Coordinated participation of calreticulin and calnexin in the biosynthesis of myeloperoxidase.

Myeloperoxidase (MPO) is a neutrophil lysosomal hemeprotein essential for optimal oxygen-dependent microbicidal activity. We have demonstrated previously that calreticulin, a luminal endoplasmic reticulum protein, functions as a molecular chaperone during myeloperoxidase biosynthesis, associating reversibly with the heme-free precursor apopro-MPO. Because the membrane-bound endoplasmic reticulum protein calnexin is structurally and functionally related to calreticulin, we assessed the role of calnexin in myeloperoxidase biosynthesis. Like calreticulin, calnexin coprecipitated exclusively with glycosylated MPO precursors and with apopro-MPO but, in contrast to calreticulin, also with the enzymatically active, heme-containing precursor pro-MPO. To determine if calnexin participated in heme insertion into MPO, we compared the kinetics of chaperone association with MPO precursors using stable transfectants expressing cDNA encoding wild type MPO or mutated forms that do not acquire heme. Transfectants expressing mutant cDNA had prolonged association of MPO-related precursors with calreticulin and especially with calnexin. These studies demonstrate that 1) both calreticulin and calnexin associated with glycosylated apopro-MPO; 2) only calnexin associated selectively with the enzymatically active, heme-containing precursor pro-MPO; and 3) mutants unable to incorporate heme had prolonged association with calnexin. These findings represent the first evidence of a specialized role for calnexin in facilitating protein maturation in the endoplasmic reticulum of myeloid cells.

Myeloperoxidase (MPO 1 ; donor: H 2 O 2 oxidoreductase, EC 1.11.1.7) is a heme-containing lysosomal protein present exclusively in cells of neutrophil and monocyte lineage (1). In concert with hydrogen peroxide generated by the NADPH-dependent oxidase, MPO is a critical component of the most efficient microbicidal system in human neutrophils (1).
Biosynthesis of MPO is restricted to the promyelocyte stage in myeloid development and has been studied extensively using human promyelocytic cell lines (for review, see Refs. 2 and 3). We reported previously that calreticulin (CRT), a high capacity, low affinity calcium-binding protein located in the endoplasmic reticulum (ER) (4,5), interacts transiently with apopro-MPO, the heme-free protein precursor of MPO (6). Based on these studies we suggested that CRT functions as a molecular chaperone during MPO biosynthesis. Several studies have confirmed the ability of CRT to function in a similar fashion in the biosynthesis of a variety of unrelated proteins (for review, see Ref. 7).
Calnexin (CLN), a membrane-bound molecular chaperone located in the ER, shares significant structural and functional features with CRT (8,9). The regions of greatest similarity between CRT and CLN are those that extend into the lumen of the ER (8,10,11) and thus most likely to contribute to their shared capacity to interact with nascent glycoproteins. Published studies of the role of CRT and CLN in the biosynthesis of a given protein have not identified any functional differences in their capacity as molecular chaperones (7). We undertook these studies to determine to what extent CLN participated in MPO biosynthesis and if there were significant differences between these interactions and those we had observed previously between apopro-MPO and CRT.

EXPERIMENTAL PROCEDURES
Reagents-The human leukemia cell line PLB-985 (12) was acquired from Dr. Timothy Ley (Washington University, St. Louis, MO) and maintained in RPMI 1640 medium supplemented with 2 mmol/liter glutamine, penicillin-streptomycin, and 5% heat-inactivated fetal calf serum with 5% Serum-plus (JRF Biosciences, Lenexa, KS). Cells were free of mycoplasma infection. Tissue culture medium was obtained from the University of Iowa Cancer Center. For biosynthetic labeling, methionine-free RPMI (Life Technologies, Inc.) was supplemented with 1 mmol/liter pyruvate, 1 mmol/liter glutamine, antibiotics, and 10% dialyzed fetal calf serum. [ 35 S]Methionine (1,320 Ci/mmol) and ␦-[ 14 C]aminolevulinic acid (48.2 mCi/mmol) were obtained from Amersham Corp. and DuPont, respectively. Monospecific rabbit polyclonal antiserum against MPO has been described previously (13). Antiserum against CRT was generated using baculovirus-expressed recombinant CRT (6,14) and is available commercially (Affinity Bioreagents, Golden, CO), whereas that against CLN was raised against a synthetic peptide derived from the carboxyl terminus of CLN and was kindly provided by Drs. John J. M. Bergeron and David Y. Thomas of McGill University (Montreal, Canada). There is no cross-reactivity of the antisera against CRT and CLN (data not shown). Protein A was purchased from Life Technologies, Inc. and radiolabeled with 125 I at a core facility at the Iowa City Veterans Affairs Medical Center. Additional chemicals and reagents were obtained from Sigma.
Biosynthetic Labeling-PLB-985 cells were grown at 37°C in an atmosphere of 5% CO 2 in the medium described above. For biosynthetic labeling cells were suspended at 5.0 ϫ 10 5 /ml in methionine-free RPMI with 10% dialyzed fetal calf serum for 60 min. After methionine depletion, 25 mCi/ml [ 35 S]methionine was added and the cells cultured for the specified time interval. At the end of the labeling period, cells were collected by centrifugation for subsequent analysis, as described previously (15). In pulse-chase experiments, cells were resuspended in medium made 1 mM with unlabeled methionine and chased for the specified time interval. In experiments in which the heme group was radiolabeled, cells were cultured for 16 -20 h in the presence of medium supplemented with 15 mCi/ml ␦-[ 14 C]aminolevulinic acid.
Immunoprecipitation-Immunoprecipitations were done under ei-ther denaturing or nondenaturing conditions, as described previously (6). For both conditions, 100 ml of lysed cells or 700 ml of culture supernatant was used for immunoprecipitations as described previously (15). Samples were incubated with nonimmune serum for 30 min at 4°C followed by a 10% suspension of washed protein A-containing Staphylococcus aureus to clear the sample of radiolabeled proteins which might nonspecifically be precipitated. The cleared sample was incubated subsequently with primary antiserum. The antigen-antibody complexes were recovered with protein A-bearing S. aureus and the pellets washed serially with 1 ml each of 0.5% Triton X-100 in Trisbuffered saline (10 mmol/liter Tris buffer, pH 7.5, with 150 mmol/liter NaCl), 2 mmol/liter urea in 0.5% Triton X-100 in Tris-buffered saline, 1 mg/ml bovine serum albumin in 0.5% Triton X-100 in Tris-buffered saline, and Tris-buffered saline. After the final wash the antigen-antibody complex was released from protein A by heating at 100°C for 5 min in the presence of SDS sample buffer (62 mmol/liter Tris, 2 mmol/ liter EDTA, 5% ␤-mercaptoethanol, 2.3% SDS, pH 6.9). When immunoprecipitations were done under denaturing conditions, the cleared lysate was denatured by making the sample 2% SDS and heating to 100°C for 2 min. Prior to the addition of the primary antibody, the SDS concentration of the sample was reduced to 0.2% by the addition of dilution buffer (50 mM Tris HCl, pH 7.4, 190 mM NaCl, 6 mM EDTA, and 2.5% Triton X-100) and cooled on ice. Because chelation of calcium inhibits CLN association with glycoproteins (16 -18) EDTA was omitted from the dilution buffer when immunoprecipitating CLN. The presence of EDTA in the buffer had no effect on coprecipitation of MPO-related proteins with CRT (data not shown). In contrast, immunoprecipitations done under nondenaturing conditions omitted the addition of SDS and heating and proceeded directly with addition of the primary antibody to the cleared sample, as described above.
In sequential immunoprecipitations, protein-protein complexes recovered by immunoprecipitation under nondenaturing conditions were made 2.3% SDS and heated to 100°C for 2 min. Samples were cooled to room temperature and the SDS concentration reduced to 0.2% by the addition of dilution buffer before addition of the antibody for the second immunoprecipitation.
Analysis of Immunoprecipitated Proteins-Immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis. The resultant gel was fixed, soaked in 1 mmol/liter sodium salicylate (19), dried, and subjected to autoradiography. Quantitation of immunoreactive signals was generally done using PhosphorImager SF (Molecular Dynamics, Sunnyvale, CA). In some cases the relative intensity of the signal seen in autoradiography was quantitated using densitometry on a Shimadzu CS-9000U Dual Wavelength Flying Spot Scanner (Shimadzu Scientific, Tokyo, Japan).
Replicate Experiments-All experiments were performed in triplicate or greater. In the case of experiments with transfected K562 cells, replicate experiments were also performed with cells derived from different transfections.

CLN Associates with Biosynthetic Precursors of MPO-We
have demonstrated previously that CRT associates with apopro-MPO early in MPO biosynthesis (20). Like CRT, CLN serves as a molecular chaperone in the biosynthesis and folding of glycoproteins (for review, see Ref. 7). To determine if CLN associated with MPO precursors, PLB-985 cells were pulse labeled with [ 35 S]methionine and chased for 0 or 24 h in the presence of excess unlabeled methionine. Cell lysates were immunoprecipitated under nondenaturing conditions with monospecific antiserum directed against CLN or MPO. Immediately after pulse labeling, anti-CLN antiserum immunoprecipitated a 90-kDa protein that comigrated with that immunoprecipitated with MPO antiserum (Fig. 1). Sequential immunoprecipitations using antibody to CLN followed by that to MPO confirmed that the 90-kDa protein coprecipitating with CLN under nondenaturing conditions was immunochemically related to MPO (see below). Like CRT, CLN did not associate with mature MPO subunits. After the 24-h chase, during which time MPO precursors undergo proteolytic processing to the subunits of mature MPO (21,22), anti-CLN did not coprecipitate the 59-kDa heavy subunit of mature MPO (Fig. 1). In fact, the only precipitated protein under these conditions was CLN.
Neither CRT nor CLN Interacts with Nonglycosylated Ap-opro-MPO-Several studies demonstrate that CRT and CLN share an affinity for GlcMan 9 GlcNAc 2 oligosaccharides (23)(24)(25) and that this property is critical to their role as molecular chaperones for glycoproteins in the ER (7, 24 -29). The primary translation product of MPO is glycosylated at five asparagine residues cotranslationally (22). In the presence of tunicamycin, an antibiotic inhibiting cotranslational N-linked glycosylation (30), promyelocytic cells synthesize a nonglycosylated 80-kDa MPO precursor that does not associate with CRT (20) and fails to be processed into mature or enzymatically active protein. 2 Similar to CRT, CLN failed to associate with the nonglycosylated form of MPO precursor synthesized in the presence of tunicamycin (Fig. 2), indicating that the MPO precursors require either glycosylation per se, glycosylation-dependent folding, or both for firm association with CRT and CLN. The Association of CLN and MPO Precursors Is Transient-An essential characteristic of molecular chaperones is the transient nature of their association with folding intermediates in the ER (31). To examine the relative association of MPO precursors with CRT and with CLN, PLB-985 cells were pulse labeled for 60 min (Fig. 3A) or for 120 min (Fig. 3B) and immunoprecipitated sequentially with antisera to CRT, CLN, and MPO. At 1 h, most of the 90-kDa MPO precursor was associated with CRT, with lesser amounts associated with CLN or free of CRT or CLN (61, 3.7, and 18.2%, respectively). In contrast, after 120 min of labeling, only 27.2% of the MPO precursor was associated with CRT, whereas 31.2% was CLNassociated, and 41% was free of CRT and CLN. Taken together, these data suggest that MPO precursors associate first with CRT and subsequently with CLN. nolevulinic acid, and the presence of enzymatic activity can distinguish the two glycoproteins (21). We demonstrated previously that CRT associates exclusively with apopro-MPO, the heme-free precursor of MPO (20). Consistent with that observation, immunodepletion of CRT-associated MPO precursors did not decrease the amount of 14 C-labeled pro-MPO from PLB-985 cells (Fig. 4A). To determine if CLN also associated with only the apo-form of MPO, PLB-985 cells cultured in the presence of ␦-[ 14 C]aminolevulinic acid were immunoprecipitated with antiserum against MPO or CLN (Fig. 4B). As noted previously, radiolabeled heme was incorporated into the 90-kDa pro-MPO and the 59-kDa heavy subunit of mature MPO. In contrast to the results when CRT-associated MPO precursors were recovered, a fraction of the heme-containing pro-MPO coprecipitated with CLN. Only ϳ13% of the pro-MPO present was associated with CLN, indicating either that the association was restricted to a subpopulation of the pro-MPO or that the CLN⅐pro-MPO complex was relatively unstable under the conditions of immunoprecipitation.
Association of CRT and CLN with Mutated Forms of MPO-Taken together these data indicate that CRT and CLN each associated with fully glycosylated apopro-MPO transiently during MPO biosynthesis. Furthermore, CLN associated also with pro-MPO, the heme-containing, enzymatically active precursor. It is not clear from these data whether CLN facilitates heme insertion into the peptide backbone of apopro-MPO or if heme insertion triggers dissociation of CLN from the newly formed pro-MPO. To examine this question in more detail, we compared the interactions of CRT and CLN with MPO precursors expressed in a K562 cell line transfected with wild type or specifically mutated cDNA for MPO (33).
We identified previously a missense mutation in exon 10, whereby an arginine at codon 569 is replaced by a tryptophan (R569W), as one genotype responsible for hereditary MPO deficiency (34). When transfected into K562 cells, mutant cDNA with R569W exhibits a maturational arrest in MPO biosynthesis; heme is not inserted into the mutant apopro-MPO, neither pro-MPO nor mature MPO is formed, and the cells expressing this mutation lack any peroxidase activity attributable to MPO (33). To assess the impact of R569W and its defective heme insertion on the association of CRT and CLN with MPO precursors, we examined MPO biosynthesis in K562 cells trans-fected with cDNA for normal and mutant MPO.
To verify that chaperone association of MPO precursors in PLB-985 cells was mirrored accurately by the transfected K562 cells, PLB-985 cells and K562 cells expressing normal (pREP-MPO) and mutant (pREP-R569W) cDNA for MPO were pulse labeled for 1 h and immunoprecipitated directly with MPO antiserum or immunoprecipitated serially with antisera against CRT or CLN followed by MPO antiserum (Fig. 5A). In both pREP-MPO and pREP-R569W cell lines, CRT and CLN coprecipitated a 90-kDa MPO precursor.
To compare the transient nature of the association of the complex in cells expressing normal MPO with those expressing the R569W mutation, cells were pulse labeled for 30 min and chased for 0 -180 min prior to immunoprecipitation under nondenaturing conditions with antisera against MPO, CRT, or CLN. During biosynthesis of MPO in pREP-MPO cells, CRT and CLN each associated transiently with MPO precursors (Fig. 5B). In pREP-R569W cells there was slightly less MPO precursor associated with CRT than there was in pREP-MPO, but more striking was the delay in dissociation of the CLN⅐MPO complex seen in pREP-R569W cells compared with cells expressing the cDNA for normal MPO. At 180 min of chase, 34% of the apopro-MPO associated with CLN at time zero was still in the complex compared with only 5% in the pREP-MPO cells. Because the R569W mutation results in the failure of heme to be incorporated into apopro-MPO and thus form pro-MPO (33), the prolonged association of the apopro-MPO⅐CLN complex suggests that heme insertion may be necessary for the release of CLN from the MPO precursor. Alternatively, the prolonged stability of the apopro-MPO⅐CLN complex and the failure of heme insertion may both reflect the misfolded state of the precursor resulting from the R569W missense mutation.
Because the replacement of arginine with tryptophan may significantly alter structure in addition to blocking heme insertion, we created pREP-H502, mutating the histidine that is the distal heme-binding site in MPO (35,36). After pulse labeling, the pREP-H502 product appeared to be less stable than normal MPO, as less than 10% of immunoreactive MPO was produced relative to that made by pREP-MPO (Fig. 6A). The mutant precursor in pREP-502H lacked peroxidase activity and was not processed into the subunits of mature MPO (Fig. 6A). In addition, the association of CLN with MPO precursors was prolonged in pREP-H502 cells (Fig. 6B). At 5 h of chase, 15.7% of the MPO precursor associated with CLN was still bound to CLN in the pREP-MPO cells, whereas 31.3% was still complexed to CLN in pREP-H502 cells (n ϭ 3). Thus, as seen with cells expressing R569W, alterations in heme insertion were associated with delayed dissociation of the CLN⅐MPO precursor complex. DISCUSSION MPO is the only heme or lysosomal protein demonstrated to date to associate with molecular chaperones during its biosynthesis (7). However, as with many other glycoproteins, quality control during its biosynthesis is mediated by transient association with ER resident proteins. The data presented here indicate that, like several other glycoproteins, precursors of MPO associated sequentially with different ER resident proteins, in this case two structurally similar proteins, CRT and CLN.
In specific situations in which they have been compared, CRT and CLN have shown striking similarity in the substrates to which they bind (23)(24)(25), consistent with their structural homology and similar lectin specificities. Several studies have demonstrated sequential interactions of the folding intermediates of the same protein with CRT and CLN during its biosynthetic maturation (37,38) as well as coincident interaction of a given substrate with CRT and CLN (7, 39 -42), although significant differences in substrate specificity for CRT and CLN have also been noted (27,37,43,44). Despite these similarities between CRT and CLN, we found evidence of functional discrimination in their ability to interact with biosynthetic precursors of MPO in the ER. In the case of CRT, only glycosylated apopro-MPO associated in a complex; heme-containing pro-MPO or mature MPO subunits did not coprecipitate with CRT under any conditions. CLN, on the other hand, bound not only apopro-MPO but also pro-MPO; as with CRT, neither subunit of mature MPO coprecipitated with CLN. Furthermore, mutations that inhibited the insertion of heme into the peptide backbone of apopro-MPO, R569W and H502, resulted in prolonged association of CLN and the mutant apopro-MPO. CLN may be subserving an ER "quality control" function for MPO biosynthesis, as these two particular mutations may have produced significant misfolding. As a result, the misfolded precursor would be unable to accommodate heme and would associate with CLN in a more stable complex. Alternatively, these findings may indicate that heme insertion is necessary for the normal dissociation of CLN from MPO precursors. Our studies cannot distinguish between these two possible interpretations of the findings.
The active site in MPO is unusual and has been the subject of considerable study in the past (36,(45)(46)(47)(48)(49)(50)(51)(52). Data derived from the crystal structure of mature MPO indicate that a protoheme IX derivative is associated with the protein backbone via a methylsulfonium bond with methionine 409 and with histidines at positions 261 and 502, the distal and proximal coordination positions, respectively (35). The data presented support the hypothesis, previously proposed by us and others (32,53,54), that heme insertion is a prerequisite for proteolytic maturation of MPO precursors into the enzymatically active subunits of mature MPO. On a broader level, the data may suggest that CLN serves a highly specialized function, i.e. to facilitate insertion of the prosthetic group into a hemeprotein, a function previously not attributed to molecular chaperones in the ER and not shared by the structurally related ER protein CRT.