Separation of decay-accelerating and cofactor functional activities of Kaposi's sarcoma-associated herpesvirus complement control protein using monoclonal antibodies

Abstract

Complement is an essential part of the innate immune system, which clears pathogens without requirement for previous exposure, although it also greatly enhances the efficacy and response of the cellular and humoral immune systems. Kaposi's sarcoma-associated herpesvirus (KSHV) is the most recently identified human herpesvirus and the likely aetiological agent of Kaposi's sarcoma, primary effusion lymphoma and multicentric Castleman's disease. We previously reported that the KSHV complement control protein (KCP) was expressed on infected cells and virions, and could inhibit complement through decay-accelerating activity (DAA) of the classical C3 convertase and cofactor activity (CFA) for factor I (FI)-mediated degradation of C4b and C3b, as well as acting as an attachment factor for binding to heparan sulphate on permissive cells. Here, we determined the ability of a panel of monoclonal anti-KCP antibodies to block KCP functions relative to their recognized epitopes, as determined through binding to recombinant KCP containing large (entire domain) or small (2–3 amino acid residue) alterations. One antibody recognizing complement control protein (CCP) domain 1 blocked heparin binding, DAA and C4b CFA, but was poor at blocking C3b CFA, while a second antibody recognizing CCP4 blocked C3b CFA and 80% DAA, but not C4b CFA or heparan sulphate binding. Two antibodies recognizing CCP2 and CCP3 were capable of blocking C3b and C4b CFA and heparan sulphate binding, but only one could inhibit DAA. These results show that, while KCP is a multifunctional protein, these activities do not completely overlap and can be isolated through incubation with monoclonal antibodies.

Introduction

The complement system is a key component of the innate immune system, which provides the first line of defence against invading pathogens. It is comprised of a number of serum pro-enzymes and proteins that interact in an amplification cascade (reviewed by Walport¹). Complement activation can occur through specific antibody–pathogen binding (classical pathway), through lack of inhibition by surfaces of pathogens (alternative pathway) or through recognition of foreign microbial carbohydrate configurations (lectin pathway). Central to complement activation is the cleavage of pro-enzymes to enable the formation of the C3 and C5 convertase enzymatic complexes, with the release of smaller chemoattractant and anaphylatoxin fragments. The covalent attachment of C4b and C3b to pathogen and infected cell surfaces also enhances recognition by phagocytes and dramatically increases the humoral response to those pathogens.^2,3 In order to protect host cells from autologous complement attack, a number of soluble and membrane-bound complement regulators have evolved to limit inflammation to the infected site.

An important group of complement regulators are encoded in the regulators of complement activation (RCA) gene cluster on chromosome 1 (locus 1q32). All of these proteins, including membrane cofactor protein (MCP; CD46); complement receptor 1 (CR1; CD35); decay-accelerating factor (DAF; CD55); factor H (FH) and C4b binding protein (C4 BP), contain 4–35 complement control protein (CCP) domains and share significant homology as well as complement inhibition mechanisms.⁴ CCP domains consist of approximately 60 amino acids and contain conserved proline and hydrophobic residues organized into a compact hydrophobic core surrounded by short β-strands.⁵ This structure is stabilized by four invariant cysteine residues that form two pairs of disulphide bonds in a 1–3 and 2–4 manner.

RCA proteins regulate C3 or C5 convertases via two mechanisms. Firstly, they may accelerate the decay of convertase enzyme complexes through dissociation of C2a from C4b or dissociation of factor Bb from C3b. Secondly, they may act as cofactors to induce or enhance the cleavage of either C3b or C4b to inactive fragments, incapable of forming convertases, through recruitment of the serine protease FI. However, the precise mechanism of action for these events, including binding specificity, location of binding on C3b or C4b, and requirement for conformational changes in proteins, is still under investigation.

As complement does not require previous exposure to be fully effective and has been shown to accelerate and enhance the adaptive immune responses to pathogens, viruses have developed a broad range of complement evasion strategies which we are just beginning to discover. For example, not only does human cytomegalovirus up-regulate the expression of host RCA complement regulators on infected cells, but progeny virions acquire these cell surface-expressed host regulators upon egress.^6–8,8,9 Human immunodeficiency virus, human lymphotropic virus 1 and vaccinia virus (VV) have also been shown to incorporate host regulators into their envelope during virus egress.^8,9,9–13 However, to date only two virus families, Herpesviridae and Poxviridae, have been found to encode their own complement inhibitors (reviewed by Blue et al.¹⁴).

Kaposi's sarcoma-associated herpesvirus (KSHV) is a member of the gamma-2 (rhadinovirus) genus of the gammaherpesvirus subfamily, and it is the most recently discovered oncogenic virus in humans.¹⁵ It is widely regarded as the cause of three types of neoplasia: Kaposi's sarcoma, multicentric Castleman's disease and primary effusion lymphoma (reviewed by Schulz¹³). Like other members of the herpesvirus family, KSHV encodes a number of proteins that share significant homology with host proteins,¹⁶ many of which are immunomodulatory (reviewed by Rezaee et al.¹⁷). One such protein, which is homologous to human host RCA proteins, is encoded by open reading frame 4. We have previously shown that the protein encoded by this gene is expressed following induction of the lytic cycle of infection in KSHV-infected lymphoma cells, endothelial cells and virions,^18,19 and that this protein is a potent regulator of the classical complement pathway in vitro, as well as enhancing viral infection through binding to cell surface heparan sulphate.^18,20

The KSHV complement control protein, or KCP, consists of four N-terminal CCP domains, a di-cysteine motif, a long serine/threonine (S/T)-rich domain and a C-terminal hydrophobic region sufficient to act as a transmembrane (TM) region.^3,19,20 The full protein is 550 amino acids in length, but KSHV-infected cells also express two alternatively spliced forms of KCP in which either the S/T region or the S/T region plus the di-cysteine region is removed. All three isoforms retain the CCP and TM regions and retain complement regulatory activity.¹⁹

The complement-binding and heparin-binding regions of the viral complement regulator herpes simplex type 1 glycoprotein C (the first described viral complement inhibitor) have been investigated, both in vitro and in vivo;^21–23 however, this protein does not contain CCP domains. Recently, we and other investigators have dissected the functional regions of KCP following the deletion of single or multiple domains (or following exchange of CCP domains with non-inhibitor CD21),^18,24 an approach that has also been employed to investigate the functional regions of the vaccinia virus complement control protein (VCP).¹² VCP is also the only viral complement inhibitor to date for which monoclonal antibodies have been used to map the functional regions.²⁵

Here we investigate the ability of a panel of monoclonal anti-KCP antibodies to block decay-accelerating activity (DAA), to block the ability to mediate C3b or C4b cleavage by factor I (FI) [cofactor activity (CFA)], or to block KCP binding to heparan sulphate. These studies expand upon our recent use of site-directed mutagenesis to validate a structural model of KCP²⁶ and allow a direct comparison with the structural requirements for the function of homologous inhibitors encoded by the poxvirus family and the human host.

Materials and methods

Cell culture

CHO cells, obtained from the European Collection of Animal Cell Cultures (ECACC; Salisbury, UK), were maintained in RPMI 1640 medium, supplemented with 5% fetal bovine serum, 1%l-glutamine and 1% non-essential amino acids. CHO cells engineered to stably express recombinant and wild-type forms of KCP at the cell surface or secreted as hybrid forms fused to the human immunoglobulin G1 (IgG1) Fc region have been previously described.^18,20 Stably transfected CHO cells expressing KCP were selected with cell medium containing 400 µg/ml hygromycin B and were cloned to homogeneity (high expression) by limiting dilution and screening for expression.

Recombinant forms of KCP

CHO cells expressing full-length wild-type KCP at the cell surface¹⁹ were used for primary screening of antibody-producing hybridoma cells, and truncated, cell surface-expressed forms of KCP were also engineered. Recombinant forms of KCP either missing the fourth, or third and fourth, C-terminal CCP domains were created by designing polymerase chain reaction (PCR) primers that inserted a NotI restriction enzyme site into the hinge region between CCP domains, followed by subcloning of the cDNA into an expression vector that adds (in-frame) the minimum required signal for glycophosphoinositol (GPI) anchor addition.¹⁹ A recombinant form of KCP that lacks the N-terminal CCP1 domain was created by designing primers that added the GPI signal two amino acid residues after CCP4 and replacing the wild-type signal sequence and CCP1 domain with the signal sequence from CD33 (SigPigPlus vector; R & D Systems, Abingdon, UK).

Soluble recombinant forms of KCP expressed as Fc fusion proteins were also used to map monoclonal antibody binding sites including KCP CCP1–4 or 2–4 domains or KCP CCP1–4 domains where individual CCP domains were exchanged with equivalent domains from CD21 (previously described in Spiller et al.¹⁸). Further definition of monoclonal antibody binding sites was also achieved using KCP-Fc fusion proteins differing from the wild-type CCP1–4 sequence by two or three amino acid residues (point mutations for structural mapping); the construction, expression and purification of these proteins have been described elsewhere.²⁶ DNA sequencing confirmed the integrity of all recombinant forms, and Fc fusion proteins were purified with a hi-trap protein A-sepharose column (GE Healthcare, Amersham, UK) according to the manufacturer's instructions, and standardized to 1 mg/ml stocks in phosphate-buffered saline (PBS).

Generation of specific anti-KCP monoclonal antibodies

The monoclonal anti-KCP antibodies, BS-B6, -E7, -F8, -H10 and -Jll, were generated in-house (Cardiff University). Briefly, wild-type KCP-Fc was mixed with Freund's incomplete adjuvant and used to immunize BALB/c mice via intraperitoneal inoculation. After four boost inoculations, mouse spleens were removed and used to create hybridoma cells using SP2/0 fusion partners (ECACC) by standard techniques, and positive splenocyte–myeloma fused cells were selected by flow cytometry. Specific anti-KCP monoclonal antibody-secreting cells were identified by binding to CHO cells expressing full-length human KCP,¹⁹ as detected by phycoerythrin (PE)-conjugated rabbit anti-mouse immunoglobulin antibody (Dako UK Ltd., Cambridge, UK). All antibodies were of the murine IgG1 isotype as determined using the Isostrip typing kit (Roche Applied Science, Burgess Hill, UK). Large quantities of each monoclonal antibody were made by the ascites production method (purchased as a service from Eurogentec, Seraing, Belgium), purified by high-performance liquid chromatography (HPLC) using a hi-trap protein G-sepharose column (GE Healthcare) as per the manufacturer's instructions, and standardized to 4 mg/ml stocks.

Surface plasmon resonance (SPR) analysis (Biacore)

The affinity of each monoclonal antibody to bind to immobilized KCP CCP1–4 domains was determined using surface plasmon resonance (Biacore 2000; Biacore International AB, Uppsala, Sweden). To eliminate analytical complications arising from co-operative binding of dimeric Fc fusion forms of KCP, recombinant KCP was released by digestion with factor Xa, as a specific enzyme recognition site had been engineered between KCP and the Fc fusion protein (R & D Systems). The optimal cleavage conditions were determined to be 12 hr at room temperature, using 0·5 units of factor Xa per µg of KCP. Each of four flow cells on a CM5 sensor chip were activated with 20 µl of 0·2 m 1-ethyl-3-(3 dimethylaminopropyl) carbodiimide with 0·05 m N-hydroxy-sulfosuccinimide at a flow rate of 5 µl/min, and then monomeric recombinant KCP (0·01 mg/ml in 10 mm Na-acetate buffer, pH 4·5) was injected over flow cells to reach 2000 resonance units (RU). In all cases, unreacted groups were blocked with 20 µl of 1 m ethanolamine, pH 8·5. A negative control was prepared by activating and subsequently blocking the surface of flow cell 1 without the addition of monomeric KCP.

The association kinetics were studied for a range of monoclonal antibody concentrations using the standard flow buffer (10 mm Hepes-KOH, pH 7·4, supplemented with 150 mm NaCl and 0·005% Tween 20). Protein solutions were injected for 300 seconds to achieve saturation during the association phase at a constant flow rate of 30 µl/min. The sample was injected first over the negative control surface and then over immobilized KCP flow cells and analysed for a dissociation phase of 200 seconds at the same flow rate. Signals were normalized by subtracting the non-specific signal measured by control flow cell 1. Between each different concentration of analyte tested, the flow cell surfaces were regenerated with a 30-µl injection of 2 m NaCl with 100 mm HCl. All sensograms were analysed using the biaevaluation 3·0 software (Biacore) to calculate affinity constants.

Flow cytometry analysis

Confluent layers of CHO cells expressing recombinant cell-surface forms of KCP (or control cells transfected with empty vector) were disaggregated by incubating with flow cytometry buffer [FCB: PBS containing 1% bovine serum albumin (BSA), 15 mm ethylenediaminetetraacetic acid (EDTA) and 30 mm NaN₃; pH 7·35] at room temperature for 15 min. All cell suspensions were standardized to 1 million cells per ml and 10⁵ cells were incubated with varying concentrations (0·4–40 µg/ml) of monoclonal (or isotype-matched control) antibody for 30 min on ice. Unbound antibody was removed by centrifugation (2 min at 1000 g; 4°), followed by resuspension in FCB (repeated three times), and bound antibody was detected with 1/100 PE-conjugated goat anti-mouse immunoglobulin antibody (Dako) diluted in FCB. Each antibody concentration was analysed in triplicate for each test cell line and mean cellular fluorescence was measured with a Becton-Dickinson FACScalibur (BD Biosciences, Oxford, UK). Results were confirmed by at least one replication of the experiments.

Enzyme-linked immunosorbent assay (ELISA) analysis

Epitope mapping was also determined by monoclonal antibody binding to immobilized recombinant forms of KCP-Fc. Recombinant fusion proteins having all four KCP CCP domains (with or without altered clusters of residues), truncation of CCP1, or exchange of single CCP domains with CD21 were immobilized on Nunc 96-well Maxi-sorp plates (Nunc, Roskilde, Denmark) at a concentration of 10 µg per well using carbonate-based coupling buffer (0·1 m, pH 8·5). Non-specific binding was blocked by preincubation of coated plates with PBS containing 0·05% Tween 20, 5% dehydrated milk powder and 1% BSA. Doubling dilutions of monoclonal antibodies (diluted in blocking solution) were incubated with coated plates (in triplicate) at 37° for 1 hr, and then unbound antibody was removed by washing with PBS containing 0·05% Tween 20, before bound antibody was detected with 1/1000 peroxidase-conjugated donkey anti-mouse immunoglobulin antibody (minimum cross-reaction; Jackson Immunoresearch Laboratories, West Grove, PA) diluted in blocking buffer. Peroxidase was detected using dissolved OPD tablets (Dako) following the manufacturer's instructions and measuring the absorbance of the plates at 490 nm using an MRX revelation plate reader (Dynex Magellan Biosciences, Chantilly, VA). Results were confirmed by at least one replication of the experiments.

The inhibition of KCP-heparan sulphate binding by monoclonal antibodies was also determined by ELISA. Ninety-six-well plates were coated with 10 µg/ml streptavidin (Invitrogen, Paisley, UK) and kept overnight at 4°, followed by incubation with blocking solution (0·05 m phosphate, 75 mm NaCl and 3% fish gelatin) for 1 hr at room temperature and then incubation with 10 µg/ml biotinylated heparan sulphate (Sigma-Aldrich, Gillingham, UK) for 1 hr at room temperature. KCP-Fc (100 µg/ml; 0·52 µm) was preincubated with monoclonal antibody at ratios of 1 : 20, 1 : 10, 1 : 5 and 1 : 1 (antibody:heparin-binding sites; assuming two heparin binding sites per KCP monomer) for 1 hr at room temperature prior to addition to the heparan sulphate-coated 96-well plates for a further 1 hr. Binding of KCP-Fc was detected using peroxidase-conjugated goat anti-human Fc antibody diluted in fish-gelatin blocking buffer (1/1000) and peroxidase activity was detected as described above. Inhibition of heparan sulphate binding by monoclonal antibodies was observed as decreasing optical density at 490 nm relative to KCP-Fc binding in the absence of preincubation with monoclonal antibodies. Each point was determined in triplicate and results were confirmed by at least one replicate of the experiment on a different day.

Proteins

C1,²⁷ C2,²⁸ C3 and C4,²⁹ and FI³⁰ were purified from plasma as described previously. C3b and C4b were purchased from Complement Technology Inc. (Tyler, TX). C1 and C2 were functionally pure as they were devoid of other complement factors. FI, C4, C3 and C3b were at least 95% pure, as determined by Coomassie staining of proteins separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE). All purified proteins were stored in aliquots at −80°. Protein concentrations were determined by measuring absorbance at 280 nm or from amino acid analysis following 24 hr of hydrolysis in 6 m HCl. C3b and C4b (both from Complement Technology Inc.) were labelled with ¹²⁵I using the chloramine T method,³¹ and the specific activity was determined to be 0·4–0·5 MBq/µg of protein. C4 and C3 were treated with methylamine to hydrolyse the internal thioester bond, thus changing the conformation to C4met and C3met, respectively. Proteins were prepared by incubation with 100 mm methylamine, pH 8–8·5, at 37° for 1 hr, followed by dialysis in 50 mm Tris-HCl, pH 7·5, and 150 mm NaCl or 10 mm Hepes-KOH, pH 7·4, supplemented with 75 mm NaCl.

FI cofactor activity (CFA) assay

Wild-type KCP-Fc (0·3 µm) was preincubated for 1 hr with 40, 4 or 0·4 µg of each monoclonal antibody (representing molar ratios of 10 : 1, 1 : 1, and 1 : 10 relative to the number of copies of KCP) at room temperature. After the incubation period, KCP-Fc–antibody complexes were incubated with C3b (75 µg/ml) or C4b (25 µg/ml), FI (60 nm), trace amounts of I^125-labelled C3b or C4b and buffer (50 mm Tris-HCl and 150 mm NaCl, pH 7·4) to a final volume of 50 µl, at 37° for 2 hr. The assay was stopped by addition of SDS loading buffer containing the reducing agent dithiothreitol (DTT), and proteins were separated on a 10–15% SDS-PAGE gradient gel. Radiolabelled proteins were then visualized by autoradiography using a phosphoimager (Molecular Dynamics, Sunnyvale, CA) and bands corresponding to C3b and C4b cleavage fragments were quantified by densitometry analysis. Inhibition of CFA by monoclonal antibodies was observed as decreased presence of C4d or C3b fragments, relative to levels seen with 0·3 µm KCP-Fc in the absence of preincubation with antibody. Experiments were repeated at least three times and the mean of densitometry results (relative to no addition of monoclonal antibody) was used to determine mean and standard deviation for inhibition.

Decay-accelerating activity (DAA) for the classical pathway C3 convertase

Generation of the sheep erythrocyte targets that are coated with the classical C3 convertase has been described in detail elsewhere.²⁰ Briefly, sheep erythrocytes are sensitized with a specific rabbit polyclonal antibody, and purified complement components C1 and C4 are then sequentially incubated with these cells, with excess unbound proteins removed by intervening wash steps. Purified C2 is then added to complete the convertase, followed immediately by addition of dilution series of test and control proteins, and incubation at 30° with constant shaking for 5 min. Guinea pig serum is then added (1/50 dilution in 40 mm EDTA veronal-buffered saline) as a source of terminal complement components, and lysis of target cells (measured as released haemoglobin at A405 nm) is directly proportional to the number of remaining C3 convertase sites. Soluble recombinant DAF-Fc²⁰ was used as a positive control in these experiments and all results for inhibitors were compared with the amount of lysis observed in the absence of added inhibitors to determine the degree of decay acceleration.

Inhibition of DAA with monoclonal antibodies

Wild-type KCP-Fc (0·3 µm in 100 µl) was preincubated for 1 hr with 40, 8, 4, 0·4 or 0·04 µg of each monoclonal antibody at room temperature. The amount of wild-type KCP-Fc used in these assays was empirically determined to inhibit 75% of the erythrocyte lysis in the above assay. After the preincubation period, complexes were added to target cells following addition of C2, and prior to addition of diluted guinea pig serum (as described above). Data for all points were determined in triplicate and experiments were repeated at least once for confirmation.

Results

Generation of mouse monoclonal anti-KCP antibodies

BALB/c mice were repeatedly immunized with wild-type KCP-Fc (CCP1–4 domains fused to the human IgG1 Fc region), prior to removal of their spleens, mechanical release of lymphocytes and fusion with immortal SP2/0 mouse B-lymphocyte fusion partner cells. Positive clones were selected by flow cytometry using CHO cells expressing membrane-bound, wild-type KCP (to avoid anti-human IgG1-reactive clones). Positive hybridoma cells were cloned out to homogeneity through three consecutive limiting dilution procedures (0·3 cells per well) by flow cytometry (see ‘Materials and methods’). The affinity of these antibodies for KCP was determined by SPR analysis (Table 1) and the disassociation constants for five selected monoclonal antibodies ranged from 7·3 × 10⁻⁸ m (J11; lowest affinity) to 1·36 × 10⁻⁹ m (H10; highest affinity).

Table 1.

SPR analysis of monoclonal antibody binding to monomeric Kaposi's sarcoma-associated herpesvirus complement control protein (KCP)

Monoclonal antibody	kassociation	kdissociation	kD
B6	3·72 × 104	7·53 × 10−4	2·55 × 10−8
E7	1·03 × 105	1·94 × 10−3	2·1 × 10−8
F8	7·65 × 103	7·7 × 10−4	1·0 × 10−8
H10	2·26 × 105	2·85 × 10−4	1·36 × 10−9
J11	2·25 × 104	1·62 × 10−3	7·3 × 10−8

All values are in molarity. Values represent the average of binding in the presence of physiological concentrations of NaCl (150 mm) to monomeric KCP immobilized to three separate CM5 chips. k_association, association rate constant; k_dissociation, dissociation rate constant; k_D, dissociation equilibrium constant.

Mapping anti-KCP binding epitopes

To map the location of antibodies to particular CCP domains, the concentration required to achieve binding saturation for flow cytometry analysis was determined for CHO cells expressing full-length KCP or truncated GPI-anchored recombinant forms containing CCP domains 1–3, 1–2, or 2–4 (Fig. 1). Antibody BS-F8 has previously been investigated and served as our control.¹⁸ Saturated binding to full-length KCP was found at approximately 1 µg/ml for all antibodies, except J11, which required 8 µg/ml, and B6, which did not reach saturation even when added at 40 µg/ml (despite having similar affinity constants by SPR analysis; Table 1). Only the binding of F8 was affected by the removal of CCP1 (although it was not completely abrogated), while the majority of B6 binding and all H10 binding was dependent on the presence of CCP4. Only F8 was able to bind a recombinant KCP form missing CCP3 and CCP4 (Fig. 1d; all the remaining symbols at the bottom of the figure overlap).

Figure 1.

Binding of monoclonal anti-Kaposi's sarcoma-associated herpesvirus complement control protein (KCP) antibodies to wild-type (full) KCP (a), or recombinant cell surface glycophosphoinositol (GPI)-anchored forms lacking complement control protein 1 (CCP1) (b: GPI-CCP234), CCP4 (c: GPI-CCP123), or CCP3 and CCP4 (d: GPI-CCP12) as determined by flow cytometry analysis. Average mean cellular fluorescence values are shown following incubation of cells with increasing concentrations of each antibody (the standard deviations for samples measured in triplicate are also shown).

As the recombinant GPI-anchored forms may be influenced by steric hindrance of their proximity to the cell surface when expressed as truncated, GPI-anchored forms, we also assessed antibody binding to soluble recombinant forms. Soluble KCP included all four CCP domains, CCP2–4, or forms in which individual CCP domains had been exchanged with CCP domains from CD21 (previously used to determine individual CCP contribution to function¹⁸). Maximum binding was observed when 50 ng/ml of all antibodies was incubated with immobilized KCP CCP1–4, except for J11, which required 5 µg/ml for maximum binding (Fig. 2), and B6, which only gave low readings even at 50 µg/ml (data not shown). Binding of F8 appeared to be split between CCP1 and 2 (dependence on CCP2 > CCP1), H10 recognized an epitope confined to CCP4, and E7 binding required both CCP2 and CCP3. While removal of CCP1 did not affect J11 binding, exchange of CCP4 was found to enhance binding, but binding was reduced with the removal of CCP2 or CCP3 (CCP3 > CCP2). These findings are consistent with those obtained by flow cytometry (Fig. 1).

Figure 2.

Binding of monoclonal anti-Kaposi's sarcoma-associated herpesvirus complement control protein (KCP) antibodies E7 (a), F8 (b), H10 (c) and J11 (d) to recombinant soluble KCP-Fc immobilized to 96-well plates, as detected by enzyme-linked immunosorbent assay (ELISA) methods. Plates were coated with recombinant KCP containing all four CCP domains (KCP) or lacking the first CCP domain (Del1), or with recombinant forms in which KCP CCP domain 2, 3 or 4 was replaced (swap2, etc.) with an irrelevant matched CCP domain from CD21. Bound monoclonal antibody was detected by peroxidase-conjugated specific secondary antibody activity (420 nm), following incubation of coated plates with increasing concentrations of each monoclonal antibody. Each point was determined in duplicate, and the experiment was replicated on at least three separate occasions.

In an attempt to map antibody binding to clusters of amino acids on the surface of KCP, we measured antibody binding by ELISA using recombinant forms of KCP CCP1–4 with two or three amino acids altered by site-directed mutagenesis (previously used in molecular modelling studies).²⁶Table 2 shows the summary of these findings; binding of B6 was too low to give reliable results again and has been excluded from the table. H10 bound all recombinant forms, although mutations of charged residues in CCP4 are under-represented in this cohort. As previously reported, F8 bound two groups of residues that make up a positively charged groove that starts in CCP1 and continues into CCP2.¹⁸ E7 and J11 were both reliant on positively charged residues at the hinge region between CCP2 and CCP3 (K131, K133 and H135), although J11 additionally required negatively charged E99 (CCP2), E152 (CCP3) and D155 (CCP3) for maximum binding, indicating that their binding sites do not completely overlap. Interestingly, neither E7 nor J11 was influenced by the loss of positively charged R136 and K138, given their proximity to K131/K133/H135.

Table 2.

Monoclonal anti-Kaposi's sarcoma-associated herpesvirus complement control protein (KCP) antibody binding to KCP containing point mutations

Point mutation	CCP location	E7	F8	H10	J11
None (wild type)	−	+	+	+	+
D21A/D23A/E28A	1	+	+	+	+
R20Q/R33Q/R35Q	1	+	0	+	+
K64Q/K65Q/K88Q	1–2 hinge	+	0	+	+
M113A/M120A	2	+	+	+	+
K131Q/K133Q/H135Q	2–3 hinge	0	+	+	0
R136Q/K138Q	3	+	+	+	+
E99Q/E152Q/D155Q	2 (E99) and 3	+	+	+	0
H158A/H171A/H213A	3 and 4 (H213)	+	+	+	+
F195A/F207A/L209A	4	+	+	+	+

Binding is shown as + for comparable binding to wild-type KCP or 0 for no detectable binding. All binding studies were performed in triplicate and repeated in a minimum of two separate experiments. Locations of the point mutations are indicated based on molecular modelling.²⁶ Monoclonal B6 bound poorly to wild-type KCP, making comparisons difficult; therefore, results have not been included in this table.

CCP, complement control protein.

Antibody blocking of heparan sulphate binding

We previously found that F8 can inhibit virus binding and that this is probably a result of interference with KCP binding to heparan sulphate on the cell surface. Therefore, we tested the ability of each monoclonal antibody to block KCP-Fc binding of immobilized heparan sulphate, following preincubation (Fig. 3). F8 had the highest ability to block heparin binding, while H10 had very little or no heparin-binding blocking activity. Interference of E7 and J11 with KCP–heparan sulphate binding appeared intermediate, with E7 being slightly better than J11 over all experiments. These data suggest that, the closer to the N-terminus of KCP antibodies bind, the better they are at inhibiting heparin binding.

Figure 3.

Inhibition of Kaposi's sarcoma-associated herpesvirus complement control protein (KCP)-Fc binding to immobilized heparan sulphate following preincubation with monoclonal anti-KCP antibodies. One representative experiment is shown; each point was determined in triplicate. Binding of wild-type KCP-Fc to heparan sulphate was detected with a peroxidase-conjugated secondary antibody specific for the human Immunoglobulin G1 (IgG1) portion of KCP-Fc, and results are shown as the extent of binding relative to KCP-Fc binding in the absence of preincubation with monoclonal antibodies. The antibody:KCP ratio represents the ratio of antibody relative to each molecule of KCP, taking into account that KCP-Fc is dimeric (i.e. the 1 : 1 ratio is two molecules of monoclonal antibody for each dimeric KCP-Fc molecule).

Antibody inhibition of classical pathway C3 convertase DAA

KCP regulates complement through two separate mechanisms: DAA and FI CFA. Previously, we found that KCP accelerates the dissociation of C2a from C4b, resulting in abrogation of lytic pore formation (C5b6–9).²⁰ We investigated whether preincubation of soluble KCP with antibodies differentially inhibited the ability of soluble KCP-Fc to protect antibody-sensitized sheep erythrocytes from lysis (Fig. 4). Both E7 and F8 neutralized the DAA of KCP, resulting in complete lysis of the erythrocytes at a molar ratio of 1 : 1. H10 partially blocked KCP DAA, reaching maximum inhibition at a 2-fold ratio to KCP, J11 was poor at inhibiting KCP and could only block 60% of DAA at a 10 : 1 ratio, and B6 showed no significant inhibitory activity. As we previously found that the ability of KCP to accelerate the decay of the alternative pathway C3 convertase (C3bBb) was negligible,²⁰ we did not investigate antibody inhibition of DAA for this pathway.

Figure 4.

Inhibition of classical C3 convertase decay acceleration by preincubation of wild-type Kaposi's sarcoma-associated herpesvirus complement control protein (KCP)-Fc with monoclonal antibodies. Results of inhibition are shown as the relative decrease in the ability of KCP-Fc to protect target cells from lysis [i.e. KCP inhibition in the absence of antibody = 0%, and complete loss of KCP-mediated decay-accelerating activity (DAA) = 100%, resulting in maximal lysis of target cells], when KCP-Fc was preincubated with increasing molar ratios of monoclonal antibodies (1 : 10 to 10 : 1). Monoclonal antibodies included B6 (grey triangle), J11 (grey square), H10 (black circle), F8 (black triangle), and E7 (black square). All assays were performed in triplicate and the analysis was repeated at least twice. Error bars represent standard deviation.

Antibody inhibition of FI-cofactor-mediated cleavage of C3b and C4b

To determine the extent of antibody-mediated inhibition of FI-cofactor cleavage of C3b and C4b, antibody blocking functional studies were performed. Soluble KCP was preincubated with a 10-, 1- or 1/10-fold concentration of each antibody prior to addition to radiolabelled C3b or C4b and FI. In the absence of antibody (lane labelled KCP + FI), KCP cleaved the alpha chain of C3b releasing two prominent bands of 46 and 43 kDa [iC3b(1) and (2), respectively, in Fig. 5a]. Pre-incubation with B6, at any concentration, did not inhibit KCP CFA. However, E7, H10 and J11 significantly inhibited production of the 46- and 43-kDa C3b-cleavage products at 10 : 1 antibody to KCP ratios, while F8 was markedly less effective (Fig. 5b). In contrast, E7, F8 and J11 significantly inhibited C4b cleavage (as measured by the production of the 46-kDa C3d fragment), while the ability of H10 to block C4b cleavage was negligible. Similar results were observed for inhibition of cleavage of radiolabelled C4b, with the exception that H10 did not significantly inhibit the release of the 46-kDa C4d fragment (Figs 5a and b).

Figure 5.

Representative autoradiographs showing generation of (a) C3b cleavage products (46 and 43 kDa) or (b) C4b cleavage product (C4d: 46 kDa), in the absence of factor I (no FI), in the presence of Kaposi's sarcoma-associated herpesvirus complement control protein (KCP) and FI (KCP + FI), or in the presence of FI and KCP following preincubation with a 10-fold molar excess of anti-KCP monoclonal antibody (B6, E7, F8, H10 or J11). (c) Summary of inhibition of KCP C3b or C4b cofactor activity (CFA) as determined by relative decrease of iC3b (46 kDa species) or C4d autoradiograph band (determined by densitometry) following preincubation of each monoclonal antibody (i.e. KCP inhibition in the absence of antibody = 0%, and complete loss of KCP-mediated CFA = 100%, resulting in no production of degradation bands). The graph represents the average decrease from all performed experiments (repeated at least three times) for a 10-fold molar excess of antibody compared with KCP-Fc, and standard deviation.

Discussion

The use of blocking monoclonal antibodies to define functional regions of host complement inhibitors is a well-established method of determining structural requirements for function without the risk of artefacts being induced through large- or small-scale mutations in recombinant proteins. Coyne et al.³² combined panels of monoclonal antibodies from four separate laboratories to define the functional regions of CD55, and similar approaches have been used to map functional regions of CD46,³³ FH,^34–36 CD35³⁷ and CD59.³⁸ Monoclonal antibodies have also been used to define the regions of host complement inhibitors that bind to pathogens (e.g. CD46 binding to measles virus).^39,40 Studies using blocking monoclonal antibodies to map functional sites of viral complement inhibitors are not as common. Monoclonal antibodies have been used to map heparan sulphate binding in herpes simplex virus glycoprotein C,^41,42 as the ability of these antibodies to inhibit C3b binding has also been examined.^43,44 VCP is the only virally encoded CCP-containing complement inhibitor to be investigated in this manner. Isaacs et al.²⁵ found that anti-CCP1 antibodies did not block VCP complement inhibition, whereas function-blocking antibodies recognized CCP2, CCP4, or the junction between CCP3 and CCP4. However, they did not test the ability of the antibodies to block CFA independently from DAA, nor has anyone investigated the ability of monoclonal antibodies to block VCP binding to heparan sulphate.

We previously showed that the complement-regulating function of KCP was localized to the four CCP domains,²⁰ and here we have further defined the required regions of KCP using a panel of monoclonal antibodies to inhibit various mechanisms of KCP function. We have found that KCP is expressed on KSHV virions,¹⁸ as well as KSHV-infected cells.^18,20 Our data suggest that KCP functions to aid entry of virus into target cells through binding of heparan sulphate,⁴⁵ as well as to inhibit complement activation. Examining a panel of five anti-KCP antibodies, we found that heparan sulphate-binding, decay-accelerating and cofactor activities were differentially blocked by different monoclonal antibodies, indicating that these functions do not completely overlap (Fig. 6).

Figure 6.

Schematic summary of the mapped epitopes for antibody binding relative to functional regions of Kaposi's sarcoma-associated herpesvirus complement control protein (KCP) (including functional regions previously identified by domain deletion and site-directed mutagenesis^18,26,45). The N-terminal part of KCP containing the four complement control protein (CCP) domains (numbering to the right) is shown, and the dashed line at the bottom of the figure indicates the alternatively spliced serine/threonine-rich region leading to the conserved transmembrane region. Binding sites for the monoclonal antibodies are shown to the left (B6 not shown but is suspected to bind a partially hidden epitope in CCP3). CCP domains required for heparin, C3b and C4b binding, and C3b and C4b cofactor activity (CFA), as well as classical C3 convertase decay-accelerating activity (DAA), are shown to the right. Predicted N-glycosylation sites on the CCP domains are also indicated (octagons).

In general, the closer to the N-terminus of KCP an antibody bound, the more effective it was at blocking heparan sulphate binding. This makes sense from an evolutionary perspective, as the target cell binding sites on proteins are most effective when positioned at the furthest protruding point from the virion. This observation confirms site-directed mutagenesis studies performed based on molecular modelling of KCP, which found that replacing positively charged arginine residues 20, 33 and 35 or lysine residues 64, 65 and 88 with uncharged, polar glutamine residues (R20Q/R33Q/R35Q or K64Q/K65Q/H88Q) resulted in loss of heparin binding by KCP. These residues are located in CCP1 and the CCP1–2 hinge region, respectively, and it is interesting to note that F8 (which blocks heparan sulphate binding most effectively) was also unable to bind these mutated forms of KCP. Monoclonal antibodies E7 and J11 had an intermediate ability to block heparan sulphate binding, which may reflect a steric hindrance effect or may reflect the ability of heparan sulphate to additionally bind other positively charged regions of KCP. The latter possibility is unlikely as heparin binding was previously found to be unaltered for K131Q/K133Q/H135Q,⁴⁵ which is an area of positive charge bound by both Jll and E7 (Table 2).

KCP inhibits complement through two mechanisms: DAA and CFA. Monoclonal antibodies differed in their ability to block KCP inhibition of complement activation through these two different mechanisms. KCP DAA was blocked most efficiently by E7 and F8 (Fig. 4), which fits reasonably well with previous domain deletion and site-directed mutagenesis data: DAA requires CCP1–3 and the site-directed mutants that lose binding to E7 (K131Q/K133Q/H135Q) and F8 (R20Q/R33Q/R35Q or K64Q/K65Q/K88Q) show a 10-fold decrease and complete loss of DAA, respectively. In contrast, H10 also inhibited DAA to a lesser degree and, although the exact location of epitope binding could not be more precisely mapped with the available site-directed mutants, we have previously found that the entire CCP4 domain (which contains the H10 binding site) could be removed without detriment to the DAA.¹⁸ The inability of J11 to efficiently block DAA (Fig. 4) contrasts with the loss of J11 binding to site-directed KCP mutants K131Q/K133Q/H135Q and E99Q/E152Q/D155Q, as E7 also failed to bind the former mutant and both mutations resulted in significant to complete loss of KCP DAA. Comparison of the epitope mapping of J11 with that of E7 using exchanges in CCP domains between KCP and CD21 (Fig. 1) and binding to point mutants of KCP (Table 2) suggests that J11 has a higher reliance on CCP3 than CCP2 for binding relative to E7. However, this finding does not explain the inconsistency between J11 blocking of DAA compared with the site-directed mutants of KCP, but might suggest that even small alterations to clusters of amino acid residues may dramatically alter the overall conformation and function of a molecule. The ability of monoclonal anti-KCP antibodies to block CFA also varied (Fig. 5): H10 blocked FI-mediated cleavage of C3b, but not C4b, and F8 blocked cleavage of C4b, but was poor at blocking C3b cleavage. These results also show similarity with the results of previous studies: the H10-binding epitope is confined to CCP4, which was previously shown to be critical for C3b CFA, but dispensable for C4b CFA.¹⁸ Of the two KCP mutants that fail to bind F8, CFA inhibition by F8 was more similar to that previously found for R20Q/R33Q/R35Q, which retained all C3b CFA activity and reduced C4b CFA activity compared with K64Q/K65Q/K88Q, which had no C3b or C4b CFA activity at all.⁴⁵ Comparison of inhibition by J11 with the activity previously reported for the mutants that failed to bind J11 also yielded mixed results: preincubation of KCP with J11, K131Q/K133Q/H135Q and E99Q/E152Q/D155Q completely abrogated C4b CFA (except for E99Q/E152Q/D155Q, which had 50% activity) and C3b CFA, while J11 was poor at inhibiting DAA compared with K131Q/K133Q/H135Q and E99Q/E152Q/D155Q, which had greatly diminished DAA. Of all the antibodies tested, E7 was the most versatile as it inhibited DAA, C3b and C4b CFA and, for the most part, heparan sulphate binding, despite the fact that E7 and J11 appeared to share the K131/K133/H135 epitope (Table 2).

With the exception of E7, most monoclonal antibodies blocked two or three of the four KCP functions: binding of H10 left C4b CFA and heparan sulphate binding intact, J11 removed all functions but DAA, and F8 let most of the C3b CFA activity remain. Despite the relatively high affinity of B6 for KCP, as measured by SPR analysis, B6 did not appear to bind well to KCP in flow cytometry or ELISA studies and showed no blocking activity of any description; this may relate to altered conformation of KCP (and exposure of a partially hidden epitope) when bound to the CM5 chip for binding analysis which was not common to native KCP. The differential abilities of antibodies to block DAA and CFA indicate that there are common and unique requirements for C3b and C4b interactions with KCP, and, more importantly, distinct regions required for DAA and CFA (which may reflect areas of direct FI binding sites on KCP that are not required for DAA). Our future studies will determine the major immunogenic epitopes of KCP recognized by antibodies from KSHV-infected patients, to determine if antibodies arising from natural infection block all KCP functions, correlate with disease status, or are selective in their ability to neutralize the complement evasion mechanisms of KSHV.