APMIS 1001615-622, 1992
Antigenic change in a human IgG4-specific CH3epitope upon binding of a monoclonal antibody against a neighboring IgG4-sPecific epitope ANNA KRISTIN ROLSTAD, TERJE E. MICHAELSEN and JAN KOLBERG Department of Immunology, National Institute of Public Health, Oslo, Norway
Rolstad. A. K., Michaelsen, T. E. & Kolberg, J. Antigenic changes in a human IgGCspecific CH3 epitope upon binding of a monoclonal antibody against a neighboring IgG4-specific epitope. APMIS 100: 615-622, 1992. Two sets of monoclonal antibodies (mAbs) probably reacting with two different epitopes in the CH3 domain of the human IgG4 molecule were studied. We observed that the commercially available mAb HP 601 1 inhibited the antigen binding of the three mutually inhibitable mAbs, 40-A2, 41-E8 and 43F11 (40-series), made by us. However, the 40-series mAbs, including those with similar affinity such as mAb HP6011, were not able to inhibit mAb HP 601 1. When the 40-series mAbs were preincubated with IgG4, the mAb HP 6011 could partially displace these antibodies. This one-way inhibition indicates that upon binding mAb HP 601 1 changes the antigenic structure of the IgG4 molecule by disrupting the epitope for the 40-series mAbs. A steric hindrance of this epitope by mAb HP 601 1 is more unlikely, since the small Fab fragment of mAb HP 601 1 also inhibited the reaction of the 40series mAbs. Key words: Monoclonal antibodies; antigenic change; IgG4. Terje E. Michaelsen, Department of Immunology, National Institute of Public Health, Geitmyrsveien 75, 0462 N-Oslo 4, Norway. I
The four human IgG subclasses are structurally closely related and show more than 95% amino acid sequence identity between homologous domains, while the hinge regions show only 6%70% homology in sequence and vary considerably in length (8, 11, 15). There is also a difference in effector functions between the subclasses (1). Because of the great degree of homology, the individual IgG subclasses only express a few subclass-specific epitopes, and hence antibodies specific for these epitopes are difficult to produce. Some of these subclassspecific epitopes are probably directly or indirectly involved in effector function differences between the subclasses. Amino acid residues unique for human IgG4 are present in the CH1,
CH2and c H 3 domains of the molecule (1 I), and monoclonal antibodies (mAbs) specific for the IgG4 isotype have been made against each of these domains (10). In this report we have studied the expression of human IgG4 subclassspecific epitopes in the c H 3 region by employing mAbs produced by us and commercially available mAbs. Unexpectedly, inhibition experiments showed that the commercially available mAb HP 601 1 was able to block the binding of the whole group of 40-series antibodies made by us, but not vice versa. The 40-series antibodies cross-inhibited each other, and thus apparently react with the same epitope. MATERIALS AND METHODS
Received September 3, 1991. Accepted January 16, 1992.
Production of monoclonal antibodies Three different hybridoma cell lines were made in
615
ROLSTAD et al.
one fusion experiment. BALB/c mice were immunized with 20 pg IgG4 Fc in 0.25 ml of PBS mixed with 0.25 ml Freund’s incomplete adjuvant, followed by booster injections two and four weeks later with the same immunogen mixture. Four weeks later the mice were given 20 pg IgG4 Fc in PBS intraperitoneally on days one, two and three. On day five spleen cells were fused with NSO myeloma cells by standard methods with polyethylene glycol 1450 (Eastman Kodak Company, USA) as the fusion agent. Selected hybrid cells were cloned by limiting dilution and injected into pristan-primed BALBlc mice for production of antibody-containing ascites. One different hybridoma cell line, antibody 89-C7, was made in another fusion experiment using the same method as above but with intact human IgG4 as immunogen. The monoclonal antibody HP 601 1 (RJ4) specific for cH3 IgG4 was delivered by the WHO. The monoclonal antibody GB7B specific for CH2 IgG4 was kindly provided by Dr R. Jefferis, University of Birmingham, England (see Table 1). Isotype determination of the monoclonal antibodies The monoclonal antibodies purified on Protein A Sepharose (Pharmacia, Uppsala, Sweden) were isotyped using an ELISA technique. Sheep antibodies specific for mouse IgG isotypes biotinylated by Biotin-X-NHS (Calbiochem, USA) according to Wofsy (20), mixed with streptavidin-conjugated alkaline phosphatase (ALP) (Sigma, USA, type VII) (5), were used as specific reagents. The mAbs 40-A2, 41-E8 and 43-FI1 were of IgG2a isotype (Table 1). MAb 89-C7 was typed as IgGl . The mAbs HP 601 1 and GB7B were also of IgGl isotype (10). Tests .for antibody specificity The specificity of the monoclonal antibodies was tested in an ELISA system on microtiter plates (Nunc-Immuno-Plate, Maxisorb, Denmark) coated with myeloma proteins of the four different human IgG isotypes (1 pg/ml, 150 pUwe11). The antibodies were shown to be specific for IgG4 subclass (Table 1). The epitope specificities were further evaluated by
testing them on microtiter plates coated with IgG4 Fab-, Fc- and pFc’-fragments (1 pg/ml, 150 pl/well). The fragments were prepared by papain and pepsin digestion as described (14). Our monoclonal antibodies in the 40-series and HP 6011 all reacted with IgG4 pFc’, while 89-C7 reacted with the IgG4 Fab fragment and GB7B with IgG4 Fc but not with IgG4 pFc’ (IgG4 Fc, non pFc’) (Table 1). Quantitation of IgG in ascites The IgG concentration in ascites was measured by single radial immunodiffusion according to Mancini et al. (12), employing sheep antiserum specific for mouse IgG. The sheep antiserum used was shown to react equally well with all mouse IgG subclasses. Antibody affinity Antibody affinities of the different monoclonal antibodies were compared by testing on microtiter plates coated directly with a human IgG4 myeloma protein (1 pg/ml, 150 pl/well), and plates coated indirectly with IgG4 (1 pg/ml) by catching with sheep anti-human IgG Fab (1 pg/ml). After incubation with dilutions of monoclonal antibodies for two h at 37°C and washing, biotinylated sheep anti-mouse IgG antibodies mixed with streptavidin-ALP were added and incubation continued. This biotinylated sheep antimouse IgG conjugate was shown to react equally well with IgGl and IgG2a (data not shown). After further washing the enzyme substrate disodium p-nitrophenyl phosphate (NPP) (1 mg/ml in 10% diethanolamine buffer, pH 9.8, containing 10 mM MgClJ was added and incubation continued for 30-60 min before absorbance at 410 nm was recorded (Dynatech MR 700, Germany). The high affinity antibodies are seen to the left in Fig. 1; the low affinity antibodies to the right. Antibody affinity was also evaluated using an inhibition test where the concentration of IgG4 required for 50% inhibition of the antibodies was taken as a measure of affinity (2).
ELISA inhibition tests Cross-inhibition of the different monoclonal antibodies was performed in two different ways using microtiter plates coated with IgG4 myeloma protein (1 pg/ml, 150 pl/well) in both cases. Various concen-
TABLE 1. SDecificitv o f monoclonal antibodies Monoclonal Isotype Subclass Fragment (domain) antibody specificity specificity 40-A2 IgG2a IgG4 PFc’ (cH3) IgG2a 4 1-E8 IgG4 pFc‘ (cH3) 43-F11 IgG2a IgG4 PFc’ (cH3) HP 601 I IgG 1 IgG4 pFc‘ (cH3) (RJ 4) 89-C7 IgG 1 IgG4 Fab (CHI) IgG4 Fc non pFc’ (C,2) GB7B IgG 1
616
ANTIGENIC CHANGE IN IgG4
trations of the hybridoma antibodies to be tested for inhibitory capacity were added to the coated microtiter plates and incubated for one h. The antibody which we wanted to inhibit was then added as a biotin conjugate at an optimal concentration and the incubation continued for two h. Following this the plate was washed and streptavidin-ALP added, and there was a two h incubation period. After further washing the enzyme substrate NPP was added and incubation continued for 30-60 min before absorbance at 410 nm was recorded. The other inhibition test took advantage of the fact that the antibodies HP 601 I and 89-C7 were of the
g
0.90
-
0.72
-
0
A
F
t 0.54
-
0.30
-
0.18
-
0.00
-
0
E
a
51 a a
.&mL
0.001
0.01
0.1
1
10
100
mouse IgGl isotype, while our antibodies in the 40series were of the IgG2a isotype. The inhibitory antibody and the antibodies to be inhibited were discriminated simply by developing the microtiter plate with sheep antibodies specific for the respective mouse isotypes. Various concentrations of the hybridoma antibodies to be tested for inhibitory capacity were added to IgGCcoated plates and incubated for one h before the second antibody at a fixed optimal concentration was added, and incubated further for two h. After washing, biotinylated sheep anti-mouse IgGl and anti-mouse IgG2a, respectively, mixed with streptavidin-ALP were added and incubated for two h. Incubation with the substrate NPP and reading of the plates were performed as described above. This procedure could measure inhibitory capacity and binding activity of both antibodies simultaneously by parallel development with anti-mouse IgGl and IgG2a antibodies. ELISA replacement tests The antibodies’ ability to replace each other was tested in ELISA by incubating various concentrations of one antibody, washing away excess antibodies and adding various concentrations of the second antibody. The second antibody’s ability to replace the first antibody was measured by developing the ELISA plates with antibody conjugates specific for the isotype of the first antibody.
1000
Concentration >f antibodies, pg/ml
RESULTS 1.1 0.9
g
-
0.8
-
2
0.7
1.0
0
o 0.6 g 0.5 0
*::
0.4
-
a 0.3 -
a
0.2
0.1
-
0
0.001
0.01
0.1
1
10
100
1000
Concentration of antibodies, pg/rnl
Fig. 1. Dose response curves of the antibodies, reflecting the antibody affinity of the mAb. Microtiter plates were coated with human IgG4 (Hen.), 1 pg/ ml directly (A) or indirectly with sheep anti-human IgG Fab (B) and dilutions of the mAb added. After incubation and washing, biotinylated sheep antimouse IgG antibodies were added and incubation continued before the enzyme substrate NPP was added.
H P 6011 inhibits the 40-series of antibodies which mutually inhibit each other The 40-A2, 41-E8 and 43-Fll antibodies could mutually inhibit each other in ELISA inhibition tests (Fig. 2A, B and C). The mAb HP 601 1 also efficiently inhibited mAbs 40-A2, 41-E8 and 43-F11 (Fig. 2A, B and C). The Fab fragment of mAb HP 6011 inhibited the 40series of antibodies (data not shown). On the other hand, the IgG4 C,Zspecific GB7B antibody did not compete with the 40-series antibodies (Fig. 2A, B and c ) , or with HP6011 (Fig. 2D), while as a positive control it was shown to inhibit itself (Fig. 2E). The mAb HP 6011 was shown to inhibit the native unconjugated 40series of antibodies by using the other inhibitory test where we employed sheep anti-mouse IgGl and IgG2a antibodies, respectively. From Fig. 3 it can be seen that mAb HP 6011 even inhibits the 40-A2 antibody which has the same affinity as HP 6011 against the antigen, IgG4 (Fig. 1).
617
ROLSTAD et al.
60 11, and employing sheep anti-mouse IgG 1 and IgG2a antibodies, respectively. No inhibition by the 40-series against HP 601 1 was observed even with the high affinity antibody 40-A2 (Fig. 4).The mAb 89-C7 of isotype IgGl, reacting with IgG4 Fab, was included as a negative control (Fig. 4).
The 40-series of antibodies do not inhibit HP 6011 The 40-A2,41-E8 and 43-F11 antibodies were tested for inhibitory capacity against HP 6011. None of the antibodies in the 40-series inhibited HP 601 1 (Fig. 2D). The inhibitory capacity was also tested using the native unconjugated HP
0.75 0.60
0.45
0.6
-
n 0.30 L
0.4
-
0.2
-
0
v
e
m 0
43-Fll
Y)
.n
U
0.15
&
0.00 0.01
g 0
1
0.1
10
0.0 1
100
0.1
1
10
100
1
10
100
1
1.6
r
.t ID V C
1.2
fz
s
0.6
a
0.4
A
40-A2
P
a 0.4
0 .o
0.0
0.0 1
1.oo
5 0
- A 41-E8 43-F11 - *0 G878 HP6011
I
1
0.1 '
' " 1 " 1 ' 1
'
"'""1
100
10 '
"1""1
""""I
F
w
0.60
c
a
fz0
0.40
2
0.20
H
0.00
t
4042 A 4l-EO P 43-Fl1 0 HP6Oll 0078
*
0.01
0.1
1
10
100
Concentration of antibodier, pg/rnl
618
0.1
Concentratlon of antibodies, pg/rnl Fig. 2. ELISA inhibition against biotin-labeled
0.80
V
0.01
mAb 40-A2 (A), mAb 41-E8 (B), mAb 43-Fll (C), MAb HP6011 (D) and GB7B (E). Microtiter plates were coated with human IgG4 (Hen.), I pg/ml. The different antibodies tested for inhibitory capacity were added at the concentrations indicated in the figure, and incubated for one h before an optimal concentration of biotinylated mAb 40-A2 (1 : 16,000), mAb 41-E8 (1 : 8,000), mAb 43-F11 (1 :20,000), MAb HP6011 (1 20,000) and mAb GB7B (1 : 8,000) in A, B, C, D and E, respectively, was added and incubation continued for two h. Plates were further developed as described in Materials and Methods.
ANTIGENIC CHANGE IN IgG4
HP 6011 and the 40-series bind to different sets of epitopes
ELISA inhibition tests revealed that the HP 601 1 and 40-A2/41-E8/43-F11 antibodies could bind simultaneously provided that the HP 601 1 antibody was added after mAb 40-A2/41-E8/ 43-Fll (Fig. 4). Furthermore, the 40-A2 antibody was not substituted with HP 6011 antibody, but stayed bound to the IgG4 antigen under these conditions (Fig. 4). High concentrations of mAb 40-A2 have no influence on binding of either mAb HP 6011 or mAb 89-C7
E
!2 1.0 d
' I0 0.8
0.6
0.4 0.2
The IgG4 epitopes are not disturbed by plastic adherence In some tests, the IgG4 antigen was caught on the ELISA plates with the Fab part oriented inwards by coating with sheep antibodies specific for the Fab part of human IgG. We observed the same kind of one-way inhibition under these experimental conditions (data not shown). The 40-series antibodies are not subclones There was a marked difference in the electrophoretic mobility in agarose gel electrophoresis (data not shown), indicating that they were not subclones. Furthermore, the antibodies showed different antibody affinity (Fig. l), also indicating that they were not subclones.
1.2
88
(Fig. 4). HP6011 could partly replace 40-A2 (Fig. 5). The 40-A2 antibody could not replace HP6011, and 89427 could not replace 40-A2 in similar experiments (data not shown).
1 1
0 1 :32000
1:3200
1:320
Dilution of HP 6011 First mAb/ inhibiting
-E-
HP6011
Second mAb/ inhibited
none
4HP6011
40-A2
+ 40-A2
none
+ HP6011
none
-& HP6011
40-A2
+ 40-A2
none
Fig. 3. ELISA inhibition against mAb 40-A2 attempted by mAb HP 6011. Microtiter plates were coated with human IgG4 (Joh.), 1 pg/ml. Different dilutions of mAb HP 6011 (mouse IgG1) were added and incubated for one h before mAb 40-A2 (mouse IgG2a) (1 : 128,000) was added and incubation continued for two h. Open symbols indicate further developing with sheep anti-mouse IgG2a antibodies and filled symbols developing with sheep anti-IgG1 antibodies.
The one-way inhibition is not due to differences in affinity MAb 40-A2 and HP 6011 showed equally high affinity; mAb 43-Fll showed somewhat lower affinity, while mAb 41-E8 showed the lowest affinity (Fig. 1). The results were the same when unpurified ascites and purified antibodies were employed. A similar pattern was seen with two different IgG4 proteins (Hen. and Joh.). The measurements when using plates coated directly and indirectly (catching) with IgG4 revealed two differences: the 41-E8 antibody had a relatively higher affinity when indirectly coated IgG4 was employed (Fig. 1A) and the 89-C7 antibody had lowest affinity with directly coated IgG4 (Fig. 1B). Otherwise the results were similar for directly and indirectly coated IgG4 (Fig. 1A and 1B). Using an inhibition test described by Devey et al. (2), the same antibody affinity pattern of the antibodies was obtained as referred to above (data not shown).
DISCUSSION By immunizing BALB/c mice with Fc fragments of an IgG4 myeloma protein, we have made three hybridoma antibodies: mAb 619
ROLSTAD et al.
40-A2, 41-E8 and 43-Fll. They were all shown to be specific for human IgG4 pFc', and they could all mutually inhibit each other and thus probably react with the same or closely situated epitopes. Another commercially available IgG4 pFc'-specific antibody, HP 6011 (RJ4), efficiently inhibited mAb 40-A2, 41-E8 and 43-Fl l, while none of the latter antibodies could inhibit mAb HP 601 1. However, when human IgG4 was preincubated
I
1.2 1 .o
z
0.8
R
~~
1
with the 40-series antibodies and the excess antibodies washed away, mAb HP 6011 could partly replace the 40-series antibodies. The reverse replacement could not be achieved. These observations indicate that mAb 40-A2/ 41-E8/43-Fll and mAb HP 6011 react with two different sets of epitopes on the IgG4 molecule. There is, however, the possibility that HP 6011 can inhibit the 40-series antibodies, but not vice versa, if HP 601 1 has a much higher affinity than the other antibodies. This was not the case, at least for mAb 40-A2, which was tested to have nearly the same affinity as mAb HP 6011 (Fig. 1). Furthermore, the mAb 41-E8 with the lowest antibody affinity can inhibit the other two antibodies in the 40-series which have higher affinity. There may be at least two possible explanations for this peculiar one-way inhibition: 1) the first reacting mAb is sterically hindering the access of the second mAb to its epitope, or 2) the first mAb induces a conformational change in the antigen molecule upon binding, thereby changing the structure of the second epitope sufficiently to prohibit binding to the second antibody (the 40-series antibodies). We favor the
[ I 1 \I 1
t
o.2 01:10000
1:lOoo
1:loo
Dilution of 4042
--
First mAb/ inhibiting
Second mAb/
inhibited
-ff 40-A2
none
++40-A2
HP 601 1
-A- 40-A2
89-67
_--E
........
...... ......................... 8
............ -0- 40-A2 -C 40-A2
40-A2
none HP 6011 89-67
Fig. 4. ELISA inhibition against mAb H P 6011 and
89-C7 attempted by mAb 40-A2. Microtiter plates were coated with human IgG4 (Hen.), 1 pg/ml. Different dilutions of mAb 40-A2 (mouse IgG2a) (ascites) were added and incubated for one h before mAb HP 601 1 (1 : 30,000) and 89-C7 (1 : 10,000) (both mouse IgG l), respectively, were added and incubation continued for two h. Open symbols indicate further developing with sheep anti-mouse IgG2a antibodies and filled symbols developing with sheep antimouse IgGl antibodies. All the experiments shown in the figure were done simultaneously.
620
9
0 0 pg/rnl A 0.6 pg/rnl v 6 pg/rnl 0 60 pg/rnl
0.2 0.10 I
I -
ANTIGENIC CHANGE IN IgG4
second explanation as it is difficult to envisage that HP 6011 can sterically block the epitope of the 40-series antibodies and not vice versa. The inhibition was also observed when using the small Fab fragment of mAb HP 6011 (data not shown), further weakening the steric hindrance theory. A reservation must be expressed in the event that the immunization procedure in the two cases has favored a kinetic maturation of the HP6011 antibody, resulting in a much higher on-rate for this antibody compared to the 40-series antibodies. High on-rate antibodies would effectively compete with antibodies having low on-rate and otherwise the same affinity. Such antibodies are characteristic for maturation of an immune response (6). The ELISA tests were usually performed by coating microtiter plates with IgG4 proteins. This adsorption step might place constraints on the availability of the epitopes reacting with HP6011 and the 40series, or deform the epitopes sufficiently to explain our observation. This possibility was unlikely since microtiter plates coated with sheep antibodies specific for human IgG Fab and then reacted with human IgG4, leaving in particular the flexibility and availability of cH3 epitopes, resulted in the same one-way inhibition. Thus artificial constraints on the availability of the epitopes can not explain our results. Parham (19) showed that a mAb, upon binding to its antigen (MHC class I), led to loss of affinity for a second antibody reacting with the same domain of MHC. Diamond et al. (3) described the opposite phenomenon, in which binding of one mAb to rat histocompatibility antigen led to enhanced affinity for a second mAb. Mazza & Retegui (13) also described how mAb to human growth hormone induced an allosteric conformational change in the antigen leading to higher affinity of a second antibody against another epitope, which had been modulated in a positive way. The Fab fragment of the antibody was as effective, and thus it could not be due to the formation of a cyclic complex observed with bivalent antibodies (4, 7, 17). Neither could it be due to the recognition of a new epitope in the Fc fragment on the enhancing antibody as proposed by Nemazee & Sato (1 8). It is surprising that the three mAbs made in
one fusion experiment all reacted with the same IgG4 subclass-specific epitope. The possibility that they might be subclones from a common mother clone, as they all came from the same fusion experiment, was excluded since they all were different both with respect to antibody affinity and electrophoretic mobility. However, it can not be formally excluded that they could be clones developed from a mother clone by somatic mutations. The appearance of so many separate clones raised against the same subclass specific epitope could perhaps indicate that this epitope is better exposed and/or more immunogenic on the human IgG4 Fc fragment (used for immunization) than on the intact IgG4 molecule. Three subclass-specific amino acids are present in the CH3 domain of the human IgG4 molecule: Gln 335, Arg 409 and Leu 445 (1 1). Arg 409 is probably not exposed on the surface of the IgG4 and thus not available for reactions with antibodies (9). It is therefore plausible that Gln 355 and Leu 445 are involved in the two sets of subclass-specific epitopes dealt with in this study. Using computer graphic models it has been shown that these two amino acids are exposed on the surface with a C,-C, distance of 0.908 nm (R@mming,personal communication). Our observations might indicate that there is an interplay between IgG4 subclass-specific epitopes in the cH3 region. Such an interplay might be of potential importance for subclass differences in effector functions. It is here interesting to note that IgG4 is the only IgG isotype which does not activate complement (16) and the only IgG isotype which binds to mast cells (1). We thank Svein Flaaten for skilful technical assistance with the fusions and the cell cultivations.
REFERENCES 1 . Burton, D. R.: Immunoglobulin G: Functional sites. Molec. Immunol. 22: 161-206, 1985. 2. Devey, M. E., Bleasdale, K.. Lee, S. & Rath, S.: Determination of the functional affinity of IgGl and IgG4 antibodies to tetanus toxoid by isotypespecific solid-phase assays. J. Immunol. Meth. 106: 119-125, 1988.
62 1
ROLSTAD et al.
3. Diamond A. G., Butcher, G. W & Howard, J. C.: Localized conformational changes induced in a class I major histocompatibility antigen by the binding of monoclonal antibodies. J. Immunol. 132: 1169-1175, 1984. 4. Ehrlich l? H., Moyle, W R., Moustafa, Z. A. & Canfield, R. E.: Mixing two monoclonal antibodies yields enhanced affinity for antigen. J. Immunol. 128: 2709-2713, 1982. 5. Engvall, E. & Perlmann, l?: Enzyme linked immunosorbent assay, ELISA. 111.Quantitation of specific antibodies by enzyme-labelled anti-immunoglobulin in antigen-coated tubes. J Immunol. 109: 129-135, 1972. 6. Foote J, & Milstein, C.: Kinetic maturation of an immune response. Nature 352: 530-532, 1991. 7. Holmes, N. J. & Parham, El: Enhancement of monoclonal antibodies against HLA-A2 is due to antibody bivalency. J. Biol. Chem. 258: 1580-1 586, 1983. 8. Huck, S., Fort, l?, Crawford, D. H., Lefranc, M-l? h Lefranc, G.: Sequence of human 3 heavy chain constant region: comparison with the other human C genes. Nucleic Acids Res. 14: 1779-1789, 1986. 9. Jefferis, R.: Human IgG subclass-specific epitopes recognized by murine monoclonal antibodies. Monogr. Allergy vol. 20. Karger, Basel, 1986, pp. 26-33. 10. Jefferis, R., Reimer, C. B., Skvaril, K , de Lange, G., Ling, N . R., Lowe, J. et al.: Evaluation of monoclonal antibodies having specificity for human IgG subclasses: results of a IUIS/WHO collaborative study. Immunol. Lett. 10: 223-252, 1985. 11. Kabat, E. A . , Wu, I: T , Reid-Miller, M . , Perry H. M. & Gottesman, K. S. (Eds). In: Sequence of Proteins of Immunologic Interest, 4th edn., National Institute of Health, Bethesda, Md, 1987, pp. 293-322. 12. Mancini, G., Carbonara, A. 0. & Heremans, J. E:
622
13.
14.
15.
16.
17.
18.
19.
20.
Immunochemical quantification of antigens by single radial immunodiffusion. Immunochemistry 2: 235-254, 1965. Mazza, M . M . & Retegui, L. A.: Monoclonal antibodies to human growth hormone induce an allosteric conformational change in the antigen. Immunology 67: 148-153, 1989. Michaelsen, I: E. & Haug, E.: Human IgG subclass-specific rabbit antisera suitable for immunoprecipitation in gel, ELISA and multilayer hemagglutination techniques. J. Immunol. Meth. 84: 203-220, 1985. Michaelsen, I: E., Frangione, B. & Franklin, E. C.: Primary structure of the “hinge” region of human IgG3. Probable quadruplication of a 15 amino acid residue basic unit. J. Biol. Chem. 252: 883-889, 1977. Michaelsen, I: E., Garred, l? & Aase, A . : Human IgG subclasses induce complement-mediated cytolysis differently upon changes in antibody affinity, complement concentration, antigen concentration and epitope density. Eur. J. Immunol. 21: 11-16, 1991. Moyle, W R., Anderson, D, M. & Ehrlich, I? H.: A circular antibody-antigen complex is responsible for increased affinity shown by mixtures of monoclonal antibodies to human chorionic gonadotropin. J. Immunol. 131: 1900-1905, 1983. Nemazee, D. A. & Sato, K L.: Enhancing antibody: A novel component of the immune response. Proc. Natl. Acad. Sci. USA 79: 3828-3832, 1982. Parham, l?: Changes in conformation with loss of alloantigenic determinants of a histocompatibility antigen (HLA-B7) induced by monoclonal antibodies. J. Immunol. 132: 2975-2983, 1984. Wofsy, L.: Methods and applications of haptensandwich labeling. Methods Enzymol. 92: 472488, 1983.
Antigenic change in a human IgG4-specific CH3epitope upon binding of a monoclonal antibody against a neighboring IgG4-sPecific epitope ANNA KRISTIN ROLSTAD, TERJE E. MICHAELSEN and JAN KOLBERG Department of Immunology, National Institute of Public Health, Oslo, Norway
Rolstad. A. K., Michaelsen, T. E. & Kolberg, J. Antigenic changes in a human IgGCspecific CH3 epitope upon binding of a monoclonal antibody against a neighboring IgG4-specific epitope. APMIS 100: 615-622, 1992. Two sets of monoclonal antibodies (mAbs) probably reacting with two different epitopes in the CH3 domain of the human IgG4 molecule were studied. We observed that the commercially available mAb HP 601 1 inhibited the antigen binding of the three mutually inhibitable mAbs, 40-A2, 41-E8 and 43F11 (40-series), made by us. However, the 40-series mAbs, including those with similar affinity such as mAb HP6011, were not able to inhibit mAb HP 601 1. When the 40-series mAbs were preincubated with IgG4, the mAb HP 6011 could partially displace these antibodies. This one-way inhibition indicates that upon binding mAb HP 601 1 changes the antigenic structure of the IgG4 molecule by disrupting the epitope for the 40-series mAbs. A steric hindrance of this epitope by mAb HP 601 1 is more unlikely, since the small Fab fragment of mAb HP 601 1 also inhibited the reaction of the 40series mAbs. Key words: Monoclonal antibodies; antigenic change; IgG4. Terje E. Michaelsen, Department of Immunology, National Institute of Public Health, Geitmyrsveien 75, 0462 N-Oslo 4, Norway. I
The four human IgG subclasses are structurally closely related and show more than 95% amino acid sequence identity between homologous domains, while the hinge regions show only 6%70% homology in sequence and vary considerably in length (8, 11, 15). There is also a difference in effector functions between the subclasses (1). Because of the great degree of homology, the individual IgG subclasses only express a few subclass-specific epitopes, and hence antibodies specific for these epitopes are difficult to produce. Some of these subclassspecific epitopes are probably directly or indirectly involved in effector function differences between the subclasses. Amino acid residues unique for human IgG4 are present in the CH1,
CH2and c H 3 domains of the molecule (1 I), and monoclonal antibodies (mAbs) specific for the IgG4 isotype have been made against each of these domains (10). In this report we have studied the expression of human IgG4 subclassspecific epitopes in the c H 3 region by employing mAbs produced by us and commercially available mAbs. Unexpectedly, inhibition experiments showed that the commercially available mAb HP 601 1 was able to block the binding of the whole group of 40-series antibodies made by us, but not vice versa. The 40-series antibodies cross-inhibited each other, and thus apparently react with the same epitope. MATERIALS AND METHODS
Received September 3, 1991. Accepted January 16, 1992.
Production of monoclonal antibodies Three different hybridoma cell lines were made in
615
ROLSTAD et al.
one fusion experiment. BALB/c mice were immunized with 20 pg IgG4 Fc in 0.25 ml of PBS mixed with 0.25 ml Freund’s incomplete adjuvant, followed by booster injections two and four weeks later with the same immunogen mixture. Four weeks later the mice were given 20 pg IgG4 Fc in PBS intraperitoneally on days one, two and three. On day five spleen cells were fused with NSO myeloma cells by standard methods with polyethylene glycol 1450 (Eastman Kodak Company, USA) as the fusion agent. Selected hybrid cells were cloned by limiting dilution and injected into pristan-primed BALBlc mice for production of antibody-containing ascites. One different hybridoma cell line, antibody 89-C7, was made in another fusion experiment using the same method as above but with intact human IgG4 as immunogen. The monoclonal antibody HP 601 1 (RJ4) specific for cH3 IgG4 was delivered by the WHO. The monoclonal antibody GB7B specific for CH2 IgG4 was kindly provided by Dr R. Jefferis, University of Birmingham, England (see Table 1). Isotype determination of the monoclonal antibodies The monoclonal antibodies purified on Protein A Sepharose (Pharmacia, Uppsala, Sweden) were isotyped using an ELISA technique. Sheep antibodies specific for mouse IgG isotypes biotinylated by Biotin-X-NHS (Calbiochem, USA) according to Wofsy (20), mixed with streptavidin-conjugated alkaline phosphatase (ALP) (Sigma, USA, type VII) (5), were used as specific reagents. The mAbs 40-A2, 41-E8 and 43-FI1 were of IgG2a isotype (Table 1). MAb 89-C7 was typed as IgGl . The mAbs HP 601 1 and GB7B were also of IgGl isotype (10). Tests .for antibody specificity The specificity of the monoclonal antibodies was tested in an ELISA system on microtiter plates (Nunc-Immuno-Plate, Maxisorb, Denmark) coated with myeloma proteins of the four different human IgG isotypes (1 pg/ml, 150 pUwe11). The antibodies were shown to be specific for IgG4 subclass (Table 1). The epitope specificities were further evaluated by
testing them on microtiter plates coated with IgG4 Fab-, Fc- and pFc’-fragments (1 pg/ml, 150 pl/well). The fragments were prepared by papain and pepsin digestion as described (14). Our monoclonal antibodies in the 40-series and HP 6011 all reacted with IgG4 pFc’, while 89-C7 reacted with the IgG4 Fab fragment and GB7B with IgG4 Fc but not with IgG4 pFc’ (IgG4 Fc, non pFc’) (Table 1). Quantitation of IgG in ascites The IgG concentration in ascites was measured by single radial immunodiffusion according to Mancini et al. (12), employing sheep antiserum specific for mouse IgG. The sheep antiserum used was shown to react equally well with all mouse IgG subclasses. Antibody affinity Antibody affinities of the different monoclonal antibodies were compared by testing on microtiter plates coated directly with a human IgG4 myeloma protein (1 pg/ml, 150 pl/well), and plates coated indirectly with IgG4 (1 pg/ml) by catching with sheep anti-human IgG Fab (1 pg/ml). After incubation with dilutions of monoclonal antibodies for two h at 37°C and washing, biotinylated sheep anti-mouse IgG antibodies mixed with streptavidin-ALP were added and incubation continued. This biotinylated sheep antimouse IgG conjugate was shown to react equally well with IgGl and IgG2a (data not shown). After further washing the enzyme substrate disodium p-nitrophenyl phosphate (NPP) (1 mg/ml in 10% diethanolamine buffer, pH 9.8, containing 10 mM MgClJ was added and incubation continued for 30-60 min before absorbance at 410 nm was recorded (Dynatech MR 700, Germany). The high affinity antibodies are seen to the left in Fig. 1; the low affinity antibodies to the right. Antibody affinity was also evaluated using an inhibition test where the concentration of IgG4 required for 50% inhibition of the antibodies was taken as a measure of affinity (2).
ELISA inhibition tests Cross-inhibition of the different monoclonal antibodies was performed in two different ways using microtiter plates coated with IgG4 myeloma protein (1 pg/ml, 150 pl/well) in both cases. Various concen-
TABLE 1. SDecificitv o f monoclonal antibodies Monoclonal Isotype Subclass Fragment (domain) antibody specificity specificity 40-A2 IgG2a IgG4 PFc’ (cH3) IgG2a 4 1-E8 IgG4 pFc‘ (cH3) 43-F11 IgG2a IgG4 PFc’ (cH3) HP 601 I IgG 1 IgG4 pFc‘ (cH3) (RJ 4) 89-C7 IgG 1 IgG4 Fab (CHI) IgG4 Fc non pFc’ (C,2) GB7B IgG 1
616
ANTIGENIC CHANGE IN IgG4
trations of the hybridoma antibodies to be tested for inhibitory capacity were added to the coated microtiter plates and incubated for one h. The antibody which we wanted to inhibit was then added as a biotin conjugate at an optimal concentration and the incubation continued for two h. Following this the plate was washed and streptavidin-ALP added, and there was a two h incubation period. After further washing the enzyme substrate NPP was added and incubation continued for 30-60 min before absorbance at 410 nm was recorded. The other inhibition test took advantage of the fact that the antibodies HP 601 I and 89-C7 were of the
g
0.90
-
0.72
-
0
A
F
t 0.54
-
0.30
-
0.18
-
0.00
-
0
E
a
51 a a
.&mL
0.001
0.01
0.1
1
10
100
mouse IgGl isotype, while our antibodies in the 40series were of the IgG2a isotype. The inhibitory antibody and the antibodies to be inhibited were discriminated simply by developing the microtiter plate with sheep antibodies specific for the respective mouse isotypes. Various concentrations of the hybridoma antibodies to be tested for inhibitory capacity were added to IgGCcoated plates and incubated for one h before the second antibody at a fixed optimal concentration was added, and incubated further for two h. After washing, biotinylated sheep anti-mouse IgGl and anti-mouse IgG2a, respectively, mixed with streptavidin-ALP were added and incubated for two h. Incubation with the substrate NPP and reading of the plates were performed as described above. This procedure could measure inhibitory capacity and binding activity of both antibodies simultaneously by parallel development with anti-mouse IgGl and IgG2a antibodies. ELISA replacement tests The antibodies’ ability to replace each other was tested in ELISA by incubating various concentrations of one antibody, washing away excess antibodies and adding various concentrations of the second antibody. The second antibody’s ability to replace the first antibody was measured by developing the ELISA plates with antibody conjugates specific for the isotype of the first antibody.
1000
Concentration >f antibodies, pg/ml
RESULTS 1.1 0.9
g
-
0.8
-
2
0.7
1.0
0
o 0.6 g 0.5 0
*::
0.4
-
a 0.3 -
a
0.2
0.1
-
0
0.001
0.01
0.1
1
10
100
1000
Concentration of antibodies, pg/rnl
Fig. 1. Dose response curves of the antibodies, reflecting the antibody affinity of the mAb. Microtiter plates were coated with human IgG4 (Hen.), 1 pg/ ml directly (A) or indirectly with sheep anti-human IgG Fab (B) and dilutions of the mAb added. After incubation and washing, biotinylated sheep antimouse IgG antibodies were added and incubation continued before the enzyme substrate NPP was added.
H P 6011 inhibits the 40-series of antibodies which mutually inhibit each other The 40-A2, 41-E8 and 43-Fll antibodies could mutually inhibit each other in ELISA inhibition tests (Fig. 2A, B and C). The mAb HP 601 1 also efficiently inhibited mAbs 40-A2, 41-E8 and 43-F11 (Fig. 2A, B and C). The Fab fragment of mAb HP 6011 inhibited the 40series of antibodies (data not shown). On the other hand, the IgG4 C,Zspecific GB7B antibody did not compete with the 40-series antibodies (Fig. 2A, B and c ) , or with HP6011 (Fig. 2D), while as a positive control it was shown to inhibit itself (Fig. 2E). The mAb HP 6011 was shown to inhibit the native unconjugated 40series of antibodies by using the other inhibitory test where we employed sheep anti-mouse IgGl and IgG2a antibodies, respectively. From Fig. 3 it can be seen that mAb HP 6011 even inhibits the 40-A2 antibody which has the same affinity as HP 6011 against the antigen, IgG4 (Fig. 1).
617
ROLSTAD et al.
60 11, and employing sheep anti-mouse IgG 1 and IgG2a antibodies, respectively. No inhibition by the 40-series against HP 601 1 was observed even with the high affinity antibody 40-A2 (Fig. 4).The mAb 89-C7 of isotype IgGl, reacting with IgG4 Fab, was included as a negative control (Fig. 4).
The 40-series of antibodies do not inhibit HP 6011 The 40-A2,41-E8 and 43-F11 antibodies were tested for inhibitory capacity against HP 6011. None of the antibodies in the 40-series inhibited HP 601 1 (Fig. 2D). The inhibitory capacity was also tested using the native unconjugated HP
0.75 0.60
0.45
0.6
-
n 0.30 L
0.4
-
0.2
-
0
v
e
m 0
43-Fll
Y)
.n
U
0.15
&
0.00 0.01
g 0
1
0.1
10
0.0 1
100
0.1
1
10
100
1
10
100
1
1.6
r
.t ID V C
1.2
fz
s
0.6
a
0.4
A
40-A2
P
a 0.4
0 .o
0.0
0.0 1
1.oo
5 0
- A 41-E8 43-F11 - *0 G878 HP6011
I
1
0.1 '
' " 1 " 1 ' 1
'
"'""1
100
10 '
"1""1
""""I
F
w
0.60
c
a
fz0
0.40
2
0.20
H
0.00
t
4042 A 4l-EO P 43-Fl1 0 HP6Oll 0078
*
0.01
0.1
1
10
100
Concentration of antibodier, pg/rnl
618
0.1
Concentratlon of antibodies, pg/rnl Fig. 2. ELISA inhibition against biotin-labeled
0.80
V
0.01
mAb 40-A2 (A), mAb 41-E8 (B), mAb 43-Fll (C), MAb HP6011 (D) and GB7B (E). Microtiter plates were coated with human IgG4 (Hen.), I pg/ml. The different antibodies tested for inhibitory capacity were added at the concentrations indicated in the figure, and incubated for one h before an optimal concentration of biotinylated mAb 40-A2 (1 : 16,000), mAb 41-E8 (1 : 8,000), mAb 43-F11 (1 :20,000), MAb HP6011 (1 20,000) and mAb GB7B (1 : 8,000) in A, B, C, D and E, respectively, was added and incubation continued for two h. Plates were further developed as described in Materials and Methods.
ANTIGENIC CHANGE IN IgG4
HP 6011 and the 40-series bind to different sets of epitopes
ELISA inhibition tests revealed that the HP 601 1 and 40-A2/41-E8/43-F11 antibodies could bind simultaneously provided that the HP 601 1 antibody was added after mAb 40-A2/41-E8/ 43-Fll (Fig. 4). Furthermore, the 40-A2 antibody was not substituted with HP 6011 antibody, but stayed bound to the IgG4 antigen under these conditions (Fig. 4). High concentrations of mAb 40-A2 have no influence on binding of either mAb HP 6011 or mAb 89-C7
E
!2 1.0 d
' I0 0.8
0.6
0.4 0.2
The IgG4 epitopes are not disturbed by plastic adherence In some tests, the IgG4 antigen was caught on the ELISA plates with the Fab part oriented inwards by coating with sheep antibodies specific for the Fab part of human IgG. We observed the same kind of one-way inhibition under these experimental conditions (data not shown). The 40-series antibodies are not subclones There was a marked difference in the electrophoretic mobility in agarose gel electrophoresis (data not shown), indicating that they were not subclones. Furthermore, the antibodies showed different antibody affinity (Fig. l), also indicating that they were not subclones.
1.2
88
(Fig. 4). HP6011 could partly replace 40-A2 (Fig. 5). The 40-A2 antibody could not replace HP6011, and 89427 could not replace 40-A2 in similar experiments (data not shown).
1 1
0 1 :32000
1:3200
1:320
Dilution of HP 6011 First mAb/ inhibiting
-E-
HP6011
Second mAb/ inhibited
none
4HP6011
40-A2
+ 40-A2
none
+ HP6011
none
-& HP6011
40-A2
+ 40-A2
none
Fig. 3. ELISA inhibition against mAb 40-A2 attempted by mAb HP 6011. Microtiter plates were coated with human IgG4 (Joh.), 1 pg/ml. Different dilutions of mAb HP 6011 (mouse IgG1) were added and incubated for one h before mAb 40-A2 (mouse IgG2a) (1 : 128,000) was added and incubation continued for two h. Open symbols indicate further developing with sheep anti-mouse IgG2a antibodies and filled symbols developing with sheep anti-IgG1 antibodies.
The one-way inhibition is not due to differences in affinity MAb 40-A2 and HP 6011 showed equally high affinity; mAb 43-Fll showed somewhat lower affinity, while mAb 41-E8 showed the lowest affinity (Fig. 1). The results were the same when unpurified ascites and purified antibodies were employed. A similar pattern was seen with two different IgG4 proteins (Hen. and Joh.). The measurements when using plates coated directly and indirectly (catching) with IgG4 revealed two differences: the 41-E8 antibody had a relatively higher affinity when indirectly coated IgG4 was employed (Fig. 1A) and the 89-C7 antibody had lowest affinity with directly coated IgG4 (Fig. 1B). Otherwise the results were similar for directly and indirectly coated IgG4 (Fig. 1A and 1B). Using an inhibition test described by Devey et al. (2), the same antibody affinity pattern of the antibodies was obtained as referred to above (data not shown).
DISCUSSION By immunizing BALB/c mice with Fc fragments of an IgG4 myeloma protein, we have made three hybridoma antibodies: mAb 619
ROLSTAD et al.
40-A2, 41-E8 and 43-Fll. They were all shown to be specific for human IgG4 pFc', and they could all mutually inhibit each other and thus probably react with the same or closely situated epitopes. Another commercially available IgG4 pFc'-specific antibody, HP 6011 (RJ4), efficiently inhibited mAb 40-A2, 41-E8 and 43-Fl l, while none of the latter antibodies could inhibit mAb HP 601 1. However, when human IgG4 was preincubated
I
1.2 1 .o
z
0.8
R
~~
1
with the 40-series antibodies and the excess antibodies washed away, mAb HP 6011 could partly replace the 40-series antibodies. The reverse replacement could not be achieved. These observations indicate that mAb 40-A2/ 41-E8/43-Fll and mAb HP 6011 react with two different sets of epitopes on the IgG4 molecule. There is, however, the possibility that HP 6011 can inhibit the 40-series antibodies, but not vice versa, if HP 601 1 has a much higher affinity than the other antibodies. This was not the case, at least for mAb 40-A2, which was tested to have nearly the same affinity as mAb HP 6011 (Fig. 1). Furthermore, the mAb 41-E8 with the lowest antibody affinity can inhibit the other two antibodies in the 40-series which have higher affinity. There may be at least two possible explanations for this peculiar one-way inhibition: 1) the first reacting mAb is sterically hindering the access of the second mAb to its epitope, or 2) the first mAb induces a conformational change in the antigen molecule upon binding, thereby changing the structure of the second epitope sufficiently to prohibit binding to the second antibody (the 40-series antibodies). We favor the
[ I 1 \I 1
t
o.2 01:10000
1:lOoo
1:loo
Dilution of 4042
--
First mAb/ inhibiting
Second mAb/
inhibited
-ff 40-A2
none
++40-A2
HP 601 1
-A- 40-A2
89-67
_--E
........
...... ......................... 8
............ -0- 40-A2 -C 40-A2
40-A2
none HP 6011 89-67
Fig. 4. ELISA inhibition against mAb H P 6011 and
89-C7 attempted by mAb 40-A2. Microtiter plates were coated with human IgG4 (Hen.), 1 pg/ml. Different dilutions of mAb 40-A2 (mouse IgG2a) (ascites) were added and incubated for one h before mAb HP 601 1 (1 : 30,000) and 89-C7 (1 : 10,000) (both mouse IgG l), respectively, were added and incubation continued for two h. Open symbols indicate further developing with sheep anti-mouse IgG2a antibodies and filled symbols developing with sheep antimouse IgGl antibodies. All the experiments shown in the figure were done simultaneously.
620
9
0 0 pg/rnl A 0.6 pg/rnl v 6 pg/rnl 0 60 pg/rnl
0.2 0.10 I
I -
ANTIGENIC CHANGE IN IgG4
second explanation as it is difficult to envisage that HP 6011 can sterically block the epitope of the 40-series antibodies and not vice versa. The inhibition was also observed when using the small Fab fragment of mAb HP 6011 (data not shown), further weakening the steric hindrance theory. A reservation must be expressed in the event that the immunization procedure in the two cases has favored a kinetic maturation of the HP6011 antibody, resulting in a much higher on-rate for this antibody compared to the 40-series antibodies. High on-rate antibodies would effectively compete with antibodies having low on-rate and otherwise the same affinity. Such antibodies are characteristic for maturation of an immune response (6). The ELISA tests were usually performed by coating microtiter plates with IgG4 proteins. This adsorption step might place constraints on the availability of the epitopes reacting with HP6011 and the 40series, or deform the epitopes sufficiently to explain our observation. This possibility was unlikely since microtiter plates coated with sheep antibodies specific for human IgG Fab and then reacted with human IgG4, leaving in particular the flexibility and availability of cH3 epitopes, resulted in the same one-way inhibition. Thus artificial constraints on the availability of the epitopes can not explain our results. Parham (19) showed that a mAb, upon binding to its antigen (MHC class I), led to loss of affinity for a second antibody reacting with the same domain of MHC. Diamond et al. (3) described the opposite phenomenon, in which binding of one mAb to rat histocompatibility antigen led to enhanced affinity for a second mAb. Mazza & Retegui (13) also described how mAb to human growth hormone induced an allosteric conformational change in the antigen leading to higher affinity of a second antibody against another epitope, which had been modulated in a positive way. The Fab fragment of the antibody was as effective, and thus it could not be due to the formation of a cyclic complex observed with bivalent antibodies (4, 7, 17). Neither could it be due to the recognition of a new epitope in the Fc fragment on the enhancing antibody as proposed by Nemazee & Sato (1 8). It is surprising that the three mAbs made in
one fusion experiment all reacted with the same IgG4 subclass-specific epitope. The possibility that they might be subclones from a common mother clone, as they all came from the same fusion experiment, was excluded since they all were different both with respect to antibody affinity and electrophoretic mobility. However, it can not be formally excluded that they could be clones developed from a mother clone by somatic mutations. The appearance of so many separate clones raised against the same subclass specific epitope could perhaps indicate that this epitope is better exposed and/or more immunogenic on the human IgG4 Fc fragment (used for immunization) than on the intact IgG4 molecule. Three subclass-specific amino acids are present in the CH3 domain of the human IgG4 molecule: Gln 335, Arg 409 and Leu 445 (1 1). Arg 409 is probably not exposed on the surface of the IgG4 and thus not available for reactions with antibodies (9). It is therefore plausible that Gln 355 and Leu 445 are involved in the two sets of subclass-specific epitopes dealt with in this study. Using computer graphic models it has been shown that these two amino acids are exposed on the surface with a C,-C, distance of 0.908 nm (R@mming,personal communication). Our observations might indicate that there is an interplay between IgG4 subclass-specific epitopes in the cH3 region. Such an interplay might be of potential importance for subclass differences in effector functions. It is here interesting to note that IgG4 is the only IgG isotype which does not activate complement (16) and the only IgG isotype which binds to mast cells (1). We thank Svein Flaaten for skilful technical assistance with the fusions and the cell cultivations.
REFERENCES 1 . Burton, D. R.: Immunoglobulin G: Functional sites. Molec. Immunol. 22: 161-206, 1985. 2. Devey, M. E., Bleasdale, K.. Lee, S. & Rath, S.: Determination of the functional affinity of IgGl and IgG4 antibodies to tetanus toxoid by isotypespecific solid-phase assays. J. Immunol. Meth. 106: 119-125, 1988.
62 1
ROLSTAD et al.
3. Diamond A. G., Butcher, G. W & Howard, J. C.: Localized conformational changes induced in a class I major histocompatibility antigen by the binding of monoclonal antibodies. J. Immunol. 132: 1169-1175, 1984. 4. Ehrlich l? H., Moyle, W R., Moustafa, Z. A. & Canfield, R. E.: Mixing two monoclonal antibodies yields enhanced affinity for antigen. J. Immunol. 128: 2709-2713, 1982. 5. Engvall, E. & Perlmann, l?: Enzyme linked immunosorbent assay, ELISA. 111.Quantitation of specific antibodies by enzyme-labelled anti-immunoglobulin in antigen-coated tubes. J Immunol. 109: 129-135, 1972. 6. Foote J, & Milstein, C.: Kinetic maturation of an immune response. Nature 352: 530-532, 1991. 7. Holmes, N. J. & Parham, El: Enhancement of monoclonal antibodies against HLA-A2 is due to antibody bivalency. J. Biol. Chem. 258: 1580-1 586, 1983. 8. Huck, S., Fort, l?, Crawford, D. H., Lefranc, M-l? h Lefranc, G.: Sequence of human 3 heavy chain constant region: comparison with the other human C genes. Nucleic Acids Res. 14: 1779-1789, 1986. 9. Jefferis, R.: Human IgG subclass-specific epitopes recognized by murine monoclonal antibodies. Monogr. Allergy vol. 20. Karger, Basel, 1986, pp. 26-33. 10. Jefferis, R., Reimer, C. B., Skvaril, K , de Lange, G., Ling, N . R., Lowe, J. et al.: Evaluation of monoclonal antibodies having specificity for human IgG subclasses: results of a IUIS/WHO collaborative study. Immunol. Lett. 10: 223-252, 1985. 11. Kabat, E. A . , Wu, I: T , Reid-Miller, M . , Perry H. M. & Gottesman, K. S. (Eds). In: Sequence of Proteins of Immunologic Interest, 4th edn., National Institute of Health, Bethesda, Md, 1987, pp. 293-322. 12. Mancini, G., Carbonara, A. 0. & Heremans, J. E:
622
13.
14.
15.
16.
17.
18.
19.
20.
Immunochemical quantification of antigens by single radial immunodiffusion. Immunochemistry 2: 235-254, 1965. Mazza, M . M . & Retegui, L. A.: Monoclonal antibodies to human growth hormone induce an allosteric conformational change in the antigen. Immunology 67: 148-153, 1989. Michaelsen, I: E. & Haug, E.: Human IgG subclass-specific rabbit antisera suitable for immunoprecipitation in gel, ELISA and multilayer hemagglutination techniques. J. Immunol. Meth. 84: 203-220, 1985. Michaelsen, I: E., Frangione, B. & Franklin, E. C.: Primary structure of the “hinge” region of human IgG3. Probable quadruplication of a 15 amino acid residue basic unit. J. Biol. Chem. 252: 883-889, 1977. Michaelsen, I: E., Garred, l? & Aase, A . : Human IgG subclasses induce complement-mediated cytolysis differently upon changes in antibody affinity, complement concentration, antigen concentration and epitope density. Eur. J. Immunol. 21: 11-16, 1991. Moyle, W R., Anderson, D, M. & Ehrlich, I? H.: A circular antibody-antigen complex is responsible for increased affinity shown by mixtures of monoclonal antibodies to human chorionic gonadotropin. J. Immunol. 131: 1900-1905, 1983. Nemazee, D. A. & Sato, K L.: Enhancing antibody: A novel component of the immune response. Proc. Natl. Acad. Sci. USA 79: 3828-3832, 1982. Parham, l?: Changes in conformation with loss of alloantigenic determinants of a histocompatibility antigen (HLA-B7) induced by monoclonal antibodies. J. Immunol. 132: 2975-2983, 1984. Wofsy, L.: Methods and applications of haptensandwich labeling. Methods Enzymol. 92: 472488, 1983.
