Keywordr: Sox genes; SRY-box; Neuronal development; Chick. 1. Introduction ..... and Sox-1 and thus define these Sax proteins as all being members of a ...
Mechanisms of Development 49 (1995) 23-36
ELSEVIER
Embryonic expression of the chicken Sox2, Sox3 and Soxll suggests an interactive role in neuronal development
genes
Dafe Uwanoghoa, Maria Rexb, Elizabeth J. Cartwrighta, Gina Pearlb, Chris Healy”, Paul J. Scottingb, Paul T. Sharpe*a aDepartment of Craniofacial Development. UMDS, Guy’s Hospital, London Bridge, London SE1 9RT. UK bDepartment of Biochemislry. Queen’s Medical School, University of Nottingham, Clifton Boulevard, Nottingham NG7 2UH, UK
Received21 February 1994;revision received 26 July 1994; accepted 26 August 1994
Abstract
Three chicken Sax (SRY-like box) genes have been identified that show an interactive pattern of expression in the developing embryonic nervous system. &ox2 and &ox3 code for related proteins and both are predominantly expressed in the immature neural epithelium of the entire CNS of HH stage 10 to 34 embryos. cSoxl1 is related to cSox2 and cSox3 only by virtue of containing an SRY-like HMG-box sequence but shows extensive homology with Sox-4 at its C-terminus. cSoxl1 is expressed in the neural epithelium but is transiently upregulated in maturing neurons after they leave the neural epithelium. These patterns of expression suggest that Sox genes play a role in neural development and that the developmental programme from immature to mature neurons may involve switching of Sox gene expression. cSoxl1 also exhibits a lineage restricted pattern of expression in the peripheral nervous system. Keywordr:
Sox
genes; SRY-box; Neuronal development; Chick
1. Introduction Our understanding of the control of embryonic processes at the level of gene transcription has increased dramatically following the discovery of large families of developmentally-regulated transcription factors containing highly conserved DNA binding motifs. The highly conserved HOX/HOM complexes and their role in positional specification in development, together with the many other homeobox genes, zinc-linger encoding genes, etc. that have putative roles in many different embryonic processes has led to the realisation that members of these large families can have related and interactive roles in development. The discovery of the sex determining gene SRY and its murine counterpart, Sry, has led to the identification of a number of related autosomal genes with similar DNA-binding motifs. These SRY-like Box (Sax) genes appear to constitute a large family of developmentally l
Correspondingauthor.
Elsevier Science Ireland Ltd. SSDI 09254773(94)00299-3
regulated genes capable of encoding putative transcription factors (Gubbay et al., 1990; Ner, 1992; Coriat et al., 1993; Laudet et al., 1993). The DNA-binding SRYlike HMG-box motif found both in the Sry and Sox genes, is related to the HMG-box originally described in the high mobility group proteins that are non-sequence specific DNA binding proteins, and in the RNA polymerase I transcription factor UBF (Jantzen et al., 1990). The HMG-box motifs found in Sox gene products show sequence-specific DNA binding although all appear to recognise the same consensus binding sequence in vitro, A/TA/TCAAAG, which is recognised at high affinity by many different HMG-box proteins (Sinclair et al., 1990; Nasrin et al., 1991; Oosterwegal et al., 1991a,b; Travis et al., 1991; van de Wetering et al., 1991; Waterman et al., 1991; Denny et al., 1992; Harley et al., 1992). Sequence-specific HMG-box proteins appear to bind in the minor groove of DNA and induce a strong bend in the helix (Ferrari et al., 1992; Giese et al., 1992). Expression of a number of HMG-box genes shows them to be expressed in a wide range of embryonic and adult cells.
D. Uwanogho et al. /Mechanisms of Development 49 ( 1995) 23-36
24
TCFa and LEFI are expressed in embryos and in specitic adult cells; TCFa in T cells and LEFI in lymphocytes (Travis et al., 1991; van de Wetering et al., 1991; Waterman et al., 1991). SOX-5is expressed only in adult testis (Denny et al., 1992) whereas Sox-4 is a transcriptional activator in lymphocytes (van de Wetering et al., 1993). Sox genes appear to be highly evolutionarily conserved since genes with similar HMG-box related sequences have been identified in Drosophila, reptiles, birds and mammals (Coriat et al., 1993; Laudet et al., 1993). Furthermore within the family as a whole there are identifiably distinct sub-classes of sequence, both in the HMG-box and in other regions of the protein. We describe here the cloning and expression analysis of three Sox genes isolated from an embryonic chicken cDNA library. Two of these genes, cSox2 and cSox3 are related by sequence and expression pattern whereas the third, cSoxf 1 appears to be a member of a different sub-class
1
61 121 191 241
60 120 100 240 300 14
that is related to Soxl. cSox2 and cSox3 are both expressed in undifferentiated cells of the neural epithelium in the CNS. cSoxl1 shows a related expression pattern, with a low level of expression in undifferentiated proliferating cells of the neural epithelium and an increased staining in neurons after they leave the neural epitheliurn. The decrease in &ox2 and cSox3 expression and transient increase in cSoxl1 expression correlates with the switch from proliferating to differentiating cells such that the change in Sox gene expression may play a role in commitment to a differentiated phenotype. These reciprocal expression patterns suggest that these Sox gene products may act as switches in neuronal development .
60 120 160 240 18 241 19
360 58 361 59
360 34 361 35
300 38
420 78 480 98
420 54 480 74
600 138
540 94 600 114
661 IS9
941 219
960 234 961 235 1021 255 1081 275 1141 295
1140 294
314
1201 315
1260
1261 1321
1320 1378
Fig. I. Nucleotide and conceptual amino acid sequence of cSox2. The boxed region is the HMG-box sequence. Underlined are the three regions conserved between cSox3 and murine Sax-1, Sex-2 and SOX-3. Nucleotide numbering starts from the end of the cDNA clone.
1081 299
1140 312
1141 1201 1321 1381 1441 1501 1561 1621 1681 1741 1801
1200 1320 1380 1440 1500 1560 ,620 1680 if40 lSO0 1860
Fig 2. Nucleotide and conceptual amino acid sequence of cSox3. The boxed region is the HMG-box sequence. Underlined are the three regions conserved between cSox2 and murine Sox-Z, Sox-2 and Sax-3. Also underlined is the putative polyadenylation signal sequence. Nucleotide numbering starts from the end of the cDNA clone.
D. Uwanogho et al. /Mechanisms of Development 49 (1995) 23-36
2. Results
60 120 180 240 300 360 420 19
2.1. Sequence Five cDNAs isolated from a mixed HH stage 14 and 17 (Hamburger and Hamilton, 1951) chick embryo library following probing with a pool of PCR generated HMG-box sequences were restriction mapped, sequenced in full and compared with published-sequences of murine HMG-box sequences and the full sequences of Sox-I to Sox-5 genes (R. Lovell-Badge, pers. commun.). Two of these were found to be overlapping clones of the same gene that was identified as the chick homologue of mouse Sox-2, cSox2. Two other clones also overlapped with each other and were derived from the chick homologue of mouse Sox-3, &0x3. Amino acid homology between cSox2 and Sox-2 is 81% and between cSox3 and Sox-3 is 68%. The fifth cDNA had an HMG-box amino acid sequence identical to that of a murine PCR product named Sox-I1 and this was provisionally named cSoxl1 until identity can be confirmed (Wright et al., 1993). The overlapping cSox2 cDNA clones overall consist of a 1.38 kb insert which was found to contain a single open reading frame of 315 amino acids with an HMGbox of 79 amino acids; 32 amino acids from the Nterminus. Translation is assumed to start from the first inframe methionine following an inframe stop since this is conserved between both chick and mouse Sox-2 and Sox-3 genes. However this start does not have a good consensus start sequence whereas the second inframe ATG (three amino acids downstream) has an excellent start consensus sequence match of CAACAUG (Cavener, 1987) (Fig. 1). This clone is assumed to be incomplete at the 3 ’ end since no polyadenylation sequence or signal is detectable. The cSox3 cDNA clones overall have a 1.86 kb insert that contains a single open reading frame of 312 amino acids. The 79 amino acid HMG-box is located 42 amino acids from the first inframe methionine which is assumed to be the translation start by virtue of the conservation described above for cSox2. cSox3 also has a good start consensus sequence of AGCAAUG around the second ATG present three amino acids further downstream (Fig. 2). cSox2 and cSox3 have HMG-boxes sharing 93% amino acid homology and outside this DNA binding region they also share three other highly conserved sequences (Fig. 4A). At the N-terminus both share a 14 amino acid sequence from the methionine start with only four of the first 14 amino acids being different with a consensus sequence of MYXMXETEXKXPXP (Fig. 4B). Fifty amino acids downstream of the HMG-box in cSox2 and forty-six amino acids downstream of the HMG-box in cSox3 there is a 41-44 ammo acid stretch that is highly conserved between the two proteins (Fig.
25
600 79
781 140 841 160 901 180 961 200
220
1080 239
1081 240
1140 259
10.21
1141 260 1201 280 1261 300 1321 320
1381 34* 1441 360 1501 380 1561 1621 1681 1741 ISOl 1861 1921
1620 1680 1740 1800 1860 1920
Fig. 3. Nucleotide and conceptual amino acid sequence of cSoxl1. The boxed region is the HMG-box sequence. Solid underlined are the two regions of homology with Sex-4. Dashed underlined is the acidic region and dot underlined is the proline-rich region. The arrow marks the position of the sequence 5’ of which is derived from genomic DNA. Nucleotides are arbitarily numbered from the 5’ extent of the genomic DNA sequenced. - - - putative acidic activation region; -.-.- proline-rich region.
4C, 4D). This region consists of an imperfect leucine/methionine repeat with the whole region having properties of a hydrophobic structure reminiscent of leucine zippers. Twenty amino acids further downstream in cSox2 and twenty-five in cSox3 is
D. Uwanogho et al. /Mechanisms of Developmenl 49 11995) 23-36
26
Fig. 4. (A) Primary structure of the cSoxl1 and Sol-4 proteins and the cSo.x_7and cSox3 proteins highlighting the areas of homology. (B) Comparison of the N-terminal sequences of chick and mouse Sox proteins. (C) Alignment of the leucine/methionine repeat regions of chick and mouse Sox proteins. (D) Homology between regions toward the C-terminus of chick and mouse So.u-213proteins.
another highly conserved 39-42 amino acid region of the proteins which is particularly rich in serine, 26% in cSox2 and 20% in cSox3. The cSoxl1 cDNA contains a 1.5 kb insert that contains a single translatable sequence of 396 amino acids, but since the reading frame continues to the 5’ extent of the clone without any inframe stop codons, genomic sequence 5 ’ to the end of the cDNA was used to identify the putative translation start (Fig. 3). The presumed methionine start was therefore identified from genomic sequence as the first inframe codon. The predicted start sequence of AGCGAUG does not however show a good match with the vertebrate consensus sequence of CANCAUG (Cavener, 1987). The 79 amino acid HMG-box is 46 amino acids from the N-terminus and is 64% homologous with cSox2 and 63% with &0x3. The cDNA is also incomplete at the 3 ’ end since no poly-A tail was detected and although a putative polyadenylation signal sequence of AATTA is located 15 bp from the end of the cDNA the 2.4 kb size of the transcript detected on Northern blots (section 2.2.) would suggest that the clone has at least 500 bp missing from either or both of the 5’ and 3 ’ ends. The cSoxl1 sequence does not share any common regions outside the HMG-box with either cSox2 or cSox3. cSoxl1 does however show two regions that are also found in Sox-4 (van de Wetering et al., 1993) namely the first four amino acids at the
cSoxl1 sox-4
translation start, MVQQ and forty-seven amino acids at the C-terminus which show 68% identity (Figs. 4A, 5A). There is a highly acidic region between 204-231 where twenty out of twenty-eight amino acids are acidic (71%). This acidic sequence shows homology with the activation region of the HMGl and 2 proteins (Bustin et al., 1990). Close to the C-terminus of cSoxZ1 there is a short proline-rich sequence (286-300) which is similar to sequences found in the rat DBP transcriptional activator (Mueller et al., 1990) Drosophila brahma (Tamkun et al., 1992) and murine Hoxb4 (Graham et al., 1988) proteins (Fig. 5B), although no obvious function can be ascribed to this region. 2.2. Temporal expression Northern blots of poly-A+ RNA from embryos and adult tissues were probed in turn with each cDNA insert of cSox2, cSox3 and cSoxl1 (Fig. 6). A single transcript was detected for cSox2 of 1.8 kb and cSox3 of 1.8 kb and two transcripts of 2.4 and 3.4 kb for cSoxl1. The transcript size for cSox3 is close to the size of the cDNA suggesting that the cDNA is almost full length. The origin of the two cSoxl1 transcripts is at present not known and, although the entire 1.5 kb cDNA sequence is found in a single exon (unpublished), we cannot at present rule out differential splicing of alternative 5 ’
VDElX,DSFSEGSLGSEFEE’PDYCTPELSEMfAFSDLVFTY LDRDLDFNFEPGSGSEFEPPDYCTPEVSEMISGDWLESSISNLVE’TY Fig. 5. (A) The C-terminal sequences of cSoxfl and SOX-4.
D. Uwanogho et al. /Mechanisms of Development 49 (1995) 23-36
3.4 kb3.4 kb1.8 kb1.8 kb-
3.4 kb2.4 kb-
Fig. with lane lane adult
6. Northern blots of embryonic and adult tissue RNA’s probed CSOXclones. Lane I, Stage 24; lane 2, Stage 25; lane 3. Stage 28; 4, Stage 31; lane 5, Stage 32; lane 6. Stage 36; lane 7, Stage 37; 8. adult liver; lane 9, adult kidney; lane 10, adult heart: lane 11, skeletal muscle. A, cSox2; B. cSos3; C, cSoxl1.
exons. The sequence of the putative cSoxl1 protein is thus based on a single exon sequence and would most closely match the size of the 2.4 kb transcript. All three genes were expressed in embryos at HH stages 24, 25, 28, 31, 32, 36 and 37. All three genes show similar temporal expression in embryos where expression is detectable at stages 24-3 1, when it begins to decline. By stage 37 no cSox3 expression can be detected while cSox2 and cSoxll are expressed at low levels. By stage 39 no expression of these two genes can be seen (not shown). No expression was detected in the adult tissues tested, namely; liver, brain, spleen, gonad, skeletal and cardiac muscle. In embryos all three genes showed maximum expression between stages 24-31. After stage 3 1 the expression of all three genes falls. At all stages cSox2 and cSox3 show similar levels of expression which are higher than cSoxl1.
Fig. 7. Expression of &0x2, cSox3 and cSoxl1 in whole chick embryos (HH stage 17). Whole mount in situs were performed using digoxigenin-labelled antisense riboprobes. cSox2 is expressed in the CNS and all regions of the brain (A). &ox3 shows a very similar expression pattern to cSox2, being strongly expressed in the CNS, brain and pharyngeal clefts (B). cSoxl1 is expressed in the entire embryo (0
21
28
D. Uwanogho er al. /Mechanisms of Development 49
csox 2
A I
’
.
.
( 1995) 23-36
csox 11
csox 3 B
c
D. Uwanogho
et al. /Mechanisms
2.3. Spatial embryonic expression Whole mounts Expression in embryos was investigated in whole mount in situ hybridisation and on sections using digoxigenin-labelled RNA probes. On whole mounts of HH stage lo-17 embryos expression of all three genes looked similar (although cSoxl2 staining was relatively high in many tissues), with expression in a band along the length of most of the CNS (staining was not always visible in the most posterior regions of the CNS) and in all compartments of the brain (Fig. 7). A number of other regions were also positive including the pharyngeal clefts and the nasal placodes (Figs. 7A, 7B, 7C). Expression was investigated in more detail on tissue sections. Tissue sections cSox2 and &ox3 transcripts were first detected in the neural plate shortly before it closed to form the neural tube. By HH stage 16 cSox2, cSox3 and cSoxl1 were seen at similar levels in neuronal tissue although cSox1 I was seen at equally high levels throughout most embryonic tissues until HH stage 19 (most notably in mesoderma1 structures, Fig. 8C). From HH stage 16 onwards all three genes were seen to be expressed throughout the length of the CNS (except for the most caudal regions in younger embryos) including all compartments of the brain. In general, cSox2 and cSox3 were detected in the proliferative neural epithelium whilst cSox1 I was detected more strongly in more mature neurons. The level of general staining over other tissues was higher with cSoxl1 than with cSox2 and cSox3 especially at early stages (a high level of cSoxl1 in the surface epidermis appears to be the main cause of the apparently less specific staining seen in whole mounts). cSox2 and cSox3 were seen only at low levels in tissues other than the CNS and gut (data not presented). cSoxl1 was relatively highly expressed in the peripheral nervous system and in the gut epithelium. Expression in the developing neural tube. cSox2, cSox3 and cSoxl1 transcripts were all detected in the early neural tube (Figs. 8A, 8B, 8C). cSox2 and cSox3 staining was specific to the neural tube while cSoxl1 stained most other embryonic tissues with equal if not greater intensity than the neural tube (especially mesodermal structures). As development progressed cSox2 and
of Development
49 (1995)
23-36
29
cSox3 transcripts were detected only in the neural epithelium and later in the ependymal layer (Figs. 8D-L). cSox2 and cSox3 expression decreased as development proceeded (10 days, HH stage 36 onwards) and was absent by 14 days (data not presented) which correlates with decreasing proliferation from 8 days onwards (Hamburger, 1948). Early in the differentiation of the neural tube (HH 17) cells immediately lateral to the neural epithelium in the subventricular zone (inner mantle layer) began to exhibit much higher levels of cSoxl1 transcripts (Fig. 81). This staining was initially most pronounced ventrally in the expanding basal plate (HH17-HH24) but became more pronounced dorsally in the expanding alar plate between stages HH26 and HH34 (8 days) (Fig. 8L). Once the inner mantle layer had disappeared in the basal region, the level of expression throughout the basal plate never increased above background levels. Later still (HH stage 34-36, 8-10 days) cSoxl1 expression was seen predominantly in a broad region across the dorsal l/3 of the neural tube in the dorsal horn region where sensory neurons develop (Fig. 80). The level of staining for cSoxl1 in the alar plate decreased after 8- 10 days of development. This progression from basal to alar expression of cSoxl1 followed by a general decrease in staining of the neural tube, correlates with the relative stage of maturation of neurons in those regions. In order to further analyse the relationship between cSoxl1 expression and the maturation of neurons in the neural tube, we carried out double labelling of sections for cSox2 or cSoxl1 expression and (using immunohistochemistry) neurofilament protein, a marker of relatively mature neurons (Figs. 9J, 9K). This showed that there is a group of cells in the subventricular zone in which neither cSox2 transcripts or neurofilament are detected, and it is these same cells which express elevated levels of cSoxl1 (this is particularly well illustrated by comparison of Fig. 8L with Figs. 9J and 9K). A further feature of the expression pattern of cSoxl1 in the neural tube was its absence from the floor plate by HH 17 whilst cSox2 and cSox3 transcripts were clearly detected in this region of the neural tube (Figs. 8D-I). Expression in the developing brain. By HH stage 13 the entire brain is clearly distinguishable and the major compartments well established. In embryos of this stage the entire neuronal layer was positive for transcripts of all three genes. The pattern of expression varied in dif-
Fig. 8. Expression of cSox2, cSox3 and cSoxlf on transverse wax sections of chick neural tube at the level of the developing hind limb (except panels M, N and 0 at the level of the forelimb). All panels show results of in situ hybridization using digoxigenin-labelled antisense riboprobes with blue colour development of alkaline phosphatase and no counterstain. (Sense probes gave no detectable signal for all clones). Probe used is shown at the top of each column. Panels A, B and C from HH stage 13, panels D. E and F from HH stage 15, panels G, H and 1 from HH stage 21, panels J, K and L from HH stage 26 and panels M, N and 0 from HH stage 36. d, dorsal root ganglion; n, neural tube; s, sympathetic ganglion; sv, subventricular zone; v, ventricular zone. Scale bars correspond to 100 pm.
30
D. Uwanogho
et al. /Mechanisms
of Developmenr
49 (1995)
23-36
Fig. 9. Expression of cSox2, cSox3 and ~50x11 on wax sections from chick embryos. All panels show results of in situ hybridization using digoxigenin-labelled antisense riboprobes with blue colour development of alkaline phosphatase and no counterstain. (Sense probes gave no detectable signal for all clones). Panels A, B and C show transverse sections through the hindbrain region of HH stage 28. Panels D, E and F show sagittal sections through the base of the midbrain from HH stage 28. Panels G, H and I show sagittal sections through the forebrain of HH stage 28. Panels J and K show transverse sections through the wing region of a HH stage 26 neural tube double stained with monoclonal antibodies specific for nemofilament (see experimental procedures for details), arrow heads indicate subventricular zone negative for eSux2 and neurofilament, but positive for cSox-il. Panel L shows a transverse section through the hindgut region of an HH stage 36 embryo. Panel M (dark field optics to show all tissues) and N show transverse sections through the adrenal gland and gut of a HH stage 36 embryo. Panel 0 is a diagram showing mid-sagittal section through HH stage 28 head. The line through the hindbrain shows the plane of section seen in panels A-C. Numbers in top right hand corner of panels describe the probe used: 2, CSOXZ;3, CSOXS;11, CSOXII; d, dorsal root ganglion; e, endoderm lining of gut; ep, enteric plexus; fb, forebrain; g, gut; hb, hindbraln; I, lumen; m, non-neuronal mesodenn; mb, midbrain; ml, mantle layer; rp, roof plate; v, ventricular zone. Scale bars correspond to 100 pm.
ferent compartments as the brain developed, as described in detail below. l Hindbrain. By HH stage 27 additional discrete layers of cells were apparent in both midbrain and hindbrain outside of the neural epithelium. At this stage cSox2 and &ox3 expression was restricted to the ventricular zone (Fig. 9A, 9B). Staining for cSoxl1 was present in the ventricular zone but stronger in the adjacent more peripheral layer of cells (Fig. 8L). This staining pattern is consistent with that seen in the more caudal
neural tube. As the neural tube opens out into the hindbrain the staining pattern seen in the neural tube is retained. Thus the base of the hindbrain which is formed by this opening and flattening of the neural plate exhibited cSox2 and cSox3 expression in the ventricular epithelial cells, below which was a layer of cells strongly positive for cSoxZ1, with the most basal layer of cells being negative for cSox2 and cSox3 and only weakly positive for cSoxl1 (analogous to the most lateral cells of the neural tube).
D. Uwanogho
et al. /Mechanisms
The roof of the hind brain is formed from expansion of the ventricular neural epithelium but did not exhibit significant staining with any of the probes (Figs. 85,8K, 8L), in contrast to the epithelium of the neural tube and base of the hindbrain. l Midbrain. The pattern of expression seen in the midbrain was similar to that seen in the base of the hindbrain. The base of the midbrain again exhibited a pattern of Sox gene expression consistent with that seen in the neural tube. cSox2 and cSox3 expression was seen in the ventricular epithelial cells, below which was a layer of cells strongly positive for cSoxZ1, with the most basal layer of cells being negative for cSox2 and cSox3 and only weakly positive for cSoxl1. The roof of the midbrain was initially moderately positive for all probes. cSoxll expression subsequently increased in the alar region of the midbrain by HH stage 24 as a stronger staining single cell layer. The fibrous layer formed immediately outside of this cSoxlZ positive layer by HH stage 27. This cSoxl1 layer was continuous with the thicker layer at the base of the midbrain corresponding to the staining seen in the base of the hindbrain and the cSoxl1 positive cells of the neural tube. As the mid brain developed through HH stages 24-34 cSox2 and cSox3 remained expressed in cells of the ventricular zone but appeared absent from all cells outside of this (Figs. 9D, 9E) whilst cSoxl1 was strongly expressed in most cells outside of the ventricular zone (Fig. 9F) and weakly expressed in the ventricular zone. l Forebrain. Following general distribution of all three cSox transcripts at HH stage 23 of development their expression in the forebrain by HH stage 24 was similar to that seen in the basal regions of the hindbrain and midbrain. cSox2 and cSox3 remained expressed in the ventricular zone whilst cSoxl1 expression increased dramatically outside of this layer, although expression in the outermost layers was not so strong (Fig. 9M). Expression outside of the central nervous system. In general cSox2 and cSox3 were expressed only at low levels in tissues outside of the central nervous system. Ex-
Fig. IO. Boxed regions
illustrate
areas shown in panels D-l.
oj Development
49 (1995)
23-36
31
pression was however detected in a few locations most notably in the retina and gut epithelium (data not presented). cSoxll was much more evident in other locations especially in the gut epithelium (Fig. 9L) and peripheral nervous system (Figs. 80 and 9L). In the gut all three cSox genes were expressed in the epithelial lining of the digestive tract. High levels of cSoxli transcripts were detected in ganglia of the sympathetic chain (Fig. 80) and the enteric plexus (Fig. 9L) but were not detected at such elevated levels in the sensory dorsal root ganglia (Figs. 8L, 80). cSoxl1 was also detected in cells of the adrenal medulla at elevated levels comparable to that seen in the sympathetic ganglia (Fig. 9M). All three cSox transcripts were also detected at above background levels in certain other ectodermal tissues such as the nasal placode and otic vesicle, in regions of neuronal development (unpublished observations). 3. Discussion We have isolated and characterised three developmentally-regulated chicken Sox genes that show an interactive expression pattern in the developing nervous system. All three genes code for proteins with HMG-box motifs at their N-termini and have a serinejthreonine rich region toward their C termini. The genes have been named according to homology with published murine and human sequences. Thus cSox2 is highly homologous to murine Sox-2 (R. Lovell-Badge, pers. commun.) and cSox3 to human, mouse and marsupial Sox-3 sequences. However the overall amino acid identities between cSox3 and other Sox-3 genes is reduced as a result of cSox3 lacking the poly glycine and alanine stretches found in the mammalian sequences (Stevanovic et al., 1993, Foster and Graves, 1994). 5 ’ of the HMG-box cSox3 is 49% identical with Sox3, when aligned to exclude the poly glycines and is approximately 85% identical in the rest of the protein with Sox3 when the poly alanine regions are excluded. The HMG-box sequence of cSoxl1 is similar to the PCR-derived HMG-box sequence Sox-11 from Wright et al. (1993) and outside the box also shows extensive identity with a murine Sox-11 cDNA sequence (P. Koopman, pers. commun.). Two of these proteins, cSox2 and cSox3 are highly related both with respect to their HMG-boxes which are 93% homologous and also three other conserved regions of the proteins. cSoxZ1 has an HMG-box that is 64-67% related to cSox2 and cSox3 and does not share any of the other three conserved regions. cSox2 and cSox3 code for proteins of similar size, 3 15 and 313 amino acids (34 kDa), respectively with overall similar amino acid compositions (Fig. 4A). At their Ntermini, both proteins have a 14 amino acid sequence with a consensus of MYXMXETEXKXPXP in which one of the differences is a conservative change, position 2, met/leu. This sequence shows no particular structural
32
D. Uwanogho et al. /Mechanisms of Development 49 11995) 23-36
characteristics that might suggest a possible function and shows no homology with motifs in other proteins in the database with the exception of murine Sox-l (R. Lovell-Badge, pers. commun.) (Fig. 4). Although both c&x2 and &ox3 protein sequences are assumed to start at the first inframe methionine because of the conservation of the first two amino acids in both chick and mouse Sax-2 and Sax-3 (MYXM - - ) in each case the best consensus start sequence corresponds with the second ATG, three amino acids downstream. Downstream of the HMG-boxes of cSox2 and c&x3 are two highly conserved regions which are also present in Sax-2, Sex-3 and Sox-1 and thus define these Sax proteins as all being members of a distinct sub-group. cSoxl1 codes for a protein of 396 amino acids (42.5 kDa) assuming translation starts from the first in frame methionine codon. Although this sequence does not have a good Kozak consensus match it is likely to be the translation start due to the first four amino acids being the same as those in Sox-4, namely, MVQQ. There are two other possible alternative starts at 18 and 26 amino acids downstream but neither of these have good Kozak consensus sequences. Outside the HMG-box, cSoxl1 shows no homology with cSox2 or cSox3 but shows two regions of homology with Sax-4, the first four amino acids (above) and the C-terminal 47 amino acids. Within this C-terminal region 20 out of the 28 amino acids are identical between cSoxl1 and Sox-4 and the overall similarity is 82% (Figs. 4A, 5A). This extensively conserved domain implies that these proteins are related and as with cSox2, cSox3 and Sax-1, form a distinct sub-class of Sax proteins. The region immediately downstream of the cSoxl1 HMG-box is overall basic, consisting of 20% lysine between 143-199. This region is followed by a short very acidic stretch, between 204-23 1, consisting largely of glutamic and aspartic acid (75%). This basic-acidic organisation is characteristic of many transcription activation domains including HMG proteins and reveals significant homology with the activation domain of HMGl and 2 (Laudet et al., 1993). The cSoxl1 cDNA isolated did not contain a poly-A tail and the size of the transcript on Northern blots (2.4 kb) suggests there is about 1 kb of sequence missing from the clone. The missing sequence contains the 425 bp of N-terminal coding sequence identified from genomic sequence and 5 ’ and 3 ’ untranslated sequence. Genomic clones have been isolated for all three genes and preliminary mapping (not shown) suggests all three genes are intronless. In situ analysis of these genes has revealed a temporally and spatially restricted pattern of expression predominantly in the nervous system. All three genes, CSOXZ, cSox3 and cSoxl1, are expressed in the CNS whilst only cSoxl1 exhibits high levels of expression in the PNS. In the neural tube cSox2 and cSox3 expression is
restricted to the proliferative layer where their transcripts were always localised in the ventricular zone. By 8 days of development the levels of cSox2 and cSox3 transcripts decrease and are no longer detectable by 14 days consistent with the decrease and halt of proliferation in the ependymal layer (Hamburger, 1948). cSoxl1 is also initially expressed in the neural epithelium and ependymal layers but is then expressed at much higher levels in cells which have migrated out to a more lateral position. The strongest expression of cSoxl1 was seen in regions where cells had most recently migrated, in the basal plate from HH17-HH24 and then later in the alar plate HH26-HH34. The level of cSoxl1 transcripts decreased in these regions concomitant with cessation of proliferation and migration. Double labelling with antibodies against neurotilament shows that there is a group of cells in the subventricular zone in which neither cSox2 nor cSox3 transcripts or neurofilament are detected, and it is these same cells which express elevated levels of cSoxl1. (This is particularly well illustrated by comparison of Fig. 8L with Figs. 9J and 9K). Thus the expression of these CSOXgenes appears to correlate closely with neuronal maturation in the CNS, with all three genes being expressed in proliferating neuroepithelia, followed by downregulation of cSox2 and cSox3 and concomitant upregulation of cSoxl1. As the cells continue to mature cSoxl1 expression is again downregulated when neurotilament protein, a marker of neuronal maturation, appears. Anteriorly, as the neural tube opens out into the hindbrain the neural epithelium forms the base of the brain with the roof forming above it by extension from the dorsal/lateral edge of the neural epithelium. Thus the positive subventricular layer for cSoxl1, seen in the neural tube, now lies beneath the ventricular epithelial layer while cells further beneath the epithelial layer are equivalent to the most lateral cells in the neural tube and are weakly stained for cSoxl1. A similar pattern of staining is seen in the midbrain with a corresponding pattern in the forebrain, that is, a weakly positive ventricular layer, a strongly staining layer immediately adjacent to this and weak staining in the most peripheral layers. In the alar region of the midbrain which forms the optic tecturn cSoxl1 expression appears in a defined temporal progression. As cells migrate out from the ventricular layer they form an expanding cSoxl1 positive layer between the ventricular layer and the fibrous layer. It therefore appears that cSox2 and cSox3 expression is associated with immature cells of the CNS whilst upregulation of cSoxl1 expression correlates with migration and the early stages of maturation (studies on cell differentiation in vitro support this hypothesis; unpublished observations). It is interesting that the roof of the hindbrain never stains positive for any of our probes since it is derived from the neural epithelium which is positive for all three
D. Uwanogho et al. /Mechanisms of‘ Developmen! 49 ( 1995) 23-36
probes. The roof plate is destined to become the highly vascularised choroid plexus. It therefore appears that this tissue derived from the neural epithelium develops into non-neuronal tissues which does not express this group of Sox genes, though its progenitor cells in the neural epithelium were positive for all three CSOXgenes. cSoxl1 is expressed at a relatively low level in the ventricular layers of the CNS compared to its level of expression in cells which have recently migrated to a more peripheral position. The level of expression in the neural epithelium and ventricular zone does not appear to be significantly different to that seen for &ox2 and cSox3, though the in situ data is by no means quantitative. In cells which have left the ventricular zone the difference between cSox2/cSox3 and cSoxl1 expression is marked since cSox2/cSox3 expression is reduced and the level of cSoxl1 is increased. Since these changes correlate so closely it seems possible that the relative levels of expression are directly related. It may be that a transcriptional regulator results in these changes in expression and that alteration in the level of expression of one of these cSox genes leads to changes in expression of the other genes. For example, cSoxl1 expression may be increased under the control of an as yet unknown factor, and the resulting increase in cSoxZ1 product might switch off cSox2 and cSox3 expression. Experiments using transgenie mice and ectopic gene expression in vitro should address this question. Outside of the CNS we see lower levels of expression of all three genes in a variety of other tissues. Most notably we see cSoxl1 expressed in the sympathetic chain and certain other peripheral ganglia. Since many of these cells are derived from the neural crest and we do not see such levels of expression in other crest derivatives such as the neurons of the dorsal root ganglia, it appears that cSoxlZ expression is lineage restricted in neural crest derivatives and may therefore play a role in defining those lineages. In particular the expression pattern of cSoxl1 we have observed appears to be confined to the sypathoadrenal lineage, which it has been suggested develops from a common progenitor sublineage of the neural crest. This lineage restriction is further supported by the observation that cSoxl1 transcripts are seen at equally high levels in the adrenal medulla, another site where sympathoadrenal progeny of the neural crest are found. This association between the sympathoadrenal lineage and cSoxl1 expression can be studied with relative ease both in vivo and in vitro. Neural crest cells in vivo and in culture can be studied for coexpression of the cSox genes and of catecholamines which are a characteristic feature of this sublineage. The expression patterns of these cSox genes show similarity to several vertebrate homologues of genes which encode products involved in neuronal development in Drosophila. In Drosophila several of these genes appear
33
to play a sequential role in the development of neurons. The genes of the Achaete-scute complex (AS-C) encode helix-loop-helix transcription factors and are members of the proneural group of genes which give cells the potential to become neurons. Notch encodes a cell surface receptor with multiple EGF-like repeats and is a member of the neurogenic group of genes involved in the decision whether to become a neuron. Prosper0 is another Drosophila gene which encodes a homeobox transcription factor which is not expressed in dividing neuroblasts or mature differentiated neurons but is expressed in young post-mitotic neurons. It is apparent then that these genes play progressively later roles in neuronal development (reviewed in Jan and Jan, 1992; Jimenez and Modolell, 1993). The expression patterns of the vertebrate homologues of these genes suggest that they share these sequential roles in neuronal development. The AS-C homologue Mash-Z is expressed early in neuronal development and its expression decreases early during neuronal maturation in both the CNS and PNS (Lo et al., 1991; Guillemot and Joyner, 1993). Notch homologues are also largely expressed in early neuronal precursors of the ventricular layer (Weinmaster et al., 1991; Reaume et al., 1992; Lardelli et al., 1994) whilst the prosper0 homologue, Prox-I is predominantly expressed in the young neurons of the subventricular layer (Oliver et al., 1993), that is in the same region of the CNS as cSoxlZ. Thus the expression of cSox2 and cSox3 is consistent with an early role similar to that of AS-C whilst cSoxl1 shows a pattern of expression in the CNS closely related to that of Prox-I. cSoxl1 also appears to be expressed at a similar stage of maturation in the PNS in relation to Mash-l, with increasing expression after Mash-l in neural crest derivatives, although Prox-I expression was not detected in these cells. The expression of cSoxl1 and Mash-l in the PNS also correlates in terms of cell type since Mash-Z expression is also restricted to the sympathoadrenal lineage, again supporting the proposition that expression of Mash-l might activate cSoxlZ in these cells. The inactivation of Mash-l by insertional mutagenesis in transgenic mice has shown that its expression is required for neuronal development in sympathetic and other PNS neurons (Guillemot et al., 1993). Since the expression patterns of cSoxl1 and Mash-l appear related it will be interesting to see the effect of such a null mutation for Mash-Z on cSoxl1 expression and vice versa. Our data would suggest that cSoxl1 is downstream of Mash-l since Mash-l expression is restricted to early stages of neuronal development in both the CNS and PNS (Lo et al., 1991) whilst cSoxl1 expression is upregulated later. It is not clear if Prox-I and cSoxlZ expression patterns are genuinely different since Oliver et al. (1993) found that Prox-1 expression levels were exceptionally low and therefore difficult to
34
il. Uwanogho
ef
al.
/Mechanisms
detect. Prox-1 might therefore also be expressed in the PNS but was undetected due to its very low level of expression. It seems then that these c&x genes, like the vertebrate homologues of Drosophila proneural and neurogenic genes, exhibit patterns of expression consistent with roles analogous to characterised genes in neuronal development in Drosophila. Expression of all three cSox genes at low levels in other non-neuronal tissues such as gut epithelium and muscle indicates other as yet undefined roles for these genes in cellular development during embryogenesis. The expression pattern of the mouse homologues Sox2, Sox3 and Soxll is being studied by others and apparently shows a similar pattern to that described here (R. Lovell-Badge; P. Koopman, pers. comms.). cSox2 and cSox3 fall into a clear subgroup according to both primary sequence and pattern of expression, suggesting a different functional role for Sox genes with this particular type of overall structure. The relationship of cSoxl1 expression to that of cSox2 and cSox3 does however show that, like the Hox genes and Pou domain genes, several of the Sox genes appear to play a role in development of the same embryonic tissues; in this case the CNS. 4. Experimental procedures 4. I. Isolation and sequencing of cSox genes An amplified, mixed stage (HH stage 14-17) chick embryo cDNA library in lambda ZAP11 derived from 1 x lo6 primary recombinants was screened at high stringency with a mixture of HMG-box fragments generated by PCR from chick genomic DNA (Coriat et al., 1993). Duplicate positive clones were purified and excised into pBluescript with R408 helper phage. A number of plasmid inserts were sequenced directly with an HMG-box primer (Coriat et al., 1993) to characterise the HMG-box sequence. Two clones pB4-1 and pB1 l-l with inserts of 1.7 kb and 1.4 kb, respectively were overlapping clones of the same gene, cSox2. Clones pB18-1 and pB7-1 with inserts of 1.5 and 1.7 kb were overlapping clones of cSox3. A fifth cDNA (pB12-1) with an insert of 1.5 kb had an HMG-box sequence identical to a murine PCR product named Sox-11 and this was provisionally named cSoxfZ (Wright et al., 1993). The cDNA inserts of these five clones were sequenced in full in both directions using nested Exonuclease III deletions (Erase-a-base-Promega) and dideoxy sequencing with Taq polymerase (USB) (Sanger, 1977). Sequences were initially analysed with the aid of the LKB DNAsis software and more detailed analysis of the conserved amino acid motifs with the “BLAST” programme (Altshaul et al., 1990). A genomic clone of cSoxl1 was isolated by screening an EMBW library of 1.0 x IO6 primary recombinants
of Development 49 (1995) 23-36
at high stringency with the pB12-1 insert. The genomic clone insert was restriction mapped, Southern blotted and probed with the pB12-1. Two PstI fragments of 1.2 kb and 1.1 kb were identified as covering the entire coding region and these were subcloned into pBluescript KS+ (Stratagene). The 1.2 kb fragment which contained the 5 ’ sequence was sequenced using a primer specific to cSoxl1 sequence at the 5 ’ end of the cDNA insert to obtain the most 5’ sequence of the gene not present in the cDNA. Genbank accession numbers: U12532, U12534 and U12467. 4.2. Expression Poly A+ RNA was extracted from chick embryos at HH stages 24,25,28, 31, 36 and 37 and from adult liver, kidney, brain, spleen, gonad, skeletal and cardiac muscle using guanidinium thiocyanate (Chomczynski and Saachi, 1987) and magnetic beads (Promega). A 4 pg aliquot of RNA was electrophoresed in 1.2 M formaldehyde/1.3% agarose gels and transferred to GeneScreen Plus membranes (NEN). The blots were stained with a solution of 0.03% methylene blue to reveal relative loadings, destained in 1% SDS/l x SSC prior to hybridisation and confirmed by /3-actin probing. Blots were hybridised in turn with labelled cDNA inserts from clones pB1 l-l, pB18-1 and pB12-1 at high stringency (50% formamide, 42°C). Filters were exposed to Hyperfilm-B for 5 days at -70°C before developing. Transcript sizes were estimated from the positions of /3actin, 18s and 28s ribosome bands (which could be seen on the methylene blue stained blots). Whole mount in situ hybridisation
Embryos at HH stage 17-21 were probed with both antisense and sense cSox2, cSox3 and cSoxll-specific digoxigenin-labelled riboprobes (Wilkinson, 1992). The embryos were dissected from the yolk and fixed overnight, at 4OC, in 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS). The embryos were then dehydrated through a series of methanol solutions (25, 50, 75, 100%). All the following steps were carried out in 1 ml for 5 min at room temperature, unless otherwise stated. The embryos were rehydrated and washed twice in PBST (PBS & 0.1% Tween-20), endogenous peroxidase in the embryos was blocked by washing in 6% hydrogen peroxide. According to size, embryos were digested in 10 &ml proteinase K, for 5-10 min, then post-fixed in 4% PFA10.2% gluteraldehyde. The embryos were prehybridised for 3 h in 50% formamide, 5 x SSC, 50 &ml tRNA, 1% SDS, 50 &ml heparin, at 70°C. For hybridisation the digoxigenin-labelled RNA probes were added to fresh prehybridisation solution at 500 rig/ml, and hybridised overnight at 70°C. The embryos were washed twice in solution 1 (50%
D. Uwanogho et al. /Mechanisms of Development 49 (199s) 23-36
formamide, 5 x SSC, 1% SDS) for 30 min at 70°C; once in a 1:1 mixture of solutions 1 and 2 (0.5 M NaCl, 10 mM Tris pH 7.5,O. 1% Tween-20) for 10 min at 70°C and three times in solution 2 for 5 min at room temperature. The embryos were treated with 100 &ml RNase A in solution 2 for 30 min at 37°C and twice in solution 3 (50% formamide, 5 x SSC) for 30 min at 65°C. The embryos were blocked with 10% sheep serum for 90 min at room temperature to prevent non-specific binding of antibody, and then incubated overnight at 4°C with preabsorbed anti-digoxigenin antibody (alkaline phosphatase conjugated). The embryos were washed with TBST (136 mM NaCl, 25 mM KCl, 25 mM Tris pH 7.5, 0.1% Tween-20) at room temperature for 24 h before the colour reaction was initiated. The embryos were incubated with 4-nitro blue tetrazolium chloride and 5bromo-4-chloro-3-indolyl-phosphate in NTMT (100 mM Tris pH 9.5, 100 mM NaCl, 50 mM MgCl*, 0.1% Tween-20) at room temperature. After completion of the colour reaction the embryos were washed in PBS and photographed. Wax section in situ hybridisation The in situ hybridisation protocol was the same in principle as that described by Wilkinson and Green (1990), modified as follows. Both antisense and sense &0x2, &ox3 and cSoxl1 -specific digoxigenin-labelled riboprobes were made, precipitated twice and resuspended in hybridisation buffer. Embryos were fixed in 4% paraformaldehyde, embedded in paraffin wax, and 6 pm sections were collected on TESPA coated slides as described by Angerer and Angerer (1991). The sections were dewaxed in histoclear (Cellpath plc), rehydrated through graded alcohols and incubated for 5 min in 2 x SSPE. The sections were digested with 20 mgiml Proteinase K at room temperature for 7.5 min, rinsed in 2 x SSPE, postfixed, and treated with 0.2 M hydrochloric acid for 15 min. The slides were incubated for 5 min in 2 x SSPE before being acetylated for 10 min in 0.1 M triethanolamine with 50 mM acetic anhydride. Hybridisation ( < 200 ng DIG-labelled probe/slide) was carried out overnight at 42°C. The slides were then washed under stringent conditions (65”C, 2 x SSC, 50% formamide) and treated with RNase (20 &ml). After washing in PBST (Phosphate buffered saline, pH 7.3, 0.1% Tween 20) the slides were incubated in blocking buffer (1% blocking reagent (Boehringer Mannheim) in PBST) for 30 min, and were then incubated overnight at 4°C in alkaline phosphatase-conjugated anti-digoxigenin antibody (Boehringer Mannheim), diluted 115000in PBST containing 2% chick serum (Sera labs). Unbound antibody was washed off with PBST and the bound antibody detected by a standard immunophosphatase reaction using 4-nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl-phosphate as substrates. The development time varied from l-10 days.
35
After development the sections were mounted in Depex (BDH). Immunohistochemistry Immunohistochemistry was carried out immediately following the in situ hybridisation protocol, all treatments were at room temperature. Colour development of the in situ hybridisation was stopped by immersing slides in 1 x TE (10 mM Tris/HCl pH 8: 1 mM EDTA). These were then overlayed with PBS, 20% serum for 30 min. This was replaced with primary antibody mixture; monoclonal 3AlO 1:20 dilution (Furley et al., 1990), monoclonal RT97 1:1000 dilution (Anderton et al., 1982) in PBS, 5% serum and incubated for l-2 h. Following a brief rinse in PBS they were overlayed with biotinylated sheep anti-mouse secondary antibody (Boehringer Mannheim) diluted 1:100 in PBS and incubated for 30 min. Following a brief rinse in PBS they were overlayed with AP-streptavidin (Boehringer Mannheim) diluted l/500 in PBS for 30 min. Following a brief rinse in PBS they were overlayed with Fast Red (prepared according to suppliers instructions) and the appearance of colour was observed (colour develops in 15- 120 min). Sections were mounted in Mowiol (Calbiochem). Acknowledgements We thank Robin Lovell-Badge and Peter Koopman for providing Sox gene sequences prior to publication and Andy Brass for detailed analysis of the cSox peptide sequences. Dafe Uwanogho and Maria Rex were funded by SERC studentships and Elizabeth J. Cartwright by a grant from the British Heart Foundation. References Altshaul, S.F., Gish, W., Miller, W., Myers, E.W. and Lipman, D.J. (1990) J. Mol. Biol. 215, 403-440. Anderton, B.H., Breinburg, D., Downes, M.J., Green, P.J., Tomlinson, B.E., Ulrich, J., Wood, J.N. and Kahn, J. (1982) Nature 298, 84-86. Angerer, L.M. and Angerer, R.C. (1991) Methods Biol. 35, 37-71. Bustin, M., Lehn, D.A. and Landsman, D. (1990) Biochim. Biophys. Acta 1049, 231-243. Chomczynski, R. and Saachi, N. (1987) Anal. Biochem. 162, 156. Cavener ( 1987). Coriat, A-M., Muller, U., Harry, J.L., Uwanogho. D. and Sharpe, P.T. (1993). PCR Methods Appl. 2, 218-222. Courey, A.J. and Tjian, R. (1988) Cell 55, 887-898. Denny, P., Swift, Brand, N., Dabhade. N., Barton, P. and Ashworth, A. (1992). Nucl. Acid Res. 20, 2887. Ferrari, S., Harley, V., Pontiggia, A., Goodfellow, P., Love&Badge, R. and Bianchi, M.E. (1992) EMBO J. 11, 4497-4506. Foster, J.W. and Graves, J.A. (1994) Proc. Natl. Acad. Sci. USA. 91, 1927-1931. Furley, A.J., Morton, S.B., Manalo, D., Karagogeos, D., Dodd, J. and Jessell, T.M. (1990) Cell 61, 15-170. Giese, K., Cox, J. and Grosschedl, R. (1992) Cell 69, 185-195. Graham. A., Papalopulu, N., Lorimer, J., McVey, J.H., Tudenham, E.G.D. and Krumlauf, R. (1988) Genes Dev. 2, 1424-438.
36
D. Lrwanogho et al. /Mechanisms of Development 49 (1995) 23-36
Guillemot, F. and Joyner, A.L. (1993) Mech. Dev. 42, 171-185. Guillemot, F., Lo, L-C., Johnson, J.E., Auerbach, A., Anderson, D.J. and Joyner, A.L. (1993) Cell 75, 463-476. Gubbay, J., Collignon, J., Koopman, P., Capel, B., Economou, A., Muensterberg, A., Vivian, N., Goodfellow, P. and Love&Badge, R. (1990) Nature 346, 245-250. Harley, V.R., Jackson, D.I., Hextall, P.J., Hawkins, J.R., Berkovitz, G.D., Sockanathan, S., Lovell-Badge, R. and Goodfellow, P.N. (1992) Science 255, 453-456. Hamburger, V. (1948) J. Comp. Neurol. 88. 221-283. Hamburger, V. and Hamilton, H.L. (1951) J. Morph. 88, 49-92. Jan, Y.N. and Jan, L.Y. (1992) Curr Opin Genet Dev. 2. 608-613. Jantzen, H-M., Admon, A., Bell, S.P. and Tjian, R. (1990) Nature, 344, 830-836. Jimenez, F. and Modolell, J (1993) Curr Opin Genet and Dev. 3, 626-632. Lardelli, M., Dahlstrand, J. and Lendahl, U. (1994) Mech. Dev. 46. 123-136. Laudet, V., Stehelin, D. and Clevers, H. (1993) Nucl. Acid Res. 21, 2493-2501. Lo, L-C., Johnson, J.E., Wuenschell, C.J., Saito, T. and Anderson, D.J. (1991) Genes Dev. 5, 1524-1537. Mueller, C.R., Maire, P. and Schibler LJ. (1990) Cell, 61, 279-291. Nasrin, N., Buggs, C., Kong, X.F., Carnazza, J., Goebl, M. and Alexander-Bridges, M. (1991) Nature 354, 317-320. Ner, S.S. (1992) Curr. Biol. 2, 208-210. Oliver, G., Sosa-Pineda, B., Geisendorf, S., Spana, E.P., Doe, C.Q. and Gruss, P. (1993) Mech. Dev. 44, 3-16. Oosterwegal, M., Wetering, M.v.d, Dooyes, D., Klomp, L., Winoto, A., Gergopoulos, K., Meijtink, F. and Clevers. H. (1991a) J Exp. Med. 173, 1133-1142.
Oosterwegel, M., Wetering, M.v.d. Holstege, F., Prosser, H.M.. Owen, M.J. and Clevers. H. (1991b) Int. Immunol. 3, 1189-l 192. Reaume A.G., Conion R.A.. Zirngibl R., Yamaguchi T.P. and Rossant J. (1992) Dev. Biol. 154, 377-387. Sanger, F., Nicklen, S. and Coulson, A.R. (1997) Proc. Natl. Acad. Sci:USA, 74, 5463. Sinclair, A.H., Palmer, M.S., Hawkins, J.R., Griffiths. B.L., Smith, M.J., Foster, J.W., Frischauf, A-M., Lovell-Badge, R. and Goodfellow, P.N. (1990) Nature 346, 240-244. Stevanovic, M., Lovell-Badge, R., Collignon, J., and Goodfellow. P.N. (1993) Human Mol. Gen. 2, 2013-2018. Tamkun, J.W., Deuring, R., Scott, M.P., Kissinger, M.. Pattatucci. A.M., Kaufman, T.C. and Kennison, J.A. (1992) Cell, 68. 561-572. Travis, A., Amsterdam, A., Belanfer, C. and Grosschedl, R. (1991) Genes Dev. 5, 880-894. van de Wetering, M., Oosterwegel, M.. Dooijes, D. and Clevers, H. (1991) EMBO J. IO, 123-132. van de Wetering M., Oosterwegel M., van Norren K. and Clevers H. (1993) EMBO J. 12, 3847-3854. Waterman, M.K., Fischer, W.H. and Jones, K.A. (1991) Genes Dev. 5, 656-669. Weinmaster G., Roberts V.J. and Lemke G. (1991) Development, 113, 199-205. Wilkinson, D. and Green, J. (1990) In Rickwood, D. and Cockcroft, D.L. (eds) In situ hybridisation and three dimensional reconstruction of serial sections: a practical approach. IRL Press, Oxford, pp. 155-171. Wilkinson D.G. (1992). In situ hybridisation. A practical approach. IRL Press, Oxford pp 75-84. Wright, E.W., Snopek, B. and Koopman, P. (1993) Nucl. Acids Res. 21,744.
