Chronic inflammatory demyelinating polyneuropathy is a debilitating autoimmune disease characterized by peripheral nerve demyelination and dysfunction. How the autoimmune response is initiated, identity of provoking Ags, and pathogenic effector mechanisms are not well defined. The autoimmune regulator (Aire) plays a critical role in central tolerance by promoting thymic expression of self-Ags and deletion of self-reactive T cells. In this study, we used mice with hypomorphic Aire function and two patients with Aire mutations to define how Aire deficiency results in spontaneous autoimmune peripheral neuropathy. Autoimmunity against peripheral nerves in both mice and humans targets myelin protein zero, an Ag for which expression is Aire-regulated in the thymus. Consistent with a defect in thymic tolerance, CD4+ T cells are sufficient to transfer disease in mice and produce IFN-γ in infiltrated peripheral nerves. Our findings suggest that defective Aire-mediated central tolerance to myelin protein zero initiates an autoimmune Th1 effector response toward peripheral nerves.
Chronic inflammatory demyelinating polyneuropathy (CIDP) is the most common acquired chronic autoimmune neuropathy and affects 1 in 10,000 individuals (1). The pathogenic steps resulting in immune destruction of the peripheral nervous system (PNS) are not well understood, in part because of the scarcity of robust animal models. We recently reported that spontaneous autoimmune peripheral neuropathy develops in NOD mice harboring a G228W point mutation in the autoimmune regulator (Aire) gene (NOD.AireGW/+ mice) (2). Aire plays a critical role in central tolerance by upregulating the ectopic expression of a wide array of tissue-specific self-Ags in medullary thymic epithelial cells (mTECs) (3) and promoting the negative selection of developing thymocytes that recognize these Ags with high affinity (4). NOD.AireGW/+ mice have hypomorphic Aire function in that mTECs express tissue-specific self-Ags at ∼10% of normal levels (2). NOD.AireGW/+ mice are protected from early lethal autoimmune diseases (e.g., exocrine pancreatitis, pneumonitis) but remain susceptible to a distinct set of autoimmune diseases that include autoimmune peripheral neuropathy.
Patients with autoimmune polyendocrinopathy syndrome type 1 (APS1) have genetic mutations in Aire and develop autoimmunity. Recently, CIDP was recognized as a potential novel component of APS1 in two unrelated children with progressive sensory loss, motor weakness, and confirmed mutations in Aire (5). We show in this study that NOD.AireGW/+ mice develop autoimmune peripheral neuropathy that strongly resembles CIDP. We use NOD.AireGW/+ mice to identify myelin protein zero (P0) as a major Aire-regulated PNS Ag and demonstrate defective tolerance to P0 in both Aire-deficient mice and humans.
Materials and Methods
NOD.AireGW/+ mice were generated as previously described (2). NOD.scid mice were purchased from The Jackson Laboratory. Mice were housed in a pathogen-free barrier facility at University of California, San Francisco and at the University of North Carolina, Chapel Hill. Clinical neuropathy and diabetes were assessed as described previously (6). For the group of mice used in neuropathy incidence studies, mice that developed diabetes were maintained on insulin (i.p. injection daily) until 22 wk of age. NOD.AireGW/+ mice with clinical neuropathy were used as serum and splenocyte donors. Experiments complied with the Animal Welfare Act and the National Institutes of Health guidelines for the ethical care and use of animals in biomedical research.
H&E staining was performed as described in (2). Immune infiltration was scored in a blinded fashion with 0, 1, 2, 3, and 4 scores indicating no, <25, 25–50, 50–75, and >75% infiltration, respectively. Luxol fast blue staining was performed as described previously (7). Toluidine blue staining and transmission electron microscopy was performed on sciatic nerve sections from mice perfused with 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer. Toluidine blue staining was performed on semithin sections of epon-embedded sciatic nerves. Transmission electron microscopy was performed as described (in Ref. 8).
Immunofluorescence staining was performed using diluted (1:600) mouse and human sera on OCT-embedded NOD.scid sciatic nerve sections as described previously (9).
Adoptive transfer of whole spleen and lymph node cells from neuropathic NOD.AireGW/+ mice was performed as described previously (6). Diabetic NOD.AireGW/+ mice were excluded as donors. For each NOD.scid recipient mouse, 10 × 106 whole spleen and lymph node cells were transferred. CD4+ and CD8+ T cells were isolated by staining with CD4-FITC (Southern Biotechnology Associates) and CD8a-APC-Cy7 (BD Biosciences). Populations were purified using a MoFlo (DakoCytomation) cell sorter and plated in the presence of anti-CD3 and anti-CD28 (1 μg/ml in PBS) for 96 h. For some experiments, anti-CD4 and anti-CD8 microbeads (Miltenyi Biotec) were used for positive selection on an MS or LS column. For the last 24 h, 1 μg/ml IL-2 (20 U/ml) was added to culture. A total of 7 × 106 CD4, 7 × 106 CD8, or 3 × 106 CD8 T cells was transferred. Cotransfers of 7 × 106 CD4 and 7 × 106 CD8 T cells were also performed.
Intracellular cytokine staining
Infiltrating immune cells were isolated from sciatic nerves by mincing and digestion in 2 mg/ml collagenase. Cells were stimulated with PMA/ionomycin and stained as described previously (9).
Sciatic nerve extracts were prepared in CHAPS buffer. Immunoblots and immunoaffinity purification were performed as described (in Ref. 10). Bound Ag was eluted using 100 mM triethylamine (pH 11.5) and neutralized with 1 M phosphate buffer (pH 6.8).
Fusion protein purification/competition studies
P0-maltose binding protein (MBP) fusion protein was produced as described previously (10). Human P0 cDNA clone (IMAGE) (>90% homology to mouse P0) was subcloned into pMAL-c2X (New England Biolabs) using EcoRI (5′) and HindIII (3′) restriction sites. These sites were introduced onto human P0 amplicons generated with primers 5′-GAATTCATGGCTCCCGGGGCTC CCT-3′ and 5′-AAGCTTCTATTTCTTATCCTTGCGAGACTCC-3′. Competition studies were performed by preincubating sera with serial dilution of recombinant P0-MBP or MBP prior to immunoblotting. ELISAs were performed using 1:600 serum dilutions and 50 ng/well recombinant protein.
Recombinant P0-MBP or MBP alone (50 ng/well) was immobilized in wells, and diluted (1:600) serum specimens were used for assays. Rat anti-mouse alkaline phosphatase was used to detect mouse Ig.
Thymic stromal preparations were performed as described (in Ref. 10). mTECs were sorted using the markers propidium iodide−, CD45−G8.8+, and Ly51int
Proliferation assay using [3H]thymidine incorporation was performed as described previously (11). P0 180–199 SSKRGRQTPVLYAMLDHSRS and hen egg lysozyme (HEL) 11–25 AMKRHGLDNYRGYSL peptides were purchased from Genemed Synthesis. A total of 5 × 105splenocytes/well was cultured in HL-1 medium with nonessential amino acids (BioWhittaker), 2 mM l-glutamine, 1 mM sodium pyruvate, and 55 mM 2-ME with peptide at three different concentrations and without peptide. Cultures were pulsed with 1 μCi [3H]thymidine and incubated for an additional 18 h. Cells were harvested on a glass-fiber filter and incorporation of [3H]thymidine measured using a liquid scintillation counter. Stimulation index was calculated by dividing counts per minute with Ag by counts per minute without Ag.
Radioligand binding assay
Human P0 cDNA clone (IMAGE) was in vitro transcribed and translated with [35S] labeling and assays performed as described (9). Autoantibody index was calculated as [cpm sample − cpm negative control]/[cpm positive standard − cpm negative standard] × 100. Samples were positive if >3 SD above mean for healthy control group.
Patients and controls were included in the study after written informed consent was obtained. Study protocol was approved by the institutional review board at University of California, San Francisco.
Data was analyzed with Prism software (GraphPad) using unpaired t tests. Log rank tests were used for comparison of survival curves. A p value ≤0.05 was considered significant.
Hypomorphic Aire mice spontaneously develop autoimmune peripheral neuropathy
By 22 wk of age, ∼80% of female NOD.AireGW/+ mice develop spontaneous neuropathy (Fig. 1A) (2) that is not seen in wild-type (NOD.WT) littermates. Affected mice display bilateral weakness of the hind limbs that progresses to severe paralysis affecting all limbs (Fig. 1B, Supplemental Video 1). This neuropathy is associated with immune cell infiltration in NOD.AireGW/+ sciatic nerves (Fig. 1C, 1D), but not in brain or spinal cord (Supplemental Fig. 1A).
Progression of neuropathy is associated with upregulation in sciatic nerves of CXCL10 (IP-10), CCL2 (MCP-1), and CCL5 (RANTES) (Supplemental Fig. 1B), three chemokines that have been linked with inflammatory neuropathies (11, 12). Multifocal areas of demyelination are seen on Luxol fast blue-stained sections of NOD.AireGW/+ sciatic nerve (Fig. 1E). Additionally, decreased density of myelinated axons is seen on toluidine blue-stained cross sections of sciatic nerve in neuropathic NOD.AireGW/+ mice (Fig. 1F). Electron microscopy of these same sciatic nerves demonstrates demyelination of individual axons in NOD.AireGW/+ mice (Fig. 1G). The symmetric nerve dysfunction, mononuclear cells infiltrating peripheral nerves, and multifocal demyelination recapitulate many of the features of CIDP.
In addition to developing autoimmune peripheral neuropathy, NOD.AireGW/+ mice also develop spontaneous autoimmune diabetes. The incidence of diabetes in NOD.AireGW/+ mice is the same as in NOD.WT mice (Supplemental Fig. 2A) (2). Histology of sciatic nerves from five diabetic (nonneuropathic) NOD.AireGW/+ mice did not demonstrate signs of immune infiltration in the sciatic nerve (Supplemental Fig. 2B, 2C). Sera from two diabetic NOD.AireGW/+ mice were negative for P0 autoantibodies by ELISA (data not shown). Rarely, diabetic NOD.AireGW/+ mice maintained on insulin also develop neuropathy. Histology of sciatic nerves from two mice with both diabetes and neuropathy demonstrated moderate immune infiltration (Supplemental Fig. 2C).
CD4+ T cells transfer autoimmune peripheral neuropathy and produce IFN-γ in peripheral nerves
To determine the cellular composition of immune cells infiltrating the PNS of NOD.AireGW/+ mice, we stained sciatic nerves from NOD.WT and NOD.AireGW/+ mice with Abs against CD4, CD8, F4/80, B220, and CD11c cell-surface markers. Within NOD.AireGW/+ nerve infiltrates, CD4+ Th cells and F4/80+ macrophages were more frequent than CD8+ cytotoxic T cells, B220+ B cells, or CD11c+ dendritic cells (Fig. 2A and data not shown). Few cells expressing these markers were seen in NOD.WT nerves.
To demonstrate a cellular basis for the pathogenesis of autoimmune peripheral neuropathy, we transferred spleen and lymph node cells from neuropathic NOD.AireGW/+ mice (between 19 and 34 wk of age) into immunodeficient Prkdcscid/scid NOD mice (NOD.scid mice). Histology of donor sciatic nerves demonstrated immune infiltration in all nerves tested (Fig. 2B, 2C). Eight out of eight recipients of whole spleen and lymph node cells (10 × 106 cells per recipient) from neuropathic NOD.AireGW/+ mice developed neuropathy by 10 wk posttransfer (Fig. 2D), suggesting that immune cells are a pathogenic entity.
To determine the relative importance of CD4+ helper and CD8+ cytotoxic T cell subsets, we transferred purified CD4+ and CD8+ populations into NOD.scid recipients. Similar to recipients of whole spleen and lymph node cells, 12 out of 12 recipients of CD4+ cells (7 × 106 cells per recipient) developed neuropathy by 10 wk posttransfer (Fig. 2D). This neuropathy incidence in recipients of 7 × 106 CD4+ cells was not significantly different from recipients of 10 × 106 whole spleen and lymph node cells. Thus, CD4+ cells are sufficient to transfer neuropathy.
None of the four recipients of 3 × 106 CD8+ cells were neuropathic up to 12 wk posttransfer (Fig. 2D). However, three out of seven recipients of 7 × 106 CD8+ cells developed neuropathy within 12 wk (Fig. 2D). The incidence of neuropathy in recipients of 7 × 106 CD8+ cells was significantly less than recipients of 10 × 106 whole spleen and lymph node cells (p = 0.008) and significantly less than recipients of 7 × 106 CD4+ cells (p = 0.0193). This finding suggests that CD8+ cells can also transfer neuropathy, but less effectively than CD4+ cells. Finally, eight out of eight recipients of both 7 × 106 CD4+ and 7 × 106 CD8+ cells combined developed neuropathy by 10 wk posttransfer. The incidence of neuropathy in recipients of both CD4+ and CD8+ cells was not significantly different from either recipients of 10 × 106 whole spleen and lymph node cells or from recipients of 7 × 106 CD4+ cells (Fig. 2D). Histological examination of sciatic nerves from recipients demonstrated immune infiltration within sciatic nerves in all mice with clinical neuropathy at the time of sacrifice (Fig. 2E).
We stained CD4+ T cells infiltrating NOD.AireGW/+ sciatic nerves for intracellular IFN-γ, IL-4, and IL-17 cytokines to determine the frequency of subsets producing these cytokines. IFN-γ production was seen in ∼40% of CD4+ T cells (Fig. 3A, 3B), and IL-4 and IL-17 production was rarely seen (Fig. 3B). No infiltrating CD4+ T cells were detected in NOD.WT sciatic nerves. A comparable percentage of CD4+ T cell subsets were seen in the spleen of NOD.AireGW/+ and NOD.WT mice (Fig. 3C).
Within the sciatic nerves, an average of 1700 IFN-γ–producing CD4+ T cells were isolated from each neuropathic NOD.AireGW/+ mouse (Fig. 3D). In contrast, <100 IL-4– and IL-17–producing CD4+ T cells were present on average in the sciatic nerves. Within the spleen, an average of 5.9 × 106 IFN-γ–producing CD4+ T cells was seen in NOD.AireGW/+ mice compared with 2 × 106 IFN-γ–producing CD4+ T cells in NOD.WT mice (Fig. 2G). Fewer than 1 × 106 IL-4– and IL-17–producing CD4+ T cells were seen in NOD.AireGW/+ and NOD.WT spleens. Taken together, these data are consistent with a strong Th1 effector response in the sciatic nerves in the model.
P0 is a major Aire-regulated PNS autoantigen
Serum autoantibodies from Aire-deficient mice have been successfully used to identify tissue-specific self-Ags that are targets of autoimmunity (9, 10). On indirect immunofluorescence, serum autoantibodies in neuropathic NOD.AireGW/+ mice react against sciatic nerves in a streaklike pattern (Fig. 4A). Immunoblot with sera from neuropathic NOD.AireGW/+ mice against whole sciatic nerve extract demonstrated an oligoclonal pattern of reactivity in NOD.AireGW/+ sera, which was predominantly against a 25–30-kDa Ag (Fig. 4B). This 25–30-kDa Ag was immunopurified using autoantibodies from sera of NOD.AireGW/+ mice to bind Ags from whole sciatic nerve extract and subjected to peptide mass fingerprinting. Microsequenced peptides derived from the 28-kD PNS-specific protein P0 with a confidence score approaching 100% (Fig. 4C).
To confirm the specificity of autoantibodies in NOD.AireGW/+ sera for P0, we performed a competition assay in which recombinant P0-MBP tag fusion protein was incubated with serum samples prior to immunoblotting. Titration of increasing concentrations of recombinant P0-MBP fusion protein into serum samples resulted in decreasing intensity of the 25–30-kDa band on immunoblot, whereas the equivalent concentrations of negative control protein (MBP tag alone) did not (Fig. 4D). Furthermore, autoantibodies in sera from NOD.AireGW/+ mice had significantly increased reactivity to P0-MBP fusion protein by ELISA than autoantibodies in sera from NOD.WT littermates (Fig. 4E).
NOD.AireGW/+ mice have a quantitative decrease in the expression of Aire-regulated tissue-specific self-Ags in mTECs (2). Because P0 is a major PNS autoantigen in NOD.AireGW/+ mice, we sought to determine whether P0 is quantitatively decreased in NOD.AireGW/+ mTECs. cDNA generated from sorted mTECs was used to interrogate the relative expression levels of tissue-specific self-Ags in NOD.AireGW/+ compared with NOD.WT mTECs. To validate our cDNA samples, we demonstrated that insulin 2 (Ins2), but not glutamic acid decarboxylase 67, is quantitatively decreased in NOD.AireGW/+mTECs compared with NOD.WT (Fig. 4F), as has previously been reported (4). Similar to Ins2, P0 expression in NOD.AireGW/+mTECs was ∼10% of NOD.WT expression (Fig. 4F).
The decreased expression of P0 in NOD.AireGW/+mTECs led us to hypothesize that autoimmune peripheral neuropathy results from defective negative selection of P0 specific T cells in the thymus. To test this hypothesis, splenocytes from neuropathic NOD.AireGW/+ mice (between 20 and 22 wk of age) were incubated with irrelevant HEL peptide or P0 peptide 180–199. Splenocytes from NOD.AireGW/+ and WT mice proliferated to the same extent in response to HEL peptide (Fig. 4G, left panel). At all three concentrations of P0 180–199, in contrast, NOD.AireGW/+ immune cells proliferated to a significantly greater degree than WT immune cells (Fig. 4G, right panel).
Autoreactivity to P0 in APS1 patients
CIDP was recently reported in two unrelated APS1 patients with confirmed mutations in Aire (5). One patient harbored compound heterozygous Aire mutations (R257X/R203X), and the other patient was homozygous for an Aire R139X mutation. The diagnosis of CIDP was based on: 1) weakness, sensory loss, and absent reflexes; 2) evidence of demyelination; and 3) the presence of infiltrating CD4+ T cells and macrophages on nerve biopsy. Surprisingly, autoantibodies against a number of neuropathy-associated Ags were negative (5). Autoantibodies against P0, however, were not previously tested in these patients.
Indirect immunofluorescence using sera from these two patients demonstrated autoantibodies that bound sciatic nerve in a streaklike pattern (Fig. 5A) reminiscent of that seen in NOD.AireGW/+ mice. We then employed a radioligand binding assay to test for P0-specific autoantibodies in the sera from these two patients. Although none of the sera from the healthy controls reacted against P0, sera from the two APS1 patients with CIDP were positive for autoantibodies against P0 (Fig. 5B). Additionally, 2 out of 12 patients with APS1 but without a clinical history of neuropathy also demonstrated serum reactivity to P0, raising the possibility that these patients may be at risk for developing CIDP.
In this study, we show that mice and humans with defective Aire function share loss of tolerance to P0, which correlates with the development of spontaneous autoimmune peripheral neuropathy. We show for the first time, to our knowledge, that P0 is an Aire-regulated Ag in mTECs and that decreased P0 expression in mTECs is linked to increased immune cell reactivity toward P0 in the periphery. Finally, we show that CD4+ T cells are sufficient to transfer neuropathy and that the autoimmune response in the peripheral nerves is dominated by a Th1 effector response.
The spontaneous autoimmune peripheral neuropathy in NOD.AireGW/+ mice shares a number of features with human CIDP. First, CD4+ T cells and F4/80 macrophages are abundant immune cell types in CIDP patient nerve biopsies (13, 14). Similarly, CD4+ T cells and F4/80-expressing macrophages are also prevalent in the immune infiltrates of NOD.AireGW/+ sciatic nerves (Fig. 2A). Second, CD4+ Th cells in CIDP patients are skewed toward the production of IFN-γ (15). Similarly, ∼40% of CD4+ Th cells in the sciatic nerve infiltrates of neuropathic NOD.AireGW/+ mice produce IFN-γ after in vitro stimulation (Fig. 3A, 3B), and <1% produce IL-17 or IL-4. Third, increased expression of the chemokine CXCL-10 in sural nerve biopsies is associated with CIDP (12). In neuropathic NOD.AireGW/+ mice, CXCL-10 expression is also increased in the sciatic nerves (Supplemental Fig. 1B). Finally, demyelination is a hallmark of CIDP, and extensive demyelination is characteristic of sciatic nerves from neuropathic NOD.AireGW/+ mice (Fig. 1E–G). Thus, autoimmune peripheral neuropathy in NOD.AireGW/+ mice resembles human CIDP in immune response and pathological changes in affected nerves.
The identification of P0 as a peripheral nerve autoantigen in Aire deficiency is significant because dominant protein Ag(s) in human autoimmune peripheral neuropathy are not well defined. α and β tubulin, P0, P2, PMP22, and connexin 32 have all been reported to be candidate Ags targeted by the autoimmune response in CIDP (reviewed in Ref. 16). P0 is a peripheral nerve-specific protein that makes up the majority of myelin in the PNS (17). Loss of P0 due to genetic mutations (rather than autoimmune destruction) can result in Charcot Marie Tooth type IB, a heritable demyelinating peripheral neuropathy (Online Mendelian Inheritance in Man 118200). In addition, P0 can induce experimental autoimmune neuritis upon immunization with adjuvant in mice (18) and is an autoantigen in other models of autoimmune peripheral neuropathy (11, 19).
We demonstrate in this study that P0 is an Aire-regulated tissue-specific self-Ag in the thymus and that immune cells in the spleen of Aire-deficient mice have increased reactivity toward P0. These data delineate a model a model in which tolerance toward peripheral nerves is maintained by Aire-mediated upregulation of P0 Ag in the thymus. Negative selection of P0-specific T cells in the thymus prevents escape of P0-reactive T cells into the periphery, which can cause autoimmune peripheral neuropathy. In this study, we did not test whether T cells infiltrating the sciatic nerves of NOD.AireGW/+ mice are also reactive to P0 because of the low number of immune cells that we are able to isolate from sciatic nerves. Further study, including the generation of Ag-specific clones from infiltrating T cells, will be needed to define the myelin-specific T cell response in the model.
We found that CD4+ T cells were sufficient to transfer neuropathy to immunodeficient mice in a 6–10-wk time frame. Previously, P0-specific T cell lines have been shown to transfer neuropathy into Lewis rats by 6 d (20). The slower kinetics of neuropathy development in our study can potentially be due to multiple factors, including the polyclonal repertoire of the T cells, species-specific differences between mice and rats, and lack of B cells in NOD.scid recipients.
A caveat of this study is the small number of patients with APS1 and CIDP. APS1 is a rare condition with a frequency of much <1 in 100,000 in the North American population. In addition, it appears that CIDP is an infrequent manifestation within APS1 subjects. We have attempted to identify additional subjects with APS1 and biopsy-proven CIDP within larger patient collections available in Scandinavia, but were unable to identify additional patients. Importantly, however, the two patients in this study are unrelated and have different Aire mutations, which strengthens the possibility that our findings are associated with Aire.
Finally, findings in this study may inform the development of more specific and effective immune therapies for autoimmune peripheral neuropathy. Current treatments consist primarily of glucocorticoids, i.v. Ig, and plasmapheresis, which result in nonspecific immunosuppression (21). Furthermore, these treatments are ineffective in one third of patients (22). The demonstration in this study that CD4+ T cells are sufficient to transfer disease and that infiltrating CD4+ T cells produce IFN-γ suggests that therapies targeting this cell lineage or Th1 subsets may be efficacious in treating CIDP. Also, the identification of P0 as an important autoantigen in Aire-mediated autoimmune peripheral neuropathy may help guide the development of Ag-based therapies, such as Ag-specific DNA vaccination or infusion of Ag-coupled cells (23). Such therapeutic approaches that address underlying pathogenesis may be more efficacious in treating inflammatory neuropathies.
The authors have no financial conflicts of interest.
This work is supported by the National Institutes of Health, the Neuropathy Association (to M.S.A.), and the Burroughs Wellcome Fund (to M.S.A.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- autoimmune regulator
- autoimmune polyendocrinopathy syndrome type 1
- chronic inflammatory demyelinating polyneuropathy
- hen egg lysozyme
- insulin 2
- maltose binding protein
- medullary thymic epithelial cell
- myelin protein zero
- peripheral nervous system
- Received February 9, 2012.
- Accepted March 12, 2012.
- Copyright © 2012 by The American Association of Immunologists, Inc.