The Ca2+ Affinity of the Plasma Membrane Ca2' Pump Is Controlled by Alternative Splicing*

The plasma membrane Ca2+ pump is a calmodulin- regulated P-type ATPase that is an essential element in controlling intracellular Ca2+ concentration. Studies on the gene structure of this pump have revealed an alternate splice option that changes the structure of the cal- modulin-binding domain. This change in the structure of the enzyme results in a reduced calmodulin affinity. Tests of the enzyme's activity in the presence of a high calmodulin concentration, approximating that found in-side living cells, show that this reduced calmodulin af- finity causes a reduced apparent affinity of the enzyme for Cas+. This shift in the Ca2+ activation occurs in a Ca2+ concentration range crucial to cellular function and is probably the physiologically important consequence of the alternate splice. V,, apparent a

The plasma membrane Ca2+ pump is the sole high affinity mechanism for removal of Ca2+ from the cytosol to the extracellular space. The activity of this pump is stimulated by direct interaction with calmodulin, which increases both the V , , and the apparent affinity for Ca2+. In the presence of Ca2+, calmodulin binds to a n autoinhibitory domain (C domain), which is a 28-residue region near the carboxyl terminus of the enzyme. Since calmodulin binds tightly only when it is loaded with Ca2+, stimulation of the enzyme by calmodulin will be regulated by changes in the intracellular concentration of Ca2+. Because of this, an alteration in the affinity of the pump for Ca2+-calmodulin might lead to an enzyme with altered Ca2+ sensitivity under the conditions in a living cell.
Such a change in the calmodulin affinity can be obtained by the utilization of an alternate splice. At least four different genes code for this pump, and alternate splices affecting at least two areas of the pump introduce additional variation in its properties (1). In all four genes, an alternative splice option Grant GM 28835. The costs of publication of this article were defrayed * This work was supported in part by National Institutes of Health in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. exists in the middle of the sequence coding for the C domain, which causes a substantial change in the structure of the downstream portion of the domain. The most common form of the C domain is the b form, which has a net plus charge of 6 or 7, while the a form, expressed in a limited range of tissues, has a net plus charge of 3 or 4. Little is known about the properties of the proteins coded for by the different genes and splices, and most of the available information has been gained by inference, by experiments using synthetic peptides (2) and inactive products from short genetic constructs (3). Studies utilizing synthetic peptides corresponding to the C domain showed that the peptide representing the a form has about a 10-fold lower affinity for Ca2+-calmodulin and was also about 10-fold less effective as an inhibitor of the activated Ca2+ pump (2). Recent successes at overexpression of the Ca2+ pump in COS cells (4,5) and the development of a useful assay for the enzyme expressed in COS cells ( 6 ) have opened the way to studies relating the structure of the pump to its biochemical activity. We will focus here on the functional consequences of an alternative splice in the middle of the sequence coding for the C domain.

II
MATERIALS AND METHODS Chemi~als-~~CaCl~ was obtained from DuPont NEN. Calmodulin was purchased from Sigma. All other chemicals used in this study were of reagent grade.
Construction of Full-length hPMCA4b and hPMCA4a and Their IFuncated Versions, 4bfct92) and 4afct56bFull-length hPMCA4b was put together from the partial clones (4) and cloned into expression vector pMM2. A NsiI and KpnI fragment of hPMCA4a obtained from a partial clone was substituted for the 3' end of hPMCA4b to obtain a full-length hF"CA4a. 4btct92) and 4a(ct56) were constructed by the strategy described for dl20 (6).
IFansfection-Transfection of COS-1 cells was performed as described (6) except that the M6 isolate of COS-l cells obtained from the Genetics Institute, Cambridge, MA was used in the transfection and 15 x lo6 cells were used per 150-cm2 flask.
Isolation of Microsomes from COS-1 Ceh-crude microsomal membranes were prepared as described (6). Ca2+ Dunsport Assay-Calcium influx into microsomal vesicles was measured at 37" C by rapid filtration through Millipore membrane filters (0.45-pn pore size, type H A ) essentially as described in the previous paper (6) except that the higher specific activity of the microsomes allowed reduction of their concentration to 4-7 pg/ml. The vesicles were preincubated at 37" C in the presence of the appropriate concentrations of calmodulin for 3 min, and Ca2+ uptake was initiated by the addition of ATP. Controls from pMM2-transfected cells were subtracted from each data point.

RESULTS AND DISCUSSION
We have overexpressed in COS cells the full-length a and b versions of the human plasma membrane Ca2+ pump, isoform 4, which are called hPMCA4a and hPMCA4b, and also the truncated versions 4a(ct56) and 4b(ct92). These truncated versions contain the 28 residues of the C domain, but all residues downstream of that are omitted. Fig. 1 shows a Western blot of membranes from COS cells expressing these four constructs, showing that enzymes of the expected molecular weight are produced. All four constructs express proteins that are identical up to residue 1104. Downstream of this residue, the structures of the a and b isoforms are different because of the insertion of a 178-base pair sequence into the mRNA of the a form. Fig. 1 (lower panel) shows the sequences of the carboxyl-terminal portions of all four constructs used in this study.
When hPMCA4b and hPMCA4a were expressed in COS cells and their activity measured as a function of calmodulin con-

41
Control of Ca2+ Pump's Affinity for Ca2+ centration, it was found that hPMCA4a was less responsive to calmodulin than hPMCA4b. As shown in Fig. 2, the concentration of calmodulin required for half-maximal activation of hPMCA4a was about 7 times higher than that required to activate hPMCA4b. A 3-5-fold stimulation by calmodulin was observed under these conditions. More relevant to the function of the pump is its Ca2+ affinity under conditions approximating those in a living cell. In order to approximate intracellular conditions, it is necessary to use a high calmodulin concentration. The total calmodulin concentration is about 3.6 1.1~ in red cells (7) and about 2 1.1~ in thyroid and in skeletal muscle (8). In the tissues where it is most abundant (brain, testis) it approaches 30 p~ (8). The fraction of calmodulin that is free can vary widely, from 5% of the total to more than 90% (9). We measured the dependence of Ca2+ uptake on free Ca2+ at two high free calmodulin concentrations that might be encountered in living cells. Fig. 3 shows the Ca2+ response of hPMCA4b and hPMCA4a a t 0.3 and 1.5 1.1~ calmodulin. Under both circumstances, hPMCA4b was more responsive to Ca2+ than hPMCA4a. At 1.5 p~ calmodulin the K,+ for Ca2+ was 0.25 1.1~ for hPMCA4b and 0.54 1.1~ for hPMCA4a.
Since these differ in a range crucial to cellular function, we can expect these two isoforms to regulate intracellular Ca2+ differently.
To determine which portion of the pump's carboxyl-terminal sequence was responsible for the difference in Ca2+ affinity, we also measured the Ca2+ response of the truncated mutants in the presence of 0.3 1.1~ calmodulin (Fig. 4). The difference in the KX for Ca2+ was still present, indicating that the difference in Ca2+ affinity came from the 9 residues in the C domain, which is the only difference between 4b(ct92) and 4a(ct56).
These experiments demonstrate that the alteration of a few residues in a region not directly involved in Ca2+ transport can cause a substantial change in the effective Ca2+ afinity. This alteration in the enzyme structure directly affected only the calmodulin affinity, which in turn reduced the sensitivity of the plasma membrane Ca2+ pump to Ca2+. The expression of such an altered isofonn may imply altered intracellular Ca2+ signaling and a higher basal level of Ca2+.
Isoform hPMCA4b is widespread, while hPMCA4a is seen primarily in brain, smooth muscle, and perhaps heart (10, 11). Since these tissues make a particularly heavy use of the intracellular Ca2+ signal, the presence of hPMCA4a in them may be related to the existence of higher average Ca2+ levels in some of the cell types in these tissues.
It has recently been reported that alternative splicing of two different receptors produce four or five different isoforms, which couple via different G proteins (12,131. Thus, alternative splicing of mRNA, generating multiple isoforms from a single gene, appears to be a general mechanism for the fine tuning of intracellular signaling.