Purification, cloning, and expression of a novel, endogenous, calcium-sensitive, 28-kDa phosphoprotein.

In gastric parietal cells, cholinergically induced increases in intracellular free calcium concentrations have been well characterized, but little is known about the signaling events beyond the initial rise in intracellular calcium. In the present study, we report the isolation of a 28-kDa protein, which is rapidly phosphorylated in intact, enriched parietal cells in response to both the cholinergic agonist, carbachol, and the calcium ionophore, ionomycin. A combination of in situ 32P labeling and one- and two-dimensional gel electrophoresis was used to acquire sufficient quantities of protein to obtain partial amino acid sequence. Cloning of the pp28 cDNA revealed a novel protein which we have named CSPP28 based on its calcium-sensitive phosphorylation. There are three CSPP28 mRNA species (1.7, 2.2, and 3.3 kilobases) that are widely distributed throughout a variety of rabbit tissues. Recombinant CSPP28 was phosphorylated by both crude parietal cell homogenate and purified CaM kinase II in a calcium/calmodulin-dependent manner. We propose that CSPP28 may play an important and ubiquitous role in the calcium signaling pathway.

In gastric parietal cells, cholinergically induced increases in intracellular free calcium concentrations have been well characterized, but little is known about the signaling events beyond the initial rise in intracellular calcium. In the present study, we report the isolation of a 28-kDa protein, which is rapidly phosphorylated in intact, enriched parietal cells in response to both the cholinergic agonist, carbachol, and the calcium ionophore, ionomycin. A combination of in situ 32 P labeling and one-and two-dimensional gel electrophoresis was used to acquire sufficient quantities of protein to obtain partial amino acid sequence. Cloning of the pp28 cDNA revealed a novel protein which we have named CSPP28 based on its calcium-sensitive phosphorylation. There are three CSPP28 mRNA species (1.7, 2.2, and 3.3 kilobases) that are widely distributed throughout a variety of rabbit tissues. Recombinant CSPP28 was phosphorylated by both crude parietal cell homogenate and purified CaM kinase II in a calcium/calmodulin-dependent manner. We propose that CSPP28 may play an important and ubiquitous role in the calcium signaling pathway.
Intracellular signaling is generally mediated by activation of specific receptors leading to alterations in intracellular concentrations of different second messengers, including calcium, inositol phosphates, diacylglycerol, and cyclic AMP. These second messengers modulate many physiological processes that involve the phosphorylation of enzymes, receptors, and substrates by multifunctional protein kinases, namely calcium/ calmodulin-dependent protein kinase II (CaM kinase II), calcium/phospholipid-dependent protein kinase, and cyclic AMP-dependent protein kinase (for reviews, see Refs. [1][2][3]. Although there is an abundance of information about second messengers and second messenger-dependent protein kinases, much less is known about the specific protein kinase substrates in these signaling pathways. In many secretory cells, cholinergic stimulation of muscarinic receptors activates phospholipase C, which hydrolyzes phosphoinositol 4,5-bisphosphate to liberate inositol 1,4,5bisphosphate and diacylglycerol (2). Similarly, in HCl-secreting gastric parietal cells, cholinergic agonists elevate inositol 1,4,5bisphosphate concentrations which, in turn, stimulate the rapid release of calcium from internal stores (4 -8). The cascade of signaling events following the rise in intracellular free calcium concentrations in parietal cells and in other cell types remains obscure. It is clear, however, that protein phosphorylation is a critical component of second messenger-dependent cascades. In parietal cells, at least three different proteins with molecular masses of 28, 36, and 66 kDa are phosphorylated in response to cholinergic stimulation (4,8,9). These phosphorylation events appear to occur by way of different protein kinaseactivating mechanisms. Since the 36-and 66-kDa phosphoproteins are phosphorylated in isolated intact parietal cells following addition of phorbol ester under calcium-chelating conditions, it appears that these phosphoproteins are protein kinase C substrates (4,8,9). In contrast, the 28-kDa (pp28) protein is not phosphorylated in response to protein kinase C activators, but is phosphorylated in response to either cholinergic agonists or the calcium ionophore, ionomycin (9). Furthermore, pp28 phosphorylation is inhibited by chelation of intracellular calcium with the cell permeant form of BAPTA 1 (9). Until now, none of these phosphoproteins have been identified or characterized.
The present study describes the purification of pp28 and the isolation and initial characterization of a full-length cDNA clone. Since previous work determined that the 28-kDa protein was present in parietal cells in low abundance, a novel purification strategy was developed that utilized a combination of in situ 32 P labeling and preparative two-dimensional gel electrophoresis. These techniques allowed for the isolation of sufficient quantities of highly purified protein to obtain partial amino acid sequence information which was then used to clone the open reading frame of pp28. Recombinant pp28 was phosphorylated in a calcium-dependent manner by crude parietal cell homogenate and purified CaM kinase II. Our data suggest that pp28 represents a novel phosphoprotein. We have designated this protein as CSPP28, a calcium-sensitive phosphoprotein of 28 kDa (10).

Isolation of Parietal Cells and Gastric
Glands-Parietal cells were prepared from fundic mucosae of male 2-3-kg New Zealand White rabbits as described previously (5,11). This method yields approxi-* This work was supported by National Institutes of Health Grants DK 31900 (to C. S. C.), DK 48370, and DK 43405 and a Department of Veterans Affairs Merit Review (to J. R. G.). The costs of publication of this article were defrayed 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U35428.
In Situ 32 P Labeling and Isolation of pp28 Marker from Parietal Cell Extracts-Since pp28 was in low abundance, a novel methodology was developed to purify sufficient quantities for sequencing. In situ 32 Plabeled pp28 from enriched parietal cells was used as an internal marker throughout the purification. Protein for pp28 purification was extracted from gastric glands rather than parietal cells because this phosphoprotein was previously detected in both chief and parietal cells (9) and protein yields were substantially higher in glands as compared to parietal cells. Parietal cells (ϳ10 6 cells/ml) were labeled with carrierfree [ 32 P]orthophosphate, as described previously (9). Aliquots (1.0 ml) of cells were transferred to microcentrifuge tubes and incubated with 3 M ionomycin (Calbiochem), 5 min, 37°C. Incubations were terminated by rapid centrifugation, followed by a brief wash with cold phosphatebuffered saline and lysis in boiling 0.3% SDS, 1% 2-mercaptoethanol. SDS-solubilized extracts were precipitated with 4 volumes of acetone (30 min, 23°C) to reduce the SDS concentration prior to electrofocusing (9,13).
Microsequencing of pp28 -Peptides for sequencing were prepared by two different procedures: in-gel V8 protease digest (14) and in-gel tryptic digest (15,16). The in-gel V8 protease digest was transferred to a polyvinylidine difluoride membrane (Millipore Corp., Bedford, MA). An 18-kDa peptide fragment was isolated from the membrane and microsequenced at the Emory University Core facility (Atlanta, Georgia). Following in-gel tryptic digest, peptides were isolated by reverse phase chromatography on a Pharmacia SMART system (Pharmacia Biotech, Sweden) using the integrated Peak detector with simultaneous monitoring at two wavelengths of 215 and 280 nm on a RPC C2/C18 SC 2.1/10 column (0 -40% acetonitrile in 0.065 to 0.05% trifluoroacetic acid, flow rate of 100 l/min). These peptide peaks were sequenced at the Ludwig Institute for Cancer Research in Uppsala, Sweden.
After amplification of the cDNA, a range of PCR products was generated. These products were reamplified with the same degenerate sense primer SP-2 plus a degenerate antisense primer that was designed from a tryptic digest-derived amino acid sequence SFEEKVE (TCIACRTTRTCRTCYAAIGA, ASP-1). Upon reamplification, a 294base pair product was generated, subcloned into pBluescript and sequenced.
PCR Screening-PCR screening was performed as described by Friedman and colleagues (17) using a subdivided Lambda Zap II cDNA Library (Stratagene, La Jolla, CA) prepared from Ͼ95% pure rabbit parietal cells. To facilitate screening, the amplified library was divided into 20 aliquots of 50,000 plaque-forming units. To disrupt the phage, 5 l from each library aliquot were diluted with 30 l of water, heated (70°C, 5 min), and immediately placed on ice. Sense (CAAAGGGTG-GCAAGATGTAAC, SP5) and antisense (GTCTTCCAGCTTTTTGGT-GAT, ASP-5) primers were designed from the 294-bp sequence. Because the product size was small (130 bp) and to decrease nonspecific annealing, thermal cycling times were shortened (95°C for 20 s, 55°C for 10 s and 72°C for 10 s, 35 cycles). Ethidium bromide-stained (0.5 g/ml) agarose gels (3% w/v) were used to detect the correct product size. Two aliquots from the rabbit cDNA library were found to contain product appropriately sized.
Library Screening-The 294-bp fragment obtained in the initial PCR amplification of cDNA with SP-2 and ASP-1 was PCR labeled with digoxigenin dNTPs (Life Technologies Inc.) and used as a probe to screen 50,000 plaques from each of the two positive library aliquots. Recombinant Protein Expression-Recombinant protein was expressed in a prokaryotic system using the pET19b expression vector (Novagen, Madison, WI). The 555-bp open reading frame of pp28 was amplified by PCR from parietal cell cDNA with a sense primer (GGC-CATATGGACCGCGGCGAGCAAGGTCGT) containing a 5Ј NdeI restriction site and an antisense primer (CCGGATCCTCACAGGCCCTC-CTGTGTCTG) containing a 3Ј BamHI restriction site. The resulting full-length sequence was inserted into pET19b in-frame with the 5Јpolyhistidine (His-tag) sequence. Ligated plasmids were transformed into BL21(DE3)pLysS (Novagen) and selected with carbenicillin (50 g/ml) and chloramphenicol (34 g/ml). For protein production, bacterial cultures (1 liter, log phase) were induced with 1 mM isopropylthio-␤-D-galactoside for 2 h at room temperature. Bacteria were isolated by centrifugation, resuspended in binding buffer (5 mM imidazole, 0.5 M NaCl, and 20 mM Tris-HCl pH 7.9), and sonicated. Homogenates were centrifuged (39,000 ϫ g, 20 min) and supernatants passed over a Hisbind nickel chelate resin column (2 ml). Recombinant protein was eluted with 1 M imidazole buffer. Purity was verified by SDS-PAGE.

RESULTS
Purification and Protein Sequencing-Based on our previous knowledge that: 1) pp28 was a low abundance, acidic phosphoprotein that was present in both parietal cells and chief cells; 2) pp28 was phosphorylated in response to elevated calcium; and 3) pp28 exhibited a characteristic migratory pattern on analytical two-dimensional-gels (9), we designed new strategic approaches to obtain sufficient protein for microsequencing. Thus, parietal cell proteins were radiolabeled in situ with carrier-free [ 32 P]orthophosphate, then stimulated with ionomycin. Relatively large amounts of radiolabeled proteins (ϳ1 mg/gel) were resolved using preparative two-dimensional SDS-PAGE. pp28 spots were located by autoradiography, pooled with ammonium sulfate-precipitated proteins from ionomycin-stimulated glands and resolved by molecular mass to a single radiolabeled peak detected by Cerenkov counting. Peak fractions were pooled then resolved according to molecular mass and pI using preparative two-dimensional-PAGE. By combining oneand two-dimensional preparative electrophoresis protocols and using radiolabeled pp28 as an internal marker, sufficient amounts of Coomassie Blue-stained pp28 spots were obtained for initial microsequencing (Fig. 1, top left).
Initial attempts to microsequence intact protein transferred to polyvinylidine difluoride membranes were unsuccessful, apparently because pp28 was N terminally blocked. In-gel V8 protease (14) and in-gel tryptic digests (15,16) were used to obtain internal amino acid sequence for these digests. Six to eight spots of radiolabeled pp28 (ϳ1-2 g/gel) prepared as described above (Fig. 1, top) were used for each protocol. The V8 protease (not shown) digest yielded a major 18-kDa peptide from which 25 amino acids were sequenced. Fragments from the in-gel tryptic digest were resolved with a Pharmacia SMART system and selected peptide peaks were sequenced as shown in (Fig. 1, bottom). Amino acid sequence information FIG. 1. Purification of CSPP28. Top left, Coomassie Blue stain of a two-dimensional gel of 32 P-labeled protein that was separated on a Bio-Rad Prep cell (see "Materials and Methods"). Both Coomassie Blue-stained spots (arrow) were used for sequencing because initial amino acid composition analyses of the two spots indicated that they were identical (C. S. Chew and A. C. Petropoulos, unpublished observations). Top right, autoradiograph of Coomassie Blue-stained two-dimensional gel depicted on the left. The arrow identifies 32 P labeled CSPP28. Bottom, chromatogram of peptides resolved with a Pharmacia Smart system after in-gel tryptic digest of CSPP28. Letters indicate peaks that were used for amino acid sequencing.
derived from both protocols is summarized in Fig. 2.
Molecular Cloning of CSPP28 -Based on the amino acid sequence of the V8 digest, two degenerate oligonucleotides were designed and used as sense primers for the initial 3Ј-RACE. PCR products ranging in size from 1.5 kb to 200 bp were isolated and re-amplified with the degenerate sense primers along with a degenerate antisense oligonucleotide designed from peptides obtained from the tryptic digest (Fig. 2, bottom). This re-amplification generated a specific 294-bp fragment (Fig. 3A). The deduced amino acid sequence from the 294-bp product contained 5 of the 6 sequenced tryptic fragments (Fig. 3A).
The 294-bp fragment was then used to screen a parietal cell cDNA library, resulting in the identification of CSPP28 clones of 1400 and 1860 bp. These clones were identical from their 5Ј ends through the open reading frame to the first polyadenylation signal sequence site. The 1400-bp clone was polyadenylated after the first polyadenylation signal sequence, whereas the 1863-bp clone contained an additional polyadenylation signal sequence which was 460 bp beyond the first signal sequence (Fig. 3A). Since the clones did not contain an initiating methionine, additional sequence was obtained using a 5Ј-RACE strategy. The resulting additional 219 bp contained a putative start codon and an additional 5 nucleotides upstream of the start codon, which contained a Kozak consensus sequence. The deduced amino acid sequence from the 555-nucleotide open reading frame had a predicted molecular mass of 19.8 kDa. All 7 peptide sequences resolved from the tryptic and V8 digests of purified pp28 were present, without error, in the deduced amino acid sequence. In our initial searches of the GenBank there was no significant homology between CSPP28 and any other known protein (10). However, more recent searches have detected a 95% amino acid sequence identity (Fig. 3B) between CSPP28 and the predicted amino acid sequence of an overexpressed cDNA transcript (D52) initially isolated from an infiltrating ductal breast carcinoma (19).
Northern Blot Analysis-The distribution of CSPP28 RNA in various rabbit tissues was determined by using Northern blot analysis. Messenger RNA from gastric glands, chief cells, and parietal cells was screened with the full-length CSPP28 probe. Three messages of 3.3, 2.2, and 1.7 kb (Fig. 4) were found in both parietal cells and chief cells. The same size transcripts were detected in total RNA from pancreas, spleen, liver, colon, brain, duodenum, jejunum, ileum, antrum, and fundus (Fig. 5), as well as in lung, kidney, and skeletal muscle (data not shown).
In Vitro Phosphorylation-Analysis of the CSPP28 amino acid sequence did not detect any consensus phosphorylation sites. However, since all available evidence suggested that CSPP28 phosphorylation was calcium dependent (4, 9), we used in vitro phosphorylation analyses to determine whether FIG. 2. Sequence from in-gel digests. Top, sequence of the first 25 amino acid residues of the 18-kDa fragment obtained from the V8 digestion. Bottom, sequences of peptide fragments obtained from the in-gel tryptic digest. Letters identifying peptide fragments also correspond to peaks shown in Fig. 1 (bottom).   FIG. 3. Molecular cloning of CSPP28. A, cDNA and deduced amino acid sequences of CSPP28 derived from PCR cloning and library screening. The 5Ј end of the clones obtained from the library are depicted by the single arrowhead. Amino acid sequence from the V8 digest is double underlined. Amino acid sequences from the tryptic digest are underlined. The cDNA sequence upstream of the 5Ј end of the double arrows (22) was obtained by 5Ј-RACE. The deduced amino acid sequence from the 294-bp fragment, obtained from the initial PCR cloning, is in bold text. B, sequence comparison of CSPP28 and D52, the deduced sequence of a mRNA message from a breast cell carcinoma library. The upper lane is the D52 sequence and the lower is CSPP28. C, sequence comparison of CSPP28 and deduced amino acid sequence from C. elegans cosmid F13E6. The upper lane is the cosmid sequence and the lower is CSPP28.
CSPP28 phosphorylation could be increased in a calcium/calmodulin-dependent manner. Fig. 6 shows that recombinant CSPP28 was phosphorylated by gastric gland extracts in a calcium-dependent manner and its phosphorylation was enhanced by addition of calmodulin. In addition, purified CaM kinase II phosphorylated recombinant CSPP28 in a calciumand calmodulin-dependent manner. CSPP28 phosphorylation by both cell extracts and purified CaM kinase II was strongly inhibited by a CaM kinase (281-302) II pseudosubstrate inhibitor peptide (40 M) (Fig. 6).

DISCUSSION
Although receptor-mediated elevation of intracellular free calcium concentrations is a universal second messenger signaling event and a number of calcium-dependent protein kinases have been identified, little is currently known about the downstream substrates for these kinases. This is particularly so in gastric parietal and chief cells, in which only the calcium dependence for cholinergically modulated secretion of HCl has been well characterized (4, 5, 8, 20 -24). In this work, we define a novel combination of methodologies, based on preparative one-and two-dimensional gel electrophoresis in conjunction with in situ 32 P labeling, which allows for the isolation of sufficient amounts of low abundance agonist-responsive phosphoproteins for microsequencing and cDNA cloning. Through the use of such strategies we have successfully identified and partially characterized CSPP28, a novel acidic phosphoprotein member of the calcium signaling cascade. Northern analyses indicate that CSPP28 mRNA is widely distributed throughout the gastrointestinal tract as well as in brain. Thus, CSPP28 may serve an important and ubiquitous function in calcium signaling cascades in a variety of cell types.
Previous work in our laboratory demonstrated that CSPP28 is rapidly and transiently phosphorylated in parietal and chief cells upon cholinergic stimulation. CSPP28 is also strongly phosphorylated when intracellular free calcium concentrations are elevated by calcium ionophores. The calcium dependence of CSPP28 phosphorylation was further demonstrated by a complete inhibition of this response upon chelation of intracellular and extracellular calcium using a combination of EGTA and the cell-permeant form of BAPTA (9). A phosphoprotein with similar molecular weight and calcium sensitivity has also been detected in two-dimensional gel analyses of extracts of other cell types from several species, as well as in cultured cell lines. For example, Williams and colleagues have reported calciumdependent phosphorylation of a pancreatic aciniar cell protein in mice and guinea pigs with properties similar to those of CSPP28 (25). Also in a colonic epithelial cell line (HT-29), Cohn and colleagues (26) have detected a similar protein which is phosphorylated upon activation of H1 receptors.
A potential role for CaM kinase II in parietal cell secretion was proposed by Tsunoda and colleagues (27) who found that the CaM kinase II inhibitor, KN-62, strongly suppressed cholinergic stimulation, but not histaminergic stimulation of parietal cell accumulation of [ 14 C]aminopyrine, an indirect measure of HCl secretion (28). In addition, we were unable to detect either phorbol ester-or cAMP-induced increases in CSPP28 phosphorylation in intact parietal cells (9). On the basis of these data and observations that CaM kinase II is present in parietal cells (29 -31), we hypothesized that CSPP28 is a CaM kinase II substrate. Our results thus far support this hypothesis that both purified CaM kinase II and parietal cell extracts increase phosphorylation of recombinant CSPP28 in that this phosphorylation exhibited calcium/calmodulin dependence. In addition, CSPP28 phosphorylation was strongly inhibited by a specific CaM kinase (281-302) II pseudosubstrate inhibitor peptide.
Although the cDNA and deduced amino acid sequence of CSPP28 was initially not found to have significant homology to any protein in the GenBank (10), more recent searches have detected a 95% amino acid sequence identity (Fig. 3B) between CSPP28 and the predicted amino acid sequence of an overexpressed cDNA transcript (D52) initially isolated from an infiltrating ductal breast carcinoma (19) ZR-75-1), HeLa, CACO-2, and KATO III carcinoma cell lines (19). Moreover, significant homology has also now been found between CSPP28 cDNA and two expressed sequence tags 139973 and 132820 from a human placenta library. 2 A 52% identity over 71 bp has also been found between the deduced amino acid sequence of CSPP28 and the predicted amino acid sequence of the cDNA cosmid originating from the Caenorhabditis elegans genome mapping project (Fig. 3C) (33). The close homology between these diverse predicted sequences suggests a strong conservation between species and further supports a potentially important role for CSPP28 in cellular signaling events.
Another relevant observation is that the cDNA transcript D52 was reported to be 3.3 kb in size. Thus, D52 cDNA is similar in size to the largest of the CSPP28 messages that we detected in a number of rabbit tissues (Fig. 5). D52 cDNA also contains polyadenylation signal sequence sites at nucleotides 1671-1676 and 2171-2176, suggesting possible messages of 1.7 and 2.2 kb. These data strongly suggest that D52 cDNA transcribes the same messages as CSPP28 and that there are three different CSPP28 messages present in multiple cell types.
In parietal cells the 2.2-and 1.7-kb messages coded for identical open reading frames, differing only in the length of their 3Ј-untranslated regions (UTR). These differences in the 3Ј-UTRs are potentially important. Recent work by Kislauskis and colleagues (34), suggests that differences in 3Ј-UTRs may direct the targeting of protein expression to different cellular compartments. Another potentially important function of the 3Ј-UTR may be in message stability (32,35). The nucleotide sequence ATTTA has been suggested as an mRNA destabilizing motif (32,35). In the 3Ј-UTR of the 2.2-kb message, there are six copies of the ATTTA motif. The 1.7 kb contains three motifs, whereas the 3Ј-UTR of the D52 message has 15 (19). The reasons for the presence of messages with different stabilities is unclear. One possibility is that these differences may modulate relative abundance of particular messages in individual tissues.
In summary, we have developed specific methods to obtain the cDNA and protein sequence of a novel calcium-sensitive phosphoprotein, CSPP28, and have shown that this protein is phosphorylated in vitro by CaM kinase II. Further studies are necessary to determine the in vivo site or sites on which CSPP28 is phosphorylated and to define unambiguously the specific protein kinase(s) that mediate CSPP28 phosphorylation. Since CSPP28 mRNA is distributed in many tissues and across species, it may be an ubiquitous mediator of the calcium signaling pathway.