Proc. Natl. Acad. Sci. USA Vol. 92, pp. 10443-10449, November 1995

Review Meiosis in Drosophila: Seeing is believing Terry L. Orr- Weaver Whitehead Institute for Biomedical Research and Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02142

ABSTRACT Recently many exciting advances have been achieved in our understanding of Drosophila meiosis due to combined cytological and genetic approaches. New techniques have permitted the characterization of chromosome position and spindle formation in female meiosis I. The proteins encoded by the nod and ncd genes, two genes known to be needed for the proper partitioning of chromosomes lacking exchange events, have been identified and found to be kinesin-like motors. The effects of mutations in these genes on the spindle and chromosomes, together with the localization of the proteins, have yielded a model for the mechanism of female meiosis I. In male meiosis I, the chromosomal regions responsible for homolog pairing have been resolved to the level of specific DNA sequences. This provides a foundation for elucidating the molecular basis of meiotic pairing. The cytological techniques available in Drosophila also have permitted inroads into the regulation of sisterchromatid segregation. The products of two genes (mei-S332 and ord) essential for sister-chromatid cohesion have been identified recently. Additional advances in understanding Drosophila meiosis are the delineation of a functional centromere by using minichromosome derivatives and the identification of several regulatory genes for the meiotic cell cycle. The investigation of meiosis in Drosophila has two powerful advantages. The first is the ability of researchers to visualize the meiotic divisions in female and male meiosis (Figs. 1 and 2). The second is a valuable collection of mutations affecting meiosis, an inheritance accumulated over seven decades. The cytology of Drosophila meiosis has been investigated extensively during the past 20 years. In males, the meiotic cells are accessible to light and electron microscopy, so chromosome behavior as well as spindle structure and kinetochore ultrastructure have been described (1-8). In females, the early events of meiosis leading to homolog pairing, synaptonemal complex formation, and recombination were visualized in detail by electron microscopy of serial sections of oocytes (911). This painstaking analysis provided an image of the structure of the synaptonemal complex and recombination nodules.

However, until recently the cytology of the chromosomes in later meiosis was not well characterized. The oocyte yolk obscures the meiotic spindle and chromosomes in the light microscope. One of the most significant recent advances in Drosophila meiosis is the improvement of female meiotic cytology to permit analysis of the later stages of meiosis (12). Confocal imaging of whole mount oocytes has provided insights into the mechanisms that normally partition chromosomes during the meiotic divisions and the effects of mutations on those processes. Frequently in biology exceptional cases provide important insights into fundamental mechanisms. This has certainly been true for Drosophila meiosis (Fig. 3). Genetic studies demonstrated that during the first meiotic division in Drosophila females homologs pair and segregate by the common mechanism of recombination, synaptonemal complex formation, and presumably chiasmata formation (13). However, there is an exception. The tiny fourth chromosome does not undergo exchange and yet segregates faithfully. Larger chromosomes also partition properly when they fail to undergo exchange. A set of genes was identified that are necessary for the segregation of nonexchange chromosomes in the female, and the recent molecular and cytological analysis of some of these mutants has provided key information concerning the segregation of all chromosomes during female meiosis. Meiosis in Drosophila males is also an exception to the general mechanism of homolog segregation. Recombination does not occur in males, and synaptonemal complex is not formed. Recent studies of the molecular basis of homolog pairing in males are likely to enhance our understanding of mechanisms of pairing in many systems. Inroads into some poorly understood areas of meiosis have been opened - in Drosophila by particular mutants. The mechanisms responsible for sister-chromatid attachments have been elusive. Mutations in two Drosophila genes result in premature sister-chromatid separation in meiosis. These genes are likely to play a direct role in controlling sister-chromatid cohesion, and their protein products have been identified recently. A minichromosome derivative that is transmitted faithfully has made the centromere amenable

10443

to molecular analysis, and centromere function has been delineated to a 220-kb region. Finally, several genes regulating the meiotic cell cycle have been identified in the past 3 years in Drosophila.

Female Meiosis

The investigation of meiosis in many organisms explained the requirement for recombination for homologous chromosome segregation (14, 15). During meiosis I the homologs pair, and the resulting association between them is necessary for bipolar attachment of the bivalent on the spindle. The tension caused by the forces toward the two poles being counteracted by the forces holding the homologs together results in stable attachment of the bivalent to opposite poles (16). What holds the homologs together? The synaptonemal complex, which during pachytene forms a zipper between the homologs, cannot provide stable homolog attachment because it dissociates during diplotene. Chiasmata are cytologically defined structures observed in many organisms that seem to hold the homologs together until their segregation at anaphase I. Chiasmata appear to be the visible relics of exchange events because there is a clear correlation between chiasmata and the position and number of exchange events (14, 15). However, the physical distinction between a chiasma and a crossover is not clear. Chiasma could be a recombination intermediate such as a Holliday junction or a resolved exchange event in which the two DNA duplexes are held in a cross structure. In summary, the requirement for recombination in proper homolog segregation can be explained by exchange events becoming chiasmata that stably hold the homologs together, ensuring their bipolar attachment to the spindle. It has not been possible to observe chiasmata in Drosophila; however, it is clear that exchange is important for proper chromosome segregation (13). The large chromosomes have about one exchange event per arm during meiosis. Mutations reducing recombination cause aberrant segregation of the homologs in meiosis I. The presumption is that in Drosophila females exchange events lead to chiasmata that serve as stable homolog attachments until the metaphase I/anaphase I transition.

10444

Review: Orr-Weaver

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

FIG. 1. Cytology of Drosophila female meiosis. The images are confocal micrographs of oocytes fixed and labeled with antibodies to visualize meiotic stages. (A) Metaphase I arrest in a mature stage 14 oocyte. The chromosomes are stained with anti-histone antibodies and are shown in yellow. The spindle is stained with anti-a-tubulin antibodies and is shown in red. The unusual female spindle has tapered, narrow poles. The nonexchange fourth chromosomes have migrated toward the poles away from the chromosomal mass at the metaphase plate. (Photograph courtesy of W. Theurkauf, State University of New York, Stony Brook.) (B) Metaphase II. The chromosomes are labeled with 7-amino-actinomycin D and are shown in red. The spindles are labeled with anti-a-tubulin antibodies and are green. The two meiosis II spindles lie in a line, and the chromosomes are on the metaphase plates. (C) Early anaphase II. Labeling as in B. The sister chromatids are separating on each spindle. In B and C mature oocytes were activated to complete meiosis in vitro, making it easier to capture these rapid stages. (Photographs in B and C are from A. Page, Whitehead Institute.)

If chiasmata

serve as

physical attach- (heterologous chromosomes) could segre-

ments between the homologs, how do the gate from each other, and she characternonexchange fourth chromosomes segre- ized the parameters of this segregation.

gate faithfully? If an additional chromo- She found that distributive segregation is some fails to undergo exchange (for ex- influenced by the availability, size, and

ample, because exchange is suppressed by multiple inversions on a balancer homolog) it also will segregate correctly. The segregation of the fourth chromosome and other nonexchange chromosomes was investigated extensively by Grell (17, 18) and termed the distributive pairing system (Fig. 3). Grell found that in Drosophila females chromosomes that lack homology

shape of the chromosomes. For example, the effect of free X duplication chromosomes of increasing size on the segregation of the fourth chromosomes was investigated (18). These free duplications contain the centromeric region of the X together with varying amounts of X euchromatic DNA and are monosomic. If they pair with and segregate from another

chromosome such as the fourth, then the other fourth homolog will move randomly to a pole. This can result in gametes that contain two fourth chromosomes. Thus, segregation of the free duplication from the fourth chromosome can be scored by the appearance of gametes with the free duplication but no fourth chromosomes and gametes with two fourth chromosomes but no duplication. Grell found that when the X duplication chromosome was equivalent in size to the fourth chromosome, the duplication segregates efficiently from the fourth chromosome. In further analysis of the segregation of nonexchange chromosomes, Hawley et al. (19) also found that heterologous chromosomes could segregate from each other during female meiosis. However, in a series of experiments with free duplications they observed that homology can be important in the proper segregation of nonexchange chromosomes (19). This led to an alternative nomenclature, illustrated in Fig. 3, in which the segregation of nonexchange homologous and heterologous chromosomes is distinguished. There is genetic evidence for two pathways for segregating nonexchange chromosomes in that mutations in theAxs, ald, and mei-S51 genes affect the segregation of homologous nonexchange chromosomes but do not affect heterologous chromosomes (for review, see ref. 20). There is a large collection of Drosophila mutants that affect homolog segregation during the first division of female meiosis. Most of these are defective in recombination (13). The reason that reduced recombination leads to aberrant segregation is that it appears that female meiosis has a limited capacity for segregating nonexchange chromosomes. Thus, as more chromosome pairs fail to undergo exchange, the nondisjunction frequency increases. Recent advances come from the analysis of two genes needed for the segregation of nonexchange chromosomes. The nod (no distributive segregation) gene is required only for the segregation of nonexchange chromosomes (21, 22). The gene has been cloned and found to encode a member of the kinesin family of microtubule motor proteins. The N-terminal domain of the NOD protein is homologous to the motor domain of the kinesin heavy chain (23). Mutations that disrupt NOD function result in changes in conserved amino acids, demonstrating the significance of the homology (24, 25). For example, a dominant mutation, nodDTw, is the consequence of an amino acid change in the predicted ATP binding domain. Although it has not been possible to demonstrate motor activity with the purified NOD protein, it is hypothesized to be a plus-end-directed motor based on its conserved structure with kinesin. The ncd (nonclaret disjunctional) gene is necessary for the segregation of nonex-

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

Review: Orr-Weaver

pi rt

rl,

, AL~~~~~~~~~

?F

j

A

AL

FIG. 2. Cytology of Drosophila male meiosis. Shown are phase micrographs of testis squashes fixed and stained with orcein to visualize the chromosomes. (A) Prophase I in a wild-type male. The sex chromosome bivalent is at the top and has the characteristic structure resulting from pairing at a specific site. The large autosomes are paired tightly along their length, and the tiny fourth chromosomes appear as dots. (Photograph by W. Miyazaki, Whitehead Institute.) (B) Anaphase I in a wild-type male. Equal numbers of chromosomes have partitioned to the two poles, and the sister chromatids can be seen to be attached now only at their centromere regions. (C) Anaphase II in a wild-type male. The sister chromatids have segregated equally to the two poles. (D) Anaphase I in a mei-S332 mutant. Only one pole is in focus. The sister chromatids prematurely separated, with the exception of the pair shown by the arrow. (E) Anaphase II in a mei-S332 mutant. Sister chromatids distribute unequally to the two poles and lagards are observed (arrow). (Photographs in B-E courtesy of A. Kerrebrock, Whitehead Institute.) (F) Diagram of sisterchromatid cohesion in meiosis. Early in meiosis I the sister chromatids are attached along their entire length. In most organisms chiasma result from exchange events, as indicated by the crossed chromatid structure. These chiasmata hold the homologs together, and they may be stabilized by cohesion on the chromatid arms distal to the site of the chiasma. After the metaphase I/anaphase I transition (arrow) cohesion is lost on the arms, and the sister chromatids remain attached at their centromeric regions until the metaphase II/anaphase II transition.

change and exchange chromosomes (26). The ned gene also encodes a protein with kinesin homology, but in this case the conserved motor domain lies at the C terminus of the protein (27, 28). A surprising observation was that, in contrast to kinesin, the purified NCD protein has a minus-end-directed motor activity on microtubules in vitro (29, 30). NCD is also capable of bundling microtubules in an ATP-dependent manner (29). A major breakthrough in our understanding of the mechanism of homolog segregation in female meiosis I came from characterization of the cytology of the spindle in oocytes (Fig. 1) (12). The female meiotic spindle differs from that of male meiosis or mitosis in that the microtubules are not organized from the pole but rather from the chromosomes themselves. There are no apparent centrosomes. After pachytene in prophase I the chromosomes are held tightly condensed in a karyosome; individual chromosomes cannot be visualized. The synaptonemal complex is initially present after karyosome formation but then dissociates (7). Prior to nuclear envelope breakdown short microtubules appear associated with the nuclear envelope. After envelope breakdown these appear to be captured by

the chromosomal mass. Consequently, a bipolar spindle is formed in which the greatest microtubule mass is at the center, with the chromosomes, rather than at the poles (12). In several mutants individual chromosomes move away from the spindle and are isolated in the cytoplasm (see below). Consistent with the role of the chromosomes in organizing the spindle, these single chromosomes are able to organize bipolar spindles. At metaphase I in the stage 14 oocyte, the spindle elongates and is very narrow and tapered at the poles (Fig. 1A). Strikingly, the nonexchange fourth chromosomes move out of the chromosomal mass at the metaphase plate and are localized on the spindle near the poles (Fig. 1A) (12). The fourth chromosomes are observed near the poles in living oocytes as well; thus, this position is not an artifact of fixation (W. Theurkauf, personal communication). If the dosage of the nod gene is reduced, other nonexchange chromosomes also will move out of the chromosomal mass toward the pole (12). The extent of migration to the poles is a function of size; the X chromosomes do not migrate as close to the poles as do the tiny fourth chromosomes.

10445

Mutations in the nod and ned genes cause alterations in the position of chromosomes or spindle structure that give insights into the mechanism of segregation of exchange and nonexchange chromosomes. In nod mutants the spindle is normal but nonexchange chromosomes are frequently observed in the cytoplasm outside the spindle (12). These wandering chromosomes can reattach to the spindle and, when they do so, attach randomly to either pole. Thus, the cytological observations account for the nondisjunction and chromosome loss observed in the mutants. A variety of spindle and chromosomal defects is observed in female meiosis in ned mutants (31). In metaphase I the spindle is diffuse. Later in metaphase I/anaphase I, broad, diffuse, and multipolar spindles are observed. The chromosomes are frequently located off the spindle, in the cytoplasm. The dynamics of the spindle have been characterized recently by studying living oocytes from ned mutants. In real time it is observed that in ned mutants the spindle assembles more slowly, and the spindles that do form are not stable (W. Theurkauf, personal communication). The NCD protein is localized along the spindle microtubules during all of the stages of meiosis (32). Thus, the NCD protein appears to be essential for forming and stabilizing the meiotic spindle, possibly by virtue of its capability to bundle microtubules (29, 31, 32). NOD protein localization provides much information concerning its function. The protein is localized all along the chromosomes in prometaphase- and metaphase-arrested oocytes, the only stages examined to date (33). Moreover, the nonmotor domain of the NOD protein binds DNA with a preference for (A+T)rich regions (33). A set of deletions spanning a small free duplication chromosome was used to map the sites at which NOD acts (34). This minichromosome, Dp1187, is transmitted faithfully during mitosis and meiosis in wild-type females, and it is lost at a slight but significant frequency in nod heterozygotes. However, deletions in the minichromosome result in a marked loss of transmission when the dosage of the nod gene is reduced. Consistent with the immunolocalization of the protein along the length of the chromosomes, deletions throughout the noncentromere region of the minichromosome enhance loss in nod heterozygotes. In general, transmission frequency increases with increasing size of the chromosome, although some regions interact more strongly with NOD than others. The cytological and molecular observations converge to provide a model for meiosis I segregation in females that has been articulated most explicitly by Hawley and colleagues (20, 35). The general features of the model are that the chromo-

10446

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

Review: Orr-Weaver Female

Exchange Chromosomes

Female Nonexchange Chromosomes

functions may be integrally linked to the Male Nonexchange Chromosomes unusual female meiotic spindle. In con-

Distributive Segregation Achiasmate

Achiasmate Homologous Segregation

Other Designations

Heterologous Segregation

(eg.X-*X Bal (eg. XX4- Y A A Dp4 *--4,4 XX 4-*044 DpK *--* XX) DpX *-#-4,4)

Recombination

Representative

Deficient Mutants

Mutants

[FAxsald

3mei-l

mei-13

nod

I

nod Dub

ord

FIG. 3. Summary of Drosophila meiosis I. The classes of homologs segregating in female and male meiosis are indicated together with representative examples of mutations affecting segregation. At the bottom the types of segregation affected in the mutants are indicated by the extent of the boxes. The alternative names for the segregation of nonexchange chromosomes in females are shown, along with some examples of each class of segregation. Homologous chromosomes that fail to exchange because they are in trans to a balancer (Bal) undergo achaismate homologous segregation. Free duplications segregating from homologous normal chromosomes also are classified as achiasmate homologous segregation. The segregation of the X chromosomes from the Y, heterologous compound chromosomes, or free duplications from nonhomologous chromosomes are termed achiasmate heterologous segregation. Note that the ord gene causes meiosis I nondisjunction of all classes of chromosomes because it results in premature separation of the sister chromatids. See the text for references for the mutants. somes pair early in meiosis and form a synaptonemal complex. The fact that there is a karyosome stage gives unique properties to Drosophila female meiosis. Whereas in most organisms the homologs appear to repulse each other during diplotene/diakinesis, these stages do not occur in Drosophila females. It is proposed that the tight packing in the karyosome maintains homolog associations even for the nonexchange bivalents. When the karyosome breaks down and the spindle forms, achiasmate chromosomes might be predicted to dissociate and move to the poles. The NOD motor protein, which is localized along the chromosome arms, could act to push the achiasmate chromosomes away from the poles, forcing them together into the chromosomal mass on the metaphase plate (33). This hypothesis is consistent with the location of the NOD protein and the observation of stray nonexchange chromosomes in nod mutants. It requires that the direction of the NOD motor is toward the plus ends of microtubules, a prediction that has not yet been verified. The model proposes that nonexchange chromosomes that have homology pair early in meiosis. This is certainly the case for the fourth chromosome, as it is observed to have synaptonemal complex (9). Hawley et al. (20) propose that other nonexchange chromosomes can pair through homologous heterochromatin prior to karyosome formation. However, chromosomes that lack apparent homology are capable of accurately segregating from each other in female meiosis. Hawley

advances two hypotheses to account for the segregation of heterologous chromosomes (20). In the first, the chromosomes somehow interact to result in the proper centromere alignment to the two poles, and this association is stabilized by the karyosome. This explanation does not necessarily require that the interaction between the heterologous chromosomes be a physical attachment. The second hypothesis accounts for segregation by the narrow female meiotic spindle. The premise is that the tapered poles have limited accommodation for chromosomes. So the orientation of the heterologous chromosomes could be random within the karyosome prior to spindle formation. Once the spindle is formed the nonexchange chromosomes would begin to orient and move to the pole. If one heterolog had moved toward one pole, that pole would be crowded, thus excluding its partner and resulting in its orientation to the other pole. Male Meiosis In Drosophila males recombination does

trast, the male meiotic spindle resembles a mitotic spindle with a microtubule organizing center at each pole, rather than a tapered spindle organized by the chromosomes as in female meiosis (3, 6). In addition, the chromosomes do not assume a karyosome structure in the male and are condensed and visibly paired throughout the meiosis I division (Fig. 2A). Little is known about the trans-acting genes required for male homolog segregation, and some of the most interesting mutants have been lost (36). However, recently there have been exciting advances in our understanding of the cis-acting elements responsible for homolog pairing. The mechanism by which homologous chromosomes find each other and pair during meiosis is not understood. Drosophila males provide the advantage that specific sequences promote pairing, thus giving a defined focal point from which to elucidate the molecular basis of pairing. Moreover, the ability to visualize pairing cytologically in testis squashes permits pairing to be addressed independently of segregation (Fig. 2A). Cooper (37) demonstrated that the X and Y chromosomes pair at a specific site, termed the collochore. He used deletions to map the position of this site to the base of the long arm of the X chromosome and the short arm of the Ychromosome. In the past 5 years McKee and colleagues (38, 39) have delineated the basis of X-Y pairing to a molecular level. The repeated rDNA genes are present on the X and Y chromosomes in the vicinity of the collochore defined by deletions. McKee and Karpen (38) found that a single copy of the rDNA repeat inserted on the X chromosome is capable of restoring X-Y pairing and segregation to an X chromosome deleted for most of the X heterochromatin, including the collochore. By generating a set of deletions within the rDNA repeat and measuring pairing and segregation, a 240-bp repeated sequence within the intergenic region was found to be sufficient for pairing and segregation (39). Increased copies of this repeat enhance X-Y chromosome interaction. Interestingly, insertion of the rDNA repeat onto an autosome does not promote pairing with the Y chromosome (38). Thus, the chromosomal domain in which the rDNA repeat is inserted must also influence ho-

not occur and synaptonemal complex is molog pairing. not detectable (4, 5). Nevertheless, these These experiments establish that the sex nonexchange homologs pair and segregate chromosomes interact via a specific pair-

faithfully during meiosis I. Mutations in the genes affecting the segregation of nonexchange chromosomes in the female have no effect on male meiosis (Fig. 3, with the exceptions of Dub and ord discussed below); thus, distributive segregation does not account for the mechanism of male meiosis I. The distributive segregation

ing site, a block of homology present on theX and Ychromosomes. However, there are unique properties of the rDNA sequence responsible for this interaction. Other blocks of homology shared between the X and Y chromosomes such as the Stellate (Ste) repeats are not sufficient for proper segregation (38). One possibility

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

Review: Orr-Weaver for the special characteristics of the rDNA

10447

That fact that homolog pairing in male chromatid cohesion early versus later in

repeat is that the presence of this se- meiosis requires homology, even for the meiosis (43). Early in meiosis I the sister quence within a nucleolus somehow con- sex chromosomes, suggests that some chromatids are held together along their

fers pairing ability. This explanation is unlikely, since there are deletions within the rDNA that prevent detectable nucleolus formation yet still permit proper X-Y segregation (39). Chromosomal regions needed for autosomal pairing and segregation in Drosophila males have also been investigated. In the most extensive study, McKee et al. (40) used a set of transpositions in which segments of the second chromosome are inserted onto the Y chromosome. They examined cytologically whether these chromosomal segments caused the Y chromosome to pair with the second chromosome. Using genetic assays, they also tested whether the transposed chromosome 2 segment could direct the Y chromosome carrying it to segregate from the homologous chromosome 2 The results from these experiments contrast with the simple, specific pairing site used in segregation of the sex bivalent. Consistent with previous reports, McKee et al. (40) found that the heterochromatin of the second chromosome does not promote pairing and segregation. While the euchromatin does act as a pairing region, its ability to do so is a function of the size of the euchromatic block. Thus, increasingly larger transpositions were found to be more effective in pairing and segregating with chromosome 2. These observations indicate that either there are not specific pairing sites on the autosomes or they are distributed very frequently along the chromosome. Although many regions along the second chromosome confer pairing ability, there is an exceptional region at the base of the left arm of the second chromosome that is particularly effective at promoting segregation (40). By deficiency mapping, this autosomal pairing site has been shown to colocalize with the histone gene cluster (B. McKee, personal communication). In Drosophila, the histone genes are repeated tandemly in 100-200 copies (41). Thus, this repeated gene cluster appears to play a role in autosomal pairing analogous to that of the rDNA cluster in sex chromosome pairing. The distinction is that this is not the sole sequence that promotes pairing of the second chromosomes. It is unclear why these two tandemly repeated sequences function as pairing sites whereas the highly repetitive DNA found in heterochromatin does not. The different chromatin configuration in heterochromatin may impede pairing. The ability of tandem repeats of specific sequences to promote pairing in males could reflect the increased density of sequence homology or, more likely, the properties of specific chromatin proteins bound to these repeats.

functions promoting homolog pairing could be shared between males and females, a point emphasized by McKee (40). Such common functions have not been described. However, the previous screens for meiotic mutations were not saturating, and these genes may not have yet been isolated. A recently identified gene Dub (Double or nothing), is intriguing because, to my knowledge, it is the first mutation that affects homolog segregation in meiosis I of males and females (Fig. 3) (42). This suggests that the Dub gene could provide a common pairing function in both sexes, or it could be involved in another aspect of homolog partitioning used in both sexes. The Dub mutation is a conditional dominant allele that leads to predominantly meiosis I nondisjunction in males. In mutant females exchange chromosomes nondisjoin, but nonexchange chromosomes are affected most severely. Nonexchange chromosomes nondisjoin during meiosis I in Dub mutant females, whether they are homologous or heterologous. The Dub mutation also is a recessive conditional lethal, producing phenotypes consistent with extensive cell death or mitotic abnormalities. The possible effect of Dub on mitosis raises the potential for the Dub gene product to have a general function in segregation in all divisions. Further interpretation of the role of the Dub gene in male and female meiosis requires analysis of the loss-of-function phenotypes. Sister-Chromatid Cohesion The cytological techniques available to Drosophila researchers have permitted investigation of a critical aspect of chromosome segregation that has remained elusive in most other systems: the functions that hold replicated sister chromatids together until the metaphase/anaphase transition (43). Mutations in genes encoding proteins required to hold sister chromatids together would cause premature separation of the sister chromatids in meiosis or mitosis, depending on which division they affected. Premature separation of the sister chromatids is identified most definitively if it can be directly visualized (Fig. 2D). Two strong candidates for proteins having a primary role in maintaining sister-chromatid cohesion in meiosis are the Drosophila ord and mei-S332 genes. Mutations in these genes cause aberrant meiotic chromosome segregation in females and males, and they have been observed to cause premature sisterchromatid separation in male meiosis (4448). Premature sister separation occurs at different times in the two mutants. Cytology suggests different control of sister-

entire length, as they are in mitosis. This cohesion along the chromatid arms has been postulated to stabilize chiasmata in organisms with chiasmate meiosis (14). At the metaphase I/anaphase I transition cohesion along the arms of the sister chromatids is lost, and they remain attached only at the centromere regions until the metaphase II/anaphase II transition (Fig. 2F). Thus, based on the cytological observations, either there are different functions controlling sister-chromatid cohesion early and late in meiosis or the release of cohesion is spatially controlled so that it persists in the centromere region until anaphase II. Even in apparent null mutations of meiS332 sister-chromatid cohesion is normal until late in anaphase I (45, 46). This observation suggests that the MEI-S332 protein might act specifically to maintain cohesion in the centromere regions of the chromosomes (Fig. 2 D and E) (46). In contrast to meiosis, sister-chromatid cohesion is released in one step in mitosis at the metaphase/anaphase transition. Mutations in the mei-S332 gene have no effect on mitosis, and the gene is transcribed strongly during developmental stages when meiosis is occurring but not during the larval stages when mitosis occurs (49). These results argue that mei-S332 plays a specific role in meiosis to hold sister chromatids together at their centromere regions. The mei-S332 gene has been shown recently to encode a novel 44-kDa protein that is highly charged (49). The MEI-S332 protein was localized by fusing it to the green fluorescent protein from jellyfish (49). The MEI-S332-GFP fusion protein is fully functional, because it rescues meiS332 mutants. MEI-S332-GFP localizes to the centromeric region of meiotic chromosomes from prophase I until metaphase II. Strikingly, as the sister chromatids separate in anaphase II and cohesion is lost, MEI-S332-GFP is no longer detectable on the chromosomes. These results indicate that MEI-S332 binds to centromere regions and holds sister chromatids together and that it must dissociate to

permit segregation. In contrast to the specific effect of mei-S332 on meiosis, mutations in the ord gene affect several aspects of meiosis as well as mitosis (47-50). Premature sisterchromatid separation is observed in male meiosis as early as prometaphase I (45, 48). Null alleles give missegregation consistent with ratios predicted from premature separation and random segregation of the four sister chromatids through both meiotic divisions (48). This is true for male and female meiosis. Thus, it appears that the ord gene is necessary to maintain sister-chromatid cohesion beginning early

10448

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

Review: Orr-Weaver

in meiosis I when the sister chromatids are held along their lengths. One prediction is that the ORD protein may be localized on the chromosome arms as well as at the centromere regions. In addition to loss of sister-chromatid cohesion, most mutations in the ord gene cause a reduction of recombination in female meiosis, and the exchange that does occur is not sufficient to ensure proper segregation (47, 48). This is consistent with the prediction that premature loss of sister-chromatid cohesion would result in loss of chiasmata. ord may have several independent functions in meiosis, or the effects on recombination and cohesion may be related. Moreover, strong ord mutations result in mitotic nondisjunction in the male germline (48, 50). However, these alleles show almost no effect on somatic mitotic divisions, so the requirement for the gene may be restricted to divisions in the germ line (J. Wu, W. Miyazaki, and T.L.O.W., unpublished results). The ord gene has been isolated recently (S. Bickel, D. Wyman, W. Miyazaki, D. Moore, and T.L.O.-W., unpublished results). There are different forms of the ord transcript in testis and ovaries. A common open reading frame shared by the testis transcripts predicts a 55-kDa protein. The six ord mutations cause either stop codons or missense changes within the C-terminal half of this open reading frame. Since these affect female as well as male meiosis, at least the C-terminal half of the protein is common to both sexes. It remains possible that there may be female or mitotic forms of the protein with differing N termini. These potentially different forms of the protein could account for the pleiotropic effects of ord mutations. Centromere Structure

Ultimately it will be necessary to determine how segregation proteins interact with the chromosomes to partition them correctly during meiosis. In particular it is essential to identify the proteins that act at the centromere. The prerequisite for achieving this goal is to define the centromere at a molecular level. Drosophila has yielded a recent important advance in this area. The centromeres of the chromosomes in higher eukaryotes are complex, and they have eluded molecular characterization because they are embedded in heterochromatin. By studying the structure and function of a minichromosome, Karpen and colleagues (51, 52) have been able to localize a centromere. The free duplication Dp187 is a deletion derivative of the X chromosome that is only 1.3 Mb in size and is stably transmitted (51, 52). Deletions within the 1 Mb of heterochromatin on this minichromosome were generated, mapped, and tested for their effects on transmission (52, 53). The genetic assay for transmission re-

quires stability in germ-line mitotic divisions, meiosis, and somatic mitotic divisions. This analysis defined a region of 220 kb essential for normal transmission of the minichromosome (52). This interval is an island of complex DNA that is single copy or moderately repetitive DNA. Completely normal transmission requires an additional 200 kb of satellite DNA on either side of the essential region. Interestingly, transmission of the minichromosome through females is considerably more sensitive to the effect of deletions than is transmission in males. The delineation of a centromere will permit a comparison of cis-acting requirements for meiosis and mitosis. It will be interesting to determine whether different centromere functions are needed during the first meiotic division. These minichromosome deletions may make it possible to map the sites of MEI-S332 action and to determine whether MEI-S332 localizes to the kinetochore or flanking centric heterochromatin. Cell Cycle Regulation of Meiosis A complete understanding of meiosis requires that we know the mechanisms by which chromosomes pair and segregate but also that we elucidate the regulation of the steps of meiosis. Meiosis can be viewed as a variant of the normal mitotic cycle in which an extra, specialized division with homolog pairing and segregation is inserted after DNA replication. In addition, in multicellular organisms, female meiosis is arrested during oogenesis to permit growth and differentiation of the oocyte. Developmental signals subsequently trigger the completion of meiosis. In the case of Drosophila, the oocyte is arrested at metaphase I and the signal to complete meiosis involves movement into the uterus, although the molecular basis of this activation is unknown. The identification of the cell cycle regulators of meiosis is still preliminary, but recently several control genes have been isolated in

Drosophila. Several meiotic cell cycle control genes have been identified because they also have a role in the mitotic cell cycle. The twine gene was isolated as a homolog of string (stg), a mitotic cell cycle regulator, and it is needed for early steps in the meiotic cell cycle in males (54, 55). twine encodes a tyrosine phosphatase. twine and string are homologs of the cdc25 gene of Schizosaccharomyces pombe, a positive activator of cdc2. The germ-line mitotic divisions take place in twine mutant males, but meiosis does not occur (54, 55). Although no spindles are formed, some aspects of meiosis such as chromosome condensation are observed in twine mutants (56). The coordination of the two meiotic divisions in males requires the proper dosage of the product of the roughex (rux)

gene (57). In loss-of-function mutations of roughex the two meiotic divisions occur, but they are followed by another inappropriate division. Overexpression of roughex blocks the meiosis II division. Reducing the dosage of Cyclin A (CycA) or twine can suppress the roughex phenotype; thus, it appears that roughex acts through Cyclin A/cdc2 to control the second meiotic division. The effect of roughex mutations on female meiosis has not been investigated. The meiotic arrest of oocytes is signaled in part by exchange events (58). In recombination-deficient mutants, mature oocytes are not arrested at metaphase I but are observed in anaphase I, metaphase II, or anaphase II. An exchange event on one chromosome pair is sufficient to cause arrest, so it is not simply that chromosomes are physically constrained to the metaphase plate by chiasmata. However, it is not the exchange itself that causes arrest. Jang et aL (59) produced an "all-compound" strain in which the two homologs of each of the major chromosomes were attached to the same kinetochore. Although these compound chromosomes undergo exchange, there is no metaphase I arrest. Thus, it appears that the tension resulting from an exchange (and presumably chiasma) between two chromosomes with separate kinetochores sends a signal for meiosis arrest (59). In twine mutant females meiosis initiates but fails to arrest at metaphase I, suggesting twine may be involved in the signaling process (55, 56). twine mutant oocytes undergo a series of aberrant nuclear divisions and DNA replication. The Drosophila oocyte is activated to complete meiosis by movement into the uterus, and fertilization is not required for this process. The molecular basis of activation remains a mystery, although the oocyte becomes hydrated upon oviposition. Moreover, the completion of meiosis can be efficiently activated in vitro by swelling in hypotonic buffers (Fig. 1 B and C). Eggs from mothers mutant for the gene grauzone or cortex fail to complete meiosis and arrest at metaphase II (A. Page and T.L.O.-W., unpublished results). These mutations may disrupt the signals for the completion of meiosis. Future Directions

The cytological techniques and genetics available in Drosophila have permitted great strides in our understanding of homolog pairing and segregation in meiosis I and mechanisms of sister-chromatid cohesion. Other important proteins will be identified by cloning genes known to be essential for meiosis. Insights into the function of these proteins will result from analysis of their localization on meiotic chromosomes or spindles. Furthermore, antibodies to these proteins will facilitate investigation of the interaction and rela-

Review: Orr-Weaver

tionships between meiotic proteins by examining protein behavior in mutants. The foundation for the isolation of proteins acting at the kinetochore has been laid by the delineation of the centromere. In addition to these advances in our understanding of meiotic mechanisms, given the pace of the mitotic cell cycle field, it is anticipated that we will soon have a clearer image of the control of the meiotic cell cycle. I thank S. Endow, R. S. Hawley, and G. Karpen for providing reprints and preprints. I am grateful to B. McKee and W. Theurkauf for permission to cite unpublished results. A. Kerrebrock, W. Miyazaki, A. Page, and W. Theurkauf kindly provided cytology photographs, and H. Protzman prepared the figures. G. Karpen, B. Reed, A. Page, S. Bickel, D. Moore, and A. Kerrebrock critically read the manuscript. This work was supported by National Science Foundation Grant MCB 9316168. 1. Cooper, K. (1950) in Biology of Drosophila, eds. Demerec, M. (Wiley, New York), pp. 1-61. 2. Goldstein, L. S. B. (1981) Cell 25, 591602. 3. Lin, H.-P., Ault, J. G. & Church, K. (1981) Chromosoma 83, 507-521. 4. Meyer, G. F. (1960) in Proceedings of the European Regional Conference on Electron Microscopy, eds. Houwink, A. L. & Spit, B. J. (Nederlandse Vereniging voor Electronenmicroscopie, Delft, The Netherlands), pp. 951-954. 5. Rasmussen, S. W. (1973) Z. Zellforsch. 140, 125-144. 6. Cenci, G., Bonaccorsi, S., Pisano, C., Verni, F. & Gatti, M. (1994) J. Cell Sci. 107, 3521-3534. 7. Church, K. & Lin, H. P. P. (1982) J. Cell Biol. 93, 365-373. 8. Church, K. & Lin, H. P. P. (1985) Chromosoma 92, 273-282. 9. Carpenter, A. T. C. (1975) Chromosoma 51, 157-182. 10. Carpenter, A. T. C. (1975) Proc. Natl. Acad. Sci. USA 72, 3186-3189. 11. Carpenter, A. T. C. (1979) Genetics 92, 511-541. 12. Theurkauf, W. & Hawley, R. S. (1992) J. Cell Biol. 116, 1167-1180.

Proc. Natl. Acad. Sci. USA 92 (1995) 13. Baker, B. & Hall, J. (1976) in The Genetics and Biology ofDrosophila, eds. Ashburner, M. & Novitski, E. (Academic, New York), pp. 351-434. 14. Darlington, C. D. (1932) RecentAdvances in Cytology (Churchill, London). 15. John, B. (1990) Meiosis (Cambridge Univ. Press, Cambridge, U.K.). 16. Nicklas, R. B. (1974) Genetics 78, 205213. 17. Grell, R. F. (1962) Genetics 47, 17371754. 18. Grell, R. (1976) in The Genetics and Biology ofDrosophila, eds. Ashburner, M. & Novitski, E. (Academic, New York), pp. 436-486. 19. Hawley, R. S., Irick, H., Zitron, A. E., Haddox, D. A., Lohe, A., New, C., Whitley, M. D., Arbel, T., Jang, J., McKim, K. & Childs, G. (1993) Dev. Genet. 13, 440467. 20. Hawley, R. S., McKim, K. S. & Arbel, T. (1993) Annu. Rev. Genet. 27, 281-317. 21. Carpenter, A. (1973) Genetics 73, 393428. 22. Zhang, P. & Hawley, R. S. (1990) Genetics 125, 115-127. 23. Zhang, P., Knowles, B. A., Goldstein, L. S. B. & Hawley, R. S. (1990) Cell 62, 1053-1062. 24. Rasooly, R. S., New, C. M., Zhang, P., Hawley, R. S. & Baker, B. S. (1991) Genetics 129, 409-422. 25. Rasooly, R. S., Zhang, P., Tibolla, A. K & Hawley, R. S. (1994) Mol. Gen. Genet. 242, 145-151. 26. Davis, D. G. (1969) Genetics 61, 577-594. 27. Endow, S. A., Henikoff, S. & Soler-Niedziela, L. (1990) Nature (London) 345,81-83. 28. McDonald, H. B. & Goldstein, L. S. B.

(1990) Cell 61, 991-1000. 29. McDonald, H. B., Stewart, R. J. & Goldstein, L. S. B. (1990) Cell 63, 1159-1165. 30. Walker, R. A., Salmon, E. D. & Endow, S. A. (1990) Nature (London) 347, 780-782. 31. Hatsumi, M. & Endow, S. (1992) J. Cell Sci. 101, 547-559. 32. Hatsumi, M. & Endow, S. A. (1992)J. Cell Sci. 103, 1013-1020. 33. Afshar, K, Barton, N. R., Hawley, R. S. & Goldstein, L. S. B. (1995) Cell 81, 129-138. 34. Murphy, T. D. & Karpen, G. H. (1995) Cell 81, 139-148. 35. Hawley, R. S. & Theurkauf, W. E. (1993) Trends Genetics 9, 310-317.

10449

36. Lindlsey, D. & Zimm, G. (1992) The Genome of Drosophila melanogaster (Academic, New York). 37. Cooper, K. W. (1964) Proc. Natl. Acad. Sci. USA 52, 1248-1255. 38. McKee, B. & Karpen, G. (1990) Cell 61, 61-72. 39. McKee, B. D., Habera, L. & Vrana, J. A. (1992) Genetics 132, 529-544. 40. McKee, B. D., Lumsden, S. E. & Das, S. (1993) Chromosoma 102, 180-194. 41. Moore, G. D., Sinclair, D. A. & Grigliatti, T. A. (1983) Genetics 105, 327-344. 42. Moore, D. P., Miyazaki, W. Y., Tomkiel, J. E. & Orr-Weaver, T. L. (1994) Genetics 136, 953-964. 43. Miyazaki, W. Y. & Orr-Weaver, T. L. (1994) Annu. Rev. Genet. 28, 167-187. 44. Davis, B. (1971) Mol. Gen. Genet. 113, 251-272. 45. Goldstein, L. S. B. (1980) Chromosoma 78, 79-111. 46. Kerrebrock, A. W., Miyazaki, W. Y., Birnby, D. & Orr-Weaver, T. L. (1992) Genetics 130, 827-841. 47. Mason, J. M. (1976) Genetics 84,545-572. 48. Miyazaki, W. Y. & Orr-Weaver, T. L. (1992) Genetics 132, 1047-1061. 49. Kerrebrock, A. W., Moore, D. P., Wu, J. S. & Orr-Weaver, T. L. (1995) Cell 83, in press. 50. Lin, H. P. & Church, K. (1982) Genetics 102, 751-770. 51. Karpen, G. H. & Spradling, A. C. (1990) Cell 63, 97-107. 52. Murphy, T. D. & Karpen, G. H. (1995) Cell 82, 599-609. 53. Le, M.-H., Duricka, D. & Karpen, G. H. (1995) Genetics 141, 283-303. 54. Alphey, L., Jimenez, J., White-Cooper, H., Dawson, I., Nurse, P. & Glover, D. (1992) Cell 69, 977-988. 55. Courtot, C., Fankhauser, C., Simanis, V. & Lehner, C. F. (1992) Development (Cambridge, U.K) 116, 405-416. 56. White-Cooper, H., Alphey, L. & Glover, D. M. (1993) J. Cell Sci. 106, 1035-1044. 57. Gonczy, P., Thomas, B. J. & DiNardo, S. (1994) Cell 77, 1015-1025. 58. McKim, K. S., Jang, J. K., Theurkauf, W.E. & Hawley, R. S. (1993) Nature (London) 362, 364-366. 59. Jang, J. K, Messina, L., Erdman, M. B., Arbel, T. & Hawley, R. S. (1995) Science 268, 1917-1919.