Di-methyl H4 Lysine 20 Targets the Checkpoint Protein Crb2 to Sites of DNA Damage^,

Abstract

Histone lysine methylation is an important chromatin modification that can be catalyzed to a mono-, di-, or tri-methyl state. An ongoing challenge is to decipher how these different methyllysine histone marks can mediate distinct aspects of chromatin function. The fission yeast checkpoint protein Crb2 is rapidly targeted to sites of DNA damage after genomic insult, and this recruitment requires methylation of histone H4 lysine 20 (H4K20). Here we show that the tandem tudor domains of Crb2 preferentially bind the di-methylated H4K20 residue. Loss of this interaction by disrupting either the tudor-binding motif or the H4K20 methylating enzyme Set9/Kmt5 ablates Crb2 localization to double-strand breaks and impairs checkpoint function. Further we show that dimethylation, but not tri-methylation, of H4K20 is required for Crb2 localization, checkpoint function, and cell survival after DNA damage. These results argue that the di-methyl H4K20 modification serves as a binding target that directs Crb2 to sites of genomic lesions and defines an important genome integrity pathway mediated by a specific methyl-lysine histone mark.

Post-translational modification of histone proteins has emerged as a key mechanism for controlling chromatin structure and function (1). These marks can function by either directly altering the structure of chromatin or by serving as a binding scaffold for the recruitment of regulatory factors. Of the possible modifications, methylation of histone lysine residues offers a unique platform for the control of chromatin function (2, 3). Histone lysine methylation can be catalyzed to a mono-(me1), di-(me2), or tri-methylated (me3) state. The emerging paradigm is that these different methyl-lysine histone marks can direct distinct avenues of chromatin function by serving as binding targets for chromatin effector proteins. Many chromatin-binding proteins contain methyl-lysine-binding motifs, such as chromo and tudor domains, that can discriminate between the mono-, di-, and tri-methylated modifications (4). The recognition of a specific methyl-lysine target by a binding module is thought to be essential for the proper localization and function of many chromatin regulatory factors.

To ensure their survival after genomic insult, cells initiate a number of molecular pathways leading to checkpoint-mediated cell cycle arrest and the repair of damaged DNA (5). Essential to these processes is the rapid localization of a variety of factors to genomic lesions. Recently several distinct histone modifications have been shown to be important marks for the recruitment of key damage response proteins (6). In the fission yeast Schizosaccharomyces pombe, recruitment of the checkpoint protein Crb2 to double-strand breaks (DSBs)2 requires two independent modifications: methylation of histone H4K20 (7) and C-terminal H2A phosphorylation (referred to as H2AX phosphorylation (8)). Phosphorylated H2AX in yeast or the equivalent modification in mammalian cells, defined as γH2AX, functions as a damage-specific chromatin mark with multiple roles in genome integrity (6, 9). In fission yeast loss of H2AX phosphorylation abolishes localization of Crb2 to DSBs, impairs the G₂ to mitosis (G₂/M) DNA damage checkpoint, and induces genome instability (8). Disruption of the H4K20 methylating enzyme Kmt5/Set9 or its catalytic substrate H4K20 impairs Crb2 recruitment, checkpoint function, and genome stability in a manner very similar to phospho-H2AX deficiency (7, 10). Note that as outlined by the unified nomenclature for the naming of histone lysine methyl-transferases (11), we now utilize the name Kmt5 to refer to the fission yeast H4K20 methylase previously know as Set9 (7). Unlike phosphorylated H2AX, the methyl H4K20 modification itself does not appear to be specifically targeted to genomic lesions but is constitutively present even in the absence of DNA damage (7, 12). It has been proposed that alterations in chromatin structure at sites of DNA damage render pre-existing methyl H4K20 available for Crb2 binding (7). The dual requirement for both a methylated and a phosphorylated histone residue in targeting is shared between Crb2 and its related factors human 53BP1 and budding yeast Rad9 (12–18).

Crb2 and related proteins are defined by the presence of two distinct motifs with modification binding potential: a pair of tandem tudor domains and C-terminal BRCT (BRCA1 C-terminal) repeats (see Fig. 1A and Ref. 19). BRCT domains are known phospho-binding modules, and this motif in Crb2 can directly recognize phosphorylated H2AX (20). Genetic experiments argue that the requirement for H4K20 methylation in Crb2 targeting is mediated through its tandem tudor domains (10), suggesting that this motif directly binds the methylated H4K20 residue. Strikingly, the tudor domains of the mammalian 53BP1 protein do indeed preferentially bind the di-methyl H4K20 modification (12, 21). The structure of the 53BP1 ·H4K20me2 complex has revealed a binding cage that is remarkably conserved in Crb2 (Ref. 12 and Fig. 1A). Like Crb2, 53BP1 localization to DSBs requires both H4K20 methylating enzymes and an intact tudor motif (12, 16, 22). Together these studies suggest a model whereby tudor domain recognition of the di-methylated H4K20 chromatin mark is required to direct Crb2 and 53BP1 to sites of DNA damage.

FIGURE 1.

The tudor domains of Crb2 preferentially bind the di-methylated H4K20 modification. A, bottom, schematic representation of Crb2 (not drawn to scale) with relevant mutations indicated. Top, protein sequence alignment of the tudor domains from fission yeast Crb2, human 53BP1, and budding yeast Rad9. The sequence from only the methyl-lysine-binding pocket is shown. Identical residues are shaded black, and similar residues are shaded gray. Cage residues are indicated with asterisks. B, the Crb2 tudor domains specifically interact with H4K20me2. Peptide pulldowns were performed as described in the text with H4 tail peptides modified at Lys²⁰ as indicated (top) and recombinant Crb2 tudor domain fragments as labeled left. Input represent 10% of the total protein used in the assay.

H4K20 methylating enzymes have been defined as important regulators of genome integrity in both fission yeast and higher organisms (7, 10, 12, 23–26). How the specific H4K20 methylation states contribute to this role is not completely understood. In fission yeast a complete abolishment of H4K20 methylation induces genome instability, but whether the mono-, di-, or tri-methylated residue is the essential modification is unknown (7, 10). The preferential binding of the 53BP1 tudor motif to H4K20me2 argues that di-methylation is the critical modification, but this has not been directly tested. Further, whether di-methyl H4K20 is indeed the essential modification for Crb2 and 53BP1 targeting in cells is still unclear.

Here we have sought to investigate the modification state specificity for H4K20 methylation in controlling fission yeast genome stability. We demonstrate that the tudor domains of Crb2 directly and preferentially bind the di-methylated H4K20 modification. Disruption of the tudor-H4K20me2 interaction results in a loss of Crb2 localization to DSBs, impaired checkpoint function, and DNA damage hypersensitivity that is analogous to inactivation of the Kmt5 H4K20 methylase. By generating several Kmt5 methylation state mutants, we further show that di-methylation of H4K20, but not tri-methylation, is required for Crb2 targeting, checkpoint function, and cell survival after genomic insult. These results argue that di-methylation of H4K20 mediates fission yeast genome stability by directing the checkpoint protein Crb2 to sites of genomic lesions.

EXPERIMENTAL PROCEDURES

Strains and Manipulations—Relevant strains are listed in Table 1. Yeast cell growth, phenotyping, checkpoint analysis, localization experiments, and generation of mutants were performed essentially as described (7). A cesium-137 γ-irradiation source with a dose rate of ∼3 Gy/min (localization experiments) or ∼10 Gy/min (phenotyping and checkpoint analysis) was utilized for irradiation experiments.

TABLE 1.

Fission yeast strains

Strain	Genotype
YSLS611	h- ura4D18 crb2Δura4 kmt5Δkan leu1-32::leu1+ pJK148-REP81-GFP-crb2+
YSLS702	h- crb2Δkan leu1-32::leu1+ pJK148-REP81-GFP-crb2+
YSLS703	h- crb2Δkan leu1-32::leu1+ pJK148-REP81-GFP-crb2T215A
YSLS704	h- crb2Δkan leu1-32::leu1+ pJK148-REP81-GFP-crb2Y378Q
YNTG67	h- crb2Δnat leu1-32::leu1+ pJK148-REP81-GFP-crb2+ kmt5F164Y::CYCTermkan
YNTG68	h- crb2Δnat leu1-32::leu1+ pJK148-REP81-GFP-crb2+ kmt5F178Y::CYCTermkan
YNTG69	h- crb2Δnat leu1-32::leu1+ pJK148-REP81-GFP-crb2+ kmt5F195Y::CYCTermkan
YNTG73	h- crb2Δnat leu1-32::leu1+ pJK148-REP81-GFP-crb2+ kmt5::CYCTermkan
YNTG74	h- crb2Δnat leu1-32::leu1+ pJK148-REP81-GFP-crb2+ kmt5Δkan
YSLS710	h-
YSLS711	h- crb2T215A
YSLS712	h- crb2Y378Q
YSLS716	h- crb2Δkan
YSLS718	h- kmt5Δkan
YSLS719	h- kmt5Δkan crb2T215A
YSLS720	h- kmt5Δkan crb2Y378Q
YSLS725	h- crb2T215A-Y378Q
YSLS541	h- chk1-HA kmt5Δkan
YARA36	h- chk1-HA
YARA37	h- chk1-HA crb2T215A
YARA38	h- chk1-HA crb2Y378Q
YARA41	h- chk1-HA crb2Δkan
YNTG75	h- kmt5::CYCTermkan chk1-HA
YNTG76	h- kmt5F164Y::CYCTermkan chk1-HA
YNTG77	h- kmt5F178Y::CYCTermkan chk1-HA
YNTG78	h- kmt5F195Y::CYCTermkan chk1-HA
972	h-
YSLS252	h- kmt5Δkan
YSLS298	h- kmt5::CYCTermkan
YNTG39	h- kmt5F164Y::CYCTermkan
YNTG41	h- kmt5F178Y::CYCTermkan
YNTG43	h- kmt5F195Y::CYCTermkan

Antibodies—Western blotting and antibody characterization were performed as described (7). Similar results were routinely achieved utilizing two different antibodies directed against H4K20me1 (LP Bio AR-0134 and Abcam ab9051), H4K20me2 (LP Bio AR-0135 and Abcam ab9052), and H4K20me3 (LP Bio AR-0136 and Abcam ab9053). Polyclonal anti-Kmt5 has been previously described (7).

Binding Experiments—A N-terminally tagged glutathione S-transferase expression construct for the Crb2 tudor domain amino acids 354–494 was generated by PCR cloning. Purification of glutathione S-transferase-tagged proteins and peptide pulldowns were performed as described (21) except that peptides were cross-linked to SulfoLink resin (Pierce).

RESULTS

The Crb2 Tudor Domains Preferentially Bind H4K20me2—Localization of the checkpoint protein Crb2 to ionizing irradiation (IR)-induced DSBs requires H4K20 methylation, and genetic experiments argue that this requirement is mediated through the tudor motifs of Crb2 (7, 10). This suggests that the tudor domains of Crb2 are methyl H4K20 recognition modules, and NMR experiments do argue that the tudor motif of Crb2 can recognize H4 tail peptides di-methylated at Lys²⁰ (12). However, neither the specificity of this interaction toward the di-methyl H4K20 modification nor its relevance to Crb2 function was demonstrated. To examine the methyl binding potential of the Crb2 tudor motif, a peptide pulldown assay was employed (Fig. 1). Purified recombinant glutathione S-transferase-tagged tudor domain was mixed with immobilized H4 tail peptides with Lys²⁰ either unmodified or mono-, di-, or tri-methylated. After precipitation and washing, Western blotting with anti-glutathione S-transferase antibody was used to detect peptide bound tudor domain protein. Fig. 1B shows that the Crb2 tudor domains preferentially bound the H4K20me2 peptide. Quantitation of the immunoblot image revealed a 4–5-fold enrichment of the tudor domain protein with the di-methyl peptide relative to the unmodified or mono- or tri-methylated peptides (data not shown). To test the specificity of this interaction, the ability of a mutant tudor domain protein, Y378Q, to bind the methylated H4K20 peptides was also analyzed. Tyr³⁷⁸ is one of the critical residues that forms the methyl-lysine-binding cage in Crb2 (Fig. 1A and Ref. 12). The corresponding mutation in either human 53BP1 (16) or budding yeast Rad9 (14) has been show to disrupt histone binding and localization to DSBs. Fig. 1B shows that binding of the Crb2 tudor domain to H4K20me2 is specific because the interaction is abolished by the Y378Q mutation. We conclude that the tudor domains of Crb2 preferentially bind the H4K20me2 modification similar to that of the 53BP1 tudor motif (12, 21).

The Tudor Domain-H4K20me2 Interaction Is Required for Crb2 Targeting to DSBs—The tudor domains of Crb2 can directly bind H4K20me2 (Fig. 1), and H4K20 methylation is required for Crb2 targeting to DSBs in fission yeast cells (7, 10). This argues that localization of Crb2 to sites of DSBs is mediated by tudor domain recognition of the di-methylated H4K20 residue. If this hypothesis is correct Crb2 localization to DSBs should be compromised by the tudor-Y378Q methyl-lysine binding mutation similar to that of methyl H4K20 deficiency. To directly test this prediction, live cell imaging of green fluorescent protein-tagged Crb2 was used to examine the ability of Crb2Y378Q to localize to IR-induced DSBs (Fig. 2). In this assay the green fluorescent protein-tagged crb2 allele is under the control of the moderately expressed nmt81 promoter and integrated into the fission yeast genome as the sole source of the protein (7). As a control, Crb2T215A was also analyzed. This mutation severely compromises Crb2 phosphorylation after DNA damage but not its ability to localize to IR-induced DSBs (27). The cells were grown in minimal medium and processed for imaging either immediately before (0 Gy; Fig. 2) or immediately after (40 Gy; Fig. 2) IR exposure. To examine the stability of Crb2 foci a portion of each irradiated sample was allowed to recover for 1 h prior to image analysis (Fig. 2, 1 Hour Post 40 Gy). Fig. 2 shows that as expected both wt Crb2 and Crb2T215A readily localized to DSBs, as represented by sites of focal enrichment after irradiation. In contrast, cells deficient in either the methyl-lysine-binding motif (Crb2Y378Q) or H4K20 methylation (Δkmt5) displayed a very similar ablation of Crb2 DSB targeting both immediately and 1 h after irradiation (Fig. 2). These results indicate that the tudor H4K20me2 binding module of Crb2 is required for Crb2 localization to IR-induced DSBs.

FIGURE 2.

Localization of Crb2 to IR induced DSBs requires the tudor methyl-lysine-binding motif. Strains harboring a green fluorescent protein-tagged crb2 allele were processed for live cell imaging either immediately before (0 Gy), immediately after (40Gy), or 1 h after exposure to IR (1 Hour Post 40 Gy). The left panels show representative images from either untreated (0 Gy) or cells immediately after irradiation (40 Gy). Right panel, quantitation of Crb2 foci. The data are averaged from three independent experiments in which 100–300 cells were counted for each point.

The Tudor Domain-H4K20me2 Interaction Is Required for Genome Integrity and Checkpoint Function—The results from Figs. 1 and 2 argue that tudor domain recognition of the dimethyl H4K20 modification mediates targeting of Crb2 to DSBs. We next sought to examine the role of this interaction in controlling fission yeast genome stability and DNA damage checkpoint function. The loss of H4K20 methylation results in hypersensitivity to genotoxic challenge and impairment of the G₂/M DNA damage checkpoint (7, 10). From this it would be expected that mutation of the tudor methyl-lysine-binding motif (crb2Y378Q) should yield a similar pattern of defects whether introduced alone or in combination with a loss of the H4K20me2-binding target (Δkmt5). The series of experiments presented in Figs. 3 and 4 argue that this prediction is correct and support our model of H4K20me2-dependent Crb2 recruitment.

FIGURE 3.

Loss of the tudor-H4K20me2 binding motif impairs fission yeast genome integrity. A and B, the crb2Y378Q mutation induces hypersensitivity to DNA damage, and the Crb2 tudor domain and the H4K20 methylase Kmt5 function in the same genetic pathway. In A the indicated strains were grown in liquid culture, and 5-fold serial dilutions were spotted onto either rich medium or rich medium containing the indicated amounts of CPT. Cells spotted onto rich medium were then either untreated (control) or irradiated with the indicated UV or IR dose. All of the plates were then incubated at 30 °C. In B quantitative IR survival curves were performed. Strains grown in liquid culture were treated with the indicated amount of IR (x axis), and 600–800 cells were plated in duplicate onto rich medium plates. Colony formation was measured after incubation at 30 °C, and survival was plotted relative to the un-irradiated 0 Gy sample (y axis). The data are from a single representative experiment.

FIGURE 4.

Loss of the tudor-H4K20me2-binding motif impairs the G₂/M DNA damage checkpoint. A, the crb2Y378Q mutation impairs checkpoint arrest similar to loss of H4K20 methylation. The indicated strains were grown in rich medium and G₂ synchronized by lactose gradient sedimentation. G₂ cells were then either untreated (0 Gy) or IR exposed (500 Gy) and allowed to recover at 30 °C in liquid medium. The cell aliquots were methanol fixed every 15 min (time indicated on x axis), and mitotic progression was assessed by 4′,6′-diamino-2-phenylindole staining (y axis). B, Chk1 phosphorylation is impaired in crb2Y378Q cells. Strains (labeled at the top) harboring a Chk1-HA3 allele were treated with the indicated dose of IR (right) and processed for Western blotting with anti-HA. The slower migrating band marked on the left indicates phosphorylated (P) Chk1.

To examine the requirement for the Crb2 methyl-lysine-binding motif in genome integrity, the Y378Q mutation was introduced into the endogenous crb2⁺ allele, and phenotypic analysis was undertaken. Fig. 3 demonstrates that cells harboring the crb2Y378Q or Δkmt5 mutations, either alone or in combination, display an essentially identical pattern of damage sensitivity. These mutants are hypersensitive to IR, UV light, and the topoisomerase I poison camptothecin (CPT) but not to the replication inhibitor hydroxyurea (data not show). The nonadditive relationship between the crb2Y378Q and Δkmt5 mutations argues that the Crb2 tudor domains and the Kmt5 methylase function in the same genetic pathway. Consistent with previous results (7, 10), the loss of either H4K20 methylation or Crb2 tudor domain function results in only a partial inactivation of crb2⁺ (Fig. 3A). Cells harboring a complete disruption of crb2⁺ (Δcrb2) are markedly more sensitive to DNA damage than either crb2Y378Q or Δkmt5 cells. Previously we demonstrated that the loss of H4K20 methylation (Δkmt5) combined with a Crb2 phosphorylation defect (crb2T215A) yields a nearly complete abolishment of crb2⁺ function (7). Given that the Crb2 tudor domain and Kmt5 function within the same genetic pathway, it would be expected that combining the crb2Y378Q and crb2T215A mutations would result in a similar ablation of crb2⁺ function. Fig. 3A demonstrates that this prediction is correct because crb2T215A-Y378Q cells display a level of damage sensitivity similar to that of Δcrb2 and crb2T215A-Δkmt5 cells. Note that unlike with IR or UV irradiation, crb2T215A cells are nearly as sensitive to CPT as Δcrb2 cells (Fig. 3A). Because of this the enhancement of the crb2T215A phenotype by the crb2Y378Q or Δkmt5 mutations is much less pronounced with CPT than with IR or UV irradiation. These results argue that the H4K20me2-binding motif of Crb2 is required for genome stability and underscore the genetic similarities between the Crb2 tudor domain and the H4K20 methylase Kmt5.

We next examined the role of the Crb2 tudor domain in the G₂/M DNA damage checkpoint (Fig. 4). The cells were synchronized in G₂ by size separation, and their rate of mitotic progression was monitored by 4′,6′-diamino-2-phenylindole staining after either no treatment or IR exposure. In the absence of irradiation all cells progressed through mitosis at a similar rate (see 0 Gy; Fig. 4A). As previously demonstrated (7) methyl H4K20 deficient Δkmt5 cells arrested after IR but prematurely released into mitosis ∼45 min prior to wt cells (see 500 Gy; Fig. 4A). Critically, cells harboring a disruption of the tudor H4K20me2-binding motif (crb2Y378Q) displayed an identical premature checkpoint release (Fig. 4A). With both Δkmt5 and crb2Y378Q mutations, >90% of those cells that released within the first 2 h after IR exposure underwent a catastrophic mitosis (data not shown). This is indicative of cells re-entering cell cycle progression with damaged DNA (data not shown). To investigate checkpoint function at the molecular level, IR-induced phosphorylation of the checkpoint kinase Chk1 was examined. Fig. 4B demonstrates that abolishment of either H4K20 methylation in Δkmt5 cells or tudor domain function in crb2Y378Q cells resulted in a similar decrement in Chk1 activation as monitored by loss of the slower migrating hyperphosphorylated band. These results indicate that disruption of either the di-methyl H4K20 recognition module of Crb2 or its binding target yield essentially identical defects in checkpoint function.

Di-methyl H4K20 Is Required for Genome Integrity and Crb2 Function—Inactivation of the Kmt5 methylase in fission yeast cells completely abolishes H4K20 methylation, compromises Crb2 targeting, checkpoint function, and induces genome instability (see above and Refs. 7 and 10). The preferential binding of the Crb2 tudor domains to the di-methylated H4K20 residue (Fig. 1) suggests that this mark is the essential modification for genome integrity, but this hypothesis has not been directly tested. We next sought to generate a series of Kmt5 methylation state mutants that would allow us to examine the requirement for each methyl H4K20 modification in genome integrity. Based upon the di-methyl binding preference, we reasoned that the loss of tri- or mono-methylation without an effect upon di-methylation should not affect damage sensitivity. In contrast disruption of H4K20 di-methylation would be expected to render cells sensitive to genomic insult and compromise Crb2 function. Figs. 5 and 6 demonstrate that we successfully generated a number of Kmt5 methylation state mutants and argue that H4K20 di-methylation, but not tri-methylation, is required for genome stability, Crb2 targeting to DSBs, and checkpoint function.

FIGURE 5.

Di-methylation of H4K20, but not tri-methylation, is required for fission yeast genome stability. A, Kmt5 methylation state mutations. Protein sequence alignments were performed with SET domain sequence from the indicated human (h) or S. pombe (sp) histone methyl-transferases. Only the relevant portion of the alignment is shown. Identical residues are shaded black. Residues corresponding to the known Tyr³⁰⁵ Phe/Tyr switch of human KMT7 are shaded gray. Relevant mutations are labeled across the top. B, Western blotting of Kmt5 methylation state mutants. Whole cell extracts were prepared from the labeled strains (top) and processed for Western blotting with the polyclonal antibodies marked on the left. Note that the Kmt5 immunoblot was performed with extracts independently prepared from those used for histone modification immunoblotting. C–E, DNA damage sensitivity of Kmt5 methylation state mutants. CPT phenotyping (C) and survival curve analysis with either IR (D) or UV (E) irradiation were performed as outlined in Fig. 3. The survival curve data are averaged from three independent experiments.

FIGURE 6.

Di-methylation of H4K20, but not tri-methylation, is required for Crb2 targeting and checkpoint function. A, H4K20 di-methylation is required for Crb2 localization to IR induced DSBs. Green fluorescent protein-Crb2 imaging experiments were performed as in Fig. 2. The data are averaged from two independent experiments. B and C, the G₂/M DNA damage checkpoint is impaired by loss of H4K20 di-methylation. Checkpoint (B) and Chk1 phosphorylation (C) assays were performed as described for Fig. 3.

Enzymatic and structural work has identified a number of key residues that can determine methylation state specificity in SET domain histone lysine methyl-transferases (28). In particular a Phe/Tyr switch has been identified that can control methylation state specificity (29). To generate Kmt5 methylation state mutants SET domain protein sequence alignments were used to identify potentially important catalytic residues (Fig. 5A and data not shown). Phe¹⁹⁵ of Kmt5 was identified as a likely switch residue because it corresponded to Tyr³⁰⁵ of human KMT7/Set7, a known Phe/Tyr switch (Fig. 5A and Ref. 28). Because of the general importance of bulky hydrophobic residues in SET domain catalysis, a number of such residues conserved between Kmt5 and the related human KMT5B/C enzymes were also disrupted (Fig. 5A). Mutations were introduced into the endogenous kmt5⁺ gene, and Western blotting was performed to examine H4K20 methylation status (Fig. 5B). Consistent with the idea of a Phe/Tyr switch, mutation of Phe¹⁹⁵ to Tyr, but not Trp, resulted in a near complete loss of H4K20me3 with minimal affect upon H4K20me1 or H4K20me2 (Fig. 5B and data not shown). We also identified two independent mutations, F164Y and F178Y, which resulted in both a loss of H4K20me3 and a significant decrease in H4K20me2 (Fig. 5B). As evident by the weak detection in F164Y and F178Y cells relative to Δkmt5 cells, these mutations did not completely abolish H4K20me2. F164Y and F178Y cells also displayed a minimal decrease in H4K20me1, but this effect was much less pronounced than that upon H4K20me2 or H4K20me3. Despite analyzing a number of independent mutations, we were unable to generate any kmt5 mutants that specifically compromised H4K20me1 and/or H4K20me2 without also decreasing H4K20me3 (data not shown). We next examined Crb2 function and genome integrity in the Kmt5 methylation state mutants.

Our model of H4K20 methylation-dependent Crb2 recruitment argues that the di-methyl residue is the essential modification for fission yeast genome integrity. From this we would predict that tri-methyl deficient kmt5F195Y cells would not be hypersensitive to genomic insult. In contrast kmt5F164Y and kmt5F178Y cells compromised for both di- and tri-methylation of H4K20 should display genome instability after DNA damage. Fig. 5 (C–E) demonstrates that this prediction is correct. The kmt5F164Y and kmt5F178Y mutations render cells hypersensitive to DNA damage, whereas kmt5F195Y cells are not hypersensitive. Consistent with their incomplete loss of H4K20me2 (Fig. 5B), the sensitivity of kmt5F164Y and kmt5F178Y cells is less pronounced than that of Δkmt5 cells that are completely devoid of H4K20me2. Although the data do not allow us to directly address the role of H4K20me1 in genome stability, these results argue that di-methylation of histone H4K20, but not tri-methylation, is required for fission yeast cell survival after genomic challenge.

We next directly examined Crb2 targeting and the G₂/M DNA damage checkpoint in the Kmt5 methylation state mutants (Fig. 6). Consistent with the phenotypic results, Crb2 localization and checkpoint function in tri-methyl deficient kmt5F195Y cells was similar to that of wt cells (Fig. 6). In contrast the kmt5F164Y and kmt5F178Y mutations, which disrupt both di- and tri-methyl H4K20, significantly compromised Crb2 localization to IR-induced DSBs (Fig. 6A). Checkpoint function was also impaired as kmt5F164Y and kmt5F178Y cells arrested after IR exposure but prematurely re-entered the cell cycle 20–30 min sooner than wt cells (Fig. 6B). Consistent with the incomplete loss of H4K20 di-methylation (Fig. 5B), the checkpoint defect induced by the kmt5F164Y and kmt5F178Y mutations was less severe than that seen in Δkmt5 cells, which released ∼45 min prior to wt cells (Fig. 4A). Note that because of technical limitations we were unable to reliably process more than four strains simultaneously in the checkpoint assay and could not directly compare Δkmt5 cells versus the methylation state mutants. IR-induced Chk1 phosphorylation was also impaired by the kmt5F164Y and kmt5F178Y mutations but was again less pronounced than that seen in Δkmt5 cells (Fig. 6C). These results indicate that H4K20 di-methylation, but not tri-methylation, is required for Crb2 localization to DSBs and DNA damage checkpoint function.

DISCUSSION

Earlier work has suggested a model whereby targeting of the fission yeast checkpoint protein Crb2 to DSBs requires a tudor domain-methyl H4K20 interaction (7, 10, 12). Here we have demonstrated that the tandem tudor domains of Crb2 directly bind the H4K20me2 modification. Ablation of this interaction either by loss of the di-methyl H4K20-binding target or the tudor methyl-lysine-binding motif disrupts Crb2 localization to DSBs, impairs checkpoint function, and renders cells sensitive to genomic insult. Together with published work (10, 12), the results of this study strongly argue that di-methylated H4K20 functions as a chromatin-binding target for the tudor domains of Crb2 and that this interaction is required for fission yeast genome integrity. This model of histone methylation-dependent recruitment appears to have been conserved during eukaryotic evolution because the tudor domains of the Crb2-related human 53BP1 protein preferentially bind H4K20me2 (12, 21). Mammalian H4K20 methylating enzymes are also required for localization of 53BP1 to DSBs and genome integrity (12, 22, 25). Together this work defines an important genome-targeting pathway mediated by a specific histone methyl-lysine modification state.

A critical unresolved issue is to understand how the tudor-H4K20me2 interaction functions to target Crb2 and 53BP1 specifically to DSBs. Unlike the damage-specific phospho-H2AX modification, di-methylated H4K20 itself does not appear to be targeted to genomic lesions. In both fission yeast and mammals all three methylated H4K20 residues are readily detectable during unperturbed cell growth, bulk levels do not increase after DNA damage, and neither H4K20 methylation nor methylating enzymes display any significant relocalization to sites of DSBs (7, 12). Further, the di-methyl modification (but not the mono- or tri-methylated residues) appears to be extremely abundant with as much as 90% of the total H4 di-methylated at Lys²⁰ in mammals and Drosophila (22, 25, 30). This argues that di-methyl H4K20 functions as a genome wide modification that does not require specific localization to sites of damage. We have previously suggested that alterations in local chromatin structure at DSBs render pre-existing methyl H4K20 accessible for Crb2 binding (7). A similar mechanism has been proposed for 53BP1 because conditions that promote chromatin accessibility reduce 53BP1 nuclear diffusion (16). An alternative but not mutually exclusive hypothesis has also been suggested in which lesion site-specific H2AX phosphorylation mediates the tudor-H4K20me2 interaction (10, 12). This idea is supported by the dual requirement for both modifications in Crb2 targeting (7, 8) and the apparent functional redundancy between H4K20 methylation and H2AX phosphorylation (10). Further studies are required to fully resolve the mechanism that governs the specificity of the tudor-H4K20me2 interaction.

In both fission yeast and mammals H4K20 methylation functions as an important histone mark for genome stability. Interestingly, in budding yeast neither the methyl H4K20 modification nor methylating enzymes are readily detectable (7). Rather methylation of histone H3 Lys⁷⁹ appears to have an analogous role in mediating budding yeast genome integrity through the Rad9 checkpoint protein (14, 15, 18, 31). Fission yeast does not have detectable H3K79 methylation (see supplemental data). In mammals both modifications are present, and an initial report linked H3K79 methylation to 53BP1 targeting (16). More recent reports argue a predominant role for di-methylation of H4K20 (12, 22). This suggests that targeting of 53BP1 may involve multiple methyl-lysine histone residues. Both modifications are thought to function as pre-existing histone marks that are not specifically targeted to genomic lesions (12, 16). Because of this we suggest that the relative abundance of each modification will be key to determining their functional role as related to 53BP1. Published work argues that di-methyl H4K20 fulfills the requisite abundance requirement (22, 25, 30). To our knowledge the abundance of methylated H3K79 in mammals is unknown.

Di-methylation of histone H4K20 functions in DNA damage checkpoint control and genome integrity, but the role of mono- and tri-methylation is not completely understood. In mammals tri-methyl H4K20 is a mark of heterochromatin and gene repression (32, 33). Surprisingly H4K20 methylation is not required for heterochromatin function in fission yeast (7), but unpublished observations suggest that tri-methyl H4K20 may be associated with transcriptional repression.3 H4K20 mono-methylation has been linked to both transcriptional activation (34–36) and repression (23, 37, 38), but the functional significance of this association is not yet fully understood. Interestingly a very recent report has demonstrated that mouse cells harboring a double knock-out of the KMT5B/C H4K20 methylases are almost completely devoid of H4K20 di- and tri-methylation but display a dramatic increase in bulk mono-methylation (25). These cells are hypersensitive to genomic insult but exhibit a surprisingly moderate impairment in 53BP1 targeting. The 53BP1 tudor domains can recognize mono-methyl H4K20, although with an ∼2.7-fold lower affinity relative to the di-methylated residue (12). Because of this and the fact that the abundance of the target modification seems to be key to 53BP1 targeting (see above), this suggests that mono-methyl H4K20 may be able to partially substitute for the di-methyl modification. This and other intriguing aspects of histone methyl-lysine-mediated genome stability await further investigation.