Expression of Calreticulin in Escherichia coli and Identification of Its ...

CHEMISTRY THEJOURNALOF BIOLOGICAL 0 1991 by The American Society for Biochemistry and Molecular Biology. Inc.

Vol. 266, No. 32, Issue of November 15, pp. 21458-21465,1991 Printed in U.S. A.

Expression of Calreticulin in Escherichia coli and Identification of Its Ca2' Binding Domains* (Received for publication, June 19,1991)

Shairaz BakshS and MarekMichalakg From the Cardiovascular Disease ResearchGroup and the Departments of Biochemistry and Pediatrics, University of Alberta, Edmonton, Alberta T6G 2S2, Canada

Recombinant calreticulin and discrete domains of calreticulin were expressed in Escherichia coli, using the glutathione S-transferase fusion protein system, and their Ca2+binding properties were determined. Native calreticulinbound 1 mol of Ca2+/mo1 of protein with high affinity, and also bound approximately 20 mol of Ca2+/mo1 protein of withlow affinity. Both Ca2+ binding sites were present in the recombinant calreticulin indicating that proper folding of the protein was achieved using this system. Calreticulin is structurally divided into three distinct domains: the N-domain encompassing the first 200 residues; the P-domain which is enriched in proline residues (residue 187-317); andthe C-domain which covers the carboxyl-terminalquarter of the protein (residues 310-401), and contains a high concentration of acidic residues. These domains were expressed in E. coli,isolated, and purified, and their Ca2+ bindingpropertieswere analyzed. The C-domain bound approximately 18 molof Ca2+/molof protein with a dissociation constant of approximately 2 mM. The P-domain bound approximately 0.6-1 mol of Ca2+/ mol of protein with a dissociation constant of approximately 10 PM. The P-domain and theC-domain, when expressed together as the P+C-domain, bound Ca2+ with both high affinity and low affinity, reminiscent of both full lengthrecombinant calreticulin and native calreticulin. In contrast the N-domain, did not bind any detectable amount of 4sCa2+. We conclude that calreticulin has two quite distinct types of Ca2+binding sites, andthat these sites are in different structural regions of the molecule. The Pdomain binds Ca2+with high affinity andlow capacity, whereas theC-domain binds Ca" with low affinity and high capacity.

Calreticulin is a Ca" binding protein which was first identified almost 20 years ago by Ostwald and MacLennan (1974) as the high affinity Ca2+binding protein of skeletal muscle

* This work was supported by the Program Grant in Molecular Biology of Membranes from the Medical Research Council of Canada, the Alberta Heart and Stroke Foundation, and the Alberta Heritage Foundation for Medical Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Recipient of a Studentship from the Heart and Stroke Foundation of Canada. f Scholar of the Alberta Heritage Foundation for Medical Research. Correspondence should be addressed to thisauthor.

sarcoplasmic reticulum (SR)' membranes. Thisprotein is, however, only a minor component of the SR in skeletal (Michalak et al., 1980) and cardiac muscle(Fliegel et al., 1989b), and in these tissues the major Ca2+binding/storage protein is considered to be calsequestrin (for review see MacLennan et al., 1983). In contrast, calreticulin has been identified as amajor Ca2+binding (storage) protein of smooth muscle SR and nonmuscle ER (Milner et al., 1991). Both calreticulin and calsequestrin are high capacity Ca2+binding proteins; calsequestrin binds 30-50molof Ca2+per mol of protein with a low affinity of approximately 1 mM (MacLennan et al., 1983). MacLennan's group have shown that calreticulin, in addition to a single high affinity site may also possess a number oflow affinity Ca" binding sites (MacLennan et al., 1972; Ostwald and MacLennan, 1974). On the basis of this observation, and the tissue distribution of the two proteins we have proposed that calreticulin may be a nonmuscle functional analogue of calsequestrin (Milner et al., 1991). Calsequestrin (Fliegel et al., 1987; Scott et al., 1988) and calreticulin (Fliegel et al., 1989a; Smithand Koch, 1989; McCauliffe et al., 1990; Murthy et al., 1990) have both been cloned and their amino acid sequences deduced from the cDNAs encoding them. Importantly,the only similarity in the amino acid sequences of these two proteins occurs in the clusters of acidic residues found at their respective C termini and even this similarity is limited (Fliegel et al., 1989~). Calreticulin has been proposed to be divided structurally into distinct domains (Fliegel et al., 1989a; Smith andKoch, 1989). The first domain, which encompasses the NHz-terminal half of the molecule, contains two helices followed by a sequence predicted to form eight antiparallel &sheets connected by protein loops (N-domain). This region leads into a prolinerich sequence (P-domain) which is followed bythe C-terminal quarter of the protein (C-domain). The C-domain of calreticulin is acidic; in the last 57 residues, 37 areasparticor glutamic acid (Fliegel et al., 1989a; Smith and Koch, 1989; McCauliffe et al., 1990; Murthy et al., 1990). The C-domain also terminates with the KDEL ER retention signal. The functional implications of these distinct domains within the sequence of calreticulin are not clear at present. However, in analogy with observations on the Ca2+binding behavior of calsequestrin (Ohnishi and Reithmeier, 1987) we proposed that the C-domain of calreticulin might contain the low The abbreviations used are: SR, sarcoplasmic reticulum; ER, endoplasmic reticulum; GST, glutathione S-transferase; SDSPAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; FPLC, fast protein liquid chromatography; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid; PBS, phosphate-buffered saline; PIPES, 1,4-piperazinediethanesulfonic acid; MOPS, 4-morpholinepropanesulfonic acid; PCR, polymerase chain reaction; bp, base pair(s); HPLC, high performance liquid chromatography.

21458

ea2+Binding to Calreticulin

21459

affinity andhighcapacityCa2+binding site (Fliegel et al., this cDNA clone see Fliegel et al., 1989a). The following three oligo1989a). Although a high affinity Caz+ binding site has been deoxynucleotides with 5"flanking EcoRI restriction site were synthesized and used as primers. identified in calreticulin (Ostwald and MacLennan, 1974) the Primer 1: 5'-ATGAATTCGGAGCCCGTCGTCTACTTCAA-3' protein amino acid sequencecontains no EF-hand consensus sequences (Fliegel et al., 1989a; Smith and Koch, 1989), and, Primer 2: 5'-GGGAATTCACTACCAGTCATCCTCCAGGGA-3' therefore, the location of this Ca2+ bindingsite in the protein Primer 3: 5'-GGGAATTCAGAGACATTATTGGCTCTGCG-3' has yet to be determined. Primer 4: 5'-TGGAATTCAGGCATAGATGTTAGCGTCGGG-3' In our laboratory, we are carrying out studies to elucidate the structural and functional properties of the different cal- The nucleotide sequence of primer 1corresponds to nucleotides 1-20 reticulin domains. To avoid the difficulties inherent in at- of the cDNA encoding the first six amino acids of the mature calreticulin (Fliegel et al., 1989a) and was used as a 5' (forward) tempting to purifylargeamountsof the protein wehave expressedrecombinantcalreticulin. A plasmidvector has primer. Primer 2 corresponded to the nucleotide sequence 531-549 (encoding amino acid residues 178-182); primer 3 to the nucleotide recently been constructed which allows expression proteins sequence 1242-1264 (36 bp behind the stop codon of the calreticulin as fusion proteins withthe enzyme glutathione S-transferase clone), and primer 4 to nucleotides 849-870 (encoding amino acid (GST), thereby enabling affinity purification under nonde- residues 284-290) (Fliegel et al., 1989a). Primer 2 and primer 3 were naturing conditions(Smith and Johnson, 1988). This system used as 3' (reverse) primers for the synthesis of N-domain and the is able to provide large amounts of both a full length protein full length calreticulin, respectively. Primer4 was used as a 3' and of specific domains.In the present study usingthe GST (reverse) primer for the synthesis of the N+P-domain. PCR reactions carried out in a buffer containing 50 mM KCI, 10 mM Tris, pH fusion protein system we have expressed functional full lengthwere 8.3, 0.1% gelatin, 200 FM of each dNTP, 1.5 mM MgCl,, 2.5 units of mature calreticulin as well as several specific domainsof this TuqI polymerase, and 100 pmol of the appropriate primer. PCR protein in order to identify the domainsinvolvedinCa2+ products were purified by polyacrylamide gel electrophoresis, cut with EcoRI restriction endonuclease, and ligated into anEcoRI restriction binding to this protein. Wehave shown that nativeand site of the phosphatase-treated pGEX-3X plasmid. The following recombinant calreticulins bind Ca2+ with both high and low affinities. We have confirmed our earlier suggestion that the plasmids were generated pGEX-CRT, pGEX-N, and pGEX-N+P encoding full length mature calretriculin, N-domain, and N+P-dolow affinity and high capacity Ca2+ binding sites are localized main of the protein. to the C-domainof the protein. Importantly, we have also Plasmids containing P-domain,C-domain, and P+C-domain of the localized the high affinity and low capacity Ca2+ bindingsite protein, designatedpGEX-P,pGEX-C, and pGEX-P+C, respectively, to the P-domain of calreticulin.This observation should pro- were generated by subcloning of cDNA restriction fragments (see Fig. vide important information concerning the structural require- 4). BclI-BamHI DNA fragment (nucleotides 416-815) encoded the Pments associatedwithEF-handlackinghigh affinity Ca2+ domain; BamHI-BamHI fragment (nucleotides 815-1865) encoded the C-domain; and BclI-SmaI fragment (nucleotides 416-1653) enbinding sites in proteins. coded P+C-domain (Fliegel et d., 1989a). The SmuI restriction site is not shown in Fig. 4 since it is located in the 3"noncoding region of the cDNA encoding calreticulin (for restriction map see Fliegel et Materials-PIPES, MOPS, and TritonX-100 were purchased from al., 1989a). The restriction fragments were purified by acrylamide gel Sigma..'5CaC12 was obtained from Du Pont-New England Nuclear. electrophoresis and subcloned into theBamHI site(P-and C-domain) Peroxidase-conjugated rabbit anti-goat IgGs and peroxidase-conju- or BamHI/SmaI site (P+C-domain) of the pGEX-3X. Plasmid were gated goat anti-rabbit IgGs were from Bio-Rad and Boehringer Mann- than transformed, first into DH5a E. coli followed by the transforheim, respectively. Nitrocellulose and Immobilon (polyvinylidene di- mation into BNNlO3 E. coli. All expression experiments were carried fluoride) membrane filters were from Schleicher and Schuell and out using the BNN103 E. coli host. Expression and Isolation of Recombinant Proteins-GST fusion Millipore, respectively. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) reagents and molecular weight markers proteins were expressed in BNN103 E. coli host in LB medium were from Bio-Rad. Restriction endonuclease and DNA-modifying containing 50 pg of ampicillin/ml. Cultures were grown to the midenzymes were obtained from Boehringer Mannheim, Bethesda Re- log phase (Am = 0.6-1.0) followed by the induction of the expression (final search Laboratories, and Bio/Can Scientific. Isopropyl-1-thio-o-D- of fusion proteins with isopropyl-1-thio-8-D-galactopyranoside galactopyranoside and factor Xa were obtained from Boehringer concentration of 0.1 mM) for 4 h. Cells were spun down at 3,500 X g Mannheim. Plasmid pGEX-3X, Mono Q 5/5 FPLC column and for 15 min, and the pellet was resuspended in PBS containing 0.1% glutathione-Sepharose 4B were from Pharmacia LKB Biotechnology Triton X-100 and lyzed using the French Press set at 1,000 p.s.i. Inc. Dialysis bags (6.4 mm in diameter)were from Spectra/Por, Fisher followed by centrifugation at 10,000 X g for 10 min. E. coli extracts Scientific. Fresh canine pancreas was a generous gift of Dr. B. Jugdatt containing fusion proteins were assayed for protein content using the (University of Alberta). The following E. coli hosts were used in this method described by Bradford (19761, frozen, and stored at -70 "C study: DH5a, GM48, and BNNlO3. DH5a were obtained from Be- before further use. GST and GST-fusion proteins were purified by one-step glutathithesda Research Laboratories. GM48 and BNN103 were generous gifts of Drs. B. Lemire and J. H. Weiner (Department of Biochemis- one-Sepharose 4B affinity chromatography (2 cm X 12 cm; 40-ml bed try, University of Alberta), respectively. All chemicals were of the volume). Samples of the E. coli extract (2 mg/ml) were applied onto the glutathione-Sepharose 4B column equilibrated with PBS containhighest grade available. ing 1%Triton X-100. The first flow-through was reapplied onto the Construction of Plasmid Encoding Different Fusion Proteins-For the expression and isolation of recombinant proteins, a GST fusion column, and thiswas followedby a wash with 400 ml of PBS. Fusion system was used consisting of pGEX-3X plasmid (Smith and John- proteins were eluted from the affinity column with a buffer containing son, 1988). This plasmid encodes GST followed by the factor Xa 5 mM glutathione and 50 mM Tris-HCI, pH 8.0. Fusion proteins used cleavage site, unique cloning sites (EcoRI, S m d ,and BamHI) anda in this study consisted of approximately 15-18% of the E. coli lysate stop codon. For the purpose of this study we have expressed the proteins (25-30 pg of fusion protein/ml of E. coli culture). Over 90% of the fusion protein was purified to homogeneity by one-step glutaprotein in three domains: N-terminal domain (N-domain),amino acids 1-182; proline-rich domain (P-domain), amino acids 182-290; thione-Sepharose 4B column chromatography. In some experiments the GST-full length calreticulin fusion proand C-terminal domain (C-domain), amino acids 330-401 (see Fig. 4). In addition, N+P-domain (amino acid residues 1-290 encoding tein was cleaved with factor Xa. Factor Xa digestion was carried out both N-domain and P-domains) and P+C-domain (amino acid resi- in a buffer containing 100 mM NaCl, 1 mM CaC12,and 50 mM Trisdues 182-401 encoding both P- and C-domains) were subcloned and HCI, pH 8.0. The enzyme was used at 1:500 dilution (w:v) and the expressed. reaction was carried out a t room temperature for 16-18 h and stopped PCR was used to synthesize cDNA encoding full length mature by the addition of phenylmethylsulfonyl fluoride (final concentration calreticulin, N-domain and P+N-domain. As a template we used the of 2.5 mM). Digested sample was directly loaded onto theglutathioneBamHI-cut pcDX plasmid containing cDNA of 1865 bp in length Sepharose 4B column and theflow-through, containing recombinant encoding skeletal muscle calreticulin (for the nucleotide sequence of calreticulin was collected, concentrated using Amicon concentrator, EXPERIMENTAL PROCEDURES

21460

Ca2+Binding to Calreticulin

and dialyzed against 50 mM NaCI, 100 mM KHzPO4, 1 mM EDTA, pH 7.1. Dialyzed sample was applied directly onto a Mono Q FPLC column and pure calreticulin was obtained as described earlier (Milner et al., 1991). Native calreticulin was isolated by a selective ammonium sulphate precipitation procedure in the presence of the mixture of protease inhibitors followed by a Mono Q FPLC chromatography as described by Milner et al. (1991). Caz+Binding to Purified Proteins-Ca2' binding was measured by equilibrium dialysis. The protein was first dialyzed against 10 mM MOPS, pH 7.0, 150 mM KCI, and 0.1 mM EGTA. Protein aliquots (-200-400 pg) were then dialyzed for 18 h a t 4 "C against 10 mM MOPS, 150 mM KCI, 0.1 mM EGTA, 0.64 pCi of 45Ca2+/mlcontaining different concentrations of cold Ca2+. In some experiments 3 mM MgClzwas also included in Ca" binding buffers. The free Ca2+ concentration was calculated on the basis of the EGTA association constant reported by Fabiato and Fabiato (1979). The calculation of free Caz+was accomplished by the use of a computer program that corrected for ionic strength, pH, EGTA, and K'. Caz+binding assays were carried out in triplicateusing two or threedifferent preparations of native calreticulin and recombinant proteins. SDS-Polyacrylamide Gel Electrophoresis, Immunoblotting, and Analysis of "Ca2+ Binding-SDS-PAGE was on 10 or 12.5% polyacrylamide gels as described by Laemmli (1970). After gel electrophoresis, gelswere stained with either Coomassie Blue or with the carbocyanine dye "Stains-All." For 45Ca2+binding analysis, proteins were transferred electrophoretically onto nitrocellulose membrane (Towbin et al., 1979) and incubated with 45Ca2+(Maruyama et al., 1984). Immunoblotting was carried out as described by Michalak et al. (1990). Antibody binding was detected with appropriate peroxidase-conjugated second antibodies and a standard peroxidase color development reaction. Goat anti-rabbit skeletal muscle calreticulin antibodies used in this study were characterized by Milner et al. (1991). Standards were Bio-Rad low range molecular weight proteins (phosphorylase b (97,400),bovine serum albumin(66,200),ovalbumin (42,700), bovine carbonic anhydrase (31,000), soybean trypsin inhibitor (21,500),and lysozyme (14,400))or Bio-Rad low range prestained markers (phosphorylase b (106,000), bovine serum albumin (80,000), ovalbumin (49,000), carbonic anhydrase (32,000), soybean trypsin inhibitor (27,000), and lysozyme (17,000)). Miscellaneous-All recombinanttechniques were conducted according to standard protocols (Ausubel et al., 1989). Protein was determined by the method of Lowry et al. (1951) or Bradford (1976). NHZ-terminal sequence analysis of native and recombinant proteins was carried out on protein electroblotted to Immobilon (polyvinylidene difluoride) membrane (Matsudaira, 1987) or purified by reversed-phase HPLC.

A

R

C

97

66 42 31 21 14

1 2 3 4 5 6

S 1 2 3456

S 1 2 3 4 5 6

FIG. 1. SDS-PAGE analysis of native and recombinant calreticulin. Native purified calreticulin, E. coli lysate, and recombinant proteins were separated by SDS-PAGE, transferedelectrophoretically to nitrocellulose membranes, and incubated with anti-calreticulin antibody or with 45Ca2+as described under "Experimental Procedures.'' A, Coomassie Blue staining; B, immunostaining with goat anti-calreticulin; C, autoradiography of A after electrophoretic transfer and incubation with 4sCaZ+.Lane 1, purified pancreatic calreticulin; lane 2, affinity chromatogzaphy-purified recombinant GST, lane 3, E. coli lysate containing GST-calreticulin fusion protein; lane 4, glutathione-Sepharose-purifiedGST-calreticulin fusion protein; lane 5,factor Xa-cleaved GST-calreticulin fusion protein; lane 6, Mono Q FPLC-purified recombinant calreticulin; S , Bio-Rad low molecular weight standards. In C prestained molecular weight markers were used. The position of molecular weight marker proteins isindicated.

is shown in Fig. lA, lane 5. Under the conditions used complete cleavage of the fusion protein was obtained. Recombinant calreticulin was separated from the GSTby chromatography on a glutathione-Sepharose 4B affinity column, followed by Mono Q FPLC. Both native calreticulin and the recombinant protein eluted at the same salt concentration from the Mono Q FPLC column (0.3 M NaCl) (Milner et al., 1991), andbothproteinshad the same mobility in SDSPAGE, corresponding to an apparent molecular weight of 60,000 (Fig. LA, lanes 1 and 6 ) . In order to firmly establish the identity of the recombinant calreticulin NH2-terminal amino acid sequence analysis was carried out. The NH2terminalamino acid sequence of recombinant calreticulin corresponded to that of native calreticulin, and the amino acid composition of recombinant and native calreticulins was virtually identical, confirming the identity of the recombinant protein (data not shown). Full length calreticulin-GST fusion protein, as expected, RESULTS cross-reacted with goat anti-rabbit calreticulin antibodies Cloning, Expression, and Purification of Recombinant Cal- (Fig. lB), whereas recombinant GST did not (Fig. lB, lane reticulin-Fig. lA shows SDS-PAGE of recombinant calreti- 2). Native purified calreticulin (Fig. lC, lune 1 ) , the purified culin expressed in E. coli as a fusion protein with GST. Upon GST-calreticulin fusion protein (Fig. lC, lane 4 ) , and factor lysis of the cells by French Press the recombinant GST- Xa cleaved recombinant calreticulin (Fig. IC, lane 5 ) all bound calreticulin fusion protein was a major component of the E. Ca2+by 4sCa2+overlay technique. Recombinant GST did not coli soluble extract (Fig. LA, lane 3). The solubility of the bind any Ca2+ in either 4sCa2+overlay (Fig. lC, lane 2) or fusion protein suggested that itwas probably correctly folded equilibrium dialysis. Acidic proteins which bind Ca2+,such as calreticulin (Fliewithin E. coli and might, therefore, be able to bind Ca2+. The GST-calreticulin fusion protein purified by affinity gel et al., 1989b; Krause etal., 1990)and calsequestrin (Campcolumn chromatography migrated in SDS-PAGE with a mo- bell et al., 1983), stain blue with Stains-All, whereas other (Campbell et al., 1983). Recombinant bility corresponding to a polypeptide of 80 kDa (Fig. LA, lane proteinsstainpink 4 ) . This mobility does not correspond to thepredicted size of calreticulin was also identified using this stain. Fig. 2 shows the fusion protein (72 kDa; 46-kDa calreticulin + 26-kDa Stains-All staining of the same protein samples, separated by recombinant GST). The anomalous mobility of the fusion SDS-PAGE, which were shown in Fig. 1. Native calreticulin, protein is reminiscent of the anomalous mobility of native GST-calreticulin fusion protein, and recombinant calreticulin calreticulin in SDS-PAGE (60 kDa in Laemmli SDS-PAGE all stained blue with Stains-All. All other proteins, including uersus a molecular weight of 46,600 predicted for the protein recombinant GST, stained pink. Ca2+Binding to Recombinant Calreticulin-Ca2+ binding to from its amino acid sequence) (Krause et al., 1990; Milner et purified native and recombinant calreticulin was determined al., 1991). The pGEX-3X plasmid encodes GST (26 kDa) (Fig. lA, by an equilibrium dialysis assay (Fig. 3). Using at least a lane 2) followed immediately by a factor Xa recognition site. 10,000-fold range of Caz+ concentrationwe found that native Therefore, the fusion protein can be cleaved with factor Xa calreticulin bound Ca2+with high affinity (11.4 PM) and low and the recombinant protein purified. SDS-PAGE of the capacity (1mol/mol of protein) as well as with low affinity (2 GST-calreticulin fusion protein after cleavage by factor Xa, mM) and high capacity (17 mol/mol of protein) (Fig. 3A).

Calreticulin

S 1

2

3

4

5

to Ca2+Binding

6

FIG. 2. Stains-all staining of native and recombinant calreticulin. Proteins were separated by SDS-PAGE and stained with the cationic carbocyanine dye Sbins-All, as described by Campbell et al. (1983).Protein samples were the same as in Fig. 1. The position of molecular weight marker proteins is indicated.

21461 Expression of Distinct Domains of Calreticulin as Fusion Proteins-In order to further characterize Ca2+ binding to calreticulin, and to identify specific domains in the protein that might be responsible for Ca2+binding we utilized a fusion protein approach. Structurally, calreticulin can be divided into three distinct domains (Fliegel et al., 1989a; Smith and Koch, 1989) (Fig. 4): the N-domain containing the first 200 amino acids, the C-terminal quarter of the protein (residues 310-401; C-domain), and the sequence connecting these two regions which is enriched in proline residues (residues 187317; p-domain). The following domains were expressed in E. coli as GST fusion proteins: N-domain, P-domain, C-domain, P+C-domain, and N+P-domain (Fig- 4, details on the construction of the recombinant molecules are under “Experimental Procedures”). GST fusion proteins were expressed at high levels in BNN103 E. coli after induction with isopropyl-1-thio-0-Dgalactopyranoside (Fig. 5, lanes c, e, g,i, k, and m). Upon lysis of the cells by French Press inthe presence of Triton X-100, all of the fusion proteins (except for the N-domain) were found to be present in the high speed supernatant fraction. This is in contrast to many proteins expressed in E. coli, which are recovered in insoluble inclusion bodies. The solubility of the fusion proteins supportsthe suggestion that they might be properly folded within E. coli and therefore might be able to bind Ca2+.The N-domain was somewhat less soluble than the otherregions of calreticulin but approximately 20% of this fusion protein was still found in thesupernatant fraction. The GST-fusion proteins were purified by one-step affinity chromatography on glutathione-Sepharose 4B (Fig. 5A, lanes d, f, h, j , 1, and n). Fig. 5, B and C, shows that the C-domain (lane f ), P-domain (lane h ) , and P+C-domain (lane I ) all bound 45Ca2+and reacted with anti-calreticulin antisera. These fusion proteins also stained blue with Stains-All further confirming their involvement in Ca2+binding to the protein (Fig. 6).Incontrast, the N-domain (lane j ) did not bind detectable amounts of 4sCa2+,reacted only weakly with the goat anti-calreticulin antibodies (Fig, 5, and c), and did

:::\L O.Oo0 0 0.020

2

4

6

8

1

0

B (mol/mol)

0.0

2.0

4.0

6.0

8.0

~_.”., ..._., FIG. 3. Ca2+binding to native and recombinant calreticulin. R

lmnl Imnn

I

Ca2+binding to native and recombinant calreticulins was carried out Procedures.” SDS-PAGE of the proteins used for Ca2+binding is shown in Fig. l A , lanes I and 6. Scatchard analysis of the Ca2+binding data is presented for the native calreticulin ( A ) and the recombinant calreticulin ( B ) .B, bound; BIF, bound/free. The data shown in this figure and in Figs. 7 and 8 represent the compilation of values measured in at least two sets of determinations in triplicate using at least two different protein preparations.

by equilibrium dialysis as described under“Experimental

This is in agreement with earlier observations (Ostwald and MacLennan, 1974). Importantly both Ca2+binding sites were present in the recombinant protein indicating that proper folding of calreticulin was achieved (Fig. 3B). Ca2+binding to the low affinity site (approximately 21 mol of Ca2+/mo1of protein) exhibited a binding constant of approximately 2 mM (Fig. 3B). The high affinity Ca2+binding site of the recombinant calreticulin exhibited a maximal binding capacity of approximately 1 mol of Ca2+/mo1of protein with a binding constant of approximately 11PM (Fig. 3B). Ca2+binding to thelow affinity high capacity site in calreticulin was reduced by approximately 60% when measured in the presence of 3 mM M$+ (data notshown). The high affinity low capacity site, however, was not effected by the inclusion of 3 mM MgC12 (data not shown). This observation probably explains why Van et al. (1989) and Waisman et al. (1987) did not detect a low affinity, high capacity Ca2+binding site in calreticulin, since these authorsincluded high concentrations of M$+ in the buffers used for Ca2+binding studies.

not stain with Stains-A11 (Fig. 6)* ca2+ B i d i n g to the Fusion Proteins-Fig. 7 shows caz+ binding to the recombinant C-domain and P-domain. The recombinant C-domain bound up to 18 mol of Ca2+ per mol of protein pig. 7) and did not bind significant amountsof Ca2+at Ca2+concentrations below 300 PM Ca2+(Fig. 7). This result indicates the that high capacity Ca2+binding site of calreticulin is located in the C-domain of the protein. In contrast. the P-domain bound maximum levels of only 0.6-1 mol of Ca2+per mol of protein (Fig. 7), and this binding was saturated between 300 to 500 PM Ca2+ (Fig. 7). The N+Pdomain showed the same binding properties as the P-domain (data notshown). In order to further analyze Ca2+binding to calreticulin the P- and C-domains were expressed as one entity (P+C-domain). The results obtainedso far indicated that this domain should contain both the high affinity and the low affinity Ca2+binding sites. Fig. 8 shows Caz+ binding to the P+Cdomain. As expected this domain bound Ca2+with high affinity (6 pM) and low capacity (1.3 mol of Ca2+/mo1of protein), and with low affinity ( 2 mM) and high capacity (18 molof Ca2+/mo1of protein) (Fig. 8). DISCUSSION

In this report we describe the expression of a full length recombinant calreticulin in E. coli and show that itbinds Ca2+ with high and low affinities, as does the native protein. We

Ca2’

21462 30 160

150

140

20

10

@ 130

Binding to Calreticulin

120

110 90

KBEQNIDCGG 190

310

FIG.4. Putative domain organization of calreticulin. The amino acid 300 270 sequence of 290 the mature280 skeletal muscle calreticulin shown shown in A was de390 of duced from the nucleotide sequence cDNA encoding the protein (Fliegel et al., 1989a).In B the calreticulin polypeptide was divided into specific domains, which were subcloned in to unique restriction sites in the pGEX-3X plasmid as described under “Experimental Procedures.” The residue numbers in B correspond to the residue numbers inA. P-, C-, and P+C-domains were subcloned as restriction fragments. The N- and N+Pdomains were sublconed using PCR technology. Two ovals in the N-domain depict the approximate location of the two predicted helices in the N-terminal end of the protein. KPEDWD repeats localized in the P-domain are bored.

A 9 I3 4:

-?q

40

F W-

m

50

60

100

G Y V F J X P m DQ!DM!GDSE

YNIMFGPDIC

GPGTKKVEW 1-

NKDIRCKDDE STaLYTLIVR

170 . 200 240 210 230 PDNTYEVXID NSQVESGS~ DDWUFWPXK IFDPLV~KPEuwobuurrDD PTD-

180 260

220

I(PER1PDPDA

250

320

DGEWEPPVIQ NPPEYRGEWW RQIDNPDYXG TWIEPEIDNP EYSPDANIYA 380

80

70

IESKEKSDFG KFVLSSCIM GDQEFDKGLQ TSQDARFYAL SARFEPFSNK GQPLWQFI’V

370 330 YAEEFGNETW

360 340

350

YDSFAVLGLDLWVKSGTIF DNFLITNDER

401 EEEREEDEE D F D D m E D E DEEDFDEEEE EARAGpAI[DEL

GVTKTAEKW FDXQDEEQRL

I

B

66

3

42

2 S a b c d e f g h i i k l m n

a b c d e t g h i

j

k l m n

31

C

a

S a b c d e f g h i j k l r n n

FIG.5. Expression,purification,andcharacterizationof calreticulin fusion proteins. E. coli lysates and purified proteins were separated by SDS-PAGE, transfered electrophoretically to nitrocellulose membranes and incubated with anti-calreticulin antibody, or with “Ca2’ as described under “Experimental Procedures.” A, Coomassie Blue staining; B, immunostaining with goat anticalreticulin; c, autoradiography of A after electrophoretic transfer and incubation with 4’Ca2+.Lane a, purified pancreatic calreticulin; lane b, affinity chromatography purified recombinant GST; lanes c, e, g,i, k, and m, E. coli lysates containing fusion proteins; lane d, f, h, j , 1, and n, glutathione-Sepharose 4B-purified fusion proteins. Lanes c and d, full length recombinant calreticulin; lanes e and f, C-domain; lanes g and h, P-domain; lanes i and j , N-domain; lanes k and 1, P+Cdomain; lanes m and n, P+N-domain; S, Bio-Rad low molecular weight standards in A and prestained Bio-Rad low molecular weight markers in C. The position of molecular weight marker proteins is indicated.

have further confirmed our earlier suggestion that the high capacity, low affinity Ca2+binding site is present in the Cterminal domain of calreticulin (Fig. 9, C-domain). Importantly, we have also specifically localized the high affinity

b

c

d

e

f

g

h . i

i

k

I

m

n

FIG.6. Stains-all staining of fusion proteins. Proteins were separated by SDS-PAGE and stained with the cationic carbocyanine dye Stains-All, as described by Campbell et al. (1983).Protein samples were the same as in Fig. 5. The position of molecular weight marker proteins is indicated.

0 0

1000

2000

3ooo

4ooo

ICal W

FIG.7. Ca2’ binding to GST-calreticulinfusionproteins. Ca2+binding to GST-fusion proteins was carried out by equilibrium dialysis as described under “Experimental Procedures.” Binding of “Ca2+ to the C-domain, Q Ca2+binding to the P-domain, W. SDSPAGE of the C- and P-domains used for Ca2+binding is shown in Figure 5A lanes f and h, respectively.

Binding

to Calreticulin

ea2+

.s % *

z . 3

100

w

4

$ 0 0

1000

2000 3000 ICal M I

4000

5000

FIG. 8. Ca2+ binding to the GST-P+C-domain fusion protein. Ca2+ binding tothe GST-P+C-domain of calreticulin was carried out by equilibrium dialysis as described under “Experimental Procedures.” Inset, Scatchard analysis of Caz+ binding data. SDSPAGE of the protein used for Ca2+binding is shown in Fig. 5A, lane 1. B, bound; BIF, boundlfree. C-domain P-domain N-domain NH;

KDEL-COOHighaffinity and low capacily rile

Lawrffinily and high capa