Anat Embryol (1992) 185:583 588

Anatomy and Embryology 9 Springer-Verlag1992

Cell death in cranial neural crest development Pete Jeffs, Karen Jaques, and Mark Osmond Department of Anatomy, University of Cambridge, Downing Street, Cambridge, CB2 3DY, UK Accepted January 17, 1992

Summary. The rhombencephalic neural crest, crucial to the patterning and development of m a n y craniofacial structures, migrates laterally f r o m the dorsal hindbrain, but not as a continuous sheet. We have used a vital dye to demonstrate a discontinuous pattern of cell death in the dorsal midline of the avian rhombencephalon associated with the migration of the neural crest. Whilst cell death commences in the dorsal midline of the presumptive mesencephalon at stage 8, two distinct domains of cell death are apparent in the rhombencephalon by stage 1 I. The rostral domain lies over primary rhombomere RhA1 and r h o m b o m e r e rh3, while the caudal domain occurs on the neural midline between the otic vesicles, in the region of rh5. Using a marker for the neural crest, we show that the rostral and caudal domains of cell death correlate with the absence of neural crest migration from rh3 and rh5. Thus segment-specific cell death in the dorsal region of particular r h o m b o m e r e s m a y account for their subsequent failure to contribute to the cranial neural crest.

Key words: Avian embryo - Hindbrain - Neural crest - Cell death - Cell migration pathways

Introduction Primitive segmentation of the vertebrate brain, once the subject of m u c h debate (Waters 1891), has recently aroused renewed interest (Lumsden and Keynes 1989; Lumsden 1990; Wilkinson 1990; Fraser et al. 1990). This is due in part to the characterisation of a series of developmentally important genes first isolated in Drosophila, whose vertebrate analogues have been shown to possess spatial and temporal expression patterns consistent with a role in the development of a segmented rhombencephalon (Holland et al. 1988; H u n t et al. 1991); see Fig. 1. Offprint requests to: P. Jeffs

Although the segmentation of the chick hindbrain into r h o m b o m e r e s occurs rapidly between stages 8 and 13 (Vaaga 1969), r h o m b o m e r e formation is not the outcome of a simple sequential segmentation process. Initially, the hindbrain is divided into three primitive and

gVII

gl>

(xl

cxl

r

-r

-r

-r

Fig. 1. The rhombencephalon, showing the relationship between the cranial nerve roots, their parent rhombomeres and the domains of Krox-20 expression, which in the mouse embryo is restricted to rhombomeres rh3 and rh5. The rostral expression boundaries of Hox 2.6, 2.7 and 2.8, which correlate with the boundaries between rhombomeres rhT/rh6, rh5/rh4, rh3/rh2 respectively (Hunt et al. 1991), are indicated by the top ends of the hatched bars. The neural crest, originating from the dorsal midline throughout the entire rhombencephalon (Anderson and Meier 1981), migrates into the mesoderm adjacent to all rhombomeres except for rh3 and rh5. The figure shows only the final arrangement of rhombomeres, and thus RhA1 referred to in the text (and which gives rise to rhl and rh2).has already disappeared. See introduction for explanation. The figure has been redrawn and modified from Lumsden and Keynes (1989) and Hunt et al. (1991)

584

transitory segments, RhA, RhB and RhC. These are then subdivided to give the rhombomeres, for example RhA is divided into RhA1 and rh3. RhA1 then gives rise to rhl and rh2. The cranial nerves (and their ganglia) derive their identity from their parental rhombomeres: cranial nerves V (trigeminal), VII (facial) and IX (glossopharyngeal) thus grow out from rhombomeres rh2, rh4 and rh6 respectively (Lumsden 1990). No nerve roots are, however, derived from rh3 or rh5. The cranial neural crest (CNC), migrating from the rhombencephalon, contributes not only to the development of these cranial ganglia (Noden 1984), but also to the anterior part of the chondrocranium (Hall 1987), connective tissues associated with facial musculature (Noden 1988), and to the patterning of these tissues (Noden 1988). Rhombomeres rh3 and rh5 are unique not only in their absence of nerve roots; each represents an expression domain of Krox-20 (Wilkinson 1990) and neither appears to contribute to neural crest-derived structures in mouse (Hunt et al. 1991), rat (Hall 1987), avian (Anderson and Meier 1981), or snapping turtle (Meier and Packard 1984) embryos, although the neural crest is made, in each case, at all levels of the rhombencephalon. Patterns of cell death in the trunk of the developing chick embryo appear to be associated with early development of the neural crest (Jeffs and Osmond 1992). We therefore decided to examine the head-region of the chick embryo at the stages associated with early development of the cranial neural crest, to determine whether localised patterns of cranial cell death might occur. In the trunk, cell death occurs along specific migration pathways. Here, we wished to ask whether more extensive regions of cranial cell death might correlate with the known absence of neural crest cells in the mesoderm adjacent to rhombomeres rh3 and rh5 during hindbrain development.

Materials and methods Egg preparation. Fertilised eggs of Leghorn Chickens from Winter Egg Farms (Thriplow, Fowlmere, Cambs.), were incubated in a Western humidified incubator at 38 ~ C, until the embryos had reached the required developmental stage (8 to 12). All stages refer to those of Hamburger and Hamilton (1951).

NBS staining. Nile Blue sulphate dye has been used frequently to visualise patterns of cell death during embryogenesis (Saunders et al. 1962; Jeffs and Osmond, 1992; see also Hinehliffe 1981 and Bowen 1981 for reviews). As a control for the staining of dead cells, we repeated the classic staining experiments performed on developing chick limb-buds (Saunders et al. 1962) and were able to visualise anterior, posterior and inter-digital necrotic zones (see Hinehliffe 1981) at the appropriate developmental stages (data not shown). Embryos were removed from the egg and placed in P&C saline (Pannett and Compton 1924). The amniotic membrane was removed with the aid of a Wild stereo-dissecting microscope and the embryo then transferred to a plastic dissecting dish containing 10 ml of P&C saline, with 30 gl of Nile Blue sulphate (NBS) (Sigma, Lot No. 35F-3496) made to a concentration of 1 mg/ml in double-distilled H20. Embryos were gently agitated on a Rotatest

R/100 (Luckham) shaker for 30 to 40 min at room temperature, after which they were removed from the NBS solution, and washed in 10 ml of P&C saline for approximately 1 h under the same conditions. Embryos were photographed as whole-mounts on a Leitz standard objective microscope using Kodak Vericolor ASA 160 colour negative film.

HNK-1 staining. Embryos to be stained with HNK-1 monoclonal antibody serum (a gift from Dr. Claudio Stern, Oxford), were fixed for a minimum of 48 h at 4 ~ C, in absolute ethanot, after NBS staining. Embryos were gently rehydrated in a graded series of alcohols, from 90% to 50%, changing solution each hour, then washed in PBS overnight at 4 ~ C. The next day, the embryos were given 5 x 10 min washes with PBS, followed by 3 h blocking with 1.0% BSA in PBS. Excess blocking solution was poured off and the embryos incubated in concentrated serum at 4 ~ C for 5 days. The embryos were washed in PBS (5 x 10 rain washes), and then incubated with 1 : 100 dilution of Sigma Goat anti-mouse g-chain fluorescein conjugate, in 0.3% BSA in PBS and photographed using Fuji Push Film 1600-D on a Leitz standard microscope.

Histology. Embryos to be analysed histologically were treated with NBS, as described above, then fixed in buffered formal saline (4% paraformaldehyde in PBS), dehydrated in a graded series of alcohols and cleared in cedar wood oil (BDH). For embedding, specimens were transferred from cedar wood oil to molten wax (Fibrowax) at 60~ C for 2 h, with one change of wax, then transferred to a small plastic mold containing molten wax which was subsequently allowed to set. The wax block was carefully trimmed around the tissue before cutting 8-gin sections with a Reichert manual microtome. Sections were mounted on glycerine-albumencoated glass slides, cleared with xylene, rehydrated and stained with Harris' haematoxylin and eosin. Finally, slides were quickly dehydrated in an ascending series of alcohols and xylene, then mounted with DPX mounting medium (Gurr).

Fig. 2a-h. Nile Blue sulphate (NBS) was used to visualise patterns of cell death during early chick hindbrain development (a-c). a Stage 8-9. Dying cells in the region of the neural crest of the mesencephalon (M), and RhA prior to extensive hindbrain segmentation. The exact rostral extent of the hindbrain (RhA) is difficult to determine at this stage (Vaaga 1969). b Stage 9-10. The region of cell death has moved caudally, extending from the mesencephalon to RhAI and rh3 (see text). The large arrow (right) indicates regions of the neuro-epithelium which also stain with NBS. c Stage 11. Two domains of cell death are visible in the rhombencephalon. The rostral domain (CDr) spans RhAx and rh3, and the caudal domain (CDc) spans rh5 (CD, and CDc are defined in Fig. 3 c). Rhombomere identities are clear from the position of the otic vesicle ( 0 7 ) which lies adjacent to rh5 and rh6 (not labelled). No cell death is apparent in rh4. d-f Cranial neural crest (CNC) migration visualised with monoclonal antibody H N K - I d Stage 8. H N K 1 binds to CNC cells in the mesoderm adjacent to the mesencephalon and the rostral half of RhAx. No mesodermal staining is seen caudal to RhA 1. e Stage 9-i 0. The rhombencephalic CNC is visible in the adjacent mesoderm. Signal is absent from the mesodermal regions adjacent to rh3 and rh5. This is seen more clearly as development proceeds, f Post stage 11. HNK-1 reactivity moves ventrally, but is still never observed adjacent to rh3 or rh5. Histology of stage 10 embryos shows condensed cell bodies in regions appropriate to a neural crest origin of cell death, g Condensed cell bodies in the neural crest (arrowed NC) and also in the neural tube (arrowed NT). (Compare to b, large arrow, where NBS staining in the neural epithelium is clearly seen), h Condensed cell fragments restricted to the neural crest in a region proximate to the otic vesicle

585

Fig. 2 a-h

586 Results

Nile Blue sulphate staining This was used, as described previously (Saunders et al. 1962; Bowen 1981 Jeffs and Osmond 1992), to visualise patterns of cell death in the developing chick hindbrain. The complex nature of hindbrain development (Vaaga 1969; Anderson and Meier 1981), is summarised in Fig. 3. The onset of cell death occurs prior to stage 9 (Hamburger and Hamilton 1951) of development, corresponding to 5 7 somites. At this time, cell death appears as a band of cells lying on the dorsal midline of the neural folds of the developing mesencephalon (Fig. 2 a). This single cluster of dying cells extends as far rostrally

a

stage 9 RP

b

stage 10

RhA, rh3 c

rh5 .OT

CO r

stage 11

CDc

Fig. 3a-e. Development of the avian rhombencephalon, neural crest migration and patterns of cell death. The neural crest is represented (on one side of the embryo alone) by hatched regions. Domains of dying cells are represented by stippling. Tissue blocks on either side of the neural tube represent somitomeres, the partially segmented head mesoderm. The figure is redrawn and modified from Anderson and Meier (1981). a Stage 9 (5-7 somites). The rostral plate (RP) neural crest moves into the mesoderm adjacent to the mesencephalon (m) and the cranial part of the rhombencephalon, RhA, the most rostral primary hindbrain segment. Cell death is restricted to the mesencephalon and the border with RhA. b Stage 10 (10-12 somites). By this stage, the rhombencephalic neural crest begins to migrate over the adjacent mesoderm, except in the region of rh3 and rh5. Cell death occurs caudal to the mesencephalon and extends into RhA1 and rh3, formerly RhA. c Stage 11 (15 somites). Cranial neural crest migration is essentially complete by this stage. The otic vesicle (07) is visible (shown only on one side) and is surrounded by the otic neural crest on the rostral side and the post-otic neural crest caudally. Cell death at this stage is tightly localised in the rhombencephalon to RhA1 and rh3 (rostral domain - CD,) and a caudal domain at rh5 (CDc). CDc and the caudal part of CDr correspond to regions which do not give rise to cranial neural crest

as the cranial half of the mesencephalon. The caudal extent is difficult to determine due to the difficulty in establishing the exact location of the mesencephalon/ rhombencephalon border at this stage (Vaaga 1969), but appears to overlap RhA, the most cranial presumptive rhombomere (see Fig. 3 a). By stage 10, (approximately 10-12 somites), two domains of cell death are seen: a rostral domain in the mesencephalon and a caudal domain in the region destined to form RhA1 and rhombomere rh3 (Fig. 2b, see Fig. 2g for histology). In some specimens examined these regions were contiguous. No cell death is apparent caudal to RhA1/rh3, until the somitic region is reached (see Jeffs and Osmond 1992), where cell death occurs in neural epithelium adjacent to fully formed somites. In stage 11 embryos, (14-16 somites), cell death in the mesencephalon is restricted to the most caudal part alone. As in stage 10 embryos, many dying cells are detected in the rhombencephalon spanning RhA1 and rh3 (Fig. 2c), the rostral necrotic zone, CDr. Cell death ceases abruptly in the region of rh4. Caudal to rh4, cell death is detected only in the dorsal midline tissue between the cranial part of the developing otic vesicles, which corresponds to rh5 (Fig. 2c, and Fig. 2h for histology), the caudal necrotic zone, CDc. At the latest stage examined (stage 12), cell death is no longer apparent in the dorsal midline. Instead, specific staining with Nile Blue sulphate occurs in the region of the developing cranial nerve routes, VII and IX, and in the caudal half of the otic vesicle, through which the neural crest is known to migrate (Anderson and Meier 1981 ; Meier and Packard 1984; Lim et al. 1987). These patterns indicate that by stage 11, cell death is localised to the regions o f r h 3 and rh5. Since the neural crest is thought not to migrate from these regions (Anderson and Meier 1981; Meier and Packard 1984; Hall 1987; Hunt et al. 1991); we used a whole-mount technique with the monoclonal antibody HNK-1 (Tucker et al. 1984; Rickmann et al. 1985; Bronner-Fraser 1986; Loring and Erickson 1987), a well-characterised marker for the neural crest, to establish whether a spatial and temporal relationship exists between cell death and neural crest migration in the rhombencephalon at these developmental stages.

HNK-1 staining Stage 9 (5-7 somites). At this stage, very strong expression of the H N K - I epitope is seen in the mesoderm surrounding the prosencephalon and mesencephalon (Fig. 2d). This is consistent with previous reports that the HNK-1 epitope in the trunk is restricted to cells which have penetrated the mesoderm (Newgreen et al. 1990). There is strong expression at the mesencephalon/ rhombencephalon border, although not in the neural epithelium itself. By stage 10 (10 12 somites), H N K - I binding appears in the rhombencephalic mesoderm (Fig. 2 e). The pattern is most striking at this stage because of the absence of binding in the mesoderm adjacent to rhombomeres rh3

587 and rh5. Strong binding is seen first at rh6 and then rh4 (data not shown). HNK-1 binding adjacent to rh6 is correlated with neural crest migration through the caudal part of the otic vesicle (Fig. 2f). Stage 11 and later (14 somites onwards). At this stage, the neural crest migrates ventrally beneath the ectoderm (Noden 1988) and, in the region rostral to the otic vesicle, migrates towards the visceral arches. HNK-1 labelling is seen on either side of the otic vesicle, in close association with the cranial ganglia and the paths of the facial (VII) and glossopharyngeal (IX) nerves (Fig. 2f).

Histology To test the prediction that cell death in the region of the rhombencephalon was of neural crest origin, we prepared serial sections of stage 10-11 embryos, which had previously been stained with NBS (see Jeffs and Osmond 1992). In these embryos, highly condensed cell bodies were seen in the dorsal lip of the neural folds in this region (Fig. 2g), and in the neural tube itself, (Fig. 2g, h). Darkly stained cells, correlating with the presence of NBS-stained whole-mount material were also seen in the neural tube and, in particular, in the region of RhA, although the significance of this is unknown (see Fig. 2 b). Discussion

We have used Nile Blue sulphate staining (Saunders et al. 1962; Hinchliffe 1981 ; for a detailed discussion of the technique see Bowen 1981, also Jeffs and Osmond 1992) to demonstrate a series of developmentally regulated patterns of cell death in the cranial regions of the chick embryo. Using HNK-1 as a marker for the paths of cranial neural crest migration, we have demonstrated a spatial and temporal correlation between the absence of neural crest cells and discrete zones of cell death in the cranial neural midline. Whilst studies of cell death in cranial structures exist (Martin-Partido et al. 1986), none of these studies (to our knowledge) has described discrete patterns of cell death in the rhombencephalon/mesencephalon during the stages of neural crest migration. Further, although it is well known that the CNC moves as a discontinuous sheet of cells over the cranial mesoderm (Anderson and Meier 1981 ; Meier and Packard 1984; Hall 1987; Hunt et al. 1991), no plausible explanation for this discontinuity, nor the mechanism of guidance, has so far been proposed. A tacit inference is that this discontinuity may result from non-permissive interactions with the cranial mesoderm (Anderson and Meier 1981). Our results provide the first evidence that cell death may contribute to the absence of neural crest cells adjacent to rhombomeres rh3 and rh5. If this is so, cell death may thus contribute to patterning the initial neural crest migration routes in the developing vertebrate head. One limitation of HNK-1 as a marker for cranial neural crest cells is that the epitope is only expressed once the cells have reached the cranial mesoderm. Since

it is not possible to stain CNC cells prior to migration from the neural folds, it is important to establish whether the CNC originates along the whole length of the cranial region, even though its subsequent migration is discontinuous. Anderson and Meier (1981) addressed this issue with SEM studies; our results from HNK-1 staining are in agreement with theirs. The onset of neural crest migration into the mesoderm adjacent to the mesencephalon and the cranial part of the rhombencephalon at stage 8 is also seen with the electron microscope. Whilst gaps in the pattern of neural crest migration in the mesoderm are seen adjacent to rh3 and rh5, neural crest cells are also seen to remain as a continuous sheet over the midline during these stages. Anderson and Meier (1981) point out that the midline neural crest cells at the level of rh3 and rh5 remain condensed and non-migratory at this stage. The continued presence of CNC cells on the midline at rh3 and rh5 is consistent with our hypothesis that it is cells of the neural crest which are dying. Kuratani (1991) has shown that by stage 15, rh3 and rh5 themselves stain with antibody directed towards the HNK-1 epitope, and this is consistent with the hypothesis that HNK-l-expressing cells do not migrate into the adjacent mesodem at these axial levels. Whilst the caudal region of cranial cell death (CDo) correlates precisely with the absence of neural crest cells in the adjacent mesoderm, the rostral region (CDr) is clearly wider than rh3. This additional cell death, as well as that occurring in the mesencephalon at earlier stages, may represent either a response to over-production of the CNC, or alternatively, may indicate an early selection of CNC precursors. Evidence from clonal culture of single quail cranial neural crest cells indicates that about 20% of these cells give rise to monopotent lines (Baroffio et al. 1991). If the results from in vitro analysis are applicable to the in vivo system, these data may indicate that early differentiation of some CNC cells does indeed take place. This might support a role for the early establishment of a system of selection amongst precursor cells. As stated above, the existence of CDr and CDc may contribute to the patterning of the early CNC migration. That the establishment of the proper migration routes is important has been underlined by recent studies on head development (Hunt et al. 1991). For example, specific Hox genes are thought to be expressed early in the development of the neuroepithelium and the CNC. This expression is maintained by the CNC throughout migration, resulting in the transfer of a Hox code to the cranial ganglia and branchial mesenchyme that reflects the CNC's rhombomeric origin. Early rhombomere-specific cell death signals may thus contribute to early pattern formation in the cranial neural crest. This idea is further supported by the expression patterns of Hox 7 class genes (Hill et al. 1989; Tabin 1991) which correlate with regions of cell death in the interdigital necrotic zones of the mouse limb. If, for example, Hox 7.1 is involved in specifying the location of developmental cell death, it is interesting to note that it is also expressed in the neural crest of the rhombencephalon (Hill et al. 1989).

588

Acknowledgements. We thank Professor Ruth Bellairs and Drs. Rob White, Tom Weaver and Nicki Winston for their criticisms and discussion; Dr. Claudio Stern for the gift of the HNK-1 antibody, and advice on whole-mount staining procedures; John Bashford for photographic expertise; and Marie Wilkins and Jill King for help with histology. PSJ was supported by "Action Research for the Crippled Child" Post-Doctoral Fellowship S/P/1945, KJ by an MRC studentship and MKO by an AFRC grant.

References Anderson CB, Meier S (1981) The influence of the metameric pattern in the mesoderm on migration of cranial neural crest cells in the chick embryo. Dev Biol 85:385-402 Baroffio A, Dupin E, Le Douarin N (1991) Common precursors for neural and mesectodermal derivatives in the cephalic neural crest. Development 112: 301-305 Bowen ID (1981) Techniques for demonstrating cell death. In: Bowen ID, Lockshin RA (eds) Cell death in biology and pathology. Chapman & Hall, London New York, pp 379444 Bronner-Fraser ME (1986) Analysis of the early stages of the trunk neural crest migration in avian embryos using monoclonal antibody HNK-I. Dev Biol 115:44-55 Fraser S, Keynes R, Lumsden A (1990) Segmentation in the chick embryo hindbrain is defined by cell lineage restrictions. Nature 334:431-435 Hall BK (1987) Tissue interactions in head development and evolution. In: Maderson PA (ed) Developmental and evolutionary aspects of the neural crest. Wiley, New York, pp 215-259 Hamburger V, Hamilton HL (1951) A series of normal stages in the development of the chick embryo. J Morphol 88: 49-82 Hill RE, Jones PF, Rees AR, Sime CM, Justice M J, Copeland NG, Jenkins NA, Graham E, Davidson DR (1989) A new family of mouse homeo-box containing genes: molecular structure, chromosomal location and developmental expression of Hox7.1. Genes Dev 3:26-37 Hinchliffe JR (1981) Cell death in Embryogenesis. In: Bowen ID, Lockshin RA (eds) Cell death in biology and pathology. Chapman & Hall, London New York, pp 35-78 Holland PWH, Hogan BLM (1988) Expression of Homeo box genes during mouse development: a review. Genes Dev 2: 773782 Hunt P, Whiting J, Muchamore I, Marshall H, Krumlauf R (1991) Homeobox genes and models for patterning the hindbrain and branchial arches. Development [Suppl] 1 : 187-196 Jeffs PS, Osmond MK (1992) A segmented pattern of cell death during development of the chick embryo. Anat Embryo1 185 : 589-598

Kuratani SC (1991) Alternate expression of the HNK-1 epitope in rhombomeres of the chick embryo. Dev Biol 144:215-219 Lim TM, Lunn ER, Keynes RJ, Stern CD (1987) The differing effects of occipital and trunk somites on neural development in the chick embryo. Development 100:525-533 Loring JF, Erickson CA (1987) Neural crest migratory pathways in the trunk of the chick embryo. Dev Biol 121:220-236 Lumsden A (1990) The development and significance of hindbrain segmentation. Sem Dev Biol 1 : 117-125 Lumsden A, Keynes R (1989) Segmental patterns of neuronal development in the chick hindbrain. Nature 337:424-428 Martin-Partido G, Alvarez IS, Rodriguez-Gallardo L, Navasscu6s J (i 986) Differential staining of dead and dying embryonic cells with a simple new technique. J Microsc 142:101-106 Meier S, Packard DS (1984) Morphogenesis of the cranial segments and distribution of neural crest in the embryos of the snapping turtle, Chelydra serpentina. Dev Biol 102: 309-323 Newgreen DF, Powell ME, Moser B (1990) Spatiotemporal changes in HNK=I/L2 glycoconjugates on avian embryo somite and neural crest cells. Dev Biol 139:100-120 Noden A (1984) Craniofacial development: new views on old problems. Anat Rec 208 : 1-13 Noden A (1988) Interactions and fates of avian craniofacial mesenchyme. Development [Suppl] 107:121-140 Pannett CA, Compton A (1924) The cultivation of tissues in saline embryonic juice. Lancet 1 : 381-384 Rickmann M, Fawcett JW, Keynes RJ (1985) The migration of neural crest cells and the growth of motor axons through the rostral half of the chick somite. J Embryol Exp Morphol 90: 437455 Saunders JW, Gasseling MT, Saunders LC (1962) Cellular death in morphogenesis of the avian wing. Dev Biol 5:147-178 Tabin CJ (1991) Retinoids, homeoboxes and growth factors:toward molecular models for limb development. Cell 66:199-217 Tucker GC, Aoyama H, Lipinski M, Turz T, Thiery JP (1984) Identical reactivity of monoclonal antibody HNK-1 and NC-I: conservation in vertebrates on cells derived from neural primordium and on some leucocytes. Cell Differ 14:223-230 Vaaga S (1969) The segmentation of the primitive neural tube in chick embryos (Gallus domesticus). In: Brodal A, Hild W, Ortmann R, Schiebler TH, T6ndury G, Wolff E (eds) Advances in anatomy, embryology and cell biology. Springer, Berlin Heidelberg New York, pp 8-57 Waters BH (1891) Primitive segmentation of the vertebrate brain. Q J Microsc Sci 33:457476 Wilkinson DG (1990) Segmental gene expression in the developing mouse hindbrain. Sere Dev Biol 1:127-134