Maize cap1 Encodes a Novel SERCA-type Calcium-ATPase with a Calmodulin-binding Domain*

A cDNA (CAP1) isolated from maize roots shares sequence identity with genes encoding P-type Ca 2 1 -AT-Pases and restores the growth phenotype of yeast mutants defective in Ca 2 1 -pumps. CAP1 was transcribed and translated in the yeast mutant. Furthermore, the membrane-integrated product formed a Ca 2 1 -depend-ent phosphorylated intermediate and supported Ca 2 1 transport. Although CAP1 shares greater sequence identity with mammalian “endoplasmic reticulum-type” Ca 2 1 -pumps, it differs from these genes by having features of calmodulin (CaM)-regulated Ca 2 1 -pumps. CAP1 from yeast microsomes bound CaM, and the CAP1-de-pendent Ca 2 1 transport in yeast was stimulated by CaM. Peptides from the C terminus of CAP1 bound CaM. Anti-CAP1 antibodies specifically recognized a maize microsomal polypeptide that also bound CaM. A similar polypeptide also formed a Ca 2 1 -dependent phosphoenzyme. Our results suggest that cap1 encodes a novel form of CaM-regulated Ca 2 1 -ATPase in maize. CAP1 appears to be encoded by one or two genes in maize. CAP1 RNA is induced only during early anoxia, indicating that the Ca 2 1 -pump may play an important role in O 2 deprived maize cells. Hybridization— DNA gel blot from three different regions of CAP1 clone, the 5 9 sequence coding for hydrophilic loop, 3 used probes. hybridization and washings were done at moderate or high stringency. mutant K616 cells transformed with p426Gal1 or pCAP1 in SC-Ura medium glucose or galactose as carbon CaCl at m of trans- formed

Previous studies have shown that an elevation of cytosolic calcium ([Ca 2ϩ ] i ) precedes molecular and whole plant responses to oxygen deprivation in maize (1,2). Furthermore, anoxia-induced [Ca 2ϩ ] i elevation in maize cells falls into two distinct patterns, differing in their magnitude and kinetics (1). Tight regulation of [Ca 2ϩ ] i is essential to the proposed role for Ca 2ϩ as a cellular messenger (3). The maintenance of [Ca 2ϩ ] i at submicromolar levels and regulation of spatio-temporal patterns of Ca 2ϩ signals are mediated by various Ca 2ϩ transporters, such as Ca 2ϩ -pumps and proton-coupled antiporters (see e.g. Ref. 4). Hence, there has been a great interest in molecular cloning of Ca 2ϩ transporters and characterizing their role in Ca 2ϩ -mediated signaling pathways (5)(6)(7)(8)(9). As part of our analysis of the pathway and components of Ca 2ϩ -mediated anoxiasignaling in maize, we initiated studies to identify Ca 2ϩ stores (10) and transporters that may regulate the cytoplasmic Ca 2ϩ signal. Here, we report the isolation and characterization of a cDNA clone (CAP1) that encodes a calmodulin-binding Ca 2ϩ -ATPase in maize roots.
Ca 2ϩ -ATPases belong to two distinct classes, differing in their size, sequence, cellular location, and regulation by calmodulin (CaM). 1 The "ER-type" pump (located on the ER membranes) is a ϳ90-kDa polypeptide, lacks a CaM-binding domain, and is not dependent on CaM for its regulation. The "PM-type" pump (located on the plasma membrane) is a 138-kDa molecule, possesses a C-terminal CaM-binding region, and is regulated by CaM. Homologs of ER-type as well as PM-type Ca 2ϩ -pumps (but with the CaM-binding domain at the N terminus) have been cloned from plants, recently (5)(6)(7)(8)(9). In maize, a calmodulin-regulated Ca 2ϩ -pump, related in size and antigenicity to the animal PM-type Ca 2ϩ -ATPase, was purified from endomembranes of young etiolated shoots (11)(12)(13). On the other hand, biochemical evidence was presented for the existence of an intriguing ER-type pump with CaM-binding properties, in the enriched ER membranes of young etiolated maize shoots (14). Our results indicate that cap1, by virtue of its sequence identity to ER-type Ca 2ϩ -pumps and at the same time possessing a C-terminal CaM-binding domain characteristic to PM-type pumps, may encode a novel chimeric Ca 2ϩ -ATPase such as the one implied by previous studies (14). In addition, we have functionally characterized the CAP1 protein using a yeast expression system and identified a putative cognate protein in maize microsomes. Furthermore, our expression analysis suggests that the abundance of CAP1 mRNA is low in maize roots and is mildly increased under anoxia.
Isolation and Analysis of cDNA Clones-A cDNA library constructed in ZAP vector (Stratagene) from the root tissue of 6-h anoxically * This work was supported by National Research Initiative Competitive Grants Program (NRICGP) Grant 96-35100-3143 from the United States Department of Agriculture (to M. M. S. and C. C. S.). 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) AF096871.
§ To whom correspondence should be addressed: Dept. of Crop Sciences, University of Illinois, Urbana, IL 61801. Tel.: 217-333-9743; Fax: 217-333-6064; E-mail: subbaiah@uiuc.edu. treated 3-day-old maize cv. B73 seedlings (17) was screened using a partial cDNA clone LCA1 that encodes a putative Ca 2ϩ -ATPase in tomato (9). Fifteen positive clones were obtained by screening 2 ϫ 10 5 plaques, and the one containing the longest insert (3.3 kilobase pairs, CAP1) was sequenced using dideoxy chain termination method (United States Biochemical Corp.). Sequence comparison and analyses were carried out using BLAST search program available in the public domain and MacDNASIS Pro program from Hitachi Software Engineering America.
Cloning and Sequencing of 5Ј-RACE-PCR Products-A 22-base primer derived from the 5Ј-end of CAP1 cDNA (20 bases inside) was 32 P-end-labeled and used to reverse transcribe the missing 5Ј-end of CAP1 mRNA from maize root total RNA preparations. Precise products that could later be PCR-amplified were obtained only at priming temperatures Ͼ65°C, using the thermostable reverse transcriptases (Ther-moScript RT, Life Technologies, Inc.; or Carboxydothermus hydrogenoformans polymerase, Roche Molecular Biochemicals). The primer extension products were separated on 6% acrylamide-urea sequencing gels. 32 P-Labeled 10 and 100 base ladders were co-run to size the products. The 140-base-long product was eluted, PCR-amplified using a hot-start method (PLATINUM Taq polymerase, Life Technologies, Inc.), and cloned into Topo-TA vector (Invitrogen).
Expression of CAP1 Product in Yeast and Complementation of Yeast Mutants-The entire open reading frame of CAP1 and a part of the 3Ј-untranslated region were PCR-amplified from the original cDNA clone. An initiation codon at the 5Ј-prime and EcoRI sites at both termini were introduced during amplification. The product was ligated into the EcoRI site of a yeast expression vector p426Gal1 between the galactokinase (Gal1) promoter and CYC1 termination sequence. The junctions of the recombinant plasmid was sequenced to confirm the orientation. The plasmid with correct orientation of the insert was used to transform wild type and mutant yeast strains, and the transformants were selected for uracil prototrophy on SC-URA plates. The Ura ϩ colonies were used for complementation studies.
Overexpression of CAP1 Fragments in Escherichia coli and Purification of Recombinant Proteins-Two fragments of CAP1, one that encodes a major part of the central hydrophilic loop (i.e. residues 342-576, with a calculated molecular mass of 25 kDa) and the other from the C-terminal region comprising residues from 947 to 1038 (ϳ14 kDa), were PCR-amplified and cloned into an expression vector pQE-30 (Qi-aExpress, Qiagen). The vector contains an N-terminal histidine tag downstream to the T5 promoter and the Lac operator. The chimeric plasmids were transformed into appropriate hosts, and expression was induced by isopropyl-1-thio-␤-D-galactopyranoside. The recombinant proteins were purified using a nickel-affinity column according to manufacturer's protocols (Qiagen). The expression of the C-terminal region was limited by its apparent toxicity to E. coli. The affinity purification of this peptide on nickel columns was also not successful, since a large number of host proteins co-eluted with the recombinant product due to hydrophobic interactions. However, the peptide could be purified to homogeneity using CaM-Sepharose chromatography (see "Results"). A calmodulin cDNA isolated from roots of maize inbred B73Ht seedlings (Ref. 18, gift from Dr. G. Feix, Germany) was PCR-amplified and subcloned into pQE-30. The clone was overexpressed, and the recombinant CaM was purified to homogeneity using the QiaExpress system. The identity and orientation of all the cloned DNA fragments was confirmed by sequencing.
CaM Mobility Shift Assays-Synthetic peptides identical to parts of the CAP1 C terminus were made using an Applied Biosystems Peptide Synthesizer (Perkin-Elmer, Applied Biosystems Division) at the Peptide Synthesis Facility of the University of Illinois, Urbana. Purity and sequence of the peptides were verified by mass spectrometry, high pressure liquid chromatography, and microsequencing. The peptides were chosen based on their secondary structure (mean hydrophobicity and the hydrophobic moment of axial helical projections). The ability of the peptides to interact with calmodulin was studied using gel mobility shift assays (19). Maize recombinant calmodulin or calmodulin purified from maize roots was used for the assay. CaM was purified to homogeneity from maize roots as described (20).
Circular Dichroism Spectroscopy of Calmodulin and CAP1 Peptides-The binding of CAP1 synthetic peptides to calmodulin was also measured using CD spectra. Spectra of maize recombinant calmodulin (3 M) and different molar equivalents of peptides were obtained in a buffer containing 5 mM Tris-Cl, 1 mM CaCl 2 (or 2 mM EGTA), 1 mM DTT at pH 7.2 using a Jasco J-720 model spectropolarimeter (Jasco Inc.). The signal was recorded using a 1-cm path length, a sensitivity of 20 millidegrees, a resolution of 0.5 nm, and at a scan speed of 20 nm/min. The results presented are averages of three separate scans.
Analysis of Phosphorylated Intermediate-Yeast microsomes prepared as described (7) or maize root microsomal proteins fractionated on CaM-Sepharose columns were used for the analysis of phosphorylation intermediate. The reaction mixture (0.2 or 0.6 ml) contained 5-15 g of protein, 10 nM [␥-32 P]ATP (10 Ci), 1 mM EDTA, 0.5 mM EGTA, 100 mM KCl, 25 mM HEPES-BTP, pH 6.0. The effects of Ca 2ϩ and La 3ϩ were tested by adding CaCl 2 and LaCl 3 to give a final free ion concentration of 100 and 50 M, respectively, as estimated by the MAXCHELATOR program (21). The reaction was terminated, and proteins were resolved in acidic phosphate gels (22,23).
Ca 2ϩ Transport Assay with Yeast Membrane Vesicles-Membrane isolation from K616 cells transformed with vector alone or with pCAP1 and the measurement of 45 Ca uptake were carried out, essentially as described by Liang and Sze (24). The transport buffer contained 250 mM sucrose, 25 mM HEPES-BTP, pH 7.5, 10 mM KCl, 3 mM MgSO 4 , 0.4 mM sodium azide, 100 M EGTA, 5 M gramicidin, and 10 M 45 CaCl (ICN). The final specific activity was 2 Ci/10 nmol Ca 2ϩ per ml. Under the conditions of pH, temperature, ATP, EGTA, and total calcium of our assay, the calculated free Ca 2ϩ concentration varied between 10 nM and 50 M (21). Membrane vesicles equivalent to 15 g of protein were used in 0.25 ml of reaction mix. Uptake was initiated by the addition of 3 mM ATP and incubated at 25°C for up to 20 min. The reaction mixture was spotted on pre-wet GS filters (Millipore) under vacuum, and the filters were washed in ice-cold rinse buffer containing 250 mM sucrose, 25 mM HEPES-BTP, pH 7.5, and 200 M Ca 2ϩ . When used, the calmodulin inhibitor W7 was added at a final concentration of 100 M and the Ca 2ϩ ionophore A23187 at 1 M.
Antibody Generation and Protein Gel Blot Assays-Monoclonal antisera were raised against the 24-kDa central hydrophilic loop expressed in E. coli. Calmodulin-binding proteins purified from maize microsomes were concentrated by trichloroacetic acid precipitation and resolved in 8 or 6 -12% gradient SDS-acrylamide gels. Gels were stained in silver or Coomassie Blue. Proteins resolved in adjacent lanes were used for immunoblot analysis. Protein transfer onto polyvinylidene difluoride (Bio-Rad) and antibody incubations were done as described previously (25).
Preparation of Horseradish Peroxidase-coupled CaM (HRP-CaM) and CaM Overlay Assays-Maize recombinant calmodulin was conjugated to activated horseradish peroxidase (Pierce) by primary amine coupling and was used to probe CaM-target interactions. The eluant from CaM-Sepharose column was concentrated using trichloroacetic acid precipitation. Proteins were separated on SDS-acrylamide gels and blotted onto a polyvinylidene difluoride membrane. Subsequent blocking, incubation with HRP-CaM, and washings of the blot were carried out as described (26). The peroxidase signal was visualized by chemiluminescence.
DNA and RNA Gel Blot Hybridization-Maize genomic DNA was isolated, digested with appropriate enzymes, and subjected to DNA gel blot analysis according to standard protocols (27). Fragments from three different regions of CAP1 clone, namely the 5Ј-end, sequence coding for the central hydrophilic loop, and the 3Ј-end, were used as probes. The hybridization and washings were done at moderate or high stringency. Yeast mutant K616 cells transformed with p426Gal1 or pCAP1 in SC-Ura medium containing glucose or galactose as the carbon source. CaCl 2 at 5 mM was added to support growth of pGal1 transformed cells. Total RNA was isolated from 250-ml cultures grown for 20 h. RNA gel blot analysis was carried out as described previously (2).

RESULTS
cap1 Encodes a Putative P-type Ca 2ϩ -ATPase-The clone, CAP1, was isolated by screening a cDNA library made from the root tissue of anoxic maize seedlings, with a tomato cDNA encoding an ER-type Ca 2ϩ -ATPase (LCA1, 9). As shown in Fig.  1, the full-length CAP1 sequence can be translated into a polypeptide of 1,049 amino acids with a calculated molecular mass of 113,099 Da (GenBank TM accession number AF096871). The deduced protein is 63% identical to the ER-type Ca 2ϩ -ATPase reported from rice (5) (Fig. 1). The sequence identity was 46% with SERCAs (sarco/endoplasmic reticulum calcium-ATPases, see Ref. 28) and 56 -64% with plant homologs from tomato and Arabidopsis (7,9). The sequence identity is greater than 75% in the essential sequence motifs (see below). The CAP1 protein includes all the domains highly conserved in P-type Ca 2ϩ -ATPases. The 10 transmembrane domains predicted by hydropathy analysis are analogous to those in SERCA-pumps (TM1 to TM10 in Fig. 1). A hydrophilic domain between TM4 and TM5 of CAP1 contains a potential aspartyl phosphorylation site within the CSDK motif and two ATPbinding domains ( Fig. 1) characteristic of all P-type ion pumps (29). Six residues within TM4, -5, -6, and -8 required for high affinity Ca 2ϩ transport (28) are all conserved in the maize clone ( Fig. 1). The transmembrane domains in CAP1 have different degrees of identity to the animal SERCAs and the plant Ca 2ϩpumps reported thus far. TM5 and TM6 of CAP1 are 80 and 88% identical to those of SERCAs, 92-96% with those in tomato LCA1 and Arabidopsis ECA1 products, and 100% with those in rice Os-Ca-ATPase. However, CAP1 differs in its sequence and structure (ϳ30% sequence identity) from the PM-type Ca 2ϩpumps including those recently reported from cauliflower (8) and Arabidopsis (6).
The sequence shown in Fig. 1 is compiled from the original cDNA, CAP1 (which is short only by 19 amino acids from the complete sequence), and an extension product of the 5Ј-sequence by rapid amplification of cDNA ends (RACE)-PCR reaction. CAP1 is shorter at the N terminus by 25-32 residues than its plant homologs from rice, tomato, and Arabidopsis, but it shows a longer C-terminal tail (by ϳ40 residues) relative to these SERCA-type pumps.
CAP1 Complements Yeast Mutants Defective in Ca 2ϩpumps-The putative Ca 2ϩ -pumping function of CAP1 was tested by expressing the clone in yeast mutants that lack the Golgi-Ca 2ϩ -pump (pmr1) or both Golgi-and vacuole-located Ca 2ϩ -ATPases as well as calcineurin (pmr1 pmc1 cnb1). As shown in Fig. 2, A and B, wild type yeast with functional endogenous Ca 2ϩ -pumps grew on plates containing 10 mM EGTA or MnCl 2 . However, mutants in the Golgi-located Ca 2ϩ -ATPase (pmr1; Fig. 2, A and B) or triple mutants lacking both the endomembrane Ca 2ϩ -pumps (pmr1 pmc1 cnb1, also known as K616; Fig. 2A, data not shown) failed to grow on EGTA or 3 mM MnCl 2 , as reported earlier (7,30). CAP1 restored the growth of both pmr1 and triple mutants on EGTA plates ( Fig.  2A). The growth complementation was observed only in the presence of galactose ( Fig. 2A), the inducer of Gal1 promoter under whose control the CAP1 cDNA was inserted. This suggested that CAP1 was transcribed as well as translated, and the product was assembled as a fully functional Ca 2ϩ -pumping enzyme in yeast membranes. The transformants could not grow on MnCl 2 (Fig. 2B, data not shown for K616), indicating that CAP1 cannot transport Mn 2ϩ . Rabbit SERCA-pump was also specific for Ca 2ϩ transport and failed to catalyze Mn 2ϩ transport (31). However, yeast PMR1 or Arabidopsis ECA1 products restored the growth of pmr1 on Mn 2ϩ (7,32). The specificity of CAP1 protein for Ca 2ϩ was further indicated by its inability to restore the growth of the pmr2 mutant (deficient in Na ϩ /Li ϩ efflux activity) on high lithium-containing medium (Fig. 2C).
CAP1 Is Transcribed and Translated in the Yeast Mutant K616 -We have determined whether the phenotypic complementation of yeast mutants by CAP1 was indeed due to the expression of CAP1 cDNA in a galactose-dependent manner. Total RNA preparations from K616 transformants grown in the presence of glucose or galactose were probed with the 5Ј-end of CAP1 cDNA. The results, presented in Fig. 3A, show that CAP1 transcripts are detectable only in pCAP1-transformed mutant cells, and only when grown on galactose. The absence of CAP1 transcripts in glucose-grown cells is consistent with the repression of GAL1 promoter by glucose and the inability of glucose to support the growth of the mutant in the absence of extracellular Ca 2ϩ ( Fig. 2A). A strong inducibility of CAP1 in K616 cells was also indicated by the detection of CAP1 signals using only 7.5 g of total RNA per lane.
We tested for the presence of CAP1 protein in the microsomes of transgenic yeast mutant. For this, monoclonal antibodies were produced against a recombinant protein homologous to the central hydrophilic loop of CAP1 deduced amino acid sequence (residues 342-576). This fragment was overexpressed in E. coli as a His-tagged protein; and the recombinant protein was purified using a nickel column. The antisera raised against CAP1 protein recognized an ϳ110-kDa protein in the membranes isolated from K616 yeast cells transformed with pCAP1 (Fig. 3B). The lower molecular mass band detected (ϳ80 kDa) in the yeast membranes could be a proteolytic product of the 110-kDa CAP1 product, as neither of the cross-reactive bands was present in the membranes from the mutant transformed with the vector alone (Fig. 3B).
The CAP1 Product Forms a Ca 2ϩ -dependent Phosphorylated Intermediate Characteristic of P-type ATPases-We asked whether the CAP1 product expressed in yeast mutants can form a phosphorylated intermediate (E-P) characteristic of Ca 2ϩ -ATPases, i.e. a hydroxylamine-sensitive and Ca 2ϩdependent acyl-phosphate intermediate. As shown in Fig. 4, the membranes from K616 cells expressing CAP1 showed one 32 P-labeled band (ϳ110 kDa) in the presence of Ca 2ϩ . Furthermore, no signals were seen in membranes from yeast cells transformed with vector alone (Fig. 4), indicating that the ϳ110-kDa phosphoprotein is most likely identical to the CAP1 product. The labeling of the 110-kDa band was dependent on a short incubation time (ϳ15 s, data not shown), which is consistent with the rapid turnover rate of the phosphorylation intermediate common for all P-type ATPases (33). Incubating the reaction product with hydroxylamine completely removed the label (Fig. 4), confirming that the phosphorylation was by an acyl phosphate-linkage characteristic to P-type ATPases. The phosphate labeling was dependent on the presence of Ca 2ϩ in the reaction mixture and enhanced by lanthanum (Fig. 4), consistent with the slow turnover rate of the intermediate in the presence of lanthanum. Thus, the E-P assay further confirmed that the CAP1 product may be a functional Ca 2ϩ -ATPase. Ca 2ϩ -dependent Manner-Since the CAP1 product appeared to be divergent from the known ER-type Ca 2ϩ -ATPases by possessing an extended C-terminal tail rich in positively charged residues interspersed by hydrophobic amino acids, it is possible that the CAP1 product may be a CaM-regulated Ca 2ϩ -ATPase. The CaM binding nature of the CAP1 protein expressed in the yeast mutant K616 was assessed using CaM-affinity chromatography. Detergent-solubilized microsomal proteins (0.75-1 mg) were loaded on a 2-ml column of CaM-Sepharose in the presence of Ca 2ϩ . The column was washed with excess column buffer and eluted with 10 mM EGTA. The second and third column fractions of the eluent showed enrichment of polypetides at 100-, 60-, 40-, and 33-kDa region as detected by silver staining (data not shown). The flow-through (40 g of protein) and the eluent (3-5 g protein) were probed by anti-CAP1 antisera. As shown in Fig. 5, a cross-reactive band at 105-110 kDa was apparent only in the eluent and none in the flow-through. The results indicate that CAP1 product expressed in K616 is capable of interacting with calmodulin and is most likely regulated by CaM in vivo.
Microsomes from the CAP1-expressing Yeast Mutant Mediate Calmodulin-stimulated Ca 2ϩ Transport-To determine further whether the CAP1 complementation of yeast mutants defective in Ca 2ϩ -pumps was indeed resulted by the restoration of calcium transport, we have tested the Ca 2ϩ transport activity of microsomes isolated from the triple mutant K616 expressing CAP1. The mutant transformed with vector alone showed background Ca 2ϩ uptake (40 -55% of the CAP1 transformed cells). This is apparently driven by H ϩ /Ca 2ϩ antiporter as indicated by its abolition by gramicidin (24,34) or nitrate (data not shown). The addition of bafilomycin (24,34) was not tested, as gramicidin alone was effective in eliminating the antiporterdriven Ca 2ϩ transport in the mutant at the pH/ionic conditions of our assays. Gramicidin was routinely added to our assays to eliminate interference from the antiporter activity. The CAP1mediated transport activity was dependent on the presence of ATP in the assay buffer and was abolished by heating the membranes at 90°C for 3 min (data not shown). The Ca 2ϩ uptake increased up to 5 min (Fig. 6A) and reached a plateau within 10 min (data not shown). Addition of the ionophore A23187 released most of the membrane-accumulated 45 Ca 2ϩ , indicating that the transport was against a concentration gradient (Fig. 6A). Ca 2ϩ dependence assay of the transport indicated that the Ca 2ϩ uptake by CAP1 saturates at ϳ5 M free Ca 2ϩ in the medium (Fig. 6B).
The CaM regulation of CAP1 was further confirmed by testing the effect of exogenous CaM or W7, a CaM inhibitor on the CAP1-restored Ca 2ϩ transport activity of the triple mutant. The results, presented in Fig. 6C, show up to a 2-fold stimulation of Ca 2ϩ transport activity by an external addition of CaM, indicating that the CAP1 activity may be calmodulin-regulated. The calmodulin inhibitor, W7, decreased the 45 Ca transport by 50% even in the absence of supplemental calmodulin (data not shown) indicating that CAP1 product may be associated with endogenous CaM. With supplemental CaM in the assay buffer, W7 caused a very high background retention of 45 Ca 2ϩ on the filters. Therefore, its effect on the Ca 2ϩ transport activity of the microsomes in the presence of CaM could not be reliably ascertained (data not shown). In summary, these results provide further evidence that CAP1 is a calmodulin-regulated Ca 2ϩ -ATPase.
Peptides from CAP1 C Terminus Bind Ca 2ϩ /Calmodulin-Given the CaM binding nature of recombinant CAP1 and CaM stimulation of its activity, we expected to find a potential CaMbinding domain in its sequence. Taking analogy to the animal calmodulin-regulated pumps and the presence of an extended C terminus into consideration, we searched for a putative CaMbinding domain at the C terminus of CAP1. Although not conserved in their primary sequence, CaM-binding domains most commonly form basic amphipathic ␣-helices (35). The C terminus of CAP1 showed helical structures with clusters of positively charged amino acids. Three overlapping regions (between 995 and 1049 residues, Fig. 7A) were selected based on the properties of their helical wheel projections. Synthetic peptides were made corresponding to these sequences and tested for their ability to shift the mobility of calmodulin (recombinant or purified from roots) in native acrylamide gels (19). Of the three peptides tested, only the two distal peptides (peptides P2 and P3, Fig. 7A) with an overlapping sequence of KQKASSER-RLTFD bound CaM and shifted its mobility in a Ca 2ϩ -dependent manner (Fig. 7, B-D). Between P2 and P3, the latter caused a greater retardation in the CaM mobility in the presence of urea (Fig. 7B). However, electrophoresis of CaM⅐peptide complexes in the absence of urea led to a complete shift of CaM by P2 at equimolar concentrations (Fig. 7C). The two peptides (P2 and P3) that caused a mobility shift were highly basic in their overall charge (pI ϭ 12) in contrast to the P1 peptide (pI ϭ 6.36) that failed to induce a shift. However, P 2 and P 3 did not bind to other acidic proteins such as soybean trypsin inhibitor or bo-vine serum albumin (data not shown), under identical conditions. This indicated that the two CAP1 peptides bound calmodulin by specific hydrophobic interactions and not due to electrostatic attraction.
The binding of CaM to its targets induces structural changes in CaM as well as the target peptides/proteins. Binding is accomplished by a change in conformation of the central ␣-helix of CaM to a random coil allowing the globular lobes of CaM to engulf the target peptide (36). In turn, the target peptides may attain a greater helicity upon CaM binding. Hence, spectropolarimetry has been a useful tool to follow the helicity changes in CaM⅐target complexes (19). CAP1 peptide and CaM interactions were monitored using CD spectra. Calmodulin displayed a CD spectrum typical of helical proteins with minima at 222 and 208 nm and a maximum at 190 nm (data not shown). The peptides themselves showed random conformation as indicated by the spectra of uncomplexed peptides (data not shown). Fig.  8A presents the difference spectra resulting after the subtraction of the spectrum of individual peptide⅐CaM complex from that of CaM alone. When co-incubated with calmodulin, peptides P2 and P3 induced a change in the molar ellipticity of CaM at 208 and 222 nm, whereas P1 induced nonspecific changes (Fig. 8A). The changes induced by P2 and P3 indicate a decreased helicity of CaM and an increase in the helicity of the two peptides (19). Furthermore, the interaction was dependent on Ca 2ϩ , as these conformational changes were abolished by EGTA (Fig. 8B). The absence of a helicity increase in P1 confirmed that CAP1 residues included in this peptide are not involved in the interaction of the pump with CaM, in accordance with our results from CaM-gel shift assays (data not shown). Furthermore, CD spectral analysis also showed that P3 caused greater decrease in the helicity of CaM than P2 did (Fig. 8B), confirming that in addition to KQKASSER-RLTFD other residues in the P3 peptide are important for the interaction of the pump with calmodulin. A recombinant peptide homologous to the C terminus that encompasses the CaMbinding peptides was expressed in E. coli. This longer peptide can be purified to homogeneity by CaM-affinity chromatography and interacts with CaM in filter binding assays (data not shown). Since this peptide is toxic to E. coli and could not be overexpressed, synthetic peptides will be used to fine map the CaM-binding domain.
Antibodies Against a Recombinant CAP1 Polypeptide Recognize a Single Maize Microsomal Protein Purified by CaM-Affinity Chromatography-We attempted to identify and characterize the CAP1 cognate protein from maize root tissue, using CAP1 antisera. Although most of the monoclonal antisera gave a reactivity at high titer with the recombinant polypeptide expressed in E. coli (Ͼ16,000-fold dilution) or yeast (1:4000), none of the clones recognized any polypeptide in total maize microsomes or after fractionation on sucrose gradients (data not shown). This indicated that the cognate protein, if present, is in very low abundance in maize tissues. Since our results with the CAP1 product suggested that it is a CaM-binding protein (Figs. 5-8), we enriched maize root microsomal proteins on CaM-affinity chromatography and then probed with CAP1 antisera. Microsomal CaM-binding preparations showed a major polypeptide of ϳ110 kDa and 5-6 additional bands of varying sizes (Fig. 9A). The CAP1 antibodies cross-reacted only with the 110-kDa polypeptide (Fig. 9B), indicating that the CAP1 cDNA encodes a single microsomal protein that binds CaM by itself or through interaction with a calmodulin-binding protein.
A 110-kDa Protein from the CaM-Sepharose Binding Fraction of Maize Microsomes Directly Interacts with Ca 2ϩ /Calmodulin-The ability of putative CAP1 cognate protein to in-FIG. 6. Characterization of Ca 2؉ transport activity driven by CAP1 in yeast microsomes. A, yeast microsomes from vector or CAP1transformed cells were isolated and assayed for ATP-dependent Ca 2ϩ transport activity as described under "Experimental Procedures." At the end of 5 min, 1 M of the ionophore A23187 was added in a replicate assay. Vector-transformed cells showed no ATP-stimulated activity in the presence of gramicidin (routinely added at 50 M to the assay medium). The calculated free Ca 2ϩ in this assay was 57 nM. However, the linearity of transport was maintained only for the first 5 min, even at higher free Ca 2ϩ concentrations tested (1 or 5 M). B, Ca 2ϩ dependence of 45 Ca transport driven by CAP1. CaCl 2 /EGTA mixtures were used to buffer free Ca 2ϩ concentration as described under "Experimental Procedures." C, effect of supplemental calmodulin was tested using recombinant maize calmodulin. Bovine serum albumin was substituted for CaM to provide equimolar concentration of protein in all the tubes. Background activity in the absence of ATP was subtracted from each corresponding assay with ATP. Values are means of duplicate samples from a single assay and represent results of three separate membrane preparations.
teract independently with CaM was tested by the overlay of HRP-labeled CaM on gel blots of microsomal CaM-Sepharose eluents. With brief exposures to x-ray film, only a 110-kDa band was found to bind calmodulin in these assays (Fig. 10A), although longer exposures revealed additional CaM-binding proteins in the preparation (data not shown). The binding was abolished when CaCl 2 was replaced by EGTA from the binding buffer (data not shown). Calcineurin was used as a positive control in the assay, and only the 58.6-kDa large subunit (which is known to be the CaM-binding subunit of the holoprotein) was bound by calmodulin (Fig. 10A). Taken together, CaM-affinity chromatography, CaM overlay assay, and immunoblotting experiments reveal that the CAP1 cognate protein in maize tissues is a calmodulin-binding protein.
The 110-kDa polypeptide migrated faster in the presence of Ca 2ϩ -chelator EGTA in acrylamide gels, i.e. as a ϳ100-kDa polypeptide (Fig. 10B). This mobility shift, which is characteristic of many Ca 2ϩ -binding proteins (e.g. Refs. 37 and 38 and references therein), indicated that the CAP1 cognate polypeptide binds Ca 2ϩ , in addition to calmodulin. This is in accordance with the putative function of the CAP1 product as a Ca 2ϩ -pumping enzyme and conservation of the putative Ca 2ϩbinding residues (Glu 291 , Glu 759 , Asn 784 , Thr 787 , Asp 788 , and Glu 919 ) within the predicted transmembrane regions, TM4, -5, -6, and -8 (Fig. 1).
The CaM-Sepharose Fraction Has a 110-kDa Polypeptide That Forms a Ca 2ϩ -dependent Phosphorylated Intermediate-We also investigated if the 110-kDa CaM-binding protein of maize microsomes can form Ca 2ϩ -dependent acyl-phosphate intermediate. As shown in Fig. 10C, the CaM-binding preparation showed one 32 P-labeled band in the presence of Ca 2ϩ . The phosphoprotein was similar in its molecular size (ϳ110 kDa) to that of the anti-CAP1 cross-reactive polypeptide. The labeling of the polypeptide was also dependent on a short incubation time, sensitive to hydroxylamine dependent on the presence of Ca 2ϩ in the reaction mixture and enhanced in the presence of lanthanum (Fig. 10C). These results suggest that the putative CAP1 cognate protein in maize membrane may be a functional Ca 2ϩ -ATPase.  7. Peptides from the C terminus of CAP1 bind to calmodulin. A, synthetic peptides (P1, P2, and P3) identical in their sequence to the C terminus of deduced CAP1 were used to test the affinity of CAP1 product to calmodulin. The location of each peptide in the CAP1 sequence is shown by the number of start and end residues. The sequence overlap within the peptides is indicated by an underline. B, interaction of CAP1 peptides with calmodulin as studied by native urea-acrylamide gel electrophoresis. 150, 300, 750, and 1500 pmol of P1 (lanes 2-5), P2 (lanes 6 -9), or P3 (lanes 10 -13) were incubated with 300 pmol of calmodulin in the presence of 4 M urea and 100 M CaCl 2 for 1 h. The peptide-protein complexes were analyzed in a native 4 M urea, 12.5% acrylamide gel as described by Erickson-Viitanen and DeGrado (19). Lanes 1 and 14 show the migration of free CaM. Electrophoresis of free peptides in the same gel did not result in any bands. C, interaction of CAP1 peptides with calmodulin as studied in native gels lacking urea. 150 and 300 pmol of P1 (lanes 2 and 3), P2 (lanes 4 and 5), or P3 (lanes 6 and 7) were incubated with 300 pmol of calmodulin in the presence of 100 M CaCl 2 for 1 h. The peptide-protein complexes were analyzed in native acrylamide gels (19). Lane 1 shows the migration of free CaM. D, Ca 2ϩ dependence of the CaM interaction with CAP1 peptides. The protein and peptide incubations were as in lanes 1-5 of B, except that Ca 2ϩ was replaced by 2 mM EGTA in the binding and gel running buffers (to lower the free Ca 2ϩ concentration to near zero). Results are shown for P2 and were similar with P1 and P3 peptides. Lane 1 is free CaM and lanes 2-5 are 150, 300, 750, and 1500 pmol of peptide incubated with 300 pmol of CaM.

CAP1 Transcripts Are Low Abundant and Induced by Anoxia in Maize
Roots-Our interest is to elucidate the role of this chimeric Ca 2ϩ -pump in intracellular signaling, particularly in the Ca 2ϩ -mediated signaling of anoxia. As a first step toward this goal, we investigated the abundance and induction patterns of CAP1 message in maize roots by different environmental stresses. Gel blot hybridization and RT-PCR experiments indicated that that CAP1 transcripts are of very low abundance in maize tissues, as was the concentration of its cognate protein in the microsomes. Consistent detection of signals in the RNA gel blots required the purification of poly(A) RNA. Specific hybridization signals could often be detected by loading Ͼ40 g of total RNA per lane (data not shown). There was only single transcript class of about 6 kilobase pairs long indicating that there could be a processing regulation of the primary transcript. Despite the low abundance of its transcripts in maize roots, the CAP1 cDNA was isolated by screening Ͻ2 ϫ 10 5 plaques from 6-h anoxic maize root cDNA library. This indicated that the CAP1 transcript abundance might increase during anoxia. We tested the induction of CAP1 message under anoxia as well as under other abiotic stresses. The CAP1 transcripts were induced and maintained at 2-3-fold greater than aerobic levels for 2-4 h of anoxia (Fig. 11A). Furthermore, the faster migration of CAP1 transcripts at 4 h of anoxia is also suggestive of a potential processing regulation of the primary transcript. Heat shock also caused a mild induction of CAP1, but salt, cold, and osmotic treatments were ineffective (Fig.  11A).
CAP1 Appears to Be Encoded by a Single Copy Gene in Maize Genome-Genomic DNA gel blot hybridization was done to investigate if there is a gene family encoding CAP1 homologs in the maize genome. Probes from three different regions of CAP1 clone (including the central hydrophilic loop) were used for hybridization at medium or low stringency. Results indicate that there are only one or two genes that encode CAP1 in the maize genome (Fig. 11B). This prediction is in agreement with the protein gel blot analysis (Fig. 9). The Southern hybridization results also suggested that there is restriction fragment length polymorphism among the genotypes tested (Fig. 11B). DISCUSSION In mammalian cells, molecular and biochemical evidence shows that the CaM-regulated Ca 2ϩ -pump is located on the plasma membrane and is divergent in its sequence and structural features from the SERCA-type Ca 2ϩ -pumps. Evidence has accumulated that there are calmodulin-regulated Ca 2ϩ -ATPase activities in plant cells (39). However, unlike in animal cells, such activities have been reported both from purified plasma membrane preparations (40 -42) as well as enriched endomembranes (e.g. Refs. 13, 22, and 42-44). The presence of multiple calmodulin-regulated Ca 2ϩ -pumping activities and their distribution to different cellular compartments indicate a crucial role for these enzymes in plant cell signaling. Isolation and functional characterization of genes encoding these multiple Ca 2ϩ -pumps would facilitate the unraveling of mechanisms of cellular Ca 2ϩ homeostasis and attendant pathways of cellular communication. Recently, Malmstrom et al. (8) and Harper et al. (6) have cloned cDNAs for CaM-regulated Ca 2ϩ -pumps from cauliflower and Arabidopsis (BCA1 and ACA2, respectively). These clones are related to the mammalian plasma membrane Ca 2ϩ -ATPases, although the CaM-binding domain in the plant clones is located at the N terminus. CAP1 presents yet another novel type of CaM-binding Ca 2ϩ -pump in plants. Sequence homology to SERCAs, growth and functional complementation of yeast mutants deficient in Ca 2ϩ -ATPases, binding affinity of synthetic peptides and transgenic CAP1 protein to calmodulin, CaM-stimulated Ca 2ϩ transport activity in yeast microsomes, antigenic identity of CAP1 product with a Ca 2ϩand CaM-binding microsomal protein from maize roots, properties of the phosphoenzyme formed by cognate proteins, provide strong support to our proposal that CAP1 encodes a calmodulin-binding Ca 2ϩ -ATPase in maize. It differs from the canonical mammalian CaM-regulated Ca 2ϩ -pumps in that it has greater overall sequence identity with the SERCA-type Ca 2ϩ -ATPases (e.g. Refs. 7, 29, and 45). At the same time, it is similar to the animal PM-type pumps in having a CaM-binding domain in the C-terminal tail. Thus, CAP1 shares features of both the PM-and ER-type Ca 2ϩ -pumps of mammalian systems. However, it is very divergent from BCA1 or ACA2 (the plant homologs of PM-type pumps) in the overall sequence as well as in the location of its CaM-binding domain. Thus, the presence of multiple CaM-regulated activities distributed on more than one cellular membrane and novel type of genes encoding these activities indicate that plant signaling pathways or components involved may not always fit the animal paradigm.
Previously, evidence was presented for the presence of CaMregulated Ca 2ϩ -ATPases in young maize seedling shoots (11,13,14). However, the polypeptides that were attributed to belong to Ca 2ϩ -pump were of two different size ranges. Briars et al. (11) purified an enzyme on CaM-affinity columns, and this preparation showed a 140-kDa polypeptide that crossreacted with antisera for the mammalian CaM-binding Ca 2ϩ -ATPase. Later, this polypeptide was confirmed to be a Ca 2ϩpump by reconstitution studies as well as phosphoenzyme analysis (12,13). On the other hand, Logan and Venis (14) identified a 102-kDa polypeptide of maize microsomes as CaMbinding Ca 2ϩ -ATPase based on its cross-reactivity with anti-SERCA antisera as well as binding to CaM. However, no further studies have been reported on this protein. The product of CAP1 clone has a predicted mass of 113 kDa, which is also in the size range of the polypeptide recognized by CAP1 antisera in maize. This is much smaller than the 140-kDa Ca 2ϩ -pump purified by Theodoulou et al. (13). The cross-reactivity of the 140-kDa protein with the antisera for mammalian PM Ca 2ϩpump further indicates that these two pumps are most likely divergent even in their sequence. In contrast, the 102-kDa FIG. 9. Antibodies to CAP1 24-kDa polypeptide recognize a single polypeptide in the calmodulin-binding proteins of maize root microsomes. A, SDS-polyacrylamide gel electrophoresis analysis of root microsomal calmodulin-binding proteins. Microsomal proteins from 3-day-old maize seedling roots were fractionated on CaM-Sepharose column, and the bound proteins (5 g) were resolved in a 6 -12% gradient acrylamide gel. The proteins were stained in silver. B, protein gel blot analysis of root microsomal CaM-binding proteins. Monoclonal antisera raised against the purified 24-kDa CAP1 product were used to probe a gel blot of the CaM-binding fraction of maize root microsomal proteins (5 g). Probing an identical blot with preimmune serum or anti-histidine antisera did not give any signal. Size markers are shown on the left side of the figures.
CaM-binding putative Ca 2ϩ -ATPase reported by Logan and Venis (14) is in the size range of the CAP1 product. Furthermore, these authors showed that a similar polypeptide crossreacted with antibodies for a region conserved in all SERCAtype-pumps, including the CAP1 (Refs. 5, 7, 9, and 46; Fig. 1).
Our studies further indicate that the CAP1 protein possesses its CaM-binding domain at the C terminus, and the evidence is 3-fold. CaM-affinity chromatography indicated that a recombinant polypeptide corresponding to the last 100 residues of the C terminus bound calmodulin in the presence of Ca 2ϩ (data not shown). CaM mobility shift assays demonstrated that shorter peptides within this C-terminal tail bound to CaM in calciumdependent manner and retarded its mobility in acrylamideurea gels (Fig. 7). Furthermore, co-incubation of CaM with CAP1 peptides resulted in characteristic conformational changes typical to CaM⅐target complexes, as revealed by the CD spectra (Fig. 8). These studies were also indicative of the amino acid residues involved in the interaction of CAP1 with CaM. Of the three peptides tested, only P2 and P3 were effective in binding to CaM in gel shift as well as spectropolarimetric assays. The overlapping sequence KQKASSERRLTFD appears to be critical for CaM binding, although CaM gel FIG. 11. CAP1 is expressed in low abundance in maize roots and may be encoded by a single gene. A, RNA gel blot analysis of CAP1 transcripts. Maize seedlings were exposed to none (C), 2, 4, and 24 h of submergence (anoxia), 1 h of temperature shock at 10 or 40°C, 5 h of salt stress with 0.1 M NaCl (SS), or osmotic shock (OS) with 0.1% polyethylene glycol 8000 for 4 h. Poly(A) RNA was purified from the root tissue of treated seedlings and resolved in formaldehyde-agarose gels (2 g per lane). The gel was blotted and probed with 32 P-labeled CAP1 cDNA. The blot was subsequently hybridized to a constitutive probe, 1055, and CAP1 signals were normalized as described earlier (2). The normalized intensities of CAP1 are shown in the grid at the bottom. B, genomic DNA gel blot analysis of CAP1. 20 g of genomic DNA from the maize inbred B73 or Mo17 was digested with EcoRI (E), HindIII (H), or HpaI (Hp), fractionated in a 0.8% agarose gel and blotted onto nylon membrane. The blot was probed with a fragment of CAP1 that encodes the central hydrophilic loop. Probes from 5Ј-and 3Ј-ends of CAP1 hybridized to single bands of different sizes in the two genotypes. retardation assays and CD spectra indicate additional residues in the extreme C terminus may enhance the interaction.
A high degree of sequence similarity between CAP1 and SERCAs indicates that the CAP1 protein may be located on the ER membranes. However, the tomato LCA1 product, despite its similarity to the mammalian ER pumps, is not ER-localized but is distributed on the tonoplast and the plasma membrane (45). This has also been the case with the cauliflower calmodulinregulated BCA1 product, which is a PM-type pump (42). It is not known if the CAP1 product is also localized on more than one membrane in maize cells. Nevertheless, the products of ECA1 and ACA2 from Arabidopsis are confined to the ER/ intracellular membranes. In fact, biochemical evidence suggests that the CaM-regulated Ca 2ϩ pumping activity in maize seedlings is predominantly distributed on the endomembranes, either the ER or the tonoplast (Refs. 13, 14, 47, and 48; but also see Ref. 40).
The low abundance of the CAP1 protein and transcripts in maize tissues indicates a tight regulation of CAP1 expression. Furthermore, regulation of the Ca 2ϩ transport activity by calmodulin suggests the involvement of CAP1 product in a feedback attenuation of cytosolic Ca 2ϩ concentration during cell stimulation. A stringent regulation of Ca 2ϩ sequestration from the cytosol should allow the Ca 2ϩ -dependent signaling processes to continue without the cell attaining cytotoxic levels of free Ca 2ϩ . Induction of CAP1 transcripts in maize roots only during the early hours of anoxia indicates such a regulation of Ca 2ϩ homeostasis in the O 2 -deprived maize cells.