The Development of Autoimmune Inflammatory Arthropathy in Mice Transgenic for the Human T Cell Leukemia Virus Type-1 env-pX Region Is Not Dependent on H-2 Haplotypes and Modified by the Expression Levels of Fas Antigen¹

Rheumatoid arthritis (RA)³ is a systemic, chronic inflammatory disorder that mainly affects the joints (1). The disease is autoimmune in nature, and certain class II MHC Ags, in particular, products of certain HLA-DR4 and HLA-DR1 genes, are associated with a large proportion of the patients (2). RA patients often develop autoantibodies and/or cellular immunity against various self-substances including IgG (3), type II collagen (IIC) (4, 5, 6), and heat shock proteins (7). Immune reactions against these molecules have been suspected to play important roles in the pathogenesis of the disease. Moreover, augmented expression of various cytokines, including IL-1, IL-2, IL-6, TNF-α, TGF-β, and IFN-γ, was found in the joints of RA patients (8, 9, 10). Since these cytokines promote synovial cell growth and stimulate immune reactivity, they may also be involved in the development of the disease.

Human T cell leukemia virus type-1 (HTLV-I) is the causative agent of adult T cell leukemia (11). This virus encodes a transcriptional transactivator, Tax, in the env-pX region that transactivates transcription from the cognate viral promoter as well as many host cell promoters (12), including cytokines (13, 14, 15, 16), cytokine receptors (17, 18), and immediate early transcriptional factor genes (19, 20). Previously, we reported that transgenic mice (pX-Tg; C3H/HeN background), carrying the HTLV-1 env-pX region under its own long terminal repeat promoter, developed chronic inflammatory polyarthropathy at a high incidence (21). The arthritis developed as early as 4 wk of age and usually affected multiple joints, including fingers, ankles, and knees. The histopathology of the lesions closely resembled that of RA in humans, showing marked synovial and periarticular inflammation with articular erosion caused by invasion of granulation tissue (22). Immunologically, these mice produced Abs against IgG, IIC, and heat shock proteins accompanied by IgG hypergammaglobulinemia (23). The content of agalactosylated forms of IgG N-linked sugar chains was increased in those transgenic mice, as in RA patients (24). Furthermore, various genes, including inflammatory cytokine genes such as IL-1α, IL-1β, IL-6, TNF-α, and IFN-γ genes, immediate early genes, such as c-fos and c-jun, and class I and class II MHC genes were activated in the transgenic joints (23, 25), probably due to the transcriptional transactivating activity of the tax gene. Since these pathological findings were very similar to those of RA, a possibility that HTLV-I is one of the etiologic agents of RA was suggested (21). Actually, epidemiological studies support this notion (26). Therefore, pX-Tg mice are thought to be an important disease model to elucidate pathogenesis of RA in humans.

Irradiated, nontransgenic mice transferred with pX-Tg mouse bone marrow cells developed arthritis, and on the contrary, normal mouse bone marrow cells transferred into irradiated transgenic mice suppressed development of arthritis (S.S., and Y.I., unpublished data). Moreover, development of arthritis was greatly reduced in athymic pX-Tg mice carrying the nu genes. These observations, together with elevated levels of autoantibodies in the circulation (23), suggest the involvement of autoimmunity in arthritis in those mice. In relation to this, we recently found that T cells from pX-Tg mice are refractory against anti-Fas Ab treatment (27). Since Fas-mediated apoptosis of T cells has been suggested to be important in eliminating autoreactive T cells in the periphery, this finding suggested a possibility that these transgenic mice may become autoimmune due to the defect in Fas-mediated apoptosis of T cells, as is suggested in lpr/lpr mice (28).

In this study, to elucidate the pathogenesis of the autoimmunity, we backcrossed pX-Tg mice with different inbred strains of mice and examined effects of genetic backgrounds on the disease development. Involvement of MHC locus was closely examined to compare with RA in humans and IIC-induced arthritis in mice. Furthermore, to examine the involvement of Fas-mediated apoptosis of peripheral lymphocytes in the development of arthritis, effects of a deficiency of the fas gene and T cell specific overexpression of Fas (APO-1/CD95) on the disease development were studied.

Transgenic mice used in this study were those described by us previously (21), in which the pX and env regions of the HTLV-1 genome with its own promoter were introduced into fertilized mouse ova ((C3H/HeN × C57BL/6J)F₁). These mice were backcrossed with C3H/HeN, BALB/cAn, or C57BL/6J mice. Female mice at the age of 2–3 mo from backcross generations 10–12 were used for the experiments, unless specified in the figure legends. The transgene was detected through dot-blot hybridization using DNA prepared from mouse tails (23), and the littermates were used as controls.

B.10 A5R(H-2ⁱ⁵), B.10 HTG (H-2^g), B.10 D2n (H-2^d), and BALB.B (H-2^b) mice were obtained from National Institute of Genetics (Shizuoka, Japan) (29). pX-Tg mice with C57BL/6J background were crossed with B.10 A5R, B10. HTG, or B.10 D2n mice. The transgenic F₁ mice were again crossed with respective strain of mice, and the resulting transgenic F₂ mice were used for the experiments. pX-Tg mice with BALB/c background were crossed with BALB.B mice for two generations to produce H-2^b transgenic mice. H-2 haplotypes of the offspring were determined by FACS analysis using FITC-conjugated anti-mouse H-2D^b (clone KH95) and anti-mouse H-2D^d mAbs (PharMingen, San Diego, CA).

Transgenic mice carrying the fas gene were produced by injecting a transgene, in which the murine fas cDNA (30) was ligated downstream to the murine proximal lck promoter (31), into fertilized C3H/HeN mouse eggs (Y.Y., Y.I., and S.Y., unpublished data). These transgenic mice were crossed with C3H/HeN-pX-Tg mice, and the incidence of arthritis in the heterozygous hybrid mice was examined.

C3H/HeN-lpr/lpr mice were obtained from Japan SLC (Shizuoka, Japan) and crossed with C3H/HeN-pX-Tg mice for two generations to obtain lpr homozygous mice. The genotypes were determined by PCR using lpr-specific primers (32).

These mice were kept in specific pathogen-free conditions in an environmentally controlled, clean room of the Laboratory Animal Research Center, Institute of Medical Science, University of Tokyo. The experiments were conducted according to the institutional ethical guidelines for animal experiments and safety guidelines for gene manipulation experiments.

Each joint was examined weekly for swelling, redness, and rigidity, and the severity was graded from 0 to 3 for each paw, as described previously (23). The maximum grade was 12 for each mouse.

IL-1β and MHC Ag mRNAs were analyzed by Northern blot hybridization as previously described (23). Briefly, total RNA was prepared from the joints of the knees, ankles, and digits by the acid guanidinium thiocyanate-phenol-chloroform method (33), and the poly(A)⁺ RNA was purified with an oligo(dT)-cellulose column. For Northern blot hybridization, 5 μg poly(A)⁺ RNA was electrophoresed for each lane and hybridized with ³²P-DNA probe (10⁹ dpm/μg) labeled with multiprime DNA labeling system (Amersham, Boston, MA) and [³²P]dCTP (3000 Ci/mmol, NEN). IL-1β cDNA (CDMmIL-1β (34)) was kindly provided by Dr. Tetsuo Sudo (Toray Industry, Kanagawa, Japan), and the XhoI fragments (2.0 and 1.3 kb) were used as the probes. Mouse class I (H-2K^b genomic) (35) and class II MHC (Eα^d genomic) (36) were obtained from Dr. Masafumi Takiguchi (Institute of Medical Science, University of Tokyo) and Dr. Jun-ichi Miyazaki (Osaka University), respectively, and the XbaI-BglII (2.0 kb) and SalI (3.3 kb) fragments were used as the probes. β-actin was used as a control. The intensity of the bands on the autoradiogram was estimated by the BAS 2000 system (Fuji-Film, Tokyo, Japan).

The pX-mRNA (Tax) was measured by RNase protection assay (21). Total RNA (40 μg) was hybridized at 46°C overnight with an antisense RNA probe made with SP 6 RNA polymerase with the HpaI-ClaI fragment from pHLX-I that had been subcloned into pGEM-4 (Promega, Madison, WI) as a template. After RNase A and T₁ treatment at 37°C for 1 h, the RNA sample was electrophoresed on a polyacrylamide gel. In the RNase protection assay, pX mRNA was detected as a 181 b band, and env mRNA as a 191 b band.

Serum Ab levels against IgG rheumatoid factor (RF) were measured by ELISA, as described (23). Briefly, polyvinyl microtiter plates (Dynatech Laboratories, Alexandria, VA) were coated with 50 μl of heat-denatured rabbit IgG (10 μg/ml) in Tris-buffered saline (TBS; 25 mM Tris-HCl, 140 mM NaCl, pH 7.4) overnight at 4°C. After washing the plate with TBS, the plates were blocked with 1% skim milk (Sigma, St. Louis, MO), 5 mM EDTA, 0.02% NaN₃, and TBS (blocking buffer) for 1 h. Then, 50 μl of blocking buffer-diluted mouse serum was added to each well and incubated for 1 h at room temperature, followed by washing three times with 0.05% Tween 20-TBS. Alkaline phosphatase (50 μl) conjugated goat anti-mouse IgG Ab (Zymed, San Francisco, CA) (1/500 dilution) in the blocking buffer was added as the second Ab and incubated at 37°C for 1 h. After washing with Tween 20-TBS, 100 μl of 1 mg/ml p-nitrophenylphosphate (Sigma) in 50 mM NaHCO₃-5 mM MgCl₂ (pH 9.5) was added, and the absorbancy at 415 nm was measured by ELISA microreader (Colona, MTP-120, Ibaragi, Japan) after incubation at 37°C for 1 h.

The pX-Tg mouse was originally produced by injecting the HTLV-I-env-pX DNA into the pronuclei of fertilized (C3H/HeN × C57BL/6J) F₁ mouse ova (21). While the founder mouse did not develop arthritis, after backcross with C3H/HeN mice for several generations, the progeny became susceptible to the development of arthritis (Table I). The proportion of the affected mice gradually increased with advanced backcrossing, and after 11 generations, the incidence was 25% at 3 mo and 41% at 6 mo of age. The incidence was not different between male and female mice. The severity of the disease did not vary among the affected mice with different backcross generations. Since genetic factors have been suggested to be involved in the development of RA, we examined the effects of genetic background on the development of arthritis in pX-Tg mice.

As shown in Table II, the incidence of chronic arthritis changed greatly depending on the genetic background of the pX-Tg mice. The incidence increased when these mice were backcrossed to BALB/cAn mice. On this background, the disease onset was earlier than on C3H/HeN background, and the incidence was ∼64% at 3 mo and 72% at 6 mo of age. In contrast, when backcrossed with C57BL/6J mice, the incidence decreased greatly and was 0% at 3 mo and only 2% at 6 mo of age, with no significant difference between male and female mice. The average severity score of the affected mice was significantly higher on BALB/cAn than that on C3H/HeN background (Table II). These results clearly demonstrate that genetic background greatly affects the development of chronic arthritis in pX-Tg mice.

In the previous report, we showed that various genes, including inflammatory cytokine genes, immediate early genes, and MHC genes, are activated in pX-Tg mice (23, 25). Since similar activation was observed in mice transgenic for the tax gene (K. Habu, J. Nakayama-Yamada, M. Asano, S. Saijo, and Y. Iwakura, unpublished data), this activation is thought to be due to the transcriptional transactivating activity of Tax. Thus, we examined the expression of IL-1β and MHC genes in transgenic mice with different genetic backgrounds to test whether or not genetic background affects the efficiency of the tax-induced transcriptional activation. Fig. 1, A and B, show data from one of the three representative results. In mice with C3H/HeN background, the expression of the IL-1β gene was elevated more than 10-fold in arthritic, transgenic mice and 1.5- to 2-fold in nonarthritic, transgenic mice, consistent with the previous report (Fig. 1,A) (23). The class I and class II MHC gene expression was also elevated in both arthritic and nonarthritic transgenic mice (Fig. 1,B). The class II MHC molecule in C57BL/6J mice could not be detected because we used the Eα genomic DNA as the probe to detect the expression of the class II MHC mRNA, which is absent in this strain of mice due to the ∼500-bp deletion in the promoter to first exon region of the Eα gene (37). After measurement of the intensity of the bands of the RNase protection assay and Northern blot hybridization assay, the average stimulations in transgenic mice with different genetic background are shown in Fig. 1 C. Activation of these genes was similarly observed in transgenic mice with BALB/cAn, C3H/HeN, and C57BL/6J background. The activation levels were not statistically different among mice with different genetic backgrounds, both in arthritic and nonarthritic joints, although the activation of IL-1β gene was rather lower in BALB/cAn arthritic, transgenic mice than in C3H/HeN mice. Only nonarthritic mice were analyzed in the case of C57BL/6J mice because arthritic mice were not available. These results show that transactivation of host genes by Tax can similarly be induced on different genetic backgrounds.

Since autoimmunity is involved in the development of arthritis in pX-Tg mice, it seemed likely that these genetic background effects on arthritis may involve genes that affect immunological reactivity. Fig. 2 shows the levels of anti-Ig autoantibodies (RF) in the serum of the transgenic mice. The concentrations of IgG type RF were found to be significantly different with different genetic backgrounds, the highest in BALB/cAn, intermediate in C3H/HeN, and lowest in C57BL/6J background mice. The levels of this autoantibody were positively correlated with the incidence of arthritis in these animals. These observations are consistent with the notion that autoantibodies produced in these transgenic mice are involved in the development of the disease, and that the efficiency of Ab production is different among inbred strains.

It is known that MHC loci are involved in the development of arthritis in some RA in humans and collagen-induced arthritis in mice. Thus, it seemed possible that H-2 loci were responsible for the strain difference described above. Then, we examined the effects of different H-2 loci on the development of arthritis in pX-Tg mice. As shown in Fig. 3, when pX-Tg mice with C57BL/6J background were crossed with B10.D2/n mice, the incidence of arthritis remained at a low level, although the H-2 loci are the same as those of BALB/cAn mice. Other lines of B10 congenic mice, B10.HTG and B10.A(5R), also developed arthritis at a low incidence. On the contrary, when H-2 loci of the BALB/cAn background Tg mice were substituted with BALB.B, in which the H-2 loci are the same as those of C57BL/6J, the incidence of arthritis remained at a high level. The incidence of arthritis was not significantly different whether the H-2 locus was homozygous or heterozygous. Thus, the data presented in Fig. 3 contain data from both homozygous mice and heterozygous mice. These results clearly indicate that H-2 loci are not the major determinants of the susceptibility to Tax-induced arthritis.

Recently, we found that T cells from pX-Tg mice were refractory to anti-Fas induced apoptosis, and we suggested that this abnormality may be involved in the development of autoimmunity in those mice. To examine this possibility, we analyzed the effects of Fas deficiency, as well as its overexpression, on the development of arthritis. For this, C3H/HeN-lpr/lpr mice, in which the fas gene is defective due to the insertion of a retrotransposon, and C3H/HeN-fas-Tg mice, in which Fas was overexpressed on T cells, were employed. When C3H/HeN-pX-Tg mice were crossed with C3H/HeN-lpr/lpr mice, the incidence of arthritis in homozygous C3H/HeN-lpr/lpr-Tg mice significantly increased as compared with that in C3H/HeN-pX-Tg mice (Table III). In contrast, the incidence decreased when C3H/HeN-pX-Tg mice were crossed with C3H/HeN-fas-Tg mice. The incidence of arthritis in C3H/HeJ-lpr/lpr was 0/34 and 0/43 in fas-Tg mice at 6 mo of age. These results indicate involvement of Fas-Fas ligand system in the development of arthritis.

To elucidate the pathogenesis of the autoimmunity that developed in pX-Tg mice, we assessed the effects of the genetic background of these mice on the development of arthritis and found that the incidence of arthropathy differed significantly with genetic backgrounds. It was highest in BALB/cAn, intermediate in C3H/HeN, and lowest in C57BL/6J background mice. The incidence correlated well with the RF levels in the serum, consistent with the autoimmune nature of the disease. Interestingly, we found that the incidence did not depend on the H-2 haplotypes of the mice. The incidence was high when the genetic background was BALB, irrespective of the H-2 haplotypes, b/b, b/d, or d/d. On the contrary, incidence was low on the C57BL background. The low incidence of arthritis in C57BL/6J is not due to the lack of the I-E molecule, because the incidence of arthritis in BALB.B (H-2^b) mice, in which I-E molecule is also absent, was as high as that in BALB/c mice.

Although incomplete depletion of self-reactive T cell clones in the thymus is one of the possibilities for the development of autoimmunity in these mice, these results suggest that the autoimmunity in pX-Tg mice is not caused by a defect in negative selection, since the efficiency of the negative selection against a specific epitope should be different among different MHC contexts (38). In this connection, we showed that the T cell populations carrying specific TCR Vβs, such as Vβ 3, 5.1, 5.2, 11, 12, and 17a, which were reactive with endogenous mammary tumor virus (MMTV) superantigens, were deleted in the splenocytes and lymph node cells of these transgenic mice as in nontransgenic, wild-type mice, indicating that negative selection in the thymus proceeds normally in these transgenic mice (39).

These results form a clear contrast to IIC-induced arthritis in mice, in which only mice with H-2^q or H-2^r haplotypes are susceptible (40). Regarding this, David and coworkers (41, 42, 43) showed that T cell clones reactive to the arthritogenic epitopes of IIC were deleted in the thymus of most mouse strains, but this deletion was incomplete in mice carrying IIC-sensitive class II MHC molecules. They proposed that the T cells that escaped from negative selection in the thymus were activated in the periphery by arthritogenic foreign Ags and cross-reacted against components of the cartilage through Ag mimicry (41). Therefore, the pathogenesis of autoimmune development in pX-Tg mice seems to be different from that in IIC-induced arthritis. The finding that specific MHC haplotypes are not involved in pX-induced arthritis suggests that arthritogenic epitopes in pX-Tg may be heterogeneous and different among different MHC backgrounds.

In humans, it was reported that an epitope presented on DRβ-chains of DR4 and DR1 haplotypes was highly conserved in RA patients (44). However, the association of these specific HLA haplotypes with RA is not absolute. It should be noted that some patients are without these HLA genes and that HLA association is not found in some populations (1). Thus, there seems to be at least two different types of arthritis, one is restricted by MHC haplotypes and the other is not. pX-Tg mice may provide a model for patients of the latter category.

MMTV-reactive T cell populations, including Vβ11- and Vβ12-positive ones, were nearly undetectable in the spleen or lymph nodes of pX-Tg mice, as well as in wild-type mice, due to the negative selection in the thymus. However, we found that Vβ11- and Vβ12-positive T cell clones were expanded in the arthritic joints, and some of these clones were reactive to IIC (Kotani et al., unpublished data). Since MMTV-reactive T cell populations that escaped from the negative selection in the thymus should be anergic, the expansion of these populations in the affected joints suggests that the tolerance is broken in these sites. Moreover, these mice were susceptible to IIC immunization, and arthritis developed even in mice with C3H background, which is usually unresponsive to IIC-induced arthritis (23). Thus, in contrast to wild-type mice, IIC-specific T cells that are escaped from the negative selection in the thymus seem to be reactive with synovial tissues of these transgenic mice to cause arthritis.

To explain the augmented reactivity of the peripheral T cells of pX-Tg mice to IIC, we assessed roles of decreased Fas sensitivity of T cells, which we reported previously (27), in the development of autoimmunity. We found that the incidence of arthritis decreased when Fas was overexpressed in pX-Tg mice, and inversely, the incidence increased when these mice were crossed with lpr/lpr mice (Table III). These observations suggest that this abnormality of T cells against Fas-mediated apoptosis plays an important role in the development of autoimmunity. This seems reasonable because Fas-mediated apoptosis is crucial for eliminating activated autoreactive T cells in the periphery (27, 45). Since it was reported that the efficiency of Fas/Fas ligand system is different among mouse strains, it is possible that this difference of sensitivity to Fas-mediated apoptosis is involved in the strain difference of the susceptibility to arthritis (46).

However, it is clear that the resistancy of T cells against Fas-mediated apoptosis cannot completely explain the pathogenesis of autoimmunity and of arthritis, since C3H/HeN-lpr/lpr mice do not develop arthritis spontaneously. Thus, other mechanisms than the Fas-mediated apoptosis of T cells should also be involved in the development of autoimmunity. In this context, the fact that various cytokines are overproduced in the joints of these transgenic mice is worth noting (23). Since T cell anergy can be broken by the presence of excess cytokines, such as IL-2 or IL-1 (47, 48), it is possible that T cell tolerance is broken in the transgenic joints because of these excess cytokines. Then, if T cell reactivity is modified differently among mouse strains by this mechanism, this could explain the strain difference of the susceptibility to arthritis. Thus, at least two possibilities are suggested for the strain difference; one is the difference of T cell susceptibility to Fas ligand, and the other is that of T cell reactivity to cytokines that induce tolerance break. We are now trying to map the pathogenic genes by producing C57BL/6J and BALB/cAn hybrid pX-Tg mice. At any rate, these pX-Tg mice, especially BALB/cAn background mice that show very high incidence of arthritis, should be useful as a disease model to elucidate pathogenesis of autoimmunity and development of arthritis.

We thank all the members of the Laboratory Animal Research Center, Institute of Medical Science, University of Tokyo, for valuable discussions and excellent animal care.