Tibor T. Glant, MD, PhD, Department of Orthopedic Surgery Rush-Presbyterian-St Luke's Medical Center Cohn Research Building, Room 708, 1735 W. Harrison Street, Chicago, IL 60612, USA. E-mail: email@example.com
Mucosal tolerance is a natural mechanism that prevents immunological reactions to antigens by altering the activity of immune cells of pathogenic clones without modulating the entire immune system. This ‘natural immune suppression’ can be exploited when antigen(s) of the target organ in an autoimmune disease is used for mucosal treatment. Being inspired by the experimental results in animal models, clinical trials using type II collagen for mucosal treatment have been conducted in rheumatoid arthritis. High-density proteoglycan (aggrecan) is another major macromolecular component in articular cartilage, and may be a candidate autoantigen for provoking immune reactions in patients with rheumatoid arthritis. Indeed, like type II collagen, systemic immunization of genetically susceptible mice with proteoglycan (PG) aggrecan induces progressive autoimmune polyarthritis. Here, we investigated whether intranasally applied PG can be effective in suppressing PG-induced arthritis (PGIA) in BALB/c mice. We found that nasal administration of 100μg PG exerted a strong suppressive effect on both the incidence and severity of the disease, most probably by reducing responsiveness towards the immunizing PG antigen. When we transferred PGIA into genetically matched but immunodeficient SCID mice, we were able to establish a tolerized state, but only if the recipient SCID mice received lymphocytes from tolerized animals and intranasal treatment with PG was continued. Without nasally administered antigen, the transferred anergic cells recovered and arthritis rapidly developed in a severe form. Intranasal PG treatment of recipient SCID mice was ineffective when cells from non-tolerized arthritic donors were transferred, in which case the regular weekly ‘tolerizing’ dose of PG made the disease worse. Our results suggest that mucosal treatment in an already existing disease may result in paradoxical outcomes.
One of the most appealing therapeutic approaches in autoimmune diseases is an antigen-specific immune suppression which does not modulate the immune system as a whole, but wherein the activity of antigen-specific T cell clones is significantly altered. Mucosal tolerance is a physiological phenomenon that prevents harmful immunological reactions to ‘non-dangerous’ antigens such as protein components of food or airborne pollen and dust. In a number of animal models of autoimmune diseases such as diabetes , myasthenia gravis [2,3], autoimmune uveoretinitis , experimental allergic encephalomyelitis [5,6] or collagen-induced arthritis [7–9], mucosal tolerance has proved to be effective in suppressing the disease. Based on the success of these animal experiments, several clinical trials have been initiated [10,11], occasionally with promising results . While oral administration of an antigen is the most easily applied treatment and may down-regulate the disease, relatively large doses are necessary due to gastric and intestinal degradation of the antigen. Delivery of a protein via the respiratory mucosal route has advantages as the antigen is less likely degraded, therefore a significantly smaller quantity may achieve the same efficacy as the dose used for oral tolerance .
Systemic immunization with high-density cartilage proteoglycan (PG) aggrecan induces progressive polyarthritis in susceptible mouse strains [14–18]. PG-induced arthritis (PGIA) shows many similarities to human rheumatoid arthritis, such as repeated inflammatory episodes, formation of synovial pannus, which erodes articular cartilage and bone and female preponderance [14,15,17]. The mechanism of the disease is based upon autoimmune reactions that develop against the mouse (self) PG during systemic immunization with human PG.
To determine whether nasal tolerance can be induced in PGIA, BALB/c mice immunized with PG were treated nasally with human PG throughout the experiment, and incidence and severity of arthritis were monitored. Next, lymphocytes from arthritic donor BALB/c mice (nasally tolerized or non-tolerized) were transferred adoptively into MHC-matched (H-2d) severe combined immunodeficient (SCID)  mice. This transfer system allowed us to gain insight into the mechanisms of nasal suppression, and determine the requirements for establishing and maintaining tolerance in mice with PGIA.
MATERIALS AND METHODS
Antigens, animals and immunization
High-density PG (aggrecan) was purified from human cartilage by CsCl gradient centrifugation and depleted of glycosaminoglycan side chains, as described . Female BALB/c mice (National Cancer Institute, NCI, Friedrich, MD, USA) were immunized intraperitoneally with cartilage PG . The first antigen injection (100 μg protein) was given in complete Freund's adjuvant, and the same doses of antigen were injected as second and third boosts in incomplete Freund's adjuvant on days 21 and 42. In nasally treated groups of mice the antigen was given intranasally 10 days prior to the first injection, repeated daily for 4 consecutive days and then once a week during the immunization. Mice received various doses (10 μg and 100 μg) of PG, 100 μg ovalbumin (OVA; Sigma, St Louis, MO, USA) in 25 μl PBS (phosphate buffered saline, pH 7·4) or 25 μl PBS alone distributed equally among the two nostrils.
The intranasal treatment regimen was optimized in preliminary experiments. The most ‘optimal’ intranasal dose of PG (100 μg) and treatment protocol (intranasal tolerization prior to the immunization, necessary intervals between intranasal treatments) were determined using five to six mice in each group. Due to the large number of combinations and variability of results, animals of these preliminary results are not included in this paper. Typically, we used 12 mice in the first experiment and then the appropriate experiment was repeated once or twice using 15 animals in each group.
Female SCID mice with BALB/c background (NCI/NCrC.B-17-scid/scid), aged 8–12 weeks, were purchased from NCI and maintained under germ-free conditions at our Comparative Research Center. All animal experiments were approved by the Institutional Animal Care and Use Committee. In order to exclude animals with ‘leaky’ immune systems, SCID mice were tested for serum immunoglobulins using enzyme-linked immunosorbent assay (ELISA) prior to the experiments .
Cell isolation and transfer of arthritis into SCID mice
Single-cell suspensions were prepared in Dulbecco's modified Eagle's medium (Sigma) from spleens of arthritic BALB/c mice by mechanical dissociation. As described above for nasal treatment, the optimal cell numbers for transfer, the route of cell injection (intraperitoneal versus intravenous), the dose of cartilage PG administered along with cells, and intervals between injections were determined in preliminary experiments. In all transfer experiments 1 × 107 spleen cells were injected intraperitoneally along with 100 μg of PG into SCID mice. Another group of SCID recipients, in addition to the intraperitoneal injection, also received a weekly dose of 100 μg PG intranasally. Cell transfer was repeated on day 7, whereas the nasal administration of PG antigen was continued (once a week) throughout the entire experiment. Twelve SCID mice were used in each transfer group and experiments were repeated once with 15 mice.
Clinical assessment of arthritis
Immunized BALB/c mice were examined twice a week, and recipient SCID mice daily. The appearance of the first clinical symptoms (swelling and redness) was recorded as the time of onset of arthritis. Joint swelling was scored (from 0 to 4 of each paw) and expressed as the acute arthritis score, which is a summarized score for the four paws of one animal at a given time point [17,21,22]. Typically, in the primary form of PGIA, BALB/c mice developed swelling and redness in one or more limbs 7–14 days after the third injection of PG [14,17,22]. In the transfer system, arthritic SCID mice developed a more ‘uniform’ disease with the involvement of essentially all peripheral joints, beginning 1–2 days after the second cell transfer. Mice were sacrificed, and limbs were dissected, fixed in neutral formalin, decalcified and embedded in paraffin. Sections were stained with haematoxylin and eosin for histopathological analysis.
Measurements of PG-specific antibodies, T-cell responses and cytokine production
At the end of experiments, blood samples were collected from the retrobulbar venous plexus. Maxisorp immunoplates (Nalgene Nunc International, Denmark) were coated with human or mouse cartilage PGs (0·1 μg protein/100 μl/well) for ELISA as described [18,23,24]. Sera were applied at increasing dilutions from 1:12 500 to 1:62500, and the titre of isotypes of PG-specific antibodies was determined using peroxidase-conjugated rat antimouse IgG1, IgG2a or IgG2b (Zymed, San Francisco, CA, USA), or rat antimouse IgG3 (Accurate Chemical & Scientific Corp., Westbury, NY, USA) secondary antibodies, as described [24–26]. The optimal dilutions of isotype-specific second antibodies were determined in preliminary experiments. Serum antibody levels were normalized to mouse isotype standards. The control immunoglobulin isotypes were purified from irrelevant (non-PG specific) monoclonal antibody-containing ascites fluids, and immobilized on the microplate's surface at linear concentrations ranging from 0·2 to 200 ng/well.
Antigen-specific T-cell proliferation was measured in quadruplicate samples of spleen cells (3 × 105 cells/well) in the presence of 25 μg human PG protein/ml. Interleukin (IL)-2 secretion was determined by IL-2 bioassay using CTLL-2 cells pulsed with supernatants from 24 h-cultured spleen cells. Proliferation of CTLL-2 cells and antigen-specific T-cell proliferation were assessed on days 2 and 5, respectively, by measuring incorporation of 3H]-thymidine . The antigen-specific response was expressed as Δ counts per minute (cpm). Antigen (PG)-specific production of interferon-γ (IFN-γ), IL-10, IL-4 and transforming growth factor-β (TGF-β) were determined in media harvested from antigen (PG)-stimulated spleen cells (2·5 × 106 cells/ml) on day 4. To detect TGF-β production, spleen cells were cultured in serum free HL-1 medium (Biowhittaker, Walkersville, MD, USA). Cytokine concentrations were measured using capture ELISA from R&D Systems (Minneapolis, MN, USA). TGF-β was measured after acid treatment of samples by using TGF-β ELISA kit (Promega, Madison, WI, USA) as described .
The percentage of CD4+CD25+ T cells was determined by staining spleen cells with FITC-labelled anti-CD4 antibody, and a biotin-labelled anti-CD25 antibody followed by CyChrome-labelled streptavidin (BD PharMingen, San Diego, CA, USA), and analysing double-labelled fluorescent cells on a FacScan flow cytometer (Beckton Dickinson, San Jose, CA, USA). For intracellular CTLA-4 (cytotoxic T lymphocyte antigen-4) analysis, spleen cells were first stained with fluorescence-labelled antibodies, and then fixed, permeabilized using Cytofix/Cytoperm kit (BD PharMingen) and stained with PE-labelled anti-CTLA-4. For flow cytometry, we usually used cells from three to five animals of each group.
Statistical analysis of data was performed using SPSS v7·5 software package (SPSS, Chicago, IL, USA). The Mann–Whitney and Wilcoxon tests were used for intergroup comparisons. Significance was set at P < 0·05.
Nasal tolerance in BALB/c mice with PGIA
BALB/c mice were treated intranasally with cartilage PG (10 or 100 μg), OVA (100 μg) or vehicle (PBS) for 4 consecutive days beginning 10 days before PG immunization. To maintain tolerance, mice were given the same doses of antigen or PBS once a week throughout the entire experimental period. Nasally PBS-treated control BALB/c mice developed arthritis with 100% incidence and a cumulative arthritis score of 6·8 ± 1·5 after the third intraperitoneal injection of PG (Fig. 1). In contrast, BALB/c mice intranasally tolerized with 100 μg of PG developed arthritis much later with reduced incidence (30%) and severity (arthritis score: 2·5 ± 1·1) (Fig. 1). The intergroup variance was less than 10% between experimental groups. Higher amounts of intranasally applied antigen (e.g. 200 μg) had no additional suppressive effect on the disease, whereas low doses (1–10 μg) seemed to be insufficient to generate tolerance (data not shown).
The optimum time frame of the intranasal antigen application was tested using 100 μg dose of PG. When the first intranasal treatment with PG was applied at the time of the first or after the second intraperitoneal injection of PG, disease severity was less affected cumulative scores: 5·5 ± 1·1 (n = 12) and 6·3 ± 1·9 (n = 12), respectively] and the incidence was reduced by only 30% in both groups (data not shown). Mice (n = 12) that received intranasal PG first only after the third intraperitoneal PG injection developed anaphylactic reaction resulting in over 50% mortality. Mice that survived exhibited a poor physical condition and delayed onset of arthritis.
Cellular and humoral immune responses to PG in nasally tolerized mice
PG-immunized BALB/c mice tolerized intranasally with PG had significantly reduced T-cell responses measured in vitro either as antigen specific proliferation or IL-2 production (Fig. 2). While nasal treatment with a reduced amount of PG (10 μg) appeared to be slightly suppressive on in vitro T-cell responses (Fig. 2), the in vivo tolerizing effect was very modest (results not shown).
Serum antibody levels to either self (mouse) or immunizing (human) PG showed strong correlation with the clinical appearance of the disease (Fig. 3). Both IgG1 and IgG2a isotypes against both mouse and human PG were suppressed significantly in mice that received 100 μg PG intranasally (Fig. 3). Serum IgG2b levels were highly comparable to IgG2a, but the PG-specific IgG2b level was only half to two-thirds of the serum IgG2a levels. IgG3 isotype to either mouse or human PG was hardly detectable, usually 500–1000 times lower than the IgG1 in the same animal (not shown). Although the BALB/c strain has a genetic predisposition to Th2 type responses , PGIA causes a shift to Th1 type response, i.e. an increased production of IgG2a isotype prior to the onset of arthritis [24,29]. However, in the nasally tolerized group, both IgG1 and IgG2a (and IgG2b) isotypes of anti-PG antibodies were suppressed (Fig. 3).
CD4+CD25+ cells and T-cell cytokines have limited functions in PG-specific tolerance
To investigate whether suppressive CD4+CD25+ cells play a role in nasal tolerance to PGIA we compared this cell population in control and nasally tolerized mice. We found no difference in the percent of CD4+CD25+ cells between the two groups, and the intracellular CTLA-4 expression was also identical (Fig. 4). There was no difference in antigen-induced cytokine production between the tolerized and nontolerized groups, with respect to levels of IL-4, IL-10, IFN-γ and TGF-β (data not shown).
Transfer of arthritis and nasal tolerance into SCID mice
In agreement with our previous observations [30,31], only PG-stimulated spleen cells from arthritic BALB/c mice were able to transfer the disease . SCID mice which received spleen cells alone (without PG) or PG antigen alone (without cells) did not show any symptoms of inflammation, i.e. the transfer was negative . On the other hand, there was no difference in incidence or severity (not shown) whether arthritic cells were coinjected with PG intravenously or intraperitoneally.
When spleen cells from arthritic donor BALB/c mice were transferred into SCID mice along with PG intraperitoneally, recipient mice developed arthritis with 100% incidence (Fig. 5). The clinical appearance of the disease and the histopathological abnormalities in affected joints in all positive recipients were identical and indistinguishable from those described in donor BALB/c mice [14,17]. However, arthritic SCID mice developed a more uniform clinical picture of arthritis (essentially all peripheral joints become inflamed) with a robust involvement of the interphalangeal, metacarpo- and metatarso-phalangeal joints. To investigate whether the established state of nasally induced tolerance (Fig. 1) can be transferred adoptively into SCID mice, we transferred splenocytes from control (PBS- or OVA-treated) or the PG nasally tolerized BALB/c donors to SCID mice along with PG injected intraperitoneally. Unexpectedly, SCID mice receiving cells from nasally tolerized BALB/c mice developed arthritis with similar or even higher severity (Fig. 6a) and incidence (not shown) than those inoculated with cells from non-tolerized arthritic donors. In contrast, when SCID mice received spleen cells from nasally tolerized BALB/c mice and weekly doses of intranasal PG, there was a significant reduction in arthritis severity (Fig. 6b). Together, these results indicate that a successful transfer of PGIA and tolerance into SCID mice requires repeated intranasal administration of antigen.
Recently, a number of clinical trials have been initiated in autoimmune diseases using oral administration of tolerizing proteins such as type II collagen in rheumatoid arthritis [10,12,32], myelin basic protein in multiple sclerosis , and insulin in insulin-dependent diabetes . While the results of these clinical studies showed a promising trend toward clinical improvement, the mechanisms of tolerance are not fully understood and the results are frequently controversial [34,35]. Oral treatment of rheumatoid arthritis is probably in the most advanced stage of these clinical studies. The source of heterologous (bovine or chicken) cartilage type II collagen, the duration of the disease and previous or ongoing anti-inflammatory medications are important factors that influence the efficacy of the oral administration of type II collagen. Moreover, immune reactions to type II collagen can be detected only in a subset of patients with rheumatoid arthritis, indicating the heterogeneity of cartilage-derived autoantigens in this disease. Therefore, as proposed by Trentham , other noncollagenous cartilage constituents, such as human cartilage glycoprotein gp-39, link protein or PG (aggrecan) should be tested either alone or in combination, for their abilities to induce tolerance.
In addition to type II collagen [7,8,13], it has been reported recently that nasally applied cartilage gp-39 reduced the severity of experimentally induced arthritis . In this study, we suppressed inflammatory reactions in joints and immune responses to PG in BALB/c mice with PGIA by nasal administration of cartilage PG. The incidence was reduced significantly in intranasally treated BALB/c mice and affected animals developed significantly less severe arthritis (Fig. 1). However, effective mucosal tolerization in autoimmune models was achieved only if the treatment began before the systemic immunization of the animal [2,7,8], and the disease suppression was more evident if the nasal administration of the antigen was continued throughout the entire experiment . Tolerance was found effective only in a very rare case when mice were given the antigen intranasally after systemic immunization . In PGIA, the most effective suppression was achieved when the nasal treatment with PG was introduced days before systemic immunization began, and was continued throughout the entire experimental period. If the nasal treatment, for example, was applied first after the second injection of the immunizing PG, i.e. just before the onset of the disease, little or no suppressive effect was observed. Therefore, taking into account that animal models are never identical with human disease, it is unlikely that mucosal tolerance can be induced in the same way in human patients with chronic autoimmune diseases.
By definition, immune cells are tolerized by antigens that are not present in the thymus or bone marrow at the time of their differentiation . A number of studies have suggested that mucosal tolerance can be achieved by either ‘bystander’ suppression or induction of anergy [39,40]. An antigen-specific suppression by Th2 and/or Th3 regulatory cells is thought to be the major mechanism of oral tolerance when multiple low doses of antigen are used . In contrast, anergy can be induced by oral treatment of animals with high doses of antigen [40,42,43]. This may be the case when high doses of type II collagen are used for oral treatment of patients with rheumatoid arthritis. Anergy induction by mucosal tolerization is antigen-specific, whereby antigen-primed T cells are driven to the state of unresponsiveness. As antigen-specific T cells are present, anergy can clearly be distinguished from clonal deletion. We found that while intranasally administered PG had a strong suppressive effect at the dose of 100 μg, it was ineffective at a low dose (10 μg). Tolerized mice exhibited significantly reduced antigen-specific T-cell proliferation and IL-2 production in vitro (Fig. 2), which was accompanied by lower levels of PG-specific antibodies in the circulation (Fig. 3).
PGIA is a Th1-dominated autoimmune disease [24,29], and Th2-supported cytokines can significantly decrease disease severity . However, we found no changes in PG-induced Th1-(IFN-γ) or Th2 (IL-4, IL-10)-related cytokine productions in nasally PG-tolerized animals, nor was the PG-specific IgG1/IgG2a antibody ratio altered. CD4+CD25+ cells were recognized recently as naturally anergic and/or suppressive regulatory cells in several autoimmune diseases , possibly in the TGF-β-producing Th3 population . However, we found no changes in the number of CD4+CD25+ T cells or intracellular CTLA-4 expression of these cells in nasally PG-tolerized mice (Fig. 4), and the PG-specific TGF-β production was also comparable between the nasally treated and non-tolerized groups (data not shown). As PG-reactive T cells were present in nasally treated mice (Fig. 2), the clonal deletion of these antigen-specific T cells in donor animals could be excluded. This was even more evident when lymphocytes from PG-tolerized mice were transferred to SCID mice without continuation of intranasal treatment (Fig. 6a). In summary, the mechanism of nasal tolerance in PGIA appears to be a PG-specific anergy rather than deletion or active suppression of particular T-cell clones.
Cell transfer was used to further characterize once already established tolerance [42,46–51], but nasal treatment in recipients was either not continued or modified in the previous transfer studies [50,51]. We transferred PGIA into SCID mice, and continued the nasal treatment in recipient animals. It is important to note that PGIA could be transferred into SCID mice only if the splenic T cells were coinjected with PG . As recipient SCID mice had matched genetic backgrounds, both the mucosally and peritoneally administered antigens could be processed by the recipient's antigen presenting cells, thus providing optimal conditions for disease development. The ‘transfer of tolerance’ into SCID mice was successful, however, only when the nasal administration of the antigen was continued (Fig. 6). Although SCID mice lack functional T and B cells, the mucosal antigen processing and mucosal functions might remain intact, such as in those experiments which investigated oral tolerance toward antibody response [49,51]. Based on the effect of continued nasal administration of PG, transferred lymphocytes might have been able to traffic to the mucosa and remained anergic, while lymphocytes did not receive an anergizing signal (Fig. 6a), proliferated and expanded in the host [49,51].
The results of disease and tolerance transfer described here suggest that antigen-specific T cells were present in nasally tolerized donor mice, however, these cells were ‘paralysed’ functionally. As T-cell unresponsiveness is a reversible phenomenon both in vitro and in vivo thus antigen-specific anergized T cells, after the termination of tolerization following a few cell divisions, might become active again . Homeostatic polyclonal T-cell proliferation, i.e. the restoration of the original T-cell pool, occurs in severe lymphopenic conditions such as those following irradiation, chemotherapy, or upon transfer of a low number of lymphocytes into SCID mice. In such an autologous or syngeneic condition, which allows the recognition of a wide range of self-MHC-associated/(self)peptide ligands by matched TCRs, lymphocytes are signalled to replicate in a rapid manner and repopulate the lymphoid organs [53–57]. If the anergizing mucosal antigen exposure is not sustained/present in the recipient mouse, the low number of antigen-specific transferred T cells may undergo several rounds of division, anergic clones can revert [43,52] and expand , thus ultimately contribute to arthritis development in the host animal. We hypothesize that transferred arthritogenic but temporarily anergic clones can revert after a few cell divisions due to the discontinuation of nasal antigen exposure and the presence of antigen coinjected with cells. When non-tolerized cells are transferred, nasal administration of the antigen has no suppressive effect, rather it challenges the disease (Fig. 6b). Therefore, nasal antigen exposure has opposite effects on activated and anergized T cells. In conclusion, while induction of mucosal tolerance is a logical approach, it might become a ‘double-edged sword’ in treatment of autoimmune diseases, especially in humans where individual responsiveness varies to a great extent.
The authors thank the members of the Departments of Immunology and Biochemistry (Rush University, Chicago, IL) for helpful comments, discussion and criticism; Dr Joshua J. Jacobs, Leslie Manion-Patterson and members of the Department of Orthopedics (Rush University, Chicago, IL) for providing human cartilage samples; and David Gerard, Sonja Velins and members of the Rush Comparative Research Center for expert technical assistance.