The morphogenesis of rhombomeres (neuromeres) caudal to the preotic sulcus during neurulation in rat embryos is described. A model is proposed to explain the development of the characteristic neuromeric sulci and interneuromeric gyri based on the cytoskeletal elements and the kinetic behaviour of the neural epithelium.
Evidence obtained from a study of control, cytochalasin D-treated and colchicine-treated embryos, at the electron microscopic level, supports the proposed model. The longitudinally expanding cranial neural epithelium bulges between microtubule blocks present within the interneuromeric gyri, causing a bulge to develop along the line of least resistance, away from the microfilament-rich luminal border of the neuromeric sulcus region.
Neuromeres have been observed as a segmental arrangement of sulci and gyri within the early neural tube of all vertebrate embryos. Their characteristic morphology is illustrated in Fig. 1. This study addresses the question of how the sulci and gyri develop from an initially straight neural epithelium, and how their structure is maintained.
Observations on a variety of vertebrates (chick, fish, urodeles) led Kallen (1956) to propose that the formation of neuromeric sulci was the result of mitotic patterning within the neural epithelium. His model is based on transverse sections of the neural epithelium during the formation of a neuromere, which he termed a ‘proliferation centre’. According to Kallen, the furrowing of the luminal border results from the redistribution of neuroepithelial cells, due to the mitotic patterning, while the bulging at the basal border is brought about by the formation of proportionally more cells in the centre of the region, where the mitotic activity is highest. This interpretation relies strictly on the kinetics of the cells within the neuromere and their movement across the epithelium between the basal and luminal borders, and takes no account of the possible migration of cells within the neural epithelium (along the length of the embryo), or of possible regulation of shape through a periodicity of structural components.
We have suggested that a localized increase in the mitotic index is the initial means of identifying a neuromere, and once the neuromere is expressed morphologically as a surface sulcus the mitotic index returns to its basal level (Tuckett, 1984; Tuckett, Lim & Morriss-Kay, 1985). Kallen’s ‘proliferation centre’ appears to refer to the localized concentrations of mitotic activity which we have observed in both transverse and coronal section.
This study investigates the possibility of an alternative hypothesis for the development of neuromeres which embraces both the mitotic patterning (orientation of the mitotic spindle axes) and the cytoskeletal components of the neural epithelium. This hypothesis is illustrated diagrammatically in Fig. 2. It proposes that growth of the neural tube is generated longitudinally by cell division in this plane, but that elongation is prevented by the fixed nature of the tube within the embryo. In consequence, the lengthening epithelium bulges outwards into the adjacent mesenchyme since the basal border (lacking in microfilaments) is more easily deformable than the microfilament-bound luminal border. The repeating pattern of sulci and gyri results from the presence of a series of blocks to deformation (gyri) between the bulging regions (sulci). This hypothesis is based on evidence from a previous study (Tuckett & Morriss-Kay, 1985), as follows: (i) cell division within the neural epithelium at relevant stages is predominantly (98%) orientated with the mitotic spindle axis parallel to the long axis of the embryo, so as to increase the length of the neural epithelium; (ii) rostral to the preotic sulcus cells generated in the midbrain and upper hindbrain appear to flow rostrally to provide an extrinsic source of cells for the rapidly expanding forebrain, suggesting the existence of a block to cell movement at or close to the preotic sulcus. Since we are now suggesting a series of blocks to cell movement between neuromeric sulci, the present hypothesis relates only to the clearly segmented caudal metencephalon and myelencephalon, i.e. the region caudal to the preotic sulcus.
To examine the feasibility of this model for the development of neuromere morphology, and to investigate the morphogenetic nature of the proposed blocks to cell movement between neuromeres, transmission electron microscopy was employed. The function of microtubules and microfilaments in maintaining the morphology was examined by treating embryos in vitro with colchicine which binds to tubulin and prevents microtubule assembly, and cytochalasin D which inhibits microfilament assembly; each may also disrupt existing microtubules and microfilaments respectively.
MATERIALS AND METHODS
Wistar strain rat embryos were explanted in Tyrode’s saline on day 10 of pregnancy (day of positive vaginal smear = day 0). A total of 39 embryos was used. In the 34 embryos which were subsequently to be cultured only Reichert’s membrane was opened; in the remaining 5 embryos the extraembryonic membranes were removed before fixation. At the time of fixation the embryos had between 12 and 16 somite pairs.
Whole embryo culture
Embryos were cultured at a temperature of 38 °C in 60 ml Pyrex glass bottles containing 2 ·5 ml immediately centrifuged, heat-inactivated rat serum (Steele & New, 1974), 2 ·5 ml Tyrode’s saline and 10 μl penicillin/streptomycin (5000 i.u. ml-1 and 5000 μg ml-1). The bottles were gassed with 5 % CO2:5 % O2:90 % N2 (New, Coppola & Cockroft, 1976a,b) prior to sealing. The bottles were continuously rotated at 30r.p.m. At the end of the culture period, the embryos were thoroughly washed in Tyrode’s saline and the extraembryonic membranes were removed before fixation.
Cytochalasin D-treated embryos
A stock solution of 1 ·0 mg ml-1 cytochalasin D (Sigma, London) was prepared in a 10% solution of dimethylsulphoxide (DMSO; Sigma, London) and stored at –20°C. 7 ·5 μl of the stock solution was added to the culture medium (final concentration of 0 ·15 μg ml-1). Ten embryos at the 11- to 13-somite stage were cultured in the presence of cytochalasin D for 2h. Five control embryos were cultured in the presence of 7 ·5 μl of a 10 % solution of DMSO for 2 h.
14 embryos at the 11- to 13-somite stage were treated in culture for 3 h with colchicine (Sigma, London) which had been added to the culture medium so as to produce a final concentration of 0 ·2 μgml-1. This concentration was found to cause a sufficiently high mitotic arrest without adversely affecting the gross morphology of the embryo; the duration of colchicine treatment was less than half the cell-cycle time which we have determined previously (Tuckett & Morriss-Kay, 1985), and thus only a small number of cells were observed to be arrested at the metaphase stage. Subsequent treatment was the same as for the untreated, control embryos. Five embryos were cultured in unsupplemented culture medium for 3 h.
After removal of the membranes, embryos were fixed in 2 ·5 % cacodylate-buffered osmium tetroxide, washed and dehydrated through graded alcohols before embedding in Spurr resin. Thin sections of 80 –90 nm thickness were collected on copper grids and double stained with uranyl acetate at 40°C for 45 min and lead citrate for 5 min at room temperature.
Semi thin sections of 1 μm thickness were cut immediately adjacent to the thin sections. These sections were collected on glass slides and stained with either equal parts of 1 % methylene blue and 1 % azure II in 1 % borax, or 1 % toluidine blue in 1 % borax.
No differences were observed between the control cultured and the noncultured embryos, and thus the subsequent use of the term ‘control’ will apply to either of these two states.
In control embryos the neural epithelium displayed a typical pseudostratified epithelial arrangement. All cells had luminal border attachments and those cells not in mitosis also had basal border attachments. The characteristic morphology of the neuromere was clearly seen at the light microscopic level (Fig. 3A). Within the sulcus the cells had narrow necks and overlapping apices which were more easily visible at the electron microscopic level (Fig. 4A); the narrow cell apices were associated with the presence of apical microfilament bundles which were aligned parallel with the cell surface between the intercellular junctions (Fig. 4B). At the interneuromeric junction region (gyrus), the luminal border had a somewhat different morphology. The cell apices were broader than in the sulcus region, although some overlapping did still occur (Fig. 5A); microfilament bundles were not continuous along the luminal border but were associated laterally with the intercellular junctions (Fig. 5B). At the basal border of the outward bulging neural epithelium, the cell processes formed a broad attachment with the basement membrane (Fig. 4C) whereas at the interneuromeric junction, the basal attachments of the inward bulging neuroepithelial cells were narrow (Fig. 5C). At high power these narrow basal cell processes were shown to contain microtubules (Fig. 6A,B) which were aligned perpendicular to the luminal border. The microtubules had a diameter of 25 nm and their orientation within the gyrus region as a whole displayed a characteristic fan shape, which was reflected in the arrangement and orientation of cells within the gyrus. The fan-shaped cellular arrangement can be seen at the light microscopic level in Fig. 3A. The basement membrane and mesenchyme formed a short angled cleft between adjacent neuromeres (Fig. 5C).
Cytochalasin D-treated embryos
The shape and organization of the epithelium of cytochalasin D-treated embryos differed from that of the control embryos as follows. There was a marked thinning of the epithelium which was noticeable at the light microscopic level (Fig. 3B); this was probably associated with the change in cellular morphology which was more noticeable at the electron microscopic level (Fig. 7A,B). The cells all had broad apices which protruded into the lumen; there was no difference in the apical appearance of sulcus and gyrus cells. The sulci and gyri were more gently curved, with the inward and outward bulges of a similar size; there was no sharply angled basal cleft formation at the inward bulge.
Within the outward bulging epithelium (sulcus) microtubule groups were present perpendicular to the cell surface, though not in a clearly organized fanshaped orientation. The microtubule groups were more diffuse in nature, and could not be resolved at the lowest magnification at which they are discernible in control embryos. Fig. 7C illustrates the diffuse nature of the microtubule groups following cytochalasin treatment, and may be compared with Fig. 6A,B which is taken at the same magnification and illustrates the more ordered arrangement of microtubules in control embryos. Assuming that their position has not been affected by the cytochalasin treatment, it would seem that the original sulci (Fig. 3A) have everted and that they now appear as gyri (Fig. 3B); and similarly the microtubule groups originally present within the gyri (Fig. 3A) are now present within the sulcus (Fig. 3B).
Two other features which characterized the cytochalasin D-treated embryos were the increased number of cytoplasmic vacuoles and intercellular spaces compared with the control embryos.
At the light microscopic level a number of cells was seen to be arrested at the metaphase stage of mitosis, and those cells which were not arrested were characteristically beginning to round-up (Fig. 3C). These changes in cellular morphology resulted in narrowing of the neural epithelium but not to the same extent as following cytochalasin D treatment. Although the cells were more rounded than their control counterparts there was a narrowing of cell apices within the sulci, suggesting that microfilaments were contracting along the luminal border at the centre of a neuromere (Fig. 8A). The apical border of the gyrus cells (Fig. 8B) was arranged in a similar manner as in the control embryos (Fig. 5A). The most pronounced effect of colchicine treatment on neuromere structure was the lack of microtubule groups between adjacent neuromeres; the fan-shaped gyrus cells lost their elongate form although the unarrested cells maintained their apical and basal contacts. As a result of the loss of cell elongation, the basal cleft between adjacent neuromeres deepened significantly (Figs 3C and 8C; compare with Figs 3A and 5C of the clefts of a control embryo).
As with the cytochalasin D-treated embryos, there were quite large intercellular spaces but these were more likely to be associated with the rounding-up of the cells and concomitant loss of junctional contacts rather than a direct effect on intercellular junctions per se.
The electron micrographs of the control embryos illustrated here show that in the coronal plane there is an alternating pattern of cells with characteristic cytoskeletal components, correlated with the pattern of sulci and gyri of the neuromeres. In the sulci, the apical regions of the cells show narrow necks with overlapping apices, and a continuous line of microfilament bundles in association with intercellular junctions, parallel to the epithelial surface. This line becomes discontinuous in the interneuromeric (gyrus) regions, where fan-shaped groups of cells form the angle between adjacent sulci. These cells have long narrow basal regions and broaden towards the apical surface, and are rich in microtubules orientated perpendicular to the surface (i.e. along the length of the cells).
Microtubules and microfilaments have long been recognized as playing an essential role in cell shape, cell movement, and cytoplasmic organization. The results of the colchicine and cytochalasin D experiments reported here suggests that they play important roles in maintaining the shape of neuromeres. Reversal of neuromere shape from a series of sulci between the microtubule-rich interneuromeric regions to a series of gyri was observed in embryos cultured in medium containing cytochalasin D. This experiment clearly demonstrates the role of apical microfilament bundles in maintaining both the shape of the sulcus cells with their narrow neck regions and overlapping apices, and the shape of the sulcus region as a whole. Curvature generated and maintained by the microfilament bundles is expressed as an outward bulging of the neural tube into the adjacent mesenchyme, which must enable this to occur through its own deformability.
Microfilament bundles orientated parallel to the neuroepithelial surface play an essential role during neurulation, either by generating curvature, as interpreted by Baker & Schroeder (1971) or by thickening and contracting the neural plate, as interpreted by Jacobson & Gordon (1976). They are present in the cranial neural epithelium of rat embryos at the time of epithelial curvature (Morriss & New, 1979), and their disappearance from the cytoplasm as a result of cytochalasin treatment corresponds to the loss of neuroepithelial curvature (Morriss-Kay & Tuckett, 1985).
In the presence of colchicine, the fan-shaped groups of gyrus cells lost their elongate form but maintained their apical contacts, causing the normally shallow indentations of the basal neuroepithelial surface to deepen considerably. The cells lost their polarized shape, becoming rounded. Microtubules have been shown to play an essential role in the establishment and maintenance of cell polarity and cell shape in a variety of cell types including the amphibian egg after fertilization (Kirschner, Gerhart, Hara & Ubbels, 1980; Gerhart et al. 1981) and in the extension of neurites and maintenance of neurite structure (Seeds, Gilman, Amano & Nirenberg, 1970; Kirschner, 1982; Fulton, 1984). The present experiments suggest that while microtubules are not essential for the maintenance of the periodic neuromere structure, they provide a series of stiffened annular regions on which the adjacent microfilament-rich sulci can contract; these annular regions also form a morphological break between the cells of adjacent sulci, and the fan shape of the cell groups provides the necessary compensatory curvature between sulci.
The mechanism underlying the generation and maintenance of neuromeric periodicity is not understood. It is likely to involve the adjacent primary mesenchyme, since this is involved in regionalization of the brain at the time of neural induction (Spemann, 1938, in amphibian embryos). The cranial mesenchyme is itself organized into a segmental pattern of somitomeres, before and during neuromeric development (Anderson & Meier, 1981; Meier & Tam, 1982). Bernfield, Banerjee, Koda & Rapraeger (1984) have demonstrated that during glandular morphogenesis, change in shape of the epithelium is governed by the adjacent mesenchyme, via molecular remodelling of the basement membrane. We have been unable to find evidence for any longitudinal periodicity of glycosaminoglycans in the neuromeric basement membrane or subsequent mesenchyme (Tuckett, 1984); the possibility that other extracellular matrix components may show some neuromere-related periodicity is currently under investigation.
Reflecting their segmental pattern, neuromeres have a close correlation with the pattern of subsequent development of cranial motor nerve nuclei, in general in a one-to-one relationship (Adelmann, 1925; Bartelmez & Dekaban, 1962; Vaage, 1969). This observation, taken together with the ultrastructural features reported here, suggests that they may represent clonal compartments. Jacobson (1983), in a study of cell lineage in late blastula-stage frog embryos, found groups of blastomeres whose descendants contributed to specific compartments within the brain and spinal cord which were determined at, but not before, the 512-cell stage. In the rhombencephalon there were boundaries along the dorsal and ventral midlines, a division into dorsal and ventral compartments, and a transverse boundary separating the rhombencephalon from the mesencephalon (but not from the spinal cord). Within compartments there was much mixing of cells. Assuming that similar compartments are also determined in mammalian presumptive neural plate cells, it would seem likely that mixing within compartments becomes further restricted as progressively finer aspects of pattern are determined, and that neuromeres themselves represent clonal compartments (each divided into four quadrants as defined above). The fan-shaped cell groups of the gyrus regions would, according to this hypothesis, represent intercompartmental boundaries, across which mixing cannot occur.
The experiments reported here provide evidence in support of the hypothesis for neuromere development and maintenance as proposed in the introduction. Outward curvature of the neuromeric sulci depends on the presence of a line of apical microfilament bundles, in the absence of which the direction of curvature is reversed. The gyrus regions contain cells rich in microtubules which are required to maintain their fan-shaped organization as viewed in coronal sections. We have no information on the mechanism underlying the periodic nature of the pattern, but suggest that neuromeres represent semi-autonomous clonal compartments, and that the gyrus cells provide a physical barrier to cell movement between compartments as well as a skeletal framework for the support of the contracting sulcus epithelium.
Thanks are due to Mr Martin Barker for technical assistance, and to Mr Tony Barclay for photographic assistance. This study was supported by the Medical Research Council and the University of Oxford.