Carol Irving/ M. Angela N ieto / Romita DasGupta/ Patrick Charnay/ and David G. Wilkinson*'^
* Division of Developmental Neuiobiology, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 lAA, United Kingdom; Unstituto Cajal, CSIC, Doctor Arce 37, 28002 Madrid, Spain; and t Biologie Moléculaire du Développement, INSERM U368, Ecole Normale Supérieure, 46666 Rue d ’Ulm, 75230 Paris, Cedex 05, France
After segmentation of the vertebrate hindbrain, expression of the zinc-finger gene Krox-20 and the receptor tyrosine kinase gene Sek-1 is precisely restricted to rhombomeres (r) 3 and 5. This precise segmental expression is likely to reflect a critical requirement for these rhombomeres to acquire a distinct and homogeneous identity and raises the question as to how this relates to the intermingling and restriction of cell movement during segmentation. We have analysed Krox-20 and Sek- 1 expression in the mouse and chick hindbrain at single-cell resolution using whole-mount in situ hybridisation and immunocytochemistry. We find that, in the mouse, the presumptive r3 and r5 expression domains each arise as narrow stripes that then broaden, suggestive of a recruitment of cells to an r3/r5 identity and/or a segmental regulation of cell proliferation. In addition, we find that expression of these genes initially occurs in fuzzy domains, and that these are progressively restricted to segmental domains, although occasional "violating" cells are observed even after segmentation. We propose that the establishment and maintenance of these segmental domains may involve both a dynamic regulation
of r3/r5 identity and the restriction of cell movement across rhombomere boundaries. © i996 Academic Press, inc.
INTRODUCTION
Studies of cell differentiation and lineage have provided compelling evidence that the vertebrate hindbrain is pat terned through a process of metameric segmentation. This segmentation is manifested by the formation of rhombom eres, a series of bulges demarcated by morphologically dis
tinct boundaries (Lumsden and Keynes, 1989; Heyman et
al, 1993), which in the chick form in an order that is invari
ant, but not in a simple rostrocaudal sequence (Vaage, 1969). Clonal analysis reveals extensive intermingling and dis persal of cells in the hindbrain, and prior to segmentation the progeny of individual cells often contribute to two
rhombomeres (r) (Fraser et al, 1990). In contrast, most
clonal progeny marked after boundary formation are re
stricted to individual rhombomeres (Fraser et al, 1990), but
even at this stage some clones straddle boundaries (Birg bauer and Fraser, 1994), indicating that the spatial restric tion of cell lineage is not absolute. This partial compart-
* To whom correspondence should be addressed. Fax: 0181-906- 4477. E-mail: [email protected].
mentalisation may contribute to the maintenance of territo ries with distinct identities, as later reflected in segment- specific patterns of neuronal differentiation (Lumsden and Keynes, 1989; Clarke and Lumsden, 1993; Gilland and Baker, 1993). The formation of boundaries and the restric tion of cell movement may involve the specification of re gions with alternating cellular properties. Grafting experi ments reveal that a boundary does not form when r3 and r5 are juxtaposed, and these cell populations mix with each other relatively freely (Guthrie and Lumsden, 1991; Guthrie
et al, 1993). In contrast, boundaries always form between
either r3 or r5 and any of r2, r4, or r6, and there is little intermingling between these odd- and even-numbered rhombomeres.
Clues regarding the genetic basis of hindbrain patterning have come from studies of genes with rhombomeric expres sion, including those encoding transcription factors such as the zinc finger protein Krox-20 and members of the Hox family and several receptor tyrosine kinase (RTK) genes of
the Ephfamily (reviewed by Hunt and Krumlauf, 1992; Wil
kinson, 1993). Intriguingly, expression of the Krox-20and
the Sek-1RTK genes is up-regulated in presumptive r3 and
r5, alternating domains that correlate with cellular proper-
26
0012-1606/96 $12.00 Copyright © 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.
Restriction of Sek-1 and Krox-20 27
ties that underlie subdivision of the hindbrain (Guthrie and Lumsden, 1991; Guthrie et a l, 1993). Indeed, in mice with null mutations in the Krox-20 gene, definitive r3 and r5 fail to form and morphological segmentation is disrupted (Schneider-Maunoury et a l, 1993; Swiatek and Gridley, 1993). Less is known regarding the role of Sek-1, although recent work showing that ligands for several Eph-related RTKs are membrane-hound (Bartley et a l, 1994; Beckmann
et a l, 1994; Cheng and Flanagan, 1994; Davis et a l, 1994)
suggests that it may mediate cell contact-dependent signal ling. Early in mouse hindbrain development Sek-1 is ex pressed in a broad domain, and later it is up-regulated in rhombomeres 3 and 5 at a time similar to the Krox-20 gene (Nieto et a l, 1992). In addition, Sek-1 is expressed in a complex pattern in other tissues, including segmental ex pression in presumptive somites.
After segmentation, Krox-20 and Sek-1 gene expression is precisely restricted to r3 and r5, and this restriction may reflect a critical requirement for each rhombomere to have a distinct and homogeneous identity. This raises the ques tion as to how the spatial restriction of segmental gene expression is established and maintained during the inter mingling of cells prior to and after segmentation. Towards addressing this question, we have analysed expression of
the Sek-1 and Krox-20 genes in the hindbrain at single-cell
resolution. Our findings provide new insights into two as pects of hindbrain segmentation. First, we find that in the mouse the presumptive r3/r5 expression of these genes is initiated in narrow domains that then broaden, with a shift in the relative number of expressing cells corresponding to pre-r3 compared with pre-r5. Thus, although morphological
segmentation subdivides r2-r6 into compartments of roughly equal size, at least in this system this is preceded by a dynamic establishment of presumptive r3 and r5 terri tories. Second, prior to segmentation the Krox-20 and Sek- 1 expression domains are fuzzy and then become progres sively sharpened as definitive r3 and r5 form. We discuss mechanisms that might underlie the establishment and sharpening of these r3/r5 gene expression domains in the hindbrain.
MATERIALS AND METHODS
Production and Purification of A ntibodies
Mouse Sek-1 polypeptide sequences were expressed as fusion proteins with glutathione transferase by suhcloning fragments of Sek-1 cDNA into the pGEX expression vector (Smith and Johnson, 1988). Two fusions were made: the C-terminal part of the extracellular domain {Bglll-EcoKV
fragment, residues 1293-1553) for the raising of anti-Seks antibody and the C-terminus of the intracellular domain (EcoRI-NcoI fragment, residues 2783-3195) for the raising of anti-Sek] antibody. Large scale preparations of these fu sion proteins were prepared and purified on glutathione- agarose columns as described (Smith and Johnson, 1988). Polyclonal antibodies were raised in rabbits by injection of 500 irg fusion protein at 28-day intervals until an anti-Sek- 1 immune reaction was detected.
The anti-Seki Ah was purified by affinity chromatography using an Affi-Gel 10 affinity support (Bio-Rad) to which
^200 ^97 "6 9 "4 6 " 3 0 " 9 7 " 6 9 " 4 6
FIG. 1. Detection of Sek-1 by Western blot analysis and immunoprécipitation, (a) Embryo and brain extracts were Western-blotted and probed with affinity-purified anti-Sek, antibody, raised against the C-terminus of Sek-1. (Lane 1) 10.5-day mouse embryo; (lane 2) adult mouse brain; (lane 3) stage 18 chick embryo; (lane 4) adult Xenopus brain. A polypeptide of 107 kDa is detected, as indicated by the arrowhead, (b) Whole extracts and immunoprécipitations of extracts with anti-Sek, or anti-Sekg antibodies were analysed by Western blotting and probing w ith anti-Sek, antibody. The anti-Sek^ antibody was raised against an extracellular domain of Sek-1. (Lane 1) Anti-Sek, immunoprecipitate of mouse brain extract; (lane 2) anti-Sek^ immunoprecipitate of mouse embryo extract; (lane 3) anti-Sekg immunoprecipitate of mouse brain extract; (lane 4) preimmune serum immunoprecipitate of mouse brain extract; (lane 5) mouse brain extract. Anti-Sekg antibody immunoprecipitates the 107-kDa polypeptide detected by anti-Sek, antibody (arrowhead). In addition, im muno globulin (Ig) is revealed by the secondary detection reagent. The migration of size markers are indicated to the right of each panel.
28 Irving et a l
m
<1
FIG. 2. Whole-mount detection of Sek-1 in mouse and chick embryos. Whole-mount in situ hybridisation of embryos with Sek-1 probe or immunocytochemical staining w ith anti-Sek, antibody was carried out. (a) In situ hybridisation to detect Sek-1 RNA in 9.5-day mouse embryo, (b) Immunocytochemical detection of Sek-1 in 9.5-day mouse embryo, (c) Immunocytochemical detection of Sek-1 in 18 somite chick embryo, (d-f) Higher power views, corresponding to (a-c), of expression in forming somites. Rostral is to the top in (c-f), but due to curvature of the mouse embryo is to the bottom for the tail region in (a,b). r, rhombomeres; o, otic vesicle; m, early mesoderm; ps, presumptive somites; s, recently formed somite; n, notochord.
FIG. 3. Sek-1 expression in the mouse hindbrain. Whole-mount staining of mouse embryos to detect Sek-1 protein (a-e) or RNA (f-k) was carried out. Flat mounts of the hindbrain (a-e, g-i), a lateral view of a whole mount (f), and transverse sections (j,k) are shown. Flat mounting was carried out such that the dorsal/lateral edge (DL) is lateral, and the ventral/medial midline (VM) is medial. Rostral (R) is to the top and caudal (C) to the bottom in (a-i), and dorsal (D) is to the top and ventral (V) to the bottom in (j,k). The stages of the embryos are: (a) 0 somites (7.75 day); (b) 4 somites; (c) 6 somites,- (d) 8 somites; (e) 12 somites,- (f,g) 0 somites (7.75 day); (h) 5 somites; (i) 20 somites, showing only the right side of the hindbrain,- (j) 20 somites, transverse section through r3; (k) 20 somites, transverse section through r5. Arrowheads indicate boundaries of Sek-1 expression in the hindbrain, p, presumptive rhombomere; r, rhombomere,- m, early mesoderm; ps, presumptive somite.
Restiiction of Sek-1 and Kiox-20 29
30 Irving et al.
3 mg purified fusion protein was bound according to the manufacturer's protocol. Serum from rabbits im m unised w ith the fusion protein was diluted w ith an equal volum e of PBS and purified by passing it through the colum n three times, followed by washing and elution w ith acid buffer. Eluted fractions were neutralised and those containing im m unoglobulin were pooled.
Immunoprécipitation of Sek Protein from Cell Lysates
D issected tissues were wrapped in foil in a m inim um volum e of PBS, frozen in liquid N2, and then ground to a
powder. The powder was dissolved in RIPA buffer con taining protease inhibitors. One hundred micrograms of to tal protein was diluted to 100 fA in RIPA buffer and was incubated w ith 4 lA of anti-Sek Ah at 4°C for 3 hr. Protein G-Sepharose beads (Pharmacia) in RIPA buffer were added and shaken for 2 hr at 4°C. The beads were washed three tim es in RIPA buffer, three tim es in 20 mM T ris-H C l, pH 7.5, and diluted in 50 fA sample buffer. An aliquot was boiled for 5 m in and 20 /xl was loaded on a SDS-PAGE gel. Immunoprecipitated proteins were detected by Western blot analysis.
Western Blot Analysis
After SDS-polyacrylam ide gel electrophoresis, protein samples were transferred to a nitrocellulose membrane by w et electrophoretic transfer as described (Harlow and Lane, 1988). The membrane was blocked in 10% sheep serum in PBTw (PBS, 0.1% Tween-20) for 1 hr and then incubated in 1/3000 diluted primary antibody in 10% sheep serum for 3 hr at 4*^0. The membrane was washed in PBTw four tim es for 10 m in at room temperature w ith rocking. The m em brane was then incubated in 1/10^ diluted secondary anti body (goat anti-rabbit-HRP; Sigma) in 10% sheep serum for 3 hr at 4°C and then washed in PBTw four tim es for 10 m in w ith rocking. The bound antibody was detected using ECL chem ilum inescence (Amersham), according to the manu facturer's instructions, and exposed using ECL hyperfilm (Amersham).
Wbole-Mount Immunocytochemistry
Embryos were dissected in ice-cold PBS and fixed for be tween 3 hr and overnight at 4°C. Of a range of fixatives tested, 2% TCA proved to give optimal results after staining w ith anti-Sek antibody. Fixed embryos were washed three tim es in PBT for 10 m in and then incubated in 0.05% hydro gen peroxide for 8 hr at 4°C. The embryos were washed in PBT three tim es for 10 m in, followed by blocking in 10% sheep serum in PBT for 1 hr. They were then incubated in 1/1000 diluted primary anti-Sek; antibody in 10% sheep serum, PBT overnight at 4°C, w ith rocking. After washing five tim es for 1 hr, the embryos were incubated in 1/500 diluted secondary Ab (goat anti-rabbit-AP; Promega) in 10%
sheep serum, PBT overnight at 4°C. The embryos were washed in TBT five tim es for 1 hr at RT, and then detection was carried out using BCIP/NBT as substrate. To observe gene expression at a single-cell resolution, the hindhrain neural plate or neural tube was dissected away from other tissues, if required a cut was made through its dorsal edge, and the neural epithelium was flattened out lik e a kipper and m ounted under a coverslip.
Whole-Mount in Situ Hybridisation
In situ hybridisation w ith digoxigenin-labelled probes and
detection w ith AP-conjugated antibody were carried out as described (Wilkinson and N ieto, 1993).
RESULTS
Detection of Sek-1 Protein Expression
In previous work Sek-1 expression was studied by in situ
hybridisation w ith radioactive probes against tissue sec tions, but this does not allow analysis of Sek-1 expression at single-cell resolution. To enable analysis of Sek-1 protein expression at a single-cell resolution w e have raised poly clonal antibodies against a fusion protein of the C-terminus of the intracellular domain of m ouse Sek-1; this region was selected because of its substantial amino acid sequence di vergence (<43% identity) relative to other described m em bers of the Eph-related family, but is w ell conserved (82 and 96% identity, respectively) between Sek-1 hom ologues in
Xenopus (Xu et a l, 1995; W inning and Sargent, 1994) and
chick (Cek-8; Sajjadi and Pasquale, 1993) compared to the mouse. The specificity of this reagent, designated anti-Selq antibody, is indicated by the Western blot and im m unocyto chem ical staining patterns described below.
In Western blot analyses the anti-Selq antibody detects a 107-kDa polypeptide, the predicted size of Sek-1 protein (Gilardi-Hebenstreit et a l, 1992), in extracts of 10.5-day m ouse embryos (Fig.la, lane 1) and adult m ouse brain (lane 2). This antibody also cross-reacts against a 107-kDa protein present in stage 18 chick embryos (Fig. la, lane 3) and adult
Xenopus brain (lane 4). To confirm that anti-Sek; antibody
is detecting Sek-1 protein w e used an antibody, anti-Sek:, raised against an extracellular domain of Sek-1, w hich also detects a 107-kDa polypeptide in embryo and brain extracts (not shown). Anti-Sek; antibody was used to probe a W est ern blot of m ouse embryo and brain extracts immunoprecip itated w ith anti-Sek; antibody (Fig. lb, lane 1 ), anti-Sek: anti body (lanes 2 and 3) or preimm une serum (lane 4). Both antibodies immunoprecipitated a polypeptide that comi- grates w ith the protein detected in total brain extracts (Fig.lb, lane 5), indicating that they detect Sek-1.
Whole-mount im m unocytochem ical staining of m ouse embryos w ith anti-Sek; antibody revealed a pattern very similar to that of Sek-1 mRNA: expression of R N A and protein is detected in the early mesoderm, notochord, pre-