DDBJ/EMBL/GenBank accession no. U89331
The abnormalities seen in Turner syndrome (monosomy X) presumably result from haploinsufficiency of certain genes on the X chromosome. Gene dosage considerations lead to the prediction that the culpable genes escape X inactivation and have functional homologs on the Y chromosome. Among the genes with these characteristics are those residing in the pseudoautosomal regions (PAR) of the sex chromosomes. A pseudoautosomal location for a dosage-sensitive locus involved in stature has been suggested based on the analyses of patients with deletions of a specific segment of the short arm PAR; hemizygosity for this putative locus probably also contributes to the short stature in Turner individuals. We have isolated a gene from the critical deleted region that encodes a novel homeodomain-containing transcription factor and is expressed at highest levels in osteogenic cells. We have named the gene PHOG, for
Turner syndrome is a common developmental disorder in females, characterized by four cardinal features: (i) short stature, (ii) ovarian failure, (iii) a variety of major and minor malformations and (iv) a high rate of embryonic lethality (1 ). It is classically caused by complete monosomy for the X chromosome (a 45,X karyotype; ref. 2 ), although certain X chromosome structural abnormalities that result in partial X monosomy can also give rise to the Turner phenotype (1 ,3 ). As in other aneuploid disorders, the phenotypic abnormalities are most easily understood as resulting from altered dosage of certain genes. Genes expected to have an altered dose in Turner individuals are those which escape the process of X inactivation and which have functional homologs on the Y chromosome (4 ). Although such an idea for the pathogenesis of Turner syndrome was proposed over 30 years ago (3 ), no individual genes have emerged as strong candidates for involvement in the Turner phenotype. A gene pair, RPS4X/Y, has been proposed to play a role in the disorder (5 ), but evidence arguing against such a role has recently been reported (6 ).
A growing number of genes have the above-noted property of escape from X inactivation and expression from the Y chromosome (7 ). The prototypes of such genes are those in the pseudoautosomal regions (PAR) of the X and Y chromosomes, located at the distal ends of both the short and long arms (8 ). Besides the gene dosage considerations, another reason for implicating pseudoautosomal genes in Turner syndrome comes from observations of certain rare non-Turner patients who exhibit short stature. Ogata and colleagues have investigated patients having structural changes in the X and/or Y which result in deletions of portions of the short arm PAR. It was found that individuals who were hemizygous for the distal 750 kb of this region were short, while those with deletions elsewhere within the PAR were of normal stature (9 ,10 ). It was concluded that haploinsufficiency for a gene(s) in the distal deleted interval results in short stature, and these studies refined the localization of such a gene proposed a few years earlier based on a different set of X deletion patients (11 ).
Since virtually all Turner patients are monosomic for the short arm PAR, and since short stature is a cardinal feature of Turner syndrome, it seems reasonable to hypothesize that the dosage-sensitive locus described by Ogata et al. contributes to the short stature in Turner individuals. Currently no cloned genes have been assigned to the critical deleted interval, and so we have searched for expressed sequences in this region. In this paper we describe the isolation and characterization of a novel pseudoautosomal gene. Because of its location, its expression pattern, and the nature of the predicted protein, we believe it to be a very attractive candidate for involvement in the short stature phenotype of Turner syndrome. In addition we provide mapping data on the mouse homolog of this gene that could have important implications regarding the phenotypic differences of humans and mice with monosomy X.
The expression of the gene was examined in a variety of human fetal and adult tissues, with the results summarized in Figure 4 . Northern analysis (Fig. 4 A) revealed a faint band at 4.8 kb in heart, skeletal muscle and liver, but transcripts were undetectable in most tissues, even after the month-long exposure shown in the figure. In contrast, three prominent transcripts were observed in RNA from bone marrow stromal fibroblasts: besides the 4.8 kb transcript that (after accounting for a poly A tail) matches the length of the composite cDNA of Figure 2 A, transcripts of 7.5 and 2.0 kb were also seen. We cannot immediately account for these transcripts based on our cDNA clones, although several possible explanations exist, including partially processed transcripts, alternative polyadenylation/splicing, and transcripts from related genes. The stromal fibroblasts are pluripotent stem cells with the capacity to differentiate into several different cell types, among them osteoblasts and chondrocytes (17 ). Separate cultures of these cells from different individuals all gave strong signals by northern analysis and/or RT-PCR, as did primary cultures of trabecular bone cells, which have an osteoblast-like phenotype (18 ). Figure 4 B shows results of RT-PCR experiments that demonstrate expression from both of these primary bone-derived cultures, as well as from two osteosarcoma cell lines, (Saos-2, U-2 OS) and a chondrosarcoma cell line (SW1353). Several other tissues showed expression by RT-PCR, though northern signals were weak or absent (compare with Fig. 4 A). Thus the highest and most consistent levels of expression that we have observed is in osteogenic cells. For this reason we have named the gene PHOG, for
A search of sequence databases with the PHOG sequence revealed strong similarity to a mouse homeobox-containing gene that has very recently been cloned. An alignment of the protein sequences of PHOG and the mouse gene, OG-12 (19 ) is shown in Figure 5 . The homeodomains of the predicted proteins are identical, and there is substantial amino acid similarity (80%) in the C-terminal segments.
The corresponding nucleotide similarities are 81% (homeobox) and 76% (3' coding region). Upstream of the homeodomains the human and mouse proteins are quite different: other than three small blocks of near amino acid identity, the two proteins are almost completely dissimilar. In addition, both the human and mouse proteins have small polyglycine stretches, but these are in different locations in the two proteins. Nonetheless because of the overall sequence similarity, as well as an identical intron-exon organization for the two genes, there is little doubt that OG-12 is the mouse homolog of PHOG. Indeed the differences in the N-terminal region are reminiscent of the trend of extensive divergence of human pseudoautosomal genes and their mouse homologs (20 ).
The genes expected to show altered dosage in 45,X individuals is but a subset of all the genes on the sex chromosomes. These genes are the ones expressed in two doses in both sexes: from both the X and Y in males, and from both the active and inactive X in females (4 ). Among the growing list of such genes, most probably play no role in Turner syndrome, because one gene copy is sufficient for normal development to occur. This conclusion is based on the observation that hemizygous carriers of distal Xp deletions typically display no phenotypic abnormalities other than short stature (11 ,24 ). The various abnormalities of Turner syndrome thus likely result from haploinsufficiency for a relatively small number of genes. The identification of these dosage-sensitive genes, and the determination of which part of the phenotype they are involved in, has been difficult. This difficulty has been in part due to the phenotypic variability of the disorder and the resulting lack of genotype-phenotype correlations in Turner patients with partial monosomy X. This is why we think the pseudoautosomal deletion patients studied by Ogata et al. (9 ,10 ) are so important; not only are they deleted for genes expected to have altered dosage in Turner syndrome, but their phenotype is a cardinal feature of the disorder. This seems more than coincidence, and it thus seems reasonable to hypothesize that the same locus that leads to short stature in these individuals when present in single dose also plays a role in the short stature of Turner individuals, though in the latter other genes may also contribute.
Since PHOG lies in the critical deleted region of the PAR, it satisfies the genotype-phenotype criteria for a candidate stature locus. Because of its tissue-restricted expression, it has not been possible to examine the X inactivation status of PHOG in the usual manner. This is normally done by studying expression in rodent-human hybrid cell lines that retain active or inactive human X chromosomes, but since such hybrids are typically derived from fibroblasts, in which PHOG is not expressed (see Fig. 4 B), such studies could not be performed. Since the gene is pseudoautosomal, one would predict that, like all other short arm pseudoautosomal genes studied to date (25 -28 ), it would escape X inactivation, and thus have an altered dose in Turner patients. The nature of the protein encoded by PHOG (a homeodomain-containing transcription factor) and its predominant expression in osteogenic cells furthermore make PHOG a plausible candidate for involvement in linear skeletal growth. Lastly, there are precedents for dosage effects of genes encoding transcription factors (29 ,30 ), including homeobox-containing genes (31 ).
Taken together, the above considerations make a strong case for the hypothesis that haploinsufficiency of PHOG contributes to the short stature phenotype of Turner syndrome. The proof of such a hypothesis may be difficult, since the complete Turner phenotype (even its growth component) likely results from the interactions of a number of genes. Finding mutations of PHOG in patients with short stature would provide strong evidence for its proposed role, but given the many etiologies of short stature, such patients may be difficult to identify. Another way to support PHOG's candidacy would be to take advantage of an animal model. However, the monosomy X mouse does not have the somatic features of Turner syndrome (32 ). Although reduced weight gain of certain XO mice has been reported, this was attributed to a paternal X imprinting effect (33 ). The autosomal location that we have determined for OG-12, the mouse homolog of PHOG, could have an important implication: if PHOG indeed has a dose-dependent role in growth in humans and OG-12 has a similar function in mice, then the lack of a growth abnormality in XO mice could at least in part be explained by the retention of two copies of OG-12 in these animals. The obvious extension of this line of thinking is to examine a mouse with a single copy of the PHOG homolog; such studies have the potential not only for inferring the role of PHOG in humans, but also for generating a mouse model for part of the Turner phenotype.
Yeast artificial chromosome (YAC) clones were obtained from the Genome Center at Washington University. Multiple tissue northern blots and RNAs were purchased from Clontech. The cell lines Saos-2, U-2 OS and SW 1353 were obtained from the American Type Culture Collection. Poly A RNAs were prepared from these sources using the FastTrack kit from Invitrogen. Trabecular bone cells were prepared from pieces of normal bone under IRB approved procedures as described previously (18 ). Bone marrow stromal fibroblasts were cultured from normal bone marrow fragments as described (34 ). Total RNA was prepared from these two types of cultured cells, and poly A RNA was isolated by oligodT cellulose chromatography. DNAs from mouse interspecific backcross animals (23 ) were obtained from the Jackson Laboratory.
YAC clones from the distal PAR that were partially characterized by the Washington University Genome Center were further analyzed for STS content and by pulsed field electrophoresis to rule out chimerism. Clones yWXD2539 and yWXD2540 were chosen for further experiments. Cosmid subclones with human inserts were produced from these YAC clones as described (35 ). Restriction mapping and hybridization to known pseudoautosomal STSs allowed the assembly of several small cosmid contigs, including the one containing PHOG that contains the locus DXYS28. The exon organization of PHOG was determined by restriction mapping, hybridization to different cDNA fragments, and sequencing of genomic DNA subclones.
DNAs from pools of 4-6 cosmids were separately digested with EcoRI, BamHI and BglII, and fragments were shotgun cloned into the exon trapping vector pSPL3 (12 ). Pooled plasmid DNAs were then used with the exon trapping kit from Gibco BRL, according to the manufacturer's instructions. PCR amplification products were subcloned into the vector pAMP10, and double-stranded templates were sequenced using the Sequenase kit from US Biochemicals. The sequences obtained were used to search sequence databases using BLAST.
Poly A RNA (0.5 [mu]g) from trabecular bone cells was used as a template for cDNA synthesis using the cDNA synthesis primer from Clontech's Marathon cDNA Amplification Kit. Following adaptor ligation, two rounds of PCR were performed using an antisense primer from the trapped exon, along with the nested primers provided in the kit. A 1.2 kb fragment was amplified and subcloned into the TA cloning vector pCR2.1 (Invitrogen), and was designated clone 5'-R. A 0.3 kb fragment from the 3' end of this clone was used to screen a cDNA library constructed as described (36 ) from bone marrow stromal fibroblast RNA. Sequential screens were performed using probes derived from 3' cDNA clones, and sequencing of these clones allowed the assembly of the cDNA contig whose sequence is presented in Figure 3 .
Poly A RNA (2 [mu]g) from bone marrow stromal fibroblasts was subjected to northern blot analysis as described (36 ). This filter strip was hybridized along with the commercial multiple tissue blots as described (36 ), using the 5'-R cDNA fragment. The autoradiogram shown in Figure 4 A was exposed for 4 weeks with an intensifying screen. The same filters were rehybridized with a [beta]-actin probe, and were exposed for 6 h. For RT-PCR experiments, cDNA was synthesized in 20 [mu]l reactions using 0.5 [mu]g poly A RNA and random hexamer primers; 1 [mu]l of the cDNA reaction was then used in 20 [mu]l PCR reactions using a sense primer from exon 2 (5'-GGACCACGTAGACAATGACAAGG-3') and an antisense primer from exon 4 (5'-ACTTGGCTCTCCGGTTCTGGAAC-3') at a concentration of 0.5 [mu]M each. Thirty five cycles of amplification were carried out by denaturation at 94oC for 45 s, annealing at 65oC for 45 s and extension at 72oC for 1.5 min (the extension time for the first cycle was 15 min). PCR products were electrophoresed on a 5% polyacrylamide gel and photographed after ethidium bromide staining. Control PCR reactions were done using primers for the MIC2 gene as described (37 ).
A dinucleotide repeat polymorphism was found within intron 2 of the OG-12 gene which allowed us to distinguish domesticus and spretus alleles. PCR was done using primers flanking this polymorphism; the sense primer was 5'-CTGGTAATGAGAYTGTCTTGC-3' and the antisense primer was 5'-CAGTTCAGGGGACACTGAAG-3'. PCR was done using 50 ng of DNA from each of 188 backcross animals [both the BSB and BSS backcrosses (23 )], under the following conditions: initial denaturation at 94oC for 4 min, followed by 35 cycles of 94oC for 45 s, 58oC for 45 s and 72oC for 1.5 min. This resulted in the amplification of a 138 bp M.domesticus allele and a 171 bp M.spretus allele. The PCR products were run on a 10% polyacrylamide gel and visualized by ethidium bromide staining. Typing data were sent to the staff at the Jackson Laboratory for linkage analysis.
We thank the staff at the Jackson Laboratory for data analysis and assistance in figure production. We also thank C. Epstein, W. Miller, F. Chehab, E. Salido and L. Shapiro for comments on the manuscript. This work was supported by a Basil O'Connor Starter Scholar Research Award from the March of Dimes (#5-FY96-0562) to J.W.E., who was also supported in part by a Child Health Research Center grant (HD28825) to UCSF.
+Present address: Departments of Pediatrics and Human Genetics, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA23298, USA
Human Molecular Genetics
Pages
Introduction
Results
Cloning of a new pseudoautosomal gene
Expression studies
Mapping of the mouse homolog
Discussion
Materials And Methods
Cells and nucleic acids
Characterization of genomic sequences
Exon trapping
cDNA clones
Expression studies
Genetic mapping of the mouse PHOG homolog
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Tue Jul 15 11:13:42 BST 1997. Part of the OUP Journals World Wide Web service. Copyright Oxford University Press, 1996