OUP user menu

Clinical, Pathological, and Immunologic Aspects of the Multiple Sclerosis Model in Common Marmosets (Callithrix jacchus)

Bert A. 't Hart PhD, Luca Massacesi MD
DOI: http://dx.doi.org/10.1097/NEN.0b013e31819f1d24 341-355 First published online: 1 April 2009


The efficacy of many new immunomodulatory therapies for multiple sclerosis (MS) patients has often been disappointing, reflecting our incomplete understanding of this enigmatic disease. There is a growing awareness that, at least in part, there may be limited applicability to the human disease of results obtained in the widely studied MS model experimental autoimmune encephalomyelitis in rodents. This review describes the experimental autoimmune encephalomyelitis model developed in a small neotropical primate, the common marmoset (Callithrix jacchus). The model has features including clinicopathologic correlation patterns, lesion heterogeneity, immunologic mechanisms, and disease markers that more closely mimic the human disease. Several unique features of experimental autoimmune encephalomyelitis in marmosets, together with their outbred nature and close genetic and immunologic similarities to humans, create an attractive experimental model for translational research into MS, particularly for the preclinical evaluation of new biologic therapeutic molecules that cannot be investigated in rodents because of their species specificity. Moreover, this model provides new insights into possible pathogenetic mechanisms in MS.

Key Words
  • Experimental autoimmune encephalomyelitis
  • Genetics
  • Immunology
  • Magnetic resonance imaging
  • Marmoset
  • Pathology


Multiple Sclerosis: An Autoimmune Disease?

Multiple sclerosis (MS) is a relapsing remitting and/or chronic progressive inflammatory disease of the central nervous system (CNS) (1, 2). Although the trigger(s) of MS are not known, many scientists believe that autoreactive T cells and antibodies directed against CNS components have a major pathogenic role in the induction of inflammation and tissue destruction in the CNS and therefore in the progression of neurological impairment and disability.

A strong argument supporting the autoimmune concept of MS is the similarities with the effector phase of the autoimmune response in experimentally induced autoimmune encephalomyelitis (EAE). Indeed, inoculation of myelin or myelin components in a suitable proinflammatory formulation creates EAE models with an MS-like clinical and pathological presentation in a variety of laboratory animals, including mice, rats, guinea pigs, and primates (3-5).

EAE: An Autoimmune Animal Model of MS

It is pertinent to emphasize at the start of this discussion that EAE is not MS. Whereas EAE is a classical autoimmune disease model, autoimmunity in MS is only 1 contributor to a highly complex pathogenic process that also involves neurodegeneration. It is therefore not surprising that despite the many similarities between MS and EAE, major differences also exist. Multiple sclerosis seems to develop spontaneously in individuals who are genetically predisposed to the disease. Although an infectious trigger has been proposed, none has been identified despite extensive studies over many decades. By contrast, EAE is induced by evoking a CNS-targeting autoimmune reaction in healthy animals with the requisite immunogenetic background, but which would have not developed the disease spontaneously. A second important difference is the time interval between disease induction and onset of clinical signs, which varies from a few weeks in most rodent models to several months in EAE in nonhuman primates; development of MS may take decades. Another major difference is that most EAE models are driven by CD4+ T cells, whereas the dominant T-cell type in MS lesions is CD8+. Finally, many immunomodulatory therapies that work well in EAE are inactive or may even exert opposite effects in MS patients. Nevertheless, despite these differences, the EAE model has been instrumental for developing the autoimmune concept of MS on which current immunomodulatory treatments have been based.

The most thoroughly investigated EAE models are those induced in inbred strains of mice and rats. Although these models have been tremendously important for the unraveling of pathogenic mechanisms, their validity as a model for therapy development has been debated (6, 7). Indeed, there is a long list of treatments that were found to work well in rodent EAE models but then failed or even showed detrimental effects when they were tested in MS patients (8, 9). This has created a need for a disease model in nonhuman primates that are more closely related to humans. The Table summarizes some salient differences between EAE models in rodents and nonhuman primates and MS.

View this table:

The frequent failure of new immunomodulatory treatments is not an exclusive problem in EAE/MS because this has been observed also in other immune-mediated inflammatory disorders, including rheumatoid arthritis (10, 11) and organ transplantation (12). One explanation for discrepant effects of immunomodulatory treatments between EAE and MS may be the remarkable immunologic differences between an inbred and often pathogen-free laboratory mouse and an MS patient (13). Comparable discrepancies have been well recognized by transplantation immunologists and have led to the situation that new immunomodulatory therapies are tested first in nonhuman primates before they are evaluated in clinical trials (12). We propose a similar policy in therapy development for autoimmune-based inflammatory diseases, that is, to include preclinical tests in relevant disease models in nonhuman primates, thereby bridging the gap between the classical disease models in rodents and the disease in humans (14).

Whereas EAE models in rodents have been frequently reviewed elsewhere, there are only a few reviews on EAE models in nonhuman primates. The obvious explanation for this is that compared with the many scientists who work with rodent EAE models, only very few have access to nonhuman primate models. Nevertheless, the development of biologic drugs that do not work in rodents creates an urgent need for nonhuman primate EAE models.

The aim of this review is to discuss how nonhuman primates can contribute to our understanding of MS and the development of more effective therapies. The fact that nonhuman primates are similarly susceptible and exposed to infectious pathogens as humans and are genetically and immunologically closely related to humans can provide insight into the 2 major MS susceptibility factors-genetics and environment. Moreover, we will discuss how the pathological similarity between the model and MS can enhance insight into the early pathogenic events in CNS white matter lesion formation.

EAE in Nonhuman Primates

Since the beginning of the past century, it was observed that humans injected with rabies vaccine, cultured in rabbit embryo containing nervous tissue, occasionally developed acute disseminated encephalomyelitis. Around 1930, Rivers et al (15) conducted a series of experiments to investigate the underlying pathogenic mechanism. In these early studies, rhesus monkeys were injected with brain homogenates to assess whether they would develop acute disseminated encephalomyelitis aiming to exclude rabies infection etiology. The observation that inoculation of nervous tissue evoked an acute disseminated encephalomyelitis-like disease in monkeys can be regarded as the first documented successful EAE experiment. Subsequent studies by Kabat et al (16), also in rhesus monkeys, revealed that the neuropathologic disorder had been caused by an immune reaction against rabbit myelin contaminants of the vaccine. In subsequent years, several research groups, including ours, have further refined the rhesus monkey EAE model (17-20). All efforts to create a representative model of MS in rhesus monkeys have failed, however, because in all cases, there was an acute or even hyperacute disease course that caused death within a few days after onset. Postmortem histological examination usually showed a markedly destructive CNS pathology, which is atypical of MS.

In 1995, Massacesi et al (21) reported the discovery of an intriguing new EAE model in the common marmoset (Callithrix jacchus) that was induced with whole human myelin in complete Freund adjuvant (CFA) supplemented with an infusion of heat-killed Bordetella particles. This innovative EAE model shared more clinical and neuropathologic similarities with MS than the EAE model in macaques. Shortly thereafter, we reproduced this exciting new model in the Netherlands Primate Center, where it is still fully operational. After several modifications, in particular, the substantial reduction of adjuvant costimulation and the use of myelin from MS patients, a highly reproducible MS-like disease model was created that confirmed its striking similarity with MS (22). In its most characteristic presentation, the disease affects 100% of marmosets from an outbred colony and follows a relapsing-remitting/secondary progressive course. With magnetic resonance imaging (MRI) and histology, the development of MS-like lesions can be detected in brain and spinal cord white matter (23, 24).

The Marmoset EAE Model

Clinical and Pathological Features

In the marmoset EAE model, neurological signs are associated with lesion onset but, as in MS, there is a marked discrepancy between the clinical presentation and the lesion load detected with brain MRI; this phenomenon is known as clinicopathologic paradox. Indeed, some monkeys may have an impressive lesion load in brain white matter without outward clinical signs, whereas others may be completely paralyzed but show only a few brain lesions (23). Possible explanations of this paradox are the same as for MS: lesions in the spinal cord or in the CNS gray matter, or small encephalic lesions located in so-called strategic sites, such as the brainstem, may have a greater clinical impact than large lesions in the cerebral white matter (25).

This may not be the only explanation for the heterogeneity among lesions and their clinical correlates, however. The presence of anti-myelin oligodendrocyte glycoprotein (MOG) antibodies may also affect the features of lesions and clinical expression (26, 27); they might also enhance antigen cross presentation to autoreactive CD8+ T cells, as has been observed in a mouse diabetes model (28). Finally, the activation of cytotoxic T cells specific for (an) epitope(s) encompassed within MOG Peptide 34-56 was found to be critical to the clinical expression of EAE in marmosets (29).

Serial MRI of marmoset EAE shows that the progression of clinical signs may involve lesion formation in any functional system in the CNS (spatial dissemination) and the development of lesions over time. Thus, the model approximates the typical asynchronous development of MS lesions, that is, they are disseminated in time and space (30). Consequently, a variety of lesion stages can be found in the same individual. Moreover, although they are not detectable with in vivo MRI, there can be an impressive amount of demyelination in the cerebral and spinal cord gray matter that is discernible by immunohistochemical staining (31). Gray matter demyelination was often associated with atrophy (32). Possibly because of the similar brain vein distribution similarities among primates, the spatial distribution of the lesions is comparable to that in MS (i.e. periventricular, juxtacortical, etc). Using classification criteria proposed for the staging of MS lesions (33), early active white matter lesions were simultaneously observed with late active and inactive lesions. Monocyte infiltrates are usually present not only in early active lesions (i.e. those showing blood-brain-barrier [BBB] damage and early active demyelination), but also in late active lesions that show a recovered BBB with less or no active demyelination. Part of the lesions with a healthy BBB appears as remyelinated scars without evidence of inflammation (22).

Pathological Features of Different Marmoset EAE Models Vary According to the EAE-Inducing Stimulus

The marmoset has yielded different new EAE models that vary in their clinical and pathological features. In several of these models, there are striking similarities to MS lesions.

EAE Induced With Myelin of Human or Mouse Origin

The original marmoset EAE model was induced with MS patient (21, 22) myelin in CFA, but later studies show that essentially the same results could be obtained using mouse (27) myelin. This model is characterized by multiple large demyelinating lesions that are mainly centered around venules and that are scattered throughout the white matter of the brain and spinal cord. In some cases, several hundred small lesions can be found, but more frequently, there are large demyelinated areas formed by the confluence of smaller lesions. The heterogeneity of such large lesions can be visualized by MRI and by immunostaining of the tissues, such as with the antibody myeloid-related protein 14, which stains recently immigrated macrophages (22). Further histological analysis of this model with markers for MS lesion maturation stages confirmed the pathological heterogeneity of the model (22). In the same brain, several lesion stages could be found ranging from early and late active to chronic inactive with or without remyelination. Detailed immunostaining of early lesions revealed that these lesions contain infiltrated T cells and macrophages that express a variety of proinflammatory and immunomodulatory molecules (34). Deposition of antibody and complement on myelin sheaths was often observed (22, 35). Based on these characteristics, the lesions resemble pattern II MS lesions, as defined by Lucchinetti and colleagues (36).

EAE Induced With Myelin From MOG-Deficient Mice

Mice in which the MOG gene has been deleted normally develop and lack marked abnormalities in their CNS white matter (27). Biozzi ABH mice immunized with myelin isolated from these MOG-deficient mice emulsified with CFA develop chronic relapsing EAE with a less severe course than usual and have markedly reduced CNS inflammation with almost absent demyelination (37). Similarly, marmosets immunized with MOG-deficient mouse myelin displayed only early lesion formation, that is, inflammation and some demyelination, which is apparently induced as a consequence of autoimmune responses to myelin components other than MOG, such as myelin basic protein (MBP) and myelin proteolipid protein (PLP); both of these major myelin proteins are only mildly encephalitogenic in marmosets (26,38). The absence of serum antibodies binding to conformationally intact MOG probably explains the reduced degree of demyelination in these models because infusion of such antibodies amplifies the demyelination (39). In one of 5 monkeys, the damage was sufficient to induce acute EAE, butmonkeys prone to a progressive disease course remained asymptomatic. The fact that chronic relapsing EAE in Biozzi mice could be restored by the addition of recombinant mouse MOG underlines the involvement of MOG in the induction of the immune processes that drive EAE progression in this model. This makes EAE models induced with recombinant MOG of particular interest for the study of the immune mechanisms that underlie chronic disease.

Recombinant Human MOG-Induced EAE

Lesions in this model are usually fewer but enlarge rapidly with progression of the disease (30). This dynamic disease pattern, like that of MS, is also characterized by lesions that are disseminated in time and space. The main lesion type in the recombinant human MOG (rhMOG)-induced model is an early active lesion that consists of a sharply demarcated demyelinated area filled with many macrophages that contain PLP-positive myelin degradation products in phagocytic vacuoles (Fig. 1A). Although evidence of early axonal injury such as accumulation of β-amyloid precursor protein in axons and reduced phosphorylation of neurofilaments was detected in these lesions, axons are usually spared (40). Late active lesions in these animals display much less inflammatory activity and lack these markers (Fig. 1B), suggesting that the axonal injury does not continue when the inflammation subsides. This led us to hypothesize that inflammatory mediators that are formed in these active lesions (e.g. reactive oxygen species) may cause adenosine triphosphate depletion and dysregulation of Ca2+ homeostasis in the neuroaxonal system (23). Dutta and colleagues (41, 42) have shown that this might indeed be the case in MS. Antibody and complement deposition on myelin sheaths can be found in early active lesions in the marmoset model (35). Although axons in this model seemed to be protected against complement-dependent cytotoxic mechanisms by the expression of the complement regulatory factor CD55, axons in the rhesus monkey EAE model lack CD55 expression and were more susceptible to complement-dependent cytotoxicity. This discrepancy may also explain the different sensitivity of rhesus monkeys and marmosets to EAE. Using immunogold-labeled peptides, Raine and colleagues (43) provided evidence suggesting that a proportion of the deposited antibodies is directed against MOG peptides. The histological characteristics of the lesions in this model also type them as resembling the Lucchinetti pattern II lesions in MS (36).


Heterogeneous neuroimaging features of brain lesions in the recombinant human myelin oligodendrocyte glycoprotein (rhMOG) marmoset experimental autoimmune encephalomyelitis (EAE) model. A representative slice of 1 monkey that was scanned during clinically active rhMOG-induced EAE using different magnetic resonance imaging techniques. Three regions of interest are enlarged in the inserts. Note that a large portion of Lesion 2 is located in the gray matter. This figure illustrates the lesion heterogeneity that can be found in the same monkey with respect to size, anatomical localization, and neuroimaging characteristics. See text for detailed descriptions. Gd-DTPA, gadolinium-diethylenetriamine pentaacetic acid; MTR, magnetic transfer ratio; T2W, T2-weighted.

MOG Peptide-Induced EAE

Another type of EAE model can be induced with Peptide 34-56 from rhMOG in CFA in marmosets (29) and in rhesus monkeys (44). In contrast to the model in C57BL6 or Biozzi ABH mice, the peptide-induced marmoset model reproduces MS-like inflammation and marked demyelination in gray as well as white matter. The antibodies formed in this model are confined to the overlapping peptides MOG24-46 and MOG34-56 and cross react only poorly with rhMOG. Antibodies that bind conformational MOG epitopes (those considered necessary for mediating demyelination [39]) have not been found in this model. This strongly suggests that the large demyelinated lesions observed in postmortem T2-weighted (T2W) images of fixed brains are caused by the activity of MOG34-56-induced T cells (29). Recent unpublished data show that EAE in marmosets can also be induced with MOG34-56 peptide in incomplete Freund adjuvant (IFA), that is, mineral oil without mycobacteria. This highly refined EAE model is also characterized by widespread inflammation, demyelination, and axonal injury. Whether the pathological features of the lesions in both models correspond to the Lucchinetti pattern I-like MS lesions remains to be determined.

In Vivo Characterization of Brain Lesions With MRI

Proton nuclear MR provides a variety of noninvasive imaging techniques for the visualization of pathological changes in CNS tissues (45). The most commonly used diagnostic MRI techniques in MS have also been implemented in the marmoset EAE model (46, 47). The T2W images, which are particularly sensitive to water, are often used to determine the total number and spatial distribution of lesions in the brain white matter. However, because many conditions are associated with edema, T2 images give little information as to the underlying pathological processes. The T1W images and magnetic transfer ratio (MTR) images (the latter reflecting the ratio between the tissue concentrations of free protons vs protons bound to macromolecules) provide more information on the extent of CNS tissue loss. To assess where the BBB has become more permeable as an indirect marker of inflammation, intravenously injected contrast agents such as gadolinium-diethylenetriamine pentaacetic acid (Gd-DTPA) are used; these provide signal intensity in T1W images.

Magnetic resonance imaging has been used in the marmoset EAE model to address 3 important subjects, namely: 1) to evaluate disease activity in the CNS in presymptomatic monkeys; 2) to investigate the relationships between MRI changes and defined neuropathologic changes; and 3) to create useful diagnostic tools for the preclinical activity assessment ofnew therapies. A detailed histopathologic evaluation of MRI-detectable MS-like lesions in the marmoset EAE model demonstrated the simultaneous presence of contrast-enhancing and nonenhancing lesions in the same animals, thereby demonstrating the heterogeneity of the lesions (22, 30, 43).

MRI in Presymptomatic EAE

Serial T2W images in the rhMOG-induced EAE model showed that lesion dissemination in time and space occurs weeks before clinical signs become evident (23). This offers the opportunity to examine the developmental kinetics of individual lesions over a period of several weeks. Figure 1 shows 3 selected lesions in a single brain slice from a monkey with clinically active MRI: Lesions 1 and 3 are confined to the white matter and Lesion 2 is located at the border between the white matter and the gray matter. The depicted MRI parameters illustrate the heterogeneity of these lesions. The contrast-enhanced upper half of Lesion 1 is visible as the most hypointense area on the MTR image. The lower half of this lesion is clearly visible on the T2W image, where contrast is mainly based on proton density effects. This part of the lesion is visible on the T1 image as a hyperintense area, but is invisible on the T2- and contrast-enhanced T1 images. Interestingly, both the gray matter and the white matter part of Lesion 2 are visible on the T2W image, but in the quantitative T1- and T2 images, as well as in the contrast-enhanced T1W images, the white matter part is invisible. Finally, Lesion 3 is of equal size on the T2W- and MTR images, whereas only the lower part of this lesion is visible on the T1-relaxation time (RT) image and only a vague shadow can be seen on the T2-RT image. Thus, lesion development in the marmoset EAE model follows a dynamic pattern of some parameters, particularly MTR and contrast enhancement of T1W lesions (30).

Correlations of MRI and Histology

Figure 2 illustrates MRI and histological characteristics of the 2 types of T2 lesions in the marmoset EAE model. The early active lesion in Figure 2A displays contrast enhancement (intravenous gadolinium) on T1 images, a hyperintense T1 signal (caused by the high field strength of 4.7 T), a hyperintense T2 signal, and a reduction of MTR. This MRI pattern characterizes recent lesions which, when examined histologically, display marked inflammation and demyelination but with preservation of axons. As previously discussed, however, histological evidence of early axonal injury can be found. The lesion type at the other extreme of the spectrum can also be visualized as hyperintensity on T2W scans, but is more prevalent in long-lasting chronic marmoset EAE. This nonenhancing lesion lacks histologically detectable inflammatory activity but displays marked demyelination and destruction of most axons (Fig. 2B). Intermediate lesion stages between these extremes, as in MS, can also be demonstrated in the marmoset EAE model (22).


Neuroimaging and histopathologic characterization of the 2 prevalent lesion types in the recombinant human myelin oligodendrocyte glycoprotein (rhMOG)-induced experimental autoimmune encephalomyelitis (EAE) (A and B). In vivo magnetic resonance (MR) images are shown in (a); postmortem MR (ex vivo) images are shown in (b): Luxol fast blue-periodic acid-Schiff (PAS) stains of lesions (with centers indicated by asterisks) are shown at low power (c) and high power (d); immunohistochemistry for recent infiltrates macrophages (myeloid-related protein 14 [MRP-14]-positive) in lesions are shown in (e). (A) The most prevalent lesion type seen at the peak of EAE in the brain of an rhMOG-immunized monkey. This typical early active lesion shows primary demyelination and an abundance of recently infiltrated MRP-14-positive macrophages. The presence of myelin particles within the macrophages (arrowheads) indicates that these are actively engaged in demyelination. (B) A typical late active demyelinated lesion without significant inflammatory activity. Macrophages are present (arrowheads), but these are mostly PAS positive and MRP-14 negative, indicating a late phenotype (22). Based on the in vivo and ex vivo brain magnetic images, the 2 lesion types can be distinguished in vivo with the MR imaging techniques shown (see text). Abbreviations are as in the legend to Figure 1. Reprinted from Blezer et al (30) with permission from John Wiley and Sons.

Data on histopathologic evaluation of MRI-visualized lesions from the brain autopsy of a patient with MS who accidentally died a few hours after a brain MRI scan indicated that not only lesions that show contrast enhancement on T1-weighted images, but also many nonenhancing lesions may contain lymphocytes and macrophages (48). This finding indicates that MRI often visualizes T2-nonenhancing CNS hyperintensities that correspond to histopathologically active lesions.

Histopathologic analyses of contrast-enhancing lesions detected by intravenous Gd-DTPA in the marmoset EAE model also show that areas with inflammatory infiltrates are smaller than areas of BBB damage observed in vivo by MRI. In addition, the area of in vivo-enhancing BBB damage is smaller than the hyperintense area observed at the direct scans using the most water-sensitive T2 sequences (Fig. 2). These data suggest that the lesions are formed as concentric rings around a center made by an inflammatory infiltrate with associated BBB damage. The inner peripheral ring is characterized by BBB damage only, and the outer peripheral ring is caused by edema that surrounds the inner ring and the core center. Analysis of the T2 lesion volume kinetics has shown that a phase of volume expansion when brain lesions can increase 100-fold within only 2 to 4 weeks is usually followed by a phase when the lesions shrink and sometimes completely disappear (30).

Similar morphological and kinetic patterns have been observed in MS. Indeed, the volume kinetics of a number of persisting lesions visualized by frequent MRI suggests that the marmoset EAE and the MS lesions behave in a similar way, thereby underlining the close similarity between the marmoset model and MS (own unpublished data).

MRI in Therapy Trials

Magnetic resonance imaging detection of new brain lesions provides a valid surrogate marker of the disease course in MS; therefore, brain MRI is often used as end point of clinical trials in MS. In clinical trials, the currently widely used therapies interferon-β and Copaxone (a random copolymer of amino acids present in MBP) showed a reduction of MRI-detectable lesion activity in addition to effects on the relapse rate. It has become clear, however, that despite these encouraging effects, both treatments have little effect on the progressive accumulation of neurological impairment. For this reason, other treatments are currently under investigation in preclinical and clinical phases. The marmoset EAE model provides an exquisite system for a wide-range efficacy analysis of a new therapeutic agent on clinical progression, on MRI lesions, and on critical immunopathogenic processes.

An illustrative example of a preclinical trial executed in the marmoset EAE model is the efficacy evaluation of a fully human antibody against a shared subunit of interleukin 12 (IL-12) and IL-23 (IL-12p40) molecule. After demonstrating the neutralizing properties of the antibody against both human and marmoset IL-12p40, the efficacy was tested in the marmoset EAE model in a prophylactic paradigm (49). Strong suppression by the antibody on clinical and pathological features of the disease was observed. To test the effect of the antibody on established disease, we used a novel therapeutic study design based on MRI-guided immunotherapy development in the marmoset EAE model (47). In brief, treatment was started once brain lesion formation in rhMOG-immunized monkeys was confirmed using gadolinium-enhanced T1W and T2W images. Existing and newly forming lesions were examined in follow-up scans recorded at approximately 14-day intervals using semiquantitative techniques, including T1 (before and after Gd-DTPA infusion) and T2 relaxation time and MTR images. Monkeys reaching the predetermined clinical end point (i.e. showing complete paralysis of the hindlimbs) were killed, and the brains and spinal cords were then processed for histological examination. The results of this study showed that the anti-IL-12p40 antibody suppressed the activity of preexisting lesion and delayed the onset of neurological deficit. The suppression of CNS inflammation was confirmed by histopathologic examination (50).

Marmoset EAE as a Model of MS

The relevance of EAE as a model for MS has been criticized (6). Induction of neurological disability in the EAE model requires the activation of adaptive as well as innate immunopathogenic mechanisms by formulation of CNS myelin antigens together with bacterial or viral antigens that interact with the innate receptor CFA. This is illustrated by the SJL mouse EAE model, in which induction of clinical signs with PLP139-151 peptide depended on formulation of the peptide with bacterial peptidoglycan, a ligand of the innate receptors TLR2 and Nod1/2 (51).

Evidence that neurological disability requires a prior infectious insult in MS is lacking. Interestingly, however, it has been observed in primates that neutrophils and macrophages carry peptidoglycan into the CNS (52); peptidoglycan was also detected in cervical lymph nodes and spleen of EAE-affected marmosets. In this way, bacterial antigens and myelin-loaded antigen-presenting cells (APCs) come together in the CNS-draining lymph nodes. Whether the conditions under which autoreactive B and T cells are primed are fundamentally different between EAE models and the MS patient is an open question at present.

Antibodies: Specificity and Pathogenic Role in the Marmoset EAE Model

The pathogenic role of antibodies in MS has become the subject of considerable debate after the finding by Berger et al (53) that MS patients who are seropositive for anti-MOG (and anti-MBP) antibodies at the time of their first clinical event may develop a relapse earlier than patients who lack such antibodies. A variety of studies in the EAE model support a pathogenic role of anti-MOG antibodies. In mouse (54), rat (55), and marmoset (26) EAE models, anti-MOG antibodies amplify demyelination in vivo and enhance myelin phagocytosis by macrophages in vitro (56). Moreover, McFarland et al (38) reported that marmosets immunized with MP4 (a chimeric protein of PLP and MBP) developed EAE only when serum autoantibodies against MOG had been formed.

Our recent findings in marmosets confirm that clinical expression of EAE is associated with the formation of antibodies against conformationally intact MOG (27). The demyelinating capacity of anti-MOG antibodies seems to be confined to binding of specific conformational epitopes (57,58); these likely include N-linked carbohydrates at position Asn31 (59). The fine specificities of anti-MOG antibodies in marmosets immunized with recombinant rat MOG were assessed by von Büdingen et al (39, 60) and were mapped on a homology model of the extracellular domain of human MOG (61). In brief, major B-cell epitopes in rat MOG were identified at Peptides 13-21 and 65-75, whereas Peptides 28-35 and 40-45 were minor epitopes. It is of note that these studies did not confirm the identification of a critical B-cell epitope in MOG54-76 (Fig. 3), suggesting that this epitope may be accessible for antibody binding in the rhMOG protein, but not in conformationally intact native MOG.


Myelin oligodendrocyte glycoprotein (MOG)54-76 is a critical epitope of anti-recombinant human MOG (rhMOG) antibodies. Immune sera from rhMOG-immunized monkeys were preincubated for 1 hour at 37°C with saline (white bar), a pooled mixture of 23-mer peptides spanning the extracellular domain of human MOG1-125 (black bar), or a mixture of all peptides but one of the MOG peptides or other encephalitogenic protein indicated. The preincubated sera were subsequently tested for reactivity with ELISA plate-bound rhMOG. Reactivity of the sera with ELISA plate-bound MOG was observed only when MOG54-76 was omitted from the preincubating mixture, demonstrating that this peptide encompasses one or more epitopes for anti-rhMOG antibodies. αB, αB-crystallin; MBP, myelin basic protein; pMOG, MOG peptide tested.

T cells: Specificity and Pathogenic Role in the Marmoset EAE Model

The chronic progressive disease course in the rhMOG-induced EAE model in marmosets may be driven by the possibly continuous recruitment of de novo activated autoreactive T and B cells into the CNS. This concept of epitope spreading has been well documented in rodent EAE models (62, 63).

Our recently published study suggests that lesions and neurological signs in marmoset EAE may be induced by distinct immune mechanisms (27). Five unrelated marmoset twins were immunized with mouse myelin in CFA; for 1 sibling of each twin, the myelin was isolated from wild-type C57BL/6 mice, for the other, from MOG-deficient C57BL/6 mice (27). In 4 of the 5 twins, only the siblings immunized with MOG-containing myelin from wild-type mice developed chronic progressive EAE, whereas the siblings immunized with MOG-deficient myelin remained asymptomatic. Histological analysis showed that inflammation and demyelination were present in both siblings, although the amount of demyelination was higher in monkeys immunized with MOG-containing myelin. We interpreted these results as indicating that autoimmunity against MOG is not essential for initiating the acute phase of the disease, but is critical for developing a progressive disease course.

Prior studies in Biozzi ABH mice similarly showed that the impaired induction of chronic EAE with MOG-deficient myelin is completely restored when recombinant mouse MOG is added to the MOG-deficient myelin inoculum (37). This implies that the information needed for the induction of chronic EAE is encrypted in the nonglycosylated extracellular domain of MOG. To assess the exact pathogenic role of MOG in the marmoset EAE model, monkeys were sensitized with rhMOG1-125 (29, 64). A single immunization with 100μg of Escherichia coli-translated rhMOG in CFA induced progressive EAE in 100% of randomly collected marmosets from an outbred colony. The disease course considerably varied between individual monkeys; some animals developed neurological signs within 2 weeks, although in others, this occurred only after several months. This could not be explained by a late onset of the autoimmune reactions because active brain lesion formation could be detected early in the asymptomatic period. The model raised 2 questions: 1) how the 100% disease incidence in an outbred model can be explained, and 2) by which (immune) mechanism the variable disease course is regulated.

EAE Initiation

The fine specificity analysis of the MOG-reactive T cells that were present in the lymphoid organs of rhMOG-immunized marmosets at the height of the disease was tested using a set of overlapping 23-mer peptides encompassing the entire rhMOG sequence. It was observed that all monkeys shared a strong proliferative response of CD4+ T helper 1 cells against the epitope MOG24-36 (29, 64). The major histocompatibility complex (MHC) Class II restriction element was also elucidated (see later). The observation that EAE could be transferred to a naive recipient with MOG24-36-specific CD4+ T cells (65) indicates that the ubiquitous T-cell reactivity against this peptide underlies the 100% EAE incidence in the outbred marmoset EAE model. Indeed, sensitization of 4 unrelated marmosets against Peptide MOG14-36-induced clinical EAE characterized by perivascular mononuclear cell infiltrates within CNS white matter, but without demyelination (64).

EAE Progression

The phenomenon of epitope spreading implies that the epitope-specific autoreactive T and B cells diversify with disease progression. The opposite situation was observed, however, in rhMOG-immunized marmosets; monkeys with a more rapidly progressing disease course displayed a wider epitope diversification of the T-cell response than monkeys with a slower disease progression (29). To explain this, we hypothesized that certain peptides from processed rhMOG presented by certain MHC specificities may activate T cells that induce prompt exacerbation of clinical signs (Fig. 4). The analysis of prevalent T-cell reactivities in a large panel ofrhMOG-immunized monkeys identified 3 peptides (i.e. MOG4-26, MOG34-46, and MOG74-86) that were associated with a rapid EAE progression (29). The pathogenic role of 2of these peptides was further tested by immunizing twin siblings with MOG34-56 or MOG74-96 in CFA, followed by booster immunizations with the same peptides in IFA at 28-day intervals until neurological signs were observed. The combined data of 3 independent experiments showed that 10of 11 marmosets sensitized against MOG34-56 developed full-blown clinical EAE (29). By contrast, all 7 monkeys sensitized against MOG74-96 remained asymptomatic, although a few small lesions were detected in CNS white matter on histological examination. Importantly, a single immunization of 3 MOG74-96-sensitized monkeys with MOG34-56/IFA induced full-blown clinical EAE within 3 weeks in 2 monkeys. Whereas in the 4 nonchallenged monkeys, only mild pathological changes were observed, the 2 monkeys with overt clinical EAE displayed marked inflammation, demyelination, and axonal injury similar to that observed in the rhMOG-induced EAE model. Importantly, the MOG34-56-immunized monkeys not only had inflammatory perivascular cuffs in CNS white matter but also had widespread demyelination in the optic nerve, cerebral and spinal cord white matter, and cerebral cortex. This implies that the essential clinical and pathological changes observed in the original MS myelin-induced EAE model can be reproduced in monkeys immunized with the single 23-mer peptide MOG34-56 in CFA. Antibodies do not seem to play animportant pathogenic role. The only detectable serum reactivity was against MOG24-46 and MOG34-56, indicating that the epitope was in the overlapping sequence, that is, MOG34-46. This is not an important B-cell epitope (61).


Hypothetical relationship between epitope diversification and chronic experimental autoimmune encephalomyelitis. Antigen-presenting cells (APCs) take up recombinant human myelin oligodendrocyte glycoprotein (MOG) and process it into small peptide fragments. Major histocompatibility complex (MHC) Class II molecules expressed by the APC bind peptides from the pool on the basis of affinity. Because all monkeys share the invariant Caja-DRB*W1201 allele that binds MOG peptide24-36, a CD4+ T-cell response against this peptide could be measured in all monkeys (64). Caja-DR molecules encoded by the 2 polymorphic Caja-DR3 and Caja-DRW16 lineages bind variable sets of peptides from the pool. We hypothesize that MHC Class II molecules of monkeys with a fast-progressing aggressive disease courses select peptides such as MOG34-56 that can trigger autoaggressive T cells and which not only induce inflammation, but also demyelination and axonal injury.

Studies in rhesus monkeys may explain the dominant pathogenic role of MOG34-56 in the nonhuman primate EAE model. One hundred percent of macaques immunized with MOG34-56/CFA developed clinical EAE and inflammation, demyelination, and axonal injury in cerebral and spinal cord white matter (44). The fine specificity of MOG34-56-induced T-cell lines generated from these monkeys was defined with a panel of truncated synthetic peptides at 39-47 (WYRPPFSR) was found. Using the National Center for Biotechnology Information Blast database, sequence homology with a peptide from the UL86 open reading frame encoded major capsid protein of human cytomegalovirus (CMV) (986-993; WLRSPFSR). Subsequent analyses revealed that T cells from monkeys immunized with the CMV-derived peptide CMVmcp981-1003 cross react ex vivo with MOG34-56 and infiltrate the CNS white matter upon in vivo restimulation with MOG34-56. Importantly, captive populations of marmosets (66) and rhesus monkeys (67) are naturally infected with CMV; as in humans, the infection exacerbates under immunocompromised conditions in nonhuman primates (68, 69).

It has not been formally demonstrated, but it is tempting to speculate that nonhuman primates use similar immune mechanisms to control CMV latency as are used in humans (70). Interestingly, the major capsid protein that is encoded within the UL86 open reading frame is the fourth dominant CD4+ T-cell epitope of CMV; it engages an impressive 1% of the total human CD4+ T-memory repertoire (71). This may imply that individuals equipped with the requisite MHC-DR alleles contain a considerable population of CMV-induced memory T cells specific for the CMVmcp/MOG34-56 mimicry motif. The experiment with marmosets sensitized against low pathogenic MOG74-96 peptide previously described suggests that this may confer an increased risk for exacerbation of encephalitis (29).

In summary, EAE in marmosets often initially shows a relapsing remitting course, followed by a chronic progressive phase that leads to axonal damage and irreversible neurological deficits. Because the time span is much shorter than in MS, this implies a more aggressive disease course. The natural history profile is, however, very similar to that of MS. Although we do not know yet what characterizes the shift to a progressive phase in either the human disease or the animal model, our data indicate that the persistence of inflammation in the lesions may represent one of the major mechanisms that causes perpetuation of the disease process and ultimate disability.

MHC Influence on EAE in Marmosets

Although the association of MS susceptibility with MHC has been known for many years, the underlying mechanisms are poorly understood. The strongest genetic influence in MS is exerted by the MHC Class II region, in particular, the HLA-DRB1*1501/-DRB5*0101 haplotype (72). Other alleles may, however, positively or negatively influence particular aspects of the pathogenic process (73, 74). For example, a potential influence of HLA-DP alleles in progressive MS has been reported (75). Interestingly, a strong influence of the Mamu-DPB*01 allele on the susceptibility of rhesus monkeys to MBP-induced EAE (76) was observed. This allele was found to restrict the activation of a TH1 cell line specific for MBP61-82 from an EAE-affected monkey (76). Interestingly, an MBP61-82-specific TH1 cell line isolated from the naive repertoire of a nonrelated rhesus monkey induced mild meningoencephalitis when it was injected into a nonirradiated autologous monkey (77). Moreover, certain MHC class I alleles (HLA-A3 and HLA-B7) affect the MS course by the activation of autoreactive CD8+ T cells (78).

We have taken advantage of the outbred nature of marmosets to analyze the multifaceted MHC association. The most critical autoimmune reactions for EAE development are directed against MOG, and several epitopes that partly overlap with those in MS and rodent EAE models have been identified (Fig. 5). Although elucidation of the genetic control of the marmoset EAE model is far from complete, the data obtained thus far support our hypothesis that different immunopathogenic aspects may be controlled by distinct sets of MHC alleles (79).


Mapping of dominant T- and B-cell epitopes in recombinant human myelin oligodendrocyte glycoprotein (rhMOG)-immunized marmosets T-cell and antibody epitopes with confirmed pathogenic significance in the marmoset experimental autoimmune encephalomyelitis (EAE) model are mapped on the amino acid sequence of mouse MOG (revised from Kerlero de Rosbo and Ben-Nun 113 with permission from Elsevier), in which dominant T-cell reactivities in multiple sclerosis (MS) patients and rodent EAE models are plotted. The figure shows that the ubiquitous T-cell reactivity in marmosets to MOG24-36 has not been found in MS or in rodent EAE models. By contrast, T-cell reactivity against MOG34-56 is broadly detected in rodent EAE models as well as in MS patients. T-cell reactivity against MOG74-96 is shared between marmosets and MS patients but is rarely detected in rodent EAE models. T-cell reactivity against MOG94-106 is not common in MS, but is shared with EAE models in SJL mice and Lewis rats (113,114). Antibody reactivity in rhMOG-immunized marmosets seems to be mainly directed against MOG54-76 (Fig. 3).

Genomic analysis showed the MHC Class II region of Old-World primates, such as rhesus monkeys (80). The MHC Class II region of the marmoset (acronym Caja [81]) was found to comprise monomorphic Caja-DRB*W1201, polymorphic Caja-DRB*W16 and Caja-DRB1*03 sequences, and oligomorphic Caja-DQA and -DQB sequences (82). Subsequent transcriptomic analysis revealed that the Caja-DRB1*03 locus encodes pseudogenes that can form chimeric Caja-DR molecules by recombination of Caja-DR3 Exon 2 elements with Caja-DRB*W16 Exon 1 and 2 elements (83). Differences also exist in the MHC Class I region between marmosets and rhesus monkeys or humans; marmosets lack classical MHC Class Ia lineage members, equivalent to Mamu-A and -B. Thus far, only the equivalents of Mamu-G (at least 5 alleles) and Mamu-E (only 2 alleles) have been found. Whether Caja-G alleles are true orthologs of Mamu-G and HLA-G or rather function as classical HLA-C-like molecules is debated (84, 85).

Using anti-MHC Classes I and II blocking antibodies in combination with exchange of autologous APCs (Epstein-Barr virus-transformed B cells) by APCs from unrelated donors, it was demonstrated that the monomorphic Caja-DRB*W1201 allele encodes the restriction element for activation of encephalitogenic MOG24-36-specific CD4+ T cells (64). As previously discussed, this common specificity may explain the 100% EAE incidence. The situation with the second important reactivity (i.e. to MOG34-56) seems considerably more complex because both CD4+ and CD8+ T cells are stimulated to proliferate by this peptide (29). Interestingly, the observation that MOG35-55 peptide is strongly encephalitogenic in mice expressing the HLA-DR2 allele HLA-DRB1*1501 demonstrates that this allele might be a restriction element for the activation of encephalitogenic T cells in MS patients (86). The failure of APC from unrelated donors to support proliferation of MOG34-56-specific T-cell lines demonstrates that this peptide is not presented via invariant Caja-DRB*1201 molecules (unpublished data). Whether the requisite Caja-DR restriction elements of MOG34-56-specific CD4 cells can be found in the polymorphic lineages Caja-DRBW16, which is specific for New-World monkeys, or Caja-DRB1*03, which is evolutionary conserved, remains to be established.

We have recently reported that MOG34-56-reactive cytotoxic CD8+ T cells may also play a dominant pathogenic role in the marmoset EAE model (29). This implies that MHC Class I polymorphisms may also influence the disease course. The cytolytic activity of MOG34-56-induced T cells toward MOG peptide-pulsed B cells is peptide specific, but they recognize peptide presented by B-blastoid lines from unrelated donors (unpublished data). Whether invariant Caja-E molecules are involved in peptide presentation to cytotoxic T cells needs to be proven.

Compartmentalization of the Immune Response in the CNS

Recently, we suggested that the immunopathogenic mechanisms that give rise to EAE comprise 3 compartments: 1) the peripheral compartment, comprising the secondary lymphoid organs in which autoreactive T cells are primed by infectious pathogens; 2) the central compartment, comprising the brain and spinal cord in which autoreactive T cells and antibodies find their targets; and 3) the draining lymph node compartment, that is, the cervical lymph nodes for the brain and the lumbar lymph nodes for the spinal cord (87).

Compartment 1

The beneficial effect of immunotherapies that exert their main activities in the peripheral compartment (e.g. the B cell-depleting antibody rituximab, or natalizumab, which blocks lymphocyte entry into Compartment 2) illustrates that this compartment is not only involved in the initiation of MS, but also in advanced stages of the disease.

Compartment 2

Whether T cells are activated and expand only in CNS draining lymph nodes or whether they also are activated and expand within the CNS has not been resolved. The restriction of the T-cell repertoire observed in MS lesions and in cerebrospinal fluid cells, but not in peripheral blood mononuclear cells collected from MS patients, supports the hypothesis of intrathecal antigen stimulation and expansion of these cells (88-90). By contrast, there is little doubt about compartmentalization of the B response during the course of MS because intrathecal production of immunoglobulin molecules with oligoclonal profiles that persist for the entire disease course is a well-established diagnostic hallmark in MS (91). Moreover, intrathecal expansion of B cells in MS patients is well documented (92, 93).

Compartment 3

During the course of EAE (94) and MS (95), an increment of APC loaded with myelin antigens has been observed within the CNS-draining cervical and lumbar lymph nodes. Their location within T-cell activation areas of the lymph nodes suggests that new autoreactive T-cell specificities may be activated at those sites. Moreover, data of Weller et al (96) in a rat cryolesion model support the concept that cervical lymph nodes are a potential source of newly primed autoreactive T cells.

The important question determining to what extent each of the 3 compartments can be targeted therapeutically (e.g. by small molecules or large monoclonal antibodies) can be investigated in the marmoset EAE model. The histopathology of the lesions in Compartment 2 indicates that persisting inflammation and demyelination are frequently associated with an intact BBB, suggesting that after the acute phase has waned, the BBB may recover (48). The similar lesion morphology observed both in MS and in marmoset EAE and the availability of relevant MRI and histologic analyses for systematic and longitudinal examination of brain lesion stages offer the opportunity of investigating these critical issues for therapy development (22, 30).

As in other autoimmune diseases, the immune response in MS and in EAE is at least in part sustained by a tertiary lymphoid tissue compartmentalized in the target organ (97). In addition, longitudinal evaluation by MRI indicates that after the contrast-enhancing phase in marmoset EAE, many new lesions persist, whereas others disappear (30). A similar lesion course can also be observed in MS because virtually all new brain lesions after an enlarging phase shrink, and about 30% (particularly the smaller ones) may disappear (97, 98).

To explain why some lesions are chronically activated and others are not, it has been hypothesized that during the course of the disease, a specific event facilitates persistence of the inflammatory infiltrates in the CNS. What is/are the event/s that can orient the destiny of a lesion to chronicity or to heal? This key question can also be investigated in the EAE model. Some evidence points to the recruitment into the lesions of professional APC, such as (Epstein-Barr virus-infected) B cells (99, 100) or dendritic cells (62, 101). Another explanation may be that the homeostatic regulation of resident CNS APC, such as microglia, by inhibitory C-type lectin receptors is disturbed, thereby changing the default state of resident APC from tolerogenic to immunogenic (102, 103).

Summary and Further Questions to be Addressed by the Marmoset EAE Model

The data obtained in the marmoset EAE model suggest an alternative paradigm of MS pathogenesis from current models. Although a large part of the model is still hypothetical, supporting evidence is accumulating.

We propose that virus infections do not directly trigger MS, but rather create a repertoire of potentially autoreactive memory T cells. Memory T cells in CNS draining lymph nodes acquire encephalitogenic capacities when they come into contact with APC that present myelin antigens. Consistent with the insightful primary lesion hypothesis formulated by Wilkin (104), we believe that MS patients are genetically predisposed high responders to these antigens, which are released from the CNS by an antecedent pathogenic event. The accumulation of myelin-loaded APC in cervical lymph nodes of EAE-affected animals (94) and the observation that surgical removal of cervical and lumbar lymph nodes abrogates the typical chronic relapsing EAE course of Biozzi ABH mice (105) provide support for this hypothesis. Because drainage of CNS antigens to cervical and lumbar lymph nodes is not a specific feature of MS but occurs in many types of CNS injuries, the specific conditions in which a CNS insult leads to chronic neuroinflammatory disease need to be identified.

The immune processes inside the CNS draining lymph nodes are still poorly understood. Although the concept that T-cell activities induced in cervical lymph nodes are mainly anti-inflammatory holds (106), the induction of T cells with the capacity to exacerbate EAE has also been observed (107). There is some evidence that alterations in the posttranslational modification of CNS glycoproteins may abrogate their binding to inhibitory receptors on APCs, resetting T-cell activation from a tolerizing into a proinflammatory default state (102). Moreover, also the presence of bacterial antigens may create a proinflammatory environment within CNS draining lymph nodes (52).

The experiments with MOG-deficient myelin in mice (37) and marmosets (27) reveal an important role for anti-MOG autoimmunity in the progressive phase of chronic EAE models. Although autoimmunity against myelin antigens other than MOG can cause initial myelin damage and self-limiting disease in marmosets, development of chronic disease depends on the presence of anti-MOG antibody formation (27, 38, 108). Indeed, anti-MOG antibodies clearly exacerbate EAE induced with MBP or PLP (26), especially antibody species that bind conformationally intact MOG (39). We also recently found that mild CNS inflammation can be turned into fulminant EAE by the activation of anti-MOG Tcells, particularly those directed against MOG34-56 (29). Recent unpublished data also show that inflammatory demyelination of the CNS white matter together with neurological deficit can be induced by repetitive immunization with MOG34-56 in IFA (unpublished data). Autoreactive T cells in naive C57BL6 or Biozzi ABH mice are not activated under these conditions, suggesting that this aspect may be specific to primates. A fundamental difference between primates and pathogen-free laboratory rodent strains is that the primates are continuously exposed to environmental pathogens, including the herpes viruses (herpes simplex virus, CMV, human herpes virus 6, and Epstein-Barr virus) that cause persistent latent infections. Data obtained in rhesus monkeys hint at a role of anti-CMV T cells cross reacting with MOG34-56 in the pathogenesis of EAE (44). In humans, it has been shown that recurrent exacerbations of latent CMV infection create a vast repertoire of memory T cells, estimated to comprise ±10% of the total memory pool. About 25% of the anti-CMV reactivity of CD4+ memory T cells is directed against the UL86 open reading frame-encoded major capsid protein (71), which shares a mimicry motif with MOG34-56 (44). The MOG34-56-induced marmoset T cells share phenotypic (CD3+CD4/8+CD56+CD16) and functional (peptide-specific cytolytic activity) similarities with anti-CMV effector memory cells (29, 70, 109). Intriguingly, a phenotypically similar subset of CD4+CD56+ cytotoxic T cells with the capacity to lyse oligodendrocytes has been identified in MS (110, 111), although the cytolytic activity was found to be MHC independent (112). It is therefore tempting to speculate that the demyelination observed in the MOG34-56-induced EAE model is caused by CMV-induced CD4+CD56+ T cells exerting cytolytic killing of oligodendrocytes. However, solid proof is lacking and much work needs to be done to prove that this is the case.


The authors thank Mr H. van Westbroek for the artwork.


  • Much of the work discussed in this review article was supported by grants from the European Community and the Italian Multiple Sclerosis Foundation.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
View Abstract