Research Article

Protein Kinase A Effects of an Expressed PRKAR1A Mutation Associated with Aggressive Tumors

Elise Meoli,1 Ioannis Bossis,1 Laure Cazabat,3,4,5 Manos Mavrakis,2 Anelia Horvath,1 Sotiris Stergiopoulos,1 Miriam L. Shiferaw,1 Glawdys Fumey,3,4,5 Karine Perlemoine,3,4,5 Michael Muchow,1 Audrey Robinson-White,1 Frank Weinberg,1 Maria Nesterova,1 Yianna Patronas,1 Lionel Groussin,3,4,5 Je´roˆme Bertherat,3,4,5 and Constantine A. Stratakis1

1Section on Endocrinology and Genetics, Program in Developmental Endocrinology and Genetics, and 2Section on Organelle Biology, Program in Cell Biology and Metabolism, National Institute of Child Health and Human Development, NIH, Bethesda, Maryland; 3Institut National de la Sante´et de la Recherche Me´dicaleU567, De´partementd’Endocrinologie, Me´tabolismeand Cancer, Institut Cochin; 4Centre National de la Recherche Scientifique Unite´Mixte de Recherche 8104; and 5Centre de Re´fe´rence des Maladies Rares de la Surre´nale,Service d’Endocrinologie, Hoˆpital Cochin, Universite´Paris 5, Paris, France

Abstract PRKAR1A germ-line or somatic mutations that lead to tumors a Most PRKAR1A tumorigenic mutations lead to nonsense are associated with increased PKA activity (1–4). RI is the main PKA subunit mediating PKA type I (PKA-I) activity in endocrine mRNA that is decayed; tumor formation has been associated PRKAR1A with an increase in type II kinase A (PKA) subunits. and other tissues (3); most mutations that have been The IVS6+1G>T PRKAR1A mutation leads to a protein lacking identified led to nonsense mRNA, which was not made into protein exon 6 sequences [R1A#184-236 (R1A#6)]. We compared through the process known as nonsense mRNA–mediated decay a in vitro R1A#6 with wild-type (wt) R1A. We assessed PKA (2). It has been assumed that reduction of RI levels, a consequent activity and subunit expression, phosphorylation of target decrease in PKA-I, and an increase in type II PKA (PKA-II) are molecules, and properties of wt-R1A and mutant (mt) R1A;we responsible for increased cAMP-responsive kinase activity in PRKAR1A observed by confocal microscopy R1A tagged with green tissues and primary and transformed cell lines bearing fluorescent protein and its interactions with Cerulean-tagged mutations (5–7). An overall increase in response to cAMP and V catalytic subunit (CA). Introduction of the R1A#6 led to higher PKA-II activity were also seen in mouse cells with 50% of aberrant cellular morphology and higher PKA activity but the wild-type (wt) RIa protein level (8–10). PRKAR1A no increase in type II PKA subunits. There was diffuse, The first mutation that led to an expressed RIa variant cytoplasmic localization of R1A protein in wt-R1A– and that was associated with inherited tumors was described in 2002 R1A#6-transfected cells but the former also exhibited (11). The expression of the mutant (mt) protein lacking exon 6 discrete aggregates of R1A that bound CA; these were absent [R1aD184-236 (R1aD6)] was associated with increased kinase in R1A#6-transfected cells and did not bind CA at baseline or activity because it led to increased phosphorylation of cAMP- in response to cyclic AMP. Other changes induced by R1A#6 responsive element binding protein (CREB) in transfected cells. PRKAR1A included decreased nuclear CA. We conclude that R1A#6 Recently, we reported more mutations leading to leads to increased PKA activity through the mt-R1A decreased expressed mt-RIa variants; they, too, were associated with in vitro binding to CA and does not involve changes in other PKA increased PKA activity (12, 13). subunits, suggesting that a switch to type II PKA activity is not In the absence of decreased R1a protein levels, how do expressed PRKAR1A necessary for increased kinase activity or tumorigenesis. mutations lead to increased PKA activity? PKA, when not [Cancer Res 2008;68(9):3133–41] stimulated by cAMP, exists as a tetrameric holoenzyme that consists of a homodimer of regulatory subunits that bind two inactive Introduction catalytic subunits, one catalytic molecule to each regulatory subunit (14). The accepted model of PKA activation suggests that PRKAR1A Inactivating mutations of the coding for the cooperative binding of two cAMP molecules to each regulatory regulatory subunit type 1A (RIa) of cyclic AMP (cAMP)–dependent subunit results in dissociation and the consequent release of the A (PKA) have been found in sporadic tumors and in two catalytic subunits, which, in turn, are free to phosphorylate the germ line of the majority of patients with the ‘‘complex of serine-threonine residues of target (14, 15). There are four myxomas, spotty skin pigmentation, and endocrine overactivity’’ or encoding the different regulatory subunits (RIa,RIh, RIIa, ‘‘’’ (1, 2). and RIIh) and three encoding the catalytic subunits (Ca,Ch, and Cg); of these, only RIa,RIIa, and Ca are widely expressed whereas the remaining have mostly tissue-specific expression (14). PKA-I activity is mediated mainly through the expression of RIa, whereas Note: Supplementary data for this article are available at Cancer Research Online PKA-II depends on both RIIa and RIIh expression in endocrine and (http://cancerres.aacrjournals.org/). E. Meoli, I. Bossis, and L. Cazabat contributed equally to this work. most other, nonneural, tissues; the balance between PKA-I and PKA- Current address for I. Bossis: School of Veterinary Medicine, University of II has been proposed to be critical for the control of cellular growth, Maryland, College Park, MD 20742. proliferation, and differentiation (14, 16). The best known function Requests for reprints: Constantine A. Stratakis, Section on Endocrinology and Genetics, Program in Developmental Endocrinology and Genetics, National Institute of the regulatory subunits is inhibition of the catalytic subunits, but of Child Health and Human Development, NIH, Bethesda, MD 20892. Phone: 301-496- an increasing body of evidence supports additional functions, 4686; Fax: 301-402-0574; E-mail: [email protected]. I2008 American Association for Cancer Research. including some that may be PKA-independent (15). We recently doi:10.1158/0008-5472.CAN-08-0064 showed direct binding of mammalian target of rapamycin by RIa www.aacrjournals.org 3133 Cancer Res 2008; 68: (9). May 1, 2008

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. Cancer Research

Figure 1. IMRO (A and C) and CAR20.15 (B and D) fibroblasts before (A and B) and after the introduction of the R1aD6 construct. Transfected cells (C and D) showed long extrusions, a more prominent nucleus, and decreased cytoplasm. Gradually the transfected IMRO and CAR20.15 cell lines became apoptotic and did not sustain large numbers of transfected cells. Although similar morphologic changes were also seen after the introduction of R1aD6 in HeLa and HEK293 cells (see Fig. 4), these cell lines were successfully propagated and used for the experiments described in this report.

in vitro (17) and interactions of RIa with an outer mitochondrial mRNA quantification. Total RNA from cells was extracted with TRIzol membrane protein (18). The formation of heterodimers between reagent (Invitrogen) and was purified with the RNeasy Mini Kit (Qiagen). PKA-I and PKA-II subunits (RIa and RIh, and likely between RIIa The quantitative real-time reaction was carried out and analyzed with ABI and RIIh) has also been reported (19) and nuclear localization of Prism 7900HT Sequence Detection System (Applied Biosystems). The primers and probes for PRKAR1A, PRKAR1B, PRKAR2A, PRKAR2B, and regulatory subunits may point to additional, possibly PKA- PRKACA (BioServe Biotechnologies) have been published elsewhere (7). All independent, roles (16, 17, 20). results were normalized against the expression of a Thus, for the expressed variants of RIa, the possibilities for (GAPDH). All points for the standard curves and samples were done in mechanisms associated with tumorigenesis can vary considerably: quadruplicates. Dysregulated kinase activity may be caused by altered binding of Antibodies and expression constructs. All antibodies for this study the catalytic subunit, an increase in PKA-II, heterodimer formation, have previously been described (17); commercially available antibodies for or altered sensitivity to cAMP (11–13); in addition, both direct and the PKA subunits RIa, RIIa,RIh, RIIh, and Ca and for CREB and indirect interactions of RIa and/or PKA with other molecules and extracellular signal–regulated kinase (ERK)-1/2 were also those that we Hin signaling pathways may be affected. Expressed PRKAR1A muta- have previously used (7, 11). HA-RIa was prepared by subcloning a dIII/ Xho PRKAR1A tions seem to be associated with a more aggressive clinical I fragment of the hemagglutinin (HA)-tagged human cDNA from pREP4-HA-RIA (11), as we described elsewhere (17). The green phenotype (11–13), an observation that makes understanding their fluorescent protein (GFP) and Ca-Cerulean constructs were made as we effect on PKA function of paramount importance to the have reported elsewhere (17); the HA-tagged PRKAR1A was also previously development of therapies directed toward cAMP/PKA signaling reported (11). and related tumorigenicity. PKA and cAMP responses. PKA activity was measured as previously We report here our investigation of the first reported, naturally described (1, 3, 4, 7), using [g-32P]deoxy-ATP, in cell extracts that had been occurring and pathogenic PRKAR1A mutation that led to an snap-frozen in liquid nitrogen. All determinations of PKA activity were expressed variant, R1aD184-236; we refer to it from here on as done twice for each sample, which were corrected for protein concentration R1aD6 because it leads to deletion of the sequence coding for exon (per milligram of total protein), and then an average value was calculated 6 of the PRKAR1A cDNA (11). We confirmed the association of this for each experiment. For cAMP responsiveness, transfected cell lines were treated with protein with increased kinase activity and showed that this was À forskolin (10 5 mol/L), 8-bromo-cAMP (8-Br-cAMP; 2 mmol/L), or vehicle due to decreased binding to Ca. and observed continuously during 20 min after stimulation. Transfection of the empty GFP vector was used as a control for the specificity of the signal. Materials and Methods Immunoblotting, immunofluorescence, immunoprecipitation, and Cell culture. HeLa, H295R, HEK293, IMRO-90, and COS cells were nuclear extracts. For immunoblotting (Western), equivalent amounts of obtained from the American Type Culture Collection. A primary protein were separated by SDS-PAGE and then transferred onto a fibroblastoid cell line (CAR20.15) was established, following standard nitrocellulose membrane. Density for each band was analyzed with a methods, from a skin biopsy from a patient with Carney complex and a densitometer. Protein loading was normalized by probing the same germ-line PRKAR1A-inactivating mutation (c.491_492delTG/p.Val164fsX4) membrane with anti–h-actin. that we have published elsewhere (1). Only HeLa and HEK293 cells were For immunofluorescence, cells were grown on coverslips and fixed in 4% used for immunoblotting, immunoprecipitation, and confocal microscopy formalin (15 min), followed by blocking in 0.1% saponin, 1% bovine serum experiments. albumin-PBS (10 min) and sequential incubations with the primary and

Cancer Res 2008; 68: (9). May 1, 2008 3134 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. An Expressed PRKAR1A Mutation

Figure 2. A and B, successful introduction of the GFP constructs containing wt-R1a and R1aD6 in HeLa and HEK293 cells; *, P < 0.05; NS, nonsignificant. Y axis represents arbitrary optical densitometry units. C, in both cell types, PKA activity in response to cAMP and after addition of protein kinase inhibitor was greater in R1aD6-transfected cells (Y axis represents counts per minute per milligram of protein; *, P < 0.05); D, free PKA activity was increased in HEK293 and had a tendency to be higher in HeLa cells (*, P < 0.05; Y axis represents a unitless ratio).

secondary antibodies as in Western blotting. Cells were washed and the UCSF Chimera software from the Resource for Biocomputing, mounted on slides with Fluoromount-G (Southern Biotechnology Asso- Visualization, and Informatics at the University of California, San Francisco, ciates) in preparation for microscopy as we described elsewhere (17). CA (25), as suggested elsewhere (22, 23). For GFP-RIa and Ca coimmunoprecipitations, captured immunopreci- Statistics. PKA activity, mRNA expression, and optical densitometry data pitates were washed thrice with lysis buffer and subjected to Western were obtained in at least duplicate measurements and an average was blot analysis for RIa and Ca, as we have reported elsewhere (17). calculated for each value. Comparisons were made by ANOVA using the Nuclear extracts and cytosolic preparations were prepared using standard PDIFF procedure of the Statistical Analysis System (SAS; SAS Institute) to methods (17). compare differences between treatment means. Differences were consid- Confocal microscopy and photobleaching. Cells were seeded over- ered significant at P < 0.05. night in Lab-Tek chambers (Nalge Nunc) and cotransfected with the plasmids of interest, using Lipofectamine 2000 (Invitrogen). Confocal microscopic images of cells 15 to 24 h posttransfection were captured on a Results Zeiss 510 or Zeiss ConfoCor-2 inverted microscope using the 413-nm line of Cell morphology, PKA activity, and PKA subunit mRNA and a Kr laser with a 430- to 470-nm emission filter for Cerulean and the 488-nm protein levels. After introduction of mt-R1aD6, IMR-90 and the line of an Ar laser with a 505- to 530-nm emission filter for enhanced GFP primary cell line CAR20.15 (Fig. 1A and B) both assumed a more  (17). Images were captured with a Plan-Apochromat 63 oil immersion compact, randomly oriented phenotype with an overall decrease in objective (numerical aperture, 1.4). Cells expressing both proteins were cytoplasm and a prominence of the nucleus (Fig. 1C and D). These selected for z-sectioning. Z stacks were taken using a pinhole of 1 airy unit changes were especially obvious in IMR-90 cells (Fig. 1A and C). for both channels. Images were analyzed with ImageJ and Zeiss Image a Examiner software and prepared by Adobe Photoshop 7.0. Fluorescence loss Introduction of mt-RI in H295R, HeLa, and HEK293 cells also in photobleaching experiments were done by repeatedly photobleaching a caused changes in morphology (see below) but did not affect their small region of interest and monitoring fluorescence depletion in distant viability. For the rest of the experiments, only HeLa and HEK293 regions over time, as previously described (17, 21). cells were used. R1A protein modeling. Modeling of RIa sequence was done after the In both cell lines, adequate expression of the RIa constructs was ENSP00000351410 RIa 1–381 amino acid sequence.6 Crystal structures for obtained as shown by both mRNA and protein studies (Fig. 2A and 7 R-C binding exist for bovine RIa and mouse Ca (22, 23). Bovine RIa and B). PKA assays showed that what was suggested by transfections in mouse Ca are highly homologous to their human counterparts: Bovine R1a COS cells (11) was true in both these and human cell lines: (not including exon 6) has only three changes from the human molecule, Groussin et al. had shown that introduction of the R1aD6 and the mouse Ca is 98% homologous with human Ca with only two construct in COS cells increased forskolin-induced cAMP-respon- nonconservative changes. Amino acids 184 to 236 were highlighted on the existing structure protein sive element activity significantly more than did the introduction of C database (PDB) #1RL3 (24). Molecular graphics images were produced using the wt-R1a. Figure 2 shows that introduction of the R1aD6 construct in HEK293 cells had the same effect as the wt-R1a on total kinase activity at baseline, but led to significant increases in 6 See http://www.ensembl.org. cAMP-stimulated kinase acticity. The data in COS, H295R, and 7 See http://www.rcsb.org/pdb/cgi/pdbId=2QCS. HeLa cells were similar but we also found significant differences www.aacrjournals.org 3135 Cancer Res 2008; 68: (9). May 1, 2008

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. Cancer Research between PKA assays done 24 or 48 hours posttransfection. Figure effects on PRKAR2A mRNA (Fig. 3A). The wt and mt constructs 2D shows that the ratio of free PKA versus total PKA was differed only in their effects on PRKAR1B (P < 0.001; which, significantly greater in HEK293 cells (and had the tendency to be however, was barely expressed) and tended to be different in their significantly higher in HeLa cells) 24 hours after transfection with effects on PRKACA (P = 0.08). There were no significant the R1aD6 construct versus wt-R1a. In addition to CREB corresponding effects on the protein content of whole-cell extracts phosphorylation, we also saw greater phosphorylation of ERK1 for the same subunits both at 24 and 48 hours after transfection and ERK2 in COS, H295R, and HeLa cells (data not shown), (Fig. 3B), with the exception of endogenous R1a: The cells consistent with previous findings that had shown increased transfected with wt-RIa had modestly higher endogenous RIa pERK1/2 and pCREB as a marker of abnormal PKA activity in levels at 48 hours after transfection (Fig. 3B, top left; P = 0.04); the the lymphocytes and adrenal cells and tumors of patients with ratio between RI and RII subunits, however, was not different at germ-line PRKAR1A-inactivating mutations (3, 6, 11). 24 or 48 hours. The introduction of the wt-R1a and R1aD6 constructs in both Cellular distribution of R1A and R1A#6; binding to CA and HeLa and HEK293 cells had significant effects on the mRNA of other PKA subunits. The increase of kinase activity caused by three other PKA subunits (PRKACA, PRKA2B, and PRKAR1B) but no R1aD6 without a dramatic effect on the whole-cell content of the

Figure 3. A, mRNA levels of PKA subunits in HEK293 cells after introduction of the wt-R1a and R1aD6-GFP constructs. *, P < 0.05. B, protein levels of PKA subunits in HEK293 cells 24 and 48 h after introduction of the wt-RIa and R1aD6-GFP constructs. *, P < 0.05; Y axis represents arbitrary optical densitometry units.

Cancer Res 2008; 68: (9). May 1, 2008 3136 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. An Expressed PRKAR1A Mutation

Figure 4. Morphology of HeLa and HEK293 cells after introduction of the wt-RIa and RIaD6-GFP constructs: Puncta are seen in both cell types only after transfection with the wt-R1a construct, whereas both wt-R1a and R1aD6 have diffuse cytoplasmic localization and a small nuclear pool; introduction of the mt-RIa caused morphologic changes (projections, irregular cytoplasmic shape) similar to what were seen in IMRO and CAR20.15 cells (Fig. 1A–D) and reported elsewhere after the introduction of a catalytic subunit GFP construct (35).

other PKA subunits suggested an effect on the holoenzyme and/or immunofluorescence studies suggested that RIIa, but not RIIh its catalytic activity that did not involve a major imbalance (Fig. 5A, bottom)orRIh (data not shown), may form heterodimers between PKA-I and PKA-II isozymes. We then determined the with mt-RIa. Indeed, in immunoprecipitation experiments, wt-R1a cellular compartmentalization of wt-R1a and mt-R1a at baseline and mt-RIa showed that they both formed heterodimers with and after activation of the cAMP signaling pathway; a construct of RIIa when immunoprecipitated with Ca (Fig. 5A, bottom) or with the human Ca with Cerulean was also used so that we could RIa (data not shown), but there were no quantitative differences observe any interactions between wt-RIa, mt-RIa, and Ca in in repeated experiments between the two constructs. doubly transfected cells. We then used confocal microscopy to study the interactions of First, diffuse cytoplasmic localization was observed with both wt-R1a and mt-RIa with Ca-Cerulean in cotransfections: Again wt-RIa and mt-RIa constructs, but obvious puncta, dense there was little, if any, interaction between mt-RIa and Ca whereas aggregates, were observed in the cytoplasm only with the wt-RIa several of the puncta containing wt-RIa showed that they protein (Fig. 4). Confocal microscopy showed, in addition, some contained Ca (Fig. 5C and D). We also conducted fluorescence nuclear localization of R1a for both constructs; however, loss in photobleaching experiments, in which we repeatedly perinuclear puncta that were obvious at baseline disappeared after photobleached (i.e., noninvasively abolished) the fluorescence in a forskolin treatment, and nuclear translocation occurred only with region in the cytoplasm of cells containing wt-RIa or mt-RIa the wt-R1a construct; there were no distinct perinuclear puncta in cotransfected with Ca-Cerulean both at baseline and after cells transfected with mt-RIa, and the mobility of the R1aD6 exposure to 8-Br-cAMP. There were no significant differences construct was not enhanced by forskolin or 8-Br-cAMP (Fig. 5A, between wt-RIa and mt-RIa in these experiments (data not top). Cell morphology was different in cells transfected with shown). the wt-RIa construct (Fig. 4), as we noted above with other cells Projection of the mt-RIa deletion on the existing wt-RIa-Ca (Fig. 1). These data have been replicated with a different GFP interaction models (22, 23) showed that deletion of sequences construct and in a number of other cell types (data not shown). corresponding to exon 6 would affect not only interactions with We next examined the mt-R1a interactions with Ca and other cAMP (which was predictable from elimination of the cAMP- subunits: Immunofluorescence studies suggested that R1aD6 was binding domain A of RIa) but also one of the major grooves likely to not bind Ca, at least not as effectively as wt-R1a (Fig. 5A, forming around Ca in the RIa-Ca dimer (Fig. 6A). bottom). Immunoprecipitation experiments confirmed this obser- Nuclear and cytoplasmic R1A and CA. Transfections with the vation: When cells transfected with wt-R1a or mt-R1a were wt-R1a and mt-R1a constructs showed a small but detectable pool immunoprecipitated with Ca, only the wt-RIa–transfected cells of R1a in the nucleus in both HEK293 and HeLa cells. cAMP showed any immunostaining with the RIa-specific antibody induced wt-R1a mobility and entry into nucleus both before and (Fig. 5B, top). When immunoprecipitated with RIa, cells transfected after introduction of the Ca construct, whereas it had no effect on with mt-RIa showed significantly less immunostaining for Ca than mt-R1a; introduction of Ca did not change the lack of response to cells transfected with wt-RIa (Fig. 5B, bottom). Interestingly, cAMP (Supplementary data). Western blotting of cytosolic and www.aacrjournals.org 3137 Cancer Res 2008; 68: (9). May 1, 2008

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. Cancer Research nuclear extracts of HEK293 cells showed that all PKA subunits increased cAMP-responsive element activity had been shown in could be found in the nucleus; Ca was mostly nuclear and COS cells transfected with R1aD6, it was not known how this was regulatory subunits were mostly cytosolic, as expected. There were mediated. Our studies are significant for both what they showed differences in the cytosolic and nuclear localization of endogenous and what they did not show: We first showed that the mechanism RIa and Ca when preparations from wt-R1a– and mt-RIa– responsible for the observation of increased PKA activity due to the transfected cells were compared: The nuclear content of Ca in mt- R1aD6 mutation involves a Ca that is not bound by mt-R1a and, RIa–transfected cells was significantly lower (P < 0.001); mt-R1a thus, is most likely inappropriately regulated. Absence of amino also did not affect the cytosolic endogenous RIa, which was acids 184 to 236 of the RIa protein abolished part of cAMP-binding induced by the wt-RIa construct, whereas both wt-R1a and mt-RIa domain A (which extends between amino acids 143 and 260) but increased cytosolic endogenous Ca (Fig. 6B–D). retained cAMP-binding domain B (amino acids 261–374) and the entire amino (NH2) part (22–24) including the dimerization/ docking domain. Recognition of RIa by Ca is mediated through Discussion specific amino acid interactions that include the critical Tyr205 The present investigation is the first attempt to study the PKA (26, 27) that is absent in R1aD6(Fig.6A); Glu200, Leu201, and Ile204 effects of a naturally occurring PRKAR1A mutation that has been are also absent in R1aD6 and play an important role in RIa-Ca associated with an aggressive clinical phenotype (11). Although binding (27). Therefore, deficient binding of Ca by R1aD6 could

Figure 5. A, different distribution and mobility of the wt-R1a and R1aD6-GFP constructs in HeLa cells in response to cAMP (see text for details). Immunohistochemical fluorescence shows that wt-R1a-GFP colocalizes (arrows) with Ca (1), whereas there is less R1aD6-GFP colocalizing with Ca (2); both wt-R1a and R1aD6-GFP constructs colocalized with RIIa—only the R1aD6-GFP–transfected cells immunostained with RIIa (arrows) are shown (3); finally, there was no obvious colocalization with RIIh of neither construct—only the R1aD6-GFP–transfected cells immunostained with RIIh are shown (4). B, immunoprecipitation experiment where three different preparations of HEK293 cells (1, wt-R1a GFP-transfected cells; 2, R1aD6-GFP–transfected cells) were immunoprecipated with a Ca-specific antibody: In all preparations, Ca recognizes what was immunoprecipitated (positive control); the RIa-specific antibody detects the endogenous R1a in all lanes, but only the wt-R1a GFP is present in R1a-Ca complexes; RIIa is shown in all lanes of Ca immunoprecipitates as expected. C and D, differential distribution and colocalization (arrows) with a Ca-Cerulean construct of the wt-R1a (C) and R1aD6-GFP (D) constructs in HeLa cells: The puncta of R1a colocalize with Ca, whereas there is no colocalization with the diffusely present R1aD6 molecules.

Cancer Res 2008; 68: (9). May 1, 2008 3138 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. An Expressed PRKAR1A Mutation

Figure 6. The deletion in R1aD6 highlighted within an available model of R1a when bound to cAMP (A) and in a complex with Ca, structures were adapted from refs. 22–24. R1aD6 lacks the most conserved region of R1a (26), one that is essential for both cAMP binding and interaction with Ca. Differences in cytoplasmic and nuclear RIa (B) and Ca (C) after introduction of the wt-R1a and R1aD6-GFP constructs in HEK293 cells. D, representative blots from one experiment for all PKA subunits: There is R1a present in the nuclear extracts of all lanes, whereas there is no R1h in the nuclear extracts and little in cytosolic extracts (lanes 1 and 4, from wt-R1a-transfected cells; lanes 2 and 5, from R1aD6-transfected cells; lanes 3 and 6, from mock transfections); *, P < 0.05; NS, nonsignificant. Y axis represents random optical densitometry units.

have been predicted from existing models, with the caveat that the increases in expression of RIIa or RIIh (8–10, 28), the question latter are derived from interactions of the murine Ca and the remained whether these were simply compensatory changes (29). bovine R1a (which are highly homologous molecules). The R1aD6 studies showed that, in at least the case of one Second, we also showed that the introduction of R1aD6 in cells mutation associated with aggressive human tumors, an in vitro did not alter their content of other PKA subunits, and more increase of PKA-II could not be shown. specifically, it did not lead to an increase in PKA-II subunits Several other findings are worthy of discussion, however, and (Fig. 6A). The balance between PKA-I and PKA-II isozymes has require further investigation: Wt-R1a formed cytosolic aggregates, been considered critical for PKA control of growth, proliferation, puncta, that have been seen by us (17, 30) and by other and differentiation (14, 16). Although both human and mouse investigators and in various (although not all) types of cells, tagged studies confirmed that R1a reduction was associated with with GFP (17, 30, 31) and untagged (32, 33). Zaccolo et al. (34) www.aacrjournals.org 3139 Cancer Res 2008; 68: (9). May 1, 2008

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. Cancer Research showed similar puncta with the RIIa-GFP, but not Ca-GFP, RII subunits because, interestingly, Ca diffuses away from PKA-II constructs. GFP tagged to the NH2 or the COOH terminus of R1a but not from PKA-I (31). In addition, R1aD6-transfected cells had resulted in similar formations (17, 30), and R1aD6-GFP completely less endogenous cytosolic R1a levels (Fig. 6B–D). Thus, the net lacked them (Fig. 6A). Fluorescence loss in photobleaching effect of R1aD6 introduction was to further decrease any R1a-Ca experiments showed the dynamic relationship between the puncta binding that would normally take place in these cells in both the and the R1a and Ca cytosolic pools (17). We found that these cytoplasm and the nucleus. It remains unclear how decreased formations in HeLa, MNTI, and HEK293 cells associated with late nuclear Ca presence affects other functions of PKA in the nucleus endosomes and autophagosomes in cultured cells (17, 30). of R1aD6-transfected cells. Thus, R1aD6 not only did not bind Ca but also lacked the cAMP, either directly (8-Br-cAMP) or generated by forskolin, ability to associate with these cellular formations. This was not did not affect R1aD6 binding to Ca. Recent data indicate that due to a defect in dimerization: R1aD6 bound both wt-R1a and cAMP-bound PKA-I holoenzyme is stable yet catalytically active with itself (Fig. 6B–D); this could also have been predicted by and that maximum cAMP levels do not lead to immediate or the retention of the dimerization/docking domain (26) by the mt complete dissociation of the tetramer (31, 35, 44), consistent with protein. Binding to the Ca was not necessary for these our data. R1aD6 lacks cAMP-binding domain A (Fig. 6A) but one aggregates to form: We have shown elsewhere (17), and it was could hypothesize that it would respond to cAMP: cAMP binding evident from our studies here, too (Fig. 5), that these puncta to R1a is an orchestrated event that is assisted by the tertiary formed by wt-R1a did not always associate with endogenous structure of the PKA holomer; one cAMP molecule binds first to (or transfected) Ca. cAMP-binding domain B of the inactive PKA tetramer–bound Another significant observation was that cells transfected with R1a. However, complete separation of the A and B domains R1aD6 changed morphology consistent with earlier published data (so that cAMP-binding domain B is ‘‘available’’ for binding) is not (34, 35) and in other settings (36, 37), whereas cells transfected with possible without formation of the R1a-Ca complex (27, 31). wt-R1a retained their original shape and contour (Figs. 1 and 5). Because R1aD6 is not capable of efficiently forming a dimer with These morphologic changes have been linked to excess Ca (34, 35) Ca, chances are that it would not efficiently bind cAMP, which or increases in PKA activity (36–38) and are consistent with the explains why it did not respond to forskolin and 8-Br-cAMP in observation that R1aD6 did not efficiently bind Ca. our experiments; apparently, increasing the amount of available Nuclear localization of mammalian R1a has been seen before Ca also did not change that, as suggested by our cotransfection but has not been adequately studied, despite the knowledge that experiments. the yeast PKA regulatory subunit Bcy1p is primarily nuclear (39). We conclude that R1aD6, a variant of R1a that lacks cAMP- Our studies showed that both wt-R1a and mt-R1aD6, as well as binding domain A, an ‘‘ancient signaling domain conserved in endogenous R1a, are present in nuclei of HEK293 and HeLa cells, every genome’’ (26), does not bind Ca and has other unique consistent with observations in mouse oocytes (33), cardiac properties when introduced in human cell lines. R1aD6 also myocytes from various species and mouse embryonic cells causes Carney complex and aggressive human tumors, and from (40, 41), cancer cell lines (20), and in other settings (31, 35), that, one can speculate that a ‘‘free’’ Ca, and not a switch to and despite some evidence to the contrary (42). The role of PKA-II, is the cause of tumorigenicity in states of PRKAR1A nuclear PKA-I remains unclear, but regulation of cell division and inactivation. Further studies are needed to confirm this coordination of this event with the cell cycle are likely (40). suggestion that changes at least some of the dogmas related to Interestingly, because R1aD6 did not bind Ca and our studies on the role of PKA in cancer. the Ca content of total cell lysates did not show Ca changes A (Fig. 6 ), we looked at cytosolic and nuclear Ca and found it to Acknowledgments be decreased in R1aD6-transfected cells (Fig. 6B–D). Our hypothesis is that increased PKA activity in R1aD6-expressing Received 1/7/2008; revised 1/30/2008; accepted 1/31/2008. Grant support: U.S. National Institutes of Health, National Institute of Child cells is due to the unregulated, mostly PKA-II–bound, cytosolic Health and Human Development intramural project Z01-HD-000642-04 (C.A. Ca;Ca that is not bound by R1a would be unstable and Stratakis), research funds to J. Bertherat (Hopital Cochin, Paris, France), and Groupement d’Inte´reˆtScientifique-Institut National de la Sante´et de la Recherche unavailable to enter the nucleus, an event that is brought about Me´dicaleInstitut des Maladies Rares and the Plan Hospitalier de Recherche Clinique by simple diffusion (43) and would not be prevented by cytosolic to the Comete Network grant AOM 02068 (J. Bertherat and C.A. Stratakis).

References (17q22–24) in sporadic adrenocortical tumors: 17q regulatory subunit type 1A of develops losses, somatic mutations, and protein kinase A endocrine and other tumors: comparison to Carney 1. Kirschner LS, Carney JA, Pack SD, et al. Mutations of expression and activity. Cancer Res 2003;63:5308–19. complex and other PRKAR1A-induced lesions. J Med the gene encoding the protein kinase A type I-a 5. Sandrini F, Matyakhina L, Sarlis NJ, et al. Regulatory Genet 2004;41:923–31. regulatory subunit in patients with the carney complex. subunit type I- of protein kinase A (PRKAR1A): a tumor- 9. Griffin KJ, Kirschner LS, Matyakhina L, et al. Down- Nat Genet 2000;26:89–92. suppressor gene for sporadic thyroid cancer. Genes regulation of regulatory subunit type 1A of protein 2. Kirschner LS, Sandrini F, Monbo J, Lin JP, Carney JA, Cancer 2002;35:182–92. kinase A leads to endocrine and other tumors. Cancer Stratakis CA. Genetic heterogeneity and spectrum of 6. Robinson-White AJ, Leitner WW, Aleem E, Kaldis P, Res 2004;64:8811–5. mutations of the PRKAR1A gene in patients with Carney Bossis I, Stratakis CA. PRKAR1A inactivation leads to 10. Kirschner LS, Kusewitt DF, Matyakhina L, et al. A complex. Hum Mol Genet 2000;9:3037–46. increased proliferation and decreased in mouse model for the Carney complex tumor syndrome 3. Robinson-White A, Hundley TR, Shiferaw M, Bertherat human B lymphocytes. Cancer Res 2006;66:10603–12. develops neoplasia in cyclic AMP-responsive tissue. J, Sandrini F, Stratakis CA. Protein kinase A activity in 7. Robinson-White A, Meoli E, Stergiopoulos S, et al. Cancer Res 2005;65:4506–14. PRKAR1A-mutant cells, and regulation of mitogen- PRKAR1A mutations and protein kinase A interactions 11. Groussin L, Kirschner LS, Vincent-Dejean C, et al. activated protein kinase (MAPK) ERK1/2. Hum Mol with other signaling pathways in the adrenal cortex. J Molecular analysis of the cyclic AMP-dependent protien Genet 2003;12:1475–84. Clin Endocrinol Metabol 2006;91:2380–8. kinase A (PKA) regulatory subunit 1A (PRKAR1A) gene in 4. Bertherat J, Sandrini F, Matyakhina L, et al. Molecular 8. Griffin KJ, Kirschner LS, Matyakhina L, et al. A patients with Carney complex and/or primary pigmented and functional analysis of PRKAR1A and its locus transgenic mouse bearing an antisense construct of nodular adrenocortical disease (PPNAD) reveals novel

Cancer Res 2008; 68: (9). May 1, 2008 3140 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. An Expressed PRKAR1A Mutation

mutations and clues for pathophysiology: augmented 23. Kim C, Cheng CY, Saldanha SA, Taylor SS. PKA-I 34. Zaccolo M, De Giorgi F, Cho CY, et al. A genetically PKA signaling is associated with adrenal tumorigenesis in holoenzyme structure reveals a mechanism for cAMP- encoded, fluorescent indicator for cyclic AMP in living PPNAD. Am J Hum Genet 2002;71:1433–42. dependent activation. Cell 2007;130:1032–43. cells. Nat Cell Biol 2000;2:25–9. 12. Horvath A, Bossis I, Giatzakis C, et al. Large deletions 24. Wu J, Brown S, Xuong NH, Taylor SS. RIa subunit of 35. Prinz A, Diskar M, Erlbruch A, Herberg FW. Novel, of the PRKAR1A gene in Carney complex: phenotype PKA: a cAMP-free structure reveals a hydrophobic isotype-specific sensors for protein kinase A subunit correlations and implications for laboratory and diag- capping mechanism for docking cAMP into site B. interaction based on bioluminescence resonance energy nostic testing. Clin Cancer Res 2008;14:388–95. Structure 2004;12:1057–65. transfer (BRET). Cell Signal 2006;18:1616–25. 13. Greene EL, Horvath AD, Nesterova M, Giatzakis C, 25. Pettersen EF, Goddard TD, Huang CC, et al. UCSF 36. Olson MF, Krolczyk AJ, Gorman KB, Steinberg RA, Bossis I, Stratakis CA. In vitro functional studies of Chimera—a visualization system for exploratory re- Schimmer BP. Molecular basis for the 3¶,5¶-cyclic naturally occurring pathogenic PRKAR1A mutations search and analysis. J Comput Chem 2004;25:1605–12. adenosine monophosphate resistance of Kin mutant that are not subject to nonsense mRNA decay. Hum 26. Das R, Esposito V, Abu-Abed M, Anand GS, Taylor SS, Y1 adrenocortical tumor cells. Mol Endocrinol 1993;7: Mut. In press 2008. Melacini G. cAMP activation of PKA defines an ancient 477–87. 14. Bossis I, Stratakis CA. PRKAR1A:normaland signaling mechanism. Proc Natl Acad Sci U S A 2007;104: 37. Puck TT, Webb P, Johnson R. Cyclic AMP and the abnormal functions. Endocrinology 2004;145:5452–8. 93–8. reverse transformation reaction. Ann N Y Acad Sci 2002; 15. Taylor SS, Kim C, Vigil D, et al. Dynamics of signaling 27. Gullingsrud J, Kim C, Taylor SS, McCammon JA. 968:122–38. by PKA. Biochim Biophys Acta 2005;1754:25–37. Dynamic binding of PKA regulatory subunit RIa. 38. Porter SE, Dwyer-Nield LD, Malkinson AM. Regula- 16. Cho-Chung YS, Pepe S, Clair T, Budillon A, Nesterova Structure 2006;14:141–9. tion of lung epithelial cell morphology by cAMP- M. cAMP-dependent protein kinase: role in normal and 28. Amieux PS, Howe DG, Knickerbocker H, et al. dependent protein kinase type I isozyme. Am J Physiol malignant growth. Crit Rev Oncol Hematol 1995;21:33–61. Increased basal cAMP-dependent protein kinase activity Lung Cell Mol Physiol 2001;280:L1282–9. 17. Mavrakis M, Lippincott-Schwartz J, Stratakis CA, inhibits the formation of mesoderm-derived structures 39. Griffioen G, Anghileri P, Imre E, Baroni MD, Ruis H. Bossis I. Depletion of type IA regulatory subunit (RI) of in the developing mouse embryo. J Biol Chem 2002;277: Nutritional control of nucleocytoplasmic localization of protein kinase A (PKA) in mammalian cells and tissues 27294–304. cAMP-dependent protein kinase catalytic and regulato- activates mTOR and causes autophagic deficiency. Hum 29. Amieux PS, McKnight GS. The essential role of RIa in ry subunits in Saccharomyces cerevisiae. J Biol Chem Mol Genet 2006;15:2962–71. the maintenance of regulated PKA activity. Ann N Y 2000;275:1449–56. 18. Liu J, Matyakhina L, Han Z, et al. Molecular cloning, Acad Sci 2002;968:75–95. 40. Reinitz CA, Bianco RA, Shabb JB. Compartmentation chromosomal localization and studies in PKA regulatory 30. Mavrakis M, Lippincott-Schwartz J, Stratakis CA, of the type I regulatory subunit of cAMP-dependent subunit type 1A (PRKAR1A)-mutant cells and tissues of Bossis I. mTOR kinase and the regulatory subunit of protein kinase in cardiac ventricular muscle. Arch human peripheral-type benzodiazepine receptor- and protein kinase A (PRKAR1A) spatially and functionally Biochem Biophys 1997;348:391–402. PRKAR1A-associated protein PAP7. FASEB J 2003;17: interact during autophagosome maturation. Autophagy 41. Linask KK, Greene RM. Subcellular compartmen- 1189–91. 2007;3:151–3. talization of cAMP-dependent protein kinase regula- 19. Taske´nK, Ska˚lhegg BS, Taske´nKA, et al. Structure, 31. Martin BR, Deerinck TJ, Ellisman MH, Taylor SS, tory subunits during palate ontogeny. Life Sci 1989;45: function, and regulation of human cAMP-dependent Tsien RY. Isoform-specific PKA dynamics revealed 1863–8. protein kinases. Adv Second Messenger Phosphoprotein by dye-triggered aggregation and DAKAP1a-mediat- 42. Constantinescu A, Diamond I, Gordon AS. Ethanol- Res 1997;31:191–204. ed localization in living cells. Chem Biol 2007;14: induced translocation of cAMP-dependent protein 20. Barradeau S, Imaizumi-Scherrer T, Weiss MC, Faust 1031–42. kinase to the nucleus. Mechanism and functional DM. Intracellular targeting of the type-Ia regulatory 32. Mucignat-Caretta C, Caretta A. Clustered distribution consequences. J Biol Chem 1999;274:26985–91. subunit of cAMP-dependent protein kinase. Trends of cAMP-dependent protein kinase regulatory isoform 43. Harootunian AT, Adams SR, Wen W, Meinkoth JL, Cardiovasc Med 2002;6:235–41. RIa during the development of the rat brain. J Comp Taylor SS, Tsien RY. Movement of the free catalytic 21. Mavrakis M, Richa R, Mary L, Lippincott-Schwartz J. Neurol 2002;451:324–33. subunit of cAMP-dependent protein kinase into and out Fluorescence imaging techniques for studying Drosoph- 33. Duncan FE, Moss SB, Williams CJ. Knockdown of the of the nucleus can be explained by diffusion. Mol Biol ila development. Curr Protcol Cell Biol. In press 2008. cAMP-dependent protein kinase (PKA) type Ia regula- Cell 1993;10:993–1002. 22. Kim C, Xuong NH, Taylor SS. Crystal structure of a tory subunit in mouse oocytes disrupts meiotic arrest 44. Yang S, Fletcher WH, Johnson DA. Regulation of complex between the catalytic and regulatory (RIa) and results in meiotic spindle defects. Dev Dyn 2006;235: cAMP-dependent protein kinase: enzyme activation subunits of PKA. Science 2005;307:690–6. 2961–8. without dissociation. Biochemistry 1995;34:6267–71.

www.aacrjournals.org 3141 Cancer Res 2008; 68: (9). May 1, 2008

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research. Protein Kinase A Effects of an Expressed PRKAR1A Mutation Associated with Aggressive Tumors

Elise Meoli, Ioannis Bossis, Laure Cazabat, et al.

Cancer Res 2008;68:3133-3141.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/68/9/3133

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2008/04/29/68.9.3133.DC1

Cited articles This article cites 42 articles, 11 of which you can access for free at: http://cancerres.aacrjournals.org/content/68/9/3133.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/68/9/3133.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/68/9/3133. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2008 American Association for Cancer Research.