Thalamus pathology in multiple sclerosis: from biology to clinical application

Abstract

There is a broad consensus that MS represents more than an inflammatory disease: it harbors several characteristic aspects of a classical neurodegenerative disorder, i.e. damage to axons, synapses and nerve cell bodies. While the clinician is equipped with appropriate tools to dampen peripheral cell recruitment and, thus, is able to prevent immune-cell driven relapses, effective therapeutic options to prevent the simultaneously progressing neurodegeneration are still missing. Furthermore, while several sophisticated paraclinical methods exist to monitor the inflammatory-driven aspects of the disease, techniques to monitor progression of early neurodegeneration are still in their infancy and have not been convincingly validated. In this review article, we aim to elaborate why the thalamus with its multiple reciprocal connections is sensitive to pathological processes occurring in different brain regions, thus acting as a “barometer” for diffuse brain parenchymal damage in MS. The thalamus might be, thus, an ideal region of interest to test the effectiveness of new neuroprotective MS drugs. Especially, we will address underlying pathological mechanisms operant during thalamus degeneration in MS, such as trans-neuronal or Wallerian degeneration. Furthermore, we aim at giving an overview about different paraclinical methods used to estimate the extent of thalamic pathology in MS patients, and we discuss their limitations. Finally, thalamus involvement in different MS animal models will be described, and their relevance for the design of preclinical trials elaborated.

Introduction to the disease

General remarks

Multiple sclerosis (MS) is a complex, multifactorial, polygenic disease, influenced by various factors including age, gender, hormones and environment. The most widely accepted hypothesis is that auto-reactive T and B cells induce myelin damage, neuroinflammation and neurodegeneration. However, primary oligodendrocyte dysfunction was implicated as well as potential disease-triggering factor [1]. Despite an unknown etiology, the (histo-) pathological hallmarks of MS lesions are well defined. They include focal demyelination, oligodendrocyte loss, inflammation and neuronal damage which can be seen in various brain regions including diverse white and gray matter areas [2, 3]. MS affects patients of all ages, but symptoms are most likely to appear first in individuals between 20 and 50 years of age. The estimated prevalence of MS is about 2.5 million people worldwide, and it is two to three times higher in women than in men. The diagnosis of MS has traditionally relied upon accumulation of clinical and paraclinical information which at the same time can help to eliminate alternative diagnoses [4].

Clinical aspects

The diagnosis “Multiple sclerosis” requires evidence of lesions in at least two separate areas of the central nervous system (CNS), including brain, spinal cord, and optic nerves (dissemination in space), and evidence that new lesions developed at different time points. This dissemination in time is defined by the simultaneous presence of contrast enhancing (=active) and non-enhancing lesions that are T2w hyperintense, or by novel clinical symptoms at least 30 days after the disease onset. Initial symptoms are diverse but the most frequent ones are visual disturbances, paresthesias, ataxia and muscle weaknesses. Due to the fact that inflammatory foci can arise in virtually any brain area, symptoms can also be related to other brain functions than listed above. For example, cognitive impairment is also a frequently observed phenomenon in MS patients [5]. MS can present in different clinical courses. A clinically isolated syndrome (CIS) is a patient’s first neurological episode caused by inflammation and demyelination. The long-term risk of developing clinically definite MS (CDMS) after a CIS is 60–80 % when lesions consistent with MS are seen on MRI, and about 20 % when the brain scan is normal. Relapsing remitting MS (RRMS) is the most common disease course affecting about 85 % of all MS patients. RRMS means that symptoms appear (i.e. a relapse) and then fade away either partially or completely (i.e. remitting). Benign MS is usually a subset of RRMS and comprises patients who accumulate little physical disability over many years or even remain clinically stable. Secondary progressive MS (SPMS) is characterized by chronically progressive clinical worsening over time. This progressive course may develop slowly after an initial clinically isolated syndrome but usually follows a period of a well-defined RRMS disease course. During the transition from RRMS to SPMS, relapse frequency decreases but quickening of neurodegeneration can be observed. In about 15 % of patients, classical relapses cannot be clearly delineated, despite clinical deterioration, a disease course called primary progressive MS (PPMS). A PPMS patient’s rate of progression may vary over time with an occasional plateau or even temporary improvement, but the overall progression remains continuous [6].

Histopathological hallmarks

Notwithstanding the traditional focus on white matter tracts as predominantly affected regions in MS, we now know that there is significant gray matter involvement in this disease, in both, early and late phases of the disease process [7, 8]. Although cortical demyelination has drawn much attention, it has been shown in parallel that the disease process also involves various subcortical areas such as basal ganglia, hypothalamus, hippocampus, spinal cord or the thalamus [8, 9]. On the histopathological level, gray matter lesions are not just characterized by loss of myelin, but display typical features of neurodegeneration such as neuronal loss and axonal transection [8, 10, 11]. Interestingly, such neurodegenerative events are not restricted to demyelinated lesions but can also be found diffusely in normal appearing gray and white matter. The pathogenesis of gray matter lesions is not well understood and may differ from that of white matter lesions. For example, purely gray matter lesions lack many typical inflammatory features of their white matter counterparts [12–14].

Pathological studies from the beginning of the twentieth century were the first to demonstrate the involvement of gray matter pathology in MS. This finding was, however, disregarded for many years. In 1962, Brownell and Hughes [15] reported that almost one-third of all lesions were located entirely or partially within the cortex. The use of myelin immunohistochemistry has greatly expanded the ability to detect and classify gray matter lesions, and consecutively improved our knowledge of gray matter pathology. Lately, the relationship between clinical disability, disease progression and gray matter pathology was extensively investigated. These clinical studies came up with surprising and equally interesting findings. In a 5-year follow-up study of 181 MS patients by Horakova and colleagues [16], gray matter volume loss could predict Expanded Disability Status Scale (EDSS) progression at different time points over 0–36 months, whereas white matter volume loss was less potent to do so. Fisniku and colleagues [17] concluded similarly in their 20-year follow-up study; in MS patients with a relative long and homogeneous disease duration, gray matter atrophy was more marked than white matter atrophy and reflected disease subtype as well as disability to a greater extent compared to white matter atrophy or T2w lesion load. Beyond, results from Lavorgna and colleagues [18] implicate that gray matter volume and EDSS are the best long-term predictors of disease progression in RRMS patients with a relative long and mild disease. In a recent publication by Hagemeier and colleagues [19], deep gray matter volume was the strongest EDSS predictor in a cohort of 149 RR- and 61 SPMS patients, indicating that not only cortical atrophy has a clinical effect but also the deep gray matter plays an important role in MS. Furthermore, Batista and colleagues [20] demonstrated that gray matter atrophy predicts cognitive dysfunction in MS. In summary, there is sound evidence that gray matter pathology is an important predictor for irreversible clinical disability. A better understanding of how gray matter structures are destroyed in MS will pave the way for the development of new therapeutic strategies.

One of the deep gray matter structures harboring diverse functional sub-nuclei is the thalamus. Clinical studies underpin the assumption that the thalamic network is severely impaired in MS patients. The findings include reduction in cerebral blood flow [21–23], a decrease of normalized thalamic volume [24, 25], or reduced cerebral glucose metabolism rates [26, 27]. Reports about thalamus pathology at the histopathological level, however, are sparse. The thalamus—as an integral part of the diencephalon—is implicated in diverse neuronal networks. Thalamic axons are well known to transmit information between various subcortical and cortical areas. As such, damage to the thalamus and its connections potentially impairs a wide range of neurologic functions that may clinically translate into significant cognitive, physical or mental disability [28]. To be more specific, atrophy of the thalamus, determined with MRI, can help identify which patients with clinically isolated syndrome are at risk for developing clinically definite MS [29–31]. Thus, MRI-based volumetry of the thalamus could be used to determine which CIS patients are at highest risk for a second attack. Second, it was found that atrophy in the cortex and subcortical deep gray matter, including the thalamus, was significantly related to patients’ declining cognitive abilities. Thus, thalamus atrophy is a strong predictor for cognitive decline in MS patients [20, 25, 32, 33]. Furthermore, thalamic atrophy is correlated with long-term accumulation of EDSS-rated clinical disability in patients with MS [34, 35].

In principal, thalamus damage in the brain of MS patients can occur due to various different underlying mechanisms, but two aspects are highlighted in this review article. First, demyelinating lesions can form directly within the thalamus which results in thalamic neurodegeneration and malfunction as systematically analyzed by Vercellino and colleagues [8]. Results of Rocca and colleagues [35] suggest a second mechanism: they observed that white matter lesions outside of the thalamus contribute to thalamus tissue loss. Thus, both local inflammatory demyelination and changes secondary to axonal transection of fibers passing through areas of diseased white matter can account for thalamic abnormalities in patients with MS. Consequently, it was suggested that the thalamus might be a critical barometer of diffuse neuronal pathology in the MS brain. As a highly interconnected brain structure, the thalamus might be able to “integrate” tissue damage which occurs somewhere else in the brain and, thus, reflecting global neuronal health or dysfunction. If this holds true, two distinct conclusions can be drawn: First, amelioration of thalamic pathology by pharmacological intervention might significantly prevent the progress of physical and mental disability in MS patients. Second, measurement of thalamic atrophy may become an ideal MRI outcome for neuroprotective MS clinical trials.

Before we will address, which mechanisms might be operant during indirect thalamus damage in MS patients, we will introduce to the most important aspects of thalamus anatomy and then give an overview of current treatment options in MS.

Neuroanatomy of the thalamus

The thalamus forms the largest part of the diencephalon and is eponymous for other diencephalic components such as the epithalamus and hypothalamus. Topographically, the medial surface of the thalamus constitutes the upper part of the lateral wall of the third ventricle and is connected to the corresponding surface of the opposite thalamus by a flattened gray band, the so called inter-thalamic adhesion. Although the thalamus may appear to be a maddening jumble of nuclei, three basic types of thalamic nuclei can be distinguished: first relay nuclei, second association nuclei, and third nonspecific nuclei. Relay nuclei receive input from the periphery and forward that information to the cortex [36]. The sketch in the left part of Fig. 1 shows the location of some of the thalamic nuclei on the surface as seen from an angle looking down and toward the outside wall of the right half. Relay nuclei can be divided into three functional groups: (1) sensory relay nuclei that receive input from the peripheral sensory receptors through their respective pathways and project to sensory areas of the cortex (ventral posterolateral (VPL), ventral posteromedial (VPM), medial geniculate and lateral geniculate nuclei; [37]), (2) motor relay nuclei that interconnect with motor structures and project to motor areas of the cortex (ventral lateral (VL) and ventral anterior (VA) nucleus; [38, 39]), and (3) limbic nuclei that interconnect with the different structures of the limbic system (anterior nucleus, lateral dorsal nucleus and dorsomedial nucleus). Thus, the VPL, VPM and medial geniculate and lateral geniculate nuclei can be considered as “sensory thalamus”, whereas the VL and VA as “motor thalamus”. The association nuclei as the second principal type of thalamic nuclei receive most of their input from the cerebral cortex and project back to the cerebral cortex in the association areas where they appear to regulate activity. The pulvinar is the largest of these association nuclei, occupying the posterior part of the dorsal tier of the thalamus [40]. The third principal type of thalamic nuclei is the nonspecific nuclei, including many of the intra-laminar and midline thalamic nuclei that project quite broadly through the cerebral cortex, and may be involved in general functions such as alerting [41].

Fig. 1.

The left part of the picture shows a schematic drawing of distinct nuclei within the thalamus, highlighting the connection of relay nuclei with distinct cortical regions. On the right side, a coronal brain T2-weighted MRI slice is shown at two different levels. Arrows depict the anterior and posterior part of the thalamus, respectively. LGN lateral geniculate nucleus, MGN medial geniculate nucleus, VPM ventral posteromedial nucleus, VPL ventral posterolateral nucleus, VL ventral lateral nucleus, VA ventral anterior nucleus; adapted from M. Trepel, Neuroanatomie- Struktur und Funktion

Due to the focus of the article and restrictions in length, the present section can only provide a brief overview of the functional anatomy and connectivity of the thalamus. We will focus on the description of principal thalamic relay nuclei notwithstanding the importance of association nuclei and nonspecific nuclei.

Lateral and medial geniculate nucleus

The lateral geniculate nucleus (LGN), also called the lateral geniculate body, can be subdivided into a dorsal lateral geniculate nucleus (DLG) and a ventral lateral geniculate nucleus (VLG). The former forms the main relay between the retina and the primary visual cortex [42]. The output of the DLG is primarily directed towards the primary visual cortex, terminating in layer IV, while there are additional inputs to layers I and VI.

The medial geniculate nucleus (MGN), also called the medial geniculate body, forms the most caudal extension of the thalamus and its caudal half is situated laterally alongside the mesencephalon. Together with the LGN, the MGN comprises the Metathalamus. The MGN represents the thalamic relay between the inferior colliculus and the auditory cortex, and is thus the principal auditory relay nucleus of the thalamus. This nucleus consists of several sub-nuclei that all have different functions within the auditory system [43].

Ventral posterior complex (VPL + VPM)

The ventral posterior complex is the main relay for sensory inputs to reach the cerebral cortex and can be divided into at least three main parts: the ventral posterolateral nucleus (VPL), receiving spinal somatosensory inputs, the ventral posteromedial nucleus (VPM), receiving trigeminal somatosensory inputs, and the parvicellular medial part which forms the main thalamic relay for gustatory and visceral ascending pathways. Notably, the VPL and VPM receive afferents from various other subcortical areas, among them a serotonergic input from the dorsal raphe nucleus and a GABAergic input from the thalamic reticular nucleus.

Ventral lateral and ventral anterior nuclei

The ventral lateral (VL)–ventral anterior (VA) complex occupies an extensive nuclear area [44]. Cytoarchitectonically, the VA/VL complex is cell sparse and contains relatively large neurons which make this region relatively easy to identify. The VL/VA complex receives afferents mainly from the cerebellum and the basal ganglia, and is in reciprocal contact with somatomotoric and premotoric cortical areas. Additional subcortical inputs derive from the vestibular nuclei. Cerebellar inputs originate from the deep cerebellar nuclei. Furthermore, this complex receives a third strong input from the somatomotoric cortex and from a wide array of frontal and parietal cortical areas. As already mentioned above, the VA/VL complex must be considered the main motor thalamic relay to the cerebral cortex. The largest volume of VA/VL is devoted to the cerebellar influences on the motor thalamo-cortical system. This cerebello-thalamo-cortical pathway is most probably devoted to the coordination of multi-joint movements [45].

Limbic relay nuclei

The limbic system has connections with three major nuclei of the thalamus: the anterior nucleus, the lateral dorsal nucleus (LD), and the dorsomedial nucleus (DM) [46]. Both the anterior and the LD nuclei can be considered relay nuclei, but the DM on the other hand is an association nucleus that is closely tied to limbic system function. The anterior nucleus of the thalamus receives its driver input from the mammillary body through the mammillothalamic tract (also called bundle of Vicq d’Azyr). The mammillary bodies, anatomically a part of the diencephalon, belong to the limbic system circuitry and are directly linked to the hippocampus. Information from the anterior nucleus of the thalamus is then relayed to the cingulate cortex, the main cortical area dedicated to the limbic system as well as to the prefrontal and parietal cortices. The LD of the thalamus is closely related to the anterior nucleus but receives its driver input primarily from the entorhinal cortex. Similar to the anterior nucleus, the LD projects to the cingulate and parietal cortices. Together, the anterior and LD nuclei play a pivotal role in consolidation of memories, motivation, and direction of attention to a specific stimulus. This is particularly important to understand why thalamic pathology correlates with cognition deficits in MS patients (in detail discussed below). The DM has connections with various structures of the limbic system and influences motivation through connection with the cingulate gyrus. It inhibits inappropriate behavior and mediates executive function through connections with the prefrontal and orbitofrontal cortices.

Current treatment options in MS

It is widely accepted that the common pathogenic pathway in MS involves the activation of the peripheral immune system finally targeting CNS myelin and neurons. In consequence, therapeutic strategies include modalities aiming at antagonizing the various immune elements that are involved in this multi-faceted cascade. During the last three decades, major advances could be achieved in the field of MS immunotherapy. The immunotherapeutic modalities can be divided into two main groups: those affecting the acute stages (relapses) of the disease, and long-term treatments that aim at decreasing relapse frequency (i.e. disease modifying treatments). Corticosteroids are mainly used to reduce the acute inflammation that spikes during a relapse. These drugs inhibit lymphocyte proliferation [47], induce apoptosis in peripheral blood leukocytes [48], ameliorate the synthesis of pro-inflammatory cytokines [49], and reduce the expression of cell surface molecules required for immune function [50]. Furthermore, it is believed that corticosteroids stabilize the blood–brain barrier (BBB), for example, by decreasing the expression of angiopoietin-1 and vascular endothelial growth factor-A, both well known to regulate the permeability of the BBB [51]. Alternatively, plasmapheresis (removal of blood components) might be applied to help combat severe symptoms of relapses in patients who are not responding to corticosteroids. In contrast, disease-modifying drugs are prescribed with the aim to reduce the relapse frequency. Currently, this group includes β-interferons (Avonex, Betaseron, Extavia, and Rebif), fingolimod (Gilenya), glatiramer acetate (Copaxone), mitoxantrone (Novantrone), natalizumab (Tysabri), alemtuzumab (Lemtrada), and teriflunomide (Aubagio). Furthermore, dimethyl fumarate (Tecfidera) was recently approved for the treatment of RRMS.

Beta-interferons balance the expression of pro- and anti-inflammatory agents in the brain and reduce the number of inflammatory cells that cross the BBB. Fingolimod, which is in vivo converted to its active form fingolimod phosphate [52], suppresses lymphocyte emigration from lymphoid tissues into the circulation and thereby reducing the immune cell load within the brain parenchyma. Glatiramer acetate is a mixture of random polymers of four amino acids, which is believed to mimic the antigenic properties of the myelin basic protein, a component of the myelin sheath of nerves with which it competes for presentation to T cells. Mitoxantrone is a type II topoisomerase inhibitor. It disrupts DNA synthesis and DNA repair in both healthy and cancer cells. Hence, it suppresses the proliferation of T cells, B cells, and macrophages, and consequently impairs antigen presentation as well as decreases the secretion of pro-inflammatory cytokines. Natalizumab is a humanized monoclonal antibody against the cell adhesion molecule integrin α4 and reduces the ability of inflammatory immune cells to pass through the BBB. Alemtuzumab is a monoclonal antibody that binds to CD52, a protein present on the surface of mature lymphocytes, but not on the stem cells from which these lymphocytes are derived. After treatment with alemtuzumab, these CD52-bearing lymphocytes are targeted for destruction. Teriflunomide belongs to a class of drugs called pyrimidine synthesis inhibitors. By inhibiting dihydroorotate dehydrogenase (catalyzes de novo biosynthesis of pyrimidine) and diminishing DNA synthesis, teriflunomide has a cytostatic effect on proliferating B and T cells. Dimethyl fumarate is rapidly attacked by the detoxifying agent glutathione (GSH). GSH depletion and subsequent induction of the anti-inflammatory stress protein heme oxygenase-1 (HO-1) is thought to be one of the mechanisms responsible for the immune-modulatory actions of dimethyl fumarate. Other postulated mechanisms of action include direct cytoprotective effects through the induction and activation of nuclear factor (erythroid-derived 2)-like 2 (Nrf2) and subsequent induction of an antioxidant response [53].

Knowing that neurodegeneration and neuroinflammation play a key role in the disease course of MS, attention has shifted towards the development of neuroprotective strategies to ameliorate long-term neurological disability. Future clinical studies have to show to what extent the currently approved disease modifying treatments are able to ameliorate the accumulation of irreversible clinical disability. Among others, a potential candidate in this field is fingolimod. In addition to its peripheral action, fingolimod may cross the BBB and target CNS cells expressing S1P receptors, including astrocytes and oligodendrocytes [52]. By this interaction, fingolimod might directly halt neurodegenerative processes or induce remyelination. In line with this assumption was the finding of the FREEDOMS study that fingolimod improved the risk of disability progression in MS patients [54]. In addition, reductions in brain volume were smaller with fingolimod. Other novel therapeutic strategies that specifically target the neurodegenerative aspect of MS include blockade of Na⁺ voltage-gated Ca²⁺ channels, Lingo-1 antagonism which is assumed to promote remyelination and axonal regeneration, erythropoietin (a hemopoietic growth factor commonly used to treat anemia), cannabis or statins [55].

As pointed out in this section, several approved therapies are available to decrease relapse frequency in RRMS patients. This bright hope for overtly inflammatory forms of MS contrasts with ongoing challenges in progressive forms, which include PPMS and SPMS. Many therapies with efficacy in relapsing MS have produced disappointing results in clinical trials of purely progressive MS. Indeed, primary neuroprotection has proven to be a harder nut to crack than anti-inflammation.

One of the challenges in developing neuroprotective therapies for progressive MS is a twofold predicament: What biologic pathway should be targeted, and what imaging metric should be used to test therapies that perturb these pathways? In the following section, we will briefly address some aspects underlying mechanisms of neurodegeneration in MS before we will move on describing why the thalamus might be an ideal imaging metric during neuroprotective clinical trials.

The cause of neurodegeneration in MS

While demyelinating foci within the thalamus might impair thalamus tissue integrity, indirect tissue damage might be operant as well (see next section). An important question is, thus, whether or not gray matter atrophy is indispensably linked to gray matter demyelination. On the one hand it has been shown that apoptotic neurons and transected neurites are significantly increased in the demyelinated compared to myelinated cortex [10]. In line with this finding there is evidence for substantial cortical neurodegeneration and generalized cell loss in progressive MS in association with meningeal inflammation and subpial demyelination [56]. On the other hand, neurodegeneration can as well be observed in the absence of demyelination [8, 57] implicating that both, direct and indirect mechanisms might result in neurodegeneration. Furthermore, examples exist in which inflammation causes neurodegeneration, neurodegeneration causes inflammation, inflammation and neurodegeneration appear to occur independently of one another and in which inflammation protects against neurodegeneration [58].

In preclinical studies, there are various mechanisms described to directly impact neurodegeneration in the inflamed brain, amongst them TRAIL (tumor necrosis factor related apoptosis inducing ligand), a lymphocyte-derived factor which induces apoptosis [59], FAS which induces apoptosis by binding to FAS ligand (CD95L) [60], perforin which induces membrane instability and subsequently apoptosis [61], or excess of the neurotransmitter glutamate that induces axonal calcium increase and thus axonal damage [62]. Furthermore, the extent and kinetics of remyelination are believed to be equally important predictors of neurodegeneration in MS. For example, it has been found that nerve conduction is restored with remyelination [63], demyelinated but not remyelinated lesions are characterized by acute axonal damage [64], experimental inhibition of remyelination in rodents results in a significant increase in the extent of axonal degeneration and loss [65], and extensive remyelination on the histopathological level correlates with functional recovery [66]. Shadow plaques, representing fully remyelinated lesions, demonstrate that complete repair of MS plaques is in principal possible [67], although it is common to observe only partial repair at the edges of lesions [68]. The reason why in some patients remyelination is widespread, while it remains sparse in others remains puzzling. However, it is an important prerequisite for developing effective therapeutic approaches [69]. A different possible cause of neurodegeneration in MS is brain parenchymal degeneration due to distant damage of afferent or efferent fibre tracts, such as Wallerian or trans-neuronal degeneration [70]. This aspect is less well studied but might be an important component of the disease and will, therefore, be discussed in the following part of the manuscript.

Trans-neuronal degeneration as underlying mechanism of thalamic atrophy

In some neurodegenerative disorders populations of neurons destroyed by a particular disease are embedded in functional networks. In Alzheimer’s disease, as well as in olivo-ponto-cerebellar atrophy, progressive supranuclear palsy, amyotrophic lateral sclerosis (ALS), primary autonomic failure of the Shy-Drager type, and other system degenerations, the main feature of the affected neuronal populations is their anatomical interconnectivity. One hypothesis to explain the specificity of such neurodegenerative processes was that neuronal populations at risk in each disorder share a common metabolic abnormality, e.g. a defect in neurotransmitter metabolism. In Parkinson’s disease, dopaminergic neurons in various brain regions are affected, such as those located in the substantia nigra, the locus coeruleus or the dorsal vagal complex [71]. Nevertheless, the melanin pigmented, dopaminergic neuroendocrine neurons in the hypothalamus remain spared in Parkinson’s disease [72], while neurons in other brain regions that do not synthesize dopamine are yet affected [73]. In other neurological disorders, affected populations of neurons apparently do not share any common neurotransmitter at all, such as degenerating glutamatergic upper motoneurons and cholinergic lower motoneurons in ALS, or pre- and post-ganglionic sympathetic neurons that are affected in Shy-Drager disease. The first are cholinergic, the latter adrenergic neurons. One has to conclude that additional mechanisms or a combination of other mechanisms must account for the characteristic pattern of neuronal involvement seen in the above-mentioned system degenerations. Alternatively, neuronal connectivity might determine the patterning of nerve cell loss. For example, the connectivity of the degenerating upper and lower motor neurons in ALS is well known, as is the functional connectivity of pre- and post-ganglionic nerve cells that degenerate in the primary autonomic failure of the Shy-Drager syndrome. The connectivity of the neurons most characteristically involved in other forms of system degeneration, such as Alzheimer’s disease, olivo-ponto-cerebellar atrophy or progressive supranuclear palsy, is more complex but equally well established [74].

In principal, three distinct biological phenomena can cause neuronal degeneration after disrupted neuronal connectivity namely (1) anterograde trans-neuronal degeneration, (2) retrograde trans-neuronal degeneration, or (3) Wallerian degeneration. To understand the underlying pathophysiology, it is essential to recognize that neurons not only communicate by passing an electrical signal from one to another but at the same time propagate trophic factors by means of anterograde and retrograde trans-neuronal transport [75]. The first evidence for trans-neuronal transport came, in fact, from studying the pathogenesis of viral disease and is in our days well established in the context of transmission of toxic agents through the nervous system [76, 77]. Trophic factors, such as BDNF or FGF-2, can be released by neurons [78, 79], and these trophic factors within the synaptic cleft can regulate pre- and postsynaptic neuronal function and survival [80–82]. Thus, the interruption of a neuronal network might result in growth factor depletion and in consequence neuronal cell death. The vital importance of an intact neuronal connectivity for the maintenance of neuronal integrity is further highlighted by a phenomenon called “trans-neuronal degeneration”. The term trans-neuronal degeneration describes the death of neurons resulting from the disruption of input from or output to other nearby neurons. Such damage might occur in an anterograde or retrograde fashion, indicating the direction of the degeneration relative to the original site of damage. Anterograde trans-neuronal degeneration is caused by a loss of input; it occurs when a damaged neuron causes the degeneration of a postsynaptic neuron associated with a similar function as the presynaptic neuron. This type of degeneration is often termed “dying forward”, and is also referred to as trans-synaptic degeneration. Anterograde trans-neuronal degeneration is, for example, operant during the death of pyramidal cells in the pirifom cortex (part of the rhinencephalon situated in the telencephalon) which undergo classical apoptosis within 24 h after bulbectomy [83]. Another example for anterograde trans-neuronal degeneration is elevated intraocular pressure during glaucoma causing progressive retinal ganglion loss and, subsequently, significant distant degeneration within the lateral geniculate nuclei. Such degenerative changes include deterioration of the dendrite architecture [84], shrinkage and loss of neurons [85, 86], reduction in neural metabolism [87], or changes in the expression pattern of several synaptic plasticity markers [88]. In contrast, retrograde trans-neuronal degeneration is degeneration caused by loss of trophic support from the target. It occurs in presynaptic cells that are sending inputs to injured postsynaptic cells. This type of neuronal cell death is often termed “dying backward”. Such type of degeneration is best described for retinal ganglion cells including retrograde degeneration of the retinal ganglion cells in patients with cerebral infarction [89] or experimental visual cortex ablation [90]. Dying backward degeneration has as well been described in other CNS regions such as the spinal cord [91, 92] or the midbrain [93].

Although it was postulated that trans-neuronal degeneration is generally not seen in adult animals [94], we now know that this is not the case. Johnson and colleagues [95] were able to show that trans-neuronal retrograde degeneration being operant in retinal ganglion cells following extensive striate cortical removal in young-adult macaque monkeys. Kataoka and colleagues [96] demonstrated nigral degeneration following striato-pallidal lesions, Ginsberg and colleagues [97] showed neurodegeneration in the septum after fimbria–fornix transection in adult rats, and DeGiorgio and colleagues [98] demonstrated that unilateral neurotoxin lesion of adult rat caudate-putamen and globus pallidus resulted in delayed, trans-neuronal degeneration of GABAergic substantia nigra pars reticulata neurons.

As pointed out above, trophic reciprocal interactions between target neurons and their afferents are thought to regulate neuronal survival and to determine the shape and size of axonal and dendritic processes [99]. However, it is unclear whether afferents that innervate more than one target neuronal population need to be trophically supplied by all of these populations to survive or whether a hierarchy among targets exists. Thus, the loss of the main target possibly leads to cell death, whereas removal of minor projection sites may only cause somatic and/or terminal modifications or even not affect the morphology of afferents at all. Experiments using wv/wv mutants (wv mutation lies in the kinase domain of the proto-oncogene c-kit), which have a complete granule cell loss in the vestibule-cerebellum, show that the granule cell target is not vital for vestibular ganglion neuron survival [100]. On the other hand, severe degeneration of inferior olivary neurons is evident in experimental models of cerebellar Purkinje cell loss. In these models, inferior olivary neurons die despite the continued presence of their other regularly innervated synaptic counterparts, such as the deep cerebellar interneurons and target cells in the deep cerebellar nuclei [101].

Disturbed axonal transport as underlying mechanism of thalamic atrophy

After having spotted the phenomenon that pre- and postsynaptic neurons support each other by releasing trophic factors, disturbed axonal transport might have an impact on neurodegenerative changes in the affected target areas. Axonal transport is an essential process in neurons, analogous to shipping goods, by which energetic and cellular building supplies are carried downstream (anterograde) and waste is carried upstream (retrograde) by molecular motors which act as cargo porters. Neuronal cell bodies and dendrites contain ribosomes and rough endoplasmic reticulum, but axons contain significantly less of the apparatus for the synthesis of new proteins [102]. Therefore, most proteins used in structural maintenance and synaptic transmission in the axon derive from the perikaryon. The process of axonal transport of proteins was first convincingly demonstrated by Droz and Leblond [103], who performed autoradiography on the sciatic nerve after systemic injections of radioactive amino acids. A wave of labeled proteins was observed traveling down the sciatic nerve. Subsequently, more detailed studies on the kinetics of the transport and the nature of the transported material have demonstrated two main phases of axonal transport, namely fast and slow axonal transport. In addition, axons also transport material in a retrograde direction. The retrograde transported material consists mainly of small membrane-bound vesicles and tubular structures moving along microtubules, at roughly the same rate as fast anterograde transport. There is good evidence that components of the axonal transport are disturbed in MS. The accumulation of amyloid precursor protein (APP), which is transported along axons in an anterograde fashion, indicates that there might exist a dysfunction of the axonal transport mechanisms in MS [104, 105]. In addition, abnormal accumulation of neurofilaments (NFs) occurs in MS which may cause structural instability to the axon [106]. Anterograde transport of proteins is mediated via the kinesin superfamily proteins (KIFs). There are presently 45 known members of the KIF family, 38 of which are neuronally expressed and the majority are involved in intracellular transport of newly synthesized proteins [107]. Hares and colleagues [108] observed that the levels of distinct kinesin proteins, among KIF5A, KIF21B and KIF1B were significantly decreased in MS tissues. Decreased axonal transport rates were, furthermore, reported at the onset of optic neuritis in EAE [109]. One has to keep in mind that decreased axonal transport rates might as well facilitate axonal survival as recently shown for local mitochondrial immobilization mediated by syntaphilin [110] and, thus, might be a pro-survival mechanism. Nevertheless, there is a growing body of evidence that axonal transport is disturbed in the brain of MS patients. Future systematic studies have to show whether this disturbed axonal transport mechanisms impacts on disease progression and accounts for to the spread of MS pathology in a ‘tract-specific’ pattern.

Wallerian degeneration as underlying mechanism of thalamic atrophy

A distinction must be drawn between trans-neuronal and Wallerian degeneration. Both mechanisms might impact on the integrity of target structures. Most structural axonal proteins are synthesized in the neuronal cell body and actively transported along the axon. Interruption of this supply causes a degenerative process known as Wallerian degeneration in the distal portion of the axon. Wallerian degeneration was originally described by Augustus Waller in 1850 based on his sentinel observations in transected glossopharyngeal and hypoglossal nerves [111]. Waller found that the distal nerve stump undergoes typical morphological alterations after transection which results in total nerve fiber fragmentation followed by disintegration. Although Waller`s description of degeneration was based on studies with transected peripheral nerves, the main features of Wallerian degeneration are also observed in many types of CNS insults. They are also present in the course of neurodegenerative and demyelinating diseases, suggesting a common triggering mechanism. Previous studies provided evidence that Wallerian degeneration, occurring as a consequence of axonal destruction in focal MS lesions, significantly contributes to early axonal pathology in MS patients [112, 113] and might be one characteristic for tissue alterations seen in the so called “Normal appearing white matter” or “diffusely abnormal white matter” [114]. Interestingly, there is evidence that Wallerian degeneration contributes to the recruitment of inflammatory cells into the CNS by altering the local microenvironment [115]. This mechanism might trigger the formation of new inflammatory foci.

Summing up, whether or not trophic reciprocal interactions are relevant for neuronal survival most likely depends on the neuroanatomical site of injury, and future studies have to show whether such mechanisms are relevant in the context of thalamus atrophy in MS patients. A recent study performed by Kolasinski and colleagues [116], however, strongly support this hypothesis. In their studies, the authors focused on two white matter tracts and their associated cortical and thalamic structures based on their distinct neuroanatomy and on their documented involvement in MS. First, the optic radiations between the lateral geniculate nucleus and primary visual cortex and second the component of the anterior thalamic radiations between the mediodorsal nucleus of the thalamus and prefrontal cortex (compare section “Neuroanatomy of the thalamus” in this review article). In line with the hypothesis that tract-specific patterns of pathology exist the authors performed a correlative analysis of markers of neurodegeneration both within and across the two distinct thalamo-cortical projection systems, specifically investigating the relationship between (1) MRI cortical thickness, (2) cell profile density in the thalamic nucleus and (3) the intensity of myelin staining within the connecting tract white matter. Interestingly, they found that all intra-tract associations were significant, such that the greater the degeneration in the cortical region, the greater the reduction in myelination in the connected non-lesional tract white matter and the greater the cell loss in the associated thalamic nucleus [116]. In support of these findings, lower retinal nerve fiber layer thickness was found to be associated with lower thalamus volume [117]. These data demonstrate the relevance of functional anatomical connectivity to the spread of MS pathology in a ‘tract-specific’ pattern.

In the following section, we aim at highlighting the most relevant findings of thalamus pathology in MS brains.

Thalamus involvement in multiple sclerosis

Histopathological alterations in the thalamus of MS patients

In 1983, Gilbert and colleagues [118] found in two out of five cases small plaques in the thalamus and brain stem. Comparable to cortical gray matter lesions, deep gray matter lesions are very difficult to detect in Luxol fast blue-stained sections, whereas they are easily recognizable using immunohistochemistry against myelin proteins. This fact may account for the low frequency of thalamic lesions observed in this study. Twenty-five years later, Vercellino and colleagues [119] performed a systematic neuropathological study focusing on the structural deep gray matter (i.e. caudate, thalamus, putamen, pallidum, hypothalamus, amygdala, and claustrum) in MS. In contrast to previous reports [15], they found that deep gray matter demyelination can be frequently observed in post-mortem MS brains, especially in the caudate and medial and anterior thalamic nuclei. In the 14 brains examined, the authors found 11 lesions within the thalamus. Most lesions involving the deep gray matter were leukocortical lesions, thus affecting white matter bundles and gray matter regions. In this study, lesions within the thalamus commonly seemed to follow the ventricular surface. Besides demyelination and inflammation, the authors observed clear signs of neurodegeneration including neuronal loss, neuronal shrinkage, and acute axonal damage, the latter evidenced by anti-amyloid precursor protein immunohistochemistry. These findings are in agreement with another study from Cifelli and colleagues. They additionally observed that neuronal loss might be the substrate for thalamic atrophy [24]. In the study of Vercellino and colleagues, neuronal loss was not just evident within the demyelinated thalamus but as well within the normal appearing one. Furthermore, mild diffuse inflammatory alterations were also frequently observed in the non-demyelinated deep gray matter distant from any focal demyelinating lesions. Such inflammatory changes have been hypothesized to be the consequences of a spreading of the autoimmune inflammatory process outside focal lesions [120] but may as well represent secondary responses to widespread anterograde and retrograde axonal degeneration.

One important aspect of this study is that inflammation within the deep gray matter shares similar pathological features with inflammation in the WM, however, is less intense, with a preponderance of activated microglia, scarce myelin-laden macrophages, and a lower degree of axonal injury and frequency of BBB disruption. Therefore, similar to cortical demyelination, deep gray matter demyelination is probably not easy to detect by conventional MRI because tissue destruction is less pronounced and the contrast between affected and non-affected deep gray matter regions is low. Post-mortem MRI studies evaluating the sensitivity of MRI for deep gray matter demyelination are lacking so far.

In summary, involvement of the deep gray matter in MS, particularly the thalamus, is frequent. As discussed above, neuronal loss and axonal pathology in the thalamus may relate to direct damage to neurons during inflammation or might be related to retrograde and anterograde degeneration of neurons, the axons of which had been transected in distant WM lesions. The later mechanism applies especially for neuronal loss observed within the normally myelinated thalamus.

MRI volumetry as a tool to study thalamus pathology in MS

MRI not only allows the non-invasive quantification of entire brain volume but also the quantification of specific gray matter nuclei volumes, e.g. of the thalami [28]. The degree of thalamic atrophy can indirectly be determined by evaluating the size of the third ventricle (compare Fig. 2). An enlarged third ventricle, if not resulting from an internal hydrocephalus, may thus be attributed to atrophy of the thalamus. However, more specific conclusion can be drawn, if thalamic volume is determined directly. A considerable number of studies observed thalamic atrophy in MS patients [32, 121, 122] which accumulates on a background of normal age-related reduction in thalamic volume [123]. Just as global brain atrophy starts early during MS disease course, thalamic atrophy was described already at earliest disease stages, such as in CIS patients [31, 124], in patients with early RRMS [125] or in an early PPMS patient cohort [126]. In pediatric MS patients, the only gray matter structures being reported to be atrophic are the thalamus and the globus pallidus [127, 128]. Findings of thalamic volumetry might have direct clinical implications: decrease of thalamic volume may serve as a predictor of the transformation from CIS to MS [28, 30, 31]; decline of cognitive performance in MS patients correlates with thalamus atrophy [20, 25]; thalamic volume loss in MS patients correlates with disability as determined by the EDSS score [34, 35, 125, 129]. In a recent study, Magon and colleagues [34] identified the thalamus to be the most relevant deep gray matter structure in predicting EDSS-disability. Moreover, their results demonstrate that within the thalamus, the sum volume of the VA, VL, and VP is the best predictor of EDSS scores. This is not surprising given that fact that the EDSS score mostly measures sensory and motoric functions which are represented in the VA, VL and VP, respectively.

Fig. 2.

Axial T2-FLAIR (a) and coronal T1w (b) MRI of a 56-year-old female patient with a secondary progressive MS. Widespread confluent white matter lesions (a arrowheads), widening of the lateral ventricles (b asterisks). The enlarged third ventricle with convex borders (b thin arrow) may be an indirect sign of thalamic atrophy. Normal width of superficial sulci including the sylvian fissure (b thick arrow) indicates that there is no generalized cerebral atrophy. c Axial T1w MRI of a 43-year-old female patient with a primary progressive MS and a progressive cognitive decline. The enlarged third ventricle with convex borders (c arrow) is an indirect sign of thalamic atrophy which is clearly visible despite an overall low image quality

Despite these promising findings that thalamus volume changes reflect distinct clinical phenotypes, there are some caveats in volumetric MRI studies that need to be taken into account: Brain volume loss estimations might be affected by pathophysiological factors such as age and hydration status [130], inflammation, and immune-modulatory therapies [131]. Technical variations such as field strength and inhomogeneity, image contrast and signal-to-noise ratio influence the data. Increased iron deposits are described in MS patients’ thalami, possibly influencing the signal intensity of T1w images and causing bias on deep gray matter segmentations. Furthermore, the algorithms applied in the various automatized procedures are not comparable with each other, neither are the volumetric outcomes between different software packages, causing variability between studies [132].

Diffusion imaging as a tool to study thalamus pathology in MS

While volumetric MRI studies provide rather morphologic information, diffusion-weighted imaging (DWI) allows insights into the tissue integrity and structure of the brain parenchyma. Diffusion tensor imaging (DTI) additionally enables the estimation of fibre tracks and thus the visualization of the connectivity between various brain regions. In simple terms, diffusion, described by mean diffusivity (MD), reflects the molecular movement of water molecules that is dependent on cell size and structure and normally restricted by cell boundaries. In myelinated fiber tracts, the diffusivity has a strong directionality reflected by high values of the fractional anisotropy (FA)—a parameter that is reduced in both, demyelinated lesions and NAWM of MS patients. Studies in animal models have shown that two other coefficients, the parallel diffusivity (D_∥; synonymous axial diffusion, AD) and the perpendicular diffusivity (D_⊥; synonymous radial diffusion, RD) provide additional information on white matter structures that is more specific to underlying histological processes as compared to FA or MD [133]. D_∥ may reflect diffusivity along the axon (i.e. axonal integrity) whereas D_⊥ represents diffusivity perpendicular to the axon (i.e. myelination). A reduction of the mean diffusivity parallel to the fiber tracts (D_∥) has been observed after axonal damage in different animal models such as peri-contusional traumatic injury and in experimental autoimmune encephalomyelitis (EAE). On the other hand, an increase of the diffusivity perpendicular to the fiber tracts (D_⊥), and to a minor extent also to D_∥, has been demonstrated in a number of demyelinating and dysmyelinating animal models [134]. The main clinical application of DWI has been in the study and treatment of neurological disorders, especially for the management of patients with acute stroke [135]. Because DTI can additionally reveal abnormalities in white matter fibre structures and provide models of brain connectivity by fiber tracking algorithms, it is rapidly becoming a standard for white matter disorders [136, 137].

Numerous studies have applied DTI to characterize microstructural damage in cerebral WM. Most authors demonstrated abnormal low fractional anisotropy and high radial diffusivity in MS patients suggestive for an alteration in the myelin architecture [138, 139]. Axial diffusion, however, has been reported to be both decreased and increased [140, 141]. A decreased axial diffusion is usually regarded as a sign of axonal damage [142]. Such DTI changes are known to occur in CIS as well as very early stages of MS, and to precede visible structural changes [143, 144]. DTI measurements focusing on the thalamus as region of interest detected abnormalities as well, including decreased thalamic fractional anisotropy in early MS [145], or a progressive increase of thalamic mean diffusivity combined with progressive decrease of thalamic fractional anisotropy in PPMS [146]. These results are in line with others based on different MS cohorts [147]. From a clinical perspective, thalamic DTI changes could be correlated with EDSS progression and cognitive decline [148].

Perfusion MRI as a tool to study thalamus pathology in MS

Dynamic susceptibility contrast enhanced perfusion MRI (DSC-MRI) requiring contrast agent injection or non-invasive arterial spin labelled perfusion MRI (ASL-MRI) can indirectly address brain activity by detecting associated changes in blood flow, relying on the fact that cerebral blood flow and neuronal activation are coupled: When a certain brain region is active, the local blood supply also increases. In MS in contrast to actively inflamed lesions, chronically reduced blood flow has been reported both for the whole brain [21] and for deep gray matter in particular, including the thalamus [24, 149]. The recent observation of increased thalamic blood flow in verbal memory impaired CIS patients might indicate that during the early disease stages compensatory upregulation of local blood flow might occur [150], which fails in more advanced disease stages [23]. Furthermore, Ota et al. [21] reported that thalamic blood flow decreased with increasing white matter lesion burden. Since the thalamus and the largest portion of the white matter do not generally share local blood supply, this finding may be interpreted as a decreased thalamic activity in the context of retrograde and anterograde degeneration rather than common regional haemodynamic changes. Definitely, future haemodynamic studies are warranted to clarify the relevance of altered thalamus perfusion in MS patients.

Functional MRI as a tool to study thalamus pathology in MS

Functional MRI (fMRI) measures neural activity based on changes in blood flow and deoxy-hemoglobin levels (blood oxygen level dependent; BOLD). In simple terms, the blood flow and oxygen metabolism in active cerebral areas cause MRI signal changes that serve as a surrogate for brain activity. Two general approaches are used: (1) activation fMRI which measures BOLD modification during specified tasks, and (2) resting state fMRI which correlates the synchrony of low frequency fluctuations of the BOLD signal in various regions while the brain is at rest. This technique can be used to determine functional connectivity of neural networks [151]. Simultaneous co-activation of cerebral areas indicates functional connectivity both in resting state and after given tasks. Some authors reported an increased connectivity within the thalamic resting state and motor network, pointing towards a reorganization of this network [152–154]. Increased subcortical motor connectivity in MS may therefore reflect a remodeling of the subcortical motor network compensating tissue damage. However, further studies are necessary to investigate the changes of the functional thalamic networks in MS.

MR spectroscopy as a tool to study thalamus pathology in MS

MR spectroscopy (MRS) quantifies metabolic parameters including markers i.e. specific for neurons (N-acetylaspartate; NAA) or astrocytes (myo-inositol; mI). It is therefore assumed that decreased levels of NAA and increased levels of mI reflect neuronal damage and gliosis. While spectroscopic data dealing with thalamic changes in particular are sparse, some authors reported spectroscopic changes suggestive for thalamic damage [24, 155, 156]. Noteworthy, Geurts and colleagues [157] found a significant correlation between total cerebral T2 lesion load and increased thalamic mI levels, possibly reflecting local gliosis. Since great care was taken in this study to exclude lesions detectable on the T2-weighted images, thalamic gliosis is less likely to be due to local inflammatory processes. A possible explanation for their finding could be that remote structural damage to WM networks induces trans-synaptic axonal degeneration and accompanying gliotic alterations within the thalamic nuclei [157]. While data concerning MRS are promising, further studies dealing with thalamic spectroscopic changes are needed for a better understanding of thalamic damage in MS, and its functional significance.

In summary, the paraclinical assessment of thalamus pathology is a powerful tool to study underlying mechanisms of neurodegeneration in MS. The best way to visualize and quantify thalamic pathology, whether inflammatory or neurodegenerative, has not been worked out and, thus makes the comparison of different studies somewhat difficult. However, it is our strong belief that higher resolution MRT will be routinely available in the near future which will further help to visualize subtle changes of tissue integrity and function within the deep gray matter including the thalamus.

Thalamus involvement in animal models of multiple sclerosis

With respect to the various animal models applied to study MS pathogenesis and disease progression, relatively little is known about the involvement of the thalamus in these models, which might be due to its complex neuroanatomical architecture. However, this is important for our understanding which animal model should be applied to study distinct aspects of thalamus pathology in MS. To be more precise, if lesion formation within the thalamus is a characteristic feature of a distinct MS animal model, the contribution of trans-neuronal or Wallerian degeneration for thalamus pathology in this model is most likely hard to dissect. For such a study, animal models are warranted which do not show a direct involvement of the thalamus. In this section, we briefly outline the various animal models used to study MS pathogenesis and to develop new therapeutic opportunities. For more detailed information, we refer to recently published review articles from our group and others [158–161].

Frequently applied MS animal models

The similarities between MS pathology and viral demyelinating disorders of the CNS as well as epidemiological observations have made the infectious etiology of MS an attractive hypothesis. Electron microscopic and virological studies have supported this by revealing the presence of viruses in MS brain tissues. Three distinct viruses are commonly used to study viral mediated demyelination namely Semliki Forest virus (SFV), Theiler’s murine encephalomyelitis virus (TMEV), and the Mouse hepatitis virus (MHV). SFV is neuro-invasive, crosses the BBB, and infects predominantly oligodendrocytes and neurons [162]. TMEV is a neurotropic murine picornavirus, which, depending on the strain, causes either acute encephalitis or a persistent demyelinating disease. Since demyelination is also observed in athymic nu/nu mice, direct viral damage to myelin appears to contribute to the observed pathology in TMEV [163]. Infection of rodents with MHV may result in lethal encephalitis or paralytic demyelinating disease resembling the human disease MS [164]. Demyelination in MHV-infected animals has been attributed to the cytolytic effects of viral infection on myelin-producing oligodendrocytes but more recent evidence supports an immune-pathological mechanism for demyelination.

Immunization of susceptible animals with CNS antigens gives rise to a spectrum of inflammatory disorders collectively named experimental autoimmune encephalomyelitis (EAE). Despite major differences in disease course and pathology, EAE is still the most intensely used experimental MS model. Immunization of susceptible animals with CNS tissues or related peptides and adjuvant elicits either a chronic paralysis from which the animals do not recover, or chronic-relapsing neurological episodes with progressive disability. EAE induced by immunization is referred to as actively induced EAE. Alternatively, EAE can be induced following adoptive transfer to naive recipients of lymph node or spleen cells, or specific T cell lines and clones from immunized animals. This is termed adoptive transfer or passive EAE. While many animal species and strains are susceptible to EAE, the availability of a plethora of biological tools to probe the disease and the availability of genetically engineered animals makes the mouse ideal as a host for an experimental model of MS. Most EAE studies are either carried out using C57Bl/6 animals immunized with MOG_35–55 peptide or SJL animals immunized with PLP_193–151 peptide. Histopathological alterations in EAE include peripheral cell recruitment, demyelination, reactive gliosis and neurodegeneration. Such alterations are predominant in the spinal cord, cerebellum and brain stem.

A third common group of animals models used to study MS-related aspects are toxin models. In principal, toxin-mediated demyelination can be induced by either focal application or systemic administration of the toxin. While focal demyelination is usually induced by injection of lysolecithin (also called lysophosphatidylcholine) or ethidium bromide, the copper-chelator cuprizone is used for systemic demyelination. In certain mice strains, the ingestion of the copper chelator cuprizone induces a consistent and highly reproducible chemically induced demyelination, predominantly of the corpus callosum. This commissural white matter tract represents the most frequently investigated region in this animal model. After 5–6 weeks of cuprizone treatment, the corpus callosum is almost completely demyelinated, a process called ‘‘acute demyelination’’. Acute demyelination is followed by spontaneous remyelination during subsequent weeks when mice are fed normal chow. In contrast, remyelination is highly impaired/delayed when cuprizone administration is prolonged (12–13 weeks or even longer), a process called ‘‘chronic demyelination’’. This fact makes the cuprizone model an attractive tool to study remyelination biology [165]. Notably, cuprizone-induced demyelination does not require the recruitment of peripheral immune cells. Thus, this model is ideal to study non-immune related aspects of MS.

Thalamic alterations

As outlined above, various animal models exist to study MS pathogenesis and concomitant disease progression. Since MS is a very complex and heterogeneous disease, all of these models are valuable and represent important aspects of MS.

In viral models, the thalamus appears to be involved during the disease process. MHV viral RNA has been detected within the thalamus [166] which is not surprising due to the intense connections of the thalamus with other parts of the CNS. Thalamic alterations have well been documented in the Theiler’s murine encephalomyelitis virus model [167, 168]. In the study of Pirko and colleagues [169], quantitative MRI analysis revealed a correlation between the degree of T2/T2* thalamic hypo-intensity and rotarod-defined disability.

MOG_35–55 induced EAE in non-obese diabetic (NOD) mice results in lesions of the thalamus visualized noninvasively by MRI [170]. In C57BL/6 MOG_35–55 induced EAE, it had been observed that mast cells accumulate at the thalamic border [171]. Similar results have been reported in a rat model using Lewis rats immunized with spinal cord homogenate emulsified in Freund’s complete adjuvant [172, 173]. Beyond, increased NGF levels were found in the thalamus of EAE-diseased rats [174–176]. Furthermore, in EAE animals, there were higher thalamic expression levels of the mast cell markers, c-kit and CD40L, as well as astrocyte marker GFAP. Meuth and colleagues [177] were able to show that neuronal TWIK-related acid-sensitive potassium (TASK) channels, which regulate efflux of potassium ions and by this can orchestrate neuronal death through intracellular potassium depletion, are lower expressed in the thalamus of rats undergoing MOG-induced EAE. In summary, the thalamus appears to be affected in distinct EAE models, however, such changes are relatively mild as compared to the pronounced pathology observed in other brain regions such as the spinal cord or the cerebellum [178]. Whether or not the intense EAE lesions distant to the thalamus induce thalamic neurodegeneration by the above-mentioned processes was not yet systematically addressed.

Already in 1971, Kesterson and Carlton [179] described edematous vacuolization, creating spongiform tissue alterations 5 days after initiation of the cuprizone diet which was widespread but prominent in distinct brain regions, amongst them the thalamus. Since then, the thalamus as a region affected by the cuprizone intoxication received minor attention. Most studies focused on the white matter demyelination in this model. Nevertheless, it is obvious that cuprizone intoxication induces profound cellular changes within the thalamus, such as reduction in oligodendrocyte numbers [180], microglia activation and demyelination (see Fig. 3), or decreased levels of several metabolites, i.e. N-acetylaspartate or N-acetyl-aspartyl-glutamate [181].

Fig. 3.

Effect of cuprizone on myelination and microglia activation of the thalamus in male mice demonstrated by means of anti-PLP and anti-Iba1 immunohistochemistry, respectively. a Schematic illustration of the thalamus in mice. The inner region delineated by the green line in the left part of the picture shows the thalamus. Within the thalamus formation, the area highlighted in yellow in the right part of the picture delineates the ventral posterolateral nucleus (VPL), the ventral posteromedial nucleus (VPM), the ventral anterior (VA) and ventro-lateral (VL) nucleus of the thalamus, respectively. b Anti-PLP staining of a control mouse, c anti-PLP staining of a mouse treated 5 weeks cuprizone (0.25 %). The same regions are shown directly below at higher magnification. Note the significant demyelination within the four thalamic relay nuclei. Arrowhead points at an afferent/efferent fiber tract, whereas the arrow points at a neuronal rich gray matter part of the respective thalamus region. Both sub-compartments are affected. d Anti-Iba1 staining of a control mouse, e anti-Iba1 staining of a mouse treated 5 weeks cuprizone (0.25 %). Note the severe hypertrophy and hyperplasia of microglia cells within the four thalamic relay nuclei

In summary, the thalamus is presumably directly involved in the disease process induced by global cuprizone intoxication. This means that cuprizone directly attacks oligodendrocytes within the thalamus and, in consequence, induces demyelination and gliosis. In contrast, the thalamus appears not to be directly involved in the disease process in the EAE model. Bearing in mind that thalamus atrophy in MS patients might occur either because lesions develop within the thalamus or because of trans-neuronal degeneration, the cuprizone model is a valuable tool to study the first aspect, whereas the EAE model is best suitable to study the later one. Future studies are now warranted to address to what extent the intense lesions within the cerebellum and spinal cord in EAE induce atrophy of the thalamus, which mechanisms are operant and how such trans-neuronal deleterious effects can be ameliorated by pharmacological intervention.

Conclusion and future perspectives

The paired thalamic nuclei are gray matter structures on both sides of the third ventricle, and they are involved in a wide range of neurologic functions that include motor, sensory, integrative, and higher cortical functions. Thalamic location, widespread cortical and subcortical connections, and vulnerability to MS-related pathologic changes from the earliest clinical disease stages make the thalamus a critical structure for examining neurodegeneration in MS. As pointed out in this review article, all currently approved MS therapeutics primarily target inflammation. However, the awareness that MS is as much neurodegenerative as inflammatory shows that an optimized therapeutic approach should specifically tackle the promotion of neuroprotection and repair to prevent irreversible chronic disability. This underpins the necessity to carefully choose primary and secondary outcomes during the design of future trials of neuroprotection. Given that progressive pathology within the thalamus has been shown in all different MS disease types, that thalamus damage correlates with various clinical parameters, and that thalamic volume loss has also been detected in pediatric MS patients, the measurement of thalamic atrophy may become an ideal MR biomarker in MS neuroprotective clinical trials [31]. Future preclinical and clinical studies have to provide definitive proof whether or not the thalamus indeed serves as a surrogate biomarker for the amount of global neurodegeneration in the entire brain of MS patients, and which mechanisms are operant during this process. It is our strong belief that mechanisms discussed in the review article are operant during thalamus pathology. This hypothesis should be tested in the future using appropriate experimental model systems. For example it would be seminal to know whether the ventral posterior complex of the thalamus, which is the main relay for sensory inputs to reach the cerebral cortex, displays signs of neurodegeneration in MOG-induced EAE. Since the spinothalamic tract, which transmits sensory information from the spinal cord to the thalamus is frequently demyelinated in this model, one should expect degeneration of this part of the thalamus. Another experimental approach would be to induce focal lesions within thalamic afferent pathways by for example stereotactic lysophosphatidyl choline (LPC) injection and subsequently studying the presence or absence of thalamus pathology. Once such models are established, mechanistic studies will tell us which cell types and factors are involved in this process.