Molecular Immunology, Vol. 29, No. 1, pp. 83-93, 1992 Printed
in Great
0
Britain.
0161-5890/92 $5.00 + 0.00 1991 Pergamon Press plc
A CALCIUM-BINDING MONOCLONAL ANTIBODY THAT RECOGNIZES A NON-CALCIUM-BINDING EPITOPE IN THE SHORT CONSENSUS REPEAT UNITS (SCRs) OF COMPLEMENT Clr* SHERRY L. WARD and KENNETH C. INGHAM~ Biochemistry Laboratory,
American Red Cross Biomedical Research and Development MD 20855, U.S.A.
(First received 15 December
1990; accepted in revised form
22 March
Rockville,
1991)
Abstract-Clr is a Ca*+-binding serine protease that interacts with two other plasma proteins, Clq and Cls, to form Cl, the first component of the complement cascade. A monoclonal antibody, BG6, has been produced which binds to Cl? only in the presence of Ca2+, requiring 3-5 PM Ca2+ for half-maximal binding. The antibody reacts with native and heat-denatured Clf, and with zymogen Clr, but does not cross-react with ClS or Clq. BG6 did not significantly affect the esterolytic activity of Cl? toward a synthetic thioester substrate nor the hemolytic activity of Cl reconstituted from subcomponents in the presence of the antibody. A tryptic fragment of Cl? which consists of the C-terminal y region of the A chain disulfide-linked to the B chain (YB) binds in a Ca2+-dependent manner to BG6-Sepharose. Western blotting experiments have further localized the epitope to the y region of the A chain, which is composed of two short consensus repeat (SCR) units. The N-terminal CIregion contains the only previously determined Ca2+ -binding site in the Cl! molecule. Equilibrium dialysis experiments confirmed that Clf-yB does not bind Ca2+, and showed that antibody BG6 and the yB/BG6 complex do bind Ca2+. Thus, the Ca’+-dependent nature of this interaction is due exclusively to binding of the metal ion to the antibody. Equilibrium dialysis and immunoblotting have further localized the Ca 2+-binding site to the Fab fragment of BG6, indicating that the metal-induced conformational change resides in or near the variable region of the IgG. BG6 may set a precedent for the preparation of Ca2+-dependent antibodies to non-Ca2+-binding epitopes in other proteins.
to form the two-chain activated serine protease C If. Cl r is a multidomain protein of 86,000 Da (Cooper, 1985; Arlaud et al., 19876; Sim et al., 1977). It is very similar in size and domain structure and shares 40% sequence homology with Cls (Villiers et al., 1985). Both proteins contain domains that are similar to those found in other proteins; a serine protease domain, an epidermal growth factor-like domain, two short consensus repeat (SCR) units, and two internally homologous motifs that are unique to Clr and Cls (Leytus et al., 1986; Arlaud et al., 1987a). The assembly and function of the Cl complex is dependent on the presence of Ca’+, which binds to all three subcomponents (Cooper, 1985; Arlaud et al., 1987a). The Ca’+-dependent interaction of ClS with itself, and with ClF, occurs through the N-terminal tl region of each protein. Ca2+ inhibits the spontaneous intra-dimer autoactivation of Clr, as well as the activation of Cls by ClF (Villiers et al., 1980), but not the autoactivation of Clr-yB (Lacroix et al., 1989). The fact that the enzymatic properties of the C-terminal region of Clr can be affected by Ca’+-binding to its N-terminal region suggests the occurrence of global conformational changes induced by the metal ion. The initial objectives of the present study were to map these Ca2+-induced conformational changes, and perhaps further localize the Ca’+-binding site(s) in Cl?.
INTRODUCTION Cl, the first component of the classical complement pathway, is a complex of three different proteins: Clq, and two molecules each of Clr and Cls (Cooper, 1985; Reid, 1986; Schumaker et al., 1987; Arlaud et al., 1987a; Muller-Eberhard, 1988). Clr and Cls are proenzymes that form a tetramer, Clr,s,, in the presence of Ca’+, which interacts with the collagenous arms of Clq to form Cl. The binding of Clq to immune complexes, or to other activators, causes proenzyme Clr to cleave itself
*Supported in part by the National Institutes of Health grant HL 21791. tTo whom correspondence should be addressed at: American Red Cross, 15601 Crabbs Branch Way, Rockville, MD 20855, U.S.A. Abbreviations-ELISA, enzyme-linked immunosorbent assay; EDTA, ethylenediamine-tetraacetic acid; DTT, dithiothreitol; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; BSA, bovine serum albumin; MAb, monoclonal antibody; Ab/Ag, antibody/antigen; Tris, Tris[hydroxymethyl]aminomethane; SCR, short consensus repeat; RCM, reduced and carboxymethylated; Fab, antibody fragment that consists of the light and heavy chain variable regions, the light chain constant region, and the first constant region of the heavy chain; IgG, immunoglobulin G. 83
S. L. WARD and K. C. IN~~HAM
84
Metal-dependent antibodies have been used to identify and characterize such Ca’+-induced conform~tional transitions in other plasma proteins (Ohlin and Stenflo, 1987; Taylor et al., 1987; Ohlin et al., 1988; Orthner et al., 1989). A monoclonal antibody demonstrating Ca’+dependent binding to Cl? was obtained. We expected that the epitope for this antibody would be in the cx region, the only previously determined site of Ca2+ binding in Cli: (Villiers et al., 1980; Busby and Ingham, 1987). However, the Ca *+-dependent interaction of this antibody with Clf was ultimately localized to the y region of the protein. From equilibrium dialysis experiments, we confirmed that Clf-yB does not bind Ca2+, and showed that the Caz+-dependent Ab/Ag interaction is due to Ca2+-binding by the Fab portion of the antibody. Our results suggest that closer investigation of the metal-binding properties of monoconal antibodies may be required when using a Ca2+-dependent antibody to define a Ca2+-induced conformational change in an antigen. Furthermore, it may be reasonable to screen hybridoma supernatants for Ca2+-dependent antibodies, regardless of whether the antigen is a Ca2’-binding protein. MATERIALS
AND METHODS
Enzyme-linked
immunosorbent assay (ELISA)
Plates were coated overnight at 4” with 1 pg Clr/well in 100 ~1 of coat buffer (0. I M Tris-base, pH 8.0, and 0.15 M NaCl) containing either 2 mM CaCl, or 10 mM EDTA. All subsequent steps were performed at room temperature. Cl? adsorbed to the ELISA plates equally well in either Ca’+ or EDTA-containing buffer. To prevent nonspecific binding, Cl r-coated wells were incubated with 10 mg/ml BSA in incubation buffer (0.05 M Tris-base, pH 7.4, containing 0.1 M NaCl, 0.02% Tween 20 (v: v), and either 2 mM CaCl, or 10 mM EDTA) for I hr. Aliquots of cell supernatants (60-70 ~1) were then added to duplicate wells containing an equal volume of incubation buffer with either 2mM Cal+ or 10 mM EDTA and incubated for 2 hr. The wells were washed thoroughly between each incubation step with 0.02 M Tris-base, pH 7.2, containing 0.9% NaCl, 0.05% Tween 20 (v:v), and either 2 mM CaCl, or 10 mM EDTA. Mouse antibody that bound to Cl? was identified using rabbit anti-mouse IgG-AP. Production and char~&teriz~tion of rno~o~lon~~~ntibod~e~~
Freund’s adjuvants were from Difco. Tissue culture media, sera, and reagents, rabbit anti-mouse IgG (Fcspecific) alkaline phosphatase (AP), AP substrate tablets, Ponceau S, and Iodoacetamide were from Sigma. Hyclone fetal bovine serum was used for the initial hybridoma production. Goat anti-mouse Ig subclassspecific seras were from ImmunoVision, Inc., Springdale, AK. Horseradish peroxidase (HRP)-labeled anti-goat and anti-mouse IgG were from BioRad. TPCK-treated trypsin was purchased from Worthington Biochemical Corporation, and immobilized papain was from Pierce, 4sCaCl, was obtained from New England Nuclear. Eq~librium dialysis cells and membranes were from Be&Art Products, Pequannock, NJ. Isolation of complement proteins Cl?, Cl:, and Clq were purified from Cohn fraction I of human plasma as described elsewhere (Busby and Ingham, 1987). Zymogen Clr was a gift from Dr Andrea Tenner, and was isolated as previously described (Tenner and Frank, 1987). The partially purified Cl-inhibitor preparation was provided by Dr Milan Wickerhauser (Wickerhauser et al., 1987). Protein concns were determined spectrophotometrically using the values A (280nm, lo!)= 11.6, 8.0, 13.9, 10.5, and 6.82 for Clf, Cl?-a, Cli-yB, ClS, and Clq, respectively (Busby and Ingham, 1987; Busby and Ingham, 1988). Preparation of reduced and carboxymethylated
samples were dialyzed into HzO, and then into 0.02 M Tris-base, pH 7.5, containing 0.5 M NaCl, 2 mM CaCl,, and 0.02% Na azide.
Cl?
ClF was incubated with 24mM DTT both in the presence and absence of 4.5 M guanidine HCl, at 37” for 150 min. Iodoacetamide (40 mM) was added and the mixture incubated in the dark for 15 min at room temp. Excess DTT was added to quench the reaction, and the
Mice were immunized intraperitoneally with 62.5 fig of Cl? in complete Fruend’s adjuvant. Fourteen days later they were boosted with 62.5pg of protein in incomplete Fruend’s adjuvant. On day 20 the blood of the immunized animals was tested for reactivity to Cl? using a solid-phase ELISA. The mouse with the highest antibody titre to Clr was immunized for the final time on day 31, Four days later the spleen was removed and fused with Sp2/0-Ag14 mouse myeloma cells (Shulman et al., 1978) using 50% polyethylene glycol 1500 (Kohler and Milstein, 1975). Cell supernatants from wells showing hybridoma growth, after HAT selection, were screened using the ELISA described above. Two hybridoma wells contained IgGs that bound to Cl? in the presence of Ca 2+, but not in EDTA. These cells were cloned by limiting dilution, and the positive clones identified using the same ELISA. The mouse Ig subclass of these Ca2’-dependent monocional antibodies was determined using goat anti-mouse Ig subclass specific seras and anti-goat Ig HRP in an ELISA.
This was performed using a Pharmacia fast-protein liquid chromatography (FPLC) system, and flow rates of l-2 ml/min. The elution profiles represent the absorbance of the proteins at 280nm. Proteins were coupled to CNBr-Activated Sepharose 4B (2 mg protein/ml gel), according to the manufacturer’s directions (Pharmacia). Columns were equilibrated in TBS f Ca2+ (0.02 M Tris-base, pH 7.5, containing 0.5 M NaCl, 2 mM CaCI,, and 0.02% Na azide). Cell-culture media (39 ml) was diluted I/5 in 0.1 M HEPES, pH 7.5, containing 0.75 M NaCI, and 10mM CaCl,; 0.45 nm filtered, and applied to a 7 ml column of Cl?-Sepharose.
A calcium-binding monoclonal antibody The column was washed with 16 volumes of the equiliwas eluted with bration buffer, and antibody TBS + EDTA (0.02 M Tris-base, pH 7.5, containing 0.5 M NaCl, 10 mM EDTA, and 0.02% Na azide). CH was diluted to 0,14mg/ml in TBS + Ca’+, 0.45 pm filtered, and then 10 ml was applied to a 3.5 ml column of BG&Sepharose, which was washed and eluted as above. BG6-Sepharose was also used for purification of the fragments Cl?-y B and CY,from a tryptic digest (l/30 trypsin) of Clf. Conditions were as described above except that only 1 mM CaCl, and 5 mM EDTA were included in the chromatography buffers. A subsequent chromatography step on C 1S-Sepharose (2 mg protein/ml gel) was then used to obtain the purified fragments. Preparation of Fab fragments of BG6
85
The dialysis cells had a capacity of 100 ~1 on each side of the membrane, but only 90 ,ul was added per side so that mixing could be achieved. The cells were separated by a dialysis membrane with a 6000 kDa exclusion limit. Prior to use the dialysis membrane was boiled in 2 mM EDTA, and then washed extensively in chelexed ED buffer. “CaCl, was diluted in chelexed buffer containing the appropriate concn CaCl,, and then added to the cells. Proteins were added to opposite sides of different cells. The cells were rotated at 4” for 48 hr. Samples (3 x 15 p 1) were removed, added to 5 ml of scintillation cocktail, and counted in a Beckman Model LS 2800 scintillation counter. Volumes on each side of the cells were equal when the samples were removed. Blank cells containing only buffer and 45CaC1, were assayed to determine that the cells had reached equilibrium. K, values were calculated using the equation: Kd = (F/B) (nP,, - B) where F and B refer to the concentration of free and bound Ca2+, POis the total concn of antibody, and n = 2 was assumed for the number of binding sites per mole of antibody.
MAb BG6 was purified from cell-culture media as described above. BG6 (1.2mg/ml) was dialyzed into 20 mM HEPES, pH 7.0, 0.15 M NaCl, 10 mM EDTA, and 0.02% Na azide, and then incubated with immobilized papain at 35” for 3 hr. Immobilized papain was removed by filtration, and the digest was dialyzed against 20mM Na acetate, pH 5.5, containing 10 mM NaCl. The Fab fragment was purified using a 10-100 mM NaCl gradient on a Mono S column (Pharmacia).
SDS-PAGE was performed using a Pharmacia PHAST gel system and 8-25% gradient gels. Proteins were visualized on the gel using either the Coomassie Blue or silver stain technique suggested by the manufacturer.
Western blotting
Hemolytic titrations
ClF (1 pg/lane) and autodigested Clf (5.5 @g/lane) were separated on 9% SDSpolyacrylamide gels under nonreducing conditions, transferred electrophoretically to nitrocellulose (NC), and visualized after brief incubation with 0.2% Ponceau S in 10% acetic acid. Duplicate NC strips were processed in buffers containing either 10 mM EDTA or 2 mM CaCl,. To prevent nonspecific binding, NC was incubated with 3% milk and 2% BSA in blot buffer (0.02 M Tris-HCl, pH 7.5, containing 0.5 M NaCl, 0.05% Tween 20 (v:v), and either 2 mM CaCl,, or 10 mM EDTA) for 2.5 hr at 37”. Cell-culture media was diluted l/5 in blot buffer + 2% milk, and incubated with the NC for 1.5 hr at room temp. The NC strips were washed 4 x 10 min. Rabbit anti-mouse IgG HRP, diluted l/3000 in blot buffer + 2% milk, was incubated with the NC for 1 hr. The HRP substrate was diaminobenzidine tetrahydrochloride (Cappel).
Cl? was pre-incubated with antibody BG6, or Fab fragments of BG6 (at a lo-fold excess of Ab over the concentration of Clf), or with buffer alone for 15 min at room temperature. Cl was then reconstituted from its subcomponent proteins by incubating physiological concns of Clq (7Opg/ml) and ClS (32 fig/ml) with 11 pgg/ml of soluble or Ab-complexed Clf for 15 min at 37” in Veronal-buffered saline. Dilutions of these samples were then added to antibody-sensitized sheep erythrocytes (EAC4 cells) (Rapp and Borsos, 1970), and the assay was performed as previously described (Tenner and Frank, 1986). The amount of hemolysis was determined by measuring the absorption at 412 nm after the cells were sedimented. The Cl hemolytic activity is presented as the % of total lysis which is the (GDexr - ODbutrcrconfr0l)/(GDtotal lysis ODttu,r contro1)*
Equ~iibr~u~ dialysis
Identification of a Ca2+-dependent antibody to Clf
Metal ions were removed from the buffers by chromatography on Chelex 100 and stored in pre-rinsed polypropylene containers. The proteins used in equilibrium dialysis (ED) experiments were concentrated using Centricon 30 cells from Amicon. EDTA was added to a final concn of 2 mM, and the proteins were dialyzed against 0.02 M Tris-base, pH 7.5, containing 0.15 M NaCl, 2 mM EDTA, and 0.02% Na azide. They were then dialyzed extensively against chelex-treated ED buffer (0.02 M Tris-base, pH 7.4, containing 0.15 M NaCl, and 0.02% Na azide).
Monoclonal antibodies (MAbs) were prepared by immunization of mice with human Cl?. Screening of hybridoma supernatants using duplicate ELISA’s, performed with buffers containing either 2 mM CaCl, or 10 mM EDTA, led to the identification of several hybridomas producing monoclonal antibodies which bind to Clr, two of which (BG6 and CF4) bind only in the presence of Ca2+. Cells producing the Ca’+-dependent antibodies were cloned by limiting dilution, and a number of stable clones, all of the mouse IgG,, subclass, were obtained. Antibodies derived from BG6 and CF4 seem
Gel electrophoresis
RESULTS
86
S.
L.
WARD
and K. C.
INGHAM
to compete for the same site in Clf, as determined by the equivalent inhibition of Cl? binding to BG6-Sepharose in the presence of either soluble BG6 or CF4 (data not shown). The clone BG6-5E5 was used in the studies described in this paper, and antibody derived from it will
Table
Protein adsorbed to ELISA plate
A,,,/20 min Media BG6
be called BG6. An ELISA was used to demonstrate that BG6 reacts in a concn-dependent, saturable manner with Clf in the presence of Ca’+, but not in EDTA-containing buffer (Fig. 1A). At 2 mM CaCl, and 1 fig of Cl F per well, half-maximal binding was observed at 4 nM BG6. Panel
Cl? + 2 mM CaCl, Cli: + 10 mM EDTA Proenzyme C IF 37” Cl? (LTT) 65” Cl? (HTT) Cl? + DTT RCM ClF BSA ClS
0.450 0.014 0.322 0.459 0.443 0.443 0.000 0.019 0.021 0.047
c (4
0.5
I 0.01
0.10
1.00
10.00
Antibody (ug)
I
I
I
I
0,001
0.010
0.100
1.000
I 10.000
[Ca*+] (mM)
0.4
0.1
0
0.001
0.010
0.100
1 .ooo
[Metal] (mM) Fig. 1. (A) The monoclonal antibody BG6 interacts with Cl? in a concn-dependent manner in the presence of Ca2+, but not in the presence of EDTA. Increasing amounts of affinitypurified antibody were added to triplicate wells of Cl?-coated ELISA plates in the presence of 2 mM CaCl, (0) or 10 mM EDTA (@). (B) The half-maximal concn of Ca2+ required for the binding of BG6 was determined using an ELISA. ELISA plates were coated with either 0.12 ,uM ClT- (0) or Cl?-yB (a) in the presence of EDTA. Metal-free antibody in the various concns of Ca*+ was then added to triplicate wells. Wells incubated with BG6 in the presence of EDTA were used as the blank. (C) The BG6/ClP interaction is also mediated by metal ions other than Ca*+. Different metal ions were substituted for Ca2+ in an ELISA
as described
in part B.
and cross-reactivity using an ELISA
of BGh
0.000 0.007 0.000 0.009 0.003 0.001 0.000 0.003 0.006 0.000
All ELISA buffers contained 2 mM CaZi, except when Clr was assayed in the presence of EDTA. Low-temperature transition (LTT) Clf was prepared by heating Cl? at 38” for 30 mm in the presence of 1 mM EDTA. High-temp transition (HTT) Clf was heated at 65” for I hr in the presence of 1 mM EDTA. Reduced Cl? was prepared by incubation with 2 mM dithiothreitoI at 38” for 2.5 hr and adsorbed to the microtitre wells in the presence of the reducing agent. The preparation of RCM Cl? is described under Methods. Results are reported as the average absorbance at 410nm for three wells incubated with AP substrate for 20 min.
_I--“‘--*--’
0
determined
Clq
0.1
0
I. Specificity
B of Fig. 1 shows that between 3 and 5 ,uM Ca”+ was required to obtain half-maximal binding of antibody BG6 to Clf (or to the yB fragment). Other metals, including Mg2+, Ba2+, Mn’+, Co2+ and Zn2”, were able to substitute for Ca2+ as shown in Fig. 1C. The specificity of antibody BG6 was also examined using an ELISA (Table 1). The epitope for this Ca2+dependent antibody is also present on the single-chain proenzyme Cl r. heat-denatured Clr, and Cl? that was adsorbed to the wells in the presence of DTT retained the ability to interact with BG6. However, fully reduced and carboxymethylated Cli: did not bind BG6. BG6 did not cross-react with BSA, nor with the other Cl subcomponent proteins ClS and Clq. In fact, when human serum is immunoblotted with BG6, the only protein that reacts with the antibody appears at the same molecular weight as Cl? (data not shown), indicating that BG6 is quite specific for Clf. The Ca*+-dependent Ab/Ag interaction was also demonstrated by affinity chromatography. As shown in Fig. 2A, BG6 from cell-culture media bound to Cl?Sepharose in the presence of Ca’+, and was efficiently eluted with EDTA. The heavy and light chains of antibody BG6 are the only protein bands evident in the EDTA eluant (lane 3 of the figure insert). Cli: (lane I) was not detectable in the purified antibody. Clii binds to BG6-Sepharose in the same dent manner, as shown in Fig. 2B. Partially (lane 1) was passed over BG6-Sepharose. nating protein passed through the column only Clf was eluted with EDTA (lane 3).
Ca2+-depenpurified Cl? A contami(lane 2), and
A calcium-binding
monoclonal
87
antibody
0.18
EDTA /LJ I/
201
401 7 110
301 Volume
(ml)
0.08
0.06
20 Volume (ml)
80
90
Fig. 2. Affinity chromatography of BG6 on ClT-Sepharose (A), and Clf on BG6-Sepharose (B). The figures show representative elution profiles from the affinity columns, and the figure inserts show the reduced proteins separated by SDS-PAGE. The first lane of each gel shows the molecular weight markers: 99.2, 66.5, 45.0, 31.0, 21.5, and 14.7 kDa. (A) BG6containing cell-culture media (lane 2) was applied to Clf-Sepharose, washed with Ca2+ -containing buffer, and the antibody eluted with 10 mM EDTA (lane 3). Lane 1 of the gel shows Cl? that was immobilized on the Sepharose. (B) Partially purified ClF (lane 1) was applied to BG6-Sepharose. Bound protein that was eluted with 10 mM EDTA contained only Cl? (lane 3). The unbound fraction contained an unidentified contaminant (lane 2). Localization
of the epitope for BG6 in Clf
When Clf is digested with trypsin, the A chain is cleaved into three fragments (a, /I, and y), very similar to those obtained upon autodigestion (Arlaud et al., 1980). The y fragment remains attached to the B chain through a disulfide bridge, to form the fragment y B (Arlaud et al., 1980; Arlaud et al., 1986). When such a digest (lane 2 of Fig. 3) was passed over BG6-Sepharose, only Cl? and Clf-y B bound, and were eluted with EDTA (lane 4). Lane 3 shows the proteins in the
unbound fraction: Cl?-u (at 31 kDa), trypsin, and soybean trypsin inhibitor (SBTI). While the small fi fragment is not visible by Coomassie staining, it can be seen in the unbound fraction on silver-stained gels. These results suggest that the epitope for BG6 is located in the y B fragment of the Clf molecule, and not in the c1region as was originally expected. Sequential column chromatography on BG6- and CIS-Sepharose was used to isolate the yB (lane 5) and c1fragments (lane 6) from a tryptic digest of Cl?. Digestion of CIF with trypsin produced two sizes of y B fragments: the major fragment
88
S. L. WARD
and K.C.
INGHAM (A)
YB)
a)
P)
Fig. 3. Electrophoretic analysis of fractions obtained by affinity chromatography of trypsin-digested ClF on BG6-Sepharose. The figure shows the nonreduced proteins separated by SDS-PAGE and stained with Coomassie Blue. Lane 1 contains Cl?. Lane 2 shows trypsin-digested Clf which contains, in the order of decreasing mol. wt. Clf, ;JB, 51,SBTI, trypsin, and /j’ (not visible). The unbound fraction (lane 3) contains cx, SBTI, and trypsin. Bound proteins, eluted with 10mM EDTA, are Cl? and yB (lane 4). Additional chromatography on ClSSepharose was used to obtain the purified Clr fragments, yB (lane 5) and s( (lane 6).
+ of 56 kDa, and the minor component of 51 kDa (lane 5). The molar yield of yB was almost 50% of theoretical. To confirm the unexpected result that the epitope for the Ca’+-dependent antibody resides in the yB fragment of Cl?, the reactivity of BG6 was tested using Western blotting. Again, BG6 reacted with Cl? in the presence of Ca’+, but not in EDTA-containing buffers (Fig. 4A). When autodigested Cl? was tested, only residual intact Cl? and the y B fragments were detected by BG6, and only in the presence of Ca*’ (Fig. 4B). A disulfide in the Cl i: molecule appeared to be required to preserve the integrity of the epitope, since BG6 failed to blot either chain of reduced Clf, even though protein staining confirmed that transfer had occurred (not shown). The results of the ELISA also suggested that a disulfide in the antigen was required for BG6 binding in that incubation of Clf with DTT prior to coating of the wells had no effect on subsequent antibody binding, whereas complete reduction and carboxymethylation abolished binding (Table 1). In an attempt to separate the A and B chains of Cl? without reducing all of the intra-chain disulfide bonds, Clf was reduced and carboxymethylated (RCM) in the absence of guanidine-HCl. This partially reduced Clr (Fig. 5A, lane 2) was applied to BG6-Sepharose. About 23% of the protein bound to BG6 in a Ca*+-dependent manner. The bound protein consisted of approximately equimolar amounts of A and B chain (lane 3), suggesting that the two chains associate by some mechanism other than the inter-chain disulfide that is no longer intact. When Clr was reduced and carboxymethylated in the presence of 4.5 M guanidine-HCl (lane l), the protein lost the with
a molecular
weight
-
Fig. 4. Western blotting localizes the epitope of the Calf-dependent antibody BG6 to the yB fragment of Cl?. In both panels, the left lane shows the protein bands present on the nitrocellulose as stained with Ponceau S. (A) BG6 bound to Clf, but only in the presence of Cal+. BG6 did not react with reduced Cl? (not shown). (B) In autodigested Clr, only intact Cl? and fragments of C I? containing the 7 B region of the protein were detected with BG6. Autodigested Cl1- was prepared by incubating C 1f at 38” for 5 hr in the presence of 1 mM EDTA.
ability to interact with the immobilized antibody. These data, in addition to the ELISA and immunoblotting results, are evidence that an inaccessible disulfide in Clf is necessary for BG6 binding. As shown in Fig. 5B, the A chain of RCM ClT- that was prepared in the absence of denaturing agents retained the ability to interact with BG6 in a Western blot when separated by SDS-PAGE under nonreducing conditions (lanes 2 and 3). Neither chain of the fully reduced sample reacted with BG6 (lane 1). When the yB fragment is reduced and carboxymethylated in the same manner as Clf, the antibody epitope is destroyed. However, if the reduction of yB is effected with only 0.1-0.2 mM DTT (instead of 24 mM), a faint band that corresponds to y can be detected by immunoblotting (data not shown). This suggests that at least one of the disulfides in the y region has a greater susceptibility to reduction in the y B fragment than in the intact Clf molecule. In any case, it is clear that the epitope for antibody BG6 resides in the C-terminal y region of the A chain of c1r.
A calcium-binding (B)
(A)
monoclonal
2
3
89
Cl-inhibitor (Cl-Inh) binds at the active site in the B chain of Clf, forming a stable complex of 190 kDa (Arlaud et al., 1979). A partially purified preparation of Cl-Inh was incubated with ClT. When the incubation mixture was passed over BG6-Sepharose, both ClF and the Cl?-Cl-Inh complex bound to, and were eluted from, the column (Fig. 6D, lane 3). The unbound fraction contained numerous proteins that were contaminants of the crude Cl-Inh preparation (lane 2). When the Cl-Inh preparation alone was passed over BG6-Sepharose, neither Clf nor Cl-Inh was detected in the eluant (not shown). Efict
1
antibody
1
2
3
Fig. 5. Electrophoretic and transblotting analysis of reduced and carboxymethylated Clt localizes the epitope for BG6 to the A chain of Cl?. Panel A, Coomassie-stained proteins on an SDS-PAGE gel. Panel B, proteins immunoblotted with BG6 after transfer to nitrocellulose. Lane 1, fully reduced and carboxymethylated Cl?. Lane 2, partially reduced and carboxymethylated Cl?. Lane 3, that fraction of material shown in lane 2 that bound to BG6-Sepharose and was eluted with EDTA.
Effect of other complement proteins The effect of other proteins that bind to Clf on the interaction of BG6 with Cl? was characterized by affinity chromatography on BG6-Sepharose, using SDS-PAGE to analyze bound and unbound fractions. The experiment of Fig. 6A shows that ClS does not bind to BG6-Sepharose (lane 3); more than 99% of the protein passed through the column. However, a small amount of Clf that was present as a contaminant, presumably in the form of the tetramer Clf,S2 did bind to the column, together with an approximately equal amount of CIS (lane 4). In a subsequent application of the unbound ClS, no detectable protein bound to BG6Sepharose. This not only confirms that ClS does not cross-react with antibody BG6, but indicates that the epitope in Cl? is still accessible to the antibody when Cl? is bound to ClS in the tetramer. In the experiment of Fig. 6B, equimolar amounts of ClF and ClS were preincubated together for 30min at room temp in the presence of Ca’+, to obtain the Cl?,& tetramer (lane 1). The tetramer bound to BG6-Sepharose (lane 3) confirming that the interaction of Clf with CIS does not mask the epitope. The binding of Clr to antibody BG6 does not appear to disrupt the dimeric structure of Clf. Cl? was immobilized on BG6-Sepharose in the presence of 1 mM Ca*+, and then eluted with EDTA (Fig. 6C, lane 2). In a separate experiment, using the same conditions, soluble BG6 was added to the immobilized Cl?, followed by elution of the column with EDTA (lane 3). In this case, both Clf and BG6 were eluted from the column, suggesting that a second epitope for the antibody was present on the immobilized Cl?, and therefore, that Clf dimers were bound to BG6-Sepharose.
of BG6 on Cl7 and Cl activity
The effect of BG6 on the enzymatic activity of Cl? was determined by monitoring spectrophotometrically the hydrolysis of the synthetic thioester substrate Z-GlyArg-SBzl (McRae et al., 198 1). When Cl? was preincubated with a IO-fold excess of BG6, the esterolytic activity was inhibited by no more than 15% (not shown). The effect of BG6 was also tested in a more complex hemolytic assay (Rapp and Borsos, 1970; Tenner and Frank, 1986) comparing Cl that had been reconstituted in the presence or absence of antibody. At a dilution of Cl producing 45% lysis of the erythrocytes, Cl formed in the presence of a lo-fold excess of BG6 produced 44% lysis, and Cl formed in the presence of a IO-fold excess of BG6 Fab fragments produced 52% lysis. This lack of effect of BG6 on the hemolytic activity of Cl was observed in three different experiments. Thus, binding of BG6 to the SCR units of Clf does not appear to interfere with the assembly or function of the Cl complex. However, it was not conclusively demonstrated that BG6 remains bound to Clf in the Cl complex. Ca2+-binding studies The above results establish that BG6 binds specifically, and in a Ca*+-dependent manner, to the y region of the A chain of Cl?. Since this region of the protein has never been reported to bind Ca2+, and the yB fragment has in fact been reported to not bind Ca2+ (Villiers et al., 1980), equilibrium dialysis experiments were performed to characterize the Ca’+-dependence of this Ab/Ag interaction, and to determine which component of the complex binds Ca’+, Dialysis cells that contained BG6, y B, or BG6 and yB were dialyzed against cells that contained 45CaC12with varying concns of unlabeled CaCl,. Blank cells containing only buffer and 45CaC12were assayed to determine that the cells had reached equilibrium. The equilibrium dialysis results are presented in Table 2. Cl?-yB consistently failed to bind 4SCa2+under the conditions used in these experiments. On the other hand, antibody BG6 and the BG6/yB complex consistently bound 45Ca2+, BG6 with a Kd of 149 +_ 18 ,uM, and the complex with a Kd of 65 +_11 PM. These results indicate that it is the antibody that binds Ca’+, and that the affinity of the antibody for Ca2+ increases approximately two-fold upon complexing with Cl+B.
90
S.
L.
WARD
and
K. C. IN(;HAM
(A)
1
1
2
3
2
3
Fig. 6. Analysis by affinity chromatography of the effect of CIS, BG6, and Cl-Inh on the interaction of Cl? with BG6-Sepharose. The figures show the nonreduced proteins separated by SDS-PAGE on 825% gradient gels. The first lane in each panel contains the molecular weight markers: 99.2, 66.5, 45.0, 31.0, 21.5, and 14.7 kDa. (A) CIS does not bind to BG6-Sepharose: lane 1, Cl?; lane 2, CIS; lane 3, unbound fraction which contained about 99% of the applied CIS. The small amount of protein that was eluted with EDTA was concentrated and shown to contain approximately equal amounts of Cl? and ClS (lane 4). (B) The tetramer, Clr,S, (lane I), binds to BG6-Sepharose: lane 2, unbound fraction; lane 3, bound fraction. (C) A second epitope for BG6 is present on Clf immobilized on BG6-Sepharose. Cl? was immobilized in the presence of 1 mM Ca* ‘, the column was washed, and bound Clf was eluted with EDTA (lane 2). When soluble BG6 was added to the immobilized Cl?, both Cl? and BG6 were eluted with EDTA (lane 3). (D) Clf complexed with Cl-Inh retained the ability to bind to BG6Sepharose. Cl? (0.46mg) (lane 1) was incubated with a crude preparation of Cl-Inh (-0.12 mg), and then applied to BG6Sepharose. A number of proteins which were present as contaminants in the Cl-Inh preparation passed through the column (lane 2). Only Cl? and the higher molecular weight Cl?-Cl-Inh complex, were eluted with EDTA (lane 3).
An additional estimate of the affinity of the BG6/Clr of a complex for Ca2+ was obtained by a modification previously described affinity chromatography technique (Chaiken, 1979; Dunn and Chaiken, 1975). Cl? was bound to BG6-Sepharose in TBS + 1 mM CaCl,, Table 2. Equilibrium Protein Antibody BG6 (n = 7) Clf-yB (n = 5) BG6/yB complex (n = 5) Cl? (n = 2) Rabbit anti-mouse IgG (n = 1)
dialysis
results
Kd + S.D. (PM) 149 + 18 No detectable binding 65+11 15 io.4 No detectable binding
Protein concns were 5-12 PM, and Ca*+ concns were varied from 2 to 200 PM. The number of Ca*+-binding sites were assumed to be 1 per mole of ClT: and 2 per mole of antibody BG6. Numbers in parentheses indicate the number of experiments leading to the averages and standard deviations (S.D.) shown.
washed, and then eluted with either EDTA, chelexed buffer, or with increasing concns of Ca’+. The elution volumes (V,) increased as the [Ca’+] was increased. A plot of l/( V, - V,) ’ vs [Ca’+] produced a straight line, from which an apparent Kd of 14 PM was calculated (Fig. 7). Localization Fab
qf the Ca’+-binding
fragments
of
BG6
site in BG6
appear
as
a
doublet
on
SDS-polyacrylamide gels with mol. wts of 48 and 45 kDa. Both fragments bind to Cl? in a Ca’+-dependent manner in a modified Western blot assay (Fig. 8). BG6 Fabs (lanes 1) and intact BG6 (lanes 2) immobilized
on nitrocellulose were incubated with Clr in the presence of either 2 mM Ca2+ or 10 mM EDTA. Cl?, which was bound to protein on the nitrocellulose, was then identified using a different MAb to Cl?. Binding of this second MAb was detected using Fc-specific anti-mouseIgG-AP. This Fc-specific IgG could not bind directly to the Fab on the nitrocellulose, but only to the Fc region
A calcium-binding
monoclonal
91
antibody
that the Ca*+-binding site in BG6 is located in the Fab region of the antibody. indicate
DISCUSSION
I 0
I
I
I
50
100
150
pM Caz+ Fig. 7. Affinity chromatographic determination of an apparent Kd for the Ca2+-dependent interaction of antibody BG6 with Clf. Cl? (0.23 mg) was immobilized on BGdSepharose in TBS containing 1 mM CaCl,. The column was washed with the same buffer, and then eluted in separate experiments with either EDTA, chelexed buffer, or with increasing concns of Ca*+. V, is the elution volume, and I’, is the total volume determined using buffer containing 5 mM EDTA (assuming no interaction of Cl? and BG6). An apparent Kd of 14 PM was calculated as the y intercept/slope of the plot (Chaiken, 1979).
of the second MAb. Cl? reacted with BG6 Fab only in the presence of Ca 2+. The weaker reaction observed with intact BG6 in EDTA is due to binding of the detection antibody directly to the Fc region of BG6 on the nitrocellulose. BG6 Fabs also bound 45Ca2+in an equilibrium dialysis experiment, with an affinity similar to that observed for the intact antibody. At 19.6 FM Fab and 101 PM CaCl,, the & of the Fab for Ca2+ was 124 yM. These results
1
2 Silver stain
1 Ca*+
2
1
2
EOTA
Fig. 8. The Fab fragment of BG6 binds to Clf in a Ca*+-dependent manner in a modified Western blot assay. In this experiment, BG6 and Fab fragments of BG6 were transferred electrophoretically to nitrocellulose (NC). The first panel shows the silver-stained proteins on the NC. The NC was incubated sequentially (in the presence of either 2 mM CaCl, or 10 mM EDTA) with Clf (20 pgg/ml), then with a second MAb to Cl? (VIIC6-5D2), and finally with AP-labeled anti-mouseIgG (Fc-specific).
To our knowledge, this is the first documentation of a monoclonal antibody to the complement protein Cl? (Heinz and Loos, 1989; Burger, 1985). The antibody, BG6, binds to Cl? only in the presence of Ca2+, and was selected during the screening process because of this metal-dependent reactivity. Other non-Ca*+-dependent monoclonal antibodies were also generated to Clf, but have not yet been characterized. The Ca’+-dependent nature of the interaction of antibody BG6 with Clf was verified by ELISA, affinity chromatography, and Western blotting. EDTA completely abolished the binding in all cases. BG6 reacts specifically, and in a concentrationdependent manner, with Cl? adsorbed to plastic. The antibody does not cross-react with BSA, Clq, or even with ClS, a protein sharing significant homology with Cl?. This specific, Ca2+-dependent interaction of antibody and antigen provides a useful system for the purification of BG6, Cl?, and proteolytic fragments of Cl? using affinity chromatography. The epitope for antibody BG6 resides in the y B fragment of Cli: as determined by Western blotting, and also by the Ca ‘+-dependent binding of y B from a tryptic digest of CIF to BG6-Sepharose. The conclusion that BG6 does not bind to Cl&a, which contains the only known Ca2+-binding site in the protein, is consistent with our data showing that binding is not affected by the protein/protein interaction of Cl? with ClS to form the tetramer Cl?,&, or by the irreversible low-temp unfolding transition, both of which occur in the OLportion of the protein (Busby and Ingham, 1987; Busby er al., 1986). Analogously, two events that occur in the B chain of the protein, the interaction of Cl7 with Cl-Inh (Arlaud et al., 1979), and a high-temp unfolding transition (Busby and Ingham, 1987), do not affect the interaction of BG6 with Cl?. Antibody BG6 also had no effect on the activity of Clt towards a synthetic peptide substrate. These data suggested that the epitope for the antibody resides in the y portion of Cl?, a region that is composed of two SCR units. This assignment correlates with the fact that Cl?-y, by analogy with ClS-y, is the most stable region of the protein toward heat (Medved et al., 1989). The fact that BG6 failed to immunoblot either chain of fully reduced ClT indicated that the epitope is conformation-specific, Further experiments showed that a buried disulfide, other than the inter-chain disulfide, is necessary to maintain the integrity of the epitope for BG6. When partially reduced and carboxymethylated ClF was immunoblotted with BG6, the antibody reacted with the A chain, thereby establishing the epitope for BG6 within the y region of Clr. In spite of the proposed models that Clf dimer interaction occurs through a y-y (Weiss et a/., 1986) or a y B-y B (Arlaud et al., 1986) association, the binding of BG6 to the y region of Cli: did not disrupt Cli dimer structure. Since antibody BG6 was not found to cross-
I
in Great
0
Britain.
0161-5890/92 $5.00 + 0.00 1991 Pergamon Press plc
A CALCIUM-BINDING MONOCLONAL ANTIBODY THAT RECOGNIZES A NON-CALCIUM-BINDING EPITOPE IN THE SHORT CONSENSUS REPEAT UNITS (SCRs) OF COMPLEMENT Clr* SHERRY L. WARD and KENNETH C. INGHAM~ Biochemistry Laboratory,
American Red Cross Biomedical Research and Development MD 20855, U.S.A.
(First received 15 December
1990; accepted in revised form
22 March
Rockville,
1991)
Abstract-Clr is a Ca*+-binding serine protease that interacts with two other plasma proteins, Clq and Cls, to form Cl, the first component of the complement cascade. A monoclonal antibody, BG6, has been produced which binds to Cl? only in the presence of Ca2+, requiring 3-5 PM Ca2+ for half-maximal binding. The antibody reacts with native and heat-denatured Clf, and with zymogen Clr, but does not cross-react with ClS or Clq. BG6 did not significantly affect the esterolytic activity of Cl? toward a synthetic thioester substrate nor the hemolytic activity of Cl reconstituted from subcomponents in the presence of the antibody. A tryptic fragment of Cl? which consists of the C-terminal y region of the A chain disulfide-linked to the B chain (YB) binds in a Ca2+-dependent manner to BG6-Sepharose. Western blotting experiments have further localized the epitope to the y region of the A chain, which is composed of two short consensus repeat (SCR) units. The N-terminal CIregion contains the only previously determined Ca2+ -binding site in the Cl! molecule. Equilibrium dialysis experiments confirmed that Clf-yB does not bind Ca2+, and showed that antibody BG6 and the yB/BG6 complex do bind Ca2+. Thus, the Ca’+-dependent nature of this interaction is due exclusively to binding of the metal ion to the antibody. Equilibrium dialysis and immunoblotting have further localized the Ca 2+-binding site to the Fab fragment of BG6, indicating that the metal-induced conformational change resides in or near the variable region of the IgG. BG6 may set a precedent for the preparation of Ca2+-dependent antibodies to non-Ca2+-binding epitopes in other proteins.
to form the two-chain activated serine protease C If. Cl r is a multidomain protein of 86,000 Da (Cooper, 1985; Arlaud et al., 19876; Sim et al., 1977). It is very similar in size and domain structure and shares 40% sequence homology with Cls (Villiers et al., 1985). Both proteins contain domains that are similar to those found in other proteins; a serine protease domain, an epidermal growth factor-like domain, two short consensus repeat (SCR) units, and two internally homologous motifs that are unique to Clr and Cls (Leytus et al., 1986; Arlaud et al., 1987a). The assembly and function of the Cl complex is dependent on the presence of Ca’+, which binds to all three subcomponents (Cooper, 1985; Arlaud et al., 1987a). The Ca’+-dependent interaction of ClS with itself, and with ClF, occurs through the N-terminal tl region of each protein. Ca2+ inhibits the spontaneous intra-dimer autoactivation of Clr, as well as the activation of Cls by ClF (Villiers et al., 1980), but not the autoactivation of Clr-yB (Lacroix et al., 1989). The fact that the enzymatic properties of the C-terminal region of Clr can be affected by Ca’+-binding to its N-terminal region suggests the occurrence of global conformational changes induced by the metal ion. The initial objectives of the present study were to map these Ca2+-induced conformational changes, and perhaps further localize the Ca’+-binding site(s) in Cl?.
INTRODUCTION Cl, the first component of the classical complement pathway, is a complex of three different proteins: Clq, and two molecules each of Clr and Cls (Cooper, 1985; Reid, 1986; Schumaker et al., 1987; Arlaud et al., 1987a; Muller-Eberhard, 1988). Clr and Cls are proenzymes that form a tetramer, Clr,s,, in the presence of Ca’+, which interacts with the collagenous arms of Clq to form Cl. The binding of Clq to immune complexes, or to other activators, causes proenzyme Clr to cleave itself
*Supported in part by the National Institutes of Health grant HL 21791. tTo whom correspondence should be addressed at: American Red Cross, 15601 Crabbs Branch Way, Rockville, MD 20855, U.S.A. Abbreviations-ELISA, enzyme-linked immunosorbent assay; EDTA, ethylenediamine-tetraacetic acid; DTT, dithiothreitol; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; BSA, bovine serum albumin; MAb, monoclonal antibody; Ab/Ag, antibody/antigen; Tris, Tris[hydroxymethyl]aminomethane; SCR, short consensus repeat; RCM, reduced and carboxymethylated; Fab, antibody fragment that consists of the light and heavy chain variable regions, the light chain constant region, and the first constant region of the heavy chain; IgG, immunoglobulin G. 83
S. L. WARD and K. C. IN~~HAM
84
Metal-dependent antibodies have been used to identify and characterize such Ca’+-induced conform~tional transitions in other plasma proteins (Ohlin and Stenflo, 1987; Taylor et al., 1987; Ohlin et al., 1988; Orthner et al., 1989). A monoclonal antibody demonstrating Ca’+dependent binding to Cl? was obtained. We expected that the epitope for this antibody would be in the cx region, the only previously determined site of Ca2+ binding in Cli: (Villiers et al., 1980; Busby and Ingham, 1987). However, the Ca *+-dependent interaction of this antibody with Clf was ultimately localized to the y region of the protein. From equilibrium dialysis experiments, we confirmed that Clf-yB does not bind Ca2+, and showed that the Caz+-dependent Ab/Ag interaction is due to Ca2+-binding by the Fab portion of the antibody. Our results suggest that closer investigation of the metal-binding properties of monoconal antibodies may be required when using a Ca2+-dependent antibody to define a Ca2+-induced conformational change in an antigen. Furthermore, it may be reasonable to screen hybridoma supernatants for Ca2+-dependent antibodies, regardless of whether the antigen is a Ca2’-binding protein. MATERIALS
AND METHODS
Enzyme-linked
immunosorbent assay (ELISA)
Plates were coated overnight at 4” with 1 pg Clr/well in 100 ~1 of coat buffer (0. I M Tris-base, pH 8.0, and 0.15 M NaCl) containing either 2 mM CaCl, or 10 mM EDTA. All subsequent steps were performed at room temperature. Cl? adsorbed to the ELISA plates equally well in either Ca’+ or EDTA-containing buffer. To prevent nonspecific binding, Cl r-coated wells were incubated with 10 mg/ml BSA in incubation buffer (0.05 M Tris-base, pH 7.4, containing 0.1 M NaCl, 0.02% Tween 20 (v: v), and either 2 mM CaCl, or 10 mM EDTA) for I hr. Aliquots of cell supernatants (60-70 ~1) were then added to duplicate wells containing an equal volume of incubation buffer with either 2mM Cal+ or 10 mM EDTA and incubated for 2 hr. The wells were washed thoroughly between each incubation step with 0.02 M Tris-base, pH 7.2, containing 0.9% NaCl, 0.05% Tween 20 (v:v), and either 2 mM CaCl, or 10 mM EDTA. Mouse antibody that bound to Cl? was identified using rabbit anti-mouse IgG-AP. Production and char~&teriz~tion of rno~o~lon~~~ntibod~e~~
Freund’s adjuvants were from Difco. Tissue culture media, sera, and reagents, rabbit anti-mouse IgG (Fcspecific) alkaline phosphatase (AP), AP substrate tablets, Ponceau S, and Iodoacetamide were from Sigma. Hyclone fetal bovine serum was used for the initial hybridoma production. Goat anti-mouse Ig subclassspecific seras were from ImmunoVision, Inc., Springdale, AK. Horseradish peroxidase (HRP)-labeled anti-goat and anti-mouse IgG were from BioRad. TPCK-treated trypsin was purchased from Worthington Biochemical Corporation, and immobilized papain was from Pierce, 4sCaCl, was obtained from New England Nuclear. Eq~librium dialysis cells and membranes were from Be&Art Products, Pequannock, NJ. Isolation of complement proteins Cl?, Cl:, and Clq were purified from Cohn fraction I of human plasma as described elsewhere (Busby and Ingham, 1987). Zymogen Clr was a gift from Dr Andrea Tenner, and was isolated as previously described (Tenner and Frank, 1987). The partially purified Cl-inhibitor preparation was provided by Dr Milan Wickerhauser (Wickerhauser et al., 1987). Protein concns were determined spectrophotometrically using the values A (280nm, lo!)= 11.6, 8.0, 13.9, 10.5, and 6.82 for Clf, Cl?-a, Cli-yB, ClS, and Clq, respectively (Busby and Ingham, 1987; Busby and Ingham, 1988). Preparation of reduced and carboxymethylated
samples were dialyzed into HzO, and then into 0.02 M Tris-base, pH 7.5, containing 0.5 M NaCl, 2 mM CaCl,, and 0.02% Na azide.
Cl?
ClF was incubated with 24mM DTT both in the presence and absence of 4.5 M guanidine HCl, at 37” for 150 min. Iodoacetamide (40 mM) was added and the mixture incubated in the dark for 15 min at room temp. Excess DTT was added to quench the reaction, and the
Mice were immunized intraperitoneally with 62.5 fig of Cl? in complete Fruend’s adjuvant. Fourteen days later they were boosted with 62.5pg of protein in incomplete Fruend’s adjuvant. On day 20 the blood of the immunized animals was tested for reactivity to Cl? using a solid-phase ELISA. The mouse with the highest antibody titre to Clr was immunized for the final time on day 31, Four days later the spleen was removed and fused with Sp2/0-Ag14 mouse myeloma cells (Shulman et al., 1978) using 50% polyethylene glycol 1500 (Kohler and Milstein, 1975). Cell supernatants from wells showing hybridoma growth, after HAT selection, were screened using the ELISA described above. Two hybridoma wells contained IgGs that bound to Cl? in the presence of Ca 2+, but not in EDTA. These cells were cloned by limiting dilution, and the positive clones identified using the same ELISA. The mouse Ig subclass of these Ca2’-dependent monocional antibodies was determined using goat anti-mouse Ig subclass specific seras and anti-goat Ig HRP in an ELISA.
This was performed using a Pharmacia fast-protein liquid chromatography (FPLC) system, and flow rates of l-2 ml/min. The elution profiles represent the absorbance of the proteins at 280nm. Proteins were coupled to CNBr-Activated Sepharose 4B (2 mg protein/ml gel), according to the manufacturer’s directions (Pharmacia). Columns were equilibrated in TBS f Ca2+ (0.02 M Tris-base, pH 7.5, containing 0.5 M NaCl, 2 mM CaCI,, and 0.02% Na azide). Cell-culture media (39 ml) was diluted I/5 in 0.1 M HEPES, pH 7.5, containing 0.75 M NaCI, and 10mM CaCl,; 0.45 nm filtered, and applied to a 7 ml column of Cl?-Sepharose.
A calcium-binding monoclonal antibody The column was washed with 16 volumes of the equiliwas eluted with bration buffer, and antibody TBS + EDTA (0.02 M Tris-base, pH 7.5, containing 0.5 M NaCl, 10 mM EDTA, and 0.02% Na azide). CH was diluted to 0,14mg/ml in TBS + Ca’+, 0.45 pm filtered, and then 10 ml was applied to a 3.5 ml column of BG&Sepharose, which was washed and eluted as above. BG6-Sepharose was also used for purification of the fragments Cl?-y B and CY,from a tryptic digest (l/30 trypsin) of Clf. Conditions were as described above except that only 1 mM CaCl, and 5 mM EDTA were included in the chromatography buffers. A subsequent chromatography step on C 1S-Sepharose (2 mg protein/ml gel) was then used to obtain the purified fragments. Preparation of Fab fragments of BG6
85
The dialysis cells had a capacity of 100 ~1 on each side of the membrane, but only 90 ,ul was added per side so that mixing could be achieved. The cells were separated by a dialysis membrane with a 6000 kDa exclusion limit. Prior to use the dialysis membrane was boiled in 2 mM EDTA, and then washed extensively in chelexed ED buffer. “CaCl, was diluted in chelexed buffer containing the appropriate concn CaCl,, and then added to the cells. Proteins were added to opposite sides of different cells. The cells were rotated at 4” for 48 hr. Samples (3 x 15 p 1) were removed, added to 5 ml of scintillation cocktail, and counted in a Beckman Model LS 2800 scintillation counter. Volumes on each side of the cells were equal when the samples were removed. Blank cells containing only buffer and 45CaC1, were assayed to determine that the cells had reached equilibrium. K, values were calculated using the equation: Kd = (F/B) (nP,, - B) where F and B refer to the concentration of free and bound Ca2+, POis the total concn of antibody, and n = 2 was assumed for the number of binding sites per mole of antibody.
MAb BG6 was purified from cell-culture media as described above. BG6 (1.2mg/ml) was dialyzed into 20 mM HEPES, pH 7.0, 0.15 M NaCl, 10 mM EDTA, and 0.02% Na azide, and then incubated with immobilized papain at 35” for 3 hr. Immobilized papain was removed by filtration, and the digest was dialyzed against 20mM Na acetate, pH 5.5, containing 10 mM NaCl. The Fab fragment was purified using a 10-100 mM NaCl gradient on a Mono S column (Pharmacia).
SDS-PAGE was performed using a Pharmacia PHAST gel system and 8-25% gradient gels. Proteins were visualized on the gel using either the Coomassie Blue or silver stain technique suggested by the manufacturer.
Western blotting
Hemolytic titrations
ClF (1 pg/lane) and autodigested Clf (5.5 @g/lane) were separated on 9% SDSpolyacrylamide gels under nonreducing conditions, transferred electrophoretically to nitrocellulose (NC), and visualized after brief incubation with 0.2% Ponceau S in 10% acetic acid. Duplicate NC strips were processed in buffers containing either 10 mM EDTA or 2 mM CaCl,. To prevent nonspecific binding, NC was incubated with 3% milk and 2% BSA in blot buffer (0.02 M Tris-HCl, pH 7.5, containing 0.5 M NaCl, 0.05% Tween 20 (v:v), and either 2 mM CaCl,, or 10 mM EDTA) for 2.5 hr at 37”. Cell-culture media was diluted l/5 in blot buffer + 2% milk, and incubated with the NC for 1.5 hr at room temp. The NC strips were washed 4 x 10 min. Rabbit anti-mouse IgG HRP, diluted l/3000 in blot buffer + 2% milk, was incubated with the NC for 1 hr. The HRP substrate was diaminobenzidine tetrahydrochloride (Cappel).
Cl? was pre-incubated with antibody BG6, or Fab fragments of BG6 (at a lo-fold excess of Ab over the concentration of Clf), or with buffer alone for 15 min at room temperature. Cl was then reconstituted from its subcomponent proteins by incubating physiological concns of Clq (7Opg/ml) and ClS (32 fig/ml) with 11 pgg/ml of soluble or Ab-complexed Clf for 15 min at 37” in Veronal-buffered saline. Dilutions of these samples were then added to antibody-sensitized sheep erythrocytes (EAC4 cells) (Rapp and Borsos, 1970), and the assay was performed as previously described (Tenner and Frank, 1986). The amount of hemolysis was determined by measuring the absorption at 412 nm after the cells were sedimented. The Cl hemolytic activity is presented as the % of total lysis which is the (GDexr - ODbutrcrconfr0l)/(GDtotal lysis ODttu,r contro1)*
Equ~iibr~u~ dialysis
Identification of a Ca2+-dependent antibody to Clf
Metal ions were removed from the buffers by chromatography on Chelex 100 and stored in pre-rinsed polypropylene containers. The proteins used in equilibrium dialysis (ED) experiments were concentrated using Centricon 30 cells from Amicon. EDTA was added to a final concn of 2 mM, and the proteins were dialyzed against 0.02 M Tris-base, pH 7.5, containing 0.15 M NaCl, 2 mM EDTA, and 0.02% Na azide. They were then dialyzed extensively against chelex-treated ED buffer (0.02 M Tris-base, pH 7.4, containing 0.15 M NaCl, and 0.02% Na azide).
Monoclonal antibodies (MAbs) were prepared by immunization of mice with human Cl?. Screening of hybridoma supernatants using duplicate ELISA’s, performed with buffers containing either 2 mM CaCl, or 10 mM EDTA, led to the identification of several hybridomas producing monoclonal antibodies which bind to Clr, two of which (BG6 and CF4) bind only in the presence of Ca2+. Cells producing the Ca’+-dependent antibodies were cloned by limiting dilution, and a number of stable clones, all of the mouse IgG,, subclass, were obtained. Antibodies derived from BG6 and CF4 seem
Gel electrophoresis
RESULTS
86
S.
L.
WARD
and K. C.
INGHAM
to compete for the same site in Clf, as determined by the equivalent inhibition of Cl? binding to BG6-Sepharose in the presence of either soluble BG6 or CF4 (data not shown). The clone BG6-5E5 was used in the studies described in this paper, and antibody derived from it will
Table
Protein adsorbed to ELISA plate
A,,,/20 min Media BG6
be called BG6. An ELISA was used to demonstrate that BG6 reacts in a concn-dependent, saturable manner with Clf in the presence of Ca’+, but not in EDTA-containing buffer (Fig. 1A). At 2 mM CaCl, and 1 fig of Cl F per well, half-maximal binding was observed at 4 nM BG6. Panel
Cl? + 2 mM CaCl, Cli: + 10 mM EDTA Proenzyme C IF 37” Cl? (LTT) 65” Cl? (HTT) Cl? + DTT RCM ClF BSA ClS
0.450 0.014 0.322 0.459 0.443 0.443 0.000 0.019 0.021 0.047
c (4
0.5
I 0.01
0.10
1.00
10.00
Antibody (ug)
I
I
I
I
0,001
0.010
0.100
1.000
I 10.000
[Ca*+] (mM)
0.4
0.1
0
0.001
0.010
0.100
1 .ooo
[Metal] (mM) Fig. 1. (A) The monoclonal antibody BG6 interacts with Cl? in a concn-dependent manner in the presence of Ca2+, but not in the presence of EDTA. Increasing amounts of affinitypurified antibody were added to triplicate wells of Cl?-coated ELISA plates in the presence of 2 mM CaCl, (0) or 10 mM EDTA (@). (B) The half-maximal concn of Ca2+ required for the binding of BG6 was determined using an ELISA. ELISA plates were coated with either 0.12 ,uM ClT- (0) or Cl?-yB (a) in the presence of EDTA. Metal-free antibody in the various concns of Ca*+ was then added to triplicate wells. Wells incubated with BG6 in the presence of EDTA were used as the blank. (C) The BG6/ClP interaction is also mediated by metal ions other than Ca*+. Different metal ions were substituted for Ca2+ in an ELISA
as described
in part B.
and cross-reactivity using an ELISA
of BGh
0.000 0.007 0.000 0.009 0.003 0.001 0.000 0.003 0.006 0.000
All ELISA buffers contained 2 mM CaZi, except when Clr was assayed in the presence of EDTA. Low-temperature transition (LTT) Clf was prepared by heating Cl? at 38” for 30 mm in the presence of 1 mM EDTA. High-temp transition (HTT) Clf was heated at 65” for I hr in the presence of 1 mM EDTA. Reduced Cl? was prepared by incubation with 2 mM dithiothreitoI at 38” for 2.5 hr and adsorbed to the microtitre wells in the presence of the reducing agent. The preparation of RCM Cl? is described under Methods. Results are reported as the average absorbance at 410nm for three wells incubated with AP substrate for 20 min.
_I--“‘--*--’
0
determined
Clq
0.1
0
I. Specificity
B of Fig. 1 shows that between 3 and 5 ,uM Ca”+ was required to obtain half-maximal binding of antibody BG6 to Clf (or to the yB fragment). Other metals, including Mg2+, Ba2+, Mn’+, Co2+ and Zn2”, were able to substitute for Ca2+ as shown in Fig. 1C. The specificity of antibody BG6 was also examined using an ELISA (Table 1). The epitope for this Ca2+dependent antibody is also present on the single-chain proenzyme Cl r. heat-denatured Clr, and Cl? that was adsorbed to the wells in the presence of DTT retained the ability to interact with BG6. However, fully reduced and carboxymethylated Cli: did not bind BG6. BG6 did not cross-react with BSA, nor with the other Cl subcomponent proteins ClS and Clq. In fact, when human serum is immunoblotted with BG6, the only protein that reacts with the antibody appears at the same molecular weight as Cl? (data not shown), indicating that BG6 is quite specific for Clf. The Ca*+-dependent Ab/Ag interaction was also demonstrated by affinity chromatography. As shown in Fig. 2A, BG6 from cell-culture media bound to Cl?Sepharose in the presence of Ca’+, and was efficiently eluted with EDTA. The heavy and light chains of antibody BG6 are the only protein bands evident in the EDTA eluant (lane 3 of the figure insert). Cli: (lane I) was not detectable in the purified antibody. Clii binds to BG6-Sepharose in the same dent manner, as shown in Fig. 2B. Partially (lane 1) was passed over BG6-Sepharose. nating protein passed through the column only Clf was eluted with EDTA (lane 3).
Ca2+-depenpurified Cl? A contami(lane 2), and
A calcium-binding
monoclonal
87
antibody
0.18
EDTA /LJ I/
201
401 7 110
301 Volume
(ml)
0.08
0.06
20 Volume (ml)
80
90
Fig. 2. Affinity chromatography of BG6 on ClT-Sepharose (A), and Clf on BG6-Sepharose (B). The figures show representative elution profiles from the affinity columns, and the figure inserts show the reduced proteins separated by SDS-PAGE. The first lane of each gel shows the molecular weight markers: 99.2, 66.5, 45.0, 31.0, 21.5, and 14.7 kDa. (A) BG6containing cell-culture media (lane 2) was applied to Clf-Sepharose, washed with Ca2+ -containing buffer, and the antibody eluted with 10 mM EDTA (lane 3). Lane 1 of the gel shows Cl? that was immobilized on the Sepharose. (B) Partially purified ClF (lane 1) was applied to BG6-Sepharose. Bound protein that was eluted with 10 mM EDTA contained only Cl? (lane 3). The unbound fraction contained an unidentified contaminant (lane 2). Localization
of the epitope for BG6 in Clf
When Clf is digested with trypsin, the A chain is cleaved into three fragments (a, /I, and y), very similar to those obtained upon autodigestion (Arlaud et al., 1980). The y fragment remains attached to the B chain through a disulfide bridge, to form the fragment y B (Arlaud et al., 1980; Arlaud et al., 1986). When such a digest (lane 2 of Fig. 3) was passed over BG6-Sepharose, only Cl? and Clf-y B bound, and were eluted with EDTA (lane 4). Lane 3 shows the proteins in the
unbound fraction: Cl?-u (at 31 kDa), trypsin, and soybean trypsin inhibitor (SBTI). While the small fi fragment is not visible by Coomassie staining, it can be seen in the unbound fraction on silver-stained gels. These results suggest that the epitope for BG6 is located in the y B fragment of the Clf molecule, and not in the c1region as was originally expected. Sequential column chromatography on BG6- and CIS-Sepharose was used to isolate the yB (lane 5) and c1fragments (lane 6) from a tryptic digest of Cl?. Digestion of CIF with trypsin produced two sizes of y B fragments: the major fragment
88
S. L. WARD
and K.C.
INGHAM (A)
YB)
a)
P)
Fig. 3. Electrophoretic analysis of fractions obtained by affinity chromatography of trypsin-digested ClF on BG6-Sepharose. The figure shows the nonreduced proteins separated by SDS-PAGE and stained with Coomassie Blue. Lane 1 contains Cl?. Lane 2 shows trypsin-digested Clf which contains, in the order of decreasing mol. wt. Clf, ;JB, 51,SBTI, trypsin, and /j’ (not visible). The unbound fraction (lane 3) contains cx, SBTI, and trypsin. Bound proteins, eluted with 10mM EDTA, are Cl? and yB (lane 4). Additional chromatography on ClSSepharose was used to obtain the purified Clr fragments, yB (lane 5) and s( (lane 6).
+ of 56 kDa, and the minor component of 51 kDa (lane 5). The molar yield of yB was almost 50% of theoretical. To confirm the unexpected result that the epitope for the Ca’+-dependent antibody resides in the yB fragment of Cl?, the reactivity of BG6 was tested using Western blotting. Again, BG6 reacted with Cl? in the presence of Ca’+, but not in EDTA-containing buffers (Fig. 4A). When autodigested Cl? was tested, only residual intact Cl? and the y B fragments were detected by BG6, and only in the presence of Ca*’ (Fig. 4B). A disulfide in the Cl i: molecule appeared to be required to preserve the integrity of the epitope, since BG6 failed to blot either chain of reduced Clf, even though protein staining confirmed that transfer had occurred (not shown). The results of the ELISA also suggested that a disulfide in the antigen was required for BG6 binding in that incubation of Clf with DTT prior to coating of the wells had no effect on subsequent antibody binding, whereas complete reduction and carboxymethylation abolished binding (Table 1). In an attempt to separate the A and B chains of Cl? without reducing all of the intra-chain disulfide bonds, Clf was reduced and carboxymethylated (RCM) in the absence of guanidine-HCl. This partially reduced Clr (Fig. 5A, lane 2) was applied to BG6-Sepharose. About 23% of the protein bound to BG6 in a Ca*+-dependent manner. The bound protein consisted of approximately equimolar amounts of A and B chain (lane 3), suggesting that the two chains associate by some mechanism other than the inter-chain disulfide that is no longer intact. When Clr was reduced and carboxymethylated in the presence of 4.5 M guanidine-HCl (lane l), the protein lost the with
a molecular
weight
-
Fig. 4. Western blotting localizes the epitope of the Calf-dependent antibody BG6 to the yB fragment of Cl?. In both panels, the left lane shows the protein bands present on the nitrocellulose as stained with Ponceau S. (A) BG6 bound to Clf, but only in the presence of Cal+. BG6 did not react with reduced Cl? (not shown). (B) In autodigested Clr, only intact Cl? and fragments of C I? containing the 7 B region of the protein were detected with BG6. Autodigested Cl1- was prepared by incubating C 1f at 38” for 5 hr in the presence of 1 mM EDTA.
ability to interact with the immobilized antibody. These data, in addition to the ELISA and immunoblotting results, are evidence that an inaccessible disulfide in Clf is necessary for BG6 binding. As shown in Fig. 5B, the A chain of RCM ClT- that was prepared in the absence of denaturing agents retained the ability to interact with BG6 in a Western blot when separated by SDS-PAGE under nonreducing conditions (lanes 2 and 3). Neither chain of the fully reduced sample reacted with BG6 (lane 1). When the yB fragment is reduced and carboxymethylated in the same manner as Clf, the antibody epitope is destroyed. However, if the reduction of yB is effected with only 0.1-0.2 mM DTT (instead of 24 mM), a faint band that corresponds to y can be detected by immunoblotting (data not shown). This suggests that at least one of the disulfides in the y region has a greater susceptibility to reduction in the y B fragment than in the intact Clf molecule. In any case, it is clear that the epitope for antibody BG6 resides in the C-terminal y region of the A chain of c1r.
A calcium-binding (B)
(A)
monoclonal
2
3
89
Cl-inhibitor (Cl-Inh) binds at the active site in the B chain of Clf, forming a stable complex of 190 kDa (Arlaud et al., 1979). A partially purified preparation of Cl-Inh was incubated with ClT. When the incubation mixture was passed over BG6-Sepharose, both ClF and the Cl?-Cl-Inh complex bound to, and were eluted from, the column (Fig. 6D, lane 3). The unbound fraction contained numerous proteins that were contaminants of the crude Cl-Inh preparation (lane 2). When the Cl-Inh preparation alone was passed over BG6-Sepharose, neither Clf nor Cl-Inh was detected in the eluant (not shown). Efict
1
antibody
1
2
3
Fig. 5. Electrophoretic and transblotting analysis of reduced and carboxymethylated Clt localizes the epitope for BG6 to the A chain of Cl?. Panel A, Coomassie-stained proteins on an SDS-PAGE gel. Panel B, proteins immunoblotted with BG6 after transfer to nitrocellulose. Lane 1, fully reduced and carboxymethylated Cl?. Lane 2, partially reduced and carboxymethylated Cl?. Lane 3, that fraction of material shown in lane 2 that bound to BG6-Sepharose and was eluted with EDTA.
Effect of other complement proteins The effect of other proteins that bind to Clf on the interaction of BG6 with Cl? was characterized by affinity chromatography on BG6-Sepharose, using SDS-PAGE to analyze bound and unbound fractions. The experiment of Fig. 6A shows that ClS does not bind to BG6-Sepharose (lane 3); more than 99% of the protein passed through the column. However, a small amount of Clf that was present as a contaminant, presumably in the form of the tetramer Clf,S2 did bind to the column, together with an approximately equal amount of CIS (lane 4). In a subsequent application of the unbound ClS, no detectable protein bound to BG6Sepharose. This not only confirms that ClS does not cross-react with antibody BG6, but indicates that the epitope in Cl? is still accessible to the antibody when Cl? is bound to ClS in the tetramer. In the experiment of Fig. 6B, equimolar amounts of ClF and ClS were preincubated together for 30min at room temp in the presence of Ca’+, to obtain the Cl?,& tetramer (lane 1). The tetramer bound to BG6-Sepharose (lane 3) confirming that the interaction of Clf with CIS does not mask the epitope. The binding of Clr to antibody BG6 does not appear to disrupt the dimeric structure of Clf. Cl? was immobilized on BG6-Sepharose in the presence of 1 mM Ca*+, and then eluted with EDTA (Fig. 6C, lane 2). In a separate experiment, using the same conditions, soluble BG6 was added to the immobilized Cl?, followed by elution of the column with EDTA (lane 3). In this case, both Clf and BG6 were eluted from the column, suggesting that a second epitope for the antibody was present on the immobilized Cl?, and therefore, that Clf dimers were bound to BG6-Sepharose.
of BG6 on Cl7 and Cl activity
The effect of BG6 on the enzymatic activity of Cl? was determined by monitoring spectrophotometrically the hydrolysis of the synthetic thioester substrate Z-GlyArg-SBzl (McRae et al., 198 1). When Cl? was preincubated with a IO-fold excess of BG6, the esterolytic activity was inhibited by no more than 15% (not shown). The effect of BG6 was also tested in a more complex hemolytic assay (Rapp and Borsos, 1970; Tenner and Frank, 1986) comparing Cl that had been reconstituted in the presence or absence of antibody. At a dilution of Cl producing 45% lysis of the erythrocytes, Cl formed in the presence of a lo-fold excess of BG6 produced 44% lysis, and Cl formed in the presence of a IO-fold excess of BG6 Fab fragments produced 52% lysis. This lack of effect of BG6 on the hemolytic activity of Cl was observed in three different experiments. Thus, binding of BG6 to the SCR units of Clf does not appear to interfere with the assembly or function of the Cl complex. However, it was not conclusively demonstrated that BG6 remains bound to Clf in the Cl complex. Ca2+-binding studies The above results establish that BG6 binds specifically, and in a Ca*+-dependent manner, to the y region of the A chain of Cl?. Since this region of the protein has never been reported to bind Ca2+, and the yB fragment has in fact been reported to not bind Ca2+ (Villiers et al., 1980), equilibrium dialysis experiments were performed to characterize the Ca’+-dependence of this Ab/Ag interaction, and to determine which component of the complex binds Ca’+, Dialysis cells that contained BG6, y B, or BG6 and yB were dialyzed against cells that contained 45CaC12with varying concns of unlabeled CaCl,. Blank cells containing only buffer and 45CaC12were assayed to determine that the cells had reached equilibrium. The equilibrium dialysis results are presented in Table 2. Cl?-yB consistently failed to bind 4SCa2+under the conditions used in these experiments. On the other hand, antibody BG6 and the BG6/yB complex consistently bound 45Ca2+, BG6 with a Kd of 149 +_ 18 ,uM, and the complex with a Kd of 65 +_11 PM. These results indicate that it is the antibody that binds Ca’+, and that the affinity of the antibody for Ca2+ increases approximately two-fold upon complexing with Cl+B.
90
S.
L.
WARD
and
K. C. IN(;HAM
(A)
1
1
2
3
2
3
Fig. 6. Analysis by affinity chromatography of the effect of CIS, BG6, and Cl-Inh on the interaction of Cl? with BG6-Sepharose. The figures show the nonreduced proteins separated by SDS-PAGE on 825% gradient gels. The first lane in each panel contains the molecular weight markers: 99.2, 66.5, 45.0, 31.0, 21.5, and 14.7 kDa. (A) CIS does not bind to BG6-Sepharose: lane 1, Cl?; lane 2, CIS; lane 3, unbound fraction which contained about 99% of the applied CIS. The small amount of protein that was eluted with EDTA was concentrated and shown to contain approximately equal amounts of Cl? and ClS (lane 4). (B) The tetramer, Clr,S, (lane I), binds to BG6-Sepharose: lane 2, unbound fraction; lane 3, bound fraction. (C) A second epitope for BG6 is present on Clf immobilized on BG6-Sepharose. Cl? was immobilized in the presence of 1 mM Ca* ‘, the column was washed, and bound Clf was eluted with EDTA (lane 2). When soluble BG6 was added to the immobilized Cl?, both Cl? and BG6 were eluted with EDTA (lane 3). (D) Clf complexed with Cl-Inh retained the ability to bind to BG6Sepharose. Cl? (0.46mg) (lane 1) was incubated with a crude preparation of Cl-Inh (-0.12 mg), and then applied to BG6Sepharose. A number of proteins which were present as contaminants in the Cl-Inh preparation passed through the column (lane 2). Only Cl? and the higher molecular weight Cl?-Cl-Inh complex, were eluted with EDTA (lane 3).
An additional estimate of the affinity of the BG6/Clr of a complex for Ca2+ was obtained by a modification previously described affinity chromatography technique (Chaiken, 1979; Dunn and Chaiken, 1975). Cl? was bound to BG6-Sepharose in TBS + 1 mM CaCl,, Table 2. Equilibrium Protein Antibody BG6 (n = 7) Clf-yB (n = 5) BG6/yB complex (n = 5) Cl? (n = 2) Rabbit anti-mouse IgG (n = 1)
dialysis
results
Kd + S.D. (PM) 149 + 18 No detectable binding 65+11 15 io.4 No detectable binding
Protein concns were 5-12 PM, and Ca*+ concns were varied from 2 to 200 PM. The number of Ca*+-binding sites were assumed to be 1 per mole of ClT: and 2 per mole of antibody BG6. Numbers in parentheses indicate the number of experiments leading to the averages and standard deviations (S.D.) shown.
washed, and then eluted with either EDTA, chelexed buffer, or with increasing concns of Ca’+. The elution volumes (V,) increased as the [Ca’+] was increased. A plot of l/( V, - V,) ’ vs [Ca’+] produced a straight line, from which an apparent Kd of 14 PM was calculated (Fig. 7). Localization Fab
qf the Ca’+-binding
fragments
of
BG6
site in BG6
appear
as
a
doublet
on
SDS-polyacrylamide gels with mol. wts of 48 and 45 kDa. Both fragments bind to Cl? in a Ca’+-dependent manner in a modified Western blot assay (Fig. 8). BG6 Fabs (lanes 1) and intact BG6 (lanes 2) immobilized
on nitrocellulose were incubated with Clr in the presence of either 2 mM Ca2+ or 10 mM EDTA. Cl?, which was bound to protein on the nitrocellulose, was then identified using a different MAb to Cl?. Binding of this second MAb was detected using Fc-specific anti-mouseIgG-AP. This Fc-specific IgG could not bind directly to the Fab on the nitrocellulose, but only to the Fc region
A calcium-binding
monoclonal
91
antibody
that the Ca*+-binding site in BG6 is located in the Fab region of the antibody. indicate
DISCUSSION
I 0
I
I
I
50
100
150
pM Caz+ Fig. 7. Affinity chromatographic determination of an apparent Kd for the Ca2+-dependent interaction of antibody BG6 with Clf. Cl? (0.23 mg) was immobilized on BGdSepharose in TBS containing 1 mM CaCl,. The column was washed with the same buffer, and then eluted in separate experiments with either EDTA, chelexed buffer, or with increasing concns of Ca*+. V, is the elution volume, and I’, is the total volume determined using buffer containing 5 mM EDTA (assuming no interaction of Cl? and BG6). An apparent Kd of 14 PM was calculated as the y intercept/slope of the plot (Chaiken, 1979).
of the second MAb. Cl? reacted with BG6 Fab only in the presence of Ca 2+. The weaker reaction observed with intact BG6 in EDTA is due to binding of the detection antibody directly to the Fc region of BG6 on the nitrocellulose. BG6 Fabs also bound 45Ca2+in an equilibrium dialysis experiment, with an affinity similar to that observed for the intact antibody. At 19.6 FM Fab and 101 PM CaCl,, the & of the Fab for Ca2+ was 124 yM. These results
1
2 Silver stain
1 Ca*+
2
1
2
EOTA
Fig. 8. The Fab fragment of BG6 binds to Clf in a Ca*+-dependent manner in a modified Western blot assay. In this experiment, BG6 and Fab fragments of BG6 were transferred electrophoretically to nitrocellulose (NC). The first panel shows the silver-stained proteins on the NC. The NC was incubated sequentially (in the presence of either 2 mM CaCl, or 10 mM EDTA) with Clf (20 pgg/ml), then with a second MAb to Cl? (VIIC6-5D2), and finally with AP-labeled anti-mouseIgG (Fc-specific).
To our knowledge, this is the first documentation of a monoclonal antibody to the complement protein Cl? (Heinz and Loos, 1989; Burger, 1985). The antibody, BG6, binds to Cl? only in the presence of Ca2+, and was selected during the screening process because of this metal-dependent reactivity. Other non-Ca*+-dependent monoclonal antibodies were also generated to Clf, but have not yet been characterized. The Ca’+-dependent nature of the interaction of antibody BG6 with Clf was verified by ELISA, affinity chromatography, and Western blotting. EDTA completely abolished the binding in all cases. BG6 reacts specifically, and in a concentrationdependent manner, with Cl? adsorbed to plastic. The antibody does not cross-react with BSA, Clq, or even with ClS, a protein sharing significant homology with Cl?. This specific, Ca2+-dependent interaction of antibody and antigen provides a useful system for the purification of BG6, Cl?, and proteolytic fragments of Cl? using affinity chromatography. The epitope for antibody BG6 resides in the y B fragment of Cli: as determined by Western blotting, and also by the Ca ‘+-dependent binding of y B from a tryptic digest of CIF to BG6-Sepharose. The conclusion that BG6 does not bind to Cl&a, which contains the only known Ca2+-binding site in the protein, is consistent with our data showing that binding is not affected by the protein/protein interaction of Cl? with ClS to form the tetramer Cl?,&, or by the irreversible low-temp unfolding transition, both of which occur in the OLportion of the protein (Busby and Ingham, 1987; Busby er al., 1986). Analogously, two events that occur in the B chain of the protein, the interaction of Cl7 with Cl-Inh (Arlaud et al., 1979), and a high-temp unfolding transition (Busby and Ingham, 1987), do not affect the interaction of BG6 with Cl?. Antibody BG6 also had no effect on the activity of Clt towards a synthetic peptide substrate. These data suggested that the epitope for the antibody resides in the y portion of Cl?, a region that is composed of two SCR units. This assignment correlates with the fact that Cl?-y, by analogy with ClS-y, is the most stable region of the protein toward heat (Medved et al., 1989). The fact that BG6 failed to immunoblot either chain of fully reduced ClT indicated that the epitope is conformation-specific, Further experiments showed that a buried disulfide, other than the inter-chain disulfide, is necessary to maintain the integrity of the epitope for BG6. When partially reduced and carboxymethylated ClF was immunoblotted with BG6, the antibody reacted with the A chain, thereby establishing the epitope for BG6 within the y region of Clr. In spite of the proposed models that Clf dimer interaction occurs through a y-y (Weiss et a/., 1986) or a y B-y B (Arlaud et al., 1986) association, the binding of BG6 to the y region of Cli: did not disrupt Cli dimer structure. Since antibody BG6 was not found to cross-
I
