Experimental autoimmune encephalomyelitis (EAE) is a valuable model for studying immunopathology in multiple sclerosis (MS) and for exploring the interface between autoimmune responses and CNS tissue that ultimately leads to lesion development. In this study, we measured gene expression in mouse spinal cord during myelin oligodendrocyte gp35–55 peptide–induced EAE, using quantitative RT-PCR, to identify gene markers that monitor individual hallmark pathological processes. We defined a small panel of genes whose longitudinal expression patterns provided insight into the timing, interrelationships, and mechanisms of individual disease processes and the efficacy of therapeutics for the treatment of MS. Earliest transcriptional changes were upregulation of Il17a and sharp downregulation of neuronal and oligodendrocyte marker genes preceding clinical disease onset, whereas neuroinflammatory markers progressively increased as symptoms and tissue lesions developed. EAE-induced gene-expression changes were not altered in mice deficient in IKKβ in cells of the myeloid lineage compared with controls, but the administration of a selective inhibitor of soluble TNF to mice from the day of immunization delayed changes in the expression of innate inflammation, myelin, and neuron markers from the presymptomatic phase. Proof of principle that the gene panel shows drug screening potential was obtained using a well-established MS therapeutic, glatiramer acetate. Prophylactic treatment of mice with glatiramer acetate normalized gene marker expression, and this correlated with the level of therapeutic success. These results show that neurons and oligodendrocytes are highly sensitive to CNS-directed autoimmunity before the development of clinical symptoms and immunopathology and reveal a role for soluble TNF in mediating the earliest changes in gene expression.
Multiple sclerosis (MS) is a chronic inflammatory disease of the CNS that is characterized by complex pathological processes, notably blood–brain barrier breakdown, neuroinflammation, immune cell infiltration, demyelination, and progressive neurodegeneration, as well as repair processes, such as remyelination (1). It is an autoimmune disease associated with adaptive immune responses to myelin Ags, as supported by current genetic evidence (2) and the effectiveness of therapies, such as anti-α4 integrin (3) and anti-CD20 (4) mAbs, in decreasing relapse incidence in relapsing-remitting MS patients. However, little is known about the sequence, timing, and interrelationships between different pathological processes that lead to the histological hallmarks of disease, and major questions exist as to whether neuroinflammation and demyelination processes are primary or secondary in disease development (5).
Experimental autoimmune encephalomyelitis (EAE) is an inducible animal model for MS in which myelin-specific autoimmune T cells trigger histopathological lesions similar to those seen in early MS. EAE has been used extensively for investigating mechanisms of autoimmune-triggered pathology in the CNS and for identifying and validating therapeutic targets for the treatment of MS (6). Clinical symptoms are generally taken to mark the onset of disease, because this coincides with autoimmune effector CD4+ T cell infiltration into the CNS parenchyma (7, 8). However, recent reports that structural and functional changes take place within CNS tissues before the development of symptoms have changed the conventional view that disease is a result of bystander immune damage to myelin and axons. In particular, enhanced glutamate transmission, synaptic degeneration, and dendritic spine loss occur in the striatum during the presymptomatic phase of disease (9), and changes in transcripts associated with the innate immune response and neuronal functions start several days before clinical onset, as detected by microarray analysis of mouse EAE tissues (10). Further, in vivo imaging studies elegantly demonstrate intraluminal crawling of encephalitogenic T cells (8) and perivascular clustering of microglia (11) prior to the onset of clinical symptoms and histopathological features. Together, these data suggest that subclinical vascular and neuronal alterations take place in the CNS tissue during the development of EAE that might predispose it to immunopathology.
These observations led us to hypothesize that differential expression of CNS cell-selective gene markers within the mouse spinal cord during the development of EAE might allow us to monitor individual pathogenic processes in a temporal and spatial manner and provide a tool for probing the effectiveness of experimental MS therapeutics in a preclinical setting. By following the expression of genes marking major cell targets of pathology in MS and EAE at different stages of EAE in mice, using quantitative RT-PCR, we defined a small panel of genes that sensitively monitored early changes in oligodendrocytes and neurons and innate immune activation. In this way, we show that the soluble form of TNF participates in the initiation of neuroinflammatory, myelin, and neuronal changes. Furthermore, proof of principle that the gene panel shows drug-screening potential was obtained using mRNA from tissues from mice treated with the approved MS drug glatiramer acetate (GA), where normalization of gene marker expression directly correlated with the effectiveness of disease treatment. These results suggest that expression analysis of marker genes in preclinical models is a useful tool for further investigating and understanding mechanisms of neuropathology and for predicting the effectiveness and penetration of novel therapeutics intended for the treatment of MS.
Materials and Methods
Mice with a selective deletion of IKKβ in macrophage/microglia (mIKKβKO) were generated by crossing IKKβF/F mice (12, 13) (kindly donated by Michael Karin, University of California at San Diego, La Jolla, CA) with mice that express a CD11b promoter-driven Cre recombinase specifically in cells of myeloid origin, including microglia (14) (V. Kyrargyri and L. Probert, manuscript in preparation). Female wild-type, mIKKβKO and IKKβF/F mice (all in C57BL/6 background), aged between 8 and 12 wk, were used to induce EAE. Animals were bred and maintained under specific pathogen–free conditions in the Experimental Animal Facility of the Hellenic Pasteur Institute. All animal procedures were approved by institutional review boards and national authorities and conformed to European Union guidelines.
EAE induction, evaluation, and tissue sampling
Myelin oligodendrocyte gp35–55 peptide (MOG; kindly provided by Theodore Tselios, University of Patras, Rio, Greece) was synthesized as previously described (15). EAE was induced by s.c. tail base injection of 30 μg MOG emulsified in CFA (Sigma-Aldrich, St. Louis, MO) supplemented with 400 μg H37Ra Mycobacterium tuberculosis (Sigma-Aldrich) on day 0. Mice also received i.p. injections of 200 ng pertussis toxin (PTx; Sigma-Aldrich) on days 0 and 2 postimmunization. Mice were assessed daily and scored for clinical signs of disease: 0, normal; 1, limp tail; 2, hindlimb weakness; 3, hindlimb paralysis; 4, forelimb paralysis; and 5, moribund or dead (0.5 gradations represent intermediate scores). Moribund animals were sacrificed. Control mice were either naive (referred to as “non-EAE”) or injected with CFA and PTx only (referred to as “adjuvant”). Mice were housed under the same controlled environmental conditions and allowed free access to food and water throughout the experiment. Animals were sacrificed at different time points of disease. For the preonset time point, healthy mice were sacrificed on the day the first mouse in the EAE control group showed a clinical symptom (days 11–13). For the disease onset and peak, we observed a slight variability in disease development among mice (days 11–15), so the day of sacrifice was defined for individual mice as either the day of the first clinical symptom or the first day after the maximum score, respectively. For the chronic time point, mice were sacrificed at day 30. Whole spinal cords were isolated fresh for RNA extraction or following perfusion-fixation for histopathological analysis. Non-EAE mice used as baseline controls for all experiments were sacrificed on day 0.
In vivo treatments
Groups of C57BL/6 mice were injected s.c. on the hind flanks with GA (COPAXONE; Teva Neuroscience; 25 mg/kg) emulsified in IFA on day 7 preimmunization, on the day of immunization with MOG for the induction of EAE, and then every 7 d until the end of the experiment (16). A nontreated EAE group was used as disease control. Other groups were treated with twice-weekly s.c. injections of XPro1595 (Xencor; 10 mg/kg) (17, 18), etanercept (Amgen; 10 mg/kg) (19), or saline vehicle as control starting on the day of immunization.
Total RNA isolation and quantitative RT-PCR
Supplemental Table I) were Il6 (Mm_Il6_1_SG), Cyfip2 (Mm_Cyfip2_1_SG), App (Mm_App_1_SG), Il6st (Mm_Il6st_1_SG), Ctsz (Mm_Ctsz_1_SG), B2m (Mm_B2m_2_SG), Tmem1 (Mm_Tmem1_2_SG), Ptma (Mm_Ptma_1_SG), Ntn1 (Mm_Ntn1_1_SG), Ntn2l (Mm_Ntn3_1_SG), Unc5a (Mm_Unc5a_1_SG), Pdgfra (Mm_Pdgfra_1_SG), Lingo (Mm_Lingo1_1_SG), Ccl2 (Mm_Ccl2_1_SG), Msn (Mm_Msn_1_SG), Twist1 (Mm_Twist_1_SG), Nono (Mm_Nono_1_SG), Lin7c (Mm_Lin7c_1_SG), Evc2 (Mm_Evc2_1_SG), Ywhah (Mm_Ywhah_1_SG), and Klf7 (Mm_Klf7_1_SG).RNA) + B. The coefficient of determination (R2) was >0.99. Gene-expression analyses were performed using two housekeeping genes (Gusb and Gapdh) independently for normalization; similar results were obtained. All results were analyzed using LightCycler software version 3.5 (Roche). QuantiTect Primer Assays (QIAGEN) were used for Il17a (Mm_Il17a_1_SG), H2Ab1 (Mm_H2-Ab1_1_SG), Cxcl16 (Mm_Cxcl16_1_SG), Mbp (Mm_Mbp_1_SG), Olig2 (Mm_Olig2_1_SG), Nrg1 (Mm_Nrg1_11_SG), Snap25 (Mm_Snap25_2_SG), Grin1 (Mm_Grin1_2_SG), Tnf (Mm_Tnf_1_SG), Bmi1 (Mm_Bmi1_1_SG), Ninj1 (Mm_Ninj1_1_SG), Gapdh (Mm_Gapdh_3_SG), and Gusb (Mm_Gusb_1_SG). Additional QuantiTect Primer Assays (QIAGEN) used for screening the gene candidates (
Mice were transcardially perfused with ice-cold PBS, followed by 4% paraformaldehyde in PBS under deep anesthesia. CNS tissues were postfixed in the same fixative overnight at 4°C, embedded in paraffin, and processed for standard histopathological and immunohistochemical analyses. Inflammation was visualized by H&E, demyelination was visualized by Klüver-Barrera (Luxol fast blue), and axonal damage was visualized by Bielchowsky’s silver staining. Immunohistochemistry was performed as previously described (20) using mouse mAb to neurofilament H nonphosphorylated (SMI-32) (1/1000; Covance), rabbit anti-ionized calcium binding adaptor molecule 1 (anti-Iba1) (1/500; WAKO), rabbit anti-CD3 Ab (1/300; Biodynamics), or rat anti–Ly-6G mAb (NIMP-R14) (1/100; Abcam); appropriate biotinylated secondary Abs; avidin-biotin complex; and 3, 3′-diaminobenzidine (all from Vector Laboratories). Ag retrieval was done for CD3 and Iba1 in 10 mM EDTA buffer (pH 8.5) and for Ly-6G in 10 mM citrate buffer (pH 6) for 1 h at 95–100°C in a household food steamer.
All statistical analyses were performed with Sigma Stat 3.5 for Windows. All data are given as mean ± SEM. For quantitative RT-PCR analyses, the Student t test and Mann-Whitney rank-sum test were used to identify significant changes in gene expression between different groups during the development of EAE. The Mann–Whitney rank-sum test was performed to determine significant differences in clinical scores between nontreated and GA-treated mice at different time points after MOG immunization (Fig. 5). The p values < 0.05 were considered statistically significant.
Differential expression of selected marker genes in spinal cord during MOG-induced EAE predicts the onset of clinical symptoms
To investigate molecular changes taking place in the spinal cord during autoimmune demyelination and to identify gene-expression markers that would predict and monitor pathogenic and tissue repair processes independently of clinical symptoms, we screened gene expression in the spinal cord of mice during the development of MOG-induced EAE by quantitative RT-PCR. We originally checked the expression of >30 gene candidates that were selected according to one or more of the following criteria: 1) identification in a comparative microarray study of brain from mouse models of MS, cerebral stroke, and Alzheimer’s disease (21), particularly genes showing strong dysregulation early in pathology and, therefore, having possible predictive value; 2) dysregulation in microarray studies of MOG-induced EAE spinal cord (22, 23) or MS lesions (24–29); 3) involvement in CNS cell compartments targeted in EAE and MS (e.g., myelin, neurons, synapses); and 4) established functional contribution to EAE or MS pathogenesis (Supplemental Table I).
Expression patterns of gene candidates were measured in spinal cord mRNA samples taken at different time points during the development of MOG-induced EAE in C57BL/6 mice, as determined by clinical symptoms. Thus, EAE was induced in mice by immunization with MOG, and whole spinal cord was collected at preonset, onset, peak, and, in some experiments, chronic disease (Supplemental Table I). Expression patterns of the original gene candidates during the different phases of disease could be generally classified into three main categories. 1) Genes with upregulated expression levels compared with non-EAE controls. All genes in this category have known roles in inflammation and immune responses, specifically B2m, Ccl2, Cxcl16, H2-Ab1, Il6, Il17a, and Tnf. These results are consistent with previous microarray data for MOG-induced EAE spinal cord that show upregulated expression of B2m and MHC class I and II genes (10, 22, 23, 30), Cxcl16 and Tnf (10), Il6 and Ccl2 (30), as well as for MS lesions showing upregulated expression of B2M and MHC class I and II genes (23, 25, 26), TNF (29), and IL17A (25). 2) Genes with downregulated expression levels compared with non-EAE controls that correlated with severity of symptoms. These genes include those associated with oligodendrocytes (Lingo1, Mbp, Olig2), neurons (App, Grin1, Il6st, Klf7, Lin7c, Ninj1, Nrg1, Ntn1, Ntn3, Ptma1, Snap25, Unc5a), or plasticity and repair processes (Bmi1, Olig2, Pdgfra). Several homologs of genes associated with human CNS diseases also showed this pattern of expression (Cyfip2, Evc2, Tmem1, Ywhah). Previous microarray studies in MS reported a reduction in genes encoding myelin proteins, including myelin basic protein in normal-appearing white matter and lesions from MS patients (24–26, 28). Expression of neuronal genes is also generally decreased in MS (25), although variations in SNAP25 expression are seen among patients (27, 28), and decreased expression is seen in MOG-induced EAE spinal cord at disease peak (22). 3) Genes with undetectable changes in expression compared with non-EAE controls (Ctsz) or whose deregulation correlated inconsistently with the severity of clinical symptoms (Msn, Twist1).
Based upon these results, we narrowed down the number of genes to a mini-panel of genes representing major pathological processes in MS and EAE, demyelination, inflammation, and neuron damage and also showed strong and reproducible expression patterns during disease progression, as well as several genes that could predict disease onset (Fig. 1). In our initial experiments, we compared the expression of the selected genes at all time points against that of the housekeeping gene Gusb (Fig. 1A). To minimize the possibility that the choice of Gusb for normalization introduced bias in our expression results, in subsequent experiments we independently compared marker gene expression for key markers and time points (non-EAE, preonset, and onset) with that of Gapdh (Fig. 1B) and Gusb (Supplemental Fig. 1); similar results were obtained. As expected from previous studies, upregulated expression of inflammatory gene markers Il17a, H2-Ab1, and Cxcl16 was detectable in the spinal cord from the presymptomatic time point and increased as disease progressed (Fig. 1, Supplemental Fig. 1). Unexpectedly, expression of two myelin gene markers, Mbp and Olig2, as well as Bmi1, a marker of cell plasticity, was markedly downregulated at the presymptomatic time point, and expression continued to decrease until the peak of disease (Fig. 1, Supplemental Fig. 1, data not shown). Mbp encodes myelin basic protein, a major myelin component, whereas Olig2 is a transcription factor that controls oligodendrocyte development (31). Similarly, expression of all selected neuron-specific genes, Nrg1, Ninj1, Snap25, and Grin1, was markedly downregulated at the presymptomatic time point and, as with the myelin genes, expression continued to decrease until the peak of disease (Fig. 1, Supplemental Fig. 1). Snap25, Grin1, and Nrg1 encode proteins that are involved in synaptic functions.
To confirm that differential expression of the selected genes was specific for EAE and not a result of the immunization protocol (i.e., among genes regulated by CFA and PTx) (10), we next compared their expression in spinal cords isolated from mice immunized with adjuvant, with or without MOG Ag. None of our selected genes showed expression changes in adjuvant controls compared with non-EAE baseline controls, whereas those in EAE mice showed the characteristic pattern described above when normalized against Gapdh (Fig. 2) or Gusb (Supplemental Fig. 1). Thus, expression of inflammatory markers progressively increased from preonset in MOG-immunized mice but not adjuvant-immunized mice. Further, expression of Mbp, Snap25, Nrg1, and Grin1 was downregulated from the presymptomatic time point in MOG-immunized mice but not in adjuvant-immunized mice. These results show that the gene-expression changes are disease specific and not induced by adjuvant effects.
Differential expression of marker genes precedes lesion development in MOG-induced EAE
To complement the results of the gene-expression analysis, neuropathological analysis was performed on individuals from the EAE and non-EAE groups that showed equivalent clinical symptoms to mice used for gene-expression analysis. Spinal cord sections were stained with H&E for inflammation, Klüver-Barrera (Luxol fast blue) for demyelination, and Bielschowsky’s silver stain for axonal damage. Prior to the onset of clinical symptoms, no spinal cord lesions were detectable using standard histopathological methods (Fig. 3, second column from left), but Iba1+ microglia with branched morphology were distributed throughout the parenchyma, indicative of a neuroinflammatory response (Fig. 4A, second column from left); CD3+ T cells (Fig. 4B, 4C, arrow) and Ly-6G+ neutrophils (Fig. 4D, arrow) were present in the leptomeninges and associated with blood vessel walls. At the onset of clinical symptoms, typical EAE lesions composed of focal immune cell infiltrates (Fig. 3A, third column from left, arrow), demyelination (Fig. 3B, arrow), and axonal damage (Fig. 3C, arrow) were observed in the white matter of the spinal cord. At this time point, strongly immunostained, branched Iba1+ microglia were again distributed throughout the parenchyma (Fig. 4A, third column from left), whereas CD3+ T cells (Fig. 4B, 4C, arrows) and Ly-6G+ neutrophils (Fig. 4D, arrow), in addition to being present in the leptomeninges and associated with blood vessels, were located within lesions. At the disease peak, the lesions were more extensive and axonal damage was more severe compared with lesions at disease onset (Fig. 3, right panels, arrows), as shown by immunostaining with SMI-32, a marker for nonphosphorylated neurofilaments (Fig. 3D, arrows). In normal CNS, SMI-32 stains cell bodies and dendrites but not axons, which are heavily phosphorylated. During conditions of neurodegeneration, axons also become positive for nonphosphorylated neurofilaments; therefore, SMI-32 staining of axons in the white matter of the EAE spinal cord indicates axonal damage. Numerous activated Iba1+ microglia/macrophages with phagocytic morphology (Fig. 4A, right panels), CD3+ T cells (Fig. 4B, 4C, arrow), and Ly-6G+ neutrophils (Fig. 4D, arrow) were distributed throughout the parenchyma. These results show that changes in the expression of marker genes coincide with immune cell recruitment to leptomeninges and blood vessel walls, as well as microglia activation in the spinal cord, prior to the development of inflammatory lesions and clinical EAE.
Prophylactic treatment with GA prevents deregulated expression of selected gene markers as correlated with amelioration of clinical symptoms
To investigate whether the gene panel could be useful for evaluating the effectiveness of experimental therapies, we measured differential expression of the selected genes during development of MOG-induced EAE in mice that had been treated prophylactically with GA. We chose to use GA because it is effective in ameliorating MOG-induced EAE in both prophylactic and therapeutic protocols, shows protective properties in CNS tissues (32), and is an established drug for the treatment of MS (33). Most mice treated prophylactically with GA by s.c. injection every 7 d, starting from 7 d preimmunization with MOG, showed reduced disease incidence, delayed onset, and significantly milder clinical scores compared with those in the nontreated group (Fig. 5, Table I), although not all mice benefited from GA treatment. Therefore, we correlated spinal cord gene-expression results with the clinical score of each individual mouse at the time of sacrifice (Fig. 6) and compared gene expression in groups of mice at the onset and peak of disease with non-EAE controls to be able to evaluate data statistically (Supplemental Fig. 2).
As described above, Il17a and innate immune markers were upregulated from the presymptomatic time point, and expression of the latter continued to increase as clinical symptoms developed (Fig. 1). In mice that were treated successfully with GA (score 0), EAE-induced upregulation of Il17a, H2-Ab1, Cxcl16, and Tnf expression was prevented at all time points analyzed, as shown in individual mice at onset and peak (Fig. 6), as well as in groups of mice at disease peak compared with non-EAE controls (Supplemental Fig. 2). In mice that were not treated successfully with GA (scores ≥ 0.5), EAE-induced upregulation of the inflammatory genes was not prevented (Fig. 6).
We next examined the effect of GA treatment on expression of oligodendrocyte and neuronal marker genes. As described above, both oligodendrocyte (Mbp, Olig2) and neuronal (Nrg1, Ninj1, Snap25, Grin1) marker genes were significantly downregulated from the presymptomatic phase of disease and remained at low levels during disease onset and peak (Fig. 1). Interestingly, in mice that were treated successfully with GA, EAE-induced downregulation of both oligodendrocyte (Mbp, Olig2) and neuronal (Nrg1, Snap25, Ninj1) genes was prevented at disease onset and peak, as shown in individual mice (Fig. 6) and groups of mice compared with non-EAE controls (Supplemental Fig. 2). In mice that were not treated successfully with GA, expression of these genes was downregulated to similar levels as in nontreated EAE mice (Fig. 6).
EAE-induced changes in marker gene expression are not prevented in mice with macrophage/microglial IKKβ deficiency
To study the mechanisms underlying altered oligodendrocyte and neuron gene expression during the early stages of EAE, we tested the contribution of cells of the myeloid lineage, including microglia, which are known to play a critical role in EAE pathogenesis (34). Because the proinflammatory activities of microglia in EAE (35) and in a model of kainic acid–induced neuronal cell death (36) are mediated via activation of the transcription factor NF-κB, we induced EAE in mice with a selective deletion of IKKβ in macrophage/microglia (mIKKβKO) and in IKKβF/F control mice and measured gene expression at different time points. The pattern of marker gene expression in IKKβF/F control mice was similar to that in C57BL/6 mice at all time points of EAE tested (Figs. 1, 7). Further, no differences in expression of marker genes tested were measured in mIKKβKO mice compared with IKKβF/F controls at any of the time points studied (Fig. 7), indicating that macrophage/microglial NF-κB activity is not necessary for the altered expression of our selected innate immune, oligodendrocyte, and neuron marker genes during EAE onset.
EAE-induced changes in neuronal, oligodendrocyte, and innate immune genes are delayed by selective inhibition of soluble TNF
To investigate the possibility that soluble factors mediate the downregulation of oligodendrocyte and neuronal gene expression from early in EAE, we administered a dominant-negative TNF analog, XPro1595, peripherally to mice to selectively block the effects of soluble TNF (17, 37) and compared its effects with those of blocking both soluble and transmembrane TNF using etanercept, a TNFR II–IgG1 Fc fusion protein (19). The increased expression of Il17a, as well as a small increase in the expression of innate immune marker genes, H2-Ab1 and Cxcl16 as detectable against Gapdh (Fig. 8), but not Gusb expression (Supplemental Fig. 3), in the spinal cord of vehicle-treated controls at preonset, was prevented in XPro1595-treated mice. Importantly, XPro1595, but not etanercept, also reduced the sharp EAE-induced downregulation of myelin and neuronal marker gene expression. Thus, reduced expression of Mbp, Olig2, Nrg1, Snap25, and Grin1 at preonset and of Olig2 and Nrg1 at onset was prevented by XPro1595 (Fig. 8, Supplemental Fig. 3), but not etanercept (Supplemental Fig. 3), compared with vehicle-treated groups. The protective effect of XPro1595 indicates that soluble TNF plays a role driving changes in neuron, oligodendrocyte, and innate immune gene expression from very early stages of EAE, prior to the onset of clinical symptoms. The absence of protection by etanercept, which also blocks soluble TNF, points to a possible counteracting protective effect of transmembrane TNF or the differential ability to access CNS tissues during the presymptomatic phase of disease.
The hallmark pathological features of MS, blood–brain barrier damage, neuroinflammation, demyelination, and neurodegeneration, are intimately interrelated, but the disease-initiating events are not known, and details of how individual pathological processes are causally related remain unclear. EAE is an excellent tool for investigating mechanisms relevant to MS immunopathology because the initiating trigger, T cell–mediated immune responses to CNS Ags, is known, and the downstream effector cascade results in a disease that recapitulates many clinical and pathophysiological features of MS (6). In this study, we wished to define a small number of genes whose longitudinal expression patterns in spinal cord would provide insight into the timing and interrelationships of individual disease processes during the development of EAE and allow us to investigate pathogenic mechanisms and evaluate the effectiveness and penetrance of experimental therapies. From >30 original gene candidates, chosen mainly for the cell selectivity of their expression and/or their strongly deregulated expression from early time points in mouse models of CNS inflammatory conditions (21), we defined a small panel of genes whose expression patterns allowed us to follow changes in neurons, oligodendrocytes, and neuroinflammation in the spinal cord at distinct stages of EAE and under different experimental conditions. Our results reveal several important new aspects of EAE that could be relevant for our understanding of MS pathogenesis: 1) marked downregulation of neuron and oligodendrocyte genes occurs on a background of innate immune activation before immune cell infiltration of CNS parenchyma and the development of clinical symptoms, suggesting that alterations in these cells may be a primary feature of disease; 2) EAE-induced changes in gene expression were not altered in mice conditionally deficient in IKKβ in cells of the myeloid lineage, indicating that microglial NF-κB activity is not required for mediating them; 3) peripheral administration of XPro1595 to mice from the day of immunization with MOG delayed changes in oligodendrocyte, neuron, and inflammatory gene expression, revealing a role for TNF as a soluble mediator of early CNS changes during the presymptomatic phase of EAE; and 4) normalization of gene marker expression in the spinal cord of EAE mice can be used to evaluate the efficacy of therapeutics for MS.
Among the first detectable changes in the spinal cord after immunization with MOG was increased expression of Il17a, which was used as a marker for effector T cell infiltration, and sharply reduced expression of all myelin, neuronal, and cell plasticity marker genes tested, including those selected for the final screening panel. These changes preceded the onset of clinical symptoms and the appearance of typical inflammatory and demyelinating lesions in the spinal cord, at a stage when CD3+ T cells and other immune cells, such as neutrophils, showed restricted localization to the meninges and vasculature (38). The expression of Il17a was transient, with mRNA levels peaking at onset and dropping at peak clinical score, a pattern consistent with the reported immigration of autoreactive T cells prior to the onset of clinical symptoms, as well as their subsequent regression (39). IL-17A is the signature cytokine of Th17 cells, which, together with IFN-γ–producing Th1 cells, constitute the pathogenic effector T cells that induce EAE (40). The possibility that Il17a is also expressed by infiltrating neutrophils cannot be excluded. The expression of innate immune markers, such as H2-Ab1, Cxcl16, and Tnf, steadily increased as clinical disease progressed. A previous microarray study of spinal cord from NOD mice by Baranzini et al. (10) showed that adjuvant induces early nonspecific changes in a number of innate immune genes, but none of our chosen gene markers was regulated in adjuvant controls. Our results reveal a clear sequence in gene-expression changes in the spinal cord during MOG-induced EAE, starting with downregulation of neuron- and oligodendrocyte-associated genes against a background of meningeal inflammation and enhanced Il17a, followed by a steady increase in neuroinflammatory gene expression correlating with glial cell activation and lesion development.
The expression of myelin, neuronal, and plasticity markers was rapidly reduced prior to the onset of clinical symptoms and remained shut down at all subsequent time points tested, indicating that neurons, oligodendrocytes, and, perhaps, precursor cells react rapidly to myelin-specific autoimmune responses prior to infiltration of effector T cells into the CNS. Our finding of early downregulated expression of neuronal genes is consistent with a previous study (10) in NOD mice, which described early downregulation of genes involved in locomotory behavior and potassium transport. As a possible correlate in MS, levels of the amino acid N-acetylaspartate, which is primarily localized to neurons in the adult CNS, were significantly lower in normal-appearing white matter of MS patients compared with controls, as seen by magnetic resonance spectroscopy (41). Similarly for myelin, a rapid downregulation in the expression of Plp1, which encodes the myelin component proteolipid protein, was observed in lesion and nonlesion areas of spinal cord in Theiler’s virus–induced encephalomyelitis, another mouse model for MS (42). Evidence that oligodendrocytes and neurons might play primary roles in the development of neuroinflammation and neurodegeneration in the CNS is increasing. In humans, mutations in the oligodendrocyte gene PLP1 are responsible for Pelizaeus–Merzbacher leukodystrophy (43). Also, altered expression of Mbp in mice causes axonal abnormalities without necessarily affecting the structure of myelin (44). The altered expression of oligodendrocyte genes also can enhance neuroinflammation, as shown in mice conditionally deficient for the peroxisome gene peroxin-5 in oligodendrocytes (45). Finally, downregulation of Olig2, a master regulator of oligodendrocyte development (31), might compromise myelin repair in later stages of EAE. Overall, reduced expression of Mbp and other myelin genes from very early in EAE might enhance the susceptibility of both myelin and axons to immune-mediated damage and limit the potential for remyelination in later stages of disease.
The reduction in expression of neuronal genes can reflect altered transcription from neuronal cell bodies located within the spinal cord and, in the case of genes like Grin1, whose transcripts are enriched in the synaptic neuropil (46), a loss of dendrites and synapses. Three of the neuronal gene markers chosen in this study–Snap25, Grin1, and Nrg1–encode proteins that are critical for the maintenance of synaptic transmission and normal structure of nerve terminals. Specifically, Snap25 encodes synaptosomal-associated protein 25 and plays a key role in exocytosis of synaptic vesicles (47), Grin1 encodes a basic subunit of the NMDA glutamate receptor, and Nrg1 encodes Neuregulin 1, which performs multiple functions in neurons and is a strong candidate gene for schizophrenia (48). Reduced expression of these genes during EAE would be expected to alter neuronal functions and possibly susceptibility of neurons to damage. Previous studies support a functional involvement of neuronal changes in the development of EAE. Blockade of glutamate signaling with AMPA antagonists is known to ameliorate clinical EAE (49). Further, enhanced glutamate transmission and expression of the AMPA receptor GluR1 subunit have been detected in the striatum starting from the presymptomatic phase of EAE (9). Recent evidence shows that synaptic abnormalities occur in the primary somatosensory cortex prior to T cell infiltration and microglial activation and are prevented by the peripheral administration of the soluble TNF inhibitor, XPro1595 (50). It is possible that early changes in the expression of functionally relevant neuronal genes, including those described in this article, underlie such effects.
The mechanisms by which a CNS-directed autoimmune response in EAE, and possibly MS, induce gene-expression changes in neurons and oligodendrocytes prior to immune cell infiltration of CNS parenchyma are not known. Meningeally located immune cells, which include activated T cells and neutrophils (38), might mediate their effects through intermediate cells, such as microglia or the induction of soluble factors. Microglia are activated in early MS lesions (51), and paralysis of microglia in transgenic mice delays the onset and reduces the severity of EAE by inhibiting the development and maintenance of inflammatory CNS lesions (34). Changes in microglia have been detected during the preclinical phase of EAE; they form perivascular clusters in response to fibrinogen leakage at the blood–brain barrier and mediate neuron damage (11). Our observation that activated microglia and deregulated expression of neuronal and myelin genes co-occurred in presymptomatic spinal cord would be consistent with a role for microglia as mediators of early subclinical changes in CNS tissues prior to T cell infiltration of the neuropil. However, when we induced EAE in mice selectively deficient in IKKβ, the main activating kinase for NF-κB in the canonical pathway (52), in cells of the myeloid lineage, our observed changes in marker gene expression were not prevented. This finding indicates that NF-κB, a principal transcriptional mediator of microglial activation (35, 36), is not necessary for these specific effects. In contrast, selective inhibition of soluble TNF, a major proinflammatory product of activated T cells, macrophages, and glia, by the peripheral administration of XPro1595 to mice from the day of EAE induction, delayed changes in innate immune, myelin, and neuronal marker genes in the presymptomatic phase. For example, EAE-induced changes in expression of H2-Ab1, Mbp, Nrg1, Snap25, and Grin1 were prevented in the presymptomatic phase of disease by XPro1595 administration. We (37) and other investigators (53) showed that XPro1595 protects mice against clinical MOG-induced EAE when administered in prophylactic and therapeutic protocols. There, protection was not paralleled by a reduction in inflammatory lesions in the spinal cord, but with reduced overall neuroinflammation and preservation of myelin and neuron integrity during full-blown disease. Together, these results show that soluble TNF mediates changes in oligodendrocytes and neurons that might alter their sensitivity to damage during EAE. Further work is needed to determine whether the failure of etanercept to inhibit EAE-induced gene-expression changes reflects the additional protective properties of transmembrane TNF or whether XPro1595 and etanercept show differential accessibility to CNS tissues in early disease.
In addition to our characterization of early molecular events in EAE, we investigated whether the gene panel could be used as a tool for screening therapies designed for the treatment of MS or other inflammatory CNS diseases. As a proof-of-principle, we treated mice with GA and monitored gene expression at different time points of disease. GA is a random polymer of four amino acids enriched in myelin basic protein and is an approved drug for treating relapsing-remitting MS (33). GA is thought to work primarily by modulating autoimmune T cell responses by altering their maturation or acting as a decoy for myelin. It inhibits Th17 and Th1 responses (54) and boosts CD4+CD25+ regulatory T cells in the EAE model (16). It also promotes several other beneficial effects in the inflamed CNS, such as remyelination, oligodendrogenesis (32, 55), neuroaxonal protection, and neurogenesis (56–58). Previous gene-expression studies in PBMCs obtained from untreated versus GA-treated MS patients showed that GA reduced the expression of proinflammatory cytokines, such as TNF (59). We found that prophylactic GA administration normalized the expression of all of the gene markers selected for the mini gene panel and that this correlated with the amelioration of clinical symptoms. Because GA is unlikely to cross the intact blood–brain barrier (60), our observation that early changes in oligodendrocyte and neuronal gene expression were normalized by GA treatment adds further strong support to the hypothesis that such changes can be indirectly signaled by immune cells prior to their infiltration into the CNS parenchyma.
In summary, our results show that expression analysis of a small number of genes in the spinal cord of mice during the development of EAE can provide information concerning the timing, sequence, and mechanisms of major pathological processes during the development of an autoimmune disease and provides a tool for further evaluating the effectiveness of novel therapeutics at a preclinical level. The finding that transcriptional changes in endogenous CNS cells, such as oligodendrocytes and neurons, occur before infiltration of the CNS parenchyma by immune cells in EAE suggests that such changes might play a primary role in sensitizing the CNS to subsequent infiltration by activated T cells. Our results open up the possibility of finding new therapeutic approaches to target the CNS-localized mediators and processes that are responsible for promoting T cell infiltration and immune-mediated damage in diseases like MS.
D.E.S. is an employee of Xencor and holds stock and stock options in the company. The other authors have no financial conflicts of interest.
We thank Prof. Hans Lassmann for hosting and training M.K. in techniques for the neuropathological analysis of mouse CNS lesions, as well as for critical reading and valuable comments on the manuscript. We also thank Prof. Theodore Tselios for providing MOG peptide, Dr. Clementine E. Karagiorgiou for providing GA, Vassiliki Kyrargyri for providing groups of mIKKβKO and IKKβF/F mice, and Fotis Badounas for providing Abs.
This work was supported by the European Commission through NeuroproMiSe Integrated Framework 6 Project Grant LSHM-CT-2005-018637 and NeuroSign Capacities Framework 7 Project Grant 264083, by the Hellenic Republic Ministry of Education–General Secretariat of Research and Technology through the National Action Cooperation Project Grant “Multiple Sclerosis Therapy” 09SYN-21-609 (to L.P.), and by European Cooperation in Science and Technology (COST) Action Inflammation in Brain Disease (NEURINFNET) Project Grant BM0603 through a short-term scientific mission fellowship to M.K.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- experimental autoimmune encephalomyelitis
- glatiramer acetate
- ionized calcium binding adaptor molecule 1
- myelin oligodendrocyte gp35–55 peptide
- multiple sclerosis
- pertussis toxin.
- Received March 12, 2013.
- Accepted February 21, 2014.
- Copyright © 2014 by The American Association of Immunologists, Inc.