|Home | About | Journals | Submit | Contact Us | Français|
PRKAR1A encodes the regulatory subunit type 1-alpha (RIα), of the main mediator of the cAMP effects in the eukaryotic cells – cAMP dependant Protein Kinase A (PKA). Inactivating PRKAR1A mutations are known to be responsible for the multiple neoplasia and lentiginosis syndrome Carney complex (CNC). To date, at least 117 pathogenic variants in PRKAR1A have been identified. The majority of them are subject to non-sense mediated mRNA decay (NMD), leading to PRKAR1A haploinsufficiency and, as a result, activated cAMP signaling in the affected tissues. In recent years it became apparent that CNC may be caused not only by RIα haploinsufficiency, but also by the expression of altered PRKAR1A protein, as proven by analysis of the relatively small proportion of expressed mutations in the gene, consisting of aminoacid substitutions and in-frame genetic alterations. In addition, a new subgroup of mutations that potentially escape NMD and result in CNC through altered rather than missing protein has been analyzed – frame-shifts in the 3′end of the coding sequence that shift the stop codon downstream of the regular one and result in the expression of longer than the wild type protein. The PRKAR1A mutation detection rate in CNC patients is recently estimated at above 60%; PRKAR1A mutation negative CNC patients are characterized by significant phenotypic heterogeneity. In this paper, we present a comprehensive analysis of all 117 known to date pathogenic PRKAR1A sequence variations, and discuss their molecular context and clinical relevance.
Cyclic AMP (cAMP) dependent Protein Kinase A (PKA, EC 184.108.40.206) is the major mediator of the cAMP effects in the eukaryotic cells. In its inactive state, the PKA holoenzyme is a tetramer comprised of dimer of two regulatory subunits bound to two catalytic subunits (Kinderman, et al., 2006; Tasken, et al., 1997). Upon elevated cellular cAMP concentrations, two molecules of cAMP bind each of the regulatory subunits, which leads to the dissociation of the tetramer into regulatory dimers and two free catalytic subunits (Tasken, et al., 1997). The free catalytic subunits represent active serine threonine kinases which in turn phosphorylate series of cellular targets that regulate downstream effector enzymes, ion channels, and transcription of specific genes that mediate cell growth and differentiation (Shabb, 2001).
In the human, 4 different regulatory (PRKAR1A, PRKAR1B, PRKAR2A, and PRKAR2B) and 3 catalytic subunits (PRKACA; PRKACB, PRKACG) have been identified so far. It is now assumed that the multiple combinations of homo- or heterodimers of regulatory subunits with associated catalytic subunits may, at least in part, account for the specificity of the transduction of the signal generated by numerous different hormones, neurotransmitters, and other signaling through one sole second messenger, cAMP.
The only PKA subunit in which mutations have been found to lead to human disease is PRKAR1A: inactivating PRKAR1A mutations cause the multiple neoplasia and lentiginosis syndrome Carney complex (CNC) (Casey, et al., 1998; Kirschner, et al., 2000a). CNC (MIM# 160980) is an autosomal dominant disorder associated with skin lesions, cardiac and other myxomas, and different types of endocrine tumors including primary pigmented nodular adrenocortical disease (PPNAD), growth hormone–secreting pituitary tumors, gonadal and thyroid neoplasias; rarely, malignant psammomatous melanotic schwannomas (PMS) may also be present (Carney, 1995; Carney, et al., 1986; Stratakis, et al., 1998). The vast majority of the reported PRKAR1A mutations lead to premature stop codons and subsequent degradation of the mutant mRNAs species by NMD (Kirschner, et al., 2000b). Inactivating PRKAR1A mutations lead to excess PKA signaling in affected tissues (Robinson-White, et al., 2006a).
PRKAR1A is located on chromosome 17q23-q24 and extends to a total genomic length of approximately 21 kB. The gene is composed of 11 exons, 10 of which coding, with a coding region of 1143 bp starting from exon 2. The RIα protein consists of 384 AA organized in a dimerization/docking domain at the amino-terminal, followed by a PKA inhibitor site, two tandem binding domains for cAMP at the carboxyl terminus (cAMP:A and cAMP:B), and a linker region that contains the main docking site for the C subunit (Zawadzki and Taylor, 2004). Until recently, it was widely accepted that, in contrast to RIIα, which typically is docked at discrete cellular locations through A Kinase Anchoring Proteins (AKAPs), RIα tends to diffuse in the cytoplasm (Kinderman, et al., 2006; Skalhegg, et al., 1994; Wong and Scott, 2004). In the past several years, however, evidence has accumulated that RIα can be dynamically recruited as well localized at certain compartments of the cell such as the plasma membrane, mitochondria, cytoskeleton, and centrosomes, and AKAPs capable of binding it have been described (Huang, et al., 1997a; Huang, et al., 1997b).
PRKAR1A orthologues have been characterized in many species, and high level of preservation can be tracked back to invertebrates (Aedes aegipty, 73% homology) and protozoa (Ciona intestinalis, 63% homology). The high degree of amino acid sequence conservation underlines the significance of PRKAR1A, the functional “receptor” for cAMP. Indeed, both cloning and gene knockout studies on the individual regulatory subunits have confirmed the essential role of PRKAR1A in maintaining cAMP control of PKA activity through compensatory mechanisms assuring balanced regulation of the free catalytic subunits (Amieux and McKnight, 2002).
Since the identification of PRKAR1A as the gene causing CNC, more than a hundred different pathogenic mutations have been reported, and their molecular and functional analysis has been investigated in several major studies (Bertherat, et al., 2009; Greene, et al., 2008; Groussin, et al., 2002; Kirschner, et al., 2000a; Veugelers, et al., 2004). Herein we review all described to date PRKAR1A sequence variations, including 17 that have not been published before, in an attempt for better understanding of their molecular context and clinical relevance.
To date, a total of 117 different PRKAR1A mutations have been identified in 387 unrelated families of various ethnic origin; they are summarized in Table 1. The molecular changes involve single base substitutions and small (up to 15bp) deletions, insertions or combined rearrangements that are spread along the whole open reading frame of the gene; in addition, several relatively large deletions are reported (Blyth, et al., 2008; Horvath, et al., 2008a). A schematic representation of the PRKAR1A mutations’ type and location is shown on Figure 1.
The mutations in PRKAR1A are spread along the whole coding sequence, without significant preference for an exon or a domain. Most of them are unique – they are identified in single families only (Bertherat, et al., 2009). To date, only 3 mutations have been found in more than three unrelated pedigrees: c.82C>T, c.491_492delTG and c.709(-7-2)del6; these mutations have been seen in kindreds with different racial and ethnic background, suggesting that they are likely to result from more than one independent mutation events.
A graph representing the relationship between the molecular type of the PRKAR1A mutations’ and their functional effect is shown on Figure 2. Relatively small proportion (20/117, 17.1%) of the unique mutations result in the expression of an altered protein; these group convenes the 11 missense substitutions, 6 frameshift mutations affecting the last exon of the gene and thus escaping NMD, one in-frame deletion that eliminates exon 3 in frame, and one splice variant that leads to exon 7 skipping. The vast majority of mutations (97/117, 82.9%) result in premature stop codon (PSC) generation caused by nonsense and frameshift changes upstream of the last gene exon; the mutant mRNAs are degraded through NMD. In this second group fall also the majority of the splice variants – disrupted donor or acceptor sequences lead to exon skip and/or retaining of part of the intron, and, thus, direct or by frame-shift incorporation of PSC. The different mutations’ type functional effect and relation to the phenotype are discussed below.
As mentioned, PSC are caused by nonsense, frame shift and splice variants located before the last exon of PRKAR1A. The expression of the shortened protein is abolished by the NMD, as shown by our experiments (Kirschner, et al., 2000b); thus, the overall effect of this type of mutation is a 50% reduction in cellular PRKAR1A levels. The resulting higher proportion of catalytic to regulatory subunits causes increased signaling and activation of the downstream cellular processes such as proliferation and differentiation, (Robinson-White, et al., 2006b). Hence, NMD mutations cause disease in the affected tissues through a mechanism of haploinsufficiency. This type of mutations is found in the majority of the CNC patients and the resulting phenotypes vary in a very wide range both as number of manifestation and severity of their expression. The factors to affect this variability, besides age, gender and environmental agents are likely to include other genetic determinants with modifying effects. The later is supported by family clustering of particular sets of manifestations as well as our recent studies showing influence of genetic variants in other cAMP-related molecules on the severity of the CNC phenotype (manuscript in preparation). Although the uniqueness of the most of the mutations limits genotype/phenotype analysis, it is now accepted that phenotype/genotype correlation is rare among carriers of NMD mutations in PRKAR1A - they all share haploinsufficiency at the protein level.
The expressed missense substitutions in PRKAR1A are listed in Table 2. While the pathogenic effect of the NMD mutations is almost certain to destroy the PRKAR1A function, the expressed mutations that lead to altered protein clearly require estimation of their pathogenic potential. Eight of these mutations – six aminoacid substitutions (S9N, R74C, R146S, D183Y, A213D and G289W), and the two in-frame deletions of exons 3 and 7 - have been subject of detailed investigation of their effect in vitro (Greene, et al., 2008; Meoli, et al., 2008). The expressed mutations were spread over all functional PRKAR1A domains and every one of them led to variable extends of decrease of the cAMP binding and increase in the cAMP-specific PKA activation (Greene, et al., 2008; Meoli, et al., 2008). As expected, the highest impact on the protein function was measured for the in-frame deletion that eliminates exon 3, and, respectively, the primary binding site for the catalytic subunit (Greene, et al., 2008); this mutation leads to significant elevation in the PKA activity even in the absence of cAMP. Similar, although milder effect on the protein is caused by the R146S, which disrupts the secondary binding site for the catalytic subunit (residues 138-148 and 232-247 within the cAMP binding domain A (Greene, et al., 2008). Expressed mutations affecting the cAMP binding domains (A183Y, A213D and G289W), are shown, in line with their position, to significantly decrease the cAMP binding; the proposed mechanism of action is through conformational changes that alter the enzymatic function. Conformational changes are also proposed to take place for the last two of the studied expressed mutations: S9N, located in the dimerisation/docking domain, and R74C, which is positioned in the linker region, outside any known functional sequence.
Notably, substantial correlation between the in vitro studies and the in silico modeling of the effect of the expressed missense substitutions on the protein function, along with the homology data, was observed (See Table 2). In silico prediction of the effect of the missense substitutions on the protein function (http://coot.embl.de/PolyPhen/) has estimated highest potential to impact the protein function generally for mutations residing in the cAMP binding domains (G168S, D183Y, A213D and G289D, Table 2).
Separate attention requires the missense substitution affecting the first coding aminoacid - M1V. In silico modeling predicted very severe effect on the protein function (See Table 2). Consistent with the above, the in vitro studies have shown significant effect of the mutation on both cAMP binding and cAMP specific PKA activation. However, although the mutant mRNA is found to be expressed in equal to the wt levels in carriers of the mutation, it is not clear if the mutant protein is expressed and stable in vivo (Kirschner, et al., 2000b). The possibility of alternate expression from a surrogate initiation site has been explored – an in-frame ATG in the context of a relatively good match for Kozak sequence is located 141 bp downstream from the original initiation codon; however, Western blot did not detect shorter protein forms.
Analysis of the phenotype characteristics in the carriers of expressed PRKAR1A mutations revealed severe CNC phenotype in the carrier of the in-frame deletion of ex3 and the c.708 +1G>T mutation, expressed both as number of manifestations and severity of their expression. Apart from this finding, no genotype-phenotype connection could be characterized within this relatively small group, which is expected taken the limited number of affected patients and the variable number CNC manifestations. Nevertheless, the finding of expressed PRKAR1A mutant variations that cause disease related changes in vitro, confirms that altered PRKAR1A function, not only haploinsufficiency, is enough to lead to the disease phenotype.
Six of the frameshift mutations (c.1055_1058delGACC, c.1067_1070delAACGins5, c.1076_1077delTTins13, c.1083delA, c.1131_1132delTG, and c.1142_1145delTCTG) are located in the last exon of PRKAR1A and are predicted to escape NMD. Each of the mutations causes a frame shift that abolishes the wild type termination codon and generates a new one further downstream. Compared to the wild type PRKAR1A, the predicted mutant proteins are longer and carry a different 3′ AA sequence, resulting from codon rearrangement. The functional significance of the elongated proteins versus the wild-type R1α is subject of another report under submission.
Several relatively large deletions in the region of PRKAR1A have been described so far (Blyth, et al., 2008; Horvath, et al., 2008a), including the above discussed expressed deletion that eliminates exon 3 in frame (c.178_348del171/p.E60_K116del). This mutation has been shown to result to the in vivo expression of shorter PRKAR1A protein lacking part of the linker region connecting the dimerisation/docking domain and the first cAMP binding domain; the deleted region includes the binding site for the catalytic subunit/inhibitor (Horvath, et al., 2008a). Only one patient was found to carry this mutation, and he presented with very severe phenotype, both as number of CNC manifestation and level of their expression (Horvath, et al., 2008a). In vitro studies with expression vectors harboring PRKAR1A ORF lacking exon 3 have shown extreme effect on the protein function (Greene, et al., 2008). The second identified deletion was located in the upstream regulatory region of PRKAR1A and did not affect the ORF of the gene; this mutation is expected to lead to decreased PRKAR1A mRNA levels but no other effects on the protein; the molecular phenotype is predicted to be PRKAR1A haploinsufficiency, consistent with the majority of PRKAR1A mutations causing CNC. The third noteworthy deletion encompasses significantly larger genomic region from 17q24.2-q24.3 and eliminates, along with PRKAR1A, 13 more genes (Blyth, et al., 2008). This mutation was identified as a de novo genomic rearrangement in a 12 years old patient presenting with posterior laryngeal cleft, moderate growth and mental retardation, microcephaly, and, as the only CNC related manifestation - multiple freckles and lentigines. However, other CNC manifestations can be underrepresented due to the severity of the main phenotype, or, more likely may occur later in the development. Although larger rearrangements involving PRKAR1A region are reported in the literature, in most of the cases they led to extremely severe phenotype and premature death that prevented comprehensive analysis of relatively mild CNC manifestation with later onset (Bridge, et al., 1985; Levin, et al., 1995; Olney, et al., 1999).
Twenty-eight splice variants are identified to date in the close proximity of the exon-intron junctions of PRKAR1A (Table 3). As mentioned, the vast majority of them lead to a frame-shift and subsequent incorporation of PMS. Notably, for the most of the splice mutations (23, 82%), the in silico modeling predicted complete abrogation of the junction formation; in 2 of them accompanied by a shift of the predicted junction with 1 base (see Table 3). For the remaining five variants (c440+4delG, c.502G>A, c.550G>A, c.763delAT, and 891+3A>G) a significant decrease in the probability score to form junction compared to the wt was estimated (see Table 3). It is noteworthy that three of these last variants affected directly exonic sequences and their effect on the splice-site was complemented by a change in the approximate coding sequence, which additionally increased their pathogenic impact. Combined with the family data that show segregation of these variants with the CNC phenotype, all described 28 splice changes were classified as pathogenic.
Interestingly, some splice-site alleles may express both mutant and wild type mRNA molecules in a proportion depending on the type of the mutation and its location relevant to the junction. Thus, the relative decrease of the cellular wt PRKAR1A for these mutations is expected to be lower, and, accordingly, to lead to a milder phenotype. This assumption is in line with a recent analysis of a “mild” splice variant (c.709(-7-2)del6), which is one of the very few mutations where incomplete penetrance of PRKAR1A mutation is seen, and, in the affected individuals, only PPNAD and lentiginosis – the most common disease manifestations - have been diagnosed (Groussin, et al., 2006).
Four nonpathogenic unique synonymous substitutions in the third base of the codon have been described so far in PRKAR1A: c.87G>A/p.A29A, c.204G>A/p.L68L, c.318G>C/p.T106T, and c.492G>A/p.V164V. From them, the first one has been seen in at least five unrelated kindreds, and the remaining three are detected in single families only. Multiple intronic variants in close proximity of the junctions have been described, from them the most frequent being c.349-5dup5 (detected in approximately 15% of the studied alleles), followed by c.769-24A>G and c.891-34G>T (in approximately 12% and 10% of the studied alleles, respectively). However, the extend to which different research groups analyze the exon flanking sequences widely vary and the frequency of these variants may therefore be significantly underreported; furthermore, both missense variants and splice changes may be as well underreported. All these polymorphic variants are seen in healthy control individuals and in both CNC families, and in the later, they do not show segregation with the disease status.
Since the first description of the disease, numerous CNC patients have been reported from all ethnic groups and presenting with various number, combination and severity of manifestations. The most recently re-evaluated diagnostic criteria for CNC are listed in Table 4; a definite diagnosis is given when two or more major manifestations are present. Soon after the identification of PRKAR1A and the initial patient screenings, the mutation detection rate was estimated to approximately 50% (Figure 3, A). After a decade of comprehensive phenotype analysis and re-evaluation of the diagnostic criteria, it is now assumed that mutations in PRKAR1A are responsible for the disease in more than 60% of the CNC patients (Figure 2B). The largest to date study of 185 families, detects PRKAR1A mutations in 114 (62%) (Bertherat, et al., 2009). Interestingly, this percent rises up to 80 among CNC patients presenting with Cushing syndrome due to PPNAD (Cazabat, et al., 2007). Among the mutation carriers, Cushing syndrome due to PPNAD is seen in 70% of the female patients before 45 years, but in only 45% of the male patients, likely reflecting the generally higher frequency of Cushing syndrome in females (Grumbach, et al., 2003). These numbers, however, may represent an underestimate of the true incidence of Cushing syndrome, since autopsy of CNC patients detects PPNAD in almost 100% of the cases (Stratakis, et al., 2001).
When opposed to the PRKAR1A mutation carriers, CNC patients with no genetic defect in PRKAR1A presented with disease overall later in life and more frequently in a sporadic fashion suggesting implication of different genetic factors (Bertherat, et al., 2009). Further, several phenotype subgroups could be distinguished among the PRKAR1A mutation negative individuals: with typical CNC manifestations (9%); with PPNAD and lentigines only (11%); with isolated bilateral micronodular hyperplasia (12%); (see Figure 3, B). Recently, a genome wide based study on the last group identified inactivating mutations in two of the human genes encoding phosphodiesterases – PDE11A and PDE8B (Gunther, et al., 2004; Horvath, et al., 2006a; Horvath, et al., 2008b). Another CNC locus on region 2p16 has been suggested by linkage analysis, however, so far no other gene has been identified, and the genetic searches for additional molecular factors continues (Stratakis, et al., 1996; Stratakis, et al., 2001).
The overall penetrance of CNC among PRKAR1A mutations’ carriers is above 95% by the age of 50; only few PRKAR1A carriers are reported who do not meet the diagnostic criteria. Only two mutations so far have been found to result in incomplete penetrance of CNC – the splice variant c.709(-7-2)del6 and the initiation alternating substitution p.M1Vp. Even more interesting is the observation that, when expressed, these two mutations lead to relatively mild CNC phenotype restricted mostly to PPNAD, in some cases accompanied by lentigines. These findings were confirmed in all 12 unrelated families affected by c.709(-7-2)del6 and in the two unrelated kindreds where c.1A>G/p.M1V was seen (Groussin, et al., 2006). The explanation of this mild, and, in some cases, missing phenotype is seek in the light molecular effect of these two mutations – as a splice variant, c.709(-7-2)del6 does not affect 100% of the molecules, and, respectively, leads to lower level of PRKAR1A insufficiency. The molecular effect of M1V substitution is not clear, but the same substitution in the first codon is suggested to be a mild mutation for other genetic disorders such as Smith-Lemli-Opitz syndrome (Witsch-Baumgartner, et al., 2005). Taking into account the particular sensitivity of the adrenal cells to even minor changes in the PKA activity and/or PRKAR1A levels, it is suggested that adrenal will be the first tissue to react on these PRKAR1A mutations.
Sequencing of the entire coding region of PRKAR1A has not identified alterations in at least 200 unrelated chromosomes. In addition, genotyping of 745 unrelated healthy individuals enrolled in New York Cancer Prospective study (Mitchell, et al., 2004) for 51 unique PRKAR1A mutations (Supplementary Table 1) did not detect the presence of any of them. Thus, PRKAR1A mutations are concluded to be very rare in general population. It is an essential parallel to be made with the gene encoding phosphodiesterase type 11A (PDE11A) – despite the proven association between pathogenic mutations, including null alleles, with adrenal neoplasias, such mutations are seen in general population with lower frequency (Horvath, et al., 2006a; Horvath, et al., 2006b).
Until recently, genotype-phenotype correlation studies were limited by the high variability of the CNC phenotype, both as number and combinations of manifestations, and the relatively limited series of patients. After more than 20 years of analysis and studying of CNC patients, as well as the fusion of the major databases from the National Institutes of Health, Bethesda, MD, and the Hospital Cochin, Paris, France, more than 380 patients are genotypically characterized and clinically evaluated. An overall phenotypic segregation could be seen when the PRKAR1A mutation carriers were opposed to the mutation negative CNC patients: the former presented more frequently and earlier in life with pigmented skin lesions, myxomas, thyroid and gonadal tumors (Bertherat, et al., 2009). Further, subgroups of patients presenting with particular genotype-phenotype correlation could be outlined. The first such a group was presented by patients with isolated PPNAD, in some cases accompanied with lentiginosis. Among these patients, the following tendencies were observed: (1) patients diagnosed before 8 years of age were rarely carriers of PRKAR1A mutation; (2) most of the patients with isolated PPNAD and presence of PRKAR1A mutation were carriers of either the c.709 (-7-2) del6(ttttta) mutation (p<0.0001) or the c.1A>G/p.M1V substitution affecting the initiation codon of the protein. These observations are in line with the discussed above mild, mainly adrenal or, in some cases absent phenotype associated with these two PRKAR1A mutations.
The second group of CNC patients with particular genotype-phenotype correlation was comprised by individuals with myxomas (affecting all locations - skin, heart, and breast), PMS, thyroid tumors, and LCCSCT. In these patients, PRKAR1A mutations were seen substantially more often. Related to this is the acknowledgment that certain tumors presented at significantly younger age in the PRKAR1A mutation carriers: cardiac myxomas (p=0.02), thyroid tumors (p=0.03) and LCCSCTs (p=0.04) (Bertherat et al., 2009). Within the group of PRKAR1A carriers the mutations located in exons associated more frequently with lentigines, PMS, acromegaly and cardiac myxomas, compared to the intronic ones (p = 0.04); this observation is in line with the above discussion of a milder phenotype associated with splice variants. And, finally, the “hot spot” c.491-492delTG mutation was most significantly associated with lentigines, cardiac myxoma, and thyroid tumors when opposed to all other PRKAR1A defects combined (p = 0.03).
Clinical and genetic studies of CNC patients have revealed considerable heterogeneity both in mutational spectrum and in phenotypic variation. Patients with two or more diagnostic criteria from Table 4 or with isolated PPNAD are referred to genetic analysis of PRKAR1A. Once a mutation is identified, genotyping for the particular variant is recommended for all at-risk relatives. The proportion of de novo mutations in PRKAR1A is estimated to be approximately 20% (our unpublished results). Germline mosaicism has not been seen so far and may not be frequent, and thus not of diagnostic concern currently.
To date, most PRKAR1A studies are focused on its coding regions and on screening for point mutations and small rearrangements detectable by PCR-based techniques. The identification of large deletions in the gene, and the significant proportion of patients with characteristic CNC phenotype and not detectable PRKAR1A mutations have triggered the application of alternative technologies for detection of non-coding variations and larger re-arrangements that may account for the disease phenotype (our on-going project).
Apart from PRKAR1A, three other genes from the cAMP signaling pathway are known to date to cause related or overlapping to CNC phenotype. Germline mutations in PDE11A and PDE8B associate with iMAD (Horvath, et al., 2006a; Horvath, et al., 2008b). Somatic activation mutations in G-alpha stimulatory subunit of the G-protein coupled receptor (GNAS) are known to result in McCune Albright syndrome – a postzygotic somatic multiple neoplasia characterized by polyostotic fibrous dysplasia (POFD), cafe-au-lait skin pigmentation, and peripheral precocious puberty (Weinstein, et al., 1991). Identification of genetic defects in cAMP-signaling molecules in patients with similar and/or overlapping phenotype underlines the importance of this signaling in the normal functioning of the endocrine and other cells. Identification of genetic defects in cAMP-signaling molecules in patients with similar and/or overlapping with CNC features underscores the importance of this signaling pathway in endocrine and other tumorigenesis; on-going studies aim at characterizing this involvement better.