| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | ADVANCED SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Laboratory of Developmental Biology, Institute of Molecular and Cellular Biosciences, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan
2 Biomedicinal Information Research Center (BIRC), National Institute of Advanced Industrial Science and Technology (AIST), 2-42 Aomi, Koto-ku, Tokyo 135-0064, Japan
 |
Abstract |
|---|
 Top Abstract IntroductionResultsDiscussionExperimental proceduresReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionResultsDiscussionExperimental proceduresReferences |
|---|
|
The predominant evidence, however, is that the signaling pathway through histone modifications are frequently complex (Berger 2007). First, some modifications affect more than one downstream modification (Fig. 1B and Table 1). Second, paired modifications occur, in which each of two modifications activates or inactivates the other (Fig. 1B and Table 1). These observations suggest that individual histone modifications are not a simple code for mediating downstream biochemical and biological functions, but rather are part of a complex signaling pathway or machinery (Fig. 1B).
|
The robustness of the signaling has also been shown with in vivo analyses of deletion and point mutations of histone residues in yeast, a more tractable system for in vivo point mutational analyses than higher eukaryotes (Fig. S1). These analyses showed that deletion of the N-terminal histone tail of any one of the four core histones in yeast does not result in lethality (Durrin et al. 1991), and that single point mutations of acetylatable residues in the histone H4 tail reportedly involved in transcriptional activation only induce moderate defects (Durrin et al. 1991). Furthermore, comprehensive point mutational analysis of the four core histones in yeast recently showed that most of the single point mutations in modifiable residues did not yield any cellular phenotypes within the Spt(–) phenotype for transcription initiation or sensitivity to 6-azauracil for transcription elongation, hydroxyurea for DNA replication, and methyl-methanesulfonate for DNA repair (Matsubara et al. 2007; Figs 1C and S2). Interestingly, signaling in the N-terminal unstructured tail region was more robust against point mutations than signaling in the structured core domain (Figs 1D and S2), suggesting that these regions may have different signaling mechanisms. This observed robustness of histone signaling system against point mutations suggests a complexity of the nuclear signaling pathway through histone modifications that cannot be easily explained by the histone code hypothesis. Therefore, several major questions arise regarding the molecular rules governing the relationship between the complexity of the histone signaling pathway and its robustness.
The first question concerns the overall architecture of the signaling pathway. Although biochemical and biological studies have suggested that signaling through histone modifications is complex and includes bypasses, the overall architecture of the signaling pathway has not been analyzed in detail. The second question concerns the mechanism(s) responsible for robustness against mutations. This question also encompasses how the signal is read out from histone modifications, a process that is also resistant to the effects of point mutations. The third question regards the relationship between the structural characteristics of histone proteins and the continued robustness against point mutations.
Because these problems are related to the complex architecture of the signaling network (Fig. 1B), they clearly fall outside the scope of the histone code hypothesis, which attempts to simply establish one-to-one relationships from the viewpoint of "cause and effect". The histone code hypothesis does not purport to analyze the characteristics of the overall structure of the histone modification network. Therefore, a new framework is needed to analyze the properties of histone modifications as an integrated system. Since the collection of histone modification interactions appears to constitute a complex signaling network (Fig. 1B), the overall architecture (topology) of the histone modification network must be analyzed to develop a more relevant understanding of the signaling through histone modifications.
Recent advances in mathematical network theory have enabled the analysis of the overall architecture of complex biological networks (Barabasi & Oltvai 2004; Albert 2005). In mathematical network theory, the network is treated as a collection of nodes and links, which are system elements and their interactions, respectively. Small world (Watts & Strogatz 1998) and scale-free (Barabasi & Albert 1999) networks have frequently been used as models for the analysis of complex biological networks. Each model network has characteristic values of network parameters, such as average path length, clustering coefficient and degree distribution (Albert & Barabasi 2002). A small world network, which has a large cluster coefficient and a small average path length (Watts & Strogatz 1998), is positioned between completely regular and completely random networks. The scale-free network is characterized by node connectivity that follows a scale-free power-law distribution (Barabasi & Albert 1999). The scale-free network is tolerant of random deletion of some nodes (Albert et al. 2000). However, when highly connected nodes (hubs) are eliminated, the network topology is rapidly transmuted. Metabolic, transcriptional regulation, signal transduction and protein–protein interaction networks have all demonstrated scale-free properties (Barabasi & Oltvai 2004; Albert 2005).
In general, the functional characteristics of nodes and links need not be considered in network theory. For example, temporal and spatial conditions that are required to activate nodes and links in the biological systems are not reflected in the network. The analysis of the network, however, can provide not only an overall insight into the characteristics of the network but also a clue to understand the functional characteristic(s) of nodes. It has been suggested that highly connected hubs in a scale-free network are critical factors for network function. For example, the observation that ATP constitutes a highly connected node in the metabolic network infers that ATP is likely to be an important factor (metabolite) in cell systems (Jeong et al. 2000). It has also been reported that highly connected proteins are the ones that are most important for cell survival (Jeong et al. 2001). The results of these network studies were obtained from networks that were prepared without considering the temporal and spatial aspects of nodes and links, and their biological functions. Furthermore, the "cause and effect" type relationship, which provides the basis for the histone code hypothesis, was not an essential part of these former analyses. Analyses based on network theory are significantly different from those used in standard methods in biology.
In the present study, we analyzed the characteristics of the histone modification signaling pathway on the basis of network theory, showing that the histone modification network also has scale-free properties (Horikoshi 2008). Our analysis suggests that the characteristics of the unstructured histone tails are suitable for producing a complex scale-free network, leading to a robust signaling pathway between chemical modifications in the unstructured histone tail region. In this signaling pathway, the combination of the modified and unmodified residues seems to contribute to the signal readout from the network.
|
Results |
|---|
|
TopAbstractIntroductionResultsDiscussionExperimental proceduresReferences |
|---|
To analyze the network of histone chemical modifications, we initially prepared a table of direct interactions of histone modifications (Table 1). The direct interaction data was obtained from an intensive search of such data relating to human, mouse, fission yeast and budding yeast in the literature. When compiling the cross-talk data in the Table 1, we carefully adopted "direct" interactions between histone post-translational modifications for which we had both biochemical and biological evidence (Fig. S3). Five types of modifications, acetylation, methylation, phosphorylation, ubiquitination, and citrullination, were found in the direct interactions so far reported. When preparing the histone modification network (histone "modification web"), modifications were considered to be nodes and direct interactions between histone chemical modifications to be links. A node can be definitely specified with a histone subunit, a residue number and a type of modification (e.g. H3-K4me; Turner 2005) for all types of modifications except methylation.
Since methylation has a further classification, such as mono-, di-, or tri-methylation, the full specification of histone methylation requires the class of methylation in addition to the normal set of specifiers. Because some studies on histone methylation did not consider the class of the histone methylation, two types of specifications for methylation-relevant nodes are found in Table 1, namely nodes with methylation class (methylation-classified nodes; approximately 60% of methylation-relevant nodes) and nodes without methylation class (methylation-unclassified nodes; approximately 40% of methylation-relevant nodes). It would be inappropriate to prepare a histone modification web by combining different types of nodes. When nodes with methylation class are utilized in the graph, methylation-unclassified nodes cannot be incorporated, resulting in the loss of a substantial amount of information about methylation-relevant interactions (Fig. S4). In order to maximize the amount of information in the graph, we did not include information about the methylation class in the present study. Because most of the experimental data used in the present study, including the analysis of the robustness of signaling through histone modifications, were obtained from biochemical and biological experiments at the residue level, such as from point mutations, the network architecture analysis at the residue level is likely to be valid even in the absence of the methylation classification.
We applied the following five rules when preparing the Table 1. First, we did not distinguish between inter- and intranucleosomal interactions, because it remains unclear whether interactions among histone chemical modifications occur in an inter- or intranucleosomal manner. However, the mobile characteristics of histone tails make internucleosomal reactions likely to occur. Second, although it is possible that histone modifications affect the modification of non-histone proteins and vice versa, we only considered the modification network among histone proteins in the present study. Third, although we could not exclude the possibility that unmodified residues in one or more histone proteins regulate the chemical modification of a histone tail, only interactions between histone modifications were considered. Fourth, only the relationship originating from one modification was considered, although, theoretically, a set of modifications can simultaneously regulate a downstream modification; indeed, results obtained from biochemical and biological experiments have suggested that one modification has a dominant effect on the subsequent modifications (Table 1). Fifth, we assumed that the network could function as an integrated network, although the network is composed of links that were independently characterized.
Histone "modification web" and its properties
We prepared individual histone modification webs for yeasts (fission yeast and budding yeast) and for mammals (human and mouse), as well as a combined web for mammals and yeasts, using the direct interaction data for these histone modifications (Table 1, Figs 2A and S5). The yeast histone modification web was not further analyzed because it was too small (only 11 nodes and 8 links). Mammalian and the combined histone modification webs were mathematically analyzed. Validity of the combination of the interactions can be given by the fact that core histones and their interactions are well-conserved between yeast and human. The amino acid sequences of histones are approximately 90% conserved among eukaryotes. Those residues chemically modified in yeast histones are completely conserved in human histones (Fig. S6). In addition, most histone modification enzymes are evolutionarily conserved among eukaryotes (Klose & Zhang 2007), suggesting that the essential modifications are well-conserved. Furthermore, functional conservation of histone modification has been reported across widely divergent species (Trojer & Reinberg 2007; Fig. S5). All these indicate that the graphs composed of the histone modification interactions are well-conserved from yeast to human. Histone modification enzymes and factors containing modification-recognition domains are more prevalent in mammals than in yeasts. In addition, mammals have various cell types that are regulated in temporally- and spatially-dependent manners. These observations suggest that the mammalian histone modification web is more complex than that of yeasts and that the yeast histone modification web is a subset of the mammalian histone modification web.
|
The combined histone modification web had a clustering coefficient of 0.31 and an average path length of 2.5. These values are marginally larger than those of random networks of the same size. The network heterogeneity (0.71) is, however, substantially larger than that of a random network, suggesting the presence of network hubs (Dong & Horvath 2007). Indeed, there are some highly connected nodes in the histone modification web, such as H3-K4me, H3-K9me, H3-K9ac, H3-S10ph and H3-K14ac (Fig. 2A). Furthermore, these nodes form a cluster in which every two nodes are connected, except H3-K9ac and H3-K14ac (Fig. 2A). The relatively small clustering coefficient suggests that the present histone modification web is not a typical small world network (Watts & Strogatz 1998). Analysis of the degree distributions of the network (Albert & Barabasi 2002), namely in-degrees [Pin(k)], out-degrees [Pout(k)], and the degree distribution of weakly connected components [Pweak(k)], revealed that each P(k) follows a power law (Fig. 3 and Table 2), although the range of the degree distribution was rather small. Together, the network parameters suggest that the histone modification web has the properties of a scale-free network. The mammalian histone modification web has similar network parameters (clustering coefficient = 0.34, average path length = 2.3, and network heterogeneity = 0.66) to those of the combined histone modification web. Furthermore, its degree distributions follow a power law (Fig. S7 and Table S1).
|
|
Pseudo–mirror symmetry structure of the histone modification web
By varying the graph layout of the histone modification web on the basis of functional characteristics of links, a substructure with a pseudo–mirror symmetry emerged from the graph (Fig. 2B). This structure and its implications become clearer when the direct interactions for a histone acetylation with an ambiguous residue specification (red dotted arrow in the Fig. 2B) are added to the histone modification web (Fig. 2B and Table 1). Although the network architecture of the upper-left (orange box) and upper-right (blue box) regions was similar, their functions could be regarded, respectively, as activation and inactivation subnetworks for the acetylation of histones H3/H4 (Fig. 2B).
Interestingly, the two subnetworks may each inhibit signaling in the opponent subnetwork. Since the input signal into the activation or inactivation subnetwork inhibits the function of its opponent (or antagonistic) subnetwork, a clear signal for the acetylation of histones H3/H4 or its inhibition is generated. For example, when a signal that activates histones H3/H4 is inputted into the histone modification web, the modifications that inhibit histones-H3/H4 acetylation are prevented and acetylation of histones H3/H4 is ensured.
A feedback mechanism from the acetylation status of histones H3/H4 is also implemented in the network. For example, the acetylation of H3-K14 inhibits the methylation of H3-K9, leading to inhibition of the inactivation subnetwork; this route can be considered an anti-inactivation feedback (Fig. 2B). Positive feedback routes from acetylated histones H3/H4 are also found in the network: H3-K14ac and H3-K9ac activate H3-S10ph and H3-K4me which, in turn, promotes the acetylation of histones H3/H4 (Fig. 2B).
Robustness of the histone modification web
Next, we tried to explain the observed robustness of the histone modification web on the basis of the histone modification web architecture. First, the scale-free property of the histone modification web suggests that the random failure of a particular node is unlikely to destroy the network. Second, the pseudo–mirror symmetry (Fig. 2B), which provides complementary functions to each subnetwork, reduces the signal transduction in the opponent subnetwork, supporting a stable response of the network. Third, the feedback pathway enhances the response of the network. Fourth, the highly connected cluster of hubs in the histone modification web provides tolerance to a hub defect; if one hub fails, other hubs in the cluster can compensate for the lost function of the defective hub. In combination, these properties would make the histone modification web able to compensate for the effects of mutations. For example, if H3-S10ph is removed from the web, other links can compensate for its function (Fig. S8A). The mutation of H3-K4 can be compensated by a feedback mechanism if histones H3/H4 are acetylated at a basal level (Fig. S8B); the basal acetylation of H3-K9 and -K14 can facilitate the phosphorylation of H3-S10, which in turn facilitates acetylation of H3-K14 and inhibits the inactivation subnetwork.
Since histone modifications can be regulated in a temporally- and spatially-dependent manner in living cells, the all nodes and links are not always available. It is therefore reasonable to expect that compensation pathways will change depending on temporal and spatial conditions.
Distribution of nodes and links on the N-terminal histone tail in the histone modification web
Next, we investigated the histone modification web from a structural viewpoint. Because the histone modification web is tolerant of mutations of residues in the N-terminal long tail (Figs 1D and S2), we analyzed the correlation between structural characteristics and the properties of the nodes and links on the unstructured N-terminal histone tails. The number of links per node was inversely correlated with the distance between the N-terminus and the node, suggesting that preferential link formations with nodes at the near N-terminus occurred in the course of generating the histone modification web (Fig. 4A). This phenomenon can be explained by the structural characteristics of the tail region. The N-terminus provides an area widely accessible to interactions with various core histone residues, due to the mobile and flexible characters of the unstructured region (Fig. 4B). Residues close to the N-terminus may, therefore, have a greater chance to form links. In addition, because the N-terminal histone-tail region is at a distance from the nucleosome core particle, the binding of a chromatin-associated factor around the N-terminal histone tail is unlikely to encounter steric hindrance, whereas the chance of steric hindrance increases when the binding site is closer to the nucleosome core particle. This would explain the preferential formation of links by residues near the N-terminus.
|
|
Discussion |
|---|
|
TopAbstractIntroductionResultsDiscussionExperimental proceduresReferences |
|---|
The histone modification web connects upstream signaling pathways and downstream DNA-mediated reactions and functions as a junction between them; robust signaling activity is, therefore, critical to ensure a stable response for various nuclear processes. In the present study, we analyzed the characteristics of the histone modification web using graph theory. Because all known direct interactions include fewer than 40% of the histone modifications identified so far, the information presented in the present histone modification web is inevitably limited. This limitation is, however, not a special characteristic of the histone modification web. Complex networks in the real world, such as the World Wide Web, the metabolite network and the signal transduction network in the cell, are large in size and extremely complex; it has been impossible in practice to obtain full information about these networks. Despite the incomplete data, these networks have been successfully analyzed using network theory, resulting in novel insights into the characteristics of these networks (Barabasi & Oltvai 2004; Albert 2005).
Similarly, the present analysis revealed unique characteristics of the histone modification web. First, we found that the histone modification web has scale-free properties, which are likely to be due to the characteristics of the N-terminal unstructured histone tail (see below). Second, the web includes some signaling pathways for feedback regulation, a characteristic widely found in various biological systems (Barabasi & Oltvai 2004; Albert 2005). Third, the structure of the histone modification web contains a pseudo–mirror symmetry of activating and inactivating subnetworks for acetylation of histones H3/H4. Because it is unlikely that pseudo–mirror symmetry could be generated from a random process, there must be some principles that generate the symmetric structure. For example, the combination of binary-switch–like processes for two adjacent nodes (or two nodes on the same residue) would produce a symmetric regulatory network (Fig. S9).
Robustness of the histone modification web and its evolutionary implications
Our analysis suggested that the combination of scale-free properties, pseudo-mirror symmetry structure, feedback pathways, and hub clusters of the histone modification web (Fig. 2A,B) contributes to the robustness of signaling through histone modifications. Indeed, a theoretical analysis of the histone modification web suggested that these characteristics could compensate for the failure of nodes within the network (Fig. S8). The bypasses predicted in Fig. S8 can be tested by comparing the histone modification patterns of modifiable residues between wild-type and mutant species.
Because the nodes in our web are mainly located at the N-terminal histone tail regions (Fig. 2C), the analysis in the present study mostly reflects the properties of the nodes located in these regions. The robustness of the network in response to point mutations in other regions and molecules (for instance, modifications of core regions of histones and other chromatin-associated factors, or DNA methylation) could be partly explained by the results of the present study, if these modifications also belong to the histone modification web. Further analysis, and compilation of additional relevant data, will refine and extend the overall structure of the histone modification web.
The robustness of the histone modification signaling system against point mutation(s) [even in the case of N-terminal deletions of core histones (Durrin et al. 1991)] means that several modifiable residues can be mutated without affecting the function of the system, a phenomenon that seems, at first glance, to be inconsistent with the high degree of primary sequence conservation in core histones. We suggest that this high evolutionary conservation enables conservation of the whole architecture of the histone modification web, which seems to be a critical condition for survival under ever-changing environmental conditions over many generations. When a point mutation or deletion of a histone modifiable residue causes a defect in a normal signaling pathway, a compensation pathway(s) would be activated to maintain cell function. However, as described above, the nodes and links that are needed for signal compensation are not always available. Therefore, the nodes and links for the compensation pathway would, in general, be activated via the expression and/or activation/inactivation of components, such as histone modification enzymes and factors containing modification-recognition domains. Since these compensation processes would likely cause a delay in signaling and engender additional "costs" for the cell, the accumulation of mutations in the histone modification network would reduce the chances of cell survival. Therefore, a histone modification web without mutations would be selected for during evolution, resulting in the high degree of primary sequence conservation in core histones. A robust histone modification signaling system would help an organism overcome the temporary loss of a histone modification, and ensure the stable response of the system to ever-changing conditions.
Generation of the scale-free network of histone modification in the unstructured region
The analysis of the positional effect on properties of nodes suggested that links were preferentially formed at residues close to the N-terminus (Fig. 4), which is consistent with the unstructured property of the histone tail. Interestingly, this "preferential attachment" is one of the two conditions required to generate a Barabasi-Albert type (BA type) scale-free network (Barabasi & Albert 1999; Barabasi & Oltvai 2004; Albert 2005). The unstructured region also seems to satisfy the second condition, the ability of the network to grow by adding new nodes.
Because the flexibility and exposed side chains of the unstructured regions provide good substrates for various enzymes (Dyson & Wright 2005), unstructured tail regions can easily have multiple nodes. Supporting this, it has been reported that histone deacetylase Hst2 prefers denatured substrates to native substrates (Khan & Lewis 2005). In addition to offering a structural framework to accommodate many nodes for signaling, the unstructured regions seem to promote the formation of numerous links between nodes. In order to form a link, a modification site needs to interact with at least two proteins, namely a modification enzyme and a histone–modification recognition domain (Taverna et al. 2007; Fig. 5A). Because the structures of the "node-binding" sites of these two proteins are likely to be different from each other, the conformational flexibility of the unstructured regions around the node is required for specific interaction with these two distinct proteins. Several examples reveal that the histone tail region can adopt distinct conformations to interact with different chromatin-associated factors (Fig. 5B,C), showing that the flexible tail region is competent to interact with various factors through a change in its conformation.
|
Proteins with fully or partially denatured structures are called natively unfolded proteins. It is predicted that about one third of eukaryotic proteins are natively unfolded proteins (Fink 2005). The novel concept proposed here could therefore be applied to other natively unfolded proteins, introducing the notion of a general principle for the function of natively unfolded proteins.
The ability of flexible regions to acquire a scale-free network of post-translational modifications can be applied to all unstructured regions of proteins, so that scale-free networks may exist in unstructured regions of other proteins. Indeed, the unstructured regions of p53 and platelet-derived growth factor receptor (PDGFR) show similar properties to those of the histone modification web. Targeted point mutations in the modifiable residues of these proteins resulted in only modest phenotypic changes (Schreiber & Bernstein 2002; Fischle et al. 2003; Toledo & Wahl 2006), suggesting that unstructured regions of p53 and PDGFR include a scale-free network of chemical modifications. This model could be tested for p53 and PDGFR by using an experimental procedure similar to that reported here, and such analyses of the modification webs of p53 and PDGFR would serve to extend the concept of the histone modification web into other biological systems.
The nucleus is where biological signals are translated into gene expression. In order to regulate switching between the ON and OFF states of gene expression, the enormous amount of signaling information transmitted to the nucleus must be processed in a robust manner under ever-changing environmental conditions. It is therefore reasonable to assume a general mechanism for the processing the large amounts of biological signaling information in the nucleus. Interestingly, many nuclear proteins, such as histones, TATA box-binding protein (Hoffmann et al. 1990), and CTD of RNA polymerase II (Phatnani & Greenleaf 2006), harbor intrinsically unstructured regions (Wright & Dyson 1999). As suggested from the present analysis, flexible regions would seem to afford an appropriate structural platform for the generation of a modification network system (modification web) having a scale-free property. We would like to propose that the modification webs based on unstructured regions could function as routers for the modification signals; the signal router would collect a variety of biological signals, process them, and then transduce their processed forms, leading to the activation or inactivation of gene expression. Although structured domains would also receive and process biological signals, the amount of information processed by such domains would be small compared to those processed by unstructured regions. The flexible and dynamic properties of unstructured regions make them appropriate for the receiving and processing of large amount of the signaling information, respectively (Figs 4 and 5). A signal router based on unstructured regions (signal routing tail) could constitute a platform for the processing of large amounts of biological signaling information.
Signal readout mechanism deduced from the architecture of the histone modification web
Tolerance for most single point mutations in the core histones (Matsubara et al. 2007) suggests that a single chemical modification is not responsible for the signal readout, but rather a set of chemical modifications are recognized as an output signal by signal-readout machinery. Since the histone modification web is composed of activation and inactivation relationships (links) between modifications (nodes), not only the modified residues but also the unmodified residues should be considered as part of the information in the signal readout process. Indeed, some examples of the simultaneous recognition of modified and unmodified residues have already been observed. One of the most frequently occurring patterns is the combination of H3-K4me and unmodified H3-R2 (Fig. 5D). We investigated the so far reported three-dimensional structures of two different H3-K4me recognition domains, the double chromodomain of CHD1 (PDB ID: 2B2T; Flanagan et al. 2005) and the BPTF PHD finger of NURF (PDB ID: 2F6J; Li et al. 2006), in a peptide complex with H3-K4me. As expected, these two domains seem to recognize unmodified H3-R2 as well as modified H3-K4me (Fig. 5E). The three-dimensional structures of these complexes suggested that the methylation of H3-R2 in the histone peptides would decrease the affinity for these recognition domains (Fig. 5F). Indeed, the peptide with H3-R2me and H3-K4me shows lower affinity for CHD1 double chromodomain (Flanagan et al. 2005). The interaction between HP1 and a peptide containing H3-K9me has a similar recognition mode: phosphorylation of H3-S10 of the histone H3 peptide, which inhibits the methylation of H3-K9, interrupted its interaction with HP1 (Fischle et al. 2003).
The model of the simultaneous recognition of modified and unmodified residues can explain an enigmatic recognition pattern of the histone peptide of the WD40 protein WDR5 (Couture et al. 2006). Although WDR5 was previously thought to recognize H3-K4me in the histone tail, the crystal structure clearly showed that WDR5 accommodates only unmodified H3-R2 in the deep pocket of WDR5 (Couture et al. 2006). WDR5 can be regarded as indirectly recognizing H3-K4me through recognizing the unmodified H3-R2, because unmodified H3-R2 can be generated by an inhibitory signal from H3-K4me (Figs 2A and 5D).
Prediction of missing interactions in the histone modification web
More data on direct histone modification interactions are needed to fully understand the signaling mechanism(s) through histone modifications. The pseudo-mirror symmetry structure would be a reasonable working framework to begin searching for missing nodes and links in the histone modification web, because this structure seems to play an important role in regulating the acetylation of histones H3/H4. For example, the pseudo-mirror symmetry structure suggests that there is a positive link from H3-K4me to H3-S10ph (Fig. 6A). This hypothetical link is consistent with the activation of histones H3/H4 acetylation by H3-K4me. Further analysis may reveal the existence of the link.
|
Comprehensively understanding the molecular mechanisms of signaling through histone modifications will require not only the preparation of the "full-scale" histone modification web but also the reconsideration of the five rules we used to prepare the present histone modification web, because the five rules were introduced to simplify the histone modification network (see the section "Histone modification network" in Results). The resulting simple histone modification web was quite useful for extracting its key critical characteristics, but removal of some or all of the five rules would lead to a more generalized histone modification web containing comprehensive information about the histone modification network. Below we discuss the likely characteristics of such a generalized histone modification web.
In order to prepare the generalized (or semi-generalized) histone modification web, the following conditions are required: (i) discrimination of the internucleosome and intranucleosome interactions, (ii) inclusion of interactions between histones and non-histone proteins, (iii) inclusion of the effect of the simultaneous recognition of the modified and unmodified residues by the histone-modification recognition domains (Figs 5, 6B and S10) and the effect of the simultaneous regulation of downstream modifications by a set of modifications, and (iv) combination of the histone modification web and dynamics of signaling through histone modifications. These conditions would expand the histone modification web in the following ways. First, discrimination of internucleosome and intranucleosome interactions would open up the possibility of theoretically analyzing the spatial propagation of histone modification patterns along a chromatin fiber. A high-resolution genome-wide analysis of histone modifications and an in vitro reconstitution system for histone modification analysis would be required to achieve this. Second, inclusion of interactions between histones and non-histone proteins may identify signaling pathways that bypass histone modifications. In addition, the analysis of the interaction between the histone modification web and other modification networks in the cell would lead to the elucidation of a "network" of multiple networks. The analysis of the layered structure of biological systems (networks) will reveal novel mechanism(s) of biochemical and biological regulation of the system. Third, inclusion of the effect of the simultaneous recognition of modified and unmodified residues by the histone-modification recognition domains (Figs 5, 6B and S10), and the effect of the simultaneous regulation of a downstream modification by a set of modifications, is essential for a generalized histone modification web. In the generalized method, each node should represent the state of all modifications in a nucleosome and each link should represent the relationship between two states of the nucleosome with histone modifications (Fig. S11); this will require a substantial amount of data about histone modifications of a nucleosome in various states. Fourth, it is important to analyze the dynamics of the signaling in the histone modification web. As described in the Introduction section, we did not take into consideration the temporal and spatial characteristics of nodes and links, because the histone modification web was analyzed using network theory which only focuses on the static structure of the histone modification web. In order to fully understand the function of the histone modification web in living cells, the temporal and spatial characteristics of nodes and links (the dynamic characteristics of the nodes and links) would have to be considered. The integrated study of the static and dynamic characteristics of the histone modification web will be the subject of future studies.
Finally, the histone modification web is a good model system for the comprehensive analysis of biological network systems because it is a relatively small network in the cell, and extensive point mutational analyses have already been carried out (Matsubara et al. 2007) that would make it possible to comprehensively analyze the functional characteristics of the histone modification web. Analysis of the histone modification web could lead to a novel understanding of signaling not only in the histone modification web, but also in other biochemical and biological networks in the cell.
|
Experimental procedures |
|---|
|
TopAbstractIntroductionResultsDiscussionExperimental proceduresReferences |
|---|
First, each histone modification interaction was converted into an "elementary" graph in accordance with the criteria shown in Fig. S3. As an example, phosphorylation of H3-S10 is known to promote the acetylation of H3-K14. Two modifications (phosphorylation and acetylation) were considered to be nodes of the network, and the relationship (or interaction) between H3-S10ph and H3-K14ac was regarded as a link. In this case, a direct arrow from H3-S10ph to H3-K14ac represents the promotion of H3-K14 acetylation by H3-S10 phosphorylation (a bar-headed arrow was used to represent inhibitory relationships). Second, all the prepared elementary graphs were integrated to construct the overall graph shown in Figs 2A and S4 (2A and S4A are methylation-unclassified webs; S4B is a methylation-classified web). We list references describing the relationships among histone modifications in Table 1.
Calculation of degree distribution
We calculated degree distribution through making no distinction between normal and bar-headed arrows in the histone modification web during the calculation process; only the direction of the arrows was taken into account in order to represent the structure of the web (Barabasi & Oltvai 2004). Pweak(k) is the probability that a node in a network is connected to k other node(s). Pin(k) and Pout(k) are the corresponding probabilities that only incoming and outgoing links will be considered. Other network parameters of the histone modification web were analyzed using the program Cytoscape (Shannon et al. 2003) with a network analysis routine (Assenov et al. 2008).
|
Acknowledgements |
|---|
|
Footnotes |
|---|
These authors contributed equally to the work. 
* Correspondence: horikosh{at}iam.u-tokyo.ac.jp |
References |
|---|
|
TopAbstractIntroductionResultsDiscussionExperimental proceduresReferences |
|---|
Albert, R. (2005) Scale-free networks in cell biology. J. Cell Sci. 118, 4947–4957.
Albert, R. & Barabasi, A.L. (2002) Statistical mechanics of complex networks. Rev. Modern Physics 74, 47–97.[CrossRef]
Albert, R., Jeong, H. & Barabasi, A.L. (2000) Error and attack tolerance of complex networks. Nature 406, 378–382.[CrossRef][Medline]
An, W., Kim, J. & Roeder, R.G. (2004) Ordered cooperative functions of PRMT1, p300, and CARM1 in transcriptional activation by p53. Cell 117, 735–748.[CrossRef][Medline]
Assenov, Y., Ramirez, F., Schelhorn, S.E., Lengauer, T. & Albrecht, M. (2008) Computing topological parameters of biological networks. Bioinformatics 24, 282–284.
Barabasi, A.L. & Albert, R. (1999) Emergence of scaling in random networks. Science 286, 509–512.
Barabasi, A.L. & Oltvai, Z.N. (2004) Network biology: understanding the cell's functional organization. Nat. Rev. Genet. 5, 101–113.[CrossRef][Medline]
Berger, S.L. (2007) The complex language of chromatin regulation during transcription. Nature 447, 407–412.[CrossRef][Medline]
Carrozza, M.J., Li, B., Florens, L., Suganuma, T., Swanson, S.K., Lee, K.K., Shia, W.J., Anderson, S., Yates, J., Washburn, M.P. & Workman, J.L. (2005) Histone H3 methylation by Set2 directs deacetylation of coding regions by Rpd3S to suppress spurious intragenic transcription. Cell 123, 581–592.[CrossRef][Medline]
Cheung, P., Tanner, K.G., Cheung, W.L., Sassone-Corsi, P., Denu, J.M. & Allis, C.D. (2000) Synergistic coupling of histone H3 phosphorylation and acetylation in response to epidermal growth factor stimulation. Mol. Cell 5, 905–915.[CrossRef][Medline]
Couture, J.F., Collazo, E. & Trievel, R.C. (2006) Molecular recognition of histone H3 by the WD40 protein WDR5. Nat. Struct. Mol. Biol. 13, 698–703.[CrossRef][Medline]
Cuthbert, G.L., Daujat, S., Snowden, A.W., Erdjument-Bromage, H., Hagiwara, T., Yamada, M., Schneider, R., Gregory, P.D., Tempst, P., Bannister, A.J. & Kouzarides, T. (2004) Histone deimination antagonizes arginine methylation. Cell 118, 545–553.[CrossRef][Medline]
Daujat, S., Bauer, U.M., Shah, V., Turner, B., Berger, S. & Kouzarides, T. (2002) Crosstalk between CARM1 methylation and CBP acetylation on histone H3. Curr. Biol. 12, 2090–2097.[CrossRef][Medline]
Dong, J. & Horvath, S. (2007) Understanding network concepts in modules. BMC System Biol. 1, 1–20.[CrossRef]
Downs, J.A., Allard, S., Jobin-Robitaille, O., Javaheri, A., Auger, A., Bouchard, N., Kron, S.J., Jackson, S.P. & Côté, J. (2004) Binding of chromatin-modifying activities to phosphorylated histone H2A at DNA damage sites. Mol. Cell 16, 979–990.[CrossRef][Medline]
Durrin, L.K., Mann, R.K., Kayne, P.S. & Grunstein, M. (1991) Yeast histone H4 N-terminal sequence is required for promoter activation in vivo. Cell 65, 1023–1031.[CrossRef][Medline]
Dyson, H.J. & Wright, P.E. (2005) Intrinsically unstructured proteins and their functions. Nat. Rev. Mol. Cell Biol. 6, 197–208.[CrossRef][Medline]
Fink, A.L. (2005) Natively unfolded proteins. Curr. Opin. Struct. Biol. 15, 35–41.[CrossRef][Medline]
Fischle, W., Wang, Y. & Allis, C.D. (2003) Binary switches and modification cassetes in histone biology and beyond. Nature 425, 475–479.[CrossRef][Medline]
Flanagan, J.F., Mi, L.Z., Chruszcz, M., Cymborowski, M., Clines, K.L., Kim, Y., Minor, W., Rastinejad, F. & Khorasanizadeh, S. (2005) Double chromodomains cooperate to recognize the methylated histone H3 tail. Nature 438, 1181–1185.[CrossRef][Medline]
Forneris, F., Binda, C., Vanoni, M.A., Battaglioli, E. & Mattevi, A. (2005) Human histone demethylase LSD1 reads the histone code. J. Biol. Chem. 280, 41360–41365.
Gu, W. & Roeder, R.G. (1997) Activation of p53 sequence-specific DNA binding by acetylation of the p53 C-terminal domain. Cell 90, 595–606.[CrossRef][Medline]
Guccione, E., Bassi, C., Casadio, F., Martinato, F., Cesaroni, M., Schuchlautz, H., Luscher, B. & Amati, B. (2007) Methylation of histone H3R2 by PRMT6 and H3K4 by an MLL complex are mutually exclusive. Nature 449, 933–937.[CrossRef][Medline]
Hoffmann, A., Sinn, E., Yamamoto, T., Wang, J., Roy, A., Horikoshi, M. & Roeder, R.G. (1990) Highly conserved core domain and unique N terminus with presumptive regulatory motifs in a human TATA factor (TFIID). Nature 346, 387–390.[CrossRef][Medline]
Horikoshi, M. (2008) 1. From histone modification to nucleosome disassembly 2. Histone "modification web". Regulatory Mechanisms in Eukaryotic Transcription, Keystone Symposia. Poster, 133 Oral, B.4.
Jacobs, S.A. & Khorasanizadeh, S. (2002) Structure of HP1 chromodomain bound to a lysine 9-methylated histone H3 tail. Science 295, 2080–2083.
Jeong, H., Mason, S.P., Barabasi, A.L. & Oltvai, Z.N. (2001) Lethality and centrality in protein network. Nature 41, 411.[CrossRef]
Jeong, H., Tombor, B., Albert, R., Oltavi, Z.N. & Barabasi, A.L. (2000) The large-scale organization of metabolic networks. Nature 407, 651–654.[CrossRef][Medline]
Joo, H.Y., Zhai, L., Yang, C., Nie, S., Erdjument-Bromage, H., Tempst, P., Chang, C. & Wang, H. (2007) Regulation of cell cycle progression and gene expression by H2A deubiquitination. Nature 449, 1068–1072.[CrossRef][Medline]
Khan, A.N. & Lewis, P.N. (2005) Unstructured conformations are a substrate requirement for the Sir2 family of NAD-dependent protein deacetylases. J. Biol. Chem. 280, 36073–36078.
Kirmizis, A., Santos-Rosa, H., Penkett, C.J., Singer, M.A., Vermeulen, M., Mann, M., Bahler, J., Green, R.D. & Kouzarides, T. (2007) Arginine methylation at histone H3R2 controls deposition of H3K4 trimethylation. Nature 449, 928–932.[CrossRef][Medline]
Klose, R.J. & Zhang, Y. (2007) Regulation of histone methylation by demethylimination and demethylation. Nat. Rev. Mol. Cell Biol. 8, 307–318.[CrossRef][Medline]
Kornberg, R.D. & Lorch, Y. (2007) Chromatin rules. Nat. Struct. Mol. Biol. 14, 986–988.[CrossRef]
Kouzarides, T. (2007) Chromatin modifications and their function. Cell 128, 693–705.[CrossRef][Medline]
Latham, J.A. & Dent, S.Y. (2007) Cross-regulation of histone modifications. Nat. Struct. Mol. Biol. 14, 1017–1024.[CrossRef][Medline]
Li, H., Ilin, S., Wang, W., Duncan, E.M., Wysocka, J., Allis, C.D. & Patel, D.J. (2006) Molecular basis for site-specific read-out of histone H3K4me3 by the BPTF PHD finger of NURF. Nature 442, 91–95.[Medline]
Lo, W.S., Trievel, R.C., Rojas, J.R., Duggan, L., Hsu, J.Y., Allis, C.D., Marmorstein, R. & Berger, S.L. (2000) Phosphorylation of serine 10 in histone H3 is functionally linked in vitro and in vivo to Gcn5-mediated acetylation at lysine 14. Mol. Cell 5, 917–926.[CrossRef][Medline]
Luger, K., Mäder, A.W., Richmond, R.K., Sargent, D.F. & Richmond, T.J. (1997) Crystal structure of the nucleosome core particle at 2.8 Å resolution. Nature 389, 251–260.[CrossRef][Medline]
Matsubara, K., Sano, N., Umehara, T. & Horikoshi, M. (2007) Global analysis of functional surfaces of core histones with comprehensive point mutants. Genes Cells 12, 13–33.
McGinty, R.K., Kim, J., Chatterjee, C., Roeder, R.G. & Muir, T.W. (2008) Chemically ubiquitylated histone H2B stimulates hDot1L-mediated intranucleosomal methylation. Nature 453, 812–816.[CrossRef][Medline]
Milne, T.A., Briggs, S.D., Brock, H.W., Martin, M.E., Gibbs, D., Allis, C.D. & Hess, J.L. (2002) MLL targets SET domain methyltransferase activity to Hox gene promoters. Mol. Cell 10, 1107–1117.[CrossRef][Medline]
Nakagawa, T., Kajitani, T., Togo, S., Masuko, N., Ohdan, H., Hishikawa, Y., Koji, T., Matsuyama, T., Ikura, T., Muramatsu, M. & Ito, T. (2008) Deubiquitylation of histone H2A activates transcriptional initiation via trans-histone cross-talk with H3K4 di- and trimethylation. Genes Dev. 22, 37–49.
Nakayama, J., Rice, J.C., Strahl, B.D., Allis, C.D. & Grewal, S.I.S. (2001) Role of histone H3 lysine 9 methylation in epigenetic control of heterochromatin assembly. Science 292, 110–113.
Nelson, C.J., Santos-Rosa, H. & Kouzarides, T. (2006) Proline isomerization of histone H3 regulates lysine methylation and gene expression. Cell 126, 905–916.[CrossRef][Medline]
Nightingale, K.P., Gendreizig, S., White, D.A., Bradbury, C., Hollfelder, F. & Turner, B.M. (2007) Cross-talk between histone modifications in response to histone deacetylase inhibitors: MLL4 links histone H3 acetylation and histone H3K4 methylation. J. Biol. Chem. 282, 4408–4416.
Nishioka, K., Chuikov, S., Sarma, K., Erdjument-Bromage, H., Allis, C.D., Tempst, P. & Reinberg, D. (2002a) Set9, a novel histone H3 methyltransferase that facilitates transcription by precluding histone tail modifications required for heterochromatin formation. Genes Dev. 16, 479–489.
Nishioka, K., Rice, J.C., Sarma, K., Erdjument-Bromage, H., Werner, J., Wang, Y., Chuikov, S., Valenzuela, P., Tempst, P., Steward, R., Lis, J.T., Allis, C.D. & Reinberg, D. (2002b) PR-Set7 is a nucleosome-specific methyltransferase that modifies lysine 20 of histone H4 and is associated with silent chromatin. Mol. Cell 9, 1201–1213.[CrossRef][Medline]
Owen, D.J., Ornaghi, P., Yand, J.C., Lowe, N., Evans, P.R., Ballario, P., Neuhaus, D., Filetici, P. & Travers, A.A. (2000) The structural basis for the recognition of acetylated histone H4 by the bromodomain of histone acetyltransferase GCN5p. EMBO J. 19, 6141–6149.[CrossRef][Medline]
Phatnani, H.P. & Greenleaf, A.L. (2006) Phosphorylation and functions of the RNA polymerase II CTD. Genes Dev. 20, 2922–2936.
Pray-Grant, M.G., Daniel, J.A., Schieltz, D., Yates, J.R. III & Grant, P.A. (2005) Chd1 chromodomain links histone H3 methylation with SAGA- and SLIK-dependent acetylation. Nature 433, 434–438.[CrossRef][Medline]
Rea, S., Eisenhaber, F., O’Carroll, D., Strahl, B.D., Sun, Z.W., Schmid, M., Opravil, S., Mechtler, K., Ponting, C.P., Allis, C.D. & Jenuwein, T. (2000) Regulation of chromatin structure by site-specific histone H3 methyltransferases. Nature 406, 593–599.[CrossRef][Medline]
Schotta, G., Lachner, M., Sarma, K., Ebert, A., Sengupta, R., Reuter, G., Reinberg, D. & Jenuwein, T. (2004) A silencing pathway to induce H3-K9 and H4-K20 trimethylation at constitutive heterochromatin. Genes Dev. 18, 1251–1262.
Schreiber, S.L. & Bernstein, B.E. (2002) Signaling network model of chromatin. Cell 111, 771–778.[CrossRef][Medline]
Shahbazian, M.D., Zhang, K. & Grunstein, M. (2005) Histone H2B ubiquitylation controls processive methylation but not monomethylation by Dot1 and Set1. Mol. Cell 19, 271–277.[CrossRef][Medline]
Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., Amin, N., Schwikowski, B. & Ideker, T. (2003) Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504.
Shi, X., Hong, T., Walter, K.L., et al. (2006) ING2 PHD domain links histone H3 lysine 4 methylation to active gene repression. Nature 442, 96–99.[Medline]
Sims, R.J. 3rd & Reinberg, D. (2008) Is there a code embedded in proteins that is based on post-translational modification? Nat. Rev. Mol. Cell Biol. 9, 815–820.[CrossRef][Medline]
Strahl, B.D. & Allis, C.D. (2000) The language of covalent histone modifications. Nature 403, 41–45.[CrossRef][Medline]
Suganuma, T. & Workman, J.L. (2008) Crosstalk among histone modifications. Cell 135, 604–607.[CrossRef][Medline]
Taverna, S.D., Ilin, S., Rogers, R.S., Tanny, J.C., Lavender, H., Li, H., Baker, L., Boyle, J., Blair, L.P., Chait, B.T., Patel, D.J., Aitchison, J.D., Tackett, A.J. & Allis, C.D. (2006) Yng1 PHD finger binding to H3 trimethylated at K4 promotes NuA3 HAT activity at K14 of H3 and transcription at a subset of targeted ORFs. Mol. Cell 24, 785–796.[CrossRef][Medline]
Taverna, S.D., Li, H., Ruthenburg, A.J., Allis, C.D. & Patel, D.J. (2007) How chromatin-binding modules interpret histone modifications: lessons from professional pocket pickers. Nat. Struct. Mol. Biol. 14, 1025–1040.[CrossRef][Medline]
Toledo, F. & Wahl, G.M. (2006) Regulating the p53 pathway: in vitro hypotheses, in vivo veritas. Nat. Rev. Cancer 6, 909–923.[CrossRef][Medline]
Trojer, P. & Reinberg, D. (2007) Facultative heterochromatin: is there a distinctive molecular signature? Mol. Cell 28, 1–13.[CrossRef][Medline]
Turner, B.M. (2005) Reading signals on the nucleosome with a new nomenclature for modified histones. Nat. Struct. Mol. Biol. 12, 110–112.[CrossRef][Medline]
Utley, R.T., Lacoste, N., Jobin-Robitaille, O., Allard, S. & Cote, J. (2005) Regulation of NuA4 histone acetyltransferase activity in transcription and DNA repair by phosphorylation of histone H4. Mol. Cell. Biol. 25, 8179–8190.
Wang, H., Cao, R., Xia, L., Erdjument-Bromage, H., Borchers, C., Tempst, P. & Zhang, Y. (2001a) Purification and functional characterization of a histone H3-lysine 4-specific methyltransferase. Mol. Cell 8, 1207–1217.[CrossRef][Medline]
Wang, H., Huang, Z.Q., Xia, L., Feng, Q., Erdjument-Bromage, H., Strahl, B.D., Briggs, S.D., Allis, C.D., Wong, J., Tempst, P. & Zhang, Y. (2001b) Methylation of histone H4 at arginine 3 facilitating transcriptional activation by nuclear hormone receptor. Science 293, 853–857.
Watts, D.J. & Strogatz, S.H. (1998) Collective dynamics of "small-world" networks. Nature 393, 440–442.[CrossRef][Medline]
Wright, P.E. & Dyson, H.J. (1999) Intrinsically unstructured proteins: Re-assessing the protein structure-function paradigm. J. Mol. Biol. 293, 321–331.[CrossRef][Medline]
Zhang, Y., Griffin, K., Mondal, N. & Parvin, J.D. (2004) Phosphorylation of histone H2A inhibits transcription on chromatin templates. J. Biol. Chem. 279, 21866–21872.
Zhao, K., Harshaw, R., Chai, X. & Marmorstein, R. (2004) Structural basis for nicotinamide cleavage and ADP-ribose transfer by NAD+-dependent Sir2 histone/protein deacetylases. Proc. Natl. Acad. Sci. USA 101, 8563–8568.
Zhou, Y. & Grummt, I. (2005) The PHD finger/bromodomain of NoRC interacts with acetylated histone H4K16 and is sufficient for rDNA silencing. Curr. Biol. 15, 1434–1438.[CrossRef][Medline]
Zhu, P., Zhou, W., Wang, J., Puc, J., Ohgi, K.A., Erdjument-Bromage, H., Tempst, P., Glass, C.K. & Rosenfeld, M.G. (2007) A histone H2A deubiquitinase complex coordinating histone acetylation and H1 dissociation in transcriptional regulation. Mol. Cell 27, 609–621.[CrossRef][Medline]
Received: 21 January 2009
Accepted: 1 April 2009
This article has been cited by other articles:
|
|
 |
|
  M. Sakamoto, S. Noguchi, S. Kawashima, Y. Okada, T. Enomoto, M. Seki, and M. Horikoshi Global analysis of mutual interaction surfaces of nucleosomes with comprehensive point mutants Genes Cells, November 1, 2009; 14(11): 1271 - 1330. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | ADVANCED SEARCH | TABLE OF CONTENTS |
