Proc. Natl. Acad. Sci. USAVol. 92, pp. 10796-10800, November 1995Medical Sciences

Gene conversions and unequal crossovers between CYP21(steroid 21-hydroxylase gene) and CYP21P involvedifferent mechanisms

(recombination/polymerase chain reaction/spermatozoa)

MARiA-TERESA TUSIE-LUNA* AND PERRIN C. WHITEt:*Unidad de Genetica de la Nutrici6n-Instituto de Investigaciones Biomedicas, Universidad Nacional Aut6noma de M6xico-Instituto Nacional de Pediatria,Mexico D.F. 04510; and tDepartment of Pediatrics, University of Texas Southwestern Medical Center, Dallas, TX 75235-9063

Communicated by Jean D. Wilson, University of Texas Southwestern Medical Center, Dallas, TX, August 22, 1995 (received for reviewMay 19, 1995)

ABSTRACT Most cases of congenital adrenal hyperpla-sia, the inherited inability to synthesize cortisol, are caused bymutations in the steroid 21-hydroxylase gene (CYP21). Steroid21-hydroxylase deficiency is unusual among genetic diseasesin that -95% of the mutant alleles have apparently beengenerated by recombination between a normally active gene(CYP21) and a linked pseudogene (CYP21P). Approximately20% of mutant alleles carry DNA deletions of 30 kb that havepresumably been generated by unequal meiotic crossing-over,whereas 75% carry one or more mutations in CYP21 that arenormally found in the CYP21P pseudogene. These latter muta-tions are termed "gene conversions," although the mechanism bywhich they are generated is not well understood. To assess thefrequency at which these different recombination events occur,we have used PCR to detect de novo deletions and gene conver-sions in matched sperm and peripheral blood leukocyte DNAsamples from normal individuals. Deletions with breakpoints ina 100-bp region in intron 2 and exon 3 were detected in speriDNA samples with frequencies of -1 in 105-106 genomes outwere never detected in the matching leukocyte DNA. Geneconversions in the same region occur in -1 in 103-105 genomesin both sperm and leukocyte DNA. These data suggest thatwhereas deletions occur exclusively in meiosis, gene conversionsoccur during both meiosis and mitosis, or perhaps only duringmitosis. Thus, gene conversions must occur by a mechanismdistinct from unequal crossing-over.

Congenital adrenal hyperplasia is an inherited disorder ofcortisol biosynthesis. Although five different enzymes arerequired to synthesize cortisol in the adrenal cortex, steroid21-hydroxylase deficiency accounts for >90% of congenitaladrenal hyperplasia (1). This defect is one of the most commoninborn errors of metabolism in humans; the severe "classic"form of 21-hydroxylase deficiency occurs in 1 in 10,000-15,000individuals, whereas a milder "nonclassic" form is found in upto 2-3% of several ethnic groups such as Ashkenazi (EasternEuropean) Jews (2).There are normally two 21-hydroxylase genes in humans:

CYP21 is the functional gene whereas CYP21P is a pseudogenethat carries small deletions, insertions, and point mutationsthat prevent synthesis of a functional enzyme. These genes are-98% identical in nucleotide sequence (3,4). Both are locatedwithin the HL,A major histocompatibility complex locus onchromosome 6p21.3, adjacent to and alternating with the C4Aand C4B genes encoding the fourth component of the serumcomplement (5).

Steroid 21-hydroxylase deficiency is unusual among geneticdiseases in that -95% of mutant alleles have apparently been

generated through intergenic recombination. Twenty percentof alleles have a 30-kb deletion including the 3' end of CYP21P,all of C4B, and the 5' end of CYP21 (6). These chromosomesthus carry a single nonfunctional chimeric gene with 5' and 3'ends corresponding to CYP21P and CYP21, respectively. Pre-sumably the normal tandem duplication of C4 and CYP21 isprone to misalignment during meiosis, leading to unequalcrossing-over.

Approximately 75% of alleles have apparently been gener-ated by another type of intergenic recombination. Thesechromosomes do not carry deletions, but instead CYP21 carriesone or more deleterious mutations normally found in CYP21P(7). These mutations are referred to as "gene conversions,"although the mechanism by which they are generated is notwell understood (8).The very high proportion of intergenic recombinations seen

among 21-hydroxylase deficiency alleles suggests that thesegenes contain one or more "hotspots" for recombination.However, interpretation of allele frequencies is complicated bythe presence of founder effects. For example, about half ofdeletions are associated with the rare extendedHLA haplotypeA3;Bw47;DR7 (6), whereas the majority of nonclassic allelescarry an apparent gene conversion, V281L, on the HLA haplo-type B14;DRI (9). Because gene conversions differ in the degreeof enzymatic impairment they confer (10), it is possible thatascertainment biases also affect observed allele frequencies.To overcome these problems, we used serial dilutions of

DNA and PCR amplification to determine the frequencies ofde novo gene conversions and deletions of CYP21 in sperm andleukocyte DNA from normal individuals. We found thatdeletions occur only during meiosis, whereas gene conversionsoccur at a considerably higher frequency and mainly duringmitosis.

MATERIALS AND METHODSDNA Preparation. Five healthy male volunteers between 30

and 43 years old without family history of congenital adrenalhyperplasia donated semen and blood samples. Semen sampleswere left at room temperature to liquify and sperm cells werepelleted. Cells were washed twice with phosphate-bufferedsaline, and DNA extraction in 10 mM Tris HCl, pH 8.0/5 mMEDTA/0.5% SDS/0.02% proteinase K/20 mM dithiothreitolwas followed by phenol/chloroform extraction and ethanolprecipitation. Blood samples were frozen and then werethawed in hypotonic medium, and leukocyte DNA was ex-tracted in the same manner as for sperm DNA except thatdithiothreitol was omitted. Before PCR amplifications, DNAsamples were digested with Taq I restriction endonuclease andprecipitated with sodium acetate and ethanol. Aliquots of

tTo whom reprint requests should be addressed.

10796

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 92 (1995) 10797

Table 1. Primers and probes used for PCR amplification and detection of de novo gene conversionsand deletions

No. Use Specificity Sense Sequence (5' to 3')1 Primer CYP21, 1316 + ATCTGGTGGGAGAA\AGC2 Primer CYP21, 1473 - AGAGCAGOGAjIAGTCI3 Primer CYP21P, 1418 - TGTCTGCAGGAGGAGC4 Probe CYP21P, 1392 + TCCAGCCCCCAQCTCCTC5 Primer CYP21P, 1309 + GGGGCATATC1TCAGGAGAA£6 Primer Both, 1495 - GGTGAGCTTCTTGTGGGCTTT7 Primer CYP21, 1477 - CAGGGAGIAGETCT8 Probe CYP21, 1392 + TCCAGCCCCCAACTCCTC

The position of the 5' nucleotide of each oligonucleotide is listed in the Specificity column; numberingis for CYP21 (3). Gene-specific positions in each sequence are underlined; the backslash in the sequenceof oligonucleotide 1 denotes the position of a 2-nt insertion in CYP21P.

these digests were analyzed by genomic blot hybridization todocument DNA quality and to ensure that no subject carrieda heterozygous deletion of CYP21 or a homozygous deletion ofCYP21P.

Detection of Gene Conversion Events. A two-stage hemi-nested PCR amplification strategy was used, similar to themethod described for detection of hybrid T-cell receptor genes(11). Serial dilutions of each DNA sample were made rangingfrom 500 ng to 1 pg. All amplifications employed either manualor Ampliwax (Perkin-Elmer) "hot start" protocols and had aninitial denaturation step of 94°C for 3 min and a final extensionstep of 72°C for 10 min. First amplifications used primers 1 and2 (Table 1) and were run for 35 cycles of 94°C for 30 sec, 54°Cfor 30 sec, and 72°C for 1 min. Reaction products were loadedon 2% agarose gels and were visualized by staining withethidium bromide. A 104 dilution from each first amplificationwas used for the second PCR round. Second amplificationsused primers 1 and 3, primer 1 being end-labeled with[y-32P]ATP, and were run for 20 cycles of 94°C for 30 sec, 57°Cfor 30 sec, and 72°C for 60 sec. Reaction products were loaded on8% polyacrylamide sequencing gels and were visualized by au-toradiography.

Additionally, products of the first-stage PCR were directlysubcloned into the pCR1000 vector (Invitrogen). Approxi-mately 104 independent colonies from each subject werescreened by hybridization in duplicate with a radiolabeledoligonucleotide probe (no. 4, Table 1) recognizing the pointmutation A -> G 13 bp before the end of intron 2. Thesequence of each positive clone was determined with T7polymerase (Sequenase; United States Biochemical).

Detection ofDe Novo Deletions. A two-stage PCR protocolsimilar to that described above was used. Serial dilutionsranged from 1 ,tg to 1 pg. First amplifications used primers 5and 6 (Table 1) and consisted of 35 cycles of 94°C for 1 min,60°C for 1.5 min, and 72°C for 2 min. Second amplificationsused primers 5 and 7 and consisted of 35 cycles of 94°C for 1min, 66°C for 1.5 min, and 72°C for 2 min. Products of thesecond amplification were run on agarose gels, transferred tonylon membranes, and hybridized with a CYP21-specific in-ternal oligonucleotide probe (no. 8, Table 1). Products ofpositive reactions were cloned into pCR1000 and individualclones were sequenced to confirm the location of the break-points.

RESULTSDetection ofDe Novo Gene Conversion Events in Blood and

Sperm Samples. The most common mutation causing classic21-hydroxylase deficiency, occurring in -25% of all affectedalleles (7, 12), is a A -- G mutation in intron 2 ("i2g") at nt656, 13 bp before the end of the intron. This mutation affectssplicing of pre-mRNA (13). We used a two-stage hemi-nestedPCR strategy to detect this mutation occurring de novo insperm or leukocyte DNA from five normal individuals. In the

first PCR, a 156-bp segment of CYP21 was amplified by usinggene-specific primers in intron 2 and exon 3. Moleculescarrying i2g were then detected by allele-specific PCR usingthe same sense primer and an antisense primer complemen-tary to i2g at its 3' end (Fig. 1).To determine the sensitivity and specificity of this proce-

dure, serial dilutions from 500 ng to 1 pg were made from aDNA sample from an individual known to be a heterozygouscarrier of the i2g mutation. A positive signal was obtained at1 pg (Fig. 2). As this quantity of DNA is considered tocorrespond to one haploid genome, it suggests that the tech-nique has nearly single-molecule sensitivity. A water controlshowed no amplification. As a more robust negative control,serial dilutions were made of a DNA sample from an individualwho was homozygous for the HLA haplotypeAl;B8;DR3. Thishaplotype, which occurs on -5% of normal chromosomes,invariably carries a complete deletion of CYP21P (5, 14) andso this sample should not contain de novo recombinationsbetween CYP21 and CYP21P. As expected, no such recombi-nations were observed. To rule out the possibility of in vitrorecombinations being generated during the PCR, 500 ng of thisnegative control sample was spiked with 30 pg of a cosmidclone carrying CYP21P (-2 x 106 copies, representing an -4:1ratio of CYP21P to CYP21 in each reaction), and serialdilutions were made. Again, no amplification was detected inthe second, allele-specific PCR.

In contrast, de novo i2g mutations were observed in bothsperm and leukocyte DNA from all five normal individuals(Fig. 3). Because these mutations were not observed in theindividual who lacked CYP21P, they were presumed to be theresult of gene conversion. The frequencies at which theserecombinations occurred varied over 2 orders of magnitudeamong individuals; the lowest amount of DNA at which thesemutations could be detected ranged from 1 to 100 ng. TheseDNA quantities correspond to mutation frequencies of 1 in

P1

P2

P1

2 5' 3'

P3

CYP21i2g

CYP21

i2g

CYP21

FIG. 1. Amplification strategy for de novo gene conversion events.Two stages of PCR amplification are carried out. In the first ampli-fication, two CYP21-specific primers are used; P1 is located in intron2 and P2 is located in exon 3. The products of the first amplificationare CYP21 fragments, some of which contain internal mutationsresulting from de novo gene conversion events. A 104 dilution of thefirst amplification is used as a template for the second (allele-specific)reaction, in which the radiolabeled sense primer is P1 and the antisenseprimer, P3, is complementary at its 3' end to the i2g mutation.

Medical Sciences: Tusi6-Luna and White

10798 Medical Sciences: Tusie-Luna and White

Positive NegativeSpikednegative

5 1&1&1I1 1&1O 1 55434 3lel 1 5x1010 13115ylOlOlOl 4101 ~~~~~~~~~5x1O1001010 1 5101 0ei

1st

2nd

FIG. 2. Detection of gene conversion events in control patients byPCR. Serial dilutions of DNA were used to amplify gene conversionevents in blood DNA derived from a patient known to be a heterozy-gous carrier of the i2g mutation in CYP21 (positive control), anindividual carrying a homozygous deletion of CYP21P (negativecontrol), and the DNA of the latter individual mixed with 30 pg ofcosmid DNA containing CYP21P (spiked negative control). The firstrow shows ethidium bromide staining of the products of the firstamplification (CYP21 fragments) according to the procedure de-scribed in Fig. 1. The amount of DNA in each reaction is noted inpicograms. The shorter products observed in some reactions arepresumed to be nonspecific; they were never amplified in the secondround of PCR. The second row shows autoradiograms of the productsof the second amplification (CYP21 fragments carrying i2g). Thepositive control assay shows amplification at the 1-pg level, approxi-mately equivalent to the amount ofDNA contained in one haploid cell.

103-105. Despite these individual variations, the matchingleukocyte and sperm samples from each individual had similarlevels of de novo i2g gene conversions.

This frequency of de novo gene conversions was confirmedby an alternative approach in which CYP21 gene segments

1 2

were cloned directly after amplification by a single PCR.Clones carrying i2g were identified by colony hybridizationwith an allele-specific oligonucleotide. Screening of _104colonies from each PCR resulted in the identification of fourindependent gene conversion events in three samples. Se-quencing of these clones and examination of gene-specificnucleotide positions revealed that the gene conversions haddistinct breakpoints, with conversions ranging in size from1-24 bp (the minimum and maximum sizes of the smallestconversion, based on involvement of polymorphic nucleotides)to 54-61 bp.

Detection ofDe Novo Deletions in CYP21. A priori, apparentgene conversions could be the result of double crossoversbetween CYP21 and CYP21P. Therefore, we compared thefrequencies of gene conversions and single unequal crossovers(i.e., deletions) with breakpoints in the same genetic region inintron 2 and exon 3. De novo deletions were identified with atwo-stage PCR procedure (Fig. 4). In the first PCR, a CYP21P-specific primer in intron 2 and a nonspecific primer in exon 3were used to amplify CYP21P as well as any CYP21P/CYP21chimeric molecules that were present. This first step increasedthe sensitivity of the procedure and ensured that all DNAsamples were amplified equally well. The second PCR includedthe same CYP21P-specific primer in intron 2 and a nestedCYP21-specific primer in exon 3, thus amplifying only chimericmolecules. Serial dilutions of leukocyte and sperm DNAsamples were tested.

Chimeric CYP21P/CYP21 genes with deletion breakpointsin the intron 2-exon 3 region were reproducibly detected insperm DNA samples from three of the five individuals studied(Fig. 5). Two individuals had positive signals when 100 ng ofDNA was included in the reaction, whereas one subject had apositive signal only with 1000 ng of DNA. These results areconsistent with frequencies of de novo deletions of 1 in105_106 in these individuals; the remaining two subjects ap-parently have deletion frequencies of <1 in 106. These differ-

3 4 554324 3 2 5 4 3 2 5 4 3 2 5 4 3 5 4 1l 1

1stsperm

..4. .... ..

::....: .:

..: :.

2ndsperm

St

blood

* .... ........... .:.

2nd *.... ..

blood

kf4 ff. :.:::~,

..... .. .... ..::: .: :: .: :..::. :: ::: :.: .:*. : ... .

,.. ....:*:...: ;.................................. .................... ..... ..... .... . :

....

FIG. 3. Detection of de novo gene conversion events in blood and sperm samples from five healthy donors. Serial dilutions ranging from 100ng to 1 pg (noted above each lane in picograms) were used. The first and third rows show ethidium bromide staining of the products of the firstamplification of sperm and blood samples, respectively. The second and fourth rows display autoradiograms of the products of the secondamplification of these samples.

Proc. Natl. Acad. Sci. USA 92 (1995)

....... .....

Proc. Natl. Acad. Sci. USA 92 (1995) 10799

P5

1 -_________.__

P6

P5

2P7

CYP21 P/CYP21

CYP21 P/CYP21

FIG. 4. Amplification strategy for de novo crossover events. Atwo-stage PCR was carried out with a CYP21P-specific sense primer(P5) located in intron 2 and an antisense primer located in exon 3 thatcould recognize either CYP21 or CYP21P (P6). The products of thefirst amplification are a segment of CYP21P and chimeric CYP21P/CYP21 segments. A 104 dilution of the first PCR is used as a templatefor the second PCR, in which only chimeric genes are amplified byusing the same CYP21P-specific sense primer (P5) and a CYP21-specific nested primer (P7) located in exon three.

ences in recombination frequency are unlikely to be due tointer- or intraassay variability, since similar results were ob-tained with a second set of semen samples obtained from thesame individuals (data not shown).De novo deletions were not detected in leukocyte DNA from

any individual, even though CYP21P sequences could beamplified from 1 pg (nominally one haploid genome) of theseDNA samples (data not shown). Thus, unequal crossing-overevents were detected only in DNA from cells that had under-gone meiosis.PCR products from positive sperm samples were cloned and

sequenced. Twenty-four independent clones were analyzed;the deletion breakpoints were rather evenly distributed overthe amplified region (Fig. 6).

DISCUSSIONGenetic recombination in higher eukaryotes is essential formaximizing genetic diversity among individuals and also for suchimportant biological processes as development of immunoglob-ulin and T-cell receptor repertoires. Although recombination in

1

1stsperm

2

BASE# 29 34 41 60 64 69 70 95 142

CYP21 G G C A G C A A C

CYP21 P T A A G T G G G G

12% T A C A G C A A C

12% T A A A G C A A C

25% T A A G G G A C

50% T A A G T G G G C

FIG. 6. Sequence analysis of de novo crossover events identified insperm DNA samples. Twenty-four independent clones were se-quenced. Nucleotide changes between CYP21 and CYP21P in theregion amplified are shown. The sites and frequencies of differentbreakpoints are indicated. Nucleotides are numbers from the 5' end ofthe amplified fragment; the end of intron 2 is position 108.

immunoglobulin genes is relatively well understood, many ques-tions remain about other forms of recombination, includingwhether there are specific sequences that promote or inhibitrecombination (15), specific chromosomal regions in which var-ious types of recombination are more or less likely to occur (16),and specific protein factors that are required for recombination.These questions are difficult to address by studying recombina-tion in mammalian progeny, because of the small numbers ofoffspring available and the time and expense required to ascertainrecombinant offspring. The direct study of meiotic recombinationin spermatozoa is attractive because millions of meioses can beexamined in a single semen sample. Previous studies have de-tected meiotic recombinations by assaying (3-galactosidase activ-ity in sperm from transgenic mice (17) and by using PCR to typeflow-sorted single spermatozoa for alleles at several linked loci(18). With methods analogous to those employed in the presentstudy, mitotic recombinations have been detected in T-cell re-

3 4 5

10 10 10 3101eo 10 10 10 131010 1 10 010 1010 10 I 10110 01310 10 1 tO 10 10 30101eo

R* .....r..-

..... 1. ,.t

2nd

sperm

2ndblood

FIG. 5. Detection of de novo crossover events in sperm samples from healthy donors. Serial dilutions of DNA from matching blood and spermsamples from five healthy donors were amplified according to the procedure described in Fig. 4. The amount of DNA in each dilution is notedin picograms. The first row shows ethidium bromide staining of the PCR products from the first amplification of the five sperm samples (the firstamplification of the blood samples, which gave very similar results, is not shown; note that photographic negatives of the ethidium bromidefluorescence were used in this figure). The second and third rows show the second amplifications of the five sperm samples and the correspondingblood samples, respectively. These products were hybridized with a radiolabeled CYP21-specific oligonucleotide (P8, Table 1) that should detectonly CYP21P/CYP21 chimeric products in this experiment. Such products were observed only in sperm samples.

("VDf3i D

Medical Sciences: Tusie'-Luna and White

10800 Medical Sciences: Tusi6-Luna and White

ceptor genes (11), and meiotic recombinations have been de-tected in murine H-2 class II genes (19) and human minisatellites(20).CYP21 and CYP21P represent an interesting experimental

system for studying recombination in humans. Due to the longtandem duplication in which these genes are located, disease-causing mutations due to intergenic recombinations are rela-tively frequent and different types of recombinations (i.e.,deletions and gene conversions) occur in the same geneticregion. We confirmed that the deletions observed in thesegenes were generated by unequal meiotic crossing-over. Geneconversions, however, take place at equal frequencies insomatic cells and gametes, suggesting that gene conversionsoccur mainly in mitosis and that meiotic recombination (i.e.,double crossing-over) contributes little, if at all, to this process.Unlike studies of gene conversion in meiotic yeast, it was notpossible in the present study to simultaneously examine thedonor and recipient of the gene conversion, and so it is notcertain whether this type of intergenic recombination is, likegene conversion in yeast, nonreciprocal. The observed fre-quency of gene conversions is consistent with the reported rateof de novo intron 2 gene conversions ('1% of alleles) inpatients with 21-hydroxylase deficiency (7, 21). The frequencyof 21-hydroxylase deficiency alleles in the general populationis -2%; the allele frequency of de novo gene conversions inintron 2 in the general population can be estimated by multi-plying these two figures and is thus -2 x 10-i (i.e., 1 in 5 x 103).The methodology used in the present study should be able

to answer several additional questions concerning recombina-tion. The first is whether deletions or gene conversions withinCYP21 and CYP21P, or more generally within the 30-kbtandem duplication containing these genes, take place withincertain discrete regions, or hotspots. It has been suggested thatsequences resembling bacteriophage A chi sites, which arepresent at relatively high frequencies within CYP21/CYP21P,might promote recombination (22, 23), but this hypothesis hasnot been directly tested. Second, the HLA complex as a wholemay be subject to a relatively high rate of gene conversions, forthe teleological reason that intergenic recombination is animportant mechanism for increasing diversity of transplanta-tion antigens and thus maximizing the versatility of the im-mune response. To test this hypothesis, recombination fre-quencies between CYP21 and CYP21P could be compared withthose of other tandemly duplicated genes located outside ofthe HLA complex. For example, as compared with CYP21 andCYP21P, the steroid 1 1,B-hydroxylase and aldosterone synthasegenes (CYPJJBJ and CYPllB2) on chromosome 8q22 are asimilar distance apart (40 kb vs. 30 kb) (24, 25) and have onlyslightly lower sequence identity (95% vs. 98%) (26). These genesare known to undergo unequal crossing-over, and the resultingduplication causes an autosomal dominant form of hypertension,glucocorticoid-suppressible hyperaldosteronism (24, 25). Be-cause both CYPJJBJ and CYP1IB2 encode active enzymes, geneconversions are difficult to ascertain in individuals but should bereadily detectable by the approach used in the present study.

Finally, the existence of reproducible individual differencesin recombination frequency is intriguing and needs to beconfirmed in a larger number of subjects. In particular, oneindividual in our study had a high frequency of de novo geneconversion and a low frequency of deletions, suggesting thatdifferent factors may influence these two types of recombina-tion. Individual differences could be due to nongenetic factors,polymorphisms linked to CYP21 and CYP21P, or geneticfactors remote from CYP21. Of nongenetic factors, the mostobvious is donor age, but our donors' ages varied within onlya 13-year range, and frequencies of neither type of recombi-nation were well correlated with age. Exposure of agriculturalworkers to environmental toxins increases somatic recombi-

nation in T-cell receptor genes (27), but none of our subjectshad any known exposure of this type. As regards polymor-phisms linked to CYP21, it has been suggested that deletionsof CYP21P, which occur on -5% of normal chromosomes, maypredispose the trans chromosome to unequal crossing-overand deletion (28). Consistent with this hypothesis, one indi-vidual in this study (subject 1) with a high rate of de novounequal crossing-over indeed carried a deletion of CYP21P.Other flanking polymorphisms may also be involved, as evensingle nucleotide changes are known to influence gene con-version rates at human minisatellites (15). If, instead, furtherstudies suggest that factors remote from CYP21 influenceeither type of recombination, a panel of brother pairs in whichdeletion and gene conversion frequencies are known could beused to test candidate loci for their effects on these processes.

We thank Leigh Pascoe, Francis Barany, and Alejandro Zentella forhelpful discussions. This work is supported by Grant DK37867 fromthe National Institutes of Health (P.C.W.), Grant 2087-M9302 fromthe Consejo Nacional de Ciencia y Tecnologia (M.-T.T.-L.) and by aBiotechnology Career Development Fellowship from the RockefellerFoundation (M.-T.T.-L.).

1. White, P. C. (1994) Vitam. Horm. (New York) 49, 131-195.2. Speiser, P. W., Dupont, B., Rubinstein, P., Piazza, A., Kastelan, A. & New,

M. I. (1985) Am. J. Hum. Genet. 37, 650-667.3. Higashi, Y., Yoshioka, H., Yamane, M., Gotoh, 0. & Fujii-Kuriyama, Y.

(1986) Proc. Natl. Acad. Sci. USA 83, 2841-2845.4. White, P. C., New, M. I. & Dupont, B. (1986) Proc. Natl. Acad. Sci. USA

83, 5111-5115.5. White, P. C., Grossberger, D., Onufer, B. J., Chaplin, D. D., New, M. I.,

Dupont, B. & Strominger, J. L. (1985) Proc. Natl. Acad. Sci. USA 82,1089-1093.

6. White, P. C., Vitek, A., Dupont, B. & New, M. I. (1988) Proc. Natl. Acad.Sci. USA 85, 4436-4440.

7. Speiser, P. W., Dupont, J., Zhu, D., Serrat, J., Buegeleisen, M., Tusie-Luna,M. T., Lesser, M., New, M. I. & White, P. C. (1992) J. Clin. Invest. 90,584-595.

8. Donohoue, P. A., Van Dop, C., McLean, R. H., White, P. C., Jospe, N. &Migeon, C. J. (1986) J. Clin. Endocrinol. Metab. 62, 995-1002.

9. Speiser, P. W., New, M. I. & White, P. C. (1988) N. Engl. J. Med. 319, 19-23.10. Tusie-Luna, M. T., Traktman, P. & White, P. C. (1990) J. Biol. Chem. 265,

20916-20922.11. Lipkowitz, S., Stern, M. H. & Kirsch, I. R. (1990)J. Exp. Med. 172,409-418.12. Higashi, Y., Hiromasa, T., Tanae, A., Miki, T., Nakura, J., Kondo, T.,

Ohura, T., Ogawa, E., Nakayama, K. & Fujii-Kuriyama, Y. (1991) J.Biochem. (Tokyo) 109, 638-644.

13. Higashi, Y., Tanae, A., Inoue, H., Hiromasa, T. & Fujii-Kuriyama, Y.(1988) Proc. Natl. Acad. Sci. USA 85, 7486-7490.

14. Carroll, M. C., Palsdottir, A., Belt, K. T. & Porter, R. R. (1985) EMBO J.4, 2547-2552.

15. Monckton, D. G., Neumann, R., Guram, T., Fretwell, N., Tamaki, K.,Macleod, A. & Jeffreys, A. J. (1994) Nat. Genet. 8, 162-170.

16. Watkins, W. S., Zenger, R., O'Brien, E., Nyman, D., Eriksson, A. W.,Renlund, M. & Jorde, L. B. (1994) Am. J. Hum. Genet. 55, 348-355.

17. Jasin, M. & Zalamea, P. (1992) Proc. Natl. Acad. Sci. USA 89, 10681-10685.18. Li, H., Cui, X. & Arnheim, N. (1990) Proc. Nati. Acad. Sci. USA 87,

4580-4584.19. Hogstrand, K. & Bohme, J. (1994) Proc. Natl. Acad. Sci. USA 91, 9921-

9925.20. Jeffreys, A. J., Tamaki, K., Macleod, A., Monckton, D. G., Neil, D. L. &

Armour, J. A. L. (1994) Nat. Genet. 6, 136-145.21. Wedell, A., Thilen, A., Ritzen, E. M., Stengler, B. & Luthman, H. (1994)

J. Clin. Endocrinol. Metab. 78, 1145-1152.22. Amor, M., Parker, K. L., Globerman, H., New, M. I. & White, P. C. (1988)

Proc. Natl. Acad. Sci. USA 85, 1600-1604.23. Wyatt, R. T., Rudders, R. A., Zelenetz, A., Delellis, R. A. & Krontiris,

T. G. (1992) J. Exp. Med. 175, 1575-1588.24. Lifton, R. P., Dluhy, R. G., Powers, M., Rich, G. M., Gutkin, M., Fallo, F.,

Gill, J. R., Jr., Feld, L., Ganguly, A., Laidlaw, J. C., Murnaghan, D. J.,Kaufman, C., Stockigt, J. R., Ulick, S. & Lalouel, J. M. (1992) Nat. Genet.2, 66-74.

25. Pascoe, L., Curnow, K. M., Slutsker, L., Connell, J. M., Speiser, P. W., New,M. I. & White, P. C. (1992) Proc. Natl. Acad. Sci. USA 89, 8327-8331.

26. Mornet, E., Dupont, J., Vitek, A. & White, P. C. (1989) J. Biol. Chem. 264,20961-20967.

27. Lipkowitz, S., Garry, V. F. & Kirsch, I. R. (1992) Proc. Natl. Acad. Sci. USA89, 5301-5305.

28. Sinnott, P., Collier, S., Costigan, C., Dyer, P. A., Harris, R. & Strachan, T.(1990) Proc. Nati. Acad. Sci. USA 87, 2107-2111.

Proc. Natl. Acad. Sci. USA 92 (1995)