3134 K. Takacs et al. Eur. J . Immunol. 1995.25: 3134-3141

Exacerbated autoimmunity associated with a T helper-1 cytokine profile shift in H-2E-transgenic mice

Katalin Takaics, Daniel C. Douek and Daniel M. Altmann

Transplantation Biology Group, Clinical Sciences Centre, Royal Postgraduate Medical School, Hammersmith Hospital, London, GB

Major histocompatibility complex (MHC) class I1 genes are the strongest SUS- ceptibility markers for many human autoimmune diseases. A perplexing aspect of this is that HLA alleles can confer either susceptibility or dominant protec- tion. In nonobese diabetic (NOD) mice, the strongest known diabetes susceptib- ility locus is within the MHC and is presumed to be the H-2Ag7 product. When NOD mice carry a transgenic Ead molecule allowing expression of an H-2E heterodimer, diabetes is prevented. We investigated whether, as in human auto- immunity, a single class I1 heterodimer might protect from some autoimmune diseases while predisposing to others. NOD mice are susceptible to experimen- tal autoimmune encephalomyelitis (EAE) induced by the proteolipoprotein (PLP) epitope 56-70. Susceptibility to EAE was analyzed in NOD mice which either have or lack transgenic H-2E expression. We found that H-2E expression in NOD mice has converse effects on diabetes and EAE: while diabetes is pre- vented, EAE is greatly exacerbated and leads to demyelination. Although PLP 56-70 could be presented both in the context of H-2A and H-2E, increased dis- ease severity in H-2E transgenic mice could not be attributed either to an enhanced T cell proliferative response to PLP or to differences in determinant spread. However, cytokine analysis of the response revealed important differ- ences between NOD mice and their H-2E transgenic counterparts: H-2E expres- sion was associated with reduced interleukin-4 secretion and enhanced interferon-y (IFN-y) secretion by lymph node cells, while the response of central nervous system infiltrating T cells displayed a markedly enhanced IFN-y response. Thus, whether a particular class I1 molecule confers resistance or sus- ceptibility to an autoimmune disease may depend on differential cytokine profi- les elicited by particular class IVautoantigen complexes.

1 Introduction

The strongest genetic markers of susceptibility in many human autoimmune disease are HLA genes, most com- monly alleles of HLA class 11. The mechanisms of these associations are unknown. HLA alleles can confer either susceptibility or dominant protection, as exemplified by the observation that the DQB1*0602 gene confers an enhanced risk for multiple sclerosis but a reduced risk for type I diabetes [l, 21.

Most of the murine models of autoimmune disease, whether spontaneous diseases such as the type I diabetes of NOD mice, or experimentally-induced diseases such as collagen arthritis and experimental allergic encephalomy- elitis (EAE), are also controlled by MHC genes [3, 41. In NOD mice, the strongest region of disease susceptibility maps to the H-2 complex and includes the H-2Ag7 gene

[I 146641

Correspondence: Daniel M. Altmann, Transplantation Biology Group, Clinical Sciences Centre, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London W12 ONN, GB (Fax: +44-181-259-83 03; E-mail : daltmann @ rpms.ac.uk)

Abbreviations: MSCH: Mouse spinal cord homogenate D L N Draining lymph nodes

Key words: Experimental autoimmune encephalomyelitis I H-2E I Proteolipoprotein I Transgenic

product [S]. This strain does not express an H-2E heterodi- mer because of a deletion in the H-2EA promoter [6]. A number of groups have demonstrated that when NOD mice carry a transgenic Ead molecule allowing expression of an H-2E heterodimer, diabetes is prevented [7, 81. Pro- tection does not appear to be dependent on thymic posi- tive or negative selection of particular T cell receptors, but may be a function of antigen presentation via H-2E in the periphery [9]. One possibility is that H-2E expression by APC causes determinant capture, preventing access of H- 2A molecules to the diabetogenic autoantigen epitope [lo]. This is compatible with the finding that protection by H-2E is only absolute in transgenic lines carrying multiple copies of the gene: when mice with a single copy of H-2Ek are made by crossing NOD H-2E congenics with wild-type NOD mice, a proportion of H-2Et mice develop diabetes WI. Protection from autoimmune disease by H-2E expression in mice which normally lack H-2E appears to be a wide- spread phenomenon. Protection has been demonstrated following H-2E expression in various other autoimmune models including the lupus of BXSB mice [12], collagen arthritis [ 131 and experimental allergic myasthenia gravis [14]. The mechanism of protection in these diseases is unknown, although the suppressive effects of H-2E in vari- ous models led to the notion of H-2E as an immune sup- pression gene. In EAE of Lewis rats, MHC-dependent alterations in disease susceptibility correlate with differ- ences in Tcell cytokine responses [15].

0014-298019511111-3134$10.00 + .2510 0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim, 1995

Eur. J. Immunol. 1995.25: 3134-3141 Exacerbated autoimmune disease in H-2E transgenic mice 3135

We investigated whether, as in human autoimmunity, a single class I1 heterodimer might protect from some auto- immune diseases while predisposing to others. H-2E- mediated protection from autoimmune pathology is not a universal phenomenon, since an Ea transgene which pro- tects from diabetes does not protect from sialitis (P. Chandler and E. Simpson, personal communication). A recent study reported that Biozzi AB/H and NOD mice (both lacking H-2E and sharing H-2Ag7) are susceptible to EAE induced by the proteolipoprotein (PLP) epitope amino acids 56-70, although disease severity is only mild in NOD mice [16]. We have analyzed susceptibility to EAE in NOD mice which either have or lack transgenic H-2E expression. There has been a previous study of H-2Ead transgenic A.CA mice which are of H-2‘ haplotype and express no H-2EB chain: no differences were observed between nontransgenic mice and transgenic mice express- ing H-2Ead/AB‘ hybrid molecules with regard to EAE onset, course and neuropathology 1171. Furthermore, it is clear that H-2E is not generally protective in EAE, since some H-2E’ strains are susceptible to EAE, and in PL/J mice, for example, a sub-dominant H-2E-restricted en- cephalitogenic epitope can be detected [HI.

We report here that H-2E expression in NOD mice has opposite effects on diabetes and EAE: while diabetes is prevented, EAE is greatly exacerbated.

2 Materials and methods

2.1 Mice

N0D.E mice carrying an Ead transgene have been described [8]. Full protection from diabetes in the trans- genic mice from this line has been described in detail [8, 191. Homozygous transgenic animals, NOD and BALB/c mice were bred at the Clinical Research Centre, Harrow, GB. All the mice used in this study were 7-11 weeks of age at the start of experiments.

2.2 Antigens

Mouse spinal cord homogenate (MSCH) was prepared from spinal cord of C57BL/10 mice, freeze dried and reconstituted in PBS before use. Peptide PLP 56-70 (DYEYLINVIHAFQYV) was used for EAE induction, a substituted variant (lysine for tyrosine at position 57 and 59) was used for tests in vitro since the substitution in- creases solubility of the peptide without affecting antigeni- city [20]. In epitope spread studies, overlapping peptides of MBP (19-23-mers) and PLP (16-mers) were used at a final concentration of 50 pg/ml. The peptide panels were the kind gift of Prof. N. Groome (Oxford Brookes Univer- sity, GB).

2.3 Induction of EAE

To induce EAE with MSCH, mice received 1 mg C57BL/ 10 MSCH in Freund’s incomplete adjuvant supplemented with 200 pg mycobacteria (Mycobacterium tuberculosis and M . butyricum 8 : 1) S.C. on days 0 and 7. To induce dis- ease with peptide, 200 yg PLP 56-70 emulsified in CFA

(as above) was injected S.C. on days 0 and 7. Immediately after injection of antigen and again 24 h later, mice received 200 ng Bordetella pertussis toxin (Porton Prod- ucts Ltd., Salisbury, GB) i.p. in PBS. Mice were weighed from day 11 onward and checked for neurologic signs: 0 = normal; 1 = limp tail; 2 = impaired righting reflex; 3 = partial hind limb paralysis; 4 = complete hind limb paraly- sis; and 5 = moribund. Mice exhibiting less severe signs than typically observed were scored as 0.5 less than the indicated grade.

2.4 Proliferation assays

Mice were immunized in each hind footpad with 50 pg PLP peptide emulsified in CFA. After 9 to 10 days, popli- teal draining lymph nodes (DLN) were removed and sin- gle-cell suspensions prepared in HL-1 serum-free medium (Hycor Biomedical, Irvine, CA.) supplemented with L- glutamine (2 mM), 2-mercaptoethanol (5 x lo-’ M), peni- cillin (30 IU), streptomycin (30 pg/ml). Cells (4 x lo5) were cultured in triplicate in 96-well flat-bottom microcul- ture plates in the presence of the substituted PLP 56-70 or, as a positive control, with 50 pg/ml purified protein deriv- ative of M. tuberculosis (PPD). The cultures were incu- bated for 3 days. [’HI Thymidine (0.5 pCi) was added to each well 6 h before termination, and cultures were then harvested for counting in a beta scintillation counter. In all of the assays shown, the background response of cultures in the absence of antigen was <500 cpm. For epitope spread studies, spleen cells were prepared at day 21 after immunization and screened for responses to overlapping MBP and PLP peptides at a final concentration of 50 pg/ ml.

2.5 Histology

Mice were killed by halothane anesthetic and spinal cords removed for fixation in 10 % phosphate-buffered formalin. Paraffin sections (2 pm thick) were stained with hematoxy- lin and eosin (H & E) or luxol fast blue to detect infiltra- tions and demyelinated nerve fibers.

2.6 Monoclonal antibody blocking

MHC class II-specific mAb were used to block proliferat- ive responses of DLN T cells in vitro. The peptide concen- tration was 10 pg/ml. The mAb 14.4.4s (anti-H-2Ea) and OX-6 (anti-H-2Ag7) were used as affinity-purified Ig at 4.5 Pg/ml.

2.7 Cytokine analysis

T cell proliferation assays of immunized lymph node cells (or CNS infiltrating cells) were set up as described above. Cells were stimulated with PLP 56-70 at a concentration of 50 pg/ml. After 66 h stimulation, 50 yl supernatant from each well was removed €or determination of cytokine production. The remainder of each microculture was pulsed with [3H]thymidine and cultured for a further 6 h, then incorporated radioactivity was counted. The IL-2, IL-4, IL-10, IFN-y and TNF-a content of the supernatants

3136 K. Takacs et al.

was measured by specific ELISA, measuring against stan- dard curves (Endogen, Cambridge, MA).

2.8 Isolation of mononuclear cells from the CNS

Infiltrating cells were separated by discontinuous density gradient centrifugation [21] from four NOD and five N0D.E mice showing early symptoms of EAE. Mice were anesthetised and perfused through the heart with PBS. The brains and spinal cords were dissociated by passing through a cell strainer. The dissociated tissue was centri- fuged at 200 g for 10 min and the pellet resuspended in 4 ml 70% Percoll (Pharmacia) in Hanks’ BSS. This was overlaid by equal volumes of 37% and 30% Percoll and the gradient was centrifuged at 500 x g for 15 min. The mononuclear cells were harvested from the 37-70 YO inter- face, washed in Hanks’ BSS twice and counted. Infiltrating cells (2 X lo4) separated from NOD or N0D.E CNS were stimulated in vitro with 25 &ml peptide PLP 56-70 in the presence of 2 X lo5 irradiated NOD or N0D.E spleen cells as APC. Supernatants from 60 h cultures were ana- lyzed for IL-4 and IFN-y content by ELISA.

2.9 Statistical analysis

Differences of disease severity between NOD and N0D.E mice and the cytokine data were analysed using the Mann- Whitney U-test. The differences in the frequency of dis- ease were analyzed using Yates corrected x2 test.

3 Results

3.1 EAE induction in NOD and N0D.E-transgenic mice

To investigate the effect of H-2E expression on susceptibil- ity to EAE, wild-type NOD mice and N0D.E transgenic animals were injected either with MSCH and pertussis toxin or PLP-56-70 peptide and pertussis toxin, since both have previously been reported to be mildly encephalito- genic in NOD mice [16]. Injection of MSCWpertussis toxin is a relatively aggressive regimen of EAE induction which can induce disease in a wide variety of inbred strains [22]. Using this regime, both transgenic and non- transgenic mice suffered severe disease (Fig. 1 A). H-2E transgenic NOD mice are thus not protected from EAE. In fact, though the onset of disease was slightly delayed in transgenic mice, they became more seriously paralyzed and, unlike wild-type controls, failed to recover. We did not observe relapsing and remitting disease in either group. The transgene-dependent exacerbation of disease was even more pronounced when EAE was induced by the PLP 56-70 peptide (Fig. 1B). Of 10 NOD mice, 6 developed relatively mild disease as has been previously described in this strain [ 161. In contrast, transgenic mice showed significantly more serious EAE (p < 0.001). More- over the incidence of disease was higher in the transgenic group 0, < 0.005).

Histological examination of CNS sections from coded, immunized mice was conducted blind to verify clinical dis- ease scores. Fig. 2 shows histopathological analysis of lon- gitudinal spinal cord sections from transgenic and wild-

Eur. J. Immunol. 1995.25: 3134-3141

A B

10 20 30 40 Days after disease induction Days after disease induction

Figure 1. Severity of EAE differs between NOD and N0D.E mice. EAE was induced by MSCH (A) or PLP 56-70 peptide (B) in NOD (open circles) and N0D.E mice (closed circles). In (A) 12/12 NOD mice suffered EAE with a mean score of 2.5 f 1.0 and a mean day of onset 18.6 f 2.9. All 11 N0D.E mice suffered EAE with a mean score of 3.3 f 0.7 and a mean day of onset 19.9 f 2.7. In (B), 6/10 NOD mice suffered EAE with a mean group score of 1.3 f 1.3 (mean f SD of the maximum clinical score of EAE from all animals in the group), a mean EAE score of 2.1 f 1.0 (mean f SD of the maximum clinical score from animals developing EAE within a group) and a mean day of onset 17.5 f 1.2. All 9 N0D.E mice suffered EAE (p <0.005) with a mean group score of 3.5 2 0.8 (mean f SD of the maximum clinical score of EAE from all animals in the group) (p < 0.001), a mean EAE score of 3.5 f 0.8 (mean f SD of the maximum clinical score from animals developing EAE within a group) and a mean day of onset 15.8 f 1.3.

type mice. Fields A, C and E are stained with luxol fast blue to show myelinated nerve fibers; fields B and D are stained with hematoxylin and eosin to show cellular infiltration. Fields A and B are contiguous sections from a non-transgenic mouse 16 days after immunization where normal fibres can be seen in longitudinal section. Fields C and D are contiguous sections at higher magnification from a transgenic mouse 16 days after immunization. Field C shows disarray of nerve fibers with segmental demye- lination, intrathecal edema and an infiltrate of mononu- clear cells. Field D demonstrates the mononuclear cell infiltrate at the bottom right of the section with some intra- thecal edema separating the nerve fibers. Field E is a sec- tion from a paralyzed transgenic mouse 4 weeks after immunization. It shows a central area of pallor in the absence of a mononuclear cell infiltrate, indicating chronic demyelination.

Thus, contrary to several other autoimmune animal models, H-2E expression does not confer resistance to EAE in the NOD model, but rather causes an exacerba- tion of disease leading to chronic demyelination.

3.2 Disease exacerbation is not a consequence of enhanced responses in H-2E transgenic mice

The simplest explanation for the increased disease severity was an enhanced response to PLP 56-70 in transgenic mice. To investigate this, NOD and N0D.E transgenic mice were immunized with PLP 56-70 peptide and DLN cells removed 10 days later for rechallenge with the same peptide in vitro (Fig. 3). Tcell proliferation in response to peptide was the same in the two groups. It has previously been shown in the EAE of (B1O.PL X SJL)F1 mice that, during the 21 days following immunization with an en- cephalitogenic peptide, the response may spread to other epitopes in the target molecule [23, 241. We therefore

Eur. J. Immunol. 1995.25: 3134-3141 Exacerbated autoimmune disease in H-2E transgenic mice 3 137

Figure 2. Histological examination of CNS tissue shows demyelination in transgenic mice. Longitudinal 2-pm-thick paraffin sections of formalin fixed spinal cords were stained with luxol fast blue (A, C and E) or hematoxylin and eosin (B and D) to detect infiltrations and demyelinated nerve fibers, respectively. Fields A and B are contiguous sections from a non-transgenic mouse. Fields C and D are con- tiguous sections from a transgenic mouse and field E is from a chronically paralyzed mouse.

investigated the possibility that differential epitope spread between wild-type and H-2E transgenic NOD mice may account for differences in disease severity. In long-term immunization experiments to investigate the spread of responsiveness from the immunizing epitope PLP 56-70 to other PLP or MBP epitopes, we found no evidence of epi- tope spread in either group (Fig. 4).

3.3 PLP 56-70 is presented both by H-2A and H-2E

While we found no evidence for an enhanced Tcell response in transgenic mice, the influence of H-2E and the absence of epitope spread suggested that H-2Ed was able to present PLP 56-70 which has previously been described only as an epitope presented via H-2Ag7 in either Biozzi or

NOD mice (which share this class I1 allele [16]). Indirect evidence in support of presentation via H-2Ed came from immunization of BALB/c mice (which differ at H-2A, but share H-2Ed with N 0 D . E transgenic mice) with PLP 56-70. BALB/c mice immunized with this peptide responded in vitro to challenge and moreover, developed moderately severe EAE (data not shown). Direct evidence for H-2E presentation comes from antibody blocking of DLN Tcell responses of NOD or N0D.E transgenic mice against PLP 56-70. Proliferation of the NOD Tcell line was inhibited by the anti-H-2,4 mAb OX-6, but not by the anti-H-2Eu mAb 14.4.48 (Fig. 5 A). However, presenta- tion of peptide to H-2E transgenic T cells was inhibited by either the anti-H-2A mAb or the anti-H-2Eu mAb, with an almost complete inhibition requiring both mAb (Fig. 5 B). In the experiment shown, the T cell response to PLP

3138 K. Tak6cs et al. Eur. J. Immunol. 1995.25: 3134-3141

of IFNy (< 8 nglml) in spite of the highT cell proliferation observed in each case. Each of the seven N0D.E samples contained higher concentration of this cytokine (> 15 ngl ml), suggesting a possible dominance of Thl cells in this response in H-2E transgenic mice. However, because of the variability in the IFNy response of non-transgenic NOD mice, the difference does not reach statistical signifi- cance. This degree of variability is in line with the variabil- ity generally observed in susceptibility to EAE induction. Supernatants from N0D.E T cell cultures contained little IL-4 (< 50 pglml in all except one sample) whereas T cells from NOD mice produced relatively large amounts of IL-4 (> 50 pglml in seven of nine mice). The difference between transgenic and non-transgenic mice is significant when mean IL-4 concentration normalized to cpm data is calculated (p < 0.007). Taken together, the reduced IL-4 response and enhanced IFN-y response observed in H-2E transgenic mice suggest a shift of the response induced by this peptide from mixed Thl/Th2 to predominantly Thl.

Since it has previously been shown that IL-10 production is associated with recovery from EAE [25], we also com- pared IL-10 production by Tcells from the two groups; no difference was detected. Data from EAE studies in SJL mice suggests that TNF-a production may correlate well with pathogenesis [26], although no TNF was detectable in supernatants at the time points analyzed in the present study (data not shown).

It was likely, however, that by sampling responses in peripheral lymph nodes, we had not obtained a clear rep- resentation of events at the site of pathogenesis in the CNS. In multiple sclerosis (MS), for example, there are differences between findings from analysis of PBL and CSF T cells [27]. Similarly, a difference has been reported between the cytokine profile of DLN cells and infiltrating cells separated from the CNS [28]. We therefore analyzed

h E 250000 3 v

200000

2 150000 8

2 1000Oo

2

s 50000 E

.-

0 m Y

0 25 50 75 100

Peptide concentration (Fglml)

Figure 3. Proliferative response to peptide PLP 56-70 by DLN cells. Six NOD (open circles) and six N0D.E (closed circles) mice were immunized with 50 pg peptide emulsified in CFA in the hind footpads. Popliteal lymph node cells were rechallenged in vitro with peptide 10 days later.

was lower in NOD mice than in N0D.E transgenics. This difference, however, was within the range of variability we had previously shown within groups (see Figs. 3 and 6) and did not indicate a consistently enhanced response by trans- genic mice.

3.4 Cytokine release by T cells responding to PLP 56-70

We then analyzed the cytokine profile of T cells derived from NOD and N0D.E mice responding to peptide PLP 56-70, measuring the IFN-y, TNF-a, IL-4 and IL-10 con- tent of supernatants from proliferation assay cultures by ELISA (Fig. 6). As in other experiments, there was no dif- ference between transgenic and non-transgenic mice with respect to the proliferative response to peptide. Four of nine supernatants from peptide-stimulated T cell cultures from individual NOD mice contained relatively low levels

A C

D 2 a

2 MBP peptide panel & MBP peptide panel

Figure 4. Analysis of epitope spread in NOD and N0D.E mice. Spleen cells were rechallenged in vitro with peptides from MBP (over- lapping 19-23-mers) and PLP (overlapping 16-mers) peptide panels 21 days after disease induction to detect spread of the NOD response to PLP epitopes (A) and MBP epitopes (B) and of the N0D.E response to PLP epitopes (C) and MBP epitopes (D). The sequential numbering shown for peptides on the x-axis starts with PLP 1-15 (A) and MBP 1-20 (B).

Eur. J. Immunol. 1995.25: 3134-3141 Exacerbated autoimmune disease in H-2E transgenic mice 3139

A

+OX 6 +I4 4. 1 :::r- 1 No Ag

0 1000 2000 3000

B

+OX 6 +14.4.4

+ 14.4.4s

+ O X 6

No Ab - No Ag

0 I o m 20000 30000

[3H] thymidine uptake (cpm)

Figure 5. Antibody blocking of peptide PLP 56-70-induced pro- liferation of DLN Tcells from NOD (A) or N0D.E transgenic mice (B). Anti-H-2Ea (14.4.4.S) and anti-H-2A (OX-6) mAb were added to the cultures at a concentration of 4.5 pg/ml.

the responses to T cells isolated from the CNS of mice dur- ing the onset of disease (Table 1). Cells were collected by purification over Percoll gradients on day 15, when the weight loss which precedes overt disease was first detected. In the experiments, the most severely affected mice in each group were analyzed: in the experiment shown, all three mice in the N0D.E group showed initial signs of disease and mouse number 2 in the non-transgenic group showed weight loss. Analysis of CNS infiltrates showed that the difference between the response of NOD and N0D.E mice was more pronounced than had been seen using DLN cells. Tcells from N0D.E transgenic mouse CNS make substantially more IFNy in response to PLP peptide than do Tcells from wild-type mice. The

NOD N0D.E Transgenic

1000 1 m 1000 10000

n

IFNy ng/ml 0 5 10 15 20 0 5 10 15 20

9 8 7 6 5 4 1 2 I IL-4 pg/ml . . .

0 50 100 150 200 0 50 100 150 200

IL-10 1 Jlml 0 1 2 3 0 1 2 3

Figure 6 . Cytokine production of NOD and N0D.E Tcells in response to peptide PLP 56-70. IFN-y, IL-4, and IL-10 contents of supernatants of DLN proliferation cultures were measured by ELISA.

Table 1. IFN-y release by CNS-infiltrating T cells in response to PLP 56-70 in mice at the onset of disease

Infiltrating Infiltrating cpm IFN-y) cells total ceIIs/well ( x w3) (ndml)”) ( X 10-7 (X

Mouse no. N 0 D . E 1 3.0 2.0 8.2 34.5 2 2.4 2.0 23.0 77.9 3 0.6 2.0 28.1 67.8 NOD 1 0.5 Z!.O 9.6 5.8 2 0.6 2!.0 5.8 5.7 3 0.6 2.0 7.2 n.t.b)

a) Results are shown for individual mice analyzed at day 15 after disease induction.

b) n.t. = not tested.

absolute difference is likely to be more extreme than indi- cated by the data in Table 1, since the experiment was designed such that we had selected for the more severely affected mice in the control group which also contained a majority of unaffected mice. Furthermore, for analysis of responses, we normalized cell numbersiwell between the two groups, whereas the CNS of transgenic mice in fact contained far greater numbers of infiltrating cells. In con- trast to the response by lymph node cells, IL-4 release by CNS-infiltrating cells was barely detectable, perhaps indi- cating that cells of a more Th2 phenotype do not cross the blood-brain barrier (data not shown).

4 Discussion

In a broad range of murine autoimmune models using strains which are normally H-2E-, introduction of an intact H-2E heterodimer protects from disease [7, 8, 12-14]. The original findings on prevention of diabetes in H-2Ea’ NOD mice [7, 81 were interpreted as possible evi- dence for H-2E-restricted regulatory or suppressor cells [29]. An alternative hypothesis was suggested when it was subsequently shown that expression of H-2E can have pro- found effects on TCR repertoire, both by endogenous superantigen-dependent deletion of particular TCR V@+ families and by TCR Vfl-specific positive selection. How- ever, in studies using H-2E transgenic mice with promoter mutations directing H-2E expression to different lymphoid compartments and bone marrow chimera experiments, H- 2E-mediated protection was shown to be dependent on expression by peripheral antigen presenting cells, irre- spective of thymic expression [9, 191. These observations may be explained by determinant capture on H-2E of an epitope linked to the diabetogenic peptide [lo].

H-2E-dependent protection was similarly seen in other disease models such as spontaneous lupus in BXSB mice [12], collagen-induced arthritis in BlO.RQB3 mice [13] and experimental allergic myasthenia gravis in C57BW10 mice [14]. In none of these models was it possible to define a mechanism for protection, although in one study a differ- ential effect on stimulation of Th2 and Thl cells was pro- posed [13]. Taken together, the studies implied some gen-

3140 K. Takacs et al. Eur. J. Immunol. 1995.25: 3134-3141

era1 mechanism by which the superimposition of a trans- genic H-2E molecule in mice which would normally express only H-2A, can shift the immune response away from pathogenesis. EAE is a disease in which H-2E- dependent protection might not be expected, since H-2" mice, which include EAE-susceptible strains such as PL/J, express H-2E molecules. Furthermore, an H-2E-restricted sub-dominant encephalitogenic epitope, MBP 35-47, has been defined in (PL/J X SJL)F, mice [HI.

Using the NOD EAE model, we demonstrated that dis- ease is markedly exacerbated in H-2E' mice. The simplest explanation for this would have been that by supplying more class I1 molecules for peptide presentation (albeit via H-2E rather than the H-2Ag7 molecule which has previ- ously been described as the restriction molecule for this epitope), T cell proliferation in response to the peptide was enhanced. In fact, the dose response to peptide was the same in transgenic and wild-type mice. Furthermore there was no evidence of epitope spread in either group of mice, despite chronic disease involving demyelination, presum- ably with release of antigenic material from myelin sheaths

An alternative possibility is that by altering the MHC molecules available for peptide presentation, T cell activa- tion is qualitatively different, leading to a change in cyto- kine profiles. There is a precedent for such a change in the Tcell response to a collagen peptide when presented via either H-Ab or H-2As [30]. In EAE, different cytokine profiles are associated with the onset of disease and dis- ease recovery or protection from disease by tolerization protocols. Kennedy and co-workers analyzed cytokine mRNA expression during the onset of and recovery from EAE passively transferred to SJL mice by MBP-specific lymph node cells. The onset of disease was associated with local production of inflammatory cytokines such as IFN-y, IL-1 and IL-2. Levels of these cytokines declined during recovery while the level of IL-10 increased [25]. However, studies with neutralizing antibody suggest an anti- inflammatory role for IFN-y in disease, and indeed, SJL mice can be protected with exogenous IFN-y [31]. This is in contrast to EAE of Lewis rats, where disease in suscep- tible rats is associated with high IFN-y but little IL-4 or TGF-8 release, while resistance in MHC-congenic rats is associated with high IL-4 and TGF-8, but little IFN-y [15]. In another model of mouse EAE, where disease is induced in PL/J mice by Ac-MBP 1-11, encephalitogenicity of T cell clones correlates with TNF-a production [32]. Anti- TNF-a neutralizing antibody inhibits pathogenesis in vivo [26], which may reflect an effect on pathogenic Tcells or alternatively may act at the level of the resident parenchy- mal microglia which are the main source of CNS TNF-a during EAE [33]. In our system, the availability of an H- 2E molecules for peptide presentation is associated with a change in the cytokine profile of responding Tcells from high IL-4 to high IFN-y. As would be expected from the inter-mouse variability which we and others observe in EAE severity, mice also showed relatively variable cyto- kine responses. N0D.E transgenic mice gave uniformly high (mean of 17 ng/ml) IFN-y responses, while responses in NOD mice ranged from 6.6 ng/ml to 19 ng/ml, but were generally lower than in the transgenic mice. However, IL-4 responses were significantly reduced in the transgenic mice. In the diabetes of NOD mice, IL-4 is associated with

protection from pathology while IFN-y appears to have a pro-inflammatory role in the islets of Langerhans. In our analysis of Tcells infiltrating the CNS, a strong IFN-y response was correlated with susceptibility. It is note- worthy that both transgenic and non-transgenic mice show equally strong proliferative responses to PLP by DLN cells from the periphery, yet significantly more cells cross the blood-brain barrier and home to the CNS in transgenic mice. This may be an important consequence of their differential cytokine profile. The more markedly Thl phe- notype of cells from transgenic mice is likely to enhance IFN-y-mediated induction of VCAM-1 on endothelium at the blood-brain barrier [34, 351.

In summary, we demonstrated that H-2E expression may exacerbate some autoimmune diseases while protecting from others. Disease exacerbation was associated with a shift in cytokine release of anti-PLP T cells responding to a single peptide in the context of different class I1 mole- cules. Thus, whether a particular class I1 molecule confers disease resistance or susceptibility may be determined by differential cytokine profiles elicited by specific class II/ autoantigen complexes.

The authors would like to thank Drs. E. Simpson, A. Cooke and 7: Lund for making transgenic mice available for this study, and Prof. N. Groome and Dr. l? Byfield for peptides. K. Takacs is supported by the Multiple Sclerosis Society of Great Britain and Northern Ire- land. D. C. Douek is supported by the Wellcome Trust.

Received July 13, 1995; in final revised form September 5 , 1995; accepted September 5 , 1995.

5 References

1 Todd, J. A., Bell, J. I. and McDevitt, H. O., Nature 1987.329: 599.

2 Spurkland, A, , Ronningen, K. S . , Vandvik, B., Thorsby, E. and Vartdal, F., Human Immunol. 1991. 30: 69.

3 Fritz, R. B., Skeen, M. J., Chou, C. J. , Garcia, M. and Ego- rov, I. K., .I. Immunol. 1985. 134: 2328.

4 Huse, A , , Wooley, P. H., Luthra, H. S. , Stuart, J. M. and David, C. S. , Fed. Proc. 1984. 43: 1820.

5 Wicker, L. S. , Miller, B. J., Coker, L. Z . , McNally, S. E., Scott, S. , Mullen, Y. and Appel, M. C., .I. Exp. Med. 1987. 165: 1639.

6 Hattori, M., Buse, J. B., Jackson, R. A. , Glimcher, L., Dorf, M. I., Minami, M., Makino, S., Moriwaki, K . , Kuzuya, H., Imura, H., Strauss, W. M., Sidman, J . G. and Eisenbarth, G. S. , Science 1986.231: 733.

7 Nishimoto, H., Kikutani, H., Yamamura, K. and Kishimoto, T., Nature 1987. 328: 432.

8 Lund, T., O'Reilly, L., Hutchings, P., Kanagawa, O., Simpson, E., Gravely, R., Chandler, P., Dyson, J., Picard, J. K., Edwards, A, , Kioussis, D. and Cooke, A., Nature 1990. 345: 727.

9 Bohme, J., Schuhbaur, B., Kanagawa, O., Benoist, C. and Mathis, D., Science 1990. 249: 293.

10 Hongkui, D., Apple, R., Clare-Salzer, M., Trembleau, S. , Mathis, D., Adorini, L. and Sercarz, E., J . Exp. Med. 1993. 178: 1675.

11 Podolin, P., Pressey, A. , DeLarato, N. H., Fischer, P. A., Peterson, L. B. and Wicker, L. S. , .I. Exp. Med. 1993.178: 793.

12 Merino, R., Iwamoto, M., Fossati, L., Muniesa, P., Araki, K., Takahashi, S. , Huarte, J., Yamamura, K., Vassali, J. and Izui, S. , J . Exp. Med. 1993. 178: 1189.

Eur. J. Immunol. 1995.25: 3134-3141 Exacerbated autoimmune disease i n H-2E transgenic mice 3141

13 Gonzales-Gay, M. A., Nabozny, G. H., Bull, M. J., Zanelli, E., Douhan, J. III., Griffiths, M. M., Glimcher, L. H., Luthra, H. S. and David, C. S., J. Exp. Med. 1994.180: 1559.

14 Christadoss, P., David, C. S., Shenoy, M. and Keve, S. , Immu- nogenetics 1990.31: 241.

15 Mustafa, M., Vingsbo, C., Olsson, T., Ljunghdahl, A., Hoje- berg, B. and Holmdahl, R., Eur. J . Immunol. 1993.23: 3089.

16 Amor, S., Baker, D., Groome, N. andTurk, J. L., J . Immu- nol. 1993. 150: 5666.

17 Cross, A. H., Ishikawa, S. , Raine, C. S. and Diamond, B., Lab. Invest. 1992. 66: 598.

18 Zamvil, S. S. , Mitchell, D. J., Powell, M. B., Sakai, K., Roth- bard, J. B. and Steinman, L., J . Exp. Med. 1988. 168: 1181.

19 Parish, N. M., Chandler, P., Quartey-Papafio, R., Simpson, E. and Cooke, A, , Eur. J . Immunol. 1993. 23: 2667.

20 Smith, R. M., Morgan, A. and Wraith, D. C., Immunology 1994. 83: 1.

21 Zeine, R. and Owens, T., J. Neuroimmunol. 1992. 40: 57. 22 Montgomery, I . N. and Rauch, H. C., J. lmmunol. 1982. 128:

23 Lehmann, P. V., Forsthuber, T., Miller, A. and Sercarz, E. E.,

24 Cross, A. H., Tuohy, V. K. and Raine, C. S . , Cell. Immunol.

25 Kennedy, M., Torrance, D. S. , Picha, K . S. and Mohler, K.

421.

Nature 1992. 358: 155.

1993. 146: 261.

M., J. Immunol. 1992. 149: 2496.

26 Ruddle, N. H., Bergman, C:. M., McGrath, K . , Lingenheld, E. G., Grunnet, M. L., Padula, S. J. and Clark, R. B., J. Exp. Med. 1990. 172: 1193.

27 Steinman, L., Weisman, A. and Altmann, D., Molecular Med- icine Today 1995. I : 79.

28 Renno, T., Zeine, R., Girard, J . M., Gillani, S. , Dodelet, V. and Owens, T., Intl. Immunol. 1994. 6: 347.

29 Oliviera, D. B. G. and Mitchison, N. A, , Clin. Exp. Immunol. 1990. 75: 167.

30 Murray, J. S., Pfeiffer, C., Madri, J. and Bottomly, K., Eur. J. Immunol. 1992. 150: 559.

31 Billiau, A , , Heremans, H., Vandekerckhove, F., Dijkmans, R., Sobis, H., Meulepas, €3. and Carton, H., J. Zmmunol. 1988. 140: 1506.

32 Powel, M. B., Mitchell, D., Ledermann, J., Buckmeier, S. , Zamvil, S . , Graham, M., Ruddle, N. H. and Steinman, L., Int. Immunol. 1990. 2: 539.

33 Renno, T., Krakowski, M., Piccirillo, C., Lin, J. and Owens, T., J . Immunol. 1995. 154: 944.

34 Baron, J. L., Madri, J. A., Ruddle, N. H., Hasim, G. and Janeway, C. A, , J. Exp. Metl. 1993. 177: 57.

35 Kuchroo, V. K., Martin, C. ,4., Greer, J. M., Ju, S. T., Sobel, R. A. and Dorf, M. E., J. Immunol. 1993. 151: 4371.