The expression and distribution of laminin in the developing nervous system

Neuronal survival and differentiation in both the developing peripheral and central nervous systems depend upon a variety of epigenetic factors. In addition to stimulating the rapid growth or regeneration of neurites by embryonic neurons (Collins, 1980), extracellular matrix molecules produced by a variety of cells can greatly potentiate the survival-promoting effects of neurotrophins such as the nerve growth factor (Edgar and Thoenen, 1982). The non-collagenous, extracellular matrix glycoprotein laminin is unique in that it evokes such effects on neurons both in vitro (Baron-Van Evercooren et al. 1982; Edgar et al. 1984) and in vivo (Kalcheim et al. 1987). Furthermore, laminin tissue culture substrates alone have been shown to enhance the expression of the enzyme, tyrosine hydroxylase, that regulates the synthesis of catecholaminergic neurotransmitters (Acheson et al. 1986), and such substrates also permit the survival of neuronal precursor cells, without the need for the presence of a neurotrophin (Ernsberger et al. 1989).

Laminin isolated from the Engelbreth-Hohn-Swarm (EHS) mouse sarcoma tumour (Timpl et al. 1979) has been cleaved by proteolysis to reveal that the site which evokes neurite outgrowth and neuronal survival is located towards the end of the long arm of this cruciform molecule (Edgar et al. 1984). Because tertiary protein structure and sites sensitive to proteolysis appear to be necessary for biological activity, then further attempts to increase the resolution of mapping of active site(s) have been largely unsuccessful (Edgar et al. 1988; Deutzmann et al. 1990). However, inhibition with domain-specific antibodies do show that the site responsible for neurite outgrowth, neuronal survival (Edgar et al. 1984) and induction of tyrosine hydroxylase (Acheson et al. 1986) is located at or near the large globular domain at the end of the long arm. Many cells are able to interact with this or a closely juxtaposed site (Aumailley et al. 1987; Dillner et al. 1988). However, the greater sensitivity to denaturation and/or proteolysis of the neurite-promoting activity of this region indicates that stimulation of neurite outgrowth requires laminin structures different from, or in addition to, those required for cell adhesion (Deutzmann et al. 1990).

Laminin was originally identified as being essentially only located in all basement membranes of the adult (Timpl et al. 1979). In this location, it is likely that long arm cell binding sites are available for interaction with their responsive cells: epitopes found on the long arm of laminin are accessible outside the basal lamina, indicating that this part of the molecule is exposed in the adult (Schittny et al. 1988; Abrahamson et al. 1989). It should be noted that a major potential cell binding site in the short arms of laminin is masked not only within the basement membrane (Schittny et al. 1988; Abrahamson et al. 1989), but also by the short arm globular domains (Nurcombe et al. 1989). Thus, in addition to sites overtly expressed by basement membrane-bound laminin, this matrix molecule also possesses sites which have the potential to interact with cells either after activation by proteolytic release, or alternatively, if the laminin were not confined to basement membranes.

Variants of laminin lack the epitopes recognised by blocking antibodies, owing to a different laminin subunit composition (Edgar et al. 1988). At least some of these variants can nevertheless stimulate neurite outgrowth (Edgar et al. 1988), indicating that antibody inhibition of EHS tumour laminin is likely to be by steric hindrance, the blocking antibodies recognising an epitope unique to EHS laminin which is close to (but not the actual) active site. Laminin variants may lack this epitope, but because some of them do promote neurite outgrowth they may be expected to share the active site which is likely to be non-antigenic (Edgar et al. 1988). One unfortunate practical consequence of this observation is that it is unlikely that antibodies blocking the activities of EHS tumour laminin will be of use in attempts to block the biologically important functions of other laminin variants in vivo (see below).

In contrast to its situation in the adult, it has been reported previously that laminin immunoreactivity in many different embryonic tissues appears as punctate deposits independent of basement membranes (Ekblom et al. 1980; Dziadek and Mitrangas, 1989). Such immunoreactivity is most noticeable in the interstitial extracellular matrix or mesenchyme, where this is a prominent feature of the tissue (Ekblom et al. 1980; Dziadek and Mitrangas, 1989). However in the nervous system, punctate deposits of laminin-immunoreactive material have been noted on the surface of neural cells, there being comparatively little interstitial extracellular matrix in the central nervous system (Liesi, 1985; Cohen et al. 1987; McLoon et al. 1988). Because this laminin appears transiently at the times when axonal growth is occurring, it is possible that laminin is one of the components within the immediate vicinity of the axonal growth cone which promotes axonal extension (see Bixby et al. 1987). This likelihood is strengthened by the observation that developing neurons also express receptors capable of inducing neurite outgrowth in response to intact mouse EHS tumour laminin at this time (Cohen et al. 1989).

Immunohistochemical analysis of laminin in the adult sciatic nerve has shown a differential distribution of laminin subunits (Sanes et al. 1990). Thus, while the laminin of the endoneurium was [A⁻B1⁺B2⁺M⁺S⁻], as determined by subunit-specific antibody cross-reactivities, that of the perineurium was [A⁺B1⁻B2⁺M⁻S⁺]. Significantly, preliminary experiments have indicated that the interactions of laminin variants with neural cells are influenced by the laminin subunit composition, so that while both A chain- and M chain-containing human laminins can support neurite outgrowth from sympathetic neurons, the rate of neurite growth is much faster on the A chain variant (unpublished observations; see also Hunter et al. (1989), for evidence that the S chain variant may actually inhibit neurite growth). In the light of these observations, it is clear that until the molecular composition, and consequently the biological activity, of the laminin responsible for any immunoreactivity seen in vivo is established, then it is not clear to what extent a response demonstrated by EHS tumour laminin in vitro can be used to confirm that an endogenous laminin variant is active and therefore likely to stimulate axonal growth: the variant(s) expressed in the developing nervous system may lack, or alternatively, have exposed different active sites. From the above discussion, it may be seen that this could either be due to their subunit structure, or because they are not in a basement membrane, or because they have been subjected to proteolytic degradation exposing novel active sites.

Not only the expression of laminin receptors and response (Cohen et al. 1989) are developmentally regulated: it is now clear that laminin gene expression is down-regulated in several tissues, as reflected by the decrease in steadystate levels of the mRNA coding for laminin subunits during development (Senior et al. 1988; Kücherer-Ehret et al. 19906; Ekblom et al. 1990). Thus, in the sciatic nerve of the developing postnatal mouse it may be seen that mRNA levels decrease from birth to adulthood (Table 1). It is however, unknown what factors regulate laminin gene expression in this or other tissues, nor have the molecular mechanisms been established which determine the differential expression of laminin subunits (Cooper and Mac-queen, 1983; Ohno et al. 1986; Kleinman et al. 1987; Liesi and Risteli, 1989; Hunter et al. 1989; Paulsson and Saladin, 1989; Ehrig et al. 1990; Ekblom et al. 1990). Significantly, there is currently little evidence for a developmentally regulated differential expression of the A or Bl and B2 subunits, which has been suggested to be responsible for alternative localisations of laminin either within, or extraneous to, basement membranes (see McLoon et al. 1988). Indeed, most tissues display very low levels of the laminin A chain, relative to the B chains, throughout development (Kücherer-Ehret et al. 19906; Ekblom et al. 1990).

One consequence of high levels of laminin expression during development is likely to be that laminin accumulates at a rate faster than that with which it can be incorporated into basement membranes: it has been noted previously that laminin synthesis is not the ratedetermining step in basement membrane production (Clark and Bunge, 1989). In the peripheral nervous system it been shown that the Schwann and other cells enshea-thing the axons can synthesise and secrete laminin (Cornbrooks et al. 1983; McGarvey et al. 1984; Jaakkola et al. 1989). However, in the developing sciatic nerve it was demonstrated by immuno-electronmicroscopy that most of this laminin accumulates extracellularly where it is found in the interstitial extracellular matrix (mesenchyme), rather than being localised exclusively in the basement membranes as in the adult (Kücherer-Ehret et al. 1990a).

Although the laminin of the developing sciatic nerve is apparently associated mainly with interstitial collagen fibres (Kücherer-Ehret et al. 1990a), this association is extremely weak or even artifactual, possible arising from the tissue fixation procedures used in histochemistry: most of the laminin is freely extractable from unfixed tissue into physiological buffer solutions (Table 2). Similarly, it has been noted that the laminin and other basement membrane components are easily extractable from a variety of embryonic tissues, indicating that the phenomenon of soluble laminin is likely to be widespread throughout the tissues of the embryo (Dohrmann et al. 1986; Dziadek and Mitrangas, 1989).

One consequence of the change in levels of laminin gene expression during development is that this brings about an effective change in laminin distribution from the mesenchyme or interstitial extracellular matrix, to the basement membrane. Thus neural cells (and those of other tissues) have extensive access to laminin during development, although they have little or no contact with basement membranes in the adult. Furthermore, because the laminin is not sequestered into basement membranes during development, then other cell binding sites on the molecule might be able to interact with their receptors, such sites being masked within the basement membrane in the adult (Schittny et al. 1988; Abrahamson et al. 1989). Finally, laminin located outside basement membranes is more likely to be susceptible to proteolysis (see Dziadek and Mitrangas, 1989), giving rise to the possibility that cryptic cell binding sites within the molecule may be activated (Nurcombe et al. 1989).

Given the presence of laminin variants which are likely to have biologically active sites different from, or in addition to, EHS tumour laminin, then it is clear that laminin has a greater potential to influence the behaviour of neural and other cells during development than might be expected from its structures and distribution in the adult.

This work was supported by the Wellcome Trust.