GENOMICS
11,
154-164
(1991)
Physical Mapping of the Loci Gabra3, DXPas8, CamLl, and Rsvp in a Region of the Mouse X Chromosome Homologous to Human Xq28 CYNTHIA Institute
for Molecular
J. FAUST AND GAIL E. HERMAN’
Genetics and Department Received
March
of Pediatrics, 5. 1991;
pulsed-field gel electrophoresis, a 3 million-bp map containing the X-linked loci Gubra3, DXPasB, CamLl, and RSVPhas been constructed for a segment of the mouse X chromosome homologous to human Xq28. Detailed mapping was performed using single and double digestions with rare-cutter restriction enzymes. Gabra3 and DXPae8 have been shown to be physically linked within a maximal distance of 1600 kb, DXPas8 and CamLl within 750 kb, and CamLl and Rsvp within 450 kb. In addition, several CpG islands have been detected in the region encompassing CamLl and RSVP.These studies confirm a gene order of ten-Gabra3-DXPaeB-CamLlRsvp-tel determined by genetic mapping in interspecific backcrosses (A. S. Ryder-Cook et al., 1989, EMBO J. 7: 3017-3021; G. E. Herman et al., 1991, Genomics9: 670677). Physical distances for the loci studied agree with the calculated genetic distances. Assuming that there is conserved linkage between man and mouse in the region, the physical mapping data presented here may help to clarify the uncertain gene order for some human Xq28 loci. q iesl Press, Inc.
INTRODUCTION
The conservation of loci on the X chromosomes of and mouse is predicted by Ohno’s Law, which states that the movement of genes between the sex chromosomes and autosomes is restricted due to dosage compensation (Ohno, 1973). The order of Xlinked loci, however, is not necessarily conserved. Comparisons of X-linked loci in mouse and man reveal that at least five chromosomal rearrangements are required to account for the gene order in the two species (Davisson, 1987; Amar et al., 1988). Within each syntenic block, few, if any, rearrangements appear to have occurred. man
1 T O whom correspondence should be addressed at Institute for Molecular Genetics, One Baylor Plaza, S911, Houston, TX 77030. 154
OSsa-7543/91$3.00
Copyright All rights
0 of
by Academic Press, Inc. reproduction in any form reserved. 1991
revised
Houston,
Texas 77030
May 4, 1991
Using mouse interspecific backcrosses, several laboratories have mapped more than 100 loci on the murine X chromosome, including DNA probes for loci mapping to human Xq26-Xq28 (Avner et al., 1987; Brockdorff et aZ., 1987; Amar et aZ., 1988; Mullins et aZ., 1988,199O; Ryder-Cook et al., 1988; Disteche et al., 1989; Herman et al., 1991). This segment of the mouse X chromosome, which spans from Hprt to the Xlinked visual pigment gene (Rsvp), is homologous with human Xq26-Xq28. Human Xq28 contains numerous disease genes, many of which remain to be cloned (Mandel et al., 1989). Mapping in this region of the human X chromosome has been difficult because of high recombination and conflicting genetic and physical maps from different laboratories. Assuming that the gene order between man and mouse is retained in this region, maps constructed in the mouse could prove helpful in determining the correct human gene order. Even if order was not conserved, additional rearrangements detected would be important for studies of mammalian X chromosome evolution. Using an interspecific backcross, our laboratory has previously prepared a genetic map containing several loci in the region of the mouse X chromosome homologous to human Xq28 (Herman et al., 1991). Loci mapped included Gabra3, CamLl, RSVP, and DXPa.98. Gabra3 encodes the a3 subunit of the GABA, receptor, the major inhibitory neurotransmitter in the vertebrate brain. The Ll protein product of the human LlCAM and murine CumLl genes is a neural adhesion molecule and a member of the immunoglobulin superfamily (Moos et al., 1988). Rsvp encodes the murine X-linked visual pigment locus homologous to the human red and green visual pigment genes. In man, the red visual pigment gene is part of a 115kb color vision cluster that also contains one or more green pigment genes arranged in a tandem array (Nathans et al., 1986; Feil et al., 1990a). In mouse, a single X-linked visual pigment gene is believed to be present. The murine DXPas8 locus was originally defined
Using physical
Academic
Baylor College of Medicine,
PHYSICAL
MAPPING
OF
Gabra3, DXPa.98, CamLl,
by hybridization with St14 (DXS52), an anonymous human probe mapping to Xq28 (Oberle et aZ., 1985, 1987; Avner et al., 1987). St14 detects at least three loci in Xq28, although they are located within approximately 600 kb of each other (Feil et aZ., 199Ob). In the mouse, conflicting data concerning the mapping of DXPas8 have been reported (Avner et al., 1987; Herman et al., 1991). Our genetic mapping data strongly favored a gene order of ten-(Cf-9)-Gabra3-DXPasB-CamLl-(RSVP, Gdx, Cf-8)-Dmd-tel for these loci. Rsvp cosegregated with Cf-8, the murine coagulation factor VIII locus, and Gdx, a gene located approximately 40 kb 3’ to G6pd and 0.8 kb 3’to the P3 locus (Filippi et aZ., 1990). To resolve discrepancies between our genetic map and consensus human Xq28 mapping data, and to compare genetic and physical distances, we undertook physical mapping of this region of the mouse X chromosome using pulsed-field gel electrophoresis. We present in this study a physical map encompassing approximately 3 million bp that includes the four loci Gabra.3, DXPasB, CamLl, and RSVP. The results support the genetic mapping data regarding these loci and are discussed in terms of conserved linkage with loci in human Xq28. MATERIALS
Isolation and Mapping the DXPas8 Locus
AND
METHODS
of Mouse Genomic
Clones for
A l.O-kb TaqI restriction fragment from the human anonymous probe St14-1, a subclone of St14 (DXS52) (Mandel et al., 1986), was employed to screen approximately 1 million plaques from a murine NIH3T3/X FIX11 genomic library (Stratagene). Strongly hybridizing clones were mapped in an interspecific backcross using TaqI and BgZII restriction fragment length variations (RFLVs). The interspecific backcross of C57BL/6JAwdJ and Mus spretus has been previously described (Herman et al., 1991) and consists of a regional mapping panel of 120 animals with recombination events spanning the mouse X chromosome and 18 animals with single recombination events between Cf-9 and Dmd in 248 total backcross progeny.
DNA Probes The DNA probes used in this study and their sources, as well as the hybridization and washing conditions employed for each probe, are listed in Table 1. XDXPas8A and XDXPas8B are genomic clones isolated as described above. pDXP8B.l is a l.l-kb Hi&III subclone in pBluescript SK+ (Stratagene) derived from XDXPas8B. DNA probes were prepared using random hexamer labeling (Feinberg and Vogel-
AND
Rsvp
155
stein, 1984) and employed at a concentration of 1 X lo6 cpm/ml hybridization solution for conventional Southern analysis and at a concentration of 2 X lo6 cpm/ml hybridization solution for pulsed-field gel analysis. Prior to hybridization, the probes XDXPas8A, XDXPas8B, and XgMM25 (RSVP) were preannealed for 2 h at 65°C with 1 mg/ml total mouse DNA sonicated to an average molecular weight of 1 kb.
Pulsed-Field
Gel Analysis
DNA plugs were prepared from the thymus of 2to 3-week-old normal male and female progeny of B6CBA-AuJ/A or BSCBA-AwJ/A-Bps females X (C57BL/6JAweJ X CBA) males. This mating scheme results in X chromosomes derived from the inbred C57BL/6JAW” strain (Herman et al., 1991). The Bpa allele represents an X-linked dominant mutation, bare patches, segregating in some of the female parents. Thymocytes were obtained by removing the thymus from sacrificed mice and d&aggregating cells with a Cellector (E-C Apparatus Corp). Plugs were then prepared using a variation of standard methods (Smith and Cantor, 1987; Compton et aZ., 1988). Briefly, isolated thymocytes were resuspended in lysis buffer (10 mM Tris-HCl, pH 8.0,20 mM NaCl, 200 mM EDTA) and mixed with an equal volume of 1.6% Incert agarose (FMC Bioproducts) in lysis buffer to reach a final concentration of 2 X lo7 cells/ml. The mixture was poured into 2.0 X 1.0 X O.l-cm molds and chilled for 5 min. Plugs were transferred to 5 vol of lysis buffer containing 1% N-lauroylsarcosine/2 mg/ml proteinase K and incubated for 24 h at 45-50°C. The buffer was exchanged and the plugs were incubated another 24 h. Plugs were dialyzed twice in TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA) containing 1 mM phenylmethylsulfonyl fluoride (PMSF), followed by three dialyses in TE buffer without PMSF, and were stored at 4°C in 0.5 M EDTA. Plugs were prepared from splenocytes of adult animals using the above protocol except that red blood cells were lysed using hypotonic saline after initial centrifugation. Restriction digests were performed using onequarter to one-third plug (approximately l-2 X lo6 cells). Immediately prior to digestion, individual plugs were washed four times with 1 ml TE, followed by equilibration in 1 ml of the appropriate digestion buffer for 30 min. Plugs were digested for 4 h in 250~~1 reaction mixtures containing the buffer recommended by the enzyme manufacturer, 2 mM spermidine, and 30 U of enzyme (Stratagene or BoehringerMannheim). For digests with BssHII (New England Biolabs), 16 U of enzyme were employed. Partial digestions with MZuI were performed by preincubating
156
FAUST
AND
HERMAN
TABLE DNA Probe
Locus
pbGRa3 (bovine partial cDNA) St14-1 (human genomic) XDXPas8A (mouse genomic) XDXPas8B (mouse genomic) pDXP8B.l (mouse genomic) pK13 (mouse partial cDNA) hs7 (human partial cDNA) XgMM25 (mouse genomic) G28B (mouse genomic) pKpn-sac (human 5’ partial cDNA)
Gabra.3 DXPa.98 DXPas8 DXPas8 DXPas8 CamLl Rsup Rsvp Gdx
Probes
1
Used in This
Hybridization
Cf-8
Study
conditions”
Wash
I II III III II I II III II II
conditionsb A B C c C C A C C A
Source E. A. Barnard (24) J.-L. Mandel (25) This work This work This work M. Schachner (27) J. Nathans (31) J. Nathans P. Avner (3) Genetics Institute
a Hybridization conditions were as follows: I, hybridization at 65°C in 1 M NaCl, 10% dextran sulphate (Pharmacia), 1% pg/ml denatured herring sperm DNA. II, hybridization at 42°C in 40% formamide, 1 M NaCl, 10% dextran sulphate, 1% SDS, denatured herring sperm DNA. III, protocol I plus 100 rig/ml total mouse DNA sonicated to an average size of 1 kb. b Wash conditions were as follows: A, three washes with 3X SSC, 0.05% SDS. B, twice with 2X SSC, 0.1% SDS, twice with SDS. C. seauential washes of 2X SSC. 0.1% SDS: 1X SSC. 0.1% SDS; 0.5X SSC, 0.1% SDS. All washes were performed minutes. 1X SSC is 0.15 M NaCl, 0.015 M Na citrate, pH 710.
plugs for 2 h at 30°C in restriction buffer without MgCl, , followed by the addition of 10 mM MgCl, for 5 min. The reactions were terminated by the addition of EDTA to 20 n&f. Partial digestions with Sac11 were performed in a similar manner, except that preincubation and digestion were at 37°C and digestion was allowed to proceed for 20 min. Pulsed-field gels were prepared with Fastlane agarose (FMC Bioproducts) and run in 0.5 X Tris-acetate/EDTA buffer at a temperature of 10°C on a Pulsaphor apparatus (Pharmacia-LKB) using a hexagonal array of electrodes. For resolution between 50 and 150 kb, gels were run for 24 h at 170 V, with linearly ramped pulse times of 1-15 s. For resolution between 100 and 450 kb, gels were run for 24 h at 170 V, with linearly ramped pulse times of lo-25 s. For resolution between 300 to 1100 kb, gels were run at 170 V for 14 h with a pulse time of 50 s, followed by 10 h with a pulse time of 63 s. Gels resolving 800 to 2500 kb were run at 120 V for 48 h, with linearly ramped pulse times of 80-360 s. Molecular weight standards used included Schizosaccharomycespombe chromosomes (FMC Bioproducts), chromosomes prepared from Succharomyces cerevisiae strain AB1380 or Candida albicans (Smith and Cantor, 1987), and X DNA concatamers prepared according to Sambrook et al. (1989). Gels were stained with ethidium bromide to visualize molecular weight markers and treated with two 5-min washes of 0.2 N HCl at room temperature prior to Southern blotting to Gene Screen Plus membranes (DuPont). For rehybridization, filters were boiled for 30 min in a solution of 0.1X SSC, 1% SDS and exposed under film overnight to ensure removal of all probe.
SDS, and 200 and 200 pg/ml 1X SSC, 0.1% at 65°C for 30
RESULTS
Isolation
of Murk
Genomic Clones for DXPas8
The anonymous human probe St14-1 recognizes the murine DXPas8 locus (Avner et al., 1987). It detected a BglII RFLV with a C57BL/6JAWeJ allele of 9.0 kb, a constant band of 4.5 kb, and a Mus spretus allele of 3.5 kb in an interspecific backcross generated in our laboratory (Herman et al., 1991). This variation was used to map DXPa.98 between Gab&I and CamLl in the backcross. Avner et al. (1987) have described a Z’aqI RFLV produced by St14-1 that also mapped between Gabra.3 and CamLl in our cross. For pulsedfield gel analyses, the mildly repetitive human St14-1 probe often produced very weak, ambiguous signals. Mouse genomic clones for DXPas8 were isolated to enable pulsed-field gel mapping and the eventual isolation of yeast artificial chromosome (YAC) clones at this locus. A mouse NIHST3/XFIXII library was screened with a l.O-kb fragment derived from St14-1 (Materials and Methods), and nine strongly cross-reacting clones were isolated. These clones were initially screened for the presence of the previously described BglII RFLV using DNA from C57BL/6JAWMJand Mus spretus mice. As shown in Fig. 1, one of the clones, XDXPas8A, detected the appropriate fragments in a BglII digest and mapped to the appropriate location. The detection of an additional lo-kb band in digests using XDXPas8A most likely reflects additional murine genomic DNA present in the 15-kb X clone as compared to cross-hybridizing sequences in St14-1. A second clone, XDXPas8B, also detected the BgZII
PHYSICAL
MAPPING
OF
Gobra.3,
DXPas8,
CamLl,
Pulsed-Field
Cf-9
lO.O9.0-
I Gflbrajl
DXPas8
4.53.5-
CCVd.1
(RSVP,Cf-8, P3lGc.i~) Dd
FIG. 1. Restriction fragment length variation and mapping of a mouse genomic clone for the DXPasB locus. (a) BglII-digested genomic DNA was probed with the murine genomic clone XDXPas8A (Materials and Methods). B, male with C57BL/6JAWJ allele; S, male with M. spretus allele; 128 and 138 are backcross progeny used to map DXPa.98 in the interspecific cross (21). As with the human %14-l probe, female 128 is heterozygous at the DXPasB locus (lane 3) and more proximal loci and has only C57BL/6JAWeJ alleles at CamLl and more distal loci (not shown). Similarly, male 138 demonstrates a spretus allele at DXPa.98 (lane 4) and more distal loci and a C57BL/6JAWeJ allele at Gabro3 and more proximal loci. The sizes of fragments (in kilobases) are listed to the left of the panel. The band migrating slightly faster than 10 kb for the M. spretlcs allele was observed to migrate at 10 kb on other Southern blots and was always fainter than the corresponding C57BL/6JAW-’ band. (b) Schematic representation of the recombinant X chromosome from mouse 128 (a normal female) and mouse 138 (a normal male) inherited from a heterozygous F, female (B6CBA-AWeJ/A-Bpa x M. spretus). Black areas indicate the presence of a C57BL/6JAWJ allele and white areas a M. spretus allele. Only the region of interest between Cf-9 and Dmd is shown, although both animals were tested at other flanking loci and no double crossovers were detected (21).
RFLV and mapped to the correct location. Both X clones also detected the TuqI RFLV of Avner et al. (data not shown). Using standard restriction mapping techniques, we determined that XDXPas8B was entirely included in XDXPas8A. The other seven strongly hybridizing X clones did not map to the X chromosome as determined by hybridization patterns with a panel of SOmatic cell hybrids containing or lacking a mouse X chromosome, and they were not studied further. The DXPas8 h clones or a HindI subclone of XDXPas8B (Materials and Methods) were used in all pulsed-field gel mapping studies described below.
AND
RSVP
157
Gel Analysis
We have employed pulsed-field gel electrophoresis to construct a physical map including the loci Gubra.3, DXPas8, CamLl, and RSVP in the region of the mouse X chromosome homologous to Xq28. Our strategy for constructing this map was to first demonstrate physical linkage by the sequential hybridization of probes from different loci to single (overlapping) pulsed-field gel fragments, followed by the analysis of double digestion products to enable more detailed mapping of the loci. Our composite map extending approximately 3 million bp is presented in Fig. 2 and the sizes of fragments obtained for each locus are summarized in Table 2. Also illustrated in Fig. 2 are the overlapping fragments essential for construction of the map. Initial restriction digests of thymocyte plugs derived from male and female mice were performed to compare the digestion patterns between male and female DNA. Because of the more extensive methylation of the inactive X chromosome in females, digestion of female DNA with rare-cutting restriction enzymes may yield more partial digestion products, enabling one to physically link different probes. As shown in Fig. 3, digestion of DNA from female mice with the enzymes BssHII and Sac11 yielded several partial digestion fragments not observed in male DNA for the loci DXPas8, CamLl, and RSVP, although no obvious overlaps were detected. Physical linkage of CamLl and RSVP. Preliminary evidence for physical linkage between CamLl and Rsvp was the identification of a 450-kb CluI fragment that hybridized to probes for both loci (Fig. 4A). Physical linkage was confirmed by sequential hybridization of both probes to a 550-kb MluI fragment obtained by a 5-min partial digestion of female DNA (Fig. 4B). The complete MluI digestion products for each of these loci, 145 kb for CamLl and 370 kb for RSVP, add up to 515 kb. This is within a 10% margin of error of the partial digestion product of 550 kb, suggesting that a single MluI site (or several very close sites) lies between the two genes. Subsequent single and double digestions using multiple different restriction enzymes enabled more precise mapping of CamLl and RSVP near opposite ends of the 450-kb overlapping ClaI fragment (Table 2 and Fig. 2). The minimal distance between Rsvp and CamLl is approximately 170 kb. The two loci could not be further localized within the clusters of rarecutter restriction sites associated with the ends of the 450 kb overlapping ClaI fragment; therefore, the maximal distance between them is 450 kb. The high density of rare-cutter restriction sites surrounding CamLl and Rsvp also provides evidence for the existence of at least four CpG islands in the region.
158
FAUST
AND
HERMAN
1CQkb
H
FIG. 2. Composite physical map of the loci studied. The map is drawn approximately to scale, with physical distances given in kilobases. Key overlapping fragments are identified by bars below the map. ten, centromere; tel, telomere; Nr, NruI; M, MluI; N, NotI; B, BssHII; S, SocII; C, &I. For each locus, the precise locations of BssHII sites within the next flanking sites have not been determined. The BssHII sites flanking Gabro.3 are designated by hatched arrows because of the large distance (approximately 1000 kb) in which these sites could lie. Fragments resulting from partial digestion with BssHII and Sac11 are not shown on the map.
Physical linkage of DXPas8 and CamLl. The primary evidence for physical linkage of DXPas8 and CamLl was an overlapping 800 kb Not1 partial digestion fragment (Fig. 4C). The likelihood of this fragment representing the comigration of two different fragments is low since it is the approximate sum of the respective complete digestion fragments-500 kb for DXPas8 and 320 kb for CamLl. MluI/NotI and NruI/ Not1 double digests resulted in 650- and 500-kb fragments for DXPas8 and CamLl fragments of 145 and 130 kb (Table 2). The 650-kb fragment for DXPas8 results from cleavage of the 800-kb partial Not1 fragment. Since the sums of the double digests are approximately 800 kb, these results suggest that only one NruI site and one MluI site exist between the two loci, probably located in the CpG island proximal to CamLl. An interesting phenomenon observed with the 800kb Not1 partial in numerous independent gels was that the intensity of this band always equaled that of the complete digestion band. This was not due to methylation of the site on the inactive X chromosome, since the phenomenon was observed in DNA digested from males and females. Furthermore, the phenomenon was not tissue specific and was also observed in digestions of splenocyte DNA (not shown). One explanation for the intensity of this fragment could be that the Not1 site located between DXPa.98 and CamLl is methylated approximately 50% of the time. Upon Sac11digestion, probes for DXPas8 identified 620-kb complete and 680-kb partial digestion fragments in both male and female DNA (Fig. 3B). Double digestion with SacII/NotI identified a 300-kb fragment containing DXPas8. The lack of hybridization of CamLl with Sac11partials containing DXPa.98 enabled orientation and localization of these sites
within the overall map. The shortest distance between DXPas8 and CamLl is approximately 350 kb. These data, as well as the NotI/CZaI and ClaI/MZuI digestion patterns for CamLl, link DXPa.98 and CamLl within a maximal distance of 750 kb. Complete digestion with BssHII yielded a 240-kb fragment for DXPa.98. The presence of partial digestion fragments in females indicated the existence of other BssHII sites within 480 kb of the locus, although their exact positions have not been determined. Physical linkage of Gabra.3 and DXPas8. Linkage of Gabra.3 and DXPas8 was observed on 2500-kb MZuI, 2350-kb partial NruI, and 1700-kb partial Sac11 fragments (Fig. 5). The Sac11 fragment, obtained by partial digestion of female DNA for 20 min, is the approximate sum of the complete digestion products for Gab& (1125 kb) and DXPas8 (620 kb). The MluI fragment appears to be a complete digestion product, although the presence of a preferentially methylated MluI site between Gabra3 and DXPa.98 cannot be ruled out. However, the same fragments have been observed in male as well as female DNA. NruI fragments less than 2350 kb were occasionally observed after long exposures for both Gabra.3 and DXPas8 (see Table 2). The CamLl probe did not recognize either the 2500-kb MluI or 2350-kb NruI fragments, and so in constructing the map, we assumed these large fragments extend from the NruI and MZuI sites in the CpG island proximal to CamLl through the region containing Gabra3. Probes for Gabra3 recognized complete and partial digestion fragments of 1350 and 1650 kb, respectively, with NotI. Since probes for DXPa.98 did not hybridize to these fragments, we assumed that they originate from the single Not1 site lying proximal to DXPas8.
40
Gdx
560 6608
640*
780
450
450
900 1400*
900 1400*
370 550*
145 550*
2500
2500
MluI
3000
3000
390
320 800:
500 800*
1350 1650*
Not1
1350
1350
210
130
2350*
1 10od** 135od**
lOOO. d These NruI fragments were only occasionally seen as smaller fragments on gels run to resolve 800-2500 kb. They have not been detected on gels resolving lower molecular weight ranges or in double digests with other enzymes. Since the exact positions of these sites could not be determined, they are not included in Fig. 2.
240
230*
170
200* 270' 480*
100
480*
Cf-8
Rsvp
CamLl
> 1000
240
DXPa.98
290*
>lOOO’
CluI
450b 650*
BssHII
Gabra.3
Locus
Summary
TABLE
160
FAUST
AND HERMAN
Double digests with SacII/NotI reduced the size of the 1125-kb Sac11fragment encompassing Gabra.3 by approximately 100 kb, placing a Not1 site within this Sac11 fragment. Digestion with BssHII yielded 450and 650-kb fragments encompassing Gabra3, observed in both male and female DNA (Fig. 3A). It is unclear where the BssHII sites encompassing Gabra3 lie within the 1000 kb of DNA flanked by Not1 and Sac11 sites, and so an accurate minimum distance between Gabra3 and DXPas8 could not be determined. The NotI/SucII digestion data in conjunction with the Sac11 overlapping partial digestion fragment favors a maximal physical distance of about 1600 kb between the two loci. Physical linhage to other loci in the region. When we began these studies, we wished to incorporate Gdx and Cf-8 in this map to clarify the location of these loci relative to RSVP. P3 and Gdx were recently shown to be linked with Cf-8 and G6pd within 400 kb of DNA (Brockdorff et al., 1989). Analysis of BssHII, MluI, and NruI digests with probes for Gdx and Cf-8 yielded fragment sizes consistent with those of Brockdorff et al. (Table 2). However, attempts to link up either of these two loci with the rest of our map by partial digestions were not successful, suggesting the presence of strongly cleaved rare-cutter sites between Rsvp and these loci. Linkage of the loci mapped in this study with Gdx/P3 and Cf-8 will require further analysis.
DISCUSSION Using pulsed-field gel electrophoresis, we have constructed a physical map, including the four loci Gabra3, DXPas8, CamLl, and RSVP, for a region of the mouse X chromosome that is homologous to human Xq28. We were unable to physically link more than two probes on a single pulsed-field gel fragment with any of the rare-cutter restriction enzymes used. That the loci are indeed physically linked rather than merely comigrating on fragments of similar size is suggested by several lines of evidence: Overlaps were detected with two or more restriction enzymes with the exception of DXPas8 and CamLl, where a single Not1 overlap was found. Where partial digestion products linked two loci, the sizes of the complete digestion products would sum to that of the partial within a 10% margin of error. Finally, independently obtained genetic mapping data indicated close linkage for each pair of loci and strongly favored a gene order of cen-
(Cf-9)-Gabra3-DXPas8-CamLl-Rsvp-(Cf-8, GGpd,Gdx)-Dmd-tel (Ryder-Cook et al., 1988; Herman et al., 1991). The genetic mapping data are in agreement with our gene order of Gabra3-DXPas8CamLl-Rsvp obtained by pulsed-field gel analysis and enable orientation with respect to the centromere. The physical map confirms the single recombi-
nant of Ryder-Cook et al. (1988) placing Rsvp proximal to Cf-8, GGpd, P3, and Gdx. Thus, the two mapping approaches-physical and genetic-can provide complementary information and, in this case, each strengthens the validity of the other. Brockdorff et al. (1989) have demonstrated physical linkage of Cf-8, Gdx, P3, and G6pd within 400 kb of each other. The finding of a single recombinant between Rsvp and Cf-8 in more than 700 interspecific backcross progeny suggests that these loci are extremely close (Amar et al., 1988; Mullins et al., 1988; Ryder-Cook et al., 1988; Herman et al., 1991). Thus, our inability to physically link Rsvp with Cf-8 or Gdx/ P3 probably reflects the existence of completely cleaved rare-cutter restriction sites between them, perhaps in CpG islands. Similar difficulties were encountered in pulsed-field gel studies of human Xq28 attempting to link F8 and/or GGPD with the visual pigment genes and overlaps constructed utilized DNA from a 49XXXXY cell line containing three inactive X chromosomes (Arveiler et al., 1989) or from a male with a F8 gene deletion (Kenwrick and Gitschier, 1989). The physical distances that we have obtained by pulsed-field gel mapping are similar to the genetic distances calculated for these markers (Fig. 6). By comparison, in the corresponding region of human synteny, genetic distances are much larger relative to physical. This high genetic recombination probably reflects the telomeric location of the region on the human X chromosome (Hartley et al., 1984). Physical distance comparisons between loci in the region that have been studied in both mouse and human appear similar (Arveiler et al., 1989; Brockdorff et al., 1989; Djabali et al., 1990). Comparison of the murine gene order with that of human Xq28 is difficult due to the considerable controversy involving the ordering of several of the human loci. The current human Xq28 consensus gene order, based on analysis of genetic linkage, in situ hybridization, and pulsed-field gel mapping is cenGABRA3-Stl4(DXS52)-Fs-GGPD-P3 (DXS253E)-GdX (DXS254E)-LlCAM-RCP/GCPtel (Arveiler et al., 1989; Bell et al., 1989; Kenwrick and Gitschier, 1989; Patterson et al., 1989, Djabali et al., 1990; Trask et al., 1991). Discrepancies between the human and murine gene orders exist distal to DXS52, and it is possible that the gene orders of the F8, P3, GdX, and GGPD cluster relative to the LlCAM-RCP/GCP linkage group could be reversed (Herman et al., 1991). While the consensus human data imply at least one evolutionary rearrangement within the region of synteny to account for the altered gene order between the two species, a recently developed physical map of human Xq28 suggests a different gene order for some of
PHYSICAL
MAPPING
OF
Gab&
DXPas8,
CumL1,
B
Gabrad
AND
DXPas8
Q MBSMBS
161
Rsup
d
MBSMBS
-LM
- 820 - 790 - 750 686 620
650 -
600
- 560, 600 480
450 -
440 360
- 440 - 360
280 220
- 280 - 220
Rsvp
CamL 1
Q
c3f
3
YMBSMBSY
d'
MBSMBS
-LM
-LM
-820 -790 -750 -680
-
- 560,600
- 560,600
820 790 750 680
460 270 -
-440 -360
370 -
- 280 -220
- 440 - 360 - 260 - 220
170 -
FIG. 3. Pulsed-field gel analysis of the four loci studied. DNA plugs prepared from male and female thymocytes were digested and run on a single gel resolving fragments between 300 and 1100 kb. Blotting, probing, and rehybridization of the membrane were as described under Materials and Methods. The gel was sequentially hybridized with probes for the Gabtin.3 (A), DXPas8 (B), CamU (C), or Rsup (D) locus. M, Mid; B, BssHII; S, SocII; LM, range of limiting mobility of DNA; Y, yeast strain AB1380 chromosomes. Marker and fragment sizes are given in kilobases to the right and left of the panels, respectively. For fragments greater than 1000 kb or less than 280 kb (see Table 2), sizing was performed on separate gels resolving these molecular weight ranges.
162
FAUST
AND
HERMAN
B
A
C CamL 1
RSVP
CamL 1
-
- 560,600 450ilji g;
- 440
4W-
--^ -44v -364
550 -
Y
900 -
- 560,600
-766 - 660
- 440
370 -
- 360 - 220
- 360 - 220
220
-
560,600
N
N
M
550 -
CamL 1
DXPasS
RSVP
145 -
- 560,600 SW - 440 - 360
.
FIG. 4. Physical linkage of DXPas8, CarnLl, and Rsvp on overlapping pulsed-field gel fragments. DNA prepared from female thymocytes was digested and run as described in Fig. 3. Each pair of autoradiographs represents the same filter sequentially rehybridized with probes for CamLl and Rsup (A and B) or DXPasB and CamLl (C). N, NotI; M, MluI partial digestion; C, Club LM, range of limiting mobility of DNA; Y, yeast strain AB1380 chromosomes. Marker and fragment sizes are given in kilobases to the right and left of the panels, respectively.
the loci (Anne-Marie Poustka and Hans Lehrach, personal communication and X Chromosome Workshop, Oxford, 1991). This map, which utilized DNA prepared from somatic cell hybrids containing Xq28, favors a gene order of ten-GABRA3-DXS15-RCP/ GCP-GGPD-F8-tel. DXS15 is tightly linked to DXS52. This gene order is consistent with that in the mouse, although LlCAM was not included in the map.
A S*
S
DXPasS M Nr
Besides aiding in studies of chromosome evolution, further development of detailed physical maps in this region of the mouse X chromosome could have implications for the cloning of some of the numerous human disease genes located in Xq28. The accurate mapping of human disease genes using genetic linkage is often hindered by a lack of sufficient informative meioses due to the rarity of the disorder or an inability to find polymorphic markers. In human
B N
N/M
Gabra3 S’
NlNr
1700
1126
S
M
Nr
N
NIM
N/Nr
-
1660
-
1360
-
-600
620
-
-
660
-600
FIG. 6. Physical linkage of Gabra.3 and DXPm8. DNA prepared from female thymocytes was digested and electrophoresed under conditions to resolve 800 to 2500 kb (Materials and Methods). The blot was sequentially hybridized with probes for DXPa.98 (A) and Gabro.3 (B). Fragment sizes are given in kilobases to the left and right of the panels. S*, Sac11 partial digestion, S, Sac11 complete digestion; M, MluI; Nr, NruI; N, NotI; LM, range of limiting mobility of DNA. Fragment sizes were determined by comparison to S. cereuisiae and C. olbicans chromosomes.
PHYSICAL 790 Gah3
’
0.410.4
-II DXPas8
MAPPING
OF Gabra3, DXPasB,
450 A
-
Cam LI
-
Rsvp//(P3lG&
Cf - 8)
II *’
0.4+0.4
” 0.8*0.6”
11,
154-164
(1991)
Physical Mapping of the Loci Gabra3, DXPas8, CamLl, and Rsvp in a Region of the Mouse X Chromosome Homologous to Human Xq28 CYNTHIA Institute
for Molecular
J. FAUST AND GAIL E. HERMAN’
Genetics and Department Received
March
of Pediatrics, 5. 1991;
pulsed-field gel electrophoresis, a 3 million-bp map containing the X-linked loci Gubra3, DXPasB, CamLl, and RSVPhas been constructed for a segment of the mouse X chromosome homologous to human Xq28. Detailed mapping was performed using single and double digestions with rare-cutter restriction enzymes. Gabra3 and DXPae8 have been shown to be physically linked within a maximal distance of 1600 kb, DXPas8 and CamLl within 750 kb, and CamLl and Rsvp within 450 kb. In addition, several CpG islands have been detected in the region encompassing CamLl and RSVP.These studies confirm a gene order of ten-Gabra3-DXPaeB-CamLlRsvp-tel determined by genetic mapping in interspecific backcrosses (A. S. Ryder-Cook et al., 1989, EMBO J. 7: 3017-3021; G. E. Herman et al., 1991, Genomics9: 670677). Physical distances for the loci studied agree with the calculated genetic distances. Assuming that there is conserved linkage between man and mouse in the region, the physical mapping data presented here may help to clarify the uncertain gene order for some human Xq28 loci. q iesl Press, Inc.
INTRODUCTION
The conservation of loci on the X chromosomes of and mouse is predicted by Ohno’s Law, which states that the movement of genes between the sex chromosomes and autosomes is restricted due to dosage compensation (Ohno, 1973). The order of Xlinked loci, however, is not necessarily conserved. Comparisons of X-linked loci in mouse and man reveal that at least five chromosomal rearrangements are required to account for the gene order in the two species (Davisson, 1987; Amar et al., 1988). Within each syntenic block, few, if any, rearrangements appear to have occurred. man
1 T O whom correspondence should be addressed at Institute for Molecular Genetics, One Baylor Plaza, S911, Houston, TX 77030. 154
OSsa-7543/91$3.00
Copyright All rights
0 of
by Academic Press, Inc. reproduction in any form reserved. 1991
revised
Houston,
Texas 77030
May 4, 1991
Using mouse interspecific backcrosses, several laboratories have mapped more than 100 loci on the murine X chromosome, including DNA probes for loci mapping to human Xq26-Xq28 (Avner et al., 1987; Brockdorff et aZ., 1987; Amar et aZ., 1988; Mullins et aZ., 1988,199O; Ryder-Cook et al., 1988; Disteche et al., 1989; Herman et al., 1991). This segment of the mouse X chromosome, which spans from Hprt to the Xlinked visual pigment gene (Rsvp), is homologous with human Xq26-Xq28. Human Xq28 contains numerous disease genes, many of which remain to be cloned (Mandel et al., 1989). Mapping in this region of the human X chromosome has been difficult because of high recombination and conflicting genetic and physical maps from different laboratories. Assuming that the gene order between man and mouse is retained in this region, maps constructed in the mouse could prove helpful in determining the correct human gene order. Even if order was not conserved, additional rearrangements detected would be important for studies of mammalian X chromosome evolution. Using an interspecific backcross, our laboratory has previously prepared a genetic map containing several loci in the region of the mouse X chromosome homologous to human Xq28 (Herman et al., 1991). Loci mapped included Gabra3, CamLl, RSVP, and DXPa.98. Gabra3 encodes the a3 subunit of the GABA, receptor, the major inhibitory neurotransmitter in the vertebrate brain. The Ll protein product of the human LlCAM and murine CumLl genes is a neural adhesion molecule and a member of the immunoglobulin superfamily (Moos et al., 1988). Rsvp encodes the murine X-linked visual pigment locus homologous to the human red and green visual pigment genes. In man, the red visual pigment gene is part of a 115kb color vision cluster that also contains one or more green pigment genes arranged in a tandem array (Nathans et al., 1986; Feil et al., 1990a). In mouse, a single X-linked visual pigment gene is believed to be present. The murine DXPas8 locus was originally defined
Using physical
Academic
Baylor College of Medicine,
PHYSICAL
MAPPING
OF
Gabra3, DXPa.98, CamLl,
by hybridization with St14 (DXS52), an anonymous human probe mapping to Xq28 (Oberle et aZ., 1985, 1987; Avner et al., 1987). St14 detects at least three loci in Xq28, although they are located within approximately 600 kb of each other (Feil et aZ., 199Ob). In the mouse, conflicting data concerning the mapping of DXPas8 have been reported (Avner et al., 1987; Herman et al., 1991). Our genetic mapping data strongly favored a gene order of ten-(Cf-9)-Gabra3-DXPasB-CamLl-(RSVP, Gdx, Cf-8)-Dmd-tel for these loci. Rsvp cosegregated with Cf-8, the murine coagulation factor VIII locus, and Gdx, a gene located approximately 40 kb 3’ to G6pd and 0.8 kb 3’to the P3 locus (Filippi et aZ., 1990). To resolve discrepancies between our genetic map and consensus human Xq28 mapping data, and to compare genetic and physical distances, we undertook physical mapping of this region of the mouse X chromosome using pulsed-field gel electrophoresis. We present in this study a physical map encompassing approximately 3 million bp that includes the four loci Gabra.3, DXPasB, CamLl, and RSVP. The results support the genetic mapping data regarding these loci and are discussed in terms of conserved linkage with loci in human Xq28. MATERIALS
Isolation and Mapping the DXPas8 Locus
AND
METHODS
of Mouse Genomic
Clones for
A l.O-kb TaqI restriction fragment from the human anonymous probe St14-1, a subclone of St14 (DXS52) (Mandel et al., 1986), was employed to screen approximately 1 million plaques from a murine NIH3T3/X FIX11 genomic library (Stratagene). Strongly hybridizing clones were mapped in an interspecific backcross using TaqI and BgZII restriction fragment length variations (RFLVs). The interspecific backcross of C57BL/6JAwdJ and Mus spretus has been previously described (Herman et al., 1991) and consists of a regional mapping panel of 120 animals with recombination events spanning the mouse X chromosome and 18 animals with single recombination events between Cf-9 and Dmd in 248 total backcross progeny.
DNA Probes The DNA probes used in this study and their sources, as well as the hybridization and washing conditions employed for each probe, are listed in Table 1. XDXPas8A and XDXPas8B are genomic clones isolated as described above. pDXP8B.l is a l.l-kb Hi&III subclone in pBluescript SK+ (Stratagene) derived from XDXPas8B. DNA probes were prepared using random hexamer labeling (Feinberg and Vogel-
AND
Rsvp
155
stein, 1984) and employed at a concentration of 1 X lo6 cpm/ml hybridization solution for conventional Southern analysis and at a concentration of 2 X lo6 cpm/ml hybridization solution for pulsed-field gel analysis. Prior to hybridization, the probes XDXPas8A, XDXPas8B, and XgMM25 (RSVP) were preannealed for 2 h at 65°C with 1 mg/ml total mouse DNA sonicated to an average molecular weight of 1 kb.
Pulsed-Field
Gel Analysis
DNA plugs were prepared from the thymus of 2to 3-week-old normal male and female progeny of B6CBA-AuJ/A or BSCBA-AwJ/A-Bps females X (C57BL/6JAweJ X CBA) males. This mating scheme results in X chromosomes derived from the inbred C57BL/6JAW” strain (Herman et al., 1991). The Bpa allele represents an X-linked dominant mutation, bare patches, segregating in some of the female parents. Thymocytes were obtained by removing the thymus from sacrificed mice and d&aggregating cells with a Cellector (E-C Apparatus Corp). Plugs were then prepared using a variation of standard methods (Smith and Cantor, 1987; Compton et aZ., 1988). Briefly, isolated thymocytes were resuspended in lysis buffer (10 mM Tris-HCl, pH 8.0,20 mM NaCl, 200 mM EDTA) and mixed with an equal volume of 1.6% Incert agarose (FMC Bioproducts) in lysis buffer to reach a final concentration of 2 X lo7 cells/ml. The mixture was poured into 2.0 X 1.0 X O.l-cm molds and chilled for 5 min. Plugs were transferred to 5 vol of lysis buffer containing 1% N-lauroylsarcosine/2 mg/ml proteinase K and incubated for 24 h at 45-50°C. The buffer was exchanged and the plugs were incubated another 24 h. Plugs were dialyzed twice in TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA) containing 1 mM phenylmethylsulfonyl fluoride (PMSF), followed by three dialyses in TE buffer without PMSF, and were stored at 4°C in 0.5 M EDTA. Plugs were prepared from splenocytes of adult animals using the above protocol except that red blood cells were lysed using hypotonic saline after initial centrifugation. Restriction digests were performed using onequarter to one-third plug (approximately l-2 X lo6 cells). Immediately prior to digestion, individual plugs were washed four times with 1 ml TE, followed by equilibration in 1 ml of the appropriate digestion buffer for 30 min. Plugs were digested for 4 h in 250~~1 reaction mixtures containing the buffer recommended by the enzyme manufacturer, 2 mM spermidine, and 30 U of enzyme (Stratagene or BoehringerMannheim). For digests with BssHII (New England Biolabs), 16 U of enzyme were employed. Partial digestions with MZuI were performed by preincubating
156
FAUST
AND
HERMAN
TABLE DNA Probe
Locus
pbGRa3 (bovine partial cDNA) St14-1 (human genomic) XDXPas8A (mouse genomic) XDXPas8B (mouse genomic) pDXP8B.l (mouse genomic) pK13 (mouse partial cDNA) hs7 (human partial cDNA) XgMM25 (mouse genomic) G28B (mouse genomic) pKpn-sac (human 5’ partial cDNA)
Gabra.3 DXPa.98 DXPas8 DXPas8 DXPas8 CamLl Rsup Rsvp Gdx
Probes
1
Used in This
Hybridization
Cf-8
Study
conditions”
Wash
I II III III II I II III II II
conditionsb A B C c C C A C C A
Source E. A. Barnard (24) J.-L. Mandel (25) This work This work This work M. Schachner (27) J. Nathans (31) J. Nathans P. Avner (3) Genetics Institute
a Hybridization conditions were as follows: I, hybridization at 65°C in 1 M NaCl, 10% dextran sulphate (Pharmacia), 1% pg/ml denatured herring sperm DNA. II, hybridization at 42°C in 40% formamide, 1 M NaCl, 10% dextran sulphate, 1% SDS, denatured herring sperm DNA. III, protocol I plus 100 rig/ml total mouse DNA sonicated to an average size of 1 kb. b Wash conditions were as follows: A, three washes with 3X SSC, 0.05% SDS. B, twice with 2X SSC, 0.1% SDS, twice with SDS. C. seauential washes of 2X SSC. 0.1% SDS: 1X SSC. 0.1% SDS; 0.5X SSC, 0.1% SDS. All washes were performed minutes. 1X SSC is 0.15 M NaCl, 0.015 M Na citrate, pH 710.
plugs for 2 h at 30°C in restriction buffer without MgCl, , followed by the addition of 10 mM MgCl, for 5 min. The reactions were terminated by the addition of EDTA to 20 n&f. Partial digestions with Sac11 were performed in a similar manner, except that preincubation and digestion were at 37°C and digestion was allowed to proceed for 20 min. Pulsed-field gels were prepared with Fastlane agarose (FMC Bioproducts) and run in 0.5 X Tris-acetate/EDTA buffer at a temperature of 10°C on a Pulsaphor apparatus (Pharmacia-LKB) using a hexagonal array of electrodes. For resolution between 50 and 150 kb, gels were run for 24 h at 170 V, with linearly ramped pulse times of 1-15 s. For resolution between 100 and 450 kb, gels were run for 24 h at 170 V, with linearly ramped pulse times of lo-25 s. For resolution between 300 to 1100 kb, gels were run at 170 V for 14 h with a pulse time of 50 s, followed by 10 h with a pulse time of 63 s. Gels resolving 800 to 2500 kb were run at 120 V for 48 h, with linearly ramped pulse times of 80-360 s. Molecular weight standards used included Schizosaccharomycespombe chromosomes (FMC Bioproducts), chromosomes prepared from Succharomyces cerevisiae strain AB1380 or Candida albicans (Smith and Cantor, 1987), and X DNA concatamers prepared according to Sambrook et al. (1989). Gels were stained with ethidium bromide to visualize molecular weight markers and treated with two 5-min washes of 0.2 N HCl at room temperature prior to Southern blotting to Gene Screen Plus membranes (DuPont). For rehybridization, filters were boiled for 30 min in a solution of 0.1X SSC, 1% SDS and exposed under film overnight to ensure removal of all probe.
SDS, and 200 and 200 pg/ml 1X SSC, 0.1% at 65°C for 30
RESULTS
Isolation
of Murk
Genomic Clones for DXPas8
The anonymous human probe St14-1 recognizes the murine DXPas8 locus (Avner et al., 1987). It detected a BglII RFLV with a C57BL/6JAWeJ allele of 9.0 kb, a constant band of 4.5 kb, and a Mus spretus allele of 3.5 kb in an interspecific backcross generated in our laboratory (Herman et al., 1991). This variation was used to map DXPa.98 between Gab&I and CamLl in the backcross. Avner et al. (1987) have described a Z’aqI RFLV produced by St14-1 that also mapped between Gabra.3 and CamLl in our cross. For pulsedfield gel analyses, the mildly repetitive human St14-1 probe often produced very weak, ambiguous signals. Mouse genomic clones for DXPas8 were isolated to enable pulsed-field gel mapping and the eventual isolation of yeast artificial chromosome (YAC) clones at this locus. A mouse NIHST3/XFIXII library was screened with a l.O-kb fragment derived from St14-1 (Materials and Methods), and nine strongly cross-reacting clones were isolated. These clones were initially screened for the presence of the previously described BglII RFLV using DNA from C57BL/6JAWMJand Mus spretus mice. As shown in Fig. 1, one of the clones, XDXPas8A, detected the appropriate fragments in a BglII digest and mapped to the appropriate location. The detection of an additional lo-kb band in digests using XDXPas8A most likely reflects additional murine genomic DNA present in the 15-kb X clone as compared to cross-hybridizing sequences in St14-1. A second clone, XDXPas8B, also detected the BgZII
PHYSICAL
MAPPING
OF
Gobra.3,
DXPas8,
CamLl,
Pulsed-Field
Cf-9
lO.O9.0-
I Gflbrajl
DXPas8
4.53.5-
CCVd.1
(RSVP,Cf-8, P3lGc.i~) Dd
FIG. 1. Restriction fragment length variation and mapping of a mouse genomic clone for the DXPasB locus. (a) BglII-digested genomic DNA was probed with the murine genomic clone XDXPas8A (Materials and Methods). B, male with C57BL/6JAWJ allele; S, male with M. spretus allele; 128 and 138 are backcross progeny used to map DXPa.98 in the interspecific cross (21). As with the human %14-l probe, female 128 is heterozygous at the DXPasB locus (lane 3) and more proximal loci and has only C57BL/6JAWeJ alleles at CamLl and more distal loci (not shown). Similarly, male 138 demonstrates a spretus allele at DXPa.98 (lane 4) and more distal loci and a C57BL/6JAWeJ allele at Gabro3 and more proximal loci. The sizes of fragments (in kilobases) are listed to the left of the panel. The band migrating slightly faster than 10 kb for the M. spretlcs allele was observed to migrate at 10 kb on other Southern blots and was always fainter than the corresponding C57BL/6JAW-’ band. (b) Schematic representation of the recombinant X chromosome from mouse 128 (a normal female) and mouse 138 (a normal male) inherited from a heterozygous F, female (B6CBA-AWeJ/A-Bpa x M. spretus). Black areas indicate the presence of a C57BL/6JAWJ allele and white areas a M. spretus allele. Only the region of interest between Cf-9 and Dmd is shown, although both animals were tested at other flanking loci and no double crossovers were detected (21).
RFLV and mapped to the correct location. Both X clones also detected the TuqI RFLV of Avner et al. (data not shown). Using standard restriction mapping techniques, we determined that XDXPas8B was entirely included in XDXPas8A. The other seven strongly hybridizing X clones did not map to the X chromosome as determined by hybridization patterns with a panel of SOmatic cell hybrids containing or lacking a mouse X chromosome, and they were not studied further. The DXPas8 h clones or a HindI subclone of XDXPas8B (Materials and Methods) were used in all pulsed-field gel mapping studies described below.
AND
RSVP
157
Gel Analysis
We have employed pulsed-field gel electrophoresis to construct a physical map including the loci Gubra.3, DXPas8, CamLl, and RSVP in the region of the mouse X chromosome homologous to Xq28. Our strategy for constructing this map was to first demonstrate physical linkage by the sequential hybridization of probes from different loci to single (overlapping) pulsed-field gel fragments, followed by the analysis of double digestion products to enable more detailed mapping of the loci. Our composite map extending approximately 3 million bp is presented in Fig. 2 and the sizes of fragments obtained for each locus are summarized in Table 2. Also illustrated in Fig. 2 are the overlapping fragments essential for construction of the map. Initial restriction digests of thymocyte plugs derived from male and female mice were performed to compare the digestion patterns between male and female DNA. Because of the more extensive methylation of the inactive X chromosome in females, digestion of female DNA with rare-cutting restriction enzymes may yield more partial digestion products, enabling one to physically link different probes. As shown in Fig. 3, digestion of DNA from female mice with the enzymes BssHII and Sac11 yielded several partial digestion fragments not observed in male DNA for the loci DXPas8, CamLl, and RSVP, although no obvious overlaps were detected. Physical linkage of CamLl and RSVP. Preliminary evidence for physical linkage between CamLl and Rsvp was the identification of a 450-kb CluI fragment that hybridized to probes for both loci (Fig. 4A). Physical linkage was confirmed by sequential hybridization of both probes to a 550-kb MluI fragment obtained by a 5-min partial digestion of female DNA (Fig. 4B). The complete MluI digestion products for each of these loci, 145 kb for CamLl and 370 kb for RSVP, add up to 515 kb. This is within a 10% margin of error of the partial digestion product of 550 kb, suggesting that a single MluI site (or several very close sites) lies between the two genes. Subsequent single and double digestions using multiple different restriction enzymes enabled more precise mapping of CamLl and RSVP near opposite ends of the 450-kb overlapping ClaI fragment (Table 2 and Fig. 2). The minimal distance between Rsvp and CamLl is approximately 170 kb. The two loci could not be further localized within the clusters of rarecutter restriction sites associated with the ends of the 450 kb overlapping ClaI fragment; therefore, the maximal distance between them is 450 kb. The high density of rare-cutter restriction sites surrounding CamLl and Rsvp also provides evidence for the existence of at least four CpG islands in the region.
158
FAUST
AND
HERMAN
1CQkb
H
FIG. 2. Composite physical map of the loci studied. The map is drawn approximately to scale, with physical distances given in kilobases. Key overlapping fragments are identified by bars below the map. ten, centromere; tel, telomere; Nr, NruI; M, MluI; N, NotI; B, BssHII; S, SocII; C, &I. For each locus, the precise locations of BssHII sites within the next flanking sites have not been determined. The BssHII sites flanking Gabro.3 are designated by hatched arrows because of the large distance (approximately 1000 kb) in which these sites could lie. Fragments resulting from partial digestion with BssHII and Sac11 are not shown on the map.
Physical linkage of DXPas8 and CamLl. The primary evidence for physical linkage of DXPas8 and CamLl was an overlapping 800 kb Not1 partial digestion fragment (Fig. 4C). The likelihood of this fragment representing the comigration of two different fragments is low since it is the approximate sum of the respective complete digestion fragments-500 kb for DXPas8 and 320 kb for CamLl. MluI/NotI and NruI/ Not1 double digests resulted in 650- and 500-kb fragments for DXPas8 and CamLl fragments of 145 and 130 kb (Table 2). The 650-kb fragment for DXPas8 results from cleavage of the 800-kb partial Not1 fragment. Since the sums of the double digests are approximately 800 kb, these results suggest that only one NruI site and one MluI site exist between the two loci, probably located in the CpG island proximal to CamLl. An interesting phenomenon observed with the 800kb Not1 partial in numerous independent gels was that the intensity of this band always equaled that of the complete digestion band. This was not due to methylation of the site on the inactive X chromosome, since the phenomenon was observed in DNA digested from males and females. Furthermore, the phenomenon was not tissue specific and was also observed in digestions of splenocyte DNA (not shown). One explanation for the intensity of this fragment could be that the Not1 site located between DXPa.98 and CamLl is methylated approximately 50% of the time. Upon Sac11digestion, probes for DXPas8 identified 620-kb complete and 680-kb partial digestion fragments in both male and female DNA (Fig. 3B). Double digestion with SacII/NotI identified a 300-kb fragment containing DXPas8. The lack of hybridization of CamLl with Sac11partials containing DXPa.98 enabled orientation and localization of these sites
within the overall map. The shortest distance between DXPas8 and CamLl is approximately 350 kb. These data, as well as the NotI/CZaI and ClaI/MZuI digestion patterns for CamLl, link DXPa.98 and CamLl within a maximal distance of 750 kb. Complete digestion with BssHII yielded a 240-kb fragment for DXPa.98. The presence of partial digestion fragments in females indicated the existence of other BssHII sites within 480 kb of the locus, although their exact positions have not been determined. Physical linkage of Gabra.3 and DXPas8. Linkage of Gabra.3 and DXPas8 was observed on 2500-kb MZuI, 2350-kb partial NruI, and 1700-kb partial Sac11 fragments (Fig. 5). The Sac11 fragment, obtained by partial digestion of female DNA for 20 min, is the approximate sum of the complete digestion products for Gab& (1125 kb) and DXPas8 (620 kb). The MluI fragment appears to be a complete digestion product, although the presence of a preferentially methylated MluI site between Gabra3 and DXPa.98 cannot be ruled out. However, the same fragments have been observed in male as well as female DNA. NruI fragments less than 2350 kb were occasionally observed after long exposures for both Gabra.3 and DXPas8 (see Table 2). The CamLl probe did not recognize either the 2500-kb MluI or 2350-kb NruI fragments, and so in constructing the map, we assumed these large fragments extend from the NruI and MZuI sites in the CpG island proximal to CamLl through the region containing Gabra3. Probes for Gabra3 recognized complete and partial digestion fragments of 1350 and 1650 kb, respectively, with NotI. Since probes for DXPa.98 did not hybridize to these fragments, we assumed that they originate from the single Not1 site lying proximal to DXPas8.
40
Gdx
560 6608
640*
780
450
450
900 1400*
900 1400*
370 550*
145 550*
2500
2500
MluI
3000
3000
390
320 800:
500 800*
1350 1650*
Not1
1350
1350
210
130
2350*
1 10od** 135od**
lOOO. d These NruI fragments were only occasionally seen as smaller fragments on gels run to resolve 800-2500 kb. They have not been detected on gels resolving lower molecular weight ranges or in double digests with other enzymes. Since the exact positions of these sites could not be determined, they are not included in Fig. 2.
240
230*
170
200* 270' 480*
100
480*
Cf-8
Rsvp
CamLl
> 1000
240
DXPa.98
290*
>lOOO’
CluI
450b 650*
BssHII
Gabra.3
Locus
Summary
TABLE
160
FAUST
AND HERMAN
Double digests with SacII/NotI reduced the size of the 1125-kb Sac11fragment encompassing Gabra.3 by approximately 100 kb, placing a Not1 site within this Sac11 fragment. Digestion with BssHII yielded 450and 650-kb fragments encompassing Gabra3, observed in both male and female DNA (Fig. 3A). It is unclear where the BssHII sites encompassing Gabra3 lie within the 1000 kb of DNA flanked by Not1 and Sac11 sites, and so an accurate minimum distance between Gabra3 and DXPas8 could not be determined. The NotI/SucII digestion data in conjunction with the Sac11 overlapping partial digestion fragment favors a maximal physical distance of about 1600 kb between the two loci. Physical linhage to other loci in the region. When we began these studies, we wished to incorporate Gdx and Cf-8 in this map to clarify the location of these loci relative to RSVP. P3 and Gdx were recently shown to be linked with Cf-8 and G6pd within 400 kb of DNA (Brockdorff et al., 1989). Analysis of BssHII, MluI, and NruI digests with probes for Gdx and Cf-8 yielded fragment sizes consistent with those of Brockdorff et al. (Table 2). However, attempts to link up either of these two loci with the rest of our map by partial digestions were not successful, suggesting the presence of strongly cleaved rare-cutter sites between Rsvp and these loci. Linkage of the loci mapped in this study with Gdx/P3 and Cf-8 will require further analysis.
DISCUSSION Using pulsed-field gel electrophoresis, we have constructed a physical map, including the four loci Gabra3, DXPas8, CamLl, and RSVP, for a region of the mouse X chromosome that is homologous to human Xq28. We were unable to physically link more than two probes on a single pulsed-field gel fragment with any of the rare-cutter restriction enzymes used. That the loci are indeed physically linked rather than merely comigrating on fragments of similar size is suggested by several lines of evidence: Overlaps were detected with two or more restriction enzymes with the exception of DXPas8 and CamLl, where a single Not1 overlap was found. Where partial digestion products linked two loci, the sizes of the complete digestion products would sum to that of the partial within a 10% margin of error. Finally, independently obtained genetic mapping data indicated close linkage for each pair of loci and strongly favored a gene order of cen-
(Cf-9)-Gabra3-DXPas8-CamLl-Rsvp-(Cf-8, GGpd,Gdx)-Dmd-tel (Ryder-Cook et al., 1988; Herman et al., 1991). The genetic mapping data are in agreement with our gene order of Gabra3-DXPas8CamLl-Rsvp obtained by pulsed-field gel analysis and enable orientation with respect to the centromere. The physical map confirms the single recombi-
nant of Ryder-Cook et al. (1988) placing Rsvp proximal to Cf-8, GGpd, P3, and Gdx. Thus, the two mapping approaches-physical and genetic-can provide complementary information and, in this case, each strengthens the validity of the other. Brockdorff et al. (1989) have demonstrated physical linkage of Cf-8, Gdx, P3, and G6pd within 400 kb of each other. The finding of a single recombinant between Rsvp and Cf-8 in more than 700 interspecific backcross progeny suggests that these loci are extremely close (Amar et al., 1988; Mullins et al., 1988; Ryder-Cook et al., 1988; Herman et al., 1991). Thus, our inability to physically link Rsvp with Cf-8 or Gdx/ P3 probably reflects the existence of completely cleaved rare-cutter restriction sites between them, perhaps in CpG islands. Similar difficulties were encountered in pulsed-field gel studies of human Xq28 attempting to link F8 and/or GGPD with the visual pigment genes and overlaps constructed utilized DNA from a 49XXXXY cell line containing three inactive X chromosomes (Arveiler et al., 1989) or from a male with a F8 gene deletion (Kenwrick and Gitschier, 1989). The physical distances that we have obtained by pulsed-field gel mapping are similar to the genetic distances calculated for these markers (Fig. 6). By comparison, in the corresponding region of human synteny, genetic distances are much larger relative to physical. This high genetic recombination probably reflects the telomeric location of the region on the human X chromosome (Hartley et al., 1984). Physical distance comparisons between loci in the region that have been studied in both mouse and human appear similar (Arveiler et al., 1989; Brockdorff et al., 1989; Djabali et al., 1990). Comparison of the murine gene order with that of human Xq28 is difficult due to the considerable controversy involving the ordering of several of the human loci. The current human Xq28 consensus gene order, based on analysis of genetic linkage, in situ hybridization, and pulsed-field gel mapping is cenGABRA3-Stl4(DXS52)-Fs-GGPD-P3 (DXS253E)-GdX (DXS254E)-LlCAM-RCP/GCPtel (Arveiler et al., 1989; Bell et al., 1989; Kenwrick and Gitschier, 1989; Patterson et al., 1989, Djabali et al., 1990; Trask et al., 1991). Discrepancies between the human and murine gene orders exist distal to DXS52, and it is possible that the gene orders of the F8, P3, GdX, and GGPD cluster relative to the LlCAM-RCP/GCP linkage group could be reversed (Herman et al., 1991). While the consensus human data imply at least one evolutionary rearrangement within the region of synteny to account for the altered gene order between the two species, a recently developed physical map of human Xq28 suggests a different gene order for some of
PHYSICAL
MAPPING
OF
Gab&
DXPas8,
CumL1,
B
Gabrad
AND
DXPas8
Q MBSMBS
161
Rsup
d
MBSMBS
-LM
- 820 - 790 - 750 686 620
650 -
600
- 560, 600 480
450 -
440 360
- 440 - 360
280 220
- 280 - 220
Rsvp
CamL 1
Q
c3f
3
YMBSMBSY
d'
MBSMBS
-LM
-LM
-820 -790 -750 -680
-
- 560,600
- 560,600
820 790 750 680
460 270 -
-440 -360
370 -
- 280 -220
- 440 - 360 - 260 - 220
170 -
FIG. 3. Pulsed-field gel analysis of the four loci studied. DNA plugs prepared from male and female thymocytes were digested and run on a single gel resolving fragments between 300 and 1100 kb. Blotting, probing, and rehybridization of the membrane were as described under Materials and Methods. The gel was sequentially hybridized with probes for the Gabtin.3 (A), DXPas8 (B), CamU (C), or Rsup (D) locus. M, Mid; B, BssHII; S, SocII; LM, range of limiting mobility of DNA; Y, yeast strain AB1380 chromosomes. Marker and fragment sizes are given in kilobases to the right and left of the panels, respectively. For fragments greater than 1000 kb or less than 280 kb (see Table 2), sizing was performed on separate gels resolving these molecular weight ranges.
162
FAUST
AND
HERMAN
B
A
C CamL 1
RSVP
CamL 1
-
- 560,600 450ilji g;
- 440
4W-
--^ -44v -364
550 -
Y
900 -
- 560,600
-766 - 660
- 440
370 -
- 360 - 220
- 360 - 220
220
-
560,600
N
N
M
550 -
CamL 1
DXPasS
RSVP
145 -
- 560,600 SW - 440 - 360
.
FIG. 4. Physical linkage of DXPas8, CarnLl, and Rsvp on overlapping pulsed-field gel fragments. DNA prepared from female thymocytes was digested and run as described in Fig. 3. Each pair of autoradiographs represents the same filter sequentially rehybridized with probes for CamLl and Rsup (A and B) or DXPasB and CamLl (C). N, NotI; M, MluI partial digestion; C, Club LM, range of limiting mobility of DNA; Y, yeast strain AB1380 chromosomes. Marker and fragment sizes are given in kilobases to the right and left of the panels, respectively.
the loci (Anne-Marie Poustka and Hans Lehrach, personal communication and X Chromosome Workshop, Oxford, 1991). This map, which utilized DNA prepared from somatic cell hybrids containing Xq28, favors a gene order of ten-GABRA3-DXS15-RCP/ GCP-GGPD-F8-tel. DXS15 is tightly linked to DXS52. This gene order is consistent with that in the mouse, although LlCAM was not included in the map.
A S*
S
DXPasS M Nr
Besides aiding in studies of chromosome evolution, further development of detailed physical maps in this region of the mouse X chromosome could have implications for the cloning of some of the numerous human disease genes located in Xq28. The accurate mapping of human disease genes using genetic linkage is often hindered by a lack of sufficient informative meioses due to the rarity of the disorder or an inability to find polymorphic markers. In human
B N
N/M
Gabra3 S’
NlNr
1700
1126
S
M
Nr
N
NIM
N/Nr
-
1660
-
1360
-
-600
620
-
-
660
-600
FIG. 6. Physical linkage of Gabra.3 and DXPm8. DNA prepared from female thymocytes was digested and electrophoresed under conditions to resolve 800 to 2500 kb (Materials and Methods). The blot was sequentially hybridized with probes for DXPa.98 (A) and Gabro.3 (B). Fragment sizes are given in kilobases to the left and right of the panels. S*, Sac11 partial digestion, S, Sac11 complete digestion; M, MluI; Nr, NruI; N, NotI; LM, range of limiting mobility of DNA. Fragment sizes were determined by comparison to S. cereuisiae and C. olbicans chromosomes.
PHYSICAL 790 Gah3
’
0.410.4
-II DXPas8
MAPPING
OF Gabra3, DXPasB,
450 A
-
Cam LI
-
Rsvp//(P3lG&
Cf - 8)
II *’
0.4+0.4
” 0.8*0.6”
