(Received for publication, March 17, 1997, and in revised form, April 17, 1997)
From the Département de Biologie Cellulaire et
Moléculaire, Section de Biophysique des Protéines et des
Membranes, Commissariat à l'Energie Atomique et CNRS URA 2096, Centre d'Etudes de Saclay, 91191 Gif sur Yvette, Cedex, France and
the ¶ Department of Biophysics, Danish Biomembrane Research
Centre, University of Aarhus, DK-8000 Aarhus C, Denmark
Limited proteolysis by proteinase K of rabbit SERCA1 Ca2+-ATPase generates a number of fragments which have been identified recently. Here, we have focused on two proteolytic C-terminal fragments, p20C and p19C, starting at Gly-808 and Asp-818, respectively. The longer peptide p20C binds Ca2+, as deduced from changes in migration rate by SDS-polyacrylamide gel electrophoresis performed in the presence of Ca2+ as well as from labeling with 45Ca2+ in overlay experiments. In contrast, the shorter peptide p19C, a proteolysis fragment identical to p20C but for 10 amino acids missing at the N-terminal side, did not bind Ca2+ when submitted to the same experiments. Two cluster mutants of Ca2+-ATPase, D813A/D818A and D813A/D815A/D818A, expressed in the yeast Saccharomyces cerevisiae, were found to have a very low Ca2+-ATPase activity. Region 808-818 is thus essential for both Ca2+ binding and enzyme activity, in agreement with similar results recently reported for the homologous gastric H+, K+-ATPase (Swarts, H. G. P., Klaassen, C. H. W., de Boer, M., Fransen, J. A. M., and De Pont, J. J. H. H. M. (1996) J. Biol. Chem. 271, 29764-29772). However, the accessibility of proteinase K to the peptidyl link between Leu-807 and Gly-808 clearly shows that the transmembrane segment M6 ends before region 808-818. It is remarkable that critical residues for enzyme activity are located in a cytoplasmic loop starting at Gly-808.
P-type transport ATPases are members of a large family of procaryotic and eucaryotic proteins, specialized in active transport of various cations such as Na+, K+, H+, Ca2+, and Cu+ (1-6) and perhaps also aminophospholipids (7). Based on sequence homologies and transport specificities, these proteins can be divided into the following main groups: types I ATPases (heavy metal transporters), IIA ATPases (e.g. Na+,K+-ATPase or SERCA1 Ca2+-ATPase), and IIB ATPases (e.g. H+-ATPase or plasma membrane Ca2+-ATPase) (6). Ordered two-dimensional or three-dimensional membrane crystals have provided information on the overall shape of SR Ca2+-ATPase (8-10), the H+-ATPase of Neurospora crassa (11), Na+,K+-ATPase (12, 13), and H+,K+-ATPase (14), but the level of resolution obtained at present (down to 10 Å) does not allow a precise description of the topology. Thus the detailed organization of cytoplasmic regions and transmembrane segments in domains with defined functional properties still remains a matter of speculation. Based on amino acid similarities, it is reasonable to postulate that, within each group, the topology and the fundamental reaction mechanism exhibit common features (6, 15).
Directed mutagenesis and various biochemical techniques have been successfully used to pinpoint individual amino acid residues or group of residues important for transport activities. In the case of SR Ca2+-ATPase it was initially demonstrated by Clarke et al. (16) and later considerably refined (17-20) that 6 amino acid residues, presumably located in the transmembrane segments, are of primary importance for Ca2+ binding to the ATPase and/or Ca2+-dependent phosphorylation from ATP. These residues are Glu-309, Glu-771, Asn-796, Thr-799, Asp-800, and Glu-908, located in the putative transmembrane segments M4, M5/M6, and M8. A model was built in which these transmembrane segments are clustered and form a channel that can admit two Ca2+ in a single row (since these bound ions appear to be stacked inside the ATPase structure (21, 22)). All of these residues, with the exception of Glu-908 (18, 23), have been found to be essential for occlusion of Ca2+. In Na+,K+-ATPase and H+,K+-ATPase the homologous residues located in the M5/M6 segments have recently been demonstrated to be of crucial importance for cation binding and transport (19, 24-29). Thanks to these and other efforts (e.g. see Refs. 30 and 31) a consensus picture of what could be the ion pathway in the membrane is gradually emerging.
An unresolved question is whether the membranous amino acid residues
considered so far are the only ones that are directly involved in
translocation of cations. Recently, in
H+,K+-ATPase Swarts et al. (27) have
identified 3 new functionally important amino acid residues (Glu-834,
Glu-837, and Glu-839) in the C-terminal part of the subunit. In
most P-type ATPase alignments, these residues have been proposed to
reside in a cytosolic loop (L6-7), between the 6th and 7th
transmembrane segment (Ref. 6 and Fig. 1A). However in
H+,K+-ATPase, De Pont and co-workers (27, 32)
have proposed a slightly different location according to which the new
critical residues would be part of M6 (see Fig.
1B). This raises the question whether, due to
some special elaboration of structure or mechanism for cation
transport, these residues are essential in
H+,K+-ATPase only. In the present communication
we report data which suggest that this is not the case, since by a
combined protein-chemical and site-directed mutagenesis approach we
have obtained evidence that the homologous residues in SR
Ca2+-ATPase are also intimately involved in
Ca2+ binding and ATPase activity. Furthermore, since in
Ca2+-ATPase this region is accessible to proteolytic
attack, its predicted organization as a cytoplasmic loop seems to be
warranted. These results therefore emphasize the possible contribution
of non-membranous regions to the control of cation binding to the
ATPases.
Proteolytic digestion of SR (45) or purified Ca2+-ATPase vesicles (46) with trypsin was performed as described in Ref. 47. Proteinase K digestion, in 100 mM bis-Tris, pH 6.5, and 0.3 mM Ca2+, was performed essentially as described in Ref. 37 except that, in some cases, proteolysis took place for 4 min at 37 °C instead of 30 min at 20 °C, a modification which generated a higher amount of peptide p20C relative to p19C.
Electrophoresis, Blotting, and 45Ca2+ OverlaySDS-PAGE gels (48) were prepared with inclusion of either 1 mM Ca2+ or 0.02-0.1 mM EGTA in the stacking and separation gels (49). EGTA gels were made with 12.5 or 13% acrylamide instead of 11.8% to allow for less efficient polymerization (as indicated from migration of standard proteins). The protein samples, prepared without the addition of Ca2+ or EGTA (49), were treated with concentrated urea to prevent aggregation (50). Gels were stained with Coomassie Blue. Electroelution of peptides was performed as described (51). 45Ca2+ overlay (52) was performed as described in Ref. 49. Western blotting was performed as described (37) with visualization by the ECL kit (Amersham Corp.). Densitometric measurements were performed on a GS-700 imaging densitometer (Bio-Rad Laboratories).
Mutations and Expression of Ca2+-ATPase in YeastSite-directed mutagenesis was performed with the Exsite kit (Stratagene Inc.) on Ca2+-ATPase SERCA1a cDNA (53). One single mutation E309Q, one double mutation D813A/D818A (referred to later as ADA), and one triple mutation D813A/D815A/D818A (AAA) were introduced. The presence of these mutations and the absence of unexpected mutations due to polymerase chain reaction were verified by DNA sequencing. Wild type and mutant cDNAs were inserted into the yeast expression vector pYeDP60 (a gift of Dr. D. Pompon CNRS, Gif sur Yvette). Saccharomyces cerevisiae W303.1B (a, leu2, his3, trp1, ura3, ade2-1, canr, cyr+) was transformed (53, 54) and selected using ura3 complementation. Growth conditions and criteria for expression of the Ca2+-ATPase were as in Ref. 53 for the test of individual clones and as in Ref. 55 for the large scale expression and crude extract preparation. The crude extract was first centrifuged at 900 × gav for 15 min at 4 °C and then centrifuged at 10,000 × gav for 15 min, 4 °C. The supernatant was submitted to a second centrifugation at 120,000 × gav for 45 min (4 °C) in 10 mM Hepes (Tris), pH 7.4, 0.3 M sucrose, 0.1 mM CaCl2 to pellet a "light" membrane fraction which was homogenized and adjusted to a protein concentration of about 10 mg/ml. Protein concentrations (56) were measured in the presence of 0.5% SDS. The amount of Ca2+-ATPase was estimated by quantitative Western blotting as in Ref. 53 using the polyclonal antibody 577-588 (37). Typically, the light membranes expressed wild type or mutant Ca2+-ATPase at about 1 mg of Ca2+-ATPase per 100 mg of membrane proteins.
ATPase AssayATP hydrolysis was assayed at 30 °C using
an ATP-regenerating coupled enzyme system as in Ref. 53, with some
modifications. To a buffer containing 10 mM Tes (Tris), pH
7.5, 50 mM KNO3, 1 mg/ml
C12E8, 7 mM MgCl2, 0.1 mM CaCl2, 0.1 mg/ml lactate dehydrogenase (Boehringer Mannheim, catalog no. 127221), 50 mM phosphoric
acid, and 0.225 mM NADH, we added 50 µg of protein/ml of
yeast light membranes. This mixture was preincubated for 10 min at
30 °C in the presence of 5 mM NaN3, 0.05 µg/ml bafilomycin A, 0.1 mM ammonium molybdate as
inhibitors of other ATPases and 0.1 µM
N-aminocaproic acid, 0.1 µM
phenylmethylsulfonyl fluoride, and 2.8 mM
-mercaptoethanol as antiproteases. After addition of 1 mM phosphoenolpyruvate and 0.1 mg/ml pyruvate kinase
(Boehringer Mannheim, catalog no. 109045), the reaction was started by
the addition of 1 mM Na2ATP. Thapsigargin (1 µg/ml) was added after 100 s, and the rate of hydrolysis was followed for an additional 100 s. The Ca2+-ATPase
activity was calculated as the difference between the slopes obtained
in the presence and in the absence of thapsigargin and was corrected
for a small background activity obtained with control yeast light
membranes.
It has previously been documented in a number of investigations
that changes of electrophoretic mobility in Ca2+-containing
SDS gels can be used for the detection of Ca2+-binding
proteins or peptides (57-61). Examples of such changes in migration
rate are shown in Fig. 2, A and B,
where it can be seen that calmodulin in a Ca2+-containing
gel (lane 4, Fig. 2B) migrates at a much faster
rate than in a gel containing 0.1 mM EGTA (lane
4, Fig. 2A). Intact Ca2+-ATPase from
sarcoplasmic reticulum (lanes 2) moves slightly faster in
the presence of Ca2+ than in its absence (compare with the
95-kDa molecular mass marker). A more pronounced increase in migration
rate is noted for fragment B (lanes 3), the C-terminal
polypeptide (sequence 506-994) produced by tryptic cleavage of
Ca2+-ATPase, which co-migrates with the N-terminal fragment
A in the presence of EGTA but not in the presence of Ca2+.
Fragment A is further cleaved to A1 (peptide 199-505) and A2 (peptide
1-198). Some increase in migration rate in the presence of
Ca2+ is observed for A2 (compare with the 20-kDa molecular
mass marker), whereas A1 is not affected by Ca2+ (compare
with the 30-kDa molecular mass marker). The effect of Ca2+
on the migration rate of Ca2+-binding proteins and some of
their peptide fragments is due probably to retention of a more compact
and native-like structure in the presence of SDS, resulting from the
binding of Ca2+ (49, 61, 62).
In Fig. 2, C and D, we show that still shorter
C-terminal Ca2+-ATPase fragments produced by proteolysis
with proteinase K (p28C, p27C, and p20C) also change their mobility in
Ca2+-containing gels, whereas other fragments (p29/30,
p19C) are unaffected by the presence of Ca2+. As
illustrated in Fig. 3, p20C starts at Gly-808, a residue which was predicted to be close to the C-terminal border of the M5/M6
transmembrane region, and extends to the C-terminal Gly-994 while p19C
starts at Asp-818 (therefore in the L6-7 loop) and also ends at
Gly-994. Thus a marked difference in peptide behavior in SDS gels is
observed after removal of only 10 amino acids from p20C (Fig.
3B). The Ca2+-dependent change in
migration rate of p20C is observed not only in the mixture of peptides
obtained after limited proteolysis (lanes 3, Fig. 2,
C and D), but also after isolation of p20C by electroelution from the gel and renewed electrophoresis (lanes 4), indicating that Ca2+ binding is an intrinsic
property of the peptide. After electroelution p19C is still unaffected
by the presence of Ca2+ in the gel (lanes
5).
45Ca2+ overlay of proteins (52, 64-66) after
transfer to Immobilon membranes is a complementary technique with which
to detect Ca2+ binding to proteins or peptides with various
affinities for Ca2+ (49). For instance this strategy has
been successfully used to localize a Ca2+-binding region
which does not contain an EF-hand motif in the cardiac
Na+-Ca2+ exchanger (65). In Fig.
4A we show the results of a
45Ca2+ overlay experiment performed on purified
p20C and p19C. In these experiments, the 20-kDa molecular mass marker,
trypsin inhibitor (lane 3), also binds
45Ca2+ (49, 52), providing a convenient
reference in this region of the gel. It is clear that only p20C binds
45Ca2+ (Fig. 4A, lane 1),
while p19C is not labeled (lane 2), although present in
comparable amounts, as seen by Coomassie Blue staining of the blot
(Fig. 4B). Note that these experiments are performed after
extensive washing of the membrane with a non-ionic and non-denaturing detergent, C12E8, which should help to remove
SDS. In addition, 45Ca2+ overlay is performed
at low Ca2+ concentration (10-15 µM) and in
the presence of 5 mM Mg2+, which suggests that
the partially refolded binding site in p20C has retained some
specificity for Ca2+ over Mg2+.
The combined demonstration of a Ca2+-dependent
change in migration rate and of 45Ca2+ overlay
labeling is a good indication of Ca2+ binding by p20C and
not by p19C. However, this does not necessarily imply that
Ca2+ is bound at functionally important sites. By
inspection of the 808-818 sequence we have noticed some resemblance
with the Ca2+ binding site of -lactalbumin, a binding
site which in this protein is confined to a short segment (amino acid
residues 82-88; cf. Fig. 1C). In this sequence
the Ca2+ binding site is formed by three carboxylic groups
of aspartic acid residues and two carbonyl groups, located in a loop,
and two adjoining helical regions. Preliminary attempts with molecular modeling of the Ca2+-ATPase 808-818 sequence by energy
minimization according to the
-lactalbumin Ca2+ binding
site suggested that the side chains of Asp-813 and Asp-818, together
with the peptide carbonyl group between Asp-815 and Ile-816 could be
involved in Ca2+ binding to the L6-7
loop.2 We then took advantage of the
functional expression of Ca2+-ATPase in yeast (53) to test
the activity of two cluster mutants in this region, i.e.
D813A/D818A (ADA) and D813A/D815A/D818A (AAA). Note that residues
Asp-813 and Asp-815 correspond to residues found to be important in the
case of the H+,K+-ATPase (Fig. 1A).
After purification of a light membrane fraction from the
ATPase-expressing yeasts, we verified by Western blotting and
densitometric measurements that mutant and wild type ATPases had been
expressed to the same extent (data not shown). To quantify ATP
hydrolysis, we used a coupled system and followed the disappearance of
NADH in the absence or presence of thapsigargin, a specific inhibitor
of SERCA ATPase. In Fig. 5 we present histograms of specific Ca2+-ATPase activity based on several experiments.
It is seen that activity in the double mutant activity is reduced by
more than 90 %, and activity in the triple mutant drops to a
negligible level, comparable to that seen after mutation to glutamine
of the essential glutamate in M4 (E309Q; Ref. 16). These activity measurements do not point to any particular mechanism for inactivation, but phosphorylation experiments suggest that the Ca2+
binding step is indeed affected by the mutations. The inset
of Fig. 5 shows that phosphorylation by
[32P]Pi, while inhibited by Ca2+
at 126 µM and 1000 µM for wild type ATPase
and SR + control yeast membranes, is still possible for the ADA mutant
despite these high Ca2+ concentrations. Based on the
rationale of Clarke et al. (16), for the six mutations cited
in the introduction, this shows that the ADA mutant is
Ca2+-insensitive.
Our study demonstrates the involvement of the region corresponding to the L6-7 loop in the reaction mechanism of Ca2+-ATPase. It should be noted that this region is fairly well conserved, in particular among type IIA ATPases. Therefore the previous findings with the corresponding residues in H+,K+-ATPase (27) probably are of wider significance and indicate common features in the transport mechanism. However, in contrast with the interpretation proposed for H+,K+-ATPase in the latter work, there is convincing evidence for cytosolic exposure of this region in Ca2+-ATPase, since it is an easy target for proteolytic cleavage (37, 38) and it reacts with sequence-specific antibody 809-827 (67). Therefore, we do not consider it likely, as suggested by Swarts et al. (27), that these residues are embedded inside the membrane structure, but we favor a probable location of the Ca2+ liganding groups at the cytosol/membrane border. We suggest that residues of the L6-7 loop could be involved in the initial processes of Ca2+ transport. Specifically, they could act as a site for ion approach (68, 69) toward the high affinity binding pocket inside the membrane and/or be part of the gate operating during calcium occlusion (70, 71), preventing release of bound Ca2+ toward the cytosolic side. Exploration of these possibilities requires further studies.
We are grateful to Dr. J. P. Andersen, Danish Biomembrane Research Center, University of Aarhus, for communicating the results of selected mutations in loop 6-7 before publication, to Drs. D. Pompon and Ph. Urban, CNRS, Gif sur Yvette for the gift of the yeast strain and the vector, and to them and Dr. J. D. Groves, Dept. of Biochemistry, University of Bristol, for numerous helpful suggestions. We also thank the members of our Section, CEA, Gif sur Yvette for stimulating discussions, as well as Dr. J. Smith for his help with molecular modeling.