DispatchR883

an mTORC1 substrate that negatively regulatesinsulin signaling. Science 332, 1322–1326.

16. Chung, J., Kuo, C.J., Crabtree, G.R., andBlenis, J. (1992). Rapamycin-FKBP specificallyblocks growth-dependent activation of andsignaling by the 70 kd S6 protein kinases. Cell69, 1227–1236.

17. Zhang, Y., and Zheng, X.F. (2012).mTOR-independent 4E-BP1 phosphorylation isassociated with cancer resistance to mTORkinase inhibitors. Cell Cycle 11, 594–603.

18. Ducker, G.S., Atreya, C.E., Simko, J.P.,Hom, Y.K., Matli, M.R., Benes, C.H., Hann, B.,Nakakura, E.K., Bergsland, E.K., Donner, D.B.,et al. (2013). Incomplete inhibition ofphosphorylation of 4E-BP1 as a mechanism ofprimary resistance to ATP-competitive mTOR

inhibitors. Oncogene http://dx.doi.org/10.1038/onc.2013.92.

19. She, Q.B., Halilovic, E., Ye, Q., Zhen, W.,Shirasawa, S., Sasazuki, T., Solit, D.B., andRosen, N. (2010). 4E-BP1 is a key effector of theoncogenic activation of the AKT and ERKsignaling pathways that integrates theirfunction in tumors. Cancer Cell 18,39–51.

20. Shin, S., Wolgamott, L., Tcherkezian, J.,Vallabhapurapu, S., Yu, Y., Roux, P.P., andYoon, S.O. (2013). Glycogen synthasekinase-3beta positively regulates proteinsynthesis and cell proliferation through theregulation of translation initiation factor4E-binding protein 1. Oncogene http://dx.doi.org/10.1038/onc.2013.113.

1Department of Cancer and Cell Biology,University of Cincinnati College of Medicine,Cincinnati, OH 45267, USA. 2Institute forResearch in Immunology and Cancer (IRIC),Universite de Montreal, Montreal,Quebec H3C 3J7, Canada. 3Department ofPathology and Cell Biology, Faculty ofMedicine, Universite de Montreal, Montreal,Quebec, H3C 3J7, Canada.E-mail: yoonsh@ucmail.uc.edu, philippe.roux@umontreal.ca

http://dx.doi.org/10.1016/j.cub.2013.08.030

Chromosome Segregation: Learningto Let Go

To ensure accurate chromosome segregation, cohesion between sisterchromatids must be released in a controlled manner during mitosis. A newstudy reveals how distinct centromere populations of the cohesin protectorSgo1 are regulated by microtubule attachments, cyclin-dependent kinases,and the kinetochore kinase Bub1.

Jonathan M.G. Higgins

Dividing cells must convey the correctcomplement of chromosomes to theiroffspring. Eukaryotes accomplish thisby maintaining cohesion betweenreplicated sister chromatids untilchromosomes are bi-oriented on themitotic spindle. Only once this hasbeen accomplished are theattachments between chromatidsreleased, allowing them to be sortedaccurately to opposite poles of thedividing cell. Clearly then, althoughsister chromatids may be inseparableat first, they must learn to let go whenthe time comes. A report from Liu, Jiaand Yu in this issue of Current Biology[1] provides new insight into thisprocess that may have broaderimplications for our understanding ofinner centromere function.

Cohesion between sister chromatidsis maintained by cohesin complexes,together with regulators such asSororin [2]. In vertebrate mitosis,cohesin is removed fromchromosomes in two steps. Inprophase, a mechanism involvingphosphorylation of cohesin and Sororinby mitotic kinases removes the bulk ofcohesin from chromosome arms(Figure 1). Cohesin at centromeres,however, is protected by Sgo1–PP2Aphosphatase complexes thatcounteract phosphorylation of cohesin

and Sororin [3–5]. To fully separatechromatids at anaphase, the remainingcohesin is cleaved by the proteaseSeparase [2]. This raises the questionof how cleavage of centromericcohesin is limited to anaphase. Asimple possibility is that Separase onlybecomes active at anaphase, and thatSgo1 does not protect cohesin fromcleavage in mitosis. However, it hasbeen reported that Sgo1, wheninappropriately maintained at innercentromeres, preventsSeparase-mediated cohesin cleavage[6]. Also, at least in budding yeast,Sgo1–PP2A complexes may inhibitSeparase more directly [7]. Therefore,it is important to understand how thelocalization and activity of Sgo1 areregulated.

During prophase in mammalian cells,Sgo1 is found at inner centromeres(defined here as the area between thechromatin regions that containcentromeric histone CENP-A; Figure 1).As chromosomes become bi-oriented,Sgo1 appears to move outwards,relocating to two regions roughlycoinciding with CENP-A-containingchromatin underlying kinetochores[1,6,8]. This movement of Sgo1 awayfrom cohesin complexes located atinner centromeres might rendercohesin susceptible to cleavage bySeparase, and would provide a way tomake removal of cohesin favorable

only when chromosomes are correctlybi-oriented and microtubules exerttension across sister kinetochores [6].How this relocation of Sgo1 iscontrolled, however, has beenunknown.A number of ways to recruit Sgo1

to centromeres have been reported,but the relative contributions of thesepathways are debated. It is widelyaccepted that Sgo1 is brought tocentromeres when histone H2A isphosphorylated at Thr-120(H2AT120ph) by the kinetochore kinaseBub1 [9,10], though the structural basisfor this recruitment is unknown. Sgo1can also bind to the heterochromatinprotein HP1, which itself bindschromatin by recognizing histone H3trimethylated on Lys-9 (H3K9me3) [11].Although most HP1 is removed fromchromosomes during mitosis, asmall population remains at innercentromeres that could recruit Sgo1.However, other studies have found thatkey H3K9 methyltransferases are notrequired for HP1 or Sgo1 localization inmitosis [12,13], and that HP1 binds tomitotic centromeres via thechromosomal passenger complex(CPC) in a manner that excludes HP1binding to Sgo1 [14]. An alternativepotential contribution to innercentromere Sgo1 localization is bindingto cohesin itself, an interaction thatdepends on phosphorylation of Sgo1 atThr-346 by cyclin-dependent kinases(Cdk) [5]. How do these proposedmechanisms act together to controlSgo1 function?Although the dependency of Sgo1

localization on Bub1 activity is largelyunquestioned, the reason thatcentromeric cohesion depends onBub1 is less clear [15,16]. Bub1 is amitotic checkpoint protein, andlowering Bub1 levels might lead to

Sgo1P

Sgo1P

Sgo1 Sgo1 Sgo1Sgo1

SeparaseBi-orientation

Prometaphase Metaphase Early anaphase

CENP-A-containing centromeric chromatin

Kinetochore

Microtubules

H2AT120ph Cohesin

Inner centromere

Sgo1P

Sgo1P

Prophase

Release ofarm cohesin

Current Biology

Figure 1. Model for regulation of Sgo1 localization during mitosis.

During prophase, Sgo1 (green) is phosphorylated at Thr-346 and binds to cohesin complexes(red) at the inner centromere between sister chromatids (pale blue). This inner centromericaccumulation of Sgo1 requires Bub1 and binding to H2AT120ph in an as yet undeterminedmanner (curved arrows). Cohesin on chromosome arms is released through the action ofmitotickinases, but Sgo1 protects inner centromere cohesin. Once the chromosomes becomebi-oriented, Sgo1 is dephosphosphorylated at Thr-346 and Sgo1 no longer binds to cohesin.Instead, Sgo1 redistributes towards H2AT120ph (dark blue), in regions approximatelycoinciding with centromeric chromatin containing CENP-A (yellow). Once the mitoticcheckpoint is satisfied, Separase is activated and cleaves the now unprotected cohesin at innercentromeres. In anaphase, H2AT120ph begins to decline, and Sgo1 is eventually released.

Current Biology Vol 23 No 19R884

cohesion loss because the checkpointis compromised and anaphase isinitiated, rather than because Bub1 andH2AT120ph are required for Sgo1localization [15]. In their new study, Liuet al. acknowledge that inactivation ofBub1 causes a weaker cohesionphenotype than loss of Sgo1 but arguethat centromeric cohesion is flawedwhen Bub1 is depleted, even whenthe checkpoint remains active [1].However, they also find that Sgo1does not always co-localize withH2AT120ph, particularly onchromosomes that lack microtubuleattachments. On such chromosomes,H2AT120ph largely overlaps withCENP-A-containing chromatin atkinetochores whereas Sgo1 is found atinner centromeres (Figure 1). Theseresults are consistent with anadditional contribution to Sgo1localization and function beyond that ofthe Bub1–H2AT120ph pathway.

To determine the relative rolesof the Bub1–H2AT120ph andcohesin-dependent pathways, Liu et al.examined separation-of-functionmutants of Sgo1. Amutant (K492A) thatcould not co-immunoprecipitateH2AT120ph, but still bound cohesin,was no longer enriched at centromeres.Instead, it was found on chromosomearms, consistent with the effects ofdepleting Bub1. In contrast, a mutant(T346A) that could interact with

H2AT120ph but was unable to bindcohesin was found at kinetochores, butwas unable to localize to innercentromeres. Therefore, H2AT120phbinding appears important for allcentromeric enrichment of Sgo1, whilecohesin binding is importantspecifically for the accumulation ofSgo1 at inner centromeres. Notably,Sgo1-T346A (which cannot bindcohesin) was unable to restorecohesion in Sgo1-depleted cells.In contrast, Sgo1-K492A (which retainscohesin binding) was largely, thoughnot fully, able to support cohesion.

The authors propose that these twodifferent binding modes underlie theredistribution of Sgo1 observed duringmitosis. When microtubules weredepolymerized with nocodazole, theinner centromere localization andphosphorylation of Sgo1 at Thr-346were increased, and Sgo1 interactionwith H2AT120ph was decreased.Furthermore, a phospho-mimickingSgo1-T346D mutant was partiallyretained at inner centromeres, evenwhen chromosomes were bi-oriented.Cells expressing this mutant hadincreased numbers of laggingchromosomes in anaphase, consistentwith failure to fully remove cohesinfrom centromeres.

These results led to a model in whichCdk-dependent phosphorylation atThr-346 in prophase allows Sgo1

to bind and protect cohesin at innercentromeres. Bi-orientation ofchromosomes in metaphase leads todephosphorylation of Thr-346, loss ofcohesin binding, and redistribution ofSgo1 toward H2AT120ph at innerkinetochores, where it cannot preventcleavage of inner centromeric cohesinby Separase (Figure 1). Thus,microtubule attachment imposesan orchestrated change in thephosphorylation and binding partnersof Sgo1 to bring about its relocalizationand to regulate cohesion.The findings raise a number of

questions. The model provides amechanism for Sgo1 regulation bytension across bi-orientedchromosomes, but is it really tensionthat triggers Sgo1 relocation, or isstable microtubule attachment tokinetochores sufficient? What makesCdk-dependent phosphorylation ofSgo1 responsive to attachment statusand could kinetochore-bound cyclin B[6,17] play a role? Do these studiesimply that HP1 has no role in Sgo1recruitment? Not necessarily. Onepossibility is that HP1 is important forSgo1 localization prior to, but notduring, mitosis [13,14]. Alternatively,ongoing work suggests that Sgo1 canbe retained at inner centromeres inmitosis by HP1, but that this system iscompromised in a wide range of cancercells (Y. Tanno and Y. Watanabe,personal communication). Thepossibility that commonly studied celllines are defective in certain aspects ofcohesion regulation could underlieother conflicting observations in thefield, including those regarding the roleof Bub1 in cohesion regulation.A significant unresolved issue is why

inner centromeric localization of Sgo1depends on the Bub1–H2AT120phpathway. Recruitment by H2AT120phmight increase the local concentrationof Sgo1 and make binding to nearbycohesin (or HP1) more likely. However,Liu et al. find that Sgo1 does notinteract detectably with H2AT120ph innocodazole-treated cells even though,based on results with the K492Amutant, the ability to interact withH2AT120ph is required for Sgo1 toaccumulate at inner centromeres insimilar conditions [1]. Perhapstransient association with H2AT120phallows Sgo1 to pick up a bindingpartner or modification (such asThr-346 phosphorylation) that isneeded to then bind at innercentromeres.

DispatchR885

Bub1 appears to be a major conduitfor ‘outside-in’ signals from thekinetochore to the inner centromere[18], so these studies are likely to haveimplications beyond cohesion. Inparticular, H2AT120ph generated byBub1 co-operates with another histonemodification, H3T3ph generated byHaspin, to specify the inner centromerelocalization of the CPC [10,19]. Themechanism of this co-operation,however, is incompletely defined. Thenew results from Liu et al. imply thatBub1 and H2AT120ph indirectlyenhance Sgo1 binding to innercentromeres. Inner centromeric Sgo1might then provide direct binding sitesfor the CPC and/or protect cohesin toprovide binding sites for Haspin [10],and could therefore help makeCPC localization sensitive tokinetochore–microtubule attachments[20]. Further work to fully understandhow Bub1 activity enhances the innercentromeric localizationofSgo1 is likelyto provide insight into multiple aspectsof inner centromere function andchromosome segregation in mitosis.

References1. Liu, H., Jia, L., and Yu, H. (2013).

Phospho-H2A and cohesin specify distincttension-regulated Sgo1 pools at kinetochoresand inner centromeres. Curr. Biol. 23,1927–1933.

2. Nasmyth, K., and Haering, C.H. (2009). Cohesin:its roles and mechanisms. Annu. Rev. Genet.43, 525–558.

3. Riedel, C.G., Katis, V.L., Katou, Y., Mori, S.,Itoh, T., Helmhart, W., Galova, M.,Petronczki, M., Gregan, J., Cetin, B., et al.

(2006). Protein phosphatase 2A protectscentromeric sister chromatid cohesion duringmeiosis I. Nature 441, 53–61.

4. Kitajima, T.S., Sakuno, T., Ishiguro, K.,Iemura, S., Natsume, T., Kawashima, S.A., andWatanabe, Y. (2006). Shugoshin collaborateswith protein phosphatase 2A to protectcohesin. Nature 441, 46–52.

5. Liu, H., Rankin, S., and Yu, H. (2013).Phosphorylation-enabled binding ofSGO1-PP2A to cohesin protects sororin andcentromeric cohesion during mitosis. Nat. CellBiol. 15, 40–49.

6. Lee, J., Kitajima, T.S., Tanno, Y., Yoshida, K.,Morita, T., Miyano, T., Miyake, M., andWatanabe, Y. (2008). Unified mode ofcentromeric protection by shugoshin inmammalian oocytes and somatic cells. Nat.Cell Biol. 10, 42–52.

7. Clift, D., Bizzari, F., and Marston, A.L. (2009).Shugoshin prevents cohesin cleavageby PP2A(Cdc55)-dependent inhibitionof separase. Genes Dev. 23,766–780.

8. McGuinness, B.E., Hirota, T., Kudo, N.R.,Peters, J.M., and Nasmyth, K. (2005).Shugoshin prevents dissociation of cohesinfrom centromeres during mitosis in vertebratecells. PLoS Biol. 3, e86.

9. Kawashima, S.A., Yamagishi, Y., Honda, T.,Ishiguro, K., and Watanabe, Y. (2010).Phosphorylation of H2A by Bub1prevents chromosomal instability throughlocalizing shugoshin. Science 327,172–177.

10. Yamagishi, Y., Honda, T., Tanno, Y., andWatanabe, Y. (2010). Two histone marksestablish the inner centromere andchromosome bi-orientation. Science 330,239–243.

11. Yamagishi, Y., Sakuno, T., Shimura, M., andWatanabe, Y. (2008). Heterochromatin linksto centromeric protection by recruitingshugoshin. Nature 455, 251–255.

12. Koch, B., Kueng, S., Ruckenbauer, C.,Wendt, K.S., and Peters, J.M. (2008). TheSuv39h-HP1 histone methylation pathway isdispensable for enrichment and protection ofcohesin at centromeres in mammalian cells.Chromosoma 117, 199–210.

13. Perera, D., and Taylor, S.S. (2010). Sgo1establishes the centromeric cohesion

protection mechanism in G2 before subsequentBub1-dependent recruitment in mitosis. J. CellSci. 123, 653–659.

14. Kang, J., Chaudhary, J., Dong, H., Kim, S.,Brautigam, C.A., and Yu, H. (2011). Mitoticcentromeric targeting of HP1 and its bindingto Sgo1 are dispensable for sister-chromatidcohesion in human cells. Mol. Biol. Cell 22,1181–1190.

15. Perera, D., Tilston, V., Hopwood, J.A.,Barchi, M., Boot-Handford, R.P., andTaylor, S.S. (2007). Bub1 maintains centromericcohesion by activation of the spindlecheckpoint. Dev. Cell 13, 566–579.

16. Ricke, R.M., Jeganathan, K.B., Malureanu, L.,Harrison, A.M., and van Deursen, J.M. (2012).Bub1 kinase activity drives error correctionand mitotic checkpoint control but not tumorsuppression. J. Cell Biol. 199,931–949.

17. Bentley, A.M., Normand, G., Hoyt, J., andKing, R.W. (2007). Distinct sequence elementsof cyclin B1 promote localization to chromatin,centrosomes, and kinetochores during mitosis.Mol. Biol. Cell 18, 4847–4858.

18. Boyarchuk, Y., Salic, A., Dasso, M., andArnaoutov, A. (2007). Bub1 is essential forassembly of the functional inner centromere.J. Cell Biol. 176, 919–928.

19. Wang, F., Ulyanova, N.P., van der Waal, M.S.,Patnaik, D., Lens, S.M.A., and Higgins, J.M.G.(2011). A positive feedback loop involvingHaspin and Aurora B promotes CPCaccumulation at centromeres in mitosis. Curr.Biol. 21, 1061–1069.

20. Salimian, K.J., Ballister, E.R., Smoak, E.M.,Wood, S., Panchenko, T., Lampson, M.A., andBlack, B.E. (2011). Feedback control in sensingchromosome biorientation by the Aurora Bkinase. Curr. Biol. 21, 1158–1165.

Division of Rheumatology, Immunology andAllergy, Brigham and Women’s Hospital,Harvard Medical School, Smith BuildingRoom 538A, 1 Jimmy Fund Way, Boston,MA 02115, USA.E-mail: jhiggins@rics.bwh.harvard.edu

http://dx.doi.org/10.1016/j.cub.2013.08.026

Evolution: Sperm, Cryptic Choice, andthe Origin of Species

In two fruit fly species, in vivo observations of competing sperm reveal howdifferences in sperm size, female behavior and reproductive architecturepromote retention of same-species sperm. Sexual selection continues aftermating and may play an important role in speciation.

Adam K. Chippindale

Populations may diverge into separatespecies when they become physicallyisolated, each adapting to differentenvironments and genetically driftingapart for long periods of time. Butwhen there isn’t complete physicalisolation, the probability of speciationwill be greater if there are mechanismsthat inhibit gene flow betweendiverging populations. Differences in

habitat use, the timing of reproductionand mating preferences that favour likebreeding with like are factors that maypromote speciation. In some species, afemale can successfully mate andproduce offspring with a male from herown species (a ‘conspecific’ male) orwith a male from a closely relatedspecies (a ‘heterospecific’ male). If shewere to mate with both types of malewithin a short time period, their spermwould compete for fertilization

opportunities inside her reproductivetract. In sperm competition, theconspecific male tends to hold afertilization advantage, irrespective ofmating order, whereas in spermcompetition between two conspecificmales, mating order matters. Thishome court advantage in theinterspecific love triangle, calledconspecific sperm precedence,suggests a complicated interactionbetween the two different males’ejaculates and the female reproductivetract in which they compete. Suchpostcopulatory sexual selection isamong the most cryptic of biologicalprocesses known, yet is importantbecause it influences paternity andcan promote the evolution of isolation,driving populations towards newspecies [1]. In this issue of CurrentBiology, Mollie Manier, Scott