© 2004 Nature Publishing Group

776 | OCTOBER 2004 | VOLUME 5 www.nature.com/reviews/molcellbio

R E S E A R C H H I G H L I G H T S

The extremes of linear chromosomesare dangerous locations. Their struc-tural resemblance to double-strandedDNA breaks can trigger unwantedDNA-damage-response pathwayswith catastrophic results for the cell.Telomeric DNA is thereforeenveloped by a ‘cap’ of protective pro-teins that conceal these troublesomeends. Now, researchers have shed newlight on the function of one of thesetelomere-binding proteins — TRF2.

TRF2 is a telomeric double-stranded-DNA-binding protein thathas a role in the protection of chro-mosome ends, but its mechanism ofaction is unknown. A dominant-neg-ative mutation of TRF2 that displacesthe protein from its binding site leadsto telomeric association of the ATMkinase — a critical mediator of theDNA-damage-response pathway.These findings indicate that TRF2might oppose the action or therecruitment of ATM.

To investigate this further, Titiade Lange and co-workers overex-pressed TRF2 in human primary

fibroblasts, exposed these cells to ion-izing radiation (IR) and then analysedthe ATM-mediated response to DNAdamage. Microscopy analysis revealedthat an increased percentage of thesecells had entered mitosis — indicatinga failure of ATM-mediated cell-cyclearrest. Quantitative immunoblotsshowed decreased levels of the p53protein and its downstream targets,and ATM-dependent phosphoryla-tion of NBS1 — a DNA-double-strand-break-repair protein — wasalso impaired in these cells.

DNA damage that is induced bylow levels of IR is usually associatedwith activation of ATM throughautophosphorylation. However, whenthe authors co-expressed TRF2 andATM, the level of phosphorylatedATM after IR exposure was reducedcompared with controls. Similarly,endogenous ATM phosphorylationwas reduced in cells that overex-pressed TRF2.

So, does TRF2 directly interferewith ATM activation and function?Anti-ATM antibody co-precipitated

In mammals, DNA and histone methylationtogether provide an effective, long-termmechanism for silencing gene expression, buthow specific methylation patterns are‘remembered’ during cell division is unclear.In a recent paper, Sarraf and Stanchevashowed that this depends on the coupling ofthe two types of methylation during DNAreplication.

At sites of constitutive heterochromatinand transcriptionally silenced promoters,silencing is mediated by methylation ofDNA at CpG dinucleotides and ofhistone H3 at lysine 9 (H3-K9). DuringDNA replication, the methyltransferaseDNMT1 interacts with the replicationmachinery to ensure that DNA methylationpatterns are faithfully copied. By contrast,little is known about how histonemethylation is reproduced. One modelproposes that this is somehow coordinatedwith DNA methylation, but evidence has sofar been lacking.

By co-immunoprecipitation, Sarraf andStancheva showed that MBD1 — a proteinthat specifically binds methyl-CpG groups —associates with a complex that contains an H3-K9-specific methyltransferase activity,providing a possible link between DNA andhistone methylation. The other components ofthe complex were identified as the H3-K9-specific methyltransferase SETDB1 and CAF1,a protein involved in chromatin assembly. So,MBD1 bound to methylated DNA couldrecruit SETDB1 and, through its interactionwith CAF1, promote H3-K9 methylation atspecific sites during chromatin assembly.

Consistent with this, the three proteinswere shown to form a complex in vivospecifically during DNA replication. Theauthors also showed how DNA replication iscoupled to the activation of theCAF1–MBD1–SETDB1 complex. CAF1 isonly transiently associated with MBD1 andSETDB1 during S-phase and this depends onMBD1 being displaced from DNA. Specificinhibition of replication elongation showedthat this displacement depends on theprogression of the replication complex. Thisseems to knock MBD1 off the DNA strand,allowing it to bind CAF1 and promote H3-K9methylation on newly formed chromatin.

How does this relate to the silencing ofspecific genes? Sarraf and Stanchevaidentified several genomic MBD1-binding

sites, including a CpG island in the p53BP2promoter region. In HeLa cells, p53BP2 isusually transcriptionally silent, and theauthors showed that this depends on bothDNA methylation and methylation at H3-K9.Treatment with an inhibitor of DNAmethylation or with small interfering RNAsagainst either MBD1 or SETDB1 led to loss ofmethylated H3-K9 at the p53BP2 promoterand induced the expression of the gene.Similarly, reducing levels of DNAmethylation or MBD1 expression resulted inloss of methylated H3-K9 at 23 othergenomic MBD1-binding sites.

So, interactions between the DNA andhistone methylation machinery seem to beimportant for maintaining patterns ofepigenetic modification on a widespreadbasis. This provides one way of ensuring thattranscriptional silencing is transmittedaccurately through ongoing rounds of celldivision, an essential requirement for normalmammalian development.

Louisa Flintoft Assistant Editor, Nature Reviews

References and linksORIGINAL RESEARCH PAPERSarraf, S. A. & Stancheva, I. Methyl-CpG binding proteinMBD1 couples histone H3 methylation at lysine 9 bySETDB1 to DNA replication and chromatin assembly. Mol. Cell 15, 595–605 (2004)WEB SITEIrina Stancheva’s laboratory: http://www.bms.ed.ac.uk/services/staff/stancheva/index.htm

Extreme protection

T E LO M E R E S

Silenttransmission

E P I G E N E T I C S

© 2004 Nature Publishing Group

R E S E A R C H H I G H L I G H T S

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | OCTOBER 2004 | 777

Eukaryotic cells can respond to starvation byautophagy — a process whereby the cells recoveressential nutrients by the lysosomal breakdown oforganelles, proteins and other components of thecytoplasm. In response to developmental signals,‘programmed’ autophagy promotes tissue remod-elling and cell death. Reporting in twoDevelopmental Cell papers, the groups of Neufeldand Stenmark have now uncovered the signallingpathways that control autophagy in Drosophilamelanogaster.

Both groups used a combination of electronand fluorescent confocal microscopy to assayautophagy in the larval fat body of D. melanogaster,which showed a strong autophagic response whendeprived of nutrients. Stenmark and colleaguesalso studied the various phases of pro-grammed autophagy in the fat body duringthe final larval stage.

TOR (target of rapamycin) kinases, thecentral components of a conserved nutri-ent-sensing pathway, were suspected tofunction in the regulation of autophagy.To find out more, Neufeld and co-workersstudied TOR-null flies and showed that theloss of TOR activity causes the induction ofautophagy, regardless of the nutrient condi-tions. The same autophagic phenotype wasobserved when they expressed negative regula-tors of TOR in fat-body cells. Conversely, the acti-vation of TOR signalling suppressed autophagyunder conditions of nutrient deprivation. So, TORsignalling is both necessary and sufficient to suppressstarvation-induced autophagy.

Components of the phosphatidylinositol 3-kinase(PI3K) signalling pathway, when overexpressed, alsosuppressed autophagy. However, overexpression ofPI3K was unable to inhibit autophagy that wascaused by the loss of TOR function, which indicatesthat PI3K signalling requires TOR activity to sup-press autophagy.

Surprisingly, Neufeld and colleagues found that theinactivation of S6K, which is a central downstreameffector of TOR, did not induce autophagy. In fact, theautophagy phenotype of starved S6K-mutant cells wassignificantly reduced compared with starved wild-typecells. This implies not only that the suppression ofautophagy by the TOR pathway is independent ofS6K, but that S6K is a positive regulator of autophagy.

The Stenmark group also studied the signallingpathways that are responsible for programmedautophagy, and found that the ecdysone hormone,which is important for larval development, triggeredthis event in the D. melanogaster fat body. Using amarker for PI3K activity, Stenmark and colleagues

then showed that programmedautophagy coincided with a reduction in

PI3K activity and could be blocked by the over-expression of components of the PI3K pathway. So,the authors propose that ecdysone signalling down-regulates PI3K signalling, thereby inducing autophagy.

These new insights into the pathways that regulateautophagy raise new questions about its physiologicalrole. For example, does autophagy contribute to thereduced cell growth that occurs when TOR activity islow? The answer seems to be no, as the inhibition ofautophagy increased the severity of the phenotypethat is caused by the loss of TOR signalling in TOR-mutant flies, which included reductions in cell size,growth rate and survival. Neufeld and colleagues con-cluded that, when TOR signalling is reduced,autophagy is required mainly for survival and normalcell metabolism. It will also be interesting to probe therole of TOR during programmed autophagy.

Arianne Heinrichs

References and linksORIGINAL RESEARCH PAPERS Scott, R. C. et al. Role andregulation of starvation-induced autophagy in the Drosophila fat body.Dev. Cell 7, 167–178 (2004) | Rusten, T. E. et al. Programmedautophagy in the Drosophila fat body is induced by ecdysone throughregulation of the PI3K pathway. Dev. Cell 7, 179–192 (2004)

endogenous and exogenous TRF2 inprimary human fibroblasts, whichindicates that these proteins interactin vivo. Pulldown experiments usingfusion proteins that comprised differ-ent fragments of ATM showed thatTRF2 bound to a specific domain ofATM, close to serine 1981 — the mainsite of autophosphorylation. Finally,immunofluorescence of IR-treatedprimary fibroblasts that overexpressedTRF2 detected the protein at telom-eric foci, but not at chromosomal sitesof DNA damage.

Based on these results, the authorspropose a model in which TRF2 bindsto ATM and inhibits its activation bypreventing phosphorylation at S1981.Importantly, the subnuclear locationof TRF2 indicates that this inhibitionis restricted to telomeres, which wouldallow ATM to mediate vital DNArepair elsewhere in the nucleus.

Shannon AmoilsReferences and links

ORIGINAL RESEARCH PAPER Karlseder, J. et al.The telomeric protein TRF2 binds the ATM kinaseand can inhibit the ATM-dependent DNA damageresponse. PLoS Biology 2, 1150–1156 (2004)

Breakdown recovery

A U TO P H A G Y