Lipopolysaccharide Injection Induces Relapses of Experimental Autoimmune Encephalomyelitis in Nontransgenic Mice via Bystander Activation of Autoreactive CD4⁺ Cells¹

Multiple sclerosis (MS)⁴ is a chronic inflammatory demyelinating disease of the CNS. Its natural history in most patients is characterized by a course of exacerbations and remissions. Its etiology is unknown, involving both genetic and environmental factors. Current evidence strongly suggests that MS is an immune-mediated disease (1, 2, 3). The murine model of MS, experimental autoimmune encephalomyelitis (EAE), can be induced by activation or adoptive transfer of T cells that recognize myelin Ags (4). Activation of autoreactive T cells is, therefore, believed to be important for the induction, maintenance, and regulation of the inflammatory demyelination in EAE and MS (5, 6).

Epidemiological, clinical, and experimental evidence identifies autoimmunity as a potential sequel of infection (7). Exacerbation of MS is two to three times more likely to occur during or shortly after common respiratory, gastrointestinal, or urological infections (8, 9, 10, 11). Much of the search for immunological pathways connecting infection to autoimmunity has been focused on Ag-specific immune responses. A particularly plausible hypothesis is that sequence similarity between microbial and self-Ags (molecular mimicry) activates autoreactive lymphocytes (12, 13). Supporting this hypothesis are reports of T cells that recognize both microbial and myelin peptides (14, 15, 16, 17) and of EAE induced by immunization with microbial peptides (16, 17, 18). However, epidemiological studies have failed to link induction or exacerbations of MS with any particular infection (8, 9, 10, 11, 19, 20), and T cell recognition of Ag is quite degenerate, so that an individual TCR can often recognize both microbial and self-peptides (21). Evidently, cross-reactivity between a particular microbial Ag and a particular self-Ag is unlikely, in general, to induce autoimmune disease. Additional mechanisms are needed to induce pathogenic autoimmune responses (13, 22).

For full activation, T cells require costimulatory signals from APC in addition to Ag recognition. The expression of many costimulatory molecules on APC is up-regulated in response to infection (23, 24). This is one aspect of the crucial role of the innate immune system in initiating and directing adaptive immune responses, both protective and pathological (25, 26). In this study, we show that injection of LPS, a potent activator of the innate immune system, induces TCR-independent bystander activation of autoreactive CD4⁺ Th cells which induce EAE in TCR-transgenic mice and relapse of the disease in normal mice.

Peptides MBP_Ac1–11 (AcASQKRPSQRSK) and MBP_85–99 (ENPVVHFFKNIVTPR) were synthesized as described previously (17). Lysed Salmonella typhimurium strain C5 Nalr were from U. E. Schaible (Max-Planck-Insitut für Infektionsbiologie, Berlin, Germany). S. typhimurium LPS (ATCC source strain 7823, purified by gel filtration) and OVA were obtained from Sigma-Aldrich. Some experiments were also repeated with ultrapure LPS from Salmonella abortus equi (ALX-581-009; Alexis), which is free from potentially TLR2-stimulating contaminations.

Mice transgenic for a TCR that recognizes MBP_Ac1–11/I-A^u (27) were crossed onto TCR α-chain-deficient mice, resulting in mice expressing only the transgenic TCR (T⁺α⁻ mice), which were provided by J. Lafaille (Skirball Institute, New York, NY). Mice transgenic for the same myelin basic protein (MBP)-specific TCR and deficient for RAG-1 (T⁺R⁻ mice) were from J. Demengeot (Instituto Gulbenkian de Ciência, Oeiras, Portugal). TCR expression was checked as described previously (17). Other mice were purchased from The Jackson Laboratory. Mice were kept in specific pathogen-free conditions. All animal experiments were approved by the appropriate state committees for animal welfare.

Ag in CFA was injected s.c. Ag was 200 μg peptide, 10⁸ lysed S. typhimurium, or 50 μg LPS. Pertussis toxin (PT, 200 ng; Sigma-Aldrich) was injected i.v. at days 0 and 2 after immunization. EAE was scored as follows: 0, healthy; 1, limp tail; 2, partial hind leg paralysis; 3, complete hind leg paralysis; 4, tetraparesis; and 5, moribund. Mice were sacrificed when their score reached 4- 5 and their score was kept at 5 for the remainder of the experiment.

Histological analysis was performed as described elsewhere (28). Sections were stained with H&E, Luxol Fast Blue, and Bielschowsky silver impregnation to assess inflammation, demyelination, and axonal loss, respectively. In adjacent serial sections, immunohistochemistry was performed with Abs against macrophages/activated microglia (Mac-3, clone M3/84; BD Pharmingen), T cells (CD3-12; Serotec), B cells (B220, clone RA3-6B2; BD Pharmingen), and early invading macrophages (S100A9, kindly provided by C. Sorg, Münster, Germany) (29). The S100A9 Ab also recognizes polymorphonuclear granulocytes. Bound Ab was visualized using an avidin-biotin technique with diaminobenzidine as chromogen. Control sections were incubated in the absence of primary Ab and with isotype control Abs.

Single-cell suspensions were prepared from spleens in RPMI 1640 (PAA) supplemented with 10% FCS, 2 mM glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μM 2-ME (complete medium, CM) as described (17). Spleen cells (SC) were cultured at 37°C in 5% CO₂ with peptides, LPS, staphylococcal enterotoxin B (SEB, 2 μg/ml; Sigma-Aldrich), plate-bound anti-CD3 mAb (145-2C11, 3 μg/ml), and anti-CD28 mAb (37.51, 2.5 μg/ml) or in CM alone. For Transwell experiments, CD4⁺ and CD4⁻ cells were purified by MACS (Miltenyi Biotec). Purity of the separated fractions was >97% in all experiments. Transwells (0.4-μm pores; Costar) were used to separate the two populations. CD4⁺ cells were in the upper chamber. The lower chamber contained CD4⁻ APC. Cells were cultured with MBP, LPS, or CM alone for 24 h before cytokine production was analyzed flow cytometrically. SC were incubated for 1 h in CM with DMSO (0.1%) or with 50 nM cyclosporin A (CsA; Arzneimittelwerk Dresden) in DMSO before MBP or LPS was added for 24 h. For blocking experiments, SC were cultured for 24 h in the presence of LPS with or without the addition of 20 μg/ml anti-CD18 (M18/2.a.12.7), anti-CD54 (BE29G1), anti-CD80 (16-10A1), anti-CD86 (GL1), anti-ICOS-L (MIL-4495, kindly provided by R. Kroczek, Berlin, Germany), or CTLA-4-Ig 1 h before cytokine production was determined flow cytometrically.

Cells were stained with allophycocyanin, Cy5, or FITC-conjugated mAbs against CD4 (GK1.5), CD25 (2E4; BD Pharmingen), and biotinylated CD69 (H1.2F3; BD Pharmingen). Biotinylated primary mAbs were detected with streptavidin coupled to PE (BD Pharmingen). Samples were incubated with blocking anti-FcγRII/III mAb 2.4G2/75 and purified rat IgG. For analysis of proliferation, SC were incubated with CFDA-SE (Molecular Probes) at 5 μg/ml for 5 min, cultured with CM alone or Ags for 3 days, stained with Cy5-labeled anti-CD4, PE-labeled anti-B220, and analyzed by flow cytometry. For analysis of cytokine production, SC were cultured for 18 h in vitro before 5 μg/ml brefeldin A (Sigma-Aldrich) was added for the last 6 h of culture. Samples were analyzed on a FACSCalibur and analysis was performed with CellQuest software (BD Biosciences) or FCS express (De Novo software) as described previously (30).

Statistical significance was determined by a two-sided Fisher exact test.

We had shown earlier that a TCR that recognizes MBP_Ac1–11/I-A^u also recognizes a large number of microbial peptides. One of these peptides was derived from S. typhimurium and immunization with this peptide induced EAE in mice whose T cells exclusively express that receptor (T⁺α⁻ mice) (17). Immunization of T⁺α⁻ mice with lysed S. typhimurium/CFA followed by PT induced EAE with the same incidence, similar kinetics, and severity as immunization with the MBP_Ac1–11 peptide (Fig. 1 and Table I). However, presentation and recognition of the cross-reactive peptide was not necessary. Immunization with LPS from S. typhimurium also induced EAE in the T⁺α⁻ mice (Fig. 1 and Table I). Different from another transgenic model (31), the T⁺α⁻ mice did not develop EAE when immunized with PBS/CFA followed by PT (Fig. 1 and Table I). LPS-induced EAE was T cell dependent because LPS immunization did not induce EAE in littermates of the T⁺α⁻ mice that do not express the MBP_Ac1–11-specific transgenic TCR (α⁻ mice) (Table I).

No difference in the extent of inflammation and demyelination was detectable on H&E- and Luxol Fast Blue-stained sections between MBP_Ac1–11- or LPS-induced EAE. Furthermore, immunohistochemistry for T cells, B cells, and macrophages revealed no discernible differences in the composition of the inflammatory infiltrates (Fig. 2).

Only the T⁺α⁻ mice that harbor a quasi-monoclonal T cell repertoire developed EAE upon LPS immunization. Naive nontransgenic SJL, PL/J, or B10.PL mice were resistant (Table I). To determine whether previous activation and expansion of MBP-specific Th cells made genetically unaltered SJL mice susceptible to LPS-induced EAE, we immunized SJL mice with MBP_85–99. This induced monophasic EAE in 65% of the SJL mice ∼2 wk after the immunization. Almost all of the mice had recovered 40 days after the immunization (Table II). On day 45, we injected the mice with LPS or PBS followed by PT. Only 2 of the 20 PBS-injected mice suffered a relapse of EAE (Table II) and the clinicalsymptoms were mild (score 2). In contrast, immunization with LPS induced a relapse in 13 of 28 mice (p < 0.02, Table II). Similar results were obtained when we injected the mice with ultrapure LPS (data not shown), demonstrating that the LPS effect was not due to contaminating TLR2 ligands. The mean clinical score (2.5) was higher than during the first episode of EAE and some of the mice had severe symptoms (score 4). Ten of 17 mice that had developed EAE upon MBP immunization on day 0 developed an EAE relapse following the LPS immunization on day 45. Similarly, 5 of 11 mice of the mice that had not developed EAE upon MBP immunization on day 0 developed EAE following the LPS immunization (p = 0.7). Therefore, a previous EAE episode is not necessary to make SJL mice susceptible to LPS-induced EAE, rather the T cell activation induced by MBP immunization is both necessary and sufficient. In contrast to the MBP-immunized mice, none of the SJL mice that were immunized with OVA on day 0 and with LPS on day 45 developed EAE (Table II). Together, these results indicate that MBP-specific effector/memory T cells, but not naive T cells or effector/memory T cells that recognize irrelevant, non-CNS Ags such as OVA, mediate bystander-induced EAE.

MBP_85–99-immunized SJL mice had perivascular and subpial inflammatory infiltrates of T cells, macrophages, B cells, and granulocytes (Fig. 3, a–e). Mice injected with PBS at day 40 and without clinical relapse showed subpial scars at day 75, but no evidence of recent inflammation (Fig. 3, f–k), whereas mice immunized with LPS and with clinical signs of relapsing disease had extensive, highly cellular subpial lesions with polymorphonuclear granulocytes and macrophages expressing the early macrophage marker S100A9 as evidence of recent invasion of inflammatory cells into the lesion (Fig. 3, l–p).

LPS is not known to have direct effects on murine Th cells but potently induces the expression of cytokines and costimulatory molecules in cells of the innate immune system. To investigate the immunological mechanisms leading to LPS-induced EAE, we examined whether LPS-induced activation of the innate immune cells caused Th cell activation in vitro. Therefore, we cultured SC from T⁺α⁻ mice in CM alone or with MBP_Ac1–11 or LPS. LPS induced proliferation of 1.0–1.5% of the CD4⁺ T cells (Fig. 4,a). This was significantly above background proliferation (CM alone: 0.2–0.4%, p < 0.015 in five independent experiments). LPS-induced T cell proliferation depended on the presence of CD4⁻ cells and was not detectable when CD4⁻ cells were depleted from the cultures (our unpublished data). LPS also induced up-regulation of both CD69 and CD25 in T⁺α⁻ mice: >20% of the CD4⁺ T cells were CD69⁺ and ∼2% expressed CD25 after a 24-h culture in the presence of LPS (Fig. 4,b). Similar results were obtained when we used ultrapure LPS, which does not contain TLR2-stimulating contaminants (Fig. 4 b).

Effector/memory cells from previously immunized mice were more susceptible to LPS-induced bystander activation than Th cells from unimmunized mice. When SC from normal SJL mice that had been immunized with MBP_85–99 45 days earlier (henceforth MBP_85–99-immunized mice) were cultured with LPS, ∼1.5% of them produced IFN-γ or TNF-α (Fig. 4,c). The percentages of cytokine-producing Th cells were similar when SC from SJL mice that had been preimmunized with OVA were cultured with LPS in vitro (our unpublished observations). In contrast, <0.5% of the CD3⁺CD4⁺ cells from naive healthy SJL mice produced IFN-γ or TNF-α in response to LPS (Fig. 4,c). SJL mice that had been immunized with MBP/CFA s.c. followed by PT i.v. responded similarly to SJL mice that had been immunized with MBP/CFA s.c. followed by PBS i.v. Thus, previous in vivo exposure to PT was not necessary for susceptibility to LPS-induced bystander activation (Fig. 4 d).

Taken together, effector/memory Th cells of different Ag specificities are susceptible to LPS-induced bystander activation. However, only autoreactive effector/memory Th cells cause disease upon LPS-induced bystander activation (Table II).

Unsorted SC produced IFN-γ and TNF-α in response to LPS but not in CM alone (Fig. 5,a). We purified CD4⁺ cells from MBP_85–99-immunized SJL mice by MACS (purity >98%). When the purified CD4⁺ cells were cultured along with CD4⁻ cells and LPS, the fraction of cytokine-producing CD4⁺ cells was the same as in total SC (Fig. 5,a). In contrast, purified CD4⁺ cells alone did not respond to LPS (Fig. 5 a). Thus, LPS has no direct effect on Th cells. Instead, LPS-induced bystander activation of Th cells depends on the activation of CD4⁻ cells by LPS.

LPS-induced bystander activation of Th cells could be mediated either by soluble factors produced by the LPS-responsive CD4⁻ cells or by cell-cell contact between the LPS-responsive cells and the Th cells. To distinguish between these possibilities, we placed MACS-purified CD4⁺ cells from MBP_85–99-immunized SJL mice in the upper chamber of a Transwell system. The lower chamber contained CM alone, CM and LPS, or LPS and CD4⁻ cells. When CD4⁺ cells were cultured in the upper chamber and APC in the lower chamber, the CD4⁺ cells no longer produced IFN-γ or TNF-α in response to LPS (Fig. 5 a). The sorted CD4⁺ cells in the upper chamber were viable as demonstrated by the fact that they produced both IFN-γ and TNF-α upon stimulation with anti-CD3/anti-CD28 (our unpublished data). Supernatants from LPS-stimulated SC did not induce IFN-γ- or TNF-α production in the CD4⁺ cells (our unpublished data). Furthermore, addition of the selective p38 MAPK inhibitor SB203580, which inhibits IL-12- and IL-18-induced IFN-γ production of murine Th1 cells (32), had no effect on LPS-induced cytokine production in the CD4⁺ effector/memory cells (our unpublished data). Taken together, LPS-induced bystander activation of CD4⁺ cells depends on physical contact of Th cells with LPS-responsive APC.

Since LPS-induced bystander activation of CD4⁺ cells is contact dependent, we asked whether TCR-mediated signaling was necessary. CsA inhibits TCR-induced expression of IFN-γ (32). We cultured SC from naive or MBP_85–99-immunized SJL mice with SEB, MBP_85–99, or LPS in the presence or absence of CsA and determined the cytokine production of CD4⁺ cells. CsA inhibited the IFN-γ production in response to either SEB or MBP_85–99 almost completely (Fig. 5,b). In contrast, CsA did not inhibit the IFN-γ or TNF-α production of naive Th cells in response to LPS (data not shown) and only marginally inhibited the LPS-induced IFN-γ production of effector/memory Th cells (Fig. 5,b). Thus, LPS-induced bystander activation of Th cells proceeds via signaling pathways distinct from the TCR-mediated triggering of calcineurin. The APC used in these experiments could have still been loaded with Ag. To test this possibility, we performed experiments in which APC (CD4⁻CD8⁻ SC) from either naive or immunized animals were cultured with T cells from either naive or immunized animals. When Th cells from unimmunized animals (Th_n) were cultured along with APC from either unimmunized (APC_n) or immunized (APC_i) animals, there was no detectable IFN-γ production in response to either medium or MBP_85–99 and little IFN-γ production in response to LPS (Fig. 5,c, left panels). Only Th cells from immunized animals (Th_i) produced IFN-γ in response to MBP_85–99 and that response was stronger when both the Th and the APC came from immunized mice (Fig. 5,c, right panels). Supporting the data shown in Fig. 4 c, Th cells from immunized animals produced more IFN-γ in response to LPS than Th cells from unimmunized animals (Th_n) did. Interestingly, the response was slightly stronger when both the Th and the APC were from previously immunized mice. Yet, the pertinent point is that a significant percentage of Th cells from immunized animals produced IFN-γ when cultured with APC from unimmunized animals and LPS.

We next asked whether the costimulatory molecules were essential for LPS-induced bystander activation. SC from MBP_85–99-immunized SJL mice were cultured in CM alone or with LPS in the presence or absence of different mAbs or combinations thereof. Blockade of the ICAM-1:LFA-1 interaction with anti-CD54 reduced the LPS-induced IFN-γ-production by ∼25% (Fig. 6,b). Neither anti-CD80 nor anti-CD86, anti-ICOSL, or CTLA4-Ig alone influenced the LPS-induced cytokine production strongly (Fig. 6), but the combination of these reagents reduced the number of cytokine-producing CD4⁺ cells by ∼50% (anti-B7 family, Fig. 6). Addition of anti-CD54 to that combination only marginally reduced the IFN-γ- production further. Taken together, LPS-induced bystander activation of Th cells is at least partly mediated by the contact between costimulatory ligands of the B7 family on APC, some of which are up-regulated in response to LPS, and their receptors on CD4⁺ Th cells.

In this report, we show that LPS injection induces EAE relapses in genetically unaltered mice and bystander activation of Th cells. Previous reports had either demonstrated adjuvant effects of TLR ligands, such as LPS in vitro (33, 34, 35, 36, 37), or the induction of autoimmune disease through systemic injection of PT or CpG in transgenic mice (31, 38). Addition of LPS to the in vitro culture enhances the encephalitogenic potential of MBP-specific T cells upon adoptive transfer into syngeneic recipients (33, 35). In otherwise EAE-resistant B10.S mice, EAE can be induced passively if CpG or LPS plus IFN-γ are added to the in vitro culture before adoptive transfer (34). Moreover, if lymph node cells from mice or rats that had been tolerized toward myelin Ags are cultured in the presence of CpG in vitro, they become encephalitogenic upon adoptive transfer into syngeneic recipients (36, 37). Taken together, myelin-specific cells that are unable to adoptively transfer EAE if cultured with myelin Ags alone can be rendered encephalitogenic through in vitro culture with myelin Ags in the presence of CpG or LPS. Reports on the active induction of EAE by injection of Ag-nonspecific stimuli in vivo have thus far been restricted to transgenic mouse models. EAE can be induced by injection of PT alone in some, but not all, strains of mice transgenic for a MBP-specific TCR (31), and CpG injection induces EAE in a fraction of B10.S mice that express a proteolipid protein-specific TCR (38). Our finding that LPS injection induces EAE in T⁺α⁻ mice (Fig. 1) confirms and extends these reports. The important novel finding presented in this report is the active induction of EAE relapses in normal mice upon LPS injection without the need for adoptive transfer of in vitro-cultured myelin-specific cells.

Immunization with PBS/CFA followed by PT i.v. did not induce EAE in T⁺α⁻ mice or EAE relapses in SJL mice. Thus, LPS is necessary in both models and EAE is not due to the injection of PT, which has several known EAE-promoting effects, such as increasing Th1 responses and enhancing the T cells’ access to the CNS (39, 40, 41)

Bystander activation has been used to describe different phenomena. We use this term to indicate T cell activation that occurs independently of Ag recognition by the TCR. In accordance with earlier studies (42, 43, 44, 45, 46), we found that memory/effector T cells were more susceptible to TCR-independent bystander activation than naive Th cells. However, LPS-induced bystander activation of CD4⁺ cells described here differs from the cytokine-driven bystander activation depicted in the earlier studies (32, 42, 43, 44, 45, 46, 47, 48, 49) in several important aspects. Most important, soluble factors such as cytokines do not mediate LPS-induced bystander activation. This is also different from the in vitro adjuvanticity of the TLR ligands LPS or CpG, which were observed in adoptive transfer models of EAE. Those effects could be mimicked by addition of IL-12 to the culture (34, 36, 37). In contrast to a recent study that reported the detection of TLR4 mRNA in MACS-purified CD4⁺CD25⁺ T cells (50), we did not find any direct LPS effects on highly purified CD4⁺ T cells.

Instead, physical contact between LPS-responsive CD4⁻ cells and CD4⁺ Th cells is necessary for LPS-induced bystander activation of Th cells. The interaction between costimulatory molecules of the B7 family on LPS-activated CD4⁻ cells and their receptors on Th cells is an important but not the only mechanism for LPS-induced bystander activation of Th cells. Blockade of the costimulatory ICAM-1:LFA-1 interaction (51) also reduced the LPS-induced cytokine production significantly (see Fig. 6). Furthermore, enhanced costimulation via members of the TNFR family (24) may well play a role in LPS-induced bystander activation of Th cells. Taken together, our data show that under certain circumstances costimulatory signals provided by activated APC can induce T cell activation in the absence of TCR triggering. Similar observations have been made with “superagonistic” Abs against CD28 in rats (52, 53) and with pairs of Abs against human CD2 (54). The question remains, which regulatory processes usually prevent T cell activation by costimulatory signals alone and under which circumstances in vivo costimulatory signals alone suffice to cause T cell activation. It is conceivable that bystander activation in vivo may have beneficial effects. For example, bystander activation might contribute to the maintenance of T cell memory similar to what has been described for B lymphocytes (55).

Bystander activation of autoreactive Th cells occurs independently of TCR Ag recognition and fits the clinical and epidemiological data on the connection between infection and autoimmunity. A variety of infections increase the risk for exacerbations in MS patients (8, 9, 10, 11), but despite extensive efforts no specific pathogen has been identified as culprit (2, 3, 7, 19). Bystander activation of autoreactive Th cells occurs independently of TCR Ag recognition and fits the clinical and epidemiological data on the connection between infection and autoimmunity better than supposed Ag-specific mechanisms. The following scenario could link infection and autoimmunity via bystander activation: autoreactive T cells are part of the normal repertoire (56, 57, 58). The frequency of these autoreactive Th cells is one of the genetically determined factors that contribute to susceptibility to autoimmune diseases (59). Survival and expansion of autoreactive Th cells can be supported either by recognition of self-Ag and overt autoimmune attacks or, clinically silent, by the recognition of cross-reactive microbial peptides (18, 21). Once the number of autoreactive T cells has reached a certain threshold, as in the monoclonal T⁺α⁻ mice or following MBP immunization in the nontransgenic mice, or in patients who have already suffered previous episodes of MS, TCR-independent stimuli such as LPS-induced bystander activation can trigger sufficient numbers of autoreactive T cells to cause autoimmune damage. This scenario is different both from the Ag-dependent adjuvant effects of LPS (60, 61, 62, 63) and from bystander damage which can occur in virally infected mice with a monoclonal T cell repertoire (64, 65, 66). Both the Ag-specific activation of the monoclonal T cells and the simultaneous viral infection at the site of tissue damage are necessary conditions for the induction of bystander damage which can result in diabetes (64), keratitis (65), or encephalitis (66). In sharp contrast, TCR-mediated signals are not required for the LPS-induced bystander activation of Th cells described here.

Only 50% of the LPS-immunized SJL mice had EAE relapses. Similarly, <10% of the infectious episodes in the clinical studies were associated with MS exacerbations (8, 9, 10, 11). One explanation is that immunoregulatory mechanisms prevent clinically overt autoimmunity. Alternatively, the number of autoreactive Th cells, which are bystander activated, may be too small to cause damage, and finally some infectious agents may lack the potential to induce bystander activation. That only about one-quarter of MS exacerbations are associated with clinically apparent infections (8, 9, 10, 11) may be due to the fact that some infections are clinically inapparent. Alternatively, additional mechanisms, which are not triggered by infections, could cause exacerbations. The latter possibility is supported by pathological evidence suggesting that there are four distinguishable subgroups of MS (67), each of which may have different immunopathological mechanisms and different susceptibility for infection-induced immunopathology. Even so, aggressive therapy and prophylaxis of infections to prevent bystander activation of autoreactive T cells could be a useful approach in comprehensive treatment of MS patients.

Exacerbations of MS are associated with many different infections, including Gram-positive bacteria and viruses that do not possess LPS (8, 9, 10, 11). We found that bacterial lipoproteins, which are expressed by Gram-positive bacteria, can induce EAE relapses in SJL mice with similar incidence and severity as LPS (V.S. and T.K., unpublished data), and it remains to be established whether other pathogen-associated molecular patterns such as dsRNA are able to induce bystander activation of autoreactive Th cells.

In summary, LPS induces the proliferation and cytokine production of Th cells independently of TCR signaling. This bystander activation of Th cells is not mediated by soluble factors and requires the physical contact between LPS-responsive CD4⁻ cells and Th cells. LPS immunization induces exacerbations of EAE in genetically unaltered normal mice. Bystander activation of autoreactive Th cells is an Ag-independent mechanism that fits the clinical and epidemiological data on the connection between infection and autoimmunity better than Ag-specific models.

We thank Ulrich Schaible for lysed S. typhimurium, Andreas Hutloff, Richard Kroczek, and Clemens Sorg for mAbs, Juan Lafaille and Jocelyne Demengeot for mice, Joachim Listing for statistical analyses, and N. Avrion Mitchison for critical reading of this manuscript.

The authors have no financial conflict of interest.