The Plasma Membrane Calcium Pump: New Ways to Look at an Old Enzyme

The three-dimensional structure of the PMCA pump has not been solved, but its basic mechanistic properties are known to repeat those of the other Ca2+ pumps. However, the pump also has unique properties. They concern essentially its numerous regulatory mechanisms, the most important of which is the autoinhibition by its C-terminal tail. Other regulatory mechanisms involve protein kinases and the phospholipids of the membrane in which the pump is embedded. Permanent activation of the pump, e.g. by calmodulin, is physiologically as harmful to cells as its absence. The concept is now emerging that the global control of cell Ca2+ may not be the main function of the pump; in some cell types, it could even be irrelevant. The main pump role would be the regulation of Ca2+ in cell microdomains in which the pump co-segregates with partners that modulate the Ca2+ message and transduce it to important cell functions.

The three-dimensional structure of the PMCA pump has not been solved, but its basic mechanistic properties are known to repeat those of the other Ca 2؉ pumps. However, the pump also has unique properties. They concern essentially its numerous regulatory mechanisms, the most important of which is the autoinhibition by its C-terminal tail. Other regulatory mechanisms involve protein kinases and the phospholipids of the membrane in which the pump is embedded. Permanent activation of the pump, e.g. by calmodulin, is physiologically as harmful to cells as its absence. The concept is now emerging that the global control of cell Ca 2؉ may not be the main function of the pump; in some cell types, it could even be irrelevant. The main pump role would be the regulation of Ca 2؉ in cell microdomains in which the pump co-segregates with partners that modulate the Ca 2؉ message and transduce it to important cell functions.
Decades of study on the mammalian Ca 2ϩ -ATPase of the plasma membrane (the PMCA pump) (1) have revealed many properties that set it apart from the other members of the superfamily of P-type pumps (2). One property that is immediately obvious is the long C-terminal tail, which is the locale of the numerous regulatory processes. Another distinctive property of the pump is the wealth of interactors with which the pump triggers processes of general significance for the cell. The basic information on these and other general aspects of the PMCA pump has been collected in other comprehensive reviews (3)(4)(5) and will not be repeated here. In this minireview, we instead look at the pump in a novel perspective, in which the extrusion of Ca 2ϩ will not be considered as a mechanism to regulate bulk cell Ca 2ϩ but will be integrated in a complex array of processes that modulate Ca 2ϩ -dependent processes within the cell (6). We focus on the most important regulatory mechanisms of the pump. We posit that, in many cell types, the quantitative contribution of the pump to the total extrusion of Ca 2ϩ could be minor or even irrelevant. Because the main role of the pump would be to do something else, the conclusion would offer a rationale for the puzzling finding that, in some plasma membranes, the pump coexists with a much more powerful Ca 2ϩ extrusion system, e.g. Na ϩ /Ca 2ϩ exchange in cardiomyocytes (7). Naturally, the function of the pump is still essential to the well being of the cell, as underscored by the causative involvement of its malfunction in disease processes (8).
The pump contains 10 transmembrane domains, two main cytosolic loops, and a long cytosolic C-terminal tail (9,10). Separate genes encode its four basic isoforms: PMCA1 is ubiquitous and has a housekeeping role; PMCA4 is also ubiquitous but is endowed with tissue-specific roles; and PMCA2 and PMCA3 are tissue-restricted, with high levels of expression in neurons. Complex alternative splicing processes at a site in the first cytosolic loop of the pump (site A) and within the C-terminal calmodulin-binding domain (CaM-BD 3 ; site C) generate numerous pump variants with special properties. The best understood regulatory mechanism of the pump is that by CaM, which interacts with high affinity with the C-terminal tail, reducing the K m(Ca2ϩ) of the pump to submicromolar values. In the absence of CaM, the CaM-BD binds to the main body of the pump to keep it (auto)inhibited (11,12). CaM removes the domain from these sites, restoring activity. (A second, lower affinity CaM-BD has recently been identified downstream of the first in some splice variants of the pump (13).) The autoinhibition mechanism is unique among P-type pumps: other pumps of the superfamily are also inhibited in the resting state, but the mechanism involves extraneous proteins, e.g. phospholamban in the case of the SERCA pump.
Studies that preceded the purification of PMCA found that it is stimulated by phospholipids (PL). Treatment of red cell ghosts with phospholipases reduced its activity, which was restored by PL. According to one study (14), all PL were effective, whereas another study (15) found that only acidic PL (phosphatidylserine) were effective. The specificity of acidic PL was confirmed on the purified pump (16). They reduced its K m(Ca2ϩ) to even lower values than CaM (17), and it was calculated that the concentration of phosphatidylserine in the inner leaflet of the plasma membrane bilayer was, in principle, sufficient for 50% of maximal pump activity (18). A particularly effective PL was the doubly phosphorylated product of phosphatidylinositol (PIP 2 ) (19). The PL effect still has obscure aspects that will be discussed in detail below, but it has always been considered potentially important; for example, a reversible modulation of pump activity was suggested for PIP 2 (19), which is the only PL that undergoes rapid agonist-induced changes in concentration.
Protein kinases also modulate the activity of the pump by phosphorylating target residues in its C-terminal tail: PKA phosphorylates a Ser residue downstream of the CaM-BD, increasing the Ca 2ϩ affinity of the PMCA1 pump (20). The work on PKC has yielded conflicting results, particularly because the original finding had shown that the effects of phosphorylation and of CaM were additive (21), whereas others had found that the phosphorylation decreased the stimulation by CaM (22). The target residues for PKC are a Thr residue in the CaM-BD and a Ser residue farther downstream present in all pump isoforms. Surprisingly, the phosphorylation by PKC was found to increase the V max of the pump, but not its K m for Ca 2ϩ (22). It was later found that phosphorylation of the Thr residue within the CaM-BD reduced its ability to autoinhibit the pump (23). A complicating aspect of the activation by the kinases is the requirement for the inositol 1,4,5-trisphosphate (IP 3 ) and diacylglycerol (DAG) and thus for the hydrolysis of PIP 2 , which is itself strongly activatory. Therefore, the activation by the kinases should, in principle, be counteracted by the disappearance of PIP 2 . Possibly, temporal or/and spatial aspects in the activation could provide a way out of the conundrum.
Activation of the pump by calpain involves the shaving off of the C-terminal tail of the pump (24), which becomes constitutively activated and CaM-independent. This irreversible mechanism could come into play in conditions that demand a permanent increase in the activity of the pump, but it is more likely to have a role in conditions of pathology that produce cytosolic Ca 2ϩ overload.
Other described mechanisms of PMCA activity modulation are mechanistically interesting but may not have a physiological role: that by the dimerization (oligomerization) of the pump (25) through its CaM-BD (26); that linked to the membrane protein concentration (27); and that involving the association of the PMCA pump with the actin cytoskeleton (28), with G-actin being activatory and F-actin inhibitory. Interestingly, two possibly allosteric Ca 2ϩ -binding sites with nanomolar and micromolar affinities, respectively, have been found at the two sides of the C-terminal CaM-BD (29).

A General Comment on Regulation of the PMCA Pump
Activation of the pump by protein kinases implies a built-in stop mechanism: the phosphorylation by PKA is activated by the liberation of Ca 2ϩ from the endoplasmic reticulum by IP 3 because most adenylyl cyclase isoforms are Ca 2ϩ -dependent, and the phosphorylation by PKC is activated by the production of DAG. Both mechanisms demand the hydrolysis of PIP 2 and thus, in principle, the removal of its activation of the pump. Phosphatases that dephosphorylate the pump also play a role in reversing the activation. The built-in stop mechanism in the activation by the kinases has more general significance. It also concerns the activation by CaM: once the latter activates the pump, Ca 2ϩ in the microenvironment of the enzyme decreases. As a result, CaM will leave the pump, restoring autoinhibition. Thus, one should conclude that the pump can be activated only in temporally limited bursts (apart, of course, from the case of proteolysis or, perhaps, of the long-range effect of the lipid environment; see below). The time factor is relevant; for example, once an IP 3 -linked agonist provokes the hydrolysis of PIP 2 , the pump should immediately become less active. Shortly thereafter, however, it will be activated again by Ca 2ϩ (directly via CaM and indirectly via PKA) and indirectly by DAG. The spatial factor could also have a role, i.e. the pump could be distributed inhomogeneously in the plasma membrane and/or recycled in a regulated way from membrane microdomains by endocytosis. The activation/inhibition effects could thus involve only a fraction of it.

Regulation of the PMCA Pump by Phospholipids
Membrane lipids have an obvious structural role, but they are also critical actors in signal transduction: membrane-linked signaling proteins may be sequestered in domains enriched with specific interacting lipids but can also be modulated by lipid-mediated changes in membrane fluidity that could influence their lateral mobility. Special membrane domains such as rafts, caveolae, and dendritic spines are indeed loci of dynamic signal transduction activity involving Ca 2ϩ . Concerning the response of the PMCA pump to acidic PL, its active core is located closer to the cytosolic side of the membrane. The asymmetric lipid distribution in the bilayer is thus of particular interest. Phosphatidylcholine and sphingomyelin, which do not activate the pump, are located predominantly in the outer leaflet, whereas phosphatidylethanolamine, the acidic phosphatidylserine (PS), phosphatidylinositol, and its two phosphorylated derivatives (PIP 1 and PIP 2 ) are located predominantly in the leaflet facing the cytoplasm. This asymmetry is maintained by three classes of enzymes: flippases, floppases, and scramblases (for an extended review, see Ref. 30). Flippases are P-type ATPases that perform the ATP-dependent transfer of amino-PL (especially PS) from the outer to the inner leaflet of the bilayer, whereas floppases catalyze their movement in the opposite direction. Scramblases dissipate the lipid asymmetry: they may reorganize PL in the plasma membrane in response to cell stimulation (31)(32)(33). They are specifically important in the externalization of PS on the cell surface, a process that occurs early in apoptosis to make cells recognized by phagocytes (34,35). Two types of scramblases have been described: one is Ca 2ϩ -independent (36), and the other is Ca 2ϩ -dependent (37). A mechanism could thus be envisaged in which scramblases could become activated to externalize PS, e.g. in response to the increase in cell Ca 2ϩ (38 -40), removing a mechanism for the activation the PMCA pump and thus increasing the Ca 2ϩ overload. The mechanism of the flip-flop operation is still poorly understood, but structural information on the enzymes that perform it now begins to shed light (41,42). Indications have also appeared for a possible role of caspases (43,44) and of transient receptor potential channels and the Rho kinase family (45,46) in the PS exposure process.
Acidic PL interact directly with the pump. Two binding sites have been identified: one in the first cytosolic loop and the other in the CaM-BD (47,48). The contribution of the two sites to activation is still poorly understood (49). However, inhibition of Ca 2ϩ efflux has been documented in umbilical vein endothelial cells exposed to treatments that increase PS externalization (50).
Changes in the PL environment of the pump in the rapid time scale of the events linked, for example, to the action of CaM or agonists will probably not occur. Thus, PL probably are longer range "buffers" of pump activity. One exception could be the strongly activatory PIP 2 (19), which is the only PL with a concentration that changes rapidly upon cell stimulation. A reversible rapid activation cycle had indeed been proposed for PIP 2 (19). However, as mentioned, the role of PIP 2 has additional complicating aspects (see below).
In addition to interacting directly with the pump, lipids of the plasma membrane could also influence its activity in indirect ways. Pump units could be sequestered in membrane domains of particular lipid composition and/or fluidity; the fluidity of the lipid environment indeed influences pump activity both in vitro and in vivo (51)(52)(53). Plasma membrane domains such as rafts and caveolae (54 -56) are enriched in sphingolipids and cholesterol. The plasma membrane of the caveolae concentrates the PMCA pumps, but also PS and many interacting factors that modulate their activity are associated with them (6). Notably, the association of PMCA pumps with caveolae/rafts could even be isoform-specific: in intestinal smooth muscle, the full-length isoform 4b is associated with caveolin-1, whereas the truncated splice isoform 4a (see below) is instead associated with the rafts (58). The spines that protrude from neuronal dendrites also compartmentalize multiprotein Ca 2ϩ signaling complexes; their plasma membrane is enriched in lipid rafts (59).
Because the membrane PL are spatially separated from the catalytic core of the pump, it appears to be difficult to involve them directly in the operation of the enzyme cycle. The interaction of PS with cytosolic portions of the pump would demand that the latter come in contact with the surface of the membrane where the polar heads of activatory PL are located. Thus, the effect of PL would be structural: they could somehow affect the access of substrates to the active site, explaining also effects on the enzyme cycle, i.e. acceleration of the dephosphorylation of the pump (60).

Autoinhibition of the PMCA Pump
That the PMCA pump at rest is in an autoinhibited state was first indicated by the activatory effect of calpain (18,(61)(62)(63). At the time of the finding, the cleavage site by calpain was still undefined. It was then found that the protease removed the entire C-terminal tail of the pump (24), strongly indicating the latter as the portion of the molecule responsible for the autoinhibition. Direct verification was then provided by the finding that synthetic peptides representing the C-terminal CaM-BD inhibited the truncated pump (11,12). They bound to the two sequences in the main body of the pump: one next to its active site in the main cytosolic unit and the other in the cytosolic unit protruding between transmembrane domains 2 and 3. It was proposed that the C-terminal tail would interact with the main body of the pump with its CaM-BD, somehow "closing" the access route to the active site. CaM would then "open" the pump by detaching its binding domain from the two binding sites. Support for the proposal was then provided by a study in which blue and green fluorescent proteins were fused to the N and C termini of pump 4b (64). In the autoinhibited state, significant FRET was detected in the recombinant pump purified from yeast, consistent with an average distance of 45 Å between the two fluorophores. The FRET intensity decreased when the pump was activated by CaM and also by acidic PL or calpain proteolysis, indicating a rearrangement of the pump structure that separated its two ends. Later studies on the autoinhibition process have shown that even minor alternative splicing differences (see below) in the CaM-BD inhibited the pump to different extents (65,66). Other elements in the CaM-BD and in the C-terminal tail also interfere with the autoinhibition process (67, 68), e.g. mutation of the CaM anchor residue Trp-1093 of PMCA4 increased its basal activity (69). Other regions outside the CaM-BD also appear to be involved in the maintenance of autoinhibition; in particular, residues in the proposed cytosolic "stalk" regions (Asp-170 in S2) (70) could help in the stabilization of the inactive state, as their mutation decreased the autoinhibition. A recent contribution (71) underlines the role of conserved residues in the stalk region in the formation of four basic pockets, suitable for the binding of the head groups of PIP 2 . Remarkably, bound PIP 2 is protected from phospholipase C-mediated degradation.
Structural changes in the transition between the autoinhibited and activated states have also been shown by studies (72,73) of the incorporation of photoactivatable PL probes into the pump: the incorporation, which would correspond to the exposure of the pump to the surrounding PL, was greatly decreased by CaM and, albeit differently, by acidic PL. The FRET between the pump labeled in the ATP-binding site and a PL included in the micelles of the purified pump was decreased by CaM but not by acidic PL. Differences thus apparently exist between the active conformations of the pump induced by CaM and PL.

Alternatively Spliced Pump Variants Have Special Properties (and Roles)
The four basic pumps have a sequence identity of ϳ80%, but they differ significantly in some regions, very likely reflecting the specific features of each isoform (74). One important feature distinguishes PMCA2 from the other isoforms: its basic activity is nearly as high in the absence of CaM as in its presence. Because the sequence of the CaM-BD of the PMCA2 pump is identical to that of the other isoforms, the difference evidently reflects differences in the portions of the main core of the pump involved in the autoinhibition process and/or differences in the C-terminal tail outside the CaM-BD (75,76). This property of PMCA2 satisfies the Ca 2ϩ homeostasis demands of cells that require the sustained ejection of Ca 2ϩ even in the absence of activators, e.g. the outer hair cells of the inner ear. The kinetics of activation by Ca 2ϩ also differs in the four isoforms, i.e. the rate of stimulation (and of inactivation upon Ca 2ϩ removal) of PMCA1 and PMCA4 is considerably slower than that of PMCA2 and PMCA3; the first two are thus "slow" and the last two "fast" pumps. Their expression correlates with the requirements of cells in which Ca 2ϩ signals are slow or fast, respectively (77).
The primary transcripts of all PMCA pumps undergo alternative splicing at the two sites mentioned above (5). The splicing at site A is related to the localization of the pump to the apical plasma membrane (78), whereas the splicing at site C, within the CaM-BD, defines the properties of the regulation processes of the pump (74). At site C, exons are inserted in the transcript, producing variants a, c, and d (variant b has no site C inserts) and generating pumps with diverging C-terminal sequences. The insert begins in the middle of the CaM-BD. Interestingly, in variants a, c, and d, five or six of the first 10 residues of the insert that replaces the C-terminal half of the CaM-BD are the same as those of the original non-inserted CaM-BD, i.e. the original full CaM-BD is nearly reconstructed in the inserted variance. The insert in variant a is out of frame and generates a truncated pump. The sequence variations produced by the site C inserts affect both the autoinhibitory role of the C-terminal tail and its interaction with CaM. Peptides corresponding to the CaM-BD of variant a have lower affinity for CaM and decreased efficiency as inhibitors of the pump activated by calpain proteolysis than those of variant b (65,66). The PMCA4a variant is more active in the non-activated state than the PMCA4b variant, i.e. the C-terminal tail of PMCA4a is less autoinhibitory (79).
A recent study (13) on the ortholog of the PMCA pump in Arabidopsis thaliana has demonstrated that its autoinhibitory region, which is N-terminal in plant PMCA pumps, contains two (not one) adjacent CaM-binding sites. The second site, which has lower affinity for CaM and is located downstream of the first, plays a role in autoinhibition of the pump, suggesting that the regulation of the pump could be finely tuned by the Ca 2ϩ concentration. Nanomolar levels of Ca 2ϩ would partially activate the pump via the first CaM-BD, but the pump would become fully activated by the binding of CaM to the second domain only at micromolar levels of Ca 2ϩ . The presence of the second CaM-binding site has also been found in some splice variants of the animal pumps, notably the PMCA1d and PMCA4d variants. The PMCA4d variant is expressed in adult and fetal human heart cells, where it accounts for ϳ30% of the PMCA transcripts (80), and the PMCA1d variant is expressed in skeletal muscle and, less prominently, in brain (5). The bimodular regulation mechanism could thus also be operational in some splice variants of the animal pumps. A comprehensive scheme summarizing the major mechanism of PMCA pump regulation is presented in Fig. 1.

Cellular (Protein) Partners of the PMCA Pump
Numerous protein partners regulate activity, membrane targeting, and recruitment of the PMCA pump to diverse cell components (e.g. the cytoskeleton). Many studies have dealt with the interplay of the pump with proteins containing PDZ domains; these are 80 -90-amino acid domains found in hundreds of signaling proteins that recognize the C terminus of proteins to cluster them with other proteins and "target" them to signaling pathways. The first studied PDZ domain-containing proteins were members of the family of MAGUK kinases (81); the association was proposed to maintain the pumps in specific plasma membrane regions to control Ca 2ϩ homeostasis in local microdomains. Other studies have shown that pump 4b is recruited through PDZ domains to the filopodia adjacent to the actin cytoskeleton during platelet activation (82). Another MAGUK kinase, CASK (Ca 2ϩ /CaM-dependent serine kinase), has also been found to interact with pump 4b (83). CASK is a coactivator of transcription by promoters containing the T-element, which would down-regulate the T-element-dependent reporter activity by depleting Ca 2ϩ in its microenvironment. Because CASK is located at synapses, its interplay with the pump could have a role in synaptic regulation. Another PDZ protein, NHERF2 (Na ϩ /H ϩ exchanger regulatory factor 2; a member of the family of NHERF proteins, which are implicated in the targeting, retention, and regulation of a number of membrane proteins), has shown specificity for PMCA isoforms, as it interacted with variant 2b, but not variant 4b (84). The interaction was suggested to allow the clustering of the PMCA2b pump in a multiprotein Ca 2ϩ signaling complex, facilitating cross-talk among its partners. The finding that variant 4b failed to interact with NHERF2 shows that even minimal sequence differences, such as those between the C-terminal residues of variants 2b and 4b, could be critical in the interaction with partners. PDZ domains also mediate the interaction of variant 4b of cardiac cells with NOS-1 in a ternary complex with ␣-syntrophin, which binds to the pump at a site upstream of the C terminus (85). By down-regulating the production of nitrogen monoxide (NO), the interaction would confer to the pump a role in the regulation of Ca 2ϩ related to local signaling that would prevail over its role in the extrusion of Ca 2ϩ (86). It was then found that, in cardiomyocytes of PMCA4 Ϫ/Ϫ mice, NOS became delocalized from the plasma membrane to the cytosol, leading to a decrease in microdomain cGMP and to elevation of local cAMP. As a result, the L-type channel-mediated Ca 2ϩ influx and contractility were enhanced (87). PMCA4 thus had a structural role in tethering NOS to a selected plasma membrane domain. Importantly, the pump had no influence on global Ca 2ϩ homeostasis in cardiomyocytes, which was unaffected in PMCA4 Ϫ/Ϫ cardiomyocytes.
A number of other PDZ domain-containing proteins also interact with the PMCA pump, recruiting it to selected plasma membrane domains (88,89) or to actin (90). The recruitment to actin of preparations of the purified erythrocyte pump (a mixture of variants 4b and 1b) modified the pump steady-state phosphorylation level, thus influencing activity; G-actin acted as an activator, and F-actin acted as an inhibitor.
In addition to the C-terminal domain, other regions of the pump molecule also interact with partner proteins. The main intracellular loop binds the tumor suppressor Ras-associated factor 1, inhibiting the Ras-associated signaling pathway (91), and also interacts with the catalytic subunit of calcineurin, down-regulating its activity (92). The N-terminal cytosolic region, which has a low degree of sequence similarity among pump isoforms, has been used to search for the isoform specificity of interactors. The work has identified protein 14-3-3 as an inhibitory interactor that acts on isoforms 1b, 3b and 4b, but not on isoform 2b; the finding again shows that isoform 2b has properties that set it apart from the other pump isoforms (57,93).

Conclusion
The properties of the PMCA pumps discussed above are not the focus of earlier reviews. Some deal with aspects that are not well understood or not firmly established; they nevertheless offer novel perspectives on the role of the pump. Most of them are related to problems of regulation, which is not surprising, considering the unusually complex set of regulatory mechanisms that act on PMCA pumps. The first important concept to be underscored is the shift in emphasis that is now gaining momentum regarding the primary role of the pump, which is the spatially and temporally restricted control of Ca 2ϩ signaling within cellular microdomains rather than the bulk regulation of cell Ca 2ϩ (Fig. 2). Regulation of the interplay of the pump with partners co-sequestered with it in local microdomains will then transduce the local Ca 2ϩ modulation to important activities elsewhere in the cell.
The recruitment of the pump to restricted domains of the plasma membrane maximizes its effects on the locally restricted Ca 2ϩ signaling but in turn exposes the pump to the potentially important regulatory activity of the PL environment, which could even change relatively rapidly due to the action of the enzymes that move PL across the membrane bilayer. The activatory action of acidic PL is not yet fully understood, but one interesting possibility that is emerging suggests a role of the PS changes in the apoptotic death of cells. It is tempting to suggest that the removal of PS from the inner leaflet of the membrane bilayer, which occurs early in apoptosis, contributes to amplification of the pro-apoptotic Ca 2ϩ -overloading process. The activation by PL is relevant to another aspect of the function of the pump. The action of the classical activator CaM is rapid, and one can perhaps expect rapidity also of the action of the kinases primed by the products of PIP 2 hydrolysis. By contrast, the effect of PL (except that linked to the direct binding of PS) is probably slower. Changes in the PL environment, such as fluidity variations that would affect lateral mobility of pump units, could be a means to modulate pump activity in a longer time range.
The matter of pump activation has another interesting facet. As mentioned, fast modulation mechanisms, beginning with CaM, have an internal stop signal. When CaM activates the pump, it decreases Ca 2ϩ in the microenvironment surrounding the active center of the pump. CaM would thus come off the pump, terminating activation. In principle, then, CaM activation is bound to have an oscillatory behavior. The reasoning becomes more complicated in the case of the kinases: their activation demands the hydrolysis of PIP 2 . IP 3 /Ca 2ϩ and DAG would then activate the pump, but the pump at the same time had ceased to be activated by PIP 2 . Spatial and/or temporal factors in the activation process could possibly provide a reasonable way out of the paradox.
One additional point regarding the activation process must be underscored: activation of the pump is not always in the best interest of cells. If the main role of the pump is the control of Ca 2ϩ signaling in selected cell domains, its permanent activation cannot be tolerated. The correct regulation of cell Ca 2ϩ signaling demands that the activity of PMCA pumps be modulated positively or negatively; their protracted and uncontrolled activation is probably as negative as their continuous inhibition.
The last point that deserves a comment is the autoinhibition process, which is unique among P-type pumps. Some of its molecular aspects are still not completely understood, but it is gradually becoming clear that it may involve not only CaM but also modulatory PL. It is also becoming clear that changes in the autoinhibition process may underlie conditions of pathology in which the PMCA pump has a causative role.