[1] 17w Genetic Control of Multiple Sclerosis: Increased Production of Lymphotoxin and Tumor Necrosis Factor* by HLA-DR2' T Cells

[1] 228w Lymphotoxin (LT) and tumor necrosis factor-a (TNF-a) play an important role in the pathogenesis of multiple sclerosis (MS). MS is associated with the HLA-DR2, Dw2, DQb HLA class I1 haplotype. Because both LT and TNF-a are encoded in the HLA region, the HLA association of MS may be related to the production of these cytokines. To test this hypothesis, we investigated the production of LT, TNF-a, and interferon-y (IFN-y) by CD4+ T-cell lines (TCLs) specific for myelin basic protein (MBP) or tetanus toxoid (TI') isolated from MS patients and normal controls. After stimulation with specific antigen but not mitogen, TCLs from HLA-DR2+ donors produced significantly more LT and TNF-a than TCLs from DR2-donors. In contrast, HLA-DR2+ and DR2-TCLs did not differ in the production of IFN-y, a cytokine also produced by T cells but not encoded in the HLA region. Increased secretion of LT and TNF-a was unrelated to the specificity (MBP vs TT), MHC restriction (HLA-DR2 vs other DR molecules), or source (MS vs normal) of the TCLs. There was no significant association of the cytokine production with individual LT or TNF-a alleles, indicating that the increased production of these cytokines may be linked to other polymorphic genes in this region. The results suggest that the association of MS with HLA-DR2 implies a genetically determined propensity of T cells to produce increased amounts of LT and TNF-ol.

[1] 183w In initial experiments, we stimulated PBMCs from HLA-DR2+ and HLA-DR2-donors with phytohemagglutinin (PHA) and compared the production of LT, TNF-a, and IFN-y. There was no significant difference of cytokine production in these cultures (not shown). Using a more defined system., we next investigated the production of these same cytokines by 45 antigenspecific CD4+ TCLs isolated from 10 MS patients and 9 normal control donors (see Table ). The TCLs were stimulated with the relevant antigen and irradiated autologous PBMCs as antigen-presenting cells. T-cell proliferation was measured as 3H]thymidine uptake, and the culture supernatants were assayed for LT, The proliferation of HLA-DR2+ TCLs (mean antigen-induced stimulation index [SI] = 84.7 -C 24.7 SEM) and HLA-DR2-TCLs (SI = 106.4 2 35.4, p = 0.97) was similar. Furthermore, there was no correlation between the proliferation of the TCLs (SI to the specific antigen) and the production of cytokines. However, as shown in Figure 2A, the TCLs from HLA-DR2 + donors produced significantly more LT and TNF-a than the TCLs from HLA-DR2 donors. In contrast, both groups of TCLs produced similar levels of IFN-y (see Fig 2A).

[2] 253w The difference in cytokine production was only o b served when the TCLs were *stimulated with their relevant antigen but not when the TCLs were stimulated with mitogen. After stimulation with PHA, the HLA-DR2' TCLs produced 344 : L 245 pg/ml LT (mean 2 SEM) and 7,506 2 764 pglml TNF-a, whereas the DR2-TCLs produced 480 2 177 pg/ml LT ( p = 0.21) and 7,118 2 770 pg/ml TNF-cl (p = 1). Thus, PHA stimulation resulted in little LT but strong TNF-a production. Furthermore, as seen by comparison with Figure 2, the amount of LT produced by PHA-stimulated TCLs was lower than that for antigenstimulated TCLs. These differences may be related to differences in time kinetics between the different modes of stimulation or to differences in the regulation of TNF-a and LT 1241. A similar range of LT production was observed by other investigators under similar experimental conditions [25 3. The columns represent cytokine concentrations (mean 2 SEMi measured in cell coculture supernatants as described in Materials and Methods. For statistically signifcant differences, the significance level is indicated (nonparametric Mann-Whitney U test). (A) Results from all TCLr (n = 45,. (B) Separate analysis of myelin basic protein (MBPI-specific T C L (n = 31) and tetanus toxoid (TT)specific TCLr (n = 141. (C) Separate analysis of TCLr from patients ( n = 19) and normal donors ( n = 26). See the Table for ovetvieui of T-cell donors and lines. T N F a = tumor necrosis factor-a; IFN y = interferony.

[3] 65w tended to produce lower amounts of cytokines than the TCLs from normal controls (see Fig 2C). For DR2 + TCLs (LT) and DR2 -TCLs (TNF-a), the difference was not significant ( p = 0.3 and 0.4, respectively); for DR2-TCLs (LT, IFN-y), the difference was weakly significant ( p = 0.03 for both comparisons). The reasons for and the importance of these differences are not known.

[4] 107w In the group of TCLs from DR2 + donors, we distinguished between TCLs that were restricted to DR2 (as established with HLA-DR2-transfected L cells) and TCLs that were restricted to the other (non-DR2) allele of the respective donor (Fig 3 ; see Table for HLA types of donors). The DR2+, DR2-nonrestricted TCLs produced even higher levels of LT and TNF-a than the DR2 + , DR2-restricted TCLs, although the differences were not statistically significant. These results suggest that the increased production of LT and TNF-a by DR2' TCLs does not depend on the role of HLA-DR2 as the major histocompatibility complex (MHC) restriction element for antigen recognition.

[5] 178w LT is thought to be produced exclusively by T cells, whereas TNF-a may be produced by macrophages as well as by T cells. To establish the relative contribution of macrophages and T cells to TNF-a production measured in supernatants, we compared the TNF-a levels in supernatants of parallel cultures that contained either human macrophages or HLA-transfected mouse fibroblasts (L cells) as antigen-presenting cells. In a representative experiment, (TCLs + PBMCs + antigen) produced 2,225 5 178 pg/ml TNF-a, (TCLs + relevant L cells + antigen) produced 1,917 * 280 pg/ml, irradiated PBMCs alone produced 606 -t 73 pg/ml, and L cells alone did not produce detectable levels of TNF-a. A contribution by transfected mouse L cells to the TNF-a measured in the supernatants can be ruled out because mouse TNF-a would not be detected by the assay. A significant contribution by irradiated PBMCs can also be ruled out by comparing the TNF-a concentrations measured in the different cocultures. Thus, in cocultures of (antigen-specific TCLs + irradiated PBMCs + antigen), most of the TNF-a is secreted by the TCLs.

[6] 69w To establish whether there was a correlation between cytokine production and individual LT and TNF-a alleles, we compared the heterotygous subgroup TNFB*1/2 (1/2.1 or 1/2.2) with the subgroup TNFB*2/2 (2.1/2.1 or 2.2/2.2 or 2.112.2) for LT production, and the TNFA"1/2 with the TNFA*1/1 subgroup for TNF-a production. (Only 1 member of the tested group, EM in the Table, was homotygous for both of the rare alleles TNFB"1 and TNFA"2.)

[7] 88w There was no significant correlation between the cy- or TNF-a ( p = 0.41) alleles. Thus, in contrast to the positive association of LT and TNF-a production with the HLA-DR2 phenotype, there was no detectable association with cytokine polymorphisms, although there was a striking overrepresentation of TNFB"2/2 (lacking the Nco I site) in the HLA-DR2+ group (see Table ). As an incidental observation, we noted that individuals with the haplotype HLA-A1, B8 (or B44), DR3 accounted for the majority of the TNFB" 1 + and TNFA"2 + subjects.

[1] 192w Our results show that antigen-specific CD4+ T cells from HLA-DR2' donors produce increased amounts of LT and TNF-a in vitro. Both these cytokines are thought to play an important role in the pathogenesis of MS, the prototypical inflammatory demyelinating disease of the central nervous system. Because MS is associated with HLA-DR2 in North European and North American white populations, our observations indicate that in this ethnic group, at least part of the HLA association is related to a genetically determined tendency of T cells to produce increased amounts of LT and TNF-a. Both cytokines are present in acute and chronic MS lesions, where they are associated with CD3 + lymphocytes and microglia cells 126J. Furthermore, both cytokines are cytotoxic for oligodendrocytes [l l-1 31. In experimental autoimmune encephalomyelitis (EAE), secretion of LT and TNF-a by MBP-specific TCLs correlates with the encephalitogenic potential of TCLs 1271. Treatment with an anti-T N F monoclonal antibody reduced the severity of EAE transferred with MBP-specific encephalitogenic TCLs [28, 29}. Furthermore, some studies indicated that in MS patients, there is a loose correlation be-tween TNF levels in blood, serum, or culture supernatant and the clinical course 130-35).

[2] 67w Our results are consistent. with previous observations in rodent models, indicating that the cytokine profile produced by CD4 + mouse TCLs responding to the same antigen is partly controlled by the MHC class I1 genotype 136, 37). Furthermore, the MHC has a profound influence on the cytokine profile and disease severity in EAE induced by immunization with MBP peptide 63 to 88 in MHC-congenic Lewis strains 1381.

[3] 97w In principle, HLA-DR2 could be linked directly to the increased cytohne production or could serve as a marker for other genes that influence cytokine production. As to the first possibility, HLA-DR2 could exert a direct influence on cytokinc: production by its role in the trimolecular complex 139). Theoretically, HLA-DR2 might present a specti-um of peptides that, together with the presenting DR2 molecule, signal T cells to produce high amounts of LT and TNF-a. However, our observation that both DR2-restricted and nonrestricted TCLs from DK2 + individuals produced increased amounts of LT and TNF-a argues against this possibility.

[4] 65w Alternatively, HLA-DR2 may be linked to other genes that influence cytokine production. The most obvious candidates are the LT and TNF-a genes themselves. However, we found no correlation between individual cytokine alleles and cytokine production, although there was a linkage disequilibrium between HLA-DR2 and certain cytokine polymorphisms. For example, consistent with observations by others 140, 411, TNFB"2/2+ subjects were overrepresented in the DR2 + group.

[5] 226w A number of previous studies, which all relied on PBMC preparations, provided contradictory findings about the relation between LT and TNF-a production and HLAor cytokine alleles 117, 19, 25,[42][43][44][45][46]. Jacob and co-workers 142) reported that mitogenstimulated PBMCs or monocyte subpopulations from HLA-DR2 + donors, a genotype associated with lupus nephritis, produced relatively low levels of TNF-a. In contrast, the authors found no correlation between LT production by PBMCs and the MHC genotype in their mitogen-stimulated cultures 1421. Consistent with this latter result, we also found no difference in LT production between HLA-DR2 + and DR2 -individuals when we used PHA rather than the specific antigen for stimulation of PBMCs or antigen-specific TCLs. However, with regard to TNF-a, our results differ from those of Jacob and co-workers 142). For the stimulation assays with PBMCs, the discrepancy may be related to the relatively small number of samples tested in our experiments. For the experiments with antigen-specific TCLs, the discrepancy is probably explained by the different culture systems. We focused on cytokine production by antigen-specific TCLs, using antigen for specific stimulation, whereas Jacob and co-workers 1421 studied cytokine secretion by PBMCs and selected macrophage populations, using different mitogens for nonspecific stimulation. It is conceivable that in complex cocultures, interactions between various cytokines produced by different cell types show different genetic associations than are observed in a more defined system.

[6] 71w Messer and colleagues 11 71 reported that PBMCs from individuals homozygous for TNFB" 1, a very rare group, produced more LT upon stimulation with PHA than TNFB"2 homozygotes. Our analysis with antigenspecific TCLs does not support their results, since we found no significant association between LT secretion and LT alleles. Again, part of these discrepancies may be related to the different stimulation protocols and cell types used in the different studies.

[7] 88w Apart from the genes coding for LT and TNF-a, other genes in the HLA region could have an influence on cytokine production and/or play a role in the pathogenesis of MS. Candidates are, for example, the genes encoding the transporters associated with antigen processing (TAPS) or large multifunctional proteases (LMPs). However, although all these HLA-linked genes are polymorphic, none of the known polymorphisms has as yet been conclusively linked to MS , with the possible exception of a recently discovered dimorphism corresponding to a silent mutation 15 11.

[8] 126w In conclusion, our study reveals a significant association between the MS-related HLA-DR2 haplotype and the production of LT and TNF-a by antigen-specific TCLs. Since we could not relate the increased cytokine secretion to the role of HLA-DR2 as an antigenpresenting molecule or to the known cytokine polymorphisms, we speculate that HLA-DR2 may be linked to as yet unknown polymorphism(s) of these cytokine genes, or to other gene(s) in the HLA region that have a regulatory influence on cytokine production. Our results exemplify that disease-linked MHC genes may promote the production of proinflammatory cytokines by T cells. Furthermore, the data are fully consistent with the presumed pathophysiological role of LT and TNF-a in MS 111-13, 261 and encourage attempts to develop immunotherapeutical strategies aimed at these cytokines.

[9] 24w Frauke Zipp is a postdoctoral fellow of the Deutsche Forschungsgemeinschaft (DFG). The study was supported by the DFG, Sonderforschungsbereich (SFB) 2 17, and Hertiestiftung.

[1] 95w Peripheral blood mononuclear cells (PBMCs) were isolated from 10 MS patients and 9 normal individuals (Table ). Antigen-specific CD4 + TCLs were obtained using our "split-well" protocol [20, 211. In brief, 2 x 10' PBMCs were seeded in 200 pl of medium (RPMI supplemented with 2 mM Lglutamine, 100 Ulml penicillin, 100 pglml streptomycin, and 5% pooled human AB serum) in the presence of 30 pglml MBP or 4 Fglml tetanus toxoid ('IT) in 96-well roundbottom microtiter plates. Human MBP was purified as described in established protocols 122). 'IT was obtained from Behringwerke (Marburg, Germany).

[2] 131w Every 3 to 4 days the cells were fed with 100 pl of medium containing 30 Ulml recombinant human interleukin-2 (Hoffmann-La Roche, Basel, Switzerland). After 2 to 3 weeks, the contents of each well were split between two wells. Antigen was added to one of the wells, and 2 x los irradiated autologous PBMCs were added to both wells. The cultures were scanned microscopically for antigen-specific proliferation. Specifically responding populations were selected for further expansion. The MBP-specific TCLs from HLA-DR2+ and DR2-donors recognized various epitopes (mapped with synthetic peptides), including MBP 1 to 7, 29 to 48, 80 to 86, 50 to 61, and 108 to 131 121). Only specific lines with absolute counts > 1,000 cpm and an antigen-induced stimulation index >3 were included in the analysis (see Table ).

[3] 88w Duplicate or triplicate wells containing 3 x 10' antigenspecific T cells and 2 x lo5 irradiated (5,000 rad) autologous PBMCs were incubated in the presence or absence of antigen (30 pglml human MBP or 4 pglml TI'). After 48 hours, 0.22 pCi C3Hlthymidine (Amersham, Braunschweig, Germany) was added to each well. ['HlThymidine incorporation was measured with agas scintillation counting system (Matrix TM 96, Direct Beta Counter, Packard, FrankfurtIMain, Germany) that yields approximately 20% of the counts obtained with liquid scintillation counting. The cells were harvested after 18 hours.

[4] 135w T-cell Donors MB P-speci ficl TNFA (Initials) TCLIDR-2-restricted TT-specific TCL HLA Type TNFB Alleles Alleles DR2, healthy CK IM GW H W BL ES ss CF KD MB CL EM AS HWMS H K HM SA BM CA DR2, patients Non-DR2, healthy Non-DR2, patients 31 1 11 - 414 712 111 11 - 212 414 21 - 11 -31 -31 -21 -31 -21 -11 -11 -21 -21 -112 11 - 212 61 1 11 - 11 --I2 212 21 -11 -31 --I3 21 - 31 -111 11 - -11 21 - 21 -A24,A30; B5,B44; DR2,DR12 Al,A3; B7,B8; DR2,DR3 A2,A3; B7,B13; DR2,DR7 A2,A3; B7,B44; DR2,DRb A3,A24; B7,B27; DR2,DR7 A2,A3; B18,B29; DR2,DR3 A2,A32; B7,B37; DR2,DR8 Al,A2; B7,B62; DR2,DRb A l ; B8,B62; DR3,DR4 A24,A28; B44; DRl.DR6 Al,A31; B44,BbO; DR4,DRb A 1 ; B8,Bb; DR3 A2,A28; B6O,B63; DR11,DR14

[5] 85w Antigen-specific TCLs (3 x lo4) were cultured with 2 x lo5 autologous irradiated PBMCs in the presence or absence of the relevant antigen (30 pglml MBP or 4 pgIml1T). In some experiments, 5 x lo4 HLA-transfected L cells were used as antigen-presenting cells instead of PBMCs. Culture supernatants were collected after 24 hours for TNF-a and after 48 hours for LT and IFN-y determination. The supernatants were centrifuged and stored in appropriate aliquots at -80°C until cytokine concentrations were measured by enzyme-linked immunosorbent assay (ELISA).

[6] 165w LT was measured by a sandwich ELISA technique without selecting for LT-a ( = TNF-P) and LT-p. The reagents (anti-LT monoclonal antibody [Ab 9B9], polyclonal anti-LT antiserum, and human recombinant LT) were provided by Knoll AG, Ludwigshafen, Germany. The ELISA was performed in polystyrene microtiter plates (Costar, Fernwald, Germany). Plates were coated with the monoclonal antibody diluted in coating buffer (0.1 M NaHCO,, p H 8.3). Unspecific binding was blocked overnight with 1% bovine serum albumin at 4°C. Various concentrations of recombinant human LT were added as external standards. Standards and supernatants were incubated for 1 hour. Bound LT was detected with biotinylated polyclonal anti-LT antiserum. This step was followed by incubation with streptavidin-peroxidase conjugate (Boehringer Mannheim). Pierce substrate (Rockford, IL) was used as chromogenic marker. The reaction was stopped with H 2 S 0 4 . Extinction was read in a spectrophotometer (Immunoreader NJ-2000, Nunc, Wiesbaden-Biebrich, Germany) at 405 nm evaluated according to a standard curve. The LT ELISA detects concentrations 210 pgl ml.

[1] 98w man chromosome 12 1151). Since we observed no significant association with the two major LT polymorphisms detectable with the restriction enzymes Nco I 116, 171 and AspH1 1181, nor for the TNF-a alleles determined by a point mutation in the promoter 1191, we hypothesize that the increased production of LT and TNF-a by T cells from HLA-DR2' subjects is linked to other polymorphic genes of the HLA region. Our observation that the HLA-DR2 haplotype is associated with increased production of LT and TNF-(r by T cells suggests a novel mechanism for the role of HLA-DR2 in disease susceptibility.

[2] 24w HLA typing was performed using serology for HLA-A and B alleles and oligonucleotide hybridization of polymerase chain reaction (PCR)-amplified DNA for DR alleles 1231.

[3] 53w HLA-DR2 restriction of the TCLs from HLA-DR2 + donors was determined using mouse L cells transfected with human cDNA defining the specificities DRB5"O 101 (previously referred to as DR2a) and DRBl"1501 (previously referred to as DR2b). The L-cell transfectants were kindly provided by Drs D. Alunann (Harrow, UK) and J. Trowsdale (London, UK).

[4] 64w For the TNF-a ELISA the mouse monoclonal antibody 195 (Knoll AG) was used for coating and, in biotinylated form, for detection. For the IFN-y ELISA polyclonal protein Apurified rabbit anti-human antibody (Knoll AG) was used for coating and, in biotinylated form, for detection. Extinction was read at 450 nm for TNF-a and 492 nm for IFN-y. Both assays detect concentrations of 2 10 pglml.

[5] 126w Genomic D N A was isolated from peripheral blood cells. For PCR typing of the TNFB alleles, a 740-bp fragment was amplified as previously described {17]. One aliquot was digested with the restriction enzyme Nco I allowing the restriction fragment length polymorphism (RFLP) identification of the alleles TNFB"1 (cut with Nco I) and TNFB"2 (lacking the Nco I restriction site). Another aliquot was digested with AspHI, resulting in a split of the TNFB'2 allele; ie, TNFB'2.1 represents the TNFB allele that lacks the Nco I site but encodes the AspHI recognition sequence, whereas TNFB"2.2 lacks both the Nco I and the AspHI sites (Fig 1). Analyzing a large panel of individuals we found that the AspHI site is only present in the TNFB"2 allele (unpublished observation).

[6] 84w The TNF-a promoter polymorphism was tested by singlestranded conformational polymorphism (SSCP) analysis of a 107-bp amplificate using the primers described by Wilson and colleagues [19] (5' primer: 5 -A G G C A A T A G G m G AGGGCCAT-3, and 3' primer: 5-TCCTCCCTGCTCCGA 1TCCG-3) in the following 2 0 4 reaction: 200 ng genomic DNA, 0.2 mM dNTP, 1 mM MgCI,, 50 ng of each primer, and 1 unit Taq polymerase (Promega, Madison) in standard reaction buffer. The annealing temperature was 60°C.

[7] 138w For SSCP analysis, 5 p1 of amplification reaction was mixed with 5 pl of formamide, denatured for 5 minutes at 95"C, and loaded on a 15% polyacrylamide gel/ 1 x Tris-borate-EDTA buffer (TBE). Electrophoresis was performed at 200 V overnight in 1 x TBE and the alleles were visualized by silver staining. After fixation (2 x 3 min in 10% ethanol/0.5% acetic acid), the gel was incubated for 10 minutes in 0.1% AgNO,, washed twice with H 2 0 , and developed for 20 minutes with the following solution: 0.03 gm NaBH4, 1.2 ml Zipp et al: Increased Production of LT and TNF-a by HLA-DR2+ T Cells 725 formaldehyde, and 4.5 gm NaOH in 300 ml HzO. The alleles were assigned by comparison with several homotygous reference cell lines, using the classification of Wilson and colleagues [ 191.