The X inactivation centre (Xic) is a cis-acting locus thought to play a key role in the initiation of X-inactivation. We have cloned and characterized a new gene, Bpx, lying distal to the murine Xist. Bpx, which is specifically expressed in the brain,shows strong homology to genes encoding nucleosome assembly proteins and is normally X-inactivated in mice. Isolation and localization of BPX, its human homologue, has shown the gene to be located centromeric to XIST in man. The Xq13 region, whose orientation is apparently globally conserved between man and mouse, must therefore contain an inversion of at least 600 kb spanning the XIST sequence and including the CDX4 and BPX genes.
X chromosome inactivation is the process which ensures the dosage compensation of X-linked gene products between male and female mammals. X-inactivation occurs early in development and is first seen in the mouse in extraembryonic tissues where the paternal X chromosome is preferentially inactivated. In embryonic tissues on the other hand, inactivation affects either of the X chromosomes in a random fashion. X-inactivation can be arbitrarily divided into three phases: initiation, spreading and maintenance. The presence of a cis-acting locus, the X inactivation centre (Xic), thought to act during the initiation step, is required for X-inactivation to occur (for recent review see ref. 1 ).
The existence of Xic and its localization have been deduced from studies on the inactivation status of the X chromosome in cells carrying structurally abnormal X chromosomes (2 ,3 ). In the mouse, Xic candidate region has been localized between the T16H translocation breakpoint (4 ) and a breakpoint characterized in the embryonic stem cell line HD3 (5 ). The existence of a larger number of X chromosome rearrangements in man than in mouse has allowed the present size of the candidate region for XIC in man to be restricted to between 700 and 1200 kb. The proximal boundary of XIC in man is defined by the breakpoint of an X;14 translocation (6 ) whilst the distal boundary is currently defined by the breakpoint of a late-replicating rearranged X chromosome [46,X rea(X)] (7 ). Since the Xic/XIC candidate regions map to syntenic regions of the mouse and human X chromosome (8 ), it is generally thought that the more precisely defined limits of the XIC candidate region established in human will prove valid in the mouse.
The Xic/XIC candidate region is as yet poorly characterized for the presence of transcription units, although Lafrenière and colleagues have reported a novel gene on the human X, XPCT that lies 600 kb distal to XIST (9 ). Within the XIC/Xic candidate region itself, only two transcribed sequences have been characterized. One of these is the XIST/Xist gene (10 -12 ). The XIST/Xist gene possesses several characteristics which make it a potential candidate for the X inactivation centre (13 -15 ), including its unique property of being transcribed solely from the inactive X chromosome. A second gene lying within the XIC/Xic candidate region, CDX4/Cdx4, has been recently isolated (16 ).
As part of our ongoing studies of the murine Xic candidate region, we report the identification of a brain specific gene, termed Bpx, lying ~300 kb distal to Xist, which is normally X-inactivated in mice. Bpx, which shows striking homology to genes encoding nucleosome assembly proteins, is highly conserved between man and mouse. The human homologue has been mapped at least 600 kb proximally to XIST on the human X chromosome, indicating that the inversion of this region of the human and murine X chromosomes, previously found to involve both XIST/Xist (17 ) and CDX4/Cdx4 (16 ), includes Bpx and covers at least 300 kb in mouse and 600 kb in human.
In order to define the intron-exon structure of the gene, three primers pairs regularly dispersed along the cDNA sequence were used to amplify in parallel the cDNA and genomic DNA (see Material and Methods). No differences in size between cDNA and genomic DNA-PCR products were detected on 1.5% agarose gels (data not shown), indicating that no large introns are present in the gene. To discriminate between the complete absence of introns and the presence of very small introns, genomic DNA was amplified using primers which covered the whole cDNA and the amplification products sequenced. No differences were found between the genomic and the cDNA sequences (data not shown). These results indicates that Bpx is an intronless gene.
Bpx was known to be located ~30 kb from the PA-2 left YAC arm (TRP1) (17 ). The orientation of the Bpx gene on this YAC was determined by partial restriction mapping using the enzyme SalI, as a SalI site was known to be present 356 bp from the 5' end of the cDNA (Fig. 1 a). YAC PA-2 DNA was partially restricted using SalI, resolved by PFGE, blotted and Southern analysis performed using Bpx specific probes locating either 5' or 3' to the SalI site. As shown in Figure 2 , the smallest restriction fragment recognised by both the 3' Bpx probe and a probe specific for the left arm of the YAC is of smaller size (30 kb) than the smallest fragment recognised by both the 5' Bpx probe and the YAC probe (80 kb). This data indicates that the orientation of Bpx relative to YAC PA-2 is URA3 arm-5'Bpx3'- TRP1 arm. Since the orientation of YAC PA-2 relative to the X chromosome is known to be centromere-URA3/TRP1-telomere (19 ), the Bpx orientation can be established as centromere-5'Bpx3'-telomere, identical to that of the Xist gene.
When probed against a Northern blot containing multiple adult male tissues (Clontech), the P0324 cDNA probe detects a single 2.5 kb transcript in brain RNA (Fig. 3 ), as does the equivalent genomic probe IB8Xba (see above). To confirm the tissue-specificity of Bpx expression, brain, heart, spleen, liver, kidney and testis from adult male and female, 129/Sv mice were tested by RT-PCR using a set of primers (P0324S4 and P0324S5) located on either side of the stop codon. A PCR product of the expected size (147 nt) was only detected in both male and female brain samples. These results confirm that, in adult mice, Bpx is specifically transcribed from the brain of both males and females.
Comparison of the deduced amino acid sequence of Bpx with those in GenBank revealed striking similarities with murine and human nucleosome assembly proteins (NAP-1) (20 ), with the mouse brain protein DN38 (21 ) and, to a lesser extent, with the yeast NAP-1 protein (22 ). The homologous residues are clustered in two domains (the first extending from residues 98 to 174 and the second from residues 230 to 418). Within these regions, the amino acid similarity between Bpx and mouse NAP-1 reaches 77% (Fig. 1 b) and several distinct motifs found in hNRP(20 ), the human NAP-1 protein and NAP-1 from other species, are found to be shared by Bpx(Fig. 1 b). Sequences DSFFNFFSPH and KGIPDFWLT, which are highly conserved between yeast and mammalian NAP-1 proteins and probably represent functionally important domains, are for instance conserved in Bpx (238-246 and 368-377). Glycosylation may influence Bpx function since one putative site for N-glycosylation (at positions 339-341), present in hNRP, is found in Bpx. Other similarities include the presence of a threonine residue (position 162) and a serine residue (position 310) at consensus sites for phosphorylation by casein kinase II. This conservation is of interest as recent data suggest that the phosphorylation status of nucleoplasmin, an enzyme which also plays a role in nucleosome assembly, can affect its role in modulating chromatin structure (23 ). Such similarities are reinforced by the presence in the Bpx sequence of a IKKQRH motif (at position 346-352) which closely resembles the nuclear localization motif found not only in nucleosome assembly proteins such as nucleoplasmin and N1, but also in other nuclear proteins, including p53, c-abl and simian virus 40 T-antigen. A bipartite nuclear localization consensus motif (24 ) is found at position 138, which is however less conserved in hNRP.
Another similarity between the Bpx and NAP-1 proteins lies in its overall acidic nature, with 15% glutamic acid residues and 9.3% aspartic acid residues, some of which are clustered (see Fig. 1 b, from residue 212 to 222). These polyacidic stretches would be expected to facilitate the binding of the molecule to histones or other basic proteins. However, although both NAP-1 and Bpx proteins are acidic overall, the acidic domains map in part outside the regions of homology of these proteins. Of the three acidic domains in the hNRP protein (between amino acids 13 and 26, 129 and 141 and 348 and 376), only the second is located within the first domain of homology with Bpx. Similarly, a particularly acidic domain is found in the middle of the Bpxsequence (4 Asp, 7 Glu), in a region which has no homology to either the nucleosome assembly proteins or to any other protein. The acidic nature of Bpx is reflected by its hydropathicity profile, indicating that Bpx encodes a highly hydrophilic protein.
A further similarity between Bpx and hNRP lies in their 3' untranslated region. Both are AT-rich and large consisting of 309 nucleotides in hNRP and 755 in Bpx. These AT-rich 3' untranslated regions are thought to influence transcript stability and half-life (25 ).
Outside of the two domains of Bpx which have homology to NAP-1 proteins (from residues 98 to 174 and 230 to 418), no significant homology with any known protein was found in the databases. The other, more specific regions may be implicated in the restricted expression profile of Bpx in the brain.
Female mice carrying the T(X;16)16H translocation undergo non-random X-inactivation due to cell selection. In mice carrying the T16H translocation (derived from Mus musculus domesticus) and a normal X chromosome derived from the divergent mouse sub-species Mus musculus musculus (XMai), genes subjected to X-inactivation are expressed only from the domesticus X chromosome (26 ). Sequence analysis of the Bpx gene in the T16H and Mai strains permitted detection of an A to G transition in Mai mice, two nucleotides downstream of the stop codon. This single nucleotide polymorphism could be detected by single strand conformation polymorphism (SSCP) analysis, after PCR amplification using primers P0324S4 and P0324S5. To determine whether Bpx is normally inactivated in mice, brain RNA from a T16H male, a Mai female and a (T16H/Mai)F1 female were reverse transcribed and the corresponding cDNAs examined by SSCP analysis (Fig. 4 ). As shown on the right part of Figure 4 , only the T16H allele is amplified in the T16H/Mai heterozygous female. Control experiments using Xist primers have shown that, as expected, only the Mai allele of Xist is amplified in the T16H/Mai female (data not shown). Thus the Bpx gene appears to be, at least in the mouse, subject to X-inactivation and transcribed only from the active X chromosome.
When a human Northern blot was screened with the mouse P0324 probe, a single 2.5 kb band was detected specifically in the brain (data not shown). A human brain cDNA library was screened with probes corresponding to different murine cDNA clones spanning the entire cDNA sequence, allowing the sequence of BPX to be established (Fig. 1 c). The Bpx and BPX sequences are highly conserved, with 79.8% identity over the whole cDNA sequence. Within the ORF, the homology is 84.6% at the nucleotide level. At the amino acid level, the deduced proteins are 82.5% identical and 91.5% homologous. A stretch of nine GAA/GAG triplets coding for glutamic acid is found in the middle of the ORF. This region is not perfectly conserved in the murine cDNA which has only seven GAA/GAG. As trinucleotide repeats have been shown to be unstable in the human genome and are associated with several diseases, we analyzed this region for potential polymorphisms using DNA from 12 unrelated individuals. This analysis did not show any length variation in the repeated region (data not shown).
The Xist transcription unit was shown sometime ago to be oppositely orientated in man and mouse (17 ). More recently, Horn and Ashworth have reported that the Cdx4 gene, which lies 100 kb distal to Xist in mouse, is located 400 kb proximal to XIST on the human X chromosome (16 ). We were therefore interested in the localization of the human homologue, BPX. BPX was initially shown to map on the human X chromosome by the use of a hybrid containing the human X chromosome on a mouse background (data not shown). To refine the localization of BPX, a human YAC contig spanning 1.8 Mb around XIST was then screened with the BPX probe (27 ). The results are shown in Figure 5 . YACs 4566, 8028, 8038, 4565 and 8039 were shown to hybridize with the BPX probe, whereas YACs 8027, A39G7, B245H8 and 8012 remained negative. These results indicate that BPX is located 600-800 kb proximal to XIST on the human X chromosome. Thus, the size of the inversion between man and mouse is at least 300 kb in mouse and 600 kb in human.
The Xic candidate region remains poorly characterized for the presence of transcription units. Apart from Xist, only one other gene, Cdx4, had been found to map within the Xic region (16 ) prior to this report on the identification and characterization of a new brain specific gene, Bpx, lying 300 kb distal to Xist in the mouse.
The Bpx cDNA encodes a predicted protein which shows homology to nucleosome assembly proteins (NAP-1). Like the NAP-1 proteins, Bpxis highly acidic, containing 24.3% of negatively charged amino acids (aspartic and glutamic acids). The overall acidic nature of Bpx, as well as the presence of two nuclear address motifs provide arguments supporting the hypothesis that Bpx is a nuclear protein. The homology with the NAP-1 protein indeed suggests that Bpx may interact with core histones in the nucleosome.
Whilst the presence of a gene coding for an eventual nucleosomal protein lying within the Xic candidate region is highly provocative, Bpx appears to be restricted in its distribution in the adult and in the embryo to postmitotic neurons (Rougeulle et al., in preparation). The distribution of Bpx is in contrast to that of hNRP, which is ubiquitously expressed and is thought to play a role in cell proliferation. Rather than having a ubiquitous role in nucleosome assembly during DNA replication, the histone interaction of Bpx could therefore be involved more in local chromatin modifications that might be implicated in the regulation of gene expression specific to neural function. It is of interest however to note that the Drosophila scute gene, which plays an important role in the development of both the embryonic central and peripheral nervous system of larvae and adult flies, is also involved in sex determination and dosage compensation (28 ). Indeed, most of the genes involved in chromosome counting in Drosophila have been shown to play an important function in neurogenesis (29 ). Outside of those domains showing homology to NAP-1, the unique regions of the Bpx protein may be involved in its functional specificity via interactions with other proteins or DNA which permit the specific regulation of the expression of genes or domains by modifying chromatin structure.
Sequencing of the Bpx cDNA and the corresponding genomic DNA has revealed that Bpx is an intronless gene. Several characteristics indicate however that Bpx is a functional gene rather than a pseudogene. First, at high stringency, a single band is detected by Southern blot analysis on mouse genomic DNA. Secondly, Bpx is highly conserved in man, with the human homologue mapping to the syntenic region of the human X chromosome. Thirdly, when Bpx expression was assessed by SSCP in T16H/Mai heterozygote females, the T16H derived allele and only the T16H allele was detected, indicating that the gene located on the X chromosome is indeed expressed. Taken together, these data strongly suggest that the Bpx gene is a functional gene rather than a pseudogene. Bpx is not the only intronless gene known to lie on the X chromosome. Other X-linked intronless genes such as the brain specific POU-box gene Brn4 (30 ), the glutamate dehydrogenase 2 gene GLUD2 (31 ), a member of the glycerol kinase gene family (32 ) and the MYCL2 gene (33 ) have been previously reported in man and mouse. The genomic organization of the NAP-1 related genes has not been extensively studied and the presence of introns in these genes has yet to be determined. It is however of interest to note that the gene family coding for histone proteins is also intronless. Whilst the origin of the absence of introns in Bpx remains to be determined, assuming that primordial genes evolved with introns, the lack of introns in genes such as Bpx in higher eukaryotes must be due either to intron loss or to reinsertion into the genome of genetic material copied from mRNA by reverse transcription.
We have shown by SSCP analysis using the T16H translocation system, that Bpx is subject to X-inactivation in mice. It is the second gene with Cdx4, located within the present Xic candidate region, that has been shown to be normally X-inactivated (16 ). This result supports the hypothesis that the Xist expression profile, that is exclusive expression from the inactive X, is either specific to Xist, or to a highly restricted chromosomal domain including Xist. It will be of great interest to determine whether genes located proximally and close to Xist show a similar profile.
Sequence comparison of Bpx and its human homologue BPX has revealed striking homology, with 79.8% identity at the nucleotide level for the entire cDNA, which contains >1 kb of untranslated sequences. Since the homology within the ORF is 84.6%, this suggests that the 5' and 3' untranslated regions are themselves highly conserved. These regions may well be involved in the regulation of expression and/or stability of the mRNA. PCR analysis of both BPX genomic DNA and cDNA using primers which cover the whole cDNA sequence also suggest that BPX, like its murine homologue, is an intronless gene. Of note is the presence of a stretch of nine GAA/GAG triplets coding for glutamic acid found in the middle of the ORF. Such amino-acids homopolymers (poly-glu in Bpx/BPX case) are frequently found in DNA binding proteins which act as transcription factors. Such repeats have been found in the NH2 terminus of the Xnp protein which is similarly encoded by a gene lying slightly distal to this region of the X chromosome (34 ).
Using a YAC contig spanning 1 Mb on the human X chromosome (27 ), BPX has been localized 600-800 kb proximal to XIST (Fig. 5 ), suggesting that the inversion between man and mouse in this part of the X chromosome covers at least some 300 kb in mouse and 600 kb in man. It is likely therefore that BPX maps to within the same region as the proximal limit of the XIC candidate region (6 ), although testing against the X;14 translocation breakpoint carried by the hybrid W4-1A will be necessary to confirm this interpretation. If we assume, since the X-inactivation processes are similar in human and in mouse, that the limits currently defined for the human XIC are valid in mouse, then the distal limit of the actual murine Xic candidate region would be located some 300 kb distal to Xist, very close to the Bpx gene itself. A striking feature is the difference in the physical distance between Bpx and Xist in the mouse and BPX and XIST in human, of 300 and 600 kb respectively. Such a difference has already been reported for Cdx4, which is located 100 kb distal to Xist in mouse and ~400kb proximal to XIST in human (16 ). Thus, some 300 kb of the human X chromosome, lying between XIST and CDX4, seem to be absent from the murine X, suggesting that these 300 kb are not necessary in the X-inactivation process, or at least for Xic function itself. Whilst the identification of further conserved transcription units lying close to Xist will be necessary to allow the complexity of this chromosomal rearrangement to be verified, the possibility exists that the inversion is associated with the generation of the large (125 kb) inverted duplication previously identified by Lafreniere et al. (27 ) as lying between PHKA1 and DXS227. In this case, the inverted duplication may well represent the proximal boundary of the inversion.
Several human inherited diseases have been localized to the Xq13 region of the human X chromosome. Amongst these are genes involved in mental retardation. Schwartz has noted for example that 35% of mapped mental retardation genes on the X chromosome were linked to markers in the region Xq12-q21 (35 ). Further studies should determine whether BPX is involved in human diseases which have been mapped to the Xq13 region. The localization of BPX in Xq13, as well as its restricted expression to neurons makes it a good candidate for involvement in brain disorders such as mental retardation.
Genomic probe IB8Xba was used to screen 20 000 clones from a gridded adult mouse brain cDNA library kindly provided by Hans Lehrach. The P0324 cDNA probe derived from the clone isolated, which contains 1.3 kb of the 3' cDNA sequence, was used to screen an adult mouse brain 5'-stretch plus cDNA library in the [lambda]gt11 vector (Clontech) and a human brain cDNA library in the [lambda]ZapII vector (Stratagene). 5 * 105 recombinant clones from each library were plated onto 210 mm plates at a density of 105 phages per plate and colony lifts were taken with Hybond N+ membranes.
Double stranded DNA probes were labeled by random priming (36 ).
High stringency hybridizations were carried out with 0.5 M sodium phosphate, pH 7.2, 1 mM EDTA and 7% SDS at 65oC, modified from Church and Gilbert (37 ). Filters were washed in 40 mM sodium phosphate, pH 7.2, 1% SDS at 65oC. Low stringency hybridization were carried out with 25% deionized formamide, 5* Denhart, 5% dextran sulfate, 1% SDS, 50 mM sodium phosphate, pH 7.2, 5 mM EDTA, 1 M NaCl and 100 [mu]g/ml sonicated salmon sperm, at 42oC for 16 h. Filters were washed in 25% deionized formamide, 1% SDS, 50 mM sodium phosphate, pH 7.2, 5 mM EDTA, 1 M NaCl, at 42oC.
Northern blot hybridizations were carried out using the Express-Hyb solution provided from Clontech, as recommended by the manufacturer.
Double strand plasmid clones were sequenced using the Sequenase version 2.0 sequencing kit. Search for homologous sequences was carried out using the BLAST server (non-redundant DNA and protein sequence databases) at the National Center for Biotechnology Information (NCBI).
RNA from brain, heart, liver, spleen, kidney and testis were prepared according to the method described by Chomczynski and Sacchi (38 ). First strand cDNA was synthesized from 10 [mu]g of total RNA using the AMV reverse transcriptase (Promega). One-thirtieth of the cDNA was used for PCR reactions using primers P0324S4 (ATTATTTATGATGATTGGATGC) and P0324S5 (TGCTTATGTCTTCAGTAACTGC). PCR reactions were performed in 50 [mu]l containing 0.5 [mu]M of each primer, 10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl2, 1.25 mM each dNTP and 2.5 U of Thermus Aquaticus Polymerase (Amersham). Fragments were amplified under 50 [mu]l of mineral oil in an automated thermal cycler using a denaturation step of 5 min at 94oC followed by 30 cycles consisting of a 94oC denaturation (30 s), a 52oC annealing (30 s) and a 72oC extension (1 min) step completed by a final 5 min extension step at 72oC. The integrity of all RT-products was controlled by performing PCR using primers for the houskeeping gene Hprt HPRTNAF/HPRTNAR (15 ).
PCR reactions were carried out as described above (except for annealing temperatures) on 300 ng of genomic DNA and cDNA, using the following primers.Bpx analysis. P0324S1 (ACACAGGACAATATTTGCAAC)/ P0324S2 (CACTGGCTGCGACATAGA): 50oC.
MB1b2KS0 (GAAAGTCTCCCAGTGAAAGTT)/
P0324S3 (CACTTCTAATAATATTCACAGCC): 52oC.
MB3bT7-1 (CCCACTGCTTGCTGCC)/MB1b2inv (AACTTTCACTGGGAGACTTTC): 50oCBPX analysis. HB1aT3-1 (TGATTAAGAAATATGATGAGCC)/
HB1eT3-1 (ACTAACACTTCACCGACAAAC): 50oC.
HBR1 (GCTGCTTGCGGAGGAGG)/HB4bT3-2 (CTTTCACAGGTAGACTTTCAACA): 57oC.
HBr5 (GGACTTATCGGTTATGTTTTAGA)/HB1aj (CAGTTAGCCAAAAATCAGGA): 52oC.
YAC PA-2 DNA was PCR amplified using a denaturation step of 5 min at 94oC followed by 30 cycles consisting of a 94oC denaturation (30 s), a 55oC annealing (30 s) and a 72oC extension (2 min) step completed by a final 10 min extension step at 72oC, using the following primers.
MBr1 (CACCAGTTCCTCCATCCC)/MB1b2inv
MB3bT7-1/P0324S5 (GCTTATGTCTTCAGTAACTGC)
P0324S4/MBr2 (CAGTGAATGAACAGAATATGAAAT)
PCR products were cloned into the pGEM vector (Promega) according to the instructions of the manufacturer. Clones were double strand sequenced as described above.
First strand cDNA was synthesized from 5 [mu]g of total brain RNA from the T16H, Mai and (T16H/Mai)F1 strains as described above. One twenty-fifth of the cDNA was used for PCR amplifications using primers P0324S4 (ATTATTTATGATGATTGGATGC) and P0324S5 (TGCTTATGTCTTCAGTAACTGC). Labelled PCR was performed in a 10 [mu]l reaction containing 1 [mu]l of 10 * PCR buffer (Amersham), 100 ng genomic DNA or 2 [mu]l cDNA, 0.25 mM primers, 200 mM dNTPs, 0.1 [mu]l [[alpha]-32P]dCTP (3000 Ci/mmol, 10 mCi/ml, ICN). PCR involved 30 cycles at 94, 52 and 72oC for 1, 1 and 0.5 min respectively. Nine microlitres of the reaction was mixed with 9 [mu]l of 95% formamide, 20 mM EDTA 0.05% bromophenol blue and 0.05% xylene cyanol, denatured at 95oC for 5 min and cooled on ice. Two microlitres were loaded on 5% acrylamide gels containing 5% glycerol at 4oC in 0.5 * TBE for 4 h at 30 W. Gels were dried and exposed for 1-5 days at -80oC.
The Xist primers cXU8 (CAAAAACCTCTGCACTGGAA) and cXL8 (TGAGGGTATGGGATCTTGGTT) were used as control using the same conditions as those for Bpx primers.
The authors thank the `Unité de Génétique Moléculaire Humaine' for providing the human brain cDNA library. This work was supported by grants to P.A. from GREG (68/94), ARC and AFM. C.R. was supported by fellowships from ARC and AFM.
jnl.info{at}oup.co.uk
Human Molecular Genetics
Pages
Introduction
Results
Genomic organization of the Bpx gene
Bpx expression profile
Bpx has homology with nucleosome assembly proteins
Bpx is subject to X-inactivation
Cloning and localization of BPX, the human homologue of Bpx
Discussion
Materials And Methods
Probe labelling and hybridizations
Sequencing and sequence analysis
RNA isolation and RT-PCR
Analysis of exon-intron strucure of Bpx/BPX.
Sequencing of Bpx genomic region
SSCP analysis
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Thu Oct 31 15:08:29 GMT 1996. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996