Mammalian Calcium Pumps in Health and Disease

Marisa Brini and Ernesto Carafoli

Chapter Outline

Sarco/Endoplasmic Reticulum Ca²⁺ ATPase

Secretory Pathway Ca²⁺ ATPase

Plasma Membrane Ca²⁺ ATPase

Ca²⁺ Pumps in the Disease Process

Conclusions

Ca²⁺-transporting adenosine triphosphatases (ATPases; Ca²⁺ pumps) have been described in animal and plant cells and in cells of lower eukaryotes. This chapter will focus on the ATPases of animal cells and on the disease processes linked to their dysfunction. The three animal Ca²⁺ pumps belong to the large superfamily of P-type ATPases, which have been so defined because their reaction cycle is characterized by the formation of an acid-stable phosphorylated Aspartate (Asp) residue (the P intermediate) in a highly conserved sequence (SDKTGT[L/IV/M][T/I/S]).¹ The family now contains hundreds of members and eight subfamilies.² The subfamilies have been identified based essentially on transported substrate specificity, the evolutionary appearance of which having been accompanied by abrupt changes in sequence. The changes, however, do not involve eight conserved structurally and mechanistically important regions that define the core of the superfamily. Five branches have been identified in the phylogenetic tree of the superfamily: two animal Ca²⁺ pumps belong to subgroup II A (the sarco/endoplasmic reticulum Ca²⁺ [SERCA] and secretory pathway Ca²⁺ [SPCA] pumps), one to subgroup II B (the plasma membrane Ca²⁺ [PMCA] pump). All P-type ATPases, including the three that transport Ca²⁺ in animal cells, are multidomain proteins that share the essential properties of the reaction mechanism, have molecular masses varying between 70 and 150 kDa, and share the presence of 10 hydrophobic transmembrane (TM) spanning domains (however, some have only six or eight). The number of TMs being even and the N- and C-termini of all P-type pumps are on the same membrane side (i.e., the cytosol); one exception is a splice variant of the SERCA pump that has 11 TM). The P-type ATPases also share the sensitivity to the transition state analog orthovanadate and, with some specific differences (see below), to La³⁺. Other inhibitors only affect selected members of the superfamily. The three-dimensional (3D) structures of four P-type ATPases have become available following the landmark solution of the 3D structure of the SERCA pump 12 years ago³: molecular modeling on templates of the SERCA pump structure has indicated that all P-type ATPases share the general principles of 3D structure. The reaction cycle of P-type ATPases originally envisaged only the E1 and E2 steps, characterized by distinct conformations and affinities for adenosine triphosphate (ATP) and the transported ion. For example, Ca²⁺ pumps in the E1 state engage Ca²⁺ with high affinity at one side of the membrane, and in the state their E2 lowered affinity for Ca²⁺ releases it to the opposite membrane side.⁴ Later, additional intermediate states were added that made the reaction cycle much more complex, but the basic E1/E2 nomenclature has been retained. Importantly, each step of the reaction cycle is reversible, so that ATP can be produced by reversing the direction of the ion transport process. Reversal of the SERCA pump, with production of ATP, had in fact already been demonstrated in one of the first experiments on the transport of Ca²⁺ by vesicular preparations of sarcoplasmic reticulum.⁵ A simplified version of the cycle, but adapted to Ca²⁺ pumps, is shown in Figure 5-1.

Figure 5-1 A simplified reaction cycle of the P-type adenosine triphosphatases (pumps) adapted to the Ca²⁺ pumps. The two original conformational states of the pumps are envisaged. The E1 pumps bind Ca²⁺ with high affinity at one membrane side (the cytosol), the E2 pumps have much lower Ca²⁺ affinity, and release Ca²⁺ to the opposite membrane side. Adenosine triphosphate phosphorylates a conserved Asp in the active site.

Several Ca²⁺ pump isoforms have been described in animal cells, differing essentially in tissue distribution, regulatory properties, and some mechanistic peculiarities. The isoform diversity reflects the existence of separate basic gene products, but also the occurrence of complex patterns of alternative splicing that increase very significantly the number of variants of each of the three pumps. The analysis of the differential properties of the Ca²⁺ pump isoforms is a vigorously investigated topic that has important linkages to the general process of cellular Ca²⁺ homeostasis, which in animal cells is regulated by a number of nonmembrane Ca²⁺-binding proteins and of membrane-intrinsic Ca²⁺ channels and transporters. The transporters interact with Ca²⁺ with high or low affinity, and thus function either as fine tuners of cytosolic Ca²⁺ or come into play whenever the concentration of Ca²⁺ increases to levels adequate for their low affinity. The Na/Ca-exchanger of the plasma membrane and the mitochondrial Ca²⁺ uptake and release systems are the low-affinity regulators of cytosolic Ca²⁺. The three pumps, by contrast, control Ca²⁺ efficiently even in the low concentrations of the cytosol at rest. Pump activity is fundamental to the correct functioning of the machinery of animal cells: dysfunctions, genetic or otherwise, of their operation might not necessarily induce cell death, but invariably generate disease phenotypes.

Sarco/Endoplasmic Reticulum Ca²⁺ ATPase

The SERCA pump is a key mechanism to adjust the Ca²⁺ homeostasis in the endoplasmic reticulum (ER) lumen. Considering that the ER Ca²⁺ is involved in a multitude of signaling events and in housekeeping functions that control cell growth, differentiation and apoptosis, the activity of the SERCA pump is a key element in cell wellness.

The SERCA pump is inhibited by La³⁺ and orthovanadate, and the discovery of specific inhibitors such as thaspigargin,⁶ cyclopiazonic acid,⁷ and 2.5-di(t-butyl)hydroquinone⁸ represented a significant advantage in the biochemical and structural characterization of the pump.

The SERCA protein is organized in the membrane with 10 TMs: numerous mutagenesis studies and the solution of its 3D structure have clarified essential molecular details of its function, which will be summarized in this chapter. Full details are available in a number of more comprehensive reviews.9–11

Analysis of the 3D structure of the SERCA1 pump isoform has revealed that the single polypeptide chain folds in three cytosolic domains and in one transmembrane sector (M) composed of the 10 formerly predicted TMs (Figure 5-2). The three cytosolic domains have been named according to their role in the reaction cycle: the nucleotide binding domain (N) binds ATP, the phosphorylation domain (P) drives ATP hydrolysis leading the phosphorylation of the catalytic Asp, and the actuator domain (A) catalyzes the dephosphorylation of the P-domain. The A and P domains are connected to the transmembrane M domain that contains the 2 Ca²⁺ binding sites: the SERCA pump transports 2 Ca²⁺ per ATP hydrolyzed. The N domain is instead connected to the P domain. During the cycle, phosphorylation and dephosphorylation events promote conformational changes that control the access of Ca²⁺ to the two binding sites (site I and site II), which exist in high- and low-affinity states (Figure 5-3). The two sites are located near the cytoplasmic surface of the membrane, but site I faces the cytoplasmic side and site II is closer to the luminal side. Once Ca²⁺ becomes bound to site I, a conformational change increases the affinity of site II and permits the phosphorylation of the catalytic Asp by ATP, leading to the transition E2→E1→E1•2Ca²⁺ E1P. The binding of ATP crosslinks the P and N domains, permitting the interaction of P domain with the A domain, which rotates inducing the opening of the luminal gate that releases Ca²⁺ to the lumen and permits the E1P-E2P transition. The closure of luminal gate, and thus the E2P→E2 P_i transition, then occurs because a second rotation of the A domain locks it to the P domain. A highly conserved TGES motif in the A domain fills the gap between the N and P domains after the second rotation of the A domain, eventually permitting the release of Ca²⁺ into the lumen. The rearrangements of transmembrane helices M1-M6 induced by the rotation of the A domain allow protons and water molecules to enter and stabilize the empty Ca²⁺ binding sites. The rearrangements also induce the retraction of the TGES from the phosphorylation site and the entrance of one water molecule to the phosphorylation site, inducing the release of phosphate (and Mg²⁺) and the complete closure of the luminal gate.

Figure 5-2 The three-dimensional structure of the SERCA pump, showing the open configuration of the three cytoplasmic domains A, N, and P in the presence of Ca²⁺, and the more compact configuration of the cytosolic sector in its absence. The two purple spheres in the upper panel represent the two bound Ca²⁺. The E2 structure shown contains the inhibitor thapsigargin (TG). The TMs are the transmembrane domains. Several residues of importance not discussed in the text are also shown. (Modified from Toyoshima C: Structural aspects of ion pumping by Ca2+-ATPase of sarcoplasmic reticulum. Arch Biochem Biophys 476:3–11, 2008.)

Figure 5-3 The conformational changes of the cytosolic and membrane domains of the SERCA pump during the transport of Ca²⁺ across the molecule. (Modified from Toyoshima C: Structural aspects of ion pumping by Ca2+-ATPase of sarcoplasmic reticulum. Arch Biochem Biophys 476:3–11, 2008.)

Three SERCA genes generate three isoforms. Their number is increased by alternative splicing processes. SERCA1 is almost exclusively expressed in muscle tissues, specifically in fast-twitch skeletal muscles. Interestingly, the generation of truncated and less active SERCA1 variants has been described, contribute to reduce the Ca²⁺ concentration in the ER lumen and cause apoptotic cell death. SERCA2b and SERCA2a are the two major SERCA protein isoforms, with the former having a housekeeping function and the latter having a more specialized function. SERCA2a is found in slow skeletal and cardiac muscles; it is also expressed in low amounts in smooth muscles and in neurons. The splice variants SERCA2c and SERCA2d have also been found in low amounts in the heart. The expression of the SERCA3 pump, which has a limited cell-type distribution, is variable. In several cell types this expression is induced by differentiation, and it is decreased during tumorigenesis and blastic transformation. In cells of hematopoietic origin and in various epithelial cells, the SERCA2 pump gene is coexpressed with that of the SERCA3 pump. All SERCA3 splice variants (SERCA3a-f) have lower Ca²⁺ affinity than the SERCA2 pump, which raises doubts on their role in the presence of higher affinity SERCA pump variants. Differences in their spatial cellular distribution could justify their copresence. The SERCA3 pump is confined to environments with high Ca²⁺ concentration, such as those close to the plasma membrane of cardiomyocytes, at basal regions in epithelial cells, and in the membrane of acidic Ca²⁺ stores in platelets. Although the SERCA1a pump is the best characterized isoform in terms of structure-function relationship, the regulatory aspects of SERCA pumps have instead been better defined on the SERCA2 isoform.

The SERCA2a pump is the major isoform of developing and adult mammalian heart (SERCA2b is also expressed there, its level being unchanged during the development).¹² SERCA2a is the most abundant protein in the heart membrane; its increased sarcoplasmic reticulum (SR) expression during the development paralleled the increasing rate of Ca²⁺ uptake by the SR lumen and the shortening of the relaxation time in adults, with respect to neonatal heart. SERCA2a pump levels are higher in atria than in ventricles, partially accounting for the shorter duration of contraction in atrial than in ventricular tissues.

The expression and the activity of the SERCA2a pump have been studied extensively. The primary mechanism of the regulation of the pump is mediated by phospholamban (PLB).¹³ PLB is a 52-residue protein composed of a hydrophobic helical C-terminal portion inserted into the SR membrane and a hydrophilic N-terminal region that protrudes into the cytosol and contains phosphorylation sites (Ser16 and Thr 17) for PKA and, possibly, CAMKII. Dephosphorylated PLB binds to the pump, decreasing its Ca²⁺ affinity; phosphorylation by PKA, and possibly CaMKII, releases the inhibition and increases the affinity of the pump for Ca²⁺ and thus Ca²⁺ transport. The hydrophilic N-terminal portion of PLB interacts with a domain close to the active site of the pump and, within the membrane, with transmembrane helices 2, 4, 6, and 9. It is generally believed that PLN exists both as a pentamer and a monomer. It is not clear how the conversion between the two forms occurs, but several lines of evidence indicate that monomeric PLB is the active form. However, structural observations indicate that pentamers can also interact with the pump.¹⁴

Another small (31 residues) transmembrane protein, sarcolipin (SLN), originally identified as it copurified with the SERCA1a pump, has recently also received attention. SLN is predominantly expressed in the atrial compartment of the heart, and its sequence is similar to that of the transmembrane sector of PLN. Modeling studies have revealed that the SLN and PLB interactions with the SERCA pump may be similar—that is, SLN would bind to the pump in transmembrane sites of PLB.¹⁴ Studies on SLN are still limited, but SLN overexpression in ventricle cells of animal models, where SLN is essentially absent, caused a decrease in the Ca²⁺ affinity of SERCA2a and slowed relaxation, suggesting that SLN is as effective an inhibitor of SERCA pump activity as PLB. Interestingly, the overexpression of SLN in the heart of PLB-null mice caused a decrease in the affinity of the SERCA2a pump for Ca²⁺ and impaired contractility. The finding that isoproterenol, a β-adrenergic agonist, relieved the inhibition suggests that SLN could mediate the β-adrenergic response in the heart. Ablation of the SLN gene increases the affinity of the SERCA pump for Ca²⁺, resulting in enhanced rates of SR Ca²⁺ uptake.¹⁵

The SERCA2b pump is the acknowledged housekeeping isoform, and it has a dual role. By transporting Ca²⁺ from the cytosol to the ER lumen, the SERCA2b pump contributes to the maintenance of cytosolic Ca²⁺ at the low resting levels (≈100 nM), at the same time ensuring the high Ca²⁺ levels (in the mM range) in the lumen of ER that make the ER the main intracellular Ca²⁺ store that controls cellular activities (e.g., contraction, proliferation, differentiation, cell death). The pump also ensures the proper internal Ca²⁺ ambient for the ER enzymes (e.g., those involved in protein folding and lipid synthesis).

The SERCA2b pump shares 85% sequence identity with the SERCA1a counterpart but differs functionally from it and from the SERCA2a isoform, which is characterized by a unique C-terminal extension—the so-called 2b-tail, which forms a luminal sequence extension and an additional TM segment (TM11).¹⁶ The extension has regulatory properties: the longer pump has a twofold higher affinity for cytosolic Ca²⁺ and a lower maximal turnover rate. According to a model based on the SERCA1a structure, and on the NMR structure of TM11, the interaction of TM11 with TM7 and TM10 has been proposed to stabilize the pump in the Ca²⁺-bound E1 conformation, with the high-affinity Ca²⁺ binding sites facing the cytosol. The TM11 has also been proposed as a novel regulator of the SERCA pump. The co-reconstitution of the 18-residue long TM11 with the SERCA1a protein in vitro reduced the maximum reaction rate (V_max) and increased the Ca²⁺ affinity of the latter. The regulation of the SERCA pump by PLB and SLN, and by the 2b-tail in the 2b variant, resembles the interaction of the β- and γ-subunits of the Na⁺/K⁺-ATPase with the catalytic subunit α. The regulation also resembles the regulatory interaction of the PMCA pump with calmodulin, suggesting operationally similar molecular mechanisms of P-type pump regulation.¹⁷ In addition, the C-terminal portion of the ePMCa pump can to some extent replace PLB as an inhibitor of the SERCA pump. Apart PLB and SLN, which are the best-studied regulators of the SERCA pump, other proteins interact with the pump. Those interacting at the luminal side have an important role. The two chaperones calreticulin and calnexin contain a globular N-domain that binds carbohydrates—an extended P-domain that mediates the binding of ERp57 and an acidic C-terminal domain, that in the case of calreticulin binds 25 mol of Ca²⁺ per molecule of protein with low affinity (Kd, 2 mM). Luminal Ca²⁺ buffering by calnexin is less significant, and the acidic C-terminus of the protein protrudes into the cytosol. Calreticulin and calnexin have been suggested to interact through their N-domain with a putative glycosylated residue present in the C-terminal tail of the SERCA2b pump, but absent in SERCA2a. Glycosylation-independent interaction with the SERCA2b pump has been shown to occur, and molecular modeling of the SERCA2b molecule has suggested that its 2b-tail is located in luminal loops, thus making it inaccessible to interacting proteins. ERp57, a member of the PDI family of proteins with thio-oxidoreductase activity, is recruited by the SERCA2b pump–chaperone complex to catalyze the formation of an inhibitory disulfide bridge between Cys875 and Cys887 in the luminal loop L7-8 of the SERCA2b pump.¹⁸ The formation of a disulfide bridge could be a regulatory mechanism, but the proposal is controversial, because mutations of either or both Cys residues resulted in the loss of Ca²⁺ transport but not of the activity in SERCA1.¹⁹

Two additional luminal Ca²⁺ binding proteins have been shown to interact with the SERCA2 pump: the ubiquitously expressed calumenin and the histidine-rich Ca²⁺-binding protein (HRC). Both decrease the apparent Ca²⁺ affinity of the pump. HRC binds Ca²⁺ with high capacity and low affinity and could mediate both SR Ca²⁺ uptake and release through its interaction with SERCA, when the SR Ca²⁺ is low, and with triadin, which is part of the RyR Ca²⁺ release complex when it is saturated by Ca²⁺.²⁰

Secretory Pathway Ca²⁺ ATPase

The Golgi Ca²⁺ ATPase (the SPCA pump) shares with the SERCA pump the role of loading Ca²⁺ in the Golgi apparatus.21,22 The relative contribution of the SPCA and SERCA pumps to the total Ca²⁺ uptake in the Golgi apparatus is cell-type dependent. The use of the SERCA pump selective inhibitor thapsigargin, which does not inhibit the SPCA pump, has confirmed that the SERCA pump contribution is significant in numerous cell types, but not in keratinocytes that mainly use the SPCA pump. This point is interesting because SPCA pump mutations that impair the function of the pump lead to a specific human disease, Hailey-Hailey disease (discussed later). In contrast to the SERCA pump, the SPCA pump also transports Mn²⁺ and thus serves the dual function of supplying this metal to the Golgi lumen, where it acts as a cofactor for the glycosyl-transferases, and to remove it from the cytosol, where its accumulation could be toxic. The Golgi Ca²⁺ ATPase was originally discovered in yeast and named Pmr1 (plasma membrane ATPase-related) and PMA1 (plasma membrane H⁺-ATPase) by two independent groups. Later, it was detected in mammals and higher vertebrates.

Two basic SPCA genes have been found in higher vertebrates coding for the SPCA1 and SPCA2 pumps. Alternative splicing of the primary transcript has been described for SPCA1, but not for the SPCA2 pump.

The SPCA1 pump is the housekeeping Ca²⁺ and Mn²⁺ pump, because it is expressed in all cell types, even if with different levels in various tissues. The alternatively spliced SPCA1 mRNA generates SPCA1a-d proteins, which differ in the C-terminal portion. The expression of SPCA2 pump in human tissues is more restricted than that of the SPCA1 pump, suggesting a more specialized physiologic function for it. The SPCA2 pump is particularly abundant in the gastrointestinal tract, trachea, prostate, and thyroid, salivary, and mammary glands.

The SPCA pump is also predicted to be organized in 10 TMs, with a large headpiece portion protruding into the cytosol. On the basis of the SERCA pump 3D structure, the cytosolic portion is divided in the three canonical A, P, and N portions (Figure 5-4). The SPCA pump is shorter than the SERCA and PMCA pumps, because it does not have the long cytoplasmic C-terminal tail of the PMCA pump, and its intracellular loops are shorter. The SPCA pump differs from the SERCA pump in having only one Ca²⁺ binding site, corresponding to site II in the SERCA1 pump. The suggested TM packing and possibly some distant residues would define the structure of the site, making it adequate to bind Mn²⁺ with high affinity and to transport it. This structural aspect is a peculiarity of the SPCA pump. Gln783 in TM6 and Val335 in TM4 appear to be essential for Mn²⁺ transport, because their mutation (in the Pmr1 yeast pump) abolished Mn²⁺ transport. Another distinction of the SPCA pump with respect to the SERCA and PMCA pumps, which function as obligatory H⁺ exchangers, is the finding that it does not appear to countertransport H⁺. No specific SPCA pump inhibitors have been described, and no endogenous activators are known. Mutagenesis studies on the Pmr1 yeast pump have generated phenotypes that are important tools that could provide information on the role of the SPCA pump in higher eukaryotic organisms. The expression levels and activity of SPCAs change in response to altered physiologic needs. SPCA1 pump expression and activity changed in the brain after an ischemic event,²³ and the SPCA1 pump levels increased in mammary gland during lactation. This last finding is shared by the PMCA2 pump (discussed later). The role of the SPCA pump in the maintenance of Golgi Ca²⁺ homeostasis deserves a comment, because recent evidence have indicated that the Golgi apparatus is a heterogeneous and highly dynamic Ca²⁺ store.²⁴ The apparatus has been shown to be an InsP3-sensitive Ca²⁺ store, implying a role for it in the generation of local cytosolic Ca²⁺ signals. The specific distribution of the SPCA pump in the Golgi membranes appears important; it can vary with the cell type (i.e., in some cases it colocalizes with Golgi markers and in others with those of the trans-Golgi). However, a consensus has now been reached that the SPCA1 pump–containing Golgi subcompartment is insensitive (or mildly sensitive) to InsP3, and thus appears not to be involved in generating cytosolic Ca²⁺ signals. The kinetics of Ca²⁺ release from the Golgi apparatus differed from those of the ER. In particular, although the latency following agonist application and the initial rate of Ca²⁺ release were similar for the two organelles, the release of Ca²⁺ from the Golgi apparatus terminates faster than that from the ER. These findings would be compatible with distinct Ca²⁺ subcompartments in the Golgi apparatus endowed with differentCa²⁺ regulating molecular components.