OBJECTIVES To evaluate in vivo the contribution of tumour necrosis factor α (TNFα) to the chimeric transfer model of human rheumatoid arthritis synovial membrane into SCID mice (hu/mu SCID arthritis), systemic anti-TNFα treatment was performed and the clinical, serological, and histopathological effects of this treatment assessed.
METHODS Animals were treated with the rat-antimouse TNFα monoclonal antibody V1q, starting on day 1 after hu/mu engraftment, twice weekly for 12 weeks. Joint swelling, serum concentrations of human and murine interleukin 6 (IL6), and serum amyloid P (SAP) were measured. Histopathological and immunohistochemical analyses of the joints were also performed at the end of treatment.
RESULTS Neutralisation of murine TNFα induced the following effects: (a) reduction of extent and duration of the acute arthritis phase, with significant reduction of joint swelling at two weeks; (b) decrease of murine SAP concentrations after the first antibody administration; and (c) increase of murine IL6 in the serum. At the end of treatment, there was a significant reduction of the inflammatory infiltration in the engrafted joints. Because of the mild degree of joint erosion, no treatment effects could be demonstrated on the destructive process.
CONCLUSION In the lymphocyte independent hu/mu SCID arthritis, anti-TNFα treatment reduces local and systemic signs of inflammation.
- SCID mice
- rheumatoid arthritis
- synovial membrane
Statistics from Altmetric.com
The pro-inflammatory cytokine tumour necrosis factor α (TNFα) is thought to play a dominant part in the pathogenesis of rheumatoid arthritis (RA), appearing in a primary position in the cytokine cascade that sequentially activates interleukin 1 (IL1) and interleukin 6 (IL6).1 In the RA synovial membrane (SM), TNFα is predominantly produced by activated macrophages, as shown immunohistochemically.2 ,3 TNFα is also expressed at the cartilage-pannus junction,4 where it is believed to contribute to cartilage and bone destruction.5 ,6 The clinical relevance of TNFα in human and experimental arthritis is clearly indicated by the success of systemic application of anti-TNFα antibodies, which cause a rapid improvement of clinical and serological parameters of inflammation.7-9 Interestingly, recent studies have shown that anti-TNF treatment in experimental arthritis depends on T cells.10
TNFα can be clearly detected in the arthritic joints of a number of experimental models of arthritis.9 ,11-13 The pro-inflammatory role of this cytokine is stressed not only by the aggravation of collagen induced arthritis upon administration of exogenous TNFα,14 ,15 but also, and more importantly, by the chronic synovitis that develops after transgenic overexpression of human TNFα in recipient mice.16 ,17 In antigen induced arthritis, furthermore, TNFα plays a major part in the induction of arthritis, stimulating synovial fibroblasts to express metalloproteinases. Indeed, while synoviocytes isolated from mice with antigen induced arthritis transfer disease when injected into the knee joints of SCID mice, their arthritogenic potential is reduced in TNF knockout mice (R Bräuer, Jena, personal communication).
Our group has previously shown that unilateral grafting of human RA-SM into the knee joints of SCID mice induces a lymphocyte independent arthritis in which human and murine inflammatory cells, including macrophages, contribute to synovial inflammation.18Similarly to other experimental arthritides, this model is also expected to be characterised by the expression of TNFα in the inflamed SM. Based on this assumption, we therefore performed long term anti-TNFα treatment with systemic application (intraperitoneal (ip)) of anti-TNF monoclonal antibodies, to test whether TNFα neutralisation modulates the course of disease. Systemic and local parameters of disease were evaluated, including cytokine and acute phase protein levels in the serum, as well as inflammatory infiltration and tissue destruction in the joints.
HUMAN TISSUE SAMPLES
SM was obtained from RA patients (n = 2) undergoing joint surgery. Both patients fulfilled the 1987 revised ACR criteria.19SM specimens were characterised and classified according to previous studies,20 based on basic and actual disease activity (BA and AA, representing immunological, chronic processes and acute inflammatory acitivy, respectively), as well as on: (a) predominant B cell infiltration without joint destruction features (type I); (b) predominant T cell infiltration with joint destruction (type II); and (c) mixed B cell and T cell infiltration (type III), with intermediate characteristics of type I and II (for details see table1).
SCID mice (C.B.-17/lcrCrl-scid) were purchased as specific pathogen free animals from Charles River WIGA (Sulzfeld, Germany) and kept in a sterile filter cabinet throughout the experiment. Serum IgM was measured by ELISA, as described previously,21 to check for incompleteness of the SCID defect. Mice with > 10 μg/ml IgM22 were excluded from further experiments.
Implantation of human tissue into mice knee joints was as described previously,18 via introduction of a small piece of synovial tissue (1 mm3) through a lateral incision. The graft was fixed in direct contact with cartilage and the joint cavity closed by surgical suture. Table 1 describes the number of operated mice, the characterisation of the RA-SM samples, and the therapeutic modalities.
THERAPEUTIC ANTI-TNF MONOCLONAL ANTIBODY
The rat-antimouse TNFα MoAb V1q (IgD, κ)23 was used for treatment. In the L929 cytotoxicity assay, 1 ng of purified antibody neutralises 0.25 ng of recombinant murine TNFα, but not recombinant human TNFα and TNFβ, recombinant human IL1β, or recombinant murine IL1.24 After ip injection into mice, the antibody can be detected for > 5 days in the circulation.24
The MoAb V1q was administered ip at regular intervals twice weekly throughout the experiment (12 weeks), starting on day 1 after the engraftment of human tissue (see table 1 for details).
MONITORING OF DISEASE PARAMETERS
The joint diameter of the engrafted knees (consistently the left side) was measured with a calliper and compared with that of age matched SCID mice. Measurements were performed at weekly intervals by the same investigator.
Serum cytokine concentrations
At 1, 3, 10, 14, and 28 days after engraftment, blood was withdrawn from the orbital venous plexus. Serum was immediately obtained by centrifugation and stored at −70°C until analysis. Human IL6 was determined using a commercial ELISA kit (R&D Systems, Wiesbaden, Germany) with a sensitivity of 0.7 pg/ml. Murine IL6 was measured by a sandwich ELISA, using a rat-antimouse IL6 MoAb pair (Pharmingen, Hamburg, Germany). MaxiSorb 96 well microtitre plates (Nunc, Wiesbaden, Germany) were coated with antibody (clone: MP5–20F3, 4 μg/ml in carbonate buffer, pH 9.6, overnight, 4°C) and then blocked with phosphate buffered saline (PBS) (pH 7.4; 2% BSA, 0.01% Thimerosal) for 30 minutes. Recombinant murine IL6 standard (Endogen, Eching, Germany) or diluted mouse serum (1:10) were incubated for 12 hours, then a biotinylated antibody (clone: MP5–32C11, 2 μg/m) was added for six hours at 4°C. After incubation for 30 minutes with ExtrAvidin-Peroxydase 1:1000 (Sigma-Aldrich, Deisenhofen, Germany), tetramethylbenzidine (Sigma-Aldrich) was used as substrate. After 90 minutes, 1 N HCl was added and the extinction read at 450 nm. All reactants were diluted in assay buffer and incubated at room temperature, except where indicated. Washing steps with PBS and 0.2% Tween 20 were performed between all incubations. The sensitivity of the assay was determined as 4 pg/ml.
Murine serum amyloid P (SAP)
SAP was measured as described25 with some modifications. Microtitre plates were coated with dinitrophenylated keyhole limpet haemocyanin (Calbiochem, Bad Soden, Germany) 2.5 μg/ml in carbonate buffer, pH 9.6 overnight at 4°C. After a blocking step with assay buffer (50 mM TRIS/HCl pH 7.8; 0.15 M NaCl; 5 mM CaCl2; 2% BSA; 0.01% Thimerosal) for 30 minutes, standard SAP in serial dilution from 2000 ng/ml to 31.2 ng/ml, and serum 1:1000, were incubated for two hours at room temperature. Bound SAP was detected with rabbit-antimouse SAP (Calbiochem, 0.5 μg/ml, two hours, 4 °C), peroxydase labelled swine-antirabbit antibody (Dako, Hamburg, Germany) 1:1000, one hour, room temperature) and orthophenyldiamin as substrate. Extinction was read at 492 nm after addition of 1 N Hcl.
HISTOLOGY AND IMMUNOHISTOCHMISTRY
Murine knee joints were snap frozen in liquid nitrogen and embedded in 8% glycerol gelatine (Merck, Darmstadt, Germany). Sagittal cryostat sections of 7 μm were cut using a Jung Frigocut (Leica, Bensheim, Germany) and fixed onto transparent tape (Uhu, Bühl, Germany) to prevent disintegration of articular structures.26 Sections were air dried and stored at −70°C until staining.
Giemsa staining was performed to allow histopathological analysis of the joints. The degree of hyperplasia of synovial lining layer and sublining, as well as that of the inflammatory infiltration, were scored in a blinded fashion according to the criteria listed in table2.
The degree of joint destruction was determined by using a scoring system as described previously.27 Briefly, criteria used were cartilage erosion (slight=1), cartilage destruction with bone erosion (intermediate=2), and destruction of bone structures (strong=3).
Inflammatory cells were further characterised by immunohistochemistry. For this purpose, the sections were thawed (30 minutes) and fixed (15 minutes) in ice cold paraformaldehyde (4% in phosphate buffer, pH 7.4).
Human cells were stained as described previously, using an indirect peroxidase technique.28 Cryostat sections were incubated with monoclonal mouse-anti-HLA-ABC W 6/32 (Dako) followed by peroxidase labelled rat-antimouse Ig (Dako) and DAB (Sigma-Aldrich) as substrate. Counterstaining was performed with haematoxylin (Merck).
The APAAP technique with alkaline phosphatase labelled mouse F(ab)2-antirat Ig (Dianova, Hamburg, Germany) and rat APAAP (Dako) was used to detect murine macrophages and granulocytes; the ABC method with biotin labelled mouse F(ab)2-antirat Ig (Dianova) and StreptAB/ComplexHRP (Dako) was applied to detect murine IL6 and TNFα. Table 3 summarises the primary monoclonal antibodies. Control stainings were performed in parallel with isotype matched Ig (purified rat IgG1,κ; Pharmingen). Saponin (0.1%) was used to detect murine TNFα intracytoplasmatically, as described previously.29 To reduce background signals, blocking was performed using TRIS buffered saline (TBS) with 10% fetal calf serum, and washing with 0.1% Tween 20 in TBS. 3-amino-9-ethylcarbazole (AEC Substrat System, Dako) and Fast RedTR/Naphthol AS-MX phosphate (Sigma-Aldrich) were used as substrates.
Comparisons were performed using the Mann-Whitney rank sum test. A value of p<0.05 was considered significant. All statistical analyses were performed using SigmaStat for Windows (Jandel Scientific, Erkrath, Germany).
CYTOKINE EXPRESSION IN THE SYNOVIAL MEMBRANE OF HU/MU SCID MICE
The expression of murine TNFα, as well as that of IL6, could be demonstrated immunohistochemically in the inflamed knee joints of mice two weeks or one week (respectively) after local engraftment with human RA-SM tissue (fig 1), but not at later time points. Controls performed with isotype Ig were negative. In sham operated animals, neither TNFα nor IL6 were detectable. Because TNFα could not be demonstrated for longer periods, the effects of anti-TNF treatment on TNFα expression could not be investigated.
EFFECTS OF ANTI-TNF TREATMENT ON CLINICAL AND SEROLOGICAL PARAMETERS OF DISEASE
Treatment was performed accordingly to two different study designs (see table 1 for details). In experiment I, mice received 120 μg of anti-TNFα monoclonal antibodies twice weekly. This group showed an earlier decline of joint swelling than controls, with a statistically significant difference at two weeks (p < 0.05; fig 2A and B). Although these differences are very subtle, they were shown to be reliable, consistent with previous and current investigations, as well as independent of the operator. In experiment II, TNF neutralisation did not reduce joint swelling despite the high antibody dose used (300 μg; fig 2C and D). We have no explanation for this effect.
To estimate the effects of anti-TNFα treatment on the systemic components of the inflammatory response, the serum concentrations of human and murine IL6 were determined. While human IL6 could not be detected in any of the engrafted animals (data not shown), murine IL6 increased to a peak on day 3 (fig 3A). Interestingly, on this date mice treated with anti-TNF displayed even significantly higher serum IL6 concentrations than controls (p<0.05, fig 3A). These effects were stronger in the group treated with higher antibody dose.
The other systemic parameter considered was SAP, which, on day 3, was significantly increased in comparison with non-engrafted SCID mice (p<0.05; (mean (SD) 130 (48) μg/ml; n = 22). This increase declined within two weeks. Treatment with high dose anti-TNF antibodies (study design II) prevented the SAP rise on day 3 and caused a significantly faster decline of this acute phase protein compared with untreated hu/mu arthritis (p<0.05 at day 3, fig 3B).
EFFECTS OF ANTI-TNF TREATMENT ON HISTOPATHOLOGICAL PARAMETERS OF DISEASE
Untreated hu/mu mice showed prominent macrophage infiltration on day 3 after transplantation (data shown previously30). Anti-TNFα treatment induced a significant reduction of the inflammatory infiltration (p< 0.05; fig 4A) and synovial hyperplasia (p<0.05; fig 5).
With regard to joint destruction, untreated hu/mu SCID mice displayed only a mild degree of cartilage erosion (mean score 1; see methods for definitions). Thus, neither low nor high dose anti-TNF protocols could be proved to influence the degree of cartilage and bone destruction.
TNFα seems to play a central part in the pathogenesis of human RA, as well as in a number of animal models of arthritis. As this cytokine is also expressed in the SM of hu/mu SCID mice during the first two weeks, in parallel to the expression of IL6, systemic administration of an anti-TNFα monoclonal antibody was therefore performed to verify which disease parameters are sensitive to TNFα neutralisation. This treatment led to changes in the course of the inflammatory reaction—that is, a prompter reduction of local joint swelling and a faster decline of the systemic acute protein response. In the joints, systemic neutralisation of TNFα also reduced the degree of inflammatory infiltration and synovial hyperplasia. Thus, these data are consistent with the amelioration of arthritis observed in other animal models, for example murine collagen induced arthritis.9 ,11 ,31
Two features are unique to the hu/mu SCID arthritis: (1) the presence of implated (inflamed) human synovial tissue, which may contribute to the host disease by releasing human TNFα. Irrespective of the amounts of TNFα possibly produced by the human graft (which may reflect in turn the microscopic heterogeneity of the RA samples used for grafting), the release of human TNFα may not play a considerable part, as the anti-murine antibody did not cross react with human TNF; (2) the lymphocyte independent nature of arthritis (at least in terms of murine lymphocytes), which provides the possibility to examine pathogenetic and/or therapeutic aspects without the influence of specific immunity. In this study, therefore, excluding human inflammatory cells or murine T lymphocytes as possible TNF source, the disease relevant TNFα seems to originate predominanty in host activated macrophages.
In vitro experiments have shown that anti-TNF antibodies inhibit the production of IL6 and other pro-inflammatory cytokines (for example, IL1, IL8, and GM-CSF) by synovial cells.32 Somewhat surprisingly, in hu/mu SCID arthritis the neutralisation of TNF induced an increase of IL6 in the serum, suggesting that this cytokine could exert anti-inflammatory effects. Notably, however, an increase in IL6 was associated also to treatment with the higher, ineffective dose of anti-TNF treatment. Indeed, while IL6 acts as a pro-inflammatory cytokine in some systems,33 we have previously shown that the local application of recombinant human IL6 does not increase inflammation in hu/mu SCID arthritis.34 In addition, a significant inhibitory effect of IL6 has been demonstrated not only in rat adjuvant arthritis,35 but also in zymosan arthritis induced in IL6 deficient mice.36 A possible link between IL6 increase and disease amelioration may be the capacity of this cytokine to increase the release of IL1 receptor antagonist, the natural inhibitor of IL1, the latter a central mediator of joint destruction.37 Likewise, IL6 may increase the release of soluble TNF receptors,38 which, by binding to soluble TNF, neutralises its effects.
In the course of arthritis, acute phase proteins are produced in the liver as part of a systemic, acute inflammatory reaction. While C reactive protein is the main marker for disease activity in human RA,39 SAP is the main acute phase protein in mice, induced under the influence of IL1 and IL6.40 This study indicates that SAP is a useful marker of TNF mediated inflammation, in that TNF neutralisation significantly reduces the rise of this acute phase protein. Also, these early, systemic anti-inflammatory effects of anti-TNF treatment clearly precede the clinical amelioration of local synovitis, which occurs not before two weeks in terms of joint swelling (fig 2).
At the same time that SAP is significantly reduced by anti-TNFα, the IL6 concentrations in the serum increase, as discussed above. Because IL6 is the main regulator of the acute phase response, this paradoxical increase remains therefore unclear. This paradox seems true in human RA as well, because treatment of severe disease with anti-IL6 monoclonal antibody reduces the C reactive protein concentrations, however, the clinical improvement is accompanied by increased IL6 concentrations.41
Histological analyses showed that anti-TNFα treatment, in addition to inducing a decrease of joint swelling, also reduces inflammatory infiltration and synovial hyperplasia, similar to other arthritides. In contrast with collagen induced arthritis, however, this treatment has no effect on joint destruction because of the rather mild degree of cartilage damage in untreated animals.42-44 The limited feasibility of hu/mu SCID arthritis to study the anti-destructive effects of anti-TNF treatment is somewhat disappointing, as recent studies have shown that application of anti-TNFα antibodies in another model of xenogenic transfer (a graft of human synovium and cartilage into the muscle of SCID mice45) induces apoptosis in the transferred synovial cells, thereby reducing their destructive capacity on cartilage. As mentioned above, the anti-destructive process may require the mediation of IL1 rather than TNF.37 This hypothesis cannot be proved by our results.
In conclusion, in the lymphocyte independent hu/mu SCID arthritis TNFα is involved in inflammatory processes and its neutralisation reduces local and systemic parameters of inflammation.
We thank K Hofman, Institute of Clinical Immunology and Transfusion Medicine, University of Leipzig, for excellent technical work; P Windgassen, (Experimental Medicine Center, University of Leipzig), for animal housing; G Wichmann, Institute of Clinical Immunology and Transfusion Medicine, University of Leipzig, for determining the specificity of anti-TNFα monoclonal antibodies; P Stiehl, Institute of Pathology and R W Kinne, Institute of Clinical Immunology and Transfusion Medicine, University of Leipzig, for advice on histopathological examinations and helpful discussions; and E Palombo-Kinne for critical modification of the manuscript.
Funding: this study was supported by the Federal Ministry of Education, Science, Research, and Technology (grant 01ZZ9103); the Interdisciplinary Center for Clinical Research at the University of Leipzig (grant 01KS9504 A3); the German Research Council (DFG); and the Alexander von Humboldt Foundation (Bonn, Germay).
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.