Studies on cell behaviour in vitro have indicated that the chondroitin sulphate proteoglycan (CSPG) family of molecules can participate in the control of cell proliferation, differentiation and adhesion, but its morphogenetic functions had not been investigated in intact embryos. Chondroitin/chondroitin sulphates have been identified in rat embryos at low levels at the start of neurulation (day 9) and at much higher levels on day 10. In this study we have sought evidence for the morphogenetic functions of CSPGs in rat embryos during the period of neurulation and neural crest cell migration by a combination of two approaches: immunocytochemical localization of CSPG by means of an antibody, CS-56, to the chondroitin sulphate component of CSPG, and exposure of embryos to the enzyme chondroitinase ABC. Staining of the CS-56 epitope was poor at the beginning of cranial neurulation; bright staining was at first confined to the primary mesenchyme under the convex neural folds late on day 9. In day 10 embryos, all mesenchyme cells were stained, but at different levels of intensity, so that primary mesenchyme, neural crest and sclerotomal cells could be distinguished from each other. Basement membranes were also stained, particularly bright staining being present where two epithelia were basally apposed, e.g. neural/surface ectoderms, dorsal aorta/neural tube, prior to migration of a population of cells between them. Staining within the neural epithelium was first confined to the dorsolateral edge region, and associated with the onset of neural crest cell emigration; after neural tube closure, neuroepithelial staining was more general. Neural crest cells were stained during migration, but the reaction was absent in areas associated with migration end-points (trigeminal ganglion anlagen, frontonasal mesenchyme). Embryos exposed to chondroitinase ABC in culture showed no abnormalities until early day 10, when cranial neural crest cell emigration from the neural epithelium was inhibited and neural tube closure was retarded. Sclerotomal cells failed to take their normal pathway between the dorsal aorta and neural tube. Correlation of the results of these two methods suggests: (1) that by decreasing adhesiveness within the neural epithelium at specific stages, CSPG facilitates the emigration of neural crest cells and the migratory movement of neuroblasts, and may also provide increased flexibility during the generation of epithelial curvatures; (2) that by decreasing the adhesiveness of fibronectin-containing extra-cellular matrices, CSPG facilitates the migration of neural crest and sclerotomal cells. This second function is particularly important when migrating cells take pathways between previously apposed tissues.

Chondroitin sulphate proteoglycans (CSPG) are a common constituent of the extracellular matrix, both in adult tissues and in the embryo, in both vertebrate and invertebrate animals. They consist of chains of repeating disaccharide units (glucuronic acid and N-acetylgalactosamine) attached to a protein core by a serine-xylose-galactose-galactose linkage; sulphation may occur in either the 4- or 6-position of the N-acetylgalactosamine residue (Roden, 1980; Dorfman, 1981; Höök et al. 1984). Variability within the CSPG family of molecules involves core protein characteristics, number and length of glycosaminoglycan (GAG) chains, the extent to which glucuronic acid residues are epimerised to iduronic acid, and the extent and position of sulphation (Couchman et al. 1984; Hassell et al. 1986). Specific forms of CSPG are found in different tissue matrices (Couchman et al. 1984).

CSPG has been identified in rat embryos at low levels prior to and during gastrulation and at the start of neurulation, and at higher levels in late neurulation-stage embryos (Solursh and Morriss, 1977). Studies on the effects of CSPG on cell behaviour in vitro have indicated that according to the prevailing conditions, this family of molecules may participate in the control of cell proliferation, adhesion, and differentiation. However, its morphogenetic function in intact embryos is not understood. The morphogenetic effects of β-D-xyloside, an inhibitor of proteoglycan synthesis, are mimicked by heparitinase, suggesting that heparan sulphate proteoglycan(s) (HSPG) play(s) a more significant role in the control of epithelial morphogenesis at this stage (Morriss-Kay & Crutch, 1982; Tuckett & Morriss-Kay, 1989).

The form or forms of core protein present in the extracellular matrix CSPG of neurulation stage rat embryos is not known. Therefore, for immunohisto-chemical localization, we have used a monoclonal antibody to the GAG portion of CSPG. We describe here the distribution of CSPG, as localised by this antibody, from the early neural plate (late presomite) stage to the start of postneurulation morphogenesis of the cranial neural tube. In order to gain some insight into the morphogenetic function of CSPG in early craniofacial development, we have exposed embryos to chondroitinase ABC by intra-amniotic injection and subsequent culture, and by culture in medium containing the enzyme. This enzyme primarily degrades chondroitin sulphates, with weaker degradation of dermatan sulphate, chondroitin and hyaluronate. Dermatan sulphate is not present in embryos of this stage (Solursh & Morriss, 1977), and the effects of an enzyme specific to hyaluronate have been studied previously (Morriss-Kay et al. 1986).

Immunohistochemistry

Embryos were explanted from pregnant dams at a series of developmental stages from day 9 (p.m.) to day 11 (a.m.); the yolk sac and amnion were left intact on day 9 embryos and removed from older embryos to aid orientation, except that removal was sometimes incomplete to enable the staining reaction of this tissue to be observed. They were fixed in St Marie’s fixative, embedded in paraplast and sectioned at 8 pm. After rehydration, the sections were permeabilised with 0-05 % Triton X-100 in PBS, then incubated sequentially with a monoclonal antibody to chondroitin sulphate, CS-56, at a dilution of 1:50 in PBS, and fluorescein-conjugated anti-mouse IgG Fab at a dilution of 1:20 in PBS (both antibodies were obtained from ICN Biomedicals). The epitope bound by CS-56 is associated with the GAG moieties of native CSPG, as shown by several criteria including solid-phase radioimmuno-assay with purified GAGs and sensitivity of the antigen to highly purified chondroitinases ABC and AC as well as pronase; it can therefore be used to specifically identify CSPG within the heterogeneous matrix of extracellular proteoglycans (Avnur & Geiger, 1984). Control sections were prepared as above except that PBS was substituted for the primary antibody (Fig. 4B). The sections were mounted in Aquamount (Gurr) and viewed with an Olympus BH-2 microscope fitted with an indirect fluorescence attachment.

Embryo culture

Wistar strain rat embryos were explanted in Tyrode’s saline on the afternoon of day 9 of pregnancy (day of positive vaginal smear = day 0). After removal of Reichert’s membrane they were placed in culture in 60tnl glass bottles. The culture medium was 5 ml heat-inactivated rat serum containing 50i.u. streptomycin and penicillin; the gas phase contained 5 % oxygen for the first 24 h and 20 % thereafter. The bottles were rotated at 30 revs min-1 at 38°C. At the start of culture, embryos were at the late presomite/early somite stage.

Chondroitinase ABC

Chondroitinase ABC from Proteus vulgaris (Seikagaku Kogyo Co., Ltd) was obtained from ICN Biomedicals. The manufacturers’ information relevant to this study, derived from specificity studies by Yamagata et al. (1968), Saito et al. (1968) and Suzuki et al. (1968), is as follows. The enzyme catalyses the elimination cleavage of N-acetylhexosaminide linkages in chondroitin-4-sulphate, chondroitin-6-sulphate, dermatan sulphate, chondroitin and hyaluronic acid, yielding mainly disaccharides. The initial rates of degradation of chondroitin sulphate C, dermatan sulphate, chondroitin and hyaluronic acid were 1·1, 0·4, 0·2, and 0·02, respectively, relative to the rate of chondroitin sulphate A degradation. Hence the degradative effect on the hyaluronic acid component of the embryonic extracellular matrix is one fiftieth of the effect on either of the forms of chondroitin sulphate. It does not act on keratan sulphate, heparin or heparan sulphate. Evidence for the absence of a degradative effect on hyaluronate in this study is provided by fig. 7, which shows an abundance of extracellular matrix resembling that of control embryos. In an earlier study (Morriss-Kay et al. 1986), exposure of embryos to the hyaluronate-specific enzyme Streptomyces hyaluronidase resulted in abolition of intercellular spaces in the cranial mesenchyme.

Exposure to the enzyme and subsequent processing

70 embryos were cultured for 18-24 or 42 h in medium to which 1 i.u. ml-1 enzyme (diluted in 25 μl Tyrode’s saline) had been added; 43 embryos were cocultured in bottles to which 25μ1 Tyrode’s saline without enzyme had been added. The enzyme solution was added to the medium before addition of the embryos so that any protease contaminants would be inactivated by the protease inhibitors and proteins present in the serum.

56 embryos were exposed to the enzyme by microinjection of approximately 1 μl of a 10 i.u. m11 solution of chondroitinase ABC in phosphate-buffered saline (PBS) directly into the amniotic cavity, using a Leitz micromanipulator as described in Tan and Morriss-Kay (1986). 59 controls were injected with a solution of the enzyme that had been heat-inactivated by boiling for 1 h; 5 embryos were injected with PBS alone. After injection, the embryos were placed in fresh culture bottles and incubated for 18 or 24 h (i.e. to the morning or afternoon of day 10). Of the active enzyme-injected embryos, 5 were reinjected with chondroitinase solution, and 5 with heat-inactivated enzyme solution. Of the embryos that had received an injection of heat-inactivated enzyme on day 9, 5 were reinjected with this solution and 5 with the active enzyme solution. They were placed in fresh culture solution, gassed with 5 % CO2 in air, and cultured for a further 18 h (to the morning of day 11). Prior to the termination of culture, the yolk sac circulation and heartbeat were assessed by viewing the embryos with a dissecting microscope within an incubator at 38 °C. Embryos to be processed were then washed in Tyrode’s saline; the membranes were removed and the morphology of the embryos was assessed. Most were photographed at this stage, or after fixation, for later reference. 46 active enzyme-injected, 49 heat-inactivated enzyme-injected and 2 PBS-injected embryos were fixed in 4% cacodylate-buffered glutaraldehyde, and either prepared for scanning electron microscopy or embedded in Spurr resin and sectioned at 0·5-1 μm for light microscopy (further details of methods are described in Morriss-Kay & Tuckett, 1985).

As an assay for the activity of the enzyme after injection into the amniotic cavity, three day 10 embryos that had been injected with active enzyme solution and three injected with heat-inactivated enzyme solution, followed by culture for 24 h, were processed for immunohistochemistry as described above.

Immunohistochemistry

In late presomite-stage embryos (Fig. 1A), faint staining of the CS-56 epitope was present in all germ layers, as follows: apical surface of the whole ectoderm; basement membrane of the neural epithelium, extending into the lateral epiblast but not as far as the primitive streak; all mesenchyme; all endoderm, including that of the visceral yolk sac.

During early somite (convex cranial neural fold) stages, staining was undetectable in the embryo, but was present in the extraembryonic endoderm (Fig. 1B). Mesenchymal and neuroepithelial basement membrane staining reappeared as the cranial neural folds increased in height (Fig. 1C), and was very strong by the 5-somite stage (Fig. 1D), remaining strong in all subsequent stages examined (Figs 2,3). In early day 10 embryos, the foregut showed strong staining of the luminal contents, and weaker staining of the basement membrane; basement membranes of all other epithelia were strongly stained, while staining around the notochord was variable (Fig. 2, A-E). Staining within the neural epithelium was first seen near the lateral edges of the midbrain and hindbrain regions, from which neural crest cells were either emigrating or were at a late premigratory stage (Fig. 2D,E), and in the rostral (formerly lateral) part of the forebrain (Fig. 2A). Prior to crest cell emigration there was particularly bright staining of the extracellular matrix between the apposed parts of the neural and surface epithelia (Fig. 2D and the left side of 2E); where neural crest cells were emigrating, strands of brightly stained matrix provided a continuity between cells within the lateral region of the neural epithelium and the adjacent mesenchymal crest cell population (Fig. 2E, right side).

After cranial neural tube closure (Fig. 3A-D), neuroepithelial staining was distributed throughout the embryonic brain; the staining intensity was low in the roof and floor plates, but high in the walls. Staining was also strong within the epithelium of the otic pit, and in the surface ectoderm adjacent to it (Fig. 3A). All basement membranes were stained, the brightest being the neuroepithelial basement membrane and sites of apposition of two basal epithelial surfaces, e.g. otic:neural, pharyngeal ectoderm:endoderm (Fig. 3A); optic: surface ectoderm (Fig. 3C). Within the mesenchyme, differences in staining intensity may reflect neural crest (weak) and primary mesenchyme (strong) regions (Figs 3A,B); this uneven pattern was even more marked in day 11 embryos, where very weak staining was associated with the trigeminal ganglion anlagen and with the frontonasal mesenchyme, i.e. with postmigratory neural crest cells (Fig. 3C).

In the trunk region (Figs4A, 5C), the staining pattern of epithelia and basement membranes was similar to that of the head. Staining of the neuroepithelial basement membrane was continuous around the notochord, but was absent between the notochord and neural tube. Sections through somites showed unstained dermamyotome (except for the basement membrane) and stained matrix around the migrating sclerotome cells. Differential staining of the mesenchymal cell populations enabled neural crest, sclerotome and primary mesenchyme to be distinguished (fig. 5C): staining was brightest around the crest cells adjacent to the neural tube, and both lateral and medial borders of the sclerotome were usually defined by a line of brightly staining matrix. Staining of the visceral yolk sac was at first confined to the endodermal layer (Fig. 1B-D); this increased in intensity to be very bright on the apical surface by the time the yolk sac was vascularized (late day 10, Fig. 4C). By this stage, there was also some staining in the yolk sac mesoderm, though this was very faint compared with the staining around blood vessels within the embryo.

Examination of live embryos after culture

After 18 h in culture, embryos injected with heat-inactivated enzyme or PBS, and uninjected control embryos, had a similar morphology to in vivo embryos of this stage. Somite number ranged from 7 to 11 pairs; all had good heartbeats; the most advanced were beginning to turn, and had circulating blood in the yolk sac. During the subsequent 6h, embryos left in culture developed up to 15 pairs of somites, a strong yolk-sac circulation, and progressed towards or achieved cranial neural tube closure as appropriate to their somite stage. After 42 h in culture, embryos that had received two injections of heat-inactivated enzyme solution had 24-28 somite pairs, and normal morphology as seen in equivalent uninjected control and in vivo embryos.

For the first 24 h embryos exposed to chondroitinase by immersion developed similarly to the microinjected embryos (see below); these are included in the scanning electron microscopy (SEM) results. Embryos examined after 42 h of culture in medium containing chondroitinase had a weak yolk-sac circulation with narrow vessels; the abnormalities observed in the embryos themselves varied in severity, but the cranial neural tube was always closed. It was not possible to distinguish embryonic effects due directly to the enzyme from those that were secondary to the enzyme’s effects on yolk sac function.

Embryos injected with a solution containing chondroitinase ABC on day 9 had good heartbeats on day 10 and after either type of second injection when examined on day 11. On day 10 the yolk-sac circulation was slightly less strong than that of controls, and on day 11 it was feeble or static, sometimes with a blistered appearance. Somites were well defined on day 10 but those added during the second day of culture, after either type of reinjection, were poorly defined and difficult to count, and the trunk was reduced in length compared with controls. As with the embryos exposed to chondroitinase in the culture medium for 42 h, the developmental abnormalities in embryos which received two chondroitinase injections are likely to be the combined effects of poor yolk sac-mediated nutrition and more direct embryonic effects of the enzyme.

Immunohistochemistry of injected embryos

The immunocytochemical staining pattern of embryos injected with heat-inactivated enzyme solution followed by culture for 24 h was indistinguishable from that of in vivo controls (Fig. 5A,C, compared with Figs 3B and 4A respectively). Embryos cultured for 24 h following injection with active enzyme solution showed faint staining in areas that are particularly bright in the controls, and no staining elsewhere (Fig. 5B,D). Fig. 5B shows very faint staining of the apposed basement membranes of otic and neural epithelia, compared with the normal high intensity (Fig. 2C) in an embryo that is younger but at a similar morphological stage. In the trunk region, very poor staining was associated with failure of sclerotomal cells to be confined to their normal funnel-shaped pathway or to enter the potential space between the dorsal aorta and neural tube on each side (Fig. 5D compared with Fig. 5C).

Scanning electron microscopy of injected embryos

1. Embryos examined on day 10

Embryos injected intra-amniotically on day 9 with either heat-inactivated or active chondroitinase ABC showed a normal morphology after 18 h of culture (up to the 8- to 9-somite stage) by the criteria of previous studies and in comparison with in vivo embryos of the same somite stages (Fig. 6A,B). Embryos injected with chondroitinase ABC on day 9 and examined after 24 h showed slight differences from control embryos from the 9-somite stage onwards. These differences were characteristic and reproducible: at the 9- and 10-somite stages, the rostral hindbrain and midbrain neural folds were slightly more widely open than those of controls, even though cervical and caudal hindbrain regions had closed to the same extent as in controls (Fig. 6C,F). At the 10-somite stage the neural folds were not forming the normal apposition point at the midbrain-forebrain junction; instead, they were parallel in this region, and lacked the sharp-edged, incurved appearance normally seen in the midbrainμostral hindbrain neuropore at this stage. Observation of a sequence of more advanced embryos showed that these incurved neural folds came together when the embryos had approximately 16 somite pairs, making a long contact unlike the normal pattern of apposition; a teardrop-shaped midbrain/ rostral hindbrain neuropore was then formed, followed by complete closure (Fig. 6G,H, compared with Fig. 6D,E).

2. Embryos examined on day 11 (Fig. 6I,J,K,L)

Embryos given a second injection of heat-inactivated enzyme solution on day 10 (24 h after the first) and cultured for a further 18 h were indistinguishable from uninjected controls. Embryos injected with a solution containing chondroitinase ABC following heat-inactivated enzyme solution on day 9, or the reciprocal, were considerably smaller than uninjected or injected controls, but the effect on morphology was slight. Embryos given a second injection of chondroitinase solution were even more reduced in size; somites were rather indistinct, the trunk was short and had failed to turn; in the head, the pharyngeal arches were less well developed and the otic pits were still open. All embryos had closed cranial neural tubes, and caudal neuropores of similar sizes.

Embryos exposed to chondroitinase ABC which had been added to the culture medium were indistinguishable from embryos which received two chondroitinase injections (Fig. 6M).

Light microscopy (Fig. 7)

Examination of sections of the head in 9- to 12-somite stage embryos revealed that the general tissue structure of enzyme-injected embryos was similar to that of controls, and dead cells were equally rare in both types of embryo. The neural epithelium was thicker than in equivalent controls; changes normally observed at the lateral edge during the 9- to 12-somite period, associated with formation and closure of the spindle-shaped midbrain/upper hindbrain neuropore (loss of neural crest, development of a sharp mediad curvature) occurred late. The general appearance of the cranial neural epithelium was of an abnormally thick structure. The notochord, foregut and cranial mesenchyme appeared to be normal, except that there were fewer of the more solid, laterally placed mesenchymal cells which are probably neural crest.

Embryos examined after 42 h of culture (not illustrated), following one or two injections of active enzyme solution, showed very compact mesenchyme. Neuroepithelial curvatures were less pronounced than normal, and all epithelia were thicker than those of control embryos. The level of cell death in active enzyme-injected embryos was unacceptably high (10 to 25 % of nuclei were pyknotic) for detailed analysis to be worthwhile.

The primary antibody used in the immunocytochemical part of this study, CS-56, is specific for the GAG portion of CSPG, so the observed staining pattern is likely to indicate the locations and times of expression of all CSPGs present in the embryo during days 9 to 11 of development. In general, the amount of staining increased from very low levels at the start of neurulation on day 9, to very high levels on days 10 and 11. These observations are consistent with those of a previous biochemical study which showed that levels of chondroitin/chondroitin sulphates are low on day 9 but rise tenfold between day 9 and day 10 (Solursh & Morriss, 1977).

The comparatively low level of staining of the extraembryonic membranes on day 9 and day 10 also confirms the earlier biochemical measurements. The contrastingly high level of embryonic CSPG on day 10 shown by both immunohistochemical and biochemical methods suggests strongly that the abnormalities in embryos injected with intra-amniotic chondroitinase ABC on day 9 and cultured to day 10 were due to a direct effect of the enzyme on the embryo, rather than to a deleterious effect on yolk-sac-mediated nutrition. With vascularization, staining of the visceral yolk sac increased in intensity, and embryos exposed to the enzyme for a second day in culture showed some cell death and may have been affected by the relatively poor development of the yolk-sac circulation; they are therefore not taken into account when interpreting our results in terms of possible roles of CSPG in development.

The earliest detectable abnormalities in embryos developing in the presence of chondroitinase ABC appeared at the 10-somite stage, during which the V-shaped midbrainμostral hindbrain neural folds normally begin to convert to the concave form which is associated with the loss of neural crest cells from the lateral edges, and which precedes apposition of the sharp-profiled edges in the dorsal midline (Morriss & New, 1979). Neural crest cell emigration was retarded, so that the lateral edges were blunt in profile; the concave curvature was less pronounced than normal. This pattern of associated abnormality of cell behaviour and morphogenesis is similar to that seen in embryos cultured with a gas phase containing 20% or 40% oxygen during the first 24 h (Morriss & New, 1979); it is not known whether high oxygen levels are associated with reduced synthesis of CSPGs. Evidence that these results are due to the degradative action of the enzyme and not to the presence of protease contaminants is provided by the lack of any detectable difference in the abnormalities produced in embryos exposed to enzyme added to the culture medium, which being whole serum contains protease inhibitors, and embryos exposed to the enzyme by intra-amniotic injection. Either the level of protease contaminants was so low as to have no detectable effect, or there are protease inhibitors in amniotic fluid.

Some possible functions for CSPGs in early craniofacial development in rat embryos can be proposed by correlating the results of this study with normal developmental events and with studies on CSPG in other experimental systems. All of the correlations made below indicate an association between CSPG levels and the degree of cell-cell or cell-substrate adhesiveness.

1. Infiltration of CSPG into the lateral edges of the cranial neural epithelium at the time of onset of neural crest cell migration, and inhibition of crest cell emigration in chondroitinase-exposed embryos, suggest that CSPG plays a role in the conversion of crest cells from an epithelial to a mesenchymal organization. Newgreen & Gibbins (1982) have proposed that decrease in cell adhesiveness is the primary cause of the onset of neural crest cell migration. The change in organization of neural crest cells from an epithelial to a mesenchymal form is associated with a change from cell-cell to cell-substrate adhesion, with loss of N-CAM and A-CAM immunoreactivity (Thiery et al. 1982; Tucker et al. 1988). Clearly, the adhesion between emigrating crest cells and local extracellular matrix should not be too great.

2. Intense staining of the extracellular matrix appeared between the lateral neural epithelium from which crest cells were emigrating, and the adjacent surface ectoderm. Prior to emigration, these two tissues are closely apposed (Tan & Morriss-Kay, 1985); in chondroitinase-exposed embryos, in which staining was not seen in this position, surface ectoderm and lateral neural ectoderm remained in apposition. Similarly, strong matrix staining between the trunk/hindbrain dorsal aortae and the neural tube was associated with migration of sclerotomal cells between these two epithelia, while this migration pathway was not taken in embryos exposed to chondroitinase. Both observations suggest a role for CSPG in separation of two apposed epithelia in order to provide a pathway for cell migration between them.

3. Infiltration of CSPG staining into the rostral forebrain neural epithelium occurred in relation to a change in epithelial curvature. The ability to curve suggests an increase in epithelial flexibility, which may involve the movement of cells against each other. Curvature of the lateral neuroepithelial edges prior to neural tube closure may similarly be facilitated by the infiltration of CSPG here at the time of crest cell emigration. This interpretation cannot be applied to the change in curvature of the midbrainμostral-hindbrain neural folds from convex to concave, when only the basement membrane component of the neural epithelium was stained. This difference may be due to the fact that during this process there is an increase in cell-cell contact with epithelial thickening (Morriss-Kay, 1981), so that adhesiveness is more likely to be increasing than decreasing.

4. The staining of the whole neural epithelium (in spinal as well as cranial regions) observed after neural tube closure correlates with the time of onset of migration of neuroblasts into the mantle layer (Langman et al. 1966), and with the onset of neurite extension (Cajal, 1929). In the cranial region, the presence of CSPG may also be associated with the neuroepithelial curvatures which form during early morphogenesis of the embryonic brain.

5. The timing of changes in staining intensity for CSPG in the cranial mesenchyme may be correlated with neural crest cell migration: staining of the primary mesenchyme was weak at the start of cranial neurulation, but by the late convex/early V-shaped neural fold stages (6-8 somites), during which cranial neural crest cell migration begins in rat embryos (Tan & Morriss-Kay, 1985), primary mesenchyme was strongly stained. Cells that are likely to be neural crest cells (because of their position at specific times) were slightly less well stained than primary mesenchyme during the migratory phase on day 10, and faintly stained or unstained at the end of their migration pathways (frontonasal mesenchyme, trigeminal ganglion anlagen) on day 11. If staining intensity for CSPG reflects differences in adhesiveness of the extracellular matrix (see below), this pattern suggests that neural crest cells are slightly more adhesive than primary mesenchyme while migrating, and that they increase their adhesiveness considerably at the end of their migration pathways. Modulation of CSPG levels may therefore play a role in changes of cell-matrix and cell-cell adhesiveness in vivo, affecting the behaviour of neural crest cells.

In summary, the results indicate that CSPG may play a role in facilitating cell movement within and between embryonic tissues. This interpretation is consistent with in vitro studies in which CSPG or more complex chondroitin sulphate-containing proteoglycans have been observed to decrease the adhesiveness of chick embryonic fibroblasts and neural crest cells to various substrata including collagen (Knox & Wells, 1979) and fibronectin (Newgreen, 1982; Tan et al. 1987). Studies using chondroitin sulphate alone are probably not relevant, since the effect requires intact CSPG (Knox & Wells, 1979); CSPG, but not chondroitin sulphate, binds to the GAG-binding site on fibronectin (Laterra et al. 1980; Yamada et al. 1980; Rich et al. 1981). The interaction between CSPG and fibronectin may be particularly important in relation to neural crest cell migration in vivo: fibronectin is necessary for cell motility, but too much may be inhibitory, stimulating the formation of stress fibres and focal contacts (Couchman & Rees, 1979; Couchman et al. 1982). Different concentrations of CSPG have been demonstrated to modulate the adhesivity and migratory activity of neural crest cells on fibronectin substrata in vitro (Perris & Johansson, 1987; Tan et al. 1987); therefore, CSPGs may function in vivo to decrease the adhesiveness of crest cells to fibronectin to a level compatible with movement. Conversely, it has been suggested that, at very high concentrations, cell movement may be inhibited by an inability to make adhesive contacts: failure of crest cells to colonise the perinotochordal space has been attributed to high levels of CSPG in that region (Newgreen et al. 1986). This hypothesis is not consistent with our observations that crest cell emigration from neural epithelium, and sclerotomal cell migration between the dorsal aortae and neural tube, were associated with exceptionally bright staining of the CS-56 epitope and were inhibited in embryos in which CSPG levels were low following exposure to chondroitinase ABC. Also, some crest cells have been observed to penetrate into the cranial primary mesenchyme (Tan & Morriss-Kay, 1986), which showed strong CSPG staining. In fact, the highly polyanionic nature of CSPG might directly enhance migration: Sugimoto & Hagiwara (1979) found that the locomotory speeds of fibroblasts on differently charged substrata were directly proportional to the degree of negativity of the substrate.

Ruoslahti (1988) has proposed a model for CSPG inhibition of cell attachment to fibronectin, in which CSPG binding to the GAG-(heparin-) binding site on the fibronectin molecule has the effect of masking the RGD (Arg-Gly-Asp) cell-attachment site and/or interferes with the binding of membrane proteoglycans to the GAG-binding site. The functional relationship between fibronectin and the cytoskeleton is known to involve both the cell- and GAG-binding regions of the fibronectin molecule (Woods et al. 1986; Hôôk et al. 1988). The cranial mesenchyme of day 10 rat embryos is rich in fibronectin (Tuckett & Morriss-Kay, 1986); our present observations indicate that the fibronectin-rich cranial mesenchyme is also rich in CSPGs during the period of neural crest cell migration, suggesting that the fibronectin: CSPG ratio here is compatible with the formation of close contacts and with migratory behaviour. Strong fibronectin staining was also observed in the apposed basement membranes of the two pairs of epithelia discussed in (2) above (Tuckett & Morriss-Kay, 1986), providing further evidence of strong interepithelial adhesion at these sites prior to the appearance of CS-56 staining in normal embryos, and in conditions of decreased CSPG in chondroitinase-exposed embryos.

Interaction between CSPG and a substrate adhesion molecule cannot yet be invoked to explain the distribution of CSPG within the epithelium of the closed neural tube, in which neither fibronectin nor laminin have been detected at early stages (Rogers et al. 1986; Sternberg & Kimber, 1986).

This study presents evidence that CSPGs are morpho-genetically important components of the embryonic extracellular matrix of day 9 to day 11 rat embryos. We propose (1) that by decreasing the adhesiveness of fibronectin-containing substrates, CSPG facilitates crest cell emigration and migration, and sclerotomal cell migration; (2) that by decreasing adhesiveness within the neural epithelium, CSPG facilitates the migratory movements of neuroblasts, and may also provide increased flexibility during the generation of epithelial curvatures.

This study was supported by an MRC project grant. We thank Martin Barker for technical assistance, Colin Beesley for photographic assistance and Yonetaka Fukiishi for helpful comments on the manuscript.

Avnur
,
Z.
&
Geiger
,
B.
(
1984
).
Immunocytochemical localization of native chondroitin-sulfate in tissues and cultured cells using specific monoclonal antibody
.
Cell
38
,
811
822
.
Couchman
,
J. R.
,
Caterson
,
B.
,
Christner
,
J. E.
&
Baker
,
J. R.
(
1984
).
Chondroitin sulphate proteoglycans in extracellular matrices: sulphation specificity in relation to cellular differentiation
. In
Matrices and Cell Differentiation, pp. 31-46
.
New York
:
Alan R. Liss, Inc
.
Couchman
,
J. R.
&
Rees
,
D. A.
(
1979
).
The behaviour of fibroblasts migrating from chick heart expiants: changes in adhesion, locomotion and growth, and in the distribution of actomyosin and fibronectin
.
J. Cell Sci
.
39
,
149
65
.
Couchman
,
J. R.
,
Rees
,
D. A.
,
Green
,
M. R.
&
Smith
,
C. G.
(
1982
).
Fibronectin has a dual role in locomotion and anchorage of primary chick fibroblasts and can promote entry into the division cycle
.
J. Cell Biol
.
93
,
402
10
.
Dorfman
,
A.
(
1981
).
In Cell Biology of the Extracellular Matrix
(ed.
E. D.
Hay
), pp.
115
38
.
New York
:
Plenum
.
Hassell
,
J. R.
,
Kimura
,
J. H.
&
Hascall
,
V. C.
(
1986
).
Proteoglycan core protein families
.
A. Rev. Biochem
.
55
,
539
67
.
Hook
,
M.
,
Kjellén
,
L.
&
Johansson
,
S.
(
1984
).
Cell-surface glycosaminoglycans
.
A. Rev. Biochem
.
53
,
847
69
.
Hook
,
M.
,
Woods
,
A.
,
Lebaron
,
R. G.
,
Switalski
,
L. M.
&
Johansson
,
S.
(
1988
).
Molecular mechanisms of cell-substrate adhesion
. In
Self-Assembling Architecture
(ed.
J. E.
Varner
), pp.
119
127
.
New York
:
Alan R. Liss, Inc
.
Knox
,
P.
&
Wells
,
P.
(
1979
).
Cell adhesions and proteoglycans. 1. The effects of exogenous proteoglycans on the attachment of chick embryo fibroblasts to tissue culture plastic and collagen
.
J. Cell Sci
.
40
,
77
88
.
Langman
,
J.
,
Guerrant
,
R. L.
&
Freeman
,
B. G.
(
1966
).
Behavior of neuroepithelial cells during closure of the neural tube
.
J. Comp. Neurol
.
127
,
399
412
.
Laterra
,
J.
,
Ansbacher
,
R.
&
Culp
,
L. A.
(
1980
).
Glycosaminoglycans that bind cold-insoluble globulin in cell-substratum adhesion sites of murine fibroblasts
.
Proc. natn. Acad. Sci. U.S.A
.
77
,
6662
66
.
Morriss
,
G. M.
&
New
,
D. A. T.
(
1979
).
Effect of oxygen concentration on morphogenesis of cranial neural folds and neural crest in cultured rat embryos
.
J. Embryol. exp. Morph
.
54
,
17
35
.
Morriss-Kay
,
G. M.
(
1981
).
Growth and development of pattern in the cranial neural epithelium of rat embryos during neurulation
.
J. Embryol. exp. Morph
.
65
Suppl
.,
225
241
.
Morriss-Kay
,
G. M.
&
Crutch
,
B.
(
1982
).
Culture of rat embryos with /LD-xyloside: evidence of a role for proteoglycans in neurulation
.
J. Anat
.
134
,
491
506
.
Morriss-Kay
,
G. M.
&
Tuckett
,
F.
(
1985
).
The role of microfilaments in cranial neurulation in rat embryos: effects of short-term exposure to cytochalasin D
.
J. Embryol. exp. Morph
.
88
,
333
348
.
Morriss-Kay
,
G. M.
,
Tuckett
,
F.
&
Solursh
,
M.
(
1986
).
The effects of Streptomyces hyaluronidase on tissue organization and cell cycle time in rat embryos
.
J. Embryol. exp. Morph
.
98
,
59
70
.
Newgreen
,
D. F.
(
1982
).
Adhesion to extracellular materials by neural crest cells at the stage of initial migration
.
Cell Tissue Res
.
227
,
297
317
.
Newgreen
,
D. F.
&
Gibbins
,
I.
(
1982
).
Factors controlling the time of onset of the migration of neural crest cells in the fowl embryo
.
Cell Tissue Res
.
224
,
145
160
.
Perris
,
R.
&
Johansson
,
S.
(
1987
).
Amphibian neural crest cell migration on purified extracellular matrix components; a chondroitin sulphate proteoglycan inhibits locomotion on fibronectin substrates
.
J. Cell Biol
.
105
,
2511
2521
.
Ramon Y Cajal
,
S.
(
1929
).
Studies on Vertebrate Neurogenesis (Translation of the 1929 edition by L. Guth). Springfield, IL: Charles Thomas
.
Rich
,
A. M.
,
Pearlstein
,
E.
,
Weissman
,
O. G.
&
Hoffstein
,
S. T.
(
1981
).
Cartilage proteoglycans inhibit fibronectin-mediated adhesion
.
Nature, Lond
.
293
,
224
226
.
Roden
,
L.
(
1980
).
Structure and metabolism of connective tissue proteoglycans
. In
The Biochemistry of Glycoproteins and Proteoglycans
(ed.
W. J.
Lennarz
), pp.
267
371
.
New York
:
Plenum
.
Rogers
,
S. L.
,
Edson
,
K. J.
,
Letourneau
,
P. C.
&
Mcloon
,
S. C.
(
1986
).
Distribution of laminin in the developing peripheral nervous system of the chick
.
Devl Biol
.
113
,
429
435
.
Ruoslahti
,
E.
(
1988
).
Structure and Biology of Proteoglycans
.
In A. Rev. Cell Biol
.
4
,
229
255
.
Saito
,
H.
,
Yamagata
,
T.
&
Suzuki
,
S.
(
1968
).
Enzymatic methods for the determination of small quantities of isomeric chondroitin sulfates
.
J. biol. Chem
.
243
,
1536
1542
.
Solursh
,
M.
&
Morriss
,
G. M.
(
1977
).
Glcosaminoglycan synthesis in rat embryos during the formation of the primary mesenchyme and neural folds
.
Devl Biol
.
57
,
75
86
.
Sternberg
,
J.
&
Kimber
,
S. J.
(
1986
).
The relationship between emerging neural crest cells and basement membranes in the trunk of the mouse embryo: a TEM and immunocytochemical study
.
J. Embryol. exp. Morph
.
98
,
251
268
.
Sugimoto
,
Y.
&
Hagiwara
,
A.
(
1979
).
Cell locomotion on differently charged substrates
.
Expl Cell Res
.
120
,
245
252
.
Suzuki
,
S.
,
Saito
,
H.
,
Yamagata
,
T.
,
Anno
,
K.
,
Seno
,
N.
,
Kawai
,
Y.
&
Furuhashi
,
T.
(
1968
).
Formation of three types of disulfated disaccharides from chondroitin sulfates by chondroitinase digestion
.
J. biol. Chem
.
243
,
1543
1550
.
Tan
,
S. S.
,
Crossin
,
K. L.
,
Hoffman
,
S.
&
Edelman
,
G. M.
(
1987
).
Asymmetric expression in somites of cytotactin and its proteoglycan ligand is correlated with neural crest cell distribution
.
Proc. natn. Acad. Sci. U.S.A
.
84
,
7977
7981
.
Tan
,
S. S.
&
Morriss-Kay
,
G. M.
(
1985
).
The development and distribution of the cranial neural crest in the rat embryo
.
Cell Tissue Res
.
240
,
403
416
.
Tan
,
S. S.
&
Morriss-Kay
,
G. M.
(
1986
).
Analysis of cranial neural crest cell migration and early fates in postimplantation rat chimaeras
.
J. Embryol. exp. Morph
.
98
,
21
58
.
Thiery
,
J. P.
,
Duband
,
J-L.
,
Rutishauser
,
U.
&
Edelman
,
G. M.
(
1982
).
Cell adhesion molecules in early chicken embryogenesis
.
Proc, natn Acad. Sci. U.S.A
.
79
,
6737
6741
.
Tucker
,
G. C.
,
Duband
,
J-L.
,
Dufour
,
S.
&
Thiery
,
J. P.
(
1988
).
Cell-adhesion and substrate-adhesion molecules: their instructive roles in neural crest cell migration
.
Development
103
Supplement
,
82
94
.
Tuckett
,
F.
&
Morriss-Kay
,
G. M.
(
1986
).
The distribution of fibronectin, laminin and entactin in the neurulating rat embryo studied by indirect immunofluorescence
.
J. Embryol. exp. Morph
.
94
,
95
112
.
Tuckett
,
F.
&
Morriss-Kay
,
G. M.
(
1989
).
Heparitinase treatment of rat embryos during cranial neurulation
.
Anat. Embryol. (in press)
.
Woods
,
A.
,
Couchman
,
J. R.
,
Johansson
,
S.
&
Hook
,
M.
(
1986
).
Adhesion and cytoskeletal organization of fibroblasts in response to fibronectin fragments
.
EMBO J
.
5
,
665
70
.
Yamada
,
K. M.
(
1983
).
Cell surface interactions with extracellular materials
.
A. Rev. Biochem
.
52
,
761
99
.
Yamagata
,
T.
,
Saito
,
H.
,
Habuchi
,
O.
&
Suzuki
,
S.
(
1968
).
Purification and properties of bacterial chondroitinases and chondrosulfatases
.
J. biol. Chem
.
243
,
1523
1535
.