Meiotic segregation in Drosophila melanogaster females: molecules, mechanisms, and myths.

INTRODUCTION

The ground-breaking studies on meiotic nondisjunction in Drosophila melanogaster published by Calvin Bridges in 1914 (5) laid the experimental cornerstone for the chromosome theory of inheritance and began nearly a century of studies of meiosis in Drosophila females. For much of this century Drosophila females have been the preeminent model organism for the study of meiosis.

The last comprehensive reviews of meiosis in Drosophila males and females were written approximately twenty years ago (4, 62, 98). They were thorough and incisive summaries of the knowledge obtained largely by genetic studies of chromosome aberrations and meiotic mutations. Subsequently, that genetic knowledge has been enhanced, but, more crucially, studies of meiosis now involve electron and confocal microscopy and molecular biology. As a consequence, our understanding of several aspects of the meiotic process is much greater.

This review takes a different approach from its predecessors. The primary focus is on integrating the crucial elements of the genetic analysis with more recent studies of the molecular biology and cytology of meiosis. This review deals only with meiosis in Drosophila females since the meiotic processes in Drosophila males and females are so dramatically different, except for functions that ensure sister-chromatid cohesion, that both systems cannot be adequately addressed in a single review.

The critical issues discussed in this review are:

1. The extraordinary control of chiasma position likely reflects a compromise between the difficulties inherent in resolving proximal chiasmata and the low ability of very distal chiasmata to ensure disjunction.

2. Where exchange does not occur, disjunction is ensured by two achiasmate segregation systems, the achiasmate homologous system and the heterologous system. Both system depend on the nod kinesin-like protein.

3. The achiasmate homologous system is made possible because following pachytene the meiotic chromosomes condense into a dense mass where they stay until prometaphase. This detour allows the maintenance of pairing and perhaps centromere apposition at metaphase.

4. The construction of a bipolar spindle is organized by the chromosomes themselves and requires the product of the ncd kinesin protein.

5. The heterologous system of achiasmate segregation may reflect either the spatial constraints on the chromosomes to project fibers to the poles at prometaphase or a process of homology-independent alignment that can occur either in late pachytene or in the karyosome.

Drosophila female meiosis is an unusual process whose properties have crucial implications for processes of chromosome segregation. We therefore give a brief overview of the cell biology of meiosis, and of the existing formal genetic paradigms. We then discuss in more detail each step of the meiotic process.

OVERVIEW OF FEMALE MEIOSIS

Cell Biology

Each ovary consists of a bundle of ovarioles that contain oocytes arranged in order of their developmental stages (55, 67). The pro-oocyte originates in the anterior-most portion of the germarium where four mitotic cell divisions take place to generate a 16-cell cyst. One cell will become the oocyte, which grows at the expense of the sister germline cells, or "nurse" cells, contained in the cyst. The nurse cells, which subsequently become highly polytenized, supply the ooplasm with cytoplasmic organelles and macromolecules, including maternal transcripts.

The Drosophila oocyte and its development are by convention divided into 14 stages, each stage representing a continuum in oocyte growth and development beginning with stage 1 (the terminal 16-cell cluster in the germarium), and ending with the fully developed oocyte at stage 14 (when the meiotic process is arrested at metaphase I). For most of these stages [3--13] the chromosomes in the coocyte nucleus are condensed into a hollow spherical mass called a karyosome. At stage 13 the nuclear envelope breaks down and prometaphase begins.

The earliest steps of meiotic prophase in Drosophila have not been amenable to cytological analysis. However, several workers have characterized pachytene synapsis by electron microscopy (8, 10, 78, 94). At this stage, Drosophila female meiosis appears to conform to the canonical meiotic pathway in that it assembles a full-length synaptonemal complex (SC) along each bivalent. The SC is a tripartite structure consisting of one central and two lateral elements lying between paired chromosomes. In most cases the complex runs the full-length of each bivalent.

Fruit fly meiosis and the canonical meiotic process differ in the absence of diplotene-diakinesis. Instead, after pachytene the chromosomes condense into a tight mass, the karyosome, where they remain through most of oogenesis until spindle formation during prometaphase. Although SC continues to be present during the early stages in which the chromosomes are condensed into the karyosome (it disappears around stage 6), little is known about the architecture of the karyosome in the later stages of oogenesis.

Prometaphase begins at stage 13. There are no obvious centrosomes from which to assemble the poles of the meiotic spindle (112). The spindle is apparently organized by the chromosomes themselves (112). Preliminary evidence from our laboratory suggests that at least two sets of paired centromeres are required to form a bipolar spindle (see below). The organization of the karyosome itself may therefore provide the basis for assembly of a bipolar spindle.

During spindle assembly and elongation at prometaphase, the achiasmate chromosomes move precociously toward the poles in a size-dependent fashion, such that the smaller precede the larger to the poles (112). For example, in a female with nonexchange X and 4th chromosomes, the two 4th chromosomes move to opposite poles ahead of the two nonexchange X chromosomes (Figures 1, 2: see also Figure 2 in Ref. 48). The chiasmate chromosomes remain at the metaphase plate. Thus, when the oocyte arrests at metaphase I, the achiasmate chromosomes are well separated from the chiasmate bivalents. The achiasmate chromosomes do not complete their migration to the poles until anaphase begins.

[CHART OMITTED]

In normal Drosophila melanogaster oocytes, meiosis arrests at metaphase I and resumes only after passage through the oviduct. Progression into anaphase I is heralded by the movement of all chromosomes away from the metaphase plate to the poles.

Genetic Paradigms

Three major genetic systems ensure meiotic segregation in Drosophila females. As in most organisms, exchange is sufficient to ensure segregation, e.g. even in translocation heterozygotes the chiasmata that connect non-homologous centromeres are sufficient to ensure the segregation of those centromeres (43, 92). This chiasmate system for ensuring disjunction depends on euchromatic pairing, on the proper number and positioning of exchange events, and on the ability of the resulting chiasmata to ensure centromere coorientation at prometaphase.

Meiotic segregation in Drosophila can also be ensured by either of two achiasmate segregational systems (46, 48). These two secondary mechanisms of disjunction ensure the regular segregation of both achiasmate homologs and of heterologous chromosomes. Until recently, both homologous and heterologous achiasmate segregations were presumed to be the consequence of a single meiotic system, known as the distributive system (37, 44, 49). However, two lines of evidence have demonstrated the two distinct systems for segregating achiasmate chromosomes (see Table 1). First, three meiotic mutants, Axs, ald, and mei-S51, impair the segregation of nonexchange homologs, but do not prevent heterologous segregations (46, 90, 97, 124). Second, the rules for choosing segregational partners are different for the two types of achiasmate disjunction (37, but see 46). The segregation of achiasmate homologs depends on heterochromatic homologies, whereas heterologous partners are chosen by chromosome size, shape, and availability (29, 30--36, 80, but see 46). These two systems, the homologous achiasmate system and the heterologous system, are described in detail below.

Table 1 Achiasmate segregations in drosophila females

 
                     Homologous   Heterologous 
Size-dependent          --             + 
Shape-dependent         --             + 
Homology-dependent      +              -- 
Requires [axs.sup.+]    +              -- 
Requires [ald.sup.+]    +              -- 
Requires [S51.sup.+]    +              -- 
Requires [nod.sup.+]    +              + 

PAIRING, EXCHANGE, AND SEGREGATION

Genetic Approaches to Chromosome Pairing

What little is known about the mechanisms of meiotic pairing has been deduced from studies of the exchange behavior of structural rearrangements. Since these studies have been extensively reviewed (1, 98), we do not discuss them in detail here, except to note that taken together they have created a vicious paradox. The ability of heterozygous rearrangements to suppress the occurrence of exchange over long distances clearly indicates that long regions of homology (i.e. large chromosomal domains) are essential for normal levels of recombination.

However, Craymer (21) demonstrated that very short intervals of homology (\50 cytological bands on a polytene chromosome in size) are capable of recombining at extremely low frequencies, even when they occupy very different positions in the genome. If short regions of homology are sufficient for at least some level of pairing and recombination, then why should heterozygosity for a translocation breakpoint prevent recombination over much of a chromosome arm?

We propose that aberrations have little or no effect on the frequency with which exchange is initiated, but rather that aberrations exert their effects by creating discontinuities within functional domains of the SC. In yeast, for example, only those recombination events that mature in the context of the SC become functional chiasmata (26, 99).

We base this suggestion on Maguire's observations regarding exchange and synapsis in paracentric inversion heterozygotes in maize (64). She noted that the frequency of homologous pairing within inverted regions (i.e. inversion loop formation) precisely equaled the chiasma frequency. One could explain her data by suggesting that there is some probability, which is a function of the length of the inverted region and the normal levels of recombination in that region, that exchange will be initiated within the inversion. Where an SC can be formed, as a consequence of looping one homolog to create a large region of homology, the recombination intermediate will mature to a chiasma. However, where large domains of continuous SC cannot be formed, reciprocal exchange and thus chiasmata will not occur.

This hypothesis has two attractive features for meiosis in Drosophila females. The suggestion that SC formation in Drosophila occurs in large domains with defined endpoints can explain otherwise vexing data on exchange in translocation heterozygotes (42, 98). Roberts observed that heterozygosity for T(3;4)s or T(2;4)s could suppress exchange over large portions of an autosomal arm, sometimes reducing exchange to 10--20% of normal (98). Moreover, he noticed that breakpoints in the center of each autosomal arm were more effective suppressors of autosomal recombination than were breakpoints at either end. Two explanations for these data are possible. First, following Roberts, pairing could be initiated in the center of the chromosome arms. This model asserts that different breakpoints have different effects on pairing and/or recombination. Alternatively, all break-points could suppress exchange within a given chromosomal domain, but the central domains account for most of the exchange on the autosomes, as demonstrated by Lindsley & Sandler (61). Unfortunately, because Roberts did not determine the position of the residual exchanges along the arm, he could not distinguish between these two classes of models.

Hawley (42) examined X chromosomal exchange in females heterozygous for a well-marked normal sequence X chromosome and for a translocation (either a T(1;4) or a T(1;Y)). He noted that the X chromosome could be divided into four precisely mappable domains such that heterozygosity for a breakpoint within a domain suppressed recombination throughout that domain but not in adjacent domains. Specifically, he demonstrated that both homologs must be continuous between two sets of paired boundaries for normal exchange to occur within that domain. These findings have been strengthened by the observation that free duplications for the putative boundary sites can suppress exchange throughout the lengths of the intervals bounded by that site, and by the finding that boundaries can be both moved and deleted with the expected genetic consequences. Similar sites have been found by McKim and co-workers in C. elegans (73, 75, 76).

These boundaries could correspond either to sites required for the synapsis of large domains, or the domains themselves could execute synapsis (and the boundaries are just the ends of these domains). This is not to say that the translocations interfere with the initiation of recombination or with the formation of the SC, but rather that a chromosomal discontinuity can preclude the function of SC, in terms of chiasma function, over large bounded domains. (Nor is this to suggest that the appearance of short regions of SC is presaged only by exchange, a hypothesis demolished by Carpenter (10)). Specifically, this hypothesis states only that SC continuity over large domains is necessary for recombination intermediates to be converted to chiasmata.

The translocation studies of Hawley (42) demonstrate that these domains of recombination suppression by breakpoints exist, and that they have precise and mappable boundaries corresponding to regions of intercalary heterochromatin. Perhaps such regions, while corresponding to small constrictions in the polytene chromosomes, are quite large on a meiotic chromosome and thus facilitate the reinitiation of synapsis.

Secondly, this hypothesis is testable by the analysis of pachytene SC morphology in the two classes of aberration heterozygote described above. The model makes predictions similar to the results obtained by Maguire in maize (see above). One can only hope that, perhaps spurred on by this review, someone will take on such an important effort.

Cytological Studies of Pachytene Morphology in Drosophila

Drosophila oocytes contain a typical synaptonemal complex joining paired homologous chromosomes. There is a central element flanked by lateral elements approximately 110 nm apart (8, 10). The complex runs the entire length of each bivalent including the achiasmate 4th chromosomal bivalent. In 69% of the nuclei the proximal regions of each bivalent are located in one region of the nucleus, the chromocenter. In the remaining 31% of nuclei, two or more such chromocenters are observed. In contrast, the telomeres are less restricted in their positions. They are not associated with the nuclear envelope nor clustered in any region of the nucleus.

SC is longest when first completed, i.e. at the beginning of pachytene (10), then steadily decreases to about half its original length, before increasing again in later-stage cysts. Part of this change in length can be accounted for by changes in the thickness of the SC, but some SC material must either be gained, lost, or changed in density to account for all the length changes. The SC persists until after karyosome formation and then disappears.

There is no normal SC visible in meiocytes of females homozygous for …