Charged and aromatic amino acids in polyreactive antibodies
Eur. J. Immunol. 1990. 20: 2383-2387
Rosana Gonzhlez-Quintialn, Roberto BaccalaO, Pedro Maria Alzari., Clara NahmiasO, Gilbert Mazza., Michel Fougereau. and Stratis Avrameaso Unit6 d’Immunocytochimien, Unit6 d’Immunologie Structurale., Dbpartement d’Immunologie; Unit6 de Biologie MolCculaire des RBcepteursO, Ddpartement de Biotechnologie, Institut Pasteur, Pans and Centre d’Immunologie INSERM-CNRS de MarseilleLuminy., Marseille
Poly (GluaAlaqyrlO) (GAT)-induced IgG monoclonal antibodies cross-react with various self and non-self antigens through the complementarity determining regions. Comparison with IgM monoclonal polyreactive natural antibodies* Previous studies have shown that the antibodies of the preimmune repertoire are able to bind to various auto- and xenoantigens including chemical haptens. Sequence analysis of two such murine monoclonal IgM natural autoantibodies showed that they are encoded by unmutated germ-line variable regions of the light and heavy chain (V, and V,) genes which were also found in various murine immune responses, like phenyl-oxazolone ,dinitrophenyl, arsonate, phosphorylcholine and influenza virus hemagglutinin. These data raised the question as to whether induced antibodies possessing germ-line sequence are also able to react with autoantigens. To study this problem, anti-poly(Gl~~Ala~OTyr~~) (GAT) and anti-alprenolol (Alp) monoclonal antibodies, carrying similar VHand V, genes and the same IgGl isotype, were examined for their capacity to react with several self and non-self antigens. The results showed that: (a) the anti-GAT antibodies tested reacted with different autoantigens, such as murine tubulin, actin and myosin as well as trinitrophenyl (TNP) and bovine serum albumin. Similarly, one of the anti-Alp showed weak reactivities for myosin, DNA, actin and TNP; (b) in contrast two other anti-Alp antibodies did not react with any of the tested antigens. Since the major differences between the oligoreactive anti-GAT and the monoreactive anti-Alp antibodies are in the complementarity determining regions (CDR) our results suggest that the observed cross-reactions are mediated by hypervariable loops. Sequence comparison of these antibodies indicate a possible correlation between cross-reactivity and the presence of aromatic and charged amino acids in the CDR.
1 Introduction The antibody response induced by a given antigen is generally dependent on the B and T cell repertoires available at the time of immunization.These repertoires are dynamic and are the consequence of the antigenic experience of the organism [l]. However, little is known concerning the selection of these repertoires during ontogeny and their maintenance during adult life. Preimmune natural antibodies, which result from the internal activity of the naive immune system, seem to be in large part directed
[I 81731
*
2383
This work was supported by grants from the Ministry of Education and Science of Spain, the Swiss National Science Foundation, Association pour la Recherche sur le Cancer, INSERM and CNRS.
Correspondence: Rosana GonzBlez-Quintial, Scripps Clinic and Research Foundation, Department of Immunology, IMM3,lO 666 North Torrey Pines Road, La Jolla, CA 92037, USA
against self antigens [2, 31. Furthermore, it has been shown in a number of cases that they are involved in a highly connected Id network [4], and that they possess potent regulatory properties [5, 61. This connectivity may be the consequence of their ability to cross-react with several different self and non-self antigens such as cytoskeletal proteins, nucleic acids and chemical haptens [2]. Also, such multiple interactions could be the basis of selection processes and somatic events which give rise to more specific antibodies to exogenous antigens. Along these lines it has been proposed that monospecific antibodies could represent mutant forms of germ-line-encoded preimmune polyand autoreactive antibodies [2,7]. Recently, we have cloned and sequenced the complete cDNA corresponding to two natural IgM mAb (E7 and D23) obtained from normal nonimmunized adult BALB/c mice. These results [8] showed that: (a) the two antibodies are encoded by germ-line genes without mutations; and (b) mutant forms of these genes are encountered in various induced immune responses.
These data raise the possibility that post-immune antibodies with germ-line configurations might possess the capacity Abbreviations: Alp: Alprenolol GAR P ~ l y ( G l u ~ A l a ~ V T y rto ~ ~ )cross-react with several different self and non-self GAT-Abl, GAT-Ab2, GAT-Abl’: Id cascade induced by the antigen GAT, in which GAT-Abl is the responding antibody antigens. The immune response against the random syny r ~ ~ ) has been bearing the Id set, GAT-Ab2 the anti-Id antibody and GAT-Abl’ thetic terpolymer p o l y ( G l ~ ~ A l a ~ 0 T(GAT) the anti-anti-idiotypic antibody with the capacity to bind the extensively studied. Indeed, sequence analysis of mAb original Ag H1, H2,H3: CDRl, 2 and 3 of the H chain derived from the three principal stages (Abl, Ab2, Ab3/Abl’) of the GAT-induced Id cascade has shown that L1, L2, L3: CDR1, 2 and 3 of the L chain 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
014-2980/90/1111-2383$3.50+ .25/0
2384
R. GonzBlez-Quintial, R. Baccalh, P. M. Alzari et al.
oneVH (VH10) and two very homologousV, (5.1 and 1.A5) are expressed in their germ-line configurations by the majority of Abl and Abl’; Ab2 also seem to be germ-line encoded [9-141. Here we report results showing that A b l and in some cases Abl’ with anti-GAT activity cross-react with a number of other self and non-self antigens. In contrast, two mAb induced by the hapten alprenolol (Alp), which are encoded by mutated forms of the sameVH and virtually identical V, genes as the A b l and Abl’ of the GAT system [15], were found to be monospecific. Since the major differences between anti-GAT and anti-Alp antibodies are concentrated in the complementarity determining regions (CDR), our results suggest that the observed cross-reactions are mediated by the hypervariable loops.
2 Materials and methods 2.1 Antigens BALB/c muscle actin, myosin, brain tubulin and TNP-BSA were prepared as described previously [16]. Alp-BSA was prepared following the method of Vauquelin et al. [17].The synthetic random terpolymer GAT was kindly donated by Dr. C. Roth (Pasteur Institut, Paris, France). Native double-stranded (ds)DNA and BSA were purchased from Sigma Chemical Co. (St. Louis, MO) and IBF,Villeneuvela-Garenne, France. 2.2 Antibodies Sheep antibodies to mouse Ig were isolated using the corresponding immunosorbent [181 and labeled with P-galactosidase (a gift from Dr. Ullman, Pasteur Institut) using the glutaraldehyde procedure [19]. Rabbit or goat antibodies directed against mouse IgM and IgGl H chains and L chains were purchased from Southern Biotechnology Associates (Birmingham, AL). 2.3 Preimmune and induced mAb The preimmune mAb (E7 and D23) were obtained by fusing spleen cells from normal 12-week-old BALB/c mice [2]. The production and characterization of anti-Alp and GAT (Abl, Ab2 and Abl’) hybridomas have been described elsewhere [20-221. Anti-Alp mAb were purified from ascitic fluid by DEAE chromatography (Pharmacia, Uppsala, Sweden), whereas GAT system mAb were purified by D E 52 chromatography (Whatman, Maidstone, GB), followed by a Sephadex G-150 gel filtration.
Eur. J. Immunol. 1990. 20: 2383-2387 washed three times with PBS containing 0.1% Tween 20 (PBS-T) and incubated with mAb ranging from 100 ng/ml to 50 yglml in PBS-T containing 0.5% gelatin (PBS-T-G) for 1 h at 37 “C and overnight at 4 “C. After washing with PBS-T, the plates were incubated for 1 h at 37°C with P-galactosidase-labeled anti-mouse Ig antibodies (1 yg/ml). After washing, the enzyme substrate was added and the subsequent steps were performed as described elsewhere [181.
3 Results and discussion 3.1 Three IgM and four IgG germ-line-encoded mAb are polyreactive and bind to auto- and xenoantigens Induced mAb were compared with preimmune mAb for their capacity to cross-react with self and non-self antigens. Ten previously characterized mAb were considered in this study; five were from the GATsystem, three from the Alp system and two, obtained from normal nonimmunized mice, were identified as preimmune polyreactive natural autoantibodies. Fig. 1 summarizes the reported data concerning V-region genes coding for these mAb. The preimmune E7 and D23 mAb seem to be coded for by different combinations of non-mutated germ-line genes [8]. Similarly, it has been shown that mAb from the GAT-induced idioypic cascade GAT-Abl, GAT-Abl’ and GAT-Ab2 mAb are encoded by non-mutated germ-line genes [9-111. The two anti-Alp mAb 14C3 and 17C1, use, with a limited number of somatic mutations, the same VH and a very similar V, as the GAT-Abl and the GAT-Abl’,whereas the third anti-Alp, 37A4 uses the same VH associated to a different L chain [15]. Analyzed by ELISA, the anti-Alp mAb 17C1 and 14C3 seem to be monospecific and reacted only with Alp (Fig. 2). Similarly, the GAT-Ab2 (HP-22) did not react with any of the tested antigens. The third anti-Alp mAb, 37A4 showed only a weak cross-reaction with actin,TNP and myosin. In contrast, the tested GAT-Abl (G5 Bb2.2, G8 Ca1.7 and G8 Ad3.8) and also the GAT-Abl’ (22-186), like preimmune natural autoantibodies, showed significant crossreactivity, mainly with tubulin but also withTNP, BSA and actin.The anti-tubulin activity of the GAT-Abl‘ (22-186) is comparable to its anti-GATactivity. Except for the anti-Alp mAb, none of the tested mAb cross-reacted with Alp. In contrast, the antigen GAT was recognized by both the preimmune E7 and D23. In particular, the anti-GAT activity of D23 exceeded that of the GAT-Abl’ (22-186). 3.2 The observed cross-reactions are related to the structure of the CDR: possible role of charged and aromatic amino acids
2.4 ELISA Preimmune (E7 and D23), anti-GAT (G5 Bb2.2, G8 Ca1.7, G8 Ad3.8,22-186 and HP22) and anti-Alp (14C3,17C1 and 37A4) mAb were titrated using an ELISA. For the coating of microtiter plates, the protein antigens were used at 5 yg/ml;TNP-BSA and single-stranded (ss) DNA were used at 1 yg/ml and 500 yg/ml, respectively; Alp-BSA and GAT were used at 10 pg/ml. After coating, the plates were
As mentioned above, anti-GAT mAb (Abl and Abl’) and anti-Alp mAb have the same isotype (IgG1) and are encoded by very homologous VH and V, genes. The major structural differences are due to different D segments and to somatic mutations in the anti-Alp mAb, and are limited to the CDR suggesting that the analyzed GAT-induced IgG mAb cross-react with various self and non-self antigens through the CDR.
Charged and aromatic amino acids in polyreactive antibodies
Eur. J. Immunol. 1990. 20: 2383-2387
-
El 023
T I*
I*
~5
m
G9
ca 1.1
G9
a ANrIEEDIEs
w
2.2
3.8
22-186
Preirmw Preinmm
V-~IaiS
W U10.1 (3660) W 101 (Qvl
1 6 1 GAT (Abll IqGl GAT IAblI IgGl GAT IAblI
W 10 W 10
(55581 155581 (55581
I@
GAT IAbl'
W 10
155581
Wi 10 Wi 10 W 10
15558) 15558) 155581
17c1
I*l
140 31A4
161 161
anti-Alp anti-Alp anti-Alp
w22
IS1
CAT (PI21
I@ I@
W
10
Pre-
W ? (CRXIP 19)
Preirmrne
W ? -1
G5 Bb 2.2
1 6 1 CAT IAblI G9 ca 1.7 1 6 1 CAT IAbll G9 W 3.0 IgGl CAT @blI
8)
W W (&Em 11 W W -1 W W -1
22-186
I*
31A4
IqGl anti-Alp 1 6 1 anti-Alp 1 8 1 anti-Alp
w22
IqGl CAT (MJ M ?
1) 1)
GRT IAbl'J W W (QUIP 11
W 5.1 ICRzR 1) W 5.1 - 1
W ?
Figure 1. Schematic representation of variable
1)
ICRXIP ?I
regions. CDR are boxed. The amino acid differences to most homologous germ-line V genes are represented by A.
(CRXIP?)
Actin
I
1500
Table 1. Characteristics of the antibody-binding sites
mAb
1000
E7 D23
500
G5 Bb2.2 G8 0 1 . 7 G8 Ad3.8 22-186
0
-E 2 w
5 P
1500
1000
500
Pp z a
Net Arginines Aromatic charge amino acids
Tyrosines
+6 +9
4 7
20 17
14 8
+10 +10
7 7
16 16 15 18
7 7 8 12
+8
5
+7
3
37A4 17C1 146
+6 +2 +4
4
3 3
14 14 15
6
HP-22
-2
4
19
13
7 6
o
U
g
2385
Anti-GATmAb cross-react with tubulin. Sequence analysis of tubulin indicates that there are many glutamate residues in the carboxyl end of both the a and subunits [26]. Furthermore, in the same regions, the sequences GluGlu-Tyr can often be observed. It seems possible that anti-GAT mAb interact with these regions of tubulin molecules since they represent a dominant epitope of the GAT antigen.
TNP-BSA 1500
1000
500
0
rn
As shown in Fig. 3, most of the structural differences between homologous poly- and monoreactive mAb concern charged amino acids. Frequently, positively charged 750 amino acids situated in the CDR of the polyreactive GAT-induced mAb, have neutral or negatively charged 500 amino acids counterparts in the corresponding regions of 250 monospecific Alp-induced mAb, which have a prevalence of negatively charged amino acids and show in the CDR H3 n a typical pocket formed by acidic residues. This pocket 0.1 1.0 10 100 0.1 1.0 10 100 possibly involved in interactions with Alp [15], does not, pg/ml Ab however, preclude a weak interaction of one of the anti-Alp Figure2. Binding of mAb to myosin-, actin-, tubulin-, BSA-, TNP-BSA-, ssDNA-, GAT- and Alp-BSA-coatedplates. Decreas- antibodies (37A4) with actin, DNA, TNP and myosin ing concentrations of E7 (0),D23 (O), G5 Bb2.2 (+),G8 Ca1.7 (Fig. 2). The structural differences between 37A4 and the ( O ) , G8 Ad3.8 (H), 22-186 (A),HP22 (U), 14C3 (A), 17C1 (0) monospecific 14C3 and 17A4 (Fig. 3) could give interesting and 37A4 (a)were incubated with the different plates. insights into the basis of these interactions. 37A4 has two 1000,
GAT
I
Al renolol-BSA
Eur. J. Immunol. 1990. 20: 2383-2387
R. Gonzalez-Quintial, R. Baccalh, I? M. Alzari et al.
2386
H2 E7 023
HPZ
H3
M G Y I S Y S G S T Y Y N P S L @ - S L G V I W S G G S T ~ Y N A A F I S -
I
T I ~ G - WIN Y
w
N
G I I~ N Y
M
~
s
GN N
-
Y N P s LB+I
L1
L2
A S B N V V T - - - - - - Y V S W Y SSQSLLNS@T@@NYLAWY
E7
D23
S S S S
S S S S
Q Q Q Q
S S S S
I I I L
V V V V
H H H H
S S S S
-
N N N N
G G G G
N N N N
T T T T
Y Y Y Y
L L L L
Y Y E H
W W W W
S C g S S Q S L V H S - N G I T Y L H W H S C S S Q S L V H S - N G B T Y L H W H T C T A@S S V S S S - - - - - Y F H W Y
RP22
T
14c3
C
@A
S Q m I S V
------
I Y G A S N Y T G V I Y W I G V
Y Y Y Y
31A4
17C1
@PFYG@GAM------D W G m@L@L@Y F - - - - - - - 1 j W G
Y L NW L
1 y1s T
s
r
LEJSIG
v
Figure3. Amino acid representation of CDR and two flanking residues of preimmune (E7 and D23), GAT-Abl (G5 Bb3.8), GAT-Ab2 (HP-22) and anti-Alp (17C1, 14C3 and 37A4) mAb. Positively charged arginines and lysines are circled. Negatively charged aspartates and glutamates are boxed.
positively charged amino acids in the CDR H2 which are expected, and proposed that the presence of an arginine encoded by the germ-line VHlO and are substituted by anywhere in the combining site could contribute to the non-charged amino acids in the 14C3 and 17C1 monospe- binding to DNA. However, our present results show that cific anti-Alp mAb. Furthermore, in the CDR L3 and H3 of the anti-GATmAb, in spite of a net charge of +lo, and the 37A4, two additional arginines can be observed. These presence of seven arginines in the active site, do not show unpaired positive charges, concentrated in the limited important anti-DNA activity (Fig. 2).This suggests that not region of the binding site, could be important for some only the presence but also the disposition of charges in the cross-reactions. In fact, the results shown in Table 1 and in antibody-binding site relative to the disposition of the Fig. 2 suggest that a correlation can be established between opposite charges in the antigen may be essential. the net charge in the CDR and the degree of crossreactivity. Thus, the net charge of the active site of monospecific mAb is from -2 to + 4 , whereas that of 3.4 VH genes encoding polyvalent antibodies are polyreactive mAb is from +6 to +lo. However, since the frequently utilized in various immune responses net charge of the polyreactive E7 is comparable with those of 22-186 and 37A4, other parameters should be consi- It has been reported that certain clonotypes are usually dered. E7 has 20 aromatic amino acids in the active site, 14 highly represented in newly generated B cell populations of them are tyrosine which, with the hydroxyl group, could [27-31].VH1210.7 encoding the preimmune E7 and VHlOof contribute to both hydrophobic interactions and hydrogen the GATsystem are frequently utilized, in association with bonds. different D and/or J segments and with different L chain genes in various immune responses [8,12]. Having Charged amino acids may participate in the polyvalency of observed that an important number of pseudogenes homoantibodies. Heidelberger [23] reported that anti-pneumo- logous to VHlO are present in the family 5558, Schiff et al. coccal type 8 antibodies cross-react extensively with the [32] proposed that the mechanism of gene conversion specific capsular polysaccharide of streptococcus-pneu- mainly responsible for the generation of diversity in chicken monia type 19 and vice versa. He proposed that the h chains [33] might also be operating in all families of negatively charged phosphoryl-( 1,4)-fl-D-mannose-N-ace- V-gene clusters [34].The preferential utilization of “ V ~ 1 0 tyl glucosamine (type 19) interacts with the active site of like” regions may be related to the observation that a given the anti-pneumococcal type 8 antibodies, in a way similar VH can be associated with only a limited number of VL to the carboxyl groups of cellobiouronic acid (type 8). regions [35]. Certain VH such as VH10, may have a Indeed salt bridges are more energetic and act over larger propensity to associate with various VL gene products. distances than hydrogen bonds, hydrophobic and Van der Alternatively, certain germ-line genes, being able to code Waals interactions. Furthermore, charged residues possess for determined distributions of particular amino acids (as a certain mobility because of their large side chains and charged and aromatic), may generate polyvalent antibodies their interactions are not limited by a given orientation in [2, 81. Clones carrying such polyreactive receptors could be contrast to the highly linear hydrogen bonds. activated during ontogeny by elements present in the internal environment (self epitopeshdiotopes) in a way that could represent the B cell counterpart of T cell-positive selection. Thus, clones recognizing self antigens (for exam3.3 The presence of arginines in the binding site is not ple, G5 Bb2.2-tubulin) could be selected and induced to sufficient to confer anti-DNA activity proliferate when the equilibrium of the internal network is It has been proposed that charged amino acids, and perturbed by the presence of environmental antigens (e.g. arginine in particular, are important for stabilization of the GAT), which bear epitopes identical or similar to those of interactions with DNA [24]. Shlomchik et al. [25] reported self antigens. that the frequency of arginines in the CDR H3 of a collection of anti-DNA mAb was significantly higher than Received December 7, 1989; in revised form April 27, 1990.
Eur. J. Immunol. 1990. 20: 2383-2387
Charged and aromatic amino acids in polyreactive antibodies
4 References 1 Berek, C. and Milstein, C., Immunol. Rev. 1988. 105: 5. 2 Dighiero, G.. Lymberi, €?, MaziC, J. C., Rouyre, S., ButlerBrowne, G. s.,Whalen, R. G. and Avrameas, s.,J. Immunol. 1983. 131: 2267. 3 Dighiero, G., Lymberi, F'., Holmberg, D., Lundquist, I., Coutinho, A. and Avrameas, S., J. lmmunol. 1985. 134: 765. 4 Holmberg, D., Forsgren, S., Ivars, E and Coutinho, A , , Eur. J. Immunol. 1984. 14: 435. 5 Kearney, J. F. and Vakil, M., lmmunol. Rev. 1986. 94: 39. 6 Mahana, W., Guilbert, B. and Avrameas, S., Clin. Exp. Immunol. 1987. 70: 538. 7 Naparstek, Y., Andra-Schwartz, J., Manser, T., Wysocki, L. J., Bretiman, L., Stollar, B., Gefter, M. and Schwartz, R. S., J. Exp. Med. 1986. 164: 614. 8 BaccalB, R., Vo Quang, T.. Gilbert, M., Ternynck, T. and Avrameas, S., Proc. Natl. Acad. Sci. USA 1989. 86: 4624. 9 Rocca-Serra, J., Matthes, H.W., Kaartinen, M., Milstein, C., Thbze, J. and Fougereau, M., EMBO J. 1983. 2: 867. 10 Ollier, €?, Rocca-Serra, J., SommC, G., Thbze, J. and Fougereau, M., EMBO J. 1985. 4: 3688. 11 Roth, C.. Rocca-Serra, J., SommC, G., Fougereau, M. and Thkze, J., Proc. Natl. Acad. Sci. USA 1985. 82: 4788. 12 Schiff, C., Corbet, S. and Fougereau, M., Immunol. Today 1988. 9: 10. 13 Corbet, S., Milili, M., Fougereau, M. and Schiff, C., J. lmmuno/. 1987. 138: 932. 14 Tonnelle, C., Rocca-Serra, J., Moulin, A., Moinier, D. and Fougereau, M., J. Exp. Med. 1983. 158: 1415. 15 Nahmias, C., Strosberg, A. D. and Emorine, L. J., J. lmmunol. 1988. 140: 8152. 16 GonzAlez, R., Charlemagne, J., Mahana,W. and Avrameas, S., Immunology 1988. 63: 31. 17 Vauquelin, G., Geynet, €?,Hanoune, J. and Strosberg, A. D., Eur. J. Biochem. 1979. 98: 543.
2387
18 Guilbert, B., Dighiero, G . and Avrameas, S., J. lmmunol. 1982. 128: 2779. 19 Avrameas, S., Ternynck, T. and Guesdon. J. L., Scand. J. lmmunol. 1978. 8: 7. 20 Chamat, S., Hoebeke, J., Emorine, L., Guillet, J. G. and Strosberg, A. D., J. Immunol. 1986. 136: 3805. 21 SommC, G., Roth, C., Mazie, J.-C., Salem, €? and Theze. J., Eur. J. lmmunol. 1983. 13: 1023. 22 Roth, C., SommC, G., Schiff, C. and Thbze. J., Eur. J. Immunol. 1985. 15: 576. 23 Heidelberg, M., Infect. lmmun. 1983. 41: 1234. 24 Seeman, N. C., Rosenberg, J. M. and Rich, A., Proc. Natl. Acad. Sci. USA 1976. 73: 804. 25 Shlomchik, M., Mascelli, M., Shan, H., Radic, M. Z., Pisetky, D., Marshak-Rothstein, A. and Weigert, M.. J. Exp. Med. 1990. 171: 265. 26 Mandelkow, E. M., Herrmann, M. and Ruhl, U., J. Mol. Biol. 1985. 185: 311. 27 Klinman, N. R., Wylie, D. W. and Cancro, M. P., Proc. Int. Congr. Immunol. 1980. 4: 123. 28 Zharhary, D. and Klinman, N., J. Exp. Med. 1983. 157: 1300. 29 Riley, R. L., Wylie, D. E. and Klinman, N. R.. J. Exp. Med. 1983. 158: 1733. 30 Nishikawa, S.,Toshidata,T. and Rajewsky, K., Eur. J. Immunol. 1983. 13: 318. 31 Juy, D., Primi, D., Sinchez, F! and Cazenave, €?-A., Eur. J. Immunol. 1983. 13: 326. 32 Schiff, C., Milili, M. and Fougereau, M., Eur. J. Immunol. 1985. 4: 1225. 33 Reynaud, C. A., Anquez, C., Grimal, H. and Weil, J.-C., Cell 1987. 48: 379. 34 Fougereau, M. and Schiff, C., lmmunol. Rev. 1988. 105: 6Y. 35 Primi, D., Drapier, A.-M. and Cazenave, F'.-A., J. lmmunol. 1987. 138: 1607.
Eur. J. Immunol. 1990. 20: 2383-2387
Rosana Gonzhlez-Quintialn, Roberto BaccalaO, Pedro Maria Alzari., Clara NahmiasO, Gilbert Mazza., Michel Fougereau. and Stratis Avrameaso Unit6 d’Immunocytochimien, Unit6 d’Immunologie Structurale., Dbpartement d’Immunologie; Unit6 de Biologie MolCculaire des RBcepteursO, Ddpartement de Biotechnologie, Institut Pasteur, Pans and Centre d’Immunologie INSERM-CNRS de MarseilleLuminy., Marseille
Poly (GluaAlaqyrlO) (GAT)-induced IgG monoclonal antibodies cross-react with various self and non-self antigens through the complementarity determining regions. Comparison with IgM monoclonal polyreactive natural antibodies* Previous studies have shown that the antibodies of the preimmune repertoire are able to bind to various auto- and xenoantigens including chemical haptens. Sequence analysis of two such murine monoclonal IgM natural autoantibodies showed that they are encoded by unmutated germ-line variable regions of the light and heavy chain (V, and V,) genes which were also found in various murine immune responses, like phenyl-oxazolone ,dinitrophenyl, arsonate, phosphorylcholine and influenza virus hemagglutinin. These data raised the question as to whether induced antibodies possessing germ-line sequence are also able to react with autoantigens. To study this problem, anti-poly(Gl~~Ala~OTyr~~) (GAT) and anti-alprenolol (Alp) monoclonal antibodies, carrying similar VHand V, genes and the same IgGl isotype, were examined for their capacity to react with several self and non-self antigens. The results showed that: (a) the anti-GAT antibodies tested reacted with different autoantigens, such as murine tubulin, actin and myosin as well as trinitrophenyl (TNP) and bovine serum albumin. Similarly, one of the anti-Alp showed weak reactivities for myosin, DNA, actin and TNP; (b) in contrast two other anti-Alp antibodies did not react with any of the tested antigens. Since the major differences between the oligoreactive anti-GAT and the monoreactive anti-Alp antibodies are in the complementarity determining regions (CDR) our results suggest that the observed cross-reactions are mediated by hypervariable loops. Sequence comparison of these antibodies indicate a possible correlation between cross-reactivity and the presence of aromatic and charged amino acids in the CDR.
1 Introduction The antibody response induced by a given antigen is generally dependent on the B and T cell repertoires available at the time of immunization.These repertoires are dynamic and are the consequence of the antigenic experience of the organism [l]. However, little is known concerning the selection of these repertoires during ontogeny and their maintenance during adult life. Preimmune natural antibodies, which result from the internal activity of the naive immune system, seem to be in large part directed
[I 81731
*
2383
This work was supported by grants from the Ministry of Education and Science of Spain, the Swiss National Science Foundation, Association pour la Recherche sur le Cancer, INSERM and CNRS.
Correspondence: Rosana GonzBlez-Quintial, Scripps Clinic and Research Foundation, Department of Immunology, IMM3,lO 666 North Torrey Pines Road, La Jolla, CA 92037, USA
against self antigens [2, 31. Furthermore, it has been shown in a number of cases that they are involved in a highly connected Id network [4], and that they possess potent regulatory properties [5, 61. This connectivity may be the consequence of their ability to cross-react with several different self and non-self antigens such as cytoskeletal proteins, nucleic acids and chemical haptens [2]. Also, such multiple interactions could be the basis of selection processes and somatic events which give rise to more specific antibodies to exogenous antigens. Along these lines it has been proposed that monospecific antibodies could represent mutant forms of germ-line-encoded preimmune polyand autoreactive antibodies [2,7]. Recently, we have cloned and sequenced the complete cDNA corresponding to two natural IgM mAb (E7 and D23) obtained from normal nonimmunized adult BALB/c mice. These results [8] showed that: (a) the two antibodies are encoded by germ-line genes without mutations; and (b) mutant forms of these genes are encountered in various induced immune responses.
These data raise the possibility that post-immune antibodies with germ-line configurations might possess the capacity Abbreviations: Alp: Alprenolol GAR P ~ l y ( G l u ~ A l a ~ V T y rto ~ ~ )cross-react with several different self and non-self GAT-Abl, GAT-Ab2, GAT-Abl’: Id cascade induced by the antigen GAT, in which GAT-Abl is the responding antibody antigens. The immune response against the random syny r ~ ~ ) has been bearing the Id set, GAT-Ab2 the anti-Id antibody and GAT-Abl’ thetic terpolymer p o l y ( G l ~ ~ A l a ~ 0 T(GAT) the anti-anti-idiotypic antibody with the capacity to bind the extensively studied. Indeed, sequence analysis of mAb original Ag H1, H2,H3: CDRl, 2 and 3 of the H chain derived from the three principal stages (Abl, Ab2, Ab3/Abl’) of the GAT-induced Id cascade has shown that L1, L2, L3: CDR1, 2 and 3 of the L chain 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
014-2980/90/1111-2383$3.50+ .25/0
2384
R. GonzBlez-Quintial, R. Baccalh, P. M. Alzari et al.
oneVH (VH10) and two very homologousV, (5.1 and 1.A5) are expressed in their germ-line configurations by the majority of Abl and Abl’; Ab2 also seem to be germ-line encoded [9-141. Here we report results showing that A b l and in some cases Abl’ with anti-GAT activity cross-react with a number of other self and non-self antigens. In contrast, two mAb induced by the hapten alprenolol (Alp), which are encoded by mutated forms of the sameVH and virtually identical V, genes as the A b l and Abl’ of the GAT system [15], were found to be monospecific. Since the major differences between anti-GAT and anti-Alp antibodies are concentrated in the complementarity determining regions (CDR), our results suggest that the observed cross-reactions are mediated by the hypervariable loops.
2 Materials and methods 2.1 Antigens BALB/c muscle actin, myosin, brain tubulin and TNP-BSA were prepared as described previously [16]. Alp-BSA was prepared following the method of Vauquelin et al. [17].The synthetic random terpolymer GAT was kindly donated by Dr. C. Roth (Pasteur Institut, Paris, France). Native double-stranded (ds)DNA and BSA were purchased from Sigma Chemical Co. (St. Louis, MO) and IBF,Villeneuvela-Garenne, France. 2.2 Antibodies Sheep antibodies to mouse Ig were isolated using the corresponding immunosorbent [181 and labeled with P-galactosidase (a gift from Dr. Ullman, Pasteur Institut) using the glutaraldehyde procedure [19]. Rabbit or goat antibodies directed against mouse IgM and IgGl H chains and L chains were purchased from Southern Biotechnology Associates (Birmingham, AL). 2.3 Preimmune and induced mAb The preimmune mAb (E7 and D23) were obtained by fusing spleen cells from normal 12-week-old BALB/c mice [2]. The production and characterization of anti-Alp and GAT (Abl, Ab2 and Abl’) hybridomas have been described elsewhere [20-221. Anti-Alp mAb were purified from ascitic fluid by DEAE chromatography (Pharmacia, Uppsala, Sweden), whereas GAT system mAb were purified by D E 52 chromatography (Whatman, Maidstone, GB), followed by a Sephadex G-150 gel filtration.
Eur. J. Immunol. 1990. 20: 2383-2387 washed three times with PBS containing 0.1% Tween 20 (PBS-T) and incubated with mAb ranging from 100 ng/ml to 50 yglml in PBS-T containing 0.5% gelatin (PBS-T-G) for 1 h at 37 “C and overnight at 4 “C. After washing with PBS-T, the plates were incubated for 1 h at 37°C with P-galactosidase-labeled anti-mouse Ig antibodies (1 yg/ml). After washing, the enzyme substrate was added and the subsequent steps were performed as described elsewhere [181.
3 Results and discussion 3.1 Three IgM and four IgG germ-line-encoded mAb are polyreactive and bind to auto- and xenoantigens Induced mAb were compared with preimmune mAb for their capacity to cross-react with self and non-self antigens. Ten previously characterized mAb were considered in this study; five were from the GATsystem, three from the Alp system and two, obtained from normal nonimmunized mice, were identified as preimmune polyreactive natural autoantibodies. Fig. 1 summarizes the reported data concerning V-region genes coding for these mAb. The preimmune E7 and D23 mAb seem to be coded for by different combinations of non-mutated germ-line genes [8]. Similarly, it has been shown that mAb from the GAT-induced idioypic cascade GAT-Abl, GAT-Abl’ and GAT-Ab2 mAb are encoded by non-mutated germ-line genes [9-111. The two anti-Alp mAb 14C3 and 17C1, use, with a limited number of somatic mutations, the same VH and a very similar V, as the GAT-Abl and the GAT-Abl’,whereas the third anti-Alp, 37A4 uses the same VH associated to a different L chain [15]. Analyzed by ELISA, the anti-Alp mAb 17C1 and 14C3 seem to be monospecific and reacted only with Alp (Fig. 2). Similarly, the GAT-Ab2 (HP-22) did not react with any of the tested antigens. The third anti-Alp mAb, 37A4 showed only a weak cross-reaction with actin,TNP and myosin. In contrast, the tested GAT-Abl (G5 Bb2.2, G8 Ca1.7 and G8 Ad3.8) and also the GAT-Abl’ (22-186), like preimmune natural autoantibodies, showed significant crossreactivity, mainly with tubulin but also withTNP, BSA and actin.The anti-tubulin activity of the GAT-Abl‘ (22-186) is comparable to its anti-GATactivity. Except for the anti-Alp mAb, none of the tested mAb cross-reacted with Alp. In contrast, the antigen GAT was recognized by both the preimmune E7 and D23. In particular, the anti-GAT activity of D23 exceeded that of the GAT-Abl’ (22-186). 3.2 The observed cross-reactions are related to the structure of the CDR: possible role of charged and aromatic amino acids
2.4 ELISA Preimmune (E7 and D23), anti-GAT (G5 Bb2.2, G8 Ca1.7, G8 Ad3.8,22-186 and HP22) and anti-Alp (14C3,17C1 and 37A4) mAb were titrated using an ELISA. For the coating of microtiter plates, the protein antigens were used at 5 yg/ml;TNP-BSA and single-stranded (ss) DNA were used at 1 yg/ml and 500 yg/ml, respectively; Alp-BSA and GAT were used at 10 pg/ml. After coating, the plates were
As mentioned above, anti-GAT mAb (Abl and Abl’) and anti-Alp mAb have the same isotype (IgG1) and are encoded by very homologous VH and V, genes. The major structural differences are due to different D segments and to somatic mutations in the anti-Alp mAb, and are limited to the CDR suggesting that the analyzed GAT-induced IgG mAb cross-react with various self and non-self antigens through the CDR.
Charged and aromatic amino acids in polyreactive antibodies
Eur. J. Immunol. 1990. 20: 2383-2387
-
El 023
T I*
I*
~5
m
G9
ca 1.1
G9
a ANrIEEDIEs
w
2.2
3.8
22-186
Preirmw Preinmm
V-~IaiS
W U10.1 (3660) W 101 (Qvl
1 6 1 GAT (Abll IqGl GAT IAblI IgGl GAT IAblI
W 10 W 10
(55581 155581 (55581
I@
GAT IAbl'
W 10
155581
Wi 10 Wi 10 W 10
15558) 15558) 155581
17c1
I*l
140 31A4
161 161
anti-Alp anti-Alp anti-Alp
w22
IS1
CAT (PI21
I@ I@
W
10
Pre-
W ? (CRXIP 19)
Preirmrne
W ? -1
G5 Bb 2.2
1 6 1 CAT IAblI G9 ca 1.7 1 6 1 CAT IAbll G9 W 3.0 IgGl CAT @blI
8)
W W (&Em 11 W W -1 W W -1
22-186
I*
31A4
IqGl anti-Alp 1 6 1 anti-Alp 1 8 1 anti-Alp
w22
IqGl CAT (MJ M ?
1) 1)
GRT IAbl'J W W (QUIP 11
W 5.1 ICRzR 1) W 5.1 - 1
W ?
Figure 1. Schematic representation of variable
1)
ICRXIP ?I
regions. CDR are boxed. The amino acid differences to most homologous germ-line V genes are represented by A.
(CRXIP?)
Actin
I
1500
Table 1. Characteristics of the antibody-binding sites
mAb
1000
E7 D23
500
G5 Bb2.2 G8 0 1 . 7 G8 Ad3.8 22-186
0
-E 2 w
5 P
1500
1000
500
Pp z a
Net Arginines Aromatic charge amino acids
Tyrosines
+6 +9
4 7
20 17
14 8
+10 +10
7 7
16 16 15 18
7 7 8 12
+8
5
+7
3
37A4 17C1 146
+6 +2 +4
4
3 3
14 14 15
6
HP-22
-2
4
19
13
7 6
o
U
g
2385
Anti-GATmAb cross-react with tubulin. Sequence analysis of tubulin indicates that there are many glutamate residues in the carboxyl end of both the a and subunits [26]. Furthermore, in the same regions, the sequences GluGlu-Tyr can often be observed. It seems possible that anti-GAT mAb interact with these regions of tubulin molecules since they represent a dominant epitope of the GAT antigen.
TNP-BSA 1500
1000
500
0
rn
As shown in Fig. 3, most of the structural differences between homologous poly- and monoreactive mAb concern charged amino acids. Frequently, positively charged 750 amino acids situated in the CDR of the polyreactive GAT-induced mAb, have neutral or negatively charged 500 amino acids counterparts in the corresponding regions of 250 monospecific Alp-induced mAb, which have a prevalence of negatively charged amino acids and show in the CDR H3 n a typical pocket formed by acidic residues. This pocket 0.1 1.0 10 100 0.1 1.0 10 100 possibly involved in interactions with Alp [15], does not, pg/ml Ab however, preclude a weak interaction of one of the anti-Alp Figure2. Binding of mAb to myosin-, actin-, tubulin-, BSA-, TNP-BSA-, ssDNA-, GAT- and Alp-BSA-coatedplates. Decreas- antibodies (37A4) with actin, DNA, TNP and myosin ing concentrations of E7 (0),D23 (O), G5 Bb2.2 (+),G8 Ca1.7 (Fig. 2). The structural differences between 37A4 and the ( O ) , G8 Ad3.8 (H), 22-186 (A),HP22 (U), 14C3 (A), 17C1 (0) monospecific 14C3 and 17A4 (Fig. 3) could give interesting and 37A4 (a)were incubated with the different plates. insights into the basis of these interactions. 37A4 has two 1000,
GAT
I
Al renolol-BSA
Eur. J. Immunol. 1990. 20: 2383-2387
R. Gonzalez-Quintial, R. Baccalh, I? M. Alzari et al.
2386
H2 E7 023
HPZ
H3
M G Y I S Y S G S T Y Y N P S L @ - S L G V I W S G G S T ~ Y N A A F I S -
I
T I ~ G - WIN Y
w
N
G I I~ N Y
M
~
s
GN N
-
Y N P s LB+I
L1
L2
A S B N V V T - - - - - - Y V S W Y SSQSLLNS@T@@NYLAWY
E7
D23
S S S S
S S S S
Q Q Q Q
S S S S
I I I L
V V V V
H H H H
S S S S
-
N N N N
G G G G
N N N N
T T T T
Y Y Y Y
L L L L
Y Y E H
W W W W
S C g S S Q S L V H S - N G I T Y L H W H S C S S Q S L V H S - N G B T Y L H W H T C T A@S S V S S S - - - - - Y F H W Y
RP22
T
14c3
C
@A
S Q m I S V
------
I Y G A S N Y T G V I Y W I G V
Y Y Y Y
31A4
17C1
@PFYG@GAM------D W G m@L@L@Y F - - - - - - - 1 j W G
Y L NW L
1 y1s T
s
r
LEJSIG
v
Figure3. Amino acid representation of CDR and two flanking residues of preimmune (E7 and D23), GAT-Abl (G5 Bb3.8), GAT-Ab2 (HP-22) and anti-Alp (17C1, 14C3 and 37A4) mAb. Positively charged arginines and lysines are circled. Negatively charged aspartates and glutamates are boxed.
positively charged amino acids in the CDR H2 which are expected, and proposed that the presence of an arginine encoded by the germ-line VHlO and are substituted by anywhere in the combining site could contribute to the non-charged amino acids in the 14C3 and 17C1 monospe- binding to DNA. However, our present results show that cific anti-Alp mAb. Furthermore, in the CDR L3 and H3 of the anti-GATmAb, in spite of a net charge of +lo, and the 37A4, two additional arginines can be observed. These presence of seven arginines in the active site, do not show unpaired positive charges, concentrated in the limited important anti-DNA activity (Fig. 2).This suggests that not region of the binding site, could be important for some only the presence but also the disposition of charges in the cross-reactions. In fact, the results shown in Table 1 and in antibody-binding site relative to the disposition of the Fig. 2 suggest that a correlation can be established between opposite charges in the antigen may be essential. the net charge in the CDR and the degree of crossreactivity. Thus, the net charge of the active site of monospecific mAb is from -2 to + 4 , whereas that of 3.4 VH genes encoding polyvalent antibodies are polyreactive mAb is from +6 to +lo. However, since the frequently utilized in various immune responses net charge of the polyreactive E7 is comparable with those of 22-186 and 37A4, other parameters should be consi- It has been reported that certain clonotypes are usually dered. E7 has 20 aromatic amino acids in the active site, 14 highly represented in newly generated B cell populations of them are tyrosine which, with the hydroxyl group, could [27-31].VH1210.7 encoding the preimmune E7 and VHlOof contribute to both hydrophobic interactions and hydrogen the GATsystem are frequently utilized, in association with bonds. different D and/or J segments and with different L chain genes in various immune responses [8,12]. Having Charged amino acids may participate in the polyvalency of observed that an important number of pseudogenes homoantibodies. Heidelberger [23] reported that anti-pneumo- logous to VHlO are present in the family 5558, Schiff et al. coccal type 8 antibodies cross-react extensively with the [32] proposed that the mechanism of gene conversion specific capsular polysaccharide of streptococcus-pneu- mainly responsible for the generation of diversity in chicken monia type 19 and vice versa. He proposed that the h chains [33] might also be operating in all families of negatively charged phosphoryl-( 1,4)-fl-D-mannose-N-ace- V-gene clusters [34].The preferential utilization of “ V ~ 1 0 tyl glucosamine (type 19) interacts with the active site of like” regions may be related to the observation that a given the anti-pneumococcal type 8 antibodies, in a way similar VH can be associated with only a limited number of VL to the carboxyl groups of cellobiouronic acid (type 8). regions [35]. Certain VH such as VH10, may have a Indeed salt bridges are more energetic and act over larger propensity to associate with various VL gene products. distances than hydrogen bonds, hydrophobic and Van der Alternatively, certain germ-line genes, being able to code Waals interactions. Furthermore, charged residues possess for determined distributions of particular amino acids (as a certain mobility because of their large side chains and charged and aromatic), may generate polyvalent antibodies their interactions are not limited by a given orientation in [2, 81. Clones carrying such polyreactive receptors could be contrast to the highly linear hydrogen bonds. activated during ontogeny by elements present in the internal environment (self epitopeshdiotopes) in a way that could represent the B cell counterpart of T cell-positive selection. Thus, clones recognizing self antigens (for exam3.3 The presence of arginines in the binding site is not ple, G5 Bb2.2-tubulin) could be selected and induced to sufficient to confer anti-DNA activity proliferate when the equilibrium of the internal network is It has been proposed that charged amino acids, and perturbed by the presence of environmental antigens (e.g. arginine in particular, are important for stabilization of the GAT), which bear epitopes identical or similar to those of interactions with DNA [24]. Shlomchik et al. [25] reported self antigens. that the frequency of arginines in the CDR H3 of a collection of anti-DNA mAb was significantly higher than Received December 7, 1989; in revised form April 27, 1990.
Eur. J. Immunol. 1990. 20: 2383-2387
Charged and aromatic amino acids in polyreactive antibodies
4 References 1 Berek, C. and Milstein, C., Immunol. Rev. 1988. 105: 5. 2 Dighiero, G.. Lymberi, €?, MaziC, J. C., Rouyre, S., ButlerBrowne, G. s.,Whalen, R. G. and Avrameas, s.,J. Immunol. 1983. 131: 2267. 3 Dighiero, G., Lymberi, F'., Holmberg, D., Lundquist, I., Coutinho, A. and Avrameas, S., J. lmmunol. 1985. 134: 765. 4 Holmberg, D., Forsgren, S., Ivars, E and Coutinho, A , , Eur. J. Immunol. 1984. 14: 435. 5 Kearney, J. F. and Vakil, M., lmmunol. Rev. 1986. 94: 39. 6 Mahana, W., Guilbert, B. and Avrameas, S., Clin. Exp. Immunol. 1987. 70: 538. 7 Naparstek, Y., Andra-Schwartz, J., Manser, T., Wysocki, L. J., Bretiman, L., Stollar, B., Gefter, M. and Schwartz, R. S., J. Exp. Med. 1986. 164: 614. 8 BaccalB, R., Vo Quang, T.. Gilbert, M., Ternynck, T. and Avrameas, S., Proc. Natl. Acad. Sci. USA 1989. 86: 4624. 9 Rocca-Serra, J., Matthes, H.W., Kaartinen, M., Milstein, C., Thbze, J. and Fougereau, M., EMBO J. 1983. 2: 867. 10 Ollier, €?, Rocca-Serra, J., SommC, G., Thbze, J. and Fougereau, M., EMBO J. 1985. 4: 3688. 11 Roth, C.. Rocca-Serra, J., SommC, G., Fougereau, M. and Thkze, J., Proc. Natl. Acad. Sci. USA 1985. 82: 4788. 12 Schiff, C., Corbet, S. and Fougereau, M., Immunol. Today 1988. 9: 10. 13 Corbet, S., Milili, M., Fougereau, M. and Schiff, C., J. lmmuno/. 1987. 138: 932. 14 Tonnelle, C., Rocca-Serra, J., Moulin, A., Moinier, D. and Fougereau, M., J. Exp. Med. 1983. 158: 1415. 15 Nahmias, C., Strosberg, A. D. and Emorine, L. J., J. lmmunol. 1988. 140: 8152. 16 GonzAlez, R., Charlemagne, J., Mahana,W. and Avrameas, S., Immunology 1988. 63: 31. 17 Vauquelin, G., Geynet, €?,Hanoune, J. and Strosberg, A. D., Eur. J. Biochem. 1979. 98: 543.
2387
18 Guilbert, B., Dighiero, G . and Avrameas, S., J. lmmunol. 1982. 128: 2779. 19 Avrameas, S., Ternynck, T. and Guesdon. J. L., Scand. J. lmmunol. 1978. 8: 7. 20 Chamat, S., Hoebeke, J., Emorine, L., Guillet, J. G. and Strosberg, A. D., J. Immunol. 1986. 136: 3805. 21 SommC, G., Roth, C., Mazie, J.-C., Salem, €? and Theze. J., Eur. J. lmmunol. 1983. 13: 1023. 22 Roth, C., SommC, G., Schiff, C. and Thbze. J., Eur. J. Immunol. 1985. 15: 576. 23 Heidelberg, M., Infect. lmmun. 1983. 41: 1234. 24 Seeman, N. C., Rosenberg, J. M. and Rich, A., Proc. Natl. Acad. Sci. USA 1976. 73: 804. 25 Shlomchik, M., Mascelli, M., Shan, H., Radic, M. Z., Pisetky, D., Marshak-Rothstein, A. and Weigert, M.. J. Exp. Med. 1990. 171: 265. 26 Mandelkow, E. M., Herrmann, M. and Ruhl, U., J. Mol. Biol. 1985. 185: 311. 27 Klinman, N. R., Wylie, D. W. and Cancro, M. P., Proc. Int. Congr. Immunol. 1980. 4: 123. 28 Zharhary, D. and Klinman, N., J. Exp. Med. 1983. 157: 1300. 29 Riley, R. L., Wylie, D. E. and Klinman, N. R.. J. Exp. Med. 1983. 158: 1733. 30 Nishikawa, S.,Toshidata,T. and Rajewsky, K., Eur. J. Immunol. 1983. 13: 318. 31 Juy, D., Primi, D., Sinchez, F! and Cazenave, €?-A., Eur. J. Immunol. 1983. 13: 326. 32 Schiff, C., Milili, M. and Fougereau, M., Eur. J. Immunol. 1985. 4: 1225. 33 Reynaud, C. A., Anquez, C., Grimal, H. and Weil, J.-C., Cell 1987. 48: 379. 34 Fougereau, M. and Schiff, C., lmmunol. Rev. 1988. 105: 6Y. 35 Primi, D., Drapier, A.-M. and Cazenave, F'.-A., J. lmmunol. 1987. 138: 1607.
