Scand. J. Immunol 35, 677-680. 1992
Resistance to Collagen-Induced Arthritis in Biozzi Mice is Not Associated With a Defect in Autoantibodies to Accessible Epitopes on Cartilage Collagen J,-P. BOUVET. B. FRANC*, A. ZYADf & J. COUDERCf Unite dTmmunologie Microbienne. Institut Pasteur. Paris, *Laboratoire d'Anatomie Pathologique, Hopital Ambroisc Pare. Universiie Paris V. and tUnite d'Immunogenetiquc CNRS URA 1413, Institut Curie, Paris, France
Bouvet J-P, t'ranc B, Zyad A. Couderc J. Resistance lo Collagen-Induced Arthritis in Biozzi Mice is Not Associated With a Defect in Aiiloantibodies to Accessible Epitopes on Cartilage Collagen. Scand J Immtinol 1992:35:677 80 In order lo understand why most strains of Biozzi mice are unexpectedly either arthritis-resistant, or susceptible, we investigate ihe aulonntibody reactivity against matrix-embedded collagen molecules. We show a close correlation between these tissular and the IVee collagen antibodies, irrespective olarthritis susceptibility. These results arc against an antibody response restricted to hidden non-pathogenic epitopes in the resistant mice. Dr Jean-Pierre Bourct, Unite d'lnmmnotogie Miciohieiuic. liistitui Pasteur. 2N rue du Dr. Kou.x, 75724. Paris 15, France
Arthritis induced by type II collagen (CII) is associated with specific antibodies (Ab). which were first considered as the main cause of joint damage [1,2], However, further studies from dilTerent groups have indicated that other mechanisms are also involved. Holmdahl ei at. showed that CD4' Tcells reactive with CII could transfer disease to naive DBA/1 miee [3], and these cells were recently shown to include two populations acting syncrgistically in the development of arthritis [4]. The group of C. David reported a strain of mice (SWR), with a large defect in T-cell receptor V/f genes, which was resistant to arthritis despite high levels of CII autoAb (reviewed in Ref. 5). Moreover these Ab included lgG2a molecules: the subclass known as putatively the most pathogenic [6], since it strongly activates the complement cascade. In a previous study, we investigated the arthritis susceptibility of mice selected for their high (H) or low (L) Ab responsiveness to unrelated antigens [7, 8], Among these 'Biozzi mice", we observed that Iwo strains (Hii imd HG). producing high levels of lgG2a CII autoAb. were resistant to arthritis, whereas the
low responder Lii mice were severely affected [9]. These data demonstrated that most CII autoAb can be harmless, but did not rule out an eventual role of a small subset of Ab directed to the CII epilof>es which are not hidden by other macromolecules. such as proteoglycans, wilhin the hyalin matrix of the cartilage. In this study., we investigate the cartilage binding of CII Ab from the different strains of Biozzi mice. We show this binding to correlate with the titre of total CII autoAb, irrespectively of the strain susceptibility to this experimental model of rheumatoid arthritis. This rules out the hypothesis of an Ab response restricted to hidden epitopes in the resistant strains.
MATERIALS A N D METHODS Strains. Mice, bred in tbe animal unit of the Laboratory orinimunogcneticsal the Institut Curie, were from three different Bio/,zi selections: I, II [7J and G [8J: each included two diverging strains: H und L (Table I). Artbrili.s was induced after I or 2 injections of chicken CII. as already reported (9], the purity of the CII 677
678 J.p. Bouvet et ai. TABLL 1. Cll-indticed arthritis and autoantibodies in different strains of Biozzi mice CII autoantibodies Strain
MHC*
CII arthrilist
ELISAJ
Ht Hn
H-2''
HG
heterogeneous
0.9 >2
LI LII LG
heterogeneous heterogeneous
0.4 0.2 0.2
R
Immunomicroscopy§
0
• Major histocompatibility complex; +Sensitiviiy to collagen-indticed arthritis with chick collagen II (Cli). The number of crosses indicates the severity and frequency of joint involvement {R =arthritis-resislant); ^Optical absorbance at 492 nm for a 1:50 dilution of the pooled sera collected at Ihe clinical onset and/or at the maximum of antibody titre: ^Results ofthe immunomicroscopy wilh mouse cartilage. The number of crosses indicates the intensity ofthe staining (0 = no staining).
preparation (Gcn/.ymc, Boston, MA) being controlled by SDS-PAGE analysis and by lack of slaining witfi toliiidine bltie. The proportions of arthritic mice were 12/18. II..'19, and 23,24 in the Hi. Ll and Lii strains, re.spectively. The selected sera corresponded lo Ihc onsei of obvious joint involvement for each individual animal in the arthritic strains (Hi. Li, Lii), or to lhe peak of autoAb in the resislant strains (HIT, HU, LU). Five sera from each strain were pooled, and preimmune samples, as well as pools absorbed with chicken CM, served as normal controls. CII auto.Ah, Serum pools were diluted 1.100 and tesied by etizyme-liiiked immunosorbenl assay (ELISA) againsl mouse CII, purified from the cartilage ofthe mouse xiphoid process according to the method ofTrenlham ct al. [10]. Plales were coated with CII (10 /(g;ml) and overcoaled with bovine serum albumin and Tween 20. A peroxidase-labelled anti-mouse Ig was used as second Ab (Cappel, Coekranville, PA). RestiUs were expressed as optical absorbance units al 492 nm itntuuiumiicyoscopy- Ereshly exeised .\iphoid processes were frozen wilh liquid nilrogen, and sectioned for microscopic examination without fixation. Slices were coated with a 1:10 dilution of normal human scrum to prevent non-specilic binding, then incubated with dilutions of the mice sera. Peroxidase-labelled sheep (Fab); anti-mouse Ig (Biolyon, Dardilly, France) served as the second Ab. Revelation was done wilh ortho-phenylenediamine. A counterstain wilh haematoxylin allowed further visualization of lhe microscopic structures.
remained negalive. A strong positive correlation was observed between the results ofboth assays: scrum pools which had the highest levels of CII Ab ( H G , H I and Hn) provided the strongest staining in immunomicroscopy. whereas pools from L strains exhibited faintly positive (Li and Lii) or negalive (LG) results (Table I). No correlation was observed between cartilage binding and arthritis susceptibility: the serum pool from the severely affected Lri mice led to only a faint staining, whereas the apparently harmless Ab from ihc H G line provided lhe strongest positivity (Fig, I). Moreover, serum pools from Hi and Hii mice -two closely related strains, arlhritis-scnsitive ( H I ) and resistant (Hii), respectivelyshowed the same positive results. Histological features of Ab binding were similar in all cases, depending only on the staining intensity. Extracellular materials of chondroplastic areas were strongly detected, and the hyalin matrix was hazily stained (Fig. 1). Absorption of the CII Ab with pure chicken CII abolished positivily and thus confirmed specificity of the staining.
DISCUSSION
RESULTS All sera positive for CII autoAb. using HLISA, also gave positive results by immunomicroscopy. whereas the LG pool and preimmune sera
In an effort to understand the pathogeny of CIIinduced arthritis, we investigated the reason why some strains of mice are sensitive to this disease whereas related strains are resistant. Using strains issued from the selective breeding of outbred Swiss mice, according to their H or L Ab
Collagen-Induced Arthritis in Biozzi Mice
679
FIG. !. Immunoperoxtdase staining of eollagen ll-specific mouse aniibodies bound to mouse eartilagf. Normal sternum cartilage was incubated wilh serum pools from Cll-ltiimuntzed (ii) or naive (b) HG Biozzi mice, and binding was reveaied with F(ab'); anti-mouse IgG antibodies, labelled with peroxidase. The sltees were further stained with haemaloxylin. Although Hr, mtccdid nol develop arthritis, their high levels of autoantibodies included specificities to nascent CII — in the ehondroplastic areas (arrowheads) — and to nalive CII embedded in ihe hyalin matrix (brownstained in the original picture). No binding was observed with the negative control pool; the matrix was blue-stained, and the sole nuclei of chondroblasts (dark blue) were clearly apparent. Final magnification: x 300.
responses, we observed unexpected inter-strain variations in lerms of arthritis susceptibility, which allowed a new genetic approach of this experimental disease. The most striking variation was observed between the Hi and Hu lines (Table I)—issued from two independent sister selections, and bearing the same H-2'^ histocompatibility complex—which exhibited similar CII autoAb responses, whereas only one (Hi) was susceptible to arthritis [9]. Moreover when aged, the sole Hi developed a spontaneous RA-like disease associated with IgG2a CII Ab [11], suggesting that a common mechanism regulates the sensibility for both diseases in these strains. In agreement with a parallel study on SWR mice [12], we had already provided evidence that the arthritis resistance of Hit mice was not explained by a V/? gene deletion [ 13]. We show here that this resistance, and that of the HG mice, also cannot be due to a lack of Ab
response against the CII epitopes which are accessible within the cartilage matrix. CII autoAb are usually detected with a mouse antigen extracted from the cartilage matrix by a several-step method, including solubilization and pepsin digestion in acidic media [10]. Hence the purified molecule is no longer under a fully native state and the corresponding Ab can be directed against epitopes that are not apparent in vivo. Indeed, variations have been reported, in terms of recognition by autoAb, between regular and 'renatured'CI! fragments obtained after cleavage with cyanogen bromide [14]. Moreover, autoAb to in vivo hidden epitopes cannot react with tissue molecules. Nascent CII produced by chondrocytes is in a free and native form but is not accessible to Ab since it is located in ehondroplastic areas that are surrounded by the carlilage matrix. In contrast, CII molecules ofthe matrix
680
J-p.
Bouvet et al.
are covered by the proteoglyeans, and thus have few exposed epitopes, but can be reached at the cartilage border by IgG ofthe synovial fluid. Our results demonstrate that Ab from the strains (Hii and H G ) resistant to arthritis are not restricted to denatured epitopes since they heavily bound nascent CII. They also show that some of these Ab from resistant mice recognize the matrixembedded CI! molecules, indicating a specificity even for putatively arthritogenic epitopes. These results are in agreement with recent data showing that multiple CI! epitopes, recognized by different monoclonal antibodies, are accessible in vivo [15]. Passive transfers of CII arthritis require extremely high quantities of immunoglobulins, and. in our hands, could not be achieved with physiological doses of Ab from Biozzi mice (unpublished data). Moreover, the normal complement functions in these strains, and here the binding intensity, correlating with the autoAb titrc and not with the sensitivity to arthritis, are against the hypothesis of a key role of CII autoAb. It thus appears that direct injury of cartilage by complement-activating autoantibodies, directed againsi exposed epitopes ofthe matrix-embedded CII molecules, cannot be the primary mechanism of CII induced-arthritis which seems to mainly involve autoimmune CD4' Tcells[16, 17].
REFERENCES 1 Trcntham DE. Townes AS, Kang AH. David JR. Humoral and cellular sensitivity to collagen in type II collagen-induced arthritis in rats. J CItn Invest 1978:61:89-96. 2 Stuart JM. Cremcr MA, Townes AS, Kang AH. Type II collagen-induced arlhritis in rats. Passive transfer with serum and evidence that IgG antieollageii antibodies can cause arthritis. J F.xp Med l9S2;I55:t-16. 3 Holmdahl R, Klare.skog G, Rubin K, Larsson E, Wigzell H. T lymphoeytes in collagen Il-induced arthritis in mice: characterization of arthritogenic collagen ll-specific T-cell lines and clones. Scand J [mmunol I985;22:295 306. 4 Andersson M, Holmdahl R. Analysis of type II collagen-reaclive T cells in the mouse. I. Different regulation of autorcactive vs. non-autoreactive antitype II collagen T cells in the DBA-1 mouse. Eur J Immunol 1990:20:1061-66. 5 David CS. Genes for MHC. TCR and Mis determine suscepttbiltty to collagen induced arthritis. Aeta Palhol Microbiol Immunol Seand 1990:98:575 84.
6 Watson WC, Townes AS. Genetie susceptibility to murine collagen II autoimmune arthritis: proposed relationship lo the IgG2a autoantibody subclass response, complement C5, major histocompatibility eomplex jMHC) and non-MHC loci. J Exp Med 1985:162:1878-91. 7 Biozzi G. Mouton D, Sant'Anna OA, Passos HC, Gennari M. Reis MH, Ferreira WCA. Heumann AM, Bouthillier Y. Ibanez OM, StifTel C, Siqueira M. Genetics of immunoresponsivcness to natural antigens in the mouse. Curr Top Microbiol Immunol 1979;85:3l-98. S Mouton D. Siqueira M, Sant'Anna OA, Bouthillier Y, Ibanez O. Ferreira VCA, Mevel JC. Reis MH, Piatti RM, Stiffel C, Biozzi G. Genetic regulation of multispecific antibody responsiveness: improvement of'high' and 'low" characters. Eur J Immunol 1988:18:41^9. 9 Bouvet JP, Couderc J, Bouthillier Y, Franc B, Decreusefond C, Mouton D. Collagen arthritis in Biozzi mice: joint involvement is not correlated with collagen II lgG2a autoanlibodies nor restricted to only H-:" and H-2^ J Immunol 1989:143:1537-42. 10 Trentham DE. Townes AS, Kang AH. Autoimmunily to type 11 collagen: an experimental model of arthritis. J Exp Med 1977:146:857 68. 11 Bouvet JP, Couderc J, Bouthillier Y, Franc B, Ducailar A. Mouton D. Spontaneous rheumaloidlike arthritis in a line of mice sensitive to collageninduced arthritis. Arthr Rheum 1990:33:1716-22. 12 Andersson M. Goldschmidt TJ. Michaelsson E, Larsson A, Holmdahl R. T-cell receptor V/f haplotype and complement componeni C5 play no significant role for the resistance to collagen-induced arthritis in the SWR mouse. Immunology I99I;73:I91 6. 13 Vidard L, Roger T, Bouvet JP, Couderc J. Seman M. Resistance lo collagen-induced arthritis in Btozzi mice is not associated with T cell receptor V^ gene polymorphism. Eur J Immunol 1991:21:1783 85. 14 Terato K. Hasty KA, Cremcr MA. Stuart JM, Townes AS, Kang AH. Collagen-induced arthritis in mice: localization of an arlhritogenie determinant loa fragment of the type II collagen molecule. J Exp Med l985;l62:637-46. 15 Holmdahl RT, Mo JA. Jonsson R, Karlslrom K. Seheynius A. Mulliple epitopes on cartilage type IT eollagen are accessible for antibody binding in vivo. Autotmtnunity 1991:10:27 34. 16 Brawn E. Trentham DE. Experimental synovitis induced by collagen-specific T cell lines. Cell Immunol 1989:118:491 503. 17 Andersson M, Cremer MA, Terato H, Burkhardt H, Homldahl R. Analysisoftype II collagen reactive T cells in the mouse II. Different localization of immunodominant T ceil epilopes on heterologous and autologous type il collagen. Scand J Immunol 1991:33:505-10.
Received 20 December 1991 Accepted in revised form 24 January 1992
Resistance to Collagen-Induced Arthritis in Biozzi Mice is Not Associated With a Defect in Autoantibodies to Accessible Epitopes on Cartilage Collagen J,-P. BOUVET. B. FRANC*, A. ZYADf & J. COUDERCf Unite dTmmunologie Microbienne. Institut Pasteur. Paris, *Laboratoire d'Anatomie Pathologique, Hopital Ambroisc Pare. Universiie Paris V. and tUnite d'Immunogenetiquc CNRS URA 1413, Institut Curie, Paris, France
Bouvet J-P, t'ranc B, Zyad A. Couderc J. Resistance lo Collagen-Induced Arthritis in Biozzi Mice is Not Associated With a Defect in Aiiloantibodies to Accessible Epitopes on Cartilage Collagen. Scand J Immtinol 1992:35:677 80 In order lo understand why most strains of Biozzi mice are unexpectedly either arthritis-resistant, or susceptible, we investigate ihe aulonntibody reactivity against matrix-embedded collagen molecules. We show a close correlation between these tissular and the IVee collagen antibodies, irrespective olarthritis susceptibility. These results arc against an antibody response restricted to hidden non-pathogenic epitopes in the resistant mice. Dr Jean-Pierre Bourct, Unite d'lnmmnotogie Miciohieiuic. liistitui Pasteur. 2N rue du Dr. Kou.x, 75724. Paris 15, France
Arthritis induced by type II collagen (CII) is associated with specific antibodies (Ab). which were first considered as the main cause of joint damage [1,2], However, further studies from dilTerent groups have indicated that other mechanisms are also involved. Holmdahl ei at. showed that CD4' Tcells reactive with CII could transfer disease to naive DBA/1 miee [3], and these cells were recently shown to include two populations acting syncrgistically in the development of arthritis [4]. The group of C. David reported a strain of mice (SWR), with a large defect in T-cell receptor V/f genes, which was resistant to arthritis despite high levels of CII autoAb (reviewed in Ref. 5). Moreover these Ab included lgG2a molecules: the subclass known as putatively the most pathogenic [6], since it strongly activates the complement cascade. In a previous study, we investigated the arthritis susceptibility of mice selected for their high (H) or low (L) Ab responsiveness to unrelated antigens [7, 8], Among these 'Biozzi mice", we observed that Iwo strains (Hii imd HG). producing high levels of lgG2a CII autoAb. were resistant to arthritis, whereas the
low responder Lii mice were severely affected [9]. These data demonstrated that most CII autoAb can be harmless, but did not rule out an eventual role of a small subset of Ab directed to the CII epilof>es which are not hidden by other macromolecules. such as proteoglycans, wilhin the hyalin matrix of the cartilage. In this study., we investigate the cartilage binding of CII Ab from the different strains of Biozzi mice. We show this binding to correlate with the titre of total CII autoAb, irrespectively of the strain susceptibility to this experimental model of rheumatoid arthritis. This rules out the hypothesis of an Ab response restricted to hidden epitopes in the resistant strains.
MATERIALS A N D METHODS Strains. Mice, bred in tbe animal unit of the Laboratory orinimunogcneticsal the Institut Curie, were from three different Bio/,zi selections: I, II [7J and G [8J: each included two diverging strains: H und L (Table I). Artbrili.s was induced after I or 2 injections of chicken CII. as already reported (9], the purity of the CII 677
678 J.p. Bouvet et ai. TABLL 1. Cll-indticed arthritis and autoantibodies in different strains of Biozzi mice CII autoantibodies Strain
MHC*
CII arthrilist
ELISAJ
Ht Hn
H-2''
HG
heterogeneous
0.9 >2
LI LII LG
heterogeneous heterogeneous
0.4 0.2 0.2
R
Immunomicroscopy§
0
• Major histocompatibility complex; +Sensitiviiy to collagen-indticed arthritis with chick collagen II (Cli). The number of crosses indicates the severity and frequency of joint involvement {R =arthritis-resislant); ^Optical absorbance at 492 nm for a 1:50 dilution of the pooled sera collected at Ihe clinical onset and/or at the maximum of antibody titre: ^Results ofthe immunomicroscopy wilh mouse cartilage. The number of crosses indicates the intensity ofthe staining (0 = no staining).
preparation (Gcn/.ymc, Boston, MA) being controlled by SDS-PAGE analysis and by lack of slaining witfi toliiidine bltie. The proportions of arthritic mice were 12/18. II..'19, and 23,24 in the Hi. Ll and Lii strains, re.spectively. The selected sera corresponded lo Ihc onsei of obvious joint involvement for each individual animal in the arthritic strains (Hi. Li, Lii), or to lhe peak of autoAb in the resislant strains (HIT, HU, LU). Five sera from each strain were pooled, and preimmune samples, as well as pools absorbed with chicken CM, served as normal controls. CII auto.Ah, Serum pools were diluted 1.100 and tesied by etizyme-liiiked immunosorbenl assay (ELISA) againsl mouse CII, purified from the cartilage ofthe mouse xiphoid process according to the method ofTrenlham ct al. [10]. Plales were coated with CII (10 /(g;ml) and overcoaled with bovine serum albumin and Tween 20. A peroxidase-labelled anti-mouse Ig was used as second Ab (Cappel, Coekranville, PA). RestiUs were expressed as optical absorbance units al 492 nm itntuuiumiicyoscopy- Ereshly exeised .\iphoid processes were frozen wilh liquid nilrogen, and sectioned for microscopic examination without fixation. Slices were coated with a 1:10 dilution of normal human scrum to prevent non-specilic binding, then incubated with dilutions of the mice sera. Peroxidase-labelled sheep (Fab); anti-mouse Ig (Biolyon, Dardilly, France) served as the second Ab. Revelation was done wilh ortho-phenylenediamine. A counterstain wilh haematoxylin allowed further visualization of lhe microscopic structures.
remained negalive. A strong positive correlation was observed between the results ofboth assays: scrum pools which had the highest levels of CII Ab ( H G , H I and Hn) provided the strongest staining in immunomicroscopy. whereas pools from L strains exhibited faintly positive (Li and Lii) or negalive (LG) results (Table I). No correlation was observed between cartilage binding and arthritis susceptibility: the serum pool from the severely affected Lri mice led to only a faint staining, whereas the apparently harmless Ab from ihc H G line provided lhe strongest positivity (Fig, I). Moreover, serum pools from Hi and Hii mice -two closely related strains, arlhritis-scnsitive ( H I ) and resistant (Hii), respectivelyshowed the same positive results. Histological features of Ab binding were similar in all cases, depending only on the staining intensity. Extracellular materials of chondroplastic areas were strongly detected, and the hyalin matrix was hazily stained (Fig. 1). Absorption of the CII Ab with pure chicken CII abolished positivily and thus confirmed specificity of the staining.
DISCUSSION
RESULTS All sera positive for CII autoAb. using HLISA, also gave positive results by immunomicroscopy. whereas the LG pool and preimmune sera
In an effort to understand the pathogeny of CIIinduced arthritis, we investigated the reason why some strains of mice are sensitive to this disease whereas related strains are resistant. Using strains issued from the selective breeding of outbred Swiss mice, according to their H or L Ab
Collagen-Induced Arthritis in Biozzi Mice
679
FIG. !. Immunoperoxtdase staining of eollagen ll-specific mouse aniibodies bound to mouse eartilagf. Normal sternum cartilage was incubated wilh serum pools from Cll-ltiimuntzed (ii) or naive (b) HG Biozzi mice, and binding was reveaied with F(ab'); anti-mouse IgG antibodies, labelled with peroxidase. The sltees were further stained with haemaloxylin. Although Hr, mtccdid nol develop arthritis, their high levels of autoantibodies included specificities to nascent CII — in the ehondroplastic areas (arrowheads) — and to nalive CII embedded in ihe hyalin matrix (brownstained in the original picture). No binding was observed with the negative control pool; the matrix was blue-stained, and the sole nuclei of chondroblasts (dark blue) were clearly apparent. Final magnification: x 300.
responses, we observed unexpected inter-strain variations in lerms of arthritis susceptibility, which allowed a new genetic approach of this experimental disease. The most striking variation was observed between the Hi and Hu lines (Table I)—issued from two independent sister selections, and bearing the same H-2'^ histocompatibility complex—which exhibited similar CII autoAb responses, whereas only one (Hi) was susceptible to arthritis [9]. Moreover when aged, the sole Hi developed a spontaneous RA-like disease associated with IgG2a CII Ab [11], suggesting that a common mechanism regulates the sensibility for both diseases in these strains. In agreement with a parallel study on SWR mice [12], we had already provided evidence that the arthritis resistance of Hit mice was not explained by a V/? gene deletion [ 13]. We show here that this resistance, and that of the HG mice, also cannot be due to a lack of Ab
response against the CII epitopes which are accessible within the cartilage matrix. CII autoAb are usually detected with a mouse antigen extracted from the cartilage matrix by a several-step method, including solubilization and pepsin digestion in acidic media [10]. Hence the purified molecule is no longer under a fully native state and the corresponding Ab can be directed against epitopes that are not apparent in vivo. Indeed, variations have been reported, in terms of recognition by autoAb, between regular and 'renatured'CI! fragments obtained after cleavage with cyanogen bromide [14]. Moreover, autoAb to in vivo hidden epitopes cannot react with tissue molecules. Nascent CII produced by chondrocytes is in a free and native form but is not accessible to Ab since it is located in ehondroplastic areas that are surrounded by the carlilage matrix. In contrast, CII molecules ofthe matrix
680
J-p.
Bouvet et al.
are covered by the proteoglyeans, and thus have few exposed epitopes, but can be reached at the cartilage border by IgG ofthe synovial fluid. Our results demonstrate that Ab from the strains (Hii and H G ) resistant to arthritis are not restricted to denatured epitopes since they heavily bound nascent CII. They also show that some of these Ab from resistant mice recognize the matrixembedded CI! molecules, indicating a specificity even for putatively arthritogenic epitopes. These results are in agreement with recent data showing that multiple CI! epitopes, recognized by different monoclonal antibodies, are accessible in vivo [15]. Passive transfers of CII arthritis require extremely high quantities of immunoglobulins, and. in our hands, could not be achieved with physiological doses of Ab from Biozzi mice (unpublished data). Moreover, the normal complement functions in these strains, and here the binding intensity, correlating with the autoAb titrc and not with the sensitivity to arthritis, are against the hypothesis of a key role of CII autoAb. It thus appears that direct injury of cartilage by complement-activating autoantibodies, directed againsi exposed epitopes ofthe matrix-embedded CII molecules, cannot be the primary mechanism of CII induced-arthritis which seems to mainly involve autoimmune CD4' Tcells[16, 17].
REFERENCES 1 Trcntham DE. Townes AS, Kang AH. David JR. Humoral and cellular sensitivity to collagen in type II collagen-induced arthritis in rats. J CItn Invest 1978:61:89-96. 2 Stuart JM. Cremcr MA, Townes AS, Kang AH. Type II collagen-induced arlhritis in rats. Passive transfer with serum and evidence that IgG antieollageii antibodies can cause arthritis. J F.xp Med l9S2;I55:t-16. 3 Holmdahl R, Klare.skog G, Rubin K, Larsson E, Wigzell H. T lymphoeytes in collagen Il-induced arthritis in mice: characterization of arthritogenic collagen ll-specific T-cell lines and clones. Scand J [mmunol I985;22:295 306. 4 Andersson M, Holmdahl R. Analysis of type II collagen-reaclive T cells in the mouse. I. Different regulation of autorcactive vs. non-autoreactive antitype II collagen T cells in the DBA-1 mouse. Eur J Immunol 1990:20:1061-66. 5 David CS. Genes for MHC. TCR and Mis determine suscepttbiltty to collagen induced arthritis. Aeta Palhol Microbiol Immunol Seand 1990:98:575 84.
6 Watson WC, Townes AS. Genetie susceptibility to murine collagen II autoimmune arthritis: proposed relationship lo the IgG2a autoantibody subclass response, complement C5, major histocompatibility eomplex jMHC) and non-MHC loci. J Exp Med 1985:162:1878-91. 7 Biozzi G. Mouton D, Sant'Anna OA, Passos HC, Gennari M. Reis MH, Ferreira WCA. Heumann AM, Bouthillier Y. Ibanez OM, StifTel C, Siqueira M. Genetics of immunoresponsivcness to natural antigens in the mouse. Curr Top Microbiol Immunol 1979;85:3l-98. S Mouton D. Siqueira M, Sant'Anna OA, Bouthillier Y, Ibanez O. Ferreira VCA, Mevel JC. Reis MH, Piatti RM, Stiffel C, Biozzi G. Genetic regulation of multispecific antibody responsiveness: improvement of'high' and 'low" characters. Eur J Immunol 1988:18:41^9. 9 Bouvet JP, Couderc J, Bouthillier Y, Franc B, Decreusefond C, Mouton D. Collagen arthritis in Biozzi mice: joint involvement is not correlated with collagen II lgG2a autoanlibodies nor restricted to only H-:" and H-2^ J Immunol 1989:143:1537-42. 10 Trentham DE. Townes AS, Kang AH. Autoimmunily to type 11 collagen: an experimental model of arthritis. J Exp Med 1977:146:857 68. 11 Bouvet JP, Couderc J, Bouthillier Y, Franc B, Ducailar A. Mouton D. Spontaneous rheumaloidlike arthritis in a line of mice sensitive to collageninduced arthritis. Arthr Rheum 1990:33:1716-22. 12 Andersson M. Goldschmidt TJ. Michaelsson E, Larsson A, Holmdahl R. T-cell receptor V/f haplotype and complement componeni C5 play no significant role for the resistance to collagen-induced arthritis in the SWR mouse. Immunology I99I;73:I91 6. 13 Vidard L, Roger T, Bouvet JP, Couderc J. Seman M. Resistance lo collagen-induced arthritis in Btozzi mice is not associated with T cell receptor V^ gene polymorphism. Eur J Immunol 1991:21:1783 85. 14 Terato K. Hasty KA, Cremcr MA. Stuart JM, Townes AS, Kang AH. Collagen-induced arthritis in mice: localization of an arlhritogenie determinant loa fragment of the type II collagen molecule. J Exp Med l985;l62:637-46. 15 Holmdahl RT, Mo JA. Jonsson R, Karlslrom K. Seheynius A. Mulliple epitopes on cartilage type IT eollagen are accessible for antibody binding in vivo. Autotmtnunity 1991:10:27 34. 16 Brawn E. Trentham DE. Experimental synovitis induced by collagen-specific T cell lines. Cell Immunol 1989:118:491 503. 17 Andersson M, Cremer MA, Terato H, Burkhardt H, Homldahl R. Analysisoftype II collagen reactive T cells in the mouse II. Different localization of immunodominant T ceil epilopes on heterologous and autologous type il collagen. Scand J Immunol 1991:33:505-10.
Received 20 December 1991 Accepted in revised form 24 January 1992
