Isolation and function of a human endothelial cell C1q receptor

It has been shown previously that cultured human venous and arterial endothelial cells (EC) bind C1q in a time- and dose-dependent manner. Cultured human endothelial cells express an average number of 5.2 × 105 binding sites/cell. In the present study the putative receptor for C1q (C1qR) was isolated from the membranes of 1–5 × 109 human umbilical cord EC by affinity chromatography on C1q–Sepharose. During isolation, C1qR was detected by its capacity to inhibit the lysis of EAC1q in C1q-deficient serum. The eluate from C1q–Sepharose was concentrated, dialysed and subjected to QAE-A50 chromatography and subsequently to gel filtration on HPLC–TSK 3000. C1qR filtered at an apparent molecular weight of 60 kDa. Purified C1qR exhibited an apparent molecular weight of 55–62 kDa in the unreduced state and a molecular weight of 64–68 kDa in reduced form. Two IgM monoclonal antibodies (mAb) D3 and D5 were raised following immunization of mice with purified receptor preparations. Both monoclonal antibodies increased the binding of 125I-C1q to endothelial cells but F(ab')2 anti-C1qR mAb inhibited the binding of a125I-C1q to EC in a dosedependent manner. The D3 mAb recognized a band of 54–60 kDa in Western blots of membranes of human EC and polymorphonuclear leukocytes. Previously, the authors showed that C1q induces the binding of IgM-containing immune complexes to EC. Therefore, it was hypothesized that during a primary immune response generation of IgM-IC may occur, resulting in binding and activation of C1, dissociation of activated C1 by C1 inhibitor and subsequent interaction of IgM-IC bearing C1q with EC–C1qR.


Introduction
The endothelial cell layer represents a barrier between the circulation and the vessel wall, and may play an important role in processes which mediate inflammation. was present on endothelial cells, namely a receptor for Clq, was obtained by Linder 6 and further established by other, v The authors have described previously that human umbilical cord venous and arterial EC express substantial numbers of ClqR. More insight into the binding of Clq to a variety of somatic and cultured ceils and further identification of a putative ClqR has been obtained more recently. [9][10][11][12] In the present study ClqR was purified from the membranes of cultured umbilical cord EC and it appears to be closely related to Clq receptors 13 and was maintained in culture as described previously. 19 Solubilization of HUVEC and EAhy.926: Confluent layers of cells were rinsed with sterile PBS, the cells detached subsequently by incubation for 30 min at 0C in PBS containing 10 mM EDTA, and washed three times by centrifugation at 150 g and resuspension in ice-cold water containing 5 mM EDTA. Usually, 1-5 109 cells were resuspended in 5 ml water, and frozen at -80C. Thereafter the cells were frozen and thawed a total of five times.
The resultant mixture was centrifuged for 10 min at 15 000 " and the pellet containing mainly cell membranes washed three times with ice-cold PBS containing 5 mM EDTA. The washed cell membrane pellet was finally resuspended in 2 ml lysis buffer composed of 5 mM sodium phosphate, 5 mM EDTA, 150 mM NaC1, 10 mM EACA and 0.5 mM PMSF, pH 7.5 and containing 1% nonidet P40.
After incubation for 60 min at 37C with vigorous shaking, the mixture was centrifuged for 20 min at 30 000 g and the supernatant dialysed against lysis buffer containing 0.1% NP40.
Purification of ClqR: Endothelial cell membrane lysates were loaded on a column of 4ml Sepharose-Clq equilibrated in PBS containing 0.5 mM PMSF, 5 mM EDTA and 0.1% NP40, pH 7.5. Clq was isolated from pooled human serum as described previously 2 and 3 mg of Clq was coupled to 1 ml of packed Sepharose. After vigorous washing bound ClqR was liberated from the column using 1 M NaC1. Protein content in the fractions was measured by the Lowry method and conductivity was assessed at 4C. ClqR was assayed in the fractions using a haemolytic assay. The fractions containing ClqR activity were pooled, dialysed against 5 mM PMSF and subjected to ion exchange chromatography on a 1.5 10cm QAE-A50 Sephadex column equilibrated in dialysis buffer. Bound activity was stripped from the column with dialysis buffer containing 0.65 M NaC1. ClqR activity was pooled, freeze dried, resuspended in 250/1 PBS containing 5 mM EDTA and 0.5 mM PMSF and after filtration on millipore 0.2 subjected to gel filtration on TSK 3000-HPLC. Fractions of 0.3 ml/min were collected and assessed for ClqR activity.
Assay for ClqR haemotic activity: Sheep erythrocytes (E) sensitized with optimal concentrations of rabbit IgG anti-E were prepared and incubated with a suboptimal concentration of Clq, and washed at 4C with GVB2+. To assay for ClqR haemolytic activity, tubes containing 1 107 EAClq in 100/1 DGVB + + were incubated with dilutions of fractions from columns for 30 min at 0C and thereafter 0.1 ml Clq deficient serum diluted 1/50 in DGVB ++ was added to each tube followed by incubation for another 60 min at 37C. Percent haemolysis was determined following addition of 1.5 ml 0.15 M NaC1 and centrifugation. Appropriate controls for reagent blank and input were included in each assay. The amount of Clq chosen to prepare EAClq was such that EAClq in Clq deficient serum caused approximately 60-70% lysis of the ceils.
Surface iodination of intact HUVEC or purified PMN 2 was performed with 1 107 cells in 1 ml PBS at 22C by addition of 0.5 mCi Na12I, 60 #1 lactoperoxidase (1 mg/ml), and three sequential additions (10/1 each) of H202 of 0.003% every 10-20 s. Thereafter the cells were washed three times by centrifugation and resuspension in PBS to remove free 12sI, and finally solubilized in lysis buffer.
SDS-polyacrylamide gel electrophoresis (SDS-PA GE) SDS-PAGE was performed using 7.5% polyacrylamide gels. 23 Samples were mixed with an equal volume of 0.2 M Tris, 2% SDS, pH 8.0 with and without 10 mM 2fl-mercaptoethanol and boiled for 5 min. Gels were stained with Coomassie brilliant blue, dried, and subjected to autoradiography using X-ray film.
Western blot analysis was performed as described previously. 24 Five and 10 #g samples of solubilized membranes were subjected to SDS-PAGE, blotted onto nitrocellulose, reacted with monoclonal antibodies or isotype controls, washed and bound antibodies reacted with biotinylated goat antimouse Ig, followed by incubation with streptavidin alkaline phosphatase (Zymed Laboratories Inc.) for 1 h, and detected with naphthol AS-MX phosphate (Sigma, St. Louis, MO) and Fast Red (Sigma) as substrate. Every incubation step was followed by a 15 min washing step in PBS-0.5% Tween 20.
Monoclonal antibodies: Spleen cells were obtained by standard techniques from Balb/c mice immunized with three weekly injections of 50/g quantities of ClqR emulsified in complete Freund's adjuvant. The spleen cells were fused with non-secreting SP 20 myeloma cells and the fused cells selected in hypoxanthine-aminopterin-thymidine medium.
Culture supernatants of cell lines were screened by ELISA for reactivity with purified ClqR from HUVEC. From each positive well, individual clones were prepared by adding 100/1 culture medium containing 3 cells/ml to microtitre plates.
In this way two positive clones were selected (D3 and D5) for further analysis. Ascitis was prepared in Balb/c mice following injection of 1 106 cells per mouse. Limited proteolytic digestion of mouse IgM was carried out by incubation of 1 mg/ml solutions of D5 or control MoAb in 20 mM Tris, 150 mM NaC1, 20 mM cad2, pH 8 with 150 #g/ml diphenyl carbamyl chloride (DPCC, Sigma) treated trypsin (Sigma) for 5 h at 37C. Mercaptoethanol was added to 10 mM and the solution incubated for 5min at 37C, followed by 15 #g/ml trypsin inhibitor (Sigma) to stop the reaction and further incubated for 5 min. Finally the mixture was made up to 50mM iodoacetamide and left at room temperature for 10 min. After dialysis 7S fragments of IgM were obtained by gel filtration on Sepharose 4B. The 7S fragments were dialysed against sodium acetate buffer pH 3.8 and treated with 1% pepsin (w/w) for 16 h at 37C and the F(ab')2 fragments recovered after gel filtration on Sephadex G150.
Binding studies: The binding of 12SI-C1q to EC was performed as described previously. 8 To determine the effect of D3 and D5 on binding of 12SI-C1q to EC, monolayers of EC in 48-well culture wells were incubated with 100 ng 2SI-Clq in RPMI-0.5% BSA alone or in the presence of increasing concentrations of purified F(ab')2 D3 or D5 monoclonal antibodies. As a control a nonspecific F(ab')2 from mouse IgM monoclonal antibody was used. After incubation for 2 h at 4C, the wells were washed and cell bound radioactivity assessed following lysis of the cells with 100 1 1N NaOH for 1 h.

Results
The fractionation of detergent solubilized endothelial cell membranes from HUVEC by aflqnity chromatography on Sepharose-Clq is shown in Fig. 1. Most of the protein was found in the fall-through fractions, while only a small amount of protein emerged from the column between 11 and 14 mS. When the fractions from the column were tested in dilutions of 1:10 for ClqR function all the inhibitory activity was found to be associated with the protein peak in the gradient. Fractions 148-162 were pooled, dialysed and fractionated further on a QAE-A50 Sephadex column (Fig. 2). Very little detectable protein was found in the fall-through fractions and more than 80% of ClqR functional activity could be eluted from the QAE column with a step gradient of NaC1. The major peak of ClqR activity was associated with the main protein peak. To obtain some insight into the size of ClqR, fractions 52-57 were pooled, freeze dried and subjected to fractionation by HPLC on a TSK 3000 column. ClqR activity, associated with the only detectable protein peak, emerged from the column with an apparent molecular weight of 60 kDa. The fractions containing peak ClqR activity were pooled, freeze dried and part of it labelled with 2sI and analysed by SDS-PAGE and autoradiography. Under non-reducing conditions only one major band was seen with an apparent molecular weight of 55-62 kDa. Under reducing conditions the molecular weight was between 64-68 kDa (Fig. 3).    ClqR was also isolated from the membranes of EAhy.926 using the same procedure as described above for HUVEC-C1 qR. Comparable results were obtained concerning the size and functional activity of ClqR. Purified ClqR isolated from either HUVEC or EAhy.96 were both able to inhibit lysis of EAClq in a dose-dependent manner (Table 1).
ClqR induces inhibition of lysis of EAClq in Clq deficient serum by binding to EAClq and presumably by preventing the interaction with Clr and Cls because EAClq preincubated with ClqR, washed and subsequently exposed to Clq deficient serum at 37C also exhibits inhibited lysis.
Monoclonal antibodies: Immunization of BALB/c mice with purified HUVEC-ClqR and fusion of spleen cells with Sp2/0 hybridoma cells yielded two monoclonal cell lines, D5 and E3, both secreting lgM. These two mAbs reacted only with ClqR and not with Clq or its fragments ( Table 2). Western blot analysis revealed reactivity of D5 with one maior molecule of approximately 60-64 kDa in membrane lysates of both HUVEC and PMN (Fig.  4). In some experiments an additional faint band was seen at 96-98 kDa, but this band was also seen sometimes with isotype control mAb. In addition some smaller molecular weight reaction products were seen around 45 kDa. Since the antigen used for immunization was purified by aflqnity chromatography over Clq-Sepharose we also tested whether D5 or E3 reacted with Clq, heat treated C1 q, collagenous fragments of Clq or heads of Clq. No significant reactivity of either mAb with Clq or its fragments was found by ELISA.
Eect of monoclonal antibodies against C lqR on CIqR mediated binding of 125I-C1q: To determine whether the binding of 12SI-C1q to HUVEC could be influenced by F(ab')2 fragments of D5, HUVEC monolayers of HUVEC were incubated with 1251-C1q in the absence or presence of increasing concentrations of mAb (Table 3). While 37.2% of 125I-Clq bound to HUVEC in medium alone, 20-200/g F(ab'). anti CIqR caused a dose-dependent inhibition of binding. Full inhibition of binding of 12SI-C1q was not observed in any of the three experiments performed.

Discussion
The present study extends the previous observations 8 that human umbilical vein endothelial ceils express a Clq receptor that has identity or is closely related to ClqR described on lymphocytes. 9'1'14 ClqR was isolated by affinity chromatography on Clq-Sepharose followed by further purification on QAE-A50 and TSK 3000-HPLC and detected during the isolation procedure using inhibition of lysis of EAClq in Clq deficient serum. The first step yielded material which was reasonably pure but minor contaminants were mainly removed in the QAE-A50 step. The purified ClqR filtered with an apparent molecular weight of 60 kDa on TSK-3000. Although the molecular weights for lymphocyte ClqR have been reported to be in the range from 56-70 kDa, 25 these differences probably reflect the different percentages of acrylamide used for SDS-PAGE. Monoclonal antibodies against the purified endothelial cell ClqR were raised. These monoclonal antibodies reacted with purified ClqR from HUVEC and from EAhy.926 cells. The purified HUVEC ClqR was also shown to react with a polyclonal antibody (kindly donated by Dr R. B. Sim, Oxford) raised against the B-cell ClqR. On the other hand it was found that while the D5 mAb reacted also with B-cells and polymorphonuclear leukocytes (PMN), the E3 mAb only reacts with EC and PMN and not with B-cells, suggesting differences in the epitopes of endothelial cell ClqR and B-cell ClqR. On the other hand NHg-terminal amino acid sequence analysis of the first fourteen amino acids did not show any differences with the recently reported sequence of ClqR. 25 Further studies are needed to elucidate these differences.
The D5 and E3 mAbs both recognized one major band in membrane lysates of both HUVEC and PMN. The size of ClqR from both these cell types was well within the reported range size of ClqR. 26 While the D5 mAb was able to inhibit binding of 125I-Clq to EC the E3 mAb was much weaker in this respect. The results described in Table 1 indicate that endothelial cell ClqR is able to interact with immune-complex-bound Clq and prevent lysis of EA-Clq in Clq deficient serum, suggesting that assembly of an intact C1 on EAClq is prevented.
This mechanism may be of importance in viva to regulate the degree of C1 activation in an early phase of the immune response. In addition during a primary immune response mainly IgM antibodies are generated. There are no known cellular 19S IgM receptors on human phagocytic cells, but by binding and activation of C1, and subsequent removal of activated Clr and Cls from the IgM-immune complex-bound Clq, these types of immune complexes may be trapped very rapidly on vascular endothelial cells via ClqR, which, in turn, may ingest these complexes, and prevent further systemic immune complex-mediated inflammatory responses.