The Journal of Immunology

Viral Delivery of an from Haemophilus influenzae Induces Autoimmune Disease by Molecular Mimicry1

J. Ludovic Croxford, Holly A. Anger, and Stephen D. Miller2

Multiple sclerosis (MS) is an autoimmune CNS demyelinating disease in which infection may be an important initiating factor. Pathogen-induced cross-activation of autoimmune T cells may occur by molecular mimicry. Infection with wild-type Theiler’s murine encephalomyelitis induces a late-onset, progressive -mediated demyelinating disease, similar to MS. To deter- mine the potential of virus-induced by molecular mimicry, a nonpathogenic neurotropic Theiler’s murine enceph- alomyelitis virus variant was engineered to encode a mimic peptide from protease IV of Haemophilus influenzae (HI), sharing 6 of 13 aa with the dominant encephalitogenic proteolipid (PLP) epitope PLP139–151. Infection of SJL mice with the HI mimic-expressing virus induced a rapid-onset, nonprogressive paralytic disease characterized by potent activation of self-reactive ؉ PLP139–151-specific CD4 Th1 responses. In contrast, mice immunized with the HI mimic-peptide in CFA did not develop disease, ؉ associated with the failure to induce activation of PLP139–151-specific CD4 Th1 cells. However, preinfection with the mimic- expressing virus before mimic-peptide immunization led to severe disease. Therefore, infection with a mimic-expressing virus directly initiates organ-specific T cell-mediated autoimmunity, suggesting that pathogen-delivered innate immune signals may play a crucial role in triggering differentiation of pathogenic self-reactive responses. These results have important implications for explaining the pathogenesis of MS and other autoimmune diseases. The Journal of Immunology, 2005, 174: 907–917.

t present, the mechanism(s) of initiation of multiple scle- mimicry between MBP96–102, a candidate autoantigen for MS, and rosis (MS),3 a CNS demyelinating disease considered to the U24 protein of human herpesvirus-6 (HHV-6; residues 4–10), be mediated by -specific autoreactive CD4ϩ T a viral agent that may be associated with MS (11). A significant A ϩ cells, is unknown. Evidence from clinical and epidemiological number of CD4 T cells from MS patients could recognize either studies suggests that environmental factors, such as , may MBP93–105 or a synthetic HHV-6 peptide compared with cells play an important role in the etiology of MS (1). Other CNS de- from normal healthy controls. Importantly, cross-activation by the myelinating diseases, both in humans and animals, are known to be HHV-6 peptide induced Th1 differentiation of the autoreactive T associated with viral infections (2–5). There is substantial evidence cells (11). that MS is an autoimmune disease, although normal, healthy in- Theiler’s murine encephalomyelitis virus (TMEV)-induced de- dividuals also possess peripheral T cells specific for the myelinating disease is a mouse model of MS mediated by CD4ϩ within the myelin Ags, myelin basic protein (MBP), myelin oli- cells, and characterized by a chronic-progressive paralytic course godendrocyte glycoprotein, and proteolipid protein (PLP) (6–8). in SJL/J mice (12, 13). TMEV is a natural neurotropic mouse However, the mechanism(s) underlying the activation of myelin- pathogen and intracerebral (i.c.) infection of mice with TMEV specific autoreactive T cells is unknown. One putative mechanism induces an initial virus-specific Th1 response that initiates by- for initiation of autoimmune demyelinating disease is molecular stander myelin destruction, with subsequent activation of self-re- mimicry, whereby autoreactive T cells are activated by epitopes active myelin epitope-specific CD4ϩ Th1 cells via epitope spread- from infectious agents that share structural or sequence homology ing (14–16). to self Ags (9, 10). A recent study has demonstrated molecular Although epitope spreading may explain how a persistent virus infection can lead to myelin-specific autoimmunity, the major pos- Department of Microbiology-Immunology, and Interdepartmental Immunobiology tulated mechanism for initiation of viral-induced autoimmunity is Center, Northwestern University Medical School, Chicago, IL 60611 molecular mimicry, although there is currently little direct evi- Received for publication March 16, 2004. Accepted for publication November dence to support this. Previous studies have used mimic sequences 8, 2004. from infectious pathogens, to modulate experimental autoimmune The costs of publication of this article were defrayed in part by the payment of page encephalomyelitis (EAE), or administered T cell lines specific for charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. mimic epitopes to induce EAE (17–20). To provide more compel- 1 This work was supported in part by U.S. Public Health Service Grants NS-40460 ling evidence for the initiation of CNS autoimmunity via molec- and NS-23349. J.L.C. is a fellow of the National Multiple Sclerosis Society (Post- ular mimicry, we asked whether CNS disease could be initiated by doctoral Research Fellowship Award FG-1456-A-1). infecting SJL mice with a neurotropic virus expressing a peptide 2 Address correspondence and reprint requests to Dr. Stephen D. Miller, Department mimic of the immunodominant PLP epitope (PLP139–151) derived of Microbiology-Immunology, Northwestern University Feinberg School of Medi- cine, 303 East Chicago Avenue, Chicago, IL 60611. E-mail address: s-d- from an infectious pathogen. [email protected] We demonstrate here a defined model of molecular mimicry, 3 Abbreviations used in this paper: MS, multiple sclerosis; MBP, myelin basic pro- wherein infection of SJL mice with a nonpathogenic variant of the tein; PLP, proteolipid protein; HHV-6, human herpesvirus-6; TMEV, Theiler’s mu- wild-type (WT)-TMEV (BeAn strain) expressing a mimic se- rine encephalomyelitis virus; EAE, experimental autoimmune encephalomyelitis; WT, wild type; HI, Haemophilus influenzae; i.c., intracerebral; DTH, delayed-type quence (Haemophilus influenzae (HI)574–586) derived from the hypersensitivity. protease IV protein of HI, a natural bacterial pathogen of mice,

Copyright © 2005 by The American Association of Immunologists, Inc. 0022-1767/05/$02.00 908 MOLECULAR MIMICRY-INDUCED CNS AUTOIMMUNE DISEASE leads to early-onset demyelinating disease (21). This early-onset tutes of Health-approved Northwestern University Medical School animal clinical disease was associated with activation, proliferation, and facilities. All protocols were approved by the Northwestern University An- differentiation of PLP -specific autoreactive Th1 T cells. imal Care and Use Committee. Paralyzed mice were afforded easier access 139–151 to food and water. Furthermore, these data highlight the importance of studying mo- lecular mimicry in the context of an infectious pathogen, because Construction of the mimic-expressing virus immunization with the HI peptide emulsified in CFA failed 574–586 The cDNA encoding the BeAn strain of TMEV was modified by inserting to induce clinical disease. However, immunization of mice prein- ClaI sites at bp 1137 (Fig. 1A). This resulted in a 23-aa deletion in the fected with HI-BeAn with HI574–586/CFA induced a significantly leader sequence (L) of the BeAn genome. This virus was designated ⌬Cla- more severe clinical disease. This suggests that severe clinical dis- BeAn as described previously (21). Briefly, ClaI sites were introduced by ease, due to HI-BeAn-induced molecular mimicry, requires both PCR to the PLP cDNA at either end of a 30-aa sequence PLP130–159, which encompassed the immunodominant encephalitogenic PLP sequence the presence of the mimic sequence and the persistent stimulation 139–151 (Fig. 1B). A sequence from serine protease IV (HI ), naturally ex- of innate by the pathogen. This model of molec- 574–586 pressed in HI, which shares 6 of 13 aa with PLP139–151 (Fig. 1B), was ular mimicry will allow potential mimic epitopes from other in- constructed by PCR mutagenesis of the PLP139–151 sequence to introduce fectious pathogens to be tested for their ability to induce autoim- substitutions at positions 139 (H3E), 140 (C3Q), 142 munity, and has important implications for the etiology and (G3V), 147 (H3L), 149 (D3A), 150 (K3P), and 151 (F3I). Follow- ing an enzyme restriction cut with ClaI, the 30-aa piece containing the pathogenesis of MS and other autoimmune diseases. HI574–586 sequence flanked by the original PLP sequences was inserted into the ClaI site in the ⌬Cla-BeAn virus cDNA. This was designated HI-BeAn Materials and Methods cDNA (Fig. 1A). As a negative control, an OVA sequence, OVA317–346, Mice encompassing the OVA323–339 epitope, with no homology to the PLP139–151 or ⌬ HI574–586 sequence (Fig. 1B), was inserted into the Cla-BeAn parental virus Five- to 6-wk-old female SJL mice were obtained from Harlan Sprague to yield OVA-BeAn (data not shown). Viral RNA was produced from the Dawley. Mice were housed under barrier conditions at the National Insti- cDNA through the T7 promoter and transfected into BHK-21 cells, resulting

FIGURE 1. The construction of a PLP139–151 mimic expressing virus. A,AClaI insertion in the leader sequence (L) of cDNA encoding the TMEV BeAn strain 8386 resulted in a 23-aa deletion in the leader sequence of the parent virus (⌬Cla-BeAn). ClaI sites were in- troduced by PCR into the PLP cDNA at ei- ther end of a 30-aa sequence PLP130–159, which encompassed the immunodominant encephalitogenic PLP139–151 sequence. The PLP139–151 mimic sequence from the pro- tease IV protein of HI (HI574–586) was con- structed by PCR amino acid substitutions of the PLP139–151 sequence. The 30-aa sequence containing the HI574–586 sequence flanked by the PLP sequences was inserted into the ClaI site in the ⌬Cla-BeAn virus cDNA. This was designated HI-BeAn cDNA. Viral RNA was produced from the cDNA through the T7 pro- moter and transfected into BHK-21 cells re- sulting in the production of infectious virus. B, Peptide sequences of myelin and myelin mimic peptides. The mimic HI574–586 epitope shares 6 of 13 aa with the native PLP139–151 epitope (in bold type) including the primary (L145) and secondary (P148) I-As binding residues and the primary (W144) TCR con- s tact residue. I-A -restricted VP270–86 and OVA323–339 peptides used as negative con- trols do not share sequences with PLP or HI peptides. The Journal of Immunology 909 in the production of infectious virus by cells as described previously (21). Viral groups were used to quantify the numbers of positive inflammatory cells titers were measured by viral plaque assay (22). Sequencing of the HI-BeAn per area of each tissue section. Data from photomicrographs were stored as cDNA confirmed that the HI sequence was correct. 8-bit binary images in grayscale format. Quantification was determined using ImageJ software, version 1.32j (͗http://rsb.info.nih.gov/ij/͘). Thresh- Infection of SJL mice with TMEV old values of brightness and contrast were determined for the photomicro- SJL mice (n ϭ 5–8 per group) were infected by i.c. injection of 3 ϫ 107 graphs and were kept constant between each sample. Before analysis, the PFU of either WT-TMEV (BeAn 8386 strain), HI-BeAn, OVA-BeAn, or minimum and maximum size of pixels to be counted was determined ⌬Cla-BeAn, and scored at daily intervals on a clinical scale of 0–5: 0, no whereby sections with no positive staining gave a measurement of 0.0 signs of disease; 1, mild gait abnormalities; 2, severe gait abnormalities; 3, pixels. Data are presented as percentage of positive pixels per area of the paralysis in one limb; 4, more than one paralyzed limb; 5, moribund. photomicrograph.

Induction of active EAE Statistics For actively induced relapsing-EAE, mice (n ϭ 5) were immunized s.c. Clinical severity results were presented as the mean group clinical score, with 100 ␮l of a CFA emulsion containing 400 ␮gofMycobacterium and the statistical difference calculated by the Mann-Whitney nonparamet- tuberculosis H37Ra (Difco) and 100 ␮gofPLP distributed over 139–151 ric ranking test. Analysis of DTH responses and IFN-␥ ELISA were per- three sites on the lateral hind flanks and dorsally. For HI or OVA 574–586 323– formed using the two-tailed Student t test. 339 peptide priming, the same protocol was used with 100 ␮g of peptide per animal. Clinical scores were assessed on a 0–5 scale as follows: 1, lack of tail tone; 2, limp tail and hindlimb weakness; 3, partial hindlimb paralysis; Results 4, total hindlimb paralysis; and 5, moribund. Immunization of mice with HI574–586 induces PLP139–151 T cell Peptides cross-reactivity but not clinical disease PLP (HSLGKWLGHPDKF), the TMEV capsid peptide VP2 139–151 70–86 Following the immunization of mice with PLP139–151 in CFA, (WTTSQEAFSHIRIPLPH), OVA peptide (ISQAVHAAHAEINE 323–339 100% of mice exhibited a typical acute-phase disease course of AGR), and the HI peptide HI574–586 (EQLVKWLGLPAPI) (Fig. 1B) were purchased from Peptides International. The amino acid composition was EAE. In contrast, mice immunized with the PLP139–151 mimic verified by mass spectrometry, and purity was assessed by HPLC. Both peptide, HI574–586, did not exhibit clinical disease (Fig. 2A) even VP270–86 and OVA323–339 induce immune responses (delayed-type hyper- if primed twice with peptide/CFA (days 0 and ϩ7 postinfection sensitivity (DTH), proliferation, and IFN-␥ secretion) in SJL mice either (p.i.)) and treated with pertussis toxin (200 ng/day, days 0 and ϩ2 following infection with WT-TMEV or when immunized in CFA, respec- tively (21) (data not shown). p.i.) (data not shown) (19). The ability of HI574–586 immunization to induce the cross-activation of PLP139–151-specific T cells was DTH response measured by T cell proliferation, DTH, and IFN-␥ secretion. Neg- DTH responses were elicited by injecting mice s.c. with 10 ␮gofthe ative controls consisted of naive mice rechallenged with either challenge peptides, PLP139–151 or VP270–86, into alternate ears following HI574–586 or PLP139–151 peptides, or HI574–586-orPLP139–151- measurement of ear thickness using a Mitutoyo model 7326 engineer’s immunized mice rechallenged with PBS or an irrelevant TMEV micrometer (Schlesinger’s Tools). Twenty-four hours following peptide challenge, the ears were remeasured and differences in ear swelling over capsid peptide, VP270–86 (Fig. 3). Ϫ4 prechallenge thickness were expressed in units of 10 inches Ϯ SEM. Rechallenge with either HI574–586 or PLP139–151 peptide in mice immunized against PLP , induced significant DTH re- T cell proliferation and analysis 139–151 sponses (in vivo peptide rechallenge), increased T cell prolifera- Spleens were removed from infected mice (n ϭ 2) at various times fol- tion, and significant IFN-␥ secretion (in vitro peptide rechallenge) p Ͻ 0.05) (Fig. 3). Mice ,ء) lowing infection. T cell proliferation and cytokine analysis were performed when compared with naive mice as described previously (23). Proliferation was determined from triplicate wells for each peptide concentration added in vitro and then expressed as immunized with HI574–586 exhibited significant levels of T cell ⌬counts per minute. For IFN-␥ cytokine analysis, a duplicate set of pro- proliferation when rechallenged with either HI574–586 or PLP139–151 liferation wells were used to collect supernatants at 48 and 72 h, and cy- compared with naive mice (Fig. 3A). However, mice immunized to tokine concentrations were determined by ELISA (Endogen Minikits). HI574–586, in contrast to PLP139–151-immunized mice, responded only Ͻ ء Immunohistochemistry and quantitation of cellular infiltrates to DTH rechallenge with HI574–586 ( , p 0.05), but not PLP139–151 (Fig. 3B). In addition, HI574–586-immunized mice only secreted sig- Ͻ ء ␥ Five to 8 mice per experimental group were anesthetized and perfused with 1ϫ PBS on the indicated days. Spinal cords and brains were removed by nificant quantities of IFN- in response to HI574–586 ( , p 0.05), but dissection, and multiple 2- to 3-mm spinal cord blocks were immediately not PLP139–151 (Fig. 3C). Rechallenge with VP270–86 did not induce ␥ frozen in OCT (Miles Laboratories) in liquid nitrogen. The blocks were either T cell proliferation or IFN- secretion in either HI574–586-or stored at Ϫ80°C in plastic bags to prevent dehydration. Five-micrometer- PLP139–151-immunized groups, and was comparable with PBS and thick cross-sections from the lumbar and thoracic region of the spinal cord, naive control groups (Fig. 3). or longitudinal sections of the brain, were cut on a Reichert-Jung Cryocut CM1850 cryotome (Leica), mounted on Superfrost Plus electrostatically By day 14 post-PLP139–151 immunization, mice exhibit clinical charged slides (Fisher), air dried, and stored at Ϫ80°C. Slides were stained signs of disease, and the presence of inflammatory infiltrate can be using a Tyramide Signal Amplification Direct kit (NEN) according to man- identified in the CNS of these mice (24). Immunohistochemistry of ufacturer’s instructions. Sections from each group were thawed, air-dried, the spinal cord and brain, taken from mice 14 days following im- fixed in 2% paraformaldehyde at room temperature, and rehydrated in 1ϫ ␥ munization against either HI574–586 or OVA323–339, were analyzed PBS. Nonspecific staining was blocked using anti-CD16/CD32, (Fc III/ ϩ ϩ ϩ IIR, 2.4G2; BD Pharmingen), and an avidin/biotin blocking kit (Vector for the presence of CD4 and CD8 T cells, B220 B cells, or ϩ Laboratories) in addition to the blocking reagent provided by the Tyramide F4/80 monocytes/ (Fig. 4). Although mice primed Signal Amplification kit. Tissues were stained with biotin-conjugated Abs with HI574–586 exhibited cross-reactive T cells to PLP139–151 as anti-mouse CD4, anti-mouse CD8, anti-mouse B220, and anti-mouse measured by T cell proliferation, no signs of clinical disease were F4/80 (BD Pharmingen). Sections were counterstained with 4Ј,6Ј-dia- midino-2-phenylindole (Sigma-Aldrich) and then coverslipped with observed (Fig. 2A), and there was minimal presence of any in- Vectashield mounting medium (Vector Laboratories). Slides were exam- flammatory infiltrate in the spinal cord of these mice, or in mice ined and images were acquired via epifluorescence using the SPOT RT immunized against OVA323–339 in comparison with PLP139–151- camera (Diagnostic Instruments) and Metamorph imaging software (Uni- immunized mice (Fig. 4A). However, there was sparse staining for versal Imaging). Eight sections from each sample per group were analyzed ϩ ϩ at ϫ100 magnification. CD4 T cells and F4/80 monocytes/macrophages, compared Photomicrographs of immunostained sections from spinal cord, cerebel- with OVA323–339-immunized mice, in the white matter of the cer- ϩ lum, and brainstem representative of the cellular infiltrates in the different ebellum of HI574–586-immunized mice (Fig. 4B). CD8 T cells 910 MOLECULAR MIMICRY-INDUCED CNS AUTOIMMUNE DISEASE

FIGURE 2. Infection, but not immunization, of SJL mice with the my-

elin mimic, HI574–586, induces a rapid-onset, nonprogressive form of clin- ical CNS disease. A, Separate groups of SJL mice (n ϭ 5) were immunized ␮ FIGURE 3. HI574–586 immunization of SJL mice induces early myelin- on day 0 with 100 g of either the self-myelin peptide PLP139–151, the ϩ myelin mimic peptide HI , or a control nonself, nonmimic peptide, specific CD4 T cell proliferative responses, but not Th1 differentiation. 574–586 SJL mice were immunized s.c. with 100 ␮gofPLP or HI OVA , and observed for the development of clinical EAE. B, Sepa- 139–151 574–586 323–339 A ϩ rate groups of SJL mice (n ϭ 5–7) were immunized on day 0 with 100 ␮g peptides in CFA. , Splenic CD4 T cell proliferative responses were assessed 16 days thereafter in response to restimulation with PLP , of PLP or on both days 0 and 7 with 100 ␮gofHI . Additional 139–151 139–151 574–586 HI , or the immunodominant TMEV capsid peptide, VP2 . T cell groups of mice were primed on days 0 and 7 with the indicated combina- 574–586 70–86 tions of a suboptimal dose (20 ␮g) of PLP (sPLP ), 100 ␮gof proliferation was determined at 96 h, and results are expressed as counts 139–151 139–151 ϫ Ϫ3 Ϯ HI ,or100␮g of the control OVA peptide. All mice were per minute ( 10 ) SEM from triplicate cultures. Restimulation with 574–586 323–339 PLP and HI peptides induced significant T cell proliferation observed for clinical signs of EAE for 38 days. Only mice primed on day 139–151 574–586 in both PLP139–151- and HI574–586-immunized groups. Neither group re- 0 with the optimal dose of PLP139–151 developed clinical disease. C, Sep- ϭ ϫ 7 sponded to challenge with the control VP270–86 peptide. B, In vivo differ- arate groups of SJL mice (n 10) were infected i.c. with 3 10 PFU of ϩ PLP139-BeAn, HI-BeAn, WT-TMEV, or OVA-BeAn, and observed for 70 entiation of CD4 Th1 cells was assessed by DTH ear swelling assays on ⌬ Disease severity was day 14 postpriming. Results are expressed as mean 24-h ear swelling ,ء .days for clinical signs of demyelinating disease Ϯ significantly more severe in PLP-BeAn-infected mice compared with HI- (responses of naive SJL mice subtracted) SEM in groups of four to five C ␥ Disease onset was significantly earlier mice. , IFN- levels (nanograms per milliliter) from the supernatants of ,ءء .BeAn-infected mice; p Ͻ 0.05 A in PLP-BeAn- and HI-BeAn-infected mice compared with OVA-BeAn- the proliferative cultures ( ) were determined by ELISA as described in p Ͻ ء Materials and Methods and WT-TMEV-infected mice; p Ͻ 0.05. These results in each panel are . , Values significantly above control levels; representative of at least three separate experiments. 0.05. These results in each panel are representative of at least three separate experiments.

ϩ and B220 B cells were not present at this early time point in old for HI574–586-induced EAE, mice were first immunized with a ␮ either of the groups (data not shown). suboptimal dose of PLP139–151 (sPLP139–151;20 g), and then ϩ ␮ immunized on day 7 with HI574–586 (100 g) or vice versa Double priming with HI574–586 or prepriming with a suboptimal (Fig. 2B). Negative controls consisted of mice immunized with dose of PLP139–151 does not induce clinical disease ␮ ␮ HI574–586 (100 g) or a suboptimal dose of PLP139–151 (20 g),

To determine whether multiple immunizations with the mimic which received a secondary immunization against OVA323–339 peptide could induce clinical disease, mice were immunized with (100 ␮g), or two immunizations (days 0 and ϩ7) with the subop- ␮ ϩ ␮ HI574–586 (100 g) in CFA on days 0 and 7 (Fig. 2B). To test timal dose of PLP139–151 (20 g) (Fig. 2B). Positive controls con-

whether preimmunization with PLP139–151 could lower the thresh- sisted of mice immunized with an optimal dose of PLP139–151 (100 The Journal of Immunology 911

FIGURE 4. CD4ϩ T cells infiltrate the cerebellum of mice immunized with the

HI574–586 mimic peptide. Groups of SJL mice (n ϭ 8) were immunized s.c. with 100 ␮gof

PLP139–151,HI574–586,orOVA323–339 peptide in CFA on day 0. At day 14 postimmunization, mice were anesthetized and perfused, and sec- tions of spinal cord (A) or cerebellum (B) were removed. Immunohistochemistry of 5-␮m-thick slices demonstrated the presence of CD4ϩ T cells and F4/80ϩ monocytes/macrophages present in both the brain and cerebellum of

PLP139–151/CFA-immunized mice, but only in the cerebellum, but not spinal cord, of HI574–586/ CFA-immunized mice. No infiltrating cells were observed in OVA323–339-immunized mice. Quantification of immunohistochemistry was determined on representative sections using Im- ageJ software. The percentages indicated in the lower left corner of each panel indicate the per- centage of positive pixels per area of the photomicrograph.

␮g) on day 0. Only mice primed with the optimal dose of PLP (Fig. 5B). This is in contrast to responses seen in SJL mice primed ␮ 139–151 (100 g) exhibited clinical disease (Fig. 2B). with HI574–586/CFA (Fig. 3B). T cells from HI-BeAn-infected mice also secreted significant amounts of IFN-␥ in response to Ͻ ء Infection of the CNS with the HI-BeAn mimic-expressing virus VP270–86,HI574–586, and PLP139–151 ( , p 0.05) rechallenge induces early-onset, nonprogressive clinical disease associated compared with naive mice (Fig. 5C). In contrast, T cells from mice with induction of Th1 autoreactivity to PLP139–151 infected with PLP139-BeAn only responded to rechallenge with

To determine whether exposure to the mimic peptide in the context VP270–86 and PLP139–151 but not to HI574–586 (Fig. 5C). These of an active virus infection could induce clinical disease, mice results demonstrate that exposure of mice to the HI574–586 mimic were infected i.c. with 3 ϫ 107 PFU of TMEV expressing the peptide in the context of an active, persistent, replicating virus

PLP139–151 mimic peptide, HI574–586 (HI-BeAn) (Fig. 2C). Posi- infection, but not in the context of CFA immunization, leads to tive controls consisted of mice infected i.c. with 3 ϫ 107 PFU of clinical disease, concomitant with induction of a potent Th1 re- the virus expressing the immunodominant PLP139–151 peptide sponse cross-reactive with the self PLP139–151 myelin epitope. (PLP139-BeAn) or WT-TMEV. Negative controls consisted of IL-4 was not produced by restimulation of T cells from HI-BeAn- 7 mice infected with 3 ϫ 10 PFU of the virus expressing the non- infected mice with either PLP139–151 or HI574–586, indicating lack encephalitogenic peptide OVA323–339 (OVA-BeAn) (Fig. 2C). of activation of Th2 cells (data not shown). Mice infected with PLP139-BeAn or HI-BeAn exhibited clinical p Ͻ Persistent preinfection of the CNS by HI-BeAn is required for ,ءء ;.disease with a significantly earlier onset (days 5–14 p.i 0.05) in comparison to mice infected with either OVA-BeAn or severe clinical disease following the cross-reactive activation of WT-TMEV (onset, days 38–54 p.i.) (Fig. 2C). However, mice PLP139–151 T cells by HI574–586 immunization infected with HI-BeAn experienced a nonprogressive clinical dis- The difference in route of Ag exposure between the HI574–586 p Ͻ 0.05 mimic peptide-immunized and -infected groups may account for ,ء) ease, with a significantly ameliorated clinical score between days 32 and 68 p.i.) compared with PLP139-BeAn-in- the difference in disease incidence observed. The activation of res- fected mice (Fig. 2C). The clinical disease exhibited by the HI- ident CNS cells following HI-BeAn infection may increase disease BeAn-infected mice was atypical of either the usual WT-TMEV or incidence. However, mice cannot be immunized with CFA in the EAE-type diseases reported previously (20). In these mice, disease CNS, and TMEV administered by peripheral routes rapidly crosses consisted of ruffled fur, mild waddling gait, and an arched back, the blood-brain barrier into the CNS (J. K. Olson and S. D. Miller, present for the duration of the experiment, but with no worsening unpublished observations). Therefore, to determine whether acti- over time (Fig. 2C). Both WT-TMEV- and OVA-BeAn-infected vation of CNS-resident cells by persistent virus infection is nec- mice exhibited a progressive clinical disease over the duration of essary for HI-BeAn-induced autoimmunity, mice were preinfected the experiment (70 days) compared with the HI-BeAn-infected with either HI-BeAn or OVA-BeAn, and then immunized with ⌬ ϩ mice (Fig. 2C). Mice infected with the Cla-BeAn parental virus HI574–586, OVA323–339, or PBS emulsified in CFA day 14 p.i. did not exhibit disease signs (data not shown) (21). Additionally, we tested whether a transient infection of the CNS by To determine whether infection with HI-BeAn induced the ⌬Cla-BeAn, a nonpathogenic variant of WT-TMEV that fails to cross-activation of self-reactive PLP139–151-specific T cells, HI- persist, was sufficient to provide the necessary stimuli to allow BeAn-infected mice were analyzed for T cell proliferation, DTH, induction of disease following HI574–586 immunization. Mice pre- ␥ and IFN- responses to PLP139–151. Mice infected with either HI- infected with HI-BeAn and then immunized against HI574–586 de- -p Ͻ 0.01) com ,ء) BeAn or PLP139-BeAn responded similarly via T cell prolifera- veloped a significantly more severe disease tion upon in vitro challenge with HI574–586 or PLP139–151, as well pared with mice infected with HI-BeAn and subsequently as the immunodominant virus peptide VP270–86 (Fig. 5A). In ad- immunized with either OVA323–339 or PBS in CFA (Fig. 6A). All dition, mice infected with either PLP139-BeAn or HI-BeAn re- HI-BeAn-infected groups developed the typical early-onset, non- sponded significantly to DTH rechallenge with either the self progressive disease. However, only the HI574–586/CFA-immu- Ͻ ء PLP139–151 peptide or the HI574–586 mimic peptide ( , p 0.05) nized group began to develop more severe clinical disease by day 912 MOLECULAR MIMICRY-INDUCED CNS AUTOIMMUNE DISEASE

tive responses, suggesting that priming with an irrelevant Ag had no nonspecific or bystander effect upon HI-specific T cell prolif-

eration (Fig. 6B). Following rechallenge with PLP139–151 T cell responses were greatest in the HI-immunized group (Fig. 6C). However, there was not as great an effect between the groups as seen with HI rechallenge (Fig. 6B). In addition, proliferative counts were substantially lower in the PLP rechallenge group com- pared with the HI group (12 ϫ 103 vs 40 ϫ 103 cpm, respectively; Fig. 6, B and C). Splenocytes from mice infected with HI-BeAn

and immunized against HI574–586 secreted significant greater ,p Ͻ 0.05) in vitro, compared with controls ,ء) ␥-amounts of IFN

when rechallenged with either HI574–586 or PLP139–151 (Fig. 6D). Severe clinical disease is associated with lesions in the cerebellum and brainstem and the spread of inflammatory infiltrate from the thoracic to lumbar region of spinal cord Spinal cord and brain removed at day 50 p.i. was cryosectioned for immunohistochemical analysis of CD4ϩ and CD8ϩ T cells, B220ϩ B cells, or F4/80ϩ monocyte/ infiltration (Fig. 7). Mice infected with HI-BeAn and later immunized with PBS/ CFA contained significant CD4ϩ T cell infiltrates in white matter perivascular lesions in the thoracic, but not lumbar regions of the spinal cord (Fig. 7A). In addition, diffuse F4/80ϩ staining was also observed in the white matter of the thoracic spinal cord (Fig. 7A).

HI-BeAn-infected mice, subsequently immunized with HI574–586/ CFA, showed diffuse CD4ϩ and F4/80ϩ staining in the thoracic region of the spinal cord similar to that seen in PLP-BeAn mice (Fig. 7A). In contrast to the other groups, diffuse CD4ϩ and F4/ 80ϩ staining was also observed in the lumbar white matter of

HI-BeAn plus HI574–586/CFA mice and PLP-BeAn mice, suggest- ing the trafficking of inflammatory infiltrate along the spinal cord (Fig. 7A). The significantly enhanced clinical disease scores (Fig. 6A) and infiltration of CD4ϩ T cells and F4/80ϩ macrophages in

HI-BeAn plus HI574–586/CFA mice as reflected by quantitative image analyses (Fig. 7) also correlated with increased myelin loss FIGURE 5. HI-BeAn infection of SJL mice induces robust differentia- in these areas as determined by anti-PLP-FITC staining (data not tion of myelin-specific CD4ϩ Th1 cells. Separate groups of SJL mice (n ϭ shown). In contrast, HI-BeAn-infected mice that were immunized 10) were infected i.c. with 3 ϫ 107 PFU of PLP139-BeAn or HI-BeAn. A, against OVA had markedly less CD4ϩ and F4/80ϩ staining ϩ 323–339 Splenic CD4 T cell proliferative responses were assessed 16 days there- present (Fig. 7A). ϩ after in response to restimulation with PLP139–151,HI574–586, or the im- In brain sections, CD4 T cells were observed only in the white munodominant TMEV capsid peptide, VP2 . T cell proliferation was 70–86 matter of the brainstem adjacent to the cerebellum and in the main determined at 96 h and results expressed as counts per minute (ϫ10Ϫ3) Ϯ region of the brainstem of mice infected with HI-BeAn and sub- SEM from triplicate cultures. Restimulation with PLP139–151,HI574–586, sequently immunized with HI574–586/CFA similar to PLP-BeAn- and VP270–86 peptides induced significant T cell proliferation in both the ϩ PLP139-BeAn- and HI-BeAn-infected mice. B, In vivo differentiation of infected mice (Fig. 7B). F4/80 cells were observed in large quan- CD4ϩ Th1 cells was assessed by DTH ear swelling assays on day 14 postprim- tities in the white matter of the cerebellum and brainstem of the ing. Results are expressed as mean ⌬24-h ear swelling (responses of naive SJL HI-BeAn plus HI574–586/CFA group, although the HI-BeAn plus Ϯ ␥ mice subtracted) SEM in groups of four to five mice. C, IFN- levels PBS/CFA and HI-BeAn plus OVA323–339/CFA groups also con- (nanograms per milliliter) from the supernatants of the proliferative cultures tained minimal F4/80ϩ staining in the cerebellar white matter (Fig. 7B). In all groups, both CD8ϩ and B220ϩ cells were present at low ,ء .A) were determined by ELISA as described in Materials and Methods) Ͻ Values significantly above control levels; p 0.05. These results in each panel numbers in both the spinal cord white matter and in the cerebellar are representative of at least three separate experiments. white matter and brainstem (data not shown). Naive mice were negative for CD4ϩ and CD8ϩ T cells, B220ϩ B cells, and F4/80ϩ monocytes/macrophages in both the spinal cord and brain (Fig. 7). 30 p.i. By day 50 p.i., mice had reached a average clinical score of 2.3, which did not increase in severity, as late as day 110 p.i. (Fig. Discussion 6A). In contrast, preinfection of the CNS of mice with the non- This study clearly demonstrates that the viral delivery of a myelin persistent ⌬Cla-BeAn variant, or OVA-BeAn, did not predispose PLP HI mimic peptide can induce cross-activation of autoreactive ϩ mice to a more severe disease following HI574–586 or OVA323–339 PLP139–151-specific CD4 T cells and CNS autoimmune disease, immunization (data not shown). via molecular mimicry. In addition, the current results highlight HI-BeAn-infected mice subsequently immunized against with the importance of studying molecular mimicry in the context of

HI574–586/CFA developed significantly greater T cell proliferative pathogen-induced innate inflammatory immune signals. responses upon in vitro rechallenge with HI574–586 compared with The primary TCR and MHC class II binding sites of PLP139–151, the control groups (Fig. 6B). HI-BeAn-infected mice immunized a dominant encephalitogenic myelin peptide in SJL mice, have with either OVA323–339 or PBS/CFA had similar T cell prolifera- been defined by multiple amino acid substitutions (19, 25, 26). The Journal of Immunology 913

ϭ FIGURE 6. Preinfection with HI-BeAn before HI574–586 immunization induces a severe clinical disease. Separate groups of SJL mice (n 10) were ϫ 7 ␮ infected i.c. with 3 10 PFU HI-BeAn and then treated s.c. with PBS or immunized with 100 g of either HI574–586/CFA or OVA323–339/CFA at day ϩ14 p.i. (1). A, Mice were observed for clinical signs of demyelinating disease for 114 days. Mice preinfected with HI-BeAn and then immunized against

HI574–586 developed clinical disease that was significantly more severe than that observed in HI-BeAn-infected mice treated with PBS or immunized against the control OVA323–339 peptide. B and C, T cell-proliferative responses were determined at day 35 post-initial infection in response to restimulation with ϫ Ϫ3 Ϯ either HI574–586 or PLP139–151. T cell proliferation was determined at 96 h, and results are expressed as counts per minute ( 10 ) SEM from triplicate cultures. D, IFN-␥ levels (nanograms per milliliter) from the supernatants of the proliferative cultures (B and C) were determined by ELISA as described Values significantly above control levels; p Ͻ 0.05. These results in each panel are representative of at least three separate ,ء .in Materials and Methods experiments.

Analysis of these sites determined epitopes from infectious patho- correlates with disease onset (10–14 days p.i.) (Figs. 2 and 5). In gens, which shared sequence homology at the primary TCR and comparison, WT-TMEV- or OVA-BeAn-infected groups develop

MHC residues. A peptide from serine protease IV (HI574–586)se- late-onset disease and PLP139–151-specific T cell responses that are creted by HI bacteria, a natural mouse pathogen, contains struc- first evident 40–55 days p.i. arising via epitope spreading rather tural similarities to the native PLP139–151 peptide (19). Although than molecular mimicry (14, 21). Mice infected with the PLP139- this sequence has a limited sequence homology, it shares the pri- BeAn virus, a model of molecular identity, also exhibited an early- mary TCR contact site at position 144 and the primary and sec- onset clinical disease, but with more severe clinical symptoms and s ondary I-A contact residues at positions 145 and 148. The HI574–586 a progressive clinical course (Fig. 2C) (21). Interestingly, as pre- peptide has been shown to bind I-As and to activate T cell clones viously reported (21), mice infected with OVA-BeAn develop a derived from PLP139–151-primed SJL mice (19). similar late-onset disease profile as seen with WT-TMEV infec- In this study, we report that i.c. infection of mice with the HI- tion. Recent tolerance experiments indicate that disease severity BeAn variant of TMEV induces a rapid onset, nonprogressive is significantly inhibited in HI-BeAn-infected mice pretolerized form of clinical disease, wherein the early cross-reactive induction to PLP139–151, demonstrating that PLP139–151-specific T cells ␥ of IFN- -producing PLP139–151-specific Th1 cells temporally activated by cross-reactivity to HI574–586, are critical for the 914 MOLECULAR MIMICRY-INDUCED CNS AUTOIMMUNE DISEASE

FIGURE 7. Severe clinical disease in HI-

BeAn-infected plus HI574–586/CFA-immu- nized mice is associated with increased in- flammatory infiltrates in the lumbar region of the spinal cord, cerebellum, and brainstem. Groups of SJL mice (n ϭ 5–8) were infected i.c. with 3 ϫ 107 PFU of HI-BeAn and then immunized s.c. with either PBS or 100 ␮gof either HI574–586/CFA or OVA323–339/CFA at day ϩ14 p.i. At day 50 p.i., mice were anes- thetized and perfused, and sections of spinal cord were dissected into the thoracic and lum- bar regions (A), and the brainstem and cere- bellum (B) were removed. Immunohistochem- istry of 5-␮m-thick tissue slices was performed, by staining for the presence of CD4ϩ T cells and F4/80ϩ monocytes/macro- phages. Immunohistochemistry of the brain and spinal cord from PLP-BeAn-infected mice (infected i.c. with 3 ϫ 107 PFU) was used as a positive control. Quantification of immuno- histochemistry was determined on representa- tive sections using ImageJ software. The per- centages indicated in the lower left corner of each panel indicate the percentage of positive pixels per area of the photomicrograph.

4 early-onset HI-BeAn-induced disease. Furthermore, disease in- the mimic HI574–586 or self PLP139–151 peptides (Fig. 5). This is an duction does not correlate with enhanced responses to TMEV important finding, which suggests that the host can process the epitopes (Fig. 5) or to any observed differences in virus replication HI574–586 mimic sequence from the leader sequence of HI-BeAn in vivo or in vitro (data not shown). ϩ and present it in an inflammatory context to PLP139–151-specific T Following i.c. infection with HI-BeAn, CD4 T cells respond cells. Furthermore, in vivo presentation of the HI mimic equally in T cell proliferation assays following rechallenge with 574–586 sequence can induce differentiation of PLP139–151-reactive Th1 cells, as measured by IFN-␥ ELISA and in vivo DTH responses to 4 J. L. Croxford, J. K. Olson, H. A. Anger, and S. D. Miller. Initiaiton and exacer- PLP . Therefore, the early disease and associated enhanced bation of CNS autoimmune demyelination via virus-induced molecular mimicry: im- 139–151 plicaitons for MS pathogenesis. Submitted for publication. CNS inflammation (Fig. 7) are likely the result of an immune The Journal of Immunology 915

response to virus-infected cells expressing the mimic epitope in the PLP139–151 rechallenge in vitro, they do not produce significant CNS, and/or due to a response to other endogenous CNS APCs quantities of IFN-␥ (Fig. 5C), a Th1 cytokine thought to be im- presenting the self PLP139–151 epitope that are not virus infected. portant in the induction of EAE, TMEV, and MS. In addition, ␥ However, despite the presence of IFN- -producing PLP139–151-reac- HI574–586-immunized mice demonstrate minimal in vivo DTH re- tive T cells, HI-BeAn-infected mice develop only a mild, nonprogres- sponses to PLP139–151 rechallenge compared with rechallenge sive form of clinical disease, compared with PLP139-BeAn-infected with the cognate Ag (Fig. 3B). s mice. The I-A molecule has a lower affinity for the mimic HI574–586 Interestingly, in mice immunized with PLP139–151, which induces peptide compared with that of PLP139–151 (IC50, 385 and 87 nM, potent clinical disease, in vitro rechallenge with HI574–586 not only respectively) (19), and therefore, potential cross-reactive T cells may induces CD4ϩ T cell proliferation, but also significant CD4ϩ Th1 not be fully reactivated when they encounter the self-peptide in vivo. differentiation as measured by IFN-␥ secretion and DTH responses. Alternatively, the nonprogressive disease observed may be due to the Furthermore, double immunization with HI574–586 also failed to in- relative proportion of HI -specific and PLP -specific T 574–586 139–151 duce clinical disease (Fig. 2). Therefore, it appears that PLP139–151 is ϩ cells activated following HI-BeAn infection. The HI574–586 popula- a stronger immunogen than HI with regard to in vivo CD4 ϩ 574–586 tion of CD4 T cells may therefore be a subpopulation of the total Th1 differentiation. Again, this may reflect the lower affinity of the PLP population or may be separate to the PLP popula- s 139–151 139–151 HI574–586 peptide for I-A compared with that of PLP139–151,orthe tion, with only minor overlap, as we have previously proposed (10). subpopulation hypothesis discussed above (10). Therefore, rechal- The identity of the cross-reactive repertoire is currently under inves- lenge with PLP139–151 in HI574–586-immunized mice would target tigation in our laboratory. ϩ all HI574–586-specific CD4 T cells. In contrast, rechallenge with Another possible explanation could be that, whereas the cross- HI574–586 in PLP139–151-immunized mice would target only a small ϩ activated PLP139–151 Th1 cells were encephalitogenic, the HI574–586- percentage of the total PLP139–151-specific CD4 T cells. activated population could differentiate to a Th2 phenotype and may During the disease course of EAE and TMEV-induced demy- be protective against disease induction. However, rechallenge of ϩ elinating disease, inflammatory infiltrates consisting mainly of CD4 T cells isolated from either OVA-BeAn- or HI-BeAn-infected CD4ϩ T cells and F4/80ϩ monocytes can be observed in the white mice, with either PLP139–151 or HI574–586, did not induce the Th2- matter of the spinal cord, which have been shown to correlate with type IL-4 or IL-5, as measured by ELISPOT and ELISA clinical disease severity (24). Therefore, we studied the inflamma- (data not shown). Although significant numbers of CD4ϩ T cells were tory infiltrate present in the spinal cord and brain of mice immu- observed by immunohistochemistry in the white matter of the thoracic nized with HI /CFA. Throughout the spinal cord, there was region of the spinal cord in HI-BeAn-infected mice, the majority of 574–586 ϩ little presence of inflammatory infiltrates, consistent with the lack CD4 infiltrate appeared to be perivascular in nature (Fig. 7A). These ϩ ϩ of clinical disease. However, CD4 T cells and F4/80 mono- data may support the idea that T cells activated by HI-BeAn may not cytes/macrophages were present in the cerebellum of HI -, be fully activated upon encountering PLP epitopes in the CNS, 574–586 139–151 but not OVA -immunized mice. This suggests that the cer- and therefore do not efficiently traffic into the spinal cord parenchyma. 323–339 ebellum may be the primary CNS entry point for activated cells Further evidence for deficient trafficking is seen by the relative lack of from the periphery. However, HI -specific T cells may lack inflammatory infiltrate in the lumbar region of the spinal cord, where 574–586 sufficient affinity to PLP in vivo to allow further activation lesions are normally seen during EAE and TMEV disease (24). How- 139–151 ϩ in the CNS. Therefore, in the absence of activated resident CNS ever, a number of CD4 T cells do traffic through the thoracic spinal APC, such as microglia, myelin epitopes may be presented in a cord, and these may represent the proportion of T cells that have a noninflammatory context, i.e., with minimal costimulatory mole- high affinity for PLP139–151. We have previously shown that mice infected with PLP139-BeAn develop severe clinical disease associ- cule expression. This may have a negative impact on the cascade ated with widespread inflammation of the spinal cord (21, 23). In this of progressive events leading to severe clinical disease, such as ϩ secretion, adhesion molecule expression, and second- instance, it is likely that the majority of CD4 T cells are PLP139–151 specific with a high affinity for this epitope in vivo. ary recruitment of inflammatory cells. This is in contrast to the situation of potent induction of PLP139–151-directed autoimmunity Initial studies on molecular mimicry demonstrated that immu- ϩ nization of rabbits against a viral epitope, that shared 6 in HI-BeAn-infected mice, where the cross-activated CD4 T cells ␥ aa with an encephalitogenic MBP peptide, could induce T cell secrete high levels of IFN- . In addition, they also encounter the reactivity to MPB (9). Despite this autoreactive T cell component, CNS in an inflammatory state due to the presence of numerous and inflammatory infiltrate in the cerebral cortex, clinical disease innate immune molecules and up-regulation of MHC and costimu- did not develop (9). Other studies have suggested the likelihood of latory molecules directly induced by TMEV infection of CNS- molecular mimicry playing a role in the induction of autoimmu- resident microglia and (32). nity, where immunization with viral sequences mimicking self- Activation of innate immune responses by TMEV appears to be peptides can activate autoreactive T cells (17, 27, 28). Studies in a critical determinant in explaining the difference in disease sus- other infectious autoimmune models, such as myocarditis and her- ceptibility between the mimic peptide immunization vs infection pes stromal keratitis, also suggest the likelihood of molecular mim- paradigms. It is likely that the type of virus encoding the mimic icry playing a role in their pathogenesis (29–31). Immunization epitope, and its cell tropism, are important factors in setting the ϩ threshold for disease susceptibility upon infection with pathogens with the HI mimic peptide can activate PLP139–151-specific CD4 T cells, but does not induce clinical disease (Fig. 2) (19). We encoding molecular mimics. Previous reports have demonstrated demonstrate that the mechanism behind the inability of HI574–586 that a vaccinia virus encoding the whole PLP construct failed to immunization to induce clinical disease is due to the lack of Th1 induce clinical disease in SJL/J mice, although mice were more ϩ differentiation of the cross-activated CD4 T cells. This is in susceptible to later induction of PLP139–151-induced EAE (33). In marked contrast to the potent Th1 IFN-␥ response recalled by contrast, infection of mice with vaccinia virus engineered to ex-

PLP139–151 in mice infected with the HI-BeAn virus (Fig. 5C) and press an encephalitogenic epitope of MPB protected mice from further supports the importance of using infectious pathogens for EAE induction by subsequent MPB immunization (34). Vaccinia ϩ studies of molecular mimicry. Although CD4 T cells from HI574– virus was administered either by i.p. injection or by tail scarifica- 586-immunized mice proliferate significantly in response to tion (34). It is likely that different cell types will be exposed to the 916 MOLECULAR MIMICRY-INDUCED CNS AUTOIMMUNE DISEASE virus by these routes, which may influence the ability of the patho- In addition, the need for the mimic-expressing virus to persist in gen to induce clinical disease. Vaccinia virus can infect many cell the CNS is also being addressed. types, which may lack APC function and costimulatory molecule expression, as opposed to TMEV, which appears to have tropism Acknowledgments for APCs. We have recently shown that TMEV infection of mi- We thank Dr. Kaori Sakuishi (National Center of Neurology and Psychi- croglia, which normally do not express MHC or costimulatory atry, Tokyo, Japan) for consultation on and interpretation of the molecules, induces these cells to up-regulate type I IFNs, IL-12, immunohistochemical data. IL-18, MHC class I and II, and costimulatory molecules and to acquire Ag presentation function (32). References From these studies, it is possible to draw parallels to benign MS, 1. Kurtzke, J. F. 1993. Epidemiologic evidence for multiple sclerosis as an infec- where patients experience a single clinical episode with little or no tion. Clin. Microbiol. Rev. 6:382. further exacerbation. Therefore, infection of the target organ with 2. ter Meulen, V., and M. Katz. 1997. The proposed viral etiology of multiple sclerosis and related demyelinating disease. In Multiple Sclerosis: Clinical and a mimic-expressing virus during childhood may activate a popu- Pathogenetic Basis. C. S. Raine, H. F. McFarland, and W. W. Tourtellotte, eds. lation of autoreactive T cells and induce no clinical signs or only Chapman and Hall, London, p. 287. 3. Padgett, B. L., and D. L. Walker. 1976. New human papovaviruses. Prog. Med. mild symptoms. Various risk factors have been described, which Virol. 22:1. are thought to be involved in inducing or exacerbating relapses in 4. Osame, M., K. Usuku, S. Izumo, N. Ijichi, H. Amitani, A. Igata, M. Matsumoto, MS patients, such as stress, trauma, elevated temperatures, infec- and M. Tara. 1986. HTLV-I associated myelopathy, a new clinical entity. Lancet 1:1031. tions, and vaccination (35). Therefore, it is tempting to speculate 5. Gessain, A., F. Barin, J. C. Vernant, O. Gout, L. Maurs, A. Calender, and that a normally innocuous stimulus later in life may reactivate G. de The. 1985. to human T-lymphotropic virus type-I in patients those autoreactive T cells to induce severe clinical disease (10). To with tropical spastic paraparesis. Lancet 2:407. 6. Olsson, T., J. Sun, J. Hillert, B. Hojeberg, H. P. Ekre, G. Andersson, O. Olerup, test this hypothesis, mice were preinfected with HI-BeAn, and im- and H. Link. 1992. Increased numbers of T cells recognizing multiple myelin munized 14 days later with either HI574–586 or OVA323–339. Mice basic protein epitopes in multiple sclerosis. Eur. J. Immunol. 22:1083. preinfected with HI-BeAn and immunized with HI devel- 7. Scholz, C., K. T. Patton, D. E. Anderson, G. J. Freeman, and D. A. Hafler. 1998. 574–586 Expansion of autoreactive T cells in multiple sclerosis is independent of exoge- oped a more severe disease than mice infected with HI-BeAn nous B7 costimulation. J. Immunol. 160:1532. alone, or those later immunized with OVA323–339. This suggests 8. de Rosbo, N. K., M. Hoffman, I. Mendel, I. Yust, J. Kaye, R. Bakimer, that induction of severe disease requires a secondary stimulus con- S. Flechter, O. Abramsky, R. Milo, A. Karni, and A. Ben-Nun. 1997. Predomi- nance of the autoimmune response to myelin oligodendrocyte glycoprotein taining the original mimic. In addition, preinfection with a persis- (MOG) in multiple sclerosis: reactivity to the extracellular domain of MOG is tent virus that expresses a nonself, nonmimic epitope (OVA- directed against three main regions. Eur. J. Immunol. 27:3059. ⌬ 9. Fujinami, R. S., and M. B. Oldstone. 1985. Amino acid homology between the BeAn) or a transient infection with Cla-BeAn, which fails to encephalitogenic site of myelin basic protein and virus: mechanism for autoim- persist in the CNS, did not predispose mice to a more severe dis- munity. Science 230:1043. 10. Croxford, J. L., J. K. Olson, and S. D. Miller. 2002. Epitope spreading and ease upon HI574–586 immunization (data not shown). The exacer- molecular mimicry as triggers of autoimmunity in the Theiler’s virus induced bated disease was associated with an increase in HI574–586-specific demyelinating disease model of multiple sclerosis. Autoimmun. Rev. 1:251. T cell proliferation and IFN-␥ secretion, and increased numbers of 11. Tejada-Simon, M. V., Y. C. Zang, J. Hong, V. M. Rivera, and J. Z. Zhang. 2003. CD4ϩ T cells in both the thoracic and lumbar regions of spinal Cross-reactivity with myelin basic protein and human herpesvirus-6 in multiple ϩ ϩ sclerosis. Ann. Neurol. 53:189. cord. In addition, significant numbers of CD4 T cells and F4/80 12. Lipton, H. L. 1975. Theiler’s virus infection in mice: an unusual biphasic disease monocytes/macrophages were observed in the white matter of the process leading to demyelination. Infect. Immun. 11:1147. brainstem adjacent to the cerebellum and in the main region of the 13. Miller, S. D., and S. J. Gerety. 1990. Immunologic aspects of Theiler’s murine encephalomyelitis virus (TMEV)-induced demyelinating disease. Semin. Virol. brainstem of these mice, in contrast to mice infected with HI-BeAn 1:263. and immunized against OVA323–339. However, disease in mice 14. Miller, S. D., C. L. Vanderlugt, W. S. Begolka, W. Pao, R. L. Yauch, infected with HI-BeAn and subsequently immunized against K. L. Neville, Y. Katz-Levy, A. Carrizosa, and B. S. Kim. 1997. Persistent in- fection with Theiler’s virus leads to CNS autoimmunity via epitope spreading. HI574–586 was slightly less severe than in mice infected with PLP- Nat. Med. 3:1133. BeAn alone. This may be due to the differential effects of the HI 15. Karpus, W. J., J. G. Pope, J. D. Peterson, M. C. Dal Canto, and S. D. Miller. 1995. Inhibition of Theiler’s virus-mediated demyelination by peripheral immune tol- peptide on the immune system as described above. erance induction. J. Immunol. 155:947. This study demonstrates the cross-reactive potential of a myelin 16. Vanderlugt, C. L., and S. D. Miller. 2002. Epitope spreading in immune-mediated mimic sequence from an infectious pathogen, delivered by a re- diseases: implications for immunotherapy. Nat. Rev. Immunol. 2:85. 17. Wucherpfennig, K. W., and J. L. Strominger. 1995. Molecular mimicry in T combinant neurotropic virus, and suggests that molecular mimicry cell-mediated autoimmunity: viral peptides activate human T cell clones specific is a potential mechanism for the induction of a T cell-mediated for myelin basic protein. Cell 80:695. CNS autoimmune disease. This initial response can be further ex- 18. Ufret-Vincenty, R. L., L. Quigley, N. Tresser, S. H. Pak, A. Gado, S. Hausmann, K. W. Wucherpfennig, and S. Brocke. 1998. In vivo survival of viral - acerbated following reactivation of the autoreactive T cells. In con- specific T cells that induce experimental autoimmune encephalomyelitis. J. Exp. trast, immunization with the identical mimic peptide multiple Med. 188:1725. times in CFA failed to induce clinical disease highlighting the 19. Carrizosa, A. M., L. B. Nicholson, M. Farzan, S. Southwood, A. Sette, R. A. Sobel, and V. K. Kuchroo. 1998. Expansion by self antigen is necessary for importance of studying molecular mimics in the context of innate the induction of experimental autoimmune encephalomyelitis by T cells primed immune and other stimuli present during ongoing infection. Col- with a cross-reactive environmental antigen. J. Immunol. 161:3307. lectively, these data suggest that molecular mimicry could be an 20. Gautam, A. M., R. Liblau, G. Chelvanayagam, L. Steinman, and T. Boston. 1998. Viral peptide with limited homology to a self peptide can induce clinical signs of important factor in the pathogenesis of some forms of MS by ex- experimental autoimmune encephalomyelitis. J. Immunol. 161:60. panding a population of memory T cells that cross-react with my- 21. Olson, J. K., J. L. Croxford, M. Calenoff, M. C. Dal Canto, and S. D. Miller. elin epitopes. The autoreactive cells may then be further expanded 2001. A virus-induced molecular mimicry model of multiple sclerosis. J. Clin. Invest. 108:311. later in life by a number of differing stimuli, which could include 22. Lipton, H. L., J. Kratochvil, P. Sethi, and M. C. Dal Canto. 1984. Theiler’s virus reinfection with the same mimic-expressing virus. Alternatively, antigen detected in mouse spinal cord 21⁄2 years after infection. Neurology 34:1117. other unrelated stimuli, which normally may be nonpathogenic, 23. Olson, J. K., T. N. Eagar, and S. D. Miller. 2002. Functional activation of myelin- may cause the restimulation of the autoreactive T cells, either spe- specific T cells by virus-induced molecular mimicry. J. Immunol. 169:2719. cifically or by a bystander mechanism. Therefore, we are currently 24. Begolka, W. S., C. L. Vanderlugt, S. M. Rahbe, and S. D. Miller. 1998. Differ- ential expression of inflammatory cytokines parallels progression of central ner- studying the effect of repeated viral infections of the CNS and/or vous system pathology in two clinically distinct models of multiple sclerosis. the periphery either by viruses containing mimic or irrelevant Ags. J. Immunol. 161:4437. The Journal of Immunology 917

25. McRae, B. L., and S. D. Miller. 1994. Fine specificity of CD4ϩ T cell responses 30. Panoutsakopoulou, V., M. E. Sanchirico, K. M. Huster, M. Jansson, F. Granucci, to the dominant encephalitogenic PLP 139–151 peptide in SJL/J mice. Neuro- D. J. Shim, K. W. Wucherpfennig, and H. Cantor. 2001. Analysis of the rela- chem. Res. 19:997. tionship between viral infection and autoimmune disease. Immunity 15:137. 26. Kuchroo, V. K., J. M. Greer, D. Kaul, G. Ishioka, A. Franco, A. Sette, 31. Bachmaier, K., N. Neu, L. M. de la Maza, S. Pal, A. Hessel, and J. M. Penninger. R. A. Sobel, and M. B. Lees. 1994. A single TCR antagonist peptide inhibits 1999. Chlamydia infections and heart disease linked through antigenic mimicry. experimental allergic encephalomyelitis mediated by a diverse T cell repertoire. Science 283:1335. J. Immunol. 153:3326. 32. Olson, J. K., A. M. Girvin, and S. D. Miller. 2001. Direct activation of innate and antigen presenting functions of microglia following infection with Theiler’s vi- 27. Hemmer, B., B. T. Fleckenstein, M. Vergelli, G. Jung, H. McFarland, R. Martin, rus. J. Virol. 75:9780. and K. H. Wiesmuller. 1997. Identification of high potency microbial and self 33. Barnett, L. A., J. L. Whitton, Y. Wada, and R. S. Fujinami. 1993. Enhancement ligands for a human autoreactive class II-restricted T cell clone. J. Exp. Med. of autoimmune disease using recombinant vaccinia virus encoding myelin pro- 185:1651. teolipid protein. J. Neuroimmunol. 44:15. 28. Gran, B., B. Hemmer, M. Vergelli, H. F. McFarland, and R. Martin. 1999. Mo- 34. Barnett, L. A., J. L. Whitton, L. Y. Wang, and R. S. Fujinami. 1996. Virus lecular mimicry and multiple sclerosis: degenerate T-cell recognition and the encoding an encephalitogenic peptide protects mice from experimental allergic induction of autoimmunity. Ann. Neurol. 45:559. encephalomyelitis. J. Neuroimmunol. 64:163. 29. Zhao, Z. S., F. Granucci, L. Yeh, P. A. Schaffer, and H. Cantor. 1998. Molecular 35. Whitaker, J. N., and G. W. Mitchell. 1997. Clinical features of multiple sclerosis. mimicry by -type 1: autoimmune disease after viral infec- In Multiple Sclerosis: Clinical and Pathogenetic Basis. C. S. Raine, tion. Science 279:1344. H. F. McFarland, and W. W. Tourtellotte, eds. Chapman and Hall, London, p. 3.