Calreticulin is at the surface of circulating neutrophils and uses CD59 as an adaptor molecule.

Calreticulin, which has been proposed to be a C1q receptor on neutrophils, has neither a transmembrane domain nor a GPI-anchor attachment site and must utilize an adaptor molecule to attach to the plasma membrane. The expression of ecto-calreticulin on purified human neutrophils did not result from contamination by soluble or intracellular calreticulin released during cell fractionation because it was expressed on circulating neutrophils, and the expression did not increase significantly with neutrophil isolation. All neutrophils expressed calreticulin with a unimodal distribution. Treatment of neutrophils with either a cholesterol-binding agent or phosphatidylinositol-specific phospholipase C dramatically decreased ecto-calreticulin expression indicating that the adaptor molecule(s) are located in lipid rafts and have a GPI-anchor. Analysis for the co-expression of specific GPI-anchored proteins and ecto-calreticulin in cells that were deficient in specific GPI-anchored proteins, indicated that ecto-calreticulin was best associated with CD59. Calreticulin reciprocally immunoprecipited with CD59, which provided direct evidence that CD59 is an adaptor for ecto-calreticulin. Immunofluorescence and confocal microscopy demonstrated that ecto-calreticulin co-localized with a fraction of CD59 at the cell surface. Cross-linking ecto-calreticulin with antibodies induced a Ca2+ flux, which suggests that ecto-calreticulin is capable of signaling following ligand binding. Ecto-calreticulin has been associated with diverse cellular functions. An appreciation that the adaptors for ecto-calreticulin are GPI-anchored will provide a framework for understanding any common features underlying ecto-calreticulin ligation.

of CRT in the endoplasmic reticulum is due to the KDEL receptor, which interacts with the KDEL-containing C-terminal domain of CRT. In addition to its endoplasmic reticulum location, CRT has been found at other sites including its functional presence at the surface of a variety of cells. Most of these studies utilized cultured cells, such as mouse melanoma cells (1), human fetal fibroblasts (2), HeLa cells (3), in vitro activated T cells (4), bovine aortic endothelial cells (5), CEM, and Jurkat cells (6). Because there is always a fraction of dead and dying cells within a population of cultured cells, it raises the question as to whether the ecto-calreticulin (eCRT) is derived from moribund cells and stuck to the surface of viable cells. However, eCRT has been identified on the surface of freshly isolated human polymorphonuclear neutrophils (PMN) (7,8), a finding that we have confirmed (9).
Engaging eCRT induces different effects in different cells, e.g. for macrophages, ligation of eCRT with either complement C1q-, mannan-binding lectin-, surfactant protein A-, or surfactant protein D-opsonized apoptotic cells, induces phagocytosis of the apoptotic cell by a macrophage CD91-dependent mechanism (10,11). For fibroblasts, engaging eCRT by B␤ chain of fibrinogen induces mitosis (2), whereas engaging eCRT by the collagen domain of C1q induces a pro-apoptotic, anti-mitotic effect (12). CRT, which lacks a transmembrane domain, requires an adaptor molecule that is a resident of the plasma membrane to be expressed at the cell surface. The effect of eCRT engagement would then depend to a great extent on the identity of the adaptor molecule(s) of the plasma membrane.
In this study, we report that CRT is present at the plasma membrane of circulating PMN. Normal 293 cells expressed eCRT, whereas GPI-anchor-deficient 293 cells were also eCRT deficient, consistent with the putative adaptor molecule(s) for eCRT being GPI-anchored. Using immunoprecipitation and confocal microscopy, we have identified CD59 as a major adaptor protein for eCRT in PMN. Finally, cross-linking eCRT with primary and secondary antibodies induced a calcium flux in PMN, demonstrating that eCRT ligation with a natural ligand has the potential to initiate intracellular signaling.
Cells-Leukocytes were derived from finger prick blood (150 l) that was mixed with 1 ml cold HBSS 2Ϫ , 2 mM EDTA, the cells pelleted by centrifugation, and the erythrocytes lysed by ammonium chloride (8.1 g/l ammonium chloride, 1.0 g/l potassium bicarbonate, and 0.037 g/l EDTA) for 5 min. The leukocytes were then washed twice in HBSS. Dextran-sedimented leukocytes were isolated from 40 ml of ACD-anticoagulated blood obtained by venopuncture, as described (18). PMN were further fractionated from the dextran-leukocyte preparation by centrifugation through Ficoll-Paque at 3000 ϫ g for 20 min. 293 fetal kidney epithelial cells were transfected with an SV40 large T antigen plasmid and exposed to ethylmethanesulfonate. Subsequently, a normal GPI-anchor expressing cell line (293Tag1.6) and a GPI-anchordeficient expressing cell line (A293.2.2) were cloned by limiting dilution (19). PPC-1, prostate epithelial cells, were provided by S. Tomlinson (Medical University of South Carolina).
Flow Cytometry-Cells, which were impermeable and alive unless otherwise noted, were incubated for 15 min with antibodies, as noted in each figure, in FACS buffer (HBSS ϩ 5% BSA) at 4°C, followed by two washes and incubation for 15 min with secondary antibody at a dilution recommended by the manufacturer. Cells were washed once and analyzed in a FACScan TM (BD Biosciences). In all the experiments at least 10,000 events were recorded, and the results were analyzed using CellQuest Pro 4.0.1.
Immunofluorescence and Confocal Microscopy-All the staining steps were performed in 1.5-ml Eppendorf tubes at 4°C for 10 -15 min in HBSS ϩ 0.5% BSA. Before adding primary and secondary antibodies cell were blocked by incubation with HBSS ϩ 1.5% BSA for 10 min. The final concentration of primary antibodies was 10 g/ml, and the dilutions of secondary antibody were as specified by the manufacturer. After incubation with secondary antibodies, PMN were washed twice and resuspended in HBSS ϩ 1% BSA. For co-localization of eCRT and CD59 in spread cells, PMN were allowed to adhere to the slide for 5-7 min, fixed with 3.8% paraformaldehyde for 4 min, washed and mounted in PBS. Images were acquired with a Retiga EXi monochrome camera equipped on an Olympus Provis fluorescence microscope fitted with FITC and rhodamine filters. For all images a 40 ϫ 1.00 UPlan Apo oil lens was used. Acquired images were further processed using Adobe Photoshop 6.0. For confocal analysis, cells were resuspended after the final wash directly in anti-fading media (Prolong, Molecular Probes, Eugene, OR) and mounted on slides. Fifty-two sequential Z-sections at 0.15 m intervals were acquired at the resolution of 512 ϫ 512 using a Bio-Rad 1024 confocal device fitted on a Nikon Eclipse E-800 microscope with a 100 ϫ 1.4 Plan Apo oil lens. Images were further processed using AutoDeblur AutoVisualize 9.0 (AutoQuant Imaging, Troy, NY). The gain of the photo-detectors was set so that there was no detectable cross-signal from the second channel when slides with just one fluorochome were imaged.
Immunoprecipitation and Western Blotting-Isolated PMN were washed twice in HBSS and lysed in lysing buffer (25 mM Tris-HCl (pH 7.6), 150 mM NaCl, 5 mM EDTA, 0.1% Nonidet P-40, and protease inhibitor mixture (Sigma, P 2714) for 30 min on ice. The mixture was centrifuged for 20 min at 15,000 ϫ g, and the post-nuclear supernatant was pre-cleared for 12 h with protein AϩG beads (Pierce) for another 4 h with beads-control mAb antibody by end-over-end rotation. The pre-cleared supernatant was then incubated for 4 h with protein AϩG beads coupled with either anti-CRT mAb (SPA-601) or anti-CD59 (p282) or anti-CD16 (3G8). Beads were washed in lysing buffer (with Nonidet P-40 0.05%) four times (30 min total) and boiled in nonreducing loading buffer for 5 min. Samples were run on 10% NuPage Bis-Tris gels (Invitrogen), transferred on nitrocellulose paper (Hybond ECL, Amersham Biosciences), and blocked with nonfat dry milk 6% in Tris buffer with Tween 0.1% for 1 h at room temperature. Membranes were incubated with anti-CRT mAb (SPA-601), anti-CD59 (YTH53.1), or anti-CD16 mAb (3G8) for 30 min at room temperature, washed, and incubated with horseradish peroxidase-conjugated appropriate secondary antibody for an extra 30 min. Nitrocellulose membranes were washed extensively in Tris buffer with Tween 0.1% and developed with SuperSignal West Pico Chemiluminescent Substrate (Pierce) and exposed to XAR film (Kodak, Rochester, NY).
Spectrofluorimetry-Freshly purified PMN were re-suspended in HBSS 2Ϫ at a concentration of 10 7 cells/ml and loaded with Fura-2/AM (Molecular Probes) per the manufacturer's instructions for 30 min in HBSS 2Ϫ at room temperature followed by two washes in cold HBSS 2Ϫ . Cells were then re-suspended in HBSS ϩ 1.5% BSA and incubated with either anti-CD59 (MEM 43/5) mAb or anti-CRT IgY and corresponding control antibodies for 15 min at 4°C. Subsequently the cells were washed with cold buffer then warmed to 37°C for 10 min prior to starting each experiment. The addition of secondary antibody was done while monitoring Ca 2ϩ by measuring Fura-2 fluorescence at 505 nm, using 340/380 nm excitation in a FluoroView F-4500 spectrofluorometer (Hitachi Instruments, San Jose, CA) with constant stirring at 37°C. The data were analyzed with F4500 Intracellular Cation Measurement software (Hitachi).

Calreticulin Is Expressed on the Surface Circulating and
Isolated PMN-During a typical isolation procedures the viability of PMN decreases from 100% to about 92-94% as assessed by propidium iodide/acridine orange staining. Dying cells release proteins normally found in the cytoplasm, which can potentially bind nonspecifically to the surface of surrounding cells. This process could explain the presence of an endoplasmic reticulum-resident protein at the cell surface. To test this possibility, we purified PMN and checked them for CRT expression during all the steps. The expression of eCRT increased ϳ10% during the isolation of the cells (Fig. 1), but importantly eCRT was expressed on native PMN derived from finger prick blood (Fig. 1a). CRT lacks a transmembrane do-FIG. 1. eCRT is present on circulating PMN and does not increase significantly upon purification. Blood (150 l) from a finger prick was washed once in 1 ml of cold HBSS 2Ϫ ϩ 2 mM EDTA and red blood cells lysed for 5 min with ammonium chloride. The remaining leukocytes were stained at 4°C for eCRT expression and PMN gated in for analysis (left). ACD-anticoalgulated blood (40 ml) was mixed with dextran and allowed to sediment for 60 min at room temperature. Cells from the supernatant were washed once in HBSS 2Ϫ and stained for eCRT. The PMN population was gated in for analysis (middle). A 1-ml aliquot of the dextran-sedimented cells was used to further purify the PMN by centrifugation through Ficoll-Paque. The pellet that contained more than 95% PMN was analyzed for eCRT expression (right). The filled histograms represent control antibody. main and has no alternative splicing variants that would predict lipid modification to allow subsequent membrane attachment (20). 2 Thus, eCRT requires an adaptor molecule for expression at the plasma membrane. If the interaction between eCRT and its adaptor molecule were specific, modulation of the adaptor expression might be followed by concordant changes in eCRT expression.
eCRT Expression Is Not Modulated by fMLP-Because the up-or down-regulation of eCRT expression could provide insight as to adaptor protein for eCRT, we exposed PMN to fMLP concentrations ranging from 10 Ϫ8 M to 1.5 ϫ 10 Ϫ7 M, and the eCRT expression was assessed by flow cytometry. The maximum dose of fMLP resulted in less than 10% loss of eCRT (Fig.  2a). Additionally, C5a (3 and 33 nM) also failed to significantly change eCRT expression as evidenced by decrease in mean fluorescent channel of 12 and 9%, respectively.
CRT binds calcium and it has a lectin-like domain. To determine whether eCRT were bound to an adaptor in a calcium-dependent manner, we assessed the effect of exposing PMN to buffered EDTA (5 mM) for 15 min at 37°C. EDTA treatment did not significantly affect eCRT expression (Fig. 2b). This result also effectively rules out integrin-mediated binding of CRT to the cell surface, consistent with the results from Fig. 1. We did note spontaneous shedding over a 3 h incubation in HBSS at room temperature, as noted previously (8,21), and the shedding was enhanced by the additional exposure of PMN to 0.3 M KCl for 20 min (Fig. 2c), suggesting that the adaptor-eCRT interaction depends on ionic and not hydrophobic bonds.
Lipid Raft-depleting Reagents and Phosphatidylinositol-specific Phospholipase C (PIPLC) Decrease the Expression of eCRT-To continue a systematic search for the adaptor molecule(s) responsible for anchoring eCRT to the cell surface, we next evaluated if GPI-anchored proteins were involved. We pre-treated PMN with methyl-␤-cyclodextrin (MBCD) or control buffer and assessed the expression of eCRT. MBCD is a water-soluble cyclic oligomer of D-glucopyranose (6 -8 units usually) with a hydrophilic surface, which explains its water solubility, and a central non-polar cavity, which binds cholesterol. Unlike other cholesterol-binding agents, MBCD does not interact directly with the plasma membrane, but rather acts from the surface of the cell membrane by extracting specifically cholesterol. MBCD treatment depletes the cell of cholesterol in a lipid raft-independent manner but partially solubilizes proteins from lipid rafts (22). Compared with the buffer-treated control, MBCD treatment decreased the expression of eCRT on PMN by more than 60% (Fig. 3, a and b). Our results strongly suggested that the adaptor for eCRT resided in lipid rafts. Lipid rafts are heterogeneous in composition, but contain mainly GPI-anchored proteins and a few transmembrane proteins (reviewed in Ref. 23). PIPLC is an enzyme that specifically cleaves GPI-anchored proteins at their lipid anchor, making possible the distinction between transmembrane and GPIanchored proteins in lipid rafts. PMN were pre-incubated with PIPLC for 30 min, washed, and then incubated with chicken anti-CRT or mAb anti-CD16. CD16 is known to be a GPIanchored protein on PMN (24). PIPLC treatment of PMN decreased their expression of eCRT by over 50% (Fig. 3c). PIPLC is not 100% efficient in removing all the GPI-anchored proteins from cells (24) and as evidenced by the partial removal (50%) of CD16 in this experiment (Fig. 3d).
GPI-anchored Protein-deficient Cells Are Devoid of eCRT-The removal of eCRT by PIPLC strongly suggested the possi- bility of a GPI-anchored adaptor for CRT. To test this hypothesis we utilized a sub-line of 293 cells that had been chemically mutated and negatively selected using antibodies against GPIanchored proteins (19). CD59, a known GPI-anchored protein, had a normal expression on the wild type cells, and eCRT was also expressed (Fig. 4, a and c). The GPI-anchor-deficient cells failed to express CD59 and failed to express eCRT (Fig. 4, b and  d). These results provided further evidence that eCRT expression was linked to the expression of GPI-anchored proteins, but it did not help identify the relevant GPI-anchored protein(s) because all GPI-anchored proteins were missing from the mutated cells.
eCRT Expression Follows the Expression of CD59 and CD16 on the Affected PMN from Paroxysmal Nocturnal Hemoglobinuria (PNH)-like Patients-Leukocytes normally express several GPI-anchored proteins on their surface, including CD16, CD59, and CD87 (Fig. 5c), and express eCRT (Fig. 5d). Leukocytes from patients with PNH are characteristically deficient in the two complement regulatory proteins DAF (CD55) and CD59 and may be deficient in other GPI-anchored proteins as well. We compared eCRT expression in the PMN from a normal donor and two PNH-like patients. The first atypical PNH patient's PMN were DAF Ϫ (data not shown), CD87 Ϫ , CD16 ϩ , CD59 ϩ , and eCRT was expressed normally (Fig. 5, a and b). These results were consistent with, but did not prove, the finding that either CD16 or CD59 might be the adaptor for eCRT. Subsequently, we assessed a more typical PNH patient whose PMN were DAF Ϫ (data not shown), CD16 Ϫ , CD87 Ϫ , and CD59 low . In this case the PMN were eCRT low , consistent with CD59 being the adaptor and eliminating CD16 as a major adaptor (Fig. 5, e and f). These patient data were confirmed using the prostate epithelial cell line PPC-1, which were CD59 ϩ , CD16 Ϫ , and eCRT ϩ with net mean fluorescent channels of 188, 2.3, and 46.5, respectively.
Reciprocal Immunoprecipitation of CRT and CD59 -The association of CRT with CD59 expression on the PMN from PMN patients and PPC-1 cells prompted us to see if there were any physical interactions between eCRT and CD59. We immunoprecipitated eCRT using rabbit anti-CRT Ab and probed the samples using either anti-CD59 mAb YTH 53.1 or anti-CD16 mAb. CD59 co-immunoprecipitated with e-CRT, whereas CD16 did not (Fig. 6, upper panels). Performing the reciprocal immunoprecipitation with anti-CD59 mAb, we detected eCRT by immunoblotting in a band of 66 k M r (Fig. 6, lower panel), the same M r as intracellular CRT (data not shown). Interestingly, only one anti-CD59 mAb (MEM 43/5) was effective in immunoprecipitating eCRT. The other anti-CD59 mAbs that failed to co-immunoprecipitate eCRT, including YTH 53.1 and BRIC229, which bind epitope 1, and p282, which binds epitope 3 (25). These data suggest that e-CRT shields epitopes 1 and 3 of CD59, but not epitope 2.
eCRT Co-localizes with a Fraction of CD59 at the Surface of the PMN-Using fluorescence microscopy it is possible to detect the presence of two or more proteins in the same location at the same time (co-localization) in a cell. We used this approach to investigate the distribution of eCRT and CD59 on human PMN. Freshly purified cells were incubated in HBSS ϩ 1% BSA with either with rabbit anti-CRT Ab or with rat mAb anti-CD59 (YTH 53.1) for 10 min on ice followed by two washes and incubation by Cy-3 anti-rabbit and Cy-2 anti-rat secondary antibodies. The results demonstrate that eCRT is present at the surface of PMN (Fig. 7, b-d) and co-localizes with a fraction of CD59 (Fig. 7f). Similar results were obtained with three other anti-CD59 mAbs. As a negative control we used mAb CBR M1/5, which recognizes an activation-specific epitope of CD11b/CD18 (17) (Fig. 7h). In migrating PMN, most of the cross-linked proteins (if not all) tend to accumulate at the rear of the cell producing the illusion of co-localization, as is demonstrated in Fig. 7h. In additional studies, to avoid any possible false results, cells were fixed while still round and imaged using confocal microscopy followed by deconvolution. A PMN is shown after three-dimensional reconstitution using both red (eCRT) and green (CD59) channels showing all the surface staining (Fig. 8a), and two sections through the middle of the cell, which show rings of staining and prove that the cell was not permeable at the time of labeling (Fig. 8, b and c). Most of the signals came from the same location at the cell surface (Fig. 8d), whereas others were either adjacent or separate (Fig. 8, e and f ). All the cells exam-ined had more CD59 on their surface than eCRT, but different donors had different levels of eCRT expression.
Antibody Cross-linking of eCRT on PMN Induces a Calcium Flux Similar to that Induced by Anti-CD59 Cross-linking-Antibody-induced cross-linking of GPI-anchored proteins can stimulate intracellular signaling. The signaling pathway used when CD59 is cross-linked includes a Ca 2ϩ flux and the activation of tyrosine kinases (26,27). To test if cross-linking eCRT would induce a signal, cells were loaded with Fura-2/AM for 30 min, washed in cold HBSS 2ϩ , and mixed with either anti-CD59 mAb or anti-CRT IgY for 15 min at 4°C. Subsequently the cells were washed with cold buffer then warmed to 37°C for 10 min prior to addition of secondary antibody. As a negative control cells were treated with non-immune IgY, followed by secondary Ab, and no flux was seen (Fig. 9, bottom). Cross-linking eCRT induced a smaller Ca 2ϩ flux (Fig. 9, middle) than cross-linking CD59 (Fig. 9, top), and one possible explanation is that there is less eCRT-CD59 for cross-linking than there is total CD59, i.e. (eCRT-CD59 ϩ free CD59) for cross-linking, as noted in the confocal images (Figs. 8, a-c). DISCUSSION In our systematic search for the adaptor molecule(s) responsible for attaching eCRT to the plasma membrane, we found no evidence that the chemotactic factors fMLP (Fig. 2a) and C5a significantly modified the expression of eCRT on PMN, arguing against CRT being stored in a mobilizable intracellular compartment. Experiments with MBCD and PIPLC supported the existence of GPI-anchored adaptor molecule(s) for eCRT (Fig.  3). Finding that the ethylmethanesulfonate-mutated A293.2.2 cells, which lack GPI-anchors, also lacked eCRT was further evidence that adaptor(s) for eCRT were GPI-anchored (Fig. 4). However, chemical mutagenesis can induce ''silent'' mutations, Additional primary Abs were added as follows: anti-CD59 (g) or mAb that recognizes an activation epitope of MAC-1 (CBR M1/5) (h). Cells were incubated with primary Ab for 10 min at 4°C. After the incubation cells were washed once in cold HBSS ϩ 1% BSA and incubated with Cy-2 anti-rat and Cy-3 anti-rabbit secondary antibodies for 10 min on ice. Cells were washed once and allowed to adhere to the slides for 7 min and then fixed. DIC Nomarski (a, b, and d) and phase contrast images (e and g) were used either alone or overlaid with fluorescence microscopy images (b and d). Although control Ab showed no significant fluorescence (a), eCRT (b and magnified in d) displayed a punctated distribution clustered at the rear of the cell. A cell co-stained for eCRT and CD59 is shown in phase contrast microscopy (e) and fluorescence (f ). The signals from both antibodies were clustered at the rear of the cell. A similar pattern (distribution at one end of the cell) was seen when anti-CD59 and CBR M1/5 mAb were used (h). and therefore the additional evidence provided by the samples from PNH patients was helpful (Fig. 5). PNH is now thought to be an autoimmune disease with selective selection against the GPI-anchor of CD55 and CD59, and routinely some selection against the GPI-anchors of other proteins as well (28), but other non-GPI-anchored proteins are generally normal. From our PNH studies, we could conclude that there was an association with CD59 deficiency and eCRT deficiency. Specific association between CD59 and eCRT was shown by reciprocal immunoprecipitation experiments (Fig. 6). Monoclonal antibody to epitope 2 of CD59, but not to epitopes 1 and 3, co-precipitated eCRT (25). All these mAb stained PMN equivalently (data not shown), so it is most likely that CD59-associated eCRT was displaced by mAb to epitopes 1 and 3, and thus these mAbs may identify the eCRT-binding site. The confocal microscopy study showed that some, but not all, of the eCRT co-localizes with CD59 (Fig. 8). We do not know the characteristics of the other adaptor molecule(s), but our studies of the 293 cells and PNH patients argue that they must be GPI-anchored, and we specifically know from our studies with PNH patients that DAF is not involved (data not shown).
CD59 was initially described as a complement regulatory protein that interacts with C8 and C9 to inhibit cell lysis by the terminal complement pathway (15, 29 -32). It is unknown whether eCRT bound to CD59 inhibits or augments the complement regulatory activity of the latter. Erythrocytes depend heavily on CD59 to protect them from complement-mediated lysis (33), and the fact that they bear minimal levels of eCRT (9) suggests that eCRT is not required for CD59 complement regulatory activity. On the other hand, if the eCRT⅐CD59 complex inhibits the assembly of the terminal complement components on the cell surface, then eCRT may modulate complementdependent signaling to the cell (27,34,35). The calcium flux that we (Fig. 8) and others (36) have noted when eCRT is cross-linked is consistent with the finding that cross-linking any of the abundant GPI-anchored proteins on PMN (CD16, CD48, CD55, CD58, and CD59) induces a calcium flux (26).
CD91 has been associated functionally with eCRT in human macrophages, initially as a candidate receptor for soluble CRT uptake (37,38), and additionally as part of a complex with eCRT that functions for the uptake of C1q-and collectin-opsonized particles, by their respective collagen domains (10,11). However, no direct binding of eCRT to CD91 has been demonstrated to date. PMN do not express CD91 and therefore would require a different molecule as their adaptor (39). 293 cells express both CD59 and low levels of CD91, the latter of which is not GPI-anchored (39). In the GPI-anchor-deficient A293.2.2 cells we found that eCRT was also deficient (Fig. 4). Thus, in PMN and in the fetal kidney epithelial cell line 293, CD91 does not serve as an adaptor for eCRT. PIPLC treatment of human monocytes also decreased expression of cell surface CRT (data not shown), and if this were also true for monocyte-derived macrophages, it seems likely that CD91 must participate in a complex with eCRT and a GPI-anchored adaptor.
We have found that CR1 (complement receptor 1, CD35) is a functional phagocytic receptor on PMN and adhesion receptor on erythrocytes for the collagen domain of complement C1q and the homologous opsonin, mannan-binding lectin (18,40,41). However, the ability of isolated or recombinant CRT to bind C1q is also well described, giving rise to the name cC1qR (42)(43)(44)(45). In equilibrium binding assays, C1q collagen tail binding to purified CRT requires low ionic strength buffer, as does binding to CR1 (40,46); whereas C1q globular domains bind to adjacent sites in recombinant CRT in normal ionic strength buffer (8). Immobilized C1q induces PMN to make and release superoxide (47,48). The PMN receptor that mediates this response is not CR1 (48). Others have found that rabbit anti-cC1qR partially inhibits C1q-mediated superoxide production by PMN (21), a finding we have confirmed using anti-CRT IgY against a peptide of amino acid residues 24 -43 (data not shown). These data suggest that eCRT might be the receptor for this C1q-mediated PMN response.
In our studies eCRT and intracellular CRT had the same mobility on gels. However, rabbit antiserum raised against the C1q-binding protein isolated from plasma membranes (''cC1qR'') (42) cross-reacts with calreticulin, but consistently recognizes a band of 68 k M r , whereas calreticulin has a M r of 60 k (7). Thus, although it seems likely that C1q binds cellular CRT, the relationship between cC1qR and eCRT remains unclear. Although there is a second gene for CRT in humans, its expression is limited to testis, and thus cannot explain any putative differences between intracellular and eCRT in PMN (49).
Finally, the functional consequences of ligand binding to eCRT will depend upon the adaptor proteins engaged. Our identification of GPI-anchored proteins as the primary adap- FIG. 9. Cross-linking eCRT induces Ca 2؉ in human PMNs. Purified PMN were loaded with 5 M Fura-2/AM for 30 min at 37°C in PBS, washed twice, and incubated with anti-CD59 mAb or anti-CRT polyclonal Ab (IgY chicken). Cells were washed twice, resuspended in HBSS, and then secondary antibodies were added (arrows). PMN incubated with control primary rabbit or rat antibody (data not shown) or chicken IgY (lower panel) showed no Ca 2ϩ increase.