The mechanism of meiotic homologue pairing

J. theor. Biol. (1984) 106, 605-615

The Mechanism

of Meiotic

MARJORIE Department (Received

of Zoology,

Homologue

Pairing

P. MAGUIRE

University of Texas, Austin, U.S.A.

Texas 78712

13 June 1983, and in revised form 26 September 1983)

Homologous chromosome pairing involves the moving together of matching chromosomes or chromosome segments across substantial distances within a nucleus. Although the time in the life cycle of initial association of homologues varies among organisms, it may well be that similar underlying mechanisms for its occurrence prevail throughout sexually reproducing eukaryotes. The means by which pairing its accomplished is in no case understood. In the apparent absence of a long range specific force of attraction, simple partial models have been proposed which rely for the most part upon interactions of chromosome ends (telomeres) with specialized portions of the nuclear envelope. While such interactions, as well as the persistence of chromosome orientation established by mitotic anaphase poleward movement of centromere regions, may provide in many cases for closer than random positioning of some parts of homologues, the distances remaining to be traversed are still long range in physical-chemical terms. Also, the specific pairing observed in some kinds of rearranged segments is not facilitated by such circumstances, even if synapsis is initiated at available homologous telomere pairs and proceeds to completion by a “zip-up” process. A unified, more complex model is considered which is designed to accommodate the various relevant findings. It invokes the interaction of intranuclear structures with intercalary and/or terminal chromosomal pairing sites, e.g. filamentous structures which specifically bind to these, and a contractile system involving proteins such as actin and myosin to draw homologues together. Introduction

The more spectacular techniques of molecular biology so far have not provided an understanding of the intractable problems of the mechanics of chromosome behavior at meiosis. Hope for progress presently is directed mainly toward application of increased ingenuity at light microscopic and ultrastructural levels of observation, and the choice of subject material which carries relevant mutations or chromosome rearrangements for cytological and genetical analysis. Organisms differ in the details of their meiotic chromosome behavior, sometimes drastically, as in the case of the male of Drosophila melanogaster 605

0022-5193/84/040605+11 %03.00/O

0 1984 Academic Press Inc. (London)

Ltd.

606

M.

P.

MAGUIRE

where there is normally little or no crossing over, and special mechanisms to provide for normal disjunction have apparently evolved. But major features of meiosis are remarkably similar among eukaryotes where evolutionary divergence is thought to have spanned as much as two billion years. One of our problems is to decide which differences among organisms actually represent minor variations on a theme that prevails in a general way throughout most sexually reproducing eukaryotes.

Synapsis Possibly the most striking of the consistent features of meiosis is the intimate pairing of homologues which arranges matching portions in tight side-by-side, parallel array. This is called synapsis and characteristically involves the formation of a remarkable structure, the synaptonemal complex (SC) between the homologues. The SC indeed seems to represent a common theme throughout eukaryotes with minor variation in its structural anatomy among organisms (Westergaard & von Wettstein, 1972). Observations of the SC will be dealt with here mainly as they relate to the larger problem of the initial establishment of homologue pairing. For example, although the SC is normally found between homologues at the synaptic stage where matching pairing partners are available, it is also found at this stage (more sporadically) binding nonhomologous portions, most commonly where no matching partners are present for these portions. The synaptic pairing pattern seems to fit the supposition that wherever appropriately organized chromosomes approach each other to within some maximum distance at this stage, SC structure will be completed between them regardless of matching specificity, but homologues are by far most likely to achieve the required juxtaposotion in rapid time. There is reason to believe that synaptic “zipping-up” occurs from points of initiation since nonhomologous synapsis is commonly found in the vicinity of breakpoints of rearrangement heterozygotes. Most observers agree that non-homologous synapsis, if it is to occur. probably usually tends to follow homologous synapsis in time, but the stage of nonhomologous synapsis may vary from case to case. A remarkable phenomenon called “synaptic adjustment” has been reported in mammals where, for example, inversion heterozygote configurations seem consistently to include homologously synapsed loops at early synaptic stage, which then seem to be transformed to the straight configurations expected for homozygous normal sequence bivalents, even though a substantial proportion of the straightened loops must contain chiasmata (Moses et al., 1982). Necessarily, then, in this case, the final synaptic configurations involve nonhomologous synapsis of the inverted region. This form of adjustment

HOMOLOGUE

607

PAIRING

seems not to occur in maize (Gillies, 1981; Maguire, 1981) but observations of EM serial reconstructions of triploid and tetraploid Bombyx oocytes have been interpreted as evidence for synaptic revision (Rasmussen, 1977; Rasmussen & Holm, 1979). It is uncertain whether synaptic adjustment is an idiosyncratic or common phenomenon. A lesson from the mammalian systems, however, which must be born in mind, is that the final disposition of SC at synaptic stage is not necessarily the initial one. As for functions of the SC, it is commonly believed that it bears an important relationship to the crossover process although its precise role or roles remain unclear: crossover site establishment may occur immediately before, during or following completion of SC formation (see Maguire, 1972, and Beyers & Goetsch, 1982, for evidence for its occurrence at each end of this range), and it has even been suggested that the SC may serve to inhibit the formation of additional, possibly “excess”, crossover sites, so that the distribution of crossovers and crossover interference may, in some cases at least, come to depend on the mode of deployment of the SC (Egel-Mitani, Olson & Egel, 1982; Maguire, 1968). Also there is evidence (Maguire, 1982) to support the idea that a late function of the SC, before it is disassembled following synaptic stage, is to provide somehow for the establishment of the continuing sister chromatid cohesiveness which usually seems to be required for chiasma maintenance until normal anaphase I disjunction. Establishment

of Homologue

Pairing in Preparation

for Synapsis

The achievement of synapsis, once homologues are appropriately juxtaposed seems a minor problem in comparison to the question of how homologues or matching parts of homologues come to be positioned next to each other in the first place. Although homologues generally seem to be at least roughly paired in somatic and germline cells of many Dipterans, there are cases in a number of other organisms where identifiable homologous chromosomes or choromosome regions have been illustrated to be widely separated as late as premeiotic interphase (Walters, 1970) or early meiotic prophase (John, 1976), and long range specific forces of attraction capable of propelling homologues toward each other across such distances are unknown to physics (Fabergt, 1942). Is there a general solution common to most eukaryotes of this most difficult problem? In an attempt at formulating such a solution it has been noted that telomeres (the chromosome ends) at the time of initiation of synapsis tend to be associated with the inner nuclear membrane, and are, or become aggregated in a way which may facilitate the pairing of homologous telomeres (Gillies, 1975a, for review),

608

M.

P.

MAGUIRE

although the area of concentration of telomeres varies in different organisms from a small to a very large proportion of the total inner nuclear membrane surface. In some cases where details of synapsis in progress were available (from EM serial section reconstruction), patches of completed SC were inferred to have been formed earliest in distal regions which tend to encompass telomeres, and complete synapsis was thought to ensue mainly by way of the zip up process. Hence a widely respected model holds that initial homologue pairing may involve dynamic interaction of telomeres and nuclear membrane although the necessary sources of energy and guidance for direction of movement are unclear. A recent, more complex and (to many) attractive, modification of this model notes that in some organisms it can be shown that there tends to be association of nonhomologous chromosome arms of most similar length at their telomeres in mitotic cells (Bennett, 1982; Lewin, 1981). Then according to this model, the diploid complement is organized early (perhaps prezygotically) into two genomic chains, each tending to contain the various heterologues consistently in the same sequence. At meiosis, it is reasoned, appression of the two genomic chains could presumably readily accomplish homologue pairing. It seems to be implied that the initial establishment of the serial sequence of the genome chains depends in some unknown way upon matching length of adjacent chromosome arms. But it has been found that revisions of arm length by way of pericentric (between arm) inversions, so that heterozygotes for the inversions would be expected to have differing genomic chromosome sequences by this scheme, does not, in fact, produce the chaos predicted at meiotic prophase in these heterozygotes (Maguire, 1983~). If there is a general solution to the problem of homologue pairing, there are in fact a number of lines of evidence which strongly suggest that it cannot simply depend upon zipping from the initial establishment of synapsis of homologous telomeres: In some cases where synapsis in progress has been observed in EM serial section preparation reconstructions, SC patches have been found to be scattered along homologue pairs, indicating that there have been multiple sites of synaptic initiation (which may or may not include telomeres) (Gillies, 1975b; Holm, 1977). Also, intercalary pairing partner switches have been found in multivalent synaptic configurations in polyploids and polysomics (where more than two copies of homologues are present) (Henderson, 1969). More damning to general applicability of any model for homologue pairing which depends upon initial association of homologous telomeres is the high frequency of observed or inferred pairing of rearranged segments with their normal sequence counterparts in such cases as ring chromosomes (Schwartz, 1953) and insertional translocations (Maguire, 1965). In these cases homologous telomeres are totally missing

HOMOLOGUE

PAIRING

609

from the portions which nevertheless pair with seemingly high efficiency. Thus matching telomeres could not even have provided for preliminary alignment. In the case of the insertional trfnslocation cited it can be argued that a homoeologous telomeric association might conceivably have set the stage for pairing, but there is evidence against the existence of affinity between these telomeres. Statistical studies of chromosome position at synaptic completion stage suggested that the homeologues involved were no more closely positioned than heterologues, although extra (nonsynapsed) homologues were much more closely positioned than heterologues (Maguire, 1961). In another case which argues against an important role of homologous telomere association, normal recombination frequency has been reported in the vicinity of a homozygous terminal deficiency where normal telomeres are missing in both members of a pair of homologues (Rhoades, 1978). On the other hand, a role of specialized intercalary chromosome regions in the establishment of homologue pairing which is effective for crossing over can be inferred from studies of recombination frequencies in Drosophila melanogaster reciprocal translocation heterozygotes (Hawley, 1980). Still another revealing study involved maize trisomics and tetraploids which were heterozygous for a paracentric (within arm) inversion (Doyle, 1963). The results implied that preferential pairing of internally sequentially matched chromosomes had occurred. This suggests that intercalary chromosome parts contributed strongly in this case to the inspiration for initiation of pairing which is effective for crossing over. Also extremely difficult to reconcile with models which call for a determinative role of telomere association for homologue pairing are observations of chromosome behavior in organisms with zygotic meiosis. In several such cases observations suggest that immediately following karyogamy, the two parental genomes are represented by two separate clusters of chromosomes (Lu & Raju, 1970; Singleton, 1953), then at a somewhat later stage the two sets of chromosomes are found to be intermingled; next homologues may appear to be aligned at a substnatially greater than synaptic distance, and then synapsis of homologues is commenced with initial synaptic origin sites perhaps preferentially, but not exclusively, located distally. Direct observation of this process in living cells might be highly informative but has not so far been technically feasible. It may be important that the chromosomes in these cases of zygotic meiosis tend to be condensed at the stage of homologue pairing. It has often been suggested in the past that the process of synapsis of elongate chromosomes would be expected to produce interlocking bivalents if chromosomes are randomly positioned at the outset, and if synapsis is initiated at multiple sites, as in cases mentioned above. Since bivalent

610

M.

P.

MAGUIRE

interlocking is usually rare at synaptic completion, this in itself has often been taken as evidence that homologues may generally become at least roughly paired during the premeiotic or an earlier mitosis when they are relatively condensed, so that the probability of interlocking during pairing at meiosis would be considerably reduced. Some direct cytological and other evidence of pairing in premeiotic stages has been reported (Comings, 1980, for review). On the other hand, the possibility that interlocking during the process of synapsis actually occurs frequently, but is rarely observed because it is corrected after completion of synapsis (e.g. by breakage and reunion) has been suggested by some observers of EM serial reconstructions of a variety of organisms (Rasmussen & Holm, 1980) These workers argue that they have discovered the existence of mechanisms for interlock correction and that such mechanisms obviate any need for advanced pairing at a condensed stage to avoid interlocking later at meiosis. While it may eventually prove to be true that initial interlocking followed by its resolution commonly occurs, the evidence so far for this does not seem compelling. In the cases described the initial “interlocking” illustrated usually involves inclusion of a synapsed bivalent or an unsynapsed arm between as yet unsynapsed arms of another bivalent, so that actual entrapment at synaptic completion was not necessarily destined to occur, and the reported frequency of even such cases of “interlocking” during the synaptic process seems much too low to approach expectation from random distribution of elongate chromosomes at early meiotic prophase, followed by initiation of homologue synapsis at multiple sites. Only rare and at best “equivocal” cases of potential entrapment were reported in studies of Lilium synapsis sites of synaptic initiation were found in progress, where multiple (Holm, 1977). If in at least some organisms homologues must have a prearrangement to facilitate their pairing without interlocking in advance of meiotic prophase, and if telomere-nuclear membrane interaction cannot readily account for the general problem of homologue pairing, what are the remaining possibilities? It can be quickly said that evidence for the general utility of special pairing capabilities for centromeres or conspicuous heterochromatic regions is sadly lacking. Homologous centromeres may consistantly fail to associate in rearrangement or polysomic material (e.g. Burnham et al., 1972) and often appear to synapse last (Rasmussen & Holm, 1980), although centromeres may serve to some extent indirectly, by virtue of remnant orientation following mitotic anaphase movement (Avivi & Feldman, 1980). Regions which synapse regularly in spite of revised location are devoid of apparent heterochromatin (Maguire, 1965), and deletion of heterochromatin may have no apparent effect on homologue pairing

HOMOLOGUE

PAIRING

611

(Yamamoto & Miklos, 1977). Hilliker, Holm & Appels (1982) conclude that Drosophila autosomal meiotic pairing sites are specific euchromatic regions, and Lifschytz & Hareven (1982) have reported evidence that heterochromatin plays no role in homologue pairing at interphase in Drosophila although such pairing exists at this stage. Heterochromatin in some cases even seems to inhibit or slow homologue pairing to some extent (Kaltsikes & Gustafson, 1980; Naranjo & Lacadena, 1982). A Unified

Model

It seems to me that we are driven to the supposition that intercalary chromosome regions possess autonomous homologue pairing capabilities and that there is currently no good reason to hypothesize direct functional intervention of the nuclear membrane in such pairing capacity. Instead, I believe we might profitably search for intranuclear specialized structures. The existence of fibrillar connections which bind specific DNA sequences and draw together homologues (with matched spacing of these sequences) was postulated early by Holliday (1968), and Bennett et a/., (1979) have reported the finding of intranuclear bundles of microfilaments at leptotene, zygotene and premeiotic interphase in a number of cereal plant species. These bundles most often seemed to form links between masses of chromatin, and it was suggested that they might function in establishing and maintaining homologue alignment for subsequent synapsis. Another kind of observation seems to me to be exceedingly important. That is that in diverse organisms synapsis seems to be preceded by a stage at which whole homologues or long portions of homologues tend to be aligned in at least rough parallel fashion at substantially greater than synaptic distance (Gillies, 1979; Lu & Raju, 1970; Maguire, 1984; McDermott, 1971; Zickler & Sage, 1981). In fact rough alignment of more than two homologues has been seen or inferred to occur in triploids (Rasmussen, 1977; Rasmussen et ai., 1981) and in allohexaploids where both homologues and homoeologues may be aligned (Driscoll, Bieling & Darvey, 1979; Holboth, 1981; Yacobi, Mello-Sampayo & Feldman, 1982). That at least rough alignment of homologues commonly occurs in mitotic cells of Dipterans is widely recognized. In the absence of long range specific forces of attraction it is difficult to imagine a mechanism for the establishment and maintenance of such alignment short of something akin to the following (complex and demanding) scenario, which is a further elaboration of earlier schemes proposed by Holliday (1968) and by the writer (Maguire, 1974): At some time prior to the stage at which alignment at a distance is seen (perhaps one or more cell divisions earlier) homologues make contact at correspond-

612

M.

P.

MAGUIRE

ing specialized sites. At this time sequence specific binding protein (or ribonucleoprotein) filaments are formed between the homologues at matching pairing sites, and these filaments may be extended to considerable length over subsequent stages. The basis of homology would reside either in matched spacing of specialized pairing DNA sequences (Holliday, 1968), or on the presence of unique, matching sequences at corresponding locations in homologues (Maguire, 1974). The latter might be necessary to account for the high fidelity of homologous pairing observed. At synaptic stage it is hypothesized that conditions promote contraction of the connectors, perhaps by way of the strategic addition of contractile system proteins such as actin and myosin. A wide range of variations can be imagined in different organisms: the number of points of connection per chromosome could differ; the stage of installation of connectors could vary from zygotic mitosis in Dipterans, to several mitoses in advance of meiosis in some organisms where it has been suggested that increasing homologue association occurs at these stages (Brown & Stack, 1968; Juricek, 1975; McDermott, 1971) to possibly the last premeiotic mitosis in other organisms (Maguire, 1983b), to even early meiotic prophase in still other organisms. In organisms where it would have to occur at this late stage the model is complicated by the requirement of a special stirring mechanism to ensure that homologous chromosomes will become juxtaposed to allow binding, and it would probably also require either only a single site of synaptic initiation or mechanisms to avoid or correct interlocking. The movements suggested by the observations of Parvinen & Soderstriim (1976) at synaptic stage in rats might conceivably provide for the stirring. Since pairing must necessarily follow karyogamy and occur during meiotic prophase in organisms with zygotic meiosis, the condensed state of their chromosomes at this stage might conceivably provide for adequate interlock prevention, and if so the only special adaptation that seems required then for these cases (by this model) is a special stirring mechanism. In organisms with advanced establishment of homologue connectors (at least Dipterans) mitotic chromosome movements might provide sufficient stirring, e.g. in conjunction with spindle activity at prometaphase or anaphase, or with the prophase movements described by Rickards (1981). This model is obviously burdened with a great deal of complexity. In addition to the special features already mentioned, in appropriate cases, connectors would need to be replicated or reformed at each mitotic division following their initial formation. At some stages the array of relaxed connectors would be very complex. But if this situation indeed prevails, the apparent random distribution of elongate homologues at early meiotic prophase would not pose a difficult problem, for they would be preset to

HOMOLOGUE

PAIRING

613

pair. The manner of formation of the connectors would presumably usually rule out interlocking. Individual homologue pairs might even first achieve connection at different times, and in this regard the apparent synapsis in a single homologue pair seen in several cells at premeiotic mitotic prophase in maize (Maguire, 1983b) and the close associations of one or more apparent homologue pairs seen in human spermatogonia (McDermott, 1971) may be relevant. Actin and/or myosin have been reported to be associated with meiotic chromosomes (De Martin0 et al., 1980; Karsenti, Gounon & Bornens, 1978; Wahrman, 1981), and Rickards (1981) has suggested that actin and myosin may generally be involved in chromosome movement, although there is considerable controversy about the common presence of these proteins. (Difficulty of fixation for their preservation and detection may present a technical obstacle.) Studies of nuclear matrix architecture and protein composition have provided evidence for a complex three dimensional array of protein filaments, among which actin seems to be a major component (Capco, Wan & Penman, 1982). Concluding

Remarks

The challenge is clear: to demonstrate the existence of appropriate connectors (and a special stirring mechanism in advance of their formation in some cases), or to find an alternative, more suitable new model for a mechanism of homologue pairing, or to counter the objections raised here and elsewhere to currently favored models (which have not so far included means for the necessary movement of chromosomes with respect to each other), or even to reconsider the possibility of existence of a suitable long range specific force of attraction (Faberge, 1942). Two other considerations should be mentioned: It seems unlikely that simple diffusion of structures as immense as chromosomes could provide for adequate motion of chromosome segments with respect to each other to establish homologue pairing. It seems a reasonable guess that establishment of homologue pairing is a major meiotic theme and that variations on mechanisms among eukaryotic organisms will prove to be relatively minor adaptations of an initial evolutionary achievement. This work was supported by PHS Grant GM 19582. REFERENCES Avrw, L. & FELDMAN,M. BENNETT,M. D.,SMITH,J. 289.

(1980). Hum. Genet. 55,281. B.,SIMPSON,S. & WELLS,B. (1979). Chromosoma

(BerL)71,

614

M.

P. MAGUIRE

BENNETT, M. D. (1982). In: Sysfematics Associarion, Special Volume No. 20, Genome Evolution (Dover, G. A. & Flavell, R. B., eds). p. 239. New York: Academic Press. BEYERS, B. & GOETSCH, L. (1982). Mol. gen. Genet. 187, 47. BROWN, W. U. & STACK, S. M. (1968). Bull. Torrey hot. Club 95, 369. BURNHAM, C. R., STOUT, J. T., WEINHEIMER. W. H.. KOWLES, R. U. & PHILLIPS, R. L. (1972). Genetics 71, 111. CAPCO, D. G., WAN, K. M. & PENMAN, S. (1982). Cell 29, 847. COMINGS, D. E. (1980). Hum. Genet. 53, 131. DE MARTINO, C., CAPANNA, E.. NICOTRA. M. R. & NATALI, P. G. (1980). Cell Tissue Res. 213, 159. DOYLE, G. G. (1963). Generics 48, 1011. DRISCOLL, C. J., BIELIG, L. M. & DARVEY, N. L. (1979). Genetics 91, 755. EGEL-MITANI, M., OLSON, L. W. & EGEL. R. (1982). Herediras 97, 179. FABERG~, A. C. (1942). J. Genet. 43, 123. GILLIES, C. B. (1975~). A. Rev. Genet. 9, 91. GILLIES, C. B. (19756). Tray Lab. Carlsbern 40. 135 GILLIES, C. B. (1979). Genetics 91, 1. GILLIES, C. B. (1981). Chromosoma (Berl.) 83, 575. HAWLEY, R. S. (1980). Generics 94, 625. HENDERSON, S. A. (1969). In: Handbook of Molecular Cytology (Lima-de-Faria, A. ed.) p. 326. New York: American Elsevier. HILLIKER, A. J.. HOLM, D. G. & APPELS, R. (1982). Genet. Res. 39. 157. HOLBOTH, P. (1981). Carlsberg Rex Commun. 46, 129. HOLLIDAY, R. (1968). In: Replication and Recombination of Generic Material (Peacock. W. J. & Brock, R. D., eds). p. 157. Canberra: Australian Academy of Science. HOLM, P. B. (1977). Carlsberg Res. Commun. 42, 103. JOHN, B. (1976). Chromosoma (Berl.) 54, 295. JURICEK, D. K. (1975). Chromosoma (Berl.) 50, 313. KALTSIKES, P. J. & GUSTAFSON, J. P. (1980). Can. J. Genet. Cyrol. 22, 666. KARSENTI, E., GOUNON, P. & BORNENS, M. (1978). Biol. Cellulaire 31, 219. LEWIN, R. (1981). Science 214, 1334. LIFSCHYTZ, E. & HAREVEN, D. (1982). Chromosoma (Berl.) 86, 443. Lu, B. C. & RAJU, N. B. (1970). Chromosoma (Berl.) 29, 305. MAGUIRE, M. P. (1961). Erpl Cell Res. 24, 21. MAGUIRE, M. P. (1965). Genetics 51. 23. MAGUIRE. M. P. (1968). Genetics 59, 381. MAGUIRE, M. P. (1972). Genetics 70. 353. MAGUIRE, M. P. (1974). Caryologia 27, 349. MAGUIRE, M. P. (1981). Chromosoma (Berl.) 81, 717. MAGUIRE, M. P. (1982). Cytologia 47, 699. MAGUIRE, M. P. (1983a). Genetics 104, 173. MAGUIRE, M. P. (1983b). .I Hered. 74, 93. MAGUIRE, M. P. (1984). Cytologia. (In press). MCDERMO?T, A. (1971). Can. J. Genet. Cyrol. 13, 536. MOSES, M. J., POORMAN, P. A., RODERICK, T. H. & DAVISSON, M. T. (1982). Chromosoma (Berl.) 84, 457. NARANJO. T. & LACADENA, J. R. (1982). Theor. appl. Genet. 61, 233. PARVINEN, M. & S~DERSTR~M, K. 0. (1976). Nature 260, 534. RASMUSSEN, S. W. (1977). Carlsberg Res. Commun. 42, 163. RASMUSSEN. S. W. & HOLM. P. B. (1979). Carlsberg Res. Commun. 44, 101. RASMUSSEN, S. W. & HOLM, P. B. (1980). Herediras 93, 187. RASMUSSEN, S. W., HOLM, P. B.. Lu, B. C., ZICKLER, D. & SAGE, J. (1981). Curlsberg Rex Commun. 46, 347. RICKARDS, G. K. (1981). In: Mitosis/Cytokinesis (Zimmerman, A. M. & Forer, A., eds). p. 103. New York: Academic Press.

HOMOLOGUE

PAIRING

615

RHOADES, M. M. (1978). In: Maize Breeding and Genetics (Walden, D. B., ed.) p. 641. New York. John Wiley & Sons. SCHWARTZ, D. (1953). Am. Nat. 87, 19. SINGLETON, J. R. (1953). Am. .I. Bot. 40, 124. WAHRMAN, J. (1981). In: Chromosomes Today (Bennett, M. D., Bobrow, M. & Hewitt, G., eds). Vol. 7, p. 105. London: George Allen & Unwin. WALTERS. M. S. (1970). Chromosoma (Berl.) 29, 375. WESTERGAARD, M. & VON WE~TSTEIN, D. (1972). A. Rev. Genet. 6, 71. YACOBI,Y.Z.,MELLO-SAMAYO,T. &FELDMAN,M. (1982). Chromosoma (BerL)87,165. YAMAMOTO, M. & MIKLOS, G. L. C. (1977). Chromosoma (BerL) 60, 283. ZICKLER, D. & SAGE, J. (1981). Chromosoma (Berl.) 84, 305.