1344
COLLAGEN-INDUCED ARTHRITIS IN RATS Evaluation of Early Immunologic Events JOHN M. STUART, MICHAEL A. CREMER, ANDREW H. KANG, and ALEXANDER S. TOWNES When rats were injected intradermally with an oil emulsion of native type I1 collagen, they developed an inflammatory polyarthritis. The incidence and severity of arthritis increased as the amount of collagen injected was increased. Rats 4% weeks old were the most susceptible to the development of arthritis, whereas weanling and older animals were relatively resistant. There was no difference in incidence between males and females. Mononuclear cells from peripheral blood, lymph nodes, and spleen were cultured with type I1 collagen and responded maximally to a collagen concentration of 25 pg/ml. The earliest detectable response was in peripheral blood mononuclear cell cultures obtained 6 to 8 days after immunization. The response of lymph node and spleen cells tended to lag behind that of peripheral blood cells at the earlier time intervals. Antibodies were detected in sera by hemagglutination at 8 days postimmunization. Quantitation of IgM and IgG antibodies by radioimmunoassay showed good correlation with hemagglutination titers and increased binding of collagen by both classes of antibody in arthritic as compared to nonarthritic animals. It is clear that the de-
From the Veterans Administration Medical Center and Departments of Medicine and Biochemistry, University of Tennessee Center for the Health Sciences, Memphis, Tennessee 38104. Supported in part by the Veterans Administration and by Grant AM 16505 from the National Institutes of Health. John M. Stuart, M D Associate Investigator of the Veterans Administration; Michael A. Cremer, M D Andrew H. Kang, M D Alexander S. Townes, MD. Address reprint requests to John W.Stuart, MD, VA Hospital, I030 Jefferson Avenue, Memphis, TN 38104. Submitted for publication October 1I, 1978; accepted in revised form August 16, 1979. Arthritis and Rheumatism, Vol. 22, No. 12 (December 1979)
velopment of both humoral and cellular immunity to type I1 collagen is associated with the development of arthritis and may be important in the pathogenesis of this disease. We have previously reported that approximately 40% of rats injected intradermally with an oil emulsion of heterologous or homologous type I1 collagen develop an inflammatory polyarthritis after a latent period of 12 to 20 days. Other types of collagen, cartilage proteoglycans, and denatured type I1 collagen are ineffective in inducing arthritis when injected in an identical manner (1,2). If animals are studied at a time when disease is well developed (21-28 days), arthritic animals have a significantly greater immune response to type I1 collagen than similarly injected nonarthritic rats (3). However, the early sequence of events in development of the immune response and its relation to clinical disease have not been adequately investigated. The present studies were undertaken to d e h e additional factors that might influence the development of collagen-induced arthritis and to evaluate the immunologic response occurring in peripheral blood, lymph nodes, and spleen at regular intervals after injection of type I1 collagen. These studies demonstrate the important influence of the immunizing dose of collagen, route of injection, and the age of the rats on the incidence and severity of arthritis. The early appearance of antibody to type I1 collagen and the quantitation of both IgG and IgM responses reported here suggest that the role of antibody to type I1 collagen in the pathogenesis of arthritis may be more significant than heretofore recognized.
COLLAGEN-INDUCED ARTHRITIS
1345
MATERIALS AND METHODS
to 15 ml by the addition of HBSS. Aggregated material was removed by centrifugation at l00g for 30 seconds and the remaining cells were washed twice with HBSS. After the last wash they were suspended in RPMI 1640 culture media supplemented to contain 20 mM HEPES and pencillin/streptomycin (Flow Laboratories, Rockville, Maryland) at 100 units and 100 pg/ml, respectively. Media were also supplemented to contain 5% fresh pooled rat serum which was obtained from Wistar strain rat, heat-inactivated at 56°C for 30 minutes, and used within 48 hours. Peripheral blood cells were obtained by cardiac puncture of ether anesthetized animals with preservative-free heparin 50 units/ml (final concentration) as an anticoagulant. The blood was diluted with 2 volumes of HBSS and separated on a Ficoll-Hypaque gradient. Cells at the interface were removed by aspiration, washed 3 times in HBSS, and suspended in RPMI 1640 culture media supplemented as described for lymph node and spleen cells. Cell cultures were performed in sterile round bottom microtiter plates (Nunc, Neptune, New Jersey) by adding 200 pl of cell suspension to each well. Lymph node and spleen cells were cultured at a concentration of 2 x l@ cells/ml and peripheral blood cells at a concentration of 106 ce&/ml. To each of quadruplicate wells were added 25 p1 of buffer (0.02M phosphate, 0.14M NaC1, pH 7.2) (PBS) containing PHA-M at 1 :4 dilution (Difco Laboratories, Detroit, Michigan) or type I1 collagen. The collagen solution was obtained by dissolving the collagen in 0.1M acetic acid at 4°C overnight, dialyzing the solution against PBS, and passing the solution successively through a Gelman Metrigard Superfine prefilter (Ann Arbor, Michigan) and then through a 0.45 pM microporous filter (Amicon Corp., Rockford, Illinois). The 6nal concentration of collagen was determined by hydroxyproline analysis on aliquots (7). Concentrations of collagen in the range of 1.2-1.5 mg/ml were obtained after filtration of solutions in which 2 mg/ml of collagen were originally dissolved. Sterile solutions were stored at -80°C until used. Cultures were maintained for 96 hours. Sixteen hours prior to harvesting, 0.5 pCi of 'H thymidine (specific activity 2 Ci/mmole, Research Products International, Chicago, Illinois) was added to each well. Cultures were harvested by aspiration onto Whatman 3MM paper scintillation pads (A. H. Thomas, Philadelphia, Pennsylvania) using a multiple automated sample harvester (Brandel, Rockville, Maryland) and washing the filters successively with water and methanol. The filters were dried and the incorporation of label was measured by liquid scintillation spectrometry and resuIts expressed as a stimulation index (ratio of counts per minute with collagen/ counts per minute with media alone). Antibody assay. Sera were collected, heat-inactivated at 56°C for 30 minutes, and stored at -80°C until used. Antibodies were quantified by passive hemagglutination as previously described (2) and by radioimmunoassay. For radioimmunoassay, type I1 collagen was labeled by acetylation with [1-I4C] acetic anhydride using the technique described by Gisslow and McBride (8) as modified by Menzel(9). Briefly 50 mg of type I1 collagen were solubilized in 12.5 ml of 0.01% acetic acid by stirring overnight at 4°C. The pH of the solution was adjusted to 8.5 by the addition of 1M K2HP0,. The acetylating agent, 0.5 mCi of [ l - Y ] acetic anhydride (Research Products International, Chicago, Illi-
Rats. Outbred Wistar rats (Charles River Breeding Laboratory, Wilmington, Massachusetts) were housed in metal cages and fed standard Purina Rat Chow. Collagen preparation. Type I1 collagen was prepared by limited pepsin digestion of pulverized chick sternal cartilage and purified as previously described (4). Xiphoid cartilage was obtained from 3-week-old White Leghorn chicks which were made lathyritic by inclusion of 0.03% p-aminopropnonitrile fumarate in their drinking water. The xiphoid cartilage was carefully dissected free from surrounding tissues, diced, and pulverized in liquid nitrogen by use of a freezer mill (no.6700, Spex Industries, Inc., Metuchen, New Jersey). All subsequent steps were performed at 4°C. The pulverized powder was subjected to limited pepsin digestion for 24 hours in 0SM acetic acid. After separation of the residue by centrifugation at 13,OOOg for 30 minutes, type I1 collagen was precipitated by dialysis against 0.02M Na,HPO,. The precipitated collagen was redissolved in 0.05M Tris/O.ZM NaCl, pH 7.4, and passed through a column of diethylaminoethyl-cellulose (DEAE) equilibrated with the same buffer. Under these conditions, collagen elutes unretarded from the column. Collagen was precipitated from the effluent by increasing the NaCl concentration to 2.5M. The resulting precipitated collagen was dissolved in 0.4 ionic strength phosphate buffer, pH 7.6, and an equal volume of cold 4.4M NaCl was added slowly, which caused the type I collagen to precipitate and allowed its removal by centrifugation at 5,000g for 30 minutes (5). Type I1 collagen was recovered by dialysis of the supernatant against 0.02M Na2HP04,centrifugation and solubilization of the precipitate in 0.5M acetic acid. Type I1 collagen was repeatedly precipitated from 0.5M acetic acid solution by addition of 5% NaCl, centrifugation and resolubilization in 0.5M acetic acid. Finally the 0.5M acetic acid solution was extensively dialyzed against 0.1M acetic acid and lyophilized. Immunization procedures. Collagen was dissolved overnight at 4°C in O.1M acetic acid at a concentration of 4 mg/ml. The solution was clarified by centrifugation, diluted when necessary with 0.1M acetic acid, and emulsified in an equal volume of incomplete Freund's adjuvant (ICFA) (Difco Laboratories, Detroit, Michigan) for 2 minutes at high speed using a Virtis 23 homogenizer (Gardner, New York). A total volume of 0.1 ml of the cold emulsion was injected into the right hind foot pad of each rat. In studies on the influence of the route of injection, the same volume and concentration of emulsion were given either intraperitoneally or subcutaneously on the back. Rats were observed daily for the development of arthritis and scored for severity as previously described by Wood, Pearson, and Tanaka (6). Mononuclear cell cultures. Cellular sensitivity to collagen was assessed by antigen-induced 'H thymidine incorporation by mononuclear cells. Lymph node, spleen, and peripheral blood cells were each utilized. Lymph node cells were prepared as previously described (2). Spleens were gently teased apart in Hanks' balanced salt solution (HBSS) and filtered through sterile gauze. Red blood cells were removed by hypotonic shocking with 3 ml of distilled water for 20 seconds followed by 1 ml of 0.6M KC1. The volume was then adjusted
STUART ET AL
1346
INCIDENCE
I / 10
2 110
li
4/10
5/10
Figure 1. The severity of arthritis produced by various dosages of type 11 collagen in oil emulsion. The total arthritic score is the sum of the individual rat scores at the time of maximal intensity. Incidence is the number of arthritic rats/number of rats injected.
nois), in 1.0 ml of benzene was added drop by drop over a 2hour pedod. Thirty mifiutes after the addition of the acetic anhydride the pH of the mixture was adjusted to 4.0 with glacial acetic acid and the reaction products were dialyzed exhaustively against 0.1M acetic acid. The acetylated collagen was lyophilized and stored at -2OOC. Sera were diluted 1 :50 with O.05M Tris, 0.15M NaC1, pH 7.4 (Tris-NaC1). To 0.5 ml of diluted serum were added 20 pg of acetylated type I1 collagen in 10 p1 of 0.01% acetic acid containing approximately 40,000 cpm. After incubation for 18 hours at 4°C 0.1 ml of rabbit antibody specific for either rat IgG or IgM (Miles Laboratories, Elkhart, Indiana) was added (second antibody). After further incubation for 18 hours at 4OC the precipitate was centrifuged, washed twice with TrisNaCl, solubilized in 0.5 ml of 0.01M NaOH, and dispersed in 5 ml of scintillation fluid (Aquasol, New England Nuclear, Boston, Massachusetts). Control sera were obtained from normal rats or from animals immunized by intradermal injection with ovalbumin in incomplete Freund's adjuvant. All samples were performed in duplicate. Antigen binding was expressed as: cpm in precipitate % binding = x loo total cpm added
The frequency and severity of arthritis increased with increasing dosage of collagen as shown in Figure 1. Effect of the age of the rats on the production of arthritis. Fifty weanling female Wistar rats were randomly allocated to 5 study groups. Each animal received 1 mg/kg of collagen injected intradermally as described. The animals ranged in age from 3 weeks to 12 weeks and from a size range of 50-60 gm to 220-260 gm. The peak incidence of arthritis occurred in rats of about 4-5 weeks old, weighing 90-100 gm (Figure 2). Both weanling and older animals were relatively resistant to the development of arthritis. Although the onset of arthritis usually occurred within 28 days of the date of injection, the group of animals with the highest incidence was maintained for 90 days, and the 2 animals that did not develop arthritis during the routine 4-week postinjection observation period eventually developed arthritis on day 34 and day 58. Influence of sex on the development of arthritis. Twenty male Wistar rats weighing 130-150 gm (6 weeks old) were injected with 200 pg of collagen in oil emulsion. Twelve rats or 60% developed arthritis with a mean onset at 13.1 days and mean arthritis index of 6.4. These figures are comparable to the 6-week-old female group of which 50% developed arthritis with the mean onset at 12.8 days and mean arthritic index of 5.8. Route of injection. Two groups of 10 rats each were injected intraperitoneally or subcutaneously on the back with 200 pg of type I1 collagen in oil emulsion. No animal from either group developed arthritis. We have previously reported that rats injected intradermally on
The optimal concentration of rabbit antibody (second antibody) for use in the radioimmunoassay was determined from the precipitating titer of the second antibody against normal rat serum in capillary tubes and using a twofold excess of the second antibody.
RESULTS Effect of the immunizing dose of collagen and the induction of arthritis. Female Wistar rats weighing 120140 gm were divided into 4 study groups. Each group of 10 rats received 2, 20, 50, or 200 pg of type I1 collagen emulsified in ICFA. A total of 0.1 ml of the emulsion was injected in the right hind footpad of each animal.
INCIDENCE 1/10 AGE GROUP 3 w k s
S/lU
3/IU
5/10
4.5wks.
6wks.
8 wks.
n 2/10 12 wks.
Figure 2. The severity of arthritis produced by injecting rats of different ages with 1 mg/kg of type I1 collagen in oil emulsion. The total arthritic score is the sum of individual rat scores at the time of maximal disease severity. Incidence is the number of arthritic rats/number of rats injected.
COLLAGEN-INDUCED ARTHRITIS
1347
Table 1. Incorporationof 'H thymidine by lymph node cells when cultured in the presence of mitogen or various concentrations of type I1 collagen* Final concentration of type I1 collagen
Arthritic? Nonarthritic*
138 f 21 97
* 16
*
*
12,955 249 (94)t
1208 87 (8.7)
16,329 f 363 (168)
240 12 (2.5)
*
1753 f 141 (12.7)
790 f 86 (5.7)
565 f 62 (4.1)
383 f 38 (3.9)
148f5 ( 1.5)
158*7 ( 1.6)
* Results are expressed as cpm f SEM for quadruplicatecultures.
*
t Values given in parentheses are stimulation indices.
Rats were injected 3 weeks before study with type I1 collagen in incomplete Freund's adjuvant.
the back develop arthritis with an incidence and severity comparable to those reported here although the amount of collagen used previously was greater (1). Of the 10 animals weighing 150-175 gm injected intradermally in the tail with 200 pg of collagen emulsion, 5 developed arthritis with mean onset at 14.2 days and mean arthritic index of 5.0. The intradermal injection route was effective whether the footpad, tail, or skin of the back was used, whereas intraperitoneal or subcutaneous injections failed to induce arthritis. In vitro dose response of mononuclear cells to type I1 collagen. When mononuclear cells from lymph nodes of immunized rats were isolated and cultured in the presence of type I1 collagen, a typical dose response curve was observed (Table 1). Optimal stimulation occurred in the presence of 25 p g of collagen/ml of culture media. Data from representative animals are shown. Several other arthritic and nonarthritic animals were also studied with comparable results. Cells from both arthritic and nonarthritic animals responded maximally to the same concentration of type I1 collagen as did cells from peripheral blood and spleen. Cells from several normal rats studied in an identical manner failed to respond to type I1 collagen. Time course of the immune response. Cellular sensitivity to collagen was measured in rats killed at 6, 8, 10, 12, 14, 21, and 28 days after injection. The incorporation of 'H thymidine by lymph node, peripheral blood, and spleen cells was measured and the results are summarized in Figure 3. Minimal stimulation was first detected in peripheral blood in a few rats at day 6 with maximal response occurring at days 10-12. Lymph node and spleen cell reactivity tended to lag behind that of peripheral blood initially. By day 12 a dichotomy developed between arthritic and nonarthritic animals. The
response of cells from arthritic animals to type I1 collagen remained elevated while that of nonarthritic animals declined. At days 21 and 28 arthritic animals had stimulation indices greater than previous study times due to continued high incorporation of 'H thymidine in collagen containing cultures combined with a decline in background incorporation by cultures containing buffer only. Background incorporation of 'H thymidine was in general higher during the early time intervals studied. This was particularly true of lymph node and spleen cell cultures. Time course of the development of humoral immunity to collagen was measured at the same time intervals as cellular immunity but using a separate group of rats which were bled serially and kept alive. Antibody levels to type I1 collagen were determined by radioimmunoassay and by hemagglutination techniques. Since the two methods of detecting antibodies were in generally good agreement with one another, only the radioimmunoassay data are shown. Radioimmunoassay was utilized to separate the IgM and IgG portions of the response (Figure 4). When values greater than three standard deviations above the mean of control sera were considered positive, detectable levels of IgM anticollagen antibodies were present at day 6 in a few animals. At day 8 most animals had detectable levels of IgM and positive hemagglutination titers. Binding of collagen by IgM was at low levels, however, until the onset of arthritis when much higher levels were detected. Binding of collagen by IgG was detected at day 10 and, like binding by IgM, tended to be higher in arthritic animals. All animals that developed arthritis had greater than 8% binding of labeled collagen by IgM and all had greater than 10% binding of labeled collagen by IgG.
STUART ET AL
1348
Pathology of early lesions. Histopathologicstudy of joints from rats killed within 48 hours of the onset of disease (Figures 5 and 6) showed marked infiltration of the synovium with neutrophils and monocytes. There were synovial hypertrophy and exudation of cells into the joint space. A fibrinous material was noted in some sections covering the surface of the articular cartilage. Peritendinitis and periostitis had begun to develop. These histopathologic changes were consistent with the clinical observation that an apparently normal joint would become severely inflamed over a period of less than 24 hours.
2ol
T
T
r .
Peripheral Blood
J
8 #
I
..
6
a
\
I 0 I2
h-4
21
28
DAYS POST INJECTION
Figure 3. Cell-mediated response to type I1 collagen in spleen, lymph nodes, and peripheral blood was assayed by culture of mononuclear cells for 96 hours and measuring 3Hthymidine incorporation. Stimulation indices are the ratio of collagen containing cultures/cultures in media alone. Each point represents the mean f standard error of the mean for 4 rats injected at day 0 and killed at the times shown. Beginning at day 14 arthritic animals (solid circles) are separated from nonarthritic (open circles). Because animals were killed at the time of study, time points prior to 14 days include animals that may have later developed arthritis.
MY$
POST NJECTloN
Figure 4. Humoral response to type I1 collagen was measured by radioimmunoassay. Type I1 collagen labeled with 14C was incubated with sera from rats injected at day 0 and bled at the times shown. Bound collagen was separated from free collagen by heavy chain specific rabbit anti-rat second antibody and percent binding calculated as counts precipitated/total counts added X 100. Rats were bled serially and animals arthritic at any time (solid circles) are separated from nonarthritic (open circles). Each point represents the mean f standard error of the mean for at least 6 rats.
COLLAGEN-INDUCED ARTHRITIS
Figure 5. Low magnification (original X 25) view of a tarsal joint from an arthritic rat. Specimen was obtained 48 hours after onset of disease and shows exudation of cells into the joint space and marked infiltration of subsynovial area.
DISCUSSION We have shown that an oil emulsion of type I1 collagen from heterologous or homologous articular cartilage or from the vitreous body is capable of inducing an inflammatory polyarthritis in rats (1,2). Other types of collagen including types I, I11 from several species, and type IV collagen isolated from bovine anterior lens capsule are ineffective. Data presented here show a dose related response to the injected collagen. As little as 2 pg of collagen will induce arthritis in a few animals. As the dose of collagen is increased, greater frequency and severity of arthritis result. However, because of the limited solubility of collagen and the limited amount of emulsion that can be injected into the footpad of a rat, the maximum dose of collagen used in dose response experiments was 200 pg per rat. The age of the rats at the time of injection also appears to be an important variable influencing the subsequent development of arthritis. The peak incidence occurs in animals 4 to 5 weeks of age at the time of injection. Both weanling and adult animals are relatively resistant to the development of arthritis. The reasons for this are not clear. It is possible that the combination of both an adequately mature immune system and a growing host is important. Collagen synthesis and turnover are more rapid in young animals. This might permit sensitized cells or antibodies greater access to the antigen or to antigenic sites on the collagen molecule. Studies of whether collagen is a sequestered antigen or
1349
whether there is specific tolerance to collagen in normal animals would help clarify this point. Sex does not appear to be important to the development of collagen-induced arthritis. Male and female rats developed arthritis with approximately equal incidence and severity. This is in contrast to human arthritides such as rheumatoid arthritis or ankylosing spondylitis in which there are measurable dserences in the relative frequency of disease among males and females. Time course studies demonstrated that the earliest detectable cell-mediated response was in the peripheral blood. There are two possible explanations for this finding. First, it is possible that either central processing of the antigen or interaction of reactive cells in central lymphoid organs is necessary to development of the arthritic response. This is suggested by the lack of predilection for involvement of the injected side. The uninjected hind limb is frequently unilaterally involved and simultaneous bilateral involvement is common as is unilateral involvement of the uninjected hind limb. Forepaw involvement is also frequent and may develop in the absence of hind limb arthritis. Second, it is also possible that lymph node and spleen cell cultures contain suppressor cells that modify the response. Bash and Waksman have reported that rats injected in the footpad with 200 pg of ovalbumin in CFA have depressed spleen cell responses at 9 days
Figure 6. Higher magnification (original x 450) of the same specimen shown in Figure 3. Synovial proliferation is present and extends over the cartilage which is Seen in the upper right. Synovial hypertrophy is again noted with infiltration of both mononuclear and neutrophilic cells (confirmed by magnification at 100 x using oil emersion on the actual slide).
1350
postinjection. Lymph node cells showed similar inhibition but only in the presence of specific antigen (10). Cells with suppressor activity have also been found in spleen cell preparations from rats with adjuvant induced arthritis (1 1,12). When glass or plastic adherent cells were removed from spleen cell cultures of collagen-induced arthritic rats, the response of the remaining cells to both type I1 collagen and PHA was enhanced (unpublished observation). Maximal cellular and humoral immune response occurred at the time of the onset of arthritis which was usually 12-14 days after injection (mean onset of arthritis was 12.8 days). From that time the response declined in nonarthritic animals but remained elevated in arthritic animals. This suggests that in arthritic animals there is a failure of the regulation of the immune response, or a continuing release of antigen from the inflammed tissue which perpetuates the activity of sensitized cells. The relative contribution of humoral and cellmediated immune response to type I1 collagen in the pathogenesis of arthritis is not known. Both are significantly increased in arthritic animals. Data presented here show that both are temporally associated with the development of disease. In particular, levels of IgM antibody to type I1 collagen approached a peak at the time of the onset of clinical arthritis. This high level of IgM antibody in arthritic animals continued throughout the time interval studied while the lower levels in nonarthritic rats tended to decrease by 28 days. The rapidity of progression of inflammation after onset and the prominence of polymorphonuclear cells in synovial exudates early in the disease also suggest that humoral mechanisms might be involved in the development of arthritis. Evidence that disease can be transferred by cells has been presented by Trentham and David (13). They used lo’ cells pooled from lymph nodes and spleen to transfer disease to 9 of 32 naive recipients. Arthritis in the recipients was in general mild and not associated with detectable cell-mediated or humoral response to collagen. Serum transfer was also attempted but antibody levels achieved in recipients were considerably below the mean level in arthritic rats after primary immunization. Of interest is the observation of Nakamura and Weigle that experimental autoimmune thyroiditis in rabbits can be transferred either adoptively with cells or passively by sera obtained from early bleedings after primary immunization (14,15). Although they suspected that IgM antibodies might be important, they were un-
STUART ET AL
able to definitely establish why early sera were necessary for transfer of disease. Serum transfer studies in collagen-induced arthritis achieving levels of both IgM and IgG antibodies to collagen in recipients comparable to the levels detected in arthritic rats would help c l a m this point. The relationship of this animal model of arthritis to human disease is unknown. Several investigators have found antibodies in the sera and synovial fluid of patients with rheumatoid arthritis (RA) and have hypothesized that collagen autoimmunity may be responsible, at least in part, for the observed pathology (1618). Antibodies reactive with both native and denatured types I, 11, and I11 collagen have been described in rheumatoid arthritis but the greatest reactivity was found against denatured collagen (19,20). This is different from collagen-induced arthritis in rats which can be induced only with native type I1 collagen and in which antibody reactivity is greatest to the native type I1 molecule (3). Cell-mediated immunity to collagen in RA has been less extensively investigated. However, Trentham et a1 using a leukocyte inhibition factor assay found cell-mediated immunity specific for native types I1 and I11 collagen in patients with RA and suggested that type specific cell-mediated immune responses account for the localization of tissue injury (21). Studies in our laboratory using an assay for lymphocyte-derived chemotactic factor for monocytes have shown that cell-mediated immunity to collagen in RA measured by this technique resembles humoral immunity in that reactivity is not specific for one or two types of collagen and is greatest against the denatured molecule (22). Although there are differencesbetween collageninduced rat arthritis and human RA, there are also similarities. Both diseases begin as a synovitis, produce marginal erosions, and often progress to extensive destruction of cartilage and bone. Both diseases are associated with collagen autoimmunity, although the nature and specificity of that autoimmunity may be different. Further studies involving the role of collagen autoimmunity in the pathogenesis of both RA and the animal model are needed to clarify the exact mechanisms involved in disease production. Avenues for research of the potential role of collagen autoimmunity in human disease can be suggested by investigation of this animal model of arthritis in which collagen immunity is clearly involved.
ACKNOWLEDGMENT The authors wish to thank Ms Susie Keshishian for her expert technical assistance.
COLLAGEN-INDUCED ARTHRITIS
REFERENCES 1. Trentham DE, Townes AS, Kang AH: Autoimmunity to type I1 collagen: an experimental model of arthritis. J Exp Med 146:857-868, 1977 2. Stuart JM, Cremer MA, Townes AS, Kang AH: Collageninduced arthritis in rats: comparison of vitreous and cartilage derived collagens. Arthritis Rheum 22:347-352, 1979 3. Trentham DE, Townes AS, Kang AH, David J R Humoral and cellular sensitivity to collagen in type I1 collagen-induced arthritis in rats. J Clin Invest 61239-96, 1978 4. Trelstad RL, Kang AH, Igarashi S, Gross J: Isolation of two distinct collagens from chick cartilage. Biochem 9:4993-4998, 1970 5 . Trelstad RL, Kang AH, Toole BP, Gross J: Collagen heterogeneity: high resolution separation of native [al(I)],a2 and [a1(II)]3and their component a chains. J Biol Chem 247:6469-6473, 1972 6. Wood FO, Pearson CM, Tanaka A: Capacity of mycobacterial wax D and its subfractions to induce adjuvant arthritis in rats. Int Arch Allergy Appl Immunol35:456467, 1969 7. Bergman I, Loxley R Two improved and simplified methods for the spectrophotometric determination of hydroxyproline. Anal Chem 35: 1961-1965, 1963 8. Gisslow MT, McBride BC: A rapid sensitive collagenase assay. Anal Biochem 68:70-78, 1975 9. Menzel J: Radioimmunoassay for anticollagen-antibodies using I4C-labeledcollagen. J Immunol Methods 15:77-95, 1977 10. Bash JA, Waksman BH: The suppressive effect of immunization on the proliferative responses of rat T cells in vitro. J Immunol 114:782-787, 1975 11. Kourounakis L, Kapusta M A Restoration of diminished T cell function in adjuvant induced disease by methotrexate. J Rheumatol 3:346-354, 1976 12. Kayashima K, Koga T, Onoue K: Role of T lymphocytes
1351
in adjuvant arthritis. 11. Different subpopulations of T lymphocytes functioning in the development of the disease. J Immunol 120:1127-1131, 1978 13. Trentham DE, Dynesius RA, David J R Passive transfer by cells of type I1 collagen-induced arthritis in rats. J Clin Invest 62359-366, 1976 14. Nakamura RM, Weigle W O Passive transfer of experimental thyroiditis from donor rabbits injected with soluble thyroglobulin without adjuvant. Int Arch Allergy Appl Immunol32:506-520, 1967 15. Nakamura RM, Weigle WO: Transfer of experimental autoimmune thyroiditis by serum from thyroidectomized donors. J Exp Med 130262-283, 1969 16. Steffen C, Timpl R: Antigenicity of collagen and its application in the serological investigation of rheumatoid arthritis sera. Int Arch Allergy 75:243-248, 1963 17. Menzel J, Steffen C, Kolan G, Eberl R, Frank 0, Thumb N: Demonstration of antibodies to collagen and of collagen-anticollagen immune complexes in rheumatoid arthritis synovial fluids. AM Rheum Dis 3k446-450, 1976 18. Steffen C: Consideration of the pathogenesis of rheumatoid arthritis as collagen autoimmunity. Z Immunitatsforsch 139:219-227, 1969 19. Michaeli D, Fudenberg HH: The incidence and antigenic specificity of antibodies against denatured human collagen in rheumatoid arthritis. Clin Immunol Immunopathol 2~153-159,1974 20. Andriopoulos NA, Mestecky J, Miller EJ, Bradley EL: Antibodies to native and denatured collagens in sera of patients with rheumatoid arthritis. Arthritis Rheum 19:613-617, 1976 21. Trentham DE, Dynesius RA, Rocklin RE, David JR: Cellular sensitivity to collagen in rheumatoid arthritis. N Engl J Med 299:327-332, 1978 22. Stuart JM, Postlethwaite AE, Kang AH, Townes AS: Cell-mediated immunity to collagen in rheumatoid arthritis (abstract). Arthritis Rheum 22:665, 1979
COLLAGEN-INDUCED ARTHRITIS IN RATS Evaluation of Early Immunologic Events JOHN M. STUART, MICHAEL A. CREMER, ANDREW H. KANG, and ALEXANDER S. TOWNES When rats were injected intradermally with an oil emulsion of native type I1 collagen, they developed an inflammatory polyarthritis. The incidence and severity of arthritis increased as the amount of collagen injected was increased. Rats 4% weeks old were the most susceptible to the development of arthritis, whereas weanling and older animals were relatively resistant. There was no difference in incidence between males and females. Mononuclear cells from peripheral blood, lymph nodes, and spleen were cultured with type I1 collagen and responded maximally to a collagen concentration of 25 pg/ml. The earliest detectable response was in peripheral blood mononuclear cell cultures obtained 6 to 8 days after immunization. The response of lymph node and spleen cells tended to lag behind that of peripheral blood cells at the earlier time intervals. Antibodies were detected in sera by hemagglutination at 8 days postimmunization. Quantitation of IgM and IgG antibodies by radioimmunoassay showed good correlation with hemagglutination titers and increased binding of collagen by both classes of antibody in arthritic as compared to nonarthritic animals. It is clear that the de-
From the Veterans Administration Medical Center and Departments of Medicine and Biochemistry, University of Tennessee Center for the Health Sciences, Memphis, Tennessee 38104. Supported in part by the Veterans Administration and by Grant AM 16505 from the National Institutes of Health. John M. Stuart, M D Associate Investigator of the Veterans Administration; Michael A. Cremer, M D Andrew H. Kang, M D Alexander S. Townes, MD. Address reprint requests to John W.Stuart, MD, VA Hospital, I030 Jefferson Avenue, Memphis, TN 38104. Submitted for publication October 1I, 1978; accepted in revised form August 16, 1979. Arthritis and Rheumatism, Vol. 22, No. 12 (December 1979)
velopment of both humoral and cellular immunity to type I1 collagen is associated with the development of arthritis and may be important in the pathogenesis of this disease. We have previously reported that approximately 40% of rats injected intradermally with an oil emulsion of heterologous or homologous type I1 collagen develop an inflammatory polyarthritis after a latent period of 12 to 20 days. Other types of collagen, cartilage proteoglycans, and denatured type I1 collagen are ineffective in inducing arthritis when injected in an identical manner (1,2). If animals are studied at a time when disease is well developed (21-28 days), arthritic animals have a significantly greater immune response to type I1 collagen than similarly injected nonarthritic rats (3). However, the early sequence of events in development of the immune response and its relation to clinical disease have not been adequately investigated. The present studies were undertaken to d e h e additional factors that might influence the development of collagen-induced arthritis and to evaluate the immunologic response occurring in peripheral blood, lymph nodes, and spleen at regular intervals after injection of type I1 collagen. These studies demonstrate the important influence of the immunizing dose of collagen, route of injection, and the age of the rats on the incidence and severity of arthritis. The early appearance of antibody to type I1 collagen and the quantitation of both IgG and IgM responses reported here suggest that the role of antibody to type I1 collagen in the pathogenesis of arthritis may be more significant than heretofore recognized.
COLLAGEN-INDUCED ARTHRITIS
1345
MATERIALS AND METHODS
to 15 ml by the addition of HBSS. Aggregated material was removed by centrifugation at l00g for 30 seconds and the remaining cells were washed twice with HBSS. After the last wash they were suspended in RPMI 1640 culture media supplemented to contain 20 mM HEPES and pencillin/streptomycin (Flow Laboratories, Rockville, Maryland) at 100 units and 100 pg/ml, respectively. Media were also supplemented to contain 5% fresh pooled rat serum which was obtained from Wistar strain rat, heat-inactivated at 56°C for 30 minutes, and used within 48 hours. Peripheral blood cells were obtained by cardiac puncture of ether anesthetized animals with preservative-free heparin 50 units/ml (final concentration) as an anticoagulant. The blood was diluted with 2 volumes of HBSS and separated on a Ficoll-Hypaque gradient. Cells at the interface were removed by aspiration, washed 3 times in HBSS, and suspended in RPMI 1640 culture media supplemented as described for lymph node and spleen cells. Cell cultures were performed in sterile round bottom microtiter plates (Nunc, Neptune, New Jersey) by adding 200 pl of cell suspension to each well. Lymph node and spleen cells were cultured at a concentration of 2 x l@ cells/ml and peripheral blood cells at a concentration of 106 ce&/ml. To each of quadruplicate wells were added 25 p1 of buffer (0.02M phosphate, 0.14M NaC1, pH 7.2) (PBS) containing PHA-M at 1 :4 dilution (Difco Laboratories, Detroit, Michigan) or type I1 collagen. The collagen solution was obtained by dissolving the collagen in 0.1M acetic acid at 4°C overnight, dialyzing the solution against PBS, and passing the solution successively through a Gelman Metrigard Superfine prefilter (Ann Arbor, Michigan) and then through a 0.45 pM microporous filter (Amicon Corp., Rockford, Illinois). The 6nal concentration of collagen was determined by hydroxyproline analysis on aliquots (7). Concentrations of collagen in the range of 1.2-1.5 mg/ml were obtained after filtration of solutions in which 2 mg/ml of collagen were originally dissolved. Sterile solutions were stored at -80°C until used. Cultures were maintained for 96 hours. Sixteen hours prior to harvesting, 0.5 pCi of 'H thymidine (specific activity 2 Ci/mmole, Research Products International, Chicago, Illinois) was added to each well. Cultures were harvested by aspiration onto Whatman 3MM paper scintillation pads (A. H. Thomas, Philadelphia, Pennsylvania) using a multiple automated sample harvester (Brandel, Rockville, Maryland) and washing the filters successively with water and methanol. The filters were dried and the incorporation of label was measured by liquid scintillation spectrometry and resuIts expressed as a stimulation index (ratio of counts per minute with collagen/ counts per minute with media alone). Antibody assay. Sera were collected, heat-inactivated at 56°C for 30 minutes, and stored at -80°C until used. Antibodies were quantified by passive hemagglutination as previously described (2) and by radioimmunoassay. For radioimmunoassay, type I1 collagen was labeled by acetylation with [1-I4C] acetic anhydride using the technique described by Gisslow and McBride (8) as modified by Menzel(9). Briefly 50 mg of type I1 collagen were solubilized in 12.5 ml of 0.01% acetic acid by stirring overnight at 4°C. The pH of the solution was adjusted to 8.5 by the addition of 1M K2HP0,. The acetylating agent, 0.5 mCi of [ l - Y ] acetic anhydride (Research Products International, Chicago, Illi-
Rats. Outbred Wistar rats (Charles River Breeding Laboratory, Wilmington, Massachusetts) were housed in metal cages and fed standard Purina Rat Chow. Collagen preparation. Type I1 collagen was prepared by limited pepsin digestion of pulverized chick sternal cartilage and purified as previously described (4). Xiphoid cartilage was obtained from 3-week-old White Leghorn chicks which were made lathyritic by inclusion of 0.03% p-aminopropnonitrile fumarate in their drinking water. The xiphoid cartilage was carefully dissected free from surrounding tissues, diced, and pulverized in liquid nitrogen by use of a freezer mill (no.6700, Spex Industries, Inc., Metuchen, New Jersey). All subsequent steps were performed at 4°C. The pulverized powder was subjected to limited pepsin digestion for 24 hours in 0SM acetic acid. After separation of the residue by centrifugation at 13,OOOg for 30 minutes, type I1 collagen was precipitated by dialysis against 0.02M Na,HPO,. The precipitated collagen was redissolved in 0.05M Tris/O.ZM NaCl, pH 7.4, and passed through a column of diethylaminoethyl-cellulose (DEAE) equilibrated with the same buffer. Under these conditions, collagen elutes unretarded from the column. Collagen was precipitated from the effluent by increasing the NaCl concentration to 2.5M. The resulting precipitated collagen was dissolved in 0.4 ionic strength phosphate buffer, pH 7.6, and an equal volume of cold 4.4M NaCl was added slowly, which caused the type I collagen to precipitate and allowed its removal by centrifugation at 5,000g for 30 minutes (5). Type I1 collagen was recovered by dialysis of the supernatant against 0.02M Na2HP04,centrifugation and solubilization of the precipitate in 0.5M acetic acid. Type I1 collagen was repeatedly precipitated from 0.5M acetic acid solution by addition of 5% NaCl, centrifugation and resolubilization in 0.5M acetic acid. Finally the 0.5M acetic acid solution was extensively dialyzed against 0.1M acetic acid and lyophilized. Immunization procedures. Collagen was dissolved overnight at 4°C in O.1M acetic acid at a concentration of 4 mg/ml. The solution was clarified by centrifugation, diluted when necessary with 0.1M acetic acid, and emulsified in an equal volume of incomplete Freund's adjuvant (ICFA) (Difco Laboratories, Detroit, Michigan) for 2 minutes at high speed using a Virtis 23 homogenizer (Gardner, New York). A total volume of 0.1 ml of the cold emulsion was injected into the right hind foot pad of each rat. In studies on the influence of the route of injection, the same volume and concentration of emulsion were given either intraperitoneally or subcutaneously on the back. Rats were observed daily for the development of arthritis and scored for severity as previously described by Wood, Pearson, and Tanaka (6). Mononuclear cell cultures. Cellular sensitivity to collagen was assessed by antigen-induced 'H thymidine incorporation by mononuclear cells. Lymph node, spleen, and peripheral blood cells were each utilized. Lymph node cells were prepared as previously described (2). Spleens were gently teased apart in Hanks' balanced salt solution (HBSS) and filtered through sterile gauze. Red blood cells were removed by hypotonic shocking with 3 ml of distilled water for 20 seconds followed by 1 ml of 0.6M KC1. The volume was then adjusted
STUART ET AL
1346
INCIDENCE
I / 10
2 110
li
4/10
5/10
Figure 1. The severity of arthritis produced by various dosages of type 11 collagen in oil emulsion. The total arthritic score is the sum of the individual rat scores at the time of maximal intensity. Incidence is the number of arthritic rats/number of rats injected.
nois), in 1.0 ml of benzene was added drop by drop over a 2hour pedod. Thirty mifiutes after the addition of the acetic anhydride the pH of the mixture was adjusted to 4.0 with glacial acetic acid and the reaction products were dialyzed exhaustively against 0.1M acetic acid. The acetylated collagen was lyophilized and stored at -2OOC. Sera were diluted 1 :50 with O.05M Tris, 0.15M NaC1, pH 7.4 (Tris-NaC1). To 0.5 ml of diluted serum were added 20 pg of acetylated type I1 collagen in 10 p1 of 0.01% acetic acid containing approximately 40,000 cpm. After incubation for 18 hours at 4°C 0.1 ml of rabbit antibody specific for either rat IgG or IgM (Miles Laboratories, Elkhart, Indiana) was added (second antibody). After further incubation for 18 hours at 4OC the precipitate was centrifuged, washed twice with TrisNaCl, solubilized in 0.5 ml of 0.01M NaOH, and dispersed in 5 ml of scintillation fluid (Aquasol, New England Nuclear, Boston, Massachusetts). Control sera were obtained from normal rats or from animals immunized by intradermal injection with ovalbumin in incomplete Freund's adjuvant. All samples were performed in duplicate. Antigen binding was expressed as: cpm in precipitate % binding = x loo total cpm added
The frequency and severity of arthritis increased with increasing dosage of collagen as shown in Figure 1. Effect of the age of the rats on the production of arthritis. Fifty weanling female Wistar rats were randomly allocated to 5 study groups. Each animal received 1 mg/kg of collagen injected intradermally as described. The animals ranged in age from 3 weeks to 12 weeks and from a size range of 50-60 gm to 220-260 gm. The peak incidence of arthritis occurred in rats of about 4-5 weeks old, weighing 90-100 gm (Figure 2). Both weanling and older animals were relatively resistant to the development of arthritis. Although the onset of arthritis usually occurred within 28 days of the date of injection, the group of animals with the highest incidence was maintained for 90 days, and the 2 animals that did not develop arthritis during the routine 4-week postinjection observation period eventually developed arthritis on day 34 and day 58. Influence of sex on the development of arthritis. Twenty male Wistar rats weighing 130-150 gm (6 weeks old) were injected with 200 pg of collagen in oil emulsion. Twelve rats or 60% developed arthritis with a mean onset at 13.1 days and mean arthritis index of 6.4. These figures are comparable to the 6-week-old female group of which 50% developed arthritis with the mean onset at 12.8 days and mean arthritic index of 5.8. Route of injection. Two groups of 10 rats each were injected intraperitoneally or subcutaneously on the back with 200 pg of type I1 collagen in oil emulsion. No animal from either group developed arthritis. We have previously reported that rats injected intradermally on
The optimal concentration of rabbit antibody (second antibody) for use in the radioimmunoassay was determined from the precipitating titer of the second antibody against normal rat serum in capillary tubes and using a twofold excess of the second antibody.
RESULTS Effect of the immunizing dose of collagen and the induction of arthritis. Female Wistar rats weighing 120140 gm were divided into 4 study groups. Each group of 10 rats received 2, 20, 50, or 200 pg of type I1 collagen emulsified in ICFA. A total of 0.1 ml of the emulsion was injected in the right hind footpad of each animal.
INCIDENCE 1/10 AGE GROUP 3 w k s
S/lU
3/IU
5/10
4.5wks.
6wks.
8 wks.
n 2/10 12 wks.
Figure 2. The severity of arthritis produced by injecting rats of different ages with 1 mg/kg of type I1 collagen in oil emulsion. The total arthritic score is the sum of individual rat scores at the time of maximal disease severity. Incidence is the number of arthritic rats/number of rats injected.
COLLAGEN-INDUCED ARTHRITIS
1347
Table 1. Incorporationof 'H thymidine by lymph node cells when cultured in the presence of mitogen or various concentrations of type I1 collagen* Final concentration of type I1 collagen
Arthritic? Nonarthritic*
138 f 21 97
* 16
*
*
12,955 249 (94)t
1208 87 (8.7)
16,329 f 363 (168)
240 12 (2.5)
*
1753 f 141 (12.7)
790 f 86 (5.7)
565 f 62 (4.1)
383 f 38 (3.9)
148f5 ( 1.5)
158*7 ( 1.6)
* Results are expressed as cpm f SEM for quadruplicatecultures.
*
t Values given in parentheses are stimulation indices.
Rats were injected 3 weeks before study with type I1 collagen in incomplete Freund's adjuvant.
the back develop arthritis with an incidence and severity comparable to those reported here although the amount of collagen used previously was greater (1). Of the 10 animals weighing 150-175 gm injected intradermally in the tail with 200 pg of collagen emulsion, 5 developed arthritis with mean onset at 14.2 days and mean arthritic index of 5.0. The intradermal injection route was effective whether the footpad, tail, or skin of the back was used, whereas intraperitoneal or subcutaneous injections failed to induce arthritis. In vitro dose response of mononuclear cells to type I1 collagen. When mononuclear cells from lymph nodes of immunized rats were isolated and cultured in the presence of type I1 collagen, a typical dose response curve was observed (Table 1). Optimal stimulation occurred in the presence of 25 p g of collagen/ml of culture media. Data from representative animals are shown. Several other arthritic and nonarthritic animals were also studied with comparable results. Cells from both arthritic and nonarthritic animals responded maximally to the same concentration of type I1 collagen as did cells from peripheral blood and spleen. Cells from several normal rats studied in an identical manner failed to respond to type I1 collagen. Time course of the immune response. Cellular sensitivity to collagen was measured in rats killed at 6, 8, 10, 12, 14, 21, and 28 days after injection. The incorporation of 'H thymidine by lymph node, peripheral blood, and spleen cells was measured and the results are summarized in Figure 3. Minimal stimulation was first detected in peripheral blood in a few rats at day 6 with maximal response occurring at days 10-12. Lymph node and spleen cell reactivity tended to lag behind that of peripheral blood initially. By day 12 a dichotomy developed between arthritic and nonarthritic animals. The
response of cells from arthritic animals to type I1 collagen remained elevated while that of nonarthritic animals declined. At days 21 and 28 arthritic animals had stimulation indices greater than previous study times due to continued high incorporation of 'H thymidine in collagen containing cultures combined with a decline in background incorporation by cultures containing buffer only. Background incorporation of 'H thymidine was in general higher during the early time intervals studied. This was particularly true of lymph node and spleen cell cultures. Time course of the development of humoral immunity to collagen was measured at the same time intervals as cellular immunity but using a separate group of rats which were bled serially and kept alive. Antibody levels to type I1 collagen were determined by radioimmunoassay and by hemagglutination techniques. Since the two methods of detecting antibodies were in generally good agreement with one another, only the radioimmunoassay data are shown. Radioimmunoassay was utilized to separate the IgM and IgG portions of the response (Figure 4). When values greater than three standard deviations above the mean of control sera were considered positive, detectable levels of IgM anticollagen antibodies were present at day 6 in a few animals. At day 8 most animals had detectable levels of IgM and positive hemagglutination titers. Binding of collagen by IgM was at low levels, however, until the onset of arthritis when much higher levels were detected. Binding of collagen by IgG was detected at day 10 and, like binding by IgM, tended to be higher in arthritic animals. All animals that developed arthritis had greater than 8% binding of labeled collagen by IgM and all had greater than 10% binding of labeled collagen by IgG.
STUART ET AL
1348
Pathology of early lesions. Histopathologicstudy of joints from rats killed within 48 hours of the onset of disease (Figures 5 and 6) showed marked infiltration of the synovium with neutrophils and monocytes. There were synovial hypertrophy and exudation of cells into the joint space. A fibrinous material was noted in some sections covering the surface of the articular cartilage. Peritendinitis and periostitis had begun to develop. These histopathologic changes were consistent with the clinical observation that an apparently normal joint would become severely inflamed over a period of less than 24 hours.
2ol
T
T
r .
Peripheral Blood
J
8 #
I
..
6
a
\
I 0 I2
h-4
21
28
DAYS POST INJECTION
Figure 3. Cell-mediated response to type I1 collagen in spleen, lymph nodes, and peripheral blood was assayed by culture of mononuclear cells for 96 hours and measuring 3Hthymidine incorporation. Stimulation indices are the ratio of collagen containing cultures/cultures in media alone. Each point represents the mean f standard error of the mean for 4 rats injected at day 0 and killed at the times shown. Beginning at day 14 arthritic animals (solid circles) are separated from nonarthritic (open circles). Because animals were killed at the time of study, time points prior to 14 days include animals that may have later developed arthritis.
MY$
POST NJECTloN
Figure 4. Humoral response to type I1 collagen was measured by radioimmunoassay. Type I1 collagen labeled with 14C was incubated with sera from rats injected at day 0 and bled at the times shown. Bound collagen was separated from free collagen by heavy chain specific rabbit anti-rat second antibody and percent binding calculated as counts precipitated/total counts added X 100. Rats were bled serially and animals arthritic at any time (solid circles) are separated from nonarthritic (open circles). Each point represents the mean f standard error of the mean for at least 6 rats.
COLLAGEN-INDUCED ARTHRITIS
Figure 5. Low magnification (original X 25) view of a tarsal joint from an arthritic rat. Specimen was obtained 48 hours after onset of disease and shows exudation of cells into the joint space and marked infiltration of subsynovial area.
DISCUSSION We have shown that an oil emulsion of type I1 collagen from heterologous or homologous articular cartilage or from the vitreous body is capable of inducing an inflammatory polyarthritis in rats (1,2). Other types of collagen including types I, I11 from several species, and type IV collagen isolated from bovine anterior lens capsule are ineffective. Data presented here show a dose related response to the injected collagen. As little as 2 pg of collagen will induce arthritis in a few animals. As the dose of collagen is increased, greater frequency and severity of arthritis result. However, because of the limited solubility of collagen and the limited amount of emulsion that can be injected into the footpad of a rat, the maximum dose of collagen used in dose response experiments was 200 pg per rat. The age of the rats at the time of injection also appears to be an important variable influencing the subsequent development of arthritis. The peak incidence occurs in animals 4 to 5 weeks of age at the time of injection. Both weanling and adult animals are relatively resistant to the development of arthritis. The reasons for this are not clear. It is possible that the combination of both an adequately mature immune system and a growing host is important. Collagen synthesis and turnover are more rapid in young animals. This might permit sensitized cells or antibodies greater access to the antigen or to antigenic sites on the collagen molecule. Studies of whether collagen is a sequestered antigen or
1349
whether there is specific tolerance to collagen in normal animals would help clarify this point. Sex does not appear to be important to the development of collagen-induced arthritis. Male and female rats developed arthritis with approximately equal incidence and severity. This is in contrast to human arthritides such as rheumatoid arthritis or ankylosing spondylitis in which there are measurable dserences in the relative frequency of disease among males and females. Time course studies demonstrated that the earliest detectable cell-mediated response was in the peripheral blood. There are two possible explanations for this finding. First, it is possible that either central processing of the antigen or interaction of reactive cells in central lymphoid organs is necessary to development of the arthritic response. This is suggested by the lack of predilection for involvement of the injected side. The uninjected hind limb is frequently unilaterally involved and simultaneous bilateral involvement is common as is unilateral involvement of the uninjected hind limb. Forepaw involvement is also frequent and may develop in the absence of hind limb arthritis. Second, it is also possible that lymph node and spleen cell cultures contain suppressor cells that modify the response. Bash and Waksman have reported that rats injected in the footpad with 200 pg of ovalbumin in CFA have depressed spleen cell responses at 9 days
Figure 6. Higher magnification (original x 450) of the same specimen shown in Figure 3. Synovial proliferation is present and extends over the cartilage which is Seen in the upper right. Synovial hypertrophy is again noted with infiltration of both mononuclear and neutrophilic cells (confirmed by magnification at 100 x using oil emersion on the actual slide).
1350
postinjection. Lymph node cells showed similar inhibition but only in the presence of specific antigen (10). Cells with suppressor activity have also been found in spleen cell preparations from rats with adjuvant induced arthritis (1 1,12). When glass or plastic adherent cells were removed from spleen cell cultures of collagen-induced arthritic rats, the response of the remaining cells to both type I1 collagen and PHA was enhanced (unpublished observation). Maximal cellular and humoral immune response occurred at the time of the onset of arthritis which was usually 12-14 days after injection (mean onset of arthritis was 12.8 days). From that time the response declined in nonarthritic animals but remained elevated in arthritic animals. This suggests that in arthritic animals there is a failure of the regulation of the immune response, or a continuing release of antigen from the inflammed tissue which perpetuates the activity of sensitized cells. The relative contribution of humoral and cellmediated immune response to type I1 collagen in the pathogenesis of arthritis is not known. Both are significantly increased in arthritic animals. Data presented here show that both are temporally associated with the development of disease. In particular, levels of IgM antibody to type I1 collagen approached a peak at the time of the onset of clinical arthritis. This high level of IgM antibody in arthritic animals continued throughout the time interval studied while the lower levels in nonarthritic rats tended to decrease by 28 days. The rapidity of progression of inflammation after onset and the prominence of polymorphonuclear cells in synovial exudates early in the disease also suggest that humoral mechanisms might be involved in the development of arthritis. Evidence that disease can be transferred by cells has been presented by Trentham and David (13). They used lo’ cells pooled from lymph nodes and spleen to transfer disease to 9 of 32 naive recipients. Arthritis in the recipients was in general mild and not associated with detectable cell-mediated or humoral response to collagen. Serum transfer was also attempted but antibody levels achieved in recipients were considerably below the mean level in arthritic rats after primary immunization. Of interest is the observation of Nakamura and Weigle that experimental autoimmune thyroiditis in rabbits can be transferred either adoptively with cells or passively by sera obtained from early bleedings after primary immunization (14,15). Although they suspected that IgM antibodies might be important, they were un-
STUART ET AL
able to definitely establish why early sera were necessary for transfer of disease. Serum transfer studies in collagen-induced arthritis achieving levels of both IgM and IgG antibodies to collagen in recipients comparable to the levels detected in arthritic rats would help c l a m this point. The relationship of this animal model of arthritis to human disease is unknown. Several investigators have found antibodies in the sera and synovial fluid of patients with rheumatoid arthritis (RA) and have hypothesized that collagen autoimmunity may be responsible, at least in part, for the observed pathology (1618). Antibodies reactive with both native and denatured types I, 11, and I11 collagen have been described in rheumatoid arthritis but the greatest reactivity was found against denatured collagen (19,20). This is different from collagen-induced arthritis in rats which can be induced only with native type I1 collagen and in which antibody reactivity is greatest to the native type I1 molecule (3). Cell-mediated immunity to collagen in RA has been less extensively investigated. However, Trentham et a1 using a leukocyte inhibition factor assay found cell-mediated immunity specific for native types I1 and I11 collagen in patients with RA and suggested that type specific cell-mediated immune responses account for the localization of tissue injury (21). Studies in our laboratory using an assay for lymphocyte-derived chemotactic factor for monocytes have shown that cell-mediated immunity to collagen in RA measured by this technique resembles humoral immunity in that reactivity is not specific for one or two types of collagen and is greatest against the denatured molecule (22). Although there are differencesbetween collageninduced rat arthritis and human RA, there are also similarities. Both diseases begin as a synovitis, produce marginal erosions, and often progress to extensive destruction of cartilage and bone. Both diseases are associated with collagen autoimmunity, although the nature and specificity of that autoimmunity may be different. Further studies involving the role of collagen autoimmunity in the pathogenesis of both RA and the animal model are needed to clarify the exact mechanisms involved in disease production. Avenues for research of the potential role of collagen autoimmunity in human disease can be suggested by investigation of this animal model of arthritis in which collagen immunity is clearly involved.
ACKNOWLEDGMENT The authors wish to thank Ms Susie Keshishian for her expert technical assistance.
COLLAGEN-INDUCED ARTHRITIS
REFERENCES 1. Trentham DE, Townes AS, Kang AH: Autoimmunity to type I1 collagen: an experimental model of arthritis. J Exp Med 146:857-868, 1977 2. Stuart JM, Cremer MA, Townes AS, Kang AH: Collageninduced arthritis in rats: comparison of vitreous and cartilage derived collagens. Arthritis Rheum 22:347-352, 1979 3. Trentham DE, Townes AS, Kang AH, David J R Humoral and cellular sensitivity to collagen in type I1 collagen-induced arthritis in rats. J Clin Invest 61239-96, 1978 4. Trelstad RL, Kang AH, Igarashi S, Gross J: Isolation of two distinct collagens from chick cartilage. Biochem 9:4993-4998, 1970 5 . Trelstad RL, Kang AH, Toole BP, Gross J: Collagen heterogeneity: high resolution separation of native [al(I)],a2 and [a1(II)]3and their component a chains. J Biol Chem 247:6469-6473, 1972 6. Wood FO, Pearson CM, Tanaka A: Capacity of mycobacterial wax D and its subfractions to induce adjuvant arthritis in rats. Int Arch Allergy Appl Immunol35:456467, 1969 7. Bergman I, Loxley R Two improved and simplified methods for the spectrophotometric determination of hydroxyproline. Anal Chem 35: 1961-1965, 1963 8. Gisslow MT, McBride BC: A rapid sensitive collagenase assay. Anal Biochem 68:70-78, 1975 9. Menzel J: Radioimmunoassay for anticollagen-antibodies using I4C-labeledcollagen. J Immunol Methods 15:77-95, 1977 10. Bash JA, Waksman BH: The suppressive effect of immunization on the proliferative responses of rat T cells in vitro. J Immunol 114:782-787, 1975 11. Kourounakis L, Kapusta M A Restoration of diminished T cell function in adjuvant induced disease by methotrexate. J Rheumatol 3:346-354, 1976 12. Kayashima K, Koga T, Onoue K: Role of T lymphocytes
1351
in adjuvant arthritis. 11. Different subpopulations of T lymphocytes functioning in the development of the disease. J Immunol 120:1127-1131, 1978 13. Trentham DE, Dynesius RA, David J R Passive transfer by cells of type I1 collagen-induced arthritis in rats. J Clin Invest 62359-366, 1976 14. Nakamura RM, Weigle W O Passive transfer of experimental thyroiditis from donor rabbits injected with soluble thyroglobulin without adjuvant. Int Arch Allergy Appl Immunol32:506-520, 1967 15. Nakamura RM, Weigle WO: Transfer of experimental autoimmune thyroiditis by serum from thyroidectomized donors. J Exp Med 130262-283, 1969 16. Steffen C, Timpl R: Antigenicity of collagen and its application in the serological investigation of rheumatoid arthritis sera. Int Arch Allergy 75:243-248, 1963 17. Menzel J, Steffen C, Kolan G, Eberl R, Frank 0, Thumb N: Demonstration of antibodies to collagen and of collagen-anticollagen immune complexes in rheumatoid arthritis synovial fluids. AM Rheum Dis 3k446-450, 1976 18. Steffen C: Consideration of the pathogenesis of rheumatoid arthritis as collagen autoimmunity. Z Immunitatsforsch 139:219-227, 1969 19. Michaeli D, Fudenberg HH: The incidence and antigenic specificity of antibodies against denatured human collagen in rheumatoid arthritis. Clin Immunol Immunopathol 2~153-159,1974 20. Andriopoulos NA, Mestecky J, Miller EJ, Bradley EL: Antibodies to native and denatured collagens in sera of patients with rheumatoid arthritis. Arthritis Rheum 19:613-617, 1976 21. Trentham DE, Dynesius RA, Rocklin RE, David JR: Cellular sensitivity to collagen in rheumatoid arthritis. N Engl J Med 299:327-332, 1978 22. Stuart JM, Postlethwaite AE, Kang AH, Townes AS: Cell-mediated immunity to collagen in rheumatoid arthritis (abstract). Arthritis Rheum 22:665, 1979
