生物谷报道：有个常识大家都知道，真核细胞内的染色质是分布不均匀的，呈异质性。但是为什么会不均匀？它有生理学意义吗？是否与染色体位置上某些基因的活化或表达相关呢？目前仍是谜，这篇发表在Cell上的小综述认为，染色体的位置有可能与基因的活化有关，并且认为这将成为基因组学下一个研究重点和热点，同时这也提示，基因组的研究不仅仅是功能和结构，而且应该包括位置信息，也即基因组是三维的，只有完整的三维信息才能真正算是基因组学的研究。
The interior of the eukaryotic cell nucleus is structurally and functionally complex. The nuclear volume contains morphologically distinct higher-order chromatin domains, such as condensed heterochromatin, and numerous membraneless proteinaceous subcompartments, including the nucleolus and a multitude of small nuclear bodies. The physically distinct nature of each compartment not only contributes to spatially partition the nucleus but also creates distinct functional subdomains within the nucleus. The degree of structural heterogeneity is probably linked to functional complexity since higher eukaryotes with more genes generally exhibit a larger diversity of compartments than simple eukaryotes. It is likely that nuclear compartmentalization contributes to some degree to genome function in all eukaryotes.
An additional layer of spatial complexity is generated by the nonrandom spatial organization of the genome itself. In higher eukaryotes, the distribution of the genetic material of each chromosome is limited to a spatially defined nuclear subvolume in the form of chromosome territories. Chromosomes themselves are nonrandomly arranged within the nuclear space and occupy preferential positions relative to the center of the nucleus and relative to each other (reviewed in Parada and Misteli [2002]). As a consequence of the nonhomogenous nature of the nuclear space created by compartments and the nonrandom arrangement of genomes, gene loci may experience distinct local environments. Thus, the position relative to nuclear landmarks, particularly the nuclear envelope, chromatin domains, and the various proteinaceous nuclear compartments, is a fundamental property of every gene. The functional significance of spatial positioning, however, is only poorly understood.
The Nuclear Periphery: From Repression to Activation
Arguably the most prominent spatial feature of the cell nucleus is its periphery. The edge of the nuclear volume abutting the nuclear envelope has commonly been considered a zone of transcriptional repression both in yeast and in higher organisms. Part of this notion likely stems from the presence of extensive blocks of condensed, presumably transcriptionally repressed heterochromatin at the periphery of mammalian nuclei. Consistent with this notion, in human lymphocytes and fibroblasts, gene-poor chromosomes tend to be positioned preferentially toward the nuclear edge, and examples of repositioning of genes from the periphery toward the interior of the nucleus upon their activation have been reported (reviewed in Kosak and Groudine [2004]). Molecular evidence for the transcriptionally repressive nature of the periphery has come from studies in S. cerevisiae in which silenced telomeres form clusters juxtaposed to the nuclear envelope, silencing factors accumulate to form peripheral silencing compartments, and tethering of a silencing-deficient reporter to the nuclear rim facilitates its repression (reviewed in Gasser [2001]). These findings make it clear that the nuclear periphery can act as a transcriptionally repressive environment. However, recent observations now suggest that this region is not merely a silencing milieu but plays a much more complex and subtle role in gene regulation.
A qualitative argument against an exclusively repressive influence of the nuclear periphery is the simple observation that, in mammalian cells, visualization of global transcription does not reveal an underrepresentation of active transcription sites at the periphery, neither does one observe enrichment of active sites in the nuclear interior. More concrete evidence for additional roles of the nuclear periphery in gene regulation is the finding that boundary activity is linked to nuclear pore complexes (NPC) (Ishii et al., 2002). Using an unbiased genetic screen, Laemmli and colleagues identified several export factors and NPC components as strong boundary activities, i.e., these factors are essential for activation of a reporter gene by isolating it from a silent chromatin environment (Ishii et al., 2002). Boundary activity of export and NPC factors required their interaction with the nuclear rim and tethering of the reporter to the nuclear periphery, linking boundary function to peripheral positioning (Ishii et al., 2002). A role of the peripheral nuclear zone in boundary activity is also supported by the observation in Drosophila that the gypsy insulator and its binding proteins preferentially localize to the periphery (Gerasimova et al., 2000). More importantly, in strains lacking insulator activity, the gypsy element dissociates from the periphery and assumes a more internal position (Gerasimova et al., 2000).
An elegant study further extends the role of the nuclear periphery in gene regulation. Casolari et al. demonstrate coupling between nuclear architecture and gene activity, rather than silencing, and they show that spatial positioning is a functionally highly relevant, physiological, and global phenomenon (Casolari et al., 2004). In a genome-wide analysis, several NPC components, including import/export factors and pore-associated structural myosin-like proteins, were found to specifically bind to transcriptionally active genes in addition to silenced genes. The functional relevance of this activation-linked association is demonstrated by the fact that different subsets of NPC proteins selectively bind to distinct groups of genes. For example, the myosin-like pore proteins Mlp1 and Mlp2 associate with highly expressed genes such as ones involved in glycolysis and ribosome biogenesis. These differential interactions play an important role in physiological responses since, upon transcriptional challenge by switching from growth in glucose to galactose, nuclear pore components become rapidly bound to key genes of the galactose metabolism pathway. Similar to the situation for boundary activity, the observed biochemical interaction of pore components with genes and their activation is paralleled by a physical relocalization of the GAL gene cluster from an internal to a peripheral position, indicating a direct functional link between spatial position of these genes and their activity (Casolari et al., 2004). The presence of both repressed and activated genes and the observed movement of loci toward the periphery upon activation leave little doubt that the periphery of the yeast nucleus is not simply a transcriptionally repressive zone but rather a complex general gene regulation environment. The fact that both negative regulation, as for telomeres, as well as positive regulation can take place simultaneously in close proximity suggests that the involved control mechanisms act locally, possibly at the single gene level (Figure 1).
(13K)
Figure 1. Spatial Separation of Activation and Repression at the Nuclear PeripheryDifferential regulation at the nuclear periphery may be achieved by (left) formation of chromatin loops containing active loci (green). The loops are anchored via NPC components (yellow) and are readily accessible to transcriptional activators (blue). The repressed regions (black) are located in condensed peripheral heterochromatin. (Right) Alternatively, the periphery may contain distinct activating and repressive domains, which are enriched in activators and repressors, respectively.
The most likely mechanism for the simultaneous positive and negative regulation at the periphery is the differential association and looping of chromatin regions. Inactive regions might be associated with peripheral heterochromatin, whereas active genes are anchored to the NPC and looped out for ready access by transcriptional regulators in the nucleoplasm (Figure 1, left panel). Alternatively, the nuclear periphery may contain dedicated activation and inactivation centers in which repressors and activators, respectively, are concentrated and to which differentially regulated loci are tethered ( Figure 1, right panel). In either case, a likely additional function of NPCs is to define transcriptionally active and inactive regions by preventing the spreading of functionally distinct chromatin domains.
It is currently unclear how functionally similar the nuclear periphery is in yeast and mammals and how accurately one can extrapolate the findings in yeast. However, a recent study in human cells is consistent with the notion that the mammalian nuclear periphery can be a site of concurrent gene activity and repression and that regulation can occur locally, just as in S. cerevisiae. The cystic fibrosis transduction receptor (CFTR) gene and its two immediate neighbors, GASZ and CORTB2, are differentially expressed among cell types, and, remarkably, their association properties with the periphery are independent of each other (Zink et al., 2004). For example, in Calu-3 cells, the active CFTR gene is displaced a few hundred nanometers away from the extreme periphery, but the inactive GASZ and CORTBP2 genes remain at the very edge (Zink et al., 2004). This localization of neighboring mammalian genes of distinct transcriptional status at the nuclear periphery is reminiscent of the NPC-mediated boundary activity in S. cerevisiae.
Positioning Within: The Localization-Function Interplay
Localization of a gene to the most peripheral layer of the nucleus represents an extreme case of positioning. How important is the position of a gene within the nuclear volume in general? Spatial mapping of several loci indicates that many mammalian genes occupy preferential nonrandom positions relative to the nuclear center (Roix et al., 2003). Although changes in radial positioning of genes have been correlated with their activity (Kosak et al. 2002 and Kim et al. 2004), it seems unlikely that the radial position directly affects a gene's activity since, regardless of its functional status, a gene can be found at all possible spatial positions within a cell population. While it is feasible that the radial position contributes to gene function in a subtle manner, possibly reflected in the observed stochastic behavior of gene activity in vivo, it seems more probable that the preferential radial position of a locus is primarily a reflection of the nonrandom position of the chromosome on which it resides (Roix et al., 2003). Of much more functional significance appears to be the relative spatial positioning of a gene with respect to internal nuclear compartments and chromatin domains. A strong correlation between positioning near constitutive heterochromatin and gene inactivation has been observed for numerous genes, including key regulators of B and T cell differentiation (reviewed in Kosak and Groudine [2004]). The preferential spatial positioning of these loci is likely functionally and physiologically significant since it correlates with their expression profiles during the differentiation process.
Gene loci may also be nonrandomly positioned relative to proteinaceous nuclear landmarks. Actively transcribed ribosomal genes are invariably associated with the nucleolus, PML bodies are preferentially found near transcriptionally highly active genome regions, Cajal bodies have a propensity to colocalize with histone and U2 snRNA genes, and R bands of human chromosomes have recently been found to preferentially localize in close spatial proximity to nuclear compartments enriched in pre-mRNA splicing factors (Shopland et al. 2003 and Wang et al. 2004). The high frequency of these associations strongly suggests that the relative positioning is functionally relevant. The spatial proximity of splicing factor compartments to gene-rich R bands, for instance, likely facilitates the supply of pre-mRNA splicing factors to nascent RNAs synthesized from the proximal genes and thus contributes to the efficiency of processing (Shopland et al., 2003).
The central question in understanding the functional role of positioning is whether nonrandom positioning is the cause or consequence of gene function. The association of a gene locus with a particular nuclear landmark may be strictly a reflection of the gene's functional status without having regulatory relevance. Alternatively, positioning may precede changes in gene activity and might thus be a prerequisite for proper function and even serve a regulatory role. Recent results point to a model in which the functional potential of a locus facilitates its association with a functional compartment, which in turn influences the functional properties of the locus (Figure 2).
(7K)
Figure 2. The Sequence of Events Leading to Spatial PositioningThe position of a gene locus relative to a nuclear landmark (chromatin domain or compartment; orange) may be the result of a change in the functional potential of a locus (green and pink) followed by its repositioning via constrained local diffusion and association with a functionally equivalent domain. Association with the domain may reinforce or modify its functional state (red).
Detailed temporal and spatial analysis of the inactivation of the terminal transferase Dntt locus during mouse T cell differentiation supports this view (Su et al., 2004). Upon T cell stimulation, transcription from Dntt ceases, and acetylation of histone H3 on lysine 9 is lost and is gradually replaced with methylation, a mark of heterochromatin. These changes in histone modifications are paralleled by the association of the locus with heterochromatin blocks. Repositioning seems to be driven by the change in the functional status of Dntt, as it only occurs after the block in transcription. However, stable repositioning into heterochromatin occurs prior to full H3-K9 methylation, suggesting that the inactive locus is more prone to associate with a silencing region and that this region in turn contributes to the establishment of the permanent silent sate (Su et al., 2004). Similarly, a body of work on -globin suggests that positioning relative to centromeric heterochromatin domains in erythroid cells does not correlate with the locus' transcriptional activity per se but rather with its "poised" state, defined by hyperacetylation of histones (reviewed in Kosak and Groudine [2004]). These behaviors are comparable to that of the brown locus in Drosophila, which has served as a classic example for positioning effects. In this case, insertion of a heterochromatic block near brown results in the positioning of the locus near centromeric heterochromatin regions and its subsequent silencing. The heterochromatin self-association occurs independently of sequence homology between the interacting heterochromatin regions, suggesting that the initial repositioning is driven by the similarity in the functional potential of the heterochromatin regions (Sage and Csink, 2003).
The same principle appears to apply to gene activation. Osborne et al. have recently demonstrated that genes located more than 20Mb apart on the same chromosome in erythroid progenitors converge with high frequency on a single RNA polymerase II transcription domain (Osborne et al., 2004). This change in positioning was dependent on the transcriptional status of the genes. Importantly, inactive genes were also found at transcription sites; however, they were not stably recruited. This latter observation suggests again that the recruited loci had the potential to be transcribed, but only their association with a transcription domain resulted in their activity. These results paint a picture in which loci of a particular functional potential, either active or inactive, search their nuclear environment for a site that corresponds to their functional status. Once associated, the local environment contributes to establish and maintain the locus' function.
This scenario requires that gene loci are able to sample their nuclear environment in search of functionally equivalent regions, be it heterochromatin domains, transcription centers, or nuclear bodies. This prediction is consistent with the recent realization that genome regions have the intrinsic ability to undergo constrained diffusional motion, which allows them to explore a relatively large fraction of the nuclear volume (Vazquez et al., 2001). The motion of loci may be complemented by recruitment of nuclear bodies or their de novo formation near genome regions of a particular functional status. It thus seems most likely that positioning is the result of a largely self-organizing process involving the dynamic interplay between a gene's intrinsic potentiated state, its diffusional ability, and its physical interactions with functionally distinct subcompartments.
Spatial Genome Neighborhoods
In addition to nonrandom preferential association of gene loci relative to the periphery or to nuclear landmarks, gene loci may also be nonrandomly positioned relative to each other, forming defined spatial gene clusters. The classic example for such relative clustering is the nucleolus, the site of ribosomal gene transcription and rRNA processing. In most eukaryotes, ribosomal genes are located in tandem arrays located on several chromosomes. These chromosome regions congregate in three-dimensional space to form two to three nucleoli per nucleus, with each nucleolus containing genetic material from multiple chromosomes. The spatial clustering of ribosomal genes into a spatial neighborhood has generally been considered somewhat of an exception. However, even more extensive clustering has recently been discovered in S. cerevisiae tRNA genes (Thompson et al., 2003). More than 50 distinct tRNA genes belonging to five tRNA gene families and dispersed throughout the length of the linear genome on virtually all chromosomes are brought together in three-dimensional space to form a tRNA transcription and processing center near the nucleolus. Analogous to the situation for the nucleolus, this clustering might aid in the recruitment of transcription complexes needed to maintain the high level of gene activity required for the sustained production of tRNA and to ensure the efficient modification and processing of the highly abundant newly synthesized tRNAs (Thompson et al., 2003).
The observed clustering of ribosomal and tRNA genes allows for the general possibility that the genome is spatial, organized into distinct, nonrandom neighborhoods defined by specific sequence regions that congregate in three-dimensional space. An attractive idea is that sets of genes that are regulated by the same transcription factors, for example, during differentiation, might cluster around nuclear compartments enriched in these particular factors. Although examples of nonrandom clustering of chromosomes have been reported (reviewed in Parada and Misteli [2002]), the generality of formation of three-dimensional gene clusters is unclear, and their existence awaits experimental testing. An initial hint for the existence of clustering comes from the observation that, in erythroid progenitor cells, two coregulated genes, Hbb-b1 and Eraf, coalesce with high frequency onto a shared RNA polymerase II transcription domain, despite the fact that they are separated by more than 20 Mb of linear sequence (Osborne et al., 2004).
Spatial Positioning in Genome Stability
While much effort has gone into addressing how positioning affects gene expression, recent evidence suggests that the relative location of genome regions also has a prominent role in formation of chromosome translocations and in site-specific recombination. Both of these processes involve the physical juxtaposition and physical rejoining of genome regions, and thus they are sensitive to spatial positioning effects.
In the case of translocations, two chromosomes containing double-strand breaks (DSB) undergo illegitimate joining to form a chimeric chromosome. Recent experimental interrogation of the relative spatial proximity of translocation partners has revealed a remarkable correlation between translocation frequency and nonrandom relative positioning of the partners. The genome regions frequently translocated in promyelocytic leukemia, acute myelocytic leukemia, Burkitt's lymphoma, and thyroid lymphoma were all found to be preferentially positioned in closer spatial proximity than nontranslocating regions in the same cell, suggesting that the nonrandom physical position of the partners contributes to the frequency of their illegitimate joining (reviewed in Parada and Misteli [2002]). The physical basis for this bias is likely the fact that free chromosome ends can only undergo limited diffusional motion and become rapidly immobilized upon breakage (Lisby et al. 2003 and Aten et al. 2004). In this way, a broken chromosome end can only undergo illegitimate joining with its immediate, nonrandomly positioned neighbors. As a consequence, the nonrandom spatial arrangement of chromosomes in the interphase nucleus contributes to determining translocation frequency.
Nonrandom spatial proximity of genome regions also appears to influence the outcome of recombination events. A long-standing conundrum in understanding mating type switching in S. cerevisiae has been the observation of donor preference. Cells of mating-type MATa prefer to use the HML locus at the left arm of chromosome III as a template for site-specific recombination with the MAT donor locus, whereas MAT cells prefer HMR on the right arm of the same chromosome. How this preference arises has been puzzling. Bressan et al. have recently shown that the nuclear volume explored by HML is restricted in MATa cells compared to MAT cells and that the MAT donor locus is on average in closer spatial proximity to the HMR in MAT cells than to the HML locus (Bressan et al., 2004). This preferential proximity of HMR to the MAT locus is analogous to the nonrandom preferential spatial proximity of translocation-prone loci, and the findings in these two diverse systems suggest that relative spatial positioning of gene loci has significant consequences for recombination events.
Perspectives
It has become clear that chromosomes and genes are nonrandomly positioned within the three-dimensional space of the cell nucleus. Recent observations point toward functional roles of positioning, both in gene activity and genome stability. Many fundamental aspects regarding the mechanisms and significance of positioning remain to be uncovered. How essential is positioning for gene regulation? What are the molecular mechanisms that determine positioning of genes and chromosomes? Can positioning patterns be altered in response to physiological cues, and, if so, what are the cellular pathways involved? A major impediment in deducing general rules for how positioning affects gene function has been the limitation of most studies to the analysis of single genes. Use of high-throughput microscopy methods, in conjunction with pattern recognition tools, are required to begin to uncover the full impact spatial organization has on genome function.
相关文献：
1.Bressan, D.A., Vazquez, J. and Haber, J.E., 2004. J. Cell Biol. 164, pp. 361–371.
2.Aten, J.A., Stap, J., Krawczyk, P.M., van Oven, C.H., Hoebe, R.A., Essers, J. and Kanaar, R., 2004. Science 303, pp. 92–95
3.Casolari, J.M., Brown, C.R., Komili, S., West, J., Hieronymus, H. and Silver, P.A., 2004. Cell 117, pp. 427–439
4.Gasser, S.M., 2001. Cell 104, pp. 639–642
5.Gerasimova, T.I., Byrd, K. and Corces, V.G., 2000. Mol. Cell 6, pp. 1025–1035
6.Ishii, K., Arib, G., Lin, C., Van Houwe, G. and Laemmli, U.K., 2002. Cell 109, pp. 551–562
