Abstract
Induction of experimental autoimmune diseases often relies on immunization with the organ-specific autoantigens in CFA, which contains heat-killed mycobacteria. In several of these models, including collagen-induced arthritis, endogenous IFN-γ acts as a disease-limiting factor in the pathogenesis of the disease. Here we show that in collagen-induced arthritis the protective effect of IFN-γ depends on the presence of mycobacteria in the adjuvant. Omission of mycobacteria inverts the role of endogenous IFN-γ to a disease-promoting factor. Thus, the mycobacterial component of CFA opens a pathway by which endogenous IFN-γ exerts a protective effect that supersedes its otherwise disease-promoting effect. Extramedullary hemopoiesis and expansion of the Mac-1+ cell population accompanied the accelerated and more severe disease course in the IFN-γ receptor knockout mice immunized with CFA. Treatment of such mice with Abs against the myelopoietic cytokines IL-6 or IL-12 inhibited both disease development and the expansion of the Mac-1+ population. We postulate that mycobacteria in CFA stimulate the expansion of the Mac-1+ cell population by a hemopoietic process that is restrained by endogenous IFN-γ. These results have important implications for the validity of animal models of autoimmunity to study the pathogenesis and to evaluate cytokine-based therapy of autoimmune diseases.
The role of endogenous IFN-γ in the pathogenesis of experimental models of autoimmune diseases has been the subject of several studies (for review, see Ref. 1). A frequently used approach has consisted of examining changes in disease, incidence, and severity following ablation of IFN-γ by administration of neutralizing anti-IFN-γ Abs or by gene knockout technology. In some models, the outcome of such studies has been indicative of a disease-promoting role of IFN-γ. Examples are spontaneous lupus nephritis in (NZB × NZW)F1 mice (2, 3, 4) or in MRL-lpr mice (5), diabetes in nonobese diabetic mice (6, 7, 8), and orchitis following autoimmunization without adjuvant (9). Well-documented cellular and molecular pathways exist by which IFN-γ can exert this disease-promoting effect, including potentiation of autoantigen presentation by enhanced expression of the MHC, activation of macrophages to produce inflammatory mediators, and orientation of Th cell differentiation in the Th1 rather than the Th2 direction.
In several autoimmune disease models, the outcome of IFN-γ ablation procedures has been supportive of a protective rather than a disease-promoting role. Such examples are experimental autoimmune encephalomyelitis (10, 11, 12, 13, 14, 15, 16), experimental autoimmune uveitis (17, 18), and collagen-induced arthritis (CIA)3 (19, 20). So far, a satisfactory explanation, based on established cellular and biological effects of IFN-γ has not been available for this protective effect. The protective effect of IFN-γ in experimentally induced encephalomyelitis also contrasts with apparent disease-promoting role in transgenic mice overexpressing IFN-γ (21). It has been proposed that IFN-γ produced locally in affected tissues might indeed be disease-promoting, but that this effect might be overruled by systemically produced IFN-γ acting as a protective factor by its antiproliferative action on certain mononuclear cell populations or by its ability to induce suppressor cells.
We initiated the present study as we noted that all three models in which endogenous IFN-γ exerts a distinct protective role, rely on the use of the heat-killed mycobacteria-containing adjuvant, CFA, for immunization of the animals. IFN-γ is well known to be essential for effective host defense against mycobacteria, so that the use of these organisms as adjuvant can be surmised to add a level of complexity that might explain the protective effect. For our analysis of the interaction between heat-killed mycobacteria and IFN-γ, we chose CIA as a model rather than experimental autoimmune encephalomyelitis, because we found that CIA can be readily induced, albeit with less efficiency, using IFA, which does not contain mycobacteria. The outcome of our analysis is that endogenous IFN-γ acts as a disease booster of CIA induced by an IFA-assisted autoimmunization protocol, but protects against enhanced CIA induced by the classical CFA-assisted autoimmunization schedule. In addition, we obtained evidence for CIA pathogenesis to be controlled by a Mac-1+ population whose expansion is enhanced by mycobacteria but curbed by IFN-γ.
Materials and Methods
Mice and experimental conditions
The generation and the basic characteristics of the mutant mouse strain (129/Sv/Ev) with a disruption in the gene coding for the α-chain of the IFN-γ receptor (IFN-γR KO) have been described (22). These IFN-γR KO mice were back-crossed with DBA/1 wild-type mice for 10 generations to obtain the DBA/1 IFN-γR KO mice used in the present study. IFN-γR KO and wild-type mice were bred in our Experimental Animal Centre of the University of Leuven. The experiments were performed in 8- to 12-wk-old mice, but in each experiment, the mutant and wild-type mice were age-matched with a 5-day limit. The male to female ratio was kept between 0.8 and 1.3 in each experimental group.
Induction of CIA and clinical assessment of arthritis
Native chicken collagen type II (CII; Elastin Products Company, Owensville, MO) was dissolved in 0.05 M acetic acid at 2 mg/ml by stirring overnight at 6°C and emulsified in an equal volume of IFA or CFA containing 1.5 mg/ml heat-killed Mycobacterium butyricum (Difco, Detroit, MI). Mice were sensitized with a single 100-μl intradermal injection of the emulsion at the base of the tail. Mice were examined daily for signs of arthritis. The disease severity was recorded following a scoring system for each limb: 0, normal; 1, redness and/or swelling in one joint; 2, redness and/or swelling in more than one joint; 3, redness and/or swelling in the entire paw; 4, deformity and/or ankylosis.
Histological examination
Spleens and fore and hind limbs were fixed in B5. Limbs were subsequently decalcified overnight with formic acid. Four-micron-thick paraffin sections were stained with hematoxylin and eosin. Severity of arthritis was evaluated using three parameters: infiltration of mono- and polymorphonuclear cells, hyperplasia of the synovium, and pannus formation. Each parameter was scored on a scale from 0 to 3 (absent, weak, moderate, and severe).
In vivo Ab treatments
Monoclonal Abs were produced from hybridomas grown by i.p. inoculation in Pristane-primed athymic nude mice (nu/nu of nuclear magnetic resonance imaging background). Neutralizing mAb against murine IFN-γ (F3, rat IgG2a) (10) was purified by affinity chromatography on a mouse anti-rat κ-chain mAb. The neutralizing titer (end-point dilution corresponding to 50% neutralization of the antiviral effect of 30 U/ml of mouse IFN-γ on mouse L929 cells challenged with mengovirus) was 105.3 U/ml (IgG content, 1.4 mg/ml). A neutralizing rat IgG2a Ab against murine IL-12 was produced using hybridoma C17.8 (kindly provided by Dr. G. Trinchieri, Wistar Institute, Philadelphia, PA). The Ab was purified by affinity chromatography on protein G (Pharmacia, Uppsala, Sweden). Ab against murine IL-6 was prepared from ascites fluid from thymusless nude mice inoculated with the 20F3 (rat × mouse) hybridoma (American Type Culture Collection, Manassas, VA). This rat IgG1 Ab was purified by affinity chromatography on an anti-rat κ-chain Ab-Sepharose column. The neutralizing titer (endpoint dilution corresponding to 50% neutralization of the cell growth effect of 10 U of murine IL-6 per ml) was 105.5 (IgG content, 2.9 mg/ml). Irrelevant rat IgG2a was used as an isotype control and was prepared from ascites fluid of a rat plasmocytoma obtained through the courtesy of Dr. H. Bazin (University of Louvain Medical School, Brussels, Belgium). The IgG was purified by anion exchange chromatography on Hiload Q Sepharose and gel filtration on Superdex 200 (Pharmacia). Batches of anti-IFN-γ, anti-IL-12, anti-IL-6, and irrelevant IgG2a were tested for endotoxin content by a chromogenic Limulus amoebocyte lysate assay (KabiVitrum, Stockholm, Sweden) and were found to contain <2 ng/ml endotoxin. Injections were given in 200 μl of pyrogen-free saline.
Serum anti-CII Ab levels
Blood samples were taken from the orbital sinus and were allowed to clot for 1 h at room temperature and at 4°C overnight. Individual sera were tested for Abs directed to chicken CII by ELISA as described (19). A serial 2-fold dilution series of a purified standard preparation was included in each experiment to allow for calculation of the Ab content. The standard material was purified by affinity chromatography from pooled sera obtained from several IFN-γR KO and wild-type arthritic mice and contained 0.5 mg/ml IgG.
Delayed-type hypersensitivity (DTH) reactivity
At the indicated days, symptom-free mice were challenged in the right footpad with 10 μg chicken CII in 20 μl PBS. The left footpad received 20 μl PBS. DTH response was measured as percent swelling (i.e., the difference between the thickness of the right and left footpads, as percent of the thickness of the left footpad) at 24 and 48 h postchallenge.
Characterization of spleen cells
Spleens were gently cut into small pieces and passed through cell strainers (Becton Dickinson Labware, Franklin Lakes, NJ). RBC were lysed by two consecutive incubations (5 and 3 min, 37°C) of the suspension in NH4Cl (0.83% in 0.01 M Tris/HCL, pH 7.2). Remaining cells were washed, resuspended in cold PBS, and counted. Aliquots of 2 × 105 cells in 0.2 ml were preincubated (30 min) with an Fc receptor-blocking Ab (2.4G2; PharMingen, San Diego, CA; 1 μg/ml) and then stained for 30 min with PE-conjugated anti-Mac-1 Ab (CD11b+; PharMingen). PE-conjugated rat IgG (clone A 95-1; PharMingen) was used as an isotype-identical control. Cells were analyzed by a FACScan flow cytometer (Becton Dickinson, San Jose, CA). Differential leukocyte counts were obtained by centrifuging (700 rpm for 8 min) aliquots of 2 × 105 cells onto slides (Cytospin apparatus, Shandon, Cheshire, U.K.) and staining with Hemacolor (Merck, Darmstadt, Germany). From each spleen cell suspension, at least 200 cells were counted.
Results
The IFN-γR KO mutation results in higher susceptibility for CIA induced by CFA-assisted immunization, but protection against CIA induced by IFA-assisted immunization
Groups of IFN-γR KO or wild-type mice were immunized with CΙΙ in either IFA or CFA. The animals were scored daily for clinical signs of arthritis. The time of disease onset, cumulative incidences, and maximal clinical scores recorded in a total of 7 experiments are summarized in Table I⇓. The data confirmed the well-known lower incidence and the delayed onset of arthritis in wild-type mice immunized with Ag in IFA instead of CFA: only 25% of the mice developed overt arthritis when immunized using IFA, against 63% using CFA. In line with our previously reported findings (19), IFN-γR KO mice proved to be more susceptible than wild types for induction of CIA elicited with CΙΙ in CFA, as was evident from their significantly earlier times of disease onset and increased cumulative incidence and disease scores.
Cumulative incidence, day of disease onset, and clinical score of arthritis in IFN-γR KO and wild-type mice after immunization with CII in IFA or CFAa
However, in sharp contrast, when CIA was elicited with CΙΙ in IFA, none of the IFN-γR KO mice showed clinical signs of arthritis, whereas such signs did develop in 25% of the wild-type mice given the same immunization schedule. Histological examination of limbs collected on day 64 confirmed development of classical arthritic lesions in wild-type but not in IFN-γR KO mice (Fig. 1⇓): infiltration of the synovium by mono- and polymorphonuclear cells, synovial hyperplasia, and pannus formation. Results of quantitative scoring of histological sections are shown in Fig. 2⇓. All histological parameters of arthritis were lower in mutant than in wild-type mice. Furthermore, synovial hyperplasia and pannus formation were evident in limbs of clinically symptomless wild-type mice. The proportions of inflamed limbs with histological scores ≥ 2 for infiltration, hyperplasia, and pannus were 0, 0, and 8% in IFN-γR KO mice (n = 12) vs 28, 56, and 61% in wild-type mice (n = 18) and 0, 38, and 46% in symptom-free wild-type mice (n = 13).
Hematoxylin/eosin-stained sections through the hind limb joints of IFN-γR KO and wild-type mice sacrificed on day 64 postimmunization with CII in IFA. Joints of IFN-γR KO mice (A) had a normal histological appearance. Joints of wild-type mice (B–D) showed infiltration of mono- and polymorphonuclear cells in the synovium (close-up in C), hyperplasia of the synovial membrane (arrows in B), and starting pannus formation (close-up in D). Note the presence of osteoclast-like multinucleated giant cells (arrows in D). p, pannus formation; s, synovium. Magnification: ×38 (A and B); ×300 (C and D).
Histological arthritis scores of limbs from IFN-γR KO and wild-type mice immunized with CII in IFA. Of two independent experiments (Table I⇑), a total of six IFN-γR KO and nine wild-type mice were sacrificed on day 64 postimmunization, and two limbs of each mouse were examined. Macroscopically, all 12 limbs of the IFN-γR KO mice were symptomless (see Table I⇑), whereas 5 of the 18 limbs of wild-type mice showed overt signs of arthritis. Sections were scored for three parameters of arthritis: infiltration of mono- and polymorphonuclear cells, hyperplasia of the synovium, and pannus formation. The severity of each parameter was graded as 0, 1, 2, or 3 (absent, weak, moderate, severe). ∗, p < 0.01 and p ≪ 0.001 for comparison with all and only arthritic wild-type limbs (all three parameters), respectively; p < 0.01 for comparison with symptomless wild-type limbs (hyperplasia and pannus). Similar histological differences between IFN-γR KO and wild-type mice were recorded on day 80.
Absence of CIA symptoms in IFN-γR KO mice given the IFA-assisted immunization schedule is associated with reduced Ab and DTH responses to CΙΙ
Fig. 3⇓A describes the results of two experiments in which anti-CΙΙ Ab was determined in blood samples taken at different time points during the course of arthritis development in IFN-γR KO and wild-type mice given the IFA-assisted immunization schedule. Significantly lower levels of anti-CΙΙ IgG were observed in IFN-γR KO than in wild-type mice. All isotypes (IgG1, IgG2a, and IgG2b) were similarly affected (data not shown).
Ab levels and DTH reactivity to CII in IFN-γR KO and wild-type mice immunized with CII in IFA. A, Serum levels of anti-CII-IgG at the indicated time points postimmunization. Bars represent averages ±SEM for five to eight mice per group. Abs were undetectable on day 0. ∗, p < 0.05 for comparison with wild-type mice (Student’s t test). B, DTH reactivity. In two experiments, groups of six and four mice were challenged on days 63 and 102, respectively, by injection in the footpad of 10 μg of CII (right footpad) or vehicle (left footpad). DTH responses were measured as percent swelling (i.e., the difference between the increase in thickness of the right and left footpads, divided by the thickness of the left footpad ×100) at 24 and 48 h postchallenge. Bars represent averages ± SEM. ∗, p < 0.05 for comparison with wild-type mice (Student’s t test).
The cellular response to CΙΙ was assessed by measuring DTH reactivity against CΙΙ injected in the footpad on days 63 and 102 of the disease course (Fig. 3⇑B). Significant swelling was observed in wild-type but not in IFN-γR KO mice, indicating that DTH reactivity was reduced in the mutant mice. We also examined the possibility that the Th1/Th2 cytokine balance evolved differently in IFN-γR KO than in wild-type mice, as previously found to be the case when CFA is used (19). However, following exactly the same assessment system (measurement of IL-2 or IL-4 serum levels after an in vivo anti-CD3 challenge), we found the Th1/Th2 balance of IFN-γR KO mice immunized with the IFA-assisted schedule not to be different from that of similarly immunized wild-type mice (data not shown).
Endogenous IFN-γ prevents the expansion of a splenic Mac-1+ cell population, elicited by the mycobacterial component of CFA
Our results indicate that the mycobacterial component of CFA triggers events that not only enhance and accelerate pathogenesis of CIA, but in addition open a pathway by which endogenous IFN-γ can exert a protective effect that supersedes its natural disease-promoting effect. A possible clue to this mycobacterium-dependent pathway was the observation of a pronounced splenomegaly occurring in the IFN-γR KO mice immunized with CΙΙ in CFA. For instance, in one experiment in which the mice were sacrificed on day 23 postimmunization, the mean spleen weights (±SEM, n = 4) of IFN-γR KO mice and wild-type mice was, respectively, 226 ± 18 and 125 ± 12 mg (CFA-assisted immunization); 106 ± 7 and 93 ± 6 mg (IFA-assisted immunization); and 77 ± 14 and 70 ± 5 (naive mice). To clarify the possible significance of this splenomegaly for the disease course, spleen sections were examined histologically and spleen cell suspensions were analyzed morphologically and by FACScan using sets of Abs for different leukocytes: T lymphocytes (CD3+, CD4+, and CD8+), NK cells (IL-2Rβ+, CD3−, and CD8−), NKT cells (IL-2Rβ+, CD3+, and CD8−), B lymphocytes (CD19+ and B220+), and monocytes/macrophages/granulocytes (CD11β+ or Mac-1+). Histological examination performed on day 24 postimmunization with CII in CFA revealed a normal splenic architecture of the white and red pulp in both IFN-γR KO and wild-type mice. However, in IFN-γR KO mice the red pulp was extensively expanded due to extramedullary hemopoiesis with increased numbers of megakaryocytes and erythropoietic and myelopoietic cells (Fig. 4⇓, A–D). Cytospin preparations (Fig. 4⇓, E and F and Table II⇓) showed the occurrence in IFN-γR KO mice of much larger proportions of immature elements of the monocyte-macrophage and granulocytic lineages than in wild-type mice. The net numbers of T, NK, and NKT cells, as determined by FACScan, were not different from those in similarly immunized wild-type mice. As a contrast, Mac-1+ splenocytes expanded in a significantly different way in IFN-γR KO than in wild-type mice (Fig. 5⇓A). In wild-type mice, the total number of Mac-1+ cells increased for only 5 days, subsequently decreased and increased again from day 35 to reach a maximum level of 12.8 × 106 on day 41. In IFN-γR KO mice, the increase in the number of Mac-1+ spleen cells was significantly more pronounced as it reached a maximum of 36.4 × 106 cells at an earlier time point (day 21). At that point, the Mac-1+ cell population represented 47.7 ± 4.0% of total splenocytes, against 10.2 ± 2.3% in wild-type mice (n = 6). Interestingly, in both strains of mice, the maximal increase in the Mac-1+ cell population coincided with disease onset as noted in previous experiments (Fig. 5⇓A and Table I⇑). The earlier and more pronounced increase in the splenic Mac-1+ cell population in IFN-γR KO mice thus correlates with accelerated and increased disease incidence in these mice.
Histological changes in spleens and Cytospin splenocyte preparations of IFN-γR KO and wild-type mice on day 24 postimmunization with CII in CFA. A–D, Hematoxylin/eosin-stained sections of spleens of IFN-γR KO (A and C) and wild-type (B and D) mice. Note extensive extramedullary hemopoiesis in IFN-γR KO mice with megakaryocytes and myeloid and erythropoietic cells. Magnification: ×61 (A and B); ×272 (C and D); ×257 (E and F). E and F, Cytospin splenocyte preparations showing a large number of band cells (ring-shaped nucleus) in IFN-γR KO mice (E), but not in wild-type counterparts (F).
Expansion of the Mac-1+ splenocyte population following CFA-assisted immunization with CII in IFN-γR KO or in anti-IFN-γ Ab-treated mice, but not in wild-type mice. A, Numbers of Mac-1+ cells in spleens of IFN-γR KO and wild-type mice sacrificed on indicated days postimmunization. Splenocyte suspensions were analyzed by FACScan. Data points indicate means of two (days 6 and 10), three (days 13, 17, 19, 25, 35, and 48), four (days 23 and 41), or six (day 21) mice; bars represent ± SEM. ∗, p < 0.01 for comparison with wild-type mice (Student’s t test). ∗∗, p < 0.05 (Student’s t test) for comparison with age-matched naive wild-type mice (not shown on the graph). Splenocytes from IFN-γR KO and wild-type mice that were immunized with CII in IFA were also analyzed (on days 6, 10, 15, 20, 26, 39, and 42), but their number of Mac-1+ cells never reached levels higher than 12.4 × 106 (peak levels on day 10) and without significant difference between the two strains. B, Increased numbers of Mac-1+ splenocytes in wild-type mice immunized with CII in CFA but treated with anti-IFN-γ Abs. Wild-type mice received either anti-IFN-γ or irrelevant rat IgG2a (200 μg per injection) on days 0, 7, and 14 postimmunization. IFN-γR KO mice were also included for comparison. At day 20, spleens were removed, stained with anti-Mac-1, and analyzed by FACScan. The top panel shows a population of cells (indicated by gate R1) exclusively present in IFN-γR KO and in anti-IFN-γ Ab-treated wild-type mice. Gated cells are strongly positive for the expression of Mac-1 (lower panel, black areas). The numbers represent the percentages of the total number of splenocytes expressing Mac-1. The white unshaded areas represent numbers of cells stained with an isotype-matched control Ab. An increased Mac-1+ splenocyte population in anti-IFN-γ-treated vs control-treated mice was also found at days 22, 26, and 35.
Spleen leukocyte populations of IFN-γR KO and wild-type mice after immunization with CII in CFAa
In mice immunized with CΙΙ in IFA, the Mac-1+ cell population was analyzed on days 6, 10, 15, 20, 26, 39, and 42 and revealed a slight increase in both mouse strains till day 10 (11.7 ± 1.4 and 12.4 ± 0.4 × 106 for two IFN-γR KO and wild-type mice, respectively). Their number then reverted to basal level and never was significantly different between the mutant and wild-type mice (data not shown).
An increased Mac-1+ cell population was also seen in wild-type mice immunized with CII in CFA that had been treated on days 0, 7, and 14 with anti-IFN-γ Abs (Fig. 5⇑B). As we reported earlier (19), such treatment causes an accelerated onset and increased disease incidence similar to that observed in the mutant mice.
Treatment with anti-IL-12 or anti-IL-6 Ab inhibits both Mac-1+ splenocyte expansion and development of CIA
Increased numbers of immature myeloid cells (Table II⇑ and Fig. 4⇑) and enhanced expansion of the Mac-1+ cell population following CFA-assisted immunization of IFN-γR KO mice (Fig. 5⇑) might be due to the production of hemopoietic cytokines. Among the possible candidates, we primarily focussed on IL-12. Our reason for this choice was that the stimulatory effects of IL-12 on hemopoiesis are counteracted by IL-12-induced IFN-γ (23) in much the same way as IL-12 and IFN-γ affect the pathogenesis of CIA in opposite directions (24). We also considered IL-6 as a possible candidate because we found circulating levels of IL-6 to be enhanced in IFN-γR KO mice suffering from CIA (data not shown). Hence, we studied the effect of treatment with anti-IL-12 or anti-IL-6 Abs on both the expansion of Mac-1+ cells and on the development of arthritis. In each of two parallel experiments, three groups of IFN-γR KO mice were immunized with CII in CFA (day 0) and were given i.p. Ab injections on days 0, 7, and 14 (anti-IL-12 or irrelevant IgG) or every other day from day 0 till day 26 (anti-IL-6). Groups of naive IFN-γR KO mice were included. In the first experiment, two mice of each group were sacrificed on days 17, 22, and 36 to analyze the splenocyte population. In the second experiment, mice were evaluated for signs of arthritis. Fig. 6⇓A shows that anti-IL-12 Ab blocked and anti-IL-6 inhibited expansion of the Mac-1+ population (day 22). Fig. 6⇓B shows that, in parallel with the effect on the Mac-1+ population, treatments with the Abs blocked (anti-IL-12) or inhibited (anti-IL-6) development of the disease.
Inhibition by anti-IL-6 and anti-IL-12 Ab treatments of the expansion of the Mac-1+ cell population and the development of CIA in IFN-γR KO mice immunized with CII in CFA. A, FACScan of Mac-1-expressing splenocytes in naive IFN-γR KO mice and in IFN-γR KO mice immunized with CII in CFA (day 0) and injected i.p. with anti-IL-12 Ab (200 μg on days 0, 7, and 14), anti-IL-6 Ab (580 μg; every other day from day 0 to day 26), or irrelevant IgG (200 μg; days 0, 7, and 14). Scatter graphs are from pooled splenocytes of two mice sacrificed on day 22. Numbers indicated on each graph represent percent splenocytes expressing Mac-1. Similar profiles were obtained with splenocytes of mice sacrificed on days 17 and 36. B, Cumulative incidence of arthritis in IFN-γR KO mice immunized with CII in CFA and treated with anti-IL-12 Ab (n = 10), anti-IL-6 Ab (n = 6), or irrelevant IgG (n = 10). Ab treatment schedules were as in A; results of one representative experiment of three.
The IFN-γR defect coordinately affects the Mac-1+ cell population and DTH reactivity against CΙΙ
To substantiate and mechanistically clarify the connection between the Mac-1+ cell population and arthritis development, we compared DTH development to CΙΙ in IFN-γR KO mice with that in wild-type mice, both after immunization with CΙΙ in IFA or CFA. DTH reactivity was tested by the footpad swelling test on day 21, i.e., at the time of maximal expansion of the Mac-1+ cell population in the IFN-γR KO mice. Fig. 7⇓ shows that, at this early time point in pathogenesis, DTH development was restricted to IFN-γR KO mice that had received the CFA-assisted immunization schedule. This observation strengthens the concept of a relation between expansion of the Mac-1+ population, development of DTH for CΙΙ, and joint involvement.
Increased DTH to CII in IFN-γR KO mice immunized with CII in CFA. IFN-γR KO or wild-type mice were immunized with CII in either CFA or IFA. On day 21, DTH reactivity against CII was tested by the footpad swelling test (see legend to Fig. 3⇑B). Bars represent averages ± SEM. ∗, p < 0.05 for comparison with wild-type mice (Student’s t test).
Discussion
CIA, a polyarthritis that develops in DBA/1 mice following immunization with CII in either IFA or CFA, is used as a model for rheumatoid arthritis in humans (25). In virtually all studies published so far, CFA rather than IFA has been used, as it facilitates induction of the disease. This effect is generally assumed to be due to the potentiating effect of heat-killed mycobacteria on various stages of the immune response. The facilitating effect of CFA was confirmed in the present study, as we found accelerated and more frequent disease when we used mice immunized using CFA than in those immunized using IFA (Table I⇑). However, both clinically and histologically the arthritic lesions were similar whether CFA or IFA was used. Here, as in a previous study, we have analyzed the role of IFN-γ in CIA by comparing the disease course in IFN-γR KO mice with that in wild-type counterparts. We confirmed that ablation of endogenous IFN-γ aggravates CIA induced by immunization of the mice with CII in CFA. Significantly, however, we found that the opposite is true in CIA induced by an IFA-assisted immunization schedule. Under these conditions IFN-γR ablation inhibited disease progression, thus revealing a disease-promoting role of IFN-γ. Apparently, under the influence of mycobacteria from the CFA, a pathogenic pathway becomes dominant, which is counteracted by endogenous IFN-γ, resulting in the disease-mitigating effect of IFN-γ.
Inhibition of CIA by ablation of IFN-γ in mice immunized without assistance of mycobacteria was associated with suppression of both anti-CII Ab and DTH reactivity at the time of arthritis development. Therefore, in principle the disease-promoting component of IFN-γ’s action may be mediated by enhancement of both cellular and humoral immunity effector mechanisms. However, in our previous study using the CFA-assisted schedule (19), we reported that Ab formation was suppressed in IFN-γR KO mice, despite their increased disease scores. As shown here, DTH-responsiveness following CFA-assisted immunization was augmented by the IFN-γR KO mutation, an effect that correlated well with enhanced arthritis development in these mice. When IFA was used to induce CIA, DTH was significantly lower in the mutant than in the wild-type mice. Therefore, it would appear that augmented disease activity in the IFN-γR-deficient mice is mainly due to an increased effector function of cellular rather than humoral immunity.
Our current study establishes the cellular effector function in CIA that is inhibited by IFN-γ. Enhanced CIA in IFN-γR KO mice immunized by the CFA-assisted schedule was associated with pronounced splenomegaly, extramedullary hemopoiesis, and overexpansion of a Mac-1+ cell population. In these IFN-γR-deficient mice, the splenic Mac-1+ cell population not only reached a higher maximal level than in wild-type mice, but also did so at an earlier time point, which coincided with the time of disease onset. In wild-type mice, the expansion of the Mac-1+ cell population was less pronounced and also followed a biphasic time course. Remarkably, the second peak coincided with the later time of disease onset in these mice.
A key role for hemopoiesis in the pathogenesis of CIA is evidenced by the observation that GM-CSF-deficient mice are protected against the disease (26). Also, inhibitory effects of IFN-γ on hemopoiesis have been documented in several systems. The stimulatory effect of IL-12 on hemopoiesis was reported to be counteracted by IFN-γ (23), and production of CSF-1 by monocytes was found to be inhibited by IFN-γ (27). Of special relevance is a study demonstrating a dramatic effect of the IFN-γ knockout mutation on hemopoietic remodeling during infection with Mycobacterium tuberculosis bacillus Calmette-Guérin. The normal splenic architecture of the IFN-γ knockout mice was found to be effaced by expanding myeloid cells (28). Our observations indicate that a similar process takes place in IFN-γR KO mice that receive immunization schedules employing killed mycobacteria as a component of CFA. Macrophages of IFN-γ-deficient mice may perform poorly in destroying the mycobacterial cell bodies. This may result in long-term persistence of these bodies in the lymphoid system and hence also in persistent stimulation of hemopoietic cytokines. Two candidate hemopoietic cytokines, IL-6 and IL-12, were considered in particular. We found that treatment of the mice with neutralizing anti-IL-6 or anti-IL-12 Abs inhibited both the expansion of the Mac-1+ cell population and the development of arthritis. In accordance with our results, IL-6-deficient mice have been reported to be resistant against CIA (29). IL-6 is known to synergize with IL-3 (multi-CSF) in supporting the formation of multilineage blast cell colonies in murine spleen cell cultures and also to stimulate differentiation of myeloid cell lines into macrophage- and granulocyte-like cells (for review see Ref. 30). Hence requirement of IL-6 for CIA development may be due to its hemopoietic potential. IL-12 is another cytokine known to affect hemopoiesis. Specifically, in IFN-γR KO mice IL-12 administration caused rather increased bone marrow hemopoiesis and strong extramedullary hemopoiesis in the spleen (23). Our observation that anti-IL-12 Ab blocks expansion of the Mac-1+ splenic population in IFN-γR KO mice is in line with these observations. These results, obtained with anti-IL-6 and anti-IL-12 Ab treatments, are in accordance with the concept that endogenous IL-6 and IL-12 are instrumental in bringing about expansion of the Mac-1+ population, which, in turn, contributes to the development of arthritis lesions.
Why are Mac-1+ cells so important in the pathogenesis of CIA? Cells of the monocyte/macrophage lineage are known to play an important role as producers of various cytokines and other inflammatory mediators. It seems possible that the increased availability of these cells specifically in IFN-γR KO mice with CIA induced by CFA-assisted immunization leads to a higher level of infiltration of these cells into joints and thereby to a higher level of inflammatory and destructive changes. The crucial role of macrophages in destructive autoimmune diseases and the ability of IFN-γ to keep this cell population in check received support from other studies. Thus, administration of anti-Mac-1 mAb reduced the severity of CIA in DBA/1 mice and prevented its adoptive transfer to SCID mice (31). Furthermore, mice deficient in both B and T cell functions, in spite of their inability to mount Ag-specific immune responses, were found to be sensitive to induction of CIA, leading the authors to the conclusion that lymphocyte-independent pathogenic mechanisms operate in this model (32). Also, in bacterial cell wall-induced arthritis in rats, exogenous IFN-γ was reported to exert a protective effect, which correlated with an inhibitory effect of IFN-γ on chemotactic responsiveness of mononuclear phagocytes, apparently resulting from reduced expression of C5a receptors accompanying enhanced expression of Ia Ag (33). Finally, in experimental allergic encephalomyelitis, a model in which CFA is used for induction and in which endogenous IFN-γ protects against disease, the Mac-1+ cell population was shown to play an essential pathogenic role (34, 35).
Collectively, our data indicate that 1) the role of the mycobacterial component of CFA consists in stimulating myelopoiesis, potentially leading to expansion of Mac-1+ cells that act as DTH and arthritogenic effectors, and 2) that endogenous IFN-γ normally restrains the expansion of this population. When mycobacteria are omitted from the induction schedule, the stimulus for myelopoiesis is significantly reduced. Therefore, one might speculate that, under these circumstances, the inhibitory effect of IFN-γ on the number of Mac-1+ cells may be of less consequence for disease activity, while its activating effect on the aggressiveness of these cells and on Ag-specific reactivity of T and B cells becomes the dominant pathogenic factor. However, further investigation is needed to obtain evidence for these views.
Our results have important implications for the clinical relevance of the CIA model. The classical CIA model, which relies on CFA-assisted immunization, is likely to represent clinical forms of arthritis involving bacterial infections or other exogenous factors that stimulate myelopoiesis. CIA induced without the help of mycobacteria may be representative of clinical situations in which the myeloid effector population plays a less prominent role. In rheumatoid arthritis, the role of myeloid effector cells may vary from one time to another or from one patient to another. This may explain why in clinical trials systemic treatment with IFN-γ has yielded variable results ranging from no effect to transient improvement (36, 37)
Our data also provide a possible explanation for protective effects of IFN-γ in experimental autoimmune diseases other than CIA. Examples are experimental autoimmune encephalomyelitis (10, 11, 12, 13, 14, 15, 16) and experimental autoimmune uveitis (17, 18). So far, a satisfactory explanation for protection by IFN-γ in these models has not been offered. However, these models likewise rely on the use of CFA for immunization, and ablation of IFN-γ has been shown to be associated with increased influxes of monocytes/macrophages and neutrophils in the lesions (14, 15, 18). Therefore, it seems likely that the protective effect of endogenous IFN-γ in these models, like in CIA, resides in its inhibitory effect on myelopoiesis induced by the mycobacterial component of CFA. Finally, by revealing a central role for myeloid cells in the pathogenesis of CIA, our results identify myelopoiesis as a potential target for novel antirheumatic drugs that has so far not been sufficiently explored.
Acknowledgments
We thank Prof. G. Opdenakker and Prof. J. Van Damme for their critical reading of the manuscript.
Footnotes
-
↵1 Work in the authors’ laboratory is supported by grants from the Fund for Scientific Research Flanders (FWO Vlaanderen), from the Regional Government of Flanders (GOA program), and from the Belgian Federal Government (Inter-University Network for Fundamental Research). P.M. is a Postdoctoral Research Fellow of the FWO.
-
↵2 Address correspondence and reprint requests to Dr. Patrick Matthys, Rega Institute, University of Leuven, Laboratory of Immunobiology, Minderbroedersstraat 10, B-3000 Leuven, Belgium. E-mail address: Patrick.Matthys{at}rega.kuleuven.ac.be
-
↵3 Abbreviations used in this paper: CIA, collagen-induced arthritis; CII, collagen type II; DTH, delayed-type hypersensitivity; IFN-γR KO, IFN-γR knockout.
- Received May 20, 1999.
- Accepted July 7, 1999.
- Copyright © 1999 by The American Association of Immunologists
References
In this issue
