CTLA4 gene polymorphism correlates with the mode of onset and presence of ICA512 Ab in Japanese Type 1 diabetes

Diabetes Research and Clinical Practice 46 (1999) 169 – 175 www.elsevier.com/locate/diabres

CTLA4 gene polymorphism correlates with the mode of onset and presence of ICA512 Ab in Japanese Type 1 diabetes Takahiro Abe a, Hirofumi Takino a, Hironori Yamasaki a, Masako Ozaki a, Yasunori Sera a, Hideaki Kondo a, Hiroyuki Sakamaki a, Eiji Kawasaki a, Takuya Awata b, Yoshihiko Yamaguchi a, Katsumi Eguchi a,* a

The First Department of Internal Medicine, Nagasaki Uni6ersity, School of Medicine, 1 -7 -1 Sakamoto, Nagasaki 852 -8501, Japan b The Fourth Department of Internal Medicine, Saitama Medical School, Saitama, Japan Received 6 May 1999; received in revised form 11 June 1999; accepted 5 July 1999

Abstract Recently, the association of CTLA4 gene polymorphism with type 1 diabetes and AITD has been reported in several populations. CTLA4 was originally reported to regulate T-cell activity and T – B cognate interaction. To investigate the role of CTLA4 in autoimmune diseases, we examined the correlation between CTLA4 gene polymorphism and the clinical characteristics of Japanese patients with type 1 diabetes, including the mode of onset of diabetes and presence of islet-specific autoantibodies (GAD, ICA 512 Ab) in the serum. We studied 111 patients with type 1 diabetes and 445 normal subjects. CTLA4 exon 1 position 49 (A/G: codon 17: Thr/Ala) polymorphism was defined, employing PCR-RFLP. Sixty-three (57%) patients had AITD. The allele frequencies of G and A in both 111 patients (G: 65%; A: 35%) and 63 patients (G: 62%; A: 38%) were not significantly different from the control subjects (G: 63%; A: 37%). Serum samples of 69 patients were obtained within a year after onset and used for pancreas specific autoantibodies analysis. These samples were also used for further analysis between CTLA4 gene polymorphism and clinical characteristics. The allele frequencies of G and A in patients who presented with diabetic ketoacidosis (DK +) (G: 75%; A: 25%) were significantly different from those in DK− patients (G: 50%, A: 50%, P= 0.003). Allele and genotype analyses showed significant differences between DK + patients and control subjects (P= 0.014, P=0.046, respectively). Allele frequencies of G and A were not significant between patients who were positive and negative for GAD Ab, but significant for ICA 512 Ab (G: 83%, A:17% versus G: 59%, A: 41%: positive patients versus negative patients, P=0.004). Our results showed a significant correlation between CTLA4 gene polymorphism and ICA 512 Ab. Our results also indicated that CTLA4 gene polymorphism is associated with the onset mode of Japanese type 1 diabetes and the presence of ICA512 Ab. Further analysis of this polymorphism is necessary to fully understand the pathogenesis and progression of type 1 diabetes. © 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Cytotoxic T lymphocyte antigen 4; ICA512 Ab; Type 1 diabetes * Corresponding author. Tel.: + 81-95-849-7260; fax: + 81-95-849-7270. E-mail address: [email protected] (K. Eguchi) 0168-8227/99/$ - see front matter © 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 8 - 8 2 2 7 ( 9 9 ) 0 0 0 8 4 - 4

170

T. Abe et al. / Diabetes Research and Clinical Practice 46 (1999) 169–175

1. Introduction Type 1 (insulin-dependent) diabetes is an organ-specific autoimmune disease, characterized by lymphocytic infiltration into pancreatic islets and presence of pancreatic specific autoantibodies in the serum [1]. Multiple genetic predisposing factors are known to influence the autoimmunity events in type 1 diabetes [2]. Using candidate gene analysis, two susceptibility genes, HLA complex and a variable number of tandem repeat in the 5% region of the insulin gene (VNTR-INS), have been identified as IDDM1 and IDDM2, respectively [3–5]. Linkage disequilibrium mapping suggests that a type 1 diabetes locus is located close to the microsatellite marker locus D2S152 (IDDM7) on chromosome 2q31 [6]. Chromosome 2q33 contains genes encoding CTLA4 (cytotoxic T lymphocyte antigen 4) and CD28. CTLA4 and CD28, glycoprotein receptors expressed in reactive T cells, are involved in the initial T cell activation by antigen-presenting cells and subsequent regulation of cellular immunity [7]. CTLA4 is a permissive candidate gene for the pathogenesis of autoimmune diseases, because it acts as a negative regulator of T cell activation and T – B cognate interaction [8 – 12]. Recent studies have shown the involvement of CTLA4 gene in several autoimmune diseases. For example, an A–G transition polymorphism at position 49 in exon 1 of the CTLA4 gene is linked to or associated with autoimmune thyroid disease (AITD) [13] and a subgroup of Addison’s disease [14]. For type 1 diabetes, the first demonstration was made by Todd et al. [4,5] who showed a significant association with CTLA4 gene polymorphism (designated IDDM12). Such an association was subsequently confirmed by several groups [15 – 17]. In Japan, the prevalence of young patients with type 1 diabetes is 1/20 – 1/30 as compared to Caucasians [18], the subgroup of type 1 diabetes that were initially diagnosed NIDDM, but subsequently progressed to an insulin-dependent state so-called slowly progressive type 1 diabetes was reported [19]. The genetic difference might exist. Based on the immunological function evoked by CTLA4, we hypothesized that CTLA4 gene

polymorphism may contribute to the clinical characteristics of type 1 diabetes in Japanese patients. In the present study, we investigated the distribution of CTLA4 gene polymorphism in Japanese type 1 diabetes with or without AITD and analyzed the relationship between polymorphism and clinical characteristic as manifested by the mode of onset of diabetes and presence of islet specific autoantibodies including GAD and ICA 512 antibodies.

2. Materials and methods

2.1. Subjects One hundred and eleven patients with type 1 diabetes (age: 1–71 years) were investigated for CTLA4 gene polymorphism. Sixty-three of these patients also had AITD (age: 2–71 years) (Graves’ disease: n= 22; Hashimoto’s thyroiditis: n =41). Serum samples were obtained from 69 patients within 1 year of onset and subjected to pancreas specific autoantibodies analysis. The clinical characteristics of these 69 patients were analyzed according to the results of CTLA 4 polymorphism. The diagnosis of type 1 diabetes was based on the criteria of the National Diabetes Data Group [20]. Briefly, type 1 diabetes was defined based on the clinical symptoms, the presence of pancreas specific autoantibodies and insulin dependency (B 0.03 nM serum C-peptide response after intravenous glucagon stimulation or postprandial state). Also, diabetic ketoacidosis was defined by the existence of both metabolic acidosis and ketonuria at onset. We also included a control group, which consisted of 445 randomly selected healthy Japanese subjects (age: 18–71 years). The study was approved by the ethic review committees of our institutions and each subject gave an informed consent.

2.2. Restriction fragments length polymorphism analysis of CTLA4 CTLA4 exon 1 position 49 (A/G: codon 17 Thr/Ala) polymorphism was defined, employing PCR with forward (5%-GCTCTACTTCCT-

T. Abe et al. / Diabetes Research and Clinical Practice 46 (1999) 169–175

171

Table 1 Comparison of CTLA4 gene polymorphism between patients and control subjectsa Genotype

Type 1 diabetes With AITD With Graves’ disease With Hashimoto thyroiditis Control

Allele

N

GG (%)

GA (%)

AA (%)

P value

G (%)

A (%)

P value

111 63 22 41 445

50 26 8 18 177

45 26 9 17 207

16 11 5 6 61

NS NS NS NS

145 78 25 53 561

77 48 19 29 329

NS NS NS NS

(45) (41) (36) (42) (40)

(41) (41) (41) (45) (46)

(14) (18) (23) (13) (14)

(65) (62) (57) (65) (63)

(35) (38) (43) (35) (37)

a P values comparing the patients with controls are given for the x 2-tests, calculated on the 2×3 or 2×2 tables or the Fisher’s exact test.

GAAGACCT-3%) and reverse (5%-AGTCTCACTCACCTTTGCAG-3%) oligonucleotides. PCR was performed using a PCR kit (Ready To Go; Amersham Pharmacia Biotech, Piscataway, NJ), 0.2 mg genomic DNA and 25 pmol of each primer under the following conditions: initial denaturation for 4 min at 94°C, annealing for 45 s at 58°C, extension for 45 s at 72°C, denaturation for 45 s at 94°C (35 cycles), and a final extension for 4 min at 72°C in a Biomed thermocycler (Program Temp Control System PC700, Astec, Shime, Japan). PCR products were digested by Bb6-I at 37°C for 2 h and fragments were separated in 2.0% agarose gels stained with ethidium bromide. The restriction enzyme Bb6-1 could cut the sequence if a G was present at position 49 of CTLA4 gene PCR product, resulting in two discrete fragments (88/74-bp); otherwise, if an A was present at that position, no digestion of the 162-bp PCR fragment occurred, as described previously [16].

ICA 512 Ab was measured by radioligand binding assay using ICA 512 bdc cDNA encoding amino acids 256–979 of IA-2 cloned in pCR II vector (Invitrogen; Netherlands), kindly provided by G.S. Eisenbarth (Barbara Davis Center for Childhood Diabetes, Denver, CO), and 35S-methionine [22]. The antibody level was expressed as an index defined as: (cpm in test sample-negative control)/(positive control–negative control). A positive radioassay for ICA 512 autoantibody was based on the 99th percentile of sera from 184 healthy control subjects, representing an index of 0.023. The inter-assay CV and intra-assay CV were 4.0 and 1.2%, respectively.

2.4. Statistical analysis Differences in the prevalence of alleles or genotypes between patients and control subjects were examined by x 2-test or the Fisher’s exact probability test, whichever appropriate. A P value B 0.05 was considered significant.

2.3. Measurement of autoantibodies GAD Ab was detected using an anti-GAD radioimmunoassay (RIA) kit (RIP Anti-GAD Hoechst, Hoechst Japan, Tokyo), as described previously. The inter-assay coefficient of variation (CV) and intra-assay CV were 6.0 and 3.0%, respectively. At the GAD Antibody Workshop under the auspices of the Immunology and Diabetes Workshop, this assay scored 100% in lab validity, 100% in specificity and 100% in sensitivity [21].

3. Results

3.1. CTLA4 gene polymorphism in type 1 diabetes with or without AITD The frequencies of CTLA4 genotype and allele in 111 Japanese type 1 diabetes patients were not significantly different from control subjects. Sixtythree out of the 111 patients (57%) also had AITD. Among them, 26 (41%) were homozygous

172

T. Abe et al. / Diabetes Research and Clinical Practice 46 (1999) 169–175

Table 2 Clinical characteristics of 69 patientsa

Measurements of both GAD Ab and ICA 512 Ab were performed in 69 patients whose clinical characteristics are depicted in Table 2. The age at onset also appears to be an important factor to interfere with the mode of onset. However, the analysis of the CTLA4 gene polymorphism with age at onset revealed no significant relationship when the patients were divided into two subgroups ( B 15 years old and \ 15 years old) (Table 3). Allele frequencies of G and A in patients with diabetic ketosis at onset designated as DK+ (G: 75%; A: 25%) were significantly different from those in patients without diabetic ketosis at onset designated as DK− (G: 50%; A: 50%; P= 0.003; Table 3). However, the difference in genotype analysis between DK+ and DK− patients was not significant. Allele and genotype analyses showed significant differences between DK+ patients and control subjects (P= 0.014, P= 0.046, respectively; Table 3). Allele frequencies of G and A in patients who were positive for GAD Ab were not significant from those negative for GAD Ab. However, the allele frequencies were significantly different in ICA 512 Ab+ and Ab− (G: 83%, A: 17% versus G: 59%, A: 41%; P= 0.004, Table 3).

Subjects Number (M/F) Age of onset (years)

69 (27/41) 25.3 9 16.6

Complicated with AITD Without AITD

33 (48%) 36 (52%)

Number with GAD-Ab ICA512 Ab GAD-Ab or ICA512

48 (70%) 21 (30%) 51 (74%)

Age value represent the mean 9S.D. of the group. The frequency data represent the number of patients while the percentage data appear in parenthesis. a

for G, 26 (41%) were heterozygous for G/A, and 11 (18%) were homozygous for A (Table 1). The allele frequencies of G and A (G: 62%; A: 38%) were not significantly different from the control (G: 63%; A: 37%; Table 1). In 22 type 1 diabetic patients with Graves’ disease and 41 patients with Hashimoto’s thyroiditis, the frequencies of CTLA4 genotype and allele were not significantly different from control subjects (Table 1).

3.2. CTLA4 gene polymorphism in type 1 diabetes with or without ketosis at onset and islet specific antibodies Table 3 CTLA-4 gene polymorphism in patients with type 1 diabetesa Genotype N DK(+) 46 DK(−) 23 Age B15 20 Age ` 15 49 GAD Ab(+) 48 GAD Ab(−) 21 ICA512 Ab(+) 21 ICA512 Ab(−) 48 GAD Ab or ICA512 Ab(+) 51 all Ab(−) 18 Control 445 a

Allele

GG (%)

GA (%)

AA (%)

P% value

G (%)

A (%)

P’ value

P value

27 5 10 22 24 8 14 18 26 6 177

15 13 9 19 19 9 7 21 20 8 207

4 5 1 8 5 4 0 9 5 4 61

0.046† NS NS NS NS NS 0.028† NS NS NS

69 23 29 63 67 25 35 57 72 20 561

23 23 11 35 29 17 7 39 30 16 329

0.003†

0.014† NS NS NS NS NS 0.004† NS NS NS

(59) (22) (50) (45) (50) (38) (67) (37) (51) (33) (40)

(33) (56) (45) (39) (40) (43) (33) (44) (39) (44) (46)

(8) (22) (5) (16) (10) (19) (0) (19) (10) (23) (14)

(75) (50) (73) (64) (70) (60) (83) (59) (71) (56) (63)

(25) (50) (28) (36) (30) (40) (17) (41) (29) (44) (37)

NS NS 0.004† NS

P% comparison of patients with to those without. P, comparison of patients with control subjects. The P values are given for the x -tests, calculated on the 2×3 or 2×2 tables or the Fisher’s exact test. † was considered significant. 2

T. Abe et al. / Diabetes Research and Clinical Practice 46 (1999) 169–175

4. Discussion Recent studies have demonstrated that CTLA4 gene polymorphism is associated with AITD in Japanese patients [23,24] while other groups described the association between gene polymorphism and type 1 diabetes [16,17]. We expected that CTLA4 gene polymorphism would show more striking differences in Japanese type 1 diabetes complicated with AITD. The results, however, were quite the opposite to our expectations (Table 1). In contrast to our findings, a French group has recently reported a significant association between polymorphism and patients with both IDDM and AITD [25]. The difference between the two studies might be due to racial differences. We reported here a significant association between CTLA4 gene polymorphism and the mode of onset of type 1 diabetes in Japanese patients. Another independent study of Japanese patients has also reported a significant relationship between acute onset and insulin-depleted diabetes (so called typical IDDM) [23]. The results of both studies indicate that the presence of G allele of CTLA4 gene may be involved in determining the mode of onset of type 1 diabetes in Japanese patients. The role of CTLA4 gene polymorphism in the pathogenesis of autoimmune disease remains, however, unclear. Studies, using CTLA4 gene knockout mice, have provided evidence of rapid and massive T lymphocytic infiltration in most organs including lymph nodes, spleen, thymus, heart and pancreas [26,27]. We speculated that such polymorphism may alter immune function of CTLA4 protein or its gene expression. Recent studies have indicated that a polymorphic microsatellite (AT)n repeat in the 3% untranslated region of the CTLA4 gene has a strong linkage disequilibrium with the exon 1 G/A polymorphism and might affect mRNA stability [25]. These abnormalities in T cell antigen recognition might help acute destruction of pancreatic b cells leading to acute onset type 1 diabetes with DK. Genovese et al. reported that the prevalence of ICA 512 antibodies was markedly increased in DR4 positive patients and they had markedly

173

higher titers of IA-2 antibodies [28]. In the current study, we reported the positive association between ICA 512 Ab and CTLA4 gene polymorphism. In approximately 500 Belgian diabetic patients, Auwera et al. [29] showed no relationship between CTLA4 gene polymorphism and prevalence of immune markers, such as IAA, ICA, GAD Ab and ICA 512 Ab. Although we do not have convincing reasons, so far, for the difference between the two studies, our results showed a significant correlation between CTLA4 gene polymorphism and the presence of ICA 512 Ab. One possible reason for producing different results might be the patient selection in which prevalence of type 1 diabetes with AITD was much higher than that found of Caucasian (48% in our study versus 3–10%, as usually reported). Another reason for that might be due to the genetical differences between Caucasians and Japanese that we still do not know. Another interesting finding in our study was found that as the patients with DK+ at onset were subdivided based on the presence of GAD antibody (DK+ , GAD, Ab+ , G: 88%, A: 12% versus control in the allele frequency, PB 0.001). These results might indicate that CTLA4 would contribute to typical type 1 diabetes, i.e. acute onset form of diabetes with pancreas specific autoantibodies. However, there are no in vitro or animal studies that have shown either that CTLA4 function may directly associate with antibody production in autoimmune diseases or that this polymorphism would alter immunological function. A larger population sample is required to clarify the association between GAD antibody production and CTLA-4 polymorphism. 5. Conclusion In summary, we showed that CTLA4 gene polymorphism is associated with the mode of onset of type 1 diabetes and ICA 512 Ab in Japanese patients. The role of CTLA4 gene polymorphism in the pathogenesis of type 1 diabetes is, however, not delineated yet. Further analysis of this polymorphism is required to fully understand the pathogenesis and progression of type 1 diabetes.

174

T. Abe et al. / Diabetes Research and Clinical Practice 46 (1999) 169–175

Acknowledgements We thank S. Yasunaga for her excellent technical assistance. We also thank Professor F.G. Issa, from the Department of Medicine, University of Sydney, Sydney, Australia, for the careful reading and editing of the manuscript. This work was supported in part by a grant from the Ministry of Health and Welfare, Government of Japan.

References [1] G.S. Eisenbarth, Type 1 diabetes mellitus. A chronic autoimmune disease, N. Engl. J. Med. 314 (1986) 1360– 1368. [2] J.L. Davies, Y. Kawaguchi, S.T. Bennett, et al., A genome-wide search for human type 1 diabetes susceptibility genes, Nature 371 (1994) 130–136. [3] L. Hashimoto, C. Habita, J.P. Beressi, et al., Genetic mapping of a susceptibility locus for insulin-dependent diabetes mellitus on chromosome 11q, Nature 371 (1994) 161 – 164. [4] J.A. Todd, Genetic analysis of type 1 diabetes using whole genome approaches, Proc. Natl. Acad. Sci. USA 92 (1995) 8560 – 8565. [5] J.A. Todd, M. Farrall, Panning for gold: genome-wide scanning for linkage in type 1 diabetes, Hum. Mol. Genet. 5 (1996) 1443 – 1448. [6] J.B. Copeman, F. Cucca, C.M. Hearne, Linkage disequilibrium mapping of a type 1 diabetes susceptibility gene (IDDM7) to chromosome 2q31–q33, Nat. Genet. 9 (1995) 80 – 85. [7] T.L. Walunas, C.Y. Bakker, J.A. Bluestone, CTLA-4 ligation blocks CD28-dependent T cell activation, J. Exp. Med. 183 (1996) 2541–2550. [8] P. Lane, Regulation of T and B cell responses by modulating interactions between CD28/CTLA4 and their ligands, CD80 and CD86, Ann. N.Y. Acad. Sci. 815 (1997) 392 – 400. [9] J.G. Gribben, G.J. Freeman, V.A. Boussiotis, et al., CTLA4 mediates antigen-specific apoptosis of human T cells, Proc. Natl. Acad. Sci. USA 92 (1995) 811–815. [10] P. Waterhouse, L.E.M. Marengere, H.W. Mittrucker, T.W. Mak, CTLA-4, a negative regulator of Tlymphocyte activation, Immunol. Rev. 153 (1996) 183– 207. [11] E.A. Tivol, F. Borriello, A.N. Schweitzer, W.P. Lynch, J.A. Bluestone, A.H. Sharpe, Loss of CTLA-4 leads to massive lymphoproliferation and fatal multiorgan tissue destruction, revealing a critical negative regulatory role of CTLA-4, Immunity 3 (1995) 541–547. [12] Y. Wu, Y. Guo, A. Huang, P. Zheng, Y. Liu, CTLA-4B7 interaction is sufficient to costimulate T cell clonal expansion, J. Exp. Med. 185 (1997) 1327–1335.

[13] T. Yanagawa, Y. Hidaka, V. Guimaraes, M. Soliman, L.J. DeGroot, CTLA4 gene polymorphism associated with Graves’ disease in Caucasian population, J. Clin. Endocrinol. Metab. 80 (1995) 41 – 45. [14] H. Donner, J. Braun, C. Seidl, et al., Codon 17 polymorphism of the cytotoxic T lymphocyte antigen 4 gene in Hashimoto’s thyroiditis and Addison’s disease, J. Clin. Endocrinol. Metab. 82 (1997) 4130 – 4132. [15] L. Nistico`, R. Buzzetti, L.E. Pritchard, et al., The CTLA4 gene region of chromosome 2q33 is linked to, and associated with, type 1 diabetes, Hum. Mol. Genet. 5 (1996) 1075 – 1080. [16] H. Donner, H. Rau, P.G. Walfish, et al., CTLA4 alanine17 confers genetic susceptibility to Graves’ disease and to type 1 diabetes mellitus, J. Clin. Endocrinol. Metab. 82 (1997) 143 – 146. [17] M.P. Marron, L.J. Raffel, H.J. Garchon, et al., Insulindependent diabetes mellitus (IDDM) is associated with CTLA4 polymorphism in multiple ethnic groups, Hum. Mol. Genet. 6 (1997) 1275 – 1282. [18] H. Ikegami, T. Ogihara, Genetics of insulin-dependent diabetes mellitus, Endocr. J. 43 (1996) 605 – 613. [19] T. Kobayashi, T. Itoh, K. Kosaka, K. Sato, K. Tsuji, Time course of islet cell antibodies and b-cell function in non-insulin-dependent stage of type 1 diabetes, Diabetes 36 (1987) 510 – 517. [20] National Diabetes Data Group, Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance, Diabetes 28 (1979) 1039 – 1057. [21] E. Kawasaki, H. Takino, M. Yano, et al., Autoantibodies to glutamic acid decarboxylase in patients with IDDM and autoimmune thyroid disease, Diabetes 43 (1994) 80 – 86. [22] C.F. Verge, R. Gianani, E. Kawasaki, et al., Prediction of type 1 diabetes in first-degree relatives using a combination of insulin, GAD, and ICA512bdc/IA-2 autoantibodies, Diabetes 45 (1996) 926 – 933. [23] T. Awata, S. Kurihara, M. Iitaka, et al., Association of CTLA-4 gene A-G polymorphism (IDDM12 locus) with acute-onset and insulin-depleted IDDM as well as autoimmune thyroid disease (Graves’ disease and Hashimoto’s thyroiditis) in Japanese population, Diabetes 47 (1998) 128 – 129. [24] T. Yanagawa, M. Taniyama, S. Enomoto, et al., CTLA4 gene polymorphism confers susceptibility to Graves’ disease in Japanese, Thyroid 7 (1997) 843 – 846. [25] I. Djilali-Saiah, E. Larger, E. Harfouch-Hammoud, et al., No major role for the CTLA-4 gene in the association of autoimmune thyroid disease with IDDM, Diabetes 47 (1998) 125 – 127. [26] P. Waterhous, J.M. Penninger, E. Timms, et al., Lymphoproliferative disorders with early lethality in mice deficient in CTLA-4, Science 270 (1995) 985 – 988. [27] E.A. Tivol, S.D. Boyd, S. McKeon, et al., CTLA4Ig prevents lymphoproliferation and fatal multiorgan tissue destruction in CTLA-4-deficient mice, J. Immunol. 158 (1997) 5091 – 5094.

T. Abe et al. / Diabetes Research and Clinical Practice 46 (1999) 169–175 [28] S. Genovese, R. Bonfanti, E. Bazzigaluppi, et al., Association of IA-2 autoantibodies with HLA DR4 phenotypes in IDDM, Diabetologia 39 (1996) 1223–1226. [29] B.J. Auwera, C.L. Vandewalle, F.C. Schuit, et al., CTLA-

.

175

4 gene polymorphism confers susceptibility to insulin-dependent diabetes mellitus (IDDM) independently from age and from other genetic or immune disease markers, Clin. Exp. Immunol. 110 (1997) 98 – 103.