Repeated DNA of the human Y chromosome

The evolution of eukaryotes has been accompanied by an enormous increase in genome size (Britten & Davidson. 1969). This increase is largely the result of an expansion of sequences that do not code for protein structure (Galau et al. 1976). While some of this ‘non-coding’ DNA undoubtedly regulates gene expression, it is likely that most of it has either some other, as yet unknown, function or no function. A significant fraction of such ‘non-coding’ DNA is composed of various families of repeated sequences interspersed with single copy DNA. The ubiquity of this interspersion pattern (Britten & Davidson. 1971) has suggested that it may play some fundamental role in the function of the genome, such as regulation of gene expression (Davidson & Britten, 1979) or in chromosome structure (Manuelidis & Ward. 1984). An alternative view, based on the way many repeats may multiply and spread within the genome, is that repeated DNA is parasitic (Orgel & Crick. 1980) or selfish (Doolittle & Sapienza. 1980). The human Y chromosome provides a setting for examining the organization, evolutionary dispersion and possible function of various categories of repeated DNAs for several reasons. First, it has a very high proportion of repeated sequences (Kunkel. Smith & Boyer. 1979) and only a limited number of known functional genes (Goodfellow. Darling & Wolfe. 1985). Second, it has a very limited region that participates in meiotic pairing (Cooke. Brown & Rappold. 1984; Simmler et al. 1985). Thus the majority of Y-chromosome sequences are not involved in the usual meiotic recombination events. Third, the human Y chromosome is of recent evolutionary origin in that most of its sequences have homology with the X chromosome or autosomes of other primates but not with their Y chromosomes (Szabo et al. 1980; Kunkel & Smith. 1982; Burk. Patrick & Smith, 1985a; Page, Harper, Love & Botstein. 1984; Singer. 1982).

In broad terms, there are three general categories of repeated DNA: short, interspersed elements, called SINEs (Singer. 1982); long interspersed elements called LINEs (Singer. 1982) and tandemly clustered repeated elements. In this paper, we will review the distribution and organization properties of such repeats on the human Y chromosome.

The major human SINE, the Alu familv (Schmid & Jelinik. 1982). is repeated about 300IXM) times and dispersed throughout the genome (Rinehart. Ritch. Deininger & Schmid. 1981). It is homologous to an abundant small nuclear RNA (Jelinek et al. 1980) and to double-stranded heterogeneous nuclear RNA (Jelinek et al. 1980; Fritsch. Lawn & Maniatis. 1980) and is transcribed in vitro by RNA polymerase 111 (Duncan et al. 1980). These findings, together with the widespread dispersion of SINEs and the presence of short direct repeats flanking some SINEs (Singer. 1982). suggest that Alu repeats may have been dispersed about the genome via self-primed RNA polv-merase III transcription followed by reverse transcription and duplication of a target site (Jagadeeswaram. Foget & Weissman. 1981; Van Arsdell et al. 1981). It has also been suggested that Alu repeats may have initially arisen as pseudogenes of 7SL RNA (Ullu & Tschudi. 1984; Ullu & Weiner. 1985) or transfer RNA (Daniels & Deininger. 1983).

Although detailed studies have not yet been completed. it is clear that the spectrum of Alu sequences on the human Y chromosome differs from that of the rest of the genome. We have carried out preliminary studies of the Alu repeats on the human Y chromosome by using cloned Alu sequences to probe restriction digests of genomic DNA from a somatic cell hybrid in which the only cytologically detectable human chromosome is the Y (Burk et al. 1985a). Alu probes which reflect average Alu sequences, such as Blur 2 and 8 (Deininger et al. 1981). detect few homologous sequences in these digests when hybridized under conditions that detect the majority of Alu repeats in human genomic DNA. Probing under more relaxed conditions yields a significant increase in the number of hybridizing Y-chromosome sequences. Similar results have been obtained by probing recombinant DNA libraries constructed from somatic cell hybrids in which the only human chromosome is the Y (Burk et al. 1985a; Wolfe. Erickson.

Rigby & Goodfellow. 1984) with ‘typical” Alu sequences. Thus, while the human Y chromosome contains a significant number of Alu sequences, they are considerably diverged relative to the average genomic Alu sequence.

The human LINE (Adams et al. 1980), also designated the Kpn family (Shafit-Zagardo. Maio & Brown. 1982) or LI family (Voliva et al. 1983). is abundant (estimates range from several thousand (Adams et al. 1980; Manuelidis & Biro, 1982) to 100000 (Hwu, Roberts, Davidson & Britten, 1986)], widely dispersed in the genome (Shafit-Zagardo et al. 1982), and usually flanked by single-copy DNA (Adams el al. 1980). Many LINE sequences are about 6 kb (Adams et al. 1980) and colinear (Miyaki, Migita & Sakaki, 1983). However, ‘scrambled” LINE elements (Potter, 1984) and shorter elements with truncated 5’ ends also exist (Grimaldi & Singer,1983). We have shown (Schmeckpeper. Willard & Smith. 1981) that the major LINE restriction fragments in the human genome do not have a marked chromosome-specific organization, but one example of chromosome specificity in LINE organization has been reported in the African Green Monkey (Lee & Singer. 1986).

We (Schmeckpeper, Scott & Smith, 1984) and others (Kole. Haynes & Jelinek, 1983; Shafit-Zagardo, Brown. Zavodny & Maio. 1983) have detected primate LINE transcripts, the vast majority of which lack a poly(A) tail, are localized to the nucleus and are synthesized by RNA polymerase IL LINEs are often flanked by short direct repeats, and some have a poly(A)-rich sequence at their 3’ end (Grimaldi, Skowronski & Singer. 1984). These observations, coupled with the 5’ truncation of many LINEs. are consistent with their insertion into new chromosomal locations by a reverse-transcription mechanism similar to that thought to produce pseudogenes (Kole et al. 1983). However, not all LINEs have a structure compatible with this mechanism of insertion. Some lack flanking direct repeats (Grimaldi et al. 1984), some lack poly(A) tracts and several have permuted arrangements (Kole et al. 1983). Based on sequence comparisons among various LINEs it has been suggested that the family as a whole may be evolving in a manner termed ‘concerted evolution’ (Martin et al. 1985). Mechanisms proposed for ‘concerted evolution” include gene conversion, duplicative transposition and unequal sister chromatid exchange (Dover & Flavell, 1984). mechanisms which could also contribute to the dispersal of LINEs within the genome.

We have previously shown that the major LINE restriction fragments of the human X chromosome and a selected subset of autosomes are identical to those detected in whole genomic DNA (Schmeckpeper et al. 1981). Detailed restriction analyses of genomic DNA from the somatic cel) hybrid in which the only cytologically detectable human chromosome is the Y demonstrate that the LINE repeats on the Y are indistinguishable from those found elsewhere in the genome. These studies have utilized cloned LINE probes from the 3’ and 5’ halves of a complete LINE and restriction digests that reveal specific LINE organizations (Schmeckpeper et al. 1981). An example of these results is shown in Fig. 1 which shows there is no detectable difference in the restriction fragments from the X and Y chromosomes detected with a mixture of the 3’ and 5’ probes. Since we have previously demonstrated that the X-chromosome LINEs cannot be distinguished from the LINEs detected in whole genomic DNA (Schmeckpeper et al. 1981). the Y-chromosome LINEs are also representative of genomic LINEs. Knowing the abundance of LINEs on the X chromosome (Schmeckpeper et al. 1981) and comparing the relative hybridization intensities obtained in Fig. 1. it can be deduced that the abundance of LINEs on the Y is about that predicted from the size of the Y chromosome and the abundance of LINEs in the genome as a whole.

Tandemly clustered repeats, by still poorly defined mechanisms, can change their copy number at a particular locus, change their chromosomal location quickly over evolutionary time and acquire chromosome-specific characteristics which arc somehow distributed among most family members at a particular locus. Most mammalian species have a family of homologous tandemly repeated DNAs (Brutlag, 1980) clustered at centromeres and other heterochromatic sites (Miklos & John. 1979). Extensive restriction analyses reveal subsets of sequences (termed domains) which can be chromosome specific (Brown & Dover. 1980; Lee & Singer. 1982). In primates, such a repeated DNA family, alphoid DNA (Singer. 1982). exists as tandem arrays of a small unit often found near centromeres. The units within a single array are similar to each other, perhaps being maintained by unequal crossing-over (Manuelidis. 1982). These alphoid repeats have been shown to have chromosome-specific organizations (Willard. Smith & Sutherland. 1983; Wolfe et al. 1985) and the sequences seem to be evolving rapidly since many differences occur between species. As shown by Wolfe et al. (1985). the Y chromosome has a unique set of alphoid repeats which can be distinguished from the alphoid repeats of other chromosomes by sequence and by the periodicity of particular restriction sites within the tandem array Since the range of sequence divergence among Y-chromosome alphoid monomers is similar to that found on other chromosomes (Wolfe et al. 1985) and since most chromosomes have a specific set of alphoid repeats which are distinguishable by sequence and restriction site periodicity, the Y chromosomes is not unique with respect to the characteristics of its alphoid repeats.

Thus, for the major categories of repeat sequences which the Y chromosome shares with the rest of the genome, the characteristics of LINE and alphoid sequences are typical of these found elsewhere in the genome. In contrast, the predominant Alu sequences on the Y arc considerably diverged relative to the average genomic Alu sequence. Thus, if both LINEs and Alu repeats became associated with the Y chromosome by a mechanism involving RNA intermediates. they must have invaded the Y at different times or at quite different rates. Alternatively, they may have been incorporated into the Y chromosome by different mechanisms or been subject to different selective pressures.

In addition to the repeats described above, the Y chromosome contains two major sets of repeats that are specific to the Y chromosome (Cooke. 1976). As seen in Fig. 2. digestion of male genomic DNA with restriction enzyme Hael 11 yields two distinct ethidium-bromide-staining bands (3 4 and 21 kb) that are not seen in similar digests of female DNA. Since both Hae\11 fragments are male specific (Fig. 2). have a concentration within the genome proportional to the number of Y chromosomes present and are present in a somatic cell hybrid where the only human chromosome is the Y (data not shown), they must be derived from the Y chromosome. Fig. 2 also demonstrates that the 3-4 and 2-1 kb FMelll fragments, isolated from genomic DNA by a combination of physical techniques (Kunkel & Smith. 1982). do not cross-hybridize.

While the 3 · 4 and 2 · 1 kb fragments are themselves from the Y chromosome, each contains sequences which cross-react with DNA from other chromosomes. Failure to hybridize with DNA from a mouse × human somatic cell hybrid in which the only human chromosome is the X (Dorman. Shimizu & Ruddle. 1978) established that the non-Y-homologues of both the 3 · 4 and 21 kb FMcIH Y fragments are autosomal (data not shown). We have also shown (Kunkel & Smith. 1982) that the 3 · 4 kb Hae111 fragment is composed of Y-specific sequences interspersed with non-Y-specific ones and that the major sites of autosomal homology are chromosomes 1. 9 and 15 (Burk et al. 1985b). It is evident in Fig. 2 that the 2 · 1 kb Hae III fragment, isolated directly from genomic DNA. also detects homologues in Hea III digests of female DNA. To determine if the hybridization to female DNA resulted from homology to the same 2·1 kb /-Meili fragment detecting the malespecific band. Hwelll digests of male and female DNA were probed with a 2·1 kb Hea III fragment cloned into pBR322. As seen in Fig. 3. the cloned probe hybridized to the male-specific 2·1 kb band and to a 1·9 kb band in both male and female DNA. Thus, like the 3·4kb Hea III fragment, the 2·1 kb Hea III fragment, though itself derived from the Y. contains sequences with homology to other chromosomes.

The major concentration of the 1·9 kb autosomal homologues of the Y-specific 2·1 kb Hae III fragment was localized to chromosome 14 by hybridization to restriction digests of DNA from a panel of somatic cell hybrids segregating different combinations of human chromosomes (O’Brien er al. 1983).

The distribution of the 3·4 and 2·1 kb Hea III fragments along the Y chromosome was determined by in situ hybridization (Burk et al. 1985b). Fig. 4 summarizes these results and shows that both fragments are distributed throughout the length of the Y long arm. These results have been confirmed by reassociation kinetics and Southern blot analysis of DNA from individuals with various Y-chromosome deletions.

Cooke (1976) previously presented evidence that the 3·4kb Hae lll fragment was. at least in part, tandemly arranged within the Y chromosome. This conclusion was based on the generation of multimers of this fragment in partial Hae lll digests of genomic male DNA. To determine if the fragments homologous to the2·1 kb Haelll fragment were also arranged in tandem, we performed a series of partial Haelll digests of male and female DNA and hybridized the resulting Southern blots with a 2 · 1 kb Hae lII probe. The results obtained with male DNA (Fig. 5) show a ladder of fragments expected for a tandem repeat. However, the multimers are not based on a unit length of 2 · 1 kb; no band is observed at 4 · 2 kb corresponding to a dimer. Rather, prominent bands are seen at 2 · 4. 4 · 5, 4 · 8 and 7 · 2 kb. suggesting that the basic fragment length is 2· 4 kb. Similar results were obtained with female DNA (Fig. 6). Like the Y fragments, the autosomal fragments appear to be tandem clusters of a 2·4 kb repeat. This interpretation was confirmed by comparing the length of homologous fragments in male and female DNA digested with various restriction enzymes and probed with a cloned 2 · 1 kb Hae lll Y fragment (Fig. 7). With the exception of Hae lll. the restriction enzymes indicated in Fig. 7 generated a fragment of 2· 4 kb in both male and female DNA. Thus, the restriction pattern of Y and autosomal 2· 1 kb Hae lll homologues are quite similar. This observation raises the possibility that the autosomal and Y-chromosome 2·4 kb repeats are essentially identical, differing only in a few restriction sites. That this is not the case can be demonstrated by probing restriction digests of a number of independently derived 2·4 kb Y-repeat clones with radiolabelled male or female genomic DNA. The 2·4kb repeats on the Y chromosome, which contain the 21 kb Haelll fragment, have a single Mspl site which is not present in the autosomal 2·4 kb repeat (see Figure 13; Young. Willard & Smith, 1983). Thus, cloned 2·4kb repeats derived exclusively from the Y chromosome can be obtained from male genomic DNA after digestion with Mspl. If restriction digests of these cloned Y repeats do not have the same hybridization pattern when probed with male and female DNA, then there must be sequence differences between the Y and autosomal 2·4 kb repeats. When restriction digests of the Y-specific 2·4 kb Mspl clones were probed with male and female DNA, a set of fragments of about 300bp was hybridized by both probes, but larger fragments were only detected with the male probe (Fig. 8). Therefore, it is only these smaller Y-chromosome fragments that have cross-reacting homologues in female DNA. These cross-reacting sequences, with homologues within both Y and autosomal 2·4 kb fragments, will be referred to as ‘common sequences’. In contrast, the larger fragments present in these digests of Y-derived 2·4 kb cloned fragments are only detected with male genomic DNA and are therefore Probe: Male DNA limited to the Y. Thus, like the 3·4kb 7/uelII fragment (Kunkel & Smith, 1982), the 2·4 kb Y-fragment is composed of both Y-specific and non-Y-specific sequences.

The demonstration that a significant fraction of the 2·4 kb Y repeat is composed of sequences that are specific for the Y is striking, since the sequences within the 2·4kb Y repeat that have autosomal homologues (common sequences) also detect a 2·4 kb autosomal repeat (Fig. 7). Thus the common sequences are associated with distinguishable Y-specific and autosomal-specific sequences even though the unit size of 2·4 kb is not chromosome-specific. This is unlike the results previously reported for the 3·4 kb Haelll repeat, where autosomal homologues yield restriction patterns that are chromosome-specific and distinct from the Y (Burk et al. 1985b).

Comparisons among several independent Y-2·4kb Mspl clones by cross hybridization (Fig. 9) and restriction mapping (Fig. 10) suggest that there is little variation among the 2·4 kb repeats within the Y chromosome. In these digests, there are no restriction fragments that are unique to any one 2·4 kb Y fragment (Fig. 9) and only a few restriction site differences between cloned fragments (Fig. 10). Comparison of the restriction maps of the cloned fragments with the consensus map derived from analysis of genomic DNA indicates a high degree of similarity among all of the 2·4 kb Y repeats. Although these Y repeats are tandemly arrayed (Fig. 5). they are distributed along the length of the long arm of the Y (Fig. 3) and account for only about 25 % of Y longarm DNA (Cooke. 1976). Hence, they must occur as independent tandem clusters interspersed among other sequences rather than as a single tandem array. The similarity among all Y 2-4 kb repeats suggests that there is little difference among these repeats in the various tandem clusters scattered along the Y.

We have determined the base sequence of two 2·4 kb Msp1 fragments from the Y chromosome (data not shown). Comparison of the sequences of these two fragments confirms the extensive homology suggested by restriction analysis (Fig. 10). The two complete sequences determined in our laboratory show little divergence from the partial sequence of a related 2·4kb element, determined from sequence analysis of Hinf1 fragments isolated from human satellite 1 (Frommer. Prosser & Vincent. 1984). The portion of the 2·4 kb Y fragment with autosomal homology (common sequence) is about 800 bp in length. 40% GC and contains no internal redundancy. As previously reported (Frommer et al. 1984), the 2·4 kb Y fragment has a single region that has some homology to the Alu repeat family. Our analysis indicates that this A/n-like sequence occurs at one end of the common sequence and is flanked by a poly(A) stretch often associated with Alu repeats (Jelinek & Schmid. 1982). The Alu element in the Y common sequence has about 60% homology with the Alu consensus sequence (Jelinek & Schmid. 1982). The homology is greater with the right half of the Alu dimer than with the left half. In this respect, the Alu- like sequence within the 2·4 kb Y repeat is similar to the type II Alu repeats described in Galago (Daniels & Deininger. 1983). Search of the Los Alamos GenBank for homology with the 2·4 kb Y repeat detected no homology except for those with the Alu sequences. As predicted from the homology to the consensus Alu repeat, extensive homology was seen with the right, but not the left, half of each homologous Alu dimer.

It has been proposed that new Alu sequences may be generated from transcripts initiated from the RNA polymerase III promoter within the Alu element. Thus, it is possible that transcripts initiating in the Alu -like sequence within the 2·4 kb repeat might be important in the origin and dispersal of 2·4 kb repeats to various chromosomal sites and help explain the regularity of the 2·4kb repeat organization. However. direct examination of the Y common Alu sequence failed to detect homology with the consensus pol lll split promoter sequence (Fowlkes & Shenk, 1980). While the degeneracy of pol lll promoter sites makes it difficult to exclude a functional pol lll promoter by sequence analysis, there is at the moment no evidence to suggest that these Y repeats are transcribed by pol lll.

The Y-specific regions of the 2·4 kb Y repeats are 85 % AT. A significant fraction of the Y-specihc portion of these repeats is composed of a slightly degenerate 14 bp palindrome. The consensus sequence of this internal repeat is ATAATATA TATTAT.

We have also sequenced an autosomal 2·4 kb fragment. As expected, there is extensive homology between the sequences common to both autosomal and Y repeats. The autosomal common sequence has an Alu repeat Hanked by a poly(A) stretch in an analogous position to that found on the Y. The autosome-specific sequence is about 80% AT. Although both the Y- and autosome-specific sequences have a high AT content, they do not cross hybridize (Fig. 8) and have only limited sequence homology. It is difficult to devise a means for deriving one from the other by base substitution or sequence rearrangement. Their only common features seem to be their high AT content, their length and their interspersion with the common sequence.

Southern blot hybridization of restriction enzyme digests of female genomic DNA with several cloned autosomal 2·4 kb probes is shown in Fig. 11. Each clone detects the same set of fragments in the various digests, suggesting that the overall genomic organization of each repeat is similar. However, the heterogeneous smear of hybridization seen in most lanes suggests that these autosomal fragments might be more heterogeneous than the analogous fragments from the Y. This was confirmed by direct comparison of individual clones (Fig. 12). In this experiment, five cloned 2·4 kb autosomal fragments were digested with the same restriction enzyme (Avall) and each set of five digests was hybridized with a different autosomal clone after Southern blotting. Each of the autosomal 2·4 kb fragments contains sequences that cross-hybridize with all of the other autosomal 2·4 kb clones. These cross-hybridizing autosomal sequences are the previously discussed common sequences that are homologous to the Y 2·4 kb repeat. In contrast, the higher molecular weight autosome-specific fragments do not cross-hybridize. With the exception of clones pHG243 and 244, which appear to be identical, each clone yields a distinct pattern of autosomespecific fragments which only hybridize with the clone from which they were derived. The low level of hybridization seen with pHG245 is the result of a small amount of common sequence linked to the autosome-specific sequence; its self hybridization is much stronger, reflecting its autosome-specific character. Thus, unlike the 2·4 kb Y repeats which are all much the same, the 2·4 kb autosomal repeats contain several distinct autosome-specific sequences, each associated with the same common sequence.

A diagrammatic summary of our analyses of the 2·4 kb tandem repeats is presented in Fig. 13. There are at least four sets of repeats with the same overall organization. On the Y chromosome, common sequences of about 0·8 kb are interspersed with 1·6 kb sequences which are confined to the Y. The majority of the repeats on the Y are composed of the same common and Y-specific sequences. All of the autosomal repeats have the same common sequences which differ from those on the Y by the presence of an additional Hae lll site. This Hae lll site accounts for the smaller unit size of the autosomal Hae lll repeats (i.e. 1·9kb for the autosomal Hae lll repeats compared to 2·1 kb for the Y Mselll repeats). Presumably, the common sequence accounts for the similarity in genomic restriction patterns seen in Fig. 11. The autosomal 2·4 kb repeats have at least four different autosome-specific sequences. These autosome-specific sequences (16 kb) alternate with the autosomal common sequences (0·8 kb) to yield structures analogous to the Y-specific fragments. Thus, the common sequences appear capable of associating with a variety of sequences in such a way that identical tandem repeat units are formed. These findings are consistent with the notion that these common sequences have a degree of mobility within the genome and that the mechanism by which they move determines a particular structural organization.

This notion of mobility is supported by studies of gorilla sequences homologous to the human 2·4 kb repeats. Fig. 14 compares various restriction digests of human and gorilla DNA probed with a 2·4 kb Y fragment. In contrast to results with human DNA (lower panel), the pattern and hybridization intensity of restriction fragments seen in male and female gorilla DNA (upper panel) are identical. This suggests that gorilla 2·4 kb repeat homologues are primarily autosomal and are not present on the gorilla Y chromosome. This has been confirmed by reassociation kinetics and in situ hybridization. Since homologues of the Y-specific segment of the 2·4 kb Y fragment are not detected in gorilla DNA (data not shown), the hybridizing gorilla fragments must contain homologues of the common sequence. In addition to the demonstrated cross-hybridization, two indirect lines of evidence support the notion that the 2·4 kb common homologues are similar in the two species. First, the 2·4 kb autosomal common sequence in humans has a 700 bp fragment flanked by Hin fl and Bst NI restriction sites. Digestion of gorilla DNA with these two enzymes yields a fragment of about 700 bp with homology to the 2·4 kb common sequence (Fig. 15). In addition, a 790 bp fragment can be seen in this figure which also hybridizes with the 2·4 kb common sequence. An equivalent fragment is not seen in similar digests of human DNA. This suggests that, as in humans, gorillas may have more than one common sequence. Second, restriction enzymes with sites in the 2·4 kb common sequence that yield 2·4 kb fragments from human DNA also yield distinct fragments, of 2·7 and 3·0 kb. in gorilla (Fig. 16). Direct comparison of the similarity of human and gorilla common sequences requires analysis of cloned sequences. However, the gorilla repeat has proved to be difficult to clone and such a comparison cannot yet be made.

The presence of two fragments (2·7 and 3·0kb), containing a 700 or 790 bp sequence homologous to the common sequence of the human 2·4kb repeat, suggests that, as in humans, the gorilla common sequence may be associated with more than one noncommon sequence. We examined gorilla DNA to determine if homologues of the autosome-specific sequences of human 2·4 kb repeats are associated with the gorilla common sequences. Hybridization of gorilla DNA with subclones of the human 2·4 kb fragment containing only the autosomal-specific sequences demonstrates that these sequences also have identifiable repeat structures in the gorilla (Fig. 17). The repeat length of these sequences is different from that detected in human DNA and from the repeats in gorilla detected with the common sequence (Fig. 16). Thus, while both components of the human autosomal 2·4 kb repeats (the common and autosomespecific sequences) are present in the gorilla, they are not associated with each other. These observations provide additional evidence that the common sequences are capable of associating with a variety of sequences in such a way that precisely defined repeating units are generated.

The structure of the genomic repeating elements homologous with the Y 3·4kb Hae lll fragments also suggests that it may contain sequences allowing for genomic mobility. Like the 2·4kb repeat, the 3·4kb Hae lll fragments contain Y-specific sequences interspersed with non-Y-specific ones. However. unlike the 2·4 kb Y repeats, the 3·4 kb Y fragments are heterogeneous with several non-Y-specific and many Y-specific sequences (Kunkel et al. 1979). In addition, there are several distinguishable autosomal domains which contain homologues of the 3·4 kb repeat . While the Y-chromosome fragments contain all of the various sequences common to the Y and autosomes, each autosomal domain with 3·4 kb common sequence homology contains just one common sequence (Burk et al. 1985b). Thus the organizational heterogeneity of the various autosomal sequences with Y 3·4 kb homology is much greater than those with 2·4 kb Y-fragment homology. Based on our published work (Burk et al. 1985b) and analyses of cloned Y 3·4kb fragments, a model of the organization of Y 3·4 kb fragments can be constructed. As shown in Fig. 18. tandem clusters of these repeats, containing a heterogeneous mixture of sequences, occur on the Y. Their distribution throughout the length of the Y long arm indicates the existence of several separate tandem clusters. All tandem clusters on the Y are characterized by a common set of Hae lll sites at 3·4kb intervals. Based on analyses of individual cloned 3·4kb Haelll fragments, each 3·4kb fragment contains a particular non-Y-specific sequence interspersed with a number of different Y-specific sequences. The position and length of the internal non-Y-specific sequences are variable with respect to length and position. Each non-Y-specific sequence has homology with a distinct autosomal domain. The structural characteristics of these fragments are reminiscent of the scrambled repeats and some transposable elements in Drosophila (Rubin, 1983).

Most of DNA sequences within the human Y-chromosome are not associated with the Y chromosome of other primates. Recent transpositions of single copy DNA from the X chromosome (Page et al. 1984; Cooke el al. 1984) and autosomes (Burk et al. 1985a) to the human Y chromosome have been reported. In at least one case, it has been documented that DNA sequences from the non-pairing region of the human Y chromosome that have X-chromosome homology in humans have homology with the X but not the Y chromosome in other higher primates (Page et al. 1984). In addition, as presented in this report, repeated DNA sequences that are specific for the human Y chromosome (Kunkel & Smith, 1982) or have specific organizational characteristics on the human Y are not present on the Y chromosome of other primates (Cooke, Schmidtke & Gosden. 1982). Rather, those human Y-chromosome repeat sequences that have homology with non-human primates detect homologous autosomal sequences. Although the chromosomal location of these Y-repeat homologues is not known in non-human primates, they map to a few specific autosomal sites in humans. The major autosomal regions with 3·4 kb Hae lll Y-repeat homology have been mapped to chromosomes 1, 9. 15. 16, 21 and 22 (Burk et al. 1985b). while the autosomal 2·4kb repeats have a major concentration on chromosome 14. Since these major Y-chromosome tandem repeats are not known to be transcribed, it is unlikely that they have become associated with the human Y chromosome by a mechanism involving an RNA intermediate. It seems more likely that a significant fraction of the human Y chromosome originated by a series of translocations or transpositions involving the X chromosome and several autosomes. In this regard, it should be noted (Burk et al. 1985b) that the chromosomal distribution of the 3·4 kb Hae III Y-repeat autosomal homologues corresponds to autosomal sites that are frequently involved in Y-chromosome long-arm translocations (Smith. Fraser & Elliot. 1979) and are frequently associated with both homologous and heterologous somatic pairing (Ford. Callen. Roberts & Jahnke. 1983; Schmid. Grunert. Haaf & Engel. 1983).

If multiple transposition events were involved in the generation of the human Y chromosome, they must have been followed by selective amplification of particular sequences which resulted in the variety of Y-chromosome tandem repeats described here. The heterogeneity of sequences within each family of tandem repeats suggests that such amplification events must have occurred several times. The occurrence of similar clusters of each family of tandem repeats dispersed along the Y chromosome, with each cluster having the same overall organization, suggests that the events giving rise to the various Y repeats must have occurred at several independent sites. Alternatively, these repeats may have been generated as single tandem arrays and subsequently shuffled to various sites along the Y long arm. perhaps as a result of an inherent instability of the non-pairing region of the Y. The possibility of the linear instability of the Y chromosome has previously been suggested (Affara et al. 1986). Whatever the mechanism, the regularity in the organization of these repeats, which contain heterogeneous collections of sequences, and their specific chromosomal organizations in both humans and related primates, make it unlikely that they arose from purely stochastic processes (Dover. 1982) and suggest that they may play some role in the function of segregation of the Y chromosome. This notion would be strengthened if it could be demonstrated that sequences with structural characteristics similar to those described here for the human Y chromosome also occur in the Y chromosome of non-human primates.

We would like to thank Drs L. M. Kunkel. H. F. Willard and S. H. Boyer for their valuable contributions to much of the work presented here. We also thank Ms C. J. Decker and Ms C. T. Comey for their helpful discussion of the manuscript and Ms C. Dove for help in preparing the manuscript. This work was supported in part by NIH grant HD 17161 (KDS) and the Howard Hughes Medical Institute.