Study of calmodulin binding to the alternatively spliced C-terminal domain of the plasma membrane Ca2+ pump.

The C-terminal regions of the four human plasma membrane Ca2+ pump isoforms 1a-d generated from alternatively spliced RNA have been expressed in Escherichia coli, and the recombinant proteins have been purified to a very high degree. The C-termini of isoforms 1a, 1c, and 1d contain an insert encoded by an alternatively spliced exon which is homologous to the calmodulin binding domain of isoform 1b. In isoforms 1c and 1d (29 and 38 amino acid insertions, respectively), subdomain A of the original calmodulin binding site of isoform 1b is followed by the spliced-in domain, which is then followed by subdomain B of the original calmodulin binding site. The positive charges of histidine residues at positions 27, 28, and 38 of the alternatively spliced sequence are likely to be responsible for the observed pH-dependent calmodulin binding to the novel "duplicated" binding site. The affinity of calmodulin for the C-terminal domains of isoforms 1a, 1c, and 1d, which contain the histidine-rich inserts, is much higher at pH 5.9 than at pH 7.2. A synthetic peptide (I31) containing 31 amino acids of the alternatively spliced sequence (from residue 9 to 40) also binds calmodulin with strong pH dependency. Alternative splicing in the C-terminal domain is proposed to confer pH dependence to the regulation of the activity of Ca2+ pump isoforms.     

Peptide sequence analysis and molecular cloning reveal two calcium pump isoforms in the human erythrocyte membrane

The sequence of more than 1,000 amino acid residues, derived from two different isoforms, has been determined from peptides generated from purified human erythrocyte membrane Ca2(+)-ATPase (hPMCA). Several of these peptide sequences correspond to the previously reported, cDNA deduced sequence of the "teratoma" Ca2+ pump isoform hPMCA1 (Verma, A. K., Filoteo, A. G., Stanford, D. R., Wieben, E. D., Penniston, J. T., Strehler, E. E., Fischer, R., Heim, R., Vogel, G., Matthews, S., Strehler-Page, M.-A., James, P., Vorherr, T., Krebs, J., and Carafoli, E. (1988) J. Biol. Chem. 263, 14152-14159). The complete primary structure of a novel isoform (hPMCA3) has been determined by molecular cloning and nucleotide sequencing of its corresponding cDNA. This new member of the plasma membrane Ca2+ pump family consists of 1,205 amino acid residues with a calculated Mr of 133,930, and it shows 88% similarity (75% identity) with the previously sequenced pump isoform. Specific probes detect major mRNA species of 5.6 kilobases for hPMCA1, and of 7.5 kilobases for hPMCA3, on Northern blots of human K562 erythroleukemic cell RNA. A large number of peptide sequences match perfectly with only one or the other of these isoforms and all peptides (with 6 exceptions corresponding to a contaminant protein or to a third minor Ca2+ pump isoform) are found in either only one or in both of the isoforms. The two erythrocyte Ca2+ pumps display high sequence divergence in a few localized regions that may determine isoform-specific functional specializations; for example, the putative extracellular loop separating transmembrane domains 1 and 2, the highly negatively charged region previously suggested to be involved in Ca2+ binding, and the site of cAMP-dependent protein kinase phosphorylation.     

The C-terminal 165 amino acids of the plasma membrane Ca(2+)-ATPase confer Ca2+/calmodulin sensitivity on the Na+,K(+)-ATPase alpha-subunit.

The C-terminal 165 amino acids of the rat brain plasma membrane (PM) Ca(2+)-ATPase II containing the calmodulin binding auto-inhibitory domain was connected to the C-terminus of the ouabain sensitive chicken Na+,K(+)-ATPase alpha 1 subunit. Expression of this chimeric molecule in ouabain resistant mouse L cells was assured by the high-affinity binding of [3H]ouabain. In the presence of Ca2+/calmodulin, this chimeric molecule exhibited ouabain inhibitable Na+,K(+)-ATPase activity; the putative chimeric ATPase activity was absent in the absence of Ca2+/calmodulin and activated by Ca2+/calmodulin in a dose-dependent manner. Furthermore, this chimeric molecule could bind monoclonal IgG 5 specific to the chicken Na+,K(+)-ATPase alpha 1 subunit only in the presence of Ca2+/calmodulin, suggesting that the epitope for IgG 5 in this chimera is masked in the absence of Ca2+/calmodulin and uncovered in their presence. These results propose a direct interaction between the calmodulin binding auto-inhibitory domain of the PM Ca(2+)-ATPase and the specific regions of the Na+,K(+)-ATPase alpha 1 subunit that are structurally homologous to the PM Ca(2+)-ATPase. A comparison of the deduced amino acid sequences revealed several possible regions within the Na+,K(+)-ATPase that might interact with the auto-inhibitory domain of the PM Ca(2+)-ATPase.     

Primary structure of the Neurospora plasma membrane H+-ATPase deduced from the gene sequence. Homology to Na+/K+-, Ca2+-, and K+-ATPase

The gene for the Neurospora crassa plasma membrane H+-ATPase has been cloned and sequenced. The gene encodes for a protein of 920 amino acids with a molecular weight of 100,002. The coding region is interrupted by four introns: three near the amino terminus and one near the carboxyl terminus. The deduced amino acid sequence of the N. crassa plasma membrane H+-ATPase exhibits 75% homology to the amino acid sequence of the Saccharomyces cerevisiae plasma membrane H+-ATPase. Also, an amino acid comparison with the Na+/K+-ATPase from sheep kidney, Ca2+-ATPase from rabbit muscle, and K+-ATPase from Escherichia coli reveals that certain regions are highly conserved and suggest that these regions may serve essential functions which are common to the various cation-motive ATPases. This observation suggests that the phosphorylatable, cation-motive ATPases may function via a similar energy transduction mechanism.     

The amino and carboxyl termini of the Neurospora plasma membrane H+-ATPase are cytoplasmically located

Based on hydropathy analysis, the P-type cation translocating ATPases are believed to have similar topological arrangements in the membrane, but little independent evidence exists for their precise pattern of transmembrane folding. As a first step toward defining the topology of the Neurospora plasma membrane H+-ATPase, we have mapped the orientation of the amino and carboxyl termini. In three different types of experiments, both termini of the H+-ATPase were shown to be exposed at the cytoplasmic surface of the plasma membrane: 1) antibodies specific for the amino and carboxyl termini bound to permeabilized but not intact cells; 2) inside-out plasma membrane vesicles were approximately 100-fold more effective than intact cells in competing for antibody binding; and 3) trypsin, which is known to proteolyze three sites at the amino terminus and one site at the carboxyl terminus of the purified Neurospora H+-ATPase (Mandala, S. M., and Slayman, C. W. (1988) J. Biol. Chem. 263, 15122-15128), was found in the present study to cleave the same sites in inside-out plasma membrane vesicles but not in intact cells. These results indicate that the ATPase polypeptide traverses the membrane an even number of times, in support of a previously published topological model (Hager, K. M., Mandala, S. M., Davenport, J. W., Speicher, D. W., Benz, E. J., Jr., and Slayman, C. W. (1986) Proc. Natl. Acad. Sci. U. S. A. 83, 7693-7697).     

Complete primary structure of a human plasma membrane Ca2+ pump

cDNAs coding for a plasma membrane Ca2+ pump were isolated from a human teratoma library and sequenced. The translated sequence contained 1,220 amino acids with a calculated molecular weight of 134,683. All regions of functional importance known from other ion-transporting ATPases could be identified. The translated sequence also contained, near the carboxyl terminus, the calmodulin-binding domain and two domains which are very rich in glutamic acid and aspartic acid. These two domains resemble calmodulin somewhat and one of them may play a role in the binding of Ca2+. The enzyme also contains domains rich in serine and threonine, one of which has a sequence matching those of good cAMP-dependent protein kinase substrates. The carboxyl-terminal region is important for regulation by calmodulin, proteolysis, and phosphorylation. Near the amino terminus are two domains which are very rich in lysine and glutamic acid, as well as two domains resembling EF hands, one of which also has some resemblance to calmodulin. Comparison of the cloned sequence with peptide sequences from the erythrocyte Ca2+ pump showed that the two proteins have a very high proportion of identical residues but are not 100% identical, indicating that they represent different isozymes.     

Apocalmodulin and Ca2+ calmodulin bind to the same region on the skeletal muscle Ca2+ release channel.

The skeletal muscle Ca2+ release channel (RYR1) is regulated by calmodulin in both its Ca2+-free (apocalmodulin) and Ca2+-bound (Ca2+ calmodulin) states. Apocalmodulin is an activator of the channel, and Ca2+ calmodulin is an inhibitor of the channel. Both apocalmodulin and Ca2+ calmodulin binding sites on RYR1 are destroyed by a mild tryptic digestion of the sarcoplasmic reticulum membranes, but calmodulin (either form), bound to RYR1 prior to tryptic digestion, protects both the apocalmodulin and Ca2+ calmodulin sites from tryptic destruction. The protected sites are after arginines 3630 and 3637 on RYR1. These studies suggest that both Ca2+ calmodulin and apocalmodulin bind to the same or overlapping regions on RYR1 and block access of trypsin to sites at amino acids 3630 and 3637. This sequence is part of a predicted Ca2+ CaM binding site of amino acids 3614-3642 [Takeshima, H., et al. (1989) Nature 339, 439-445].     

Calmodulin binding to the C-terminus of the small-conductance Ca2+-activated K+ channel hSK1 is affected by alternative splicing

We identified three splice variants of hSK1 whose C-terminal structures are determined by the independent deletion of two contiguous nucleotide sequences. The upstream sequence extends 25 bases in length, is initiated by a donor splice site within exon 8, and terminates at the end of the exon. The downstream sequence consists of nine bases that compose exon 9. When the upstream sequence (hSK1(-)(25b)) or both sequences (hSK1(-)(34b)) are deleted, truncated proteins are encoded in which the terminal 118 amino acids are absent. The binding of calmodulin to these variants is diminished, particularly in the absence of Ca2+ ions. The first 20 amino acids of the segment deleted from hSK1(-)(25b) and hSK1(-)(34b) contain a 1-8-14 Ca2+ calmodulin binding motif, and synthetic oligopeptides based on this region bind calmodulin better in the presence than absence of Ca2+ ions. When the downstream sequence (hSK1(-)(9b)) alone is deleted, only the three amino acids A452, Q453, and K454 are removed, and calmodulin binding is not reduced. On the basis of the relative abundance of mRNA encoding each of the four isoforms, the full-length variant appears to account for most hSK1 in the human hippocampus, while hSK1(-)(34b) predominates in reticulocytes, and hSK1(-)(9b) is especially abundant in human erythroleukemia cells in culture. We conclude that the binding of calmodulin by hSK1 can be modulated through alternative splicing.     

Ca2+-dependent interaction of the trpl cation channel and calmodulin.

The transient receptor potential-like ion channel from Drosophila melanogaster was originally identified as a calmodulin binding protein (Philips et al., 1992) involved in the dipterian phototransduction process. We used a series of fusion proteins and an epitope expression library of transient receptor potential-like fusion proteins to characterize calmodulin binding regions in the transient receptor potential-like channel through the use of [125I]calmodulin and biotinylated calmodulin and identified two distinct sites at the C-terminus of the transient receptor potential-like ion channel. Calmodulin binding site 1, predicted from searching of the primary structure for amphiphilic helices (Philips et al., 1992), covers a 16 amino acid sequence (S710-I725) and could only be detected through biotinylated calmodulin. Calmodulin binding site 2 comprises at least 13 amino acids (K859ETAKERFQRVAR871) and binds both [125I]calmodulin and biotinylated calmodulin. Both sites (i) bind calmodulin at least in a one to one stoichiometry, (ii) differ in their affinity for calmodulin revealing apparent Ki values of 12.3 nM (calmodulin binding site 1) and 1.7 nM (calmodulin binding site 2), respectively, (iii) bind calmodulin only in the presence of Ca2+ with 50% of site 1 and site 2, respectively, occupied by calmodulin in the presence of 0.1 microM (calmodulin binding site 1) and 3.3 microM Ca2+ (calmodulin binding site 2) and give evidence that (iv) a Ca2+-calmodulin-dependent mechanism contributes to transient receptor potential-like cation channel modulation when expressed in CHO cells.     

Functional consequences of alterations to amino acids located in the catalytic center (isoleucine 348 to threonine 357) and nucleotide-binding domain of the Ca2+-ATPase of sarcoplasmic reticulum

The sequence of 10 amino acids (ICSDKTGTLT357) at the site of phosphorylation of the rabbit fast twitch muscle Ca2+-ATPase is highly conserved in the family of cation-transporting ATPases. We changed each of the residues flanking Asp351, Lys352, and Thr353 to an amino acid differing in size or polarity and assayed the mutant for Ca2+ transport activity and autophosphorylation with ATP or P1. We found that conservative changes (Ile----Leu, Thr----Ser, Gly----Ala) or the alteration of Cys349 to alanine did not destroy Ca2+ transport activity or phosphoenzyme formation, whereas nonconservative changes (Ile----Thr, Leu----Ser) did disrupt function. These results indicate that very conservative changes in the amino acids flanking Asp351, Lys352, and Thr353 can be accommodated. A number of mutations were also introduced into amino acids predicted to be involved in nucleotide binding, in particular those in the conserved sequences KGAPE519, RDAGIRVIMITGDNK629, and KK713. Our results indicate that amino acids KGAPE519, Arg615, Gly618, Arg620, and Lys712-Lys713 are not essential for nucleotide binding, although changes to Lys515 diminished Ca2+ transport activity but not phosphoenzyme formation. Changes of Gly626 and Asp627 abolished phosphoenzyme formation with both ATP and Pi, indicating that these residues may contribute to the conformation of the catalytic center.     

Molecular cloning of cDNA encoding the Ca2+ release channel (ryanodine receptor) of rabbit cardiac muscle sarcoplasmic reticulum

We have cloned and sequenced cDNA encoding the Ca2+ release channel (ryanodine receptor) of rabbit cardiac muscle sarcoplasmic reticulum. The cDNA, 16,532 base pairs in length, encodes a protein of 4,969 amino acids with a Mr of 564,711. The deduced amino acid sequence is 66% identical with that of the skeletal muscle ryanodine receptor, but analysis of predicted secondary structures and hydropathy plots suggests that the two isoforms exhibit the same topology in both transmembrane and cytoplasmic domains. A potential ATP binding domain was identified at residues 2619-2652, a potential phosphorylation site at residue 2809, and potential calmodulin binding sites at residues 2775-2807, 2877-2898, and 2998-3016. We suggest that a modulator binding domain in the protein lies between residues 2619 and 3016. Northern blot analysis of mRNA from a variety of tissues demonstrated that the cardiac isoform is expressed in heart and brain, while the skeletal muscle isoform is expressed in both fast- and slow-twitch muscle. No ryanodine receptor mRNA was detected in extracts from smooth muscle or any other non-muscle tissue examined. The two receptors are clearly the products of separate genes, and the gene encoding the cardiac muscle ryanodine receptor was localized to chromosome 1.     

The interaction of ethanol with reconstituted synaptosomal plasma membrane Ca2+ -ATPase

The primary effect of ethanol is on the central nervous system. However, the molecular mechanisms responsible for the physiological symptoms of ethanol intoxication are still unknown. Low concentrations of ethanol were observed to stimulate the activity of the calcium pump from reconstituted synaptosomal plasma membrane Ca2+ -ATPase (PMCA), and ethanol inhibited Ca2+ -ATPase activity at concentrations above 5%. The greatest stimulating effect was obtained with 5% (v/v) ethanol and was lipid-dependent, being 74% when the protein had been reconstituted in phosphatidylcholine (PC) and less when the reconstituted protein had previously been activated by calmodulin or after removal of a 9-kDa autoinhibitory site by controlled trypsinization. Stimulation of the pump by ethanol was lower for the native or trypsin-digested protein in the presence of phosphatidylserine than in PC. These results suggest a direct ethanol-protein interaction, because the activating effect depended on the state of Ca2+ -ATPase (native or truncated, or in presence of calmodulin). The activating mechanism of ethanol may involve opening an autoinhibitory domain located close to the calmodulin binding domain.     

Photoaffinity labeling study of the interaction of calmodulin with the plasma membrane Ca2+ pump.

Bovine brain calmodulin was labeled with synthetic peptides corresponding to the calmodulin-binding domain of the erythrocyte plasma membrane Ca(2+)-ATPase. One 20-amino acid peptide and two 28-amino acid peptides were used, carrying L-4'-(1-azi-2,2,2-trifluoroethyl)phenylalanine residues in position 9 (peptides C20W* and C28W*) and position 25 (peptide C28WC*), respectively. The localization of the contact regions between calmodulin and the N- and C-terminal portions of the peptides was the aim of this study. The three peptides were N-terminally blocked with a 3H-labeled acetyl group to facilitate the identification of labeled fragments after isolation and digestion. The binding site for phenylalanine 25 was identified in the N-terminal domain of calmodulin while the phenylalanine derivative in position 9 labeled the C-terminal domain. Fluorescence studies using the dansylated N- and C-terminal halves of calmodulin and peptide C20W corresponding to the first 20 amino acids of the calmodulin-binding domain showed that only the C-terminal lobe of calmodulin had high affinity for the peptide (KD in the nanomolar range).     

Action of long-chain fatty acids in vitro on Ca2+-stimulatable, Mg2+-dependent ATPase activity in human red cell membranes.

Human red cell membrane Ca2+-stimulatable, Mg2+-dependent adenosine triphosphatase (Ca2+-ATPase) activity and its response to thyroid hormone have been studied following exposure of membranes in vitro to specific long-chain fatty acids. Basal enzyme activity (no added thyroid hormone) was significantly decreased by additions of 10(-9)-10(-4) M-stearic (18:0) and oleic (18:1 cis-9) acids. Methyl oleate and elaidic (18:1 trans-9), palmitic (16:0) and lauric (12:0) acids at 10(-6) and 10(-4) M were not inhibitory, nor were arachidonic (20:4) and linolenic (18:3) acids. Myristic acid (14:0) was inhibitory only at 10(-4) M. Thus, chain length of 18 carbon atoms and anionic charge were the principal determinants of inhibitory activity. Introduction of a cis-9 double bond (oleic acid) did not alter the inhibitory activity of the 18-carbon moiety (stearic acid), but the trans-9 elaidic acid did not cause enzyme inhibition. While the predominant effect of fatty acids on erythrocyte Ca2+-ATPase in situ is inhibition of basal activity, elaidic, linoleic (18:2) and palmitoleic (16:1) acids at 10(-6) and 10(-4) M stimulated the enzyme. Methyl elaidate was not stimulatory. These structure-activity relationships differ from those described for fatty acids and purified red cell Ca2+-ATPase reconstituted in liposomes. Thyroid hormone stimulation of Ca2+-ATPase was significantly decreased by stearic and oleic acids (10(-9)-10(-4) M), but also by elaidic, linoleic, palmitoleic and myristic acids. Arachidonic, palmitic and lauric acids were ineffective, as were the methyl esters of oleic and elaidic acids. Thus, inhibition of the iodothyronine effect on Ca2+-ATPase by fatty acids has similar, but not identical, structure-activity relationships to those for basal enzyme activity. To examine mechanisms for these fatty acid effects, we studied the action of oleic and stearic acids on responsiveness of the enzyme to purified calmodulin, the Ca2+-binding activator protein for Ca2+-ATPase. Oleic and stearic acids (10(-9)-10(-4) M) progressively inhibited, but did not abolish, enzyme stimulation by calmodulin (10(-9) M). Double-reciprocal analysis of the effect of oleic acid on calmodulin stimulation indicated noncompetitive inhibition. Addition of calmodulin to membranes in the presence of equimolar oleic acid restored basal enzyme activity. Oleic acid also reduced 125I-calmodulin binding to membranes, but had no effect on the binding of [125I]T4 by ghosts. The mechanism of the decrease by long chain fatty acids of Ca2+-ATPase activity in situ in human red cell ghosts thus is calmodulin-dependent and involves reduction in membrane binding of calmodulin.     

Identification of the Plasma Membrane Ca2+-ATPase and of Its Autoinhibitory Domain

The effect of controlled proteolysis on the plasma membrane (PM)Ca2+-ATPase was studied at the molecular level in PM purified from radish (Raphanus sativus L.) seedlings. Two new methods for labeling the PM Ca2+-ATPase are described. The PM Ca2+-ATPase can be selectively labeled by treatment with micromolar fluorescein isothiocyanate (FITC), a strong inhibitor of enzyme activity. Both inhibition of activity and FITC binding to the PM Ca2+-ATPase are suppressed by millimolar MgITP. The PM Ca2+-ATPase maintains the capability to bind calmodulin also after sodium dodecyl sulfate gel electrophoresis and blotting; therefore, it can be conveniently identified by 125l-calmodulin overlay in the presence of calcium. With both methods a molecular mass of 133 kD can be calculated for the PM Ca2+-ATPase. FITC-labeled PM Ca2+-ATPase co-migrates with the phosphorylated intermediate of the enzyme[mdash]labeled by incubation with [[gamma]-32P]GTP in the presence of calcium[mdash]on acidic sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Controlled trypsin treatment of purified PM determines a reduction of the molecular mass of the PM Ca2+-ATPase from 133 to 118 kD parallel to the increase of enzyme activity. Only the 133-kD but not the 118-kD PM Ca2+-ATPase binds calmodulin. These results indicate that trypsin removes from the PM Ca2+-ATPase an autoinhibitory domain that contains the calmodulin-binding domain of the enzyme.     

The inhibitor peptide of the mitochondrial F1.F0-ATPase interacts with calmodulin and stimulates the calmodulin-dependent Ca2+-ATPase of erythrocytes

The binding of calmodulin to the mitochondrial F1.F0-ATPase has been studied. [125I]Iodoazidocalmodulin binds to the epsilon-subunit and to the endogeneous ATPase inhibitor peptide in a Ca2+-dependent reaction. The effect of the mitochondrial ATPase inhibitor peptide on the purified Ca2+-ATPase of erythrocytes has also been analyzed. The inhibitor peptide stimulates the ATPase when pre-incubated with the enzyme. The activation of the Ca2+-ATPase by calmodulin is not influenced by the inhibitor peptide, indicating that the two mechanisms of activation are different. These in vitro effects of the two regulatory proteins may reflect a common origin of the two ATPases considered and/or of the regulatory proteins.     

Gene knockout studies of Ca2+-transporting ATPases.

The biochemical functions of intracellular and plasma membrane Ca2+-transporting ATPases in the control of cytosolic and organellar Ca2+ levels are well established, but the physiological roles of specific isoforms are less well understood. There appear to be three different types of Ca2+ pumps in mammalian tissues: the sarco(endo)plasmic reticulum Ca2+-ATPases (SERCAs), which sequester Ca2+ within the endoplasmic or sarcoplasmic reticulum, the plasma membrane Ca2+-ATPases (PMCAs), which extrude Ca2+ from the cell, and the putative secretory pathway Ca2+-ATPase (SPCA), the function of which is poorly understood. This review describes the results of recent analyses of mouse models with null mutations in the genes encoding SERCA and PMCA isoforms and genetic studies of SERCA and SPCA dysfunction in both humans and model organisms. These studies are yielding important insights regarding the physiological functions of individual Ca2+-transporting ATPases in vivo.