Proc. Natl. Acad. Sci. USA Vol. 92, pp. 10450-10456, November 1995 Review Sex and the single cell: in yeast G. Shirleen Roeder Department of Biology, Yale University, P.O. Box 208103, New Haven, CT 06520-8103

ABSTRACT Recent studies of Sac- (Fig. 3). Concurrent with these improve- elements is closely associated at a few charomyces cerevisiae have significantly ments in cytology has been the develop- positions (Fig. 1B), postulated to be sites advanced our understanding of the mo- ment of numerous physical assays that at which Zipl and other components of lecular mechanisms of meiotic chromo- detect recombination intermediates and the central region begin to polymerize (3). some behavior. Structural components of products. By combining these cytological In some organisms, subunits called trans- the synaptonemal complex have been and physical techniques with classical ge- verse filaments have been observed to lie identified and studies of mutants defec- netic methods, it has been possible to test perpendicular to the long axis of the SC tive in synapsis have provided insight into canonical theories about the role of the and to bridge the space between lateral the role of the synaptonemal complex in synaptonemal complex (SC) and the rela- elements (6). The Zipl is pre- homolog pairing, genetic recombination, tionship of recombination to homolog dicted to form a rod-shaped dimeric mol- crossover interference, and meiotic chro- pairing and segregation. The ecule and the predicted length of the Zipl mosome segregation. There is compelling often surprising results have necessitated dimer corresponds approximately to the evidence that most or all meiotic recom- abandonment of long-standing theories distance between lateral elements in the bination events initiate with double- and inspired new ones. SC (3). Mutations that increase the length strand breaks. Several intermediates in of the Zipl dimer lead to corresponding the double-strand break repair pathway SC Structure increases in the distance between lateral have been characterized and mutants elements, suggesting that Zipl is a com- blocked at different steps in the pathway The pairing of homologous ponent of transverse filaments (9). have been identified. With the application during meiotic prophase culminates in the The RED] and HOP] encode ofgenetic, molecular, cytological, and bio- formation of the SC, which is a ribbon- associated with the lateral ele- chemical methods in a single organism, like, proteinaceous structure that holds ments of the SC and with unsynapsed axial we can expect an increasingly comprehen- homologous chromosomes in close appo- elements. The Redl and Hopl proteins sive and unified view of the meiotic pro- sition along their entire lengths (6). Yeast are not always localized continuously cess. SCs can be visualized clearly when meiotic along the length of the SC; thus, there is nuclei are surface spread, stained with not a perfect correspondence between the The budding yeast Saccharomnyces cerevi- silver nitrate, and viewed in the electron appearance of lateral elements as defined siae is an ideal organism for studies of microscope (Fig. IA). Early in the path- by silver staining and the pattern of Red 1/ meiosis for a multiplicity of reasons. When way of SC assembly, each pair of sister Hopl deposition (Hopl: ref. 10, F. Klein diploid yeast cells are starved for carbon chromatids develops a common protein- and B. Byers, personal communication; and nitrogen, almost all cells in the pop- aceous core called an axial element. Con- Redl: A. Smith and G.S.R., unpublished ulation enter meiosis, and they proceed current with the development of axial data). The Redl protein is required for through the meiotic divisions in a fairly elements, the formation of mature SC the formation of axial elements (11) and synchronous manner. This efficiency and initiates at a few sites along each chromo- may be involved in nucleating the forma- synchrony permits temporal analyses of some pair (7). At these positions, the tion of these protein cores. Unsynapsed the meiotic process using biochemical, protein components of the central region axial elements assemble in the absence of molecular, and cytological assays. Thanks of the SC assemble between the axial the Hopl protein (F. Klein and B. Byers, to sophisticated genetics and a powerful elements, which then become the lateral personal communication) and genetic transformation system, yeast meiotic mu- elements of the SC. Bidirectional SC ex- (12) and physical (13) analyses indicate tants are easy to isolate, and the corre- tension results in complete synapsis. Most that Hopl plays an important role in sponding genes can be rapidly cloned and DNA is located outside the SC and is promoting interhomolog interactions. disrupted. Artificial chromosomes and organized into chromatin loops that em- Perhaps Hopl localizes to preassembled synthetic constructs integrated into au- anate from the lateral elements. The dis- axial elements and subsequently promotes thentic chromosomes provide invaluable tance between lateral elements is 100 chromosome synapsis. Overexpression of tools in studies of recombination, ho- nm. In yeast, the average chromatin loop the RED] suppresses a subset of molog pairing, and chromosome segrega- is 500 nm in length and contains 20 kbp of mutant alleles of the HOP] gene, suggest- tion. Last, but not least, the ability to DNA (8). ing that the Redl and Hopl proteins recover and analyze all four products of The best-characterized structural-com- physically interact with each other (14, individual meioses allows detailed inves- ponent of the yeast SC is the ZipI protein. 15). tigation of the mechanisms of meiotic Zipl is localized continuously along the recombination. lengths of mature SCs (Fig. 2B) but is not Searching and Studies of yeast meiosis have benefited present in unsynapsed axial elements, in- Presynaptic Alignment from recent improvements in cytology. dicating that Zipl is a component of the Despite the small size of yeast chromo- central region (3). In the zip] null mutant, Chromosome synapsis is preceded by a somes, it is now possible to visualize mei- axial elements are full-length and homolo- homology search that results in the side- otic chromosomes in nuclear spreads (Fig. gously paired, but the distance between by-side alignment of homologs at a dis- 1), to localize proteins to chromosomes them is greater and more variable than the tance that exceeds the width of the SC (6). (Fig. 2), and to assess chromosome pairing distance between lateral elements in ma- In yeast, this presynaptic alignment has by fluorescent in situ hybridization (FISH) ture SC (Fig. 1B). Each pair of axial been visualized by FISH (Fig. 3) using 10450 Downloaded by guest on October 2, 2021 Review: Roeder Proc. Natl. Acad. Sci. USA 92 (1995) 10451

The timing of these ectopic recombination A i events (22) suggests that dispersed recom- bining sequences are physically associated throughout the period when homologous chromosomes are fully synapsed. Such ec- topic interactions may take place in chro- ...... i ..- matin loops that are distant from the protein cores engaged in SC formation. Double-Strand Cleavage Initiates Meiotic Recombination There is compelling evidence that double- strand breaks (DSBs) initiate meiotic re- combination in yeast. Meiotically induced DSBs have been observed at a number of recombination hot spots that display ele- vated levels of genetic exchange and cis- acting mutations that reduce DSBs at these sites decrease the frequency of re- combination (23-26). Analysis of whole meiotic chromosomes by pulsed-field gel electrophoresis has demonstrated that DSBs are widespread, but there are a number of preferred sites of cleavage on each chromosome (27, 28). DSBs appear and disappear with the kinetics expected for an early intermediate in the exchange process (7). Meiotic DSBs do not occur at a specific DNA sequence but rather are dispersed throughout a region of "150- 180 bp at each locus (89-91). Comparison of the yeast genetic map with the distri- FIG. 1. Electron micrographs of yeast meiotic chromosomes Meiotic nuclei from wild type (A) bution and frequency of double-strand and the zip] mutant (B) were surface spread and stained with silver nitrate (1). (Bar = gLm; cuts suggests that DSBs initiate most, if micrographs provided by Mary Sym.) not all, meiotic recombination events (29). Features of chromatin structure estab- chromosome-specific DNA probes (5, 16). volves interactions at multiple sites along lished during mitotic growth play an im- A significant, albeit reduced, level of mei- each chromosome pair (16). The total num- portant role in determining the sites of otic homolog pairing is observed in mu- ber of interactions per nucleus ('190) is meiotic DSBs (29). Sites with a high prob- tants defective in SC formation (4, 16-18). similar to the number of meiotic recombi- ability of breakage during meiosis corre- The problem of homolog alignment is nation events, leading to the hypothesis that spond to nucleosome-free regions (as de- generally assumed to be specific to meiotic meiotic recombination initiates at the sites fined by nuclease hypersensitivity) present cells; however, recent FISH studies indi- of early pairing (16, 19). The notion that in chromatin isolated from vegetative cells cate that homologs are paired in a sub- pairing connections precede (and therefore (29). In addition, cis- and trans-acting mu- stantial fraction of yeast cells prior to the possibly promote) recombination, and not tations that lead to chromatin remodeling at induction of meiosis (4, 16, 19). Changes vice versa, is supported by the following a specific locus alter the frequency and in mitotic chromatin structure that de- observations. (i) Pairing is observed in veg- distribution of DSBs (29). Almost all breaks pend on homozygosity of the affected etative cells in which the DNA presumably occur in intergenic regions that contain sequences provide additional evidence contains no strand interruptions (16). (ii) transcription promoters (29). However, for premeiotic interactions between ho- Some meiotically induced pairing is ob- transcription per se is not required for DSB mologs (S. Keeney and N. Kleckner, per- served in mutants that fail to initiate recom- formation (30) and transcription through sonal communication). The vegetative bination (4, 16-18). (iii) In a particular the promoter can interfere with recombina- pairing detected by FISH is temporarily mutant, the number of pairing connections tion hot spot activity (31). Two observations disrupted during premeiotic DNA repli- exceeds the number of exchanges initiated suggest that an open chromatin configura- cation and then reestablished (16). It has (16). tion is not the only determinant of DSB been proposed that premeiotic and mei- Surprisingly, meiotic recombination is formation (32). (i) There is not a strong otic pairing involve the formation of un- not confined to sequences at the same correlation between the level of nuclease stable side-by-side joints between intact position on homologous chromosomes. hypersensitivity and the probability of cleav- DNA duplexes (16, 19). Such reversible Two copies of a gene positioned on non- age. (ii) DSB formation can be affected by associations would provide a mechanism homologous chromosomes recombine al- sequences located several kilobase pairs to deal with the interchromosomal tangles most as frequently as allelic genes (20, 21). away. that result when uncondensed chromo- These ectopic recombination events imply Despite extensive studies of DSBs, the somes initiate pairing at multiple sites. As the operation of a genome-wide homology endonuclease(s) responsible for cleavage meiosis progresses, SC formation estab- search that allows even short stretches of remains elusive. Genetic analysis indicates lishes stable interhomolog connections DNA (-2 kbp) to find any and all homol- that several yeast mutants are defective in that can withstand the forces of meiotic ogous counterparts. Interactions between the initiation of meiotic recombination chromatin condensation. sequences on nonhomologous chromo- (33); at least five of these (rad5O, spoll, FISH using short DNA probes has pro- somes are evidently not excluded either by xrs2, mrel , and mer2) have been shown to vided evidence that premeiotic pairing in- premeiotic pairing or by SC formation. be defective in DSB formation by physical Downloaded by guest on October 2, 2021 10452 Review: Roeder Proc. Natl. Acad. Sci. USA 92 (1995)

called radSOS, has proved to be an invalu- different yeast mutants in which recombi- able tool in studies of the formation and nation events initiate at the wild-type distribution of DSBs (27-29). level, but crossing-over is reduced 2- to In three mutants, radSl (40), dmcl (41), 10-fold (2,49,50). One of these mutants is and sepl (42), DSBs are processed to zipl, suggesting that the SC can influence expose single-stranded tails with 3' ter- the direction of resolution (50). Two other mini (Fig. 4), but these tails accumulate mutants, msh4 (2) and (92), define and eventually become longer than their genes that encode homologs of the bacte- wild-type counterparts. The DMC1 (41) rial MutS protein, which is involved in the and RAD51 (40) genes encode homologs correction of mismatched base pairs. of the bacterial RecA strand exchange Since Msh4 and Msh5 play no role in enzyme. Like RecA (43), the Rad5l pro- mismatch repair, these homologs must tein coats single-stranded DNA to form a have evolved to acquire a new function, nucleoprotein filament (44) and subse- which may include interacting with Holli- quently promotes invasion of the single day junctions. strand into a homologous duplex (45). The Sepl protein, although unrelated to Recombination Nodules RecA, has also been shown to promote strand transfer in vitro (46). Thus, it is In a number of organisms, electron-dense reasonable to suppose that the accumula- structures called recombination nodules tion of DSBs in some or all of these have been observed in association with mutants results from a defect in the first meiotic chromosomes (51). Nodules are step in the repair of DSBs (i.e., strand classified as early if they are present in invasion). zygotene, when synapsis initiates. Soon after DSBs disappear, branched Nodules molecules that contain information from present during pachytene, when homologs both recombining chromosomes can be are fully synapsed, are classified as late. detected by two-dimensional gel electro- The number and distribution of late nod- phoresis (13, 47). These joint molecules ules correlate with those of crossovers, contain two full-length nonrecombinant leading to the hypothesis that these struc- strands from each parental duplex (13), but tures are multienzyme complexes that cat- digestion with a -cleaving alyze reciprocal exchange. Early nodules FIG. 2. Immunolocalization of the Zipl and are more abundant and have been postu- Msh4 proteins. Shown is a spread meiotic nu- enzyme generates crossover products (A. Schwacha and N. Kleckner, personal com- lated to mark the sites of all strand ex- cleus from a strain producing an epitope-tagged change reactions. The version of the Msh4 protein. Chromosomes munication). Thus, these molecules appear Dmcl and Rad5l were stained with a DNA-binding dye (4',6- to contain two Holliday junctions, as pre- proteins colocalize to discrete spots on diamidino-2-phenylindole) (A), anti-Zipl anti- dicted by the DSB repair model (Fig. 4fand chromosomes prior to synapsis and may bodies (B), and antibodies against an epitope- g). therefore be components of early nodules tagged Msh4 protein (C) (2, 3). Anti-Zipl and If strand exchange occurs in a region (52). Rad5l has been shown to associate anti-epitope antibodies were detected with ap- where the parental chromosomes are ge- physically with Rad52 (40), suggesting that propriate secondary antibodies tagged with flu- netically nonidentical, heteroduplex DNA the latter protein is also a component of orescein and Cy3, respectively. (Bar = 2 ,um; early nodules. Msh4 localizes to discrete photographs provided by Petra Ross-Mac- (hDNA) containing one or more mis- donald.) matched base pairs results. Unexpectedly, physical assays fail to detect hDNA until assays (18, 23, 34-36). However, it re- relatively late in meiotic prophase, around mains to be determined whether the im- the same time as mature crossover prod- plicated gene products participate directly ucts (17, 48). According to the DSB repair in DSB formation or whether they estab- model (Fig. 4), however, hDNA should be lish preconditions necessary for DSB in- produced early in the recombination pro- duction. The Rad5O protein has DNA- cess and certainly should be present in binding activity (37) and acts in conjunc- joint molecules. Thus, it is necessary either tion with Mrell in a complex that also to abandon the model or to postulate that includes Xrs2 hDNA is present at earlier times but not (36). detected in current assays. hDNA present DSB and in recombination intermediates may be Processing lost due to branch migration during DNA Recombination Intermediates extraction or it might be contained in branched molecules that fail to migrate as FIG. 3. FISH analysis of chromosome pair- The occurrence of meiosis-specific DSBs a single species during gel electrophoresis provides strong support for the DSB re- ing. Spread meiotic nuclei were painted (4, 5) (48). with composite probes specific for chromosome pair model of recombination (38, 39), as According to the DSB repair model, I (red) and chromosome III (green). DNA was diagramed in Fig. 4. Many intermediates resolution is a stochastic process and each stained blue to reveal spread chromatin. (A) postulated by the DSB repair model have recombination intermediate has an equal Nucleus in which chromosomes I and III are been demonstrated physically and mu- probability of being resolved to generate homologously paired. (B) Nucleus in which tants blocked at different steps in the either the crossover or the noncrossover both sets of homologs are unpaired. (C) Nu- repair process have been identified (Fig. configuration of flanking markers (Fig. 4 cleus in which two nonhomologous chromo- A non-null allele of the somes are accidentally associated. (D) Nucleus 4). RAD50 gene h and i). However, the behavior of certain in which chromosomes III are tightly aligned, allows the formation of DSBs, but the meiotic mutants suggests that resolution while chromosomes I are unpaired. (Bar = 5 broken molecules are not processed and in favor of crossing-over specifically re- ,um; photographs provided by Harry Scher- therefore accumulate (34). This allele, quires additional proteins. There are four than.) Downloaded by guest on October 2, 2021 Review: Roeder Proc. Natl. Acad. Sci. USA 92 (1995) 10453

rad5O, spoil xrs2, mer2, mrell is required for meiotic recombination. + Gene conversion occurs at the wild-type (a) level in zipl strains (3, 50), suggesting that the initiation of recombination is unaf- fected, despite the absence of mature SC. rad5OS In the redl mutant, in which there is no SC 5' 3' or any obvious SC precursors, recombina- (b) 3' 5, tion occurs at -25% of the wild-type rate 3' 5' (11), which is orders of magnitude above 5' 3' the mitotic background level. The ob- ; rad5l, dmcl, sepi served high frequencies of ectopic recom- bination also argue that fully synapsed (c) chromosomes are not an exclusive venue for recombination (20, 21). Finally, DSBs are observed during meiosis in haploid yeast, demonstrating that the initiation of recombination does not depend on prior (d) - interactions between homologs (53, 54). A number of observations suggest that synapsis may depend on DSB processing. DSBs with single-stranded tails appear prior to the initiation of synapsis and their (e) , processing to a later intermediate appears to be coincident with SC initiation (7). The radSOS mutation prevents the pro- cessing of DSBs to expose single-stranded tails (34) and also abolishes (34) or sub- (f) stantially reduces (4) SC formation. Of the many yeast mutants characterized, none has been found to make SC in the absence Sm As As Sm of DSBs. Finally, a recent study suggests ; msh2, pms1, that synapsis initiates at the sites of DSB processing defined by Dmcl- and Rad51-

(g) %, staining foci (93). The Dmcl and Rad5l .=X.*-- proteins are required to establish the in- terhomolog connections observed in zipl strains (Fig. 1B) and the number of these %%%ip zip1 , msh4, msh5, merl+MER2-2g connections (3) corresponds to the num- ber of Dmcl- and Rad51-staining foci (52). SC formation in dmcl and rad5l mutants is delayed and incomplete, as expected if the observed interhomolog (h) noncrossover (i) crossover connections normally serve as sites of synaptic initiation (41, 93). FIG. 4. DSB repair model of recombination (38, 39). (a) One of the two recombining DNA duplexes sustains a DSB. (b) The 5' ends of the broken duplex are exonucleolytically degraded Mismatch Correction to expose single-stranded tails with 3' termini. (c) One of the tails invades an uncut homologous duplex, resulting in the displacement of a D loop. (d) The D loop is enlarged by repair synthesis The machinery responsible for correction primed by the invading 3' end until the other single-stranded tail can anneal with complementary of mismatches during meiotic recombina- sequences in the D loop. (e) DNA synthesis primed by the second 3' end completes repair. (f) Branch migration results in the formation of two Holliday junctions; regions of asymmetric (As) tion in yeast is closely related to the Esch- and symmetric (Sm) strand exchange are indicated. (g) A mismatched , diagramed in d-f, erichia coli machinery involved in the re- is repaired by excision and resynthesis. Mismatch repair might occur earlier than diagramed (i.e., pair of replication errors. Components of immediately upon formation of hDNA). (h) If both Holliday junctions are resolved in the same the bacterial mismatch repair system in- direction (i.e., by cleavage of inner or outer strands in both cases), then the parental configuration clude the MutS protein, which binds to of flanking markers is preserved. (i) Resolution of the two Holliday junctions in opposite mismatched base pairs, and MutL, which directions results in a reciprocal crossover between markers that flank the region of strand interacts with MutS to expand the DNA exchange. For the Holliday junction shown on the right in g, the cuts indicated by the horizontal footprint (55). In yeast, genes encoding six or vertical arrows generate the products shown in h or i, respectively. Mutants blocked at particular steps are indicated. MutS (refs. 2, 56, 57, and 92; GenBank data base) and three MutL (58, 59) ho- foci on synapsed chromosomes (2) and is stranded tails appear early in prophase, mologs have been identified. Msh2 is the thus a candidate for a late nodule compo- prior to the formation of tripartite SC, and functional counterpart of bacterial MutS nent (Fig. 2C). they disappear in zygotene (7). Joint mol- (60). In contrast to E. coli, two different ecules are evident during pachytene (13). MutL proteins, Pmsl and Mlhl, are re- Relationship Between Recombination Crossover products and hDNA are de- quired for mismatch repair in yeast (58, and Chromosome Synapsis tected at the end of pachytene or soon 59). The Pmsl and Mlhl proteins physi- thereafter, as the SC disassembles (7, 17, cally associate with each other and the Temporal studies have demonstrated that 48). resulting Pmsl/Mlhl complex then inter- meiotic recombination and SC assembly Studies of yeast have challenged the acts with Msh2 protein that is bound to proceed concurrently. DSBs with single- traditional view that chromosome synapsis hDNA (61). Downloaded by guest on October 2, 2021 10454 Review: Roeder Proc. Natl. Acad. Sci. USA 92 (1995) According to the DSB repair model from opposite spindle poles or (at least undergoes at least one reciprocal ex- (Fig. 4), the invasion of single-stranded transiently) to fibers from the same pole. change. These two phenomena, referred tails results in hybrid DNA on one of the Spindle fiber attachment is stabilized only to as crossover interference and obligate two recombining DNA duplexes (i.e., when homologs become attached to op- chiasma, may be mechanistically related. asymmetric strand exchange), while the posite spindle poles and the opposing By preventing excess crossovers on large branch migration of Holliday junctions poleward forces exerted on the chromo- chromosomes, interference might ensure results in hybrid DNA on both duplexes somes are resisted by chiasmata. Chiasma that small chromosomes cross over (74, (i.e., symmetric strand transfer). In yeast, function is thought to depend on sister- 75). most non-Mendelian segregations are chromatid cohesion distal to crossovers, Crossover interference has been postu- gene conversions (i.e., 6:2 segregations) on chiasma-binding proteins positioned at lated to involve the transmission of an (33), and cis- and trans-acting mutations crossovers, or both. inhibitory signal from one crossover site to that inhibit mismatch repair result in a The SC has been postulated to play a nearby potential sites of crossing-over predominance of 5:3 segregations, indic- role in meiosis I . (75-77). The SC is an obvious conduit ative of a single heteroduplex (60, 62, 63). Remnants of SC have been observed at along which such a signal might travel. In Thus, most hDNA appears to be confined chiasmata, suggesting that SC proteins support of this hypothesis, the zipl muta- to the single-stranded tails that flank may bind chiasmata (6). In addition, stud- tion abolishes crossover interference, re- DSBs, and mismatch repair is extremely ies of maize have suggested that synapsis sulting in a random distribution of cross- efficient. is necessary to establish meiotic sister- overs along chromosomes (50). Small A number of yeast genes display a po- chromatid cohesion (68). In the zipl mu- chromosomes in the zipl mutant fre- larity gradient in which the frequency of tant, meiosis I nondisjunction events are quently fail to recombine and therefore gene conversion is high at one end and low confined to nonrecombinant chromo- missegregate (50), indicating an absence at the other (25, 64, 65). The polarity somes, indicating that chiasmata are func- of the obligate chiasma. A role for the SC gradient generally is assumed to reflect tional in the absence of tripartite SC (50). in crossover interference is further sup- the probability of heteroduplex forma- In addition, zipl strains display only a ported by two exceptional organisms, tion; the closer an allele is to the DSB, the modest defect in sister-chromatid cohe- Schizosaccharomyces pombe and Aspergil- higher the probability it will be included in sion (50). Although these results demon- lus nidulans, that fail to make SC and do heteroduplex. However, recent studies in- strate that tripartite SC is not essential for not exhibit interference (78, 79). dicate that inhibition of mismatch correc- meiosis I chromosome segregation, they In yeast, small chromosomes sustain tion (e.g., by a msh2 mutation) results in a do not preclude a role for SC components more crossovers per kilobase pair than uniformly high frequency of conversion in segregation. In the redl mutant (which large chromosomes (74,80). Chromosome throughout the gene (62, 64). One model fails to make axial elements), even recom- I is a small chromosome whose density of that accounts for these unexpected find- binant chromosomes missegregate (11). meiotic crossovers is more than twice that ings suggests that a polarity gradient re- Studies of yeast artificial chromosomes of the genomic average. When chromo- flects the direction of mismatch repair provide additional evidence that crossing- some I is bisected to produce two smaller rather than the probability of heterodu- over is not sufficient to ensure correct chromosomes, recombination in chromo- plex formation (64). According to this view, disjunction. Crossovers that occur in yeast some I sequences increases even further. mismatches near the DSB are usually re- or human sequences promote proper seg- When chromosome I is fused to a larger paired in favor of the invading strand (lead- regation of artificial chromosomes, whereas chromosome, recombination decreases. ing to gene conversion), whereas mis- crossovers in bacteriophage A DNA do not These results demonstrate the existence of matches far from the DSB are generally (69, 70). Thus, apparently only those cross- a control mechanism that responds di- repaired in favor of the invaded duplex (and overs that occur in the context of a particular rectly to chromosome size (74). therefore undetected). An alternative sequence or structure lead to functional model suggests that polarity gradients re- chiasmata. Meiotic Cell Cycle Checkpoints flect the transition from asymmetric to sym- An alternative pathway of meiosis I metric heteroduplex (Fig. 4 e and f) (62). chromosome segregation, called distribu- The events of the mitotic cell cycle are Mismatch repair enzymes are postulated to tive disjunction, promotes the segregation ordered into pathways in which the initi- be coupled to the machinery that promotes of chromosomes that cannot (or did not) ation of late events depends on the suc- symmetric strand exchange, such that ex- recombine. The distributive system re- cessful completion of earlier events (81). change is terminated (and perhaps re- quires neither crossing-over nor sequence Control mechanisms that enforce depen- versed) whenever a mismatch is encoun- homology. Two nonhomologous artificial dency are called checkpoints. Studies of tered. The mismatch repair machinery is chromosomes (71), or two genuine yeast mutants indicate that checkpoints also op- known to act as a barrier to recombination chromosomes that lack homologs (72), erate during the meiotic cell cycle and between divergent DNA sequences in yeast segregate to opposite poles in -90% of identify the transition from pachytene as a (66) and bacteria (67), supporting the hy- meioses. Disjunction is preceded by phys- critical control point. Several mutants that pothesis that mismatch repair enzymes can ical pairing between the chromosomes arrest at the start point of the mitotic cell prevent or reverse strand transfer. during meiotic prophase (73). The distrib- cycle arrest in the pachytene stage of utive system probably ensures proper dis- meiosis (82). Cells in pachytene still have Chromosome Segregation junction of those rare chromosomes that the option of reverting to mitotic cell fail to recombine in wild-type diploids but division, whereas cells that have pro- The meiosis I division is distinct from is apparently inadequate to handle multi- gressed beyond this point are irreversibly mitosis and meiosis II in that sister chro- ple pairs of nonrecombinant chromo- committed to reductional chromosome matids remain associated, while homolo- somes. segregation. gous chromosomes segregate from each Several mutants with specific defects in other. Proper reductional segregation de- Regulation of Crossover Distribution meiotic chromosome metabolism-zipl pends on crossing-over to establish chias- (83), dmcl (41), top2 (84), and sepl (42) mata, which are chromatin bridges be- The distribution of meiotic crossovers is arrest in the pachytene stage of meiosis. tween nonsister chromatids that persist nonrandom in two respects (51). (i) Two All of these mutants retain viability at the after the SC has disassembled. During crossovers on the same chromosome arm arrest point, as expected if a checkpoint is prometaphase, homologous centromeres rarely occur close together. (ii) Every pair operating. Some or all of these mutants can become attached to spindle fibers of chromosomes (no matter how small) probably arrest in response to the accu- Downloaded by guest on October 2, 2021 Review: Roeder Proc. Natl. Acad. Sci. USA 92 (1995) 10455

mulation of intermediates in the recom- 2. Ross-Macdonald, P. & Roeder, G. S. Jones, E. W. (Cold Spring Harbor Lab. bination and/or synapsis pathway. In sup- (1994) Cell 79, 1069-1080. Press, Plainview, NY), pp. 407-521. port ofthis hypothesis, the prophase arrest 3. Sym, M., Engebrecht, J. & Roeder, G. S. 34. Alani, E., Padmore, R. & Kleckner, N. of the dmcl (41), top2 (85), and zipl (3) (1993) Cell 72, 365-378. (1990) Cell 61, 419-436. mutants is alleviated by mutations that 4. Loidl, J., Klein, F. & Scherthan, H. (1994) 35. Ivanov, E. L., Korolev, V. G. & Fabre, F. block the initiation of recombination and J. Cell Biol. 125, 1191-1200. (1992) Genetics 132, 651-664. 5. Scherthan, H., Loidl, J., Schuster, T. & 36. Johzuka, K. & Ogawa, H. (1995) Genetics synapsis. The complete absence of recom- Schweizer, D. (1992) Chromosoma 101, bination does not trigger a checkpoint as 139, 1521-1520. 590-595. 37. Raymond, W. E. & Kleckner, N. (1993) evidenced by the fact that mutants defec- 6. von Wettstein, D., Rasmussen, S. W. & tive in DSB formation sporulate effi- Holm, P. B. (1984) Annu. Rev. Genet. 18, Nucleic Acids Res. 21, 3851-3856. ciently. 331-413. 38. Sun, H., Treco, D. & Szostak, J. W. (1991) 7. Padmore, R., Cao, L. & Kleckner, N. Cell 64, 1155-1161. The Rad9 and Rad24 proteins of yeast 39. Szostak, J. W., Orr-Weaver, T. L., Roth- are responsible for arrest in the G1 (86) (1991) Cell 66, 1239-1256. 8. Moens, P. B. & Pearlman, R. E. (1988) stein, R. J. & Stahl, F. W. (1983) Cell 33, and G2 (87, 88) stages of the mitotic cell Bioessays 9, 151-153. 25-35. cycle in response to DNA damage, such as 9. Sym, M. & Roeder, G. S. (1995) J. Cell 40. Shinohara, A., Ogawa, H. & Ogawa, T. DSBs. A rad9 mutation does not bypass Biol. 128, 455-466. (1992) Cell 69, 457-470. the meiotic arrest of the dmcl (41), sepl 10. Hollingsworth, N. M., Goetsch, L. & By- 41. Bishop, D., Park, D., Xu, L. & Kleckner, (42), and zipl (M. Sym and G.S.R., un- ers, B. (1990) Cell 61, 73-84. N. (1992) Cell 69, 439-456. published data) mutants; however, a rad24 11. Rockmill, B. & Roeder, G. S. (1990) Ge- 42. Tishkoff, D. X., Rockmill, B., Roeder, mutation does restore sporulation at least netics 126, 563-574. G. S. & Kolodner, R. D. (1995) Genetics to dmcl (D. Lydall, Y. Nikolsky, D. Bishop, 12. Hollingsworth, N. M. & Byers, B. (1989) 139, 495-509. and T. Weinert, personal communication). Genetics 121, 445-462. 43. West, S. C. (1992)Annu. Rev. Biochem. 61, Thus, mitotic cell cycle checkpoints do op- 13. Schwacha, A. & Kleckner, N. (1994) Cell 603-640. erate during meiosis, but meiosis-specific 76, 51-63. 44. Ogawa, T., Yu, X., Shinohara, A. & 14. Friedman, D. B., Hollingsworth, N. M. & Egelman, E. H. (1993) Science 259, 1896- factors may modify the signals generated Byers, B. (1994) Genetics 136, 449-464. and the response machinery. 1899. 15. Hollingsworth, N. M. & Johnson, A. D. 1241-1243. Studies of meiotic mutants suggest dif- (1993) Genetics 133, 785-797. 45. Sung, P. (1994) Science 265, ferences between laboratory strains of 16. Weiner, B. M. & Kleckner, N. (1994) Cell 46. Kolodner, R., Evans, D. H. & Morrison, yeast in the operation of cell cycle check- 77, 977-991. P. T. (1987) Proc. Natl. Acad. Sci. USA 84, points. For example, the zipl mutant 17. Nag, D. K. & Petes, T. D. (1993) Mol. Cell. 5560-5564. Biol. 13, 2324-2331. 47. Collins, I. & Newlon, C. S. (1994) Cell 76, sporulates in a strain background in which 65-75. arrest 18. Rockmill, B., Engebrecht, J., Scherthan, the dmcl mutant displays prophase H., Loidl, J. & Roeder, G. S. (1995) Ge- 48. Goyon, C. & Lichten, M. (1993) Mol. Cell. and vice versa (3, 41, 50, 93). The strain- netics 141, 49-59. Biol. 13, 373-382. specific behavior of zipl and dmcl sug- 19. Kleckner, N. & Weiner, B. M. (1993) Cold 49. Engebrecht, J., Hirsch, J. & Roeder, G. S. gests that meiotic arrest in these mutants Spring Harbor Symp. Quant. Bio. 58, 553- (1990) Cell 62, 927-937. is triggered by different signals or effected 565. 50. Sym, M. & Roeder, G. S. (1994) Cell 79, by different mechanisms. 20. Jinks-Robertson, S. & Petes, T. D. (1985) 283-292. Proc. Natl. Acad. Sci. USA 82, 3350-3354. 51. Carpenter, A. T. C. (1988) in Genetic Re- Overview 21. Lichten, M., Borts, R. H. & Haber, J. E. combination, eds. Kucherlapati, R. & (1987) Genetics 115, 233-246. Smith, G. R. (Am. Soc. Microbiol., Wash- Over the past several decades, the meiotic 22. Borts, R. H., Lichten, M. & Haber, J. E. ington, DC), pp. 529-548. process has been studied extensively at the (1986) Genetics 113, 551-567. 52. Bishop, D. K. (1994) Cell 79, 1081-1092. cytological level in a variety of organisms. 23. Cao, L., Alani, E. & Kleckner, N. (1990) 53. Gilbertson, L. A. & Stahl, F. W. (1994) Cell 61, 1089-1101. 11934- Due in large part to recent studies of 24. Goldway, M., Sherman, A., Zenvirth, D., Proc. Natl. Acad. Sci. USA 91, meiosis in yeast, we now have molecular Arbel, T. & Simchen, G. (1993) Genetics 11937. handles on the structures and processes 133, 159-169. 54. de Massy, B., Baudat, F. & Nicolas, A. observed microscopically. Structural com- 25. Nicolas, A., Treco, D., Schultes, N. P. & (1994) Proc. Natl. Acad. Sci. USA 91, ponents of the SC have been identified Szostak, J. W. (1989) Nature (London) 11929-11933. and enzymatic components of the recom- 338, 35-39. 55. Modrich, P. (1991) Annu. Rev. Genet. 25, bination machinery are in hand. Cytolog- 26. Sun, H., Treco, D., Schultes, N. P. & Szos- 229-253. tak, J. W. (1989) Nature (London) 338, 56. New, L., Liu, K. & Crouse, G. F. (1993) ical entities that have been the topics of 87-90. Mol. Gen. Genet. 239, 97-108. speculation and theory building for many 27. Game, J. C. (1992) Dev. Genet. 13, 485- 57. Reenan, R. A. G. & Kolodner, R. D. years are now the subjects of rigorous 497. (1992) Genetics 132, 963-973. experimental tests. 28. Zenvirth, D., Arbel, T., Sherman, A., 58. Kramer, W., Kramer, B., Williamson, Goldway, M., Klein, S. & Simchen, G. M. S. & Fogel, S. (1989) J. Bacteriol. 171, I thank Kevin Madden, Janet Novak, Beth (1992) EMBO J. 11, 3441-3447. 5339-5346. Rockmill, Petra Ross-Macdonald, and Albert 29. Wu, T.-C. & Lichten, M. (1994) Science 59. Prolla, T. A., Christie, D.-M. & Liskay, Smith for helpful comments on the manuscript. 263, 515-518. R. M. (1994) Mol. Cell. Biol. 14, 407-415. I am grateful to Nancy Hollingsworth, Nancy 30. White, M. A., Detloff, P., Strand, M. & 60. Reenan, R. A. G. & Kolodner, R. D. Kleckner, Franz Klein, Michael Lichten, Josef Petes, T. D. (1992) Curr. Genet. 21, 109- Genetics 975-985. Loidl, Hideyuki Ogawa, Tomoko Ogawa, and 116. (1992) 132, Ted Weinert for communicating unpublished 31. Rocco, V., de Massy, B. & Nicolas, A. 61. Prolla, T. A., Pang, Q., Alani, E., Kolod- data. I apologize to those investigators whose (1992) Proc. Natl. Acad. Sci. USA 89, ner, R. D. & Liskay, R. M. (1994) Science work was not cited, or cited only through re- 12068-12072. 265, 1091-1093. views, due to the brevity of this review. Re- 32. Wu, T.-C. & Lichten, M. (1995) Genetics 62. Alani, E., Reenan, R. A. G. & Kolodner, search was supported by National Institutes of 140, 55-66. R. D. (1994) Genetics 137, 19-39. Health Grant GM28904 and American Cancer 33. Petes, T. D., Malone, R. E. & Symington, 63. Nag, D. K., White, M. A. & Petes, T. D. Society Grant VM-7C. L. S. (1991) in The Molecular and Cellular (1989) Nature (London) 340, 318-320. Biology of the Yeast Saccharomyces: Ge- 64. Detloff, P., White, M. A. & Petes, T. D. 1. Dresser, M. E. & Giroux, C. N. (1988) J. nome Dynamics, Protein Synthesis, and En- (1992) Genetics 132, 113-123. Cell Biol. 106, 567-578. ergetics, ed. Broach, J. R., Pringle, J. R. & 65. Malone, R. E., Bullard, S., Lundquist, S., Downloaded by guest on October 2, 2021 10456 Review: Roeder Proc. Natl. Acad. Sci. USA 92 (1995)

Kim, S. & Tarkowski, T. (1992) Nature 74. Kaback, D. B., Guacci, V., Barber, D. & 84. Rose, D. & Holm, C. (1993) Mol. Cell. (London) 359, 154-155. Mahon, J. W. (1992) Science 256, 228- Biol. 13, 3445-3455. 66. Selva, E. M., New, L., Crouse, G. F. & 232. 85. Rose, D., Thomas, W. & Holm, C. (1990) Lahue, R. S. (1995) Genetics 139, 1175- 75. King, J. S. & Mortimer, R. K. (1990) Ge- Cell 60, 1009-1017. 1188. netics 126, 1127-1138. 86. Siede, W., Friedberg, A. S., Dianova, I. & 67. Rayssiguier, C., Thaler, D. S. & Radman, 76. Egel, R. (1978) Heredity 41, 233-237. Friedberg, E. C. (1994) Genetics 138,271- M. (1989) Nature (London) 342,396-401. 77. Maguire, M. P. (1988) J. Theor. Bio. 134, 281. 68. Maguire, M. P. (1978) Exp. Cell Res. 112, 565-570. 87. Weinert, T. A. & Hartwell, L. H. (1988) 297-308. 78. Egel-Mitani, M., Olson, L. W. & Egel, R. Science 241, 317-322. 69. Ross, L. O., Treco, D., Nicolas, A., Szos- (1982) Hereditas 97, 179-187. 88. Weinert, T. A., Kiser, G. L. & Hartwell, tak, J. W. & Dawson, D. (1992) Genetics 79. Olson, L. W., Eden, U., Egel-Mitani, M. Dev. 8, 652-665. 131, 541-550. & Egel, R. (1978) Hereditas 89, 189-199. L. H. (1994) Genes 70. Sears, D. D., Hegemann, J. H. & Hieter, 80. Kaback, D. B., Steensma, H. Y. & de- 89. Liu, J., Wu, T.-c. & Lichten, M. (1995) P. (1992) Proc. Nati. Acad. Sci. USA 89, Jonge, P. (1989) Proc. Natl. Acad. Sci. EMBO J. 14, 4599-4608. 5296-5300. USA 86, 3694-3698. 90. Xu, L. & Kleckner, N. (1995) EMBOJ., in 71. Dawson, D. S., Murray, A. W. & Szostak, 81. Murray, A. W. (1992) Nature (London) press. J. W. (1986) Science 234, 713-717. 359, 599-604. 91. de Massy, B., Rocco, V. & Nicolas, A. 72. Guacci, V. & Kaback, D. B. (1991) Genet- 82. Shuster, E. 0. & Byers, B. (1989) Genetics (1995) EMBO J., in press. ics 127, 475-488. 123, 29-43. 92. Hollingsworth, N. M., Ponte, L. & Halsey, 73. Loidl, J., Scherthan, H. & Kaback, D. B. 83. Nag, D. K., Scherthan, H., Rockmill, B., C. (1995) Genes Dev. 9, 1728-1739. (1994) Proc. Natl. Acad. Sci. USA 91, Bhargava, J. & Roeder, G. S. (1995) Ge- 93. Rockmill, B., Sym, M., Scherthan, H. & 331-334. netics 141, 75-86. Roeder, G. S. (1995) Genes Dev., in press. Downloaded by guest on October 2, 2021