Cell Imaging

Live cell imaging was performed as described previously (Loeillet et al., 2005; Palancade et al., 2005) by using a three-dimensional (3D)-epifluorescence microscope driven by MetaMorph 6.2.6 software (Molecular Devices, Sunnyvale, CA). Images were further processed using Adobe Photoshop CS (Adobe Systems, Mountain View, CA). For statistical analyses of Rad52 foci, 3D-projections of Z-stack images (n = 11; plane spacing, 0.4 μm) of live cells were used, and foci were scored by visual inspection. Quantification of the number of foci per nucleus was thereby determined for the whole cell population (circular diagrams and histogram in Figure 5C). To quantify Rad52 foci occurrence for each stage of the cell cycle, cells were staged as G1 (unbudded), S (small-budded), or G2/M (large-budded) based on differential interference contrast (DIC) images. Within each of these subpopulations, the percentages of cells exhibiting at least one Rad52 focus were quantified (histograms). In vivo nuclear import assays were performed essentially as described previously (Timney et al., 2006). Briefly, exponentially growing cells expressing the appropriate nuclear localization signal (NLS)–GFP fusion protein were metabolically poisoned in the presence of 2-deoxyglucose and sodium azide to equilibrate the reporter between the nuclear and cytoplasmic compartments. Nuclear import was induced in the presence of glucose-containing medium, and images were acquired every 20 s. Nuclear intensities were determined using MetaMorph, and they were plotted as a function of time; the slope of the linear portion of the resulting import curve (t <5 min) was normalized to the initial cytoplasmic concentration of the reporter to calculate the absolute import rate.

Open in a separate window

Figure 5.

DNA repair efficiencies in nucleoporin and ulp1 mutants. (A) Schematic figure that summarizes the principle of the assay is shown. DR, direct repeats, mediating homologous recombination. (B and C) Absolute SSA and NHEJ values were calculated as described in Materials and Methods, and they represent the number of colonies grown in galactose-relative to those grown in glucose-containing medium. The results of three independent experiments are shown along with the corresponding SD, and they are normalized to 100 for the wt value.

Protein Extraction and Analysis

Whole-cell lysates were obtained from exponentially growing cells by bead-beating in 50 mM HEPES, pH 7.5, 100 mM NaCl, 1 mM EDTA, 1% NP-40, 2.5 μg/ml aprotinin, 2.5 μg/ml pepstatin, 5 μg/ml leupeptin, and 2.5 mM phenylmethylsulfonyl fluoride. Lysates were further supplemented with an equal volume of protein sample buffer and clarified by centrifugation at 10,000 × g for 10 min. Immunoprecipitation of the Yku70–13myc protein under denaturing conditions and detection of the sumoylated Yku70 were performed as described previously (Zhao and Blobel, 2005). Western blotting was performed according to standard procedures using the following primary antibodies: anti-GFP (Roche Diagnostics, Mannheim, Germany), anti-NOP1 A66 (Tollervey et al., 1991), anti-Myc (9E10; Sigma-Aldrich, St. Louis, MO), and anti-SUMO (Zhao and Blobel, 2005). Detection was performed using enhanced chemiluminescence, and quantification was achieved through MetaMorph as described in the corresponding figure legends.

Nucleoporin and ulp1 Mutants Exhibit Similar Genetic Interaction Profiles and Repair Phenotypes

Another DNA repair-related characteristic of the Nup84 complex mutants is their synthetic lethality or synergistic interaction with mutants that have defects in DNA replication and/or repair, mainly in the Rad52 pathway (Loeillet et al., 2005). We therefore analyzed genetic interactions between the nup60Δ mutant and mutants with defects in different pathways of DNA repair. The genetic interaction profile of the nup60Δ mutant strikingly resembled that of mutants of the Nup84 complex, such as nup133Δ (Table 1). Indeed, the nup60Δ mutant showed a strongly impaired cell growth when combined with mutants of the RAD52 pathway or with rad27Δ, but not when they were associated with mutants with defects in NHEJ (yku70Δ) or DNA damage bypass (rad6Δ). The growth phenotypes of the double mutants were more pronounced with nup133Δ, which could be attributed to a poorer fitness of the nup133Δ single mutant. Finally, the mlp1Δ mlp2Δ double mutant exhibited colethality when combined with mutants carrying deletions of RAD52 or RAD27 (Supplemental Figure 1). The genetic interaction profiles of nup133Δ and nup60Δ were reminiscent of the profile described previously for a thermosensitive allele of ULP1, ulp1-I615N, which was also shown to accumulate DNA damage (Table 1; Soustelle et al., 2004). ULP1 encodes a SUMO-protease that is located at NPCs (Li and Hochstrasser, 2003; Panse et al., 2003). Analysis of the ΔN338-ulp1 mutant (Zhao et al., 2004), which lacks its nuclear envelope localization domain (Li and Hochstrasser, 2003; Panse et al., 2003), revealed a strong hyperrecombination phenotype (Figure 1B) and an increased occurrence of Rad52 foci (Figure 1A). The extent of DNA repair foci accumulation was comparable in the ulp1 and nucleoporin mutants, and it was not significantly enhanced when the ΔN338-ulp1 mutation was combined with NUP133 deletion. Finally, our systematic synthetic lethal screening of the collection of nonessential gene deletions using the nup133Δ mutation as a bait, although confirming previously described interactions of the Nup84 complex with several genes involved in DNA repair (Loeillet et al., 2005), also identified SLX5 and SLX8, two genes involved in regulating DNA repair and sumoylation-dependent processes (Mullen et al., 2001; Wang et al., 2006). Further analyses indicated that the combined deletion of NUP60 and SLX8 also leads to a strong synergistic phenotype (Table 1). Together, these data suggest that the Nup84 complex, Nup60/Mlp1–2, and Ulp1 could be involved in a common pathway that prevents the accumulation of unrepaired DNA damage through sumoylation-dependent processes.

Nuclear Envelope Localization and Stability of Ulp1 Requires Both Nup60 and the Nup84 Complex

Previously published data indicated that Ulp1 is targeted to the nuclear pore basket through interaction with Mlp1–2 and that destabilization or mislocalization of Ulp1 could account for the clonal lethality occurring in the mlp1Δ mlp2Δ mutant (Zhao et al., 2004). To determine whether the Nup84 complex could also be involved in this process, we analyzed the subcellular localization of an Ulp1–GFP fusion protein, expressed under the control of its endogenous promoter, in mutants of the Nup84 complex. As reported previously for Mlp1–2 and associated proteins (Galy et al., 2004; Zhao et al., 2004; Palancade et al., 2005), Ulp1 was asymmetrically distributed along the nuclear envelope in wild-type (wt) cells (Figure 2A). Strikingly, deletions in nucleoporins of the Nup84 complex (i.e., nup133Δ, nup120Δ) led to a major decrease in the level of NPC-associated Ulp1, with only a faint residual signal in clustered NPCs. In contrast, the signal of Nup49-GFP, a classical nuclear pore marker, was increased in NPCs clusters, an expected effect of NPC aggregation in these mutants (Figure 2A). Quantitative Western blot analysis revealed a fivefold decrease of the Ulp1 protein levels in the Nup84 complex mutants (Figure 2, B and C). Similarly, deletion of NUP60 led, as described previously (Zhao et al., 2004), to the mislocalization and destabilization of Ulp1 (Figure 2, A–C). In contrast, Ulp1–GFP localization and stability were not affected in the nup159-1 mutant or in the absence of Nup188, a structural nucleoporin that does not belong to the Nup84 complex. Notably, NPC levels of Ulp1 were partially restored upon MG132 treatment of drug-responsive (erg6Δ) nup133Δ or nup60Δ derivatives (Supplemental Figure 2), thus indicating that proteasome-mediated degradation of Ulp1 contributes to its decreased levels at NPCs. Therefore, Ulp1 targeting and/or stabilization at NPCs specifically involve both Nup60 and the Nup84 complex. Consistent with this finding, analysis of the global pattern of sumoylation in the nup120Δ and nup60Δ mutants revealed that the sumoylation status of several proteins was affected. A similar effect was also observed in the ulp1 mutant lacking its N-terminal NPC localization domain (Figure 2D). In both nucleoporins and ΔN338-ulp1 mutants, either increased or reduced sumoylation level of some proteins was observed (see bands indicated by arrowheads and stars, respectively). These effects may reflect the consequences both of Ulp1 degradation (which causes an overall decrease in Ulp1 activity) and mislocalization (which leads to an increased Ulp1 activity in the nucleoplasm) in these mutants (see Discussion).

Open in a separate window

Figure 2.

The Nup84 complex and Nup60 are required for Ulp1 localization and stabilization and for the establishment of proper sumoylation patterns. (A) Fluorescence microscopy analysis of ULP1-GFP and NUP49-GFP localization in wild-type and nucleoporin mutant strains. The DIC images are also shown. Cells were grown at 30°C, except for the nup159-1 mutant, which was grown at 22°C to visualize NPC aggregation (Belgareh et al., 1998). (B) Whole cell lysates from ULP1-GFP–tagged strains were analyzed by Western blot by using an anti-GFP antibody and an anti-NOP1 (nucleolar protein) antibody as a loading control. The positions of the Ulp1–GFP fusion protein and of Nop1 are indicated. (C) Quantification of the amounts of Ulp1 in the different nucleoporin mutants. Serial dilutions of the samples used in B were analyzed by Western blotting and the integrated intensities of the Ulp1–GFP bands were measured using MetaMorph software. The amounts are expressed as a percentage of the wt value. (D) Total sumoylated proteins from wt, nup120Δ and ΔN338-ulp1 strains were detected using an anti-SUMO antibody. Similar amounts of proteins were loaded for each lane, as indicated by the amido-black staining of the same blot (“stain”). Molecular weights (kilodaltons) are indicated. SUMO-conjugates that are decreased in the mutants are indicated by stars, and the increased conjugates are indicated by arrows.

Karyopherin-mediated Pathways Do Not Account for the Decreased Levels of Ulp1 at NPCs in Nup84 or Nup60/Mlp1–2 Complexes Mutants

We next aimed at understanding the molecular mechanisms underlying the requirement of both Nup84 and Nup60/Mlp1–2 complexes for the localization and/or stabilization of Ulp1 at the nuclear envelope. Because it has been previously suggested that Nup145-C, a subunit of the Nup84 complex, may anchor Nup60 at the NPCs (Feuerbach et al., 2002), we first analyzed the localization of Nup60 in mutants of the Nup84 complex. Deletion of NUP133, NUP120, or NUP145-C led to the accumulation of Nup60-GFP within the NPC clusters, but it did not otherwise affect its NPC localization (Figure 3A), even at restrictive temperatures (data not shown). Similarly, Mlp1 and Mlp2 became concentrated within the NPC clusters in nup133Δ and nup120Δ mutants (Palancade et al., 2005; data not shown). Conversely, deletion of NUP60 did not affect Nup133 localization (data not shown). Thus, the NPC localization of the Nup84 and Nup60/Mlp1–2 complexes seem to be two unrelated processes. Previous studies demonstrated an involvement of the Kap121 and Kap95/Kap60 karyopherins in the localization of Ulp1 at the nuclear pores (Panse et al., 2003; Makhnevych et al., 2007). We thus asked whether the phenotype of the Nup84 complex mutants could be the indirect consequences of an alteration in these karyopherin-mediated pathways. Analysis of Kap95–GFP and Kap121–GFP fusion proteins revealed that both karyopherins were properly localized at NPCs in nup133Δ and nup60Δ mutant cells (Figure 3B). In addition, analysis of the static distribution and import kinetics of Kap60/Kap95-dependent (cNLS–GFP) or Kap121-dependent (pNLS–GFP) reporters did not reveal any significant import defect in nup120Δ or nup60Δ strains (Supplemental Figure 3, A and B), even though a mild import defect was observed for the pNLS–GFP reporter in the nup133Δ strain. Finally, analysis of Rad52–YFP localization did not reveal an increased occurrence of DNA repair foci in the kap121-34 mutant strain (Figure 1A; of note, the mild increase seen in Figure 1Ac is due to the accumulation of these cells in G2). These data prompted us to compare the respective effects of karyopherin and nucleoporin mutants on the localization and cellular levels of Ulp1 tagged with GFP. In agreement with a previous study (Panse et al., 2003), karyopherin inhibition, achieved by expressing a dominant-negative form of the Yrb4 importin, or by inactivation of the kap121-34-thermosensitive mutant at semirestrictive temperature (30°C), impaired the nuclear envelope localization of mildly overexpressed GFP–Ulp1 (i.e., under the control of the NOP1 promoter; Supplemental Figure 3C). However, these treatments did not significantly affect the localization of Ulp1 tagged with GFP at its cognate locus (Supplemental Figure 3C, compare with Figure 2A). Although the increased fading at higher temperature of the GFP variant used in this study did not enable us to observe any significant change in kap121-34 cells shifted to 37°C, a similar analysis, recently performed using Ulp1 tagged with the GFP⁺ variant, revealed that shifting the kap121-34 strain to 37°C causes a decreased level of NPC-associated Ulp1 and a corresponding increase in its cytoplasmic level (Makhnevych et al., 2007). However, Western blot analysis revealed that, unlike in the nucleoporin mutants, Ulp1-GFP was not destabilized in the kap121-34 mutant even after a 3-h shift to 37°C. Rather, a slower migrating form of Ulp1-GFP was observed in kap121-34 cells (Figure 3C), suggesting the occurrence of posttranslational modification(s) of Ulp1 in this karyopherin mutant. Together, these data indicate that karyopherin and nucleoporin mutants differentially affect Ulp1 localization and stability at NPCs. The phenotypes of the nucleoporin mutants are therefore not merely caused by a defect in these karyopherin-mediated pathways. This mildly overexpressed GFP–Ulp1 fusion protein has been shown to be targeted to the nuclear envelope in an mlp1Δ mlp2Δ ulp1Δ mutant (Zhao et al., 2004). We therefore expressed this construct in the nup133Δ, nup120Δ, and nup60Δ mutants, and we analyzed the localization of GFP-Ulp1 relatively to a core nucleoporin, Nic96, which was tagged with mRFP. In all strains, Ulp1 localization at the nuclear envelope was restored (Figure 3D; data not shown). Interestingly, the overexpressed Ulp1 fusion protein no longer exhibited its asymmetrical distribution within the nuclear envelope (compare Supplemental Figure 3C, left and right). Moreover, unlike Nic96–mRFP, its localization was no longer restricted to the nuclear pores as the GFP labeling spread out of the NPC aggregates in the nup133Δ strain (Figure 3D), indicating that mildly overexpressed GFP–Ulp1 may be ectopically targeted to areas of the nuclear envelope laying between the pores. This result may reflect the ability of this overexpressed fusion protein to directly interact with membranes, an hypothesis consistent with its partial mislocalization to the plasma membrane after karyopherin inhibition (Panse et al., 2003; Supplemental Figure 3C), or, alternatively, the presence of yet uncharacterized additional inner nuclear envelope tether(s) (discussed in Makhnevych et al., 2007). Together, our data demonstrate that although the Nup84 and Nup60/Mlp1–2 complexes are required for the maintenance of endogenous Ulp1 levels at nuclear pores, their requirement for Ulp1 localization and/or stabilization can be bypassed by overexpressing a GFP–Ulp1 fusion protein.

Open in a separate window

Figure 3.

Decreased levels of Ulp1 at NPCs in Nup84 or Nup60/Mlp1–2 complexes mutants are not merely due to defects in Kap121 or Kap95 pathways. (A) Fluorescence microscopy analysis of NUP60-GFP localization in wild-type or nucleoporin mutant strains grown at 30°C. The DIC images are also shown. (B) Fluorescence microscopy analysis of KAP95-GFP (left) or KAP121-GFP (right) localization in wild-type or nucleoporin mutant strains grown at 30°C. (C) Ulp1-GFP expression levels were analyzed by Western blot by using an anti-GFP antibody in wt, nup133Δ, nup60Δ, or kap121-34 mutant strains grown at 30°C or shifted for 3 h at 37°C. The anti-NOP1 antibody was used as a loading control. (D) Fluorescence microscopy analysis of wild-type or nucleoporin mutant strains expressing NIC96-mRFP and transformed with the pNOP1-GFP-ULP1 construct. DIC images are also shown.

Restoration of Proper Ulp1 Levels at the Nuclear Envelope Partially Complements the Altered Sumoylation Patterns and the DNA Repair-related Phenotypes of the Nucleoporin Mutants

We next analyzed the consequences of the mild overexpression of this GFP–Ulp1 protein on the global pattern of sumoylation in wild-type and nucleoporin mutant strains either untreated or treated with the DNA-damaging drug methyl methane sulfonate (MMS). MMS treatment modified the sumoylation status of some proteins (Figure 4A), suggesting the involvement of sumoylation in DNA damage-induced signaling pathways. In untreated cells, GFP–Ulp1 overexpression led to a general decrease of SUMO conjugates in both wild type and nucleoporin mutants (data not shown). Overexpression of GFP-Ulp1 in the nupΔ mutants partially restored a normal sumoylation pattern in MMS-treated cells, with a notable increase of two MMS-induced sumoylated bands (Figure 4A, stars). Conversely, the enhanced sumoylation of several proteins in nupΔ, compared with wild-type cells, was corrected upon GFP-ULP1 overexpression (Figure 4A, arrows). These results show that nucleoporin mutants have altered cellular sumoylation patterns and that restoring proper Ulp1 level at the nuclear envelope counteracts this phenomenon. This result prompted us to assess the functional consequences of Ulp1 restoration on the viability of nupΔ mutants, when combined with mutations affecting DNA repair or replication. nup133Δ rad27Δ, nup133Δ rad52Δ, nup60Δ rad27Δ, and nup60Δ rad52Δ double mutants are not viable or strongly impaired in their growth at 30°C (Table 1; Loeillet et al., 2005), but they can be rescued after dissection of the corresponding diploids at 25°C. Overexpression of GFP-Ulp1 complemented the synthetic lethality or the growth defects of each of the four mutants, whereas a control plasmid could not (Figure 4B and Supplemental Figure 4A). Noteworthy, the complementation was specific for the defects that arose from the combination of nup and rad mutations, because ULP1 overexpression did not affect the growth rates of any of the single mutants (data not shown). To determine which feature of the Ulp1 protein was needed for complementation of the nupΔ radΔ colethality, different constructs were tested in this assay (Figure 4B). A plasmid lacking the GFP tag rescued cell growth as efficiently as the GFP–Ulp1 construct, indicating that the rescue is not merely due to stabilization of Ulp1 via the GFP. The catalytically inactive ulp1-C580S mutant (Li and Hochstrasser, 2003) was no longer able to rescue nupΔ radΔ colethality. Rather, this construct became toxic in the nup60Δ rad52Δ mutant and impaired viability of the nup133Δ rad52Δ mutant (Figure 4B), although it did not affect the viability of the wild type or any of the corresponding single mutants (Figure 4B; data not shown). Because this mildly overexpressed protein was properly targeted to the nuclear envelope (data not shown), it could compete at NPCs with the residual levels of endogenous Ulp1 protein that was still present in the nucleoporin mutants. Finally, expression of the ΔN338-ulp1 mutant (Zhao et al., 2004), which lacks its nuclear envelope localization domain (Li and Hochstrasser, 2003; Panse et al., 2003), did not rescue the synergistic growth defects of the double mutants (Figure 4B). However, this construct was poorly expressed (Supplemental Figure 4B); therefore, it did not allow us to discriminate whether restoration of either the expression levels or the localization of Ulp1 contributed to the complementation phenotypes. We next determined whether the DSB accumulation observed in the Nup84 complex or in nup60 mutants was due to a loss of function of Ulp1. To this end, GFP-Ulp1 was overexpressed in wild-type, nup133Δ, and nup60Δ derivatives carrying the RAD52-YFP reporter. In wt cells, mild overexpression of GFP–Ulp1 did not affect the detection and the occurrence of Rad52 foci (Figure 4C) or the level of Rad52 foci assembly after MMS treatment (data not shown). In contrast, overexpression of GFP-Ulp1 reduced the occurrence of Rad52 foci in nup133Δ and nup60Δ cells with a prominent decrease in the number of cells with multiple foci (Figure 4C and Supplemental Figure 4C). Together, our data indicate that restoring an enzymatically active Ulp1 at the nuclear envelope partially complements both the defective sumoylation of some target proteins and some of the DNA repair-related phenotypes of the nucleoporin mutants.

Open in a separate window

Figure 4.

Complementation of nucleoporin mutant phenotypes by restoration of nuclear envelope-associated ULP1. (A) Total sumoylated proteins were detected using an anti-SUMO antibody in whole cell extracts from wt, nup120Δ, or nup60Δ strains that were transformed with the pRS315 (Ø) or the pRS315-NOP1prom-GFP-ULP1 (ULP1) plasmids and treated (+), or not (−), for 2 h with 0.3% MMS. Similar amounts of proteins were loaded for each lane, as indicated by the amido-black staining (“stain”) of the same blot. Molecular weights (kilodaltons) are indicated. Stars indicate SUMO-conjugates, which were decreased in the mutants and have been restored upon ULP1 expression, whereas arrows indicate the conjugates, which were increased in the mutants and reduced upon ULP1 expression. (B) wt, nup133Δrad52Δ, and nup60Δrad52Δ mutant strains were transformed with pRS315 (Ø) or pRS315 derivatives expressing the indicated constructs under the control of the NOP1 promoter (schematized on the left). Transformants were spotted as fivefold dilutions on synthetic medium lacking leucine and plates were incubated at 25, 28, 30, or 36°C. *, nup133Δ rad52Δ strain is not viable in the presence of the GFP-ulp1[C580S] construct. (C) Rad52–YFP foci analysis in wt, nup133Δ, and nup60Δ cells transformed with either pRS315 (Ø) or pRS315-NOP1prom-GFP-ULP1 (ULP1). The percentage of cells, which show one, two, or three or more Rad52–YFP foci per nucleus, is indicated. The SD corresponds to the sum of the standard deviations for each of the three subcategories.

The Nup84 Complex and Nup60 Regulate Sumoylation Patterns and DNA Repair through Ulp1

Our data indicate that the Nup84 complex and Nup60 mutants, which exhibit decreased levels of Ulp1 at the nuclear envelope, display complex modifications of cellular sumoylation patterns that can be partially suppressed upon restoration of catalytically active Ulp1 at the nuclear envelope. Of note, both increase and decrease in the levels of SUMO-conjugates were observed in these mutants as well as in a strain expressing an N-terminal truncated allele of Ulp1 that lacks its nuclear envelope targeting domain (Figure 2D; Zhao et al., 2004). Consistent with this observation, both the level and the subcellular localization of Ulp1 have previously been shown to be major determinants in its activity toward natural and nonnatural sumoylated substrates. Indeed, disturbing one of these two parameters by overexpressing and/or truncating Ulp1 leads to significant changes in the global pattern of SUMO-protein conjugates (Li and Hochstrasser, 2003; Zhao et al., 2004). Although decreased levels of Ulp1 could lead to an oversumoylation of some of its substrates, mislocalized Ulp1, even in low amounts, could gain access to intranuclear substrates and reduce the amount of their sumoylated forms. For example, Ulp1 was shown to target the genuine substrates of its paralogue Ulp2 when mislocalized from the nuclear periphery (Li and Hochstrasser, 2003). In addition, more indirect effects of the decreased level of Ulp1 at the nuclear envelope, such as altered sumoylation of E3 ligase(s), could potentially also interfere with the sumoylation status of specific substrates. In addition to the sumoylation status of several targets, restoration of catalytically active Ulp1 at the nuclear envelope partially complemented some of the nupΔ repair phenotypes, such as colethality with rad52Δ and rad27Δ or the accumulation of Rad52 foci. However, Ulp1 overexpression poorly suppressed the NHEJ defects of the nucleoporin mutants (Supplemental Figure 4D), and it did not rescue their increased recombination frequency (data not shown). This may reflect the higher sensitivity of this reporter toward properly controlled Ulp1 activity, together with the fact that restoration of Ulp1 functions, as revealed by the aforementioned assays, is only partial. Finally, the growth defects of the nupΔ radΔ strains were enhanced upon mild overexpression of the catalytically inactive form of Ulp1. This indicates that accurate control of the sumoylation levels of some proteins is critical for the viability of nucleoporin mutants in the absence of a functional DNA recombination pathway. Similarly, it was recently reported that a mutant of the Mms21 SUMO-ligase exhibits DNA damage sensitivity (Zhao and Blobel, 2005), whereas the Slx5–Slx8 complex was suggested to be involved both in DNA integrity and sumoylation processes (Wang et al., 2006; Yang et al., 2006). Thus, our data further strengthen the connections between sumoylation and the control of genome integrity (Soustelle et al., 2004; Jacquiau et al., 2005; Motegi et al., 2006).

We thank Anita Corbett, Emmanuelle Fabre, Gérard Faye, Heidi Feldmann, Pierre-Emmanuel Gleizes, David Goldfarb, Martine Heude, Ed Hurt, Won-Ki Huh, Rodney Rothstein, Mike Rout, Roger Tsien, Laurence Vernis-Beringue, Thomas Wilson, and Rick Wozniak for reagents and/or discussion; Sophie Loeillet and Siau-Wei Baï for help with yeast handling; and Alain Nicolas for interest in this work. Many thanks to the Curie Imaging staff and to all members of the Doye, Bornens, and Nehrbass laboratories for valuable comments during the course of this work. This research was supported by a collaborative program between the Institut Curie and the Commissariat à l'Energie Atomique (PIC Paramètres Epigénétiques grant to V.D.), the Ligue Nationale contre le Cancer (Equipe Labelisée 2006 to V.D.), the Association pour la Recherche contre le Cancer (ARC to V.D.), the Ministère de l'Education Nationale, de la Recherche et de l'Enseignement Supérieur (A.C.I. “Jeunes chercheurs” to V.D.), a Cancer Center Support Grant (NCI P30 CA-08478-41 to X.Z.), and the Ministry of Science and Education of Spain (grant SAF2003-00204 to A.A.). B.P. was supported by postdoctoral fellowships from ARC and Fondation Pour la Recherche Médicale.