Am. J. Hum. Genet. 44:100-106, 1989

Selection against Lethal Alleles in Females Heterozygous forIncontinentia PigmentiBarbara R. Migeon, * Joyce Axelman, * Suzanne Jan de Beur, * David Valle, *Grant A. Mitchell,*9t and Kenneth N. Rosenbauml

Department of Pediatrics and TLaboratory of Genetics, Howard Hughes Medical Institute, Johns Hopkins University School of Medicine,Baltimore; and tDepartment of Medical Genetics, Children's Hospital National Medical Center, Washington, DC

Summary

Studies of five heterozygous females from three kindreds segregating incontinentia pigmenti indicate thatcells expressing the mutation have been eliminated from skin fibroblast cultures and in varying degreesfrom hematopoietic tissues. Clonal analysis was carried out using G6PD variants and methylation patternsat the HPRT locus. Our results confirm X linkage in these families and suggest that selection against cellsexpressing mutations that are lethal to males in utero may help ameliorate the deleterious phenotype incarrier females.

Introduction

Several X-linked diseases have been identified in whichsurvivors are predominantly females (reviewed by Wett-ke-Schafer and Kantner 1983). In such diseases, the lossof hemizygous males may be well documented; for ex-ample, most males with severe ornithine transcarbamy-lase deficiency survive only a short while after birth.In disorders where afflicted males are rarely observed,their loss is presumed to occur in utero, and the increasedfrequency of abortions in some kindreds has supportedthis assumption (Kelly et al. 1976). In this case, theabnormality in male fetuses is not rescued by maternalfactors, precluding survival to birth. Females with thesame mutant allele are more mildly affected becausethey are heterozygous, and, as a consequence of X-chro-mosome inactivation, have a population of cells thatexpress the normal allele at this locus (Davidson et al.1963). Furthermore, mosaicism of this kind might fur-ther attenuate the deleterious effects of the mutation,if mutant-type cells were eliminated as a result of aproliferative disadvantage.One disorder that is observed predominantly in fe-

Received May 11, 1988; revision received August 11, 1988.Address for correspondence and reprints: Barbara R. Migeon,

CMSC 10-04, The Johns Hopkins Hospital, Baltimore, MD 21205.i 1989 by The American Society of Human Genetics. All rights reserved.0002-9297/89/4401-0016$02.00

males is incontinentia pigmenti (IP). This disturbanceof skin pigmentation is associated with a variety of de-velopmental anomalies of the hair, teeth, eyes, and cen-tral nervous system, including varying degrees of men-tal retardation in some individuals (reviewed by Carney1976). In marked contrast to the lethality of the muta-tion in males, the disease in females is usually not aserious handicap. The few males who have been ob-served postnatally are usually sporadic cases and arepresumed to be mosaic for the mutation, because ofhalf-chromatid mutations (Lenz 1975) or new muta-tions occurring after the first cleavage of the zygote.The disorder is thought to be X linked, based on thepattern of transmission, and a relevant locus may besituated on the short arm of the X chromosome basedon the presence of de novo X-autosome translocationswith a common X breakpoint at Xp1l in six affectedfemales (reviewed in Harris et al. 1988).On the basis of preliminary studies of a single het-

erozygote, one of us previously suggested that selec-tion disfavoring the mutation (cells expressing the mu-tation) might help ameliorate the deleterious effects ofthe mutant phenotype in heterozygotes (Migeon 1978).The female mildly affected with the disorder was alsoheterozygous for glucose-6-phosphate dehydrogenase(G6PD), but she had eliminated blood cells and skinfibroblasts expressing her mother's G6PD allele. Sub-sequently, Wieacker et al. (1985) identified an affectedfemale who was also heterozygous for an RFLP at an

100

Selection in Incontinentia Pigmenti

anonymous locus on the short arm of the X chromo-some, and they fused her skin fibroblasts with mousecells. They found that the active human X chromosomein these hybrids was not random; such results are com-patible with mosaicism in this female, being skewedin favor of one of the alleles at the IP locus, but becausethe coupling phase was unknown, it was not knownwhether the mutation was disfavored.We now report studies of five heterozygotes from three

independent kindreds that indicate selection disfavor-ing the IP mutation occurs not only in skin cells, wherethe mutation is frequently expressed, but, in some cases,also in hematopoietic tissue, which has not previouslybeen considered a target for the mutation.

Strategy

To explore the possibility that selection plays a rolein ameliorating mutations at the IP locus, we identifiedheterozygotes who were also heterozygous for commonelectrophoretic variants of G6PD, which could serveas markers for analysis of X-chromosome mosaicism.We determined the G6PD phenotype of blood cells andskin fibroblast cultures from these doubly heterozygousfemales and their parents, to look for skewed patternsof mosaicism and to determine the coupling of G6PDand IP alleles. From the G6PD analysis of skin fibro-blast clones, we could directly determine the propor-tion of the two cell populations in these cultures (fig. 1).

,, t i .-

B -A 1 m

Figure I G6PD analysis of S.D. and parents. Shown are resultsof cellulose acetate electrophoresis of G6PD from red blood cells (leftpanel) and skin fibroblasts (right panel): father (lane 1), mother (lane2), S.D. (lane 3), G6PD AB heterozygote control (lane 4), G6PD Bfibroblast clone from S.D. (lane 5), uncloned culture that was thesource of S.D. clones (lane 6), G6PD A clone from S.D. (lane 7),G6PD B control (lane 8).

To look for skewing of cell populations in affectedfemales who were not heterozygous for G6PD, we usedthe restriction-enzyme analysis of DNA described byVogelstein et al. (1985, 1987). This method of clonalanalysis is based on the observation ofWolf et al. (1984a, 1984b) and Yen et al. (1984) that X-linked house-keeping loci are differentially methylated on active andinactive X chromosomes. For example, the body of theHPRT gene on the active X is extensively methylated,whereas the same sequence on the inactive X is lessmethylated. Therefore, methylation-sensitive restrictionendonucleases can distinguish the active locus from theinactive one and can be used in conjunction with RFLPsof methylation-insensitive endonucleases to distinguishthe parental origin of HPRT alleles (e.g., the BamHI12/24-kb polymorphism described by Nussbaum et al.1983). In brief, the assay shown in figure 2 consists ofprobing blots of genomic DNA, digested with restric-tion enzymes, with probes from the HPRT locus: theBamHI RFLP distinguishes the maternal X from thepaternal one (fig. 2a), and a double digest with BamHIplus methylation-sensitive HpaII distinguishes the ac-tive from the inactive X, as the BamHI fragment de-rived from the active X is relatively insensitive to HpaII

BamHI

24

B P B. Pv B 0.8

1ff Ir it,, I0 1.7 12 18 24 kb_ a

BamHI + Hpall

b c d

Figure 2 Strategy for clonal analysis using RFLPs and DNAmethylation of the HPRT locus in a female heterozygous for the 12/24-kb BamHI polymorphism (adapted from Vogelstein et al. 1985). Atleft is a map of the locus, showing BamHI (B), PstI (P), PvuII (Pv),and MspI (B*) sites. The asterisk (*) indicates the polymorphic sitethat produces a 12-kb fragment in BamHI digests and an 11.2-kbfragment in BamHI/MspI double digests, when blots are hybridizedwith HPRT intron probes (black boxes). Fragments from active X(Xa) and inactive X (Xi) chromosomes are indicated. a, Blot show-ing 12-kb and 24-kb fragments in BamHI digest from heterozygote.b-d, BamHI/HpaII double digests from the same heterozygote, show-ing results when X inactivation is random (b) or clonal (c and d),with chromosome with 24-kb fragment active in all cells (c) or withX with the 12-kb fragment active in all cells (d). In double digests,the fragments from the inactiveX decrease in size because the HpaII-MspI sites (vertical lines with round heads) are relatively unmethylatedon this chromosome. The fragments from the active X remain large,as these sites are methylated and therefore are not cut.

101

Migeon et al.

and remains intact. The HPRT locus does not have to

be near the gene of interest, as it serves to mark thechromosome in the same way as G6PD variants do.

Material and Methods

Subjects

Family l.-Proband S.D., a black female, was bornwith swirls of hyperpigmentation on her back and withreticulate hyperpigmentation on her extremities. At age

6 mo a skin biopsy revealed incontinence of pigmentin many areas of the dermis and mononuclear cellinfiltrate consisting of lymphocytes and epithelioid cells.Two other biopsies, one from the hyperpigmented area

and the other from normal skin, were used to establishcell cultures. Her 18-year-old mother, P.W., also hadswirls of hyperpigmentation on her arms, but skin bi-opsy at that age revealed no pigment incontinence (notsurprising, as this seems to disappear with age).

Family 2.-Proband R.M., a 5-year-old black female,had a hyperpigmented swirllike nevus over her nose,

face, midback, and buttocks. Her central incisors wereconical and widely spaced, and her lateral incisors weremissing. The diagnosis of IP was confirmed by histo-logical examination of the affected skin. Her 21-year-old mother, K. A., had four missing teeth, a historyof hyperpigmented whorls, and a previous miscarriage,and most likely is also heterozygous for IP. Skin fibro-blast cultures were established from both females.

Family 3.-A Caucasian female (E.S.) and her threedaughters (D.S., P.S., and T.S.) had varying manifesta-tions of IP. Table 1 presents clinical findings on affected

females in this family and summarizes phenotypes ofother heterozygotes.

Cells and Tissues

Blood samples anticoagulated with EDTA were ob-tained from all affected females and their fathers, whenpossible, for analysis ofG6PD phenotype and/or DNAanalysis.

Clones were obtained from skin fibroblasts that hadbeen subcultured only once, by plating 10 cells/60-mmdish containing MEM medium enriched with nones-

sential amino acids and 20% human serum. After 10days in the incubator, well-isolated clones were pickedusing cloning cylinders, and clonal cultures were estab-lished. When the clonal cultures had proliferated enoughto fill a 35-mm dish, they were subcultured into two

60-mm dishes; one was used for G6PD analysis.

G6PD Analysis

Extracts prepared from red cells, leukocytes, andfibroblasts were electrophoresed on cellulose acetate gels(Migeon et al. 1985).

Restriction Enzyme Analysis

DNA was purified from blood leukocytes and skinfibroblast cultures (Wolf et al. 1984b). Approximately5 x 106 fibroblasts in their second subculture were

harvested in hopes that this limited expansion of theoriginal culture would minimize effects of in vitroproliferation.

Southern blots.-Ten micrograms of high-molecular-weight DNA was digested with BamHI alone or BamHI

Table I

Clinical Features of Females with Incontinentia Pigmenti

Family and Age Neonatal Hyperpigmented Dental Mental DiagnosticSubject (Years) Vesicles Whorls Anomalies Retardation Histopathology

Family 1:S.D. .6 + + - - +P.W 18 ? + + -

Family 2:R.M. 5 - + + - +K.A. 21 - + + - ND

Family 3:D.S. 27 + + + + NDP.S...... 24 + + + - +T.S. 16 + + - - NDE.S. 48 ? + + - ND

NOTE. - + = presence of feature; - = absence of feature; ND = not done.

102

Selection in Incontinentia Pigmenti

with PvuII (10-15 u/gg DNA) and HpaII (10 u// tg DNA)under conditions suggested by BRL, electrophoresedon a 1% agarose gel, and alkaline transferred (0.5 MNaOH) to Gene Screen Plus (NEN/DuPont) followingacid depurination (0.25 M HCI). The filters were pre-hybridized with 10 ml of 1% SDS, 0.5 M NaCi, 10mM Tris, pH 7.4 at 65 C for at least 20 min. Probeswere 32P-labeled by the random primer method andwere added to the hybridization buffer at a final con-centration of 2-4 x 106 cpm/ml. Following 12-48 hhybridization at 65 C, the membranes were washedtwice for 5 min at room temperature with 2 x SSC,0.1% SDS, then twice for 30 min at room temperaturewith 1 x SSC, 0.1% SDS, then twice for 30 min at65 C with 0.15 x SSC, 0.1% SDS. Blotted filters wereautoradiographed at - 70 C in cassettes with intensify-ing screens (DuPont Lightening Plus).

Probes.-The probes used for these studies were theHPRT intron probes, pPB1.7 (Wolf et al. 1984b), aswell as the 0.8-kb MspI-PstI fragment from this insert(Fearon et al. 1986) (see map in fig. 2), which givesless background.

Results

Table 2 presents the G6PD phenotype of four femalesheterozygous for IP. Two of them (S.D. and K.A.) werealso heterozygous for electrophoretic variants ofG6PD

Table 2

G6PD Phenotype of Females with Incontentia Pigmenti

Subject and Tissue G6PD Phenotype

S.D.:RBC ............... B (paternal allele)Skin fibroblasts ...... B >>>> AClones, biopsy 1 ..... B (11) A (4)Clones, biopsy 2 ..... B (13) A (2)

P.W. (mother of S.D.):RBC ............... A+Skin ............... A

R.M.:RBC ............... BWBC .............. BSkin fibroblasts ...... B

K.A. (mother of R.M.):RBC ............... B = B'WBC .............. B>B'Skin fibroblasts ...... B >>> B'Clones ............. B (18) B'(0)

NOTE. -Numbers in parentheses denote number of clones withphenotype.

A, and G6PD B' (most likely G6PD Baltimore-Austin),respectively (see description ofG6PD variants in McKu-sick 1986, pp. 1356-1357). Studies of red cells fromS.D. at 6 mo of age and from her parents (fig. 1, lanes1-4) showed that she expressed only her father's G6PDallele (G6PD B). Her mother, P.W., also affected withIP, had a strong G6PD A band in RBCs as well as inskin fibroblast cultures, and therefore must have hadat least one A+ allele. As it could not be determinedwhether the allele transmitted to S.D. was the A - vari-ant, which is unstable in erythrocytes, we analyzed skinfibroblasts from S.D. The A - variant in skin fibroblastshas almost full activity and should be well representedin that tissue. However, we detected only trace amountsof G6PD A (fig. 1, lane 6), so that cells expressing herfather's allele predominated in that tissue as well. Thefew G6PD A clones derived from that culture confirmedthat there was skewing and showed that the G6PD Aallele in these cells has normal enzyme activity (fig. 1B,lane 7). Of interest is the fact that the pattern of mosai-cism was skewed in the biopsy specimen from normalskin as well as in that taken from areas of hyperpig-mentation. We conclude that S.D. was heterozygous forG6PD A and IP but that the paternal chromosome withG6PD B and non-IP alleles was the predominant activeX chromosome in both RBCs and skin fibroblasts.

P.W. and R.M., the proband from the second family,were most likely homozygous at the G6PD locus, asthey had only a single enzyme phenotype in all tissues;however, K.A., the mother of R.M., was informative.Although her RBCs and hair follicles expressed bothB and B' alleles and although her leukocytes showedonly slight skewing, her skin fibroblasts showed markedskewing, as all of the clones derived from the specimenexpressed only G6PD B. However, we could not deter-mine whether her G6PD B' allele was, in fact, coupledwith the IP mutation.We extracted DNAs from skin fibroblast cultures of

these females to look for RFLPs at the HPRT locus,but none was heterozygous for the 12/24-kb RFLP.However, members of the third IP family were segregat-ing this DNA polymorphism. Figures 3 and 4 presentthe results of studies of this kindred. The parents wereespecially informative, as the father was hemizygousfor the 12-kb BamHI allele at the HPRT locus (fig. 3,lane 1) and the mother, who had IP (table 1), washomozygous for the 24-kb BamHI allele (fig. 3, lanes8 and 9). Therefore, all three daughters, each affectedwith IP, were heterozygous for this RFLP, with a 24-kbmaternal allele and a 12-kb paternal allele.

Analysis of leukocyte DNA from these heterozygotes

103

Migeon et al.

I24

B B B/H

Figure 3 Pedigree of family 3 and autoradiographs showingclonal analysis of leukocytes, usingDNA polymorphism and differen-tial methylation of the HPRT locus to determine the pattern ofX-chromosome inactivation. DNA samples were digested with BamHI(B) or with BamHI plus HpaII (B/H), as described in Material andMethods, and blots were probed with the 0.8-kb MspI-PstI fragment.

revealed a nonrandom pattern of mosiacism, as in allthree females the 12-kb band from the paternal Xpredominanted. However, the degree of skewing was

variable (fig. 3). In D.S., the oldest and most severelyaffected daughter (lane 3), the loss of the 24-kb frag-ment (of maternal origin) was much more dramatic thanin her two sisters (lanes 5 and 7) and was perhaps age

related. We conclude that the paternal, non-IP-bearingX chromosome marked with the 12-kb HPRT alleleis the active X chromosome in the leukocytes of eachof these females.

Studies of skin fibroblast DNAs from the three daugh-ters revealed that the distribution of the two cell popu-

18 W :12

Figure 4 Autoradiogram showing clonal analysis of skin fibro-

blasts from family 3, using DNA polymorphism, and differential meth-

ylation of the HPRT locus. DNA from D.S. (left), P.S. (center), and

T.S. (right).The first sample for each female is a BamHPvuII di-

gest, and the second is a Bam/PvuII/Hpa1I triple digest. Digestion

withPvuIg decreases the size of the 24skb fragment to 18 kb (see

map in fig. 2), making the alleles more comparable in size.

1ations in these cell populations was nonrandom (fig.4). In all three, the 12-kb paternal allele remained rela-tively intact, whereas the 24-kb maternal allele virtu-ally disappeared when digested with the methyl-sensitiverestriction enzyme HpaII. This indicates that the pater-nal non-IP X chromosome is the active chromosomein most of their skin fibroblasts.

Discussion

Using G6PD variants and HPRT methylation asmarkers, our analysis ofmosaicism at the IP locus showsthat the population of cells whose active X carries themutant IP allele is underrepresented in heterozygotes.Most likely, the loss of these cells results from cell se-lection following random X inactivation.

There is considerable evidence that X-chromosomeinactivation is random at onset but that subsequentlythe proportion of the two mosaic populations may beskewed as a consequence of selection for alleles at oneor more loci that affect cell proliferation (reviewed inMigeon 1978). The elimination of abnormal cells fromfemales carrying X-linked mutations is well documented:abnormal platelets have been eliminated from hetero-zygotes with the Wiskott-Aldrich mutation (Gealy etal. 1980; Prchal et al. 1980), abnormal B cells are lostfrom carriers of agammaglobulinemia of the Brutontype (Conley et al. 1987; Fearon et al. 1987), and HPRT-deficient hematopoietic cells have been eliminated incarriers of the Lesch-Nyhan syndrome (Dancis et al.1968; Nyhan et al. 1970; Albertini and DeMars 1974).In fact, such elimination of specific cell populations hasprovided important clues about the nature of the mu-tant gene product with respect to the dominance orrecessiveness of the phenotype, the tissue in which themutation is expressed, and the developmental stage atwhich the relevant locus is expressed.On the other hand, selection does not always favor

the wild-type allele; in heterozygotes for the adreno-leukodystrophy mutation, the mutant allele conveys aproliferative advantage to hematopoietic cells, culturedfibroblasts, and perhaps to cells of the spinal cord, asheterozygotes are frequently disabled by this mutation(Migeon et al. 1981; O'Neill et al. 1984). Because mu-tations such as these have a dominant effect on cellproliferation, the result of selection is very noticeable.It is likely that more subtle proliferative advantagesmight contribute to the expressivity and account forsome of the range of phenotypes found among carriersof male-lethal X-linked mutations.Our studies of five heterozygotes from three kindreds

104

Selection in Incontinentia Pigmenti 105

segregating IP suggests that mutations at the IP locusmay be detrimental to the proliferation of at least somecells that express the mutation, eventually resulting inloss of these cells. We observed that cells of mutant typehave been eliminated not only from specimens of der-mal origin where the clinical abnormalities are mani-fest, but also, in some cases, and to varying degrees,from blood cells. The effect of selection is often moststriking in hematopoietic tissue, perhaps because of therelatively rapid turnover of blood cells. However, bloodleukocytes expressing the IP mutation are not invari-ably eliminated in heterozygotes (A. Harris, personalcommunication). Further evidence suggesting that thedisorder is heterogeneous is that RBCs expressing themutation were eliminated from S.D. but not from K.A.In fact, Harris et al. (1988) have recently suggested thatthere may be more than one X-linked locus responsiblefor the IP phenotype. Alternatively, in some femalesthe deleterious effect of the IP mutation may be subor-dinate to more striking selective effects mediated by al-leles at other X-linked loci. In any event, as selectionof this kind seems to act most effectively on the severelydeficient cell (Emmerson et al. 1972), we suggest thatthe protein product of the IP locus is cell autonomousin hematopoietic as well as in dermal tissues; that is,it is not readily communicated between cells.That females with de novo X-autosome transloca-

tions survive intrauterine life needs some explanation.As such females are not mosaic (in most cells the trans-location chromosome is the active X)-and thereforemost of their cells should express the mutation-onemight expect that these females would be as severelyafflicted as affected males. In fact, the females reportedby Gilgenkrantz et al. (1985) and Hodgson et al. (1985)were more severely retarded than most heterozygotes.Hodgson has suggested that in these females the pat-tern ofX inactivation may be more random in relevanttissues. Alternatively, selection against cells with an ac-tive normal X might occur late relative to the lethaleffects of the mutation on development, so that thesefemales may be protected by the normal allele, at leastfor a while, in utero. Or, conceivably, the locus disruptedby de novo translocation is different from that respon-sible for the familial cases.

Clonal analysis in females carrying an X-linked mu-tation which is lethal in males has revealed that selec-tion against cells expressing the IP mutation plays somerole in ameliorating the effects of the mutation in het-erozygotes. Our findings contribute further evidencethat the locus is X linked in the families we have stud-ied. The use of DNA polymorphisms in conjunction

with DNA methylation to detect skewed populationsin females obviates the need for cell cloning or proteinvariants. However, the method is limited to DNA se-quences that show differential methylation of loci onactive and inactive X chromosomes. At this time, themost useful loci are those that code for the housekeep-ing genes, PGK and HPRT, as differential methylationis most extensive at these loci and as there are commonpolymorphic RFLPs (Vogelstein et al. 1987). In anyevent, analysis of the mosaic phenotype in heterozy-gotes should continue to contribute new insights intothe nature of X-linked diseases.

AcknowledgmentsThis work was supported by NIH grant HD 06545. We

are grateful to Richard Shilzony and Sanjiv Patel for contri-butions to the DNA methylation studies, and to Eric R. Fea-ron for providing the 800-bp HPRT probe.

ReferencesAlbertini, R. J., and R. DeMars. 1974. Mosaicism of periph-

eral blood lymphocyte populations in females heterozy-gous for the Lesch-Nyhan mutation. Biochem. Genet. 11:397-411.

Carney, R. G., Jr. 1976. Incontinentia pigment: a world statisti-cal analysis. Arch. Dermatol. 112:535-542.

Conley, M. E., P. Brown, A. R. Pickard, R. H. Buckley, D. S.Miller, W. H. Raskind, J. W. Singer, and P. J. Fialkow. 1987.Expression of the gene defect in X-linked agammaglobu-linemia. N. Engl. J. Med. 315:564-567.

Dancis, J., P. H. Berman, V. Jansen, and M. E. Balis. 1968.Absence of mosaicism in the lymphocyte in X-linked con-genital hyperuricosuria. Life Sci. 7:587-591.

Davidson, R. G., H. M. Nitowski, and B. Childs. 1963.Demonstration of two populations of cells in the humanfemale heterozygous for glucose-6-phosphate dehydroge-nase variants. Proc. Natl. Acad. Sci. USA. 50:481-485.

Emmerson, B. T., C. J. Thompson, and D. C. Wallace. 1972.Partial deficiency hypoxanthine-guanine phosphoribosyl-transferase: intermediate deficiency in heterozygote red cells.Ann. Intern. Med. 76:285-288.

Fearon, E. R., P. J. Burke, C. A. Schiffer, B. A. Zehnbauer,and B. Vogelstein. 1986. Differentiation of leukemia cellsto polymorphonuclear leukocytes in patients with acutenonlymphocytic leukemia. New Engl. J. Med. 315:15-24.

Fearon, E. R., J. A. Winkelstein, C. I. Civin, D. M. Pardoll,and B. Vogelstein. 1987. Detection in X-linked agam-maglobulinemia by analysis ofX-chromosome inactivation.N. Engl. J. Med. 316:427-431.

Gealy, W. J., J. M. Dayer, and J. B. Harley. 1980. Allelic ex-clusion of glucose-6-phosphate dehydrogenase in platelets

106 Migeon et al.

and lymphocytes-T from a Wiskott-Aldrich syndrome car-rier. Lancet 1:63-65.

Gilgenkrantz, S., P. Tridon, N. Pinel-Briquel, J. Beurey, andM. Weber. 1985. Translocation (X;9) (pll;q34) in a girlwith incontinentia pigmenti (IP): implications for the re-gional assignment of the IP locus to Xpll? Ann. Genet.28:90-92.

Harris, A., S. Lankester, E. Haan, J. Beres, M. Hulten, J.Szollar, I. Soutter, and M. Bobrow. 1988. The gene for in-continentia pigmenti: failure of linkage studies using DNAprobes to confirm cytogenetic localization. Clin. Genet.34:1-6.

Hodgson, S. V., B. Neville, R. W. A. Jones, C. Fear, and M.Bobrow. 1985. Two cases of X/autosome translocation infemales with incontinentia pigmenti. Hum. Genet. 71:231-234.

Kelly, T. E., J. M. Rary, and L. Young. 1976. Incontinentiapigmenti: a chromosomal breakage syndrome. J. Hered.67:171-172.

Lenz, W. 1975. Half chromatid mutations may explain in-continentia pigmenti in males. Am. J. Hum. Genet. 27:690-691.

McKusick, V. A. 1986. Mendelian inheritance in man. JohnsHopkins University Press, Baltimore.

Migeon, B. R. 1978. Selection and cell communication asdeterminants of female phenotype. Pp. 417-432 in L. B.Russell, ed. Genetic mosaics and chimeras in mammals.Plenum, New York.

Migeon, B. R., H. W. Moser, A. B. Moser, J. A. Sprenkle,D. Sillence, and R. A. Norum. 1981. Adrenoleukodystrophy:evidence for X-linkage, inactivation, and selection favor-ing the mutant allele in heterozygous cells. Proc. Natl. Acad.Sci. USA 78:5066-5070.

Migeon, B. R., S. F. Wolf, J. Axelman, D. C. Kaslow, andM. Schmidt. 1985. Incomplete dosage compensation inchorionic villi of human placenta. Proc. Natl. Acad. Sci.USA 82:3390-3394.

Nussbaum, R. L., W. Crowder, W. L. Nyhan, and C. T.Caskey. 1983. A three allele restriction-fragment-lengthpolymorphism at the hypoxanthine phosphoribosyltrans-ferase locus in man. Proc. Natl. Acad. Sci. USA 80:740-753.

Nyhan, W. L., B. Bakay, J. D. Connor, J. F. Marks, and D. K.

Keele. 1970. Hemizygous expression of glucose-6-phos-phate dehydrogenase in erythrocytes of heterozygotes forthe Lesch-Nyhan syndrome. Proc. Nati. Acad. Sci. USA65:214-218.

O'Neill, B. P., H. Moser, K. M. Saxena, and L. C. Marmion.1984. Adrenoleukodystrophy: clinical and biochemicalmanifestations in carriers. Neurology 34:789-801.

Prchal, J. T., A. J. CarrollJ. F. Prchal, W. Rist, H. W. Skalka,W. J. Gealy, J. B. Harley, and A. Malluh. 1980. Wiskott-Aldrich syndrome: cellular impairments and their impli-cation for carrier detection. Blood 56:1048-1054.

Vogelstein, B., E. Fearon, S. R. Hamilton, and A. P. Feinberg.1985. Use of restriction fragment length polymorphismsto determine the clonal origin of human tumors. Science227:642-645.

Vogelstein, B., E. R. Fearon, S. R. Hamilton, A. C. Preisinger,H. F. Willard, A. M. Michelson, A. D. Riggs, and S. H.Orkin. 1987. Clonal analysis using recombinant DNAprobes from the X chromosome. Cancer Res. 47:4806-4813.

Wettke-Schafer, R., and G. Kantner. 1983. X-linked domi-nant inherited diseases with lethality in hemizygous males.Hum. Genet. 64:1-23.

Wieacker, P., J. Zimmer, and H.-H. Ropers. 1985. X inacti-vation patterns in two syndromes with probable X-linkeddominant, male lethal inheritance. Clin. Genet. 28:238-242.

Wolf, S. F., S. Dintzis, D. Toniolo, G. Persico, K. Lunnen,J. Axelman, and B. R. Migeon. 1984a. Complete concor-dance between glucose-6-phosphate dehydrogenase activ-ity and hypomethylation of 3' CpG clusters: implicationsfor X chromosome dosage compensation. Nucleic AcidsRes. 12:9333-9348.

Wolf, S. F., D. J. Jolly, K. D. Lunnen, T. Friedmann, and B. R.Migeon, 1984b. Methylation of the HPRT locus on thehuman X: implications for X inactivation. Proc. Natd. Acad.Sci. USA 81:2806-2810.

Yen, P. H., P. Patel, A. C. Chinault, T. Mohandas, and L. J.Shapiro. 1984. Differential methylation of hypoxanthinephosphoribosyltransferase genes on active and inactive hu-man X-chromosomes. Proc. Natd. Acad. Sci. USA 81:1759-1763.