Histone modifications: Now summoning sumoylation

Sumoylation may be the signal that initiates attenuation followed by gene repression.

Histones can be modified in many ways to affect gene expression, including acetylation/deacetylation (1), phosphorylation (2), methylation (3), and ubiquitylation (4). Now, in this issue of PNAS, Shiio and Eisenman (5) report that sumoylation is yet another histone modification, and, interestingly, it may regulate transcriptional repression. Although there may appear to be a bewildering array of histone modifications involved with gene regulation, there could be a fairly simple paradigm underlying them.

Eukaryotic DNA is packaged within the nucleus through its association with histone proteins (H2A, H2B, H3, and H4), forming the fundamental repeating unit of chromatin, the nucleosome. The precise architecture of chromatin dictates whether it is permissive or resistant to transcription and other DNA-templated processes, such as replication, DNA repair, and recombination. Hence, it is not surprising that mechanisms that promote changes in chromatin structure are central to transcriptional regulation. One key pathway to alter chromatin structure involves covalent modifications of the histone tails.

Small ubiquitin-related modifier (SUMO) shares 18% identity with ubiquitin and adopts a similar 3D structure (6). Ubiquitylation has a role in protein degradation, whereas SUMO does not. The size of SUMO and ubiquitin (11 and 9 kDa, respectively) clearly distinguishes them from the other known posttranslation modifications of histones, which are all small chemical groups. SUMO and its ATP-dependent pathway of conjugation to substrates are con served in all eukaryotes investigated, yeast through humans.

One important consideration regarding sumoylation of histones is that its target residue, lysine, is a putative substrate for multiple modifying enzymes (Table 1). Acetylation of lysines is dynamically opposed by deacetylating enzymes. Typically, histone acetylation is activating and histone deacetylation is repressing, and, in a symmetrical fashion, DNA-bound activators and repressors recruit these enzymes to target genes (1). By contrast, lysine methylation on histones appears to be quite stable (no demethylases have been reported despite considerable effort to identify them) and may, in fact, provide a ”memory” mark (3). Consistent with its increased stability, lysine methylation can, depending on context, contribute a ”permanent” mark of either open, active euchromatin or closed, repressed heterochromatin (7). Perhaps, then, it is not surprising that acetylation or methylation can occur on the same lysine residues to allow dynamic regulation by acetylation or more stable regulation by methylation.

Table 1. Transcription-related modifications of lysine residues in histone proteins.

Histone modification	Modifying enzymes	Demodifying enzymes	Primary transcriptional effect	Nature of modification
Acetylation	HATs	HDACs	Activation	Dynamic
Methylation	HMTs	???	Activation/repression	Static
Ubiquitylation	Rad6*	Ubiquitin proteases (UBPs)	Activation	Dynamic
Sumoylation	UBC9*	ULP-related proteases†	Repression	Dynamic

^*Although other enzymes are involved in sumoylation and ubiquitylation, UBC9 (yeast through human) is the only known E2 SUMO-conjugating enzyme, and yeast Rad6 is the ubiquitin E2 known to monoubiquitylate histones in vivo

^†Although histone desumoylation has not yet been characterized, studies with other sumoylated substrates demonstrate deconjugation by ULP proteases

The current study by Shiio and Eisenman (5) on histone sumoylation and recent reports on histone ubiquitylation (4, 8-10) bring up many questions of how these much larger polypeptides function within chromatin; i.e., are they activating or repressing, are they dynamic or static, do they occupy the same lysines, and, finally, do they oppose one another? Shiio and Eisenman provide some provocative observations and some initial answers to these questions.

SUMO has been identified bound to many proteins, but to understand histone sumoylation, it may be most informative to examine the known effects of sumoylation on DNA-binding transcription factors. Sumoylation has been shown to stimulate activity; notable examples include heat shock factor HSF1 and tumor suppressor p53 (11-13). However, sumoylation most frequently correlates with decreased transcriptional activity (e.g., Elk1, Sp3, c-Myb, and c-Jun) (11) and, thus, repression of target genes. Interestingly, Shiio and Eisenman's study indicates that sumoylation of histone H4 also correlates with transcriptional repression, at least within an artificial transfected reporter model. First, they establish that mammalian histone H4 is sumoylated both in vivo and in vitro, and this modification appears to be more efficient than sumolylation of the other core histones. They show that expression of a reporter is decreased by targeting UBC9, which conjugates SUMO to its substrate, to the reporter. Moreover, the targeted UBC9 results in reduced levels of promoter-associated acetylated histone H3 and marked elevation in heterochromatic protein 1 (HP1). Notably, HP1 is a key structural protein of heterochromatin and binds to methylated Lys-9 on histone H3. HP1 can also contribute to repression of individual euchromatic genes (14).

The amino-terminal tail of histone H4 contains five lysines, all of which may be candidates for sumoylation. However, none of the histone proteins, including histone H4, contain the putative consensus sequence for sumoylation (γ-Lys-XGlu, where γ is a large hydrophic residue and X is any amino acid). In addition, as described above, many of these lysine residues may undergo other modifications (Table 1). Consequently, it is exceedingly difficult to study the physiological significance of sumoylation. One solution to this problem, used by Shiio and Eisenman as well as by others studying sumoylation and ubiquitylation of nonhistone substrates, is to genetically fuse SUMO to the putative target protein. Thus, Shiio and Eisenman observe that SUMO-H4 associates with chromatin and can be coimmunoprecipitated with endogenous histone deacetylase 1 (HDAC1) and HP1. These data provide strong, but indirect, evidence that sumoylation of H4 associates with agents mediating gene repression. Possibly analogous to this is a recent study proposing that p300/CBP can repress transcription by recruiting HDAC6 by way of SUMO (15).

Based on Shiio and Eisenman's observations and the previous understanding of the role of histone modifications in gene regulation, a model can be envisaged for the role of sumoylation in the context of the complex interplay of activating and repressing factors at the promoter (Fig. 1). Gene activation correlates with histone acetylation by histone acetyltransferases, which are recruited within coactivator complexes to promoters by DNA-bound activators. Once the gene has been transcribed, then its activity must be attenuated and then finally repressed. The signal for recruitment of sumoylating enzymes may be acetylation itself, as suggested by the authors' observation that H4 sumoylation increases with increasing H4 acetylation. The next step is HDAC-mediated removal of acetyl groups, recruited by DNA-bound repressors. Repression then is caused by histone methyltransferase (HMT)-mediated methylation, which is required for binding of HP1, in turn providing the structural element for chromatin condensation. Thus, because Shiio and man's data place sumoylation as an event before HDAC/HP1 recruitment, sumoylation may actually be the signal that initiates a sequence of attenuation followed by repression.

Fig. 1.

Model for sumoylation function in transcription. Horizontal line represents a gene with a TATA box-containing promoter and ORF; ovals represent histone octamers/nucleosomes. Through a coactivator, a DNA-bound activator can recruit a histone acetyltransferase (HAT) that acetylates histones and promotes chromatin structure amenable to transcription. This acetylation can potentially recruit SUMO-conjugating enzymes (E2/E3) capable of modifying either histones or activators to give an attenuating effect. A corepressor and HDAC activity could then be recruited by a DNA-bound repressor (possibly even with SUMO contributing to the interaction), deacetylating histones, and making way for the addition of repression-specific methylation marks, such as H3 K9-methyl, by an HMT. Finally, methylated histones (and possibly SUMO) would recruit HP1, contributing to chromatin structure in a static repressed state.

One immediate question relevant to this model is whether histone sumoylation does indeed lead to methylation by way of deacetylation, as suggested by interaction of SUMO-H4 with HDAC and HP1 and recruitment of HP1 to the UBC9-repressed reporter. Interestingly, it was recently shown that both histone deacetylation and histone methylation activities are present in one corepressor complex (16). Thus, it is possible that sumoylation triggers direct association of both activities residing in a single complex. One easily testable question is whether histone H3 Lys-9 methylation occurs at the UBC9-repressed promoter before association of HP1.

Further research is required to determine whether histone sumoylation is relevant to repression of endogenous genes under physiological conditions. Indeed, it is critical to demonstrate sumoylation of H4 at relevant genes and to examine by time course experiments how sumoylation relates to RNA expression. In terms of the mechanisms involved, it is critical to understand the result of sumoylation and indeed, of ubiquitylation as well. Specifically, do these large polypeptide modifications cause direct structural changes in chromatin, or do they lead to the binding of effector proteins, similar to methyl-lysine associating with HP1? In the case of H4 sumoylation presented herein, the modification may directly recruit a complex possessing HDAC and HMT activity.

Finally, a very attractive idea is that the various histone modifications that occur during a biological process constitute a ”histone code” (17, 18). There are now several examples of modification patterns and sequences that relate to gene activation, some of which occur on the same histone tail or on the same amino acid (19). There is recent evidence that histone ubiquitylation is involved in gene activation (8, 9, 20). Thus, if ubiquitylation/sumoylation of histones function to activate/repress, respectively, it will be important to determine whether they occur on the same lysine residues and whether they, in a simple reciprocal fashion, oppose one another's activity. Such an antagonistic relationship occurs on both IκBα (20) and proliferating cell nuclear antigen (PCNA) (21), where ubiquitin and SUMO modify the same lysine residue. Despite the aforementioned challenges, it would be informative to map the specific lysine residues in histones that are targeted by SUMO, to determine how this modification may fit into the histone code.

Thus, histone modifications are an expanding network, and lysine has emerged as a key residue for a number of alternative modifications. Although the ”code” appears to be ever more complex, eventually it may coalesce into a coherent pathway, once we fully understand modification sequences, their relative transience or permanence, and, most revealing, their physiological relevance. Histone H4 sumoylation has thus emerged as a new potential repressive mark, used in a transient manner as a possible counterbalance to activating marks.