Understanding the evolutionary origin of the nucleus and its compartmentalized architecture provides a huge but, as expected, greatly rewarding challenge in the post-genomic era. We start this chapter with a survey of current hypotheses on the evolutionary origin of the cell nucleus. Thereafter, we provide an overview of evolutionarily conserved features of chromatin organization and arrangements, as well as topographical aspects of DNA replication and transcription, followed by a brief introduction of current models of nuclear architecture. In addition to features which may possibly apply to all eukaryotes, the evolutionary plasticity of higher-order nuclear organization is reflected by cell-type- and species-specific features, by the ability of nuclear architecture to adapt to specific environmental demands, as well as by the impact of aberrant nuclear organization on senescence and human disease. We conclude this chapter with a reflection on the necessity of interdisciplinary research strategies to map epigenomes in space and time.
Nuclear architecture represents the highest level of structural order and information content of the epigenome, or, more precisely, the multitude of epigenomic variants present in different cell types of multicellular species or even in single eukaryotes at different times and under different environmental challenges. Epigenomes may be characterized by the functional interplay of local chromatin compaction, higher-order chromatin arrangements and nuclear architecture at large within the genome. The genome has been equated with the cell’s DNA during the second half of the past century, although the early theoretical and experimentally working cytogeneticists, such as August Weismann and Theodor Boveri, to name a few, would rather have expected a solution to the riddle of inheritance in the chromatin architecture not within a single type of molecule. It has become evident that this architecture is dynamic and has an impact on regulatory networks of gene expression. Evidence for dynamic changes of a non-random chromatin order in cycling cells  and during terminal cell differentiation , as well as changes of transcriptional activities correlated with the repositioning of genes from repressive to transcriptionally favourable nuclear compartments and vice versa , emphasize the necessity of detailed studies of the dynamic nuclear architecture in order to understand cell-type-specific nuclear organization and function. We are just beginning to understand how local chromatin compaction, higher-order chromatin organization and nuclear compartmentalization control regulatory networks of gene expression and other nuclear functions, such as DNA replication and repair (for a review see ), and how disorders of nuclear architecture contribute to severe diseases, such as laminopathies  and cancer . Microscopic studies in this rising, but still relatively little explored, field of research have been greatly supported by molecular approaches. By combining a proximity-based ligation assay of DNA sequences in both cis (same chromosome) and trans (different chromosomes) with massive parallel sequencing (Hi-C), the construction of spatial proximity maps of entire genomes from a given cell type has become possible down to a resolution level of 1 Mb, with the prospect of even higher resolution in the near future . Studies of higher-order chromatin organization must be integrated with ongoing research of local chromatin compaction brought about by the interplay of DNA methylation with histone modifications and chromatin remodelling . Local modifications of chromatin, as well as non-random higher-order chromatin arrangements can be stably propagated over many cell cycles, but are also subject to changes in cycling cells, as well as during post-mitotic differentiation [2,8,9].
Despite substantial progress, a large gap must still be bridged to understand the functional interplay of the diploid genomes present in all diploid cells of the body with cell-type-specific epigenomic features. Currently discussed models of nuclear organization and function emphasize this gap . Views of higher-order organization presented below still differ. These differences may indicate to some extent the great variability of nuclear organization in different cell types and species, but also the highly speculative nature of current models, which are a telling reflection of our still insufficient knowledge about basic features of nuclear organization.
Reiterating Theodosius Dobzhansky’s famous phrase “nothing makes sense in biology except in the light of evolution” , we are convinced that understanding the evolution of the nucleus and its compartmentalized architecture provides a rewarding challenge for interdisciplinary research in the post-genomic era. The present chapter shows how much, or better how little, is presently known.
Eukaryogenesis and the evolutionary origin of the nucleus
One of the most enigmatic events in early life’s evolution is the origin of the cell nucleus as a distinct structural and functional entity of the cell, and therewith the emergence of the eukaryotic domain. While the karyogenic hypothesis proposes that the nucleus and its enclosing membrane system were gradually acquired in unknown ways within a cell lineage, the endokaryotic hypothesis acts on the assumption that the nucleus derives from an endosymbiont taken up by an engulfing host species [11–13].
Since both the ER (endoplasmic reticulum) and the nuclear envelope build a continuous compartment in extant eukaryotes, it seems plausible that the shaping of the nuclear envelope is closely interconnected with the evolution of the endomembrane system by invagination and inward separation of the cellular membrane . Accordingly, the primitive nucleus would have been constituted by the aggregation of ER cisternae covering the DNA, possibly to protect the DNA initially from shearing damage by novel molecular motors [15,16]. Very recently it has been proposed that octagonal NPCs (nuclear pore complexes) co-evolved with nuclear compartmentalization, whereby proto-NPCs could derive from modified coat protein complexes (COPII), which are involved in selective transport from the ER by generating transport vesicles . Later the acquisition of mitochondria, which most probably derived from an α-proteobacterial endosymbiont , into a host forged synergistically the origins of eukaryotic cell properties [14,19,20].
Currently, there is a controversial debate about the original nature of this putative host and the possible selective pressure that enforced the compartmentalization of the protoeukaryotic ancestor cell into a nucleus and the cytosol. It has been speculated that the host that engulfed the mitochondrial ancestor was a protoeukaryal intermediate between the first eukaryote and stem neomura capable in phagocytosis (Neomura is a speculative clade containing the archaeal and eukaryal sister domains) [14,21]. The mitochondrial seed hypothesis proposes that, after mitochondrial acquisition, spliceosomal introns as derivatives of group II self-splicing introns spread from the protomitochondrial endosymbiont throughout the host genome [22,23]. Arguably, nuclear compartmentalization facilitated the origin of spliceosomal introns from enslaved mitochondrial group II introns . The rationale of this theory is that spliceosomal pre-mRNA processing is slow in comparison with translation at ribosomes, which is fast. Incomplete maturation of pre-mRNAs due to the co-existence of DNA and the translational machinery might strongly enhance the risk of generating incorrect proteins. According to this line of reasoning, spliceosomal introns co-evolved only after acquisition of the nuclear envelope. Also, based on the theory that spliceosomal introns and spliceosomal snRNAs (small nuclear RNAs), and possibly also eukaryotic retroelements, originate from group II self-splicing introns [22,24], another recent model accounts for the generation of selective pressure for nuclear evolution. This model proposes that group II self-splicing introns from the α-proteobacterial ancestor of mitochondria massively invaded the genome of an archaeal host, leading to selective pressure for the evolution of the spliceosome and for sequestering the machineries of transcription and splicing from translation in the cytosol [19,20]. Contrary to this model it has been argued that there is no observation or theoretical basis that introns can behave sufficiently aggressively in lineages without meiotic sex, thus challenging the hypothesis of an archaeal host. It is further argued that no extant archaeal species have been identified which carry bacterial endosymbionts . However, while the explosive spread of self-splicing ancestors of spliceosomal introns as a driving force for the origin of the nuclear evolution remains to be proven, comparative genomics in fact suggest that the earliest eukaryotic ancestors accumulated numerous intron-derived non-coding DNA sequences in their genomes , leading, with few exceptions, to generally increased sizes of eukaryotic genomes over archaea and bacteria . Further support is given by a recent comparison of protein-coding and non-coding DNA ratios in numerous fully sequenced eukaryotic, as well as archaeal and bacterial, organisms. These data demonstrate that there is a linear relationship between the amount of protein-coding and non-coding DNA in archaea and bacteria, whereas the amount of non-coding DNA grows much faster in comparison with protein-coding DNA in eukaryotes . Early eukaryogenesis resulted in a remarkable accumulation of non-coding genomic DNA. Accordingly, early eukaryotes were obliged to evolve higher-order organizational principles to package increasing amounts of DNA into the nucleus and to functionally discriminate noncoding and protein-coding sequences on a structural level. Moreover, further levels of structural genome organization must have evolved in cells which express only a subfraction of genes encoded in their genomes, depending on their life cycle, developmental stage or functional specialization in multicellular organisms.
In the light of recent progress in understanding the tree of life and eukaryogenesis, it seems that opisthokonts (animals, choanozoa, fungi) and amoebozoa (all grouped together as Unikonta) diverged early from chromists and plants, resulting in a deep cleft between those eukaryotic supergroups and a multifurcated tree without ‘crown groups’ (compare Figure 1). Notably, some uncertainty about the position of the eukaryotic root remains – whether it is: (i) between unikonts and bikonts; (ii) inside the excavates; or (iii) between early diverging euglenozoa and excavates [20,29–32]. According to this view it is important to realize that the two large multicellular clades – animals and plants – branched early during eukaryotic evolution and were derived from a single-celled common ancestor. Owing to our lack of detailed knowledge about similarities and differences of nuclear organization in putatively early branching eukaryotes, it is uncertain which features of nuclear organization described below may be considered as an early achievement of eukaryotic evolution and which evolved later in a limited set of (multicellular) taxa.
Compartmentalization: a basic principle of nuclear organization
Nuclear organization implies compartmentalization at all levels from nuclear genomes (defined by the complete nuclear DNA sequence) to epigenomes (defined by differences of both chromatin compaction along the DNA sequence and higher-order chromatin arrangements). This extensive compartmentalization is the result of an evolutionary process but little is known to date with respect to the selective and neutral processes which led to the evolution of nuclei available for our analyses in extant species.
Co-expressed genes are not randomly distributed in mammals, but cluster along chromosomes more often than expected by chance [33–35]. Highly and weakly expressed genes cluster in separate chromosomal domains with an average size of 80–90 genes, suggesting domain-wide regulatory mechanisms . Transcriptome maps established from various tumour tissue types revealed non-random chromosomal regions of increased gene expression, which often correspond to experimentally verified tumour amplicons . In contrast with bacterial and archaeal microbes, genomic clustering of genes involved in the production of enzymes for a biochemical pathway is uncommon in most eukaryotes, but still some clustering of pathway members in eukaryotic genomes could be detected, suggesting a possible requirement of spatial proximity for co-regulation . The evolutionary origin of such clusters may have been driven in part by selection and in part by neutral co-evolution . Inappropriate interactions between adjacent chromatin domains can be prevented by two types of insulators, which either separate enhancers and promoters to block their interaction or create a barrier against heterochromatin spreading . In vertebrates CTCF (CCCTC-binding factor) has been identified as a versatile transcription factor with enhancer-blocking insulator activity .
A direct modification of DNA that does not change its primary sequence, which can be heritable and is reversible, is DNA methylation. The adding of methyl groups to DNA is commonly directed to carbon number 5 in the pyrimidine ring of cytosine. DNA methylation is catalysed by DNMTs (DNA methyltransferases) and occurs frequently at symmetric CpG motifs in mammals, and additionally at asymmetric motifs in plants (CpNpG and CpHpH). DNA methylation typically is associated with transcriptional repression, possibly by directly blocking transcription factor binding or by recruitment of HDACs (histone deacetylases) to impede the decondensation of higher-order conformations [42,43].
Above the level of the primary DNA sequence its organization with interacting proteins into a chromatin fibre provides the first level of higher-order genome arrangements. Although Walther Flemming had introduced the term ‘chromatin’ already in the late 19th century for a substance inside the cell nucleus with affinity to specific dyes , it took approximately 100 years to resolve its basic nature (reviewed in ). Flemming’s contemporary Friedrich Mischer isolated acidic ‘nuclein’ and basic ‘protamin’ from salmon sperm heads , before Albrecht Kossel purified ‘histones’ from bird erythrocyte nuclei . After the discovery of the double-helical structure of the DNA in 1953 , it took another two decades before Ada and Donald Olins, as well as Christopher L.F. Woodcock and co-workers recognized repeating subunits on spread chromatin fibres [45,49–51], biochemically characterized and named nucleosomes by Kornberg . In 1984 Timothy J. Richmond and co-workers determined the crystal structure of the nucleosome to a resolution of 7.0 Å (1 Å=0.1 nm) , and later it was resolved to 2.8 Å by Karolin Luger and co-workers .
The first level of chromatin compaction yields the 10 nm fibre. This fibre is built-up from DNA wrapped around the nucleosomes and can be further folded and compacted by protein–DNA and protein–protein interactions, including individual nucleosomes, the linker histone H1 as well as other proteins, into a 30 nm chromatin fibre and higher-ordered chromatin structures. PTMs (post-translational modifications) at the N-termini of all histone types can alter the degree of chromatin compaction. However, above the basic nucleosomal array – the 10 nm fibre – the structure of the 30 nm fibre has, to date, remained a controversial issue [55–59], and its presence in vivo has been doubted . Cartoons in many reviews of epigenetics present 10 nm and 30 nm thick chromatin fibres as examples for ‘open’ and ‘closed’ chromatin configurations. Yet it must be emphasized that the problem of the chromatin structure in vivo embedding transcriptionally active and silent genes respectively, remains unsolved.
Evolutionarily conserved features of higher-order chromatin arrangements
In vivo observations suggest that in the cell nucleus transcriptionally silent DNA sequences are organized into domains of facultative heterochromatin, whereas in many cases actively transcribed genes are found outside of such condensed chromatin . In opisthokont model organisms (opisthokonts are a group of eukaryotes, which includes animals, choanozoa and fungi), as well as in plants (e.g. mammals, Drosophila, yeasts and Arabidopsis) histone lysine acetylation, H3K4me and H3K79me are mostly associated with a supposedly ‘open’ chromatin state called euchromatin.
Actively transcribed genes have been widely considered to reside in fields of transcriptionally permissive euchromatin expanding between clusters of transcriptionally repressive facultative and constitutive heterochromatin [61,62]. This concept has been based on electron microscopic studies of nuclei in representative species from almost all eukaryotic supergroups (Figure 1). Although valued highly, even to the point of accepted textbook knowledge, the experimental validation of this concept remains questionable. EM (electron microscopy) micrographs in support of it (as shown in Figure 1) include staining protocols, which are not specific for DNA and therefore not suitable to visualize the nuclear topography of nuclear DNA, detect the interchromatin compartment, in contrast with EM approaches based on specific DNA-staining procedures [63,64]. Furthermore, euchromatin arguably reveals an ‘open’ configuration in contrast with the ‘closed’ configuration of heterochromatin.
The function of heterochromatin formation is multifaceted [60,65]. This ‘repressed’ chromatin state is important for nuclear processes such as chromosome condensation during mitosis or X chromosome inactivation in mammals, which involves H3K27me3. Telomeric and pericentric heterochromatin are mostly associated with low levels of acetylated histones and increased levels of H3K9me3. These particular histone modifications, H3K9me3 as well as H3K27me3, act as modules for binding of effector proteins, such as HP1 (heterochromatin protein 1)-like chromodomain proteins. HP1-family members play a key role in the formation of repressed chromatin. Such interactions have been described in various eukaryotic organisms, such as mammals, Drosophila and Caenorhabditis elegans (H3K9me3/HP1; H3K27me3/polycomb protein Pc), fission yeast (H3K9me3/Swi6), and in ciliates like Tetrahymena or Stylonychia (H3K9me3 and/or H3K27me3/Pdd1p) [66–69].
The suggestion that chromatin in the interphase cell nucleus is globally organized into repressive heterochromatin domains separated from (transcriptionally) permissive euchromatin domains is supported by both light and electron microscopic evidence in nuclei of representative species from almost all eukaryotic supergroups (Figure 1). Independent experimental support for a domain organization of CTs (chromosome territories) and a spatial genome segregation into open and closed chromatin compartments has been obtained by elegant molecular biological approaches [6,70]. The conservation of this domain organization in evolutionarily distant eukaryotic species argues for a common evolutionary origin.
CTs have become generally accepted as an evolutionarily conserved feature of higher-order chromatin organization  (Figure 2). If we neglect the possibility that the existence of CTs in distantly related taxa, such as mammals and plants, is a matter of convergent evolution, we may speculate that at least the common ancestor of unikonts and plants possessed a CT-type of chromosomal higher-order organization, suggesting that this principle of nuclear architecture shares a deep eukaryotic ancestry. The dynamic structure and topography of chromatin fibres, loops and domains, which constitute a given CT and its direct interactions with other CTs, is still not well understood and the extent of possible differences of CT organization between different cell types and species is not known. CTs in nuclei of Arabidopsis thaliana have been reported to adopt a spatial organization, which deviates from mammalian CT organization insofar that these CTs consist of heterochromatic-condensed chromocentres, from which euchromatin loops emanate . Despite general agreement that a hierarchy of chromatin loop structures exists , the actual numbers and sizes of loops made up from 10 and 30 nm thick chromatin fibres, as well as fibres of a still higher-order, are still not known. One example studied in detail revealed a mich higher compaction level . The possibility of giant loops expanding more than 2 μm away from a given CT and even across the nuclear space has been suggested but not experimentally confirmed to date.
Current models of nuclear architecture
The severe limitation of present knowledge about common principles, as well as cell-type- and species-specific differences of higher-order chromatin organization in eukaryotes outlined above, is reflected by profound differences of current models. While all current models support a territorial organization of chromosomes, they show profound differences with regard to the size distribution and compaction levels of chromatin fibres and loops, and how such a distribution may change during differentiation and environmental challenges.
The ICN (interchromatin network) model argues for intermingled chromatin loops both within and at the borders of CTs . Other models propose fields of euchromatin expanding between fields of facultative and constitutive heterochromatin. Euchromatic fields are arguably filled with chromatin loops, which expand from facultative heterochromatin . Some cartoons of this organization seem to argue for euchromatic fields mainly expanding between CTs (for examples see Figure 2 in , as well as ). Yet, such fields may penetrate into the interior of CTs as well, since transcriptionally active genes can locate either outside or at the edge, or in the interior of their CT [62,75,76]. The chromosome territory-interchromatin compartment (CT-IC) model predicts a much more striking compartmentalization of chromatin. It assumes that chromatin is basically organized within foci-like chromatin domains with a DNA content of ∼1 Mbp. These domains were first visualized as replication foci (see below), but later shown to be persistent higher-order chromatin arrangements present at any state of interphase, as well as in post-mitotic cell nuclei [77,78]. The structure of ∼1 Mbp chromatin domains has not been elucidated to date, but it has been proposed that they are built-up by rosettes of chromatin-loop domains with sizes of ∼100 kb. In contrast with models emphasizing the intermingling of chromatin fibres/loops, the CT-IC model predicts an IC largely devoid of DNA . It harbours splicing speckles and nuclear bodies and pervades the nucleus, in particular its ‘euchromatic’ regions, between and within CTs . The IC is separated from the compacted interior of ∼1 Mbp chromatin domains by a zone of decondensed chromatin with a diameter of approx. 100–200 nm, called the PR (perichromatin region) . State-of-the-art light and electron microscopic methods provide evidence that the PR serves as the functional subcompartment for transcription, co-transcriptional splicing, DNA replication and possibly also DNA repair . The CT-IC model predicts that the PR represents the major transcriptionally permissive ‘euchromatic’ compartment of the cell nucleus. Consequently, the fraction of euchromatin present in the nucleus at any given time point may be much smaller than in models, which argue for extended fields of decondensed chromatin loops expanding between clusters of compact heterochromatin [3,64].
This difference, however, could be rather a difference of degree than of a principally different higher-order chromatin organization. As a possible scenario, let us consider the IC of a cell nucleus with low overall transcriptional activity. During the transition of such a cell towards a high activity, chromatin loops might expand into the interior IC. In this way an IC devoid of chromatin loops might be transformed into a euchromatic field. This consideration shows that divergent predictions of current models may reflect different functional states. Presently, evidence for such a transformation is lacking. In order to serve scientific progress it is necessary to formulate model predictions in a way that the model can be falsified, at least for a cell type under experimental scrutiny. The ICD (interchromosome domain) model provides a case in point. It argued for an interchromatin space expanding between CTs and for the positioning of transcriptionally permissive genes at CT surfaces [80,81]. Later evidence showed that this view could not be sustained and led to the formulation of the CT-IC model [82,83]. Notably, methods available during the 1990s [FISH (fluorescence in situ hybridization) combined with epifluorescence or laser-scanning microscopy] were sufficient to test the predictions of the ICD model, whereas compelling tests of the CT-IC model are only now becoming feasible (see below).
The contribution of a nuclear matrix to nuclear organization in vivo is still a matter of conflicting opinions [84–87]. Since biochemical nuclear matrix preparations contain numerous proteins involved in major nuclear functions , the CT-IC model is compatible with the presence of a dynamic three-dimensional network of nuclear matrix core filaments pervading the IC/PR.
Evidence for gene interactions in trans has led to the concept that the respective CTs move together to interact directly , and/or that such genes may be located on giant loops, which expand from the surface of different CTs and allow the co-localization of such genes in specialized transcription factories or expression hubs at remote nuclear sites . Furthermore, the fraction of the genome present in 10 and 30 nm thick fibres and loops in a nucleus predicted by the CT-IC model is much smaller than assumed by competing models, such as the chromatin lattice model  and the ICN model . According to the CT-IC model, such fibres should be largely restricted to the PR with sizes restricted by the estimated thickness of this region in the order of approx. 100–200 nm, although occasionally fibres extending parallel to chromatin domain surfaces may exceed such a size. In both the chromatin lattice model and the ICN model, 10 or 30 nm thick chromatin fibres and loops are apparently predominant and intermingle to an extent that precludes the notion of an IC and PR as topographically and functionally distinct nuclear compartments. According to these models, the interchromatin space may be defined as the space between these fibres and loops. Some authors have argued for the presence of giant loops expanding over micrometre distances throughout the nuclear space . While substantial evidence for gene kissing events in cis and in trans has been provided by chromosome conformation capture experiments [6,90], it remains to be seen whether such events depend on the direct spatial interaction of the respective CTs or on the formation of giant chromatin loops or on both mechanisms acting in concert. Different opinions about this matter reflect different views with regard to the contribution of giant chromatin fibres and loops. Although three-dimensional FISH experiments demonstrated the occasional looping out of chromatin from CTs [75,92] or the co-localizaton of genes in cis and in trans [93,94], the contribution of giant loops to a gene kissing event was never directly depicted by microscopic observation. Notably, some previously reported kissing events have been challenged by later studies [95,96].
According to the CT-IC model, the cell nucleus may be fittingly compared with an urban landscape [9,97]. The city wall and gateways reflect the nuclear envelope and its pores, city quarters are represented by CTs, and houses and domiciles by chromatin domains and interspersed genes. The nuclear city is pervaded by a complex road system with places, streets and alleys, called the interchromatin compartment with nuclear bodies and splicing speckles. Streets are bordered by sidewalks, termed the PR, which represents the compartment for important functional interactions, such as transcription, co-transcriptional splicing, chromatin replication and possibly also repair processes . Although the structure of a city may change during history, maps typically still reveal a consistent topography of city quarters over centuries. Yet, the nuclear landscape tells a more complex story than any urban landscape. First, in contrast with the quarters and houses of a city, whose topography can be depicted by a two-dimensional map, a three-dimensional map is required to describe the topography of CTs and chromatin domains. Secondly, we would not expect to wake up in our home city and detect that houses and even entire city quarters have dramatically changed their neighbourhood overnight. In nuclei of both terminally differentiated cells and cycling cells the three-dimensional topography of chromatin may be stably maintained over long periods of time. This stability, however, is interrupted in cycling cells by mitotic events. During prometaphase, major changes of chromosome arrangements take place, yielding CT neighbourhood arrangements in daughter nuclei which differ decisively from the mother nucleus . Occasionally, periods of apparent chromatin order stability can be interrupted by periods of a surprisingly dynamic behaviour of chromatin even during interphase , as well as during post-mitotic terminal differentiation of some cell types .
Evolutionarily conserved topography of DNA replication
Nuclear processes such as DNA replication or transcription occur at specific nuclear sites , possibly within PRs at borders between chromatin domains and the IC [99,100]. The incorporation of halogenated nucleotides during S-phase allowed the visualization of these chromatin domains as so-called replication foci  in several evolutionarily distant animal species and cell types, as well as in a few single-celled eukaryotes (Figure 3), including cells from mammals, birds, Hydra vulgaris and the ciliate Stylonychia lemnae [102–105]. In nuclei of animal species, five patterns that are characteristic for successive stages of S-phase were distinguished , whereby pattern 1 is characterized by early replicating foci scattered throughout the interior of the nucleus, specifically in euchromatin regions, whereas the peripheral heterochromatin and some regions of the internal nucleoplasm are devoid of replication sites at this stage. Sites of replication are localized adjacent to the nuclear periphery in pattern 2, with fewer interior sites. In pattern 3, sites of replication are restricted to the nuclear periphery and to perinucleolar regions. Pattern 4 is characterized by sites of DNA replication that have become larger in size and fewer in number. Sites are distributed throughout the nuclear interior with only a few discrete sites at the nuclear periphery. Pattern 5 is noted towards the end of S-phase. At this stage smaller foci are associated with peripheral heterochromatin and fewer, larger foci with heterochromatin located in the nuclear interior. As already mentioned above, replication foci labelled during S-phase persist as ∼1 Mbp chromatin domains during other stages of the cell cycle and in subsequent cell generations, as well as in post-mitotic cells . They may thus represent fundamental structures of higher-order chromatin organization in evolutionarily distant species.
It has been demonstrated in the ciliate species S. lemnae that replication patterns of both the micronuclei and macronuclei exhibit evolutionarily conserved similarities, but also differences, when compared with such patterns in animals . Replication in the heterochromatic micronucleus occurs in foci-like structures showing spatial and temporal patterns similar to nuclei of higher eukaryotes, suggesting that these patterns are inherent features of nuclear architecture. The numerous gene-sized ‘nanochromosomes’ of the macronucleus are replicated in a propagating structure, called a replication band, which consists of hundreds of synchronously firing replication foci, very similar in size and shape to replication foci described in animals.
Besides the visualization of replication foci by incorporating halogenated nucleotides, similar replication patterns have been shown to occur in vivo using DNA labelling with fluorochrome-conjugated nucleotides  or PCNA (proliferating cell nuclear antigen), a central component of the replication machinery, fused with GFP (green fluorescent protein) . However, what begets the formation of distinct replication patterns in space and time is still to date a matter of speculation.
Species-specific peculiarities of chromatin organization
Interestingly some phyla have evolved remarkable deviations in their nuclear organization. The unicellular ciliates, for example, contain two different types of nuclei, one or more macronuclei and one or more diploid micronuclei. All transcripts required for vegetative growth are derived from DNA of the macronucleus, which lacks markers typical for heterochromatin. In contrast, the genome of the micronucleus is organized into heterochromatin and is transcriptionally inert [69,109,110].
The chromatin of dinoflagellates is apparently not organized into nucleosomal repeats. The protein fraction of their chromatin contains several basic histone-like proteins, which exhibit some similarities with both bacterial histone-like proteins and eukaryotic histone H1 . As a consequence the global chromatin architecture in dinoflagellate (dinokaryotic) nuclei with chromosomes in a screw-like configuration deviates from other eukaryotes  (Figure 1). Unlike most dinoflagellates, Peridinium balticum contains two nuclei. One of these nuclei was derived from engulfment of an alga (nucleomorph). Interestingly, nuclease digestion experiments suggest that this nucleomorph has preserved the organization of its DNA into nucleosomes, while the chromatin of the dinokaryotic nucleus possesses no nucleosomes . Several examples for endosymbiotic nucleomorphs, which derive from engulfment of another eukaryotic species, have been reported for other taxa besides dinoflagellates, such as the cryptophyceae (Hemiselmis andersenii as well as Guillardia theta contain nucleomorphs of red algal origin).
Evolutionary response of nuclear architecture to environmental changes
Ample evidence demonstrates non-random radial chromatin arrangements at least in somatic cell nuclei of mammals [114,115] and birds [103,116]. Gene-dense chromatin is typically packed more interior, whereas gene-poor chromatin is preferentially arranged at the nuclear periphery and around nucleoli . A recent study, however, provided a remarkable example that the evolution of higher-order nuclear organization can be responsive to environmental triggers in quite unexpected ways. This study demonstrated a fundamental rearrangement of global nuclear architecture during terminal differentiation of rod photoreceptor cells of nocturnal mammals . These nuclei exhibit an inverted pattern, in contrast with rod cell nuclei of diurnal mammals (Figure 4), i.e. heterochromatin localizes to the nuclear interior, whereas euchromatin, as well as newly synthesized RNA and the splicing machinery, are localized at the nuclear periphery. This inverted pattern is adopted by remodelling of the conventional pattern during terminal differentiation of rod photoreceptor cells. The inverted rod nuclei have been shown to act as collecting lenses, and computer simulations confirmed that columns of such nuclei are able to channel light efficiently towards the light- sensitive rod outer segments (Figure 4). This finding demonstrates a fascinating connection between an environmental trigger, the limited availability of light for nocturnal animals, and a reconfiguration of the higher-order nuclear organization to the selective pressure of survival under conditions where photons to see an enemy or detect a prey are scarce.
Plasticity of higher-order nuclear organization, senescence and human disease
In the context of human disease, clustering of genes, although essential for normal nuclear functions, may also become a source of dysregulation of gene expression patterns across genomic neighbourhoods . There is emerging evidence that the order of genes on chromosomes is non-random and that genes with co-ordinated expression profiles show a tendency to cluster-forming regulatory networks [35,39]. It is assumed that interdependent genomic loci, even if they are located on different chromosome territories, can spatially interact [6,93,94,119]. Furthermore, it has been argued that intermingling of chromatin loops between neighbouring CTs may provide opportunities for rearrangements between non-homologous chromosomes [74,120]. Cell-type-specific non-random CT neighbourhood arrangements could then lead to preferred rearrangements between neighbouring CTs, for example during the formation of tumour cells characterized by specific chromosome translocations . Evidence in favour of such a scenario is, however, limited and one needs to take into account positive selection of such cells due to a growth advantage.
A number of senescence-related changes of the nuclear architectural landscape have been described, such as aneuploidy, changes of CT positioning, as well as the relocation of heterochromatin and the formation of senescence-associated chromatin foci (for review see ). Such changes of the nuclear architecture may possibly play a role in the transcriptional repression of proliferation-associated genes. It has been reported that human lymphocyte nuclei of older people tend to contain increased amounts of heterochromatin. Concomitantly the transcriptional activity of these cells is reduced. Interestingly, a radial redistribution of CTs has been observed in non-proliferating senescent cell nuclei . While CTs in interphase nuclei of proliferating cells tend to be radially distributed in a gene-density-dependent manner, a CT-size-dependent redistribution has been observed in non-proliferating senescent cell nuclei.
Mutations in the LMNA (lamin A/C) gene, which encodes lamin A protein isoforms involved in the formation of the lamina, a matrix of intermediate filaments at the inner nuclear membrane, can cause premature aging disorders such as HGPS (Hutchinson–Gilford progeria) or atypical Werner’s syndrome, as well as other severe diseases now known as laminopathies . Nuclear lamins are involved in higher-order chromatin organization owing to the specific attachment of chromatin-bearing lamin receptors. Mutations in lamin A are correlated with a dramatic reorganization of nuclear architecture, including changes in nuclear shape and loss of peripheral heterochromatin in a progressive manner . Finally, pathological changes of higher-order chromatin arrangements and gene positions may contribute to cancer formation [5,124].
We are just beginning to explore the impact of nuclear architecture on regulatory networks of gene expression and other nuclear functions, such as DNA replication and repair. It has also become obvious that disorders of nuclear architecture lead to severe diseases. A reliable description of nuclear landscapes in different cell types and evolutionarily distant species is necessary to establish a firm fundament for further hypothesis-driven studies, including the molecular mechanism(s) involved in long-range chromatin movements.
The development of light optical nanoscopy offers the possibility to surpass the resolution limit of conventional fluorescence microscopy down to the nanometre scale. Correlative microscopy of a given nucleus, first by four-dimensional (space and time) live-cell microscopy followed after fixation by three-dimensional imaging combining light optical nanoscopy with the still superior resolution of electron microscopy, offers exciting new prospects to visualize cell-type- and species-specific differences of the nuclear architecture in much more detail . Compared with EM, light optical nanoscopy bears the promise to resolve the topography of different nuclear components labelled with different fluorophores with unprecedented resolution, including the possibility to map individual proteins, RNA molecules or DNA target regions . It may even be expected that some of the current light optical nanoscopic approaches can soon be employed for studies of living cells. These revolutionary developments in microscopy should allow decisive tests of divergent predictions of current models of nuclear architecture, such as the contribution of giant chromatin loops to gene kissing events in trans, or the existence or not of a structurally and functionally distinct IC and PR as compared with euchromatic fields of intermingling chromatin fibres. It should even become possible for the first time to visualize the chromatin compaction of individual active and silent genes. The recent combination of circular chromosome conformation capture with massively parallel sequencing (Hi-C) has allowed for the first time mapping of DNA–DNA interactions in cis and in trans at a genome-wide level . Such genome-wide mapping studies open up fascinating perspectives, but require large cell numbers. In order to generate a DNA–DNA interaction map for a given cell type, one needs first to obtain a pure cell population. This can become an exceedingly difficult task, when we consider the complex composition of tissues. Furthermore, even the complete mapping of relevant DNA–DNA interactions in a pure cell population does not substitute for microscopic observations able to visualize the topography of all nuclear components and their dynamics in living cells.
The problem of nuclear organization and function is a fitting example that big biological problems require investigations with all methods available at all accessible levels, from molecules to nuclear machineries, from genes to CTs, and so on. Generating reliable maps, be it for stars, new continents, the genome or now the epigenome, is paradigmatic for the necessity of quantitative descriptions, which depend on the development of more and more refined instruments for the localization of objects of interest with high-precision, as well as on distance measurements between several objects. Unexpected phenomena discovered in explorative, descriptive studies can open up new avenues for hypothesis-driven research. The story of genome mapping and DNA sequencing, as well as comparing the genomes from more and more specias has been full of major unexpected discoveries. An evolutionary approach towards the three-dimensional and four-dimensional mapping of epigenomes bears the promise to lead us to new exciting discoveries as well.
• The cell nucleus is thought to have evolved from the ER very early during eukaryogenesis rather than being a derivative of a putative endosymbiont.
• Common principles of hierarchical chromatin structure organization, such as DNA methylation and N-terminal post-translational modifications of histones are conserved between distantly related eukaryotes.
• The occurrence of conserved components of nuclear higher-order organization in distantly related species (i.e. CTs, non-random radial chromatin arrangements, spatiotemporal co-ordinated replication foci) suggests that such common principles of genome architecture are primarily inherent to all eukaryotes.
• Species-specific peculiarities, such as nuclear dualism in ciliates, the nucleosome independent chromatin organization of dinoflagellates or the occurrence of endosymbiont-derived nucleomorphs (e.g. in cryptophyceae) appear to be derived – not ancestral – features of chromatin organization.
• Nuclear architecture underwent major adaptive changes as a result of environmental triggers.
• Disorders of nuclear higher-order organization potentially contribute to various diseases and to senescence-related changes.
Research in the Cremer laboratory has been funded by the Deutsche Forschungsgemeinschaft (DFG), the Munich Center of Integrated Protein Science (CIPSM) and the Human Science Frontier Program (HSFP); research in the Lipps laboratory has been funded by the Deutsche Forschungsgemeinschaft (DFG) and the Center for Biomedical Education and Research (ZBAF) of the University of Witten/Herdecke. Jan Postberg was funded by the Peter and Traudi Engelhorn foundation. We thank our collegue Dr Irina Solovei for the contribution of Figures 2(D) and 4.
- © The Authors Journal compilation © 2010 Biochemical Society