Cloning and expression of the unique Ca2+-ATPase from Flavobacterium odoratum.

The 60-kDa Ca2+-ATPase from Flavobacterium odoratum is kinetically and mechanistically similar to other P-type ATPases, suggesting its use as a model system for structure-function studies of ion transport. A portion of the gene was amplified by polymerase chain reaction of genomic DNA with degenerate oligonucleotide primers, one based on the N-terminal amino acid sequence of the purified protein and the other based on a consensus sequence for the phosphorylation site of P-type ATPases. This gene fragment was used to screen a lambda library of F. odoratum 29979 DNA. Clone "C" is 3.3 kilobases in length and contains one complete and part of a second open reading frame, the first of which encodes a 58-kDa protein containing the exact N-terminal amino acid sequence of the purified protein. We have named this gene cda, for calcium-dependent ATPase. Escherichia coli, transformed with clone C, demonstrates high levels of calcium-dependent and vanadate-sensitive ATP hydrolysis activity, forms a 60-kDa phosphointermediate, and cross-reacts with antibodies to the purified Ca2+-ATPase. The gene has almost no sequence homology to even the highly conserved regions characteristic of P-type ATPases but does possess significant homology to a protein with alkaline phosphatase activity (PhoD) from Zymomonas mobilis. The putative phosphorylation site is a Walker A (P-loop) ATP binding sequence and is modified relative to P-type ATPases, suggesting that the F. odoratum Ca2+-ATPase may represent an ancestral link between the F- and the P-type ATPases or perhaps a new class of ATPases.

The 60-kDa Ca 2؉ -ATPase from Flavobacterium odoratum is kinetically and mechanistically similar to other P-type ATPases, suggesting its use as a model system for structure-function studies of ion transport. A portion of the gene was amplified by polymerase chain reaction of genomic DNA with degenerate oligonucleotide primers, one based on the N-terminal amino acid sequence of the purified protein and the other based on a consensus sequence for the phosphorylation site of P-type ATPases. This gene fragment was used to screen a library of F. odoratum 29979 DNA. Clone "C" is 3.3 kilobases in length and contains one complete and part of a second open reading frame, the first of which encodes a 58-kDa protein containing the exact N-terminal amino acid sequence of the purified protein. We have named this gene cda, for calcium-dependent ATPase. Escherichia coli, transformed with clone C, demonstrates high levels of calcium-dependent and vanadate-sensitive ATP hydrolysis activity, forms a 60-kDa phosphointermediate, and cross-reacts with antibodies to the purified Ca 2؉ -ATPase. The gene has almost no sequence homology to even the highly conserved regions characteristic of P-type ATPases but does possess significant homology to a protein with alkaline phosphatase activity (PhoD) from Zymomonas mobilis. The putative phosphorylation site is a Walker A (P-loop) ATP binding sequence and is modified relative to P-type ATPases, suggesting that the F. odoratum Ca 2؉ -ATPase may represent an ancestral link between the F-and the P-type ATPases or perhaps a new class of ATPases.
Calcium is an important component of the signal transduction process in eukaryotic cells. It is therefore necessary that intracellular calcium levels are kept low so that even small changes in concentration are detectable for signal transduction. In prokaryotes, however, calcium has not been shown to have such a role; nonetheless, intracellular calcium is maintained in the micromolar range even when extracellular levels are in the millimolar range. Although most prokaryotes use secondary transport to remove calcium from the cell, the Gramnegative Flavobacterium odoratum has been shown to possess a protein with Ca 2ϩ -ATPase activity (1) similar to the sarco-plasmic/endoplasmic reticulum Ca 2ϩ -ATPase (SERCA). 1 More recently, genes homologous to sequences for other Ca 2ϩ -transporting proteins have been cloned from several cyanobacterial species (2,3).
The Ca 2ϩ -ATPases found in eukaryotes and the genes cloned from prokaryotes all belong to a family of ATPases called Ptype. P-type ATPases transport ions across membranes at the expense of ATP; besides calcium, H ϩ (4), Na ϩ (5), K ϩ (6), Mg 2ϩ (7), Cu 2ϩ (8), and Cd 2ϩ (9) have been shown to be transported by P-type ATPases. Small regions of high sequence homology between all P-type ATPases, regardless of ion specificity, suggest common structural motifs and possibly common mechanisms for ion transduction (10). All P-type ATPases form a vanadate-sensitive phosphointermediate during the reaction cycle; the aspartyl phosphate species is acid stable and alkaline labile. Many of the P-type ATPases are comprised of a single subunit, although several contain two or three subunits in a membrane-bound complex. Sizes range from 72 kDa for the KdpB subunit of the K ϩ -transporting Escherichia coli Kdp complex (6) to 102 kDa for the MgtB Mg 2ϩ transporter of Salmonella typhimurium (7), while the eukaryotic P-type ATPases tend to be somewhat larger with SERCA at 110 kDa (11). The 60-kDa size of the protein from F. odoratum makes it the smallest known P-type ATPase.
The F. odoratum protein has been extensively characterized (12) and is found to be functionally and kinetically similar to other P-type ATPases. F. odoratum membrane vesicles transport Ca 2ϩ in an ATP-dependent and vanadate-inhibitable manner (1). The protein forms an acid-stable vanadate-sensitive phosphointermediate and exhibits K m and K i for calcium, ATP, and vanadate similar to SERCA, although the enzyme turnover is much faster (12,13). This would appear to make it an ideal candidate to use as a model system for the P-type ATPase, and SERCA in particular. While much work has elucidated the role of specific residues and domains in SERCA, the bacterial system offers the advantage of allowing high levels of expression of a mutant protein in its native environment with no interference from endogenous wild-type activity.
We have therefore set out to clone the gene for the Ca 2ϩ -ATPase from F. odoratum to analyze the structure and function of this prokaryotic ATPase. The approach employed relies on polymerase chain reaction (PCR) amplification of short regions from genomic DNA using degenerate primers based on both the N-terminal amino acid sequence of the purified protein (12) and highly conserved sequences in other P-type ATPases. The PCR products were then used to probe a genomic library for a full-length clone. The cloned gene, which we have called cda for calcium-dependent ATPase, has been expressed in E. coli and is functionally identical to the purified Ca 2ϩ -ATPase from F. odoratum (12). However, sequence analysis of the cda gene reveals that, despite its functional similarities to the P-type ATPases, it is structurally distinct. Surprisingly, the sequence contained almost none of the highly conserved regions characteristic of all known P-type ATPases.

EXPERIMENTAL PROCEDURES
Chemicals-All chemicals were obtained from Sigma (reagent grade) except where indicated. C 12 E 8 was obtained from Calbiochem.
Bacterial Strains and Vectors-F. odoratum strain (ATCC 29979) was obtained from ATCC and grown aerobically in Luria-Bertani medium, pH 7.5, at 37°C and harvested mid-log phase. E. coli strains XOLR and XL1 Blue were obtained from Stratagene, while Nova Blue and Nova Blue (DE3) E. coli cell lines were obtained from Novagen (Madison, WI); both were grown under the same conditions as F. odoratum. ZAP Express cloning vector was obtained from Stratagene, and pET-21 vector was obtained from Novagen. . PCR was performed on a Perkin-Elmer Corp. DNA thermal cycler model N801-0150 with 30 cycles of (94°, 1 min; 51°, 1 min; 72°, 2 min) using Taq polymerase and buffers from Life Technologies, Inc. and deoxynucleotides from Perkin-Elmer. Products were analyzed on a 1% agarose gel in TBE buffer (50 mM Tris, 50 mM borate, 1 mM EDTA, pH 8.3) with 0.1 g/ml ethidium bromide to visualize the DNA. The PCR reaction product mixtures were purified from excess primer and nucleotides by the Wizard PCR purification kit (Promega). The mixture was ligated into the pT7 blue vector (Novagen) according to the manufacturer's directions. After transformation into E. coli Nova Blue cells (Novagen), colonies were screened for insert size by PCR (primers T3-T7 from the vector), and the insert was sequenced by the dideoxy method using the Sequenase kit (U. S. Biochemical Corp.). A 450-bp product of the N-P reaction matched the remaining amino acid sequence obtained from the purified protein.
Screening a Genomic Library for a Clone-F. odoratum strain 29979 genomic DNA prepared according to Ref. 14 was digested with EcoRI and ligated into the ZAP Express vector (Stratagene). The plated library was screened with a PCR probe generated from primers derived from each end of the 450-bp N-P sequence (primer 2, GTGGTGGGTCT-TGTAGTTGAT; primer 1, TCCTCCTCTATCTTTAATTGC). Positive clones were rescreened twice to ensure their purity. The plaques were removed from the agar plates and subjected to excision of the internal phagemid pBK-CMV using the helper phage R408 according to the manufacturer's directions, transforming into E. coli XOLR cells, and subsequently into E. coli XL1 Blue (Stratagene). Phagemid clones were screened for insert size and approximate gene position by PCR (primer pairs T3-T7, T3-2, T3-1). Clone C (pBK-C) was predicted to contain the entire cda gene and was used in further studies. To tightly control the expression of cda, the C clone was transferred to pET-21 (Novagen) by digesting pBK-C with BamHI and XhoI and subsequent ligation. Deletion of C clone sequence 3Ј of cda was achieved using the ExSite kit (Stratagene).
Sequence Analysis of the Clone-The nucleotide sequence of clone C was determined as above, and the sequence was analyzed using the program GeneWorks (Intelligenetics, Inc., Mountain View, CA) to calculate the Kyte-Doolittle hydrophobicity plot (15). Sequence comparisons were performed by Blast (National Center for Biotechnology Information) and the GCG (Genetics Computer Group, Madison, WI) programs "motifs," "FastA," and "Pileup." The presence of a signal sequence and its cleavage site was verified by the EGCG program "sigcleave." Activity Assays-ATP hydrolysis and forward phosphorylation assays were performed on French-pressed membrane vesicles or partially purified fractions. Inside-out membrane vesicles were produced by a single passage of E. coli through an Aminco French pressure cell at 20,000 psi as described (12). Unbroken cells and debris were removed by centrifugation at 27,000 ϫ g for 30 min, and the membrane vesicles were then pelleted by ultracentrifugation at 200,000 ϫ g for 90 min. The vesicles were solubilized with 2% C 12 E 8 on ice for 60 min and then centrifuged at 360,000 ϫ g for 60 min. Ammonium sulfate was added to the supernatant at a concentration of 35% of saturation and centrifuged at 10,000 ϫ g for 20 min. The supernatant, which contained the Ca 2ϩ -ATPase, was brought first to 55% ammonium sulfate, centrifuged, and the resulting supernatant extracted with 65% ammonium sulfate. The supernatant from the 65% fraction was brought to 90% ammonium sulfate and centrifuged. This final precipitate (90% fraction) was carefully removed and resuspended in buffer A (20 mM MOPS, pH 7.5, 100 mM KCl, 0.5 mM MgCl 2 , 1 mM dithiothreitol) at 2 mg of protein/ml and dialyzed overnight against ice-cold buffer A. ATP hydrolysis was carried out with the 90% fraction incubated in 45 l of buffer A (plus 0.5 mM DCCD) with either 0.1 mM CaCl 2 (ϩ calcium) or 2 mM EGTA (Ϫ calcium) for 10 min at room temperature, and the reaction was initiated with 5 l of a 1 mM [␥-32 P]ATP (final concentration, 100 M at 100 -200 cpm/pmol). After 10 s, the reaction was terminated by the addition of 150 l of 30% trichloroacetic acid, 1 mM K 2 HPO 4 . Inorganic phosphate was extracted and counted as described (12).
ATP phosphorylations were performed as described for ATP hydrolysis except that the reactions were carried out on ice using 5 l of a 100 M [␥-32 P]ATP (at 1000 -2000 cpm/pmol). Reactions were terminated by the addition of 750 l of 10% trichloroacetic acid, the samples were centrifuged (12,000 ϫ g for 5 min), and the pellet was solubilized and analyzed by SDS-PAGE followed by autoradiography as described (12).

PCR Generation of a Partial Clone-Since the F. odoratum
Ca 2ϩ -ATPase functionally behaves like a typical P-type ATPase, it would be expected to contain the regions of high homology shared between other P-type ATPases, which are associated with such functions as phosphorylation and ATP binding. Several degenerate oligonucleotides were made for each region (ATP binding, "A"; phosphorylation, "P"; TGES transduction domain, "T") to cover the unexpected codon usage in F. odoratum. Degenerate oligonucleotides were also made to correspond to part of the 22-amino acid N-terminal sequence ("N") determined from the purified protein (12). Most primer pairs produced 2-8 major products, which were cloned into pT7 Blue (Novagen) and sequenced. The only PCR product that yielded a clone matching the entire predicted amino acid sequence for the N terminus past the primer was generated by N and P. N-A yielded no corresponding product. Based on the N-P distance in other P-type ATPases, we predicted a product of about 850 bp; however, the product was unexpectedly short at 450 bp (data not shown), suggesting major differences in the overall structure of the protein compared to other P-type ATPases.
Analysis of the Full-length Clone and Its Sequence-The 450-bp partial clone was used as a probe to screen a ZAP library of EcoRI-cut F. odoratum DNA. Several positive clones of varying lengths were obtained and analyzed by PCR prior to nucleotide sequencing. Clone C was approximately 3.3 kilobases in length and contained less than 500 bp 5Ј of the N-P sequence. The nucleotide sequence of C was determined to be 3332 bp in length, containing one full-length and part of a second open reading frame.
The first open reading frame (Fig. 1) contained the N-P probe sequence; however, the N-terminal serine did not immediately follow a methionine, suggesting the presence of a signal sequence, which would target the protein to the membrane. Several in-frame methionines upstream would produce signal se-quences of 11, 21, or 42 amino acids. The EGCG program "sigcleave" analyzed the entire sequence and predicted the presence of a signal sequence that cleaves at the N-terminal serine. The translation start site is likely to be the methionine at either Ϫ21 or Ϫ42 since both contain the hydrophobic helical region (from amino acids Ϫ13 to Ϫ1) typical of signal sequences. Starting from the N-terminal serine, the gene consists of 1581 bp, coding for a protein of 527 amino acids with a molecular mass of 58,805 Da, very close to the 60-kDa size predicted from the protein's mobility on SDS-PAGE. We therefore suggest that the first open reading frame of clone C encodes the Ca 2ϩ -ATPase protein purified from F. odoratum. We propose to call this gene cda, for calcium-dependent ATPase.
The second open reading frame continues for 1310 bp to (and probably past) the end of the clone, coding for a protein of at least 49,906 Da (436 amino acids). The GCG program "motifs" revealed homology to the insulinase family of proteins, divalent cation-dependent proteases that process peptides including insulin (17). Comparison of the predicted amino acid sequence to the GenBank and PIR protein sequence data banks revealed homologies to a number of proteins in the insulinase family, including insulinase (18), E. coli pitrilysin (19), a mitochondrial protein processing enzyme (20), and its homolog in Bacillus subtilis (21). It is therefore doubtful that this protein has a direct function in Ca 2ϩ -ATPase activity; however, it may be that this protein is responsible for cleavage of the signal sequence from the Ca 2ϩ -ATPase.
Sequence Analysis and Predicted Structure of the Ca 2ϩ -ATPase-Comparison of the cda gene's predicted amino acid sequence to GenBank and PIR protein data bases revealed no significant sequence homology to any known P-type ATPase. In fact, the only conserved sequence from the P-type ATPases that is present in the F. odoratum Ca 2ϩ -ATPase is the consensus phosphorylation site used to generate the original partial clone. However, even this sequence, starting at residue 158, is altered relative to the P-type ATPases. The GCG program "motifs" identifies this region as an ATP binding site, commonly known as a Walker A or P loop motif ((G/A)XXXXGKT) (22). Most ATP binding proteins contain a Walker A sequence; however, the P-type ATPases are an exception, containing instead the highly conserved 7-amino acid sequence DKTGT(I/L)T as part of the ATP binding site. The F. odoratum sequence instead contains DGKTGDWIT including the conserved aspartate (Asp-158), which has been implicated in the autophosphorylation of the protein as an integral step in the reaction cycle of P-type ATPases.
The F. odoratum Ca 2ϩ -ATPase exhibits moderate sequence homology (30%) with an alkaline phosphatase (phoD) from Zymomonas mobilis (23). Despite weak homology to short regions of several Ca 2ϩ -ATPases, including the Synechococcus pacL, which has been hypothesized to be a Ca 2ϩ -ATPase, the Z. mobilis sequence contains none of the highly conserved regions found in other P-type ATPases, including the phosphorylation site. The Kyte-Doolittle hydropathy plot of the F. odoratum Ca 2ϩ -ATPase (Fig. 2) definitively indicates only one transmembrane helix in the signal sequence that is cleaved off in the isolated protein. This result is confirmed by the program PredictProtein from the EMBL-Heidelberg (24,25). However, many other regions of the protein are weakly hydrophobic, and the protein partitions with the membrane during cellular disruption (Ref. 1 and data not shown), suggesting that it is a peripheral membrane protein, possibly part of a multi-subunit membrane complex.
Functional Expression of the Ca 2ϩ -ATPase in E. coli-E. coli, which has no endogenous Ca 2ϩ -ATPase activity, was transformed with pBK (no insert) or pBK-C (contains the complete open reading frame for cda and part of the putative insulinase). French-pressed membrane vesicles from XL1 Blue cells transformed by pBK-C demonstrated very high levels of calcium-dependent ATP hydrolysis (Fig. 3). The level of activity is severalfold higher, on a per protein basis, than that endogenously expressed in F. odoratum. The K m for ATP was determined to be 100 M by ATP hydrolysis assays (data not shown), similar to the 90 M value found for the purified protein (12). The activity was sensitive to vanadate, a phosphate analog that inhibits all P-type but not F-type ATPases. The 0.8 M K i for vanadate (data not shown) is also similar to the purified protein. Importantly, vesicles from cells transformed with vector alone (pBK) showed no calcium-dependent ATPase activity. Moreover, upon addition of ATP, vesicles from cells with pBK-C form a vanadate-sensitive 60-kDa phosphointermediate in the presence of calcium but not EGTA, as observed in F. odoratum membrane vesicles (Fig. 4). No phosphointermediate is observed in vesicles from E. coli transformed with pBK alone.
Western blots of vesicles from pBK-C cells cross-react with antibodies prepared against the purified F. odoratum Ca 2ϩ -ATPase, while cells with pBK (no insert) do not show a reaction (Fig. 5). The band labeled by the antibodies migrates at 60 kDa, as does the phosphorylated protein produced by the F. odoratum and the clone.
The pBK vector did not tightly control the expression of the cda gene, and induction with isopropyl-1-thio-␤-D-galactopyranoside often resulted in levels of expression that adversely affected cell viability. Therefore, the clone C insert was ligated into the pET-21 vector (pET-C), which more tightly controls expression, allowing us to regulate levels of expression. In addition, the sequence beyond the cda open reading frame of pET-C (including the second open reading frame of clone C) was deleted (pET-cda) so that only the Cda protein would be expressed. Importantly, vesicles from E. coli transformed with pET-cda demonstrated high levels of calcium-dependent, vanadate-sensitive ATP hydrolysis (Fig. 6). The cda gene product also formed a 60-kDa phosphointermediate in the presence of calcium and cross-reacted with our anti-F. odoratum Ca 2ϩ -ATPase antibody (not shown). These data verify that cda encodes the F. odoratum Ca 2ϩ -ATPase. Calcium transport assays were performed with vesicles treated with 0.1 mM DCCD and 0.5 mM N-ethylmaleimide, which inhibited the endogenous Ca 2ϩ antiporters but not the heterologously expressed Ca 2ϩ -ATPase. Unfortunately, we were unable to detect ATP-driven Calcium-dependent ATP hydrolysis activity of the 90% fraction from E. coli XL1 Blue cells transformed by the vector pBK with (pBK-C) or without (pBK) clone C were compared to F.odoratum on a per protein basis. ATP hydrolysis was carried out with the 90% fraction incubated in 45 l of buffer A (plus 0.5 mM DCCD) with 0.1 mM CaCl 2 (ϩ calcium), 2 mM EGTA (Ϫ calcium), or 0.1 mM CaCl 2 plus 100 M vanadate (ϩ vanadate) for 10 min, and the reaction was initiated with 5 l of a 1 mM [␥-32 P]ATP mixture (final concentration, 100 M at 100 -200 cpm/pmol). After 10 s, the reaction was terminated by the addition of 150 l of 30% trichloroacetic acid, 1 mM K 2 HPO 4 . Inorganic phosphate was extracted and counted as described (12).

FIG. 4. Phosphointermediate formation by the F. odoratum
Ca 2ϩ -ATPase. Calcium-dependent ATP phosphorylation of the 90% fraction from F. odoratum and E. coli XL1 Blue cells transformed by the vector pBK with (pBK-C) or without (pBK) clone C are compared by SDS-PAGE. Assays were performed in buffer A (plus 0.5 mM DCCD) with and without calcium on the 90% fraction, and gel electrophoresis was performed as described under "Experimental Procedures." The gel was loaded with sufficient protein to ensure equal amounts of Ca 2ϩ -dependent ATP hydrolysis activity for the F. odoratum (50 g) and E. coli (10 g). The fixed gels were exposed to x-ray film. Bands represent proteins labeled with 32 P. Both F. odoratum and E. coli with pBK-C demonstrate a 60-kDa phosphointermediate only in the presence of calcium. calcium uptake (data not shown). This may be due to insufficient inhibition of the endogenous Ca 2ϩ antiporters, the membrane being leaky to calcium, or the possible requirement of a second transmembrane protein for calcium transport activity.

DISCUSSION
A 1581-bp gene from F. odoratum has been cloned and sequenced. The gene encodes a protein with the same 60-kDa molecular mass and N-terminal amino acid sequence as the purified Ca 2ϩ -ATPase. The expressed gene product exhibits calcium-dependent and vanadate-sensitive ATP hydrolysis, forms a phosphointermediate in the presence of calcium and ATP, and is immunologically related to the F. odoratum Ca 2ϩ -ATPase. We therefore conclude that the first gene of clone C, which we call cda for calcium-dependent ATPase, encodes the Ca 2ϩ -ATPase, that we have previously purified.
This gene, however, appears to code for a protein very different than other P-type ATPases, despite its functional similarities. Most of the highly conserved regions found in all other P-type ATPases are missing, including the TGES transduction domain, the KGAPE fluorescein isothiocyanate binding site, and the MTGDGVNDAPAL ATP binding domain. Only the putative phosphorylation site shares homology with other Ptype ATPases, and it is altered relative to them. All other P-type ATPases conserve the 7-amino acid sequence DKTGT-(I/L)T, in which aspartate is the residue phosphorylated by ATP during the reaction cycle. In contrast, the F. odoratum Ca 2ϩ -ATPase contains the sequence DGKTGDWIT, notably with a glycine inserted between the positive and negative charges of the aspartate (Asp-158) and the lysine (Lys-160). In the yeast plasma membrane H ϩ -ATPase, insertion of a glycine in this position affected the assembly and/or stability of the H ϩ -ATPase and resulted in no H ϩ -ATPase being detected in secretory vesicles (26). Moreover, the addition of an extra aspartate (Asp-163) after the second glycine makes it unclear which aspartate (if either) is phosphorylated. The sequence GKT (preceded by an A/G five residues earlier), however, is a common motif for ATP binding in many, if not most, ATPases other than the P-type ATPases (22). This Walker A or P loop sequence is found in proteins as varied as the ␣ and ␤ subunits of the F 1 -ATPase, the myosin head protein, the ras oncogene, and many other kinases. It is puzzling that this motif is present in the F. odoratum Ca 2ϩ -ATPase rather than the standard P-type ATPase sequence, given that other proteins using the Walker A sequence do not form an aspartylphosphate intermediate as part of their reaction mechanism.
The three-dimensional structure of the F. odoratum Ca 2ϩ -ATPase may differ significantly from other P-type ATPases. The Kyte-Doolittle plot predicts only one transmembrane helix, that of the signal sequence, with the remaining protein being hydrophilic or only weakly hydrophobic. This, taken together with experiments suggesting that at least some of the protein's activity can be removed from inside-out membrane vesicles by high salt wash (data not shown), suggests that the protein is a peripheral, not an integral, membrane protein. It is highly unlikely that a peripheral membrane protein alone could transport calcium across a membrane, and therefore a second, transmembrane, component would be required. The fact that we have been unsuccessful in demonstrating calcium transport by the reconstituted protein and by E. coli expressing cda supports the theory that the Cda protein is part of a multisubunit membrane complex. The subunits of the F. odoratum Ca 2ϩ -ATPase may be analogous to the subunits of the plasmidmediated E. coli arsenate transporter (although not a P-type ATPase), in which soluble ArsA is capable of arsenate-dependent ATP hydrolysis (27), but ArsB (transmembrane protein) and ArsC are required for transport (21,28). Similarly, potassium transport in E. coli is mediated by the multisubunit FIG. 6. ATP hydrolysis activity of the cda gene expressed in E. coli. Ca 2ϩ -dependent ATP hydrolysis activity of the 90% fraction from E. coli NovaBlue (DE3) cells transformed by the vector pET-21 with the clone C (pET-C) or with the cda gene alone (pET-cda) was compared to F. odoratum on a per protein basis. The transformed E. coli were grown to an A 600 of 0.2, induced with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside, and incubated for 8 h at 30°C. ATP hydrolysis was carried out with the 90% fraction incubated in 45 l of buffer A (plus 0.5 mM DCCD) with 0.1 mM CaCl 2 (ϩ calcium), 2 mM EGTA (Ϫ calcium), or 0.1 mM CaCl 2 plus 100 M vanadate (ϩ vanadate) for 10 min, and the reaction was initiated with 5 l of a 1 mM [␥-32 P]ATP mixture (final concentration, 100 M at 100 -200 cpm/pmol). After 10 s, the reaction was terminated by the addition of 150 l of 30% trichloroacetic acid, 1 mM K 2 HPO 4 . Inorganic phosphate was extracted and counted as described (12). The antibody reacted with a 60-kDa protein in the lanes with fractions of F. odoratum and E. coli with pBK-C but not with pBK. P-type ATPase called Kdp, in which KdpA is postulated to be responsible for K ϩ transport (29) while KdpB binds ATP and is phosphorylated (30). In this case, both KdpA and KdpB are transmembrane proteins (6). In contrast to the F. odoratum Ca 2ϩ -ATPase, it is not known whether KdpB alone is active since it has never been isolated to homogeneity.
Although it has no significant homology with P-type ATPase sequences, the F. odoratum Ca 2ϩ -ATPase does show a moderate homology (overall 30%) to an alkaline phosphatase (phoD) from Z. mobilis (23), which is itself weakly homologous to short regions of several P-type ATPases, in particular Ca 2ϩ -ATPases. The F. odoratum protein is not highly homologous to the Z. mobilis protein in most of those regions. Interestingly, those regions homologous between the Z. mobilis and the Ca 2ϩ -ATPases are not in the most highly conserved regions of P-type ATPases, and the Z. mobilis is missing an obvious phosphorylation site or ATP binding site. The PhoD protein demonstrates high phosphatase activity at alkaline pH using p-nitrophenylphosphate as a substrate, as does the F. odoratum Ca 2ϩ -ATPase (12), as well as other P-type ATPases including SERCA (31,32). The Z. mobilis PhoD protein is found in the cytosol and does hydrolyze ATP but at rates that are dramatically below that of the F. odoratum Ca 2ϩ -ATPase (23).
The Z. mobilis enzyme has a similar hydropathy profile to the F. odoratum protein in that there is only one predicted transmembrane helix, at the N terminus (it is not clear whether there is a cleaved signal sequence), with weakly hydrophobic regions throughout the remainder of the protein.
Given the ease of removal from the membrane, the PhoD protein has been postulated to contain only one transmembrane helix (23). Further experimentation will reveal whether our protein is structurally similar to the PhoD protein whose sequence it conserves, or to the P-type ATPases whose function it conserves.
Perhaps because it is a peripheral and not an integral memebrane protein, the F. odoratum Ca 2ϩ -ATPase is easily expressed in E. coli, in contrast to SERCA, which has only been expressed at relatively low levels in mammalian cell cultures (33), and baculovirus-infected insect cells (34), which both contain wild-type Ca 2ϩ -ATPase. Molecular biological techniques have greatly increased our understanding of the molecular mechanism of the P-type ATPase, but unfortunately the level of expression of the ATPase in most eukaryotic expression systems is too limited to allow for extensive biochemical characterization of site-directed mutants. However, functional expression of the F. odoratum Ca 2ϩ -ATPase in E. coli at very high levels will allow for easy purification and biochemical characterization of any site-directed mutant. This makes the F. odoratum Ca 2ϩ -ATPase a very promising system for structurefunction studies.
In conclusion, the F. odoratum Ca 2ϩ -ATPase appears to be a most unique ATPase. Functionally, it behaves like a P-type ATPase, forming an alkaline-labile phosphointermediate, and displaying vanadate-sensitive activity with K m values for ATP and Ca 2ϩ similar to those for SERCA. The primary structure, however, is different from the P-type ATPases, containing none of the highly conserved regions that appear to be involved in ATP binding and hydrolysis. The putative phosphorylation site is similar but not identical to that of P-type ATPases, resembling the Walker A/P loop motif found in other ATP binding proteins including V-and F-type ATPases. The F. odoratum Ca 2ϩ -ATPase may possibly represent an ancestral link between the F-type and the P-type ATPases or a new class of ATPases. Further study is anticipated to clarify these structural puzzles and to elucidate how such apparently different structures can accomplish the same functional goals.