Copyright 0 1988 by the Society of America Perspectives

Anecdotal, Historical and Critical Commentarieson Genetics Edited by James F. Crow and William F. Dove

UNEQUAL CROSSING OVER THENAND NOW

HE Bur eye of Drosophila occupies a But the was not to rest there. Two years T rather special place in the . later, in 1925, STURTEVANTsolved the riddle of the As the experiments of STURTEVANTand MORGAN unusual properties of Bur in a publication entitled (1923) and STURTEVANT(1925) showed, it was the “The effects of unequal crossing over at theBar locus first example of unequal crossing over and also the in Drosophila.” It is a remarkable paper. At the outset, first demonstration of position effect. Over the past STURTEVANTadvances the hypothesis that both the 65 years our understanding ofposition effects has ultra: Bur mutants and the wild-type revertants arise remained largely unresolved, but our knowledge and from f + B fu’lf B fk individuals by unequal crossing appreciation of the importance of unequal crossing over at Bur. He proposed that if the site of recombi- over has grown substantially. While muchremains to nation lies to the left of Bur in one , but be learned, I think it is now clear that thephenomenon to the right of it on the other, then the ultra Bar of unequal recombination is a matter of considerable mutants should be more properly referred toas double significance for genetic . It is, therefore, the Bur because they would be genotypicallyf BB fu’ or subject of thisPerspectives on the occasion of the 65th f’ BB fu and, hence, duplicated for Bur. The wild- anniversary of its discovery. type revertants, on the other hand, would be f ‘fu or I am going to focus on two basic modes by which ffu‘ and deficient for Bur. Accordingly, double Bur unequal crossing over occurs. One involves direct and wild-type revertants are necessarily reciprocal tandem redundancy, whereas the other is mediated products of the same exchange event and should be by transposons. My purpose is to illustrate these two recovered in equal numbers. In fact, they are not. situations with examples from diverse in This is probably due to thereduced viability ofdouble order to provide a sense of the context in which they Bur and the difficulty of distinguishing it from Bur occur and to understand the questionsthey raise. alone. However, what madethe unequal crossing over Because our concept of unequal crossing over derives hypothesis so compe‘lling wasSTURTEVANT’S discovery directly from studies of the Bur mutation, it is useful and use of a new allele of Bur known as Bur-infrabar to consider these in some detail. (Bi).This mutation arose spontanteously in a Bur stock Bur is a dominant, homozygous viable, sex-linked and reduces the number of eyefacets toa value mutation that reduces, in heterozygotes, the number intermediate between B and wild type and thereby of facets inthe compound eye to abouthalf their usual made it possible to devise a crucial test ofthe unequal value. By 192 1 ZELENYhad shown that Bur was unsta- crossing over mechanism. STURTEVANTwas able to ble and could mutate at considerable frequency (6 X derive, for example, the tandem arrangement of BB’ lo”) to either wild type or to a more severe pheno- from B/B’ heterozygotes and then to recover sepa- type he referred to as ultra Bur. However, no mech- rately the B and B’ allels in their proper order with anism to explainthis anomalous behaviorseemed respect to flanking markers. obvious until STURTEVANTand MORGAN(1923) re- It would be 1 1 years, until the discovery ofpolytene ported theresults ofa disarmingly simpleexperiment. , before STURTEVANT’Sunequal crossing They marked chromosomes carrying Bur (B, 57.0 cM) over hypothesis could be confirmed. MULLER,PRO- with the flanking forked (f, 56.5 cM) and KOFIEVA-BELGOVSKAYAand KOSSIKOV (1 936) and fused (fu,59.5 cM) to produce f’ BfilfBfu’ heter- BRIDGES(1 936) found that the Bur mutation itself is ozygotes and found that the Bur’ revertant progeny a direct tandem duplication of seven bands that com- were alsorecombinant for the adjacent markers, being pose section 16A1-7 of the polytene map. Unequal eitherf +fu’ orffu. Their conclusion was unequivocal: crossing over results when the distal repeat of one “. . . reversion of Bur to normal is associated with chromosome recombines with the proximal repeat in crossing over at or near the Bur locus.” the opposite homolog to yield Bur’ revertants and

Genetics 120: 1-6 (September, 1988) 2 K. D. Tartof

total, they obtained 9 nonrecombinant B to B+ rever- tant males among 215,376 progeny for a frequency of about 4 X This value is 15 lower than f ++B B+ fu+ the comparableinterchromatid event which they I1 I tI foundto be 6 X (46 recombinantrevertants X among 78,433 progeny). Similarly,JACKSON and FINK(1 985) have shown that in Saccharomyces theoccurrence of unequal sister exchange between two copies of a direct tandem duplication of the HIS4 gene is also 10-20- fold less frequent than the comparable rate of inter- chromosomal recombination. Although the cause of f B' fu' this striking suppression of intrachromatid unequal I exchange in both flies and yeast is unknown, it does I suggest the presence of a well regulated pathway Bar+ controlling these events. JACKSON and FINKhave pro- posed that suppressing sister chromatid crossing over mightbe selectively advantageous by reducing the production of deficiencies and inversions that would I I otherwise result from intrastrand exchange between double Bar direct or inverted repeats on the same chromosome. FIGURE1 .-Unequal crossing over at Bar. The large open rec- This speculation may be particularly relevant to mi- tangle at the top represents the Bar tandem duplication whereas cro-repeats, sequences 2-10 bp in length and sepa- each smaller rectangle below denotes one copy of the Bar+ (B') rated from each other by less than l kb, that partici- locus which spans section 16A1-7 onthe polytene map. One of two pate in thegeneration of spontaneousdeletions possible misalignments of the Bur duplication is illustrated. In this through illegitimate exchange as previously discussed case, unequal exchange inf Bfu'lfBfu heterozygotes producesf B'fu' (Bur+)and f BB fu (double Bar) progeny that possess one and in these Perspectives by ANDERSON(1 987). three copies of the Bar' locus, respectively. Recently, STUART In contrast to the situation for unique genes like TSUBOTA(personal communication) has observed the presence of a Bar and HZS4, highly redundant sequences such as the 7.5-kb transposon inserted at the16A7-16A1 breakpoint in the ribosomal RNA genes (rDNA) have a propensity for middle of the Bar duplication. This suggests that the duplication itself may have been formed in the process of transposon mobiliza- unequal sister chromatidexchange. RITOSSAet al. tion. (1966) showed that in Drosophila there is a 100-200 copy array of these genes imbeddedwithin the centric double Bar progeny (Figure 1). Thus, thereversion of heterochromatin of the X chromosome and another Bar does not involve a loss of the Bar locus as initially on the short armof the Y and thatpartial deficiencies thought, but rather a loss of the duplicate copy of of rDNA result in a short-bristle phenotype known as section 16A; conversely, double Bar actually contains bobbed (bb). He subsequently madethe remarkable three doses of the Bar+ region. discovery that when X chromosome bb mutants are Since recombination was known to take place at the maintained for several generations with a Y chromo- four-strand stage, STURTEVANTconsidered the possi- some deficient for most of its rDNA (Ybb-), as in bb/ bility that unequal sister chromatid exchange might Ybb- males, phenotypically bb+ flies containing a wild- be responsible for some of the changes in Bar. How- type amount of rDNA appear (RITOSSA1968). This ever,after examining more than 36,000 offspring phenomenon has been referred to as "magnification." from homozygous B or B' females, every instance of In order to explain the very existence of bobbed eitherreversion or augmentation of Bar was also mutants, RITOSSAet al. (1966) speculated that they accompanied by crossing over between f and fu. If might arise by unequal crossing over. However, the unequal sister chromatid exchange ever occurs here, possibility that unequal exchange was responsible for it is a rather rare event. Nevertheless, PETERSONand magnification was rejected by both RITOSSA(1968, LAUCHNAN(1963) pursued the issue further in an 1972, 1973) and ATWOOD(1969). Part of the diffi- elegant series of experiments. They searched for ex- culty in understanding this phenomenon was that it ceptional nonrecombinant B+ male offspringpro- occurred only in males, a gender in which meiotic duced fromfemales that were either:(1) hemizygotes, recombination is virtually absent. However, as a result carrying f B fu on oneX chromosome and a deficiency of examining the frequency of magnification in single for Bar on the other;(2) heterozygotes, bearingf B os males rather than in populations of flies, I proposed on one X (os, outstretched small eye, 59.2 cM), and that the mechanism by which magnification occurs is the other an inversion containing f' B os+; or finally unequal mitotic sister chromatid exchange (TARTOF (3), heterozygotes of the genotype f B os#+ B os+. In 1974). I demonstrated that rDNAmagnification arises Perspectives 3 primarily in mitoticallyactive germ cells andat a within the rDNA cluster of mitotic cells (SZOSTAKand frequency such that 80%of the offspring of some 66/ Wu 1980) aswell as meiotic cells (PETES 1980). In Y66- individuals are 6bm+ (magnified wild type 66+). mitoticcells, about unequal sister chromatid This frequency of magnification is several orders of exchange event is observed per generation. In meiotic magnitude higher than expected for interchromoso- cells, the frequency of unequal sister strand exchange mal meiotic recombination events. Moreover, the Y66- is at least 10" per , while recombination be- chromosome not only induces rDNA magnification of tween nonsister is suppressed. 66 mutants but is also able to decrease the rDNA A somewhat similarsituation has alsobeen observed content of the wild-type bb+ locus, a phenomenon in the mouse. The distal ends of mammalian X and Y referred to as "reduction." It was further shown that chromosomes frequently pair and undergo reciprocal magnification and reduction are reciprocal events in exchange in gametogenesis. As a consequence, the 6b/Y66- germ cells, although 66"+ progeny are re- pattern of inheritance of markers located in this re- covered more frequently thantheir reduction-pro- gion is not strictly sex-linked and is called pseudoau- duced lethal counterparts (66""""'). This might be ex- tosomal. HARBERSet al. (1986) have isolated a mouse pected owing to selection against 66"etha' germ cells in containing a single Moloney murine leukemia a manner reminiscent of the under-representation of (M-MuLV) inserted in the pseudoautosomal double Bar recombinants observed by STURTEVANT. region of the Y. The M-MuLV proviral sequence is Finally, magnification of a b6 mutation when present readily transferred from theY to the X and back again in a ring X chromosome is diminished as might be at a frequency of about 10". Moreover, approxi- expected because single (or odd-number) crossovers mately 7% of the offspring from males homozygous are lost as double-size dicentric chromosomes. More for M-MuLV (XMm/r"") contain either noprovirus or recently, we have shown that under nonselective con- two copies of it. Since the proviral insert is flanked ditions, recovery of bb"+ and 6brletha'products is equal by a tandemly redundant sequence (repeat length and that, although the vast majority of the magnified -1.3 kb), a plausible explanation for the gain and loss bb+ progeny from 66/Y66- males arise premeiotically, of proviral DNA is unequal crossing over between the some of these magnifying events also occur at meiosis repeated elements. It is not known if some of these (HAWLEY andTARTOF 1985). events are premeiotic or involvesister chromatids. There aretwo genetic factors crucial to the process Such issues may be resolved, both in this case and in of ribosomal gene magnification and reduction. First, mammalian systems in general, by using restriction the presence of the Ybb- chromosome is required. fragment length polymorphisms (RFLPs) as chromo- Whatmakes this chromosome so mutagenic is not some markers flanking the site of unequal exchange clear. It is deficient for most of its own rDNA, but to determine precisely the source of alteration in gene this alone does not explain its behavior because Ybb- copy number. chromosomes have been constructed with a wild-type In humans, too, there is evidence for unequal cross- rDNA content and they are still effective at inducing ing over. JEFFREYS, WILSONand THEIN(1985) have magnification (HAWLEYand TARTOF1983). How- described aprobe that detects hypervariable, dis- ever, it maybe that Ybb- is deficient for a critical persed, tandemly redundant "minisatellite" regions pairing site and this leads to misalignment between whose repeat lengths vary from 16 to 64 bp and are the rDNA clusters when homologousregions of the X highly polymorphic in the human genome. In fact, and Y synapse. Second, some of the genes (mei-41, these repeat lengths are so polymorphic from one mus-ZO1, mus-108) that control meiotic recombination person tothe next that they provide a means for and DNA repair are also required for magnification- detecting individual-specific DNA. The likely source reduction, whereas others (mei-9, mus-102, mus-109) of such polymorphismis unequal crossing over. In an are not (HAWLEY andTARTOF 1983; HAWLEYet al. incisively direct experiment, JEFFREYS et al. (1988) 1985). It is interesting to note that those genes affect- examined human pedigrees with five different mini- ing rDNA magnification and reduction are also in- satellite probes to determine the rate at which new volved in post-replication repair. alleles (DNA restriction fragments) appear. For each Just as the rDNA of Drosophila may undergo un- individual probe, themutation rate per gamete varied equal sister chromatid exchange in both mitotic and from undetectable to as high as 5 x 10-2 for the most meiotic cells, similar events occur in Saccharomyces. unstable locus. By virtue of site-specific transformation, it has been The examples of unequal crossing over at Bar, in possible to insert a single LEU2 gene into thetandemly rDNA, in the pseudoautosomal region of the mouse arrayed rDNA cluster located on chromosome XI1 of and in the minisatellites ofhumans all share acommon yeast. Measurement of the frequency of increase and structural feature: direct tandem repetition of genetic decrease in LEU2 copy number has provided an une- sequence. It is conceptually straightforward to under- quivocal genetic and molecular demonstration of the stand how unequal crossing over among repeated regular occurrence of unequal sister strand exchange sequences arises whenthere are two or more identical 4 K. D. Tartof sites, and hence, substantial opportunity for misalign- overexpression of the appropriate transposase as well ment between the iterated copies. But what causes as the impact that various recombination and DNA unequal crossing over in genetically unique portions repair mutations might have on transposon-mediated of the genome where tandem duplication is not ap- unequal crossing over. parent? What sort of homology is required, and how Because small, slightly displaced transposons appear extensive must it be, to effect asymmetric exchange? to be such a significant feature of unequal crossing Considerableprogress toward answering these over, the question arises of how many times per ge- questions has beenprovided by GOLDBERGet al. nome one might expect membersof the same transpo- (1983) and DAVIS,SHEN and JUDD (1 987). They ex- son family to be sufficiently close (for instance, within amined the molecular structure of reciprocal dupli- 60 kb) to pair with each other and undergo unequal cations and deficiencies produced by interchromoso- exchange. In Drosophila melanogaster there are about mal unequal exchange in Drosophila females hetero- 70 different mobile element families, each repre- zygous for various white (w) alleles. In the experiments sented on average about 33 times in the euchromatic of GOLDBERGet al. it was found that a 7.2-kb trans- portion of the genome (YOUNG1979). Given these posable element (BEL) was present in both wa and wa4 parameters, SAM LITWIN and I have calculated, by mutants but inserted at slightly different locations in both analytical means (LITWIN1974) andby computer each mutant,about 60 kb apart and in the same simulation, the frequency distribution of two identical orientation. Further analysis demonstrated that asym- transposonslocated on the same chromosome arm metric pairing and exchange between the staggered and displaced by 560 kb. Assuming all transposons to BEL sequences are responsible for the observed du- be randomly distributed, our results predict about 13 plications and deficiencies. The frequency of unequal instances per Drosophila genome, similar to the wa/ exchange between these two white alleles is about 2 X wa4 situation wheretwo members of the same transpo- 1O-4. DAVIS,SHEN and JUDD examined unequal cross- son family reside within 60 kb or less of each other ing over in w"c/wbfheterozygotes.writ possesses an 8.7- and in the same orientation. It follows, then, that the kb roo transposon located at 0 kb on the molecular average genomic frequency of transposon-mediated map of white whereas wbf contains two roo elements, unequal crossing over per meiosis inDrosophila might one at -1.1 kb and the other at +31.9 kb. All three be(2 X X 13or about 3 X lo-'. This is a very roo inserts are oriented in the same direction. Here, high rate for a mutagenic process. too, unequal crossing over is transposon-mediated. In Since the frequency of unequal exchange is so high wric/wbf heterozygotes the frequency of unequal ex- and so ubiquitous, it is not surprising to observe its change betweenthe roo elements at 0 and - 1.1 is four impact on human disease. Either tandem duplications times higher (2 X lob4)than between the roo transpo- (as in Bar) or transposons (as in wa/wa4) may be in- sons located at 0 and+31.9 (5 X This indicates volved. that the interaction between transposons may be in- Color blindness, an X-linked disease that affects versely related to the distance that separates them. about 8% of Caucasian males, in an example of un- What is perhaps most astonishing about theseresults equal crossing overmediated by geneduplication. is that small (8-kb) displaced regions of homology, in This locus codes for the red and greenvisual pigment the form of transposons, are able to pair with each apoproteins and is organized as a head-to-tail tandem other at considerable frequency despite separationby array composed of a single red-pigment gene at the 60 kb and despitebeing surrounded by extensive 5' end followed by a variable number (one, two or regions of standard homology. At present, we do not three) of green-pigment genes (NATHANS, THOMAS understand how transposons participate in this proc- and HOGNESS1986; VOLLRATH,NATHANS and DAVIS ess. Anecdotalobservations (BURKEJUDD, personal 1988). The polymorphism for the number of green- communication)indicate that the heterozygosity of pigment genes is probably explained by unequal cross- transposon locations may beimportant because wbf ing over. Examination of genomic DNA from males homozygotes, with their two roo elements only 33 kb with two different forms of color blindness revealed apart, seem to show no evidence of asymmetric ex- the presence of an interesting pattern of re- change. This would suggest a mechanism of transpo- arrangements(NATHANS et al.1986). One class of son-mediated unequal crossing over whereby homo- color blindness, anomaloustrichromacy, has been log pairing in heterozygotes results in looped-out un- thought to result from the presence of photopigment paired transposons that recombine with each other in with an altered absorption spectrum. In fact, analysis adistance-dependent manner. Still, theextent to of the DNA from such individuals demonstrates the which heterozygosity affects this process needs to be presence of either a 5' red-3' green or a 5' green- clearly defined. It is also possible that a transposase is 3' red hybridphotopigment gene. These data are involved. The genes that control transposon-mediated most easily explained by unequal crossing over be- unequalexchange are yet tobe identified. In this tween red-pigmentand green-pigment loci. This regard it would be useful to examine the effect of seems likely given the fact that their DNA sequences Perspectives 5 are98% identical andthat the number of green- logic manifestations of duplication and mu- pigment genes is so variable. In a second type of color tations as they occur in the human LDL receptor are blindness, referred toas dichromacy, the redor green but a reflection of the mechanism by which the gene photopigment is missing as a consequence of either itself was created. It might be reasonably argued that gene deletion or red-green gene fusion. Here, too, as the numberof gene copies or transposons increases, the mutations can be most easilyexplained by unequal so should the rate of unequal exchange. But unequal crossing over, although may some- crossing over is a mutagenic event that wouldbe times occur. Similarly, hemoglobinopathies, such as expected to have negative, often lethal consequences certain a-thalassemias, hemoglobin Lepore and he- as a result of altering gene dosage or producing novel moglobinKenya, illustrate the concept of unequal gene fusions. Thus, the tendency to constantly accu- exchange between tandemly related loci (for a review mulate gene duplications may be limited by the ad- see COLLINSand WEISSMAN1984). verse impact theseextra DNA sequences usually have. Perhaps the clearest human example of unequal Likewise, increases in transposon copy number may crossing over involving displaced transposons comes also be constrained by their potential for deleterious from studies offamilial hypercholesterolemia. This effects through unequal exchange (LANGLEYet al. malady is a consequence of a defect in the gene coding 1988). forthe LDL(low density lipoprotein) receptor, a We know embarrassingly little about the role un- transmembrane glycoprotein responsible for thebind- equal crossing over plays in the genetic of somatic ing of LDLand its eventual endocytosis in coated pits. cells. Even in Drosophila, where it should be possible LEHRMANet al. (1987a,b) have described a duplication to obtain some information on the frequency of this and a deletion mutation that appear to result from process, there are few compellingfacts. However, unequal crossing over between Alu sequences located malignant cells by their very clonally amplify in different introns of the locus. rare genetic events and therefore provide a useful From the foregoing discussion is is apparent that source of biological material with which to investigate the eukaryotic genome is constantly expanding and this problem. An interesting example in this regard contracting in size. Results from Drosophila indicate concerns the homogeneously staining regions (HSRs) that, for thegenetically unique portion of the genome, of mammalian chromosomes that are found only in transposon-mediated unequal exchange may result in tumor and drug-resistant cells. They represent a form duplications or deficiencies ofat least 60 kb or so once of gene amplificationwhich evidence suggests is a in every 300 meioses. This will have special conse- product of unequal sister chromatid exchange (HOL- quences for wild-typegenes whose phenotypes are DEN et al. 1987). Although direct evidence is lacking, sensitive to dosage. In Drosophila,dosage-sensitive unequal crossing over may also be one of the several loci include Bar, the bithorax complex, runt, Notch, means by which potentially malignant cells establish homozygosity or hemizygosity for somatically reces- Enhancer-Suppressor of Hairless, modifiers of Polycomb, sive cancer genes (KNUDSON197 1). Minutes, and Enhancer-Suppressors of variegation. For the repetitous component of the genome, evidence In hisbook A History of Genetics, STURTEVANT from Drosophila, yeast, mice and humans indicates (1965) modestly regarded the Bar eye case as". . . too thatthe frequency of unequal crossing over per special to serve as a basis for any general picture of mutation. Indeed, it is a special paradigm, but meiosis is on the order of 10" for some sequences. . . ." not in the restrictive sense. I suspect STURTEVANT While it has been suggested that continual unequal would be pleased with the broad and general impor- exchange provides a means for maintaining the ho- tance of his contribution, and how things are turning mogeneity of tandemly repeated sequences(SMITH out. 1973), other processes such as gene conversion may be as important, if not more so, in this regard (JACK- KENNETHD. TARTOF SON and FINK1981 ; KLEIN and PETES1981). Institute for Cancer Research In the context of , new proteins are de- FOX ChaseCancer Center rived from older ones by . Proteins, 7701 Burholme Avenue particularly those with extracellular function such as Philadelphia, Pennsylvania 19 1 1 1 the LDL receptor, the serum albumin family and the epidermal growth factor family, are often built up from a smaller unit of amino acid sequence that is LITERATURE CITED then reiterated as a tandem array. The duplication of ANDERSON,P., 1987 Twenty years of illegitimate recombination. an entire gene, or portions of its internal structure, Genetics 115 58 1-584. may be considered simply a consequence of unequal ATWOOD,K. C., 1969 Some aspects of the bobbed problem in crossing over where the exchange event requires re- Drosophila. Genetics 61 (Suppi.): 319-327. BRIDGES,C. B., 1936 The bar "gene" a duplication. Science 83: gions of homologyeither in the form of multiple gene 210-211. copies or interspersed transposons. Thus, the patho- COLLINS,F. S., AND S. M.WEISSMAN, 1984 The 6 K. D. Tartof

of human hemoglobin. Prog. Nucleic Acid Res. Mol. Biol. 31: a subject withfamilial hypercholesterolemia. J. Biol. Chem. 3 17-462. 262: 3354-3361. DAVIS,P. S., M. W. SHENAND B. H. JUDD, 1987 Asymmetrical LITWIN,S., 1974 The distribution onthe shortest distance be- pairings of transposons in and proximal to the white locus of tween random cuts on opposite strands of DNA. J. Appl. Prob. Drosophila account for four classes of regularly occurring ex- 11: 363-368. change products. Proc. Natl. Acad. Sci. USA 84 174-178. MULLER,H. J., A. A. PROKOFIEVA-BELGOVSKAYAAND K. V. KOSSI- GOLDBERG,M. L., J-Y. SHEEN,W. J. GEHRINGAND M. M.GREEN, KOV,1936 Unequal crossing over in the bar mutant as a result 1983 Unequal crossing-over associated with asymmetrical of duplication of a minute chromosome section. C. R. (Dokl.) synapsis between nomadic elements in the Drosophila melano- Acad. Sci. USSR 1: 87-88. gaster genome. Proc. Natl. Acad. Sci. USA 80: 5017-5021. NATHANS,J., D. THOMASAND D. S. HOCNESS,1986 Molecular genetics of human color vision: the genes encoding blue, green HARBERS,K., P. SORIANO,U. MULLER AND R. JAENISCH, and red pigments. Science 232: 193-202. 1986 High frequency of unequal recombination in pseudoau- NATHANS,J., T. P. PIANTANIDA,R. L. EDDY,T. B. SHOWSAND D. tosomal region shown by proviral insertion into transgenic S. HOGNESS,1986 Molecular genetics of inherited variation mouse. Nature 324 682-685. in human color vision. Science 232 203-210. HAWLEY,R. S., AND K. D. TARTOF, 1983 Theeffect of mei-41 on PETERSON,H. M., AND J. R. LAUGHNAN,1963 Intrachromosomal rDNA redundancy in Drosophilamelanogaster. Genetics 104 exchange at the bar locus in Drosophila. Proc. Natl. Acad. Sci. 63-80. USA 50 126-133. HAWLEY,R. S., AND K. D. TARTOF, 1985 A two-stage model for PETES,T. D., 1980 Unequal meiotic recombination within tandem the control of rDNA magnification. Genetics 109 691-700. arrays of yeast ribosomal DNA genes. Cell 19: 765-774. HAWLEY,R. S., C. H. MARCUS,M. L. CAMERON,R. L. SCHWARTZ RITOSSA,F. M., 1968 Unstable redundancy of genes for ribosomal AND A. E. ZITRON, 1985 Repair-defect mutations inhibit RNA. Proc. Natl. Acad. Sci. USA 60 509-516. rDNA magnification in Drosophila and discriminate between RITOSSA, F.,1972 Procedure for magnification of lethal deletions meiotic and premeiotic magnification. Proc. Natl. Acad. Sci. of genes for ribosomal RNA. Nature New Biol. 240 109-1 1 1. USA 82: 8095-8099. RITOSSA,F., 1973 Crossing-over between X and Y chromosomes HOLDEN,J. J. A., M. R. HOUGH,D. L. REIMERAND B. N. WHITE, during ribosomal DNA magnification in Drosophilamelano- 1987 Evidence for unequal crossing-over as the mechanism gaster. Proc. Natl. Acad. Sci. USA 70: 1950-1954. for amplification of some homogeneously staining regions. RITOSSA, F. M.,K. C. ATWOOD,D. L. LINDSLEYAND S. SPIEGELMAN, Cancer Genet. Cytogenet. 29: 139-149. 1966 On the chromosomal distribution of DNA complemen- tary to ribosomal and soluble RNA. Natl. Cancer Inst. Monogr. JACKSON,J. A., AND G. R. FINK, 1981 Gene conversion between 23: 449-472. duplicated genetic elements in yeast. Nature 292: 306-3 11. SMITH,G. P., 1973 Unequal crossover and the evolution of mul- JACKSON,J. A., AND G. R. FINK, 1985 Meiotic recombination tigene families. Cold Spring Harbor Symp. Quant. Biol. 38 between duplicated genetic elements in Saccharomyces cerevisiae. 507-5 13. Genetics 109: 303-332. STURTEVANT,A. H., 1925 The effects of unequal crossing over JEFFREYS, A. J., V. WILSON AND S. L. THEIN, 1985 Hypervariable at the bar locus in Drosophila. Genetics 10: 117-147. "minisatellite" regions in human DNA. Nature 314 67-73. STURTEVANT,A. H., 1965 A History of Genetics. Harper 8c Row, JEFFREYS,A. J., N. J. ROYLE, v. WILSON AND z. WONG, New York. 1988 Spontaneous mutation rates to new length alleles at STURTEVANT, A.H., AND T. H. MORGAN,1923 Reverse mutation tandem-repetitive hypervariable loci in human DNA. Nature of the bar gene correlatedwith crossing over. Science 57: 746- 332: 278-281. 747. KLEIN, H. L., AND T. D. PETES, 1981 Intrachromosomal gene SZOSTAK,J. W., AND R. WU, 1980 Unequal crossing over in the conversion in yeast. Nature 289 144-148. ribosomal DNA of Saccharomyces cerevisiae. Nature 284: 426- KNUDSON,A. J. JR., 197 1 Mutation and cancer: statistical study of 430. retinoblastoma. Proc. Natl. Acad. Sci. USA 68 820-823. TARTOF,K. D., 1974 Unequal mitotic sister chromatid exchange as the mechanism of ribosomal RNA gene magnification. Proc. LANGLEY,C. H., E. MONTGOMERY,R. HUDSON AND N. KAPLAN, Natl. Acad. Sci. USA 71: 1272-1276. 1988 On the role of unequal exchange in the containment of VOLLRATH,D., J. NATHANSAND R.W. DAVIS,1988 Tandem transposable element copy number. Genet. Res. (in press). array of human visual pigment genes at Xq28. Science 240: LEHRMANM. A., J. L. GOLDSTEIN,D. w. RUSSELLAND M. s. 1669-1672. BROWN, 1987a Duplication of seven exons in LDL receptor YOUNG,M. W., 1979 Middle repetitive DNA: a fluid component gene caused by Alu-Alu recombination in a subject with familial of the Drosophila genome. Proc. Natl. Acad. Sci.USA 76 hypercholesterolemia. Cell 48: 827-835. 6274-6278. LEHRMAN,M. A., D. W. RUSSELL,J. L. GOLDSTEINAND M. s. ZELENY,C., 1921 The direction and frequency of mutation in the BROWN,1979b Alu-Alu recombination deletes splice acceptor bar-eye series of multiple allelomorphs of Drosophila. J. Exptl. sites and produces secreted low density lipoprotein receptor in Zool. 34: 203-233.