Functional recovery and late onset motor decline after remyelination 104

cuprizone-induced demyelination in mice. The aim of this study was to establish an animal model that mimics some aspects of progressive MS, characterized by late-onset progressive neu-rological deficits, often in the absence of active inflammation (Kremenchutzky et al., 2006;

Lassmann et al., 2007; Miller and Leary, 2007). Remyelination has been shown to be extensive in the cuprizone model (Blakemore, 1973b), but the long-term impact of myelin repair on neuro-axonal function is not known.

4.2.1 Functional recovery and late onset motor decline after remyelina-tion

Due to the localized and reversible demyelination induced in the cuprizone model, it has been difficult to detect phenotypes related to the myelination and functional state of the corpus callosum, especially after remyelination. A deficit in the prepulse inhibition of acoustic startle response has been reported, but only during cuprizone feeding (Xu et al., 2009, 2010). After recovery, a decreased anxiety and increased interactive behavior, as well

as a higher frequency of falls in the rotarod test have been reported (Franco-Pons et al., 2007; Hibbits et al., 2009). However, the interactive behaviour test should be done in naive animals and is most likely not optimal for longitudinal studies (Bolivar et al., 2007).

On the other hand, the rotarod test is not likely to be solely related to the myelination status of the corpus callosum, as differences were observed after 6 weeks of remyelination but not at 6 weeks of cuprizone feeding, when demyelination is much higher (Franco-Pons et al., 2007). In addition, it is difficult to target and follow the specific brain structures correlated with these behaviors.

Impaired performance in complex running wheel tests has been described in mice with genetically-absent or surgically removed corpus callosum (Schalomon and Wahlsten, 2002).

The corpus callosum consists of the axons of cortical projection neurons and facilitates the communication of both the cerebral hemispheres, and the associative connectivity, to exe-cute complex motor and cognitive tasks. Most of these axons are myelinated and therefore make up the largest white matter structure in the placental mammalian brain (Aboitiz and Montiel, 2003; Fame et al., 2010). One of the few methods shown to detect latent deficits in animals after remyelination in the cuprizone model is the motor skill analysis or MOSS (Liebetanz and Merkler, 2006). In this test, mice ran voluntarily in wheels with irregularly spaced crossbars, therefore requiring a constant step length adaptation and bi-hemispherical coordination.

To analyze the long-term effect and functional recovery of cuprizone-induced demyelina-tion, a motor skills evaluation was performed three times over a period of 28 weeks using the MOSS running test (Liebetanz and Merkler, 2006). In this study, we reproduced earlier findings showing that latent motor deficits in MOSS parameters, such as maxi-mum velocity (Vmax) and maximaxi-mum distance (Dmax) are still evident after 6 weeks of recovery following cuprizone diet (Liebetanz and Merkler, 2006). We further re-examined the animals 14 weeks later (i.e. 20 weeks after cuprizone treatment removal). Locomo-tor performance of remyelinated animals was indistinguishable to age-matched controls by MOSS at this time point. These results indicate that remyelination together with, most probably, neuroplastic processes, can fully compensate functional disability after a demyelinating insult.

To further investigate whether functional recovery remained stable in the long-term, we repeated MOSS analysis at 28 weeks after cuprizone removal. At this latest time point, remyelinated animals, independent of single or repeated exposure to cuprizone, displayed deteriorated locomotor performance as compared to age-matched controls in the coordi-native parameters Vmax and Dmax. Similar to earlier time points, no differences were detected in parameters assessing general motivation and/or fitness compared to controls and no differences in any of the parameters were observed between the treated groups.

These data show that mice fed with cuprizone recover completely as measured by MOSS, but develop late-onset functional deficits at advanced age. One possibility could be that after the initial deficiency detected after 6 months, the treated animals entered a steady state, and their performance level is ultimately reached by the decreased performance of control animals due to aging, and this could account for the similar performance of all groups 20 weeks after cuprizone removal. However, with time, cuprizone- treated animals exhibited a continued decline, evidenced in the last timepoint of the MOSS analysis.

Although it is not possible to describe the rate of the decline beyond the three time points analyzed, the data show that there is indeed a late-onset motor decline observed in the treated mice between 20 and 28 weeks of cuprizone treatment.

A decline in performance after a period of stability in the first 20 weeks could be explained by two independent factors: compensatory mechanisms that facilitate the functional re-covery and a steady accumulation of structural damage. A tipping of the balance between these two processes upon reaching a certain threshold may unmask latent damage. Such a mechanism has been proposed to explain the functional recovery in MS patients during relapsing-remitting stages of the disease and the irreversible clinical decline observed in the progressive phases (Trapp et al., 1998; Bjartmar and Trapp, 2003). Four components appear to be involved in maintaining this balance and contribute to the clinical readout:

axonal damage, resolution of inflammation, remyelination,and cortical plasticity.

4.2.2 Axonal damage as a driver of motor decline

Most of the axonal damage induced by cuprizone-mediated demyelination occurs during the treatment, and decreases during recovery (Hoehn et al., 2008; Lindner et al., 2009).

In our study, axonal damage, measured by APP accumulation, was at a maximum after 5 weeks of cuprizone treatment (Figure 3.26). It is intriguing, however that even at

∼ 6 months after the treatment with cuprizone was ceased, APP-positive axons are still detected. Even more interestingly, some of these APP-positive axons were still surrounded by a myelin sheath. Furthermore, we also observed enlarged mitochondria in axons in this last time point, ∼ 6 months after cuprizone had been removed from the diet, providing further evidence of persisting axonal pathology (Kiryu-Seo et al., 2010).

The axonal damage described in our model may be the result of several processes. First, it is possible that cuprizone-mediated toxicity is not selective to oligodendrocytes, but may also work directly against the axon, as many axons are lost during the treatment with cuprizone. There is mounting evidence that axonal damage most importantly decreases with remyelination efficiency and increases with the level of microglia activation (Irvine and Blakemore, 2008; Tsiperson et al., 2010; Yoshikawa et al., 2011). Inflammation level and axonal damage correlate in acute and chronic MS lesions (Bitsch et al., 2000; Kuhlmann et al., 2002). Microglia are known to perform dual roles that have beneficial positive and detrimental consequences for axonal survival. Accordingly, understanding the pathways that stimulate/inhibit one response over the other would be useful in modulating the net contribution of microglia to a demyelinating episode, in order to preserve axonal integrity (Hanisch and Kettenmann, 2007).

On one hand, microglia act as macrophages, clearing myelin debris, which is necessary for efficient remyelination to occur. On the other hand, activated microglia release cytokines and NO. These can contribute to a reversible blockage of axonal conduction, exacerbate the inflammatory effects and lead to axonal damage and subsequent loss (Smith et al., 2001;

Aboul-Enein et al., 2006). Therefore, it is possible that the low yet sustained presence of astrocytes and microglia observed in our study can contribute to the low ongoing damage.

The decrease of axonal counts observed in our study, suggests that microglia presence could be induced by an ongoing axonal degeneration and could be contributing to debris clearance without the production of harmful pro-inflammatory signals (Neumann et al., 2009).

Following axonal transection, regeneration in the CNS is considerably limited, due to the presence of inhibitory factors in the extracellular space. Oligodendrocyte molecules such as NogoA, MAG and OMgp have been identified as axonal growth inhibitors (Kotter et al., 2006; Baer et al., 2009). These three molecules bind to the Nogo receptor (NgR)

present in axons, which associates with the receptor p75 and LINGO-1 (Wang et al., 2002;

Mi et al., 2004) and is responsible for neurite outgrowth inhibition and prevents axonal regeneration in the CNS (Fournier et al., 2001). In addition, myelin debris and associated proteins have also been found to inhibit oligodendrocyte differentiation, a crucial step for remyelination to occur (Kotter et al., 2006; Syed et al., 2008). A failure in remyelination has been repeatedly associated to an increased axonal loss in the cuprizone model and progressive MS, and is currently one of the major targets in MS treatment (Smith, 2006;

Franklin and Ffrench-Constant, 2008; Irvine and Blakemore, 2008).

Further evidence that supports the idea that myelin integrity itself may be necessary for axonal maintenance stems from studies showing that mice lacking different myelin-specific proteins suffer from late-onset neurodegeneration, despite the fact the myelin appears structurally normal (Griffiths et al., 1998; Lappe-Siefke et al., 2003). In different demyelinating animal models, age can negatively affect the recruitment and differentiation of OPCs after demyelination (Shields et al., 1999; Sim et al., 2002), resulting in a decreased remyelination capacity and more extensive axonal damage (Hampton et al., 2012). In our study, we observed an extensive yet incomplete demyelination. In addition to the expected increase in the g-ratio, due to a thinner myelin sheath formed after remyelination, we observed that a population of axons (approximately 26%) remained unmyelinated.

Aside from decreasing neuronal energy demands by facilitating saltatory conduction, myelin regulates the axonal diameter (de Waegh et al., 1992), fast axonal transport (Edgar et al., 2004), and the molecular organization of the nodes of Ranvier (Peles and Salzer, 2000). Therefore, chronically demyelinated axons undergo alterations in structure and function, demand more energy and become more vulnerable to degeneration (Irvine and Blakemore, 2008; Lindner et al., 2009). Despite the axonal loss observed after cuprizone treatment, remyelination seems to contribute to the axonal preservation of remaining axons, as the majority of the APP-positive axons observed after 6 months appeared un-myelinated.

However, it is striking that, in both the cuprizone model at 6 months after treatment removal, as well as in chronic lesions from MS patients there was a constant fraction of axons that displayed APP accumulation yet had a surrounding myelin sheath. It could be possible that the axon is not myelinated along its entire length and the pathology is

caused by insufficient myelination in adjacent regions. This is difficult to prove due the fact that axons crossing the corpus callosum can be over 2 mm long (Wahlsten, 1984), and the tightly packed structure of the corpus callosum that precludes quantifying the fraction of an entire axon that is myelinated.

Another option is, that despite the fact that remyelination contributes to axonal preserva-tion and funcpreserva-tional recovery (Duncan et al., 2009), alterapreserva-tions to the newly formed myelin, such as the change in thickness, internodal length and protein or lipid composition, could compromise axonal survival in the long term. Minor alterations in myelin membrane com-position can trigger neurodegeneration (Nave and Trapp 2008). Thus, it could be possible that the changes in myelin composition occurring after episodes of demyelination and re-myelination are sufficient to induce neuronal dysfunction that only become apparent when mice age.

Interestingly, similar changes in the proteome of myelin from animals that have undergone remyelination and in myelin from old animals has been found (Manrique-Hoyos et al., 2011). This does not exclude the possibility that the minor alterations occurring in re-myelinated myelin results in subtle changes in axonal function that may only become functionally relevant after a certain age. In addition, the induction of two episodes of cuprizone-induced demyelination in our study did not increase late-onset axonal dysfunc-tion and motor decline in our model compared to a single demyelinadysfunc-tion. This finding indicates that the critical threshold is already reached after one round of demyelination and remyelination. Interestingly, in MS the onset of the progressive phase is independent of number of relapses and age is the most important risk factor that determines when chronic progressive MS is set off (Confavreux et al., 2003; Kremenchutzky et al., 2006).

While the reversible disability observed during initial inflammatory episodes is caused in part by a transient conduction block due to the edema that accompanies the infiltration through the BBB, in chronic stages or SPMS the progressive disability seems to be corre-lated with irreversible axonal degeneration. It is widely accepted that axonal loss is a key determinant for permanent disability beginning at disease onset and correlating with the degree of inflammation within lesions in patients with MS (Bitsch et al., 2000; Bjartmar and Trapp, 2003) and can be observed both in inflammatory and chronic demyelination (Trapp et al., 1998; Dutta et al., 2006). In this study, we showed that a cuprizone-induced

demyelination causes a massive axonal loss in the corpus callosum without a reduction of cortical neuronal soma. Interestingly, in MS axonal loss is more frequently found than loss of neuronal cell bodies (Trapp and Nave, 2008). We measured massive axonal loss, evidenced by the shrinkage of the corpus callosum thickness, a decrease in neurofilament signal and in axonal counts. This loss seems to continue occurring beyond 6 weeks of remyelination indicating a progressive axonal loss.

The capacity of cortical adaptation after injury has been demonstrated in both experi-mental models and in MS patients. These studies have shown cortical adaptive changes occurring concurrently with the progression of axonal injury (Reddy et al., 2000; Faivre et al., 2012). Once structural damage is so extensive that compensation mechanisms are insufficient to correctly execute a given task, functional deficit will become evident. If the axonal pathology is indeed more frequent early in disease, it is possible that the brain has a greater capacity to compensate and is able to recover from early axonal damage.

As axonal loss continues and the brain ages, a critical threshold may be reached, where compensatory mechanisms are exhausted, and clinical symptoms reappear, unmasking the long-term consequences of the demyelinating insult (Bjartmar and Trapp, 2003; Trapp et al., 1999).The mechanisms limiting the extent of compensation of axonal pathology that ultimately triggers the transition from RRMS to SPMS remain unclear.