Mechanisms of Development 53 (1995) 47-59

Comparative analysis of chicken Hoxb-4 regulation in transgenic mice

Alastair Morrisona+*, Chitrita Chaudhuri a-1, Linda Ariza-McNaughtona, Ian Muchamorea, Atsushi Kuroiwab, Robb Krumlaufas*

aDivision of Developmental Neurobiology, MRC National Institute for Medical Research, The Ridgeway. Mill Hill. London NW7 IAA, UK

bDepartment of Molecular Biology, School of Science, Nagoya University, Chikusa-ku, Nagoya 464-01. Japan

Received 27 March 1995; revision received 2 June 1995; accepted 6 June 1995

Abstract

We cloned the chicken Hoxb-4 gene and performed in situ analysis to investigate conservation in patterns of expression between the chicken and mouse. The anterior boundaries of expression for both genes in segmented tissues, such as the hindbrain and paraxial mesoderm, map to the same rhombomere (r) (r6/r7) and somite (s) (s6/s7) limits, showing a direct correlation between expression of a specific Hox gene and patterning identical axial structures in both species. Given this similarity in expression we have tested the functional activity of &-regulatory regions from the chicken Hoxb-4 gene in transgenic mice to identify and map components conserved between the species. We identified enhancers which contain conserved blocks of sequence identity and which are necessary to mediate mesodermal and neural restricted patterns of expression. However, only the neural enhancer directs the proper anterior boundary of expression (r6/r7), indicating that only a subset of the underlying molecular components regulating Hoxb-4 expression are functionally conserved between species.

Keywords: Hox genes; Rhombomeres; Transgenic mice; Chicken embryos; Gene regulation

1. Introduction

The vertebrate group 4 paralogs represent the Hox

genes related to the Drosophila Deformed (Dfd) HOM-C homeotic selector gene. This group is one of only two paralogous subfamilies containing a member from each of

the four Hox complexes: Hoxa-4, b-4, c-4 and d-4 (Featherstone et al., 1988; Graham et al., 1988; Harvey and Melton, 1988; Acampora et al., 1989; Galliot et al., 1989; Sasaki and Kuroiwa, 1990; Sasaki et al., 1990; Paro and Hogness, 1991; Geada et al., 1992). In addition to the sequence similarities between the Hox Dfd-like proteins there are considerable overlaps in their patterns of ex- pression. In the mouse hindbrain Hoxa-4, b-4 and d-4 have anterior boundaries of expression that correlate with

the formation of rhombomeric (r) segments, and which map to the junction between r6 and r7 (Wilkinson et al., 1989; Hunt et al., 1991a). In cranial neural crest these 3

genes are expressed with an anterior boundary in deriva-

* Corresponding author, Tel: 0181 9138530; Fax: 0181 9064477.

’ These authors contributed equally to this work.

tives of the fourth branchial arch and in more posterior

crest populations in the trunk (Hunt et al., 1991a). How- ever, despite having the same anteroposterior (A-P) ex-

pression boundaries, these genes display differences in

their dorsoventral (D-V) patterns of expression in the neural tube (Gaunt, 1991; Graham et al., 1991). Unlike its

three paralogs, Hoxc-4 has a more posterior boundary of

expression in both nervous system and neural crest (Geada et al., 1992). There is a close correspondence of the A-P limits of expression in the somitic mesoderm of all four genes, but the domains are slightly offset. Hoxd-4 maps to prevertebrae (PV) 1, Hoxb-4 to PV2, Hoxa-4 to PV3, and Hoxc-4 to PVS (Gaunt et al., 1989; Whiting et al., 1991; Geada et al., 1992). In other tissues there are cell-type and stage-dependent differences in their expres-

sion. Overall differences in the patterns of expression among the Dfd-paralogs imply that they have distinct forms of regulation, but similarities in some tissues, such

as the hindbrain, suggest there may be some common regulatory components.

A normal role for the Dfd-related member in the HoxB complex, Hoxb-4, has been demonstrated by gene target-

0925-4773/95/$09.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved

SSDI 0925-4773(95)00423-X

48 A. Morrison et al. I Mechanisms of Development 53 (1995) 47-59

ing in the mouse (Ramirez-Solis et al., 1993). In accord with its anterior boundary of expression in the second

cervical vertebra (CV), homozygous Hoxb-4 mutants have a partial anterior homeotic transformation of the axis (CV2) (Ramirez-Solis et al., 1993). CV2 retains the dens

but also possesses a structure resembling the anterior arch

of the atlas characteristic of CVI. Therefore, in the mouse Hoxb-4 is part of a combinatorial Hox code involved in

the patterning of vertebral structures (Kessel and Gruss,

1991), where it participates in regulating the identity of the axis (CV2). However, in other species it is uncertain

whether this gene is linked with the patterning of the same specific vertebra (axis) or more generally regulates the identity of other vertebrae. In order to address this issue in somitic mesoderm and other tissues it will be

important to investigate the patterns of expression of Hox genes in other vertebrates and determine the degree of

conservation in their expression and underlying regula-

tion. There are variety of other defects in the Hoxb-4 mu-

tants, such as a failure in fusion of the sternal rudiments, which lead to neonatal lethality and underscore the role of Hoxb-4 in multiple tissues. The phenotypic abnormalities

in the Hoxb-4 mutants are primarily detected in regions which correspond to its anterior boundaries of expression,

and posterior prevalence of other Hox genes (Duboule and Morata, 1994) could account for the lack of pheno-

types or functional roles in more posterior regions. This

correlation between function and anterior expression fo- cuses attention on the regulatory mechanisms which

modulate anterior patterns of expression. Transgenic analysis of the &-regulatory regions from

the mouse Hoxb-4 locus has identified two spatially- specific enhancers which together recreate virtually all of the major aspects of endogenous Hoxb-4 expression, in- cluding the proper anterior boundaries (Whiting et al., 1991). A 3’ flanking region of the gene (region A) func-

tions as a neural-specific enhancer directing expression

throughout the neural tube to a sharp anterior boundary at the rhombomere 6/7 junction. Region A is complex in

that it also contains one of the promoters and exons for the adjacent Hoxb-3 gene (Sham et al., 1992). Expression in mesodermal derivatives, neural crest and the posterior

neural tube is mediated by a second enhancer, region C, located in the Hoxb-4 intron and flanking exonic se- quences. In contrast, regulatory elements implicated in

the control of the mouse Hoxa-4 and the human HOXD-4

genes have been localised to their 5’ flanking regions

(Tuggle et al., 1990; Behringer et al., 1993), suggesting that paralogs do not possess similarly positioned regula- tory elements, which is in accord with the expression dif-

ferences observed between Dfd-like paralogs. However, the topographical location of regulatory regions of a par-

ticular Hox gene in other vertebrates may be maintained, as a consequence of the structural conservation of the Hox complexes. This could provide a basis for comparisons

between divergent vertebrate species to search for and identify conserved regulatory components of the Hox genes.

In order to extend our understanding of the common roles of the Hoxb-4 gene in vertebrate development, we have cloned the chicken gene and examined its pattern of

expression during embryogenesis. The domains of ex-

pression in the chick embryo map to the same rhom- bomere, neural crest and somite boundaries previously

observed in the mouse. Furthermore, on the basis that similar mechanisms might be involved in regulating these

common patterns of expression, we used a transgenic approach to test genomic fragments from the chicken Hoxb-4 locus for enhancer activity in the mouse embryo.

The results define important similarities and differences in regulation of this gene between species.

2. Results

2. I. Whole mount in situ analysis of Hoxb-4 expresion in chicken embryos

To determine the pattern of chicken Hoxb-4 expres- sion, whole mount in situ hybridisation was performed on embryos, using the full length chicken Hoxb-4 cDNA as a probe (Sasaki et al., 1990). In embryos at Hamilton and Hamburger (HH) stage 8 (5 somites) expression was first detected in segmental plate mesoderm but not in

condensed somites (s), and also in the overlying posterior neural plate and developing neural folds (data not shown).

At HH stage 10 (10 somites), expression was detected in the three most recently condensed somites, with a sharp

boundary between s6 and s7 (Fig. 1A). Expression at this anterior boundary in paraxial mesoderm persist-

ed throughout later stages, and by HH stage 14-15 (22-27 somites) a posterior boundary of expression ap- peared at somite 16 (Fig. 1B). This defines a domain of 9-10 somites which expresses high levels of Hoxb-4. In the chick, somite 7 contributes cells to the second

Fig. 1. Whole mount in situ hybridisation of Hoxb-4 in chick embryos. (A) Dorsal view of an HH stage 10 embryo showing the anterior boundary of

expression in the paraxial mesoderm between somites (s) 6 and 7. Expression is also seen in the neural tube extending just anterior to the somite ex-

pression, at a level considerably more posterior than the ultimate r6fl boundary. (B) Dorsal view of an HH stage 14 embryo showing that the sharp

anterior boundary to expression between somites 6 and 7 is maintained, and there is also a clear posterior boundary of expression at the level of somite 16. The expression in the neural tube has now moved to the r6/7 boundary (indicated by arrowhead in B,C and D). (C) Flat mount preparation of an

HH stage 14 embryo showing the sharp anterior limit to expression at the r6/7 boundary in the hindbrain. (D) Lateral view of an HH stage 15 embryo

again showing the rostra1 limit to expression at the 16/7 boundary, just posterior to the otic vesicle. Expression can also be seen in the surface ecto-

derm, and in the cranial neural crest contributing to the fourth branchial arch as well as in more posterior neural crest derivatives. Rhombomere (I);

otic vesicle (OV).

A. Morrison et ul. I Mechanisms of Development 53 (1995) 47-59 49

50 A. Morrison et al. I Mechanisms of Development 53 (1995) 47-59

cervical vertebra, which goes on to form the axis. There-

fore, in both chicken and mouse embryos the ante- rior expression of Hoxb-4 coincides with the somite in-

volved in generating the same vertebral structure (the

axis). At HH stage 10 expression was also seen in the neural

tube at an A-P level just anterior to the somite expression

(Fig. lA), which is considerably posterior to the future presumptive r6/7 boundary. However, by HH stage 14-15

(22-27 somites), expression extended more anteriorly to a

sharp boundary in the hindbrain, just posterior to the otic vesicle and there was expression in the corresponding

surface ectoderm (Fig. lB-D). Flat mount preparations of embryos at this stage demonstrated that this boundary

corresponds precisely to the junction between r6 and r7 (Fig. IC). Expression at this same segmental boundary (r6/r7) persisted and levels increased throughout later

stages of embryogenesis (data not shown). At HH14-15 expression in the cranial neural crest was also observed in the fourth branchial arch adjacent to somite 4 and in pos-

terior neural crest derivatives (Fig. ID). These boundaries of expression are identical to those of the mouse Hoxb-4 gene. Thus both the neural and mesodermal patterns of

Hoxb-4 expression are conserved in chicken and mouse

embryos.

2.2. Cloning of chicken Hoxb-4 and analysis in transgenic mice

We wanted to determine whether conserved regulatory

processes also underlie the similarities in expression and therefore characterised the regulatory regions of the chicken Hoxb-4 gene by functional analysis in transgenic mice. A cosmid containing the Hoxb-4 gene was isolated from a chicken genomic library using homeobox and

cDNA probes. Following restriction enzyme mapping a 9.25 kb XbaI-BamHI fragment containing the entire

chicken Hoxb-4 gene, with 5’ and 3’ flanking sequences

was subcloned and sequence comparisons with the cDNA used to precisely identify the positions of the exons, in- tron, 3’ untranslated region and poly A addition site. Fig. 2 shows the structure of the chicken Hoxb-4 gene aligned with the homologous mouse gene. Subfragments of this region were used to generate constructs for transgenic

analysis. Initially we tested the ability of the chicken Hoxb-4

gene itself, contained within a 6.75 kb Xbal-XhoI frag-

ment, to direct expression in mouse embryos. Construct

#l contained the B-galactosidase (lacz) gene inserted in frame into the EcoRI site in the second exon of the chicken Hoxb-4 gene (Fig. 2). The resulting fusion gene

was tested for lacZ reporter activity in transgenic em-

Mouse

I c I I A

( -) .5 kb

Chicken

CB1 CB2

Enhancer regions meso/CNS

LacZ

-meso neural r6/7

-+ pB4-LacZ

L#IExpl 1 3

2 6

3 1

4 3

5 12

6 2

7 8

6 9

Fig. 2. Comparison of the genomic organisation of mouse and chicken Hoxb-4 genes and the enhancer regions identified by transgenic analysis. The

relative positions of the promoter, exons and poly A addition site of the Hoxb-4 homologs as well as a distal promoter from the neighbouring 3’ Hoxb- 3 gene are indicated. The locations of the mouse Hoxb-4 region A and region C enhancers are shown above the locus (Whiting et al., 1991). The solid

tilled black rectangles below the chicken locus indicate the enhancer regions identified in this study and their major sites of expression. Also shown

are the relative positions of two conserved blocks of sequence homology (CBl and CB2) which lie in the mesodermal and neural enhancers of both

species. The genomic fragments used for transgenic constructs to identify the chicken enhancers are shown at the bottom, and at the right is a table

indicating the construct number and the total number of transgenic mice producing a consistent expression pattern. pB4-lacZ is the basal promoter 1acZ reporter vector (Whiting et al., 1991) used to test for enhancer activity. Restriction enzyme sites, (B) BarnHI: (Bg) B$11; (E) EcoRl; (H) HindlII; (NC)

Ncol; (P) Psrl; (Rs) Rsd; (S) SulI; (Xb) Xbal; (Xh) Xhol.

A. Morrison et al. I Mechunisms <?f Development 53 (1995) 47-59 51

bryos. Analysis of founder embryos showed that expres-

sion was observed in the neural tube up to the proper

r6/r7 boundary but staining in the mesoderm was weak

and at a position more posterior than the normal Hoxb-4

boundary (data not shown). One concern was that the chicken promoter might be

unable to function correctly in the mouse embryo, possi-

bly due to evolutionary divergence or differential re-

quirements for basal promoter transcription factors. How-

ever, studies have shown considerable sequence identity in the promoters of the Hoxbd promoters from these two

species (Gutman et al., 1994). To address this we used the

same chick XbaI-XhoI fragment to direct expression from a minimal mouse 1acZ reporter gene (pB4-1acZ; construct #8 in Whiting et al., 1991), which has been successfully

used to monitor enhancer regions of other genes (Sham et

al., 1993; Marshall et al., 1994; Studer et al., 1994). Fig. 3

shows the time course of expression from 1 of 3 stable

transgenic lines, all of which had identical patterns of

staining, generated carrying construct #2. Expression was

first detected in the posterior neural tube and mesoderm

around 9.0 days post coitum (dpc) (Fig. 3A), which is similar to the timing of the endogenous mouse Hoxb-4

gene (Wilkinson et al., 1989; Whiting et al., 1991). By

9.5 dpc levels of expression had increased, and staining

was observed in the anterior neural tube with a sharp

boundary at the r6lr7 junction in the hindbrain, corre-

sponding to the rhombomeric boundaries of expression of the endogenous chick and mouse genes (Fig. 3B,C). At

9.5 dpc there was evidence for a biphasic distribution of

expression in the neural tube, which became more pro- nounced in older embryos, with strongest staining being

observed in the anterior domains (Fig. 3B,D,E). The chicken fragment did direct expression in the

Fig. 3. Time course of fi-galactosidase expression from a transgenic line containing the entire chicken Hoxb-4 gene (construct #2). (A) 9.0 dpc, (B) 9.5

dpc (C) dorsal view of B, (D) 10.5 dpc, (E) 11.5 dpc, and (F) dorsal view of E. (B-F) expression in the neural tube extends from the rhombomere (r)6/7

boundary (indicated by the arrowhead in B, and shown more clearly in C as a dorsal view) just posterior to the otic vesicle (OV). Note the biphasic

distribution of lad staining in the neural tube, indicated by the hollow arrow(s). This first appears at 9.5 dpc (B) but becomes more pronounced by

10.5 and 11.5 dpc (D,E). The strongest staining is seen in the anterior neural domain. Expression is seen in the somites and lateral mesoderm (indicated

by arrows in B,E and F) up to the posterior limit of the forelimb bud in the lateral mesoderm and slightly more posteriorly in the somites (the two ar-

rows in E), this is caudal to the normal boundary seen with either the endogenous mouse or chicken Hoxb-4 genes.

52 A. Morrison et al. I Mechanisms of Development 53 (1995) 47-59

proper mesodermal tissues. However, the anterior bound- ary of transgene expression in somites was more posterior

than that of the endogenous Hoxb4 gene. Staining was found at the level of the posterior limit of the forelimb

bud, while endogenous expression extends rostrally be-

yond the forelimb bud (Fig. 3A,B,D-F; Whiting et al., 1991). These patterns of expression were identical to

those observed with the chicken Hoxb-4AacZ fusion gene

(construct #l), suggesting that the deficiencies in meso-

dermal expression were not simply due to species related

promoter function. Therefore, the 6.75 kb XbaI-XhoI

fragment of the chicken Hoxb-4 gene contains regulatory elements capable of mediating neural expression to the

correct rhombomere boundary, but only directs a subset

of the mesodermal patterns of Hoxb-4 expression.

2.3. The chicken intron mediates expression in the posterior neural tube and mesoderm

To localise the control regions we next examined

whether the expression patterns directed by the chicken genomic fragment involved regulatory regions topologi-

tally equivalent to those of the mouse gene. The mouse

region C enhancer had previously been shown to direct most of the mesodermal expression patterns with the cor- rect anterior boundaries, and a posterior subset of the neu- ral Hoxb-4 expression in transgenic mice (Fig. 2; Whiting

et al., 1991). Even though the chicken constructs did not

reconstruct the complete mesodermal pattern, we first

tested a 2.3 kb XbaI-EcoRI fragment which contained the

promoter, first exon, intron and 87 bp of the second exon linked to the pB4-lacZ reporter (Fig. 2, construct #3).

Analysis of a founder embryo showed expression in the

posterior neural tube and in the mesoderm (predominantly the lateral plate) (Fig. 4A). This pattern corresponds pre- cisely to that observed with the entire 6.75 kb XbaI-XhoI fragment except for the anterior neural domain, suggest- ing that thisis regulated by a separate region. We further

mapped the mesodermal regulatory region by using a 944 bp PstI-EcoRI subfragment spanning the intron

(construct #4; Fig. 4B,C). Staining was detected in the posterior neural tube and mesoderm in a pattern indistin-

guishable to that seen with construct #3, showing that the mesodermal enhancer activity lay within the intronic se-

quences.

2.4. Mesodermal regulation and sequence conservation in the Hoxb-4 intron

Despite the differences in the anterior mesodermal boundaries, the mouse region C and the chicken Hoxb-4 intron directed similar tissue-specific expression, suggest-

ing that boundaries are set by different components. We compared their DNA sequences to search for any con- served regulatory elements, and this analysis between the two species revealed a single conserved block (CBl) spanning 72 bp with 75% sequence identity (Fig. 5A). Within this region there are two TAAT core sequences,

which correspond to potential homeodomain binding sites (reviewed in Gehring et al., 1994). In parallel studies we

have recently examined the structure of the Hoxb-4 gene in a highly diverged vertebrate species (Aparicio et al.,

1995), the Japanese puffer fish (Fugu rubripes), which

has a compact genome (Brenner et al., 1993). Comparison of the sequences from the Hoxb-4 intron from the puffer fish with that of the chicken, also identified the same sin-

gle conserved block of sequence, with a 72% overall identity (Fig. 5A). Because of this conservation between

the three species we also tested the enhancer activity of the puffer fish Hoxb-4 intron in transgenic mice and sur-

prisingly none of the nine transgenic founder embryos obtained displayed expression in the neural tube or meso-

derm (data not shown). Ectopic expression of the reporter in three of the nine embryos indicated that the transgene

was capable of expression.

The observations that the chick intron only mediated a

subset of the expression pattern and that the puffer fish intron had no activity suggested that the conserved do- main may be necessary but not sufficient for regulating

mesodermal expression. In species other than the mouse it is possible that the additional regulatory components re- quired to direct the full mesodermal expression pattern

are either located elsewhere in the locus or are contained within the intron but have diverged so that they are no

longer functional between species. As a first step to test this we extended our transgenic analysis to in&de addi-

tional flanking regions of the chicken Hoxb-4 gene. Inter-

estingly, using a 6.0 kb BamHI fragment, which includes a further 2.5 kb of 3’ flanking DNA (construct #6), we

observed high levels of expression with a more rostra1

anterior boundary in the mesoderm when compared with previous constructs (Fig. 6). There was stronger staining throughout the posterior mesoderm derivatives, surface

ectoderm and neural tube, but the anterior boundary of expression at r6lr7 was unchanged.

The anterior limit of expression in paraxial mesoderm

in these embryos mapped between somites 5 and 6 (CVl), one somite more anterior than that of the endogenous

chicken or mouse Hoxb4 expression patterns. No other regions in the chicken Hoxb-4 locus were able to direct

the correct s6/s7 mesodermal boundary in transgenic mice. Although this 2.5 kb region could be involved in

Hoxb-4 regulation, it is located downstream of 1 of 2 promoters from the adjacent Hoxb-3 gene (see Fig. 2), and the anterior mesodermal boundary of expression

(s5/s6) observed corresponds to that of Hoxb-3 itself (Graham et al., 1989; Hunt et al., 1991b; Sham et al., 1992). Hence we think this region is more likely to be

involved in regulating anterior mesodermal boundaries of

Hoxb3 rather than Hoxb-4. In support of this idea, a mesodermal enhancer has been found in a similar region of the mouse Hoxb-3 gene (Kwan et al., unpublished). Therefore we favour the possibility that the critical regu- latory elements required for setting the proper boundaries

A. Morrison et al. / Mechanisms of Development 53 (1995) 47-59 53

Fig. 4. Analysis of enhancer activity in the chicken Hoxb-4 intron and requirement of conserved block for activity. (A) A 9.5 dpc founder embryo containing construct #3. (B and C) Two independent 9.5 dpc founder embryos containing construct #4. Staining from both these constructs is essen- tially identical, being seen in the neural ectoderm (ne), at a level more posterior than the normal Hoxb-4 expression limit at the 1617 boundary (indicated by the hollow arrow in A). Expression is also seen in the mesoderm (m), predominantly lateral mesoderm, but again at a position more posterior than that of the endogenous Hoxb-4 gene. (E) Expression in 9.5 dpc embryo carrying a transgene with a deletion in CBI (construct #5) as compared to a control embryo carrying construct #2 (D). Note that CBI is essential for posterior mesoderm and neural tube expression, but not for the anterior r6/7 restricted domain. Bold arrows in (D and E) denote CBI dependent domain. OV, otic vesicle.

of expression are indeed contained in the chicken or puffer fish Hoxb4 introns, but that the sequences or the factors which bind to them have diverged during verte- brate evolution such that they no longer function effec- tively in the mouse. In order to confirm a functional role for CBI in the chicken intron we prepared a construct (#5) which deleted it in the context of the full length

XbaI-XhaI fragment. In transgenic embryos carrying this deleted version the posterior mesoderm and neural tube expression were specifically abolished (Fig. 4E), while the anterior neural expression mapping to the Al7 bound- ary was unaffected. These results indicate that CBl is required for the expression mediated by the chicken en- hancer.

54 A. Morrison et al. I Mechanisms of Development 53 (1995) 47-59

A

mb4

Cb4

lb4

6

cd4

cb4

ca4

C mb4

cb4

lb4

mb4

cb4

fb4

CCATTGCCAGAGATTTACGGTCTCCTGTTTTCAGAGCCACAT~TTA~TCGCCCAT~TTTTTATGGCCT . . . . . . ...*.**.* .,.... ..**....*. . ..*... :: .*.*... .:: :... ::::::::::: :::::::::::::: .:: CCATTCGTGGAGATTTACGATCGCCTGTTTCAAGCGT ---ATAATTACATCCTCCATAAATTTTTATTCTCT

: .:: ::::::::::::: :::::: .: ..***..*. *........*..*..,.. . . . . . . . . . . . . . . . . . . . . . . . . . . . ~TGTGGGTCAAGATTTACGATCGTCTGTTTGC~G~~~~T~T~ACACCCTCCAT~TT~ATTA~AC

GATGCTCGCTCGAGCCCTGGCCGAGGCTTTMGCGGTTGCCG . . . .

CCiTT&TGGAGATTTAC;ATrdCCTG%,AGCG; . . . . . . . . . . . . . . : : :::::::::::: .:

---ATAATTACATCCTCCATAAATTTTTATTCTCT . . . . . . . . . . . . . . . . . . . . . . . . . ..#.... . . . . . . . . . .

AAAGTTCACAGCCATTCTGTGTAGACAAGAGCT-AAGAAA AACCATTAAT . . . . . . . . . . . . . . . . . . . . . . ..*....... :. : ::::.::: :::::::::: ****~***..* :::: . . . . . . . . . . . :: ::::::: AAAGTTCACAGCCATTATAAGCAGACCAGAAGCCAAGAAAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A13rGTTCACAG----hGGGG~~~G~~CC--~~-~G~TGCGAG~TTATACT~~~CC~TCATT~T

CACTTCTTTTC-TTTAAATACGTATCCTC--TCTCCTTTGT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . :: : ***. . . . . CACTTCTTTTC-TTTAAATAC-TATCCTT-TTCCC-ATTGT *** : * ..*..* *.. .** . *.... CACGTG;'AClrC,;',Cl~~C~T~~C~~~~~T~~~~~

Fig. 5. Conserved sequence blocks located in the mesodermal and neural specific enhancers of the chick, mouse and puffer fish Hoxb-4 genes. (A) Comparison of a 72 bp block of conserved sequence (CBl) within the introns of the chicken, mouse and puffer fish Hoxb-4 genes. The CBl sequences from the chicken, mouse and puffer fish Hoxb-4 genes were found 86 bp, 113 bp and 161 bp respectively from the 5’ splice site of exon 1. (B) com- parison of the introns of two chicken paralogs Hoxa-4 and Hoxd-4 with Hoxb-4. (C) Conserved sequence block (CB2) found in the 3’ neural enhancers of the chicken, mouse and puffer fish Hoxb-4 genes. (:) indicates identity with the chicken Hoxb-4 sequence and (.) indicates identity between the other two homolog or paralog sequences.

2.5. A chicken neural enhancer mediating rhombomere- restricted expression

In contrast to the regulation in mesoderm, the proper anterior boundaries of expression in the hindbrain are

reproduced by the chicken genomic fragments. Construct #6 generated neural expression with a boundary at the junction between r6 and r7, and the minimal overlap with the previous constructs (#l-4), suggested that an enhan-

cer analogous to mouse region A might be located in the 3.5 kb BamHI-XhoI 3’ flanking fragment. Therefore, we

tested this chicken fragment on the pB4-ZacZ reporter gene (construct #7). Five transgenic lines and 3 founder embryos containing this construct displayed identical neu-

ral-restricted patterns of expression, and a time course between 9.5-12.5 dpc for one of the lines is shown in Fig. 7. No staining was observed before 9.0 dpc, and when

expression appeared the anterior boundary precisely mapped to the r6/r7 junction, which corresponds to that of the endogenous mouse and chicken Hoxb-4 genes. The neural specific expression was limited to an anterior do-

main with a relatively sharp posterior boundary that mapped to the region of the forelimb bud. Expression

within this same anterior domain persisted throughout later stages (Fig. 7). In addition to the proper anterior boundary of expression the chicken enhancer also di- rected the proper D-V restricted expression expected for members of the HoxB cluster (Graham et al., 1991).

Therefore, while this chicken enhancer mediates neural-

restricted expression with the same anterior boundary as

mouse region A, it does not direct the same posterior patterns of expression in the neural tube, suggesting that

only a subset of the regulatory components have been conserved.

Further deletion mapped the chicken neural enhancer

to a 1.4 kb EcoRI-XhoI fragment (construct #8). All nine

founder embryos containing this construct displayed neu- ral-restricted staining in an anterior domain, in a manner

identical to those with construct #7 (Fig. 7G,H). Flat mounted preparations of a hindbrain clearly showed that the boundary of expression maps to r6fr7. The anterior domain of expression directed by this 3’ neural enhancer and the posterior neural staining mediated by the intron explain the biphasic nature of the transgenic expression

patterns observed in the largest constructs (#l and #2), and show that multiple regions are involved in regulating expression of the chicken Hoxb-4 gene in the mouse nervous system.

We sequenced the 1.4 kb chicken neural enhancer and

compared it with topologically equivalent domains from the mouse (region A) and puffer fish Hoxb-4 genes to search for conserved sequences that might be important for regulation. The results of this three species compari- son are illustrated in Fig. 5C and show that there is one major conserved block (CB2) of 107 bp which is located

A. Morrison et al. I Mechanisms of Development 53 (I 995) 47-59 55

Fig. 6. Mapping additional mesodermal regulatory regions. (A) Dorso-lateral view of a 10.0 dpc transgenic embryo carrying construct #6 showing a sharp anterior limit of expression at the r6/7 boundary (arrowhead) and extensive staining in the mesoderm. (B) Dorsal view of A showing expression in the paraxial mesoderm, with an anterior limit to expression at the somite 5/6 boundary (white arrow). This is at a level more anterior than the en- dogenous Hoxb-4 mesodermal boundary but which corresponds to that of the neighbouring Hoxb-3 gene.

immediately 5’ of the HindI11 site in the chicken locus

(Fig. 2). The chicken and mouse sequences showed 87% identity over this stretch, while the puffer fish sequence showed 76% identity with the chicken. This block spans

the same sequences in the mouse and human genes previ- ously shown to correspond to the first 5’ untranslated

exon of a transcript from the most distal promoter of the 3’ flanking Hoxb-3 gene (Sham et al., 1992). The conser- vation of the exon and a splice donor site at the 3’ end of CB2 suggests that this distal Hoxb-3 promoter is used in at least 4 very divergent vertebrates. Functional analysis

of this conserved block is complicated because it contains promoter elements from the Hoxb-3 gene but initial analysis of the mouse and puffer fish genes indicates that it is important in regulating r6/r7 neural activity (Gould et al., unpublished; Aparicio et al., 1995). Therefore the Hoxb-4 &-regulatory components and their correspond- ing regulatory factors, capable of directing the proper anterior neural expression in the hindbrain, have been

more highly conserved than those implicated in meso-

dermal regulation.

3. Discussion

3.1. Conserved patterns of Hoxb-4 expression We have shown by in situ hybridisation that major as-

pects of Hoxb-4 expression are conserved between the mouse and chick. In particular, the anterior limits of ex-

pression in the hindbrain map to the same rhombomeric segments (upto r6/r7) and in the paraxial mesoderm to the

same somites (upto s6/s7) in both species (Graham et al., 1988; Gaunt et al., 1989; Wilkinson et al., 1989; Whiting et al., 1991). In the mouse, Hoxb-4 is required for regulat- ing the identity of the axis or second cervical vertebra (CV2) (Kessel and Gruss, 1991), and in the chick the an- terior boundary of Hoxb-4 also marks the somite which generates CV2 (~7). Therefore, there is a direct correla- tion between the anterior expression of this Hox gene and

56 A. Morrison et al. I Mechanisms of Development 53 (1995) 47-59

Fig. 7. Localisation of the r6/r7 neural enhancer in the Hoxb-4 gene. (A-D) Time course of expression in embryos from a transgenic line containing construct #7 at 9.5, 10.5, 11.5 and 12.5 dpc respectively. Staining is only seen within a neural domain extending from a sharp anterior limit at the r6/7 boundary (indicated by arrowhead in A) caudally to the level of the forelimb bud (hollow arrow in A). This expression corresponds to the anterior component of the biphasic distribution seen in the neural tube with constructs 1 and 2 (Fig. 3). (E) Flat mount preparation of the embryo in A illustrat- ing the sharp anterior boundary at the r6/7 junction. (F) Dorsal view of one of five 10.5 dpc founder embryos containing construct #8, which also shows staining only in the neural tube up to the r6I7 boundary (arrowhead). (G) Flat mount of F. (H) A lateral view of F showing that again neural expression is absent posterior to the level of the forelimb bud (hollow arrow). Rhombomere (r), otic vesicle (ov).

the patterning of the same specific axial structures in the

hindbrain and somites in different species. In an analo-

gous manner, the anterior r4-restricted domain, but not the posterior domain, of Hoxb-I is conserved between

chicken and mouse (Murphy et al., 1989; Wilkinson et al., 1989; Sundin and Eichele, 1990; Maden et al., 1991). This evolutionary link between expression and structural homology has also been recently examined for more

posteriorly expressed Hox genes (Burke et al., 1995). In general the data argue that the conservation of the same Hox expression patterns in directly analogous anatomical structures reflects a specific requirement for a particular combination of Hox genes in regulating morphological identity.

3.2. Functional conservation of neural restricted expression

Our hypothesis has been that underlying the similarity in expression are many common cellular and molecular mechanisms involved in regulating these spatially-

restricted patterns in different vertebrates. The deletion

analysis of the chicken Hoxb-4 neural control region in

transgenic mice demonstrated that chicken cis-acting regulatory components are capable of interacting with

mouse factors to mediate the proper r6/r7 boundary of expression in the hindbrain. Furthermore, both the timing of the rhombomere-restricted expression and the progres- sive dorsoventral restriction of expression displayed by

the endogenous mouse Hoxb genes (Graham et al., 1991), were also reproduced by the chicken regulatory region. This demonstrates that the anterior neural regulatory components have been highly conserved between species. The chicken r6lr7 neural enhancer is positioned in a to-

pographically equivalent 3’ flanking region, as compared to the mouse, and sequence comparisons mapped a 107 bp core region of highly conserved sequence (CB2). This type of sequence analysis can be very valuable in defining minimal functional regulatory elements, as we have previously demonstrated by identifying retinoic acid response elements required for normal expression of the

A. Morrison et al. I Mechanisms c?f Development 53 (I 995) 47-59 57

Hoxb-I gene (Marshall et al., 1994; Studer et al., 1994).

However, the conserved sequence block in this Hoxb4 region not only contains an r6/r7 neural enhancer but a conserved distal promoter and 5’ untranslated exon for the

adjacent Hoxb-3 gene, which makes it difficult to separate the two regulatory activities. In this regard it might be

useful to make comparisons with the Hoxa4 and Hoxd4 paralogs, in order to map and determine whether the

regulatory regions that mediate r6/r7 expression are

analogous. If so this could help to delineate the precise

r6lr7 enhancer elements in the Hoxb-4 gene.

The mouse region A enhancer directs expression from

r6/r7 throughout the entire neural tube (Whiting et al.,

1991), while the chicken enhancer mediates a sharp pos-

terior boundary at the level of the forelimb (Fig. 7). Therefore, only the most anterior aspects of the neural expression are conserved between the chicken and mouse enhancers. It is perhaps not surprising that regulation of this anterior expression domain is the most conserved because mutational analysis of mouse Hox genes indicates

that it is primarily these most anterior domains which are

functionally significant (reviewed in McGinnis and

Krumlauf, 1992; Krumlauf, 1994). If posterior prevalence

of other Hox genes (Duboule and Morata, 1994) takes

functional precedence in caudal regions of the embryo, then it is possible that the posterior domains of expression do not need to be maintained or are free to diverge be- tween species. In this respect there may be considerable variation in the posterior expression and regulation of many Hox genes.

Multiple regulatory regions involved in directing neu-

ral expression might provide another potential reason to

account for posterior variation in expression controlled by

the 3’ neural-specific enhancers. As with the mouse gene,

the chicken Hoxb-4 intron directs some expression in

posterior regions of the neural tube. The posterior neural domain mediated by the intron overlaps with the anterior

domain imposed by the 3’ enhancer, generating the

biphasic nature of the neural expression pattern seen in large constructs (Figs. 3 and 7). Redundancy or compen- sation between regulatory regions directing similar poste- rior patterns could also allow drift in one of the neural components in other species.

3.3. Divergence of mesodermal regulation The Hox gene clusters are one of the best examples of

conservation between vertebrate gene families. The value

of using evolutionary comparisons based on this similar- ity in structure and expression to identify regulatory re-

gions, as detailed for Hoxb-4 neural expression above, Hoxb-I (Marshall et al., 1994; Studer et al., 1994) or Hoxd-II (Gerard et al., 1993) has clearly been demon- strated. However, this success has led us to assume that all the major aspects of regulation will be fully conserved between diverse vertebrate species. In contrast, our trans- genie analysis of regions involved in regulating meso-

dermal expression of Hoxb4 illustrates that there is con-

siderable divergence in the functional activity of control

regions between species.

The chicken Hoxb-4 intron does contain an enhancer

that mediates expression in some of the same posterior mesodermal and neural tissues, as the mouse intronic en- hancer (Fig. 3). However the anterior boundaries of ex-

pression in mesoderm are more posterior and the relative levels of expression lower. An obvious reason for the

disparity between the chicken and mouse intron activities

could be that we are missing some elements required to

set the proper anterior boundaries of expression, because

they have different positions in the chicken Hoxb-4 locus.

We attempted to exclude this possibility by testing other regions of the chicken locus but were unable to find any

additional regulatory regions that could set the proper s6/s7 boundary in somitic mesoderm. In this analysis we did find another regulatory region in 3’ flanking frag- ments which directs expression up to a rostra1 boundary at

sYs6 in paraxial mesoderm, typical of the Hoxb-3 pattern

of expression. However, we believe that this actually rep- resents a mesodermal enhancer for Hoxb-3 itself, as it is

3’ of one of the Hoxb-3 distal promoters and a similar

region has been observed in the mouse Hoxb-3 gene

(Kwan et al., unpublished). Surprisingly, the intron of the puffer fish Hoxb-4 gene

generated no consistent patterns of expression in trans- genie mice. This illustrates that as intron sequences from more evolutionary diverse vertebrates are used in trans-

genie assays they become less capable of directing the proper mesodermal patterns of expression. There is how-

ever, a highly conserved block (CBl) in the chicken,

puffer fish and mouse Hoxb-4 intron (Fig. 5A), and here

we demonstrated that CBl in the chicken intron is neces-

sary for enhancer activity in transgenic mice (Fig. 4E).

Therefore while the expression patterns mediated by the mouse and chicken introns are not identical they require

the same sequence components. The paradox of a critical

region shared between the three species and their differ- ing regulatory activities could be explained by the inabil- ity of mouse factors to recognise some of the sequence

elements in the chicken and puffer fish intron. This could be due to divergence between the species, for example in

cofactors which are necessary for the full expression pat- tern. Any binding site and its cognate factor which have evolved in concert to maintain necessary interactions may have introduced changes which render it non functional in

another vertebrate. Therefore, the partial activity of the chicken intron does not necessarily imply that different

factors are used to regulate mesodermal expression in chicken and mouse embryos.

We also cloned and sequenced the introns of the chicken Hoxa-4 and Hoxd-4 (Sasaki et al., 1990) group 4 paralogous genes to determine if CBl might also have been conserved in the paralogs during duplication and divergence of the Hox clusters. This region is conserved

58 A. Morrison et al. I Mechanisms of Deveiopmenf 53 (1995) 47-59

but to a lesser degree than between the homologs, as there is 43% identity with Hoxd-4 and 50% identity with Hoxa- 4 (Fig. 5B). Interestingly, most of the identity with the paralogs is found at the 3’ end of the block, and the addi-

tional sequences conserved between the homologs at the 5’ end may represent modifying elements specific for Hoxb-4 regulation. The CBl related region in the

paralogs could also imply that the intron of these genes

plays a role in regulating their patterns of expression. The

group 4 paralogs are all expressed in paraxial mesoderm,

but with offset anterior boundaries, and the reduced

identity in CBl between the paralogs may suggest that

some common elements/factors in combination with dif-

ferent cofactors are responsible for their differential ex- pression.

In conclusion, our analysis has further demonstrated

the value of using interspecies comparisons to identify conserved components in the reguIation of vertebrate Hox genes. For Hoxb-4 the most conserved molecular mecha-

nisms are associated with neural restricted expression and

are involved in specifying the r6/r7 anterior boundary of

expression in the hindbrain. However, we can not always

expect such a high degree of conservation, as the analysis

of regulation of mesodermal patterns of expression clearly demonstrates that there may be a considerable degree of divergence between vertebrate species.

4. Experimental procedures

4.1. Whole mount in situ hybridisation Chicken embryos were fixed in 4% paraformaldehyde

and then hybridised using a modification of the method of

Wilkinson (1992). Digoxigenin-labelled RNA probe cor-

responded to the full length Hoxb-4 gene (Sasaki and

Kuroiwa, 1990).

4.2. Transgene construction Construct 1 was made by inserting the 1acZ gene in

frame into an EcoRI restriction site within the second exon of the chicken Hoxb-4 gene within a 6.75 kb XbaI- XhoI fragment of the genomic clone ~804~x1~. The latter was made by cloning the XbaI-XhoI fragment from the cosmid pCos804D into pBluescript SK+(Stratagene). The

injection fragment was prepared by digestion with XbaI- XhoI. Constructs 2, 3, 7 and 8 were all made by inserting

the relevant DNA fragment (end filled with T4 DNA po- lymerase) into the BamHI site of the basal 1acZ reporter

pB4-lacZ (construct #8 in Whiting et al., 1991), which had also been end filled before ligation. Construct 4 was made by first digesting p804xx/c with EcoRI end filling, and then digesting with PstI. The 941 bp fragment con- taining the chicken Hoxb-4 intron was isolated, then ligated into pB4-1acZ which had been digested with NcoI, end filled, and then digested again with PstI. Construct 5 was made by inserting the XbaI-XhoI fragment with an internal M.scI deletion which removed CBl on the 3’ side

and extended on the 5’ side to the ATG, into the polylinker of pB4-1ucZ. Construct 6 was made by cloning

a 6 kb BamHI fragment from the chicken cosmid pCos804D into pB4-1acZ which had firstly been digested

with BamHI and its 5’ terminal phosphates removed using calf intestinal phosphatase. Injection fragments for con- structs 2-8 were prepared by digestion with NotI. Con-

struct variations and chicken regions were sequenced by

the dideoxy chain termination method using double

stranded DNA and Sequenase (USB).

4.3. Production of transgenic mice by microinjection and ,&galuctosidase staining

In all these experiments (CBA X C57BL10)F1 mice were used as stud males, egg donors, pseudopregnant females, vasectomised males and adult females for

breeding. Transgenic mouse production and staining re- actions for the 1acZ reporter gene were performed as pre-

viously described (Hogan et al., 1986; Whiting et al., 1991). Timed pregnancies were used to obtain embryos at

particular developmental stages. The day the vaginal plug

was observed was taken as 0.5 dpc.

Acknowledgements

We thank Sam Aparicio and Sydney Brenner for dis- cussions and help with the puffer fish sequence and analysis, Zoe Webster and Lorraine Jones for help with

animal husbandry, Alex Gould, Peter Rigby and Jonathan Gilthorpe for valuable discussions on transgenic analysis and sequence comparisons and sharing unpublished re-

sults, and other members of the Krumlauf laboratory for

discussions. A.M. was supported by an MRC Training

Fellowship, C.C. by an HFSP collaborative network grant

and I.M. by an MRC studentship.

References

Acampora, D., D’Esposito, M., Faiella, A., Pannese, M., Migliaccio, E.,

Morelli, F., Stornaiuolo, A,, Nigro, V., Simeone, A. and Boncinelli,

E. (1989) Nucleic Acid Res. 17,10385-10402.

Aparicio, S., Morrison, A., Gould, A., Gilthorpe, J., Chaudhuri, C..

Rigby, P.W.J., Krumlauf, R. and Brenner, S. (1995) Proc. Natl.

Acad. Sci. USA 92, 1684-1688.

Behringer, R., Crotty, D.A., Tennyson, V.M., Brinster, R., Palmiter, R.

and Wolgemuth, D. (1993) Development 117, 823-833.

Brenner, S., Elgar, G., Sandford, R., Macrae, A., Venkatesh, 8. and

Aparicio, S. (1993) Nature 366,265-268.

Burke, AC., Nelson, C.E., Morgan, B.A. and Tabin, C. (1995) Devek

opment 121,333-346.

Duboule, D. and Morata, G. (1994) Trends Genet. 10,358-364.

Featherstone, MS.. Baron, A., Gaunt, S.J., Mattei, M.G. and Duboule,

D. (1988) Proc. Natl. Acad. Sci. USA 85.4760-4764.

Galliot, B., Dolle, P., Vigneron, M., Featherstone, M.S., Baron, A. and

Duboule, D. (1989) Development 107, 343-359.

Gaunt, S.J. (1991) BioEssays 13, 505-513. Gaunt, S.J., Krumlauf, R. and Duboule, D. (1989) Development 107.

131-141.

Geada. A.M.C., Gaunt, S.J., Azzawi, M., Shimeld, S.M., Pearce, I. and

Sharpe, P.T. (1992) Development 116.497-506.

A. Morrison et (11. I Mechanisms of Development 53 (1995) 47-59 59

Gehring, W.J., Qian. Y.-Q., Billeter, M., Furukubo-Tokunaga, K.,

Shier, A.F., Resendez-Perez, D., Affolter, M., Otting, Cl. and

WurtJuich, K. (1994) Cell 78, 211-223.

Gerard, M., Duboule, D. and Zakany, J. (1993) EMBO J. 12, 3539-

3550.

Graham, A., Papalopulu, N., Lorimer. J., McVey, J., Tuddenham. E.

and Krumlauf, R. (1988) Genes Dev. 2, 1424-1438.

Graham, A., Papalopulu, N. and Krumlauf, R. (1989) Cell 57,367-378.

Graham, A., Maden, M. and Krumlauf, R. (1991) Development 112,

255-264.

Gutman, A., Gilthorpe, J. and Rigby, P.W.J. (1994) Mol. Cell Biol. 14,

8143-8154.

Harvey, R.P. and Melton, D.A. (1988) Cell 53,687-697.

Hogan, B.. Costantini, F. and Lacy, E. (1986) Manipulating the Mouse

Embryo. Cold Spring Harbor Laboratory, Cold Spring Harbor.

Hunt. P., Gufisano, M., Cook, M., Sham, M., Faiella, A., Wilkinson,

D., Boncinelli, E. and Krumlauf, R. (1991a) Nature 353, 861-864.

Hunt, P., Wilkinson, D. and Krumlauf, R. (199lb) Development 112,

43-5 I.

Kessel, M. and Gruss, P. (1991) Cell 67, 89-104.

Krumlauf, R. (1994) Cell 78, 191-201.

Maden, M., Hunt, P.N., Eriksson, U., Kuroiwa, A., Krumlauf, R. and

Summerbell, D. (1991) Development II 1,3544.

Marshall, H., Studer, M., Popped, H., Apricio, S., Kuroiwa, A., Bren-

ner, S. and Krumlauf, R. (1994) Nature 370.567-571.

McGinnis, W. and Krumlauf, R. (1992) Cell 68,283-302.

Murphy, P., Davidson, D. and Hill, R. (1989) Nature 341, 156-159.

Pare, R. and Hogness. D. (1991) Proc. Nat]. Acad. Sci. USA 88, 263-

267.

Ramirez-Solis, R., Zheng, H., Whiting, J., Krumlauf, R. and Bradley,

A. (1993) Cell 73.279-294.

Sasaki, H. and Kuroiwa, A. (1990) Nucleic Acids Res. 18, 184.

Sasaki, H., Yokoyama, E. and Kuroiwa, A. (1990) Nucleic Acids Res.

18, 1739-1747.

Sham, M.-H., Hunt, P., Nonchev, S., Papalopulu, N., Graham, A.,

Boncinelli, E. and Krumlauf, R. (1992) EMBO J. I I, 1825-I 836.

Sham, M.H., Vesque, C., Nonchev, S., Marshall, H., Frain, M., Das

Gupta, R., Whiting, J., Wilkinson, D.. Chamay, P. and Krumlauf, R.

(1993) Cell 72. 183-196.

Snider, M., Popperl, H., Marshall.H., Kuroiwa, A. and Krumlauf, R.

(1994) Science 265. 1728-1732.

Sundin, 0. and Eichele, G. (1990) Genes Dev. 4, 1267-A 276.

Tuggle, C.K.. Zakany, J., Cianetti, L., Peschle. C. and Nguyen-Huu.

MC. (1990) Genes Dev. 4, 180-189.

Whiting, J., Marshall, H., Cook, M., Krumlauf, R., Rigby, P., Stott, D.

and Allemann, R. (1991) Genes Dev. 5,2048-2059.

Wilkinson, D.G. (ed.) (1992) Whole Mount In Situ Hybridisation Of

Vertebrate Embryos. IRL Press, Oxford University Press, Oxford,

pp. 75-83.

Wilkinson, D., Bhatt. S., Cook, M., Boncinelli, E. and Krumlauf, R.

(1989) Nature 341.405-409.