Introduction

Following the publication of a number of studies testing adult stem cell therapy in patients with cardiac disease, the past year or so has seen the beginning of a comparable therapeutic effort in neurological disease: small numbers of patients with stroke have received autologous bone marrow cells [6], as have comparable numbers of individuals with amyotrophic lateral sclerosis [82]. At least one (unpublished) study of patients with multiple sclerosis has been completed, wherein the effects of intracerebral implants of autologous Schwann cells and fibroblasts were studied, apparently with no demonstrable effect (http://www.myelin.org/12082003.htm).

What is the future of cell therapy in multiple sclerosis? If there be one, does it lie with Schwann cells, with oligodendrocyte progenitors, olfactory glia, or with stem cells? Is the time ripe to try further clinical studies? If not, what are the remaining hurdles? Here we propose that remyelination treatments by cell-based therapy represent an approachable challenge offering a realistic prospect of successful implementation for the current generation of patients with multiple sclerosis.

Multiple sclerosis is particularly suitable for cell therapy

A primary demyelinating disease—with substantial secondary axon loss

A combination of factors suggests that multiple sclerosis provides an attractive and tempting testing ground for neurological cell therapy.

The first is its nature: primarily a demyelinating disease. But axon loss of course represents the principal pathophysiological cause of disability in chronic progressive disease: does this undermine the rationale of a cell-based therapy?

Rather the opposite—for two reasons. First, none of the recent experimental, imaging or neuropathological studies re-confirming the importance of axon damage have challenged the concept that disease processes in MS are primarily directed against oligodendrocytes and/or myelin, and that axons are relatively spared until late disease [12, 115]. Therefore, in the main, axon pathways remain intact. Cell therapies therefore aim ‘only’ to reinvest axons with myelin, rather than addressing the almost overwhelming challenge presented by most other neurological diseases: that of re-establishing connectivity in highly complex but fragmented axonal circuitry.

Secondly, the mechanism of axon loss is important to consider. The course of secondary progression—and by implication, of axon loss—is significantly neither related to early inflammatory disease activity [10, 31, 64] nor impeded by even the most profound immune suppressant treatments. These and other observations have fuelled the hypothesis that progressive axonal damage is (at least in part) a consequence of persistent myelin loss [12, 112, 128]. Pathological studies have indicated that chronic axon loss does not correlate with either inflammatory cell infiltrate, tumour necrosis factor expression, nitric oxide expression, or demyelinating activity, but does correlate with the overall extent of established myelin loss [10, 64]. It is seen in lesions which are demyelinated but which exhibit sparse or no inflammation, but is rare in remyelinated lesions [64]. Demyelination-induced axon loss might occur by several possible mechanisms: directly, through the loss of oligodendrocyte-derived trophic support [47, 85, 128], or sustained demyelination-induced conduction block and electrical silence [78], or indirectly through increased vulnerability of the exposed axon to injurious agents [99]. A further important driver for early cell therapy thus emerges: the restoration of a normal oligodendroglial environment in order to sustain (previously demyelinated) axons.

Supplementing spontaneous myelin repair

The second positive feature of MS, in terms of developing cell therapies, is found in the clear evidence of spontaneous if partial myelin repair in multiple sclerosis [71, 79, 96, 101]. The nature of relentlessly progressive disability in many patients implies bears witness to the insufficiency of this process, but supplementing spontaneous remyelination must appear a less fanciful proposition than imposing repair in a fundamentally non-reparative environment.

Therapeutic remyelination works experimentally

The third reason to be cheerful is the now large body of experimental evidence proving that cell therapy can achieve successful remyelination. Schwann cells, oligodendrocyte lineage cells or cell lines, olfactory ensheathing cells (OEC), and rodent embryonic, adult neural and adult bone marrow stem cells all have been shown successfully to remyelinate the rodent CNS [2, 8, 9, 18, 46, 48, 50, 54, 93, 105, 109, 120, 134], and variously demonstrated to restore normal conduction [52, 122] and/or function [56].

What problems remain?

Several! Perhaps the most pressing are readily summarised as “where?”, “when?”, “with what?” and “then what?”

Where?

Multiple inoculations of cells into widely distributed lesions in the brain and spinal cord of patients with MS is neither attractive nor realistic. Two possibilities are here relevant.

First, many plaques are clinically silent. A disproportionately large degree of disability frequently emanates from a few critical lesions in eloquent areas. Thus implantation into a very small number of carefully selected lesions—for example, the optic nerves, the spinal cord, or the superior cerebellar peduncle—could yield a commensurately disproportionate therapeutic dividend [30].

Alternatively, some experimental evidence using adult bone marrow- and brain-derived stem cells suggests that the tropism of certain reparative cells for diseased tissue can be exploited, and disseminated (or perhaps also diffuse) disease may be addressed by intravenous delivery of cells [2, 93].

When?

On the basis of “first, do no harm”, late MS seems safest: progressive disability is established, with sadly negligible hope of spontaneous recovery, and the possibility of doing damage or compromising spontaneous repair is remote.

However, arguments can be offered in favour of the early intervention—although at this stage, little disability is present and therefore there is much to lose, and the natural history is such that some patients will never develop significant disability. In addition, implanting cells into early lesions exposes remyelinating glia, and their new myelin, to ongoing inflammatory activity.

But spontaneous remyelination appears to occur maximally in acute inflammatory lesions [95, 101], suggesting (paradoxically) that these offer an optimal reparative environment. Some evidence suggests that anti-inflammatory drugs [116], or the suppression of inflammation in general [33], may impair myelin regeneration.

The clinical impact and irreversibility of accumulating axon loss in secondary progressive disease [11, 12] provide a more potent reason for earlier remyelinating intervention. Quite apart from the futility of attempting to remyelinate absent axons, it has become clear that changes in the cell surface expression of various molecules (e.g. PSA-NCAM) in chronically demyelinated axons actively inhibit myelination [23]. Accumulated myelin debris also inhibits myelination [65], and chronic astrocytosis offers a profound inhibitory effect on the migration of remyelinating glia [41].

With what?

A number of cell types may be considered. (see Fig. 1)

Fig. 1
figure 1

cell-based therapies for myelin repair in MS: potential sources

Full size image

Cells of the oligodendrocyte lineage

Oligodendrocytes are the most obvious candidates: they are the cells lost, and their normal function is to myelinate the CNS. Immature oligodendrocytes and oligodendrocyte precursors are found in fresh lesions [22, 80, 97, 100, 114, 131] and are generally considered responsible for the great majority of spontaneous remyelination [21, 40, 113, 130].

Despite their motility in vitro, oligodendrocyte progenitors however show poor survival and migration when implanted into normal white matter, although they are able to populate and remyelinate when injected into, or very close to, lesioned tissue [44]. A further difficulty is that investigations of human CNS glia have consistently demonstrated significant biological differences from rodent cells, so that data concerning rodent OPCs cannot be directly extrapolated to human glia. Thus, notwithstanding the success of rodent glial progenitors in experimental myelin repair, initial studies of adult human oligodendrocyte progenitors suggested a very limited capacity for remyelination (in the irradiated rodent spinal cord) [119]. However, oligodendrocyte lineage cells can successfully be derived from ES-cells, and can repair myelin, with demonstrable functional recovery [59, 89]—but human embryonic stem cells as a source of therapeutic material bring a number of problems (see below). Improved methods for isolating adult human oligodendrocyte progenitors (by genetically labelling) have also emerged [106]. Interestingly, comparative studies suggest that adult human CNS-derived oligodendrocyte progenitors have a significantly greater remyelinating capacity than their foetal counterparts [129].

Schwann cells

Schwann cells make a small but significant contribution to endogenous myelin repair in multiple sclerosis, perhaps particularly in the spinal cord [55, 79, 90]. Experimental methods for preparing, from adult peripheral nerve biopsies, cultures of Schwann cells, and for purifying and expanding the cells in vitro to generate large populations of Schwann cells have been established [88, 107]. When so purified, human Schwann cells successfully lay down new myelin in the mouse [75] and the rat spinal cord [17, 62]; autologous, expanded Schwann cells successfully repair relatively large areas of demyelination in the macaque demyelinated spinal cord [5].

Autologous Schwann cell harvesting from peripheral nerve biopsy, expansion in vitro, then transplantation into patients offers the considerable attractions of relative ease of availability, and the avoidance of rejection. Furthermore, Schwann cells (and their myelin) should be resistant to continuing MS-related immunological attack. Firm evidence is required however, that expanded human Schwann cells do not form tumours in vivo, a hazard described when rodent Schwann cells immortalised by growth factor expansion were transplanted [70]; unpurified preparations of human peripheral nerve cells result in substantial fibroblast overgrowth with axon destruction [17]. The apparent inhibitory effect of astrocytes on Schwann cell mediated CNS remyelination [45, 49, 133] represents another potential problem—though genetic modification of Schwann cells to show increased and sustained surface expression of PSA on NCAM significantly improves migration [72].

Olfactory glia

Olfactory ensheathing cells, found in the olfactory bulb, nerves and epithelium, ensheath axons emanating from olfactory epithelial neurons that penetrate the olfactory bulb of the CNS. Rodent OEC’s assume a myelinating phenotype closely resembling that of Schwann cells when transplanted into lesions containing demyelinated axons [46, 54]. Human OEC’s, like rodent OEC’s, are also capable of remyelination following transplantation into the demyelinated rodent spinal cord [7, 58]. This, and the OEC’s ability to migrate in an astrocytic environment [46, 69], in stark contrast to Schwann cells, has helped generate much interest in olfactory glia in the field of CNS repair [42].

Stem cells

Stem cells have enormous therapeutic potential, not least for treating neurodegenerative disease [13, 20, 91, 127]: a consequence of their proliferative and multipotent capacities. Most studies have concentrated on using embryonic tissue as a source of stem cells [18], but to develop therapies would obviously require the use of human embryos as the stem cell source, and this raises significant practical, immunological, and insurmountable ethical concerns [15, 111]. One serious risk is that of teratoma formation (Bjorklund et al. 2002): removing this capacity from embryonic stem cells with absolute success in order to develop safe cell therapies may pose considerable problems. In addition, the emergence of significant chromosomal abnormalities in cultured human embryonic stem cells raises further concerns about their safe use [39]. The problem of rejection would also have to be circumvented. Early, optimistic suggestions were that using stem cells from embryos cloned (by cell nuclear transfer) from individual putative recipients (recently legalised uniquely in the UK) would solve this, the implication that every patient requiring a transplant would first have to be cloned seems quite unrealistic, would not bypass the major ethical difficulties associated with the use of human embryonic material—and may in any case not prove possible, as recent events in Korea have shown.

These problems have helped stimulate the largely successful search for alternative sources of stem cells [103, 104]. There is increasing evidence that adult stem cells have a greater capacity to differentiate into a wider range of cell types than previously anticipated, and the use of adult stem cells—particularly autologous in origin—would avoid many of the difficulties associated with embryonic stem cells [27, 94, 98, 110].

Neural stem cells are present in the adult rodent brain [126]; large numbers of oligodendrocyte lineage cells can be generated using neurosphere/oligosphere techniques, which, upon transplantation, successfully remyelinate axons. Neural stem cells are also present in the adult human brain [68]. Adult rodent CNS-derived stem cells repair multifocal demyelinating lesions (in EAE-affected rodents) even after intravenous delivery [39, 93].

It is now beyond doubt that adult bone marrow does indeed harbour a sub-population of potentially proliferative stem cells [25, 29, 57, 61, 66, 67, 86, 92, 94, 102], whose differentiation capacity includes glial cells and neurones [37, 60, 108, 132]. Cell fusion of bone marrow-derived cells can provide an alternative explanation for apparent trans-differentiation, and this is particularly important in some organs—the liver, for example [123, 124]. However, cell fusion cannot explain the extensive in vitro data indicating multipotent differentiation, and in vivo studies confirm transdifferentiation without fusion in a variety of tissues [53, 81, 121]. Furthermore, from a pragmatic perspective, fusion may simply be part of the means by which bone marrow-derived stem cells stimulate successful regeneration [83, 104]. Recent studies indicate that polyploidy is in fact far commoner phenomenon that previously realised; the possible occurrence of fusion does not necessarily imply diminished regenerative capacity in a putative reparative cell [14].

Bone marrow derived stem cells migrate into (presumably) normal adult human brain and transdifferentiate into highly complex, apparently functionally integrate neural cell types [28, 32, 125]. Experimentally, they show an ability to home from the circulation to various damaged tissue(s) [38, 63], including tropism in regard to areas of CNS injury [1, 25]. Directly or peripherally injected bone marrow-derived cells will repair damage, often with demonstrable functional as well as anatomical recovery, in rodent models of traumatic, degenerative and ischaemic CNS damage [4, 16, 24–26, 74, 77, 87]. Remyelination is reported not only after intra-lesional injection [3, 109], but also following peripheral injection [2].

These properties, their easy accessibility, and not least the absence lack of ethical problems, and also their significant track history in the treatment of haematological disease makes bone marrow cells strong candidates for use in cellular therapies for CNS disease. Further elucidation of the mechanisms involved may allow for mobilisation of endogenous cells, perhaps even obviating the need for transplantation.

The advantages and disadvantages of the different sources of stem cells and progenitor cells to achieve demyelination in MS are summarised in Table 1.

Table 1 Some relative advantages and disadvantages of sources of stem cells and progenitors for remyelination therapy
Full size table

Then what?

How can the effects of a trial cell therapy in MS be assessed? At present, MRI detection of new myelin is not reliably feasible, but new techniques continue to emerge, of which magnetisation transfer contrast is one promising candidate [36], 3-dimensional MRI using multiple contrasts [84], and radial diffusivity [117, 118] likewise. Magnetic resonance spectroscopy measurement of N-acetyl aspartate levels might offer means of assessing any impact on local neuron/axon survival [34, 35]. Cells can be labelled to render them MRI-visible [19, 43, 76], but from a safety perspective, even trivial manipulation of cells prior to implantation is best avoided. Furthermore, graft survival cannot be inferred from migration, since dead cells remain visible [19], and this method not only fails to show new myelin formation but may also impair the ability of other MR modalities to do so.

Serial neurophysiology may prove valuable, and monitoring conduction times may provide evidence of returning saltatory conduction in the targetted pathway(s). The optic nerve has particular advantages in this respect, but various approaches to more generalised neurophysiological assessment have been described and may prove useful for any intervention aimed at multifocal or more diffuse myelin repair [73].

Finally, robust and reproducible methods of clinical assessment need to be applied. Specific clinical outcomes measures of function, disability, and handicap must be adopted; considerable advances in clinical scale design have improved physical and functional measurement in multiple sclerosis [51], so that the tools for assessing clinical outcome, on which remyelination therapies must stand or fall, are becoming available.