REVIEWS CAFI and the yeast-specific histone chaperon (REFS 56, 57). Whereas the presence of the H3K56 e(using neighbouring has been reported in humans, its abundance seems mark as template) OR dilution (not shown) limited and its association with new histone deposi- tion is not documented. Furthermore, homologues of 感感 Rtt106 and the Lys acetyltransferase Rtt109(also known as Katl1), which acts on H3K56, have yet to be identi fied in humans. So, whether H3K56ac or an unidentified modification have similar roles in mammals remains to be investigated. Notably, newly synthesized histones H3 and H4 are present as dimers in pre-deposition complexes with histone chaperones(FIG 2a). Although this principle is clearly established for newly synthesized histones, the fate of parental H3-H4 histone dimers that are thought to be deposited as tetramers on the daughter strands might also have to be reconsidered. The fact that histones Maintenance by default emimodified nucleosome used H3 and H4 exist as stable tetramers in solution in the as a template for further modification) absence of DNA argues against the existence of parental H3-H4 dimers. However, structural data now show that HDAC the association of ASFl, and potentially also p48 and p55, with histones is incompatible with a tetrameric rrying parental marks can be detected in association with asl under conditions in which the helicase and le polymerase are uncoupledsupports the hypothesis that ASFI is involved in tetramer splitting and that it functions as an acceptor of recycled parental dimers Therefore, it is indeed possible that parental tetramers (with their own marks) are split and redistribute onto Maintenance requires interstrand daughter strands as dimers. This affects histone dynamic crosstalk OR switch (not shown at the fork and might produce either mixed tetramers that comprise parental and new dimers(FIG. 2b), or nd nucleosomes that comprise only old histones if paren tal dimers reassociate. This second scenario requires either that the old dimers are held in close contact or away from the new ones, or that some recognition event ensures that the correct old dimers are brought back together in the same particle. The spatial organization of DNA at the fork might facilitate these mechanisms 9 Parental mark C Old H3-H4 dime If modifications on the new histones are guided by New H3-H4 dimer modifications of parental histones, the way in which Figure 3 Fate of old and new H3-H4 dimers and their marks at the fork. Three parental histones are distributed to the daughter strands ill determine the degree of conservation of histone parental mark is recognized by a chromatin-binding protein, or reader protein, that in turn marks. Current models suggest that the distribution recruits a chromatin modifier, or writer protein a Random histone distribution. Parental of both parental and newly synthesized histones onto histone H3 and H4 with marks(unsplit or reassociated dimers) are distributed randomly daughter strands occurs in a random fashion(FIG 3a). To onto daughter strands and chromatin density is restored by the deposition of new H3-H4 avoid the dilution of histone marks, the maintenance of dimers. To avoid the dilution of histone marks, active maintenance requires first a modifications could be achieved by using a neighbour deacetylation step, which involves a histone deacetylase(HDAC), followed by histone ing histone as a template. A possible mechanism could modification that is guided by neighbouring parental nucleosomes(an interparticle be envisaged in which the parental mark is recognized process).b|Semi-conservative histone distribution. Parental dimers with marks segregate by a chromatin-binding protein, or reader protein, that evenly onto each daughter strand and nucleosomes are completed by the deposition of in turn recruits a chromatin modifier, or writer protein new H3-H4 dimers. After deacetylation, hemimodified' nucleosomes provide a template This has been suggested for the self-reinforcing loop for the transmission of parental marks to newly deposited H3-H4 dimers (an intraparticle in the maintenance of heterochromatin protein 1(HPI process). c Asymmetric histone distribution. Parental H3-H4 dimers with marks are redistributed onto daughter strands in an asymmetric manner. This is possibly dictated by at pericentric heterochromatinb7-(see below). Such a the intrinsic strand bias that is introduced during DNA replication, and induces a switch mechanism probably operates in repetitive regions one chromatin state to another. The maintenance of histone modifications requires in which long arrays of nucleosomes carry the same strand crosstalk. arks, but cannot apply to regions in which particular NATURE REVIEWS MOLECULAR CELL BIOLOGY 22009 Macmillan Publishers Limited All rights reservedNature Reviews | Molecular Cell Biology a Random Distribution Consequence c Asymmetric b Semi-conservative Maintenance (using neighbouring mark as template) OR dilution (not shown) Maintenance by default (hemimodified nucleosome used as a template for further modification) HDAC HDAC Maintenance requires interstrand crosstalk OR switch (not shown) R W R W Reader Writer HDAC Parental mark New mark R W Old H3–H4 dimer New H3–H4 dimer ‘Interparticle’ ‘Intraparticle’ ‘Interstrand’ HDAC HDAC regulates the nucleosome assembly that is dependent on cAF1 and the yeast­specific histone chaperone Rtt106 (ReFs 56,57). Whereas the presence of the H3K56ac mark has been reported in humans58, its abundance seems limited and its association with new histone deposi￾tion is not documented. Furthermore, homologues of Rtt106 and the Lys acetyltransferase Rtt109 (also known as Kat11), which acts on H3K56, have yet to be identi￾fied in humans. So, whether H3K56ac or an unidentified modification have similar roles in mammals remains to be investigated. Notably, newly synthesized histones H3 and H4 are present as dimers in pre­deposition complexes with histone chaperones59 (FIG. 2a). Although this principle is clearly established for newly synthesized histones, the fate of parental H3–H4 histone dimers that are thought to be deposited as tetramers on the daughter strands might also have to be reconsidered. The fact that histones H3 and H4 exist as stable tetramers in solution in the absence of DNA60 argues against the existence of parental H3–H4 dimers. However, structural data now show that the association of ASF1, and potentially also p48 and p55, with histones is incompatible with a tetrameric structure61–65. In addition, the fact that some histones carrying parental marks can be detected in association with ASF1 under conditions in which the helicase and the polymerase are uncoupled23 supports the hypothesis that ASF1 is involved in tetramer splitting and that it functions as an acceptor of recycled parental dimers. Therefore, it is indeed possible that parental tetramers (with their own marks) are split and redistribute onto daughter strands as dimers. This affects histone dynamics at the fork and might produce either mixed tetramers that comprise parental and new dimers (FIG. 2b), or nucleosomes that comprise only old histones if paren￾tal dimers reassociate. This second scenario requires either that the old dimers are held in close contact or away from the new ones, or that some recognition event ensures that the correct old dimers are brought back together in the same particle. The spatial organization of DNA at the fork might facilitate these mechanisms (FIG. 1a). If modifications on the new histones are guided by modifications of parental histones, the way in which parental histones are distributed to the daughter strands will determine the degree of conservation of histone marks. current models suggest that the distribution of both parental and newly synthesized histones onto daughter strands occurs in a random fashion (FIG. 3a). To avoid the dilution of histone marks, the maintenance of modifications could be achieved by using a neighbour￾ing histone as a template. A possible mechanism could be envisaged in which the parental mark is recognized by a chromatin­binding protein, or reader protein66, that in turn recruits a chromatin modifier, or writer protein. This has been suggested for the self­reinforcing loop in the maintenance of heterochromatin protein 1 (HP1) at pericentric heterochromatin67–70 (see below). Such a mechanism probably operates in repetitive regions in which long arrays of nucleosomes carry the same marks, but cannot apply to regions in which particular Figure 3 | Fate of old and new H3–H4 dimers and their marks at the fork. Three possibilities for the distribution of parental histones are presented. In each case, the parental mark is recognized by a chromatin-binding protein, or reader protein, that in turn recruits a chromatin modifier, or writer protein. a | Random histone distribution. Parental histone H3 and H4 with marks (unsplit or reassociated dimers) are distributed randomly onto daughter strands and chromatin density is restored by the deposition of new H3–H4 dimers. To avoid the dilution of histone marks, active maintenance requires first a deacetylation step, which involves a histone deacetylase (HDAC), followed by histone modification that is guided by neighbouring parental nucleosomes (an interparticle process). b | Semi-conservative histone distribution. Parental dimers with marks segregate evenly onto each daughter strand and nucleosomes are completed by the deposition of new H3–H4 dimers. After deacetylation, ‘hemimodified’ nucleosomes provide a template for the transmission of parental marks to newly deposited H3–H4 dimers (an intraparticle process). c | Asymmetric histone distribution. Parental H3–H4 dimers with marks are redistributed onto daughter strands in an asymmetric manner. This is possibly dictated by the intrinsic strand bias that is introduced during DNA replication, and induces a switch from one chromatin state to another. The maintenance of histone modifications requires interstrand crosstalk. REVIEWS NATuRe ReVIeWS | Molecular cell Biology VOlume 10 | mARcH 2009 | 197 © 2009 Macmillan Publishers Limited. All rights reserved