies are required, it is tempting to suggest that GNAS amplification may serve as a biomarker, predicting resistance to therapy in breast cancer. Here, we have highlighted several known mechanism of therapeutic resistance, but there are certainly additional mechanisms yet to be described. Together, these findings highlight again the diversity and complexity of Gαs and PKA signaling and their roles in the diversity of the Gαs-PKA pathway signalopathies.

1. ATP Analog Inhibitors of the Catalytic Subunit

The most commonly used small-molecule inhibitors are the high-affinity, ATP-competitive isoquinolinesulfonyl protein kinase inhibitors, such as H89, H7, and H8 (Hidaka et al., 1984; Chijiwa et al., 1990; Engh et al., 1996; Lochner and Moolman, 2006); natural product derivative KT-5720 (Kase et al., 1987); or staurosporine (Meggio et al., 1995). Although these are very effective inhibitors, they have low specificity and inhibit several other kinases in the AGC family of protein kinases and hence should not be considered specific inhibitors (Lochner and Moolman, 2006; Murray, 2008). Of course, these inhibitors also do not discriminate between the PKA isoforms, thus limiting their clinical translatability.

2. Peptide Inhibitors of the Catalytic Subunit

To overcome the concerns of specificity, derivatives of the substrate-competitive, heat-stable PKI (encoded by PKIA, PKIB, and PKIG) can be used. PKI(5-24) has low-nanomolar inhibition constants and is absolutely specific for PKA (Cheng et al., 1985). PKI(5-24) can be modified by myristylation, which allows for membrane permeation (Eichholtz et al., 1993); however, it can also be expressed recombinantly in cells to overcome delivery issues. A hydrocarbon-stapled version of a PKI-derived sequence provides another excellent tool as a membrane-permeable, highly selective inhibitor of the catalytic subunits acting with low-subnanomolar affinity (Manschwetus et al., 2019).

3. Bisubstrate Inhibitors of the Catalytic Subunit

A combination of the two cosubstrate inhibitors, ATP and peptide, would be the logical consequence, and indeed, such bisubstrate analog inhibitors termed ARC-type inhibitors have been developed by linking an adenosine analog (either an adenosine derivative or ATP inhibitor) and an arginine-rich peptide (Lavogina et al., 2010). A series of ARC-type inhibitors have been designed with low-nanomolar or even picomolar affinities and efficacy against PKA Cα and Cβ (Ricouart et al., 1991; Enkvist et al., 2006; Enkvist et al., 2007; Lavogina et al., 2010; Nonga et al., 2020). Recent work has demonstrated that ARC inhibitors can also be engineered to have greater selectivity for mutant Cβ over wild-type Cβ (Nonga et al., 2020). Although ARC inhibitors have primarily been used as tool compounds, including fluorescently conjugated ARCs, recent advances have drastically improved their pharmacokinetic properties, making them poised for future application in a therapeutic context (Lavogina et al., 2010).

4. Targeting the Regulatory Subunits with cAMP Analogs

In contrast to the ATP analog inhibitors that target the catalytic subunit, cAMP analogs have been engineered with specificity for the two classes of regulatory subunits (RI and RII). Both activators and inhibitors have been developed (Christensen et al., 2003). Achieving PKA regulatory subunit specificity has been a special challenge, as other proteins such as cGMP-dependent protein kinases, EPACs, CNG channels, PDEs, and cyclases all have cyclic nucleotide binding domains (Berman et al., 2005; Holz et al., 2008) (see section G. Other cAMP Effectors). By modifying the oxygens of the cyclic phosphate, chemists generated cAMP agonists (Sp analogs) and antagonists (Rp analogs). Global inhibition can be achieved with the Rp analogs, which bind to but do not promote dissociation of the holoenzyme (Rothermel and Parker Botelho, 1988; Christensen et al., 2003). By comparing the activity of type I inhibitors, like Rp-8-Br-cAMPS, with the activity of nonselective inhibitors, like Rp-cAMPS, it is possible to discriminate between the activities of the two holoenzymes (Gjertsen et al., 1995; Christensen et al., 2003; Farquhar et al., 2008). Similarly, the combination of different agonists can achieve some level of isoform-specific activation, but this still remains a challenge in the field (Robinson-Steiner and Corbin, 1983). However, leveraging regulatory subunit agonists and antagonists has facilitated the high-quality purification of PKA holoenzymes as well as free regulatory subunits (Bertinetti et al., 2009; Hanke et al., 2011). Unfortunately, many of these cAMP analogs suffer from poor membrane permeability, limiting their efficacy if delivered extracellularly. To overcome this, membrane-permeable versions of the cAMP analogs have been developed as prodrugs. When cleaved by cytosolic esterases, the analog is free to act inside the cell (Chepurny et al., 2013; Schwede et al., 2015). Care must be taken, however, because the effective concentration of the released nucleotide inside the cell may vary, and extremely high levels of cAMP may perturb other cyclic nucleotide signaling.

5. Inhibitors of A-Kinase Anchoring Protein Binding

PKA specificity is also highly dependent on targeting to specific sites in the cell. Targeting is typically mediated by binding to AKAPs that contain a high-affinity helical binding motif that interacts with the D/D domains of the regulatory subunits. Naturally, peptides have been developed to disrupt this interaction, nonselectively perturbing the interactions with both type I and type II interactions (Carr et al., 1992; Herberg et al., 2000). Over time, this led to development of peptides specific to type I or type II, although these peptides still suffered from limited cell permeability (Calejo and Taskén, 2015). Now, isoform-specific, cell-permeant stapled peptides have been engineered that can selectively disrupt the targeting of type I and type II holoenzymes (Wang et al., 2014, 2015; Kennedy and Scott, 2015; Bendzunas et al., 2018). Unfortunately, these peptides still lack clinical utility because of their unfavorable pharmacokinetics and relative inability to distinguish among specific AKAP interactions (Calejo and Taskén, 2015). Reagents have also been developed to disrupt other AKAP binders, such as PDEs and phosphatases, but as AKAPs have multiple binding partners, it has been difficult to translate this disruption to direct modulation of cellular consequences (Bucko and Scott, 2020; Omar and Scott, 2020). To begin to answer these difficult questions of microdomain dynamics, a promising new tool has been developed using AKAP targeting sequences as a means to localize drug delivery to specific PKA microdomains, such as those present at the centrosome. Although this approach, called local kinase inhibition, is still in its infancy, conceptually it holds a lot of promise in understanding AKAP interactions more directly and ultimately enhancing the specificity of PKA modulation (Bucko et al., 2019). Finally, small-molecule AKAP disrupters represent another promising approach with potential for clinical translation. Protein-protein interactions have been notoriously difficult to target with small molecules, but the advances in high-throughput screening have made this approach more feasible (Calejo and Taskén, 2015). Several groups have applied these approaches recently to identify disrupters of AKAP interactions (Gold et al., 2013; Schächterle et al., 2015). Although there are real challenges, huge potential lies in the ability to apply these small-molecule disrupters to specific AKAP complexes in diseases settings; for instance, disruption of the PKA/AKAP interactions that mediate immune suppression in T cells in cancer (see section 5. Gαs–Protein Kinase A Induced Therapeutic Resistance in Cancer).

6. Emerging Approaches

As we discussed, many of the Gαs-PKA pathway signalopathies are driven by specific hotspot point mutations, like GNAS R201C or PRKACA L206R, so generating mutation-specific drugs could be a viable therapeutic option. Recently, this strategy has shown clinical promise, most notably by targeting the mutant cysteine of KRAS G12C with drug electrophiles (Ostrem et al., 2013). It has been proposed that this same method could also be applied to target GNAS R201C mutants in cancer (Visscher et al., 2016). Although PKA is not amenable to targeting with drug electrophiles, the PKA Cα L206R mutation does have reduced affinity for its endogenous inhibitor PKI compared with wild-type PKA Cα, whereas the small-molecule inhibitor H89 still retains its efficacy. This opens the possibility of exploiting this differential binding to selectively target PKA Cα mutants. However, significant challenges remain, as H89 retains its efficacy because it is an ATP-competitive inhibitor. As alluded to previously, this class of drug is susceptible to multiple off-target effects on other kinases, making it a liability in the clinical setting (Luzi et al., 2018). In an effort to identify drugs that do not act as ATP-competitive inhibitors of PKA, high-throughput screening platforms based on fluorescence polarization have been developed and proven capable of identifying allosteric agonists and antagonists (Saldanha et al., 2006; Brown et al., 2013). Some promise has also been shown for antisense oligonucleotides targeting RIα in combination with chemotherapy in cancer (Goel et al., 2006; Almeida et al., 2012). The mechanism is not completely understood, but the compensatory increase in RIIβ protein could be important in restoring the balance of type I and type II holoenzyme signaling (Nesterova et al., 2000). Similarly, although many of the PKA Cα mutations have been linked to altered substrate profiles and decreased preference toward canonical substrates. It is plausible that restoring activity toward key substrates may also serve as an additional therapeutic avenue (Lubner et al., 2017).