(Received for publication, December 4, 1996, and in revised form, February 6, 1997)
From the Department of Physiology, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205
PMR1, a P-type ATPase cloned from the yeast
Saccharomyces cerevisiae, was previously localized to the
Golgi, and shown to be required for normal secretory processes
(Antebi, A., and Fink, G.R. (1992) Mol. Biol. Cell 3, 633-654). We provide biochemical evidence that PMR1 is a
Ca2+-transporting ATPase in the Golgi, a hitherto unusual
location for a Ca2+ pump. As a starting point for
structure-function analysis using a mutagenic approach, we used the
strong and inducible heat shock promoter to direct high level
expression of PMR1 from a multicopy plasmid. Yeast lysates were
separated on sucrose density gradients, and fractions assayed for
organellar markers. PMR1 is found in fractions containing the Golgi
marker guanosine diphosphatase, and is associated with an
ATP-dependent, protonophore-insensitive 45Ca2+ uptake activity. This activity is
virtually abolished in the absence of the expression plasmid.
Furthermore, replacement of the active site aspartate within the
phosphorylation domain had the expected effect of abolishing
Ca2+ transport activity entirely. Interestingly, the mutant
enzymes (Asp-371 Glu and Asp-371
Asn) demonstrated proper
targeting to the Golgi, unlike analogous mutations in the related yeast H+-ATPase. Detailed characterization of calcium transport
by PMR1 showed that sensitivity to inhibitors (vanadate, thapsigargin, and cyclopiazonic acid) and affinity for substrates (MgATP and Ca2+) were different from the previously characterized
sarco/endoplasmic reticulum and plasma membrane
Ca2+-ATPases. PMR1 therefore represents a new and distinct
P-type Ca2+-ATPase. Because close homologs of PMR1 have
been cloned from rat and other organisms, we suggest that
Ca2+-ATPases in the Golgi will form a discrete subgroup
that are important for functioning of the secretory pathway.
In eukaryotic cells, the vast bulk of cellular calcium is sequestered within intracellular calcium stores, which maintain cytoplasmic calcium ion concentrations at submicromolar levels and release calcium in response to physiological signals. A major intracellular calcium pool is the endoplasmic reticulum, which is well known for its prominent role in inositol 1,4,5-trisphosphate- and caffeine-mediated calcium release (2). Filling of this store is accomplished by a thapsigargin-sensitive Ca2+-ATPase, a member of the ubiquitous family of P-type ion pumps, and best characterized by the isoform in skeletal muscle sarcoplasmic reticulum (SERCA11; Refs. 3 and 4). There are data supporting the existence of another ionomycin-sensitive pool of intracellular calcium that is unresponsive to both inositol 1,4,5-trisphosphate and caffeine, is non-mitochondrial, and appears to be loaded by a thapsigargin-insensitive Ca2+-ATPase (5). One candidate organelle that may correspond to this pool is the Golgi; subcellular imaging of calcium using the techniques of electron probe x-ray analysis (6) and ion microscopy (7) has revealed that the Golgi apparatus sequesters calcium at concentrations of 1-2 mM, accounting for about 5% of the total intracellular calcium stores. This finding is consistent with numerous reports of an ATP-dependent Ca2+ transport activity in isolated Golgi-rich membrane fractions from tissues such as rat liver (8) and mammary gland (9), and from yeast (10).
Evidence for the role of the Golgi apparatus in regulation of cytoplasmic Ca2+ is scarce; uptake of Ca2+ into the Golgi following a mitogen-induced Ca2+ transient has been demonstrated in NIH3T3 fibroblasts (11), and in LLC-PK1 porcine kidney cells, there is a loss of Golgi calcium following treatment with arginine-vasopressin (12). There is considerable evidence, however, for the importance of Golgi calcium in the exocytic or secretory pathway. For example, calcium depletion inhibits proteolytic processing of precursors in the Golgi/TGN by inhibiting Ca2+ dependent endoproteases (13). In fact, many resident Golgi enzymes require concentrations of Ca2+ in the millimolar range for activity, including KEX2 endoprotease and guanosine diphosphatase (14, 15) in yeast. In mammalian cells, Ca2+-dependent aggregation of chromogranins and secretogranins in the low pH and Ca2+-rich environment of the TGN plays a major role in formation of secretory granules (16). Finally, calcium accumulation in the Golgi has been shown to be of particular importance in tissues involved in calcium secretion, such as lactating mammary gland (17), intestine (18-20), and tooth-forming odontoblasts (21).
The isolation of a gene encoding the putative Golgi
Ca2+-ATPase, PMR1, in the yeast
Saccharomyces cerevisiae (22) allowed a genetic approach
toward understanding the cellular role of Golgi calcium; strains
carrying a null allele of PMR1 manifest a glycosylation defect, secrete unprocessed factor, and suppress various
sec mutants blocked in endoplasmic reticulum/Golgi and
post-Golgi transport, consistent with disruptions of Golgi function
(1). These phenotypes can be reversed by addition of high
Ca2+ (10 mM) to the medium, implicating a
direct role for calcium in Golgi function. An epitope-tagged version of
PMR1 was localized to Golgi both by subcellular fractionation and
immunofluorescence staining of cells (1).
The deduced amino acid sequence of PMR1 carries all the hallmarks of the P-type ion pumps. There are 10 conserved stretches of sequence, including the region around the aspartate residue that forms the phosphorylated reaction intermediate, and regions involved in ATP binding and hydrolysis that are common to all members of this family. The predicted topology of PMR1 is that of the major P2 subgroup of P-ATPases (23), with 8-10 transmembrane spans, cytoplasmic N and C termini, and two defined cytosolic domains linking transmembrane segments 2 and 3, and 4 and 5. Notably, the full-length sequence of PMR1 does not bear a striking homology to either the sarco/endoplasmic reticulum (SERCA) or the plasma membrane (PMCA) Ca2+-ATPases, both well characterized and distinctly different members of the P-ATPase family. However, a closer examination of PMR1 in the region of transmembrane segments 5 and 6, which have been implicated in calcium transport by mutagenesis studies on SERCA and PMCA (24, 25), reveals a significant similarity to Ca2+-ATPases (22). A close homolog of PMR1 from rat has been cloned (26), and named SPCA (secretory pathway Ca2+-ATPase) for its putative functional equivalence to PMR1. The sequence similarity suggests that these two new genes may form a distinct subgroup important in the functioning of the secretory pathway.
In this work, we provide the first biochemical demonstration that PMR1 is indeed a Golgi-specific Ca2+ pump, and show that its kinetic properties and inhibitor-sensitivities are distinct from both SERCA and PMCA, justifying its inclusion in a new subgroup of Ca2+-ATPases. Furthermore, we describe a high level expression system, with low background Ca2+ transport activity, for future structure-function analysis of PMR1. This expression system is also suitable for the study of heterologous Ca2+ pumps.
Culture media contained yeast nitrogen base (Difco; 6.7 g/liter), dextrose (2%), and supplements as needed. All yeast strains used in this study were a generous gift from Kyle Cunningham (Johns Hopkins University) and have been described previously (27). YEpHR1 is a yeast 2µ plasmid carrying the PMR1 coding sequence under control of a tandem repeat of a yeast heat shock element (28), which was derived from plasmid YCp2HSE (29) as follows. The SalI to SacI fragment containing the heat shock elements and CYC1 upstream sequences from plasmid YCp2HSE was cloned into plasmid YEplac195 (30) from which the HindIII site had been destroyed by filling in. An internal HindIII site in the PMR1 coding sequence was removed by site-directed mutagenesis (see below), without change in amino acid sequence; this allowed cloning of the intact 3.6-kb HindIII fragment encoding PMR1 into the corresponding site within the SalI-SacI fragment in YEplac195 to give the expression plasmid YEpHR1.
Oligonucleotide-directed mutagenesis was by the polymerase chain
reaction, using the megaprimer method (31) as modified (32). The
following oligonucleotides were used (5
3
, base substitutions
underlined): HindIII site
(GATAAAACGGAAGCTTTGGATC), Asp-371
Glu
(CTGCTCCGAAAAACAGGTAC), Asp-371
Asn
(GTTATCTGCTCCACAAAACA). In each case, the subcloned PCR
fragment was entirely sequenced to rule out unwanted mutations.
Reconstruction of the intact PMR1 gene in YEpHR1 was by
standard cloning techniques.
Yeast cells were resuspended in Bead buffer (10 mM Tris/HCl, pH 7.4, 0.3 M sorbitol, 0.1 M NaCl, 5 mM MgCl2) containing protease inhibitors (1 mM diisopropyl fluorophosphate, 2 µg/ml chymostatin, and 1 µg/ml each leupeptin, pepstatin, and aprotinin; all from Sigma) at a density of 200 A600 units/ml. Acid-washed glass beads (Sigma) were added to the meniscus and cells were lysed on a vortex in three 1-min pulses, with intermittent chilling on ice. Lysates were centrifuged in a table-top centrifuge (Sorvall, RT6000B) at 800 rpm for 5 min, the supernatants carefully withdrawn and centrifuged for 1 h at 45,000 rpm (Beckman, Ti50). Membrane pellets were resuspended in 20 mM K-Hepes, pH 7.4, with protease inhibitors.
Subcellular Fractionation and Isolation of GolgiFractionation of yeast lysates by sucrose gradient
centrifugation was performed essentially as described by Antebi and
Fink (1), with some modifications. Strain K616 transformed with plasmid
YEpHR1 (K616/YEpHR1) was grown in 600-ml cultures at 25 °C to
A600 of 0.5-0.8 prior to a 2-h shift to
37 °C for heat-shock induced expression of PMR1. Before harvest,
NaN3 was added to 10 mM and the culture was
chilled on ice. Cells were converted to spheroplasts using yeast lytic
enzyme (10 mg/1600 OD units; ICN) for 45 min at 37 °C. Lysis buffer
contained 0.3 M sorbitol, 20 mM triethanolamine
acetate, pH 7.2, 1 mM EDTA, and protease inhibitors
(above). The spheroplasts were suspended in 3-6 ml of lysis buffer and
homogenized using 30 strokes of a Wheaton A Dounce homogenizer.
Following two short centrifugations (3 min at 2700 rpm in a Sorvall
SS34 rotor) to remove unlysed cells, the lysate (3 ml) was layered on
top of a 10-step gradient containing 3 ml each of sucrose (w/w;
18-54% in 4% increments) in 10 mM Hepes, pH 7.5, 1 mM MgCl2. Gradients were spun for 2 h at
27,000 rpm in a Beckman SW28 rotor at 4 °C. Gradients were
fractionated from top to bottom, manually, in 3-ml aliquots (fractions
1-11, respectively). For collection of Golgi membranes, fractions
containing 26, 30, and 34% sucrose were pooled, diluted to 0.8 M sucrose by addition of 10 mM Hepes, pH 7.5, 1 mM MgCl2, and centrifuged for 1 h as before. The pellet was resuspended in 0.5 M sucrose, 10 mM Mes/NaOH, pH 6.0, 0.15 M KCl containing
protease inhibitors (above) and stored at 70 °C in aliquots.
Protein concentration was
determined by a modified Lowry assay (33) following precipitation of
samples containing sucrose by 10% trichloroacetic acid, and using
bovine serum albumin as a standard. Guanosine diphosphatase was assayed
according to Abeijon et al. (15); free Pi was
measured by the method of Fiske and Subbarow (34). -Mannosidase
activity was measured using
4-methylumbelliferyl-
-D-mannopyranoside (Sigma) as a
substrate (35). NADPH-cytochrome c reductase was assayed as
described by Feldman et al. (36).
Calcium transport in sucrose gradient fractions was assayed in O-Buffer containing 10 mM Hepes/NaOH, pH 6.7, 0.15 M KCl, 5 mM MgCl2, 0.5 mM ATP, 5 mM NaN3, and 10 µM CCCP. 45CaCl2 (10 Ci/g; ICN) was added to 0.4 µCi/ml. Fractions (0.3 ml) were incubated for 12 min in 1 ml of assay mix at room temperature. Samples were collected by rapid filtration through Millipore 0.45-µm HAWP filters and washed twice with 5 ml of 10 mM Hepes, pH 7.5, 0.15 M KCl. Filters were dissolved in 1 ml of dimethylformamide and radioactivity determined by liquid scintillation counting with 10 ml of Aquasol (DuPont). Pooled Golgi membranes were incubated for 1-2 min at room temperature, in a volume of 100 µl and at a final protein concentration of 100 µg/ml. In studies involving inhibitors (vanadate, lanthanum, and cyclopiazonic acid), the "O-Buffer" mix was used. In other assays, the mix contained in addition, 10 µM non-radioactive CaCl2 and 10 µM EGTA (to give a final free Ca2+ concentration of 4.9 µM, as determined by the computer program of Fabiato and Fabiato; (37)), and 45CaCl2 was at 2 µCi/ml. Concanamycin A (Sigma; 10 nM) and CCCP (25 µM) were included in titrations of calcium and ATP, to completely eliminate activity of the vacuolar H+/Ca2+ exchanger. In experiments involving variations of pH, calcium concentration was held constant at 10 µM and EGTA was omitted due to the marked effect of pH on the buffering capacity of EGTA.
SDS-PAGE, Antibodies, and Western Blotting MethodsSDS-PAGE gels and Western blotting was performed as described previously (29). Samples (5-100 µg of protein) were precipitated by addition of trichloroacetic acid to 10% (v/v) and centrifuged for 20 min in an Eppendorf microcentrifuge (4 °C). Pellets were resuspended in sample buffer prior to loading on the gels. A C-terminal fusion of PMR1 to E. coli TrpE (pL144; Ref. 1) was used to generate polyclonal antisera from rabbit (Pocono Rabbit Farms, PA). A partial purification of IgG from the antiserum was obtained by (NH4)2SO4 precipitation, followed by dialysis against phosphate-buffered saline (38). Antibodies were used at a dilution of 1:200. Antibodies against the plasma membrane H+-ATPase were obtained from Carolyn Slayman (Yale University).
High Level Expression of PMR1 by Heat Shock Induction from Multicopy Plasmids in Yeast
Polyclonal antibodies were raised in rabbit against a fusion
protein containing the C-terminal third of PMR1 expressed in-frame with
the bacterial TrpE protein (see "Experimental Procedures" and Ref.
1). Immunoblotting of a total membrane fraction from wild type yeast
with anti-PMR1 antibody revealed a single, faint band of 104.5 kDa,
which was missing from the isogenic strain carrying a chromosomal
disruption of the PMR1 gene (Fig. 1,
lanes 1 and 2). The mobility of this band matches
the molecular weight of PMR1 predicted from its amino acid sequence
(22). The low levels of chromosomal expression of PMR1 necessitated a
strategy of overexpression to facilitate biochemical characterization
of the pump. Initially, the strong and constitutive PMA1
promoter (39) was used to direct expression of PMR1 from a yeast
multicopy (2µ) plasmid. However, introduction of this construct into
yeast was lethal, although the equivalent single-copy construct (CEN) was tolerated. This suggested that constitutive, high level expression of PMR1 is toxic to yeast. We therefore chose to use the transient and
inducible heat shock promoter described in earlier work with the
PMA1 gene (29). In addition, the heat shock element is a weaker promoter relative to that of PMA1, based on the
activity of -galactosidase fusions (28, 64). A construct of
PMR1 under control of a tandem repeat of the yeast heat
shock promoter on a multicopy plasmid (YEpHR1; see "Experimental
Procedures") gave some "leak-through" expression of PMR1 even in
the absence of heat shock (Fig. 1, lane 3), however, no
adverse effect on cell growth was apparent. A further, large
amplification of PMR1 expression was observed following a 2-h heat
shock of the cells at 37 °C (Fig. 1, lane 4).
To unambiguously demonstrate Ca2+ pump activity associated
with PMR1 expression, and to pave the way for future analysis of recombinant pumps, plasmid YEpHR1 was introduced into the yeast strain
K616 which carries a deletion of the chromosomal PMR1 gene (27). Additionally, K616 carries a deletion of PMC1,
encoding a second Ca2+-ATPase (27), resulting in virtual
elimination of background levels of Ca2+ pump activity in
all fractions of the yeast lysate (see below and Fig.
2). Finally, it was necessary to disrupt
CNB1, the gene encoding the regulatory subunit of
calcineurin, a Ca2+-, calmodulin-activated protein
phosphatase (40), to maintain viability of the double deletion of
PMR1 and PMC1 (27).
Golgi Localization and Ca2+ Transport Activity of PMR1
Following induction of PMR1 expression by a 2-h heat shock of
strain K616 transformed with plasmid YEpHR1 (K616/YEpHR1), yeast lysates were fractionated on sucrose density gradients, and individual fractions were assayed for the distribution of various organellar markers as described under "Experimental Procedures." The data, summarized in Fig. 2, show an optimal separation of Golgi membranes from vacuoles, endoplasmic reticulum, and plasma membranes. The bulk of
the total protein, representing components of the cytosol, remained in
the loading fraction on top of the gradient, while discrete bands of
membranes were recovered from the interfaces of the sucrose steps
within the gradient. Nearly all (76%) of the total units of activity
of the vacuolar marker, -mannosidase, was recovered from the top
four fractions of the gradient; in contrast, markers for the
endoplasmic reticulum (NADPH-cytochrome c reductase) and
plasma membrane (PMA1) were distributed in the denser half of the
gradient, peaking near the bottom.
Both 45Ca transport activity and PMR1 expression coincided with guanosine diphosphatase activity, a marker for the Golgi membranes (15). This confirms earlier immunological data on the localization of an hemagglutinin epitope-tagged PMR1 by Antebi and Fink (1), and demonstrates ATP-dependent Ca2+ transport into yeast Golgi vesicles. Finally, in the absence of the PMR1 expression plasmid, ATP-dependent 45Ca uptake was virtually eliminated from strain K616 (Fig. 2B), providing direct evidence for calcium transport by PMR1.
Fractions showing maximal 45Ca transport activity (fractions 4-6; corresponding to 26-34% sucrose) were pooled and the membranes collected by centrifugation. The pooled membranes were considerably enriched for PMR1 (Fig. 2C, lanes 1 and 2) and were used for further characterization of Ca2+ transport activity. Yields of pooled Golgi membranes averaged 1 mg from a small-scale culture of 600 ml (480-600 OD600 units).
Calcium Transport by PMR1 into Isolated Golgi Vesicles Is ATP-dependent and Protonophore-insensitive
45Ca transport by the pooled Golgi membranes (100 µg
of protein/ml) was time-dependent and saturable, with
half-maximal uptake occurring at 1 min (Fig. 3). Initial
rates of uptake under the conditions of this assay were >1 nmol/min/mg
with maximal levels approaching 2 nmol/mg in 20 min. This suggests that
uptake is rapidly limited by a build up of free Ca2+ within
the vesicles, since most of this calcium (90%) is readily released by
addition of the calcium ionophore A23187 (10 µM). Unlike
vesicles derived from sarcoplasmic reticulum, no enhancement of uptake
was observed following addition of either oxalate or phosphate (10 mM), possibly reflecting a lack of an anion transporter in
these membranes. In the absence of ATP, an apparent uptake of 10-15%
of maximal levels was observed (Fig. 3); however, there was no effect
of the enzyme apyrase (which would degrade endogenous ATP), and no
effect of the protonophore CCCP (which would collapse a pH gradient and
abolish H+/Ca2+ exchange). Furthermore,
addition of the calcium ionophore elicited only a small (15% after 5 min) decrease in 45Ca associated with the membranes. These
results suggest that in the absence of ATP, there is a largely
nonspecific association of 45Ca with the membrane
surface.
Inclusion of the protonophore CCCP (10 µM) in the transport buffer eliminated up to 20% of the total ATP-dependent Ca2+ transport activity. CCCP-sensitive 45Ca2+ uptake was likely due to VCX1, a vacuolar H+/Ca2+ exchanger (41), as it was also abolished by bafilomycin A (100 nM), a specific inhibitor of the vacuolar H+-ATPase. In the absence of the PMR1 expression plasmid, all 45Ca transport activity observed in the K616 mutant strain was CCCP-sensitive (not shown); the protonophore was therefore included in all the transport assays reported in this study.
Inhibitor Sensitivity of PMR1 Is Distinct from That of Plasma Membrane or Endoplasmic Reticulum Ca2+-ATPases
Vanadate is a characteristic inhibitor of P-type ATPases and is
believed to mimic the transitional state of inorganic phosphate in the
E2 conformation of the enzyme. However, sensitivity to vanadate varies
between different P-ATPases, with Ki ranging from
0.6 µM for the plasma membrane Ca2+-ATPase
from erythrocytes (42) to about 50-300 µM for different isoforms of the sarcoplasmic reticulum Ca2+-ATPase (43,
44). Sensitivity of PMR1 to vanadate more closely resembled the
intracellular Ca2+ pumps, with a Ki of
130 µM (Fig. 4A). Lanthanum, a
calcium channel blocker, inhibited Ca2+ transport with
half-maximal inhibition occurring at 55 µM (not shown).
The loss of 80% activity within one order of magnitude (10-100
µM) suggests cooperative binding of the lanthanum ion to
the ATPase.
Thapsigargin, a plant sesquiterpene lactone, which has been characterized as a potent and specific inhibitor of the SERCA family of Ca2+ pumps (45), had no effect on PMR1 activity at concentrations of up to 5 µM. Similarly, the mycotoxin cyclopiazonic acid inhibited 45Ca transport with poor affinity (Ki of 200 µM) relative to its potent effect on SERCA pumps (Ki of 10-20 nM; Ref. 46).
Characterization of 45Ca Transport by PMR1
pH DependenceGolgi vesicles, resuspended in 10 mM Mes/Tris buffer at pH 6.0 (containing 0.5 M sucrose, 0.15 M KCl for osmotic stability), were assayed for 45Ca uptake in transport buffer ranging from pH 5 to 9, as described under "Experimental Procedures." Fig. 4B shows a distinct pH optimum of 6.7, similar to the pH optimum of the SERCA1 and SERCA2 isoforms (44).
Affinity for CalciumFree calcium was adjusted by buffering
with EGTA as described by Fabiato and Fabiato (37), and the
corresponding initial velocity of 45Ca transport was fitted
as a function of a single site, with an affinity of 0.1
µM, shown in Fig. 5A. In this
respect, PMR1 resembles the plasma membrane Ca2+-ATPase of
sarcolemma and erythrocyte membranes which has a single, high affinity
calcium binding site (47). In contrast, the SERCA pump has two calcium
binding sites which display strong positive cooperativity, with a Hill
coefficient of 2, and an apparent KD of 2 µM (48).
Affinity for ATP
Assay of calcium transport as a function of ATP concentration revealed two separate affinities of 73 and 225 µM (Fig. 5B). Based on evidence from chemical modification studies and secondary structure predictions on other P-type ATPases (reviewed in Refs. 49 and 50), a single nucleotide binding domain occurs on each Ca2+-ATPase monomer. Therefore, the two separate affinities are likely to correspond to the E1 and E2 conformations of the ATPase. By analogy with more detailed analysis of nucleotide binding sites on the SERCA pump (51), it is postulated that the formation of the phosphoenzyme intermediate is coupled to a high affinity site on the E1 conformation of the enzyme. Further velocity increase by higher ATP concentrations occurs without concurrent increase in phosphoenzyme formation of the SERCA pump and is likely to correspond to a low affinity site, also referred to as the "regulatory site," on the E2 conformation.
Substitution of Asp-371 with Glu or Asn Leads to Complete Loss of 45Ca Transport Activity
In every P-ATPase tested thus far, modification of the invariant
aspartate in the highly conserved phosphorylation motif (CSDKTGTLT) results in complete loss of function, consistent with its essential role as acceptor of the terminal phosphate of ATP during ATP hydrolysis and energy transduction (52-54). In the case of the yeast plasma membrane H+-ATPase, the mutant enzymes also display a
dominant lethal phenotype when expressed constitutively, and are
retained in the endoplasmic reticulum (55, 56). It was therefore of
particular interest to analyze the effects of substitutions of the
equivalent aspartate in PMR1. Not surprisingly, 45Ca
transport activity in the mutants Asp-371 Glu and Asp-371
Asn
was abolished to background levels (Fig. 6A;
also compare with Fig. 2B). However, there was no inhibition
of the growth of yeast strains expressing the mutant PMR1 proteins
(measured at 23 °C, not shown), indicating an absence of toxic
effects. In addition, expression levels of the mutant proteins in the
Golgi fractions were nearly normal (Fig. 6, B and
C, lanes 3 and 4). However, in the
case of the Asp-371
Asn mutant, there was significant retention of
mutant polypeptide in the denser fractions of the gradient,
corresponding to endoplasmic reticulum (Fig. 6C, lanes 8-10).
We describe, in this work, the use of a reverse genetics approach to demonstrate the properties of a novel Ca2+-ATPase in yeast. With the recent completion of the yeast genome sequence, and ongoing efforts to sequence other genomes, the functional analysis of many such newly identified putative transporters will be of increasing importance. One of the first obstacles we faced was low expression of PMR1 in wild type yeast, which made its overproduction critical for subsequent biochemical characterization. However, we also found that there were limitations to overproduction, because constitutive expression of PMR1 from the powerful PMA1 promoter was toxic to yeast growth. As a workable compromise, we used the inducible heat shock promoter described in earlier work (29), on a multicopy plasmid. A second aspect of the design of this expression system was engineering of the host strain. We observed that chromosomal deletion of the PMR1 gene led to dramatic compensatory increases in expression of other Ca2+ transporters,2 making additional deletions (in PMC1 and CNB1) necessary to obtain low background levels of Ca2+ transport activity. This host strain should be of general utility for the heterologous expression of other Ca2+-ATPases.
Until now, two classes of Ca2+-ATPases have been described
in the literature, each with multiple isoforms and splice variants that
show tissue-specific distribution: the SERCA and PMCA pumps differ from
one another in their subcellular distribution, mode of regulation,
kinetic characteristics, and inhibitor sensitivities (reviewed by Refs.
57 and 58). Furthermore, sequence comparisons between these two classes
of enzymes reveal a divergence similar to that observed between
unrelated members of the P-ATPase family, suggesting a very early
divergence from one another, or possibly even a convergence in
evolution of Ca2+ selectivity. PMR1, and its putative
homolog from rat, SPCA, fall into a third group on the basis of
sequence homology, somewhat more closely related to the SERCA class
(Fig. 7). In this work, we show that the properties of
PMR1 are clearly distinct from both SERCA and PMCA pumps. Thus, PMR1
differs from SERCA in the number and affinity of the calcium binding
sites, and in the lack of sensitivity to inhibition by thapsigargin and
cyclopiazonic acid. PMR1 also differs from PMCA in the relative
insensitivity to vanadate and lanthanum. These data, together with the
sequence relations, justify the inclusion of PMR1 in a separate class, termed the secretory pathway Ca2+-ATPases. The
identification and characterization of additional members of this
family should assist delineation of this group. There is emerging
evidence for a widespread distribution of PMR1 homologs: partial clones
have been reported from the parasite Schistosoma mansoni
(59) and an algal coccolithophore.3 In
addition, the prokaryotic Ca2+-ATPases reported from
Synechococcus (60, 61) and Synechocystis (62)
resemble PMR1 and SPCA more closely (38% identity) than any other
eukaryotic Ca2+-ATPase sequence. Taken together with an
absence of a gene encoding a SERCA-type
Ca2+-ATPase in yeast, these observations raise the
intriguing possibility that the secretory pathway
Ca2+-ATPases may represent the most ancient and wide-spread
class of Ca2+ pumps, despite being the least recognized.
Our analysis (Fig. 7) has also revealed a fourth cluster of newly
identified Ca2+-ATPases that may be loosely grouped
together on the basis of sequence relations; each member of this
subgroup shares significant identity (40-50%) to the PMCA class,
although there is no additional homology between members within the
subgroup. All are intracellular in localization, typically in acidic
compartments such as vacuoles, and lack the well known
calmodulin-binding regulatory domain found in the C terminus of the
plasma membrane Ca2+-ATPases. The evolutionary relationship
between this "vacuolar" subgroup and PMCA possibly reflects a
topological and functional continuity between vacuolar and plasma
membranes; for example, in Dictyostelium, the contractile
vacuoles discharge their contents to the exterior by fusing with the
plasma membrane, and in Entamoeba, invaginations of the
plasma membrane develop into pinocytotic or phagocytotic vacuoles.
Biochemical characterization of individual members in this cluster will
be required to define the properties of the group further.
The yeast S. cerevisiae is a genetically defined system,
eminently suitable for structure-function studies on Ca2+
pumps using a mutagenic approach. As a prelude to these studies, we
report on the effects of substitutions in the catalytic site aspartate
within the phosphorylation domain of PMR1. As expected, the Asp-371 Glu and Asp-371
Asn mutations led to complete loss of
45Ca2+ transport activity, consistent with
similar functional effects of analogous mutations in the sarcoplasmic
reticulum Ca2+-ATPase (52), the
Na+/K+-ATPase (53) and the yeast
H+-ATPase (54). Additionally, the yeast
H+-ATPase mutants were misfolded (as judged by
hypersensitivity to proteases),4
exclusively retained in the endoplasmic reticulum, and had a dominant
lethal phenotype when expressed constitutively (55, 56). These results
indicated that substitutions at the phosphorylation site have large,
global effects on protein conformation, and that the defective proteins
are retained in the endoplasmic reticulum as part of a "quality
control" mechanism. In contrast to these dramatic effects,
phosphorylation site mutants in the
Na+/K+-ATPase appeared to be correctly folded,
as judged by their ability to bind cations, nucleotides, and ouabain,
although there was no evidence for localization to the plasma membrane
(63). In this work we show that phosphorylation site mutants in PMR1
lack deleterious effects on growth, and are correctly targeted to the Golgi membranes, as determined by subcellular fractionation on sucrose
gradients. However, localization appeared to be significantly retarded
relative to wild-type, since there was substantial retention of mutant
polypeptide in fractions corresponding to the endoplasmic reticulum.
Studies are currently under way to determine whether the mutant
proteins retain native conformation and partial function, such as
ability to bind calcium. In the future, PMR1 may serve as a model for
structure-function studies on the Golgi subgroup of
Ca2+-ATPases.
We thank Kyle Cunningham, Hans Rudolph, Adam Antebi and Carolyn Slayman for generously sharing yeast strains, plasmids, and reagents; Jaqueline Milne for initial help with calcium uptake; Zachary Skelding and Joseph Donofrio for experimental assistance; and Peter Maloney and Gerda Breitwieser for critical reading of the manuscript.