COSA-1揭示了动态平衡和减数分裂分频器

COSA-1Reveals Robust Homeostasis

and Separable Licensing and Reinforcement Steps Governing Meiotic Crossovers

Rayka Yokoo,1Karl A.Zawadzki,2Kentaro Nabeshima,2,4Melanie Drake,3Swathi Arur,3and Anne M.Villeneuve1,2,*

1Department of Genetics

2Department of Developmental Biology

Stanford University School of Medicine,Stanford,CA94305,USA

3Department of Genetics,University of Texas MD Anderson Cancer Center,Houston,TX77030,USA

4Present address:Department of Cell and Developmental Biology,University of Michigan Medical School,Ann Arbor,MI48109-2200,USA *Correspondence:annev@https://www.wendangku.net/doc/9f14855029.html,

DOI10.1016/j.cell.2012.01.052

SUMMARY

Crossovers(COs)between homologous chromo-somes ensure their faithful segregation during meiosis.We identify C.elegans COSA-1,a cyclin-related protein conserved in metazoa,as a key com-ponent required to convert meiotic double-strand breaks(DSBs)into COs.During late meiotic prophase,COSA-1localizes to foci that correspond to the single CO site on each homolog pair and indi-cate sites of eventual concentration of other con-served CO proteins.Chromosomes gain and lose competence to load CO proteins during meiotic pro-gression,with competence to load COSA-1requiring prior licensing.Our data further suggest a self-rein-forcing mechanism maintaining CO designation. Modeling of a nonlinear dose-response relationship between IR-induced DSBs and COSA-1foci reveals ef?cient conversion of DSBs into COs when DSBs are limiting and a robust capacity to limit cytologically differentiated CO sites when DSBs are in excess. COSA-1foci serve as a unique live cell readout for investigating CO formation and CO interference. INTRODUCTION

Genetic recombination is an integral part of the sexual reproduc-tion program by which most diploid organisms generate haploid gametes.In addition to promoting reassortment of genetic traits, crossover(CO)recombination between homologous chromo-somes plays a key mechanical role in directing their segregation at the meiosis I division(Martinez-Perez and Colaia′covo,2009). Crossing over is initiated by double-strand DNA breaks(DSBs), a subset of which is converted into COs by a specialized meiotic repair pathway that uses the homolog as a recombination partner.COs mature during the pachytene stage of meiotic prophase in the context of the synaptonemal complex(SC), a meiosis-speci?c structure that assembles at the interface between lengthwise-aligned homologs.During the subsequent diplotene and diakinesis stages,the SC disassembles and chromosomes condense and reorganize around each CO site to reveal a chiasma,a structure resulting from the CO in con-junction with sister chromatid cohesion?anking the CO site. Chiasmata maintain connections between homologs through metaphase of meiosis I,when they enable reliable biorientation of homologs toward opposite spindle poles.

Despite reliance on COs to ensure homolog segregation,most organisms make very few COs per homolog pair(on the order of one to three per chromosome arm)even though DSBs occur in substantial excess(Martinez-Perez and Colaia′covo,2009). Moreover,the distribution of COs along chromosomes re?ects a propensity of(nascent)COs to inhibit formation of other COs nearby on the same chromosome pair,a phenomenon known as CO interference(Muller,1916).These properties imply that meiotic recombination must be governed by a robust CO control system that can guarantee the formation of suf?cient COs while simultaneously limiting their numbers.

Although CO control remains poorly understood,many components of the machinery that promotes the CO outcome of meiotic DSB repair have been identi?ed.One key player is a heterodimer of conserved meiosis-speci?c MutS family members MSH4and MSH5.MSH4-MSH5is implicated in formation and/or stabilization of CO intermediates in vivo(Bau-dat and de Massy,2007;Lynn et al.,2007)and can load onto Holliday junction substrates in vitro(Pochart et al.,1997; Snowden et al.,2004).Further,localization of MSH4-MSH5to DSB-dependent foci implies function at the sites of nascent recombination events.Of note,whereas the number of Msh4 foci in S.cerevisiae corresponds well with the number of COs generated by this pathway(Hollingsworth et al.,1995;Ross-Macdonald and Roeder,1994),MSH4/MSH5foci in several other species are initially detected in signi?cant excess of COs (Higgins et al.,2004,2008;Kneitz et al.,2000;Lenzi et al., 2005;Santucci-Darmanin et al.,2000).Another key player, Zip3/ZHP-3,is a predicted SUMO or ubiquitin ligase(Lynn et al.,2007).S.cerevisiae Zip3functions in regulating SC assembly and localizes in foci that substantially overlap with Msh4,consistent with its role in CO formation(Agarwal and Cell149,75–87,March30,2012a2012Elsevier Inc.75

Roeder,2000;Macqueen and Roeder,2009).C.elegans ZHP-3also functions to promote COs,but in contrast to its S.cerevisiae ortholog,it is dispensable for SC assembly and instead partici-pates in organizing SC disassembly (Bhalla et al.,2008;Jantsch et al.,2004).Moreover,C.elegans ZHP-3initially localizes along the lengths of the SCs and gradually shrinks down to become localized at presumptive CO sites only very late in prophase.The fact that both MSH4/MSH5and ZHP-3show initial localiza-tion that differs substantially from the eventual distribution of COs raises a key question:how does CO-promoting activity become concentrated at and restricted to bona ?de CO sites?Here,we investigate the execution and regulation of CO formation during C.elegans meiosis,exploiting several advanta-geous attributes of this system.CO control is highly robust in C.elegans ,with most chromosome pairs undergoing only a single CO (Hammarlund et al.,2005;Hillers and Villeneuve,2003;Nabeshima et al.,2004).Further,whereas some organ-isms generate a signi?cant fraction of their COs using alternative pathways,essentially all COs in C.elegans are formed using the canonical MSH-4-MSH-5-dependent,interference-sensitive pathway (Zalevsky et al.,1999).Moreover,the ‘‘production line’’organization of the germline enables visualization of cyto-logical correlates of CO progression simultaneously at all stages of meiotic prophase,both under normal conditions and after acute induction of DSBs.

We identify crossover site-associated-1(COSA-1)as a cyclin-related protein widely conserved in metazoa that functions in conjunction with C.elegans MSH-4-MSH-5and ZHP-3to promote COs.GFP::COSA-1localizes in foci speci?cally at presumptive CO sites,illuminating a key transition during CO maturation and providing evidence for a self-reinforcing mecha-nism that sequesters CO factors at designated CO sites.Further,combined experimental analysis and modeling of the response of COSA-1foci to varying DSB levels reveals highly ef?cient conversion of DSBs into COs when DSBs are limiting,as well as a robust capacity to limit the number of cytologically differen-tiated CO sites to one per homolog pair when DSBs are in excess.Together,our ?ndings indicate that COSA-1foci repre-sent a reliable surrogate for the events that are distributed by the CO control system.The ability to visualize GFP::COSA-1in live worms creates an unprecedented opportunity,making it possible to apply genetic screening strategies to investigate the elusive basis of CO interference.RESULTS

COSA-1Is Required to Convert Meiotic DSBs into Interhomolog COs

The cosa-1(me13)mutation was isolated based on frequent mis-segregation of sex chromosomes:cosa-1hermaphrodites (XX)produce 38%XO male self progeny (compared to 0.2%for wild-type [Hodgkin et al.,1979]).cosa-1hermaphrodites also produce a high frequency of inviable embryos (97%,n =2737),indicative of autosomal missegregation.These segregation defects re?ect a lack of chiasmata connecting homologous chromosomes (Figure 1A).Whereas wild-type diakinesis oocytes contain six DAPI stained bodies (n =172),correspond-ing to six pairs of homologs held together by chiasmata,

cosa-1

Figure 1.COSA-1Is Required to Convert Meiotic DSBs into Inter-homolog COs

(A)Full karyotypes of individual diakinesis-stage oocytes.Six DAPI-stained bodies in the wild-type nucleus correspond to six pairs of homologs con-nected by chiasmata;12individual chromosomes (univalents)in the cosa-1-mutant nucleus re?ect a lack of chiasmata.

(B)cosa-1mutant pachytene nuclei in which pairing was assessed either by FISH at the 5S rDNA locus (chromosome V)or by immunostaining for X chromosome pairing center (X-PC)-binding protein HIM-8.A single FISH or HIM-8signal in each nucleus indicates successful pairing.

(C)cosa-1mutant pachytene nuclei,showing colocalization of SC lateral element protein HIM-3and SC central region protein SYP-1between parallel tracks of DAPI-stained chromatin.

(D)Immunolocalization of RAD-51in mid-to-late pachytene nuclei in the cosa-1mutant.RAD-51foci indicative of DSB formation are abundant in midpachytene and are greatly reduced or absent in most late-pachytene nuclei (asterisk indicates an apoptotic nucleus).

(E)Early diplotene nuclei.In WT,SYP-1and HTP-1/2are localized to reciprocal domains on each chromosome pair;in cosa-1,this indicator of CO formation is not observed,as SYP-1and HTP-1/2remain extensively colocalized.Scale bar,5m m.See also Table S1.

76Cell 149,75–87,March 30,2012a2012Elsevier Inc.

oocytes contain an average of 11resolvable DAPI-stained bodies (n =177),indicating absence of chiasmata.

Several lines of evidence together indicate that lack of chias-mata in cosa-1mutants re?ects a defect in the process of meiotic recombination,speci?cally in the conversion of initiated recom-bination events into COs.First,cosa-1mutants are pro?cient for homolog pairing,as assessed by ?uorescence in situ hybridiza-tion (FISH)at the 5S rDNA locus on chromosome V and by HIM-8immuno?uorescence (IF)to evaluate pairing at a speci?c region of the X chromosome known as the pairing center (MacQueen et al.,2005;Phillips et al.,2005)(Figure 1B).Further,cosa-1mutants are pro?cient for assembly of the SC between paired homologs,as immunostaining revealed colocalization of HIM-3(a component of the lateral elements of the SC)and SYP-1(a component of the SC central region)(MacQueen et al.,2002;Zetka et al.,1999)between parallel tracks of DAPI-stained DNA (Figure 1C).

Moreover,a combination of cytological and genetic data indi-cates that cosa-1mutants initiate recombination but fail to repair

DSBs as COs.DNA-strand exchange protein RAD-51(Colaia ′-covo et al.,2003)is detected in abundant foci beginning in late zygotene/early pachytene,indicating that DSBs form in the cosa-1mutant (Figure 1D).These foci eventually disappear at late pachytene,suggesting that DSBs are repaired but are not converted into interhomolog COs.Further,whereas COs trigger relocalization of SYP-1and chromosome axis protein HTP-1/2to

reciprocal domains during late pachytene in wild-type meiosis (Martinez-Perez et al.,2008),cosa-1mutants lack this prominent cytological indicator of COs (Figure 1E).Finally,measurement of genetic recombination frequency showed that absence of COSA-1reduces the incidence of COs to <1%of control levels (Table S1available online).

This phenotypic analysis indicates a role for COSA-1in the formation of meiotic COs.Moreover,these phenotypes closely parallel those of mutants lacking HIM-14(MSH4),MSH-5,and ZHP-3(Jantsch et al.,2004;Kelly et al.,2000;Zalevsky et al.,1999),suggesting that COSA-1functions in conjunction with these conserved CO-promoting proteins.

cosa-1Encodes a Cyclin-Related Protein Conserved in Metazoa

SNP mapping and sequencing identi?ed a G-to-A transition in the me13mutant,resulting in a premature stop at codon 148of predicted gene Y71H2AM.7(360codons total).An indepen-dently generated deletion allele (tm3298)fails to complement me13,con?rming the identity of cosa-1as Y71H2AM.7(Figure 2A).

Three notable observations arose from our search for COSA-1homologs (see Extended Experimental Procedures for bioinfor-matics analyses).First,COSA-1orthologs are found throughout the metazoan lineage (Figure 2B)but are not detected in plants and fungi,suggesting either that COSA-1arose in metazoa

or

Figure 2.COSA-1Is a Distant Member of the Cyclin Superfamily with Orthologs in Metazoa

(A)(Top)Predicted gene structure of C.elegans cosa-l with mutant alleles indicated.Gray,UTR;magenta,coding exons.(Bottom)Construct used to express GFP::COSA-1.Green,GFP coding sequence;blue,extra tags and linker sequences.(B)Phylogenetic tree depicting a sampling of metazoan species.Green indicates lineages in which COSA-1,MSH-4,and MSH-5orthologs are all present;red indicates the absence of all three from Drosophilid species.

(C)Predicted structure of residues 56–360of C.elegans COSA-1(yellow and magenta)aligned with crystal structure of residues 167–426of human cyclin B1(cyan).N-terminal residues of COSA-1and cyclin B1were removed to aid visu-alization of the two core cyclin-fold motifs.In canonical cyclins,cyclin-fold motifs consist of ?ve a helices,with well-conserved interhelical angles in the N-terminal cyclin box motif.In the predicted COSA-1structure,the N-terminal cyclin box is interrupted by an insertion of 33amino acids,modeled here as an extension of a helix 2and an additional helix (magenta,a -2.5).Predicted a helices 3–5of COSA-1align well with the cor-responding helices of cyclin B1and cyclin A,which contribute to the cyclin/CDK interface in cyclin A/CDK2(Jeffrey et al.,1995).

See also Figures S1and S6and Extended Exper-imental Procedures .

Cell 149,75–87,March 30,2012a2012Elsevier Inc.77

that divergence in plants and fungi has rendered it unrecogniz-able.Second,whereas COSA-1orthologs are present in other diptera,they are absent in the Drosophila genus(Figure2B), which also lacks both MSH4and MSH5(Schurko et al.,2010). Together with our phenotypic data,this phylogenetic distribution suggests that COSA-1and MSH-4-MSH-5may act as a func-tional module.Third,expression data for the mouse and human orthologs,provisionally named cyclin N-terminal domain-con-taining1(Cntd1),are consistent with conservation of function in meiosis.

COSA-1is predicted to have a cyclin-like structure based on analyses using PSI-BLAST and the PHYRE and I-TASSER struc-ture prediction servers(see Extended Experimental Procedures). Alignment of the predicted COSA-1structure with the crystal structure of human cyclin B1shows a high degree of similarity (Figure2C),including conservation of the region corresponding to the Cyclin/CDK interface.However,an insertion of33amino acids results in an extra predicted a helix in the highly conserved N-terminal cyclin box domain(Petri et al.,2007).This feature is also present in human CNTD1(Figure S1)and distinguishes the COSA-1/CNTD1family from conventional cyclins,indicating that it represents a distinct branch of the cyclin superfamily. GFP::COSA-1Marks Sites of Presumptive COs

We generated a transgenic strain expressing a functional GFP::COSA-1fusion to assess localization of COSA-1during meiotic prophase progression(Figures2A,3A,and S2).GFP:: COSA-1is detected as a diffuse nucleoplasmic signal beginning in early pachytene and then displays a striking localization pattern later in pachytene that is consistent with expectations for a protein that marks CO sites.Starting at the transition from mid to late pachytene and persisting through diplotene,GFP::COSA-1local-izes to6.0±0.2bright foci per nucleus(n=76),corresponding in number to the six COs—one per chromosome pair—that form during wild-type meiosis(Figure3A).At diplotene,GFP::COSA-1 localizes at the site of the single emerging chiasma on each homolog pair(Figures3A and3C).We conclude that COSA-1local-izes to CO sites;hence,the name crossover site-associated-1. COSA-1foci and RAD-51foci are detected in largely recip-rocal domains within the pachytene region of the germline.Prior to late pachytene,nuclei only have RAD-51foci,whereas in late pachytene,most nuclei only have COSA-1foci(Figure S3).Even in nuclei with both,RAD-51and COSA-1do not colocalize.This pattern is consistent with a major transition in recombination progression and indicates that COSA-1is loaded after the majority of RAD-51is removed.

GFP::COSA-1Foci Are Sites of Eventual Concentration of Multiple CO-Promoting Factors

We compared the localization pattern of GFP::COSA-1with those of two other conserved CO-promoting proteins,MSH-5 and ZHP-3.This analysis revealed that GFP::COSA-1foci repre-sent sites where MSH-5and ZHP-3also eventually become concentrated.MSH-5is?rst detected during midpachytene,ap-pearing as faint foci that accumulate in excess of eventual COs (Figure4A).Upon transition to late pachytene,MSH-5foci decrease in number to six per nucleus,colocalizing with COSA-1foci and exhibiting an increased intensity that presum-ably re?ects increased local concentration of MSH-5at these sites.ZHP-3initially localizes in stretches along the full length of the SCs but begins to concentrate to one side of each presumptive CO site during late pachytene,initially forming comet-like structures stretching from the presumptive CO site to one end of the chromosome and then progressively shrinking to foci by mid-diplotene(Bhalla et al.,2008;Jantsch et al.,2004). COSA-1begins to appear as foci just as ZHP-3begins to redis-tribute and then localizes at the head of each ZHP-3comet during late pachytene progression and colocalizes with each ZHP-3focus at diplotene(Figure3B).Thus,while both MSH-5 and ZHP-3initially exhibit broader localization patterns,both eventually become concentrated at COSA-1-marked sites. MSH-5,ZHP-3,and COSA-1not only colocalize at presump-tive CO sites,but they also exhibit interdependence for this local-ization.First,COSA-1foci were not observed in a msh-5mutant (Figure S4A);conversely,late pachytene MSH-5foci were not detected in a cosa-1mutant(Figure4B).Midpachytene MSH-5 foci were also diminished or lost in the cosa-1mutant,suggest-ing that COSA-1may facilitate or stabilize the association of MSH-5with nascent recombination events even before COSA-1is detected on chromosomes.Second,ZHP-3per-sisted along the lengths of SCs during late pachytene in the cosa-1mutant(Figure4B);conversely,six COSA-1foci were not detected in a zhp-3mutant(Figure S4B).

Kinetics of and Stage Dependence of Competence for Loading CO Proteins at IR-Induced Recombination Sites We used GFP::COSA-1to investigate changes that occur as recombination intermediates form and mature into COs,con-ducting a time course analysis to assess the temporal kinetics and developmental constraints governing recruitment of re-combination proteins to DSBs induced by ionizing radiation (IR).For these experiments,spo-11mutant worms(which lack endogenous DSBs)(Dernburg et al.,1998)expressing GFP:: COSA-1were exposed to1,000rads of g irradiation and were then assessed at various time points for localization of recombi-nation proteins.These experiments exploited the fact that C.elegans chromosomes undergo homologous pairing and synapsis in the absence of DSBs so that pachytene nuclei in the spo-11mutant are already poised to engage in interhomolog recombination once DSBs are introduced.

In untreated control spo-11germlines,most nuclei lack MSH-5and COSA-1foci,and ZHP-3remains localized along the length of the SC throughout pachytene and into diplotene, as expected given the lack of DSBs and COs(Figure5A)(Bhalla et al.,2008).However,a subset of late-pachytene nuclei have one or two atypical COSA-1aggregates in which MSH-5can also be detected;these tend to correspond to spots of higher ZHP-3concentration and may re?ect an inherent tendency of these proteins to colocalize.

By1hr post-IR,both MSH-5and RAD-51foci were detected in irradiated germlines(Figure5B and data not shown).Whereas RAD-51foci were present throughout the germline,MSH-5foci were restricted to a limited region.Induced MSH-5foci were abundant in the midpachytene region,where MSH-5foci ?rst appear during wild-type meiosis,but were not detected in late-pachytene nuclei,which appeared similar to unirradiated

78Cell149,75–87,March30,2012a2012Elsevier Inc.

spo-11controls.This suggests that DSBs per se are not suf?-cient to recruit MSH-5but must occur during the appropriate stage of meiotic prophase to be competent to load MSH-5de novo.Notably,at 1hr post-IR,there was little change in COSA-1localization,indicating that the conditions for COSA-1loading were not yet

met.

Figure 3.GFP::COSA-1Localizes to Foci Corresponding to CO Sites

(A and B)IF images of a portion of a gonad extending from midpachytene through diplotene and early diakinesis.GFP::COSA-1foci are detected from late pachytene through early diakinesis.

(A)(Left inset)Late-pachytene nuclei,each containing six bright foci.(Right insets)Diplotene nuclei,with one focus on each chromosome pair;bottom panel shows COSA-1foci positioned at the site of the single emerging chiasma on each chromosome pair.

(B)Relationship between localization of GFP::COSA-1and ZHP-3.(Upper-left inset)Six COSA-1foci in a midpachytene nucleus with ZHP-3in long stretches along the chromosomes.(Upper-right inset)COSA-1localized at one end of each comet-like stretch of ZHP-3.(Bottom inset)COSA-1and ZHP-3colocalization in a diplotene nucleus.

(C)Representative images of GFP::COSA-1localization in late-diplotene/early diakinesis nuclei,highlighting the location of GFP::COSA-1at the site of the single emerging chiasma on each chromosome pair.(Large panels)Full projections of entire nuclei showing all six bivalents;asterisk indicates a bivalent depicted in smaller panels,which shows partial projections of individual bivalents.

Scale bars,5m m except in the single bivalent panels in (C),in which scale bar is 1m m.See also Figures S2and S3.

Cell 149,75–87,March 30,2012a2012Elsevier Inc.79

At 2.5hr post-IR,a few nuclei with 6COSA-1foci were visible near the mid-to-late pachytene border (Figure 5C).These likely correspond to nuclei that had been in midpachytene at the

time of irradiation,had loaded MSH-5,had subsequently pro-gressed into late pachytene,and had thus become competent to load COSA-1.

At 4hr post-IR,the region containing nuclei with six COSA-1foci had expanded further into the late-pachytene region,and a subset of nuclei had begun to display reorganization of ZHP-3(Figures 5C and S5B).Of note,a distinct boundary is visible within the late-pachytene region between a zone contain-ing nuclei with six COSA-1foci and a zone without such nuclei,creating the impression of a wave of nuclei with six COSA-1-marked CO-sites progressing through late pachytene.

By 8hr post-IR,the wave front had reached early diplotene,and ZHP-3signals had substantially retracted toward the COSA-1foci (Figure 5A).8hr post-IR gonads stained for MSH-5and COSA-1display a striking cytological transition at the mid-to-late pachytene boundary (Figure 5D).Midpachytene nuclei exhibit abundant MSH-5foci but lack COSA-1foci,whereas late-pachytene nuclei have only six MSH-5foci,each colocalizing with COSA-1.

In summary,(1)only nuclei in midpachytene are competent for rapid de novo loading of MSH-5at IR-induced DSBs;(2)DSBs induced in germ cells that are already in late pachytene are not competent to load either MSH-5or COSA-1;(3)germ cells can acquire competence to load COSA-1upon transit into late pachytene if they had been in midpachytene at the time of DSB induction (and presumably had loaded MSH-5at that time);(4)loading of COSA-1appears to stabilize MSH-5,as MSH-5foci not colocalized with COSA-1are lost following transit to late pachytene.

Dose-Response Analysis Reveals Ef?cient DSB

Utilization and a Robust Ability to Limit COSA-1-Marked Sites to One per Homolog Pair

The consistent localization of COSA-1to six foci both in wild-type meiosis and in our IR time course experiments prompted us to conduct a dose-response analysis to investigate the rela-tionship between DSB number and COSA-1-marked sites.For these experiments,we exposed spo-11mutant worms express-ing GFP::COSA-1to different IR doses and then assessed COSA-1foci in late-pachytene nuclei ?xed at 8hr post-IR (Figures 6A and 6B).

This analysis revealed a striking nonlinear relationship between IR dose and COSA-1foci (Figure 6C).From 100to 1,000rads,the average number of foci per nucleus increased with increasing IR dose,suggesting that DSB number limits the number of COSA-1foci at doses below 1krad.At 1,000rads,90%of nuclei scored had exactly six foci,indicating that this dose is suf?cient for 99%of chromosome pairs to receive at least one DSB.The number of foci then plateaued,such that even when the dose was increased 10-fold,exactly six foci were still detected in 90%of nuclei,indicating that most excess DSBs do not yield COSA-1-marked sites.

The observed dose-response relationship closely matches that predicted by a model in which:(1)irradiation results in a random distribution of DSBs,(2)a chromosome pair with zero DSBs will have zero COSA-1foci,and (3)the presence of R one DSB on a chromosome pair yields exactly one COSA-1focus.We used these postulates and assumptions

to

Figure 4.MSH-5Colocalizes with and Depends on COSA-1

(A)IF images showing that MSH-5foci are detected in midpachytene nuclei in excess of eventual COs (cyan inset)and then decline by late pachytene,when they colocalize with GFP::COSA-1foci (yellow inset).

(B)(Left)Late-pachytene nuclei from a wild-type germline,showing comet-like localization of ZHP-3with COSA-1foci at the comet heads.(Right)Late-pachytene nuclei from a cosa-1mutant,showing persistence of ZHP-3localization along the length of the chromosomes and a lack of MSH-5foci.Scale bars,5m m;for insets,scale bars,1m m.See also Figure S4.

80Cell 149,75–87,March 30,2012a2012Elsevier Inc.

Figure5.Time Course of Localization of CO

Proteins at IR-Induced Recombination Sites

Immunolocalization of CO proteins(GFP::COSA-1,

MSH-5,and/or ZHP-3)in pachytene nuclei from

gfp::cosa-1;spo-11worms,either in the absence of

IR(A,left)or at the indicated times following expo-

sure to1kRad IR.Scale bars,5m m.

(A)Localization of COSA-1and ZHP-3or MSH-5in

late-pachytene nuclei in the absence of IR(pre-IR)

and8hr post-IR.In the unirradiated spo-11control,

ZHP-3persists along the lengths of the chromo-

somes,and the majority of nuclei lack COSA-1and

MSH-5foci;a subset of nuclei have one or two

COSA-1/MSH-5aggregates(indicated by asterisks).

8hr post-IR:six bright COSA-1foci localize at the

heads of comet-like ZHP-3signals.

(B)Mid-to-late pachytene region of a1hr post-IR

germline.Abundant IR-induced MSH-5foci are de-

tected speci?cally in midpachytene nuclei(left),

whereas MSH-5foci are not detected above baseline

in late-pachytene nuclei(right;0,1,or2MSH-5signals

colocalize with COSA-1,as in unirradiated controls).

(C)GFP::COSA-1localization in nuclei within the

late-pachytene region at2.5and4hr post-IR;?elds

also include a few midpachytene nuclei(at the left)

and a few early diplotene nuclei(at the right).Circles

indicate nuclei in which six COSA-1foci are de-

tected.At2.5hr post-IR,nuclei with six COSA-1foci

are limited to a narrow zone near the start of the late-

pachytene region.At4hr post-IR,the zone of nuclei

with six COSA-1foci has expanded,presumably

re?ecting movement into and progression through

late pachytene of nuclei that had been exposed to IR

during midpachytene.

(D)Localization of MSH-5and COSA-1at8hr post-

IR in a region spanning the mid-to-late pachytene

transition.(Left inset)Midpachytene nuclei,showing

MSH-5-only foci,in excess of eventual COs.(Right)

Late-pachytene nuclei,showing six MSH-5foci that

colocalize with six COSA-1foci.

See also Figure S5.

Cell149,75–87,March30,2012a2012Elsevier Inc.81

derive a function (see Extended Experimental Procedures for details of modeling)to model the relationship between IR dose and the mean number of COSA-1foci per nucleus:

m =6à

1àe àcr áwherein m =mean number of COSA-1foci per nucleus,r =IR dose,and c =constant describing the relationship between DSB number and IR dose.The best-?t dose-response curve generated by this function (using a value of 0.0039134for c )is shown in blue in Figure 6C,together with the observed data in red.Given the relationship:

l =cr

wherein l =mean number of DSBs per chromosome pair,this ?tted model predicts that an average of 3.9DSBs per chromo-some pair would be generated at the 1,000rad IR dose.This translates to a yield of 1DSB/17Mb/krad,which is identical

to

Figure 6.Dose-Response Analysis Reveals a Highly Nonlinear Relationship between IR-Induced DSBs and COSA-1Foci

(A)Paired IF images showing GFP::COSA-1foci in late-pachytene nuclei from gfp::cosa-1;spo-11germlines exposed to the indicated IR doses,?xed 8hr post-IR,with numbers of foci in each nucleus indicated.Scale bar,5m m.

(B)Stacked bar graph showing percentages of nuclei with the indicated numbers of COSA-1foci at different IR doses.

(C)Graph showing the highly nonlinear relationship between IR dose and the mean number of COSA-1foci per nucleus.Experimental data points are plotted in red,with error bars indicating standard deviation.Our mathematical model (m =6(1àe àcr ))is plotted in blue;see Results and Extended Experimental Procedures .

(D)Graph depicting linear relationship between IR dose and inferred mean number of DSBs per chromosome pair,calculated from our empirical data based on the postulates of our model.Empirical data points are in red,with linear regression in blue.

See also Extended Experimental Procedures .

the DSB yield previously estimated for mammalian cells based on physical detection assays (Thompson and Limoli,2000).The correspondence of these esti-mates for ef?cacy of IR in DSB induction further supports the validity of the rela-tionship between DSBs and COSA-1-marked CO sites deduced from our experimental/modeling analysis.More-over,it implies that most or all DSBs induced by IR in our system are compe-tent to enter the meiotic recombination pathway.

We also took a complementary approach,directly calculating empirical

values for l (here,the inferred mean number of DSBs per chro-mosome pair)using the data for the lower IR doses and the Pois-son equation P(0)=e àl ,wherein P(0)=the fraction of chromo-some pairs inferred to have 0COSA-1foci and thus zero DSBs based on the postulates stated above (see Extended Experi-mental Procedures ).A linear regression analysis describing the relationship between IR dose and l (Figure 6D)provided strong validation for our initial assumptions (R 2=0.997).Moreover,extrapolation of the line (slope =0.00390)yields an average of 3.9DSBs per chromosome pair at the 1kRad dose,matching the value generated above.

In summary,our dose-response data provide strong support for a model in which chromosome pairs lacking DSBs will lack COSA-1foci,whereas chromosome pairs with one or more DSBs will receive a single COSA-1focus regardless of DSB number.This indicates the operation of a robust CO control system that is both:(1)highly ef?cient at converting a single DSB into a CO when DSBs are limiting and (2)highly effective

82Cell 149,75–87,March 30,2012a2012Elsevier Inc.

at limiting the number of cytologically differentiated CO sites to one per chromosome pair when DSBs are in excess.

COSA-1Foci Exhibit Interference under Conditions that Alter CO Number

To further characterize the relationship between COSA-1foci and COs,we assessed COSA-1foci under several circum-stances previously shown to alter CO number.mnT12Fusion Chromosome Homozygotes

Worms homozygous for mnT12,an X;IV fusion chromosome,have ?ve rather than six chromosome pairs.Previous work showed that the fusion chromosome pair undergoes only one CO in the majority of meioses,indicating that interference operates on this extended chromosome to limit CO number (Hillers and Villeneuve,2003).In line with the prior CO analysis,?ve COSA-1foci were observed in the majority of late-pachy-tene nuclei in mnT12homozygotes (Figure 7A),consistent with COSA-1foci being responsive to CO interference oper-ating over distances that exceed the length of a normal chromosome.rtel-1Mutant

RTEL-1is a DNA helicase that can disassemble D loop recombi-nation intermediates in vitro,and the rtel-1mutant was reported to have a roughly 2-fold increase in COs (Barber et al.,2008;Youds et al.,2010).Despite this increase in COs detected by genetic assays,the numbers of COSA-1foci in the rtel-1mutant (6.0±0.2)were indistinguishable from wild-type controls (6.0±0.2;Figure 7A),indicating that COSA-1does not concentrate

in

Figure 7.Relationship of COSA-1to COs in Conditions that Alter CO Number

(A)Graph showing percentages of nuclei with the indicated numbers of COSA-1foci in strains with altered numbers of COs.Numbers of foci in the rtel-1(tm1866)and dpy-28(s939)mutants did not differ signi?cantly from the control (Mann-Whitney test).Worms homozygous for the mnT12(X;IV)fusion chromosome have only ?ve chromosome pairs,and mnT12undergoes only one CO in the majority of meioses;an average of 5.3COSA-1foci per nucleus was observed in mnT12homozygotes.Numbers of COSA-1foci in rtel-1;mnT12worms did not differ signi?cantly from mnT12controls.*WT control contains the gfp::cosa-1transgene in an otherwise wild-type background.(B)The rtel-1mutation does not suppress the lack of chiasmata caused by loss of cosa-1function.Graph shows percent of diakinesis nuclei with a given number of DAPI-stained bodies.As in cosa-1(tm3298)single mutants,11–12DAPI bodies were detected in rtel-1;cosa-1double mutants,re?ecting a lack of chiasmata.Numbers of oocyte nuclei scored:wild-type (n =164),rtel-1(n =125),cosa-1(n =116),rtel-1;cosa-1(n =114).

(C)Bar graph indicating genetic map distances (cM ±95%C.I.)for the unc-60dpy-11interval measured for worms of the indicated genotypes (see Extended Experimental Procedures ).**p <0.001;*p =0.01.The CO frequency in the rtel-1mutant (19.6cM)was signi?cantly elevated over wild-type (13.4cM,p =0.0002;Fisher’s exact test)and cosa-1/+(13.6cM,p =0.0006)controls,which did not differ from each other.rtel-1also differed signi?cantly from rtel-1;cosa-1/+(15.6cM,p =0.01),indicating that elevation in CO frequency was suppressed in rtel-1;cosa-1/+worms.

(D)Paired 3D volume renderings of a representative nucleus used to quantify X chromosome-associated COSA-1foci in the dpy-28mutant.Staining for chromosome axis protein HTP-3reveals the paths of synapsed chromosome pairs;arrow indicates the X chromosome,marked by X-PC-associated protein HIM-8.Scale bar,2m m.Expected incidence of X chromosomes with two or more COSA-1foci was estimated to be R 36%based on frequencies of 2-CO and 3-CO products detected by genetic assay (Tsai et al.,2008).See also Table S1.

Cell 149,75–87,March 30,2012a2012Elsevier Inc.83

foci at the sites of excess COs in the rtel-1mutant.Nevertheless, several lines of evidence suggest that the extra COs in rtel-1 mutants still require the canonical meiotic CO pathway for their formation.First,analysis of rtel-1;cosa-1double mutants showed that the rtel-1mutation does not rescue the lack of chiasmata or COs caused by loss of COSA-1(Figure7B and Table S1).Moreover,the elevation in CO frequency caused by loss of rtel-1function was partially suppressed in rtel-1/rtel-1; cosa-1/+worms,whereas cosa-1heterozygosity had no effect in the wild-type background(Figure7C).Together,these obser-vations suggest that the excess COs in rtel-1mutants are formed using the canonical COSA-1/MSH-5-dependent CO pathway (rather than an alternative pathway)despite the fact that cytolog-ically detectable COSA-1foci do not form at the extra CO sites. These?ndings are consistent with a mutually antagonistic rela-tionship between RTEL-1and COSA-1/MSH-5at prospective recombination sites.COSA-1/MSH-5may protect CO intermedi-ates from RTEL-1-mediated disassembly at CO-designated sites,whereas RTEL-1may be needed to ef?ciently disengage excess COSA-1/MSH-5-dependent intermediates at sites that were not designated to become COs.Thus,absence of RTEL-1could allow such intermediates to progress to CO products without accumulating CO proteins at high concentration.

dpy-28Mutant

DPY-28is a subunit of the condensin I complex,which associ-ates with meiotic prophase chromosomes and in?uences both the length of the chromosomes and the number and posi-tioning of DSBs(Mets and Meyer,2009;Tsai et al.,2008). Previous genetic analysis revealed a high incidence of meiotic products with two or more COs on the X chromosomes in the dpy-28(s939)mutant.However,quantitation of COSA-1 foci speci?cally on the X chromosomes in the dpy-28mutant (Figure7D)did not reveal a commensurate incidence of X chro-mosome pairs with multiple COSA-1foci.Whereas R36%of chromosome pairs are expected to have two or more COs based on the genetic data of Tsai et al.(2008),39of39X chromosome pairs analyzed had only a single COSA-1focus(p<0.0001). Thus,despite the processing of extra intermediates into COs, interference is nevertheless still operating in the dpy-28mutant to limit cytologically differentiated CO sites to one per homolog pair in most meioses.

DISCUSSION

The identi?cation of COSA-1as an integral component of a con-served CO-promoting functional module highlights the utility of C.elegans for the discovery of components of the meiotic machinery.Moreover,our ability to visualize COSA-1as an in vivo marker of CO-designated sites provides an opportunity to inves-tigate how prospective CO sites progressively differentiate,how these events are coupled to meiotic progression,and how events are regulated both to ensure the formation of COs and to limit their numbers.

Analysis of Pro-CO Proteins Suggests a Two-Step Process for CO Speci?cation

The localization dynamics of pro-CO factors during wild-type meiosis and in response to IR-induced DSBs support a model in which COs are speci?ed via a two-step process.Together, the data imply the existence of an initial‘‘CO licensing’’step, presumably involving de novo loading of MSH-5to prospective recombination sites in excess of eventual COs,that must occur during midpachytene as a prerequisite for later concentration of COSA-1at a single presumptive CO site per chromosome pair. Eligibility for de novo MSH-5loading shuts down at the mid-to-late pachytene transition,coincident with a MAP kinase-depen-dent transition in the mode of DSB repair(Hayashi et al.,2007). Further,subsequent concentration of COSA-1,MSH-5,and ZHP-3at six sites following the mid-to-late pachytene transition implies that licensing is followed by a second‘‘CO designation’’step that limits the number of licensed sites that can ultimately mature into COs.Although our data do not distinguish whether CO designation occurs at or just prior to the mid-to-late pachy-tene transition,it is clear that progression into late pachytene is required for its manifestation as cytologically differentiated COSA-1-marked sites.

Integral to this two-step model for CO speci?cation is that CO designation occurs after SC assembly,in agreement with recent ?ndings of Henzel et al.(2011)and Rosu et al.(2011).This is illus-trated both by the progressive differentiation of CO sites during wild-type meiosis and by the fact that DSBs induced in the context of assembled SCs are successful in maturing into COSA-1-marked sites in a regulated manner.Although CO designation in S.cerevisiae may be complete prior to SC assembly,we speculate that CO designation after SC assembly may be a widespread feature of metazoan meiosis,as MSH4foci in mice and humans occur in substantial excess of COs and then diminish to a number approaching the number of COs only after completion of synapsis(Baudat and de Massy,2007).

COSA-1Structure Suggests a Mechanism for Reinforcement of CO Designation

Several lines of evidence suggest that CO designation is main-tained by a self-reinforcing mechanism.First,although MSH-5 foci decrease in number upon COSA-1installation,the increased brightness of the remaining foci suggests an increase in the local concentration of MSH-5at COSA-1-marked sites.Second,COSA-1and MSH-5have a propensity to colocalize in aggregates even in the absence of appropriate substrates(i.e.,nascent recombination events).These observa-tions suggest a self-reinforcing property that could simulta-neously result in both enrichment of pro-CO factors at CO-designated sites and their depletion from sites elsewhere on the chromosome.Third,the identity of COSA-1as cyclin-related protein suggests a speci?c model for self-reinforce-ment.COSA-1may partner with a cyclin-dependent kinase (CDK)family member to form a dedicated COSA-1/CDK complex that promotes CO progression.Moreover, C.elegans MSH-5represents a likely target substrate for this proposed CDK activity,as it has15potential CDK phosphory-lation sites([S/T]P)and can be phosphorylated by a cyclin/CDK complex in vitro(Figure S6).We propose that,upon transition to late pachytene,MSH-5localized to CO-eligible sites could recruit COSA-1/CDK,which in turn could initiate a positive feedback loop that results in increased local concentration of both components.

84Cell149,75–87,March30,2012a2012Elsevier Inc.

The possibility that a CDK might play a direct role in CO progression had been suggested by the observation that CDK2colocalizes with MLH1foci in late-pachytene mouse meiocytes(Ashley et al.,2001).However,the functional signi?-cance of this localization with respect to CO formation was unknown,as apoptosis of Cdk2à/àspermatocytes prior to mid-pachytene precluded assessment of potential later roles(Ortega et al.,2003;Viera et al.,2009).Ward and colleagues(Ward et al., 2007)speculated that destruction of a B-type cyclin mediated by E3ubiquitin ligase HEI10could free CDK2to localize at CO sites, presumably in partnership with a distinct cyclin.We suggest that mammalian CDK2may indeed act locally at nascent CO sites, possibly partnered with COSA1/CNTD1.

COSA-1Foci Illuminate Robust CO Control during

C.elegans Meiosis

Previous work analyzing the meiotic behavior of fusion chromo-somes in C.elegans demonstrated the operation of a highly effective CO interference system that normally limits COs to one per homolog pair.This prior analysis probed the system by doubling the size of a chromosome,showing that the unusually large chromosome pair was nevertheless limited to a single CO in the majority of meioses(Hillers and Villeneuve,2003).This ?nding showed that the frequency of COs per Mb was malleable but did not address whether this re?ected a change in DSB number or a change in the fraction of DSBs converted into COs.Here,we altered the substrate for the CO control machinery in a complementary manner,by varying levels of IR-induced DSBs in a background that lacks endogenous DSBs,and used COSA-1foci as a readout for the response. The highly nonlinear response of COSA-1foci to IR dose high-lights the robustness of the C.elegans CO control system.First, our data and modeling provide insight into the relationship between DSBs and CO assurance,demonstrating that the C.elegans CO control system is highly ef?cient at converting DSBs into COSA-1-marked sites when DSBs are present in limiting quantities.This?nding dovetails with the recent?nding of Rosu et al.(2011)that CO formation is the preferred outcome of meiotic DSB repair when DSBs are limiting.High ef?ciency conversion of a single DSB into a COSA-1-marked CO implies that an average of four randomly distributed DSBs per chromo-some pair would suf?ce to ensure a CO for>98%of chromo-some pairs,raising the possibility that CO assurance during normal C.elegans meiosis could potentially be achieved largely through a random distribution of a modest number of DSBs. Another striking feature of the dose-response analysis is that COSA-1foci plateau at six per nucleus,even under conditions in which DSBs are5-or10-fold more abundant than the number needed to ensure a chiasma for each homolog pair in nearly all nuclei.This implies that the interference mechanism limiting the number of cytologically differentiated CO sites per bivalent is nearly impervious to a substantial excess of potential initiating events.Recently,the term‘‘CO homeostasis’’was coined to describe the ability to maintain a?xed number of CO events in the face of signi?cant reduction in DSB levels(Martini et al., 2006).Our?nding that C.elegans germ cells maintain a?xed number of COSA-1foci in the face of a substantial excess of DSBs demonstrates homeostasis operating in the opposite direction,supporting the notion that interference and homeo-stasis are inextricably linked manifestations of the same under-lying CO control mechanism.Moreover,the fact that we were able to recapitulate the entire spectrum of the dose response by a highly constrained mathematical model assuming complete CO assurance and complete CO interference indicates that these two properties can fully explain CO homeostasis in this system.

Concordance between the DSB yield inferred from our dose-response analysis and the established estimate for mammalian cells further indicates that our system can be exploited as a‘‘bio-logical dosimeter’’for DSB formation.We can use COSA-1foci as a quantitative readout of DSB yield not only to estimate DSB number in meiotic mutants,but also to evaluate ef?cacy of treatments that either provoke or inhibit DSB formation. GFP::COSA-1as a Tool to Investigate CO Interference How CO designation at one site is communicated along a chromosome pair to decommission other prospective CO sites remains a major mystery.Although the phenomenon of CO inter-ference has fascinated geneticists since its discovery a century ago,its mechanism has remained remarkably recalcitrant to experimental elucidation.

A major confounding factor in studying interference is that genetic assessment of COs can prove to be an ironically unreli-able readout for evaluating the status of CO interference,espe-cially in mutant situations.There are(at least)two pathways by which COs can be generated during meiosis(de los Santos et al.,2003),commonly referred to as class I COs,which are subject to and confer interference,and class II COs,which are interference insensitive.Thus,lack of interference among residual COs in a mutant may simply re?ect loss of events that are subject to interference.Conversely,an increase in CO number might re?ect loss of a downstream effector that func-tions to facilitate non-CO repair after CO interference has been established,as is likely the case for the C.elegans rtel-1mutant (Youds et al.,2010and this work).Cytological markers of presumptive CO sites(e.g.,chiasmata,recombination nodules, and MLH1foci)represent useful alternatives for evaluating CO distribution and interference(Anderson et al.,1999;Carpenter, 1975;Jones,1984;Zickler et al.,1992).However,methods for detecting these features are not readily adapted to serve as the basis of genetic screens.

Our analyses suggest that GFP::COSA-1foci may represent a crucial new experimental foothold for analyzing interference. Numbers of foci are robustly maintained even in the context of DSB levels that likely exceed normal levels by10-fold,a level at which previously reported SNP analysis would predict the formation of some2-CO meiotic products(Mets and Meyer, 2009).Further,the ability to limit GFP::COSA-1-marked sites to one per chromosome pair is retained in some mutants in which excess COs were detected by genetic assays.Thus, GFP::COSA-1foci appear to represent a more reliable surrogate than COs per se for the chromosomal events that are actually being sensed and/or distributed by the interference mechanism. Moreover,GFP::COSA-1foci are readily visualized in live worms. This creates an unprecedented opportunity to identify factors that contribute to the CO interference mechanism,making it Cell149,75–87,March30,2012a2012Elsevier Inc.85

possible to screen directly for impaired interference by screening for altered numbers of COSA-1foci.

EXPERIMENTAL PROCEDURES

IR time course and dose-response experiments were conducted at25 C;all other experiments were performed at20 C.Experiments using GFP::COSA-1 also included the wild-type cosa-1gene unless otherwise noted.Meiotic mutant homozygotes were derived from balanced heterozygotes by selecting progeny lacking a dominant marker associated with the balancer.

FISH was performed as in Dernburg et al.(1998).Immunostaining was performed as in(Martinez-Perez and Villeneuve,2005)with modi?cations. Dissections were performed on24hr post-L4adults unless otherwise indi-cated.Except where noted,images are projections through3D data stacks encompassing whole nuclei.

Lists of strains and antibodies and details of image acquisition are provided in the Supplemental Information.

Quantitation of GFP::COSA-1Foci

Foci were quanti?ed from deconvolved3D data stacks;only nuclei completely contained within the stack were scored.Nuclei with features indicative of apoptosis(compact and DAPI-bright)were excluded.In unirradiated gonads (Figures3and7),foci were counted in the last?ve rows of pachytene nuclei. In irradiated gfp::cosa-1;spo-11gonads,foci were counted within a three-to ?ve-row zone of late-pachytene nuclei bounded by one row of nuclei with consistent GFP::COSA-1staining on either side.Numbers of nuclei scored for each strain are:gfp::cosa-1(76),mnT12fusion chromosome(150),rtel-1 (138),rtel-1;mnT12(113),and dpy-28(79).

In the dpy-28mutant,foci were also quanti?ed speci?cally on the X chromo-some pair.A HIM-8antibody was used to mark the X-PCs,and axis marker HTP-3was used to trace the chromosome paths.Three-dimensional(3D) volume renderings were generated using Volocity5.5.1to allow rotation of images.X chromosome-associated COSA-1foci were quanti?ed only in nuclei in which X chromosome paths could be traced unambiguously,i.e.,they had a HIM-8focus near one end and exhibited a continuously traceable HTP-3 path that did not intersect with any other HTP-3tracks.Forty-nine percent of late-pachytene nuclei examined met these criteria.

Irradiation Experiments

gfp::cosa-1;spo-11worms were cultured at25 C to increase expression of GFP::COSA-1.g irradiation was performed at18hr post-L4using a Cs-137 source.Unirradiated controls were dissected with the?rst experimental time point.For time course experiments,worms were exposed to1kRad of g-irra-diation and then dissected at various time points.For dose-response experi-ments,worms were dissected8hr after exposure to various IR doses. Numbers of nuclei scored for each dose:0rad(63),100rad(68),250rad (93),500rad(64),1kRad(41),5kRad(83),and10kRad(63).

SUPPLEMENTAL INFORMATION

Supplemental Information includes Extended Experimental Procedures, six?gures,and two tables and can be found with this article online at doi:10.1016/j.cell.2012.01.052.

ACKNOWLEDGMENTS

We thank S.Mitani(National Institute of Genetics,Japan)and S.Rosu(Ville-neuve lab)for cosa-1(tm3298)and zhp-3(me95);N.Bhalla,A.Dernburg,https://www.wendangku.net/doc/9f14855029.html, Volpe,B.Meyer,S.Boulton,M.Zetka,and the Caenorhabditis Genetics Center for strains and antibodies;and J.Bessler for technical support.We thank A.D.Kaiser,M.AlQuaraishi,C.Janda,C.Thomas,D.Riordan,S.Clarke, and members of the Villeneuve lab for discussions and D.Libuda and A.Fire for comments on the manuscript.This work was supported by Stanford Genome Training Program NIH HG00044Training Grant for R.Y.,Leukemia and Lymphoma Society Postdoctoral Fellowship5381-12to K.A.Z.,and NIH grant R01GM67268to A.M.V.Received:May31,2011

Revised:October27,2011

Accepted:January15,2012

Published:March29,2012

REFERENCES

Agarwal,S.,and Roeder,G.S.(2000).Zip3provides a link between recombi-nation enzymes and synaptonemal complex proteins.Cell102,245–255.

Anderson,L.K.,Reeves,A.,Webb,L.M.,and Ashley,T.(1999).Distribution of crossing over on mouse synaptonemal complexes using immuno?uorescent localization of MLH1protein.Genetics151,1569–1579.

Ashley,T.,Walpita,D.,and de Rooij,D.G.(2001).Localization of two mamma-lian cyclin dependent kinases during mammalian meiosis.J.Cell Sci.114, 685–693.

Barber,L.J.,Youds,J.L.,Ward,J.D.,McIlwraith,M.J.,O’Neil,N.J.,Petalcorin, M.I.,Martin,J.S.,Collis,S.J.,Cantor,S.B.,Auclair,M.,et al.(2008).RTEL1 maintains genomic stability by suppressing homologous recombination.Cell 135,261–271.

Baudat,F.,and de Massy,B.(2007).Regulating double-stranded DNA break repair towards crossover or non-crossover during mammalian meiosis.Chro-mosome Res.15,565–577.

Bhalla,N.,Wynne,D.J.,Jantsch,V.,and Dernburg,A.F.(2008).ZHP-3acts at crossovers to couple meiotic recombination with synaptonemal complex disassembly and bivalent formation in C.elegans.PLoS Genet.4,e1000235.

Carpenter,A.T.(1975).Electron microscopy of meiosis in Drosophila mela-nogaster females:II.The recombination nodule—a recombination-associated structure at pachytene?https://www.wendangku.net/doc/9f14855029.html,A72,3186–3189.

Colaia′covo,M.P.,MacQueen,A.J.,Martinez-Perez,E.,McDonald,K.,Adamo, A.,La Volpe,A.,and Villeneuve,A.M.(2003).Synaptonemal complex assembly in C.elegans is dispensable for loading strand-exchange proteins but critical for proper completion of recombination.Dev.Cell5,463–474.

de los Santos,T.,Hunter,N.,Lee,C.,Larkin,B.,Loidl,J.,and Hollingsworth, N.M.(2003).The Mus81/Mms4endonuclease acts independently of double-Holliday junction resolution to promote a distinct subset of crossovers during meiosis in budding yeast.Genetics164,81–94.

Dernburg,A.F.,McDonald,K.,Moulder,G.,Barstead,R.,Dresser,M.,and Villeneuve, A.M.(1998).Meiotic recombination in C.elegans initiates by a conserved mechanism and is dispensable for homologous chromosome synapsis.Cell94,387–398.

Hammarlund,M.,Davis,M.W.,Nguyen,H.,Dayton,D.,and Jorgensen,E.M. (2005).Heterozygous insertions alter crossover distribution but allow cross-over interference in Caenorhabditis elegans.Genetics171,1047–1056.

Hayashi,M.,Chin,G.M.,and Villeneuve,A.M.(2007).C.elegans germ cells switch between distinct modes of double-strand break repair during meiotic prophase progression.PLoS Genet.3,e191.

Henzel,J.V.,Nabeshima,K.,Schvarzstein,M.,Turner,B.E.,Villeneuve,A.M., and Hillers,K.J.(2011).An asymmetric chromosome pair undergoes synaptic adjustment and crossover redistribution during Caenorhabditis elegans meiosis:implications for sex chromosome evolution.Genetics187,685–699.

Higgins,J.D.,Armstrong,S.J.,Franklin,F.C.,and Jones,G.H.(2004).The Arabidopsis MutS homolog AtMSH4functions at an early step in recombina-tion:evidence for two classes of recombination in Arabidopsis.Genes Dev. 18,2557–2570.

Higgins,J.D.,Vignard,J.,Mercier,R.,Pugh,A.G.,Franklin,F.C.,and Jones, G.H.(2008).AtMSH5partners AtMSH4in the class I meiotic crossover pathway in Arabidopsis thaliana,but is not required for synapsis.Plant J.55, 28–39.

Hillers,K.J.,and Villeneuve,A.M.(2003).Chromosome-wide control of meiotic crossing over in C.elegans.Curr.Biol.13,1641–1647.

Hodgkin,J.,Horvitz,H.R.,and Brenner,S.(1979).Nondisjunction Mutants of the Nematode CAENORHABDITIS ELEGANS.Genetics91,67–94.

86Cell149,75–87,March30,2012a2012Elsevier Inc.

Hollingsworth,N.M.,Ponte,L.,and Halsey,C.(1995).MSH5,a novel MutS homolog,facilitates meiotic reciprocal recombination between homologs in Saccharomyces cerevisiae but not mismatch repair.Genes Dev.9,1728–1739.

Jantsch,V.,Pasierbek,P.,Mueller,M.M.,Schweizer,D.,Jantsch,M.,and Loidl,J.(2004).Targeted gene knockout reveals a role in meiotic recombina-tion for ZHP-3,a Zip3-related protein in Caenorhabditis elegans.Mol.Cell. Biol.24,7998–8006.

Jeffrey,P.D.,Russo,A.A.,Polyak,K.,Gibbs,E.,Hurwitz,J.,Massague,J.,and Pavletich,N.P.(1995).Mechanism of CDK activation revealed by the structure of a cyclinA-CDK2complex.Nature376,313–320.

Jones,G.H.(1984).The control of chiasma distribution.Symp.Soc.Exp.Biol. 38,293–320.

Kelly,K.O.,Dernburg, A.F.,Stan?eld,G.M.,and Villeneuve,A.M.(2000). Caenorhabditis elegans msh-5is required for both normal and radiation-induced meiotic crossing over but not for completion of meiosis.Genetics 156,617–630.

Kneitz,B.,Cohen,P.E.,Avdievich,E.,Zhu,L.,Kane,M.F.,Hou,H.,Jr.,Kolod-ner,R.D.,Kucherlapati,R.,Pollard,J.W.,and Edelmann,W.(2000).MutS homolog4localization to meiotic chromosomes is required for chromosome pairing during meiosis in male and female mice.Genes Dev.14,1085–1097.

Lenzi,M.L.,Smith,J.,Snowden,T.,Kim,M.,Fishel,R.,Poulos,B.K.,and Cohen,P.E.(2005).Extreme heterogeneity in the molecular events leading to the establishment of chiasmata during meiosis i in human oocytes.Am.J. Hum.Genet.76,112–127.

Lynn,A.,Soucek,R.,and Bo¨rner,G.V.(2007).ZMM proteins during meiosis: crossover artists at work.Chromosome Res.15,591–605.

Macqueen,A.J.,and Roeder,G.S.(2009).Fpr3and Zip3ensure that initiation of meiotic recombination precedes chromosome synapsis in budding yeast. Curr.Biol.19,1519–1526.

MacQueen,A.J.,Colaia′covo,M.P.,McDonald,K.,and Villeneuve,A.M.(2002). Synapsis-dependent and-independent mechanisms stabilize homolog pair-ing during meiotic prophase in C.elegans.Genes Dev.16,2428–2442.

MacQueen,A.J.,Phillips,C.M.,Bhalla,N.,Weiser,P.,Villeneuve,A.M.,and Dernburg,A.F.(2005).Chromosome sites play dual roles to establish homolo-gous synapsis during meiosis in C.elegans.Cell123,1037–1050.

Martinez-Perez,E.,and Villeneuve,A.M.(2005).HTP-1-dependent constraints coordinate homolog pairing and synapsis and promote chiasma formation during C.elegans meiosis.Genes Dev.19,2727–2743.

Martinez-Perez, E.,and Colaia′covo,M.P.(2009).Distribution of meiotic recombination events:talking to your neighbors.Curr.Opin.Genet.Dev.19, 105–112.

Martinez-Perez,E.,Schvarzstein,M.,Barroso,C.,Lightfoot,J.,Dernburg, A.F.,and Villeneuve,A.M.(2008).Crossovers trigger a remodeling of meiotic chromosome axis composition that is linked to two-step loss of sister chro-matid cohesion.Genes Dev.22,2886–2901.

Martini,E.,Diaz,R.L.,Hunter,N.,and Keeney,S.(2006).Crossover homeo-stasis in yeast meiosis.Cell126,285–295.

Mets,D.G.,and Meyer,B.J.(2009).Condensins regulate meiotic DNA break distribution,thus crossover frequency,by controlling chromosome structure. Cell139,73–86.

Muller,H.J.(1916).The mechanism of crossing over.Am.Nat.50,193–221, 284–305,350–366,421–434.

Nabeshima,K.,Villeneuve,A.M.,and Hillers,K.J.(2004).Chromosome-wide regulation of meiotic crossover formation in Caenorhabditis elegans requires properly assembled chromosome axes.Genetics168,1275–1292.Ortega,S.,Prieto,I.,Odajima,J.,Mart?′n,A.,Dubus,P.,Sotillo,R.,Barbero, J.L.,Malumbres,M.,and Barbacid,M.(2003).Cyclin-dependent kinase2is essential for meiosis but not for mitotic cell division in mice.Nat.Genet.35, 25–31.

Petri,E.T.,Errico,A.,Escobedo,L.,Hunt,T.,and Basavappa,R.(2007).The crystal structure of human cyclin B.Cell Cycle6,1342–1349.

Phillips,C.M.,Wong,C.,Bhalla,N.,Carlton,P.M.,Weiser,P.,Meneely,P.M., and Dernburg,A.F.(2005).HIM-8binds to the X chromosome pairing center and mediates chromosome-speci?c meiotic synapsis.Cell123,1051–1063.

Pochart,P.,Woltering,D.,and Hollingsworth,N.M.(1997).Conserved proper-ties between functionally distinct MutS homologs in yeast.J.Biol.Chem.272, 30345–30349.

Ross-Macdonald,P.,and Roeder,G.S.(1994).Mutation of a meiosis-speci?c MutS homolog decreases crossing over but not mismatch correction.Cell79, 1069–1080.

Rosu,S.,Libuda,D.E.,and Villeneuve,A.M.(2011).Robust crossover assur-ance and regulated interhomolog access maintain meiotic crossover number. Science334,1286–1289.

Santucci-Darmanin,S.,Walpita,D.,Lespinasse,F.,Desnuelle,C.,Ashley,T., and Paquis-Flucklinger,V.(2000).MSH4acts in conjunction with MLH1during mammalian meiosis.FASEB J.14,1539–1547.

Schurko,A.M.,Mazur,D.J.,and Logsdon,J.M.,Jr.(2010).Inventory and phy-logenomic distribution of meiotic genes in Nasonia vitripennis and among diverse arthropods.Insect Mol.Biol.19(Suppl1),165–180.

Snowden,T.,Acharya,S.,Butz,C.,Berardini,M.,and Fishel,R.(2004). hMSH4-hMSH5recognizes Holliday Junctions and forms a meiosis-speci?c sliding clamp that embraces homologous chromosomes.Mol.Cell15, 437–451.

Thompson,L.H.,and Limoli,C.L.(2000).Origin,Recognition,Signaling and Repair of DNA Double-Strand Breaks in Mammalian Cells(Austin,TX: Madame Curie Bioscience Database).

Tsai,C.J.,Mets,D.G.,Albrecht,M.R.,Nix,P.,Chan,A.,and Meyer,B.J.(2008). Meiotic crossover number and distribution are regulated by a dosage compen-sation protein that resembles a condensin subunit.Genes Dev.22,194–211.

Viera,A.,Rufas,J.S.,Mart?′nez,I.,Barbero,J.L.,Ortega,S.,and Suja,J.A. (2009).CDK2is required for proper homologous pairing,recombination and sex-body formation during male mouse meiosis.J.Cell Sci.122,2149–2159.

Ward,J.O.,Reinholdt,L.G.,Motley,W.W.,Niswander,L.M.,Deacon,D.C., Grif?n,L.B.,Langlais,K.K.,Backus,V.L.,Schimenti,K.J.,O’Brien,M.J., et al.(2007).Mutation in mouse hei10,an e3ubiquitin ligase,disrupts meiotic crossing over.PLoS Genet.3,e139.

Youds,J.L.,Mets,D.G.,McIlwraith,M.J.,Martin,J.S.,Ward,J.D.,ONeil,N.J., Rose, A.M.,West,S.C.,Meyer, B.J.,and Boulton,S.J.(2010).RTEL-1 enforces meiotic crossover interference and homeostasis.Science327, 1254–1258.

Zalevsky,J.,MacQueen,A.J.,Duffy,J.B.,Kemphues,K.J.,and Villeneuve, A.M.(1999).Crossing over during Caenorhabditis elegans meiosis requires a conserved MutS-based pathway that is partially dispensable in budding yeast.Genetics153,1271–1283.

Zetka,M.C.,Kawasaki,I.,Strome,S.,and Mu¨ller,F.(1999).Synapsis and chiasma formation in Caenorhabditis elegans require HIM-3,a meiotic chro-mosome core component that functions in chromosome segregation.Genes Dev.13,2258–2270.

Zickler,D.,Moreau,P.J.,Huynh,A.D.,and Slezec,A.M.(1992).Correlation between pairing initiation sites,recombination nodules and meiotic recombi-nation in Sordaria macrospora.Genetics132,135–148.

Cell149,75–87,March30,2012a2012Elsevier Inc.87