Physical and Genetic Mapping of Polymorphic Loci in Xq28 (DXSI5, DXS52, and DXS134): Analysis of a Cosmid Clone and a Yeast Artificial Chromosome

Summary Sequences corresponding to the Xq28 loci DXS15, DXS52, DXS134, and DXS130 were shown to be present in a 140-kb yeast artificial chromosome (YAC XY58, isolated by Little et al.). This YAC clone appears to contain a faithful copy of this genomic region, as shown by comparison with human DNA and with a cosmid clone that contains probes Stl4c (part of the DXS52 sequences) and cpX67 (DXS134). cpX67 and Stl4c are contained in 11 kb and detect the same MspI RFLP polymorphism. A comparison of the YAC restriction map and pulsed-field gel electrophoresis data leads us to propose the following order of loci: DXS52(VNTR)-DXS33-DXF22S3-DXS130-DXS134-DXS52-DXS15-DXS52, this whole cluster being comprised within 575 kb. The physical proximity of the DXS15, DXS52, and DXS134 loci led us to reinvestigate recombination events that had been reported between these loci in families from the Centre d'Etude du Polymorphisme Humain. Our results do not support the assumption that this region shows increased recombination. Segregation of six Xq26-q28 RFLPs in CEPH pedi- gree 66. The following probes were used: DXSS1, 52A probe; F9, FIX-PI; DXS105, cX55.7; DXS52, Stl4-1, DXS134, cpX67; DXS1S, DX13. The same results were obtained using another set of DNAs for all individuals, except for 5 and 13, for DXSS1, F9, DXS1O5, and DXS52.


Introduction
The Xq28 region of the human X chromosome contains many disease loci. The biochemical defect of most of these diseases is poorly known, and the loci have been mapped genetically with respect to polymorphic markers in this region, notably DXS15 and the hypervariable locus DXS52. . For the localization of these disease loci, detailed and unambiguous genetic and physical maps are necessary. Physical mapping by pulsed-field gel electrophoresis (PFGE) has allowed the linking of key DNA sequences within two clusters. The first one was shown to contain the polymorphic loci DXS15, DXS33, DXS52, and DXS134 (Patterson et al. 1987;Bell et al. 1989). However, one complicating factor is the dispersion, within this region, ofthe sequence family detected by the Stl4c probe, which defines the DXS52 locus (Arveiler et al. 1989;Bell et al. 1989). The second cluster contains several well-known genes encoding the red and green color pigments, glucose 6-phosphate dehydrogenase, and coagulation factor VIII (Arveiler et al. 1989). The genetic map has been the object of some controversy concerning both the relative order of the two clusters with respect to the telomere and the recombination fraction between loci.
The cloning of large DNA fragments in yeast artificial chromosomes (YAC) appears a powerful tool to map such regions of the human genome (Burke et al. 1987), and it is important to establish at this stage the fidelity of the technique. A library of YAC clones for the Xq24-q28 region is being constructed from a somatic hybrid cell line (Little et al. 1989). A YAC clone containing sequences detected by the Stl4-l and DX13 probes (DXS52 and DXS15) has been isolated by Little et al. (1989). This clone was further shown to contain DXS130 and DXS134 (Wada et al. 1990). Here we describe the detailed physical mapping of both this clone and an independently selected cosmid clone. This allowed us to locate precisely DXS15, DXS130, and DXS134 and one polymorphic member of the DXS52 family. We report also genetic mapping information related to these loci. Material and Methods Cosmid 33-3A1 was isolated from a nonamplified 49XXXXY genomic DNA cosmid library, constructed by Heilig et al. (1987), screened with the St14-1 probe (a 3-kb EcoRI fragment at the DXS52 locus). The 49XXXXY DNA, used to make this cosmid library and used as control DNA in Southern blot experiments, originates from the lymphoblastoid cell line GM1202 (Human Genetic Mutant Cell Repository, Camden, NJ). The recombinant yeast clone XY58 has been isolated by Little et al. (1989), by screening with St14-l and DX13, and was provided to us by Dr. D. Schlessinger (St. Louis). Yeast cells were grown in minimal medium (uracil minus); yeast DNA was isolated according to the protocol of Holm et al. (1986). Isolation of genomic DNA, electrophoresis, blotting (on diazobenzyloxymethyl paper or on Hybond-N), and hybridization were performed according to a method described elsewhere (Oberle et al. 1986a). Probes were labeled by random priming (Feinberg and Vogelstein 1983). The following probes (shown with their sources) were used to hybridize genomic blots: FIX-P1 (F9 locus) (Oberle et al. 1986b), Stl4-1 (3-kb EcoRI fragment), and Stl4c (10-kb EcoRI fragment), the latter two being from the DXS52 locus (Oberle et al. 1986b); cX55.7 (DXS105), cpX6 (DXS130), and cpX67 (DXS134) ; and MN12 (DXS33) (Patterson et al. 1987), G1.3c (DXF22S3) (Bardoni et al. 1988), DX13 (DXS15), and 52A (DXS51) (Drayna et al. 1984). DNA from 3-generation pedigrees was obtained from the Centre d'Etude du Polymorphisme-Humain (CEPH). For restriction mapping of YAC XY58, high-molecular-weight DNA from YAC XY58 was digested to completion with BamHI, BssHII, MluI, NotI, NruI, PvuI, SacII, Sall, SfiI, or XhoI. Indirect end-label mapping of XY58 was carried out after partial digestion with increasing concentrations of BamHI, SacII, and XhoI. PFGE was performed using a CHEF apparatus. Agarose gels (1%) were run at 13°C in 0.5 x TBE buffer (20 x TBE = 1 M Tris, 830 mM boric acid, 10 mM EDTA) at 160 V for 15 h; pulse times were 2.5-7.5 s. Sizefractionated DNA was transferred to nylon membranes (ONCOR SUREBLOT) in denaturing solution (0.5 N NaOH, 1 M NaCl) after an initial 15-min depurina-tion in 0.2 N HCL. Filters were prehybridized in 1 M NaCl, 1% SDS, 10% dextran sulfate at 65°C for 2 h; denatured probe (2 x 106 cpm/ml) and herring sperm DNA (75 gg/ml) were added, and hybridization was for 20 h. Washing was performed as for the genomic Southern blots.

Mapping of XYS8 and Comparison with a Cosmid Clone
That Contains the Stl4c and cpX67 Sequences The St14-1 probe (a 3-kb EcoRI fragment at the DXS52 locus) was used to screen a cosmid library (Heilig et al. 1987), yielding cosmid 33-3A1 which contains a 43-kb insert. This cosmid was studied in comparison with YAC XY58, described by Little et al. (1989) as hybridizing to the St14-l and DX13 (DXS15) probes, and it was further shown to contain probes cpX6 and cpX67 (Wada et al. 1990).
Probes St14-l and cpX67 (DXS134) hybridized to restriction fragments that have identical sizes in the cosmid and YAC clones (results not shown, see fig. 1 for results on XY58). However, a 10-kb EcoRI fragment is detected by St14-1, which does not correspond to the cognate fragment (3 kb). This suggested that the cosmid and YAC clones contain the Stl4c fragment, the first probe cloned from the St14 sequence family (Oberle et al. 1985). Indeed, Stl4c hybridized to the cosmid (not shown) and to XY58, detecting a fragment of the same size as the cognate EcoRI fragment in genomic DNA ( fig. 1A). XY58 contains only a single EcoRI fragment detected by St14-1 or Stl4c, while in genomic DNA these probes detect as many as six EcoRI fragments. Thus, only a minor part of the Stl4 sequence family is present within XY58, as already has been noted by Little et al. (1989). In particular the VNTR region responsible for the multiallelic RFLP at the DXS52 locus (Mandel et al. 1986;Heilig et al. 1987) is not present in XY58 (it is 2 kb from St14-1; authors' unpublished data). Sequences corresponding to cpX6 (DXS130) and DX13 were present only in YAC XY58. The probes MN12 (DXS33) and G1.3c (DXF22S3, a member of a small dispersed X-specific sequence family), both known to be physically linked to the DXS52 locus (Patterson et al. 1987;Arveiler et al. 1989), were absent from both cosmid 33-3A1 and YAC XY58.
In order to determine the fidelity of the sequences inserted in YAC XY58, we compared the Stl4c-, cpX67-, and DX13-hybridizing fragments in genomic DNA and in YAC XY58. All Stl4c-hybridizing XY58 fragments Comparison between genomic and YAC XY58 restriction fragments detected by probes Stl4c and St14-l (DXS52), cpX67 (DXS134), and DX13 (DXS15). A, Southern blot with BamHI, HindIII, TaqI, and EcoRI digests of a control male (lanes 1), the 49XXXXY cell line GM1202 (lanes 2), and YAC XYS8 (lanes Y). Lanes 1 and 2 contain 10 gg digested DNA, and lanes Y contain 0.2 pig digested DNA. This Southern blot was hybridized with the Stl4c probe. B-D, Autoradiographs of the same blot as in fig. 1A, hybridized with the probes St14-l, cpX67, and DX13, respectively. The cpX67 hybridizing genomic DNA fragments are faint or not visible in fig. 1C; on the original X-ray film all hybridizing YAC XY58 fragments show a clearly visible corresponding genomic fragment. Sizes of marker DNA bands are indicated in kilobases. V = fragment giving a positive hybridization with pBR322 vector sequences. * = fragment which hybridizes with both Stl4c and cpX67; R = fragment hybridizing to probe R (the amplified 500-bp end fragment located at the right arm side of YAC XYS8). .6-m 2.9"-2.3m- are also present in genomic DNA. The St14-1 probe detects in XY58 a subset of the fragments that hybridize to Stl4c ( fig. 1B). The cpX67 probe also hybridized to the same fragments in genomic DNA and in XY58. cpX67 and Stl4c detected common fragments, i.e., a 6.6-kb BamHI and a 10-kb HindIII fragment (marked with an asterisk in fig. 1C). In the HindIII digest of XY58, cpX67 hybridizes to the 10-kb fragment and to a 11-kb fragment. This means that cpX67 and Stl4c are contained within an 11-kb region.
Stl4-1, Stl4c, and cpX67 detected fragments which have the same size in genomic DNA as in XY58, but this was not the case for the DX13 probe. The BamHI and TaqI fragments detected by DX13 have a different size in the YAC clone than in genomic DNA, while the 2.2-kb EcoRI fragment (which corresponds to the probe itself) and the HindIII fragment have the same size in the two DNAs ( fig. 1D). Two of the abnormal fragments (the 4.5-kb BamHI fragment and the 1.6-kb TaqI fragment, labeled V in fig. 1D) also hybridize to pBR322. These fragments thus contain both DX13 and YAC vector sequences (pYAC-4 contains pBR322 derived sequences [Burke et al. 1987]). As the library from which XY58 was isolated was constructed by cloning a partial EcoRI digest in the EcoRI site of pYAC-4 (Little et al. 1989), it appeared likely that the DX13 2.2-kb EcoRI fragment lies at one of the two extremities of the XY58 insert. We amplified the two extremities of the insert by using a recently developed technique based on the polymerase chain reaction (R. Heilig, unpublished data). We obtained a 1,200-bp fragment for the left arm (L) and a 500-bp fragment for the right arm (R). (The left arm ofthe vector contains the ampicillinresistance sequences.) L, used as probe, did not hybridize to XY58 restriction fragments which are recognized by Stl4c, Stl4-1, cpX67, or DX13. In all four digests the R probe hybridized to restriction fragments recognized by DX13 (in fig. 1D these fragments are indicated by the letter R). In the BamHI and TaqI digests, only the fragments that contain both pYAC-4 vector sequences and DX13 sequences hybridized to the R probe. Furthermore, R detects the same BgIII RFLP as DX13 (not shown). We conclude that the DX13 fragment lies at the extremity of the YAC XY58 insert (on the R side).
To further search for possible rearrangements within XY58, we compared the 43-kb insert of cosmid 33-3A1 with the corresponding region of XY58. DNA of XY58 and of cosmid 33-3A1 was digested with four restriction enzymes, and fragments hybridizing to the whole cosmid 33-3A1 were compared. All but one of the cosmid fragments which do not hybridize to vector DNA have the same size as in XY58 DNA (fig. 2). The exception was a 3.6-kb TaqT restriction fragment that also hybridizes to the Stl4c probe. This fragment and a 1.9-kb TaqI fragment observed in XY58 might correspond to alleles of a TaqI RFLP detected by the Stl4c probe. We conclude that the insert sequences in cosmid 33-3A1 are entirely contained in XY58 and that the YAC clone is not rearranged for these 43 kb. Comparison of fragment size in XY58 and cosmid 33.3A1. DNA from cosmid 33-3A1 and YAC XY58 DNA was digested with BamHI, HindIII, EcoRI, and TaqI and hybridized to the entire cosmid 33-3A1. Length of the size-marker fragments are indicated in kilobases. Y = YAC XY58 DNA (1 pg/lane); C = cosmid 33-3A1 DNA (10 ng/lane); V = fragment hybridizing to cosmid vector sequences; * = fragments that differ in size between the two clones. These fragments hybridize to Stl4c and might represent variants of the genomic DNA (see Mandel et al. 1986 . 3). Since DX13 and Stl4-l have been extensively used for pulsed field mapping studies (Patterson et al. 1987;Arveiler et al. 1989;Bell et al. 1989), it was of interest to map the sites for rare-cutter enzymes used in such studies. Agarose blocks containing XY58 DNA were digested to completion with rare-cutter restriction enzymes, analyzed by PFGE, and blotted for hybridization to the probes cpX6 (DXS130), cpX67 (DXS134), St14-1 and Stl4c (DXS52), and DX13 (DXS15), as well as to vector probes corresponding to L or R of YAC. For three additional enzymes (BamHI, SaclI, and XhoI) partial digestion and mapping by indirect end-labeling was performed (using the vector probes).
This placed the DXS134-DXS52 (Stl4c) cluster between DXS130 and DXS15, at about 40 kb from the latter (and from the right end ofthe YAC clone). A very high density of sites for "rare-cutter" enzymes was found, confirming and extending the observations of Patterson et al. (1987), especially in the region that includes the Stl4c probe, which was therefore more precisely mapped in the cosmid 33-3A1. The 5-kb segment that contains the Stl4-1 cross-hybridizing sequences contains three Sf1i sites, two MluI sites, one NotI, and one SacdI site. The latter two sites are often found in CpG (HTF) islands (Lindsay and Bird 1987). Another potential CpG island (with one BssHII site and one SacdI site) is found 12 kb from the NotI site, in the direction of DXS15.

L Stl4c Detects the Same RFLP as cpX67
Our analysis showed that Stl4c and cpX67 are comprised within 11 kb. Stl4c was previously reported to detect three MspI RFLPs (Oberle et al. 1985). The independently isolated cpX67 probe also detects an MspI RFLP with allelic fragments estimated at 3.7 and 3.4 kb ). These fragments fall in a similar size range as the allelic MspI fragments 1 and 2 detected by Stl4c (previously estimated at 4.4 and 3.6 kb). We verified that, in MspI digests of genomic DNAs, allelic fragments 1 and 2 detected by Stl4c are the same as the ones detected by cpX67 (fig. 4). It should be noted that the second MspI polymorphic system detected by Stl4c (allelic fragments 3 and 4) was found to be in very strong linkage disequilibrium with the VNTR RFLP and is within 5 kb of Stl4-1 (Oberl et al. 1985).

Reinvestigation of Two CEPH Pedigrees Reported to Show Recombination between DXSIS and DXSS2 or DXS134
Because Patterson et al. (1987) reported that the physical distance between Stl4 sequences arid DX13 could be as little as 60 kb, it was suggested that this region contains a hot spot of recombination (Bell et al. 1989). In the first genetic map of the X chromosome  the map distance between DX13 and Stl4 was reported to be 5.5 cM, while only 0.06 cM would be expected on the basis of the average relationship of 1 cM/1,000 kb. However, the only three recombinations in 54 informative meiosis occurred in the same

Stl4c cpX67
Figure 4 An MspI RFLP detected by Stl4c that is identical to the MspI RFLP detected by cpX67. A Southern blot with MspIdigested DNA of five female individuals (1-5) was hybridized to Stl4c and cpX67. The Stl4c probe detects three MspI RFLP systems: a first system with alleles 1 and 2 (4.4 and 3.6 kb), a second system with alleles 3 and 4 (2.0 and 1.6 kb), and a third system which concerns the presence/absence of allele 5 (1.0 kb) (also see Oberk et al. 1985).
family (and these were responsible for the calculated 7-cM distance between DXS15 and DXS52 in another X-chromosome genetic map based on the same data (Donis-Keller et al. 1987). In the genotype data base ofthe CEPH (version 1) we identified this peculiar family as family 1377. Since, if true, this typing could have been due to either a chromosomal inversion or some other kind of rearrangement, we retyped the family. Our DXS15 typing agreed with that reported in the CEPH data base, but the typing for DXS52 (Stl4-1 probe) differed for three DNAs, and, as a result, no recombination was found between the two loci (this could be further checked by using a probe nearer to the VNTR at DXS52, which allowed the typing on the same BgII digests as used for the DX13 typing). Recently, Bell et al. (1989) reported two recombination events in a CEPH pedigree: one between DXS134 and DXS15 and one between DXS52 and both DXS15 and DXS134. We have retyped this pedigree (fig. 5) and found one recombinant between DXS52 and both DXS15 and DXS134 but none between DXS134 and DXS15. Furthermore, individual 9 of this pedigree, reported to present a recombination event between the loci F9 and DXS105, was typed by us as being a recombinant between DXS105 and DXS52. (As an internal check, DXS52 was typed both on the TaqI blot also used for F9, DXS51, and DXS105 and on the MspI blot used for DXS134.) Thus, ofthe five recombination events that in CEPH families were reported to occur between DXS52, DXS15, and DXS134, only one has been confirmed. This weakens the support for a hot spot of recombination between these physically closely linked markers.

Discussion
Our analysis of a YAC containing well-known markers from the Xq28 region has allowed us to darify their respective positions. This illustrates well the usefulness of this new technology for genome analysis. Comparison of selected regions of the YAC clone with corresponding regions in either genomic DNA or a cosmid suggests that the YAC clone contains a faithful representation of human sequences. A single member of the Stl4 sequence family was found in XY58, that corresponds to the 10-kb segment (Stl4c) originally described by Oberk et al. (1985). Since Stl4c is at 40 or 100 kb of either ends of the clone, this confirms that the Stl4 family is dispersed, as proposed on the basis ofpulsed-field mapping studies (Arveiler et al. 1989;Bell et al. 1989). In particular, the VNTR region responsible for the multiallelic polymorphism at the DXS52 locus (Mandel et al. 1986) is not present in XY58. We have shown that DXS134 and Stl4c are adjacent fragments, within 11 kb. DXS134 previously had been reported to be within the same BssHII 300-kb fragment as an unidentified member of the Stl4 family (Bell et al. 1989). In fact, one of the three MspI RFLPs described for Stl4c (OberlM et al. 1985) is the same as the RFLP independently described for cpX67 . Stl4c is also 30 kb from the probe DX13 (DXS15). This confirms and extends the findings of Patterson et al. (1987) showing linkage within 60 kb of DXS15 and a member of the Stl4 family. It is astonishing that four independently derived probes are contained within the 140-kb region. Given the pool of mapped X-specific probes ("'200-250) from which the presented studied were drawn, an average of 1 probe/ 1,000 kb would be expected. The four probes have been isolated from two libraries by using two different strategies (Davies et al. 1981;Oberl et al. 1985;). An overrepresentation of probes for a region of chromosome 21 has also been observed (Gardiner et al. 1988). This suggests some bias in probe isolation, possibly favoring regions of the genome which are less methylated. For both libraries, mcr A + B + bacterial strains were used, which contain restriction systems that degrade DNA methylated at CpG dinucleotides. It recently has been shown that the use of mcr A+ B+ host strains and packaging extracts considerably reduces the efficiency of cloning methylated DNA (Kretz et al. 1989).
The mapping of a NotI site and a BssHII site between Stl4c and DXS15 clarifies the map of the region. Bell et al. (1989) have shown that DXS15 and DXS134 are on separate NotI and BssHII fragments, each ofthem hybridizing to the Stl4-1 probe. The cpX67-hybridizing BssHII fragment of Bell et al. (1989) appears identical to a fragment hybridizing both to the Stl4-1 probe and to G1.3c, a member of a dispersed X-specific sequence family. Furthermore, probe MN12 (DXS33) was found on a third 150-180-kb BssHII fragment, together with another Stl4 sequence Arveiler et al. 1989;Bell et al. 1989). This fragment contains the VNTR sequence (Bell et al. 1989) and can be further cleaved in some cell lines by a BssHII site that separates DXS33 from the Stl4 sequence (Arveiler et al. 1989;Bell et al. 1989). Their relative order is given by a 400-kb SfiI fragment common to DXS33 and DXS134 (Bell et al. 1989). One of the Sfil sites is within Stl4c, and the finding that the 400-kb Sf1 fragment does not hybridize to Stl4-1 is not contradictory, since the part of the Stl4c fragment at the left of the Sfil site does not hybridize to Stl4-1 (fig. 3). The three BssHII fragments must be contiguous, since a 575-kb BssHII fragment (a result of incomplete digestion) contains DXS15, DXS33, DXS52, and DX22S3 (Arveiler et al. 1989).
The final map of the region is shown in figure 6. The Stl4 family appears dispersed over a 575-600-kb region in Xq28. The signification of this organization is not known, but it might be of interest to point out that two other sequence families appear dispersed in regions of the X chromosome: the G1.3 family (one member  Figure 6 Summary of the DXS52-DXS15 region. Restriction sites for BssHII (B), NotI (N), and Sfil (S) are indicated. The BssHII site in parenthesis is cleaved only in some cell lines, while the BssHII site marked with a square was not cleaved detectably in the same studies (Arveiler et al. 1989;Bell et al. 1989). The localization of the asterisked Sfl site is only based on the size (400 kb) of the DXS33 and DXS134 hybridizing fragment and is therefore not precise. The arrowed bars indicate the range for localization of various markers. The region cloned in YAC XY58 is represented by a heavy line. B = BssHII; S = Sfil; N = NotI.
is in the Xq28 region analyzed here, but the others are in Xp22.2-p22.3 (Bardoni et al. 1988;Ballabio et al. 1989) and the ornithine amino transferase pseudogene family in Xp2l.1-11.2 (Lafreniere et al. 1989). The Stl4 sequence is conserved in evolution (Mandel et al. 1986), and this was the basis ofthe mapping of corresponding sequences on the mouse X chromosome. The presence ofprobable CpG islands within the Stl4c fragment (NotI and SacIl sites) is an additional element suggesting that the St14 family is an expressed-gene family. When the proximity of St14 sequences and DXS15 was demonstrated, it was suggested that the region might be unusually prone to recombination (Patterson et al. 1987;Brown et al. 1988;Bell et al. 1989). We know now that the VNTR polymorphism that defines genetically the DXS52 locus is 400-500 kb from DXS15. The recombination between DXS15 and DXS52 has been reported, in various studies, to be 1-5 cM Brown et al. 1988;Lehesjoki et al. 1989). In our own studies we found a maximum lod score z = 41.44 at 0.6 cM (I. Oberle, unpublished data). Here we have shown that four of the five recombination events previously reported between DXS15 and DXS52 in CEPH families cannot be confirmed. Fewer data are available for the linkage of DXS134 to either DXS15 or DXS52, and one should perhaps reinvestigate the single large family that showed rather high recombination between DXS52 and DXS134 (0 max = .15) or between DXS15 and DXS134 (0 = .08) . At present we think there is little support for the existence of a particularly high recombination rate in this region.