1
VERSION AUGUST 17, 2000
(CRITICAL REVIEWS IN EUKARYOTIC GENE EXPRESSION, IN PRESS) CHROMOSOME TERRITORIES, INTERCHROMATIN DOMAIN COMPARTMENT AND NUCLEAR MATRIX: AN INTEGRATED VIEW OF THE FUNCTIONAL NUCLEAR ARCHITECTURE
T. Cremer* (1,2), G. Kreth (3,2+), H. Koester (4), R.H.A. Fink (5), R. Heintzmann (3), M. Cremer (1), I. Solovei (1), D. Zink (1), C. Cremer* (3,2)
1) Institute of Anthropology and Human Genetics, Ludwig Maximilians University, Richard Wagner Str. 10, D-80333 Munich, Germany 2) Interdisciplinary Center for Scientific Computing (+Graduate College), Ruprecht Karls University, D-69120 Heidelberg, Germany 3) Applied Optics & Information Processing, Kirchhoff-Institute of Physics, Ruprecht Karls University, Albert-Ueberle-Str. 3-5, D-69120 Heidelberg, Germany 4) Max-Planck-Institute for Biomedical Research, D-69120 Heidelberg, Germany 5) Physiological Institute, Ruprecht Karls University, D-69120 Heidelberg, Germany
*T. and C. Cremer are corresponding authors email addresses for correspondence: T. Cremer :
[email protected]; FAX: ++49-89-2180-6719 C. Cremer :
[email protected]; Fax: ++49-6221-54-9262
Keywords: nuclear architecture – nuclear matrix – chromosome territories – interchromatin domain compartment
We dedicate this work to Professor Dr. Dr. h.c. Friedrich Vogel on the occasion of his 75th birthday.
2
Summary Advancements in the specific fluorescent labeling of chromatin in fixed and living human cells in combination with 3D (three-dimensional) and 4D (four-dimensional, i.e. space plus time) fluorescence microscopy and image analysis have opened the way for detailed studies of the dynamic, higher order architecture of chromatin in the human cell nucleus and its potential role in gene regulation. Several features of this architecture are now well established: 1. Chromosomes occupy distinct territories in the cell nucleus with preferred nuclear locations although there is no evidence of a rigid suprachromosomal order. 2. Chromosome territories in turn contain distinct chromosome arm domains and smaller chromatin foci or domains with diameters of some 300 to 800 nm and a DNA content in the order of 1 Mbp. 3. Gene dense, early replicating and gene poor, mid to late replicating chromatin domains exhibit different higher order nuclear patterns which persist through all stages of interphase. In mitotic chromosomes early replicating chromatin domains give rise to Giemsa light bands, while mid to late replicating domains form Giemsa dark bands and C-bands. In an attempt to integrate these experimental data into a unified view of the functional nuclear architecture, we present a model of a modular and dynamic chromosome territory organization. We propose that basically three nuclear compartments exist, an ”open” higher order chromatin compartment with chromatin domains containing active genes, a ”closed” chromatin compartment comprising inactive genes, and an interchromatin domain (ICD) compartment (Cremer et al. 1993; Zirbel et al. 1993) which contains macromolecular complexes for transcription, splicing, DNAreplication and repair. Genes in ”open”, but not in ”closed” higher order chromatin compartments have access to transcription and splicing complexes located in the ICD compartment. Chromatin domains which build the ”open” chromatin compartment are organized in a way which allows the direct contact of genes and nascent RNA to transcription and splicing complexes, respectively, preformed in the ICD compartment. In contrast, chromatin domains which belong to the ”closed” compartment are topologically arranged and compacted in a way that precludes the accessibility of genes to transcription complexes. We argue that the content of the ICD compartment is highly enriched in DNA depleted biochemical matrix preparations. The ICD compartment may be considered as the structural
3
and functional equivalent of the in vivo nuclear matrix. A matrix in this functional sense is compatible with but does not necessitate the concept of a 3D nuclear skeleton existing of long, extensively arborized filaments. In the abscence of unequivocal evidence for such a structural matrix in the nucleus of living cells we keep an agnostic attitude about its existence and possible properties in maintaining the higher order nuclear architecture. Quantitative modeling of the 3D and 4D human genome architecture in situ shows that such an assumption is not necessary to explain presently known aspects of the higher order nuclear architecture. We expect that the interplay of quantitative modeling and experimental tests will result in a better understanding of the compartmentalized nuclear architecture and its functional consequences. Abbreviations 3D = three dimensional 4D = four dimensional (space plus time) BrdU = bromodeoxyuridine CHRAC = chromatin accessibility complex CldU = chlorodeoxyuridine CLSM = confocal laser scanning microscopy CT = chromosome territory CTAP = chromosome territory anchor protein EM = electron microscopy FEISEM = field emission in lens scanning electron microscopy FISH = fluorescence in situ hybridization GFP = green fluorescent protein hnRNPs = heterogeneous nuclear ribonucleoproteins ICD space or compartment = interchromatin domain IdU = iododeoxyuridine LM = light microscopy or microscopic MLS = multi loop subcompartment PcG = poly comb group PSF = point spread function RNPs = ribonucleoproteins RW/GL model = random walk/ giant loop model SCD = spherical chromatin domain SPDM = spectral precision distance microscopy trxG = trithorax group Xa = active X chromosome Xi = inactive X chromosome
4
Introduction The functional architecture of the cell nucleus is an unsolved enigma of cell biology (for reviews see (Manuelidis 1985; Jackson 1995; van Driel et al. 1995; Lamond and Earnshaw 1998; Parks and De Boni 1999; Stein and Berezney 1999; Stein and Berezney 2000). On the topological level, and the level of over-all geometry, several lines of research have contributed to the study of this problem during the last decades. More than a hundred years ago Pfitzner and Balbiani claimed that chromosomes are composed of a threadlike arrangement of chromatin globules (”Pfitzner-Balbiani Chromatinkugeln”; (Pfitzner 1881; Waldeyer 1888). August Weismann hypothesized that each chromatin globule is composed of a different quality of germ plasm, a so called Id, with an unknown, but definite molecular structure (for review see (Cremer 1985). Carl Rabl and Theodor Boveri suggested that each chromosome in the cell nucleus of animal cells occupies a distinct territory (Rabl 1885; Boveri 1909). This concept was abandoned when electron microscopic (EM) studies performed in the fifties and sixties of the 20th century revealed areas of compacted and dispersed chromatin, respectively, called hetero- and euchromatin, but failed to distinguish chromosome territories (for review see (Wischnitzer 1973). (Note: The term euchromatin is often used as a synonym for transcriptionally active chromatin. However, one should be aware that euchromatin includes both active and potentially active genes.) At this time only a minority of nuclear researchers argued for a functionally relevant, nonrandom higher order chromatin architecture (for reviews see Comings, 1968; Vogel and Schroeder, 1974; Okada and Comings, 1979). Compelling evidence for the existence of chromosome territories was obtained in the eighties, employing newly developed chromosome painting procedures, and has spurred a growing interest to study the implications of a territorial chromosome organization in nuclear architecture and function (for review see (Cremer et al. 1993)). Recently, in vivo labeling of nuclear proteins, DNA and RNA, have allowed insight into the dynamics of nuclear architecture in the living cell nucleus (Robinett et al. 1996; Misteli et al. 1997; Kanda et al. 1998; Zink and Cremer 1998a; Zink et al. 1998b; Sullivan and Shelby 1999; Phair and Misteli 2000).
5
Electron microscopic (EM) studies have revealed a wealth of ultrastructural information on basic features of nuclear architecture, such as the nuclear envelope and its pores, perichromatin fibrils, interchromatin granules, interchromatin granule associated zones, coiled bodies etc. (Monneron and Bernhard 1969; Fakan and Puvion 1980; Visa et al. 1993; Puvion and Puvion-Dutilleul 1996). Unfortunately, chromatin compaction has made it difficult even at the EM resolution level to study higher order chromatin structures above the level of the beads on a string conformation of nucleosomes and DNA (Olins and Olins 1974; Kornberg and Lorch 1999). Recently, the introduction of new techniques, such as cryo-ultra-microtomy or field emission in lens scanning electron microscopy (FEISEM) allowed the ultrastructural visualization of a complex three-dimensional chromatin architecture (Woodcock and Horowitz 1995; Ris 1997; Lattanzi et al. 1998; Gobbi et al. 1999). So far, however, the details of higher order levels have remained a matter of speculation and controversy. The nuclear matrix has provided a paradigm for a compartmentalized, functional nuclear architecture (Berezney and Coffey 1974). The biochemical characterization of nuclear matrix preparations has revealed a wealth of information on matrix associated complexes of proteins and ribonucleoproteins (RNPs) involved in essential nuclear functions, such as transcription and splicing, DNA-replication and repair (Stein and Berezney 1999). The generation of specific antibodies has provided tools to elucidate the nuclear topology and over-all geometric arrangement of such complexes, but has not yet led to the definition of proteins which constitute the basic network of branched nuclear matrix core filaments described in EM studies (Nickerson et al. 1997). Notably, the existence of a topologically/geometrically defined, structurally and/or functionally important in vivo nuclear matrix has not generally been accepted to date (Hancock 2000; Pederson 2000)
Here, we attempt to integrate presently available experimental results into a unified view of the functional nuclear architecture. We start with a brief summary of the chromosome territoryinterchromatin domain (ICD) compartment model including a suggested nomenclature. Thereafter, we review the present experimental evidence for a compartmentalized nuclear architecture and discuss this evidence in light of models of chromosome territory and nuclear architecture. Being aware of the inherent challenges of any model which attempts to integrate a large body of experimental data,
6
supporters of inductive research strategies have always recommended waiting for a secure factual basis before proposing a model. However, the distinction between ”facts” and model views, it should be noted, is necessarily much less clear cut in ongoing experimental research than undergraduate textbooks often tend to make students believe. Speculative and controversial models that make clear, experimentally falsifiable predictions are an indispensable part of scientific progress. They trigger the mind to conceive of new experiments to substantiate or - more likely - to disprove a given model. We hope that this article will serve this purpose.
Summary of the chromosome territory-interchromatin domain (ICD) compartment model A summary of the chromosome territory-interchromatin domain (ICD) compartment model may be helpful for readers to obtain a comprehensive view before looking into the details of experimental findings and modeling aspects which are described thereafter. The uncertainties and controversies on higher order chromatin structures in the cell nucleus are reflected by the lack of a generally agreed upon nomenclature. In meeting discussions we have often noted that participants had different structures in mind when using terms such as chromosomal or chromatin domains, foci, granules, loops or fibers. Recently, focal chromatin aggregates with a DNA content in the order of 1 Mbp were described as a basic feature of chromosome territories studied in nuclei of living cells (see below). These larger aggregates are possibly built up from chromatin loops in the order of 100 kbp. Under these circumstances we found it most appropriate to use the term chromatin domain together with an indication of the order of DNA content in each case.
Chromatin structure modules We propose a hierarchical order of chromatin structures (Fig. 1) consisting of 1. chromosome territories (ca. 50 - 230 Mbp for human chromosome territories in G1 nuclei and ca. 100 and 460 Mbp, respectively, for replicated chromosomes in G2 nuclei); 2. ~1-Mbp chromatin domains (ranging from a few hundred kbp to several Mbp; corresponding names in the literature: subchromosomal foci, chromatin foci, chromatin granules, multiloop subcompartments) connected by chromatin linkers; and
7
3. ~100-kbp chromatin domains (range from ca. 30-200 kbp, comprising ca. 150-1000 nucleosomes and one to several genes; corresponding names: chromosomal domains, chromatin loops). This proposed view of higher order chromatin organization based on ~100-kbp and ~1-Mbp chromatin domains is certainly strongly oversimplified. Other authors have argued for a hierarchy of thinner and thicker chromatin fibers (Belmont and Bruce 1994). Such a hierarchy possibly co-exists with or is to some extent equivalent to a hierarchy of chromatin domains. Our main reason to adopt, for the time being, the proposed view (1.-3., above) emphasizing chromatin domains of these two sizes is a pragmatic one: this view has provided a straightforward approach for the quantitative modeling of chromosome territories in a way that is consistent with presently available experimental data (see below). A mammalian diploid cell nucleus in G1 contains a total of ca. 6350 Mbp DNA (Morton, 1991). This means that such a nucleus harbors some six thousand ~1-Mbp chromatin domains and some sixty thousand ~100-kbp chromatin domains. These chromatin structures together build up the chromatin compartment of the cell nucleus and may persist during the cell cycle and in terminally differentiated cells. While ~100-kbp chromatin loops have been suggested to represent structural blocks of the eukaryotic genome (Razin 1999), we assume that still higher levels of a structure-function hierarchy exist in the cell nucleus. We hypothesize that chromosome territory anchor proteins (CTAPs) required to maintain the structural integrity of chromosome territories (Ma et al., 1999) are located in the interior of ~1-Mbp chromatin domains and further that these chromatin domains can perform different functions at different times by assembling different protein complexes for transcription, DNA replication and repair. Depending on its association with specific protein complexes, a ~1-Mbp chromatin domain can be functionally defined, e.g., as a replication domain (equivalent names: replication site, replication focus), a transcription domain or a repair domain, although different ~100 kbp chromatin domains which constitute a given ~1-Mbp domain may be regulated independently and functional processes may at least occasionally occur in a given chromatin domain simultaneously.
Open and Closed Chromatin domains
8
We distinguish ”open” from ”closed” ~100 kbp chromatin domains. Although the concept of "closed” and "open” chromatin configurations has been widely discussed and is central to topological/geometrical models of gene regulation (Eissenberg et al. 1985; Hebbes et al. 1994), the structural basis for these configurations has remained elusive. To emphasize this point we will put the terms ”closed” and ”open” in quotation marks throughout the text. ”Open” chromatin domains should possess a configuration such that functionally relevant chromatin sites, e.g., transcription factor binding sites, are easily accessible for large functional protein aggregates located in the interchromatin domain (ICD) space. ”Closed” chromatin domains , in contrast, should maintain a configuration such that binding sites located in the interior of these domains are not accessible for large macromolecule complexes located in the ICD space. >From this, the ”functional surface” of a chromatin domain may be defined as the smallest enveloping surface of the domain which cannot be passed by a given macromolecular complex. Note that ”closed” chromatin domain and ”functional surface” are defined as relative terms, i.e. with regard to the size of excluded macromolecules. Instead of a two-dimensional surface one may also consider a three-dimensional border zone at the periphery of a chromatin domain which is accessible in depth for a certain protein complex in contrast to the chromatin domain interior. In the absence of detailed knowledge we use ”chromatin surface” for simplicity, but imply the possibility of such a border zone as well. A chromatin domain which is ”closed” with regard, e.g., to a large, preformed transcription complex or to a spliceosome may still allow individual proteins or small protein aggregates to penetrate into the chromatin domain interior, e.g. a certain hormone receptor proteins and associated histone acetyltransferases required for chromatin opening (Berger 1999). ”Open” and ”closed” chromatin domains, respectively, likely form transcriptionally active and inactive higher order nuclear compartments (Brown et al. 1997; Wei et al. 1998; Brown et al. 1999; Sadoni et al. 1999; Wei et al. 1999).
Interchromatin domain compartment A three-dimensionally interconnected nuclear space, called the interchromatin domain (ICD) space, estends between higher order chromatin structures. This space together with its content and its boundaries (provided by the surfaces of chromatin domains) is called the ICD compartment (Fig. 1). It
9
was first assumed that the ICD compartment starts at nuclear pores and extends between chromosome territories (Cremer et al. 1993; Zirbel et al. 1993). Later we considered that the ICD compartment has branches that further extend into the interior of chromosome territories (Cremer et al. 1995) and expand between neighbouring ~1-Mbp chromatin domains of a given chromosome territory and with its finest branches between neighbouring ~100-kbp chromatin domains. We hypothesize that the content of the ICD space is highly enriched in biochemical DNA-depleted matrix preparations and comprises complex aggregates composed of many proteins that serve nuclear functions such as transcription, splicing, DNA replication and repair. Protein aggregates of a size that precludes their diffusion into the interior of ”closed” chromatin domains may serve as storage aggregates from which individual components can be released and diffuse into the chromatin interior. Alternatively, functional protein complexes, e.g. transcription complexes preformed in the ICD space, may attach to specific DNA binding sites exposed at the surface of chromatin domains. For reasons detailed below, we propose the term ”in vivo nuclear matrix” to designate both the content and dynamic, structural organization of the ICD space and CTAPs in the nucleus of the living cell. Note that CTAPs are likely not located in the ICD space but in the interior of chromatin domains (see below).
The matrix view of nuclear architecture Since its discovery (Berezney and Coffey 1974) the nuclear matrix has been characterized as a highly complex nuclear structure with a wealth of functionally important associations. The biochemical analysis of the nuclear matrix revealed that hnRNP proteins are major protein components (Stratling and Yu 2000). The nuclear matrix is associated with active genes and essentially involved in transcription, splicing, DNA replication and DNA repair (for review see (Berezney et al. 1995). The obvious, but still controversial conclusion is that this ”chromatin free” nuclear matrix is a principal constituent of the functional nuclear architecture. Electron microscopy has provided evidence for a three-dimensional network of branching 10 nm core filaments (Nickerson et al. 1997). These core filaments may provide attachment sites for (ribonucleo-) protein complexes with distinct functional properties, but are still biochemically ill-defined. It has not been possible to date to raise antibodies that specifically and exclusively stain this network of core filaments.
10
The existence of a permanent nuclear skeleton in the living cell nucleus and its possible role in the organization of chromatin has become a focus of controversy for many years. It has often been argued that biochemical nuclear matrix preparations represent an artifact comprising a wealth of proteins and protein complexes freely diffusable in the nuclear sap. Components found adjacent to each other in an insoluble nuclear matrix preparation may reflect chance aggregations that occurred only during preparation steps but are irrelevant for the in vivo situation. This criticism is reasonable, if one views the nuclear interior as a non-compartmentalized structure, where chromatin fibers float in largely random arrangements in the nuclear sap. In contrast, we view the nucleus as a highly compartmentalized structure. In our view DNA depleted biochemical nuclear matrix preparations do not reflect a random aggregation of components from a nuclear sap, but comprise the partially intact and functional content of the ICD space in the living cell nucleus (see previous discussions of a dynamic in situ nuclear matrix in (Berezney 1984; Berezney et al. 1995). This view implicates a specific organization of ”closed” and ”open” chromatin domains (or chromatin fiber aggregations) which enables specific topological and geometrical relationships between the chromatin and the ICD compartment (see below). The major functional purpose of this compartmentalization is to regulate the access of genes to preformed functional protein complexes contained in the ICD space. Accordingly, our view implicates the formation of higher order nuclear compartments comprising sets of actively transcribed genes in ”open” chromatin domains, and inactive genes in ”closed” domains. Only genes in ”open” chromatin domains should be exposed to the ICD space and thus accessible for transcripton and splicing complexes. Accordingly, we predict a reorganization of chromatin domains and their topological relationships with the ICD compartment when gene expression patterns change. While there is growing biochemical evidence for a gene activity-related remodelling process on the nucleosomal level (Varga-Weisz et al. 1997; Wyrick et al. 1999), experimental proof for the functional importance of a dynamic topological organization for gene regulation at the level of higher order nuclear compartments is limited (Brown et al. 1997; Brown et al. 1999; Bridger et al. 2000). Thus, it may seem prudent to concentrate research efforts on models of gene regulation at the level of nucleosomes and individual chromatin loops. However, such a restriction could unduly postpone studies of a topological level of gene regulation that depends on large scale nuclear architecture,
11
studies which finally may turn out to be indispensable for the understanding of an orchestrated gene regulation during cell cycle, cell differentiation and cellular senescence and the disturbances of this regulation in malignant cells. Recent developments of DNA chips have allowed the study of expression patterns simultaneously for thousands of genes. The results have further emphasized the problem of an orchestrated gene regulation where the expression patters of hundreds or even thousands of genes may be simultaneously affected. Neglect of the functional importance of over-all nuclear geometry or nuclear topology for gene regulation may also be prohibitive for understanding the circumstances which allow the adequate expression of transgenes. The potential impact of nuclear architecture at large on gene regulation cannot be confirmed (or rejected) without the development of appropriate experimental tools and their application by interdisciplinary research teams. Presently available tools include multicolor fluorescence in situ hybridization (FISH) combined with immunocytochemical detection of nuclear proteins for 3D studies of fixed cell nuclei in cycling and terminally differentiated cells, in particular nuclei in tissue sections; chromatin and protein labeling for 4D (three space coordinates plus time) studies of nuclei in cultured, living cells; improved high resolution laser microscopy and bioinformatics for quantitative 3D and 4D evaluations; and higher order chromatin and chromosome computer modeling. Although it is safe to say that changes in chromatin structure are correlated with changes of gene activity, we still lack experimental paradigms which show unequivocally whether changes in higher order chromatin architecture are the cause (as we expect) or the consequence of an orchestrated gene expression, for example by the formation of transcription and splicing complexes at sites of active genes (Singer and Green 1997). As is often the case in the elucidation of complex problems, the framework of a simple dichotomy - changes in nuclear and chromatin architecture are either the cause or the consequence of changes in gene regulation - may turn out to be insufficient. The chromosome territory view of nuclear architecture State of evidence for distinct chromosome territories, their subchromosomal organization and intranuclear distribution
12
Compelling evidence for the existence of chromosome territories in both animal and plant nuclei was established in the '70's and '80's (Zorn et al. 1976; Stack et al. 1977; Zorn et al. 1979; Cremer et al. 1984; Manuelidis 1985; Schardin et al. 1985; Pinkel et al. 1986; Cremer et al. 1988; Lichter et al. 1988; Pinkel et al. 1988; Heslop-Harrison and Bennett 1990). Fig. 2 shows a typical example of a human cell nucleus with chromosome territories visualized by fluorescence in situ hybridization (FISH) with chromosome specific DNA libraries, a procedure termed "chromosome painting". Present evidence argues that each chromosome at least with regard to its major chromatin mass occupies its own, separate territory with distinct arm and band domains (Cremer et al. 1996; Dietzel et al. 1998a; Dietzel et al. 1998b; Visser and Aten 1999; Zink et al. 1999). Chromosome territories exhibit various shapes with occasional finger like chromatin protrusions (Fig. 3). Such protrusions may expand into the ICD space accessible between two neighbouring chromosome territories or into infoldings of an adjacent chromosome territory. Accordingly, the topology between the surfaces of neighbouring chromatin territories can be complex due to expansions and infoldings, but territories apparently remain as mutually exclusive entities., i.e. there is no evidence that chromatin loops from two differently colored (say green and red) chromosome territories could become intertwined to an extent that resembles their full intermingling (yellow) at the light microscopic level (Cremer et al. 1996; Visser and Aten 1999). Several active and inactive genes visualized with specific DNA probes were preferentially located in the periphery of painted chromosome territories in contrast to a noncoding sequence (Kurz et al. 1996). Other studies regarding the spatial distribution of early replicating, GC-rich, transcriptionally active chromatin as compared to mid to late replicating, AT-rich, transcriptionally inactive chromatin, as well as the formation of nascent RNA have provided evidence that transcriptionally active genes are also located in the chromosome territory interior (Visser et al. 1998; Verschure et al. 1999; Tajbakhsh et al. 2000). One should note, however, that the interior location within a chromosome territory is fully compatible with the location at the surface of a ∼1-Mbp or
13
∼100-kbp chromatin domain (Fig. 1). Recently, it was reported that RNA is synthesized at the surface of chromatin as defined by both fluorescence microscopy and EM morphological criteria (Cmarko et al. 1999; Verschure et al. 1999). Dietzel et al. (1999) have noted that the topology of certain Xchromosomal gene sites in the chromosome territory depends on their transcriptional activity. In summary, these findings argue for a non-random distribution of genes in chromosome territories and chromatin domains but generalized rules cannot be deduced at the moment. From the viewpoint of the chromosome territory–ICD compartment model it makes sense if active genes were located at the surface of chromatin domains. From there chromatin loops with one or several active genes could expand into the ICD space, where (as we assume in this model) preformed transcription and splicing complexes are located. Volpi et al. (2000) have recently studied the threedimensional large-scale chromatin organization of the major histocompatibility complex located on human chromosome 6. Large chromatin loops containing several Mbp of DNA were observed extending outwards from the surface of the chromosome 6 territory. The frequency with which a genomic region was observed on an external chromatin loop appeared to be related to the number of active genes in that region. Importantly, transcriptional up-regulation of genes in the major histocompatibility complex led to an increase in the frequency with which these genes were found on an external chromatin loop occasionally several microns away from the segmented surface of the painted chromosome 6 territory. It is not yet clear whether such loops may penetrate neighbouring chromosome territories, but as noted above we think it more likely that they expand into the ICD space. These data further support an association between large-scale chromatin organization of specific genomic regions and their transcriptional status. The question of to what extent chromosomes or chromosomal subregions, such as centromeres or telomeres, maintain specific positions and spatial arrangements during the cell cycle and cellular differentiation has remained as a matter of speculation and controversy. Some authors have claimed a
14
highly ordered positioning of chromosomes in prometaphase rosettes of human fibroblasts and HeLa cells, including the separation of paternal and maternal chromosome sets (Nagele et al. 1995; Nagele et al. 1998). Others have provided data clearly contrary to this claim (Allison and Nestor 1999). For human interphase nuclei some authors have argued that chromosome arrangements do not deviate significantly from a random model distribution (Lesko et al. 1995). Others noted the preferential positioning of certain chromosome types (Popp et al. 1990; Nagele et al. 1999) and provided evidence for a redistribution of chromatin during the cell cycle (Ferguson and Ward 1992; Weimer et al. 1992; Vourc'h et al. 1993) and during cell differentiation (for reviews see (De Boni 1994; Manuelidis 1990). Recently, Croft et al. (1999) reported the preferential localization of the (gene-poor) chromosome 18 territory in the nuclear periphery whereas the (gene-rich) chromosome 19 territory was mostly found in the nuclear center. This positioning changed reproducibly when cells exit from G1 to G0 and vice versa (Bridger et al. 2000). Our own unpublished data suggest a strong preference of small chromosomes towards the center and of large chromosomes towards the periphery of nuclei from different cell types and species (human and chicken; F. Habermann et al., in preparation). Systematic studies using multiple color FISH with defined probe sets in various cell types from different species are necessary in order to obtain a reliable answer to the question whether evolutionary conserved, cell cycle and cell type dependent motifs of higher order chromatin arrangements exist.
Chromosome territories contain focal chromatin aggregates Pulse labeling of cells during S-phase with halogenated thymidine analogs (BrdU, CldU, IdU) revealed chromatin aggregations of ongoing DNA synthesis, termed replication sites or foci (Nakamura et al. 1986; Nakayasu and Berezney 1989). These foci were found to have diameters between some 300 and 800 nm with an estimated DNA content of about 1 Mbp (Jackson and Pombo 1998; Ma et al. 1999). Each replication focus consists of a cluster of active replicons together with replication proteins and their auxiliary factors. This finding, however, is not unambiguously accepted, since other groups performing short labeling assays with 3H-thymidine or BrdU were not able to see distinct replication foci in EM studies (S. Fakan, personal communication).
15
Recently, chromosome territory organization was studied after double-labeling of human cells during early and mid to late S-phase with CldU and IdU (Visser et al. 1998; Zink et al. 1998b; Zink et al. 1999). Several studies demonstrated the persistence of ~1-Mbp chromatin domains labeled during S-phase at other stages of the cell cycle (Ferreira et al. 1997; Zink and Cremer 1998a; Zink et al. 1998b; Zink et al. 1999). Early and mid to late replicating chromatin domains form distinct higher order nuclear compartments in early G1 and persist throughout interphase (Sadoni et al. 1999) and references therein). Early replicating chromatin foci form R-bands in mitotic chromosomes, while midto late replicating foci form G- and C-bands, respectively (Zink et al. 1999) and references therein). It has been hypothesized that chromatin domains are capable of associating with or assemble different sets of factors, e.g. for replication, transcription or repair, at different time points during the cell cycle (Berezney and Wei 1998; Wei et al. 1998; Leonhardt et al. 2000).
When the double labeled cells were further grown for several cycles, segregation of replication labeled and unlabeled chromatids was achieved during the second and subsequent mitotic events resulting in nuclei with labeled chromosome territories side by side with unlabeled ones (Ferreira et al. 1997; Zink et al. 1998b; Zink et al. 1999). The overlap volume measured by confocal laser scanning fluorescence microscopy (CLSM) between differently colored early and mid to late replicating foci in segregated chromosome territories was small, typically a few percent only (Zink et al. 1999). In agreement with the above mentioned views of the early cytologists, these data support the concept of chromosome territories built up from ~1-Mbp chromatin domains.
16
Dynamics of chromosome territory structure in living cell nuclei It is difficult to deduce dynamic features of the large scale geometry and topology of nuclear chromatin during the cell cycle and cell differentiation in fixed cell studies. To overcome this limitation, methods that allow 4D (space + time) studies of large scale chromatin arrangements in living cells were recently developed (Robinett et al. 1996; Shelby et al. 1996; Marshall et al. 1997; Zink and Cremer 1998a; Manders et al. 1999). Zink et al. (1998b) performed microinjection of fluorochrome labeled nucleotides into nuclei of human cells during S-phase. This approach resulted in the live cell fluorescent labeling of nuclear chromatin. As noted above, further proliferation of labeled cells for several cell cycles yielded the segregation of both labeled and unlabeled chromatids, making possible for the first time the direct microsocopic observation of individual, fluorescent labeled chromosome territories in nuclei of living cells. As in the fixed cell studies described above, a composition of individual chromosome territories of chromatin foci with diameters between some 300 and 800 nm was observed. These foci were termed subchromosomal foci and are identical with the ~1Mbp domains observed in fixed cell nuclei. During observation periods of several hours we observed movements of foci and territories typically on the order of a few hundred nanometers with occasional maximum displacements up to several micrometers ((Bornfleth et al. 1999). 4D analyses suggested that both Brownian-like movements, as well as occasional directed movements of chromatin foci occur in living cell nuclei (Zink and Cremer 1998a; Bornfleth et al. 1999). Earlier studies indicated the possibility of major chromatin movements during terminal differentiation of neuronal cells (Manuelidis 1990; De Boni 1994). To allow for such movements, the nuclear matrix must have a sufficiently dynamic structure. While a continuous matrix could provide sufficient elasticity to allow smaller movements, major movements of chromatin territories may require - at least locally - a reorganization or even dissolution of a continuous matrix. The formation of a continuous matrix fiber network could help to stabilize a permanent 3D chromatin architecture in terminally differentiated cells. However, as we will discuss below in more detail, our view of the in vivo nuclear matrix does not necessarily implicate a continuous 3D-structure at any time. Instead, this view implicates that chromosome territories and their tightly associated proteins govern the topology and overall geometry of higher order chromatin architecture with other large macromolecular complexes necessarily located
17
in whatever space is left over. This space could be formed in the nucleus in a way that occasional chromatin loops penetrate into this space. However, the mean chromatin density within chromatin domains, though variable, should never drop below a certain minimum, while the mean chromatin density within the interchromatin space over scales on the order of >50 nm should never get as big as that minimum (Fig. 1, inset). This line of reasoning argues for a qualitative difference between the chromatin and interchromatin compartments, based on quantitative differences in chromatin density. Whether the data actually bear out such a dualistic behavior of chromatin density, rather than just a continuum of values, is not clear, but the apparently chromatin "free" lacunes seen between chromatin do suggest the possibility. An interchromatin space of this kind would not necessarily require a specif organization of the surrounding chromatin. In other words, DNA sequences would be randomly exposed towards the interchromatin space. In contrast to this view we hypothesize that chromatin domains are specifically organized in such a way that transcriptionally active genes or at least their promoter regions and actively transcribed parts at each given time point are exposed at chromatin domain surfaces in direct contact with the interchromatin domain (ICD) compartment.
The interchromatin domain (ICD) compartment view of nuclear architecture A space with a very variable width up to a few microns extending between aggregates of chromatin has been noted in many electron microscopic studies of the nuclear architecture. This space could also be demonstrated in confocal sections from living cell nuclei, in which green fluorescent protein (GFP) tagged histone 2B allowed the visualization of the entire nuclear chromatin (Kanda et al. 1998). Fixation of nuclei with buffered formaldehyde leaves the distribution of the chromatin in the living cell nucleus largely intact (Fig. 4A-C). The working hypothesis was established that the content of this space together with the demarcating chromatin surfaces and possibly decondensed chromatin looping out into this space constitutes a functionally important nuclear compartment, termed the ”inter chromatin domain” (ICD) compartment. We assume that protein complexes for transcription, splicing, DNA replication and repair are localized in this compartment (Cremer et al. 1993; Zirbel et al. 1993; Cremer et al. 1995; Cremer et al. 1996). For example, Fig. 4D-F reveals that variously shaped nuclear bodies that contain a variety of splicing factors and are known as speckles or interchromatin granules,
18
are located within the ICD space. Speckles are dynamic structures that can produce expansions to sites of active genes (Misteli et al. 1997). Further, we assume that chromatin domain surfaces establish a border zone between the ICD and the chromatin compartment of the cell nucleus which can only be penetrated by factors or protein aggregates below a certain size. Note that this definition is relative, e.g. a single protein may penetrate into the interior of a ”closed” chromatin domain, while a large, preformed transcription factor complex cannot. In such a scenario, genes located inside of ”closed” chromatin domains should become inaccessible to transcription protein aggregates. The opening of chromatin domains could bring the regulatory and coding sequences of a given gene in direct contact to the ICD-space and thus allow the specific binding, for example of large transcription complexes formed in this space. This scenario is compatible with experimental evidence for RNA being synthesized at chromatin surfaces (Cmarko et al. 1999; Verschure et al. 1999). RNA polymerase may migrate along a chromatin loop with active genes sufficiently expanded within the ICD space (Volpi et al. 2000). Alternatively, an immobilized transcription factor complex could spool the coding DNA through the complex. Similar considerations have been proposed with regard to replication complexes (Cook 1999). In a recent EM study it was observed that DNA replication, as well as transcription, takes place at the surface of chromatin domains (S. Fakan, personal communication). The extent to which nuclear diffusion of macromolecules and macromolecule complexes is size limited is not yet clear. Evidence based on experiments with fluorescein isothiocyanate (FITC) labeled dextrans (Lukacs et al. 2000) indicates that such macromolecules with a size up to 580 kDa are fully mobile in HeLa cell nuclei. In contrast, comparably sized DNA fragments of 250 bp and greater were nearly immobile suggesting extensive binding to immobile obstacles. Pederson and coworkers have developed an elegant method to visualize movements of endogenous polymerase II transcripts in the nuclei of living cells(Politz et al. 1998; Politz et al. 1999; Politz and Pederson 2000). The authors conclude that intranuclear trafficking of RNA occurs by diffusion throughout the ICD space of the nucleus. Available evidence supports the possibility that diffusion coefficents are considerably lower in certain chromatin compartments than in the ICD space leading to a differential accessibility of macromolecules or macromolecule complexes.
19
Based on the hypothesis that the ICD compartment exists in vivo and possesses all the major functional properties of biochemically DNA depleted nuclear matrix preparations, we propose that its topology and content provides the in vivo equivalent of nuclear matrix preparations (Tan et al. 2000). While the ICD compartment model is compatible with a continuous network of nuclear matrix core filaments, such an assumption is not a necessary condition of this model (for a possible explanation of the formation of a network in nuclear matrix preparations see below). If the concept of the ICD compartment as the equivalent of a dynamic in vivo nuclear matrix holds true, the topology and configuration of components in nuclear matrix preparations should reflect their topology and configuration in the ICD compartment of the living cell to the extent that such features can be preserved in nuclear matrix preparations. A test of this hypothesis requires bringing together light and electron microscopical evidence of chromosome territories, chromatin domains (or higher order chromatin fibers) and the in situ nuclear matrix.
Considerations on the role of the nuclear matrix in chromosome territory organization
Berezney and coworkers (Ma et al. 1999) have argued that a specific subset of proteins, termed chromosome territory anchor proteins (CTAPs), is required to maintain the structural integrity of chromosome territories. The authors noted that chromosome territories remained intact after high salt extractions (2.0 M NaCl or 0.65 M (NH4)2SO4), although these high salt extractions removed the bulk of histone proteins and other soluble nuclear proteins. RNAse A treatment of the cells alone also did not visibly affect the structural integrity of chromosome territories. A disruption of chromosome territories was, however, noted after RNAse plus 2.0 M NaCl treatment. In the light of the multiloop subcompartment (MLS) model (see below) the structural integrity of chromosome territories depends on anchor proteins in the interior of each ~1-Mbp chromatin domain necessary to keep together some ten ~100-kbp chromatin loop domains. Notably, these anchor proteins do not form a continuous backbone (see below). We predict that the CTAPs described by (Ma
20
et al. 1999) are the equivalent of the anchoring proteins predicted by the MLS model. Since CTAPs are not affected by RNAse treatment, it is clear that RNAse treatment alone should not disturb the integrity of chromosome territories. However, one should expect that high salt extraction should be able to remove the CTAPs and thus be sufficient to destroy the chromosome territory structure. Why is this not the case? Why is the combined effect of high salt extraction plus RNAse necessary for chromosome territory disruption? We consider two explanations for this surprising result. Both explanations argue that RNAse attacks the integrity of RNPs contained in the ICD space. The first explanation takes into account the possibility that in addition to CTAPs, a continuous RNP containing matrix in the ICD space is of importance and in fact sufficient to maintain chromosome territories in vivo even in the absence of CTAPs. This continuous matrix should resist high salt treatment and become only destroyed after RNAse treatment. The second explanation argues that the ICD space in the living cell nucleus contains protein and RNP aggregates only in a discontinuous, soluble form. High salt treatment of cell nuclei results in the formation of an insoluble in situ matrix. Whether or not one is willing to accept present evidence for a network of branched core filaments in the nucleus of the living cell, it seems likely that high salt treatment leads to the aggregation of additional proteins and RNPs onto this network. What matters in this context is the assumption that the in situ nuclear matrix present in high salt extracted nuclei holds together the chromosome territory structure even after the removal of CTAPs. We wish to reemphasize here that we consider the in vivo nuclear matrix as a highly dynamic structure (for relevant discussions of a dynamic view of the in situ nuclear matrix see (Berezney 1984; Berezney et al. 1995). The longstanding controvery of whether or not such a matrix is based on a permanent 3D network of branched core filaments in spite of its importance has distracted from such a dynamic view. This dynamic view implicates the possibility (but not necessity) that a continuous 3D network may be fully or partially established at some time, become dissolved at another time (e.g. to enable chromosome territory movements or the reordering of early and mid to late replicating chromatin domains), and again become reestablished later (e.g. to help to preserve a distinct 3D chromatin architecture of a terminally differentiated cell type). Most relevant in the present context is the assumption that the network contains RNA which can be cleaved by RNase A digestion. RNase A
21
digestion then would either destroy the continuous in vivo nuclear matrix network (if it exists) or prevent the aggregation of RNPs to form a continuous network during high salt treatment. In each case the nuclear matrix network surrounds chromatin domains and holds these structures together after the extraction of CTAPs. In summary, we hypothesize that the effects of RNase A and high salt treatments are directed to distinct, geometrically separated structures. The RNPs (and other proteins) possibly form a 3D nuclear matrix network in the ICD space surrounding the chromatin domains, while the CTAPs possibly provide an anchoring structure in the interior of each chromatin domain (Cremer et al. 1995). Both these distinct structures need to be dissolved in order to destroy the territorial organization of chromosomes.
Computer simulation and virtual microscopy of chromosome territory organization and of the ICD compartment Several quantitative models of nuclear genome structure have recently been developed, which explain the observed compartmentalization of nuclear chromatin with and without the assumption of a continuous nuclear or chromosomal matrix/ chromosome scaffold. These models show that chromosome territories with defined substructures can be obtained starting with a few basic assumptions concerning the ultrastructural chromatin folding motifs. Computer modeling allows one to translate chromatin folding motifs into "virtual" microscopical images. Virtual microscopical sections and 3D images of various models of chromosome territory structure were generated taking into account experimentally measured confocal point spread functions to simulate a given optical resolution. Virtual microscopy allows a direct comparison with experimental images and 3D reconstructions of chromosome territories (Münkel et al. 1999; Kreth et al. 2000 (in Press)). Virtual microscopy also helps the experimentalist to realize the limitations of certain light microscopic procedures. Neglect of these limitations can result in misinterpretations and misconceptions.
22
Random walk/giant loop (RW/GL) model The RW/GL model proposed by Sachs et al. (1995) suggests giant chromatin loops with a size of about 3 Mbp. Each loop is created by a random walk of chromatin fibers and held together at its basis by the attachment to a continuous backbone structure (Sachs et al. 1995; Yokota et al. 1995). A continuous backbone structure is a possibility that cannot be excluded by present experimental data. However, it is not an a priori necessity to explain the formation of chromosome territories, since they can be modeled without such an assumption. It is sufficient to assume protein ”clamps” possibly equivalent to CTAPs (Ma et al. 1999), see above) that hold together a chromatin loop at its base (Münkel and Langowski 1998).
Multi Loop Subcompartment (MLS) model The MLS model (Münkel and Langowski 1998; Münkel et al. 1999) was based on the well established finding of ca. 30 – 150 kbp sized chromatin loops. This model assumes that chromatin foci in the order of 1 Mbp are formed by rosettes of several chromatin loops (Okada and Comings 1979) each containing ~100 kbp of DNA. Rosettes are connected by chromatin linkers comprising only a small part of the total nuclear chromatin (Fig. 5). For the simulation of single chromosome territories a potential energy barrier was assumed, which in the absence of neighbouring chromosome territories provided a hindrance for individual chromatin segments to leave the territory volume. The simulation of diploid human cell nuclei with 46 chromosome territories was successfully performed without the use of such a barrier for each individual chromosome territory. The nuclear envelope was considered the equivalent of a rigid spherical shell modeled by a hard sphere potential.
Comparison of the RW/GL model and MLS model with experimental findings Calculation of the overlap volumes of early and mid to late replicating chromatin foci predicted by the RW/GL model (using a special computer simulation of this model) revealed a high degree of overlap, which was not consistent with the experimental data, demonstrating only a very minor overlap of these foci (Zink et al. 1998b; Münkel et al. 1999; Zink et al. 1999). In contrast, in the MLS model (Münkel et al. 1999) the size and mutual exclusivity of individual chromatin rosettes agreed well with
23
the size and mutual exclusivity of the light microscopically observed early and mid to late replicating chromatin foci (Zink et al. 1998b; Zink et al. 1999).
Spherical ~1-Mbp Chromatin Domain (SCD) model and modeling of the interchromatin domain (ICD) space For extended computer simulations of large scale chromatin organization in entire human diploid nuclei the SCD model was developed (Kreth et al., 2000; Kreth et al., in preparation). This model is based on the key parameters used in the MLS model (for a justification of these parameters see (Münkel and Langowski 1998; Münkel et al. 1999). However, taking into account the limited knowledge concerning the actual folding of the chromatin fiber at the ultrastructural level, it makes no assumptions on the ultrastructural chromatin topology inside the ~1-Mbp chromatin domains. Furthermore, this simplification results in a drastic reduction (several orders of magnitude) in computer time and allows for the computing of human model nuclei with 46 chromosome territories within one day using a personal computer. This has already allowed to calculate several hundred model nuclei and to study the resulting arrangements of model chromosome territories as compared to experimentally observed chromosome territory arrangements ( our unpublished data). Fig. 6 shows a 3 spherical human model nucleus with a 10 microns (µm) diameter and a volume of about 500 µm calculated according to the SCD model. Each color visualizes a distinct chromosome territory. For each chromosome territory, we assumed a chain of spherical 1-Mbp chromatin domains with 500 nanometer (nm) diameter. Between different spheres, repulsive forces were assumed (see below for a justification of this assumption). Modeling was done in such a way that it allowed for slight volume overlaps between neigbouring 1-Mbp chromatin domains. This was achieved by an increasing potential energy with a half width of ~250 nm. Chromatin linkers connecting adjacent chromatin spheres were modeled by a spring potential in order to allow a thermodynamic equilibrium distance of about 600 nm between the centers of connected chromatin domains at 37°C. In addition, a spherical
24
potential energy barrier was applied around each territory with a size corresponding to its DNA content. This potential energy barrier is essential to maintain a certain compactness of the modeled chromosome territories. It prohibits chromatin segments from leaving the territory volume and accounts in a drastically simplified way for forces which in real nuclei may arise from a variety of parameters, including the rigidity of higher order chromatin segments, chromosome territory anchor proteins and possibly a nuclear matrix network formed in the ICD space (see above). Using the parameters specified above, we achieved a good correlation of the SCD model with present experimental data concerning the variability of chromosome territories in size and shape, as well as the experimentally measured angle distribution between the centromer and the two telomers of Xchromosome territories (Dietzel et al. 1998a); G. Kreth and C. Cremer, unpublished data). Clearly, the examples of modeling chromosme territory structure and nuclear arrangements presented above are only a beginning. Multifactorial modeling needs to be performed, in which a whole range of variable quantities is assigned to the key parameters. Only then it will become possible to see which model comes closest to the observed data. However, such an approach is rendered difficult not only because of the extensive computer time required but also due to the present limitation of quantitative 3D and 4D experimental data on the higher order nuclear architecture. Improvements on both sides, quantitative modeling and quantitative, experimental tests, are necessary to achieve a better understanding of the compartmentalized nuclear architecture and its functional consequences.
A certain rigidity of higher order chromatin structures is essential for tensegrity models of nuclear architecture (see below). The variable shape and rigidity of higher order chromatin structures also contributes to the formation of an ICD compartment with a variable width of space. We have started to use this approach to model the effects of geometrical constraints of differently sized human chromosome territories on their intranuclear distribution and the frequencies of translocations between pairs of individual chromosome territories (data not shown). The present version of the SCD model
25
neglects Brownian motion of chromatin domains and chromosome territories, respectively. Observations of the ICD-space in confocal sections of nuclei from living HeLa cells with GFP-labeled H2B histons (Fig. 4) suggest a relatively stable ICD space. Depending on the efficient viscosity of large scale chromatin structures, it seems possible that interchromatin channels can open and close due to Brownian motion (R. Sachs, personal communication, and our own preliminary results of computer simulations of Brownian dynamics of the SCD model; Kreth et al., in preparation)
Possible time
scales of such an opening or closure, however, are presently not known.
To create images of virtual ”low resolution” ultrastructural sections through SCD model nuclei, the voxel data of 156 nm thick sections were calculated. As an example, Fig. 7A shows a virtual section in one color only to reflect the fact that chromosome territories in EM sections cannot be differently colored as in multiple color FISH experiments. Fig. 7B shows the same virtual section taking into account the maximum resolution of a commercial confocal fluorescence microscope. Notably, the borders of individual chromosome territories cannot be detected in these images, emphasizing the importance of specific labeling approaches which can distinguish individual chromosome territories. In contrast to the virtual EM section, individual 1-Mbp chromatin domains are no longer distinguishable in the ”light microscopical” section (compare Fig. 4). Both virtual sections, however, reveal an ICD space extending between chromosome territories and ~1-Mbp chromatin domains with diameters up to the micrometer range, while the finest branches of the virtual ICD space extending between modeled ~100-kbp chromatin domains have diameters in the nanometer range (Fig. 7C-E). The density of the ”chromatin” varies in various parts of the model sections. It is obvious that the chromatin-free space ascribed to the ICD compartment space fills up a considerable part of the ”virtual microscopic section”. A minimum of 20% of the nuclear volume was calculated for the ICD space in SCD model nuclei under the assumption that the ICD space is given by the entire spherical nuclear volume minus the total volume of the 6350 ~1-Mbp chromatin domains (each with a diameter of 500 nm) reflecting the total DNA content of a diploid mammalian cell nucleus in G1 (Morton 1991).
26
Note that the ICD space obtained in the SCD model is a consequence of a few simple assumptions of chromatin domain organization including a certain rigidity of higher order chromatin structures. According to the SCD model, the ICD space is formed as a consequence of chromosome territory architecture and its interaction with other chromosome territories and does not require the assumption of a primary, chromatin organizing role of a filamentous nuclear matrix network possibly formed within this space. This view eliminates the problem of how a chromatin organizing nuclear matrix consisting of a continuous 3D network of branched core filaments is replicated during S-phase together with chromatin to organize the two sister chromatids properly. The SCD model, however, does not exclude a secondary role of a filamentous nuclear network (if it exists in vivo) in the maintainance of a higher order chromatin architecture (see also the considerations on a role of tensegrity forces described below). In spherically shaped model nuclei with a diameter of 10 µm and a DNA content of 6350 Mbp, the volume calculated for all nucleosomes and the DNA chain with zero space in between makes up 3 only a few percent of the entire nuclear volume (500 µm ). In some tissues, however, nuclei with much smaller diameters can be observed (4 – 5 µm) resulting in a total nuclear volume of only about 3 50 µm (Mayhew and Astle 1997). In these particularly small nuclei, the volume required for most densely packed nucleosomes occupies almost half of the nuclear volume and much more for ordinary packaged chromatin. Compared to large nuclei, small nuclei with full chromatin content necessitate a higher compactness of chromatin and/or a much smaller ICD space. Small diploid nuclei possibly carry a much larger fraction of inactive genes than large diploid nuclei and provide a particularly interesting case to test the predicted topological and geometrical relationships between the ICD compartment and active or inactive genes (Fig. 1, insert).
Modeling and virtual microscopy of the ultrastructure of ~1-Mbp and ~100-kbp chromatin domains In SCD model nuclei sections through individual ~1-Mbp chromatin domains appear as small ”chromatin circles” resembling the chromatin globules or granules described by the early cytologists. Attempts were undertaken to model the possible ultrastructure of 1-Mbp and 100-kbp chromatin domains at the nucleosome level. Fig. 8 shows several three-dimensional computer models of the
27
interior spatial architecture of 1-Mbp chromatin domains exhibiting different degrees of condensation of some or all 100-kbp domains. In accordance with the MLS model we assumed that each 1-Mbp domain was built up by ten 100-kbp chromatin domains. For the purpose of modeling, we further assumed that the nucleosome fiber of each domain was held together by central elastic, tensegrity forces acting on the respective reference points of the 100-kbp domains (where the modelling process starts; Kreth et al., in preparation). The term "tensegrity" was first introduced in architecture to describe the property of skeleton structures developed by the architect R. Buckminster-Fuller. Tensegrity employs continuous tension parts and discontinuous compression parts, i.e. parts of different rigidity. These parts are interconnected in the whole structure in a way that transient mechanical stresses applied at a given site result in either reversible, or irreversible geometrical rearrangements of all other - even remote - parts, depending on the special conditions of the system. Recently, a major role of tensegrity has been proposed for cellular architecture and function (Ingber 1997; Beliakova et al. 1999; Chen and Ingber 1999). Application of the tensegrity concept to nuclear architecture requires sufficient rigidity of higher order chromatin structures to generate tensegrity forces by the interaction of these chromatin segments with each other and possibly with a nuclear matrix network. In addition to chromosome terrritory anchor proteins (CATPs), a contribution to tensegrity forces might result directly from interactions between DNA segments, e.g. via the interaction of polypurine/poly pyrimidine sequences (Wells et al. 1988; Vogt 1990; Sinden 1994). Between neighbouring, condensed 100-kbp domains shown in Fig. 8A-D, spaces can be noted, suggesting that the ICD-space with its finest branches expands between the surfaces of condensed 100kbp chromatin domains. Within these condensed domains the nucleosome density appears to be so high that the diffusion of large macromolecular complexes (with diameters in the range of, say, 20-30 nm) into the interior of these structures should be prohibited. Here, the interaction between chromatin and ICD-compartment could indeed be limited to the chromatin surface. The role of the 3D chromatin nanostructure on the diffusion of macromolecule complexes needs to be simulated more exactly. A decrease in the tensegrity force of a given 100-kbp chromatin domain is equivalent to a more ”open”
28
configuration (Fig. (8E,F), while an increase leads to a more compacted or ”closed” configuration. Notably, due to geometrical constraints the opening of a single 100-kbp chromatin domain in a 1-Mbp chromatin domain containing other 100-kbp domains in a ”closed” configuration leads to the expansion of the chromatin from the opening 100-kbp domain at the periphery of the remaining ”closed” 100-kbp domains (Fig. 8D). Such an expansion could help to expose the chromatin of the ”open” domain to a branch of the ICD compartment and facilitate the access to protein aggregates contained within this compartment. In contrast to virtual electron microscopic images, virtual light optical confocal sections through modeled chromatin domains do not reveal any details of an ”open” or a ”closed” configuration (Fig. 7C-E). This trivial consequence of the difference in resolution is explicitly mentioned here to reinforce the necessity to take into account the limits of light microscopic observations. Otherwise, conclusions based on light optical approaches, concerning e.g. the ”randomness” of chromatin arrangements or the ”diffuse” intranuclear migration of RNA, can be grossly misleading. Note that the ”practical” 3D resolution of a conventional laser scanning microscope is ca. 250 x 250 x 700 nm. New approaches of far field fluorescence microscoopy have recently been developed which have the potential to obtain 3D-images of entire cell nuclei at a resolution corresponding to a smaller voxel size. These include certain types of point spread function (PSF) engineered microscopy (Hell and Wichmann 1994; Schneider et al. 1998) or X-ray tomographic microscopy (Lehr 1997). In combination with spectral precision distance microscopy (SPDM; (Bornfleth et al. 1998; Edelmann et al. 1999), a further resolution of topological and geometrical details is expected. The SPDM approach is based on the principle, well known in astronomy, that distances between given object sites much smaller than the conventional optical resolution (as given by the point spread function) can be measured if these object sites are labeled with different spectral signatures. Recently, confocal SPDM has been applied to approach the internal topology of the BCR-ABL fusion gene in bone marrow cell nuclei of leukemia patients at a 3D ”resolution equivalent” of about 50 nm, corresponding to
29
approximately one tenth of the ”conventional” optical 3D resolution. (Esa et al. 2000 (in press)). These developments in optical engineering and image analysis, combined with multicolor labeling of specific chromatin sites, should become highly applicable in the future to test predictions of computer models of 1-Mbp and 100-kbp chromatin domains described below.
Physical forces of potential influence in maintaining large scale chromatin and interchromatin domain organization In the models described above, both attractive and repulsive forces were assumed to act between chromatin domains. In the following we will argue that such forces are not just a convenient assumption for computer modeling but are closely correlated to physical forces existing in the cell nucleus. In our view, these forces are essential for the maintainance of distinct, mutually exclusive chromosome territories with a considerable degree of dynamic flexibility (but also a considerable amount of rigidity), the formation of an ICD space, and the opening/closing of chromatin domains (Kreth et al., 2000; Kreth et al., in preparation)
Attractive forces Tensegrity forces will be found in all networks between elements of different elasticity/rigidity (Ingber 1993). Note that in a tensegrity network, ”long range” forces (e.g. elastic forces between opposing sites of the structure) are implied although the actual physical forces act on a short range basis between closely neighboring elements only. The reason for this is the multiple connections given in a tensegrity network. Thus, it is possible that ”short range” forces can combine to create ”long range” interactions. Furthermore, simple physical tensegrity model systems show that a change in the ”macrostructure” of the system (e.g. changing its overall shape) may produce subtle local changes at specific sites, i.e. influence its ”microstructure” in a site specific manner depending on the special construction of the system. In the case of chromatin, the existence of elastic forces and of stretches of different rigidity is experimentally well established (Houchmandzadeh et al. 1997). To produce the network typical for tensegrity forces, attractive forces have to exist not only between neighbouring chromatin domains
30
built up from a continuous stretch of DNA, but also between domains the DNA of which does not form linearly adjacent sequences. Note that chromatin domains can be neighbors in the nuclear space although they are genomically well separated, e.g. by many Mbp along a chromosome. Attractive forces acting between chromatin domains (or rosettes in the MLS model) coming very close to each other include van der Waals forces and hydrogen bonds between neighboring chromatin domains. In addition to hydrogen bonds between proteins, the possibility of such bonds acting between three bases (e.g. Hogsteen pairing), may be taken into account: polypurine/polypyrimidine sequences can give rise to triple-stranded DNA stretches plus a single stranded sequence; the single stranded sequence may form a triple-stranded stretch with a polypurine/polypyrimidine sequence of the DNA of a neighboring chromatin domain (Vogt 1990), or a double stranded stretch with a single stranded sequence from another polypurine/polypyrimidine site. If so, this would provide an additional mechanism for the formation of highly complex, sequence dependent chromatin structures. A computer analysis of the human genome data bank revealed a large number of highly repetitive as well as low repetitive/unique sequence motifs which might serve for such a function (Winkler 1999). Another important source of the site specific short range interactions characteristic for the formation of long range tensegrity forces might be sequence specific non-histone proteins. In Drosophila melanogaster, nuclear factors encoded by the Polycomb group (PcG) and trithorax group (trxG) genes were discovered. These factors act on specific chromosomal elements and play an important role in chromatin remodeling processes, which direct specific chromatin structures into either a ”closed” or ”open” conformation (Cavalli and Paro, 1998, and references therein). Other relevant protein candidates for control of tensegrity might be chromatin accessibility complexes (CHRAC) which induce the sliding of nucleosomes on DNA (Varga-Weisz et al. 1997). Tensegrity forces may be correlated to genetic activity (Ingber 1993; Singhvi et al. 1994). Assuming different tensegrity forces, it was possible to perform computer simulations (Kreth et al. 2000 (in Press)) of the different morphologies experimentally observed for the active (Xa) and the inactive (Xi) X chromosome territories in human amniotic fluid cell nuclei (Eils et al. 1996).
Repulsive forces resulting from Lenard-Jones and Donnan potentials
31
The other major forces assumed in modeling of nuclear genome structure are repulsive short range forces acting between neighbouring chromatin domains. Such forces may originate from different sources: a) Lenard-Jones potentials (declining with the sixth power of the distance) are widely used in solid state physics to describe short range repulsive forces. If the chromatin surfaces of two neighboring chromatin domains come very close together (less than one nm), the electron shells of their atoms will start to strongly repulse each other. These short range repulsive forces inhibit the penetration of neighbouring ~1-Mbp and ~100-kbp domains when these domains are pushed together, e.g. by Brownian chromatin movements..
b) Electrostatic interactions: Here, we want to consider especially a possible role of Donnan potentials. These may be generated as a result of negatively charged phosphate groups of very low electric mobility in the DNA, and counter ions with high electric mobility in the entire interchromatin space. Donnan potentials were expected to be formed between negatively charged chromatin domain surfaces and the ICD space (Cremer et al. 1993; Cremer et al. 1995; Cremer et al. 1996) and to result in repulsive forces between opposite chromatin domain surfaces. Although the range of repulsive forces created by Donnan potentials is much larger than the range of Lenard-Jones potentials, it is limited to a few nm. Since the width of the ICD space may be in the range of several hundred nm and some sites in the µm range (compare Fig. 4 with Fig.7 and 8), it appears at first glance unlikely that Donnan potentials play any role in the maintenance of the ICD compartment. However, we have to consider that the ICD compartment is a dynamic structure. In spite of its enormous width at some locations it becomes very narrow at many other sites. The ICD space in ”closed” higher order chromatin compartments should be largely or entirely devoid of macromolecular complexes for transcription and splicing and therefore much more narrow than the ICD space in transcriptionally active higher order chromatin compartments. As a result of statistical fluctuations of the width of the ICD-space, e.g. as a result of Brownian chromatin movements, chromatin surfaces from neighboring chromatin domains
32
(belonging even to different chromosome territories) occasionally approach each other very closely resulting in the local compression of the space between opposite chromatin domain surfaces. We argue that Donnan potentials in the order of a few mV are sufficient (see calculations and measurements described below) to maintain a minimum width of the ICD-space at these narrow sites. For the case of a zero Donnan potential, ”long range” (i.e. in the order of a few nanometers) repulsive forces should no longer exist. Consequently, opposite chromatin surfaces can be driven so close towards each other by Brownian movements that attractive forces, such as van der Waals forces and hydrogen bonds, may become predominant. This could lead to the (even irreversible) clumping together of chromatin domains (belonging even to different chromosome territories) and result in permanent loss of fine branches of the ICD space. For several reasons we believe that the maintenance of a minimum width of the ICD space (larger than the range of attractive forces between chromatin domains) is an essential feature of the dynamic, functional nuclear architecture predicted by the chromosome territory–ICD compartment model. Let us consider the predicted structural change of a higher order chromatin architecture from a genetically inactive to a transcriptionally highly active state. In the case that short range attractive forces lead to the attachment of chromatin domains as soon as they come close enough in an inactive higher order chromatin architecture with an empty ICD space, it would become necessary to overcome these attractive forces in order to regain a genetically active chromatin architecture. Clumping of chromatin domains from different chromatids and chromosome territories could interfere with chromosome and chromatid separation during mitosis and interfere with decondensation processes during the telophase/G1 transition. These functionally adverse consequences could be avoided, if an ICD space of a few nanometers could be maintained as a result of long range repulsive forces between chromatin domain surfaces even in genetically inactive chromatin, safeguarding a permanent 3D interconnectivity of the ICD space both in genetically active and inactive higher order nuclear compartments. When inactive chromatin compartments become active, narrow ICD compartments could expand as a result of transcription and splicing complexes which form de novo or enter as preformed complexes from the surrounding ICD space segments, pushing opposing chromatin surfaces of narrow ICD spaces apart.
33
In addition to their contribution to the dynamic ICD compartment structure, Donnan potentials in the suggested order of a few mV could enforce the electrostatic (Coulombic) ”trapping” of negatively charged macromolecules and (not too large) macromolecule complexes within the ICD space. Electrostatic trapping may result in an enrichment of such macromolecules, e.g. RNPs, in the ICD space and play a role in their channeled diffusion. If each individual component of functional macromolecular complexes would only show a moderate enrichment within the ICD space, it would follow from thermodynamic considerations that these complexes should form preferentially in the ICD compartment. In contrast, the ICD compartment should be depleted to a certain extent of positively charged proteins, such as the strongly basic histones.
Theoretical considerations on the size of Donnan potentials expected in nuclei of living cells The following calculations are based on the Debye-Hückel and Donnan potential theory and were performed under highly simplified assumptions (Fig. 9). They are provided here solely with the intention to show that the order of magnitude of the required physical effects seems plausible enough to justify further theoretical and experimental efforts. Repulsive forces between the opposing surfaces of two chromatin domains A and B arise as a consequence of negative charges on both surfaces. This assumption is based on the following rationale. Electrophysiological measurements showed that the interior of the interphase nucleus is negatively charged (Kanno and Loewenstein 1963; Oberleitner et al. 1993). DNA and chromatin under physiological conditions carries a small negative net charge. For RNA, DNA, as well as for entire human mitotic chromosomes in suspension the electric mobility at pH -8 2 -1 -1 7 was observed to be in the same order of magnitude, i.e. approx. –1x10 m V s . This finding suggests a negative electric net charge for entire chromosomes in the order of one to several thousand negative elementary charges (Bier et al. 1989). For simplicity, in the following model calculations we take into account negative electric charges of such an order and further assume that chromatin surfaces that demarcate the ICD-space have everywhere the same negative electric mobility. In reality, of course, the mobility depends on the viscosity and the viscosity is different for the actual cell nucleus and for the preparations where the mobility was measured. In any case, chromatin domains should repel each other. In case of an electrolyte in ICD space with monovalent ions of each sign and bulk
34
concentration c, the size of the repulsive force can be estimated as follows. According to the DebyeHückel theory the electric potential U at a distance x from the surface of chromatin domain A can be described by
equation 1: U (x ) = ∆ϕ ⋅ e − Kx where K=
1
DL
⎛ e02 ⎞ 2c ⎟⎟ = ⎜⎜ ⎝ ε r ε 0 k BT ⎠
1
2
e0 = negative elementary charge; ε0 = electric field constant; εr = relative dielectric constant; kB = Boltzmann constant; c= mobile counter ion concentration; T = absolute temperature of the ICD-space (310 Kelvin); DL = Debye length; (all values in SI-units) this length gives the distance from the surface of chromatin domain A where the potential is reduced to 1/e (37%) of ∆ϕ (see below). Fig. 9A shows the behavior of DL in dependence on the mobile counterion concentration according to eq. (1). For example, assuming two types of monovalent, mobile counter ions, e.g. Na+ and Cl-, with a 9 -1 concentration of each 150 mM and a dielectric constant εr = 80 (water), K = 1.2x10 m follows. In this case the Debye length is 0.8 nm. ∆ϕ is the maximum potential difference between the surface of chromatin domain A (x=0) and the ICD fluid at a sufficiently large distance. Usually ∆ϕ is given as a function of the surface charge density σ:
35
equation 2:
∆ϕ =
σ ε rε 0 K
The charge density σ represents the charge load per unit area in the absence of small mobile counterions. Since reliable values for σ are not available, we estimated ∆ϕ using the Donnan theory in its simplest form. This theory gives an estimate of the electric potential difference (Donnan potential) between a phase of fixed ions and a fluid phase with mobile counterions of bulk concentration c of each sign assuming a constant concentration [Y] of negatively charged, fixed phosphate groups within the chromatin domain.
equation 3:
2 ⎤ ⎡ RT ⎢ Y ⎛⎜ ⎛ 2c ⎞ ⎞⎟⎥ ∆ϕ = − ln 1+ 1+ ⎜ ⎟ F ⎢ 2c ⎜ ⎝ Y ⎠ ⎟⎠⎥ ⎣ ⎝ ⎦
where F is the Faraday constant.
In Fig. 9C the expected Donnan potential ∆ϕ is plotted as a function of the mobile counterion concentration for various fixed concentrations [Y]. In a scenario where all phosphate groups are exposed we obtain a Donnan potential estimate of ~2.7mV for the case of a nuclear concentration of monovalent counterions c ~ 150 mM (measurements by Century et al. (1970) yielded a value of c = 145 mM) and of fixed ion concentrations Y ≈ 30 mM. On the average, a large fraction of negatively charged phosphate groups will be already compensated by histones, other nuclear proteins and polyamines. However, this shielding effect might be compensated to an unknown extent by the presence of large amounts of negatively charged nonhistone proteins bound to chromatin.
36
In the following we consider the dependence of the expected repulsive force between two adjacent chromatin domains A and B as a function of the mobile counter ion concentration. For simplicity, it is assumed that chromatin domain A is fixed, while chromatin domain B can be moved. The repulsive force Frep which is exerted by chromatin domain A on chromatin domain B in a distance x can then be calculated (using equation (1) by
equation 4:
Frep (x ) = −QgradU = Q∆ϕ ⋅ Ke − Kx
gradU: gradient of the electric potential U Q: negative net charge on the surface of chromatin domain B; see above for explanation of other symbols. The maximum repulsive force F0 between the two territories expected for x = 0 is given by
equation 5:
F0 = Q∆ϕ K
Frep decreases exponentially with increasing distance x between the two chromatin domain surfaces. For x = DL, i.e. the Debye length, Frep decreases to 1/e F0. Fig. 9B shows the distance dependence Frep/F0 for different concentrations c of mobile counterions in the ICD space. The repulsive force declines to 1/e F0 at 3.2nm for a low ion concentration (10mM). This range decreases to 0.8 nm for 150mM (expected order of mean monovalent ion concentration in the cell nucleus, see above). The
37
maximum repulsive force F0 between two chromatin domains A and B can be estimated for a given mobile ion concentration, if the net charge Q on the surface of chromatin domain B is known.
For a rough estimate of the distance x between a fixed chromatin domain A and an adjacent, movable chromatin domain B produced by the repulsive electric forces Frep discussed above, we neglect Brownian movements and assume that x = 0 for t = 0. (In reality, we assume that the process will be the opposite: following mitosis, the territories approach each other until they are repelled. Due to the repulsive force Frep, the chromatin domain surface of B moves away from A with a given velocity v(t) = dx/dt. This velocity will be slowed down due to ”frictional” forces Ffric exerted mainly by other surrounding territories. For a quantitative treatment, Stokes law
equation 6:
F fric (t ) = 6πηnucl av(t ) is assumed, where a is the radius of the chromatin domain to be moved and Ffric the effective nuclear friction factor exerted on the chromatin domain, using a pseudo viscosity ηnucl). Neglect of acceleration terms (Ffric = Frep) and integration using equ. (4) results in
equation 7:
x(t ) =
⎞ 1 ⎛ Q∆ ϕ K 2 ln⎜⎜ ⋅ t + 1⎟⎟ K ⎝ 6πaη nucl . ⎠
Since Q = 6πηau (u electric mobility of a chromatin domain with net charge Q in a medium with viscosity η),
38
equation 8:
x(t ) =
⎞ 1 ⎛ u∆ϕK 2η ln⎜⎜ ⋅ t + 1⎟⎟ K ⎝ η nucl . ⎠
follows. Note that in this equation the radius a is eliminated. Thus, under the assumptions made the width of the ICD-space is independent of the size of the chromatin domain. For an estimate of the apparent pseudo viscosity ηnucl”, calculations based on optical tweezer experiments suggested ηnucl to be in the order of 50 kgm-1s-1 (Cremer et al. 1993). Recently, the movement of individual chromosome territories was quantitatively measured in undisturbed, living human cell nuclei (Zink et al. 1998). From these data, ηnucl was estimated to be in the order of 2.5-20 kgm-1s-1. Note that ηnucl is a global parameter describing the overall movement of large chromatin structures within the nucleus. Thus, it is expected to be strongly dependent on the type of structures considered. In Fig. 9d, a 9 numerical example for equ. (8) is shown, assuming as constant values ∆ϕ = -2.7 mV, K = 1.2 x 10 m 1 -8 2 -1 -1 -3 -1 -1 -1 -1 , u = 1 x 10 m V sec , η = 0.7 x 10 kgm sec , and two values for ηnucl, 50 kgm sec and -1 -1 2.5 kgm sec . After t = 100 sec the average ICD-compartment width is around 8 to 10 nm and increases to 13 to 15 nm after three hours. Due to the logarithmic dependence (equation 8), the width 2 changes very little, when u∆ϕK η/ηnucl changes by one order magnitude.
Measurements of Donnan potentials In the previous section we have argued that the existence of a Donnan potential in the range of a few mV could account for repulsive electric forces between chromatin domain surfaces sufficient to maintain a minimum width of the ICD space in the range of a few nanometers. The existence of an electrical potential difference between the nuclear interior and the cytoplasm has been shown by a variety of reports (Bustamante 1994; Kanno and Loewenstein 1963; Oberleitner et al. 1993). It was not known, however, whether these differences were due to a Donnan potential. Only a few measurements could be made so far with nuclei from living T-cell-tumor-lymphocytes (JURKAT). These
39
measurements yielded a mean potential difference of -15 mV. For technical reasons, however, we did not succeed to measure Donnan potential curves in nuclei of living cells. We therefore measured the dependence of the nuclear voltage on the charge concentration of the bath solution surrounding fixed cell nuclei which were prepared from osmotically extracted lymphocytes, human tumor T-Cell lymphocytes (JURKAT), cultured diploid human lymphocytes, and HeLa cells. Cells were fixed with either 4% buffered formaldehyde or a 3 : 1 (v,v) mixture of methanol and acetic acid. For voltage recording, a cell nucleus was held with a patch pipette and entered with a KCl-filled microelectrode. The nuclear voltage was monitored as a change of the potential and was negative in amplitude. The penetration of the cell nuclei was frequently successful as could be seen as a step in the monitored voltage (Fig. 10). The recordings were mechanically very sensitive as even very little vibrations produced a short circuit which could be seen as a drop in the voltage. The charge concentration was varied from about 10 mM to 1300 mM, covering the physiological range at about 124 mM up to very high charge concentrations. Recorded voltages were corrected for liquid junction potentials. At physiological conditions (charge concentration 124 mM) the average nuclear voltage was -17.4 mV +/1.6 mV for the formaldehyde fixed JURKAT cell nuclei and -16.8 mV+/-1.3 mV for formaldehyde fixed human lymphocyte cell nuclei. The nuclear voltages followed a curve predicted for a Donnan potential (Fig. 11). The fit with a Donnan curve yielded a concentration of fixed negative charges of 177 mM. For methanol/ acetic acid fixed JURKAT cell nuclei the concentration of fixed negative charges for the fitted Donnan curve was 76 mM. The nuclear voltage at a charge concentration similar to intracellular conditions (124 mM) was -5.2 +/- 0.4 mV for the JURKAT cell nuclei. Methanol/acetic acid fixed diploid lymphocytes were not applicable for comparison. Cell nuclei fixed with methanol/acetic acid were smaller in size than paraformaldehyde fixed nuclei. This effect was particularly notable for lymphocyte nuclei. Since it was not feasible to pick these cells and penetrate them with the microelectrode, we used methanol/ acetic acid fixed HeLa cells instead. The nuclear voltage measured at 124 mM was -5.2 mV+/-0.2 mV for HeLa cell nuclei. This value is about the magnitude reported by other authors (Oberleitner et al. 1993). Our results indicate that the measured nuclear voltage is similar for different human cell types but is strongly dependent on the fixation method. The methanol/acetic acid fixed cell nuclei showed a steady increase of the concentration of
40
fixed charges as calculated from nuclear voltages assuming a Donnan distribution with the mobile charge concentration of the external solution. This change of the concentration of fixed negative charges can be caused by several effects. Either volume changes or changes in the composition of the cell nucleus like loss of histones at high salt concentrations may increase the net charge of the nuclei. Further, a short circuit will affect higher voltage differences more than smaller ones.
The measured relation between the nuclear voltage and the charge concentration of the bath solution clearly followed that of a Donnan potential. This result rules out the possibility that changes in the tip potential of the microelectrodes or liquid-junction potentials are responsible for the results of the electrical measurements, since those artefacts show a completely different dependence on the charge concentration. We have not yet examined the role of the nuclear membrane in maintaining a diffusional resting potential; the nuclear membrane, however, is an unlikely source for a Donnan potential. This is in accordance with results published by others (Oberleitner et al. 1993). The experiments showed the same nuclear voltage despite chemical (methanol/acetic acid fixation) or mechanical (several penetrations with the micropipette) destruction of the nuclear membrane. The results of these preliminary electrophysiological measurements of fixed cell nuclei support the existence of a Donnan potential in the predicted order of several millivolts, but it remains to be seen whether this order can be confirmed for nuclei of living cells as well. In summary, a decisive role of repulsive forces due to charge in the maintainance of a minimum width of the ICD-space, when neighbouring chromatin domains are driven towards each other by Brownian forces, has not been proven so far, but requires further investigations.
Consequences of Donnan potentials for nuclear matrix formation At high salt concentrations (used in standard nuclear matrix preparations) the Donnan potential declined to almost zero values. As a consequence the repulsive forces between negatively charged (ribonucleo-) proteins of the ICD compartment may be reduced to such an extent that they aggregate. Such aggregations may or may not contain a continuous in vivo fiber network of matrix core filaments. Finally, it should be reemphasized that the lack of such an in vivo network does NOT mean that
41
nuclear matrix preparations are a meaningless "artefact" in the sense of a random aggregation of factors distributed in the nuclear sap. On the contrary, we propose that biochemical nuclear matrix preparations represent the content and the functional properties of a specific, topologically and functionally defined subcompartment of the living cell nucleus, called the ICD compartment or in vivo nuclear matrix.
Acknowledgements
This work was supported by the Bundesminister für Bildung und Forschung (BMBF; German Human Genome Project) and the Deutsche Forschungsgemeinschaft. T.C. and C.C. are indebted to Friedrich Vogel for his strong encouragement of our chromosome territory studies over many years. The ICD model was originally developed in 1993 together with P. Lichter. We thank him and the colleagues in his and our groups for many discussions, which helped to shape the views presented here. Stimulating discussions on quantitative higher order chromatin modeling with J. Langowski, Ch. Münkel and T. Knoch are also gratefully acknowledged. We thank R. Berezney, P. Becker, R. van Driel, S. Fakan, W. Hörz, T. Pederson, R. Sachs, A.P. Wolffe and L. Zech for helpful comments on earlier stages of this manuscript. In particular, we greatly appreciate discussions with Ronald Berezney concerning his views of a dynamic structure of the in situ nuclear matrix. Finally, the authors thank P. Edelmann and B. Schädler for support in image analysis and L. Hildenbrand for literature research.
Fig. 1
Chromosome territory - interchromatin domain (ICD) compartment model of nuclear architecture. Each chromosome in a cell nucleus occupies a distinct territory with a variable shape and complexly folded surface. We propose that chromosome territories (CT) are built up from ~1-Mbp chromatin domains connected by chromatin linkers. These ~1-Mbp chromatin domains in turn are composed of linker connected ~100-kbp chromatin (loop) domains (Münkel and Langowski 1998; Münkel et al. 1999). An inter-chromatin domain (ICD) space starts at the nuclear pores and extends between the mutually exclusive chromosome territories where it can form large lacunes with a width up to several micrometers. Finer branches extend into infoldings of territory surfaces and between ~1-Mbp
42
chromatin domains located in the chromosome territory interior, the finest branches possibly end between ~100-kbp chromatin domains. The ICD-space together with its demarcating chromatin domain surfaces (and chromatin loops not shown here that may expand into this space) forms the ICDcompartment of the cell nucleus. It contains macromolecular complexes for transcription, splicing, DNA-replication and repair. For a rough representation of size, higher order chromatin structures and the ICD-space were arbitrarily drawn onto the confocal section through a HeLa cell nucleus with GFPlabeled histone 2B chromatin structures (white background, compare Fig. 4). The inset shows a purely hypothetical example of topological and geometrical relationships between the ICD-compartment and active and inactive genes of ~100-kbp chromatin domains. This drastically simplified scenario assumes that factor complexes act at chromatin domain surfaces and consequently that active genes (white dots) are exposed at chromatin domain surfaces and thus have direct access to preformed factor complexes contained in the ICD-compartment, while inactive genes (black dots) are located in the chromatin domain interior and not accessible by these factor complexes. This scenario, however, does not take into account decondensed ("open") and condensed ("closed") chromatin domain configurations. For more detailed 3D models of "open" and "closed" ~100-kbp chromatin domains compare Fig. 8. We wish to emphasize that small changes in the location of genes (less than 100 nm) can possibly make a decisive difference with regard to the availability of transcription and splicing complexes located in the ICD-compartment. While positional changes of this magnitude cannot be detected by conventional light microscopic approaches, new developments of spectral precision distance microscopy and other types of high resolution laser microscopy provide the means to test this model (see text).
Fig. 2
Visualization and 3D-reconstruction of chromosome territories X and 8 in a human female lymphocyte nucleus A. Light optical serial sections through the nucleus were obtained with a laser confocal scanning microscope. From a total of 48 (250 nm) equidistant sections every fourth section is shown. FISH was performed with specific paint probes for chromosomes X (detected with Cy3, shown in green) and 8 (detected with Cy5, shown in red). The nucleus was counterstained with YOYO (shown in blue).
43
B. 3D reconstruction of the chromosome territories by surface rendering (Amira TM, TGS Inc.) is shown in combination with one median optical section of the counterstained nucleus.
Fig. 3
Optical sections at high magnification through a painted human chromosome 2 territory (A) and chromosome 18 territory (B). Light optical sections of painted chromosome territories recorded from two diploid human fibroblast nuclei. Both chromosome territories were visualised by FISH with the respective chromosome paint probes. Note the irregular shape and focal substructure of the chromosome territories showing regions with different chromatin density. Arrowheads in (B) point to finger like chromatin protrusions extending from a chromosome territory core. ICD-channels expanding into the territory interior (examples (indicated by arrows in (A)) occasionally open up to small ICD-lacunes (marked by asterix). This interpretation is presented with the caveat that structures designated as ICD-channels and lacunes do not represent areas free of chromatin but rather of non-painted chromatin due to a limited DNA complexity of chromosome paint probes and/ or the fact that the visualization of repetitive sequences was suppressed by an excess of unlabeled Cot1 DNA. In favor of this interpretation one should note that light optical serial sections through nuclei where the entire chromatin is labeled in vivo by GFPtagged histone 2B consistently demonstrate the presence of ICD-channels and lacunes (compare Fig. 4). Fig. 4
Chromatin and interchromatin domain (ICD) compartment in HeLa cell nuclei with GFP-tagged histone 2B A. Laser confocal section through the nucleus of a living cell exhibits chromatin of variable density visualized by GFP-tagged histone 2B (Kanda et al. 1998). Three nucleoli (marked by asterix) and the ICD-compartment (black) contrast sharply to the labeled chromatin.
44
B. Laser confocal section corresponding to A obtained after 3D fixation of the nucleus with freshly prepared, PBS-buffered 4% paraformaldehyde (10 min) and permeabilization with 0.5% Triton X 100 (5 min). GFP-tagged chromatin visualized in red. C. The overlay of nuclear sections shown in A and B indicates that the topology/geometry of the chromatin and the ICD-compartment seen in the living cell nucleus is maintained to a large extent in the fixed cell nucleus, although shifts of chromatin domains up to several hundred nm can be noted. For example, arrowheads indicate a rim of condensed chromatin around a large nucleolus, which has slightly widened during fixation. D-E. Laser confocal sections through a fixed HeLa cell nucleus. D. Section showing GFP-tagged chromatin, two nucleoli (marked by asterix) and the ICDcompartment (black). E. Speckles visualized immunocytochemically in the same section with with SC35 antibodies (Cy3; red). F. Overlay of the sections (D) and (E) demonstrates that all speckles are positioned in expanded regions (lacunes) of the ICD-compartment. Notably, the ICD-compartment is only partially filled by the speckles leaving space for other macromolecular complexes.
Fig. 5
Scheme of the multiloop subcompartment (MLS) model (adapted from: Münkel and Langowski (1998), with permission). Three ~100 kbp chromatin loop domains (with a contour length of approximately 1.2 µm) are exemplified. Each domain is modeled as random walk chromatin fiber held together by a loop base spring formed by a harmonic potential. About ten chromatin loop domains form a higher order subcompartment representing a ~1Mbp chromatin domain. These domains are linked by small chromatin fibers. Loop base springs in the MLS model of chromosome territory organization may reflect the function of chromosome territory anchor proteins (CTAP: Ma et al. 1999).
Fig. 6.
45
Spherical 1 Mbp chromatin domain (SCD) model of a diploid human cell nucleus with 46 chromosome territories. All 46 chromosome territories of a simulated diploid, spherical human male cell model nucleus (46,XY) are visualized with 24 different colors (using Persistence of Vision (TM) Ray-Tracer PovRay(TM))). The model nucleus is shown as a relaxed configuration after 150000 Monte Carlo steps (for further details see Kreth et al., in preparation) with statistically independently distributed chromosome territories.
Fig. 7
Virtual sections through a diploid human cell model nucleus with 46 chromosome territories simulated according to the SCD model A. Simulated ”low resolution” ultrastructural section through a digitized three dimensional SCD model of a diploid human cell nucleus (compare Fig. 6). This virtual section was obtained by an extended view projection of two median sections (thickness 156 nm) of the digitized data stack with 78 nm axial and 39 nm lateral voxel sizes. The encirclement indicates two neighboring 1-Mbp domains shown at higher magnification in C-E. B. The simulated median section shown in A. was filtered with a 250nm gaussian kernel. The result corresponds to a light optical (laser confocal) section. Note the loss of information compared to the virtual ”low resolution” ultrastructural image. Note that in both (A) and (B) an interchromatin domain space (black) is clearly recognizable. Borders between virtual chromosome territories, however, cannot be distinguished in the absence of specific coloring of chromatin domains belonging to different territories (compare Fig. 4 and 6). C-E. Virtual sections through the two encircled spherical 1-Mbp chromatin domains from 7A correspond to ”high resolution electron microscopic” (EM) images. The finest dots represent individual nucleosomes. C. The two 1-Mbp chromatin domains were simulated under the assumption that each of the ten 100kbp domains contributing to their formation was highly compacted and formed a mutually distinct
46
compartment. In the virtual section four out of ten highly compacted 100-kbp domains contributing to each 1-Mbp chromatin domain are seen (compare Fig. 8B). D-E. The nucleosome chain configuration within each of the ten 100-kbp chromatin domains was modeled according to Fig. 8E resulting in a moderate compartmentalization (D), or to Fig. 8F resulting in a weak compaction (E) and pronounced loss of compartmentalization. Below the virtual high resolution EM images (C-E), virtual ” light microscopical” (LM) images of the same model structures are shown. The assumed LM resolution of 250 nm was simulated by filtering the digitized model sections with a 250 nm gaussian kernel. Note that in the virtual LM images no internal structure can be deciphered.
Fig. 8
High resolution simulation of the internal ultrastructure of a 1-Mbp chromatin domain. Computer simulations of 1-Mbp domains were performed under the assumption that each domain is built up from ten 100-kbp domains (indicated by 10 pseudo colors). The six simulations (A – F) were performed according to different assumptions on the organization of the nucleosome chain within the 1-Mbp chromatin domain. All Visualizations were done using Persistence of Vision (TM) Ray-Tracer Pov-Ray(TM). A. 1-Mbp chromatin domain model with its nucleosome chain compacted into a 30nm chromatin fiber (visualized by cylinder segments). The 30 nm fiber is folded into ten 100-kbp sized loop domains according to the MLS model (compare Fig. 5) and occasionally interrupted by short regions of a ”beads-on-a-string” chain of individual nucleosomes (10 nm; small white dots, one indicated by an asterix) (compare Alberts et al. (1994). A transcription factor complex bound to an exposed chromatin site is shown as a brown sphere with a diameter of 30 nm (arrow). B. Each of the 100-kbp chromatin domains was modeled under the assumption of a zig-zag (random walk) nucleosome chain with volume exclusion, held together by ”tensegrity” forces directed to the barycenters of the 100-kbp domains. These tensegrity forces result in ”closed” 100-kbp domains (compare Fig. 7C). In these ”closed” 100-kbp chromatin domains, the accessibility to preformed factor complexes built up in the ICD-space is limited to the surface of each domain, while most of
47
the chromatin becomes inaccessible in the chromatin domain interior. Each dot represents an individual nucleosomes. C. ”Beads on a string” nucleosome chains building up 100-kbp chromatin domains are interspersed by several short segments of 30 nm chromatin fibers (four 30 nm fiber segments with 12.4 kbp are visualized as short cylinders). D. Simulation of a 1-Mbp chromatin domain was performed as in (B) with the exception that for one of the 100-kbp chromatin domains (denoted in yellow), a lower tensegrity force was applied. Note that the resulting relaxed chain structure of the yellow 100-kbp domain expands in the periphery of the 1-Mbp domain along the surface of other compacted 100-kbp domains. This peripheral expansion of decondensed chromatin domains would represent an ”open” 100-kbp chromatin domain configuration that allows accessibility to transcription complexes preformed in the ICD space. E. Simulation was performed as in (B) with the difference that a lower tensegrity force was applied to all ten 100-kbp domains resulting in moderately compacted 100-kbp domains (compare Fig. 7B). As a consequence of their decreased compaction, these domains expose more chromatin at the chromatin domain surface. Furthermore, factor complexes which are excluded from the interior of highly compacted, ”closed” chromatin domains (compare B) may now penetrate into the interior. F. Here, for all ten 100-kbp domains only a very low tensegrity force was applied corresponding to a random walk behaviour of the 10 nm nucleosome chains. The effects of decreasing tensegrity forces on chromatin organization noted in (E) is still more pronounced. In addition to a low compaction of all 100-kbp chromatin domains that facilitates the penetration of large factor complexes in the domain interior (compare Fig. 7E), also the compartmentalization the 100-kbp domains into distinct entities is lost (compare Fig. 7A and 8B).
48
Fig. 9
Numerical calculations to estimate the size of Donnan potentials expected in nuclei of living cells. The calculations are based on the Debye-Hückel and Donnan potential theory and were performed under highly simplified assumptions. A. Dependence of the Debye Length DL on the concentration of mobile counterions. B. Distance dependence of the normalized repulsive force Frep/F0 of two chromatin domains for different concentrations c of mobile counterions in the ICD space. C. Donnan potential plotted as a function of the concentration of mobile counterions c in the presence of various concentrations of fixed ions [Y]. D. Estimate of the distance x (t) between a fixed chromatin domain A and an adjacent, mobile chromatin domain B produced by the repulsive electric forces Frep (compare B). For this rough estimate we neglected Brownian movements and assumed that x = 0 for t = 0.
Fig. 10
Shift in the recorded electric potential at the microelectrode penetration of a fixed human cell nucleus. Ordinate: potential in mV, abscissa: time in s. Cell nuclei from diploid human lymphocytes were isolated and fixed using standard procedures. Briefly, human blood was mixed with Heparin and Ringer solution 1:1 to prevent coagulation. Lymphocytes were separated from the plasma and erythrocytes by density centrifugation using UNI-SEP tubes (WAK-Chemie, Homburg, FRG) by centrifugation for 20 min at 1000g. The lymphocytes were extracted and washed with PBS-Buffer. For isolation of cell nuclei, cells were treated 5 min. with hypotonic solution composed of 50 mM KCl at 37°C in order to rupture the cell membranes. Cell nuclei were separated from the cytosolic fraction by repeated centrifugation at 100g for 15 min and removal of the supernatant. Methanol/acetic acid (3:1) or 4% formaldehyde was dropped on a cell nuclei pellet. Cell nuclei were washed and stored in the quasi cytosolic solution (see legend of Fig. 11) in which the cells were bathed during the experiments. The graph shows an electrophysiological recording using a sharp microelectrode filled with 3M KCl (82 M Ω resistance) in this quasi cytosolic solution (see legend of Fig. 11). Repeated impalements of a given nucleus resulted in reproducible shifts of the recorded electric potential.
49
Fig. 11
Electric Potential (mV) of formaldehyde fixed tumor T-Cells measured at various charge concentrations. The graph shows the result of electrical measurements at different charge concentrations. The quasi cytosolic solutions used contained 2 mM MgCl2, 0.15 mM CaCl2 (pCa = 8), 1.1mM EGTA, 10 mM HEPES, pH = 7.2, and various concentrations of K and Na at the ratio of 11:1. Six different solutions with charge concentrations of 10, 72, 124, 240, 394 and 1300 mM were used. Error bars show ± 1 S.D. The continous line shows a calculated curve assuming a Donnan-potential.
References
Alberts B, Bray D, Lewis J, Raff M, Roberts K, Watson JD (1994) Molecular Biology of the Cell. Garland Publ. Inc., New York&London Allison DC, Nestor AL (1999) Evidence for a relatively random array of human chromosomes on the mitotic ring. J Cell Biol 145: 1-14 Beliakova NS, Sadof'ev LA, Podgornaia OI (1999) [Theory of integrating cellular tension and consequences from it]. Tsitologiia 41: 923-926 Belmont AS, Bruce K (1994) Visualization of G1 chromosomes: a folded, twisted, supercoiled chromonema model of interphase chromatid structure. J Cell Biol 127: 287-302 Berezney R (1984) Organization and function of the nuclear matrix. In: Chromosomal nonhistone proteins, vol IV: Structural associations, L.S. Hnlica. CRC pres, Boca Raton, pp 119-180 Berezney R, Coffey DS (1974) Identification of a nuclear protein matrix. Biochem Biophys Res Commun 60: 1410-1417 Berezney R, Mortillaro MJ, Ma H, Wei X, Samarabandu J (1995) The nuclear matrix: a structural milieu for genomic function. Int Rev Cytol: 1-65 Berezney R, Wei X (1998) The new paradigm: integrating genomic function and nuclear architecture. J Cell Biochem Suppl 31: 238-242
50
Berger SL (1999) Gene activation by histone and factor acetyltransferases. Curr Opin Cell Biol 11: 336-341 Bier FF, Bettag U, Rheingans T, Adrian H, Barths J, Hausmann M, Buhring HJ, Rohwer P, Dolle J, Cremer C (1989) Determination of the electrophoretic mobility of chromosomes by free flow electrophoresis. I. Morphology and stability. Electrophoresis 10: 690-697 Bornfleth H, Edelmann P, Zink D, Cremer T, Cremer C (1999) Quantitative motion analysis of subchromosomal foci in living cells using four-dimensional microscopy [In Process Citation]. Biophys J 77: 2871-2886 Bornfleth H, Sätzler K, Eils R, Cremer C (1998) High-precision distance measurements and volumeconserving segmentation of objects near and below the resolution limit in three-dimensional confocal fluorescence microscopy. J. Microsc. 189: 118-136 Boveri T (1909) Die Blastomerenkerne von Ascaris megalocephala und die Theorie der Chromosomenindividualität. Archiv für Zellforschung 3: 181-268 Bridger JM, Boyle S, Kill IR, Bickmore WA (2000) Re-modelling of nuclear architecture in quiescent and senescent human fibroblasts. Curr Biol 10: 149-152 Brown KE, Baxter J, Graf D, Merkenschlager M, Fisher AG (1999) Dynamic repositioning of genes in the nucleus of lymphocytes preparing for cell division. Molecular Cell 3: 207-217 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: 845-854 Century TJ, Fenichel IR, Horowitz SB (1970) The concentration of water, sodium and potassium in the nucleus and cytoplasm of amphibian oocytes. J. Cell Sci. 7: 5-13 Chen CS, Ingber DE (1999) Tensegrity and mechanoregulation: from skeleton to cytoskeleton. Osteoarthritis Cartilage 7: 81-94 Cmarko D, Verschure PJ, Martin TE, Dahmus ME, Krause S, Fu XD, van Driel R, Fakan S (1999) Ultrastructural analysis of transcription and splicing in the cell nucleus after bromo-UTP microinjection. Mol Biol Cell 10: 211-223
51
Comings D (1968) The rationale for an ordered arrangement of chromatin in the interphase nucleus. Am J Hum Genet 20: 440-460 Cook PR (1999) The organization of replication and transcription. Science 284: 1790-1795 Cremer C, Münkel C, Granzow M, Jauch A, Dietzel S, Eils R, Guan XY, Meltzer PS, Trent JM, Langowski J, Cremer T (1996) Nuclear architecture and the induction of chromosomal aberrations. Mutat Res 366: 97-116 Cremer T (1985) Von der Zellenlehre zur Chromosomentheorie, Naturwissenschaftliche Ekenntnis und Theorienwechsel in der frühen Zell- und Verebungsforschung. Springer Verlag, Heidelberg Cremer T, Baumann H, Nakanishi K, Cremer C (1984) Correlation between interphase and metaphase chromosome arrangements as studied by laser-uv-microbeam experiments. Chromosomes Today 8: 203-212 Cremer T, Dietzel S, Eils R, Lichter P, Cremer C (1995) Chromosome territories, nuclear matrix filaments and interchromatin channels: A topological view on nuclear architecture and function. In: Kew Chromosome Conference IV, Royal Botanic Gardens, Kew, pp 63-81 Cremer T, Kurz A, Zirbel R, Dietzel S, Rinke B, Schröck E, Speicher MR, Mathieu U, Jauch A, Emmerich P, Scherthan H, Ried T, Cremer C, Lichter P (1993) Role of chromosome territories in the functional compartmentalization of the cell nucleus. In: Symp. Quant Biol, vol 58, Cold Spring Harb., pp 777-792 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-specific library probes. Hum Genet 80: 235-246 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 De Boni U (1994) The interphase nucleus as a dynamic structure. Int. Rev. Cytol. 150: 149-171 Dietzel S, Eils R, Satzler K, Bornfleth H, Jauch A, Cremer C, Cremer T (1998a) Evidence against a looped structure of the inactive human X-chromosome territory. Exp Cell Res 240: 187-196
52
Dietzel S, Jauch A, Kienle D, Qu G, Holtgreve-Grez H, Eils R, Munkel C, Bittner M, Meltzer PS, Trent JM, Cremer T (1998b) Separate and variably shaped chromosome arm domains are disclosed by chromosome arm painting in human cell nuclei. Chromosome Res 6: 25-33 Dietzel S, Schiebel K, Little G, Edelmann P, Rappold GA, Eils R, Cremer C, Cremer T (1999) The 3D positioning of ANT2 and ANT3 genes within female X chromosome territories correlates with gene activity. Exp Cell Res 252: 363-375 Edelmann P, Esa A, Hausmann M, Cremer C (1999) Confocal laser-scanning fluorescence microscopy: In situ determination of the confocal point-spread function and the chromatic shifts in intact cell nuclei. Optik 110: 194-198 Eils R, Dietzel S, Bertin E, Schrock E, Speicher MR, Ried T, Robert-Nicoud M, Cremer C, Cremer T (1996) Three-dimensional reconstruction of painted human interphase chromosomes: active and inactive X chromosome territories have similar volumes but differ in shape and surface structure. J Cell Biol 135: 1427-1440 Eissenberg JC, Cartwright IL, Thomas GH, Elgin SC (1985) Selected topics in chromatin structure. Annu Rev Genet 19: 485-536 Esa A, Edelmann P, Trakhtenbrot L, Amariglio N, Rechavi G, Hausmann M, Cremer C (2000 (in press)) 3D-spectral precision distance microscopy (SPDM) of chromatin nanostructures after triple-color DNA labelling: A study of the BCR-region on chromosome 22 and the Philadelphia hromosome. J. of Microsc.: Fakan S, Puvion E (1980) The ultrastructural visualization of nucleolar and extranucleolar RNA synthesis and distribution. Int Rev Cytol 65: 255-299 Ferguson M, Ward DC (1992) Cell cycle dependent chromosomal movement in pre-mitotic human Tlymphocyte nuclei. Chromosoma 101: 557-565 Ferreira J, Paolella G, Ramos C, Lamond AI (1997) Spatial organization of large-scale chromatin domains in the nucleus: a magnified view of single chromosome territories. J Cell Biol 139: 1597-1610
53
Gobbi P, Falconi M, Vitale M, Galanzi A, Artico M, Martelli AM, Mazzotti G (1999) Scanning electron microscopic detection of nuclear structures involved in DNA replication. Arch Histol Cytol 62: 317-326 Hancock R (2000) A new look at the nuclear matrix. Chromosoma 109: 219-225 Hebbes TR, Clayton AL, Thorne AW, Crane-Robinson C (1994) Core histone hyperacetylation comaps with generalized DNase I sensitivity in the chicken beta-globin chromosomal domain. Embo J 13: 1823-1830 Hell SW, Wichmann J (1994) Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy. Opt. Lett. 19: 780-782 Heslop-Harrison JS, Bennett MD (1990) Nuclear architecture in plants. Trends Genet 6: 401-405 Houchmandzadeh B, Marko JF, Chatenay D, Libchaber A (1997) Elasticity and structure of eukaryote chromosomes studied by micromanipulation and micropipette aspiration. J Cell Biol 139: 1-12 Ingber DE (1993) The riddle of morphogenesis: a question of solution chemistry or molecular cell engineering? Cell 75: 1249-1252 Ingber DE (1997) Tensegrity: the architectural basis of cellular mechanotransduction. Annu Rev Physiol 59: 575-599 Jackson DA (1995) Nuclear organization: uniting replication foci, chromatin domains and chromosome structure. Bioessays 17: 587-591 Jackson DA, Pombo A (1998) Replicon clusters are stable units of chromosome structure: evidence that nuclear organization contributes to the efficient activation and propagation of S phase in human cells. J Cell Biol 140: 1285-1295 Kanda T, Sullivan KF, Wahl GM (1998) Histone-GFP fusion protein enables sensitive analysis of chromosome dynamics in living mammalian cells. Curr Biol 8: 377-385 Kanno Y, Loewenstein WR (1963) A study of the nucleus and cell membranes of oocytes with an intra-cellular electrode. Exp. Cell Research. 31: 149-166 Kornberg RD, Lorch Y (1999) Twenty-five years of the nucleosome, fundamental particle of the eukaryote chromosome. Cell 98: 285-294
54
Kreth G, Edelmann P, Münkel C, Langowski J, Cremer C (2000 (in Press)) Translocation frequencies for X and Y chromosomes predicted by computer simulations of nuclear structure, chromosome structure and function.: Kurz A, Lampel S, Nickolenko JE, Bradl J, Benner A, Zirbel RM, Cremer T, Lichter P (1996) Active and inactive genes localize preferentially in the periphery of chromosome territories. J Cell Biol 135: 1195-1205 Lamond AI, Earnshaw WC (1998) Structure and function in the nucleus. Science 280: 547-553 Lattanzi G, Galanzi A, Gobbi P, Falconi M, Matteucci A, Breschi L, Vitale M, Mazzotti G (1998) Ultrastructural aspects of the DNA polymerase alpha distribution during the cell cycle. J Histochem Cytochem 46: 1435-1442 Lehr J (1997) 3D X-ray microscopy: Tomographic imaging of mineral sheaths of bacteria Leptothrix Ochracea with the Göttingen x-ray microscope at BESSY. Optik 104: 1-5 Leonhardt H, Rahn HP, Weinzierl P, Sporbert A, Cremer T, Zink D, Cardoso MC (2000) Dynamics of DNA replication factories in living cells. J Cell Biol 149: 271-280 Lesko SA, Callahan DE, LaVilla ME, Wang ZP, Ts'o PO (1995) The experimental homologous and heterologous separation distance histograms for the centromeres of chromosomes 7, 11, and 17 in interphase human T-lymphocytes. Exp Cell Res 219: 499-506 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: 224-234 Lukacs GL, Haggie P, Seksek O, Lechardeur D, Freedman N, Verkman AS (2000) Size-dependent DNA mobility in cytoplasm and nucleus. J Biol Chem 275: 1625-1629 Ma H, Siegel AJ, Berezney R (1999) Association of chromosome territories with the nuclear matrix. Disruption of human chromosome territories correlates with the release of a subset of nuclear matrix proteins. J Cell Biol 146: 531-542 Manders EM, Kimura H, Cook PR (1999) Direct imaging of DNA in living cells reveals the dynamics of chromosome formation. J Cell Biol 144: 813-821
55
Manuelidis L (1985) Individual interphase chromosome domains revealed by in situ hybridization. Hum Genet 71: 288-293 Manuelidis L (1990) A view of interphase chromosomes. Science 250: 1533-1540 Marshall WF, Straight A, Marko JF, Swedlow J, Dernburg A, Belmont A, Murray AW, Agard DA, Sedat JW (1997) Interphase chromosomes undergo constrained diffusional motion in living cells. Curr Biol 7: 930-939 Mayhew TM, Astle D (1997) Photoreceptor number and outer segment disk membrane surface area in the retina of the rat: stereological data for whole organ and average photoreceptor cell. J Neurocytol 26: 53-61 Misteli T, Caceres JF, Spector DL (1997) The dynamics of a pre-mRNA splicing factor in living cells. Nature 387: 523-527 Monneron A, Bernhard W (1969) Fine structural organization of the interphase nucleus in some mammalian cells. J Ultrastruct Res 27: 266-288 Morton NE (1991) Parameters of the human genome. Proc Natl Acad Sci U S A 88: 7474-7476 Münkel C, Eils R, Dietzel S, Zink D, Mehring C, Wedemann G, Cremer T, Langowski J (1999) Compartmentalization of interphase chromosomes observed in simulation and experiment. J. Mol. Biol. 285: 1053-1065 Münkel C, Langowski J (1998) Chromosome structure predicted by a polymer model. Phys. Rev. E 57: 5888-5896 Nagele R, Freeman T, McMorrow L, Lee HY (1995) Precise spatial positioning of chromosomes during prometaphase: evidence for chromosomal order. Science 270: 1831-1835 Nagele RG, Freeman T, Fazekas J, Lee KM, Thomson Z, Lee HY (1998) Chromosome spatial order in human cells: evidence for early origin and faithful propagation. Chromosoma 107: 330-338 Nagele RG, Freeman T, McMorrow L, Thomson Z, Kitson-Wind K, Lee H (1999) Chromosomes exhibit preferential positioning in nuclei of quiescent human cells. J Cell Sci 112: 525-535 Nakamura H, Morita T, Sato C (1986) Structural organizations of replicon domains during DNA synthetic phase in the mammalian nucleus. Exp Cell Res 165: 291-297
56
Nakayasu H, Berezney R (1989) Mapping replicational sites in the eucaryotic cell nucleus. J Cell Biol 108: 1-11 Nickerson JA, Krockmalnic G, Wan KM, Penman S (1997) The nuclear matrix revealed by eluting chromatin from a cross-linked nucleus. Proc Natl Acad Sci U S A 94: 4446-4450 Oberleitner H, Schuricht B, Wünsch S, Schneider S, Püschel B (1993) Role of H+ ions in volume and voltage of epithelial cell nuclei. Eur. J. Physiol. 423: 88-96 Okada TA, Comings DE (1979) Higher order structure of chromosomes. Chromosoma 72: 1-14 Olins AL, Olins DE (1974) Spheroid chromatin units (v bodies). Science 183: 330-332 Parks PC, De Boni U (1999) Dynamics of structure-function relationship in interphase nuclei. Life Sci. 64: 1703-1718 Pederson T (2000) Half a century of "the nuclear matrix". Mol Biol Cell 11: 799-805 Pfitzner W (1881) Über den feineren Bau der bei der Zelltheilung auftretenden fadenförmigen Differenzierung des Zellkerns. Morphologisches Jahrbuch 7: 289-311 Phair RD, Misteli T (2000) High mobility of proteins in the mammalian cell nucleus. Nature 404: 604609 Pinkel D, Landegent J, Collins C, Fuscoe J, Segraves R, Lucas J, Gray J (1988) Fluorescence in situ hybridization with human chromosome-specific libraries: detection of trisomy 21 and translocations of chromosome 4. Proc Natl Acad Sci U S A 85: 9138-9142 Pinkel D, Straume T, Gray JW (1986) Cytogenetic analysis using quantitative, high-sensitivity, fluorescence hybridization. Proc Natl Acad Sci U S A 83: 2934-2938 Politz JC, Browne ES, Wolf DE, Pederson T (1998) Intranuclear diffusion and hybridization state of oligonucleotides measured by fluorescence correlation spectroscopy in living cells. Proc Natl Acad Sci U S A 95: 6043-6048 Politz JC, Pederson T (2000) Review: movement of mRNA from transcription site to nuclear pores. J Struct Biol 129: 252-257 Politz JC, Tuft RA, Pederson T, Singer RH (1999) Movement of nuclear poly(A) RNA throughout the interchromatin space in living cells. Curr Biol 9: 285-291
57
Popp S, Scholl HP, Loos P, Jauch A, Stelzer E, Cremer C, Cremer T (1990) Distribution of chromosome 18 and X centric heterochromatin in the interphase nucleus of cultured human cells. Exp Cell Res 189: 1-12 Puvion E, Puvion-Dutilleul F (1996) Ultrastructure of the nucleus in relation to transcription and splicing: roles of perichromatin fibrils and interchromatin granules. Exp Cell Res 229: 217-225 Rabl C (1885) Über Zellteilung. Morphologisches Jahrbuch 10: 214-330 Razin SV (1999) Chromosomal DNA loops may constitute basic units of the eukaryotic genome organization and evolution. Crit Rev Eukaryot Gene Expr 9: 279-283 Ris H (1997) High-resolution field-emission scanning electron microscopy of nuclear pore complex. Scanning 19: 368-375 Robinett CC, Straight A, Li G, Willhelm C, Sudlow G, Murray A, Belmont AS (1996) In vivo localization of DNA sequences and visualization of large-scale chromatin organization using lac operator/repressor recognition. J Cell Biol 135: 1685-1700 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 U S A 92: 2710-2714 Sadoni N, Langer S, Fauth C, Bernardi G, Cremer T, Turner BM, Zink D (1999) Nuclear organization of mammalian genomes. Polar chromosome territories build up functionally distinct higher order compartments. J Cell Biol 146: 1211-1226 Schardin M, Cremer T, Hager HD, Lang M (1985) Specific staining of human chromosomes in Chinese hamster x man hybrid cell lines demonstrates interphase chromosome territories. Hum Genet 71: 281-287 Schneider B, Bradl J, Kirsten I, Hausmann M, Cremer C (1998) High precision localization of fluorescent targets in the nanometer range by spatially modulated excitation fluorescence microscopy, fluorescence microscopy and fluorescence probes. Plenum Press Shelby RD, Hahn KM, Sullivan KF (1996) Dynamic elastic behavior of alpha-satellite DNA domains visualized in situ in living human cells. J Cell Biol 135: 545-557 Sinden RR (1994) DNA Structure and Function. Academic Press, San Diego, New York, Boston
58
Singer RH, Green MR (1997) Compartmentalization of eukaryotic gene expression: causes and effects. Cell 91: 291-294 Singhvi R, Kumar A, Lopez GP, Stephanopoulos GN, Wang DI, Whitesides GM, Ingber DE (1994) Engineering cell shape and function. Science 264: 696-698 Stack SM, Brown DB, Dewey WC (1977) Visualization of interphase chromosomes. J Cell Sci 26: 281-299 Stein ES, Berezney R (1999) 25 years of contributions to characterizing gene expression and replication within the three-dimensional context of nuclear architecture. Crit Rev Eukaryot Gene Expr 9: v-vi Stein GS, Berezney R (2000) 25 years of contributions to characterizing gene expression and replication within the three-dimensional context of nuclear architecture. Crit Rev Eukaryot Gene Expr 10: v-vi Stratling WH, Yu F (2000) Origin and roles of nuclear matrix proteins. Specific functions of the MARbinding protein MeCP2/ARBP. Crit. Rev. Eukaryot. Gene Expr. 9: 311-318 Sullivan KF, Shelby RD (1999) Using time-lapse confocal microscopy for analysis of centromere dynamics in human cells. Methods Cell Biol 58: 183-202 Tajbakhsh J, Luz H, Bornfleth H, Lampel S, Cremer C, Lichter P (2000) Spatial distribution of GCand AT-rich DNA sequences within human chromosome territories. Exp Cell Res 255: 229-237 Tan JH, Wooley JC, LeStourgeon WM (2000) Nuclear matrix-like filaments and fibrogranular complexes form through the rearrangement of specific nuclear ribonucleoproteins. Mol Biol Cell 11: 1547-1554 van Driel R, Wansink DG, van Steensel B, Grande MA, Schul W, de Jong L (1995) Nuclear domains and the nuclear matrix. Int Rev Cytol: 151-189 Varga-Weisz PD, Wilm M, Bonte E, Dumas K, Mann M, Becker PB (1997) Chromatin-remodelling factor CHRAC contains the ATPases ISWI and topoisomerase II [published erratum appears in Nature 1997 Oct 30;389(6654):1003]. Nature 388: 598-602 Verschure PJ, van Der Kraan I, Manders EM, van Driel R (1999) Spatial relationship between transcription sites and chromosome territories. J Cell Biol 147: 13-24
59
Visa N, Puvion-Dutilleul F, Bachellerie JP, Puvion E (1993) Intranuclear distribution of U1 and U2 snRNAs visualized by high resolution in situ hybridization: revelation of a novel compartment containing U1 but not U2 snRNA in HeLa cells. Eur J Cell Biol 60: 308-321 Visser AE, Aten JA (1999) Chromosomes as well as chromosomal subdomains constitute distinct units in interphase nuclei. J Cell Sci 112: 3353-3360 Visser AE, Eils R, Jauch A, Little G, Bakker PJ, Cremer T, Aten JA (1998) Spatial distributions of early and late replicating chromatin in interphase chromosome territories. Exp Cell Res 243: 398-407 Vogel F, Schroeder TM (1974) The internal order of the interphase nucleus. Humangenetik 25: 265297 Vogt P (1990) Potential genetic functions of tandem repeated DNA sequence blocks in the human genome are based on a highly conserved "chromatin folding code". Hum Genet 84: 301-336 Volpi EV, Chevret E, Jones T, Vatcheva R, Williamson J, Beck S, Campbell RD, Goldsworthy M, Powis SH, Ragoussis J, Trowsdale J, Sheer D (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: 1565-1576 Vourc'h C, Taruscio D, Boyle AL, Ward DC (1993) Cell cycle-dependent distribution of telomeres, centromeres, and chromosome-specific subsatellite domains in the interphase nucleus of mouse lymphocytes. Exp Cell Res 205: 142-151 Waldeyer W (1888) Über Karyokinese und ihre Beziehung zu den Befruchtungsvorgängen. Archiv für mikroskopische Anatomie 32: 1-122 Wei X, Samarabandu J, Devdhar RS, Siegel AJ, Acharya R, Berezney R (1998) Segregation of transcription and replication sites into higher order domains [see comments]. Science 281: 1502-1506 Wei X, Somanathan S, Samarabandu J, Berezney R (1999) Three-dimensional visualization of transcription sites and their association with splicing factor-rich nuclear speckles. J Cell Biol 146: 543-558
60
Weimer R, Haaf T, Kruger J, Poot M, Schmid M (1992) Characterization of centromere arrangements and test for random distribution in G0, G1, S, G2, G1, and early S' phase in human lymphocytes. Hum Genet 88: 673-682 Wells RD, Collier DA, Hanvey JC, Shimizu M, Wohlrab F (1988) The chemistry and biology of unusual DNA structures adopted by oligopurine.oligopyrimidine sequences. Faseb J 2: 29392949 Winkler R (1999) Computergestützte Bestimmung spezifischer DNA-Proben und deren Einsatz in der Niedertemperatur Fluoreszenzhybridisierung von Chromosomen. In. Institute of Applied Physics, University of Heidelberg, Heidelberg Wischnitzer S (1973) The submicroscopic morphology of the interphase nucleus. Int Rev Cytol 34: 148 Woodcock CL, Horowitz RA (1995) Chromatin organization reviewed. Trends Cell Biol 5: 272-277 Wyrick JJ, Holstege FC, Jennings EG, Causton HC, Shore D, Grunstein M, Lander ES, Young RA (1999) Chromosomal landscape of nucleosome-dependent gene expression and silencing in yeast. Nature 402: 418-421 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 the human G0/G1 interphase nucleus. J Cell Biol 130: 1239-1249 Zink D, Bornfleth H, Visser A, Cremer C, Cremer T (1999) Organization of early and late replicating DNA in human chromosome territories. Exp Cell Res 247: 176-188 Zink D, Cremer T (1998a) Chromosome dynamics in nuclei of living cells. Curr. Biol. 8: R321-324 Zink D, Cremer T, Saffrich R, Fischer R, Trendelenburg MF, Ansorge W, Stelzer EH (1998b) Structure and dynamics of human interphase chromosome territories in vivo. Hum Genet 102: 241-251 Zirbel RM, Mathieu UR, Kurz A, Cremer T, Lichter P (1993) Evidence for a nuclear compartment of transcription and splicing located at chromosome domain boundaries. Chromosome Res 1: 93106
61
Zorn C, Cremer C, Cremer T, Zimmer J (1979) Unscheduled DNA synthesis after partial UV irradiation of the cell nucleus. Distribution in interphase and metaphase. Exp Cell Res 124: 111-119 Zorn C, Cremer T, Cremer C, Zimmer J (1976) Laser UV microirradiation of interphase nuclei and post-treatment with caffeine. A new approach to establish the arrangement of interphase chromosomes. Hum Genet 35: 83-89
62
Figure1
63
Figure 2
64
Figure 3
65
Figure 4
66
Figure 5
Figure 7
Figure 6
67
Figure 8
68
Ffigure 9
69
Figure 10
Figure 11