- Email: [email protected]
Activation of the third component of complement (C3) detected by a monoclonal anti-C3‘g’ neoantigen antibody in a one-step enzyme immunoassay
Journal of Immunological Methods, 101 (1987) 201-207 Elsevier
201
JIM 04410
Activation of the third component of complement (C3) detected by a monoclonal anti-C3' g' neoantigen antibody in a one-step enzyme immunoassay Tom E. Mollnes and Peter J. Lachmann Institute of lmmunology and Rheumatology, The National Hospital, Oslo, Norway, and Medical Research Council Center, Cambridge, U.K. (Received 20 November 1986, revised received 19 February 1987; accepted 17 March 1987)
A previously produced and characterized rat monoclonal antibody recognizing a neoantigen in the human C3'g' fragment was used as the capture antibody in an enzyme-linked immunosorbent assay. Detection was made using a polyclonal rabbit anti-human C3d and a peroxidase-linked anti-rabbit Ig antiserum. The activity in normal human EDTA plasma was found to be 3% of that in a zymosan-activated serum pool. Fractionation experiments revealed that most of the activity in normal plasma, in vivo activated plasma and in vitro activated serum eluted in one peak with a molecular weight corresponding to iC3b. A positive correlation (P < 0.01) was found between the present assay and a previously established two-step C3d ELISA both with respect to normal plasma, individual patient samples and consecutively drawn samples following artificial in vivo activation. Complement activation assays based on specific antibodies to 'activation antigens' should be preferred whenever available since they enable direct, rapid and specific quantification of the actual fragment(s). Key words: C3; Complement activation; Neoantigen; ELISA
Introduction
Assays for quantitative detection of C3 activation products have mainly been based on radial immunodiffusion (Perrin et al., 1975), electrophoretic (Brandslund et al., 1981; Bourke et al., 1982) or nephelometric (Vergani et al., 1983) methods. In our experience most of these methods are not sufficiently sensitive and reliable to detect minor fluctuations of C3 activation products under in vivo conditions. A highly sensitive enzymelinked immunosorbent assay (ELISA) was thereCorrespondence to: T.E. Mollnes, Institute of Immunology and Rheumatology, Fr. Qvamsgt. 1, N-0172 Oslo 1, Norway.
fore constructed to quantify C3d/C3dg (Mollnes, 1985a). This assay was found convenient to evaluate C3 activation in different clinical situations (Mollnes, 1985b; Mollnes et al., 1986). However, as with previously described assays this one was also hampered by a precipitation step necessary to eliminate the influence of large C3 fragments since the antisera available reacted against native epitopes. Recently, a new principle for the detection of activation products has been introduced based on monoclonal antibodies against neoantigens expressed in the activation products. The fluidphase terminal SC5b-9 complement complex (TCC) (Mollnes et al., 1985a) and the C3 activa-
0022-1759/87/$03.50 © 1987 Elsevier Science Publishers B.V. (Biomedical Division)
202
tion products iC3b and C3c (Aguado et al., 1985) have been detected directly. Neither precipitation nor fractionation were required. The aim of the present study was to develop a C3 activation assay based on a monoclonal antiC3'g' neoantigen antibody and to compare this assay with the one previously published from our laboratory (Mollnes, 1985a). The monoclonal anti C3'g' antibody has been extensively characterized previously (Lachmann et al., 1982) as reacting against a 'g' epitope expressed in iC3b, C3dg and C3g. This antibody has been used with success to isolate complement-fixing immune complexes by coupling the antibody to Sepharose (Samuel et al., 1986).
Materials and methods
Antibodies Ascitic fluid of the rat monoclonal anti-C3'g' neoantigen antibody (clone 9) was preserved in 40% glycerol and used unfractionated in all experiments. The specificity of this antibody has been described in detail previously (Lachmann et al., 1982; Samuel et al., 1986). Rabbit anti-human C3d Ig fraction (lot l l l D ) was obtained from Dako Immunoglobulins, Copenhagen, Denmark; rabbit anti-human C3c (lot 153321B) from Behringwerke, Marburg, F.R.G., and peroxidasecoupled anti-rabbit Ig(lot 16) and anti-rat Ig (lot 10) from Amersham, U.K.
Preparation of standard and samples Activation of a normal human serum pool (n = 20, healthy blood donors) was made by adding 100 mg of zymosan A (Sigma, St. Louis, MO) to 10 ml of serum, which was incubated under continuous mixing at 37°C for 30 rain, spun at 27000 x g (Sorvall RC5B) for 30 min and stored at - 7 0 °C in 100 #1 aliquots. This activated serum was used as positive control and standard in the assay. Ethylenediaminetetraacetic acid (EDTA) plasma samples from 80 healthy blood donors (40 males and 40 females) and 20 patients were separated immediately and stored at - 7 0 ° C until tested to avoid in vitro activation. Fractionation of samples was made by passing
0.5 ml sample through a Sephacryl S-300 (Pharmacia, Sweden) column (2.5 × 100 cm) in a 0.05 M Tris/HC1 buffer, pH 7.60, containing 0.2 M NaC1, 0.01 M EDTA and 0.02% NaN 3.
Detection of C3'g' neoantigen activity in ELISA In this assay N U N C Immunoplates II (Copenhagen, Denmark) were coated with the monoclonal anti-C3'g' neoantigen antibody, diluted in phosphate-buffered saline (PBS), at 4 ° C for at least 48-h. The well volume was 100 ~tl for all steps and the optimal dilutions of the reagents are given in the results section. The plates were washed in a Dynawasher three times between each incubation using PBS containing 0.1% Tween 20 (Sigma, St. Louis, MO). This was also the buffer used for all the following incubations except for that with the substrate. The antigen was always added in triplicate and incubated at 37°C for 45 min. Then anti-C3d (Dako) was added and finally anti-rabbit Ig conjugated with peroxidase. Each incubation was performed at 37°C for 45 min. The substrate was 2,2'-azino-di-(3-ethylbenzthiazoline sulfonic acid) (ABTS, Boehringer Mannheim, F.R.G.), diluted in 0.1 M sodium acetate buffer, pH 4.0 to 0.18 mg/ml. 0.8 /~1 3% H202 was added per ml buffer immediately before use. The optical density (OD) was read on a Dynatech MR 580 at 405 nm and the results were analyzed by an Apple IIe computer with the Dynatech Immunosoft programme. Alternatively, a qualitative examination of the antigen was made by coating the ELISA plates with fractions (diluted 1:50 in PBS) after gel filtration of plasma and activated serum. AntiCYg' neoantigen diluted 1:5000 and anti-rat Ig diluted 1:1000 was then added. Incubations and development were as described above for the quantitative assay.
Stat&tical methods Correlations were calculated using Spearman's rank correlation coefficient. P values are based on two-tailed tests. Results
The monoclonal anti-C3'g' antibody was initially tested for its reactivity in ELISA using a
203
screening system for anti-complement neoantigens described in detail elsewhere (Mollnes et al., 1985b). The antibody reacted strongly with zymosan-activated serum and weakly with normal plasma, confirming the previously described neoantigen activity (Lachmann et al., 1982).
Quantitative ELISA for the C3'g" neoantigen A double-antibody ELISA with the monoclonal anti-C3'g' neoantigen antibody (clone 9) as the capture and a polyclonal anti-C3d antiserum as the indicator was constructed as described in the materials and methods section. Optimal dilutions of the reagents giving the highest signal/noise ratio were as follows: ascitic fluid of anti-C3'g' diluted 1:10000, zymosan-activated serum used as standard diluted from 1:400 (two-fold, eight steps), normal plasma and patient samples diluted 1:500, anti-C3d diluted 1:2000 and anti-rabbit Ig-peroxidase diluted 1 : 1000. The specificity of the double-antibody assay was examined by omitting the coating step (clone 9), and performing the rest of the assay as described above. These experiments were made on the same plate by coating half of the plate and
leaving the other half uncoated. Three expei'i, ments were performed using triplicate samples in each. Strong reactions using activated serum were observed on the coated part of the plate in contrast to the uncoated (Fig. 1). Normal human plasma (NHP) did not react with the uncoated part of the plate, whereas a significant signal was obtained in the assay. The standard was defined as containing 1000 arbitrary units (AU)/ml. The median value after ten experiments was found to be 30 A U / m l for NHP and the coefficient of variation (CV) was 0.19.
Fractionation of standard, normal plasma and patient plasma To examine which fragment(s) contributed to the activity in the assay and to compare the reaction patterns of in vitro and in vivo activated samples, the activated serum, the normal plasma pool and a patient sample were fractionated using a Sephacryl S-300 column. Fractions from activated serum and normal plasma were coated on ELISA plates and examined with the anti-C3'g' neoantigen, anti-C3d, and anti C3c antisera. One common peak was observed in the range of 150-200 kDa for all three antibodies. An additional low-molecular peak (approximately 30
O.D4OSnm 08 O.D. 4o5nm 0.6
e--e
0.8
\
\
Ill
\,
i--.
Anti-C3d
=--~
Anti-C3c
\
•
0.4_
\\
Anti-C3"g"
0.6\
\•
xs\ --
\
0.2
\
04
\e ~
?-%
8
0.2 0_
[
[ 125
~° ~ - - ~ I I I I I 31.2 7.8 Dilution of s t a n d a r d x l 0 -5
o I 1.9
ooe ooc I B.G.
\
I
Fig. 1. Two-fold dilutions of zymosan-activated serum (standard) tested in a double-antibody ELISA with the monoclonal anti-C3'g' as capture antibody and polyclonal anti-C3d as the second antibody (closed circles). Control experiments were made by omitting the capture antibody (open circles). Median of triplicates from three experiments are indicated. Values for background (B.G., omitting the antigen) are also indicated.
2= 0--
II
10
I
I ~ I Vo
I
I I 22
I~1~1 IgG AIb
3161
I I I Ino. Fraction
Fig. 2. ELISA plates were coated with fractions obtained after gel filtration of zymosan-activated serum. They were then reacted with anti-CYg' neoantigen, anti-C3d and anti-C3c antibodies. Elution of void volume fraction (V0), IgG and albumin (Alb) is indicated (mean of duplicate).
204
kDa) was observed in activated serum with antiC3d which was not seen using anti-C3'g'. Neither did anti-C3'g' react with other low molecular weight fractions (Fig. 2). These experiments were done four times under slightly different conditions using different preparations of activated serum giving essentially identical results each time. Normal human plasma revealed only one peak with these three antibodies corresponding to large-fragment C3 molecules. Control experiments omitting the second antibody layer were all completely negative. When zymosan-activated serum was incubated at 37 °C for 1 week a marked increase in the low molecular weight peak with anti-C3d was observed (Fig. 2). However, no activity was observed in this region using anti-C3'g'. Fragments were then examined using anti-C3'g' as capture antibody and anti-C3d (known to react with all C3 fragments containing d-epitopes) in the second antibody layer as described for the quantitative assay. Only one peak, corresponding to 150-200 kDa was observed both in activated serum, in normal plasma, and in two different patient samples (infectious diseases). However, in activated serum incubated at 37°C for 18 h a minor peak corresponding to the molecular weight of C3dg was observed in addition to the iC3b peak. The C3dg peak disappeared and the iC3b peak diminished after further incubation for 1 1.0-
OD 405rim
,,
," P a t i e n t
•
• NHP
week. Fig. 3 shows the activity in fractionated normal human plasma and the patient sample. The curves should not be compared quantitatively since the large number of fractions did not allow set-up on the same plate. Unfractionated, the patient plasma contained six times higher A U / m l values than the normal pool. Correlation with an established C3d / dg assay The double-antibody assay based on the antiC3'g' neoantigen antibody was correlated with a conventional assay for detection of C3d/C3dg (Mollnes, 1985a). This assay was based on polyethylene glycol (PEG) precipitation of large C3 fragments before the supernatant was examined for C3d in ELISA. That assay had previously been shown to correlate closely with a double-decker rocket immunoelectrophoretic method for C3d/dg quantification (Brandslund et al., 1981), but had the advantages of higher sensitivity and greater rapidity. Samples from 20 different patients in a heterogenous patient population were selected and examined in both ELISA assays at the same time. A close positive correlation was found (r~- 0.83, P < 0.0002) (Fig. 4). Furthermore, a series of plasma samples obtained from a patient undergoing extracorporeal
200_
C3d/dg I AU/ml
150--
08100--
~,,,
06-
/ • : ! elel •
50
04--
/'
02--
0 0
I 200
I 400
I
C3"g" n e o a n t i g e n [ AU/mr 600
Fig. 4. Correlationbetweentwo differentassaysto detect C3
O_ I
u 8
IAI Vo
I
i 12
i
f
I
uAI
I
IgG
18
u ,L AIb
[
n 22
i
I
~
n
Fractaon
no
Fig. 3. Fractions obtained after gel filtration of normal human plasma and a patient's plasma sample were examined in the quantitative double-antibody ELISA described. Elution of void volume fraction (V0), IgG and albumin (Alb) is indicated (median of triplicate).
activation evaluated using 20 plasma samples from a heterogeneous patient population. The abscissa shows C3 conversion (C3'g' neoantigen) measured by the method described in this paper and the ordinate shows C3 conversion measured by a previously described ELISA method for C3d/dg quantification. Upper reference limits for the assays are indicated by arrowheads (median of triplicate).
205 C3"g"neoantigen AU/rnl . _ _ . _100
C3d/dg AU/ml 300 . . . .
_80
/ /,
200
x\ '\ _60
,; ,; /
\
10C
l !
\./
I
I
/
,~2,
... o, /i ,'// i
,~ j',,,'( ....... \
"/
"z
/
I I Consecutively
_40
,, ,\
'J
I drawn
I
I I samples(n=10)
t
I
_ 20
I i
_0
Fig. 5. Correlation between the C3'g' neoantigen assay and the C3d/dg assay evaluated by a series of samples obtained consecutively from a patient undergoing extracorporeal circulation examined in both assays at the same time (median of triplicate).
circulation were examined in both assays. The complement activation pattern of this patient has been described in detail elsewhere (Mollnes, 1985b). Again, a close positive correlation was demonstrated between the two assays (r s -- 0.92, P < 0.01) (Fig. 5). Lower detection limit and reference range The lower detection limit (background + 2 SD) was found to be less than 0.02 AU/ml, corresponding to 0.002% conversion of the activated serum. The normal human plasma pool (n = 20) contained 30 AU/ml, corresponding to 3% conversion. Plasma samples from 80 healthy blood donors, 40 females and 40 males, were examined and the following reference ranges (2.5-97.5 percentile) could be defined: females 23-58 A U / m l and males 27-66 AU/ml.
Discussion
Detection of neoantigens expressed only in activated components is theoretically the most reasonable way to evaluate complement activation. These epitopes reflect the degree of activation of the corresponding component and can be quantified directly without the influence of native comPonents. Precipitation or fractionation of
samples is therefore not required. The risk of false positives due to in vitro activation should also be markedly reduced in one-step assays. Recently, two monoclonal antibodies directed against neoantigen epitopes in C3c and C3d were successfully used in assays of complement activation and immune complex formation (Aguado et al., 1985). The complement activation assays were, however, not correlated with conventional assays. We demonstrate in the present paper that the monoclonal anti-C3'g' neoantigen antibody (clone 9) can be used to design a reliable C3 activation assay which correlated significantly with a previously established C3d/dg method, but was not hampered by the disadvantages of conventional assays. A major advantage with the present assay is the high sensitivity (lower detection limit < 0.002% of activated serum). The sensitivity of the previously described C3 neoantigen assay was unfortunately not stated in the paper (Aguado et al., 1985), but estimations from their dilutions seem to indicate that the assay presented here is about ten-fold more sensitive. This has obvious implications for the examination of biological fluids with a low protein concentration as cerebrospinal fluid, which has been examined unconcentrated and diluted with success using the present assay (unpublished data). From a functional and pathophysiological point of view it can be argued that the total amount of a neoantigen epitope present may be as important as the concentration of a single particular fragment. The present assay would theoretically detect all fragments expressing a combination of the 'g' neoantigen and d-epitopes. In the present study we found that most of the activity detected is of a molecular weight corresponding to that of iC3b, which is known to express this epitope (Lachmann et al., 1982). We were hardly able to demonstrate fragments corresponding to C3dg or C3g. The latter is not surprising since the 'g' fragment very rarely reacts with polyclonal anti-C3 antibodies (Lachmann et al., 1982). The most likely explanation for the lack of reactivity against anti-C3dg in most of the samples examined is simply that no C3dg was present since the 'g' epitope is well known to be exposed in C3dg (Lachmann et al., 1982; Davis et al., 1984), and since we demonstrated C3dg in
206
activated serum incubated at 37°C for 18 h. It is easy to accept that serum activated in vitro under different conditions do not necessarily contain C3dg. However, it is surprising that C3dg was not demonstrated in the patients' plasma since it has become widely accepted that this fragment is the major physiological breakdown product after complement activation. These observations emphasize that quantification of 'activation' epitopes like the CYg' neoantigen, irrespective of which fragments the epitope is exposed on, may be more reliable as a screening method for in vivo complement activation than determination of one single fragment. Although most of the 'g' neoantigen activity corresponded to the molecular weight of iC3b, it cannot be excluded that this elution peak also contained other fragments expressing this epitope. C3d/C3dg may, under physiological conditions, be associated with albumin or other proteins (Johnson, 1984; Teisner et al., 1984), which may also explain the minor high molecular weight peak observed after the void volume fraction in Fig. 2. This is, however, of minor importance, since the control experiments definitely showed that all the activity in the double antibody assay was dependent on the 'g' neoantigen. We therefore decided to denote this assay a C3'g' neoantigen assay. For the same reasons we Prefer arbitrary units instead of percent. Arbitrary units are limited to that particular assay, in this case reflecting the total amount of CYg' neoantigen present, and thereby false comparisons between different types of assays can be avoided. We conclude that monoclonal antibodies against neoantigens are important tools in the development of assays to detect and quantify complement activation in a reliable, direct, highly sensitive and rapid fashion. There is now a need for other antibody specificities to replace conventional activation assays.
Acknowledgements We are indebted to Ms. Grethe Bergseth for her excellent technical assistance and to Ms. Karl Bertelsen for typing the manuscript.
References Aguado, M.T., Lambris, J.D., Tsokos, G.C., Burger, R., BitterSuermarm, D., Tamerius, J.D., Dixon, F.J. and Theofilopohis, A.N. (1985) Monoclonal antibodies against complement C3 neoantigens for detection of immune complexes and complement activation. Relationship between immune complex levels, state of C3, and numbers of receptors for C3b. J. Clin. Invest. 76, 1418. Bourke, B.E., Moss, I.K. and Maini, R.N. (1982) Measurement of the complement C3 breakdown product C3d by rocket immunoelectrophoresis. J. Immunol. Methods 48, 97. Brandslund, I., Siersted, H.C., Svehag, S.-E. and Teisner, B. (1981) Double-decker rocket immunoelectrophoresis for direct quantification of complement C3 split products with C3d specificities in plasma. J. Immunol. Methods 44, 63. Davis, A.E., Harrison, R.A. and Lachmann, P.J. (1984) Physiological inactivation of fluid phase C3b: isolation and structural analysis of C3c, C3dg (alpha 2D), and C3g. J. Immunol. 132, 1960. Johnson, U. (1984) The influence of polyethylene glycol precipitation and of aging plasma and serum on the measurement of C3dg/d. Acta Pathol. Microbiol. Immunol. Scand. Sect. C. 92, 193. Lachmann, P.J., Pangburn, M.K. and Oldroyd, R.G. (1982) Breakdown of C3 after complement activation. Identification of a new fragment C3g, using monoclonal antibodies. J. Exp. Med. 156, 205. Mollnes, T.E. (1985a) Quantification of the C3d split products of human complement by a sensitive enzyme-linked immunosorbent assay. Scand. J. Immunol. 21,607. Mollnes, T.E. (1985b) Early- and late-phase activation of complement evaluated by plasma levels of C3dg and the terminal complement complex. Complement 2, 156. Mollnes, T.E., Lea, T., Froland, S.S. and Harboe, M. (1985a) Quantification of the terminal complement complex in human plasma by an enzyme-linked immunosorbent assay based on monoclonal antibodies against a neoantigen of the complex. Scand. J. Immunol. 22, 197. Mollnes, T.E., Lea, T., Harboe, M. and Tschopp, J. (1985b) Monoclonal antibodies recognizing a neoantigen of poly(C9) detect the human terminal complement complex in tissue and plasma. Scand. J. Immunol. 22, 183. Mollnes, T.E., Lea, T., Mellbye, O.J., Pahle, J., Grand, O. and Harboe, M. (1986) Complement activation in rheumatoid arthritis evaluated by C3dg and the terminal complement complex. Arthritis Rheum. 29, 715. Perrin, L.H., Lambert, P.H. and Miescher, P.A. (1975) Complement breakdown products in plasma from patients with systemic lupus erythematosus and patients with membranoproliferative or other giomerulonephritis. J. Clin. Invest. 56, 165. Samuel, D.J., Amlot, P.L., Shepherd, P. and Lachmann, P.J. (1986) An efficient one-step method for isolating immune complexes from whole serum using a monoclonal anti-C3g affinity immunosorbent. Clin. Exp. Immunol. 65,458. Teisner, B., Petersen, N.E., Folkerson, J.J. and Hau, J. (1984)
207 Crossed immunoelectrophoretic analysis of split products of the third complement factor (C3d) following in vivo activation of the complement system. Electrophoresis 5, 84. Vergani, D., Bevis, L., Nasaruddin, B.A., Mieli-Vergani, G.
and Tee, D.E.H. (1983) Clinical application of a new nephelometric technique to measure complement activation. J. Clin. Pathol. 36, 793.
201
JIM 04410
Activation of the third component of complement (C3) detected by a monoclonal anti-C3' g' neoantigen antibody in a one-step enzyme immunoassay Tom E. Mollnes and Peter J. Lachmann Institute of lmmunology and Rheumatology, The National Hospital, Oslo, Norway, and Medical Research Council Center, Cambridge, U.K. (Received 20 November 1986, revised received 19 February 1987; accepted 17 March 1987)
A previously produced and characterized rat monoclonal antibody recognizing a neoantigen in the human C3'g' fragment was used as the capture antibody in an enzyme-linked immunosorbent assay. Detection was made using a polyclonal rabbit anti-human C3d and a peroxidase-linked anti-rabbit Ig antiserum. The activity in normal human EDTA plasma was found to be 3% of that in a zymosan-activated serum pool. Fractionation experiments revealed that most of the activity in normal plasma, in vivo activated plasma and in vitro activated serum eluted in one peak with a molecular weight corresponding to iC3b. A positive correlation (P < 0.01) was found between the present assay and a previously established two-step C3d ELISA both with respect to normal plasma, individual patient samples and consecutively drawn samples following artificial in vivo activation. Complement activation assays based on specific antibodies to 'activation antigens' should be preferred whenever available since they enable direct, rapid and specific quantification of the actual fragment(s). Key words: C3; Complement activation; Neoantigen; ELISA
Introduction
Assays for quantitative detection of C3 activation products have mainly been based on radial immunodiffusion (Perrin et al., 1975), electrophoretic (Brandslund et al., 1981; Bourke et al., 1982) or nephelometric (Vergani et al., 1983) methods. In our experience most of these methods are not sufficiently sensitive and reliable to detect minor fluctuations of C3 activation products under in vivo conditions. A highly sensitive enzymelinked immunosorbent assay (ELISA) was thereCorrespondence to: T.E. Mollnes, Institute of Immunology and Rheumatology, Fr. Qvamsgt. 1, N-0172 Oslo 1, Norway.
fore constructed to quantify C3d/C3dg (Mollnes, 1985a). This assay was found convenient to evaluate C3 activation in different clinical situations (Mollnes, 1985b; Mollnes et al., 1986). However, as with previously described assays this one was also hampered by a precipitation step necessary to eliminate the influence of large C3 fragments since the antisera available reacted against native epitopes. Recently, a new principle for the detection of activation products has been introduced based on monoclonal antibodies against neoantigens expressed in the activation products. The fluidphase terminal SC5b-9 complement complex (TCC) (Mollnes et al., 1985a) and the C3 activa-
0022-1759/87/$03.50 © 1987 Elsevier Science Publishers B.V. (Biomedical Division)
202
tion products iC3b and C3c (Aguado et al., 1985) have been detected directly. Neither precipitation nor fractionation were required. The aim of the present study was to develop a C3 activation assay based on a monoclonal antiC3'g' neoantigen antibody and to compare this assay with the one previously published from our laboratory (Mollnes, 1985a). The monoclonal anti C3'g' antibody has been extensively characterized previously (Lachmann et al., 1982) as reacting against a 'g' epitope expressed in iC3b, C3dg and C3g. This antibody has been used with success to isolate complement-fixing immune complexes by coupling the antibody to Sepharose (Samuel et al., 1986).
Materials and methods
Antibodies Ascitic fluid of the rat monoclonal anti-C3'g' neoantigen antibody (clone 9) was preserved in 40% glycerol and used unfractionated in all experiments. The specificity of this antibody has been described in detail previously (Lachmann et al., 1982; Samuel et al., 1986). Rabbit anti-human C3d Ig fraction (lot l l l D ) was obtained from Dako Immunoglobulins, Copenhagen, Denmark; rabbit anti-human C3c (lot 153321B) from Behringwerke, Marburg, F.R.G., and peroxidasecoupled anti-rabbit Ig(lot 16) and anti-rat Ig (lot 10) from Amersham, U.K.
Preparation of standard and samples Activation of a normal human serum pool (n = 20, healthy blood donors) was made by adding 100 mg of zymosan A (Sigma, St. Louis, MO) to 10 ml of serum, which was incubated under continuous mixing at 37°C for 30 rain, spun at 27000 x g (Sorvall RC5B) for 30 min and stored at - 7 0 °C in 100 #1 aliquots. This activated serum was used as positive control and standard in the assay. Ethylenediaminetetraacetic acid (EDTA) plasma samples from 80 healthy blood donors (40 males and 40 females) and 20 patients were separated immediately and stored at - 7 0 ° C until tested to avoid in vitro activation. Fractionation of samples was made by passing
0.5 ml sample through a Sephacryl S-300 (Pharmacia, Sweden) column (2.5 × 100 cm) in a 0.05 M Tris/HC1 buffer, pH 7.60, containing 0.2 M NaC1, 0.01 M EDTA and 0.02% NaN 3.
Detection of C3'g' neoantigen activity in ELISA In this assay N U N C Immunoplates II (Copenhagen, Denmark) were coated with the monoclonal anti-C3'g' neoantigen antibody, diluted in phosphate-buffered saline (PBS), at 4 ° C for at least 48-h. The well volume was 100 ~tl for all steps and the optimal dilutions of the reagents are given in the results section. The plates were washed in a Dynawasher three times between each incubation using PBS containing 0.1% Tween 20 (Sigma, St. Louis, MO). This was also the buffer used for all the following incubations except for that with the substrate. The antigen was always added in triplicate and incubated at 37°C for 45 min. Then anti-C3d (Dako) was added and finally anti-rabbit Ig conjugated with peroxidase. Each incubation was performed at 37°C for 45 min. The substrate was 2,2'-azino-di-(3-ethylbenzthiazoline sulfonic acid) (ABTS, Boehringer Mannheim, F.R.G.), diluted in 0.1 M sodium acetate buffer, pH 4.0 to 0.18 mg/ml. 0.8 /~1 3% H202 was added per ml buffer immediately before use. The optical density (OD) was read on a Dynatech MR 580 at 405 nm and the results were analyzed by an Apple IIe computer with the Dynatech Immunosoft programme. Alternatively, a qualitative examination of the antigen was made by coating the ELISA plates with fractions (diluted 1:50 in PBS) after gel filtration of plasma and activated serum. AntiCYg' neoantigen diluted 1:5000 and anti-rat Ig diluted 1:1000 was then added. Incubations and development were as described above for the quantitative assay.
Stat&tical methods Correlations were calculated using Spearman's rank correlation coefficient. P values are based on two-tailed tests. Results
The monoclonal anti-C3'g' antibody was initially tested for its reactivity in ELISA using a
203
screening system for anti-complement neoantigens described in detail elsewhere (Mollnes et al., 1985b). The antibody reacted strongly with zymosan-activated serum and weakly with normal plasma, confirming the previously described neoantigen activity (Lachmann et al., 1982).
Quantitative ELISA for the C3'g" neoantigen A double-antibody ELISA with the monoclonal anti-C3'g' neoantigen antibody (clone 9) as the capture and a polyclonal anti-C3d antiserum as the indicator was constructed as described in the materials and methods section. Optimal dilutions of the reagents giving the highest signal/noise ratio were as follows: ascitic fluid of anti-C3'g' diluted 1:10000, zymosan-activated serum used as standard diluted from 1:400 (two-fold, eight steps), normal plasma and patient samples diluted 1:500, anti-C3d diluted 1:2000 and anti-rabbit Ig-peroxidase diluted 1 : 1000. The specificity of the double-antibody assay was examined by omitting the coating step (clone 9), and performing the rest of the assay as described above. These experiments were made on the same plate by coating half of the plate and
leaving the other half uncoated. Three expei'i, ments were performed using triplicate samples in each. Strong reactions using activated serum were observed on the coated part of the plate in contrast to the uncoated (Fig. 1). Normal human plasma (NHP) did not react with the uncoated part of the plate, whereas a significant signal was obtained in the assay. The standard was defined as containing 1000 arbitrary units (AU)/ml. The median value after ten experiments was found to be 30 A U / m l for NHP and the coefficient of variation (CV) was 0.19.
Fractionation of standard, normal plasma and patient plasma To examine which fragment(s) contributed to the activity in the assay and to compare the reaction patterns of in vitro and in vivo activated samples, the activated serum, the normal plasma pool and a patient sample were fractionated using a Sephacryl S-300 column. Fractions from activated serum and normal plasma were coated on ELISA plates and examined with the anti-C3'g' neoantigen, anti-C3d, and anti C3c antisera. One common peak was observed in the range of 150-200 kDa for all three antibodies. An additional low-molecular peak (approximately 30
O.D4OSnm 08 O.D. 4o5nm 0.6
e--e
0.8
\
\
Ill
\,
i--.
Anti-C3d
=--~
Anti-C3c
\
•
0.4_
\\
Anti-C3"g"
0.6\
\•
xs\ --
\
0.2
\
04
\e ~
?-%
8
0.2 0_
[
[ 125
~° ~ - - ~ I I I I I 31.2 7.8 Dilution of s t a n d a r d x l 0 -5
o I 1.9
ooe ooc I B.G.
\
I
Fig. 1. Two-fold dilutions of zymosan-activated serum (standard) tested in a double-antibody ELISA with the monoclonal anti-C3'g' as capture antibody and polyclonal anti-C3d as the second antibody (closed circles). Control experiments were made by omitting the capture antibody (open circles). Median of triplicates from three experiments are indicated. Values for background (B.G., omitting the antigen) are also indicated.
2= 0--
II
10
I
I ~ I Vo
I
I I 22
I~1~1 IgG AIb
3161
I I I Ino. Fraction
Fig. 2. ELISA plates were coated with fractions obtained after gel filtration of zymosan-activated serum. They were then reacted with anti-CYg' neoantigen, anti-C3d and anti-C3c antibodies. Elution of void volume fraction (V0), IgG and albumin (Alb) is indicated (mean of duplicate).
204
kDa) was observed in activated serum with antiC3d which was not seen using anti-C3'g'. Neither did anti-C3'g' react with other low molecular weight fractions (Fig. 2). These experiments were done four times under slightly different conditions using different preparations of activated serum giving essentially identical results each time. Normal human plasma revealed only one peak with these three antibodies corresponding to large-fragment C3 molecules. Control experiments omitting the second antibody layer were all completely negative. When zymosan-activated serum was incubated at 37 °C for 1 week a marked increase in the low molecular weight peak with anti-C3d was observed (Fig. 2). However, no activity was observed in this region using anti-C3'g'. Fragments were then examined using anti-C3'g' as capture antibody and anti-C3d (known to react with all C3 fragments containing d-epitopes) in the second antibody layer as described for the quantitative assay. Only one peak, corresponding to 150-200 kDa was observed both in activated serum, in normal plasma, and in two different patient samples (infectious diseases). However, in activated serum incubated at 37°C for 18 h a minor peak corresponding to the molecular weight of C3dg was observed in addition to the iC3b peak. The C3dg peak disappeared and the iC3b peak diminished after further incubation for 1 1.0-
OD 405rim
,,
," P a t i e n t
•
• NHP
week. Fig. 3 shows the activity in fractionated normal human plasma and the patient sample. The curves should not be compared quantitatively since the large number of fractions did not allow set-up on the same plate. Unfractionated, the patient plasma contained six times higher A U / m l values than the normal pool. Correlation with an established C3d / dg assay The double-antibody assay based on the antiC3'g' neoantigen antibody was correlated with a conventional assay for detection of C3d/C3dg (Mollnes, 1985a). This assay was based on polyethylene glycol (PEG) precipitation of large C3 fragments before the supernatant was examined for C3d in ELISA. That assay had previously been shown to correlate closely with a double-decker rocket immunoelectrophoretic method for C3d/dg quantification (Brandslund et al., 1981), but had the advantages of higher sensitivity and greater rapidity. Samples from 20 different patients in a heterogenous patient population were selected and examined in both ELISA assays at the same time. A close positive correlation was found (r~- 0.83, P < 0.0002) (Fig. 4). Furthermore, a series of plasma samples obtained from a patient undergoing extracorporeal
200_
C3d/dg I AU/ml
150--
08100--
~,,,
06-
/ • : ! elel •
50
04--
/'
02--
0 0
I 200
I 400
I
C3"g" n e o a n t i g e n [ AU/mr 600
Fig. 4. Correlationbetweentwo differentassaysto detect C3
O_ I
u 8
IAI Vo
I
i 12
i
f
I
uAI
I
IgG
18
u ,L AIb
[
n 22
i
I
~
n
Fractaon
no
Fig. 3. Fractions obtained after gel filtration of normal human plasma and a patient's plasma sample were examined in the quantitative double-antibody ELISA described. Elution of void volume fraction (V0), IgG and albumin (Alb) is indicated (median of triplicate).
activation evaluated using 20 plasma samples from a heterogeneous patient population. The abscissa shows C3 conversion (C3'g' neoantigen) measured by the method described in this paper and the ordinate shows C3 conversion measured by a previously described ELISA method for C3d/dg quantification. Upper reference limits for the assays are indicated by arrowheads (median of triplicate).
205 C3"g"neoantigen AU/rnl . _ _ . _100
C3d/dg AU/ml 300 . . . .
_80
/ /,
200
x\ '\ _60
,; ,; /
\
10C
l !
\./
I
I
/
,~2,
... o, /i ,'// i
,~ j',,,'( ....... \
"/
"z
/
I I Consecutively
_40
,, ,\
'J
I drawn
I
I I samples(n=10)
t
I
_ 20
I i
_0
Fig. 5. Correlation between the C3'g' neoantigen assay and the C3d/dg assay evaluated by a series of samples obtained consecutively from a patient undergoing extracorporeal circulation examined in both assays at the same time (median of triplicate).
circulation were examined in both assays. The complement activation pattern of this patient has been described in detail elsewhere (Mollnes, 1985b). Again, a close positive correlation was demonstrated between the two assays (r s -- 0.92, P < 0.01) (Fig. 5). Lower detection limit and reference range The lower detection limit (background + 2 SD) was found to be less than 0.02 AU/ml, corresponding to 0.002% conversion of the activated serum. The normal human plasma pool (n = 20) contained 30 AU/ml, corresponding to 3% conversion. Plasma samples from 80 healthy blood donors, 40 females and 40 males, were examined and the following reference ranges (2.5-97.5 percentile) could be defined: females 23-58 A U / m l and males 27-66 AU/ml.
Discussion
Detection of neoantigens expressed only in activated components is theoretically the most reasonable way to evaluate complement activation. These epitopes reflect the degree of activation of the corresponding component and can be quantified directly without the influence of native comPonents. Precipitation or fractionation of
samples is therefore not required. The risk of false positives due to in vitro activation should also be markedly reduced in one-step assays. Recently, two monoclonal antibodies directed against neoantigen epitopes in C3c and C3d were successfully used in assays of complement activation and immune complex formation (Aguado et al., 1985). The complement activation assays were, however, not correlated with conventional assays. We demonstrate in the present paper that the monoclonal anti-C3'g' neoantigen antibody (clone 9) can be used to design a reliable C3 activation assay which correlated significantly with a previously established C3d/dg method, but was not hampered by the disadvantages of conventional assays. A major advantage with the present assay is the high sensitivity (lower detection limit < 0.002% of activated serum). The sensitivity of the previously described C3 neoantigen assay was unfortunately not stated in the paper (Aguado et al., 1985), but estimations from their dilutions seem to indicate that the assay presented here is about ten-fold more sensitive. This has obvious implications for the examination of biological fluids with a low protein concentration as cerebrospinal fluid, which has been examined unconcentrated and diluted with success using the present assay (unpublished data). From a functional and pathophysiological point of view it can be argued that the total amount of a neoantigen epitope present may be as important as the concentration of a single particular fragment. The present assay would theoretically detect all fragments expressing a combination of the 'g' neoantigen and d-epitopes. In the present study we found that most of the activity detected is of a molecular weight corresponding to that of iC3b, which is known to express this epitope (Lachmann et al., 1982). We were hardly able to demonstrate fragments corresponding to C3dg or C3g. The latter is not surprising since the 'g' fragment very rarely reacts with polyclonal anti-C3 antibodies (Lachmann et al., 1982). The most likely explanation for the lack of reactivity against anti-C3dg in most of the samples examined is simply that no C3dg was present since the 'g' epitope is well known to be exposed in C3dg (Lachmann et al., 1982; Davis et al., 1984), and since we demonstrated C3dg in
206
activated serum incubated at 37°C for 18 h. It is easy to accept that serum activated in vitro under different conditions do not necessarily contain C3dg. However, it is surprising that C3dg was not demonstrated in the patients' plasma since it has become widely accepted that this fragment is the major physiological breakdown product after complement activation. These observations emphasize that quantification of 'activation' epitopes like the CYg' neoantigen, irrespective of which fragments the epitope is exposed on, may be more reliable as a screening method for in vivo complement activation than determination of one single fragment. Although most of the 'g' neoantigen activity corresponded to the molecular weight of iC3b, it cannot be excluded that this elution peak also contained other fragments expressing this epitope. C3d/C3dg may, under physiological conditions, be associated with albumin or other proteins (Johnson, 1984; Teisner et al., 1984), which may also explain the minor high molecular weight peak observed after the void volume fraction in Fig. 2. This is, however, of minor importance, since the control experiments definitely showed that all the activity in the double antibody assay was dependent on the 'g' neoantigen. We therefore decided to denote this assay a C3'g' neoantigen assay. For the same reasons we Prefer arbitrary units instead of percent. Arbitrary units are limited to that particular assay, in this case reflecting the total amount of CYg' neoantigen present, and thereby false comparisons between different types of assays can be avoided. We conclude that monoclonal antibodies against neoantigens are important tools in the development of assays to detect and quantify complement activation in a reliable, direct, highly sensitive and rapid fashion. There is now a need for other antibody specificities to replace conventional activation assays.
Acknowledgements We are indebted to Ms. Grethe Bergseth for her excellent technical assistance and to Ms. Karl Bertelsen for typing the manuscript.
References Aguado, M.T., Lambris, J.D., Tsokos, G.C., Burger, R., BitterSuermarm, D., Tamerius, J.D., Dixon, F.J. and Theofilopohis, A.N. (1985) Monoclonal antibodies against complement C3 neoantigens for detection of immune complexes and complement activation. Relationship between immune complex levels, state of C3, and numbers of receptors for C3b. J. Clin. Invest. 76, 1418. Bourke, B.E., Moss, I.K. and Maini, R.N. (1982) Measurement of the complement C3 breakdown product C3d by rocket immunoelectrophoresis. J. Immunol. Methods 48, 97. Brandslund, I., Siersted, H.C., Svehag, S.-E. and Teisner, B. (1981) Double-decker rocket immunoelectrophoresis for direct quantification of complement C3 split products with C3d specificities in plasma. J. Immunol. Methods 44, 63. Davis, A.E., Harrison, R.A. and Lachmann, P.J. (1984) Physiological inactivation of fluid phase C3b: isolation and structural analysis of C3c, C3dg (alpha 2D), and C3g. J. Immunol. 132, 1960. Johnson, U. (1984) The influence of polyethylene glycol precipitation and of aging plasma and serum on the measurement of C3dg/d. Acta Pathol. Microbiol. Immunol. Scand. Sect. C. 92, 193. Lachmann, P.J., Pangburn, M.K. and Oldroyd, R.G. (1982) Breakdown of C3 after complement activation. Identification of a new fragment C3g, using monoclonal antibodies. J. Exp. Med. 156, 205. Mollnes, T.E. (1985a) Quantification of the C3d split products of human complement by a sensitive enzyme-linked immunosorbent assay. Scand. J. Immunol. 21,607. Mollnes, T.E. (1985b) Early- and late-phase activation of complement evaluated by plasma levels of C3dg and the terminal complement complex. Complement 2, 156. Mollnes, T.E., Lea, T., Froland, S.S. and Harboe, M. (1985a) Quantification of the terminal complement complex in human plasma by an enzyme-linked immunosorbent assay based on monoclonal antibodies against a neoantigen of the complex. Scand. J. Immunol. 22, 197. Mollnes, T.E., Lea, T., Harboe, M. and Tschopp, J. (1985b) Monoclonal antibodies recognizing a neoantigen of poly(C9) detect the human terminal complement complex in tissue and plasma. Scand. J. Immunol. 22, 183. Mollnes, T.E., Lea, T., Mellbye, O.J., Pahle, J., Grand, O. and Harboe, M. (1986) Complement activation in rheumatoid arthritis evaluated by C3dg and the terminal complement complex. Arthritis Rheum. 29, 715. Perrin, L.H., Lambert, P.H. and Miescher, P.A. (1975) Complement breakdown products in plasma from patients with systemic lupus erythematosus and patients with membranoproliferative or other giomerulonephritis. J. Clin. Invest. 56, 165. Samuel, D.J., Amlot, P.L., Shepherd, P. and Lachmann, P.J. (1986) An efficient one-step method for isolating immune complexes from whole serum using a monoclonal anti-C3g affinity immunosorbent. Clin. Exp. Immunol. 65,458. Teisner, B., Petersen, N.E., Folkerson, J.J. and Hau, J. (1984)
207 Crossed immunoelectrophoretic analysis of split products of the third complement factor (C3d) following in vivo activation of the complement system. Electrophoresis 5, 84. Vergani, D., Bevis, L., Nasaruddin, B.A., Mieli-Vergani, G.
and Tee, D.E.H. (1983) Clinical application of a new nephelometric technique to measure complement activation. J. Clin. Pathol. 36, 793.