Chromosome Dynamics in the Yeast Interphase Nucleus

See allHide authors and affiliations

Science  07 Dec 2001:
Vol. 294, Issue 5549, pp. 2181-2186
DOI: 10.1126/science.1065366


Little is known about the dynamics of chromosomes in interphase nuclei. By tagging four chromosomal regions with a green fluorescent protein fusion to lac repressor, we monitored the movement and subnuclear position of specific sites in the yeast genome, sampling at short time intervals. We found that early and late origins of replication are highly mobile in G1 phase, frequently moving at or faster than 0.5 micrometers/10 seconds, in an energy-dependent fashion. The rapid diffusive movement of chromatin detected in G1 becomes constrained in S phase through a mechanism dependent on active DNA replication. In contrast, telomeres and centromeres provide replication-independent constraint on chromatin movement in both G1 and S phases.

Most information available on the nuclear organization of chromosomes is based on the analysis of fluorescent probes in formaldehyde-fixed cells. It is generally assumed that chromosomes, which occupy distinct “territories” in mammalian nuclei (1), are relatively static when compared to metaphase chromosomes. This notion of nuclear order is supported by hybridization with specific DNA sequences: Centromeric heterochromatin is often found at the nuclear periphery or surrounding nucleoli, as are transcriptionally silent chromosomes, like the inactive X (2). In budding yeast, roughly 70% of telomeres (3) and late-firing origins (4) are positioned near the nuclear envelope in G1 phase, with telomeres clustering in three to eight foci. Furthermore, using a green fluorescent protein (GFP)–tagged (5) centromere, relative movement between two homologous yeast chromosomes was measured, showing that centromere III is confined to a subnuclear zone of restricted radius (≤0.25 μm) (6).

Nonetheless, nuclear organization may not be as static as these studies suggest. In Drosophila larval nuclei, large chromosomal movements have been observed in early G1 phase (7), and both large and small movements have been confirmed by real-time microscopy for a single locus in the premeiotic nuclei of fly spermatocytes (8). To examine the dynamics of interphase chromatin in yeast, we exploited the lacoperator-repressor interaction to detect four distinct chromosomal domains. The expression of a GFP-lac repressor fusion in cells that carry an integrated array of multimerized lacrepressor binding sites (lac op) (5) allows the tracking of specific chromosomal sites by real-time fluorescence microscopy. By combining this with a GFP-tagged copy of a nuclear pore protein (Nup49p) to label the nuclear envelope (9), we can measure movements of the chromosomal DNA in relation to the nuclear periphery or the calculated center of the nuclear plane.

Here, we have integrated a single cluster of 256lac op sites into a haploid strain ofSaccharomyces cerevisiae (GA-1320) either at a late-replicating region between two origins of replication (ARS1412 and ARS1413), located 240 kb from the left telomere of chromosome XIV, or near an early-activated origin on chromosome IV, 908 kb from the left telomere [ARS4-908 (4) (Fig. 1A)]. Both sites are within 10 kb of transcribed genes and the nearest origin of replication, and serve as examples of general chromatin structure. As markers for specialized chromosomal domains, we tagged one site 15 kb from the right end of chromosome VI, preserving the subtelomeric repeat structure (TEL VI R), and a second site 12 kb from the centromere of chromosome IV (CEN IV). The binding of GFP-lac repressor produces a bright spot of ∼0.3 μm in diameter (Fig. 1B), which can be readily distinguished from the more weakly fluorescent ring of the nuclear envelope (⊘ = 1.7 to 2 μm). Cell cycle stages are assigned by correlating the presence and the relative size of the bud, and nuclear size and shape, with DNA content, as detected by fluorescence-activated cell sorting analysis. Rapid capture of images by laser scanning microscopy (the nucleus is scanned within ∼150 ms at 1.5 s repeat intervals) allows us to monitor the movement of these loci over a 5-min period at high resolution (10).

Figure 1

Visualization of interphase chromatin movement in budding yeast. (A) Four chromosomal sites were tagged independently in a haploid S. cerevisiae strain that expresses a GFP-nuclear pore protein fusion (4) (Nup49-GFPS65T). As shown, roughly 10 kb of tandemly repeated lac op consenses (5) were integrated within 15 kb of centromere CEN IV, early replicating origin ARS4-908 (4), telomere TEL VI R, and between two internal late firing origins ARS1412 and ARS1413 [labeled ARS1413 (4)]. Two methods are used to quantify the movements of the tagged chromosomal loci. After alignment of the nuclear pore signals through the time series, absolute movement of the spot is followed frame-by-frame (Fig. 3, Tracking). Alternatively, displacement of the spot is measured relative to either the nuclear periphery (red bars) or the nuclear center (blue bars). The center is calculated using a Metamorph program that defines an optimized circle based on the nuclear pore fluorescence [see details in supplementary methods (11)]. Spot-to-pore distance was measured manually using the Zeiss confocal LSM version 2.5 software. (B) Shown are six frames taken on a Zeiss LSM 510 confocal microscope (63× objective, zoom 3) at ∼1.5 s intervals, showing the haploid G1 phase nucleus of strain GA-1325, tagged at ARS4-908. The bright internal spot represents ARS4-908 and GFP-Nup49, labels the nuclear periphery. Bar, 2 [f4]μ[/f4]m [see time-lapse movie at (11)]. Sequential images can be orthogonally displayed using the Zeiss 510 LSM software, and rotated such that the time axis (z) is displayed horizontally. (C) A 3D projection of 200 images (5 min) of the tagged locus ARS1413, ARS4-908, or CEN IV (yellow/red, as indicated) and nuclear pore (cyan) is shown for G1 phase cells. Different intensities in fluorescence are shown in false color (highest is red, lowest blue). Where indicated, 40 mM CCCP was added 15 min before and during time-lapse microscopy. Standard imaging uses cells from a culture of a GFP-lac op tagged strain at ∼5 × 106 cells/ml, on slides coated with SD agar –histidine + 4% glucose. Time-lapse image capture during 5 min on a Zeiss LSM 510 (200 scans using total 0.5 s scanning time with 1 s pause; 100× Planapo objective and zoom 2 are used unless otherwise indicated). Time-lapse series are available at (11) for each locus studied. (D) Typical measurements of the distance from the center of the ARS1413 focus to the center of the closest GFP-Nup49 signal are plotted in nm from the pore. The plotted distances were scored for movements ≥0.5 μm that occur in under 10 s (examples boxed in red). The total number of large movements scored for each condition is then divided by the total time and expressed as movements per 10 min. (E) Summary of large movement frequencies for ARS1413 under control conditions [1% dimethyl sulfoxide (DMSO)], 40 mM CCCP, 10 μg/ml nocodazole and 300 μg/ml thiabendazole (noc/TBZ), or 0.15 mM NaN3 for 15 min prior to and during time-lapse microscopy. Each value is derived from multiple time-lapse movies of different G1 phase cells. Total minutes analyzed (40 images/min) are: control, 13.3 min; CCCP, 12.4 min, Noc+TBZ, 23.8 min; NaN3, 19.4 min. The error bar indicates the range of values obtained with a 10% uncertainty of distance measurement (large movements monitored as either 0.45 μm or 0.55 μm). ARS1413 cells were also monitored as a function of culture saturation (cells ml−1), and were scanned either in the presence (hatched bars) or absence (gray bars) of 4% glucose. Only mother cells are measured in all experiments (further details in Web fig. 2).

The bright foci representing either ARS4-908 or ARS1413 are extremely dynamic in G1 phase nuclei, showing two characteristic types of movement (Fig. 1B) [Web movies (11)]. In sequential frames, we observe both short fluctuations in position (distances ≤0.2 μm) and, less frequently, larger movements (≥0.5 μm; Fig. 1D and Web fig. 1). We also observe deformations of the entire yeast envelope and oscillatory movements of pores within the nuclear membrane. If we project a series of 200 sequential 2D images horizontally, aligning the nuclear pore signals, we see that the dynamics of the tagged ARS1413 or ARS4-908 loci are distinct from overall nuclear movement or minor fluctuations in nuclear shape (compare bright yellow/red signal of the GFP lacrepressor to cyan signals; Fig. 1, B and C). The movement of the tagged origins is not due to inanimate forces, because the addition of NaN3 arrests this movement completely (Web fig. 2 and Fig. 1E). Importantly, an identical lac op tag inserted near CEN IV does not show equivalent motion (Fig. 1C).

By quantifying movements of the tagged chromosomal site relative to either the nearest point on the nuclear envelope or the center of the nuclear sphere calculated from the GFP pore signal, we eliminate any component of the movement that results from nuclear drift (Fig. 1, A and D). As a measure of large movements, we score for chromatin displacements ≥0.5 μm that occur in less than 10 s (Fig. 1D, red boxes). Their frequency is averaged over multiple time-lapse series and is expressed as large movements per 10 min.

The dynamic chromatin behavior appears not to reflect simple Brownian motion, i.e., movement resulting from random indirect forces upon the chromosomal fiber, because exposure of yeast to carbonyl cyanide chlorophenyl hydrazone (CCCP), an uncoupler of the mitochondrial F1Fo adenosine triphosphatase, largely suppresses the movement (Fig. 1, C and E, and Web movie). This protonophore also collapses plasma membrane potential, reduces intracellular ATP levels, and arrests cell growth (12). On the other hand, the frequency of large movements is not altered by the presence of the microtubule-destabilizing drugs, nocodazole and thiabendazole (+ noc/TBZ, Fig. 1E) and is therefore unlikely to be mediated by microtubule-dependent motors.

To further correlate large chromatin movements in G1with the cell's metabolic status, ARS1413-lacop dynamics are followed as cell density increases in a growing culture. The frequency of ≥ 0.5 μm movements drops fourfold in G1 phase as cell density increases, an effect enhanced when microscopy is monitored on glucose-depleted media (Fig. 1E, dark bars; Web fig. 2). The drop in mobility occurs at ∼2 × 107 cells/ml, or roughly one generation before the first changes in gene expression and chromatin structure that accompany the diauxic shift, a transition from fermentative and oxidative metabolism (13), suggesting that large-amplitude movements may correlate with intracellular ATP levels. The smaller, oscillatory movements are not abolished for ARS1413 and ARS4-908, even in stationary-phase cells.

It has been proposed that both centromeres and telomeres are clustered near the yeast nuclear periphery (3, 4). Consistently, our time-lapse analyses on cells bearing a GFP-tagged natural telomere (TEL VI R) or CEN IV (Fig. 1D) reveal an absence of large movements for both centromeric and telomeric tags, although smaller movements are readily detected at CEN IV (Web fig. 3 and Web movies). The telomere oscillates adjacent to the nuclear envelope, such that the distance from the tag to the nuclear pore rarely exceeds 0.3 μm. Nonetheless, TEL VIR moves significantly more than an integral nuclear envelope component (see below, Fig. 3C). The constraint on telomere mobility persists through S phase until late G2 phase, when the nucleus deforms to enter the daughter cell (14).

By scoring early S phase cells (shortly after bud emergence), we note that the large-amplitude movements of both tagged origins are significantly less frequent once cells enter S phase, although the smaller oscillatory movements continue (Web fig. 3). In contrast, the amplitude of centromere movement remains unchanged between G1 and S phase cells. To determine whether this suppression reflects a reduced confinement radius, we have calculated the mean squared displacement over time (MSQD), which was previously used to monitor spatial constraints on the relative movements of two homologous sites in diploid yeast or fly cells (6, 8). When the square of the displacement within discrete time intervals (ranging from 1.5 to 150 s) is plotted as a function of increasing time intervals (Δt), a monotonic increase would indicate unconstrained movement or free diffusion. A rapid increase followed by a plateau indicates that the movement is constrained within a limited area (or sphere), the radius of which can be estimated. By plotting the MSQD curves for each locus averaged over all available time intervals in either G1 or S phase cells, we obtain a reliable estimate of each locus' degree of freedom.

Figure 2 presents the MSQD graphs averaged over multiple movies of each of the four chromosomal loci we have tagged [examples of each are at (11)]. The initial slope (i.e., diffusion coefficient) and plateau values for CEN IV (Fig. 2C) are remarkably similar to those previously reported for CEN III (6) (0.05 μm2, indicating a radius of confinement ≤0.3 μm). The curves and plateau values do not change significantly between G1 and S phase for CEN IV and TEL VIR, whereas the radii of constraint calculated for ARS1413 and ARS4-908 are significantly lower in S than in G1 phase. In G1, the biphasic curve increases until time intervals of 75 and 100 s, with a plateau at 0.1 to 0.15 μm2. A plateau at this value was previously shown (6) to correspond to a radius of confinement of roughly 0.7 μm, or an occupied volume 10-fold larger than that of the yeast centromere or of origins in S phase (Fig. 2, A through C, panels S). The S-phase MSQD curves of ARS1413, ARS4-908, and CENIV are all similar, starting with a steep slope that arrives at a plateau of ∼0.05 μm2(radii of confinement ≤0.3 μm).

Figure 2

Origins are less constrained than telomeres and centromeres in G1 phase, but all sites show spatial constraint in S phase. Mean squared displacement analysis was performed as described (6, 8), by computing the square of the displacement of the indicated focus (measured from the center of the chromosomal tag to either the nuclear center or the nuclear periphery) in μm (Δd 2) as a function of the time interval Δt (s): Δd 2 = {d(t) –d(t + Δt)}2 for Δtranging from 1.5 to 150 s. The average of all Δd 2 values for each Δt value are plotted against Δt. The slope is the derivative of displacement or diffusion coefficient. The graphs represent averages from all movies available for each condition with standard deviation indicated. The number of nuclei and images analyzed for each phase areas follows : (A) G1: 10 nuclei / 2000 frames; S: 5 / 1000, (B) G1: 8/ 1600; S: 3/ 600, (C) G1: 4/ 800; S: 3/ 600, (D) G1: 3/ 600; S: 4/ 800.

The displacement measurements shown above indicate changes in spatial constraints dependent on the cell cycle, but do not identify position within the nucleus. To visualize the localization and path length of a given tagged domain over 5 min, we align the nuclear pore signals of a given time-lapse series, and follow the path of the tagged locus from frame to frame with a tracking tool. The cumulative path (Fig. 3, in red) is then projected onto a single nuclear section. The sum of the vectors' lengths indicates a minimal linear distance (in μm) traveled within 300 s (15). By projecting the entire time-lapse series into one plane and calculating the surface area that encompasses the path, a value related to the radius of confinement is obtained (Fig. 3, “Projection,” and Web fig. 5).

Figure 3

TEL VIR and CENIV are constrained to perinuclear zones, while other loci have less restricted subnuclear positions. (A) Isogenic yeast strains carrying the indicated GFP-tagged locus and GFP-Nup49 were analyzed under standard conditions as described above, in G1 and S phase. Nuclear pore signals were aligned for each time-lapse series and the position and total distance traveled by the indicated tag was measured using the AIM tool of the Zeiss LSM 510 software (release 2.8). The trajectory over 300 s (in red), is superimposed on a single nuclear focal section (Tracking). The mean length of the path in μm averaged over four or five movies is indicated, and standard deviations range from less than 1 μm (for TEL VIR) to 8 μm (ARS4-908). An intensity projection of the same stack of images is shown (Projection), and the area occupied by the accumulated signal covers on average 30 ± 4% (n = 10) of the nuclear surface for ARS1413 and ARS4-908 in G1 and 20 ± 5% (n = 8) in S phase. TEL VIR occupies 10% [± 3% (n = 4)]. (B) Tracking and trajectory measurement are shown for CEN IV in G1 and S phase cells, either in the presence of DMSO alone, or with nocodazole and thiabendazole (± noc/TBZ; Fig. 1). The amplitude and average path length changes little from G1 to S, although destabilization of microtubules shifts CEN IV internally in some S-phase nuclei. Trajectory distances are given in μm (n = 4 for S phase, n = 1 for G1). (C) Time-lapse microscopy and tracking was performed for a CFP-tagged spindle pole body component, Spc42p (17). Because Spc42p is in the nuclear envelope, we trace its movement with respect to the transmitted light image of the cell. The upper image is on the same scale as all other tracking images, while the lower transmitted light image of the entire cell is half this magnification. A white dotted line encircles the nucleus (nuc) and vac, the vacuole. Bars, 2 μm. The average of Spc42p trajectory length is 5.8 ± 1.2 μm (n = 3).

Consistent with MSQD measurements, tracking confirms that the tagged origins move less in S phase than in G1 (Fig. 3A). It also underscores the close association of TEL VIR with the nuclear envelope (3, 4). In addition, we see that the movements have no cumulative directionality (16) and that ARS4-908 and ARS1413 can “sample” large areas of the nucleus within 5 min. These loci in G1 phase cells have the least spatial constraint, whereas TEL VIR remains primarily juxtaposed to nuclear periphery. Our observations do not refute the notion of nuclear territories (1), yet they suggest that such territories are loosely defined in yeast, perhaps entailing only a few anchorage sites.

A mean linear trajectory for each locus was calculated on the basis of multiple 5-min movies and is indicated in each panel (Fig. 3). Although TEL VIR remains close to the nuclear envelope, it moves three times the distance of an integral component of the nuclear envelope, a 42 kD spindle pole body protein (17) (Fig. 3, A and C), indicating that at least one aspect of telomere movement is independent of the nuclear envelope. Exposing S phase cells to microtubule depolymerizing agents provokes a more frequent migration of CEN IV to the interior of the nucleus, but does not significantly increase the amplitude of its movement (Fig. 3B, + noc/TBZ), although nocodazole does not alter the position or amplitude of movement for the other tagged loci in G1 or S phase (Fig. 1E) (18).

To see if the S phase–specific restriction on the dynamics of ARS1413 and ARS4-908 is a direct effect of DNA replication, we inhibited DNA synthesis by the addition of hydroxyurea (HU), which depletes dNTP pools, or by aphidicolin, a specific inhibitor of DNA polymerase elongation. As expected, in control cells, the frequency of large-scale movements drops between G1 and S phase. Importantly, when S-phase cells are treated with the inhibitors, large movements of both ARS1413 and ARS4-908 are again as frequent as in G1 (Fig. 4A). Neither the loading of the MCM complex by the origin recognition complex (ORC) and Cdc6p, nor its release by initiation, is likely to be the stabilizing factor in S phase, because this assembly occurs in G1, and MCM release does not occur for late-firing origins like ARS1413 in HU (19). Besides arrest fork movement, treatment with either HU or aphidicolin activates the DNA checkpoint pathway (20). However, because the increase in S-phase dynamics in HU occurs in arad53 mutant (Fig. 4A), it is unlikely to be related to a checkpoint response.

Figure 4

Constraint of origin movement in S phase requires active DNA replication. (A) The number of large movements (≥ 0.5 μm) for ARS1413 or ARS4-908 has been quantified as in Fig. 1E, in G1- and S-phase nuclei, with addition of 2% DMSO ± 0.2 M hydroxyurea (HU) or 500 μg/ml aphidicolin. Here, we indicate for each condition the site of the lacop tag, host cell genotype, cell cycle phase, the number of movies analyzed and total time in minutes: GA-1325 (ARS4-908/wild type, G1: 10, 74.2; S: 5, 32.5), GA-1323 (ARS1413/wild type, G1: 8, 59.9; S: 3, 14.4), GA-1325 (ARS4-908/wild type, S + HU: 5, 49.4), GA-1323 (ARS1413/wild type, S + HU: 9, 55.9), GA-1323 (ARS1413/wild type, S + aphidicolin: 3, 19.9), GA-1323 (ARS1413/wild type, S + 2% DMSO: 2, 14), GA-1588 (ARS1413/mec2-1 (20) (= rad53), S: 6, 36.9). (B) Cultures of isogenic ORC2+ (GA-1323) or orc2-1 mutant cells (GA-1591) carrying the ARS1413 tag, were grown at 23°C, to 0.5 × 106 cells ml−1. Time-lapse microscopy was done on half of the culture at 23°C, while half was shifted to 30°C for 30 min and analyzed at 30°C. Center of focus to nuclear center measurements were performed using Metamorph (Universal Imaging Corporation) (11). Large movements were calculated as above, but in this experiment, G1 daughter cells were not excluded from analysis. The ±10% confidence levels are indicated as in Fig. 1E, except for orc2-1 at 23°C, in which ±5% variation is shown (25). Indicated here for each strain is the number of cells, and total imaging time in min: GA-1591/orc2-1 (G1/23°C: 3, 11; S/23°C: 3, 22.1), GA-1591/orc2-1 (G1/30°C: 5, 36.5; S/30°C: 3, 21.7), GA-1323/ORC2+ (G1/23°C: 4, 38.3; S/23°C: 3, 14.4), GA-1323/ORC2+ (G1/30°C: 3, 21.8; S/30°C: 4, 29.8).

In yeast, as in higher eukaryotes, each replication fork is predicted to assemble a complex of ∼2000 kD, and multiple such complexes appear to aggregate at sites of replication (21). If such assemblies are sufficient to impede chromatin mobility, then constraint should reflect the number of active replication forks. This was tested by monitoring relative G1 and S phase mobility in a strain carrying the yeast orc2-1 deficiency. This mutation confers a significant drop in the efficiency of origin firing (22) and a reduction in replication foci formation (23), even at 23°C, by destabilizing the conserved six-subunit ORC (22,24). Indeed, when monitored for large chromatin movements of ARS1413, the orc2-1 mutant cells lose the suppression normally detected in S phase at either 23° or 30°C (Fig. 4B). Tracking data confirm that the distances moved by the tagged ARS1413 in the orc2-1 mutant in G1 and S phases at 30°C are identical, whereas in isogenic ORC2 cells, the linear trajectory decreases by ∼25% in S phase (25).

We have documented a highly dynamic behavior of interphase chromatin in yeast, which moves distances ≥0.5 μm within seconds. We establish that movements detected for different chromosomal domains have different degrees of constraints. Telomeres move significantly less than other regions, consistent with previous data suggesting that they are tethered to the nuclear periphery (3, 4). The amplitude of movement of nontelomeric chromatin appears to be modulated by both the metabolic status of the cell and DNA replication. This could either indicate that the viscosity of the nucleoplasm changes with the cell cycle or that the movement is propelled by an active mechanism, whose activity fluctuates. Alternatively, large replication complexes may impose additional sites of anchorage in S phase. The limited movement reported for interphase chromatin in the mammalian cells may reflect the scale on which such events have been previously monitored (1, 26). Whereas a movement of 0.6 μm covers nearly a third the diameter of the haploid yeast nucleus, it is only 1/30 of the diameter of a typical human cell nucleus (average ⊘ = 15 to 20 μm). Given the speed of these movements, extremely rapid detection is required to accurately score the position of the tag.

Nonetheless, the dynamics scored here in G1 phase nuclei may be a universal characteristic of open chromatin (27), which constitutes 80 to 90% of the yeast genome. The constant sampling of local environment achieved by this movement may help regulate macromolecular interactions, as well as long-range sequence searching, like that associated with homologous recombination, which is particularly efficient in yeast. The apparent energy-dependence of the movement may implicate ATP-dependent enzymes, such as RNA polymerases or chromatin remodeling complexes in chromatin dynamics.

  • * Present address: The Salk Institute, 10010 North Torrey Pines Road, La Jolla, CA 92037, USA.

  • Present address: Euresearch, Effingerstrasse 19, 3001 Bern, Switzerland.

  • To whom correspondence should be addressed. E-mail: susan.gasser{at}


View Abstract

Navigate This Article