Chromosome Research 10: 535548, 2002. # 2002 Kluwer Academic Publishers. Printed in the Netherlands
535
The 3D structure of human chromosomes in cell nuclei
E. Luka¤sˇ ova¤1, S. Kozubek1*, M. Kozubek2, M. Falk1 & J. Amrichova¤2 1 Laboratory of Molecular Cytology and Cytometry, Institute of Biophysics, Academy of Sciences of the Czech Republic, Kra¤lovopolska¤ 135, 612 65 Brno, Czech Republic; Tel: (þ 420 5) 4151 7139; Fax: (þ 420 5) 41240498; E-mail:
[email protected]; 2 Laboratory of Optical Microscopy, Faculty of Informatics, Masaryk University, Botanicka¤ 68a, Brno, Czech Republic *Correspondence Received 3 June 2002. Received in revised form and accepted for publication by Adrian Sumner 14 August 2002
Key words: chromosome structure, confocal microscopy, mathematical models, nuclear architecture
Abstract The spatial arrangement of some genetic elements relative to chromosome territories and in parallel with the cell nucleus was investigated in human lymphocytes. The structure of the chromosome territories was studied in chromosomes containing regions (clusters) of highly expressed genes (HSA 9, 17) and those without such clusters (HSA 8, 13). In chromosomes containing highly expressed regions, the elements pertaining to these regions were found close to the centre of the nucleus on the inner sides of chromosome territories; those pertaining to regions with low expression were localized close to the nuclear membrane on the opposite sides of the territories. In chromosomes with generally low expression (HSA 8, 13), the elements investigated were found symmetrically distributed over the territories. Based on the investigations of the chromosome structure, the following conclusions are suggested: (1) Chromosome territories have a non-random internal 3D structure with de¢ned average mutual positions between elements. For example, RARa, TP53 and Iso-q of HSA 17 are nearer to each other than they are to the HSA 17 centromere. (2) The structure of a chromosome territory re£ects the number and chromosome location of clusters of highly expressed genes. (3) Chromosome territories behave to some extent as solid bodies: if the territory is found closer to the nuclear centre, the individual genetic elements of this chromosome are also found, on average, closer the centre of the nucleus. (4) The positions of centromeres are, on average, nearer to the £uorescence weight centre of the territory (FWCT) than to genes. (5) Active genes are not found near the centromeres of their own territory. A simple model of the structure of chromosome territory is proposed.
Introduction The arrangement of interphase chromosomes into separate territories provides a framework for the
investigation of the relationship between the higher-order chromatin structure and function (Cremer et al. 1988, Lichter et al. 1988, Pinkel et al. 1988). A basic question is whether gene expression
E. Luka¤sfl ova¤ et al.
536 is determined, at least in part, by the structure of chromosome territory. The studies trying to resolve this issue are aimed at determining whether particular genomic sequences occupy special positions within chromosome territories, whether these positions di¡er according to the transcriptional activity of the sequences and whether genomic regions or whole individual chromosomes occupy particular compartments within the cell nucleus (Belmont & Bruce 1994, Nagele et al. 1995, Ferreira et al. 1997, Lamond & Ernshaw 1998, Belmont et al. 1999, Cockel & Gasser 1999, Croft et al. 1999, Sadoni et al. 1999, Verschure et al. 1999, Nagele et al. 2000, Volpi et al. 2000, Chevret et al. 2000, Cremer et al. 2000, Cremer & Cremer 2001, Sadoni et al. 2001). Several studies lend support to the hypothesis that genes are not randomly positioned within chromosome territories and that the transcriptional status of certain genes is a¡ected by their nuclear topography. Kurz et al. (1996) noted that several active and inactive genes were preferentially located on the periphery of chromosome territories; Volpi et al. (2000) described the three-dimensional large-scale chromatin organisation of the major histocompatibility complex locus on human chromosome 6. These authors observed large chromatin loops containing several Mb of DNA extending outwards from the chromosome territory. Transcriptional upregulation led to an increase in the frequency with which active genes (but not inactive control genes) were found on an external chromatin loop. Dynamic repositioning of certain genes in mouse lymphocytes and during di¡erentiation of HL-60 cells depending on their transcriptional status was observed (Brown et al. 1997, 1999, Ba¤rtova¤ et al. 2002). The transcriptionally inactive genes were localized at centromeric heterochromatin clusters in contrast to transcriptionally active genes which were positioned away from them. In spite of evidence that the proximity of genetic loci to the centromeric heterochromatin leads to inactivation of their transcription activity, very little is known about the mechanisms controlling the spatial and functional distribution of chromatin within the nucleus. The relationship between the nuclear localization of chromosomes and their gene content was shown
for the ¢rst time by Jenny Croft et al. (1999) for HSA 19 and 18. The authors showed that gene-rich HSA 19 is located in the centre of the nucleus, while similarly sized HSA 18 poorly populated with genes adopt a more peripheral position in the cell nucleus. Further results from our laboratory (Ba¤rtova¤ et al. 2001, Kozubek S. et al. 2002) and Bickmore’s laboratory (Boyle et al. 2001) show that other chromosomes containing more actively transcribed genes also adopt a more internal position in the cell nucleus contrary to gene-poor chromosomes. With the progress of genome-sequencing projects, a genetic map has been developed showing the chromosomal positions of about 24 thousand genes. In addition to this, the expression pro¢les for any chromosomal regions in several normal and pathological tissue types were also detected. These expression pro¢les reveal clustering of highly expressed genes into speci¢c chromosomal regions that are largely conserved in di¡erent tissues, indicating that this arrangement may be re£ected in the higher order structure of the genome (Caron et al. 2001). In this study we have tried to determine whether the arrangement of highly expressed genes into clusters on individual chromosomes is actually translated into the structure of chromosome territories in the cell nucleus. The structure of the chromosome territories was studied in chromosomes containing clusters of highly expressed genes (HSA 9, 17) and those without such clusters (HSA 8, 13). In HSA 17, the relative positions of 5 genetic loci (from telomere p to telomere q: TP53, Iso-p, centromere, RARa, Iso-q) were investigated. The expression of genes is rather high throughout the whole HSA 17, with the exception of a small region close to the centromere where Iso-p is located (Caron et al. 2001).
Materials and methods Isolation of human lymphocytes Human lymphocytes were isolated from heparinized blood (2030 U/ml) from healthy donors by Ficoll-Hypaque (Pharmacia Biotech., Uppsala, Sweden) density-gradient centrifugation as previously described (Kozubek S. et al. 1999).
The 3D structure of human chromosomes Cell ¢xation Dehydration of the nuclei was avoided to preserve the native 3D structure. Dense cell suspension in PBS bu¡er (100 ml) was spread on a poly-Llysinated microscope slide. The cells attached to the surface of the slide (in about 5 min without drying) were ¢xed in 3.7% paraformaldehyde with 0.5% Triton X-100 and HEPEM (65 mmol/L PIPES, 30 mmol/L HEPES, 10 mmol/L EGTA, 2 mmol/L MgCl2, pH 6.9) (Neves et al. 1999) for 12 min at room temperature and thoroughly washed in PBS (3 times for 5 min). The ¢xed cells were either stored for 3 days at þ 4 C or immediately permeabilized for hybridization.
537 denaturation, hybridization and a post-hybridisation wash were performed according to the instructions of the probe manufacturers. Repeated hybridization Cover slips from the ¢rst hybridization were removed by the immersion of slides in 4 SSC/ 0.1% Triton X-100. The slides were then washed in three changes of the same solution, once in 2 SSC (for 4 min each) and immediately denatured under the same conditions as before the ¢rst hybridization. Using repeated hybridization to the same genetic elements, it can be easily demonstrated that their positions in the cell nuclei are not in£uenced by repeated denaturation.
Cell permeabilization Confocal cytometry Permeabilization of cells was performed in 0.7% Triton X-100/0.1 N HCl/PBS precooled to þ 4 C for 10 min on ice (Neves et al. 1999). Subsequently, cells were incubated with 0.1 mg/mL RNAse A in 2 SSC for 30 min at 37 C, then washed 3 times in PBS, denatured in 50% formamide/2 SSC for 20 min at 75 C and immediately hybridized. DNA probes, £uorescence in-situ hybridization FISH was performed using unique sequence probes: MBCR-ABL translocation probe (Vysis, USA) (spectrum green-BCR 22q11, spectrum orange-ABL 9q34), digoxigenated c-MYC (8q24), TP53 (17p13.1) and RB (13q14.2); t(15,17) translocation probe identifying the retinoic acid receptor alpha (RARa) gene (digoxigenated) (17q21) and PML gene (biotinylated) (15q22); probe for identi¢cation of isochromosome 17 consisting of Iso-p (17q11.2) (bitinylated) and Iso-q (17q21.3-q23) (digoxigenated), all purchased from Oncor, USA). Alpha-satellite DNA sequences of the centromeric region of HSA 9, 19 (digoxigenated), 8, 17 (biotinylated) were from Oncor; total chromosome DNA probes of HSA 8, 9, 17 were FITC labelled (Oncor, USA), that for HSA 13 was Cy3 labelled (Cambio, UK). Human telomere DNA probes for HSA 8, 9, 19 p arm (digoxigenated) and q arm (biotinylated) were purchased from Cambio, UK. Ten microlitres of a single probe or a mixture of two probes with di¡erent chemical modi¢cation was applied onto a microscopic slide. Probe
High-resolution cytometry was performed as previously described (Kozubek M. et al. 1999, 2001). A high-resolution cytometer based on an inverted completely automated Zeiss Axiovert 100 (Jena, Germany) £uorescence microscope equipped with confocal unit CARV (Atto Instruments, USA) was used. Images were captured in confocal mode using a fully programmable Micromax digital CCD camera (Princeton Instruments, USA). The whole system was controlled by a personal computer equipped with two Intel Pentium1 III processors (Intel Corporation, San Francisco, CA). The acquisition of images with FISH-stained interphase nuclei was automated, as was the on-line analysis of image quality, ¢nding cell nuclei and compression of images. It was possible to repeatedly acquire hybridized nuclei. For each hybridization, the same cells were re-allocated, re-acquired and re-analysed. The results of analyses from di¡erent hybridizations were then superimposed in the computer memory. This allowed an increase in the number of simultaneously observed genetic loci within the same nucleus. O¡-line analysis was performed using FISH 2.0 software (Kozubek M. et al. 1999, 2001) that, in addition to other functions, allows correction for chromatic aberrations. Orthogonal views (xy, xz and yz) of 3D data were used to check the results of analysis. In repeated hybridization, user-de¢ned attributes were set for the identi¢ed signals to distinguish elements pertaining to a particular
E. Luka¤sfl ova¤ et al.
538 chromosome homologue. Information about the signals was entered into text ¢les and further analysed using the statistical package Sigma Plot (Jandel Scienti¢c, CA). Typical values of precision are 2030 nm laterally and 20100 nm axially. Typical values for resolution are 250350 nm laterally and 700900 nm axially.
Evaluation of the results The £uorescence weight centres of cell nuclei were determined (Kozubek M. et al. 1999, 2001) and used for the calculation of the topographic parameters of genes, centromeres and chromosome £uorescence signals (territories). The £uorescence weight centres of the territories (FWCT) were used to determine their positions. The radial distances of all genetic elements were normalized to the local radius (radius determined in the direction of the given FISH signal in 3D space). The distances between homologous or heterologous elements were normalized to the average of the corresponding two local radii. The local radius was determined in the plane perpendicular to the xy plane passing through the signal investigated. The image in this plane was calculated and segmented using local thresholding (Kozubek M. et al. 2001). Cell nucleus sizes may slightly di¡er even in the same cell population (e.g. lymphocytes from the same donor); in these cases normalized parameters are well conserved for cells with di¡erent radii. Average values for topographic parameters were determined from at least 3 independent experiments in which 5002000 nuclei per experiment were analysed. Comparison of di¡erent data sets was performed using the Student’s t-test option in SigmaPlot (Jandel Scienti¢c, Ltd., CA). The distribution of locus-to-locus distances of both homologous and heterologous genetic loci was compared with theoretical expectation calculated according to the RS model (Kozubek S. et al. 1997, 1999). In brief, positions in the 3D space inside a sphere were generated using Monte-Carlo simulation precisely according to the measured radial distributions of the given genetic element(s) on the assumption of random angular distributions. In the next step, the distances between positions representing homologous (heterologous) elements were calculated for each generation. After 105 repeats,
the distributions of the locus-to-locus distances were determined. Random-walk function for a genetic element in a sphere In order to calculate model functions for radial distributions of genetic elements, Monte Carlo simulation of positions of N polymer links with length L were determined step by step using a random number generator (see Sachs et al. 1995). MC simulation started from the mean position at distance Ro from the centre of a sphere with radius R. Possible positions of the polymer links were restricted by the spherical volume. The position of each link was tested and the calculation repeated if the position of the link happened to be outside the sphere. After N steps, the ¢nal position of the element was identi¢ed and its distance from the centre of the sphere was determined. For R ¼ 20, we used N ¼ 30150 links with L ¼ 1; the entire distribution was derived by repeating the calculation 30 000 times. The standard deviation of the element position (se) was calculated using the number of links of the p polymer (N) and the length of one link (L): se ¼ L* N. The mean positions, as well as the standard deviations, were ¢tted to the experimental data.
Results Chromosome structure of ‘euchromatic’ and ‘heterochromatic’ territories The internal structure of the chromosome territories and their topology in the cell nucleus were investigated in detail for two ‘euchromatic’ chromosomes (HSA 9 and 17) containing a large number of expressed genes (according to Caron et al. 2001), and for two ‘heterochromatic’ chromosomes (HSA 8 and 13) using repeated dualcolour hybridization, as well as reallocation and reacquisition of a large number (*500) of 3D images for each chromosome. The positions of the £uorescence weight centre of the chromosome territories (FWCT) were visualized in parallel with centromeres and genes (c-MYC, ABL, RB, TP53 and RARa). The results for ‘euchromatic’ chromosomes 17 and 9 are shown in Figure 1A, B;
The 3D structure of human chromosomes
539
Figure 1. Structure of a chromosome in the cell nucleus. (A) HSA 17, (B) HSA 9, (C) HSA 8, and (D) HSA 13. The x-axis was set to the FWCT by the rotation of the cell nucleus and a thin section (1 mm) was cut in the central plane. The whole chromosome was then shifted to the mean position of the territory along the x-axis. The ¢gures show the positioning of genes (red and black circles) and centromeres (green circles) relative to the FWCT (blue circle) in the nucleus. The degree of variation in the x direction for the genes and centromeres is presented (upper left panel), as well as the degree of variation in the y direction (lower left panel).
those obtained for ‘heterochromatic’ chromosomes 8 and 13 are shown in Figure 1C, D. In Figure 1AD, the FWCT was positioned by 3D rotation to the x-axis and the whole chromosome was shifted along the x-axis to the FWCT mean position. In this way, the £uctuations of spatial positions of the chromosome as a whole were removed. In order to show the real values of mutual
distances between genetic elements, the xy positions of genes and centromeres are shown only for such nuclei where z co-ordinates were near the central plane. The points, therefore, represent the genetic elements in a narrow slice through the central plane of the nucleus after its rotation. In this case, genes, as well as centromeres, form distributions (inserted plots in the left panels of
E. Luka¤sfl ova¤ et al.
540
Table 1. Mean 3D distances between genetic elements with standard errors and mean centre of nucleus-to-element distances (CE) with standard errors for HSA 17 expressed in % of nuclear radius. Genetic element
C17 (%)
C17 Iso-p Iso-q RARa TP53
38.8 3.8 55.4 2.8 65.6 3.3 55.5 1.4
Iso-p (%)
46.9 2.3 64.3 3.4 63.8 2.0
Figure 1AD) that are narrower when compared with the radial nuclear distributions, owing to the fact that the £uctuations of the chromosome territory relative to the cell nucleus were removed. For the centromeres of HSA 9 and 17 the distance distributions in the x-direction are narrower and shifted towards the nuclear membrane as compared with the corresponding genes (upper left panels). The mean positions of some investigated genetic loci (radial location) of ‘euchromatic’ chromosomes (ABL 9q34; TP53 17p13.1; RARa 17q21; Iso-q 17q21.3-q23) are closer to the centre of the nucleus as compared to the FWCT; on the other hand, Iso-p 17p11.2 and centromeres of these chromosomes are located near the nuclear periphery (Table 1, Figure 1A, B). The di¡erences between the radial locations of genes and centromeres are much smaller for HSA 8 and 13 (Figure 1C, D). The nuclear location of genetic elements of HSA 17 was investigated using repeated hybridization (Figure 2A, Table 1). It was found that TP53, RARa and Iso-q are located close to the nuclear centre (at a mean distance of 50, 55 and 58% of the radius from the nuclear centre, respectively). Centromeres and Iso-p were found close to the nuclear periphery (at a mean distance of 75 and 77%). Nuclear distances between couples of genes located on di¡erent arms of HSA 17 (RARa and TP53; Iso-q and TP53) are shorter than the distances between these genes and centromere 17 in spite of the larger molecular distances between these genes as compared to gene-to-centromere distances (Figure 2B, Table 1). For example, the mean distances between C17 and the TP53 and RARa genes are 55.5 1.4 and 65.6 3.3% of the nuclear radius, respectively, while the mean
Iso-q (%)
53.2 3.1 49.8 2.1
RARa (%)
CE (%)
46.3 1.9
74.9 2.7 77.3 2.7 58.2 2.5 55.4 2.3 50.4 1.8
distance between both these genes is only 46.3 1.9%. A similar phenomenon was observed for Isoq and TP53 (Figure 2B, Table 1). In addition, the distances between both telomeres of the HSA 8, 9 and 19 were also shorter than the distances of telomeres to centromere (unpublished observation). The mean distances of telomere q and p from the centromere of the HSA 9 were 45.12 1.60 and 49.17 1.59% of the nuclear radius, respectively. The mean distance between both telomeres of HSA 9 was only 34.6 1.56%. For the HSA 8 these values are 35.96 1.51% and 45.55 1.56% for the distances between the centromere and telomere p and q, respectively. The mean distance between both telomeres of this chromosome is 32.99 1.52%. Telomeres p and q are extremely close to each other in HSA 19. In about 40% of cells, they are associated with a distance shorter than 10% of the nuclear radius (&0.5 mm). The investigated genetic loci of HSA 17 are not located in close proximity to the centromere in the cell nucleus (Figure 2C, Table 1). An exception is the Iso-p locus, which is very close to the centromere (Figure 2A, Table 1) and its mean position is closer to the nuclear membrane than is the position of the centromere. The mean distances of the centromeres of HSA 17, 9 and 8 from the FWCT of these chromosomes are shorter than the distances of genes of these chromosomes from the FWCT (Figure 2D). The greatest di¡erence between the distance of the centromere and the gene from the FWCT was found for HSA 9. In order to show the correlation between the nuclear location of the TP53 gene and the HSA 17 territory, the nuclei were divided into two groups according to the distance of the HSA 17 FWCT to the nuclear centre (chromosome-to-centre distance). The ¢rst group contained nuclei where the
The 3D structure of human chromosomes
541
Figure 2. (A) Centre-to-locus distances for genetic elements of HSA 17. The most central location was found for the TP53 gene, further for the Iso-q locus, and for the RARa gene. The centromere, as well as the Iso-p locus, was found near the nuclear periphery. (B) The locus-to-locus distances for genetic elements of HSA 17. The centromere-to-TP53 (black circles), centromere-to-RARa (red circles), and centromere-to-Iso-q (orange circles) distances are, on average, longer than TP53-to-RARa, TP53-to-Iso-q distances (yellow and green circles). (C) The positioning of TP53 relative to the centromere of HSA 17 shown in a narrow slice (1 mm) through a central plane of cell nuclei. The centromere was put onto x-axis by rotation of the whole nucleus and shifted to its mean position by shifting the whole territory. The investigated genes do not approach the corresponding centromeres. A similar phenomenon was also observed for the mutual positions of the ABL genes relative to centromere 9. (D) The distributions of distances between genetic elements and the FWCT (denoted as D) for HSA 17, 9, and 8. (E) The distribution of TP53 in the nuclei of human lymphocytes selected according to the distance of FWCT from the centre of the nucleus. Open circles the nuclei with FWCT < FWCTmean; closed triangles the nuclei with FWCT > FWCTmean; FWCTmean ¼ 56.6% of the nuclear radius. The quantity on the y-axis, probability density, is the probability normalized to unit of some quantity. For example, if we determine the probability for a genetic element to occur at different distances from the centre of the cell nucleus, we divide the whole interval into smaller intervals (e.g. 010%, 1020% . . . 90100% of the radius). In these smaller intervals we determine the probability of the occurrence using the number of positive events (genes at a given interval) and normalization to the total number of events. The resulting ratio is the probability density, i.e. probability per 10% of the radius.
chromosome-to-centre distance was shorter than 56.6% of the local radius (the mean value). Nuclei in which this distance was larger were placed into the second group. We found that the location of the TP53 gene is to some extent correlated with the
location of the chromosome territory. The distribution of genes pertaining to chromosomes located closer to the nuclear centre was also shifted in the same direction (Figure 2E) and its standard deviation was larger.
542
E. Luka¤sfl ova¤ et al.
Spatial relationships of genetic elements of chromosome 17 territory
centromere, as has already been shown by the rotation of nuclei.
To determine the spatial arrangement of di¡erent loci of HSA 17, the centromere was placed into the origin of the co-ordinate system and the chromosome territory was rotated to bring the RARa genes to the x-axis (Figure 3A); further rotation of the territory around the x-axis placed the Iso-q to the same plane with the centromere and RARa loci. In this arrangement it is evident that the Iso-p is located close to the centromere (as in metaphase chromosome). The Iso-q, positioned more distally from the centromere than RARa on the same arm of the methaphase chromosome (Figure 3B, inserted schema), is located apart from the RARa locus in the space of the nucleus; the angle between RARacentromere and Iso-qcentromere is greater than 90 . The spatial arrangement of these loci can also be deduced from their mutual distances and their positioning in the nucleus (Table 1). Taking into account these parameters, it proved possible to draw up a schema of the arrangement of a part of the chromosome (Figure 3B, Table 1). TP53, RARa and Iso-q are localized close to the nuclear centre (their mean distances range from 50% to 58% of the nuclear radius from the nuclear centre). The distance of TP53 and Iso-q from the centromere is the same (about 55% of the radius); this distance for RARa is longer (65%). The distances, Iso-qRARa, Iso-qTP53 TP53RARa, Iso-qcentromere and TP53centromere, are very similar, indicating that all these regions cannot be in the same plane of the nucleus. If TP53, centromere and RARa are in the same plane, then Iso-q should be outside this plane. It is drawn on the top of a tetrahedron, the base of which is determined by TP53, RARa and the centromere. Neither Iso-p is located in this plane. The distance of this locus from TP53 is longer than is the distance of the TP53 from the centromere, even if the molecular distance between TP53 and Iso-p on metaphase chromosome is shorter (see Figure 3, inserted schema). The nuclear distance of Iso-p from Iso-q is signi¢cantly shorter than the distance of Iso-q from the centromere. Iso-p is at about the same distance from RARa as is the distance of RARa from the centromere. Iso-p cannot, therefore, be in the same plane as TP53, Iso-q and the centromere, nor in the plane determined by Iso-q, RARa and the
A simple model of chromosome nuclear structure A simple model of chromosome territory that re£ects the basic features of chromosome structure is proposed (Figure 4A). In this model, two subdomains of chromosome territory are assumed (RIDGE and RILGE subdomains) with chromosome backbone looping according to the level of gene expression. Simulation of the 3D structure of HSA 9 in the nucleus was performed using the MC method (Figure 4B). Let us assume (in agreement with Caron et al. 2001) that approximately half of the genes of this chromosome are highly expressed. In the frame of the model, the actual positions of the territory with two equal subdomains are subject to random £uctuations around mean positions (Ro for the territory; R01 and R02 for the subdomains, Ro ¼ (R01 þ R02)/2), either independent of each other or correlated, i.e. with constant distance between subdomains (corresponding to £uctuations of the territory as a whole). These £uctuations were represented by Gaussian distributions of the positions of the territory and subdomains. At the beginning of the calculation, the positions of the territory and subdomains in the cell nucleus were attributed by a random-number generator. The calculation continued with generation of two random-walk polymers simulating the random positioning of genes (centromeres) in a sphere starting at the positions of subdomains. The radial distributions of genetic elements (ABL, C9, HSA 9 territory), the interelement distributions, and the centre-to-ABL vs. centre-to-C9 dependence were calculated (Figure 4B). The parameters of the model for the HSA 9 territory are as follows: R01 ¼ 46% for the subdomain consisting of regions with increased gene expression (RIDGE), R02 ¼ 68% for the subdomain consisting of regions with low gene expression (RILGE), Ro ¼ (R01+R02)/2 ¼ 57% for the mean position of the territory, with the variation of this position being sd ¼ 7%. The independent variation of the subdomains is about ss ¼ 11% and the variations of the elements are approximately se ¼ 17% for ABL and C9. The results of these model calculations correspond well to the experimental data (Figure 4B).
The 3D structure of human chromosomes
543
Figure 3. 3D structure of HSA 17 territory. (A) Mutual positions of 4 genetic elements of HSA 17. The centromere is placed in the origin of the coordinate system and the territory is rotated to bring the RARa locus onto the x-axis. Further rotation of the territory around the x-axis places the Iso-q locus onto the xy plane. Thus, the centromere and RARa and Iso-q are in the same plane. Iso-p is outside this plane. (B) Localization of ¢ve loci of HSA 17 in the cell nucleus and the spatial arrangement of chromosome backbone traced through the loci. The distances between the loci and their nuclear positions indicate that they do not occur in the same plane of the nucleus. The inserted schema indicates the genetic element locations in metaphase HSA 17.
544
E. Luka¤sfl ova¤ et al.
Figure 4. A model of an interphase chromosome. (A) The chromosome territory is divided into a RIDGE subdomain located more centrally (pink) and a RILGE subdomain located more peripherally (grey) involving centromere (green spot). The central backbone of the chromosome (black curve) protrudes from the RILGE to the RIDGE subdomain and vice versa. Coloured areas represent chromatin loops attached to the backbone. The subdomains can be formed by different parts of the chromosome according to the level of gene expression, which requires bending of the chromosome backbone. (B) The results of Monte-Carlo simulation of HSA 9 structure (solid lines). Radial distributions of the ABL gene and C9 centromere (blue and black circles), the territory of HSA 9 (red triangles) and the distribution of mutual distances between ABL and C9 (inverse black triangles) are appropriately described by the model. In addition, the correlation between ABL and C9 radial positions is also ¢tted (yellow squares).
Discussion In this study, we show that the radial position of a given element in the cell nucleus depends on the radial position of the corresponding chromosome territory and on the location of the element relative to the territory. The chromosome territory and all its regions behave to some extent as a solid body: if the territory is shifted closer to the nuclear centre, the gene or the centromere pertaining to this chromosome is also shifted to the centre as is
shown for TP53 (Figure 2E). Simultaneously, it was observed that the territory of the same chromosome located closer to the nuclear centre is more decondensed than the territory located closer to the membrane. The spatial £uctuations of genetic elements of the more decondensed territory are higher, indicating that decondensed chromatin is more £exible. The distribution of gene expression in the human genome for many di¡erent tissue types has been shown by transcriptome maps (Caron et al. 2001).
The 3D structure of human chromosomes These maps reveal that genes that are highly expressed in di¡erent cell types (housekeeping genes) are not dispersed homogenously through chromosomes, but are grouped into clusters (RIDGEs). Some chromosomes are rich in these clusters (HSA 19, 17, 16, 20); others do not contain such clusters at all (HSA 4, 8, 13, 18). Territories of the former chromosomes are located close to the nuclear center; they are irregular and more di¡use (e.g. HSA 9, 17). The latter are more condensed and adopt a more peripheral position (e.g. HSA 8, 13; Kozubek S. et al. 2002). In the majority of chromosomes, RIDGEs alternate with regions poorly expressed (RILGEs). The territories of these chromosomes have a distinctly polar character, are irregular and stretched from the periphery to the nuclear centre. An important factor in£uencing the nuclear location of a genetic element seems to be the concentration of highly expressed genes in the element environment on the chromosome. Each genetic element can be positioned on the transcriptome map and the density of highly expressed genes in the environment can be established according to Caron et al. (2001). If the genetic element is located in the region rich in highly expressed genes, its nuclear location is close to the nuclear centre (e.g. ABL, TP 53, RARa, Iso-q). If it is in the region poorly populated with expressed genes, its nuclear position is more peripheral (RB, Iso-p, c-MYC). The observation of Carvalho et al. (2001), implying that each individual chromosome constitutes a particular microenvironment in the interphase nucleus, which imposes speci¢c positioning of centromeric heterochromatin in the cell nucleus, is in agreement with our results. These authors found that this microenvironment consists of tight correlation between the nuclear location of centromeric a-satellite DNA and the presence of Gdark bands in the vicinity of the centromere. As follows from the presented results, the polar character of chromosome territories is determined by RIDGEs that protrude from the more condensed parts of the chromosome located in the proximity of the nuclear membrane to the nuclear centre (e.g. HSA 9). The ABL gene is located in the RIDGE occurring close to the telomere (9q34.1). The spatial £uctuations of this gene are rather broad in the central part of the nucleus, far from the centromere and the FWCT of the chromosome
545 territory (Figure 1B). The higher £uctuation of the positions of genetic elements occurring in the regions of clusters of highly expressed genes may be associated with the higher £exibility of the decondensed chromatin that is characteristic for regions with highly expressed genes. A higher degree of intermingling between chromosomes can be expected in the central region of the cell nucleus. On the other hand, c-MYC, located in the chromosomal region with low expression of genes close to the telomere (8q24), is distributed in a relatively narrow spatial volume (together with the centromere of HSA 8) around the FWCT of HSA 8 that is located more peripherally in the cell nucleus than FWCT of HSA 17 and 9 (Figure 1C). Distribution similar to the c-MYC gene was also observed for the RB gene (13q14) (Figure 1 D). Both HSA 8 and 13 are free of RIDGEs, their territories are relatively condensed and close to the nuclear membrane. Bartova et al. (2002) observed that, after di¡erentiation of promyelocytes to granulocytes, the ABL gene is signi¢cantly shifted to the nuclear membrane in contrast to the c-MYC and RB genes. This phenomenon seems to be related to the di¡erent content of highly expressed genes that determine the degree of decondensation in corresponding chromosomes. A gene pertaining to the most decondensed region of a chromosome is exposed to the most pronounced changes in nuclear localization after cell di¡erentiation and chromosome condensation. These results con¢rm our suggestions that the nuclear location of a genetic element is determined by the occurrence of clusters of highly expressed genes in the environment of its chromosome location. Distances shorter than expected from the molecular location were observed between genes located on the p- and q-arms of HSA 17 (TP53/ RARa, TP53/Iso-q). A similar phenomenon was also observed by Nikiforova et al. (2000) for RET and H4 genes located on the q- and p-arms of HSA 10, respectively. This phenomenon may be related to the hairpin-like structure of chromosomes. Such a structure of chromosomes has already been proposed by Ferguson & Ward (1992) and also follows from our further results showing that the centromeres of HSA 9 and 8 are relatively close to the nuclear membrane and both telomeres are oriented to the nuclear centre in human lymphocytes. In addition to this, the
546 nuclear distances between telomeres of both chromosomes are shorter than the distances of the corresponding telomere and the centromere. In accordance with these results Nogami et al. (2000) observed the mutual proximity of telomeres and their orientation towards the nuclear centre for HSA 12. Based on the presented ¢ndings, a simple mathematical model was developed describing several distinct features of interphase chromosome structure. In the model, RIDGE and RILGE subdomains of a chromosome are distinguished (Figure 4A). The results herein show that the actual position of a chromosome territory in individual nuclei £uctuates around a mean value. The subdomains may follow the £uctuations of the territory or may behave relatively independently. In general, a partial correlation between £uctuations of a territory and subdomains can be expected. The actual positions of genetic elements belonging to a particular subdomain acquire the £uctuations of the territory (sd) and the subdomain (ss), and show additional £uctuations (se) that depend on the length of the chromatin loops where they are located. These £uctuations concern di¡erences between cell nuclei that may appear during their formation at the end of telophase (Ferreira et al. 1997) and/or as a consequence of a slow motion of genetic elements during interphase (Zink et al. 1998). The 3D structure of the HSA 17 territory was reconstructed based on the 3D positions of 5 genetic elements. The complex 3D structure of this chromosome territory obviously results from the distribution of highly expressed genes along the chromosome. It follows from the presented results, showing the nuclear location of 5 di¡erent loci of HSA 17, that the structure of this chromosome is not a simple hairpin, rather that the chromosome backbone may be bent several times (Figure 3). The £exion of the chromosome backbone seems to be required for the spatial arrangement of the chromosome territory. It should allow the regions of highly expressed genes to approach the nuclear centre and the regions of low gene expression interspersed between them to reach the nuclear membrane. Thus, two chromosome subdomains are created to which multiple regions of RIDGEs and RILGEs contribute. The chromosome positions of RIDGEs and their separation by RILGEs
E. Luka¤sfl ova¤ et al. are obviously responsible for the £exion of the chromosome and for the structure of the chromosome territory. Therefore, two genetic elements of a particular subdomain may exist closer to each other than the elements located in di¡erent subdomains, depending on the chromosomal positions of RIDGEs and RILGEs. Indeed, we have demonstrated that the RARa and TP53 genes, as well as TP53 and Iso-q, are located closer to each other in comparison to their distances from centromere 17, despite the greater molecular distance between the genes located on the opposite arms of the chromosome. These conclusions are in contradiction to some previous reports showing that physical distances between two DNA loci of the same chromosome are related to their molecular distances (Trask et al. 1991, Yokota et al. 1995). Our results show that, in general, there is no proportionality between the physical and molecular distance of two genetic elements of the same chromosome in the cell nucleus. In reality, the 3D structure of the chromosome territory is probably more complex than would follow only from the radial £exion of the chromosome backbone (see Figure 3). The arrangement of chromatin into loops of various hierarchy, their spatial orientation in the territory and in the cell nucleus, the folding and condensation of these loops together with binding of the territory to the nuclear membrane and/or matrix, intrachromosomal and interchromosome tethering (Kozubek S. et al. 1999, Nikiforova et al. 2000, Kozubek S. et al. 2002) can play an important role in the spatial structure of chromosome territories and localization of individual genetic elements in the frame of the territory and in the cell nucleus. Our study was performed on human nonstimulated lymphocytes to exclude the possible in£uence of di¡erent stages of the cell cycle on the structure of chromosome territory and its location in the cell nucleus. These cells can be considered representative for other cells of spherical shape, as shown by Skalnı¤ kova¤ et al. (2000). Our previous results show that the nuclear location of di¡erent genetic elements in cells of colorectal epithelium and in HT 29 cells of oval shape (Koutna¤ et al. 2001) is similar to that of spherical blood cells. However, the validity of our ¢ndings for cells of other types should be further veri¢ed.
The 3D structure of human chromosomes Acknowledgements We would like to thank H. K ›riva¤nkova¤ and V. Fu ›cı¤ kova¤ for their technical help. This work was supported by the Academy of Sciences (Z5004920, S5004010, A1065203), the Ministry of Education (MSM 143300002), the Ministry of Health (NC5955) and the Grant Agency of the Czech Republic (GA202/01/0197 and GA301/01/0186). References Ba¤rtova¤ E, Kozubek S, Jirsova¤ P et al. (2001) Higher-order chromatin structure of human granulocytes. Chromosoma 110: 360370. Ba¤rtova¤ E, Kozubek S, Jirsova¤ P et al. (2002) Nuclear structure and gene activity in human differentiated cells. J Struct Biol (in press). Belmont AS, Bruce K (1994) Visualization of G1 chromosomes, A folded, twisted, supercoiled chromonema model of interphase chromatide structure. J Cell Biol 127: 287302. Belmont AS, Dietzel S, Nye AC, Strukov YG, Tumbar T (1999) Large scale chromatin structure and function. Curr Opin Cell Biol 11: 307311. Boyle S, Gilchrist S, Bridger JM, Mahy NL, Ellis JA, Bickmore WA (2001) The spatial organization of human chromosomes within the nuclei of normal and emerin-mutant cells. Hum Mol Genet 10: 211219. Brown KE, Guest SS, Smale ST, Hahm K, Merkenschlager M, Fisher AG (1997) Association of transcriptionally silent genes with Ikaros complexes at centromeric heterochromatin. Cell 91: 845854. Brown KE, Baxter J, Graf D, Merkenschlager M, Fisher AG (1999) Dynamic repositioning of genes in the nucleus of lymphocytes preparing for cell division. Mol Cell 3: 207217. Caron H, Van Schaik B, Van der Mee M et al. (2001) The human transcriptom map, Clustering of highly expressed genes in chromosomal domains. Science 291: 12891292. Carvalho C, Pereira HM, Ferreira J et al. (2001) Chromosomal G-dark bands determine the spatial organization of centromeric heterochromatin in the nucleus. Mol Biol Cell 12: 35633572. Chevret E, Volpi EV, Sheer D (2000) Mini review: Form and function in the human interphase chromosome. Cytogenet Cell Genet 90: 1321. Cockel M, Gasser, S (1999) Nuclear compartments and gene regulation. Curr Opin Genet Dev 9: 199205. Cremer T, Cremer C (2001) Chromosome territories nuclear architecture and gene regulation in mammalian cells. Nature Rev Genet 2: 292301. Cremer T, Lichter P, Borden J, Ward DC, Manuelidis L (1988) Detection of chromosome aberrations in metaphase and interphase tumor cells by in situ hybridization using chromosome speci¢c library probes. Hum Genet 80: 235246. Cremer T, Kreth G, Koester H et al. (2000) Chromosome territories, interchromatin domain compartment and nuclear
547 matrix. An integrated view of the functional nuclear architecture. Crit Rev Eucar Gene Expres 10: 179212. Croft JA, Bridger JM, Boyle S, Perry P, Teague P, Bickmore WA (1999) Differences in the localization and morphology of chromosomes in the human nucleus. J Cell Biol 145: 11191131. Ferguson M, Ward DC (1992) Cell cycle dependent chromosomal movement in premitotic human T-lymphocyte nuclei. Chromosoma 101: 557565. Ferreira J, Paolella G, Ramos C, Lamond AL (1997) Spatial organization of large-scale chromatin domains in the nucleus A magni¢ed view of single chromosome territories. J Cell Biol 139: 15971610. Koutna¤ I, Kozubek S, Zaloudı¤ k J et al. (2000) Topography of genetic loci in tissue samples: towards new diagnostic tool using interphase FISH and high-resolution image analysis techniques. Anal Cell Pathol 20: 173185. Kozubek M, Kozubek S, Luka¤sˇ ova¤ E et al. (1999) High-resolution cytometry of FISH dots in interphase cell nuclei. Cytometry 36: 279293. Kozubek M, Kozubek S, Luka¤sˇ ova¤ E et al. (2001) Combined confocal and wide-¢eld high-resolution cytometry of FISH stained cells. Cytometry 45: 112. Kozubek S, Luka¤sˇ ova¤ E, Ry¤znar L et al. (1997) Distribution of ABL and BCR genes in cell nuclei of normal and irradiated lymphocytes. Blood 89: 45374545. Kozubek S, Luka¤sˇ ova¤ E, Mare ›ckova¤ A et al. (1999) The topological organization of chromosomes 9 and 22 in cell nuclei has a determinative role in the induction of t(9,22) translocations and in the pathogenesis of t(9,22) leukemias. Chromosoma 108: 426435. Kozubek S, Luka¤sˇ ova¤ E, Jirsova¤ P et al. (2002) 3D structure of the human genome: Order in randomness. Chromosoma (in press). Kurz A, Lampel S, Nickolenko JE et al. (1996) Active and inactive genes localize preferentially in the periphery of chromosome territories. J Cell Biol 135: 11951205. Lamond AI, Earnshaw WC (1998) Structure and function in the nucleus. Science 280: 547553. Lichter P, Cremer T, Borden J, Manuelidis L, Ward DC (1988) Delineation of individual human chromosomes in metaphase and interphase cells by in situ suppression hybridization using recombinant DNA libraries. Hum Genet 80: 224234. Nagele R, Freeman T, McMorrow L, Lee H (1995) Precise spatial positioning of chromosomes during metaphase: evidence for chromosomal order. Science 270: 18311835. Nagele RG, Valesco AQ, Anderson WJ et al. (2000) Telomere associations in interphase nuclei: possible role in maintenance of interphase chromosome topology. J Cell Sci 114: 377388. Neves H, Ramos C, Gomez da Silva M, Parreira A, Parreira L (1999) The nuclear topography of ABL, BCR, PML and RARa genes: Evidence for gene proximity in speci¢c phases of the cell cycle and stages of hematopoietic differentiation. Blood 93: 11971207. Nikiforova MN, Stringer JR, Blough R, Medvedovic M, Fagin JA, Nikiforov YE (2000) Proximity of chromosomal loci that participate in radiation-induced rearrangements in human cells. Science 290: 138142.
548 Nogami M, Nogami O, Kagotani K et al. (2000) Intranuclear arrangement of human chromosome 12 correlates to largescale replication domains. Chromosoma 108: 514522. Pinkel D, Landegent J, Collins C et al. (1988) Fluorescence in situ hybridization with human chromosome-speci¢c libraries: detection of trisomy 21 and translocations of chromosomes 4. Proc Natl Acad Sci USA 85: 91389142. Sachs RK, van den Engh G, Trask B, Yokota H, Hearst JE (1995) A random-walk/giant-loop model for interphase chromosomes. Proc Natl Acad Sci USA 92: 27102714. Sadoni N, Langer S, Fauth C et al. (1999) Nuclear organization of mammalian genomes: polar chromosome territories build up functionally distinct higher order compartments. J Cell Biol 146: 12111226. Sadoni N, Sullivan KF, Weinzierl P, Stelzer EHK, Zink D (2001) Large-scale chromatin ¢bres of living cells display a discontinuous functional organization. Chromosoma 110: 3951. Skalnı´ kova¤ M, Kozubek S, Luka¤sˇ ova¤ E et al. (2000) Spatial arrangement of genes, centromeres and chromosomes in human blood cell nuclei and its changes during the cell cycle,
E. Luka¤sfl ova¤ et al. differentiation and after irradiation. Chromosome Res 8: 487499. Trask BJ, Massa H, Kenwrick S, Gitschier J (1991) Mapping of human chromosome Xq28 by two-color £uorescence in situ hybridization of DNA sequences to interphase cell nuclei. Am J Hum Genet 48: 115. Verschure PJ, van der Kraan I, Manders EMM, van Driel R (1999) Spatial relationship between transcription sites and chromosome domains. J Cell Biol 147: 1324. Volpi EV, Chevret E, Jones T et al. (2000) Large-scale chromatin organization of the major histocompatibility complex and other regions of human chromosome 6 and its response to interferon in interphase nuclei. J Cell Sci 113: 15651576. Yokota, H, van den Engh G, Hearst JE, Sachs RK, Trask BJ (1995) Evidence for the organization of chromatin in megabase pair-sized loops arranged along a random walk path in human G0/G1 interphase nucleus. J Cell Biol 130: 12391249. Zink D, Cremer T, Saffrich R et al. (1998) Structure and dynamics of human interphase chromosome territories in vivo. Hum Genet 102: 241251.
