Restoring the balance between disease and repair in multiple sclerosis: insights from mouse models

Multiple sclerosis (MS) is an autoimmune disease that disrupts brain and spinal cord function (Prineas, 1985; Raine, 1984). In many cases, the early stages of the disease are characterized by episodes of functional deficits that are followed by recovery periods with few residual symptoms (remissions). As the disease progresses, relapsing episodes ultimately lead to reduced recovery and increased functional deficits. In the majority of cases, later stages of the disease involve chronic loss of function and increasing disability. The underlying mechanisms generating functional deficits in MS appear to result from localized damage to myelin sheaths in the brain and spinal cord (Prineas, 1985; Raine, 1984) (Fig. 1).

In the vertebrate CNS, myelin is generated by oligodendrocytes. Myelin is a discontinuous fatty insulation around axons that helps lower the threshold for neuronal activation and increase the speed of action potential propagation along axons as a result of saltatory conduction. An individual oligodendrocyte can myelinate multiple segments along multiple axons and therefore damage to a relatively small number of oligodendrocytes can result in impaired axonal conduction, axonal loss (Trapp et al., 1998) and functional deficits (Bando et al., 2008).

Demyelination and oligodendrocyte loss in MS are largely a consequence of immunological attack on neural tissue. Thus, most current therapeutics are designed to suppress the immune system. Common therapeutics include immunosuppressive cytokines such as interferon β (Dhib-Jalbut and Marks, 2010), the immunomodulator glatiramer acetate (Racke et al., 2010), monoclonal antibodies such as natalizumab and alemtuzumab that inhibit immune cell entry to the CNS (Bielekova and Becker, 2010; Coles et al., 2008), and cytotoxic agents including mitoxantrone (Vollmer et al., 2010). Recent studies have shown that orally available immunomodulatory therapies such as cladribine might hold some promise for patients since it can lower the number of lesions and reduce annual relapse rates (Giovannoni et al., 2010; Rammohan and Shoemaker, 2010).

Although most of these approaches decrease the frequency of relapses, they have little effect on the progression of functional deficits in a large proportion of MS patients. Long-term improvement may require the stimulation of repair mechanisms to salvage damaged myelin and axons. Perhaps a combinatorial approach that targets immune-mediated insults and also enhances myelin repair will provide the most effective long-term benefits for patients.

The fundamental mechanisms underlying oligodendrocyte development and myelination are closely conserved between mouse and human, making the mouse an invaluable model system for myelination and demyelination studies. In both organisms, oligodendrocytes arise from neural stem cells in distinct regions of the vertebrate CNS in response to specific inductive signals, such as sonic hedgehog (Miller, 2002; Orentas and Miller, 1996; Pringle et al., 1996). These oligodendrocyte precursor cells (OPCs) then disseminate throughout the CNS in response to specific dispersal cues (Tsai et al., 2003) before populating axon-rich developing white matter, where they undergo extensive proliferation and differentiation (Miller, 2002). Myelination of receptive axons follows a specific sequence that appears to be regulated by axon-glial interactions (Emery et al., 2009; Watkins et al., 2008). Proliferation of OPCs in both mouse and human is mediated largely through platelet-derived growth factor (PDGF) (Richardson et al., 1988), and the induction of myelination reflects the removal of inhibitory cues expressed on axons and OPCs such as LINGO-1 (Mi et al., 2007; Mi et al., 2005), PSA-NCAM and NOTCH (Wang et al., 1998). Developmental studies using chimeric animals have been instrumental in defining molecular cues shared between mouse and human (Windrem et al., 2008).

Understanding the adult pathology of MS using standard mouse models is challenging. In MS patients the onset of disease is difficult to determine, the disease progression varies, and the genetic linkage, while evident, is not easily defined (Oksenberg et al., 2008). Mouse models of MS have been developed that reproduce many of the immunological aspects of the disease, but the processes of demyelination and remyelination are less extensive and more acute than in the human condition. In order to facilitate the discovery of drugs to promote myelin repair, new mouse models are required. Current mouse models can be broadly classified into three major classes: (1) genetic models where genes important for CNS myelination have been manipulated; (2) induced pathogenic models where disease is generated in the adult through selective immunization; and (3) toxin-mediated (non-immune-mediated) demyelination models where toxins such as curpizone, lysolethicin, and ethidium bromide induce focal demyelination.

Mouse models with perturbed growth factor pathways combined with targeted ablation of receptors have provided detailed insight into the molecular regulation of induction, proliferation and differentiation of OPCs (Rowitch, 2004). For example, such studies revealed the pivotal role for PDGF-AA in OPC development (Richardson et al., 1988), and the influence of fibroblast growth factor (FGF) on multiple stages of OPC development dependent on selected receptor usage (Bansal et al., 1996; Fortin et al., 2005). Manipulation of specific transcription factors such as OLIG1, OLIG2, NKX2-2, SOX10 and myelin gene regulatory factor (MRF) (Bansal et al., 1996; Emery et al., 2009; Li and Richardson, 2008; Rowitch, 2004) has provided fundamental insights into cell specification and maturation. In general, these models are less useful in defining the underlying pathological mechanisms in MS, although in some instances a role for selected transcription factors in recovery from demyelinating insults has been described (Arnett et al., 2004).

Overall, the genetic approach has been less helpful in defining the detailed mechanisms of myelin sheath formation in the CNS. Studies of naturally occurring mutants of major myelin proteins have identified crucial components of CNS myelin such as proteolipid protein (PLP) and myelin basic protein (MBP) (Lunn et al., 1995), whereas targeted knockout of enzymes responsible for myelin lipid synthesis (Coetzee et al., 1996) has revealed a role for lipids in the stabilization of newly formed myelin (Kassmann and Nave, 2008). One area in which myelin mutants have proven useful is as hosts to assess the myelination capacities of isolated populations of neural cells. Transplantation of distinct populations of rodent and human cells into nonmyelinating hosts reveals the capacity of those cells to migrate long distances and generate substantial amounts of myelin – in some cases sufficient to reverse the host phenotype (Ben-Hur and Goldman, 2008; Low et al., 2009; Windrem et al., 2008).

The development of mouse models of the immunological aspects of MS has proven extremely valuable in identifying antigenic and cellular components important in demyelination (Furlan et al., 2009; Mix et al., 2008). The disease that develops in these models depends on the specific antigenic stimulation and the genetic background of the host animal (Baxter, 2007; Krishnamoorthy and Wekerle, 2009). For example, immunization of SJL mice with a peptide of PLP results in a relapsing-remitting disease course that closely mimics the early stages of MS. By contrast, immunization of C57/BL6 mice with a peptide of myelin oligodendrocyte glycoprotein (MOG) results in a chronic disease that closely mimics the later stages of MS. The reproducible disease that occurs in these models, along with genetic knockout/knock-in technology, has illuminated major pathways of immune-mediated demyelination, including the requirement for specific subsets of T cells.

MS involves concurrent demyelination and remyelination. This is recapitulated in immune-mediated models, which necessarily demonstrate a complex phenotype (Fig. 2). To separate the distinct mechanisms involved in demyelination versus remyelination, several chemical demyelination models have been developed. The mouse strains C57/BL6 or SJL/SVJ exhibit localized demyelination when fed a chronic low dose of cuprizone, and removal of the toxin induces remyelination (Kipp et al., 2009; Lindner et al., 2009; Skripuletz et al., 2010). Alternatively, local injection of either lysolethicin (Wang et al., 2009) or ethidium bromide (Blakemore and Franklin, 2008; Talbott et al., 2007) results in focal demyelination followed by remyelination. Several complementary, but more limited, models of myelination and demyelination have recently emerged that allow for easier identification of potential molecular therapeutic targets compared with mouse models that experience changes throughout the immune and nervous systems.

The primary challenge for in vitro models is to retain sufficient relevance to the in vivo situation so that any insights are generally applicable. Myelinating slice cultures of rodent CNS have proven useful for removing the influences of the systemic immune system on myelination (Birgbauer et al., 2004; Mi et al., 2009; Notterpek et al., 1993). Unfortunately, these models are primarily derived from developing tissue and their ability to effectively mimic in vivo demyelination/remyelination is unclear (Birgbauer et al., 2004). In addition, slice preparations retain extensive cellular diversity, which complicates cellular and molecular interpretations.

In addition to slice cultures, oligodendrocyte-neuron co-cultures can be used to study the interactions between neurons and the cells that form myelin. Examples include the following co-cultures: dorsal root ganglion (DRG) neurons and Schwann cells (Paivalainen et al., 2008); DRG neurons and oligodendrocytes (Wang et al., 2007); CNS neurons and oligodendrocytes (Fex Svenningsen et al., 2003); and oligodendrocytes and spinal cord explants (Chen et al., 2010; Thomson et al., 2006). An ideal complement to the mouse models and rodent co-cultures may be the use of human cells or tissue in in vitro assays to identify potential therapeutics that might directly enhance myelin repair in MS patients. Viable human tissue for in vitro studies is limited, but human cell lines and cells derived from MS patients should become more available through new induced pluripotent stem (iPS) cell technologies (for a review, see Lako et al., 2010). Detailed myelination cell culture protocols are established for murine and rat cells (Chen et al., 2010; Emery et al., 2009; Watkins et al., 2008) and these will probably translate to human cells.

The notion that the adult vertebrate CNS is a relatively static structure with little plasticity is now essentially obsolete. New studies indicate the adult’s capacity for extensive synaptic plasticity throughout the CNS (Holtmaat and Svoboda, 2009; McCoy et al., 2009). Furthermore, it is clear that the genesis of distinct populations of neurons is ongoing in the adult (Aguirre et al., 2004; Rivers et al., 2008). Even in the absence of disease or neural damage, new myelinating oligodendrocytes are generated throughout life (Rivers et al., 2008), and this capacity is enhanced following a range of pathologies. This newly recognized potential for adult CNS repair makes it an opportune time to revise the basic concepts guiding MS research.

Rather than considering the disease to be a reflection of unique disease-initiating events such as viral infection (Kriesel and Sibley, 2005; Salvetti et al., 2009) or antigen mimicry (Blewett, 2010), it might be better to consider the disease in the setting of an imbalance between normal degradation and repair (Fig. 3). In this model the ‘normal’ CNS undergoes slow turnover, cell replacement and myelin repair in response to insults, which never manifest as functional deficits. Suppression of this repair process by additional insults, or by overwhelming natural repair as a result of pathogenic pressure, results in functional deficits and clinical manifestation of disease. If correct, this model expands the repertoire of potential therapeutic targets and suggests that combinatorial therapies that neutralize pathogenic pressure and enhance myelin repair will provide more effective long-term benefit. This endeavor requires the generation of new models of MS that allow the onset of demyelination and remyelination to be controlled and their extent accurately assessed.

What new mutant models are needed? There are two areas where development of new mouse models would be most valuable. First, models in which demyelination can be selectively induced in the adult through targeted death of oligodendrocytes would allow for a dissection of the neural responses accompanying demyelination and remyelination. One such genetic model has been developed, using Cre-loxP technology to induce the expression of diphtheria toxin A; this results in the selective ablation of specific cell types, such as oligodendrocytes (Brockschnieder et al., 2006). Unfortunately, this strategy is limited by the availability of specific Cre lines of mice and by the fact that inducible Cre lines are often capable of inducing only low levels of recombination, resulting in the incomplete expression of diphtheria toxin. Thus, some cells targeted for destruction by diphtheria toxin survive and complicate analysis. Second, the development of reporter animals in which adult myelin deposition can be accurately assessed would help define the rate-limiting steps in remyelination and allow the identification of novel therapeutic targets.

We thank Anne DeChant for help with manuscript preparation. This work was supported by grants from the NIH (NS030800) to R.H.M., and the Myelin Repair Foundation. Deposited in PMC for release after 12 months.