Essential Role of A-kinase Anchor Protein 121 for cAMP Signaling to Mitochondria*

A-Kinase anchor proteins (AKAPs) immobilize and concentrate protein kinase A (PKA) isoforms at specific subcellular compartments. Intracellular targeting of PKA holoenzyme elicits rapid and efficient phosphorylation of target proteins, thereby increasing sensitivity of downstream effectors to cAMP action. AKAP121 targets PKA to the cytoplasmic surface of mitochondria. Here we show that conditional expression of AKAP121 in PC12 cells selectively enhances cAMP·PKA signaling to mitochondria. AKAP121 induction stimulates PKA-dependent phosphorylation of the proapoptotic protein BAD at Ser155, inhibits release of cytochrome c from mitochondria, and protects cells from apoptosis. An AKAP121 derivative mutant that localizes on mitochondria but does not bind PKA down-regulates PKA signaling to the mitochondria and promotes apoptosis. These findings indicate that PKA anchored by AKAP121 transduces cAMP signals to the mitochondria, and it may play an important role in mitochondrial physiology.

Binding of extracellular ligand to G-protein-coupled receptors at the cell membrane activates adenylate cyclase and increases cAMP levels at discrete points along the membrane. cAMP binds the regulatory (R) subunits of protein kinase A (PKA), 1 dissociating the holoenzyme and releasing free catalytic subunit (C-PKA). Phosphorylation of nuclear and cytoplasmic substrates by PKA controls multiple cell functions (1)(2)(3)(4)(5)(6)(7).
S-AKAP84 and AKAP121 derive from alternatively spliced products of the same gene. They are expressed in the male germ cell lineage as well as in several other tissues (27)(28)(29)(30)(31). Hormones that activate the cAMP⅐PKA pathway induce accumulation of S-AKAP84/AKAP121 transcripts and protein (28). This suggests a positive feedback loop between membranegenerated signals and downstream effector molecules of cAMP⅐PKA signals. All splice variants share the same 525amino acid NH 2 -terminal core, which includes the anchoring domain and the R-binding domain but diverge significantly at the COOH terminus. The first 30 NH 2 -terminal residues mediate the targeting of S-AKAP84/AKAP121 to the outer membrane of mitochondria, both in male germ cells and in transfected heterologous cells (27,30). However, other localization sites have been observed. D-AKAP1, an alternative splice product of S-AKAP84 carrying additional NH 2 -terminal residues, colocalizes with both mitochondria and endoplasmic reticulum (32,33). Furthermore, S-AKAP84/AKAP121 also interacts with microtubules and associates with mitochondria in interphase and mitotic spindles during metaphase transition (34). This suggests that the same anchor protein might focus cAMP⅐PKA signaling to distinct subcellular compartments in a cell cycledependent manner.
Its location on the outer surface of mitochondria implies a role for AKAP121 in cAMP-mediated reactions at or within mitochondria. Mitochondria are the seat of a number of major cellular functions, including essential pathways of intermediate metabolism, amino acid biosynthesis, fatty acid oxidation, steroid metabolism, apoptosis, and oxidative energy metabolism (35). Several of these functions are constitutive, whereas others are finely regulated at the post-translational level. BAD is a BH3-proapoptotic Bcl-2 family member that acts at a key nodal point in the mitochondrial apoptotic pathway. Unphosphorylated BAD binds and inactivates antiapoptotic Bcl-2 homologs. This allows release of cytochrome c from mitochondria and consequent activation of the apoptotic pathway (36 -41). Phosphorylation by PKA blocks BAD association with BCL-2 and inhibits apoptosis (42)(43)(44)(45). Previous observations suggested a role of mitochondria-anchored PKA in the inhibition of apoptosis (46). Treatment with a synthetic peptide spanning the RII-binding domain of thyroid AKAP (Ht-31) decreased BAD phosphorylation at Ser 112 and increased apoptosis of FL12.5 cells. However, it was not clear whether BAD phosphorylation necessitates PKA anchored to mitochondria by AKAP121 or whether it can be carried out by PKA anchored in other membrane compartments (46). To further analyze this mechanism and investigate the role of AKAP121 in the cAMP signaling to the mitochondria, we have established a condi-tional PC12 cell line in which the expression of AKAP121 is regulated. AKAP121 induction stimulates BAD phosphorylation at Ser 155 and increases cell survival. Conversely, expression of an AKAP121 mutant that binds to mitochondria but does not anchor PKA activates the mitochondrial caspase pathway and provokes apoptosis.

MATERIALS AND METHODS
Antibodies-Anti-caspase-3 was purchased from Pharmingen; anticaspase-9, anti-BAD, and anti-phospho-specific BAD antibodies were purchased from Cell Signaling; anti-phospho-ERK, anti-ERK, anti-p21/ Waf, anti-RII, and anti-cytochrome c antibodies were purchased from Santa Cruz Biotechnology, Inc.; and anti-manganese superoxide dismutase were purchased from Bender MedSystems. Anti-RII and anti-AKAP121 polyclonal antibodies were purchased from Santa Cruz Biotechnology, Inc.; CREB and anti-phospho 136 CREB antibodies were purchased from Upstate Biotechnology Inc.
Apoptosis and Fluorescent-activated Cell Sorter Analysis (FACS)-Apoptosis was analyzed by double staining with propidium iodide and annexin (Apoptosis detection kit, Medical and Biological Laboratories). Briefly, cells were harvested at indicated times after treatment, washed twice with 1ϫ PBS, and incubated for 10 min with propidium iodide (50 ng/ml in 1ϫ PBS, Sigma) and annexin. After three washes with PBS, the cells were analyzed by fluorescence microscopy using an Axiovert microscope IX70 (Olympus) or by FACS analysis. For microscopy, apoptosis was quantified by scoring the percentage of cells stained with propidium iodide (red) and annexin (green) in the adherent cell population. To avoid unbiased counting, plates were coded, and the cells scored blind without knowledge of the treatment performed. Four to six independent experiments made in triplicate were performed for each treatment. For FACS analysis, cells were harvested in 1ϫ PBS containing trypsin and 20 mM EDTA. 3 ϫ 10 6 cells were resuspended in PBS, fixed with cold 100% ethanol, and treated with RNase-DNase-free enzyme (50 g/ml). Cells were stained with propidium iodide (50 g/ml) and annexin in a dark room for 20 min and analyzed by flow cytometry using a BD Biosciences FACScan apparatus.

RESULTS
cAMP⅐PKA Signaling Suppresses Apoptosis Induced by Serum Deprivation-To analyze the role of PKA in cell survival, we investigated the effects of a cAMP analogue in neuronal cells deprived of growth factors. In human neuroblastoma (SK-N-BE) cells, prolonged serum starvation triggers the apoptotic program that includes stress-activated protein kinases and caspases (48,49). Cells were grown to subconfluence, starved in medium containing 0.1% serum, and harvested at different times. The percentage of apoptotic cells was determined by fluorescent microscopy (see "Materials and Methods"). As shown in Fig. 1A, serum deprivation induced a time-dependent increase in cell death. As a second monitor of apoptosis, we followed the accumulation of p17, the cleaved active caspase-3 fragment. The kinetics of caspase-3 activity roughly coincided with cellular apoptosis (Fig. 1B). Activation of the cAMP⅐PKA signal transduction pathway can suppress or promote apoptosis, depending on cell type (50 -55). As shown in Fig. 1, C and D, treatment with CPT-cAMP, a potent cAMP analogue, increased survival of serum-deprived SK-N-BE cells and inhibited caspase-3 activation. The protective effects of PKA activa-tion toward trophic factor withdrawal were also demonstrated in the PC12 pheochromocytoma cell line (Fig. 1, E and F).
Conditional Expression of AKAP121 Enhances PKA Targeting to Mitochondria-We wished to determine the role of AKAP analysis of total proteins extracted from control (C) or PC12 cells stably transfected with AKAP121 vector (clones number 3 and 11). Doxycyclin was removed from the medium for 48 h. A proteolytic RII-binding fragment of AKAP121 of ϳ80 Kda was sometimes present in the extracts. C, double immunofluorescence of PC-A121 Ϫ dox cells using specific antibody directed against AKAP121 (a) and manganese superoxide dismutase (b). A field containing several cells is shown. The punctate pattern of AKAP121 colocalizes with mitochondria, as shown by the merge of both signals (c). PC-A121 ϩ dox cells were stained with anti-AKAP121 and used as control (d). A higher magnification of a single PC-A121 Ϫ dox cell is also shown (inset). Bars, 5 m. As shown in D, protein extracts from total lysate (clone number 11) or from purified mitochondria (mito) (clones number 11 and 3) were immunoblotted with anti-RII or anti-manganese superoxide dismutase antibodies. Lower panels represent densitometric analyses. The data are relative to values from control cells set as 2 for mitochondrial fractions (clone number 3 ϩ dox) and set as 50 for total lysate (clone number 11 ϩ dox) and represent the mean Ϯ S.E. of three independent experiments that gave similar results. MnSOD, manganese superoxide dismutase. in transmitting cAMP signals to mitochondria. Accordingly, we used the tet-off inducible system to regulate expression of AKAP121, an AKAP that binds and targets PKA to the cytoplasmic surface of mitochondria ( Fig. 2A). Doxycyclin downregulates transcription of a target gene in this system by inactivating the tetA transcription factor. We isolated several PC12 stable transfectants (PC-A121) that expressed AKAP121 when cultured in the absence of doxycycline for 48 h, as shown by RII overlay (Fig. 2B, left panel). Doxycyclin down-regulated AKAP121 accumulation in PC-A121 clones number 3 and 11 to the levels of PC12 controls, as demonstrated by immunoblot analysis with anti-AKAP121 antibody (Fig. 2B, right panel). Expression of the transgene was reversible since readdition of doxycyclin to the medium reduced AKAP121 concentrations (data not shown). AKAP121 expressed in PC-A121 localized principally on mitochondria, as shown by double labeling with antibody to manganese superoxide dismutase, a protein that selectively accumulates in mitochondria (Fig. 2C) (56). AKAP121 accumulation coincided with increased targeting of RII subunit to mitochondria, as demonstrated by immunoblot analysis of proteins extracted from partially purified mitochondria (Fig. 2D). The total cellular content of RII was unaffected by AKAP121 expression (Fig. 2D).
Expression of AKAP121 Protects PC12 Cells Against Apoptosis-We then assessed the biological effects of AKAP121 accumulation and increased association of PKA with mitochondria. First, we determined the growth rates of PC12 and PC-A121 cells in the presence and absence of doxycylin. As shown in Fig. 3A, cells expressing AKAP121 increased more rapidly than controls over a 96-h period. AKAP121 expression did not accelerate the cell cycle since FACS analysis and [ 3 H]thymidine incorporation showed no significant difference in the number of cells in G 1 , S, or G 2 /M when compared with controls (data not shown). We thus considered the possibility that AKAP121 might enhance cell viability. We measured the effects of AKAP121 on sensitivity to trophic factor deprivation. Cells were grown to semiconfluence Ϯ doxycyclin and then serumstarved and harvested at different times. Fig. 3B shows that control cultures (PC12 Ϯ dox and PC-A121 ϩ dox) became apoptotic more rapidly than the experimental culture (PC-A121 Ϫ dox). We conclude that AKAP121 expression protects cells against apoptosis induced by serum deprivation.
To demonstrate that AKAP121 acted through PKA, these experiments were repeated in the presence of the PKA inhibitor, H89. As shown in Fig. 3C, H89 abrogated the enhancement of survival by AKAP121 in serum-deprived cells. Parallel experiments were performed using hydrogen peroxide (H 2 O 2 ) as a proapoptotic stimulus. About 30% of PC12 and PC-A121 ϩ dox cells treated with H 2 O 2 (200 M) became apoptotic 1 day after treatment. In contrast, 20% of cells expressing AKAP121 (PC-A121Ϫ dox) showed H 2 O 2 -induced apoptosis (Fig. 3D). The release of cytochrome c from mitochondria is a critical step in the activation of downstream effectors of the apoptotic pathway. The binding of released cytochrome c to Apaf-1 induces the formation of the "apoptosome" complex and the sequential activation of the caspase cascade (38). As shown in Fig. 3, E and F, deprivation of trophic factors in PC-A121 ϩ dox cells induced a time-dependent translocation of cytochrome c from mitochondria to cytosol. Expression of AKAP121 (PC-A121 Ϫ dox) delayed cytochrome c release (Fig. 3, see 6 and 24 h). Under these conditions, activation of pro-caspase-3 was partly inhibited by AKAP121 expression (data not shown).

AKAP121 Selectively Increases PKA-dependent Phosphorylation of Endogenous BAD at Ser 155 -
These data indicate that the AKAP121⅐PKA pathway mediates, at least in part, the protective effects of cAMP on cell survival. Based on previous reports, we thought it likely that the downstream effector of PKA was likely to be BAD. BAD, a proapoptotic protein, binds and inactivates Bcl, an antiapoptotic protein located on the outer mitochondrial membrane. Phosphorylation by PKA at Ser 155 blocks association of BAD with Bcl (45). To determine the phosphorylation pattern of BAD, total proteins isolated from control or cAMP-treated cells were size-fractionated by denaturing gel electrophoresis and immunoblotted using specific antibodies to BAD phosphorylated at Ser 112 , Ser 136 , or Ser 155 . As shown in Fig. 4A, growing PC-A121 cells contain significant amounts of BAD phosphorylated at one or more of these three sites. After 6 h of serum deprivation, little or no phosphorylated BAD could be detected. Addition of CPT-cAMP 24 h after serum deprivation increased phosphorylation of BAD at Ser 155 . Phosphorylation could be detected at 15 min and was more extensive at 30 min. Expression of AKAP121 (PC-AKAP121 Ϫ dox) stimulated basal and cAMP-induced phosphorylation of BAD Ser 155 at both time points. Under these conditions, phosphorylation of BAD at Ser 136 and Ser 112 was undetectable, even after a 60-min exposure of cells to CPT-cAMP (Fig. 4, A and B, and data not shown). We next asked whether expression of AKAP121 modulates BAD phosphorylation induced by other signaling pathways. Using the phospho-BAD specific antibodies described above, we performed immunoblot analysis on total proteins from serumdeprived PC-A121 cells stimulated with neurotrophin (NGF), phorbol ester (TPA), or CPT-cAMP (Fig. 4, C-E). NGF stimulated Ser 155 and Ser 112 phosphorylation 2-fold in both control cells (ϩ dox) and in cells that expressed AKAP121 (Ϫ dox). TPA enhanced Ser 112 phosphorylation but had no effect on Ser 155 . As shown above, CTP-cAMP induced phosphorylation of Ser 155 but not Ser 112 , and this was enhanced by AKAP121 expression. AKAP121 enhancement of BAD Ser 155 phosphorylation was detectable even at very low concentrations of CPT-cAMP (50 M). As expected, CPT-cAMP stimulation of BAD phosphorylation was sensitive to the PKA inhibitor, H89. Ser 136 phosphorylation was unaffected by any of the stimuli applied (data not shown).
AKAP121 specifically stimulated phosphorylation of mitochondrial PKA substrates. Fig. 5A shows that the transient increase in phospho-CREB in cells treated with CTP-cAMP was unaffected by AKAP121 induction. Similarly, CREB-directed expression of a CRE-CAT fusion was not enhanced by AKAP121 induction during CPT-cAMP exposure (Fig. 5B), nor did AKAP121 affect NGF-dependent phosphorylation of ERK and activation of the MAPK signaling pathway (Fig. 5, C and D) (55,57). Furthermore, stimulation of the MAPK signaling pathway by cAMP, as shown by ERK phosphorylation, was likewise independent of AKAP121 (Fig. 5E). These data indicate that expression of AKAP121 selectively up-regulates PKA signaling to the mitochondria without affecting the rate or magnitude of PKA-dependent or MAPK-dependent signaling to the nucleus.
Anchoring of PKA on Mitochondria Is Critical for Survival-The experiments described above suggest that localization of PKA on mitochondria mediated by AKAP121 plays a critical role for PKA-dependent inhibition of apoptosis. To further support this notion, we generated a mutant carrying L313P and L319P substitutions within the R-binding domain of AKAP121 (AKAP121m). These mutations disrupt an amphipathic helix that is required for AKAP⅐PKA interaction in vitro and in vivo (58). The mutant transgene was subcloned into a eucaryotic expression vector under the control of the TetR promoter and stably transfected in PC12-tet-off cells as described above for wild-type AKAP121. Expression of AKAP121m was induced by CAT activity is expressed as relative units and represents a mean Ϯ S.E. of three independent experiments. C, immunoblot analysis of total proteins extracted from serum-starved or NGF-stimulated PC-A121 Ϯ dox cells by using anti-phospho-ERK or anti-ERK antibody. P-ERK, phospho-ERK. As shown in D, PC-A121 Ϯ dox cells were transiently co-transfected with pBD-Elk1 and Gal-CAT cDNA vectors, serum-deprived overnight, and stimulated with NGF (100 ng/ml). CAT activity is expressed as relative units and represents a mean Ϯ S.E. of three independent experiments. E, immunoblot analysis of total proteins extracted from serum-starved or CPT-cAMP-stimulated PC-A121 Ϯ dox cells by using anti-phospho-ERK or anti-ERK antibody.
growing the cells for 48 h in the absence of doxycyclin. Total cellular proteins were then extracted and assayed for AKAP121m by immunoblot and RII overlay analyses. As shown in Fig. 6A, wild-type and mutant AKAP121 accumulate to comparable levels after doxycyclin removal. As predicted, the affinity of AKAP121m for RII was significantly lower than wild-type AKAP121. The mutant protein remains associated with mitochondria (Fig. 6B). Expression of the mutant protein provokes the movement of PKA from mitochondria to the cytosol without significantly altering the total concentration of the kinase (Fig. 6, C and D).
The phenotype of PC12 cells that express AKAP121m is shown in Fig. 7. Expression of AKAP121m correlates with increased apoptosis, as shown by reduced cell viability and activation of mitochondrial pro-caspase 9, and the extent of apoptosis is directly related to the amount of AKAP121m expressed (Fig. 7A). The proapoptotic effects of AKAP121m are evident both in growing cells (Fig. 7B) and in cells deprived of serum (Fig. 7, C-E). Furthermore, AKAP121m impedes cAMPdependent phosphorylation of endogenous BAD at Ser 155 (Fig.  7F). The data indicate that AKAP121m protein acts in a dominant-negative fashion by displacing AKAP121/PKA from mitochondria and down-regulating cAMP signaling to these organelles.

DISCUSSION
PC12 cells deprived of serum and trophic factors undergo apoptosis (48,49). Activation of cAMP⅐PKA signaling prevents apoptosis and induces their differentiation toward neuronal cells (50 -52, 59 -61). We have sought to identify the PKA targets involved in this response. Note that PKA phosphorylates and modulates a great variety of cellular substrates local-ized in distinct cellular compartments. We report here that reversible phosphorylation of PKA substrates on or within mitochondria inhibits apoptosis. This has been achieved by establishing a PC12 line (PC-A121) that conditionally expresses AKAP121, a scaffold protein that anchors PKA to the outer membrane of mitochondria (27)(28)(29). AKAP121 induction in PC-A121 cells stimulates translocation of PKA to mitochondria. When the induced cells are treated with cAMP, there is enhanced phosphorylation of BAD at Ser 155 . Phosphorylation of BAD correlates with inhibition of cytochrome c release from mitochondria and reduced apoptosis. Our system is uniquely suited to analyze the effects of PKA and cAMP on mitochondrial physiology and apoptosis: 1) expression of AKAP121 is efficient, reversible, and temporally modulated; 2) AKAP121 increases PKA targeting to mitochondria without affecting the total concentration of cellular PKA holoenzyme; 3) AKAP121 facilitates PKA-cAMP signaling to mitochondria without affecting cAMP signaling to the nucleus or activation of MAPK signaling by cAMP or NGF. Conversely, a mutant of AKAP121 that does not bind PKA but localizes on mitochondria (AKAP121m) acts as dominant-negative. Induction of AKAP121m activates mitochondrial caspase-9 and promotes apoptosis, even in the presence of trophic factors. The mutant protein displaces endogenous AKAP121⅐PKA complexes from mitochondria sites, thus impairing physiological flux of cAMP signals from cell membrane to organelles. In particular, cAMP-dependent phosphorylation of endogenous BAD at Ser 155 is down-regulated. A similar mechanism has been postulated for the ␤-adrenergic receptor where expression of an AKAP79 mutant, which does not bind PKA but still associates with the receptor, down-regulates PKA-dependent phosphorylation of the receptor (26). The use of such proline-FIG. 6. AKAP121m, which does not anchor PKA, displaces PKA from mitochondria. As shown in A, total proteins from control (C) or PC12 cells expressing AKAP121m (L313/319P) were immunoblotted (IB) with anti-AKAP121 antibody (left panel) or subjected to RII binding assay (right panel). B, double immunofluorescence of PC-A121m Ϫ dox cells using anti-AKAP121 (AKAP121m, green) and anti-superoxide dismutase antibodies (MnSOD, red). A merge (yellow) of both signals is also presented. Bar, 5 m. As shown in C, total lysates, purified mitochondria (mito), or cytosolic fractions from PC-A121 Ϯ dox cells were immunoblotted with anti-RII antibody, anti-manganese superoxide dismutase, or anti-ERK antibodies. D, densitometric analyses of the experiments indicated in panel C.
derivative AKAP mutants represents a novel and useful approach to dissect and selectively manipulate signaling pathways traveling from cell membrane to target organelles.
Biochemical and genetic studies indicate that Ser 155 of BAD is the PKA high affinity site (43)(44)(45). However, most of these studies were performed supplying BAD as a substrate. BAD phosphorylation was measured with recombinant protein in vitro or expressing exogenous BAD in vivo. In this work, we have explored site-specific phosphorylation of endogenous BAD following activation of distinct signaling pathways. We found that PKA specifically phosphorylates BAD at Ser 155 in intact cells. This effect is potentiated by AKAP121 and inhibited by AKAP121m. Ser 112 , another potential phosphorylation site, was efficiently phosphorylated after activation of the MAPK or protein kinase C signaling pathways. No phosphorylation of Ser 136 was observed under any of our experimental conditions (43). Different signals converge to inactivate BAD through phosphorylation at various serine resides (42)(43)(44)(45). The specificity of the responding serines and the extent to which they are modified may be a critical element to discriminate the pathway activated and the intensity of the signal. In this respect, BAD may be similar to other key signaling molecules where many pathways converge, such as the cyclin-dependent kinase (CDK) inhibitor p27 (62).
Our studies demonstrate that AKAP121⅐PKA complexes play a unique role in mediating cAMP signaling to mitochondria.
The cAMP pathway influences mitochondrial physiology at multiple points, and AKAP121 appears to be an important multifaceted mediator of these effects. For example, we recently found that AKAP121 binds the 3Ј-untranslated region of mRNA encoding mitochondrial proteins and that this interaction is stimulated by PKA phosphorylation of AKAP121. 2 Thus AKAP121 assembles protein kinases, mRNA, and possibly protein phosphatases on the mitochondrial surface in proximity to heterogeneous PKA substrates and other macromolecules critical for mitochondrial function(s). FIG. 7. AKAP121m down-regulates cAMP signaling to mitochondria and promotes apoptosis. A, immunoblot analysis on total extracts from PC-A121m cells grown for 48 h in the absence (Ϫdox) or presence of the indicated amount of doxycyclin (0.5 ng/ml or 10 ng/ml). B, growth curve of PC12 cells expressing wild-type (PC-A121) or mutant AKAP121 (PC-A121m). As shown in C, viable cells following serum deprivation (S.D.) were harvested and counted. Cumulative data are expressed as the mean Ϯ S.E. of 4 -6 independent experiments made in duplicate. D, immunoblot analysis for caspase-9 of total extracts from serum-deprived PC-A121m Ϯ dox cells. Asterisks indicate processed, activated caspase-9. E, immunoblot analysis for caspase-9 of total extract from serum-deprived PC-A121m cells. F, BAD phosphorylation in AKAP121m expressing cells treated with CPT-cAMP (200 M/30 min). The densitometric analysis is expressed as the mean of two independent experiments that gave similar results. P-BAD, phospho-BAD.