Olfactory ensheathing cells (OECs) and the treatment of CNS injury: advantages and possible caveats

Abstract

One of the main research strategies to improve treatment for spinal cord injury involves the use of cell transplantation. This review looks at the advantages and possible caveats of using glial cells from the olfactory system in transplant-mediated repair. These glial cells, termed olfactory ensheathing cells (OECs), ensheath the axons of the olfactory receptor neurons. The primary olfactory system is an unusual tissue in that it can support neurogenesis throughout life. In addition, newly generated olfactory receptor neurons are able to grow into the CNS environment of the olfactory bulb tissue and reform synapses. It is thought that this unique regenerative property depends in part on the presence of OECs. OECs share some of the properties of both astrocytes and Schwann cells but appear to have advantages over these and other glial cells for CNS repair. In particular, OECs are less likely to induce hypertrophy of CNS astrocytes. As well as remyelinating demyelinated axons, OEC grafts appear to promote the restoration of functions lost following a spinal cord lesion. However, much of the evidence for this is based on behavioural tests, and the mechanisms that underlie their potential benefits in transplant-mediated repair remain to be clarified.

Introduction

Spinal cord injuries affect over 700 people per annum in the United Kingdom (data from the International Campaign for Cures of Spinal Cord Paralysis). The resulting functional deficits are often severe, largely permanent, and there is currently no effective treatment. However, at the basic research level, several strategies aimed at providing future treatments are under investigation. These include (1) encouraging the survival and growth of damaged axons using neurotrophins (Blesch et al. 1999; Bradbury et al. 2000; Ramer et al. 2000; Bamber et al. 2001), (2) targeting the downstream signalling molecules which promote axonal outgrowth, for example, cAMP (Stichel & Muller, 1998; Qiu et al. 2002), (3) neutralizing inhibitory molecules associated with the failure of axonal regeneration, e.g. enzymatic treatments for chondroitin sulphate proteoglycans (Bradbury et al. 2002), antibodies to Nogo receptors (Brittis & Flanagan, 2001; Huber & Schwab, 2000; McKerracher & Winton, 2002), and lastly (4) providing a permissive growth environment by transplanting neural cells. A range of cells have been investigated for use in transplantation. These include neural stem cells, which also comprise stem-cell-like radial glial cells and subventricular zone astrocytes (Doetsch et al. 1999; for reviews see Horner & Gage, 2000; Campbell & Gotz, 2002), and glial cells such as olfactory ensheathing cells (see reviews of Franklin & Barnett, 2000; Raisman, 2001), Schwann cells (Duncan et al. 1981; Tuszynski et al. 1998) and oligodendrocyte precursor cells (Barnett et al. 1993a).

One of the most promising candidates for cellular transplant-mediated repair of CNS lesions is the olfactory ensheathing cell (OEC). OECs reside in the olfactory system, which supports neurogenesis throughout life (Graziadei, 1973; Graziadei et al. 1979; Farbman, 1990). The olfactory system originates from the olfactory placode and is made up of the olfactory epithelium (PNS tissue) and the olfactory bulb (CNS tissue) (Fig. 1). It is still not clear whether the outer layer of the bulb of the adult rat, the olfactory nerve layer, is entirely peripheral in origin or a mixture of both peripheral and central cells (Marin-Padilla & Amieva, 1989; Valverde et al. 1992). After injury or during normal cell turnover, new olfactory receptor neurons (ORNs) are generated from basal stem cells in the olfactory epithelium, which extend axons through the cribriform plate and re-enter the olfactory bulb re-synapsing with second-order neurons in the glomerular layer (see review of Schwob, 2002). This is one of the rare situations in which peripheral axons are able to enter the adult CNS environment and form synapses, and it is thought that this unusual ability may be due in part to the specialized properties of OECs (Doucette, 1990; Farbman, 1990).

Fig. 1.

Longitudinal section through the rat head. The olfactory system is a mixture of PNS and CNS tissue. After injury or during normal cell turnover, new olfactory receptor nerves (ORNs) are generated from basal stem cells in the olfactory epithelium located in the PNS. Throughout life these neurons are able to extend axons from the PNS, enter the CNS environment of the olfactory bulb and reform synapses with the mitral cells. (Modified from H. G. Liebich.)

Glial cells as candidates for promoting CNS repair

Although several candidate cell types have been investigated for use in transplant-mediated CNS repair, OECs currently appear to be one of the most promising. In addition to their repair potential, there are other important criteria that a candidate cell needs to fulfil in order to be suitable for potential translation to the clinic; it should be feasible to obtain the relevant tissue from the patient in order to enable autologous transplants to be performed (to avoid graft rejection) and it must be possible to generate the appropriate cells in culture in sufficient numbers.

Consideration of these criteria has excluded the oligodendrocyte precursor cell (OPC). Although several rat studies have shown that rat OPCs can remyelinate experimentally created demyelinated axons (Barnett et al. 1993a; Groves et al. 1993) it has not been possible to maintain and passage cultures of human oligodendrocytes or their precursors (Scolding, 1998). To overcome this problem, a mixed-cell suspension of adult human white matter was transplanted into experimentally created demyelinated CNS lesions but these cells had a limited ability to remyelinate (Targett et al. 1996). Therefore, the use of human OPCs in transplant-mediated CNS repair seems improbable.

Unlike oligodendrocytes and their precursors, Schwann cells have many of the properties desirable for use in transplant-mediated repair. They can be readily harvested from the peripheal nerve and easily purified and grown in culture in large numbers from both rat and human tissue. It has been shown that after transplantation Schwann cells can remyelinate demyelinated axons (Li & Raisman, 1994; Kohama et al. 2001) and promote regeneration (Duncan et al. 1981; Tuszynski et al. 1998). However, the interaction of Schwann cells with astrocytes limits their usefulness in CNS repair (see review of Franklin & Barnett, 1997). This view is supported by the finding that Schwann cells invade CNS territories during development only when astrocytes are absent and that this invasion stops when an astrocyte boundary is reached (Gilmore & Duncan, 1968; Baron Van Evercooren et al. 1992). This can also be observed using in vivo models of demyelination where Schwann cells do not appear to cohabit with astrocytes after transplantation into a demyelinated white matter lesion (Shields et al. 2000) or when placed in contact with astrocytes in culture (Ghirnikar & Eng, 1994; Lakatos et al. 2000). Lastly, although it has also been shown that grafts of Schwann cells can support the in-growth of damaged axons, there is little evidence to suggest that these axons can then exit the graft and re-enter the CNS environment (Li & Raisman, 1994; Xu et al. 1995).

Although obtaining OECs from the olfactory bulb is more complicated than harvesting Schwann cells from the sciatic nerve, there is evidence that human OECs can be directly isolated from biopsies of the olfactory mucosa with no adverse effects (Feron et al. 1998). In addition, rat and human OECs are relatively easy to purify and can be grown in large numbers. OECs do not seem to exhibit the same undesirable behaviour as Schwann cells when in contact with astrocytes (Lakatos et al. 2000), they can remyelinate experimentally created demyelinated axons (Franklin et al. 1996; Imaizumi et al. 1998; Barnett et al. 2000; Kato et al. 2000) and can promote regeneration (Li et al. 1998; Ramon-Cueto et al. 1998, 2000; Nash et al. 2002). A particular advantage claimed for the OEC over the Schwann cell is that regenerating fibres are able to grow beyond the OEC graft and into the host spinal cord on the opposite side of the lesion (Li et al. 1998; Ramon-Cueto et al. 1998, 2000; Nash et al. 2002).

OECs as candidates for the promotion of CNS repair

In vitro properties

In our laboratory OECs are purified from the olfactory bulbs of 7-day-old Sprague–Dawley or Fischer rat pups using fluorescence-activated cell sorting (FACS) and marker antibodies (Barnett et al. 1993b). The growth and maintenance of large numbers of purified OECs can be obtained using a mitogen mixture of FGF2, forskolin, heregulin β1, serum-free medium conditioned by confluent cultures of cortical astrocytes (astrocyte-conditioned medium, ACM; Noble & Murray, 1984; Alexander et al. 2002). If FACS-purified OECs are cultured in ACM immediately, apoptosis is prevented and they proliferate for up to 12 days, after which they appear to differentiate into at least two subtypes (Franceschini & Barnett, 1996; Pollock et al. 1999). These subtypes exhibit different antigenic and morphological characteristics, with one resembling a Schwann cell, and the other more akin to an astrocyte (Barber & Lindsey, 1982; Pixley, 1992). Astrocyte-like OECs have a flat morphology, express intense glial fibrillary acidic protein (GFAP, an intermediate filament associated with glial cells) immunoreactivity (IR), the embryonic form of the neural cell adhesion molecule (E-NCAM) but little or no expression of p75NTR (a marker for non-myelin, forming Schwann cells). E-NCAM is the polysialylated isoform of NCAM and has been identified on cells that in adult life retain embryonic-like features, e.g. astrocytes in the hypothalamo-neurohypophysial systems (Bonfanti et al. 1992). Polysialyic acid expression is associated with a decrease in the adhesive properties of NCAM, and this may permit the cellular rearrangements that occur during neurohistogenesis and plasticity (Seki & Rutishauser, 1998; Theodosis et al. 1999). The Schwann cell-like OEC has a spindle-shaped morphology, and expresses diffuse GFAP-IR, little or no E-NCAM but high levels of p75NTR (Franceschini & Barnett, 1996). It is possible that there are other OEC subtypes and, in culture at least, the OEC has been reported to exhibit considerable cellular plasticity with a dramatic ability to change shape and size (Van Den Pol & Santarelli, 2003; Vincent et al. 2003).

Similarities and differences in properties between OECs and Schwann cells

OECs possess several characteristics that are very similar to those seen for Schwann cells (see legend to Fig. 2). However, to date there appear to be only two properties that may be used to distinguish the astrocyte and Schwann cell-like cell types. The first is the expression of erbB receptors. These receptors bind the neuregulin family of proteins, which are potent mitogens for both Schwann cells and OECs (Dong et al. 1995; Pollock et al. 1999). Using RT-PCR and immunocytochemistry rat OECs have been shown to express protein and mRNA mainly for erbB2 and erbB4, whereas Schwann cells express protein and mRNA mainly for erbB2 and erbB3 (Fig. 3; Dong et al. 1995; Pollock et al. 1999; Thompson et al. 2000), although this difference in receptor expression has not been found in the mouse (Meyer & Birchmeier, 1994). However, more convincing data to illustrate the distinct differences between OECs and Schwann cells are seen in the way they interact with astrocytes in culture. Using protocols originally devised to show the inhibitory interactions between Schwann cells and astrocytes (Ghirnikar & Eng, 1994; Wilby et al. 1999) it was confirmed that Schwann cells and astrocytes occupied distinct and non-overlapping areas in co-culture whereas OECs and astrocytes frequently occupied the same area (Fig. 4). Furthermore, astrocytes in contact with Schwann cells exhibited characteristics of hypertrophy. These include an increase in cytoplasmic area and an up-regulation of chrondroitin sulphate proteoglycans (Lakatos et al. 2000). The mechanism for these differences is not known. It has been suggested that the expression of N-cadherin on Schwann cells may mediate the adhesion of Schwann cells to astrocytes (Wilbey et al. 1999). However, OECs also express N-cadherin, which does not appear to play a role in their interaction with astrocytes (Lakatos et al. 2000).

Fig. 2.

Comparison of the properties of OECs and Schwann cells. Rat OECs and Schwann cells share: (1) antigenic and morphological characteristics – namely a spindle shape and express p75NTR, GFAP and O4 immunoreactivity (Barnett et al. 1993b; Mirsky & Jessen, 1999); (2) growth factor response to GGF, forskolin and FGF2 (Yan et al. 2001; Alexander et al. 2002); (3) connexin expression and gap junctional communication (Barnett et al. 2001); (4) transcriptional regulation during myelination in vivo (Smith et al. 2001). Rat OECs and Schwann cells may have differences in: (1) erbB receptor expression – Schwann cells express erbB2 and erbB3, and OECs express erbB2 and erbB4 (Thompson et al. 2000); (2) interaction with astrocytes – Schwann cells but not OECs induce hypertrophy in astrocytes as assessed by an increase in astrocyte cytoplasmic area and expression of chondroitin sulphate proteoglycans at the graft site (Lakatos et al. 2000; Plant et al. 2001).

Fig. 3.

Differences in erbB receptor profiles in OECs and Schwann cells. (A) RT-PCR illustrating olfactory bulb tissue (OB) and primary cultures of Schwann cells express the mRNA for erbB3. Conversely, OB and primary cultures of OECs but not Schwann cells express the mRNA for erbB4. M, molecular weight markers. (B) Immunofluorescence of OECs labelled with antibodies to erbB2, erbB3 and erbB4. These data show OECs express erbB2 and erbB4. Magnification ×200.

Fig. 4.

Schematic representation of the interaction of OECs (pink) and Schwann cells (red) with astrocytes in co-culture. The main difference in characteristics between OECs and Schwann cells is seen in their interactions with astrocytes (green). (A) Schwann cells do not mingle well and astrocytes express characteristics of hypertrophy. (B) Conversely, OECs mingle well among astrocytes and do not appear to evoke any changes in astrocytes.

Re-evaluation of the effects of OEC grafts on regeneration at the dorsal root entry zone

In addition to investigations of their effects at the sites of lesions within the spinal cord, OECs have also been tested as a means of promoting repair following dorsal root lesions, an injury that occurs clinically as dorsal root avulsion (Mackinnon & Dellon, 1988). Primary afferent fibres entering through the dorsal roots branch within the spinal cord and terminate both at their level of entry and within more rostral and caudal spinal cord segments (Brown, 1981; Fig. 5A). Following lesions of the dorsal roots, afferent fibres within the root regenerate and are able to grow back to their normal region of entry into the spinal cord. However, at this point the regenerating fibres stop as they are unable to cross from the PNS tissue of the root into the CNS environment of the cord (Ramon y Cajal, 1928; Fig. 5B). The use of OECs as an approach to solving this problem has an attractive rationale in that enabling axons from newly formed olfactory neurons to grow into the CNS environment of the olfactory bulb is thought to be a normal function of the OEC. Ramon-Cueto & Nieto Sampedro (1994) first investigated the effects of grafts of OECs at the dorsal root entry zone using Dil crystals placed on the cut dorsal root to label dorsal root fibres. They interpreted the resulting tracer distribution as evidence that regenerating fibres had entered the spinal cord (Fig. 5C). This work was extended by Nieto-Sampedro and colleagues (Navarro et al. 1999; Taylor et al. 2001; Pascual et al. 2002), who reported that the regenerating fibres were also able to re-establish functional connections with spinal cord neurons (Fig. 5D). Remarkably, functional reconnection appeared to extend even to the very specific connection between Ia muscle afferent fibres and motoneurons, as H-reflexes were reported to be restored (Navarro et al. 1999).

Fig. 5.

Schematic diagrams illustrating the normal dorsal root entry zone and different regeneration scenarios. (A) Sensory axons enter the spinal cord at the dorsal root entry zone. In the spinal cord, afferent fibres give rise to branches that travel in the dorsal columns and axon collaterals that synapse with spinal cord neurons in the grey matter. (B) When dorsal roots are lesioned, the central axons of afferent fibres regenerate along the dorsal root until they reach the dorsal root entry zone where regeneration is abortive. Grafts of OECs are reported to allow regenerating dorsal root fibres to enter the spinal cord (C), and re-establish functional connections with spinal cord neurons (D).

We have recently attempted to follow up this evidence for functional reconnection across the dorsal root entry zone, with the aim of investigating further the extent to which new connections formed by regenerating fibres are of a functionally appropriate pattern (Riddell et al. 2002). The experimental design involved two main stages (see Fig. 6). At a sterile operation, a lumbar dorsal root was sectioned and then re-anastomosed. A loop of suture thread was placed around the lesioned root so that it could be identified at the subsequent electrophysiological experiment. At the same operation a suspension of cells was injected into the dorsal root entry zone/dorsal horn, into the dorsal root, or at both of these sites (Fig. 6A). The cell suspension was prepared using FACS-purified neonatal OECs grown in culture (Barnett et al. 1993b; see above). The proportion of p75NTR-positive cells in the suspensions was generally greater than 75%. The electrophysiological investigation was carried out between 1 and 4 months later. Electrical stimulation was used to activate fibres in the lesioned root and an adjacent intact root and recordings were made from the surface of the spinal cord and within the grey matter in order to detect synaptic actions evoked by the stimulation (Fig. 6B). Although stimulation of the intact root evoked normal cord dorsum potentials and field potentials (a measure of post-synaptic activity), stimulation of the previously lesioned roots in the same animals failed to evoke any detectable post-synaptic activity. This was despite evidence that afferent fibres had successfully grown past the lesion site and reached the dorsal root entry zone. These observations were made on 11 animals, all of which received injections made into the dorsal horn/dorsal root entry zone and about two-thirds of which also received injections of cells into the dorsal roots. In separate experiments, we were able to show, by injecting cells infected with a retrovirus containing the bacterial reporter gene beta galactosidase (βgal), that the cells were appropriately distributed following injection into the spinal cord. In these experiments the cells could be detected for up to 4 weeks after the injection. The number of labelled cells appeared to reduce steadily with time but this could be due largely to a loss of βgal expression over time, for example due to splicing.

Fig. 6.

Schematic diagram to illustrate plan of electrophysiological experiments to investigate the effects of OECs on regeneration at the dorsal root entry zone. (A) A lumbar dorsal root is sectioned and re-anastomosed, and in the same operation OECs are injected into the dorsal root and dorsal root entry zone. A loop of suture thread identifies the lesioned root. (B) Three to 6 months later, an acute electrophysiological experiment is performed. The lesioned root and an intact root are stimulated electrically to activate afferent fibres and recordings are made from the surface of the cord and within the grey matter to detect post-synaptic activity.

As we did not detect functional reconnection of regenerating dorsal root fibres, we wondered whether fibres were growing into the spinal cord but failing to form functional synaptic connections with spinal cord neurons. In order to investigate this possibility we carried out further investigations on 25 animals in which anatomical methods were used to determine whether regenerating fibres entered the spinal cord. As with the electrophysiological study, lumbar dorsal roots were sectioned and re-anastomosed, and at the same operation OECs were injected into the dorsal root and dorsal root entry zone (Fig. 7). Between 2 and 6 months later the animals were investigated for anatomical evidence of fibre ingrowth. Two methods were used to visualize regenerating fibres: (1) immunocytochemistry with antibodies to the axonal protein neurofilament 200 (NF200), which is found in myelinated primary afferent fibres (not illustrated), and (2) tract tracing with biotin dextran amine (BDA), which we have found to provide extensive labelling of both myelinated and unmyelinated fibres after injection into the dorsal root ganglion (DRG) (Fig. 8A). After processing, sections of spinal cord corresponding to the region where the repaired root normally entered were examined using epifluorescence and confocal microscopy. In addition to sections from animals that received grafts of OECs, we also examined sections prepared from two groups of control animals. These animals were subjected to the same dorsal root lesions but were either injected with medium or not injected at all. Figure 8 shows examples of confocal microscope images that illustrate the observations using BDA tract tracing. Sections from animals in which the dorsal roots had been lesioned, but neither cells nor medium injected, almost never showed any evidence of fibre in-growth despite the clear presence of NF200-labelled and BDA-containing fibres at the dorsal root entry zone (Fig. 8B). Sections from animals that had been injected with medium showed evidence of limited in-growth; that is, in a small number of sections from a proportion of animals, an occasional labelled fibre could be seen to have entered the spinal cord (Fig. 8C). Sections from animals that had received OEC grafts also showed evidence of fibre in-growth but again this was very limited. Overall, there was little indication that OEC-injected animals showed fibre in-growth that was any greater in frequency or extent than for medium-injected controls. Furthermore, the BDA-labelled fibres that could be followed centrally had an abnormally simple morphology. They did not enter the dorsal columns but projected ventrally towards the dorsal horn. They showed none of the branching that typifies the central projection of afferent fibres and never penetrated further than the most superficial laminae of the grey matter (Fig. 8D). These results suggest that grafts of OECs have, at best, a marginal influence on the in-growth of regenerating dorsal root fibres. Furthermore, because in-growth was rarely seen in animals that had not been subjected to either type of injection, the limited in-growth observed is most likely due to mechanical disruption of the dorsal root entry zone. The reason for these disappointing results and the discrepancy with previous reports is not immediately obvious but one possible factor could be the source of cells used (see below).

Fig. 7.

Schematic diagram to illustrate plan of anatomical experiments to investigate the effects of OECs on regeneration at the dorsal root entry zone. (A) One or more lumbar dorsal roots are sectioned and one root re-anastomosed. At the same operation OECs are injected into the dorsal root and dorsal root entry zone. A loop of suture thread identifies the lesioned and re-anastomosed root. (B) Two to 6 months later, biotin dextran amine is injected into the DRG of the lesioned root. After allowing 2 weeks for transport of the tracer root, the animal is fixed by intravascular perfusion and the lumbar spinal cord sectioned, processed and examined using epifluorescence and confocal microscopy.

Fig. 8.

Anatomical assessment of regeneration promoted by OECs after dorsal root lesions. Confocal microscope images of transverse sections from the spinal cord of animals in which BDA was injected into the DRG. (A) Normal animal; note extensive labelling of fibres in the dorsal root, dorsal columns and of axon collaterals throughout the dorsal horn. (B) Non-injected control animal; BDA-labelled afferents can be seen to have regenerated through the peripheral dorsal root environment to the dorsal root entry zone but do not pass centrally. (C) Control animal injected with medium: note the BDA-labelled afferent fibre that has entered the spinal cord (arrow). (D) Animal transplanted with OECs; note the labelled afferents that have entered the spinal cord and regenerated as far as the superficial laminae (arrows). DR, dorsal root; DC, dorsal columns; DH, dorsal horn; II, lamina II; III, lamina III. Scale bars = 100 μm.

Which types of cells are important for effective OEC grafts?

In the dorsal root study described above and also in the original work in which remyelination of experimentally denuded axons was demonstrated (Franklin et al. 1996), cell grafts were prepared from neonatal animals and selected for culture using FACS. Others investigating the dorsal root entry zone have used adult rather than neonatal donor animals and purified the cells using immunopanning rather than FACS (Ramon & Nieto-Sampedro, 1994; Navarro et al. 1999). Although it is unlikely that these differences in cell preparation technique can explain the contrasting results, the influence that the cell preparation method and resulting mixture of cells has on the effectiveness of OEC grafts is emerging as an important issue. Axonal regeneration and/or improved functional recovery after spinal cord injury have been reported using purified OEC grafts (Ramon-Cueto & Nieto-Sampedro, 1994; Ramon-Cueto et al. 1998, 2000; Navarro et al. 1999; Nash et al. 2002; Pascual et al. 2002), non-purified cell preparations obtained from the olfactory bulb (Li et al. 1997, 1998, 2003) and even nasal olfactory tissue implanted directly after removal (Lu et al. 2001, 2002). There are, however, suggestions that the properties of OECs may be influenced by the cells with which they are transplanted. For example, meningeal cells are reported to enhance the remyelinating properties of OECs (Lakatos et al. 2003), and some consider that the fibroblasts contained in non-purified cell cultures prepared from the olfactory bulb are crucial to their regenerative properties (Raisman, 2001). The extent to which the mixture of cells included in the graft influences the resulting repair will be an important issue to pursue in future investigations.

Possible mechanisms of functional recovery after OEC grafting

There are several reports, based mainly on evidence from behavioural testing, that transplants of OECs can provide a remarkable improvement in recovery of function after spinal cord injury (Li et al. 1997, 2003; Ramon-Cueto et al. 2000; Lu et al. 2001, 2002; Nash et al. 2002). The axonal regeneration obtained using transplants of OECs, although apparently modest (e.g. Li et al. 1998; Ramon-Cueto et al. 1998, 2000), might provide a sufficient basis for the functional benefits observed. Tempting though it is to associate the regeneration described with the improved functional recovery, there is as yet no direct evidence that regenerating fibres growing through OEC grafts are able to re-establish functional connections across spinal cord lesions. Furthermore, as we describe above, previous results that appeared to provide such evidence for repair at the dorsal root entry zone are in question. There are, in addition, many other mechanisms by which OECs might bring about improvements in function; mechanisms that do not require the re-connection of damaged pathways. (1) In animals with partial transections, compensatory plastic changes (e.g. axonal sprouting) may occur in alternative pathways not directly affected by the lesion. (2) In animals with incomplete lesions or complete spinal cord transections, regeneration of descending fibres releasing monoamines at non-synaptic sites (Fuxe & Agnati, 1991) could affect the excitability of reflex pathways (Thompson et al. 1992), activity of the spinal cord central pattern generators (Brustein & Rossignol, 1999; Gimenez et al. 2000; Marcoux & Rossignol, 2000) or the tonic firing properties of motorneurons (Bennett et al. 2001a,b). (3) Transplants of OECs might improve the general environment of the lesion site and help preserve the function and viability of spared fibres in adjacent tissue; they might also release trophic substances (Boruch et al. 2001; Woodhall et al. 2001), reduce the production of inhibitory factors by glial cells or remyelinate spared axons with damaged myelin sheaths. That mechanisms such as those outlined above are likely to be invoked following spinal cord injury is demonstrated by studies in which behavioural changes have been found to occur in the absence of axonal regeneration (Houweling et al. 1998; Jakeman et al. 1998; Ribotta et al. 2000).

The possibility that OECs exert their beneficial influence by mechanisms other than functional re-connection does not diminish their value as a potential means of treatment in spinal cord injury. It does, however, have implications for the way in which OEC transplants might be applied in the clinic. Although all of the mechanisms outlined above could enhance residual function in patients with partial spinal cord injuries, only restoring synaptic transmission across the site of injury can return useful sensation and co-ordinated movement to those with complete lesions.

In conclusion, olfactory ensheathing cells represent arguably the leading candidate for transplant-mediated repair of spinal cord injury, though the mechanism(s) by which they produce their beneficial effects remain to be clarified. A better understanding of these mechanisms will inform a rational strategy for optimizing the use of OECs, indicate the type of patient most likely to benefit from their use and aid the selection of therapies with which OEC transplants could most beneficially be combined. Whatever the eventual outcome of these investigations, there is little doubt that further work on the cell biology of the OEC and its behaviour following transplantation will advance our understanding of spinal cord injury and contribute importantly to the development of ways by which the spinal cord can eventually be repaired.