Characterization of Structural Features That Mediate the Tethering of Caenorhabditis elegans Protein Kinase A to a Novel A Kinase Anchor Protein

Caenorhabditis elegans protein kinase A (PKAICE) is tethered to organelles in vivo. A unique A kinase anchor protein (AKAPCE) avidly binds the RI-like regulatory subunits (RCE) of PKAICE and stringently discriminates against RIIα and RIIβ subunits, the preferred ligands for classical AKAPs. We elucidated structural features that stabilize AKAPCE·RCE complexes and confer atypical R isoform specificity on the anchor protein. Three large aliphatic amino acids (Leu236, Ile248, and Leu252) in the tethering domain of AKAPCE(residues 236–255) are crucial for ligation of RCE. Their side chains apparently generate a precisely configured hydrophobic binding pocket that accommodates an apolar surface on RCEdimers. Basic residues (His254-Arg255-Lys256) at the C terminus of the tethering site set an upper limit on affinity for RCE. A central dipeptide (Phe243-Ser244) contributes critical and distinctive properties of the tethering site. Ser244 is essential for selective binding of RCE and exclusion of RII isoforms. The aromatic hydrophobic character of Phe243ensures maximal RCE binding activity, thereby supporting a “gatekeeper” function of Ser244. Substitution of Phe243-Ser244 with Leu-Val generated an RII-specific AKAP. RCE and RII subunits contain similar dimerization domains. AKAP-binding domains of RCE (residues 23–47) and RII differ markedly in size, amino acid sequence, and docking specificity. Four hydrophobic residues (Cys23, Val27, Ile32, and Cys44) in RCE are crucial for avid binding with AKAPCE, whereas side chains from Leu20, Leu35, Val36, Ile40, and Ile41 have little impact on complex formation. Tyr26 is embedded in the docking domain, but its aromatic ring is required for RCE-RCE dimerization. Residues 236–255 in AKAPCE also constitute a binding site for mammalian RIα. RIα (PKAIα) is tightly sequestered by AKAPCE in vitro (K D = ∼10 nm) and in the environment of intact cells. The tethering domain of AKAPCEprovides a molecular module for manipulating intracellular localization of RI and elucidating functions of anchored PKAI in eukaryotes.

Cyclic AMP-dependent protein kinases (PKAs) 1 mediate many actions of hormones and neurotransmitters that stimulate adenylate cyclase (1)(2)(3)(4). Signals disseminated by cAMP often control activities of proteins that accumulate at discrete intracellular sites (e.g. ion channel proteins in plasma membrane) (5)(6)(7)(8). A kinase anchor proteins (AKAPs) guide the transmission and targeting of cAMP signals to effector proteins in such microenvironments. Classical AKAPs have a tethering site that avidly binds regulatory subunits (RII) of PKAII isoforms 2 and distinct targeting domains that ligate AKAP⅐PKAII complexes to docking sites in organelles or cytoskeleton (5,6,9,10). Clustering of a high concentration of AKAP⅐PKAII complexes in proximity with PKA substrate/effector proteins in cytoskeleton/organelles enables efficient reception, rapid amplification, and precisely focused targeting of cAMP signals. Consequently, PKA-catalyzed phosphorylation of co-localized effector proteins is optimized (11)(12)(13). Key tenets of the preceding signaling model have been verified. Disruption of AKAP⅐PKA complexes in situ markedly diminishes the ability of cAMP to regulate such critical physiological processes as ion transport, gene transcription, apoptosis, and hormone secretion (8,(13)(14)(15)(16).
Mammals employ Ͼ20 distinct AKAPs to adapt type II PKAs for specialized functions (5)(6)(7)(8)17). Amino acid sequences of the anchor proteins are markedly divergent; thus, AKAPs are functional, not structural, homologs. Different AKAPs accumulate in distinct locations; examples include cytoplasmic surfaces of plasma membrane, mitochondria, Golgi membranes, and centrosomes (5,6,9,10,18). Routing of PKAII to specific microenvironments is governed by unique targeting/anchoring domains in individual AKAPs (9, 19 -22). Primary structures of RII-binding sites in AKAPs are not highly conserved (e.g. see alignments in Refs. 23 and 24). However, the folding pattern and physical properties of amino acids that subserve the ligation of RII subunits are universal properties of AKAPs. The RII-binding region of typical AKAPs comprises 16 -20 contiguous amino acids that have a predicted propensity to fold into an amphipathic ␣-helix (25). The ␣-helix includes six critical, precisely positioned amino acids whose large aliphatic side chains co-operatively create an extended hydrophobic surface (19). The size, hydrophobicity, and configuration of this domain generate a unique receptor site for a complementary, apolar docking surface that is produced from the folding of ϳ30 amino acid residues in RII␣ or RII␤ dimers (26 -29). Robust hydrophobic interactions between the tethering and docking surfaces drive the formation of stable AKAP⅐PKAII complexes.
The development of a comprehensive data base of information on structure, function, mechanism of assembly, and physiological significance of AKAP⅐PKAII complexes contrasts sharply with our lack of knowledge concerning intracellular targeting and tethering of RI subunits and potential functions for anchored PKAI. RI␣ is expressed in nearly all mammalian tissues, where it mediates (as PKAI␣) regulatory effects of many hormones (1)(2)(3)(4). Immunochemical staining of intact cells and biochemical assays performed on fractions derived from cell or tissue homogenates indicate that a large proportion of RI␣ (PKAI␣) is dispersed in cytoplasm (5). However, several observations suggest that PKAI can also be adapted for specialized functions by binding with anchor proteins. Substantial amounts of RI␣ (PKAI␣) are associated with the plasma membrane of human erythrocytes (30); recruited to a multi-protein complex at the "cap" site of activated T lymphocytes (31); sequestered along the fibrous sheath of mammalian spermatozoa (32); and juxtaposed with the post-synaptic myocyte cell membrane at neuromuscular junctions (33). Activity of a voltagegated, L-type calcium channel in skeletal muscle transverse tubules is increased when co-localized, anchored PKAII is activated and phosphorylates a Ser residue near the C terminus of the ␣ 1C channel subunit (8). In knockout mice that lack RII␣ and RII␤ subunits in muscle, PKAI␣ binds with the channelassociated anchor protein and sustains normal, cAMP-controlled potentiation of Ca 2ϩ flux by providing a properly localized supply of PKA catalytic subunits (34). Rescue of calcium channel regulation by immobilized PKAI␣ is apparently explained by the discovery that high affinity RII-binding sites in certain AKAPs also ligate RI␣ (PKAI␣). However, these AKAPs bind RII␣ (PKAII␣) with 25-500-fold higher affinity than RI␣ (34,35). Only small amounts of PKAI␣ are likely to be anchored by this mechanism in normal physiological contexts because most cells/tissues have concentrations of RII␣ and/or RII␤ subunits that approach or exceed the level of RI (1)(2)(3)(4)36). Nevertheless, the cited studies suggest important principles: (a) RI dimers have a binding surface available for interactions with anchor proteins, (b) PKAI␣ can be anchored in intact cells, and (c) properly targeted AKAP⅐PKAI␣ complexes can be poised to phosphorylate a co-localized effector protein and thereby, selectively regulate a key, compartmentalized physiological process.
A complementary but potentially incisive approach to advancing our understanding of the molecular basis and regulatory significance of PKAI anchoring involves discovery and systematic characterization of RI-selective, prototype AKAPs. Ideally, AKAPs to be studied should engage RI (PKAI) with high affinity (K D in the 1-100 nM range) and efficiently discriminate against RII subunits (e.g. K A (RI):K A (RII) Ն 100) in vitro and in cells. Recently, we discovered the first eukaryotic tethering protein that fulfills the specified criteria (37). AKAP CE is a novel Caenorhabditis elegans protein composed of 1280 amino acids (M r ϭ 144,000). The tethering site of AKAP CE binds the RI␣-like subunits (R CE ) of C. elegans PKAI (PKAI CE ) with a K D of 7 nM (37). Moreover, high concentrations of RII␣ and RII␤ subunits do not competitively inhibit formation of R CE ⅐AKAP CE complexes in vitro or in transfected cells. The AKAP CE polypeptide includes a RING finger domain and other putative anchoring motifs 3 that may mediate the routing of tethered PKAI CE to several intracellular docking sites (37).
The catalytic subunit of PKAI CE is 82% identical with C␣, the predominant catalytic subunit of all mammalian PKA iso-forms (38). R CE is closely related to mammalian RI␣ (ϳ60% overall sequence identity) but not RII subunits (39). In C. elegans 60% of PKAI CE is tightly associated with organelles/ cytoskeleton (39) and AKAP CE is the principal R CE -binding protein. 3 Because pathways and mechanisms of signal transduction are highly conserved between C. elegans and mammals (40 -42), it is probable that tethering and targeting of R CE subunits by AKAP CE adapts and diversifies PKAI CE for multiple functions in various compartments of C. elegans cells (39).
The discovery of a high affinity, R CE -selective anchor protein (AKAP CE ) and availability of cDNAs, protein expression systems, and specific antibodies for AKAP CE and R CE (37)(38)(39) can be exploited to elucidate molecular mechanisms that govern the selective targeting/anchoring of PKAI. Several basic but fundamental questions merit immediate consideration: What structural features in AKAP CE are essential for avid ligation of R CE ? Conversely, which amino acid side chains in R CE create a binding surface that docks with the tethering site in AKAP CE ? Which residues in R CE and the anchor protein govern binding affinity? How are RII (PKAII) isoforms excluded from the tethering site of AKAP CE ? Does the tethering domain of AKAP CE exhibit plasticity? That is, can substitution of one or a few amino acids generate a binding region that selectively sequesters RII subunits? Can the wild type R CE tethering site bind mammalian RI␣ with high affinity? In this paper we address the central questions posed above and also report that the tethering module from AKAP CE may provide a molecular tool for manipulating the intracellular location of mammalian PKI␣.

EXPERIMENTAL PROCEDURES
cDNAs and Expression Vectors-Complementary cDNAs encoding AKAP CE , AKAP75, S-AKAP84, and human RII␤ were cloned and characterized as described in previous papers (9,19,26). A cDNA clone encoding murine RI␣ was generously provided by Dr. Robert Steinberg (Department of Biochemistry, University of Oklahoma). Full-length cDNAs for R CE and RII␤ were subcloned in the bacterial expression plasmid pET14b as described previously (26,37). An analogous pET14b construct was created for expression of RI␣ by following the strategy used for RII␤ cDNA (26). Phosphorylatable Ser residues were introduced into the pseudosubstrate sites of R CE and RI␣ via site-directed mutagenesis, as described previously (37). This mutation has no effect on the properties of R CE 3 or RI␣ (43). However, it enables efficient phosphorylation of R subunits by incubation with Mg[␥-32 P]ATP and the catalytic subunit of PKA, as previously reported (37). Cloning of cDNAs in pET14b enabled IPTG-inducible high level expression of soluble R subunits that contained an N-terminal, 20-residue fusion peptide. Included in the fusion peptide are six consecutive His residues (His tag) that enabled purification of recombinant R CE , RI␣, and RII␤ proteins to near homogeneity, as described by Li and Rubin (26). A fragment of AKAP CE cDNA that encodes amino acids 176 -409 in the anchor protein was cloned into the expression plasmid pGEX-KG as described previously (37). This enabled synthesis of a GST-AKAP CE fusion protein in Escherichia coli DH5␣ that was transformed with recombinant plasmid and induced with IPTG. After induction, bacteria were disrupted in a French press, and the soluble GST-AKAP CE (176 -409) protein was purified to near homogeneity by affinity chromatography on GSH-Sepharose 4B beads (Amersham Pharmacia Biotech) as previously reported (44). The R CE tethering domain of AKAP CE encompasses residues 236 -255 (37).
Full-length AKAP CE was cloned in the mammalian expression vector pCIS2 as described by Angelo and Rubin (37). Full-length RI␣ cDNA was excised from a recombinant pBluescript SK plasmid by digestion with NotI and ApaI and ligated into the Rc/CMV mammalian expression vector (Invitrogen), which was cleaved with the same enzymes. This placed the cDNA downstream from a strong cytomegalovirus promoter and upstream from a polyadenylation signal. RI␣ cDNA was also amplified by the polymerase chain reaction, using previously described conditions (44). The 5Ј primer appended a BamHI restriction site to the amplified cDNA; the 3Ј primer contained an SpeI cleavage site. Amplified DNA was digested with BamHI and SpeI and then cloned in the mammalian expression vector pEBG (45), which was cleaved with the same enzymes. (The pEBG plasmid was generously provided by Dr. 3 R. A. Angelo and C. S. Rubin, unpublished observations. Joseph Avruch, Harvard Medical School.) This accomplished an inframe fusion of RI␣ cDNA with an upstream GST gene. The fidelity of polymerase chain reaction-mediated amplification of RI␣ cDNA was verified by DNA sequencing. Transcription of the chimeric gene is driven by the strong, constitutively active EF1␣ promoter; an optimal polyadenylation signal follows the inserted RI␣ cDNA. Mammalian cells transfected with this vector produce a dimeric GST-RI␣ fusion protein that is avidly and efficiently complexed by GSH-Sepharose 4B beads.
Mutagenesis and Expression of AKAP CE -Deletion mutagenesis was performed via polymerase chain reaction as described for AKAP75 and S-AKAP84 (19,46). Amino acid substitutions were introduced into the R CE -binding tethering domain of AKAP CE via site-directed mutagenesis, as described previously (19). Amplified cDNAs and cDNAs containing various mutated codons were cloned into the expression plasmid pGEX-KG (47) after digestion of insert and vector with EcoRI and HindIII. This placed cDNAs encoding partial AKAP CE polypeptides downstream from and in-frame with the 3Ј terminus of a GST gene in the vector. All mutants were verified by DNA sequencing. Transcription of the GST fusion gene is driven by an inducible tac promoter. High levels of chimeric GST partial AKAP CE polypeptides were produced when E. coli DH5␣ was transformed with recombinant pGEX-KG plasmids and then induced with 0.5 mM IPTG. GST-AKAP CE fusion proteins were purified to near homogeneity by affinity chromatography on GSH-Sepharose 4B, as described previously (44) DNA Sequence Analysis-cDNA inserts were sequenced by a dideoxy-nucleotide chain termination procedure (48) as described previously (9).
Mutagenesis and Expression of R CE -Deletion and site-directed mutagenesis of R CE cDNA were performed as described for RII␤ and AKAP75 in previous studies (19,26). All mutants were confirmed by DNA sequencing. E. coli BL21(DE3) was transformed with recombinant pET14b plasmids containing mutant R CE cDNA inserts. His-tagged R CE fusion proteins were induced with IPTG, isolated from E. coli, and purified to near homogeneity by affinity chromatography on Ni 2ϩ -chelate Sepharose 4B resin as described by Li and Rubin (26). Wild type, His-tagged R CE , RI␣, and RII␤ were synthesized and purified by applying the same procedures.
Production of Antibodies Directed against AKAP CE and RI␣-Preparation, characterization, and affinity purification of rabbit antibodies (IgGs) directed against AKAP CE are described in a previous publication (44). His-tagged, murine RI␣ protein (see above) was injected into rabbits (initial injection, 0.4 mg; 0.25 mg for each of three booster injections) at Covance laboratories (Vienna, VA) to generate antisera. Serum was collected at 3-week intervals.
Protein Determination-Protein concentrations were determined by using the Coomassie Plus Protein Assay Reagent (Pierce) with bovine serum albumin as a standard.
Western Immunoblot Assays-Size-fractionated proteins were transferred from denaturing polyacrylamide gels to an Immobilon P membrane (Millipore Corp.) as described previously (49). Blots were blocked, incubated with affinity purified IgGs directed against AKAP CE (1:2000, relative to serum), and washed as described previously. (26,51). Antigen-IgG complexes were visualized by an indirect enhanced chemiluminescence procedure. Membranes were incubated with peroxidase coupled with goat immunoglobulins directed against rabbit IgG heavy and light chains (Amersham Pharmacia Biotech) in 20 mM Tris-HCl, pH 7.4, containing 0.15 M NaCl and 0.1% (w/v) Tween 20 (10 ml of solution/ gel lane) for 90 min at 22°C. The filters were then washed, incubated with luminol, and exposed to x-ray film for 2-30 s. Chemiluminescence signals are captured on x-ray film. The same procedures were used for the detection of RI␣ by anti-RI␣ IgGs.
Overlay Assay for R CE , RII␤, and RI␣ Binding Activities-Overlay binding assays have been described in several papers (49,50). In brief, a Western blot is probed with 32 P-labeled R CE , RII␤, or RI␣ (using a subunit concentration of 0.3-0.7 nM and 2-3 ϫ 10 5 cpm 32 P radioactivity/ml). Complexes of 32 P-labeled R subunits with partial or full-length AKAPs are visualized by autoradiography. Results were quantified by scanning laser densitometry (Amersham Pharmacia Biotech XL laser densitometer), PhosphorImager analysis (Molecular Dynamics), or scintillation counting, as described previously (26,37,51). R subunits were labeled with 32 P by incubation with Mg [␥-32 P]ATP and the catalytic subunit of PKA as described previously (49).
Solution Binding Assays-Detailed descriptions of (a) conditions used for achieving equilibrium binding of R subunits with a GST-AKAP CE fusion protein and (b) procedures employed for separating and quantifying AKAP⅐ 32 P-R complexes and free 32 P-R subunits are provided in a recent publication by Angelo and Rubin (37). Competition binding assays were executed under the same conditions by adding various amounts of nonradioactive competitor ligand (see Fig. 8B) to the reaction mixtures (37).
FPLC Gel Filtration-Wild type or mutant R CE protein (0.9 mg in 0.2 ml of 20 mM sodium phosphate buffer, pH 7.0, containing 0.2 M NaCl, 1 mM EDTA, and 1 mM dithiothreitol (buffer A)) was applied to a 1 ϫ 30 cm column of Superose 12 (Amersham Pharmacia Biotech) equilibrated with buffer A. The column was eluted at a flow rate of 0.2 ml/min, and R CE proteins were monitored by determining absorbance of the eluate at 280 nm. The column was calibrated with the following M r standards: thyroglobin (M r ϭ 670,000), aldolase (158,000), ovalbumin (44,000), myoglobin (17,000), and vitamin B 12 (1350).
Cell Culture, Transfections, and Isolation of AKAP CE ⅐RI␣ Complexes-Hamster AV12 cells (originally derived from a subcutaneous tumor) were obtained from the American Type Culture Collection. Cells were grown at 37°C in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. The incubator contained 8% CO 2 and 92% air. Plates (10 cm) of AV12 cells that reached 50% confluence were cotransfected with 1 g of recombinant pCIS2 vector (see above), which contains full-length AKAP CE cDNA and either 1 g of recombinant Rc/CMV vector (see above), which contains a full-length murine RI␣ transgene, or 1 g of recombinant pEBG vector, which includes the chimeric GST-RI␣ gene (see above). Vector DNA was mixed with the Effectine reagent (Qiagen), and transfections were performed according to the manufacturer's protocol. Cells were exposed to the Effectine-DNA mixture for 6 h. Medium was then replaced, and the cells were harvested 42 h later. Cells were collected in 0.6 ml of lysis buffer (20 mM Hepes-KOH, pH 7.4, 20 mM NaCl, 2 mM dithiothreitol, 5 mM EDTA, 1 mM EGTA, 10 g/ml leupeptin, 5 g/ml aprotinin, 5 g/ml pepstatin, 10 mg/ml benzamidine-HCl, and 0.5 mg/ml Pefablock) and were disrupted in a Dounce homogenizer with 10 strokes of a tight pestle. Samples were centrifuged at 40,000 ϫ g for 20 min, and the supernatant fraction was harvested. All operations were performed at 4°C. (Upon homogenization and disruption of cytoskeleton, a substantial amount of AKAP CE is recovered is cytosol.) Anti-RI␣ serum (2 l) was added to aliquots of cytosol (300 g of protein) derived from cells expressing the RI␣ transgene. Samples were then incubated for 12 h at 4°C. Subsequently, 40 l of a 50% suspension of protein A-Sepharose 4B beads was added, and the incubation was continued for 2 h. Beads containing IgG, RI␣, and associated proteins were recovered by centrifugation at 3,000 ϫ g for 1 min. Next, the beads were washed three times with 1 ml of lysis buffer by repeated resuspension and centrifugation. Finally, proteins bound to the beads were eluted by boiling in gel loading buffer and analyzed by denaturing electrophoresis and Western immunoblotting as described above. Aliquots of cytosol (300 g protein) from AV12 cells expressing GST-RI␣ and AKAP CE were supplemented with 50 l of a 50% suspension of GSH-Sepharose 4B beads and incubated (with constant rotation) for 5 h at 4°C. Subsequently, the beads were recovered and washed as described above for protein A-Sepharose 4B. Finally, proteins bound to GSH-Sepharose 4B were eluted and analyzed by denaturing electrophoresis and Western immunoblotting as described above.

The Tethering Site of AKAP CE Is a Mosaic Structure: R CE Sequestration Is Governed by a Combination of Conserved Aliphatic Hydrophobic Amino Acids and Novel Hydrophilic and
Aromatic Residues-Using deletion mutagenesis, we previously mapped the R CE -binding site to a segment of AKAP CE that encompasses amino acids 236 -255 (37). To guide studies on individual amino acids that are crucial for the ligation of R CE , the sequence of the tethering region of AKAP CE was aligned with a prototype mammalian RII-binding domain (from AKAP75) and the binding region of S-AKAP84, which avidly complexes RII isoforms and binds RI␣ with low affinity (Fig.  1A). The sequences of the three tethering domains are divergent. However, like classical RII-binding regions (25), residues 236 -255 in AKAP CE are predicted to fold into an ␣-helix that contains opposing hydrophobic and hydrophilic surfaces (Fig.  1B). The positions of Leu 236 , Ile 248 , and Leu 252 in AKAP CE are invariably occupied by either Leu, Ile, Val, or Thr in previously characterized AKAPs (9,19,23,52). Moreover, each large hydrophobic side chain of the corresponding amino acids in AKAP75 (Leu 393 , Ile 405 , and Ile 409 ) (Fig. 1A) plays a key role in mediating high affinity binding of PKAII isoforms (19). The functional significance of Leu 236 , Ile 248 , and Leu 252 (and other amino acids) in the AKAP CE tethering domain was evaluated by a combination of (a) site-directed mutagenesis, (b) AKAP CE fusion protein expression in E. coli, (c) purification of fusion proteins by affinity chromatography, and (d) biochemical assays for R CE binding activity and ligand specificity.
A cDNA fragment that encodes amino acids 176 -409 in AKAP CE was cloned in the plasmid pGEX-KG to generate a template for mutagenesis and recombinant protein production. In the recombinant plasmid, the cDNA insert is preceded by a GST gene and a tac promoter. This arrangement enables IPTGinducible, high level expression of wild type and mutant GSTpartial AKAP CE polypeptides in E. coli (37). Wild type AKAP CE fusion protein (designated GST-AKAP CE 176/409) includes the complete tethering region (residues 236 -255) and binds R CE with the same affinity and specificity as full-length AKAP CE (37). 3 Effects of mutations on tethering activity were assessed by monitoring formation of 32 P-R CE ⅐AKAP CE complexes in a well established, highly sensitive, overlay binding assay (49,50). Substitution of Ala for Leu 236 diminished the R CE binding activity of GST-AKAP CE 176/409 by ϳ90% (Fig. 2). Replacement of either Ile 248 or Leu 252 suppressed R CE binding activity to an even greater extent. Under standard assay conditions, mutant fusion proteins containing Ala 248 (Fig. 2) or Ala 252 (Fig.  3) in the tethering site bound only 1-2% of the amount of 32 P-R CE that was complexed by wild type GST-AKAP CE 176/ 409. The results suggest that the large, aliphatic side chains of Leu 236 , Ile 248 , and Leu 252 contribute to a precisely configured binding pocket that accommodates a target nonpolar region in R CE dimers. Replacement of any of these large hydrophobic amino acids with Ala (which stabilizes ␣-helices but possesses only a small nonpolar side chain) evidently alters the size and/or shape of the binding pocket so that it no longer provides a complementary surface for R CE ligation. In contrast to the preceding observations, replacement of Leu 246 , Val 247 , or both amino acids with Ala has no effect on formation of partial-AKAP CE ⅐R CE complexes ( Fig. 2 and Ref. 37). Thus, incorporation of Leu, Val, or Ile residues at nonconserved positions along the tethering site ␣-helix results in side chain orientations that have little or no impact on the R CE -binding pocket.
Further analysis of the alignment of AKAP CE with AKAP75 and S-AKAP84 (Fig. 1) revealed distinctive features of the C. elegans anchor protein. The last two residues in the AKAP CE tethering domain are hydrophilic and carry positive charges; hydrophobic or neutral amino acids occupy these positions in classical RII-binding AKAPs. Mutation of His 254 -Arg 255 -Lys 256 to the neutral tripeptide Gln-Ala-Ala caused a substantial (2-3-fold) increase in binding of R CE by AKAP CE (Fig. 3). Thus, the basic, C-terminal portion of the AKAP CE tethering site may negatively regulate affinity for R CE . An intrinsic limitation in R CE binding affinity could have significant physiological consequences. A substantial amount of PKAI CE accumulates in the cytoplasm of C. elegans cells (39). By analogy with cytoplasmic FIG. 1. Conserved and divergent features of the tethering sites of AKAP CE , AKAP75, and S-AKAP84. A, the R CE tethering site of AKAP CE is aligned with the RII-selective binding sites of prototypic mammalian anchor proteins, AKAP75 (19) and S-AKAP84 (46). Asterisks indicate positions of conserved hydrophobic residues; single dots mark positions of major differences between the R CE -and RII-selective binding sites. B presents the orientations of hydrophobic and hydrophilic amino acids in the R CE -and RII-binding sites as depicted in helical wheel diagrams. Residues that are included in an extended, predominantly hydrophobic surface are underlined.
PKAI␣ in mammals, the nonanchored pool of PKAI CE may regulate metabolism (by phosphorylating cytoplasmic enzymes) and also provide catalytic subunits that enter the nucleus and control transcription of cAMP-regulated genes in response to environmental stimuli. In contrast, numerous cAMP signals are evidently routed to distinct substrate/effector proteins in C. elegans organelles/cytoskeleton via co-localized AKAP CE ⅐PKAI CE complexes (see Introduction and Refs. 37 and 39). Lowering of the inherent R CE binding affinity of the tethering domain via the inhibitory effect of the C-terminal His-Arg-Lys tripeptide may promote an optimal distribution of RI CE (PKAI CE ) between cytoplasm and AKAP CE -docking sites in cytoskeleton and organelles. This arrangement would enable cells to coordinately control a large network of intracellular target sites for cAMP, thereby ensuring maximally integrated physiological responses to signals carried by the second messenger.
A prominent and unique feature of AKAP CE is evident in the central core of the tethering domain. Classical RII-selective AKAPs contain a hydrophobic dipeptide composed exclusively of Leu, Val, or Ile residues (Leu-Val in AKAP75; Ile-Ile in S-AKAP84; Fig. 1A) that is crucial for the generation and maintenance of the RII-binding site (19,52). The appearance of a Phe 243 -Ser 244 dipeptide in the corresponding region of the AKAP CE tethering domain represents a major deviation from the previously studied paradigm. Recently, we discovered that replacement of Phe 243 with Ala results in substantial suppression but not extinction of R CE binding activity (37). This raised the possibility that a side chain with aromatic hydrophobic character was required at this position in the tethering domain to ensure avid ligation RI␣-like R CE . Retention of R CE binding activity upon substitution of Phe 243 with Trp and the unaltered binding activity of a Tyr 237 to Ala (Fig. 3) mutant are consistent with this idea. However, replacement of Phe 243 with Val yielded an AKAP CE fusion protein that bound R CE with high (but not maximal) affinity (Fig. 2). Thus, a large hydrophobic side chain at the first position in the dipeptide is sufficient to ensure R CE tethering; binding affinity may be further optimized by aromatic character in the side chain. Ser 244 introduces several properties that cannot be supplied by Leu, Ile, or Val residues incorporated into the core of the AKAP75/S-AKAP84 RII-binding regions (Fig. 1A). The Ser hydroxyl group is hydrophilic and polar, readily participates in hydrogen bonding, and can potentially undergo regulatory phosphorylation. Mutation of Ser 244 to a neutral hydrophilic residue (Asn) or Asp (to mimic phosphorylation) completely extinguished R CE binding activity (Fig. 2). Thus, Ser 244 is a unique and critical component of the tethering domain of AKAP CE .
Characterization of Amino Acids in R CE That Govern Homodimerization and Affinity for AKAP CE -Structural features of the PKAI CE regulatory subunit that mediate R CE -R CE dimerization and high affinity binding with AKAP CE were elucidated by further application of mutagenesis, protein expression, and binding assays. Wild type and mutant R CE polypeptides were expressed as His-tagged fusion proteins in E. coli and were purified to near homogeneity by affinity chromatography on nickel-chelate Sepharose 4B. An approximate location for the AKAP CE -binding domain in R CE was established by truncation FIG. 2. Identification of amino acids in AKAP CE that are critical for high affinity binding of R CE . Amino acids in the tethering site of AKAP CE were altered by site-directed mutagenesis as described under "Experimental Procedures." Wild type and mutant GST fusion proteins (apparent M r ϭ 54,000), which contain residues 176 -409 from AKAP CE , were expressed in E. coli, purified by affinity chromatography, size-fractionated by denaturing electrophoresis, and transferred to an Immobilon P membrane (see "Experimental Procedures"). Equal amounts of purified protein (0.1 g) were applied to each lane of the denaturing gel. Abilities of the wild type and mutant AKAP CE tethering domains to ligate 32 P-labeled R CE were assessed by performing overlay binding assays ("Experimental Procedures") on the Western blots. Autoradiograms are shown. The experiment was repeated three times, and similar results were obtained in each replication. analysis. Neither deletion of 15 amino acids at the N terminus nor elimination of 277 residues (i.e. amino acids 100 -376) that constitute the entire central and C-terminal portions of the cAMP-binding protein altered the ability of partial R CE proteins to interact with AKAP CE (Fig. 4A, R CE ⌬15 and R CE 1-99). Further C-terminal deletion mutagenesis revealed that smaller fragments of R CE , which include amino acids 1-80 or 1-65, bind AKAP CE with approximately the same affinity (see competition binding assays in Fig. 8) as full-length R CE (376 amino acids, R CE wild type; Fig. 4A). Thus, the site in R CE subunits that docks with the tethering domain of AKAP CE is generated from amino acids that lie between residues 16 and 64 at the N terminus.
A distinct domain, which promotes R-R homodimerization, is included in the corresponding N-terminal segments (amino acids 1-65) of mammalian RII and RI isoforms (26 -29, 53). Moreover, mutations that prevent RII␣ or RII␤ dimerization have a crucial secondary consequence: binding of RII with classical AKAPs is abolished (26 -28). This is because two copies of a short, N-terminal segment of either RII␣ or RII␤ must align (in an anti-parallel fashion) to create a single site that binds with AKAP tethering domains (29). By analogy, these observations suggested that R CE mutants, which are incapable of binding AKAP CE , may yield two levels of structural information. One subset of mutants may implicate certain amino acids in R CE oligomerization and the generation of the heterotetrameric holo-PKAI CE ; a second group of mutants could reveal individual amino acids that control the folding and/or affinity of the docking site for AKAP CE in the context of an R CE dimer. To correctly analyze and interpret experimental results, it was essential to determine whether mutated R CE proteins were dimers or monomers. This was accomplished by employing methodology illustrated in Fig. 5. Like RI␣ (54), R CE contains two N-terminal Cys residues (Cys 23 and Cys 44 in R CE ) that engage in covalent, interchain cross-linking in the absence of reducing agents. 3 Determination of the M r of dimeric regulatory subunits by denaturing electrophoresis will yield a value of 100,000 in the absence of reducing agent and 50,000 in the presence of 1 M ␤-mercaptoethanol (Fig. 5, A and B). Monomeric R CE mutants will have a M r of 50,000 in both instances (Fig.  5B). In addition, analytical FPLC gel filtration on a column of Superose 12 readily resolved the peaks of R CE dimers and monomers (Fig. 5C). Using these approaches, all R CE mutants were characterized as dimers or monomers. Results are included in Fig. 4.
An extensive series of R CE variants was prepared by sitedirected, scanning mutagenesis. Representative results that provide insights into the molecular basis for the docking of dimeric R CE with the tethering domain of AKAP CE are presented in Fig. 4 (B and C). (Numbering, alignments, and predicted secondary structure of the N-terminal regions of R CE and mammalian RI␣ and RII␣ are given in Fig. 6). Replacement of the large, branched aliphatic side chain contributed by Ile 32 with the methyl group of Ala abrogates binding of R CE with AKAP CE (Fig. 4B). Likewise, substitution of Val 27 with Ala also severely compromises (but does not completely eliminate) the tethering of R CE . In contrast, mutation of Ile, Leu, or Val to Ala at positions 20, 35, 36, 40, and 41 has no impact on the formation of AKAP CE ⅐R CE complexes (Fig. 4B). Thus, folding of the R CE -docking domain into a (predicted) ␣-helix ( Fig. 6 and text below) differentially orients side chains of Val 27 and Ile 32 for interaction with a hydrophobic binding pocket in AKAP CE . Neither introduction nor removal of charged amino acids or exchange of a basic for acidic side chain in the region bounded by Glu 16 and Gln 22 (Fig. 6) affected docking of R CE with AKAP CE (Fig. 4, B and C). However, mutation of Cys 23 to Ala elicited a significant decline (ϳ70%) in binding of R CE with the anchor protein (Fig. 4B), thereby demarcating the N-terminal end of the AKAP-binding region. Replacement of Cys 44 with Ala profoundly diminished coupling of R CE with the tethering region of AKAP CE and indicated that the R CE -docking domain spans at least 22 residues. R CE polypeptides that include single or double Cys to Ala mutations (Ala 23 and Ala 44 ) dimerize normally (Fig. 4). 3 Furthermore, the amounts of wild type and mutant R CE proteins complexed by the AKAP CE tethering domain were not altered when 0.1 M dithiothreitol was added to the binding buffer. 3 The indicated concentration of dithiothreitol reduces all Cys residues in native R subunit dimers (55). Thus, interchain disulfide bonds are not involved in shaping the configuration of the R CE -docking surface or stabilizing R CE ⅐AKAP CE complexes. 4 Cys sulfhyldryl groups are also expected to be reduced in the intracellular environment. Under these conditions, Cys side chains could co-operatively promote (Cys 44 ) or optimize (Cys 23 ) tethering of PKAI CE 4 Nonphysiological formation of interchain disulfide bonds (in solutions lacking reducing agents) was exploited to assay the ability of R CE mutants to dimerize. In all other experiments, wild type and mutant R CE subunits were studied under reducing conditions.

FIG. 4. Identification of amino acids that govern R CE docking with AKAP CE and R CE -R CE dimerization.
Wild type R CE and R CE mutants were synthesized, purified to near homogeneity, and tagged with 32 P by incubation with [␥-32 P]ATP and the catalytic subunit of PKA, as outlined under "Experimental Procedures." Western blots (on Immobilon P) that contained replicated samples of purified, wild type GST-AKAP CE 176/409 (0.1 g/lane) were cut into strips corresponding to individual lanes. Complex formation between the tethering site of AKAP CE and either wild type or mutant R CE proteins was monitored by overlay binding assays (see "Experimental Procedures"). Binding buffer contained the indicated 32 P-labeled mutant or wild type R CE proteins (2 ϫ 10 5 cpm/ml) at a final concentration of 0.5 nM. Experiments were performed three times, and similar results were obtained in each repetition. Representative autoradiograms are presented. A shows effects of deletions of residues 1-10 (R CE ⌬10), 1-15 (R CE ⌬15), 1-25 (R CE ⌬25), 1-40 (R CE ⌬40), and 100 -376 (R CE 1-99) on the ability of partial R CE proteins to dock with the AKAP CE tethering site. B and C depict binding (docking) assays performed after introducing the indicated site-directed mutations in the N-terminal region of R CE . The ability of mutant R CE polypeptides to dimerize was assayed as described in Fig. 5 and the text of "Results." via hydrophobic interactions, involvement in hydrogen bonds, or (upon ionization) neutralization of charge.
Several mutations provided insights regarding amino acids that are crucial for the fundamental property of R CE homodimerization. Substitution of Phe 59 with Ala produced R CE monomers that are unable to bind AKAP CE (Fig. 4C). In mammalian RII␤, the corresponding mutation (Phe 36 to Ala; Fig. 6 and Ref. 26) has identical consequences. Moreover, alignment of residues 52-67 of R CE with homologous segments of RI␣ and RII␣ (Fig. 6) strongly indicates that this portion of the C. elegans regulatory subunit mediates dimerization. However, not all amino acids governing oligomerization are segregated in the predicted ␣-helix bounded by residues 52 and 67 (Fig. 6). N-terminal truncations that delete residues 1-40 or 1-25 from R CE preclude dimerization (Fig. 4A). Furthermore, scanning mutagenesis revealed that replacement of Tyr 26 with a small (Ala) or large (Val) aliphatic hydrophobic amino acid ablates R CE -R CE association (Fig. 4C). In contrast, a mutation that preserved aromatic character in the side chain (Tyr 26 to Phe) enabled dimerization. Thus, an aromatic amino acid (Tyr 26 ) embedded within the AKAP CE -docking region of R CE plays a critical role in stabilizing overall dimeric structure, which in  1 and 2) and an R CE mutant (Phe 59 to Ala) (lanes 3 and 4) were analyzed as described for A above. C, samples (0.9 mg in 0.2 ml of buffer A) of purified R CE and R CE Phe 59 to Ala were applied to a calibrated column of Superose 12 (see "Experimental Procedures"). The column was eluted with buffer A at a rate of 0.2 ml/min. Protein concentration in the eluate was monitored by absorbance at 280 nm. Peaks of R CE (dimer) and the Phe 59 to Ala R CE mutant (monomer) emerged at 43 and 47 min, respectively. A mixture of the two proteins was resolved into two peaks with the expected elution times (data not shown). All R CE mutants were characterized as dimers or monomers by both their electrophoretic mobility in nonreducing SDS-polyacrylamide gels and elution times during gel filtration, as illustrated by the examples shown.

FIG. 6. Sequence alignment and structural properties of the AKAP binding (docking) and dimerization domains of R CE , RI␣, and RII␣.
Structural and functional properties of amino acids and domains at the N terminus of murine RII␣ were assigned on the basis of systematic mutagenesis/expression experiments (26 -28) and determination of the solution structure of a fragment of RII␣ (amino acids 1-44) (29). Correlations between domains in RII␣ and murine RI␣ were established by Banky et al. (53). Amino acids shown in bold constitute a proximal ␣-helix that mediates docking with AKAPs; amino acids shown in bold italics comprise a distal ␣-helix that governs homodimerization. Ile 3 and Ile 5 (italics) in RII␣ modulate the binding affinity for AKAP75/79 (27). RII␣ lacks an N-terminal extension of the AKAP-binding helix (docking region) that is present in R CE and RI␣ (dotted underline). Residues in R CE that are essential for optimal binding with AKAP CE are marked with asterisks; amino acids essential for R CE dimerization are identified with pound signs.
turn ensures association of two copies of the segment of R CE (residues 23-44) that creates a binding surface recognized by the tethering domain of AKAP CE .
Selectivity of the AKAP CE Tethering Site Is Controlled by Two Amino Acids; Mutation of Phe 243 and Ser 244 Inverts the Specificity for R Subunit Isoforms-Neither RII␣ nor RII␤ efficiently competes for the high affinity (K D ϭ 7 nM) R CE tethering site in AKAP CE (see Fig. 8 and Ref. 37). Insights into structural features that account for this selectivity were sought by assessing abilities of wild type and mutant AKAP CE proteins to ligate 32 P-R CE and 32 P-RII␤ (Fig. 3). As expected, wild type anchor protein avidly binds R CE but sequesters only a minute amount of RII␤. Various mutations in the AKAP CE tethering domain have qualitatively similar effects on association with R CE and RII␤. Binding of both ligands is sharply suppressed by substitution of Ala for Leu 252 (Fig. 3), Leu 236 , or Ile 248 . Enhanced binding of R CE and RII␤ is evident when His 254 -Arg 255 -Lys 256 is replaced with Gln-Ala-Ala (Fig. 3). Mutation of Tyr 237 to Ala slightly reduces formation of AKAP CE ⅐R CE and AKAP CE ⅐RII␤ complexes. In contrast, a rationally designed double mutation of the unique core dipeptide (Phe 243 -Ser 244 , see above) in the AKAP CE tethering region caused a striking reversal in specificity for R subunit isoforms. Replacement of Phe 243 -Ser 244 with Leu-Val yields a tethering site that avidly complexes RII␤ and efficiently excludes R CE (Fig. 3). Evidently, the aromatic and hydrophilic/polar side chains of Phe 243 and Ser 244 , respectively, subserve a "gatekeeper" function; they may provide a binding pocket that accommodates a portion of the R CE -docking domain but is incompatible with an analogous region of the RII␤ (or RII␣) interaction surface. Conversion of the gatekeeper region to a homogeneous aliphatic hydrophobic microenvironment (Leu 243 -Val 244 ) yields a mutant AKAP CE tethering domain that shares critical structural and functional properties with RII-binding sites in mammalian AKAPs. This result can be further rationalized by considering the diagrams in Fig. 1 and conclusions from previous investigations on the prototypical mammalian anchor protein, AKAP75 (19). The variant AKAP CE tethering site contains five Leu, Ile, or Val residues that align in register with Leu 393 , Leu 400 , Val 401 , Ile 405 , and Ile 409 in AKAP75. Side chains from the latter group of amino acids cooperatively interact to assemble the high affinity RII-selective binding site in the prototype mammalian anchor protein (19). This site is apparently replicated in the AKAP CE double mutant. In a reciprocal manner, the data also suggest that classical AKAPs will not form stable complexes with R CE . This idea was confirmed by demonstrating that 32 Plabeled R CE is not sequestered by the wild type, RII-selective tethering domains of AKAP75 and S-AKAP84 (Fig. 7).
The properties of the wild type and double mutant (Phe 243 -Ser 244 to Leu 243 -Val 244 ) AKAP CE proteins also demonstrate that one or two core amino acids can confer extraordinary specificity for either RI-like R CE or RII isoforms to an otherwise invariant tethering domain. Phe 243 and Ser 244 play critical but distinct roles in the R CE -selective tethering site of AKAP CE . The large aromatic side chain of Phe 243 makes a substantial contribution to binding site affinity and consequently promotes formation of stable AKAP CE ⅐PKAI complexes. Mutation of Phe 243 to Ala diminishes R CE binding activity by Ͼ90%, but isoform specificity is unchanged. However, replacement of Phe 243 with Val produces a mutant tethering site that selectively complexes R CE and exhibits only a modest decline in affinity. Thus, a bulky aliphatic amino acid at residue 243 will support high affinity R CE and RI␣ (see below) binding in the context of Ser 244 or avid RII␣/RII␤ binding in the context of Ile 244 , Leu 244 , or Val 244 (e.g. Leu 243 -Val 244 in the AKAP CE double mutant). In contrast, substitution of Ser 244 with alternative amino acids either disrupts binding of R CE or elicits inversion in specificity that produces an RII-binding protein. Therefore, Ser 244 governs the highly selective association of RI-like R CE isoforms with the AKAP CE tethering site. The Ser 244 side chain may also subserve exclusion of RII isoforms from the binding site, thereby further enhancing the degree of isoform selectivity. It is possible that the aromatic side chain of Phe 243 (which is absent in all RII binding, classical AKAPs) can potentiate RII exclusion by Ser 244 and thus contributes to R isoform specificity in an indirect manner. Conversely, substitution of Ser 244 in AKAP CE with Val introduces a bulky aliphatic side chain that promotes coupling with RII isoforms and sharply diminishes R CE ligation.
Hydrophobic residues in the wild type AKAP CE tethering site (Leu 236 , Ile 248 , and Leu 252 ) and their counterparts in classical RII-binding AKAPs apparently subserve similar functions. Their side chains collectively provide a nonpolar surface or scaffold that complements hydrophobic regions in R CE (RI␣; see below) or RII dimers. The same array of conserved hydrophobic amino acids will promote either R CE (RI␣) or RII␣/RII␤ ligation by AKAP CE or doubly mutated AKAP CE (Phe 243 -Ser 244 3 Leu-Val), respectively. Thus, it is probable that similarly configured hydrophobic subdomains of R CE , RI␣, RII␣, and RII␤ are the binding targets. Multiple hydrophobic amino acids in the docking region of RII subunits have been rigorously implicated in binding with classical AKAPs (26 -29). However, the identity of subsets of individual side chains that contribute (a) isoformspecific interactions and (b) binding surfaces shared by R CE , RI␣, and RII isoforms remains unknown. Our results and studies by Banky et al. (53) indicate that Cys 23 , Ile 25 , Val 32 , and Cys 44 (in R CE ) are candidate residues for coupling with the aliphatic hydrophobic binding pocket in the tethering region of AKAP CE . At present, the mechanism by which Ser 244 confers R CE /RI␣ selectivity is unknown. Future determination of identities of amino acids in R CE or RI␣ that interact with Ser 244 should illuminate this issue.
The observations on the roles of conserved hydrophobic residues and gatekeeper amino acids yield insights into the basic architecture and design of AKAP tethering domains. Folding of 15-20 residues into an ␣-helix and the disposition of three aliphatic hydrophobic side chains along one helical face generates a basic scaffold suitable for accommodating RI, R CE , or RII dimers. A reasonable speculation is that these properties typified a common ancestor of modern AKAPs. Isoform-specific binding selectivity could then be superimposed on the basic design by minimal changes. Subsequent mutation and selection pressure on one or two core (gatekeeper) amino acids would enable evolution of RI-R CE or RII-selective AKAPs. Additional mutual selection pressure may have been exerted by the parallel evolution of R subunits. For example, the high affinity RI␣-binding AKAP CE (see below) co-evolved with RIlike R CE . In mammals, the emergence of RII subunits appears FIG. 7. Differential binding of R CE with AKAP CE and classical mammalian anchor proteins. A Western blot was probed with 32 Plabeled R CE in an overlay binding assay ("Experimental Procedures"). Lane 1 received 0.1 g of GST-AKAP CE 176/409; lane 2 contained 0.1 g of purified His-tagged S-AKAP84 fusion protein (residues 205-451), which includes the RII tethering domain (9); and lane 3 was loaded with 0.1 g of full-length AKAP75 (26). An autoradiogram is shown. Locations of proteins on the blot were determined by staining replicate lanes with Coomassie Blue.
to have played a dominant role in the selection of AKAP binding specificities. Most of the thoroughly characterized mammalian AKAPs exhibit highly preferential binding of RII (PKAII). However, RI␣-selective and neutral (RI␣ Х RII) binding sites were recently described in a mammalian germ cell anchor protein (56). This discovery and the characterization of R CE indicate that high affinity RI␣-binding proteins are encoded in eukaryotic genomes. Further searches and analyses may yield more examples of such anchor proteins and further insights into function of anchored PKAI.
AKAP CE Is a High Affinity, RI␣-binding Protein-Alignment of the sequences of the docking region of R CE (residues 23-47; Fig. 6) and the corresponding segment of mouse RI␣ (residues 18 -42; Fig. 6) revealed 15 identities and 5 conservative substitutions (interchange of Arg for Lys, Asp for Glu, Thr for Ile, and Val for Leu) over a span of 25 contiguous amino acids. The high level of sequence similarity indicated that the AKAP CE tethering module might accommodate mammalian RI␣ dimers. Competitive solution binding assays revealed that RI␣ inhibited binding of 32 P-R CE with the same concentration dependence as nonradioactive R CE (Fig. 8). Thus, the K D for an AKAP CE ⅐RI␣ complex is approximately 10 nM. To directly characterize RI␣-AKAP CE interactions, an Ala residue in the pseudosubstrate site of RI␣ was mutated to Ser, thus generating a PKA phosphorylation site. This modification does not alter the ability of RI␣ to bind cAMP, catalytic subunits, RI␣ (dimerization), or AKAP CE (43). 3 32 P-RI␣ is avidly sequestered by AKAP CE in an overlay binding assay (Fig. 8A). Moreover, both R CE and nonradioactive RI␣ potently inhibit binding of the radiolabeled ligand (Fig. 8, A and B). Thus, the nematode (R CE ) and mammalian (RI␣) regulatory subunits seem to be tethered at the same site. To confirm these observations, solution binding assays were also performed with wild type or phosphorylated RI␣, and AKAP CE ⅐RI␣ complexes were detected by protein staining or immunochemistry. Direct, stable binding of RI␣ by the AKAP CE tethering module was evident (a representative subset of results is shown in Fig. 8, C and D).
AKAP CE Binds RI␣ in Intact Cells-The in vitro binding data suggest that the AKAP CE tethering module might ultimately provide a novel molecular tool for manipulating the intracellular location and/or function of mammalian PKAI␣. The feasibility of such experimentation depends upon the demonstration that AKAP CE ⅐RI␣ complexes are generated in intact cells. Thus, hamster AV12 cells were co-transfected with a fulllength AKAP CE transgene and expression plasmids encoding either RI␣ or GST-RI␣. Both precipitation with anti-RI␣ IgGs and binding of GST-RI␣ to GSH-Sepharose 4B beads yielded complexes that contain substantial amounts of co-purifying AKAP CE (Fig. 9). Thus, the anchor protein stably sequesters RI␣ (PKAI␣) in the context of the environment of intact cells.
Conclusions, Insights, and Implications for Current Models-We have identified and characterized crucial structural features that ensure stable association of R CE dimers (PKAI CE ) with a C. elegans anchor protein (AKAP CE ). Elucidation of these structural properties suggested further experimentation, which (a) revealed that the tethering module of AKAP CE binds mouse RI␣ (PKAI␣) with high affinity in vitro and in intact cells and (b) demonstrated that one or two core amino acids govern the differential binding of either RI-like R CE or RII isoforms by AKAPs. Concepts underlying a current structural model for the tethering of PKAs are derived from detailed studies on properties associated with the N termini of RII␣ and RII␤ subunits (26 -29). Specific functions subserved by discrete N-terminal regions of mammalian RII isoforms are indicated in Fig. 6. Short segments of polypeptide that govern subunit dimerization and create a docking surface for AKAPs are lo-cated within residues 1-50 of RII isoforms (26 -28). A turnhelix-turn-helix motif divides the N-terminal portion of RII subunits into three functional regions (29). Two sets of tandem Pro residues precede distinct ␣-helices that mediate RII-RII homodimerization (Fig. 6, helix II residues 29 -45 in RII␣) or  6) were included in competition assays. Autoradiograms are presented. B shows the ability of various amounts of nonradioactive R CE (diamonds), RI␣ (squares), R CE 1-80 (residues 81-376 deleted, asterisks), R CE 1-65 (residues 66 -376 deleted, circles), RII␤ (triangles), and the R CE Ile 32 to Ala mutant (ϫ) to inhibit the binding of 32 P-labeled R CE with the AKAP CE tethering domain. Assays were performed in solution under equilibrium conditions. In the absence of competitor, the AKAP CE fusion protein bound 33,000 cpm. C, GST-partial AKAP CE (176/475) (7 nM) was incubated to equilibrium (1 h) with either 5 nM R CE (lane 1) or 10 nM RI␣ (lane 2) at 4°C (see "Experimental Procedures"). A replicate incubation was performed without added R subunits (lane 3) as a negative control. In addition, 10 nM RI␣ was incubated with 15 nM GST to control for nonspecific binding to GSH-Sepharose 4B (lane 4). Subsequently, GST-AKAP CE (M r ϭ ϳ60,000) and tightly associated proteins were isolated on GSH-Sepharose 4B beads. After extensive washing, bound proteins were analyzed by denaturing (reducing) electrophoresis. R CE (M r ϭ ϳ50,000) and RI␣ (M r ϭ ϳ50,000) were detected by staining with Coomassie Blue. Only the relevant portions of the stained gels are shown. D, RI␣ (10 nM) was incubated in binding buffer alone (lane 1) or binding buffer containing 7 nM GST-AKAP CE fusion protein (lane 2). Proteins isolated with GSH-Sepharose 4B beads were denatured in gel loading buffer, size-fractionated by SDS-polyacrylamide gel electrophoresis, and transferred to an Immobilon P membrane. The Western blot was then probed sequentially with anti-RI␣ serum (1:1000), and peroxidase-coupled secondary antibodies directed against rabbit IgGs. Antigen-antibody complexes were visualized by an enhanced chemiluminescence procedure (see "Experimental Procedures"). Signals were recorded on x-ray film. generate a core hydrophobic binding site (helix I, residues 10 -25 in RII␣) for AKAPs. Groups of aromatic residues and amino acids with large aliphatic side chains contribute essential functional properties in helices I and II of RII␣ and RII␤ (26 -28). Ile residues upstream from the first Pro-Pro turn modulate the affinity of RII dimers for AKAPs (27). A structural model derived from solution NMR analysis (29) indicates that coalescence of the hydrophobic surfaces of the RII-docking site and the AKAP tethering region results from the formation of many specific, intermolecular hydrophobic contacts.
The N-terminal segment of R CE includes two predicted ␣-helices and shares several critical properties with corresponding regions in RII␣ and RII␤. The structure and function of the distal dimerization helix (residues 52-67 in R CE ; Fig. 6) are highly conserved between R CE and RII␣/RII␤. Substitution of Ala for Phe 59 in otherwise full-length, wild type R CE abolishes dimerization and concomitantly ablates binding with AKAP CE . These results illustrate basic principles of R subunit architecture that are conserved from nematodes to mammals: in the distal helix, individual amino acids with aromatic side chains (Phe) are indispensable for maintaining overall dimeric struc-ture of R subunits, and monomeric R subunits cannot generate a functional AKAP-docking surface from a single wild type copy of the AKAP-binding region (Fig. 6). Like RII␣/RII␤, R CE has a proximal helix that mediates high affinity binding with the tethering domain of an isoform-selective anchor protein. Robust hydrophobic interactions promote the stable tethering of both RII-and RI-like R CE dimers. Side chains from 6 of the 9 large hydrophobic amino acids in the docking (AKAP-binding) region of RII␣ engage in hydrophobic contacts with core apolar residues in the tethering domains of classical mammalian AKAPs (29). For R CE , side chains derived from Cys 23 , Val 27 , Ile 32 , and Cys 44 apparently assemble a hydrophobic surface that couples with a complementary, nonpolar binding pocket derived from Leu 236 , Ile 248 , and Leu 252 in the AKAP CE tethering site.
Striking differences between the docking regions of R CE and RII␣ are also evident. The segment of R CE that engages the anchor protein is 50% longer than the corresponding domain in RII␣. In contrast to the conserved dimerization regions, amino acid sequences of the R CE -and RII␣-docking regions are not homologous (only ϳ20% identical residues). A high proportion (66%) of large hydrophobic amino acids in the proximal helix of RII␣ directly participate in tethering, whereas side chains from Leu 20 , Leu 35 , Val 36 , Ile 40 , and Val 41 in R CE are not essential for high affinity ligation with AKAP CE . Optimal alignment of the docking regions (Fig. 6) places only two critical residues from R CE (Ile 32 and Cys 44 ) in register with hydrophobic amino acids that mediate binding of RII␣ with classical AKAPs. Three other amino acids that evidently play central roles in tethering PKA-I CE (Cys 23 , Tyr 26 , and Val 27 ) are incorporated into a segment of R CE (residues 16 -24; Fig. 6) that has no counterpart in RII␣. Conversely, R CE contains no functional analogs of the Ile 3 and Ile 5 residues, which control the binding affinity of RII␣ for AKAPs (27). Finally, two Cys residues that are conserved between R CE and mammalian RI isoforms make substantial contributions toward the stabilization of R CE ⅐AKAP CE complexes. Docking sites in RII isoforms are devoid of Cys. Reduced Cys sulfhydryls can participate in hydrogen bonds or undergo ionization, but these are atypical and infrequently observed properties. It is more likely that Cys 23 and Cys 44 side chains promote tethering by their incorporation into a hydrophobic binding surface. Thus, R CE -and RII␣-docking domains differ substantially in overall size, configuration, amino acid sequence, and the positions and identities of critical residues along the helix. This sharp divergence in N-terminal structural features accounts (in part) for R CE -selective tethering and RII exclusion by AKAP CE .
Recent investigations disclosed that several classical mammalian AKAPs bind both RII isoforms and RI␣ at the same tethering site (34,35). However, RII isoforms are bound 25-500-fold more avidly than RI␣. Banky et al. (53) thoroughly characterized properties in RI␣ that mediate the lower affinity binding with DAKAP-1 (also known as S-AKAP84/AKAP121 (9,46)). Val, Ile, and Cys residues in RI␣, which align with Val 27 , Ile 32 , and Cys 44 in R CE (Fig. 6), control coupling with DAKAP-1. Thus, several structural properties that drive high affinity binding of R CE with AKAP CE are highly conserved from C. elegans to mammals and are sufficient to enable lower affinity binding with an AKAP optimized for tethering RII isoforms. The conservation of the R CE /RI␣-docking region in the distant nematode and mammalian branches of evolution supports the idea that PKAI␣ may be adapted and targeted for specific functions by anchor proteins in a broad spectrum of eukaryotes.
Binding of R CE and RI␣ with their cognate anchor proteins declines precipitously upon substitution of critical Val or Ile FIG. 9. Formation of AKAP CE ⅐ RI␣ complexes in intact cells. A, hamster AV-12 cells were co-transfected with expression vectors encoding AKAP CE and murine RI␣ ("Experimental Procedures"). Aliquots of cytosol (300 g protein) derived from homogenates of transfected AV-12 cells were incubated with 3 l of anti-RI␣ serum or 3 l of nonimmune serum for 16 h at 4°C. Immune complexes were isolated on protein A-Sepharose 4B beads and washed extensively. Bound proteins were dissolved in gel loading buffer, size-fractionated by denaturing electrophoresis, and characterized by Western immunoblot analysis, as indicated under "Experimental Procedures." Lane 1 received proteins precipitated with nonimmune serum; lanes 2 and 4 contained proteins precipitated with anti-RI␣ IgGs. Proteins (15% of total) that were not precipitated by anti-RI␣ IgGs (lane 2) were assayed in lane 3. The blot was probed with affinity purified IgGs directed against AKAP CE (1:2000 relative to serum), and IgG-AKAP CE complexes were detected by peroxidase-coupled secondary antibodies and enhanced chemiluminescence methodology as described in Fig. 7 and "Experimental Procedures." Lane 4 was incubated with IgGs that were saturated with excess purified AKAP CE antigen. Only the relevant portion of the blot is shown. No other bands were visualized. B, AV-12 cells were co-transfected with expression vectors encoding AKAP CE and either GST-RI␣ or GST. Aliquots of cytosol (300 g of protein) derived from homogenates of transfected cells were incubated with GSH-Sepharose 4B beads for 2 h at 4°C. After extensive washing, proteins were analyzed by Western immunoblot assays as indicated above. Lane 1 received proteins precipitated from cells expressing the AKAP CE and GST transgenes. Lane 2 contained proteins precipitated from cells expressing AKAP CE and GST-RI␣ transgenes. Proteins (15% of total) that were not precipitated from cells transfected with the AKAP CE , and GST-RI␣ transgenes were applied to lane 3. The Western blot was probed with anti-AKAP CE IgGs as described above, and signals were recorded on x-ray film. residues (Val 27 and Ile 32 in R CE ) (Fig. 6) with Ala. Thus, the extended hydrophobic side chains of these amino acids are essential for diversifying functions of both PKAI isoforms. In contrast, tethering of R CE is more stringently dependent on Cys residues than immobilization of RI␣. Mutation of Cys 44 to Ala in R CE severely impairs binding with AKAP CE (Fig. 4), whereas the corresponding mutation in RI␣ has a minimal effect on the formation of complexes with DAKAP-1 (53). Binding of RI␣ with DAKAP-1 declined substantially only when Cys 37 was replaced with His, which contains a markedly different side chain. Moreover, replacement of Cys 23 with Ala reduces docking of R CE with AKAP CE by Ͼ50%; the analogous mutation in RI␣ has no impact on coupling with DAKAP-1 (53). Thus, Cys residues in the proximal N-terminal helix of R CE may be involved in cooperatively maximizing the stability of AKAP CE ⅐R CE complexes in concert with interactions mediated by Val 27 and Ile 32 . The roles of Cys 23 and Cys 44 are evidently magnified in the context of the high affinity R CE /RI␣-binding site of AKAP CE .
As this manuscript reached completion, Miki and Eddy (57) reported an analysis of the binding of RI␣ and RII␣ subunits with two sites on a germ cell anchor protein known as FSC1 (56) or AKAP82 (58). Domain B of FSC1 selectively complexes RI␣; domain A binds both RII and RI␣ with similar affinities. Binding activities and specificities were assessed after Ala or Val scanning mutagenesis In addition, mutations were introduced into a classical AKAP, Ht31 (57), to probe an RII-selective tethering site. The investigation focused on 10 central amino acids in the cited binding regions: domain B (YAN-QVASDMM), domain A (YVNRLSSLVI), and Ht31 (AASRIV-DAVI). These residues align with amino acids 239 -248 of AKAP CE (Fig. 1). Three important determinants were discerned in these regions and rules for RI-selective, RII-selective and dual binding specificities were suggested. Our current results and previous studies (9,19,23,37,52,59) are in partial accord with these suggestions. However, other aspects of our observations suggest distinctly different interpretations. Miki and Eddy (57) propose that a large hydrophobic side chain at position 10 (Met, Ile, and Ile for FSC1-A, FSC1-B, and Ht31, respectively) is essential (but nonselective) for R dimer binding. This suggestion is consistent with the previously published demonstration that conversion of the corresponding (Ile) residues in AKAP75, AKAP-KL, DAKAP200, and AKAP CE to Ala eliminates binding activity (19,37,52,59). A second, critical insight is that residue 6 (Ala, Ser, and Val in A, B, and Ht31) determines isoform specificity of the tethering site. It is proposed that Ala is required for selective tethering of RI␣; Ser enables dual binding of RI␣ and RII␣ with similar affinities, and Ile, Val, or Leu promotes RII binding. The last suggestion is consistent with previously published alignments and conclusions indicating that Leu, Ile, and Val invariably occupy these positions in RII-binding proteins (9,23,24,37,52), and mutation of these residues to Ala compromises RII binding activity in all cases (19,37,52,59). Mutagenesis/expression studies on AKAP CE document Ser 244 or Val 244 (corresponding to position 6) are crucial determinants for R isoform selectivity. The suggested rules predict that Ser at this position is characteristic of a dual specific AKAP (57). In contrast, AKAP CE is an exceptionally selective R CE (and RI␣) anchor protein. The ratio of R CE :RII binding is 50; a large excess of RII␣ or RII␤ does not efficiently inhibit R CE binding with AKAP CE in competition assays; RI␣ and R CE compete for the AKAP CE tethering site with high affinity (K D ϳ 10 nM); and AKAP CE binds RI␣ but not RII␣ in the context of intact cells. The binding properties of the wild type tethering domain of AKAP CE contravene another proposed rule (57). Occupancy of position 6 by Ala is not re-quired for establishing RI␣-selective binding activity. Finally, it is proposed that Ala at position 2 is critical for optimal binding of either RI or RII isoforms (57). However, replacement of the corresponding Ala in AKAP75, AKAP-KL, and AKAP CE with Ser has no effect on tethering (19,37,52).
Several factors may have contributed to the indicated differences in results and interpretations. We used specified concentrations of purified 32 P-R CE , 32 P-RI␣, or 32 P-RII␤ and equal amounts of nearly homogeneous wild type and mutant AKAPs in all in vitro binding studies. This enabled direct assessment of AKAP⅐R complex formation and facile comparison of binding of R CE , RI␣, and RII␤ with the tethering domain. High affinity binding of R CE and RI␣ was established by direct equilibrium binding and competition assays in solution. In studies on FSC1 A and B domains, crude testis extracts were employed as the source of RI␣ and RII␣ subunits (57). Thus, the concentrations of the two ligands were not precisely known or normalized. The possibility that substantial amounts of RII␣ and RI␣ were complexed with endogenous AKAPs in the extracts was not excluded. In addition, RI␣ and RII␣ binding activities of AKAPs were monitored in a highly indirect manner: via Western immunoblot analysis. Marked discrepancies between actual amounts of RI␣ and RII␣ bound and chemiluminescence signals can be introduced by differences in affinities of the primary antibodies. Additional limitations in quantitative analysis may have arisen from the use of different secondary antibodies and the narrowly limited linear range of chemiluminescence signals. No information was provided regarding either quantitative calibration of the indirect, antibody-based methodology used for assessment of binding activities or K D values of the various tethering domain complexes with RI␣ and RII␣. Finally, an essential criterion for establishing the significance of the various tethering domain variants involves the demonstration that stable, isoform-selective tethering occurs (as predicted) in the internal environment of intact cells. The model proposed by Miki and Eddy (57) was not evaluated by data obtained from intact cells.
Despite the caveats and substantial differences in results and interpretation regarding requirements for individual amino acid residues, our current and previous (19,37,52,59) results and the study of Miki and Eddy (57) underscore the importance of position 10 in controlling binding affinity, illuminate several shared common features in the design of all types of R subunit tethering sites, and indicate that central core residues of the tethering region are determinants of isoform specificity. However, studies on a naturally occurring R CE /RI␣-selective AKAP CE, generated via mutation and selection (at all residues) over the course of evolution, suggest that rules governing R isoform specificity and binding affinity are not fully explained at the level of single amino acids. Perhaps related but somewhat more complex rules will emerge as a more detailed understanding of cooperative interactions among key amino acids (e.g. the role of Phe 243 in modulating properties of Ser 244 ) emerges.