Journal of Cell Science 112, 525-535 (1999) 525 Printed in Great Britain © The Company of Biologists Limited 1999 JCS4591

Chromosomes exhibit preferential positioning in nuclei of quiescent human cells

Robert G. Nagele1,*, Theresa Freeman1, Lydia McMorrow2, Zabrina Thomson3, Kelly Kitson-Wind1 and Hsin-yi Lee3 1Department of Molecular Biology, University of Medicine and Dentistry of New Jersey/SOM, 2 Medical Center Drive, Stratford, New Jersey 08084, USA 2Department of Laboratory Science, Thomas Jefferson University, Philadelphia, Pennsylvania, 19107, USA 3Department of Biology, Rutgers University, Camden, New Jersey 08102, USA *Author for correspondence (e-mail: [email protected])

Accepted 20 November 1998; published on WWW 25 January 1999

SUMMARY

The relative spatial positioning of chromosomes 7, 8, 16, X predictable within remarkably narrow spatial limits. Dual- and Y was examined in nuclei of quiescent (noncycling) FISH with various combinations of chromosome-specific diploid and triploid human fibroblasts using fluorescence DNA probes and contrasting fluorochromes was used to in situ hybridization (FISH) with chromosome-specific identify adjacent chromosomes in mitotic rosettes and test DNA probes and digital imaging. In quiescent diploid cells, whether they are similarly positioned in interphase nuclei. interhomolog distances and chromosome homolog position From among the combinations tested, chromosomes 8 and maps revealed a nonrandom, preferential topology for 11 were found to be closely apposed in most mitotic rosettes chromosomes 7, 8 and 16, whereas chromosome X and interphase nuclei. Overall, results suggest the existence approximated a more random distribution. Variations in of an ordered interphase chromosome topology in the orientation of nuclei on the culture substratum tended quiescent human cells in which at least some chromosome to hinder detection of an ordered chromosome topology at homologs exhibit a preferred relative intranuclear location interphase by biasing homolog position maps towards that may correspond to the observed spatial order of random distributions. Using two chromosome X homologs chromosomes in rosettes of mitotic cells. as reference points in triploid cells (karyotype = 69, XXY), the intranuclear location of chromosome Y was found to be Key words: Mitosis, Chromosome, Chromosome topology, Human

INTRODUCTION chromosome position in interphase nuclei would imply that the nuclear context in which genes operate is critical for their Our understanding of the structure and organization of the normal expression as originally suggested by Blobel (1985). interphase nucleus have recently undergone a major revision Thus far, the results of studies on chromosome topology in due to parallel advancements including refinements in the interphase nuclei have been contradictory. A number of studies fluorescence in situ hybridization (FISH) technique, the on different cell types have shown that the intranuclear locations development of a wide variety of chromosome- and gene- of at least some chromosome territories are nonrandom and may specific DNA probes, and dramatic advances in light be both cell type- and cell cycle-specific (for reviews, see microscopy and digital imaging. For example, a number of Heslop-Harrison and Bennett, 1984; Manuelidis, 1990; Haaf and studies using FISH with chromosome painting probes have Schmid, 1991; Spector, 1993). For example, in human and shown that chromosomes occupy relatively compact, non- mouse central nervous system cells, some chromosome domains overlapping territories in interphase nuclei of both animal and exhibit similar distribution patterns (Manuelidis and Borden, plant cells (Cremer et al., 1982, 1988; Lichter et al., 1988; 1988). In human fibroblasts, the distribution of chromosome 8 Hilliker and Appels, 1989; Schwarzacher et al., 1989; Leitch centromeric regions is nonrandom and changes during the cell et al., 1990; Manuelidis, 1990; Haaf and Schmid, 1991; cycle (Popp et al., 1990; Ferguson and Ward, 1992). Spector, 1993). The question as to whether or not these Chromosome 1 centromeric DNA localizes selectively to the chromosome territories indeed exhibit a relative spatial order nuclear periphery in hemopoietic cells (Van Dekken et al., 1989). within interphase nuclei has been hotly debated over many Likewise, the inactivated X chromosome maintains a peripheral years (Comings, 1968, 1980; Heslop-Harrison and Bennett, distribution in interphase nuclei of fibroblasts (Manuelidis and 1984; Spector, 1993). The biological significance of this issue Borden, 1988). In addition, distribution patterns of centromeres cannot be overrated, particularly because control over and telomeres vary with cell type and cell cycle phase, and may 526 R. G. Nagele and others be influenced by the state of cell differentiation (Ferguson and Cell Repository). Cells were grown on glass coverslips in Dulbecco’s Ward, 1992; Manuelidis, 1985; Hadlaczky et al., 1986; Haaf and Modified Eagle Medium (DMEM) (Gibco, BRL) containing 10% fetal Schmid, 1989; Bartholdi, 1991; Vourc’h et al., 1993). Active calf serum (Gibco, BRL) and 1% penicillin/streptomycin at 37°C in a genes are also distributed nonrandomly in nuclei, concentrating 5% CO2 atmosphere. Some cultures were grown to a state of ‘high- in the nuclear periphery of mouse L and P19 embryonal density confluence’ and maintained in this condition for 2-4 days carcinoma cells and at the edges of condensed chromatin in newt without medium change. Cell cycle analysis (see below) showed that this procedure ‘locked’ cells in the G1/G0 phase of the cell cycle and erythrocytes (Hutchison and Weintraub, 1985). On the other effectively minimized individual cell and nuclear shape variations that hand, a number of other studies have failed to detect distinct would otherwise have arisen as a result of such complex behaviors as chromosome distribution patterns in nuclei, and thus support a cell migration, cell spreading, and mitosis. Confirmation of cell cycle random intranuclear distribution of chromosomes at interphase. arrest was obtained by DNA quantitation, preparation of cell cycle For example, Vourc’h et al. (1993) have shown that chromatin analysis profiles as described below, and examination of cultures for the is redistributed in the interphase nucleus of mouse lymphocytes occurrence of mitotic figures. during the cell cycle. Similarly, Lesko et al. (1995) demonstrated that the relative distributions of chromosomes 7, 11 and 17 in Cell cycle analysis interphase T lymphocytes could not be distinguished from those High-density confluent cell cultures were analyzed using the Cell obtained from random points in a truncated sphere. This Analysis System 200 (CAS 200, Beckman). The CAS 200 uses a stoichiometric Feulgen staining reaction and microscopic observation is in agreement with the data of Ferguson and Ward densitometry to determine the total DNA content of individual nuclei (1992), who measured the distances and angles between and generates cell cycle profiles for cultures (Fig. 1). The main homologs in flow-sorted T lymphocyte nuclei and reported that advantages of the CAS 200 over fluorescence cell sorters is that the the arrangement of chromosome homologs relative to each other analysis can be done on cells already affixed to microslides and a is not spatially defined. Arnoldus et al. (1989) reported the single nucleus can be analyzed by FISH. Based on the total amount interphase association of the two 1q12 heterochromatin regions of nuclear DNA, each nucleus can be assigned a specific phase of the in nuclei from the human cerebellum, but were unable to cell cycle. In the present study, the CAS 200 was also used to confirm demonstrate this association in the cerebral cortex. that cells in high-density confluent cultures were ‘locked’ in the G0/G1 Heterochromatin of the active X chromosome in cultured human phase. Data are presented as a histogram which shows the relative cells has been reported to be randomly distributed (Popp et al., numbers of cells in the G0/G1, S, and G2/M phases. For individual nuclei, results are expressed as picograms of DNA per nucleus. The 1990). Taken together, the existence of an organized replication precision of measurements of individual nuclei and the chromosome topology at interphase is still a matter for debate, stability of calibration were tested by comparing measurements of 25 and a common organizing principle that influences or dictates consecutive images of the same nucleus. interphase chromosome topology remains elusive. In the present study, we have investigated the topology of Fluorescence in situ hybridization (FISH) and digital chromosomes 7, 8, 16, X and Y in quiescent human diploid imaging and triploid fibroblasts using FISH and digital imaging. Cells grown on glass coverslips were fixed in 4% paraformaldehyde in Quiescent cells were chosen in an effort to minimize potential phosphate-buffered saline (PBS; 137 mM NaCl, 3 mM KCl, 16 mM variations in chromosome topology stemming from differences Na2HPO4, 2 mM KH2PO4, pH 7.3) for 20 minutes at room temperature. in cell type, cell cycle phase, cell and nuclear shape and Fixed specimens were washed briefly in PBS, followed by three rinses in 2× SSC (1× SSC; 150 mM NaCl, 30 mM sodium citrate) for 10 physiological state. They also provide an excellent model minutes at room temperature. For digoxigenin-labeled, whole- system because cells and their nuclei are flattened against the chromosome painting probes, cellular DNA was denatured at 70°C for culture substratum during much of the cell cycle such that the 2 minutes in 70% formamide/2× SSC and probes were prepared for intranuclear arrangement of chromosome territories is nearly hybridization following the recommendations of the manufacturer two-dimensional. This allows most fluorescent signals to be (Oncor, Inc.). Coverslips were mounted on slides with 15 µl DNA probe visualized and recorded photographically from a single focal solution and sealed with rubber cement. For chromosome-specific alpha- plane, which greatly facilitates analysis and interpretation of satellite probes, both probe and cellular DNA were co-denatured at 95°C signal distribution patterns and permits the study of a greater for 10 minutes on a heat block and hybridized overnight in a humid number of cells than would be otherwise possible using environment at 37°C. For dual FISH, subconfluent cells were hybridized confocal microscopy. Our overall results suggest an ordered with the following combinations of whole chromosome painting probes and chromosome-specific alpha-satellite probes, allowing detection with interphase chromosome topology in both diploid and triploid contrasting fluorochromes: chromosomes 1 and 17, 2 and X, 3 and X, 4 quiescent human fibroblasts in which some chromosome and X, 6 and X, 8 and X, 9 and X, 10 and X, 11 and X, 8 and 11, 11 homologs exhibit preferred relative intranuclear locations. and 14, 13 and 18, 14 and 16, 14 and 22, 16 and 21, 16 and 22, 17 and Furthermore, we provide evidence that the relative location of 18. Individual probes were labeled with digoxigenin, biotin or Cy3. After chromosomes in interphase nuclei is related to the observed hybridization, specimens were washed at stringencies recommended by spatial order of chromosomes in rosettes of mitotic cells. the manufacturer for each probe. For detection of digoxigenin-labeled probes, specimens were first treated with 4% BSA in 4× SSC containing 0.1% Tween-20 (SSC-Tween buffer) for 10 minutes at room temperature MATERIALS AND METHODS to block non-specific binding of detection reagents. Detection was performed with FITC-conjugated sheep anti-digoxigenin antibodies Cell culture (200 µg/ml; Boehringer Mannheim) for 20 minutes at 37°C. Biotinylated The following cell lines were used in this study: normal human diploid probes were detected with avidin-FITC (Sigma). After three washes in fibroblasts (AG07715, Coriell National Cell Repository), Detroit 551 SSC-Tween buffer, cells were mounted in Vectashield mounting solution (American Type Culture Collection), and triploid skin fibroblasts cells (Vector Laboratories) containing DAPI or propidium iodide as a DNA (AG05025, karyotype 69, XXY) derived from a conceptus exhibiting counterstain. Controls for FISH included hybridization without labeled multiple congenital malformations typical of triploidy (Coriell National probe or omission of the detection reagent that binds to the probe. Chromosome positioning in interphase cells 527

Specimens were examined with a Nikon Optiphot or Nikon FXA 100 microscope equipped with epifluorescence optics and a Princeton Instruments CCD camera. The DAPI or propidium iodide image was 80 used to define the morphological boundary of nuclei and mitotic 60 chromosome rosettes. For most photographs, a ×60, 1.40 numerical aperture objective was used. Some fluorescent images were recorded on 40 either Kodak T-Max 400 black-and-white film or Fujichrome 400 color

Cell Count 20 film. Separate digital images of the FITC- or Cy3-labeled chromosome- specific signals and the DAPI- or propidium iodide-counterstained nuclei 0 0 3 6 9 21 12 15 18 24 30 were acquired, aligned and processed using Metamorph (Universal 27 33 Imaging) or ImagePro Plus (Phase 3 Imaging). Images were retrieved DNA Mass (pg) from hard disks or CD-ROM and printed using a Sony dye-sublimation printer. Some were processed with Corel 6 commercial graphics Fig. 1. Cell cycle profile of quiescent human diploid fibroblasts software (Corel). showing the relative numbers of cells in the G1/G0, S and G2/M phases of the cell cycle. The DNA content of individual nuclei was Analysis of chromosome homolog distribution patterns determined using a stoichiometric Feulgen staining reaction and The distribution patterns of selected chromosomes in interphase microscopic densitometry. Results are expressed as pg DNA per nuclei of quiescent fibroblasts were analyzed by preparing homolog nucleus. Cell cycle profiles showed a general lack of S-phase cells position maps. This method takes advantage of the fact that nuclei of and, excluding tetraploid cells, >98% of cells in G1/G0. human fibroblasts in quiescent cultures are remarkably flat and uniform in shape and size. Furthermore, cells are densely packed and often mutually aligned such that, within a single viewing field, nearly physiological activity were minimized by maintaining cells at all appear to be oriented with their long axes parallel to one another high-density confluence to render them quiescent (noncycling). (cf. Fig. 2). Nuclei were in situ-hybridized with chromosome-specific The cell cycle profile of these cultures was determined using DNA probes, counterstained with DAPI or propidium iodide, and the CAS 200 and the noncycling status of cultures under images were recorded as described above. Outlines of fluorescent conditions of high-density confluence was confirmed by the chromosome territories were revealed with whole-chromosome painting probes, and the area enclosed by each nucleus was recorded lack of S-phase cells and absence of mitotic figures. More than for each cell by tracing directly from magnified projections. The final 98% of cells were in the G0/G1 phase with the usual small magnification of projected images was held constant to facilitate percentage of stable tetraploid cells (Fig. 1). To analyze the direct comparison among nuclei, and unprocessed images were used spatial positioning of selected chromosomes in nuclei, we used to generate homolog position maps. x-y coordinates were assigned to FISH and digoxigenin-labeled painting probes for each homolog by superimposing a grid over each nucleus, using the chromosomes 7, 8 and 16, an alpha-satellite DNA probe for center and long axis of the nucleus as reference points. The small chromosome X, and a Cy3-labeled, alpha-satellite DNA probe number of nuclei that overlapped or deviated markedly from the for chromosome Y. Probes for autosomes hybridized to only typical elliptical shape were excluded from the analysis (generally less one pair of chromosomes and produced two fluorescent signals than 10% of cells in a viewing field). Maps showing the frequency of in quiescent diploid nuclei. homologs being positioned at specific locations within nuclei for each chromosome were prepared using the x-y coordinates assigned to each Under conditions of high-density confluence, most cells homolog and the Corel Draw 6.0 graphics program (Corel). were densely packed and aligned with their long axes parallel to one another (Fig. 2A,B). In addition, cells and their nuclei were remarkably uniform in shape and size. Nuclei were highly RESULTS flattened (Table 1) and ellipsoidal (mean nuclear length = 24.07±2.86 µm; mean nuclear width = 12.46±2.24 µm; n = 690 Chromosome territories form a two-dimensional cells). The flattened geometry of nuclei under these conditions array in the flattened nuclei of quiescent human generates a nearly two-dimensional distribution of fibroblasts chromosome territories, which allows chromosome-specific Potential variations in chromosome topology associated with fluorescent signals to be visualized in a single focal plane in differences in cell cycle phase, cell and nuclear shape, and most cells (Fig. 2B). Further evidence for a two-dimensional

Fig. 2. (A) Phase-contrast image of quiescent diploid human fibroblasts under conditions of high-density confluence. (B) Reverse-contrast image quiescent cells showing nuclei of similar size and shape. Cells were hybridized with painting probes for chromosome 8 and counterstained with propidium iodide. Bars, 30 µm. 528 R. G. Nagele and others

Table 1. Measurements of nuclear thickness in quiescent exceptionally high degree of flattening often showed cells corresponding flattening of individual chromosome territories. Nuclear thickness (µm) Analysis of homolog separation distances reveal a Experimental Number of nonrandom chromosome technique mean s.d. nuclei As a first test for a nonrandom chromosome topology in DAPI-stained nuclei of quiescent diploid cells, we analyzed the spatial Central 2.26 0.48 50 Peripheral 2.04 0.41 50 relationship among homologs of chromosomes 7, 8, 16 and X through measurements of homolog separation distances FISH with centromere probe (point-to-point distances between the centers of homologous Central 2.29 0.43 50 Peripheral 2.12 0.43 50 chromosome territories) in specimens subjected to FISH with chromosome-specific DNA probes. The rationale for this Measurements of nuclear thickness were made at the central and peripheral approach is based on the fact that a nonrandom distribution nuclear regions of DAPI-stained cells and cells FISHed with ‘all human of chromosome homologs would be revealed by a nonrandom centromeres’ probe (Oncor) using a Nikon Optiphot microscope equipped with a K2S BIO spinning disk confocal attachment and a 60× PlanApo oil distribution of measured homolog separation distances objective (NA=1.40). (Vourc’h et al., 1993; Lesko et al., 1995). On the other hand, For DAPI-stained cells, the upper and lower limits of nuclei were random positions would result in a distribution of homolog determined using a focal series and the vertical distance traversed was separation distance measurements that reflect a random recorded from the calibrated microscope fine-focusing vernier. profile. Furthermore, statistically significant deviations from In nuclei hybridized with ‘all human centromeres’ probe, the upper limits of the focused first signal seen and the focused lower limits of the last signal a random profile would support a nonrandom spatial seen in a similar series were taken as representing the upper and lower limits distribution of chromosome homologs, the degree of which of nuclei, respectively. Analysis by t-test revealed no significant differences should be proportional to the magnitude of digression from between any of the data sets at P<0.01. randomness. Measurements of homolog separation distances were made on projected FISH images of over 500 cells for arrangement of chromosome territories in quiescent cells each of the four chromosomes examined (Fig. 3A-D). comes from the fact that they were never overlapped in Fluorescent signals from the chromosome X-specific alpha specimens treated with single or multiple whole chromosome- satellite DNA probes were small and punctate compared to painting probes. A small percentage of nuclei exhibiting an the more diffuse labeling of painted-chromosome territories

CHROMOSOME 8 CHROMOSOME 7 35 35 30 30 25 25 20 20 15 15 10 10 5 5 % OF NUCLEI % OF NUCLEI 0 0 0-2 2-4 4-6 6-8 0-2 2-4 4-6 6-8 8-10 8-10 10-12 12-14 14-16 16-18 18-20 20-22 10-12 12-14 14-16 16-18 18-20 20-22 A DISTANCE (µm) B DISTANCE (µm) CHROMOSOME 16 CHROMOSOME X Fig. 3. Distribution of measured 35 interhomolog distances of 35 chromosomes 7, 8, 16 and X in 30 30 normal human diploid cells presented 25 25 as histograms with 10 bins 20 superimposed over a random 20 distribution for two signals obtained 15 15 from simulated measurements of 10 10 2,500 ellipsoidal nuclei. Distribution 5 profiles for chromosomes 7, 8 and 16 5

% OF NUCLEI 0 deviated significantly from a random 0 % OF NUCLEI distribution at the 99.9% confidence 0-2 2-4 4-6 6-8 0-2 2-4 4-6 6-8 8-10 level, whereas chromosome X 8-10 10-12 12-14 14-16 16-18 18-20 20-22 approached a random distribution µ 10-12 µ 12-14 14-16 16-18 18-20 20-22 (Table 2). C DISTANCE ( m) D DISTANCE ( m) Chromosome positioning in interphase cells 529

Table 2. Comparison of interhomolog distances with a simulated random distribution Interhomolog distances (µm) Chromosome Kolmogorov-Smirnov Number of number mean s.d. median Z value nuclei 7 8.8 3.9 8.0 2.21 (P<0.001) 537 8 10.4 4.1 10.0 3.41 (P<0.001) 563 16 6.3 3.2 5.0 3.78 (P<0.001) 525 X 8.2 4.1 7.65 1.04 (P<0.234) 515

The mean value for a random distribution of points is 8.03 µm with s.d. 4.3 µm, as determined in a simulation of 2,500 ellipsoidal nuclei (length, 24.0 µm; width, 12.5 µm). The median homolog separation distance in this simulation is 7.8 µm. and, thus, were taken to represent the positions of Chromosome position maps reveal nonrandom, chromosome X homologs. Selection criteria for cells chromosome-specific intranuclear distributions included an ellipsoidal nuclear shape and dimensions that fall For determining the intranuclear locations of chromosomes 7, within one standard deviation of the mean length and width 8 and 16, whole chromosome painting probes were used and of nuclei. Results from measurements of homolog distances the center of the ‘painted’ chromosome territory was taken as for chromosomes 7, 8, 16 and X are presented as histograms representing the position of the chromosome. The fluorescent with 10 bins in Fig. 3. Mean homolog separation distances were clearly different among the four chromosomes examined, suggesting that these chromosomes were not Chromosome 7 Chromosome 8 randomly distributed (Table 2). To test for significant differences between experimental histograms and random simulations, we calculated the mean, s.d., median and the Kolmogorov-Smirnov (K-S) statistic for each distribution (Table 2). The K-S one-sample, two-tailed statistic (Young, 1977) was used to determine if the experimental distributions could be distinguished from a random simulated distribution. To implement the K-S statistic calculation, we used a model distribution composed of simulated measurements of the distance separating two points distributed randomly within 2,500 hypothetical nuclei. Table 2 shows that chromosomes 7, 8 and 16 are not randomly distributed (Fig. 3A-C). A B However, chromosome X homologs were found to n=390 n=250 approximate a random distribution using this method (Fig. 3D). A comparison of interhomolog distance profiles for Chromosome 16 Chromosome X chromosomes 7, 8, 16 and X is presented in Table 3, which shows that the distribution of interhomolog distances for each chromosome differs significantly from the others, again suggesting a nonrandom distribution for three of the four chromosomes examined using this approach.

Table 3. Comparison of interhomolog distance profiles among chromosomes 7, 8, 16 and X Chromosomes compared t probability value C D 7 and 8 8.14×e−7 n=378 n=326 7 and 16 2.28×e−15 7 and X 0.033 Fig. 4. Homolog position maps for chromosomes 7, 8, 16 and X − 8 and 16 3.03×e 35 showing the relative frequency at which chromosome homologs × −18 8 and X 2.0 e occupy specific intranuclear positions as defined by assigned x-y × −8 16 and X 3.89 e coordinates. Relative intranuclear locations of homologs were Data used in generating separation distance profiles for each set of revealed using chromosome-specific DNA probes. x-y coordinates chromosome homologs (see Fig. 3A-D) were analyzed using the t-test to were assigned using the exact center and long axis of the nucleus as determine differences among the experimental means. Results are presented reference points. (A,B) Most chromosome 7 and 8 homologs in terms of the probability that there is no significant difference between the occupied nonrandom positions at or near the nuclear periphery as means of measured homolog spatial separation among the chromosomes shown by clusters of points. (C) Chromosome 16 preferred a more compared. For example, a t probability value of 1.00 indicates that the two central location, with homologs clustered along lines representing the experimental means are not significantly different from each other and that x and y axes. (D) Chromosome X homologs were distributed rather both sets of data are virtually indistinguishable. uniformly throughout the nucleus. 530 R. G. Nagele and others

For example, 32% of nuclei had their chromosome 8 homologs 35 in the relative positions designated as Category 1, whereas 30 Category 5 was represented by only 2% of total nuclei. This demonstrates a clear preference for nuclei of quiescent cells to 25 have their chromosome 8 homologs in a Category 1 pattern. In 20 fact, Categories 2-4 differed only slightly from Category 1 and, 15 together, Categories 1-4 accounted for 65% of total nuclei. How is it that 65% of total nuclei can possess such similar relative 10 chromosome homolog positions, yet homolog positional maps 5 tend to obscure this fact? Part of the answer lies in the 0 orientation of individual nuclei on the culture substratum within Percentage of total nuclei Percentage 1234567 89 the x-y plane. In our chromosome position maps, we made no attempt to reorient or rotate ellipsoidal nuclei within the x-y plane in an effort to match chromosome distribution patterns. Under these circumstances, homologs of two identical nuclei, oriented at 180° angles with respect to each other in the x-y Homolog position categories plane, were recorded on the chromosome position map as four widely separated homologs. This apparent opposite nuclear Fig. 5. Categories (1-9) of relative homolog positions for polarity has the effect of strongly biasing chromosome position chromosome 8. The graph shows the percentage of total nuclei with maps toward a more random distribution. Another factor relative chromosome 8 homolog positions fitting into each of nine contributing to the apparent randomness of chromosome most frequently encountered position categories. The number of homolog distributions in such maps has its origin from mitosis. nuclei in each category was found to vary considerably. The results In a previous study (Nagele et al., 1995), we showed that demonstrate a clear preference for quiescent diploid cells to have their chromosome 8 homologs in a Category 1 pattern. chromosomes in mitotic human cells are arranged into a wheel- like array called a rosette, which persists throughout mitosis (Fig. 6A,D). In subconfluent cultures subjected to FISH with signal from chromosome X-specific alpha-satellite probe was chromosome-specific probes, adjacent newly formed daughter more punctate and its location was used to record the location nuclei often display either identical (Fig. 6E,F) or mirror-image of this chromosome. x-y coordinates were assigned by (Fig. 6G,H) chromosome homolog distribution patterns. These superimposing a grid over each nucleus and using the exact patterns occur with a slight preference for daughter cells to have center and long axis of the nucleus as reference. Results identical, rather than mirror-image, homolog distributions obtained from a minimum of 250 homologs for each (Table 4). The potential impact of cell and chromosome rosette chromosome are presented as chromosome position maps (Fig. orientation on the distributions of chromosomes 7, 8 and 16 4A-D). Random chromosome distributions would be expected distribution is shown in Fig. 7A-C. Much of the observed to generate a fairly uniform distribution of points, with an clustering of points representing preferential locations of expected tendency to avoid the central portion of the nucleus homologs closely match that predicted when taking into frequented by nucleoli. Chromosomes 7 and 8 were often account the effects of nuclear polarity and relative rosette found to occupy nonrandom positions at or near the nuclear reorientation. periphery, as indicated by clusters of points representing intranuclear regions for which these chromosomes appear to Predictable intranuclear positioning of chromosome have a preference (Fig. 4A,B). Chromosome 16 occupied a Y triploid cells more central location, with homologs clustered along lines To further test the possibility of a nonrandom chromosome representing the x and y axes (Fig. 4C). By contrast, topology in nuclei of quiescent cells, we examined the relative chromosome X was distributed rather uniformly throughout distributions of chromosomes X and Y in quiescent triploid cells nuclei with no apparent tendency to avoid the central portion (karyotype = 69, XXY) using dual FISH. These triploid cells of the nucleus, suggesting a more random relative homolog have their 69 chromosomes crowded into nuclei which do not distribution (Fig. 4D). appear to be much larger than those of their diploid counterparts, potentially restricting free relative chromosome movements Variations in nuclear orientation tend to mask an compared to that in diploid cells. We recently determined that ordered interphase chromosome topology these 69 chromosomes in mitotic cells arrange themselves into Chromosome position maps derived from nuclei of quiescent a single rosette in which chromosomes are segregated into three diploid cells showed intermittent regions containing clusters of haploid sets (Nagele et al., 1998). Here, we have taken advantage points representing frequent positioning of homologs for three of the four chromosomes (i.e. chromosomes 7, 8 and 16) Table 4. Relative orientation of chromosome arrays of examined (Fig. 4A-C). In an effort to explain the basis for this mitotic daughter cells preferential positioning, we separated relative homolog Relative orientation positions displayed by individual nuclei into several position Chromosome categories and then determined the relative number of nuclei number Same Opposite (mirror-image) falling into each category. As shown for chromosome 8, nearly 73920 all nuclei could be included in these categories, but the number 82015 of nuclei falling into each category varied considerably (Fig. 5). 16 14 12 Chromosome positioning in interphase cells 531 A Flip

Flip Flip Flip

Combine

Chromosome 8

B C Chromosome 7 Chromosome 16

Fig. 6. Chromosome rosettes persist throughout mitosis and their reorientation influences chromosome homolog distribution in diploid cells. (A) DAPI-stained mitotic chromosome rosette (blue) in prometaphase cell oriented horizontally (i.e. parallel to the culture substratum) and showing the characteristic radial arrangement of chromosomes. (B) Vertically oriented rosette (metaphase plate) and mitotic spindle revealed by immunofluorescence with anti-tubulin antibody. (C) Diagram illustrating that, following anaphase, n=390 n=378 chromosome rosettes can ‘fall’ to a horizontal orientation in either Fig. 7. Diagrams illustrating the combined impact of cell and direction with roughly equal probability. This phenomenon results in chromosome rosette orientation on the distributions of chromosomes early G1 daughter cell pairs exhibiting an equal number of identical and mirror-image chromosome distribution patterns. (D) Adjacent 7, 8 and 16. (A) Flipping and reorientation of the most commonly observed relative homolog distribution for chromosome 8 (i.e. daughter cells at late telophase/early G1 phase still connected by the spindle remnant (arrow) showing that the chromosome rosette Category 1) readily explains the observed symmetrical distribution of configuration persists throughout mitosis. (E-H) Reverse-contrast clusters of points representing homolog positions. (B,C) Clusters of FISH images showing adjacent daughter cells in cultures that are just points representing the intranuclear locations of homologs of subconfluent displaying either identical (E,F) or mirror-image (G,H) chromosomes 7 and 16 also tended to be distributed symmetrically. chromosome homolog distribution patterns. Arrows indicate the The most frequently observed relative homolog distribution patterns relative polarity of adjacent nuclei. Cells were FISHed with painting for these chromosomes are shown in the insets. For chromosome 16, probes specific for chromosomes 16 (E,G) and 8 (F,H) and two homolog position categories shown in the insets occurrred with counterstained with propidium iodide. Bars, 20 µm. roughly equal frequency. 532 R. G. Nagele and others of the XXY sex chromosome karyotype of these cells to study chromosome Y relative to the two chromosome X homologs was the relative positioning of X and Y chromosomes at interphase. predictable within narrow spatial limits (Fig. 8E). This analysis Chromosome X-specific alpha-satellite probe labeled with revealed a clear tendency for chromosomes X and Y to be as far digoxigenin was detected with FITC-labeled, anti-digoxigenin as possible away from each other. Such a relative distribution antibodies. Chromosome Y-specific DNA probes were internally would be expected if chromosome Y occupies the expected labeled with Cy3 (Fig. 8A-D). Remarkably, the location of location of a third X chromosome, and supports the idea that the 69 chromosomes in triploid cells maintain a chromosome distribution within interphase nuclei that is related to that in mitotic chromosome rosettes.

Chromosome positions in interphase nuclei are related to their positions in mitotic chromosome rosettes We further tested the possibility that the relative positioning of chromosomes in interphase nuclei is somehow linked to their positions in mitoic chromosome rosettes. To accomplish this, we carried out dual-FISH on subconfluent cultures of human fibroblasts using the following combinations of chromosome-specific probes and contrasting fluorochromes: 1 and 17, 2 and X, 3 and X, 4 and X, 6 and X, 8 and X, 9 and X, 10 and X, 11 and X, 8 and 11, 11 and 14, 13 and 18, 14 and 16, 14 and 22, 16 and 21, 16 and 22, and 17 and 18. Our objective was to first screen the various combinations in an effort to identify chromosomes that are closely juxtaposed in the chromosome rosettes of mitotic cells. From among the combinations tested, chromosomes 8 and 11 were found to be adjacent to one another in most mitotic chromosome rosettes, and a E comparable relative positioning was also clearly exhibited in Distribution of Chromosome Y Using X the majority of interphase nuclei (Fig. 9). These findings as a Reference Point in Triploid Nuclei suggest the existence of an ordered interphase chromosome topology in quiescent human cells in which at least some chromosome homologs exhibit a preferred relative intranuclear location that is somehow linked to their relative spatial arrangement in rosettes of mitotic cells.

= X = Y

Fig. 8. (A-D) Relative distributions of chromosomes X and Y in quiescent triploid cells (karyotype = 69, XXY) revealed with dual FISH. Chromosome X-specific alpha-satellite probe labeled with digoxigenin was detected with FITC-labeled, anti-digoxigenin antibodies and appears as light blue-green staining spots in these false-colorized images. Chromosome Y-specific DNA probes were internally labeled with Cy3 and appear as red spots. There was a strong tendency for both X and Y chromosomes to remain rather widely separated. Bar, 5 µm. (E) Diagram summarizing results obtained from dual FISH analysis which reveals the distribution of chromosome Y using the two chromosome X homologs as reference Fig. 9. Subconfluent human fibroblasts subjected to dual-FISH with points. The position of the Y chromosome relative to the two contrasting fluorochromes. Chromosomes 8 (yellow, FITC-labeled) chromosome X homologs was found to be quite predictable and and 11 (red, Cy3-labeled) were closely juxtaposed (yellow arrows) in showed that the preferred location of chromosome Y was most often chromosome rosettes of most mitotic cells (boxed inset) and in on the opposite side of the nucleus in triploid cells. interphase nuclei. Bar, 10 µm. Chromosome positioning in interphase cells 533

DISCUSSION provide information regarding specific intranuclear locations of chromosome territories. The favorable geometry afforded by In the present study, we investigated the relative intranuclear the flattened ellipsoidal shape of nuclei of quiescent fibroblasts spatial positioning of chromosomes 7, 8, 16, X and Y in has enabled us to assign specific x-y coordinates to represent quiescent diploid and triploid human fibroblasts. These locations of individual chromosome homologs and map their chromosomes were chosen because (1) they represent a broad relative positions with reasonable accuracy. Homolog position range in size, (2) none possess nucleolar organizers, which may maps, which show the relative locations of homologous impose additional positional constraint variables that could chromosomes described by their specific x-y coordinates, hinder direct comparison of data with chromosomes lacking clearly support preferential intranuclear locations for these structures, and (3) information on spatial positioning of chromosomes 7, 8 and 16, but not for chromosome X. these chromosomes in mitotic fibroblasts is available Furthermore, the observed chromosome distributions were not (Mosgoller et al., 1991; Leitch et al., 1994; Nagele et al., 1995). consequences of geometrical constraints imposed on Quiescent fibroblasts were chosen as a model system because chromosome territories of different size and shape. If this were they are readily subjected to FISH as intact cells, are ‘locked’ the case, larger chromosomes would be expected to congregate in the G0/G1 phase of the cell cycle, exhibit common overall in the thicker central portion of the nucleus, which was not cell shape parameters, and possess a flattened ellipsoidal observed. In fact, our results show that chromosomes 7 and 8 nucleus of uniform dimensions. The arrangement of (large chromosomes) preferred peripheral nuclear regions, chromosome territories in these flattened nuclei approached a whereas chromosome 16 (the smallest chromosome examined) two-dimensional array, which greatly facilitated mapping of frequented the central portion of the nucleus. Chromosome X their relative intranuclear positions in the x-y plane. Three of (also a large chromosome) did not exhibit a preferential the four chromosomes (i.e. 7, 8 and 16) examined in the location and was scattered throughout both peripheral and homolog position-mapping studies were found to exhibit central portions of the nucleus. Recently, Shelby et al. (1996) preferred positions in the interphase nuclei of diploid human and Abney et al. (1997) provided strong evidence for fibroblasts. In quiescent triploid cells with the karyotype 69, nonrandom chromosome positioning and showed that XXY, the ‘preferential positioning’ of the Y chromosome, chromosomes are relatively immobile for prolonged periods at relative to the two X homologs, was predictable in the majority interphase. Such immobility can account for the many of nuclei within remarkably narrow spatial limits. Lastly, after observations of nonrandom chromosome topologies in screening various combinations of chromosome-specific interphase cells. probes, chromosomes 8 and 11 were found to be closely In the present study, we imposed selection criteria that were juxtaposed in most mitotic chromosome rosettes, and were aimed at minimizing the effects of nuclear size variations on likewise positioned in interphase nuclei. This predictability relative chromosome positioning. Despite this, it is likely that confirms a nonrandom chromosome topology in these cells and the data presented here are somewhat biased against detecting suggests the existence of a specific chromosome arrangement more discrete intranuclear positions for chromosomes. One that is somehow related to the distribution of chromosomes in clear source of bias comes from the fact that ellipsoidal nuclei rosettes of mitotic cells. with otherwise identical relative homolog distributions can be positioned at 180° angles with repect to one another on the Homolog separation distances reveal nonrandom culture substratum within the x-y plane. Since nuclei were not relative homolog positioning rotated in an attempt to compensate for possible differences in As a first test for chromosome organization, we examined the nuclear polarity, this may lead to a strong bias of chromosome spatial relationship among homologs of chromosomes 7, 8, 16 position maps towards more random distributions. In addition, and X in quiescent diploid cells through measurements of their newly formed daughter cells were found to display either interhomolog separation distances. Homolog separation identical or mirror-image chromosome homolog distribution distance profiles for chromosomes 7, 8 and 16 deviated patterns (cf. Fig. 6). A similar phenomenon has been described significantly from a random distribution, with the most for nucleolar distribution patterns in insect epithelial cells pronounced deviations shown by chromosomes 8 and 16. This (Lock and Huie, 1980). Here, we have determined that suggests a nonrandom distribution for chromosomes in identical and opposite (mirror-image) chromosome distribution quiescent cells and is in general agreement with other studies patterns occur with nearly equal frequency (cf. Table 4), describing nonrandom, possibly cell type-specific, distribution indicating that these two patterns originate from the direction patterns for chromosome territories and centromeric in which the two daughter chromosome rosettes, originally in subdomains in a wide variety of cell types (Agard and Sedat, a vertical orientation during anaphase, fall to a horizontal 1983; Manuelidis, 1984; Hadlaczky et al., 1986; Hochstrasser orientation relative to each other at telophase. Thus, we show et al., 1986; Borden and Manuelidis, 1988; Manuelidis and that the observed preferential locations of homologs closely Borden, 1988; Haaf and Schmid, 1989; Popp et al., 1990; match that predicted when taking into account nuclear polarity Bartholdi, 1991; Ferguson and Ward, 1992; Haaf and Ward, and relative rosette reorientation. Despite limitations inherent 1995). in the methodology outlined here, we conclude that some chromosomes exhibit nonrandom, preferential spatial Mapping of homolog positions suggests preferred positioning in nuclei of quiescent cells as originally suggested intranuclear locations for chromosomes by Manuelidis (1985, 1990). How these relative positions are Although analysis of measured interhomolog separation influenced by variations in cell type and cell cycle phase distances can apparently distinguish between random and remain to be determined. On the other hand, based on our nonrandom chromosome topologies, this method does not findings with chromosome X in diploid cells, we cannot 534 R. G. Nagele and others eliminate the possibility that some chromosomes may occupy cellular subclones. Further work will be needed to determine more random intranuclear positions at interphase, as has been if this is the case. In addition, the structural basis for the suggested by others (Ferguson and Ward, 1992; Vourc’h et al., observed ‘cohesion’ of chromosomes 8 and 11 in mitotic 1993; Lesko et al., 1995). rosettes and corresponding interphase nuclei remains to be resolved, but it raises the possibility of a structural link Predictable positioning of chromosomes X and Y in occurring between chromosomes. triploid cells and the relationship between The functional significance of preferential positional chromosome positioning in interphase and mitotic interrelationships among specific chromosomes during cells interphase is unknown. It is tempting to speculate that the Triploid cells possess 69 chromosomes, three copies of each activities of genes on specific chromosomes are influenced by homolog, crowded into the nucleus. Taking advantage of this their specific intranuclear locations, as originally proposed by karyotype, we used quiescent triploid cells to show that the Blobel (1985). Although it is likely to be an oversimplification intranuclear location of the Y chromosome, relative to the two of a complex process, perhaps the aberrant gene expression and X homologs, was predictable in the majority of nuclei within phenotypes of cancer cells are somehow related to the remarkably narrow spatial limits. This finding provides inevitably abnormal chromosome topology resulting from their compelling evidence that, at least in triploid human cells, aneuploid condition. These questions cannot be resolved positioning of some chromosomes at interphase may be quite without further investigation into the relationship between the precise. The predictability of chromosome X and Y positioning intranuclear positions of specific chromosome territories. and the nature of their relative distribution pattern suggests an interphase chromosome topology that is spatially related to the The authors wish to thank K. Linask for helpful discussions and chromosome distribution observed in chromosome rosettes of encouragement and J. Fazekas, C. Fitzgerald, Eileen Manogue and J. mitotic cells. These rosettes are wheel-like chromosome arrays McCord for technical assistance. Supported by grants to R.N. from with centromeric domains forming the central hub and the State of New Jersey Commission on Cancer Research and the State of New Jersey Department of Human Services (Governor’s Council chromosome arms projecting out from the center in a radial on the Prevention of Mental Retardation and Developmental fashion (Costello, 1970; Chaly and Brown, 1988; Hodge et al., Disabilities). 1995; Nagele et al., 1995). Within rosettes of normal diploid cells chromosome homologs are consistently positioned on opposite sides, and heterologs also exhibit a reproducible REFERENCES spatial interrelationship (Nagele et al., 1995, 1998). This arrangement readily explains the lack of association of Abney, J. R., Cutler, B., Fillbach, M., Axelrod, D. and Scalettar, B. (1997). homologs in dividing fibroblasts, lymphocytes (Emmerich et Chromatin dynamics in interphase nuclei and its implications for nuclear structure. J. Cell Biol. 137, 1459-1468. al., 1989; Popp et al., 1990; Ferguson and Ward, 1992; Agard, D. and Sedat, J. (1983). Three-dimensional architecture of a polytene Chandley et al., 1996) and plant cells (Heslop-Harrison et al., nucleus. Nature 302, 676-681. 1989; Bennett and Bennett, 1992) and indicates that rosettes in Arnoldus, E., Peters, A., Bots, G., Raap A. and van der Ploeg, M. (1989). normal human cells are composed of two spatially distinct, Somatic pairing of chromosome 1 centromeres in interphase nuclei of ‘haploid sets’ of 23 chromosomes, with each ‘haploid set’ human cerebellum. Hum. Genet. 83, 231-234. Bartholdi, M. (1991). Nuclear distribution of centromeres during the cell forming a cohesive, ordered group of homologs inherited from cycle of human diploid fibroblasts. J. Cell Sci. 99, 255-263. one parent as has been shown in several grass species and Bennett, S. and Bennett, M. (1992). Spatial separation of ancestral genomes hybrids (Heslop-Harrison and Bennett, 1984; Leitch et al., in the wild grass Milium montianum Parl. Ann. Bot. 70, 111-118. 1991, 1994; Nagele et al., 1995). In accord with this depiction, Blobel, G. (1985). Gene gating: a hypothesis. Proc. Natl. Acad. Sci. USA 82, 8527-8529. we have recently shown that cells derived from fetuses with Borden, J. and Manuelidas, L. (1988). Movement of the X chromosome in triploidy contain mitotic rosettes composed of three tandemly epilepsy. Science 242, 1687-1691. linked haploid sets, each displaying a similar chromosome Chaly, N. and Brown, D. (1988). The prometaphase configuration and spatial order (Nagele et al., 1998). Faithful propagation of chromosome order in early mitosis. J. Cell Sci. 91, 325-335. relative chromosome positioning from one cell generation to Chandley, A. C., Speed, R. and Leitch, A. (1996). Different distributions of homologous chromosomes in adult human Sertoli cells and in lymphocytes the next within a proliferating cell line or within cells that signify nuclear differentiation. J. Cell Sci. 109, 773-776. comprise the human body raises the possibility that the Comings, D. (1968). The rationale for an ordered arrangement of chromatin chromosome positional order seen in mitotic cells may carry in the interphase nucleus. Am. J. Hum. Genet. 20, 440-460. over into interphase or at least influence the intranuclear Comings, D. (1980). Arrangement of chromatin in the nucleus. Hum. Genet. 53, 131-143. locations of chromosomes. Here, the specific distribution Costello, D. (1970). Identical linear order of chromosomes in both gametes of patterns of the sex chromosomes in quiescent triploid cells the Acoel Turbellarian polychoerus carmelensis: a preliminary note. Proc. suggests a link between the observed chromosome spatial order Natl. Acad. Sci. USA 67, 1951-1958. in mitotic chromosome rosettes and interphase chromosome Cremer, T., Cremer, C., Baumann, H., Luuedtke, E., Sperling, K., Teuber, topology. Further support for such a link comes from the V. and Zorn, C. (1982). Rabl’s model of the interphase chromosome arrangement tested in chinese hamster cells by premature chromosome observed frequent juxtaposition of chromosome 8 and 11 in condensation and laser-UV-microbeam experiments. Hum. Genet. 60, 46- both mitotic chromosome rosettes and interphase nuclei. Also, 56. a substantial fraction of cells in these dual FISH preparations Cremer, T., Lichter, P., Borden, J., Ward, D. and Manuelidis, L. (1988). exhibited one set of chromosome 8 and 11 homologs closely Detection of chromosome aberrations in metaphase and interphase tumor cells by in situ hybridization using chromosome-specific library probes. apposed, with the other set more widely separated. This Hum. Genet. 80, 235-246. suggests that these two predominant chromosome DeBoni, U. (1994). The interphase nucleus as a dynamic structure. Int. Rev. configurations may reflect the presence of two coexisting Cytol. 150,149-171. Chromosome positioning in interphase cells 535

DeBoni, U. and Mintz, A. (1986). Curvilinear, three-dimensional motion of for the centromeres of chromosomes 7, 11, and 17 in interphase human T- chromatin domains and nucleoli in neuronal interphase cells. Science 234, lymphocytes. Exp. Cell Res. 219, 499-506. 863-866. Lichter, P., Cremer, T., Borden, J., Manuelidis, L. and Ward, D. (1988). Emmerich, P., Loos, P., Jauch, A., Hopman, A., Wiegant, J., Higgins, M., Delineation of individual human chromosomes in metaphase and interphase White, B., van der Ploeg, M., Cremer, C. and Cremer, T. (1989). Double cells by in situ supression hybridization using recombinant DNA libraries. in situ hybridization in combination with digital image analysis: a new Hum. Genet. 80, 224-234. approach to study interphase chromosome tomography. Exp. Cell Res. 181, Locke, M. and Huie, P. (1980). The nucleolus during epidermal development 126-140. in an insect. Tissue Cell 12, 175-195. Ferguson, M. and Ward, D. (1992). Cell cycle dependent chromosomal Manuelidas, L. (1984). Different central nervous system cell types display movement in pre-mitotic human T-lymphocyte nuclei. Chromosoma 101, distinct and nonrandom arrangements of satellite DNA sequences. Proc. 557-564. Natl Acad. Sci. USA 81, 3123-3127. Haaf, T. and Schmid, M. (1989). Centromeric association and non-random Manuelidis, L. (1985). Indications of centromere movement during interphase distribution of centromeres in human tumor cells. Hum. Genet. 81, 137-143. and differentiation. Ann. NY Acad. Sci. 450, 205-221. Haaf, T. and Schmid, M. (1991). Chromosome topology in mammalian Manuelidis, L. (1990). A view of interphase chromosomes. Science 250, interphase nuclei. Exp. Cell Res. 192, 325-332. 1533-1540. Haaf, T. and Ward, D. (1995). Higher order nuclear structure in mammalian Manuelidis, L. and Borden, J. (1988). Reproducible compartmentalization sperm revealed by in situ hybridization and extended chromatin fibers. Exp. of individual chromosome domains in human CNS cells revealed by in situ Cell Res. 219, 604-611. hybridization and three-dimensional reconstruction. Chromosoma 96, 397- Hadlaczky, G., Went, M. and Ringertz, R. (1986). Direct evidence for non- 410. random localization of mammalian chromosomes in the interphase nucleus. Mosgoller, W., Leitch, A., Brown, J. and Heslop-Harrison, J. (1991). Exp. Cell Res. 167, 1-15. Chromosome arrangements in human fibroblasts at mitosis. Hum. Genet. 88, Heslop-Harrison, J., Leitch, A., Schwarzacher, T., Smith, J., Atkinson, M. 27-33. and Bennett, M. (1989). The volumes and morphology of human Nagele R., Freeman, T., McMorrow, L. and Lee, H. (1995). Precise spatial chromosomes in mitotic reconstructions. Hum. Genet. 84, 27-34. positioning of chromosomes during prometaphase: evidence for Heslop-Harrison, J. and Bennett, M. (1984). Chromosome order – possible chromosome order in human cells. Science 270, 1831-1835. implications for development. J. Embryol. Exp. Morphol. 83, 51-73. Nagele, R., Freeman, T., Fazekas, J., Tejera, Z. and Lee H. (1998). Heslop-Harrison, J. and Bennett, M. (1990). Nuclear architecture in plants. Chromosome spatial order in human cells: evidence for early origin and Trends Genet. 6, 401-405. faithful propagation. Chromosoma 107, 330-338. Hilliker, A. and Appels, R. (1989). The arrangement of interphase Popp, S., Scholl, H., Loos, P., Jauch, A., Stelzer, E., Cremer, C. and chromosomes: structural and functional aspects. Exp. Cell Res. 185, 297- Cremer, T. (1990). Distribution of chromosome 18 and X centric 318. heterochromatin in the interphase nucleus of cultured human cells. Exp. Cell Hochstrasser, M., Mathog, D., Gruenbaum, Y., Saumweber, H. and Sedat, Res. 189, 1-12. J. (1986). Spatial organization of chromosomes in the salivary gland nuclei Schwarzacher, T., Heslop-Harrison, J., Anamthawat-Jonsson, K., Finch, of Drosophila melanogaster. J. Cell Biol. 102, 112-123. S. and Bennett, M. (1992). Parental genome separation in reconstructions Hodge, L., Barrett, J. and Welter, D. (1995). Computer graphics of SEM of somatic and premeiotic metaphases of Hordeum vulgare × H. bulbosum. images facilitate recognition of chromosome position in isolated human J. Cell Sci. 101, 13-24. metaphase plates. Microsc. Res. Tech. 30, 408-418. Shelby, R. D., Hahn, K. M. and Sullivan, K. F. (1996). Dynamic elastic Hutchison, N. and Weintraub, H. (1985). Localization of DNase I-sensitive behavior of a-satellite DNA domains visualized in situ in living human cells. sequences to specific regions of the interphase nuclei. Cell 43, 471- J. Cell Biol. 135, 545-557. 482. Spector, D. (1993). Macromolecular domains within the cell nucleus. Annu. Leitch, A. R., Mosgoller, W., Schwazacher, T., Bennett, M. and Heslop- Rev. Cell Biol. 9, 265-315. Harrison, J. (1990). Genomic in situ hybridization to sectioned Van Dekken, H., Pinkel, J., Mullikin, J., Trask, B., van den Engh, G. and nuclei shows chromosome domains in grass hybrids. J. Cell Sci. 95, 335- Gray, J. (1989). Three-dimensional analysis of the organization of human 341. chromosome domains in human and human-hamster interphase nuclei. J. Leitch, A., Brown, J., Mosgoller, W., Schwarzacher, T. and Heslop- Cell Sci. 94, 299-306. Harrison, J. (1994). The spatial localization of homologous chromosomes Vourc’h, C., Taruscio, D., Boyle, A. and Ward, D. (1993). Cell cycle- in human fibroblasts at mitosis. Hum. Genet. 93, 275-280 dependent distribution of telomeres, centromeres and chromosome-specific Leitch, A., Schwarzacher, T., Mosgoller, W., Bennett, M. and Heslop- subsatellite domains in the interphase nucleus of mouse lymphocytes. Exp. Harrison, J. (1991). Parental genomes are separated throughout the cell Cell Res. 205, 142-151. cycle in a plant hybrid. Chromosoma 101, 206-213. Young, I. T. (1977). Proof without prejudice: use of the Kolmogorov-Smirnov Lesko, S., Callahan, D., LaVilla, M., Wang, Z. and Ts’o, P. (1995). The test for the analysis of histograms from flow systems and other sources. J. experimental homologous and heterologous separation distance histograms Histochem. Cytochem. 24, 935-941.