Deletions within human chromosome 22q11 cause a wide variety of birth defects including the DiGeorge syndrome and velo-cardio-facial (Shprintzen) syndrome. Despite the positional cloning of several genes from the critical region, it is still not possible to state whether the phenotype is secondary to haploinsufficiency of one or more than one gene. In embryological studies phenocopies of these abnormalities are produced by a variety of actions which disrupt the contribution made by the cranial and cardiac neural crest to development. The TUPLE1/HIRA gene is related to WD40 domain transcriptional regulators and maps within the DiGeorge critical region. We have cloned the chick homologue of HIRA and conducted in situ expression analysis in early chick embryos. Hira is expressed in the developing neural plate, the neural tube, neural crest and the mesenchyme of the head and branchial arch structures. HIRA may therefore have a role in the haploinsufficiency syndromes caused by deletion of 22q11.
The TUPLE1 gene encodes a protein with several features of a transcriptional control protein and sequence similarity to a series of proteins containing WD40 domains (1 ). Following the initial description of TUPLE1, named on the basis of its sequence and architectural similarity to the yeast TUP-1 gene, a very similar sequence from the same locus was isolated and named HIRA (2 ). HIRA is the major and TUPLE1 a minor splice variant of transcripts from this locus (3 ); we will refer to the gene as HIRA (human) and Hira (mouse) in recognition of this fact. HIRA is most closely related to the yeast and HIR2 histone gene transcriptional regulators, with 47% sequence similarity at the protein level.
HIRA was isolated by positional cloning from the region of chromosome 22q11 which is deleted in DiGeorge syndrome (DGS), velo-cardio-facial syndrome (VCFS) and related birth defects (1 ,2 ,4 ). The wide spectrum of abnormalities seen in association with hemizygosity of 22q11 prompted the proposal that the acronym CATCH22 be used as a mnemonic for the major pathological features which are found in various combinations in patients carrying the deletion [Cardiac defect, Abnormal facies, Thymic hypoplasia, Cleft palate, Hypocalcaemia, deletion of the DGS critical region (DGCR) at 22q11]. As HIRA mapped to the DGCR we proposed that HIRA might have a haploinsufficient effect which contributes to these phenotypes (1 ).
Further molecular mapping of the DGCR is consistent with a role for HIRA in the syndrome. One patient, ADU, with a balanced translocation and DGS has been described (5 ), and the sequence disrupted by the breakpoint has been detemined (6 ). Although the breakpoint disrupts an open reading frame (ORF) called DGCR3, no gene has been isolated encoding this ORF, and no mutations within the ORF have been reported. A transcript disrupted by the ADU breakpoint and denoted DGCR5 has been described. However, this gene appears not to code for any sizeable protein product, and no mutations of the gene have been detected (8 ). Recently a DGS patient was discovered to have an interstitial deletion which did not include the DGCR3 ORF, i.e. the sequences disrupted by the ADU rearrangement map outside the SRO, ~100 kb proximal to it (7 ). It now seems likely that the balanced translocation exerts a position effect on an adjacent gene or genes (8 ,9 ). The situation is further complicated by the description of a second SRO located distally within the frequently deleted region (10 ). Although no HIRA point mutations have yet been found in cases of DGS or VCFS without a 22q11 deletion (unpublished data; U. Atif and J. Goodship, personal communication) HIRA is hemizygous in all patients with DGS and a deletion encompassing the proximal SRO. Therefore, HIRA remains as a candidate for contributing to the haploinsufficiency secondary to hemizygosity within 22q11.
In summary, HIRA maps within the DGCR and represents a candidate for a gene having a haploinsufficient effect contributing to DGS. We examined the expression pattern of Hira during chick embryogenesis and show that the gene is expressed in a temporally and spatially specific pattern during early development, with high levels of expression in the neuroepithelium during neurulation.
One million clones from the chick embryo library were screened with the mouse MF2 cDNA clone and three positive clones obtained. The largest clone, CT3, was 4.2 kb in length and this clone was analysed in more detail. The cDNA sequence is deposited at EMBL with accession number X99375. The cDNA contained 60 bp of 5' untranslated sequence, with an initiator methionine at a position conserved between the vertebrate Hira genes. The CT3 transcript encodes a protein of 1018 amino acids which has strong sequence similarity with its human and murine counterparts (Fig. 1 a). The program GAP indicated 84% amino acid identity and 91% similarity with the mammalian Hira proteins. Database searches revealed previously reported Hira matches with WD40 domain-containing proteins, including the HIR transcriptional regulators and TUP-1. However, at both the nucleotide and protein levels a match was also obtained with the chromatin assembly factor p60 subunit (CAF1) (15 ). The alignments of the three vertebrate Hira genes, a possible Caenorhabitis elegans homologue, CAF1 and esc are shown in Figure 1 a. esc is the Drosophila melanogaster gene extra sex combs, a member of the polycomb group of proteins. It contains WD40 repeats and is thought to mediate transcriptional repression through its effects on chromatin assembly. Chick Hira, Chira, demonstrates 50% sequence similarity with the C.elegans protein (29% identity over the length of the alignment), 45% similarity with CAF1 (27% identity) and 44% similarity (17% identity) with esc.
Northern. When hybridised to mRNA from stage 10-12 and day 4 chick embryos Chira detected a major transcript of 4.5 kb, and a less abundant transcript of ~5 kb (Fig. 1 b). This is similar in size to the most abundant transcript observed in both human and murine tissues (1 ,15 ).
The Chira gene encodes an 1018 amino acid protein containing seven WD repeat motifs at the N-terminal end of the coding sequence. Alignment of three vertebrate Hira proteins with yeast HIR1 and HIR2 proteins confirms that Hira homology is greatest with the N-terminal region, of HIR1 and the C-terminal region of HIR2 as previously reported by Lamour and coworkers (2 ). The positions of the WD40 domains, the putative nuclear localization signals, and the QQQQQXXQ motifs previously identified in Hira (1 ,2 ) were all conserved in the chick. Database screening identified a possible homologue of Hira in the C.elegans genome. Overall sequence identity was 29% after alignment, with similarity at 50%. The C.elegans protein is larger with a predicted 141 amino acids N-terminal to the position corresponding to the vertebrate initiator. Although sequence similarity is greatest over the WD40 region the vertebrate proteins have homology to the C.elegans protein over their entire length. However, the QQQQQXXQ motif does not appear in the C.elegans protein. The unique N-terminal extension of the C.elegans gene product does not contain additional WD40 domains, but has strong sequence similarity to the yeast SDS22 protein. SDS22 is a leucine-rich regulator of protein-phosphatase 1 activity during the cell cycle (16 ).
There are several lines of evidence indicating that DGS is likely to be a consequence of a deficient contribution of the neural crest during development. The neural crest is a population of cells present in vertebrates which migrates extensively throughout the embryo and makes a contribution to the development of numerous neuronal and non-neuronal structures (17 ). Studies of neural crest migration in the chick and the mouse (18 -20 ) show that the rostral neural crest forms much of the developing head and branchial arches, and differentiates into mesenchymal cells (the ectomesenchyme). Using the chick-quail chimera system, it was shown that neural crest at the level of somites 1-3 migrates to the aorticopulmonary septum (21 ). Removal of the neural folds of chick embryos at this level prior to the onset of crest migration can result in malformations such as persistent truncus arteriosus, double-outlet right ventricle (DORV), overriding aorta, and transposition of the great arteries (22 ,23 ). Other defects associated with ablation of neural crest rostral to the level of somite 5 include absence or hypoplasia of the thymus and parathyroid glands (24 ). More restricted removal of premigratory neural crest tends to give the less severe outflow tract defects e.g. DORV and ventricular septal defect (VSD) (25 ). These features can be considered as an experimental phenocopy of DGS. It is particularly striking that the characteristic outflow tract defect seen in the neural crest ablation experiments are also those typical of DGS cases, but rare in the overall population of patients with congenital heart defects (26 -28 ).
It has been proposed that at least part of the positional information carried by migrating neural crest cells is determined by the so-called Hox code `address' given by the combinatorial expression of genes found in the homeobox gene clusters (29 ,30 ). It is intriguing that mice homozygous for a targeted null mutation of the Hoxa3 gene have many features seen in DGS and the neural crest ablation experiments: absence of the parathyroids and thymus, reduced thyroid gland, and an abnormal carotid artery (although the classical outflow tract defects were not seen) (31 ).
Our data demonstrate that Chira is expressed from gastrulation stages onwards with transcripts predominantly localised to the neuroepithelium during neurulation. Expression within the neuroepithelium at all axial levels becomes a major feature of the subsequent expression pattern. Relatively high levels of Chira expression are observed in regions of the embryo containing migrating neural crest cells, particularly those emanating from rhombomeres 4 and 6 and best seen at stage 12. Chira is expressed throughout the rostral mesenchyme, but with higher levels of expression seen in regions known to contain neural crest cells e.g. in the mesenchyme at the neuroepithelial and ectodermal border and in the subectodermal regions of the cranial mesenchyme.
At later stages strong expression is seen in the neural crest derived regions of the head, the branchial arches and the pharyngeal pouches. This expression is intriguing given that, as described above, the main structures affected in DGS arise from the branchial arches and pharyngeal pouches 3 and 4. Limb buds also express Chira and minor limb abnormalities are often a feature of VCFS.
Another interesting element of the expression pattern is the up-regulation of expression detected at rhombomere boundaries from around stage 16. As the rhombomere boundaries are defined by stage 12 it is unlikely that Chira has a role in establishing the boundaries, but it may have a role in their subsequent maintenance. The cells at these boundaries have special characteristics including temporal and spatially distinct expression of developmentally important genes. For instance, the pattern of Chira expression at the rhombomere boundaries is similar to that seen with Pax6 (32 ). The position of the rhombomere boundaries are known to coincide with the site of development of axon tracts and with glial cell precursors but the ultimate fate of these boundary cells remains unknown (32 ).
A number of developmentally important genes, e.g. Pax3, Pax6, Pax7 and Shh, have spatially restricted domains of expression within the neural tube. While Chira was expressed in the neural tube, where the tube was open we did not detect Chira expression in the ventral midline of the neural tube and the underlying notochord. This suggests that Chira expression is responding to dorso-ventral patterning signals within the neural tube.
It should be pointed out that both the full length probe used for northern analysis and the 5' probe used in the whole mount analysis should detect transcripts corresponding to Tuple1 (1 ), and the 2.3 kb transcript detected by Wilming and colleagues (1 ). However, at the level of northern analysis, no such corresponding transcripts were observed.
The WD-containing protein class is confined to eukaryotes. The conserved unit is usually repeated 4-8 times and each repeating unit usually ends with the sequence Trp-Asp (WD). This region folds into a propellor-like shape which is thought to be invovled in protein-protein interactions (33 ,34 ). It therefore seems likely that Hira will be a component of one or more multimeric structures. The strong conservation of the vertebrate Hira family to the two yeast proteins and HIR2 noted previously (2 ) highlights possible functional domains. In particular the strong conservation with HIR1 and HIR2 raises the possibility of a conserved function of Chira in the regulation of histone synthesis. However, the dynamic pattern of Chira expression revealed here suggests that the gene is involved in more than the regulation of histone gene transcription. The homology to the yeast histone regulators, TUP-1 and CAF1 suggests that Hira may be involved in regulation of chromatin structure. TUP-1 is known to bind directly to histones H3 and H4 (35 ) and CAF1 assembles histone H3 and acetylated H4 onto newly synthesized DNA (36 ). Alterations in local chromatin structure provides one mechanism for the maintenance of gene expression pattern over many cell generations, as shown by the action of esc. Recent data suggests Hira does indeed interact with chromatin components, in particular histone H2B (Lipinski, personal communication).
The chick homologue of HIRA, Chira is expressed in a dynamic pattern during early development. Transcripts are found in a number of tissues, including those which give rise to the structures affected in DGS. A similar expression pattern is seen in mouse embryos (unpublished data, and 1 ). Given the localization of HIRA within the DGCR, the expression pattern of Chira and Hira, and the sequence homology to general transcriptional repressors, HIRA remains an intriguing candidate for making at least some contribution to the haploinsufficiency resulting in DGS. Further studies such as the production of null mutants in mice will help to elucidate whether Hira has an essential role during embryological development and whether that role is mediated through effects on local chromatin structure.
A stage 12-15 chick embryo library in [lambda]ZAPII was kindly provided by Dr D. Wilkinson and screened with the Hira clone MF2 (1 ) using standard methodology. Inserts from positive clones were in vivo excised according to the manufacturer's instructions (Stratagene). The cDNAs were subcloned into M13mp18 and mp19, and sequenced manually (Sequenase II kit, Amersham International) using T3 and T7 primers initially, and subsequently with sequence-derived primers. This sequence was supplemented with double strand sequence obtained on an ABI 377 automated sequencer. The GCG package of sequence analysis programs available at the Human Genome Mapping Resource Centre (HGMPRC) was used to derive the predicted protein sequence. Alignments were conducted using Bestfit, Gap and Clustal at the same facility.
Fertilised eggs were obtained from Polyndon Egg Farm, Hertfordshire and incubated at 37oC to give chick embryos of Hamilton and Hamburger (11 ) stages 4-26. Embryos were dissected out into ice-cold phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde/PBS at 4oC overnight.
Digoxigenin labelled riboprobes were made using a cloned 382 bp PCR fragment from the 5' end of the chick Hira cDNA as a template (from nucleotide 187); all figures show results obtained with this probe. Full length and 3' untranslated region antisense probes gave the same results (data not shown). In situ hybridisation for stages 4-12 was carried out according to the CHAPS modified version of a previously described method (12 ). Embryos were hybridised with 1 mg/ml of probe overnight and washed at 55oC. After the colour reaction had been stopped embryos were fixed in 4% paraformaldehyde/PBS, washed in PBS and photographed on an SMZ-U Nikon microscope. Older chick embryos were hybridised using a further modification of the protocol above, in which Triton X-100 was substituted for Tween 20 and hybridisation took place at 65oC.
Embryos were cut into approximately three somite sized sections and the flat surfaces photographed using a dissecting microscope as before.
Total RNA and polyadenylated RNA were isolated as previously described (13 ) from stage 10-12 and day 4 chick embryos. RNA was electrophoresed and blotted using standard methods (14 ). 32P-labelled riboprobes were made from the chick Hira cDNA clone CT3 template and hybridised to the filters in 60% formamide at 65oC and washed to 75-80oC in 0.1% SSC. Filters were exposed to X-ray film or imaged on a phosphorimager (Molecular Dynamics).
We are grateful to the Meijers and Lipinski laboratories for useful discussion and sharing of unpublished data. The Medical Research Council, British Heart Foundation and the Child Health Research Appeal Trust provided generous support.
Human Molecular Genetics Pages
©
Introduction
Results
Isolation of chick Hira
Expression of Chira in the chick embryo
Discussion
Chira is strongly conserved with its murine and human counterparts
Expression of Chira, and its possible relevance to DGS
Implications of sequence and expression of Chira
Conclusions
Materials And Methods
Cloning of chick Hira
Embryo collection
Whole mount in situ hybridisation
Sections of whole mounts
Northern blots
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Fri Jan 31 11:45:11 GMT 1997. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996