|Home | About | Journals | Submit | Contact Us | Français|
The mammalian cell nucleus provides a landscape where genes are regulated through their organization and association with freely diffusing proteins and nuclear domains. In many cases, specific genes are highly dynamic, and the principles governing their movements and interchromosomal interactions are currently under intensive study. Recent investigations have implicated actin and myosin in chromatin dynamics and gene expression. Here, we discuss our current understanding of the dynamics of the interphase genome and how it impacts nuclear organization and gene activity.
The interphase nucleus is a highly compartmentalized organelle, in which chromosomes occupy discrete territories and various regulatory proteins are present in specific nuclear bodies and/or are diffusely distributed throughout the nucleoplasm (Figure 1; reviewed in Spector, 2003). Chromosome territories are arranged in a radial fashion whereby gene-rich chromosomes occupy a more central position in the nucleus and gene-poor chromosomes are present closer to the nuclear periphery (reviewed in Cremer and Cremer, 2001). Recent evidence has suggested that chromosome territories are not rigid compartments and that intermingling occurs between different territories (Branco and Pombo, 2006). This intermingling was found to be dependent upon transcriptional activation (Branco and Pombo, 2006). The interchromatin space between territories is occupied by an ever-increasing number of nuclear bodies, including but not limited to speckles or interchromatin granule clusters, Cajal bodies, promyelocytic leukemia (PML) bodies, paraspeckles, and the perinucleolar compartment (Figure 1; reviewed in Spector, 2003). Given the highly compartmentalized structure of the nucleus, much research has focused on examining the dynamic interactions between chromatin and nuclear bodies or their constituents and the impact of these interactions on nuclear structure and function.
An important underlying principle of nuclear organization relates to how the genome is organized and how this dynamic organization is reflected in gene expression (reviewed in Sproul et al., 2005). Although numerous studies have associated the nuclear periphery with silent chromatin (reviewed in Lanctot et al., 2007a), others have identified active genes associated with the nuclear periphery. For example, in the budding yeast Saccharomyces cerevisiae a number of active genes were found to be associated with nuclear pore complexes, suggesting that the periphery is not necessarily a silent compartment (reviewed in Brown and Silver, 2007). Gene activation has also been observed at the nuclear periphery in Drosophila melanogaster as well as in mammalian cells (reviewed in Brown and Silver, 2007; Lanctot et al., 2007a; Sexton et al., 2007). For example, expression of the β-globin locus is initiated at the nuclear periphery prior to its movement to a more internal nuclear position during erythroid maturation (Ragoczy et al., 2006). Interestingly, the locus control region (LCR) of the β-globin gene is required for its relocation to the nuclear interior. In addition, the LCR mediates the association of the locus with RNA polymerase II (Pol II) transcription factories.
Two recent studies have described approaches to alter the nuclear position of specific chromosomal regions in living cells and address the effect on transcriptional activity. In the first study, Kumaran and Spector (2008) used a lamin B1 fusion to target a stably integrated transgene array to the nuclear periphery. Locus targeting required passage through mitosis. Importantly, the kinetics by which transcription was induced from the targeted locus was similar to the kinetics for the nontargeted locus. In a second study (Reddy et al., 2008), using a similar targeting approach that involved an emerin fusion protein, the targeted locus under study as well as a large number of endogenous genes were repressed upon targeting to the nuclear periphery. Similar to Kumaran and Spector (2008), this study also found that targeting required passage through mitosis. Although these two studies come to different conclusions with regard to nuclear positioning and gene activity, they may be reconciled by the potential presence of different microdomains at the nuclear periphery. Furthermore, silencing of already activated genes upon targeting (Reddy et al., 2008) may involve different regulatory requirements than the ability of a repositioned silent transgene array to be transcriptionally induced at the nuclear periphery (Kumaran and Spector, 2008).
Genes have also been found to loop out of their chromosomal territories upon transcriptional activation (reviewed in Fraser and Bickmore, 2007). In mouse erythroid progenitor cells a set of actively transcribed genes were found to loop out into transcription factories (Osborne et al., 2004). Genes that were either distally located on the same chromosome or on different chromosomes were sometimes found to share the same factory (Osborne et al., 2007).
Distinct antigen specificity in B and T cells is attained via monoallelic rearrangements (reviewed in Spicuglia et al., 2006). Long-range chromosomal interactions between regions that are 2–3 Mb apart have been implicated in facilitating chromosomal recombination (reviewed in Spicuglia et al., 2006). Association of the variable (V) genes of the immunoglobulin loci with the diversity-joining constant [(D)JC] gene domain can occur via looping that aids in the recombination process (reviewed in Spicuglia et al., 2006). A similar mechanism facilitating recombination between T cell receptor β and T cell receptor α has been observed in thymocytes (Skok et al., 2007). Interestingly, once the recombination event has occurred the association between the two loci is no longer observed in the subsequent developmental stages (Skok et al., 2007). It is hypothesized that an interaction between the loci is prevented by the sequestration of one of the alleles to a pericentric heterochromatin compartment (Skok et al., 2007).
Not only has movement of genetic loci been observed with respect to their chromosome territory, the nuclear periphery, or a nuclear body, but specific interactions have also been observed between loci present on different chromosomes. Using the chromosome conformation capture (3C) technique, interchromosomal interactions between the regulatory regions of the TH2 cytokine locus and the promoter of the IFNγ gene have been observed in naive T cells but not in differentiated effector cells (Spilianakis et al., 2005). Abrogation of the interchromatin interaction between the IFNγ gene and the TH2 LCR was accompanied by significant changes but not complete termination of transcription. This suggests that interchromatin interactions are not the sole determinant of gene expression. Instead, such interactions may act by increasing the efficiency of transcription. In addition, fluorescence in situ hybridization (FISH) analysis for the IFNγ gene and the TH2 LCR showed that interactions occurred for only one of the two alleles in about 40% of the cells tested, and associations between both alleles were never observed above background (Spilianakis et al., 2005). It would be interesting to determine whether the two alleles in the same nucleus have different transcriptional profiles based on their association with the TH2 LCR, and if there is a difference in the transcriptional status of the IFNγ gene in cells in which no interchromatin associations are observed.
An interesting example of long-distance chromatin interactions has been identified in regard to the regulation of olfactory receptor genes. Thirteen hundred genes comprise the olfactory receptor family and only one olfactory receptor is expressed in a particular neuron; expression is monoallelic (reviewed in Shykind, 2005). It has been suggested that the choice of the specific olfactory receptor gene to be expressed in a given neuron is mediated by interactions between the gene and an enhancer sequence referred to as the H element. The H element was discovered as a cis-acting regulatory sequence ~75 kb upstream of a cluster of olfactory receptor genes (Serizawa et al., 2003). More recently, it was demonstrated that the H element on chromosome 14 can also associate with promoters of olfactory receptor genes present on different chromosomes (Lomvardas et al., 2006). Dual RNA and DNA FISH revealed that the transcriptionally active olfactory receptor gene was associated with the H element, implicating this element in the choice of which gene is activated in trans in a given neuron (Lomvardas et al., 2006). In a related study, deletion of the H element by gene targeting in mice resulted in the loss of expression of members of an olfactory receptor gene cluster as a function of distance (Fuss et al., 2007). Interestingly, in marked contrast to the previous study (Lomvardas et al., 2006), no effect was observed on the transcription of olfactory receptor genes located on different chromosomes (Fuss et al., 2007). Therefore, further analysis is necessary to integrate the findings of these two intriguing studies.
Evidence that interchromatin interactions may be mediated by specific proteins comes from studies performed on the imprinted locus Igf2/H19. In fact, the first interchromsomal interaction was observed between homologous chromosomes harboring the imprinted genes Igf2/H19 (LaSalle and Lalande, 1996). More recently, interchromosomal interactions between the imprinting control region (ICR) of the Igf2/H19 genes on chromosome 7 and the Wsb1/Nsf1 locus on chromosome 11 were observed for one of the two alleles. Importantly, the chromatin insulator protein CTCF and the ICR region located on the maternal chromosome were found to mediate the interaction (Ling et al., 2006). Another prominent example of an interchromosomal interaction is that involving the initiation of X inactivation. In this case, transient colocalization of the X inactivation centers of the homologous chromosomes precedes the initiation of inactivation of one of the homologs (Bacher et al., 2006; Xu et al., 2006). CTCF (Xu et al., 2007) as well as a recently identified X-pairing region (Xpr) (Augui et al., 2007) have been implicated in mediating this interaction. More recently, CTCF was shown to colocalize with cohesin at specific sites in the human and mouse genomes during interphase (see Minireview by D. Peric-Hupkes and B. van Steensel on page 925 of this issue).
It is hypothesized that long-range interchromatin interactions allow distant regulatory sequences and proteins to end up in close proximity to their target genes, leading to better coordination of the expression of similarly regulated genes. This mechanism may be gene specific as was found to be the case for members of the Hox gene cluster in developing mouse embryos and differentiating embryoid bodies (reviewed in Fraser and Bickmore, 2007; Lanctot et al., 2007b). Hence, an important question that arises is whether the relative positions of genes have an impact on transcription initiation and/or activity level.
A recent study examined changes in position and transcriptional status of twelve cancer-related genes in mammary epithelial cells grown in 3D culture (Meaburn and Misteli, 2008). Upon differentiation or induction of tumorigenesis, when nuclei undergo large-scale changes in nuclear organization, some but not all genes showed a change in position relative to the nuclear periphery (Meaburn and Misteli, 2008). Interestingly, such changes did not reveal a general trend correlating gene position to transcriptional activity (Meaburn and Misteli, 2008). Therefore, the influence of gene position on gene expression may be dealt with in the nucleus on a gene-by-gene basis and developmental processes may significantly influence the coordination between gene location and expression.
Much of our understanding of large-scale changes in gene positioning has been derived from analyses of fixed cells (reviewed in Lanctot et al., 2007a). The majority of live-cell imaging studies addressing chromatin movement using specific protein-DNA interactions to tag loci in yeast, Drosophila, and transformed mammalian cells have indicated that chromatin movement occurs in the range of 0.2 to 0.5 μm and is generally constrained and highly diffused (reviewed in Gasser, 2002). Recently, directed long-range movement, in the range of 1–5 μm, was observed to occur over a period of 1–2 hr for an inducible transgene in Chinese Hamster ovary (CHO) cells (Chuang et al., 2006). In a second study, directed movement over distances of 2–3 μm toward a Cajal body was observed upon transcriptional induction of a transgene array of U2 snRNA genes (Dundr et al., 2007). Interestingly, nuclear actin (Chuang et al., 2006; Dundr et al., 2007) and myosin (Chuang et al., 2006) were implicated in mediating these chromatin movements.
In this issue of Cell, Nunez et al. (2008) report an elegant study addressing long-range intra- and interchromosomal interactions between estrogen receptor (ERα) bound target sites resulting in coordinated gene expression. To deduce whether ERα induces interchromosomal interactions, the authors devised a new methodology called Deconvolution of DNA interaction by DNA selection and ligation (3D), which is a combination of 3C (Dekker et al., 2002) and Chromatin Immuno Precipitation-DNA Selection and Ligation (ChIP-DSL) (Kwon et al., 2007). When a 1.4 Mb region surrounding the ERα-regulated TFF1 gene on human chromosome 21 was subjected to 3D analysis, it revealed a series of expected intrachromosomal interactions. Surprisingly, this analysis also revealed interchromosomal interactions between TFF1 on chromosome 21 and the enhancer and promoter regions of GREB1 on chromosome 2. Chromosomal painting revealed “kissing” between chromosomes 21 and 2 in a large percentage of MCF7 cells (a human breast cancer cell line) 60 min after treatment with steroid hormone 17β-estradiol (E2). In contrast to earlier studies, which showed monoallelic interchromosomal interactions among various genes (reviewed in Lanctot et al., 2007a), the present study indicates both monoallelic (50%) and biallelic (50%) interchromosomal interactions between chromosomes 21 and 2 in both E2-treated MCF7 cells and primary human mammary epithelial cells (HMEC). Intriguingly, interchromosomal interactions between the TFF1 and GREB1 genes were nonhomologous interactions between chromosome 21 and chrosmosome 2; homologous interactions were not observed. Similar observations were made in regard to the androgen receptor (AR) regulated KLK2 gene on chromosome 19 and the TMPRSS2 gene on chromosome 21 in LNCaP prostate cancer cells. These results indicate dynamic interactions between specific genes upon steroid receptor signaling and raise the issue as to how rapidly such signals are transduced resulting in the coordinate movement and association of these genes.
Time course analysis of E2-induced ERα-dependent interchromosomal interactions revealed an amazingly rapid association, as early as 2 min post-induction, with a peak of association observed at 60 min. When simultaneous multiplex DNA-FISH analysis of ERα-targeted enhancer/promoter pairs specific for chromosome 21 was carried out in normal breast epithelial cells each of the loci on chromosome 21 fused into a single spot. However, when a combination of probes were used that recognize enhancer/promoter regions on different chromosomes (1, 2, 6, 14, 20, and 21), the ERα target regions showed convergence from approximately 20–22 hybridization signals to 7–8 hybridization signals upon E2 treatment. These results indicate that different E2-regulated gene sets may be involved in different interaction networks. However, they also raised the question as to whether the observed interchromosomal interactions were the cause or consequence of E2-induced gene expression. Treatment with α-amanitin (an RNA Pol II inhibitor) or specific short-interfering RNAs (siRNAs) to ERα or FoxA1 (required for ERα binding) abolished interchromosomal interactions. Therefore, transcription and nuclear receptor signaling are required for the observed interchromosomal interaction. Furthermore, single-cell microinjection of siRNAs to various coactivators of ERα inhibited interactions between TFF1 and GREB1. However, siRNAs against LSD1, the histone lysine demethylase, recently shown to be required for E2-dependent gene activation (Garcia-Bassets et al., 2007), blocked E2-dependent expression of GREB1 and TFF1 but did not affect ERα-dependent interchromosomal interactions. Although it is clear that ERα receptors are required for interchromosomal interactions it remains to be determined why shutting down global transcription abolishes interchromosomal interactions whereas downregulation of E2-dependent expression of GREB1 and TFF1 by LSD1 does not. Perhaps knockdown of GREB1 and TFF1 by specific siRNAs will provide further insight into this interesting observation. These results reveal intriguing interchromosomal interactions between coordinately regulated genes mediated by steroid receptors. However, the precise intranuclear signaling cascade that controls the movement and subsequent “search-and-find” mission of these genes remains to be elucidated.
Although the presence of nuclear actin was first suggested in 1969 (Lane, 1969), its role in the nucleus has been extremely controversial. However, several studies have shown actin to be a component of chromatin remodeling complexes, mRNP complexes, and all three RNA polymerase complexes (reviewed in de Lanerolle et al., 2005; Pederson and Aebi, 2005; Percipalle and Visa, 2006). Furthermore, as discussed earlier, two recent studies have implicated actin/myosin I in directing long-range movements of chromatin in interphase (Chuang et al., 2006; Dundr et al., 2007). Given the rapid repositioning of ERα-regulated genes upon estrogen treatment, Nunez and colleagues (2008) examined the potential role of actin/myosin in ERα-mediated interchromosomal interactions. Treatment of E2-stimulated breast epithelial cells with the drugs latrunculin (which blocks actin polymerization) or jasplakinolide (which inhibits actin depolymerization) prevented ERα-dependent interchromosomal interactions and activation of ERα target genes. Single-cell nuclear microinjection of neutralizing antibodies or siRNAs against ARP2/3 (actin-related protein involved in branching), nuclear myosin-I, actin-fold proteins (BAF53, 57, and 170), or dynein light chain-1 abolished E2-induced interchromosomal interactions. These data support a role for actin/myosin in ERα-dependent interchromosomal interactions and gene movement. Next, the functional consequences of disrupting these interactions were examined by RNA and DNA dual FISH analysis. Enhanced transcription was observed at interacting alleles, indicating that ERα-dependent interchromosomal interactions coordinately upregulate transcription. However, transcriptional levels were reduced but not abolished among noninteracting alleles. One interpretation of these interesting results is that the interacting alleles provide a stronger signal in recruiting the gene expression machinery. Alternatively, their movement to a new interaction site may place them in a microdomain that is more conducive to transcription.
Given that gene interactions induced by E2 exhibited enhanced transcriptional activity, Nunez et al. (2008) next investigated whether interacting genes were localized to a specific sub-nuclear domain. An obvious choice to examine were nuclear speckles also known as interchromatin granule clusters (reviewed in Lamond and Spector, 2003). These nuclear regions are enriched in pre-mRNA splicing factors as well as many other proteins involved in gene regulation (reviewed in Lamond and Spector, 2003). Nuclear speckles have been proposed to act as storage and/or assembly/modification conduits for pre-mRNA splicing factors. These factors are recruited from speckles to active genes that may reside on the periphery of speckles or at nuclear sites away from speckles (reviewed in Lamond and Spector, 2003). Proximity to nuclear speckles has been hypothesized to increase the efficiency of splicing of some genes (Huang and Spector, 1991; Moen et al., 2004).
Immuno-DNA FISH analysis using a mixture of 6 or 20 probes to ERα target genes, together with anti-SC35 (speckle marker protein) immunolabeling, showed that ERα target genes colocalized with a speckle upon E2 induction (Figure 2). Drugs affecting actin assembly/interactions (latrunculin or jasplakinolide) or siRNAs to various motor proteins not only abolished gene/gene interactions but also abolished gene/speckle interactions. Given that siRNAs against the histone demethylase LSD1 were able to inhibit E2-dependent transcription of TFF1 and GREB1 without affecting their interchromosomal interactions, the role of LSD1 in gene/speckle interactions was examined. Interestingly, siRNAs against LSD1 abolished the colocalization of TFF1/GREB1 with nuclear speckles. Based on these data, Nunez and colleagues propose a two-step mechanism for coordinately regulating the expression of genes initiating from different chromosomes. The first step involves establishment of specific interchromosomal contacts and is dependent on ERα but independent of LSD1. The second step, mediated by LSD1, involves the association of ERα-interacting loci with nuclear speckles. Hence, the authors propose that nuclear speckles may act as “hubs” where coordinately regulated genes may move in order to enhance their transcriptional activity. However, an alternative interpretation of these data is that knockdown of LSD1 results in downregulation of ERα-dependent genes (Garcia-Bassets et al., 2007) concomitant with a loss of the transcription and pre-mRNA splicing machinery that was previously recruited to the gene loci.
In summary, Nunez et al. (2008) have revealed components of a coordinated mechanism through which activated genes, present on different chromosomes, may be brought together over relatively long distances in the interphase nucleus through interaction with actin and additional motor proteins. The gene interactions result in stimulated levels of gene expression, which take place in association with the periphery of nuclear speckles via an LSD1 interaction. Therefore, nuclear organization and function are intimately linked.
Based on their data, Nunez and colleagues suggest that nuclear speckles are dynamic “hubs” for transient chromosomal interactions. As transcription and RNA processing are coordinated processes, placing genes in proximity to a nuclear speckle may result in an increased efficiency of recruitment of the pre-mRNA splicing machinery, which may then feedback on transcription. This raises the question as to whether interacting ERα target genes recruit splicing factors to their coordinate transcription site from pre-existing speckles or if the genes move to a pre-existing speckle (Figure 2). Although the authors favor the latter possibility, analyses of fixed cells cannot conclusively distinguish between these two possibilities. In order to do so, one would have to make use of systems such as the lac operator/repressor system (Janicki et al., 2004) to directly visualize the dynamics of the genes in question in living cells that also stably express a SC35 fluorescent fusion protein to mark speckles. Such a system would also allow one to further determine whether the interactions with a hub can be reversed upon removal of E2. In addition, it remains to be determined how many genes are present in a specific hub. Do hubs correspond to a subset of previously described transcription factories? Is there a specificity as to which genes come together in a specific hub, and if so, how is this specificity achieved and regulated?
Nunez et al. (2008) have established the requirement of actin-based nuclear motor proteins for hormone-induced regulation of genes through interchromosomal interactions. However, the function of these proteins in the observed events is unclear. Long nuclear actin filaments that can serve as a track for gene movements have not been observed in mammalian nuclei. Is it possible that this process involves relatively short actin filaments that are rapidly assembled and disassembled, such that at steady-state few filaments are resolvable at the current resolution of the light microscope? Certainly, electron microscopic analysis and future studies taking advantage of advances in live-cell microscopy will allow this important question to be addressed. Furthermore, actin has been shown to be involved in the association of genes in a hub. Given that previous studies have identified actin in transcription complexes (reviewed in Pederson and Aebi, 2005), could this actin be involved in providing a small localized framework upon which transcription events are enhanced? Although the present study has provided significant new insight into our understanding of E2-induced chromatin dynamics and associations, it has also opened up a series of important questions that are sure to keep investigators in the field of nuclear structure/function busy for years to come.
D.L.S. is supported by grants from NIH/NIGMS 42694 and 71407 and EY18244. We also thank James Duffy for artistic services.