Biochemistry : Zozulya and Stryer
S. Zozulya
Abstract:Recoverin, a recently discovered member of the EF-hand superfamily of Ca2+-bindig proteins, serves as a Ca2+ sensor in vision. The amino terminus of the protein from retinal rod cells contains a covalently attached myristoyl or related N-acyl group. We report here studies of unmyrtylated and myristoylated recombinant recoverin designed to delineate the biological role of this hydrophobic unit. Ca2+ induces the binding of both the unmyristoylated and myristoylated proteins to phenyl-agarose, a hydrophobic support. Binding was half-maximal at 1.1 and 1.0 FM CaW , respectively. The Hill coefficients of 1.8 and 1.7, respectively, indicate that binding was cooperative. In contrast, Ca2+ induced the binding of myristoylated but not of unmyristoylated recoverin to rod outer segment membranes. Binding to these membranes was half-maximal at 2.1 FM Ca2+, and the Hill coefficient was 2.4. Likewise, myristoylated but not unmyritoylated recoverin exhibited Ca2+-induced binding to phosphatidylcholine vesicles. These findings suggest that the binding of Ca2+ to recoverin has two effects: (i) hydrophobic surfaces are exposed, allowing the protein to interact with complementary nonpolar sites, such as the aromatic rings of phenyl-agarose; and (is) the myristoyl group is extruded, enabling recoverin to insert into a lipid bilayer membrane. The myristoyl group is likely to be an active participant in Ca2+ signaling by recoverin and related EF-hand proteins such as visinin and neurocalcin. Recoverin, a recently discovered 23-kDa Ca2+-binding protein, activates guanylate cyclase in bovine retinal rod outer segments when the Ca2+ level is lowered in the submicromolar range (1, 2). This Ca2+ sensor is a new member of the EF-hand superfamily, which includes calmodulin, troponin C, and parvalbumin (3, 4). The cDNA for recoverin has been cloned and expressed in high yield in Escherichia coli (5). Bovine recoverin is 59%o identical in amino acid sequence to visinin, a Ca2+-binding protein found in chicken cone cells (6). Homologs of recoverin and visinin have recently been found in mammalian brain (7-10). Interest in this Ca2+ sensor has been further heightened by the finding that the aminoterminal glycine of retinal recoverin is linked to a myristoyl (C14:0) or related fatty acyl group (C14:1, C14:2, or C12:0) (11). Numerous eukaryotic and viral proteins contain a covalently attached myristoyl group (12, 13). The importance of this covalent modification is evident in the viral src protein pp60V.SrC, which is oncogenic only when myristoylated (14). Human retinal recoverin (isolated as the p26 antigen in cancer-associated retinopathy) (15) and S-modulin, a recoverin-like protein from frog retinas (16, 17) are known to bind more tightly to membranes or hydrophobic supports in the presence ofCa2+ than in its absence. A similar effect has been observed with calmodulin (18) and the annexins (19), which lack a myristoyl group. Is myristoylation essential for the Ca2+-dependent binding of recoverin and homologs to membranes? We answered this question by preparing both myristoylated and unmyristoylated recombinant recoverin in E. coli and by measuring their binding to phenyl-agarose, retinal rod outer segment (ROS) membranes, and phosphatidylcholine vesicles. The Ca2+-induced binding of myristoylated but not of unmyristoylated recoverin to lipid bilayers shows that hydrophobicity conferred by this covalently attached fatty acyl group can be dynamically controlled by Ca2+. Recoverin contains a Ca2+-myristoyl switch akin to the GTP-myristoyl switch recently found in ADP-ribosylation factor, a myristoylated guanyl nucleotide binding protein that regulates protein transport (20, 21). MATERIALS AND METHODS Recoverin. Recombinant myristoylated and unmyristoylated recoverins were expressed in the overproducing E. coli strains pTrec2/pBB131/DH5aF' and pTrec2/DH5a, respectively, and were purified by a modification of a previously described procedure (5, 15). Cells from 20 liters of a bacterial culture were resuspended in 200 ml of 100 mM KCl/1 mM dithiothreitol/1 mM MgCl2/50 mM potassium Hepes, pH 7.5 (buffer A) supplied with 1 mM EGTA and disrupted by sonication. Streptomycin sulfate was added with stirring over a 10-min period to a final concentration of 0.1% from a 5% stock prepared in buffer A, and the cell debris was removed by centrifugation at 40C for 30 min at 18,000 x g. The cleared lysate was filtered through filter paper (Whatman 1), brought to 2 mM CaCl2, and applied to an 80-ml phenyl-Sepharose CL-4B (Pharmacia) column previously equilibrated with buffer A containing 2 mM CaC12. The column was washed with 400 ml of the same buffer, and the fraction containing recoverin was eluted with 100-150 ml of buffer A containing 2 mM EGTA. The eluate was diluted with 3 volumes of cold water, applied to a HiLoad 26/10 Q-Sepharose HP column (Pharmacia), and chromatographed with a gradient of 0-200 mM KCl in 500 ml of 20 mM Tris-HCl, pH 8.0/1 mM dithiothreitol/1 mM MgCl2/0.1 mM EGTA. Purified recoverin was concentrated to 10-20 mg/ml using Centriprep 10 centrifugal concentrators (Amicon), flash-frozen in liquid nitrogen, and stored at -70'C. To produce tritium-labeled myristoylated recoverin, the overproducing strain pTrec2/pBB131/DHSFa' containing the yeast N-myristoyltransferase expression vector (22) was grown in minimal medium with Casamino acids (M9CA) and supplied with [3H]-myristic acid [New England Nuclear; 39.3 Ci/mmol, 10 ,uCi/ml of culture (1 pCi = 37 GBq)] at the moment of induction with isopropyl ,8-D-thiogalactoside. The specific activity of the resulting tritiated myristoylated recoverin was 0.31 Ci/mmol. Membrane Preparations. Liposomes were formed from soybean phosphatidylcholine (type IV-S; Sigma) or from a mixture of phosphatidylcholine and bovine brain phosphatidylserine (Avanti Polar Lipids) by sonicating a lipid suspension in buffer A. Stock suspensions of liposomes (10-20 mg of lipid per ml) were stored at 40C and used within several Abbreviation: ROS, rod outer segment. 11569 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 11570 Biochemistry: Zozulya and Stryer days. ROS were isolated in the dark from frozen bovine retinas as described (23). The concentration of rhodopsin in ROS membrane suspensions was calculated from the change in absorbance at 500 nm after bleaching in 1.5% Ammonyx LO (Onyx Chemical, Jersey City, NJ); an absorbance coefficient of 40,600 M-1 cm-1 was used. Free Ca+ Concentrations. Ca2+/EGTA buffers were prepared as described (24). The free Ca2+ concentration in these buffers was determined by using the fluorescent indicator rhod-2 (Molecular Probes). Fluorescence measurements were made with an SLM 8000 spectrofluorimeter (SLM Instruments, Urbana, IL). The free Ca2+ concentration was calculated from the fluorescence emission intensity at 576 nm on excitation at 553 nm by using a Kd of 1.0 AM for the Ca2+-rhod-2 complex (24, 25). Binding of Recoverin to Membranes and Phenyl-Agarose. ROS membranes (100 jiM rhodopsin) were' bleached and washed twice in the light with 50 vol of 10 mM potassium Hepes (pH 7.5). Myristoylated or unmyristoylated recombinant recoverin (10 jiM) in buffer A containing 2 mM EGTA ("low Ca2+") or 50 AM CaCl2 ("high Ca2+") was incubated with ROS membranes (10 p.M rhodopsin) or liposomes (1 mg lipid per ml) for 20 min at 24TC. Membranes were pelleted by centrifugation for 5 min at 10,000 x g (for ROS) or 20 min at 100,000 x g (for liposomes), and the supernatant was removed. The pellet was gently rinsed with buffer and then analyzed by SDS/polyacrylamide gel electrophoresis. The Ca2+ dependence of the binding of unmyristoylated or myristoylated recoverin to phenyl-agarose was measured by incubating the protein (2 jiM) with 50 A.l ofphenyl-Sepharose 4B (Pharmacia) in 250 jul of each of a series of Ca2+/EGTA buffers (2mM EGTA) containing 40 mM NaCl, 1 mM MgCl2, and 80 mM potassium Hepes (pH 7.4) (buffer B). These mixtures were vigorously shaken for 30 min at 250C and then centrifuged for 5 min at 10,000 x g. The supernatants were removed and diluted with 10 volumes of buffer B containing 2 mM EGTA and no added Ca2+. The concentration of recoverin in each supernatant was determined from its tryptophan fluorescence intensity. 19 .... -....9 w. .1 ...g