Immediate-early gene regulation by interplay between different post-translational modifications on human histone H3

Nucleosomes are the main organizational modules of chromatin and histones are their main protein component. The high conservation of histones throughout evolution attests the basic nature of the nucleosomal design [Tsunaka et al., 2005]. Regulation of gene transcription preferentially occurs by way of post-translational modification (PTM) of the histone in amino terminal tails located outside the compact chromatin structure, as for instance, in the histone 3 (H3) protein [Cheung et al., 2000a]. Several PTMs of histones, namely phosphorylation, acetylation, methylation, and O-GlcNAc modification, regulate the contact of chromatin with DNA [Cheung et al., 2000a]. These PTMs form the basis of a histone code, a specific code that facilitates diverse cellular responses, involving gene expression and orderly completion of the cell cycle [Cheung et al., 2000a; Cosgrove and Wolberger, 2005]. In particular, phosphorylation of H3 and of several transcription factors has been found to closely correlate with immediate-early (IE)-gene transcription under diverse conditions of induction [Thomson et al., 1999; Clayton and Mahadevan, 2003].

The nucleosome response involves alterations in chromatin and nucleosome structure, relies on histone modifications, and is associated with the induction of different genes [Cheung et al., 2000a] including IE-gene transcription [Thomson et al., 1999]. The transcription of IE genes is transiently activated within minutes of cell exposure to a wide range of extracellular stimuli [Thomson et al., 1999]. IE genes encode transcription factors, such as the promoter-specific factor 1 (Sp1) [Chen et al., 1994], activator protein 1 (AP-1) [Angel et al., 1988; Fisch et al., 1989; Herr et al., 1994], and c-AMP-response element-binding protein (CREB) [Gonzales and Bowden, 2002], DNA-binding proteins and proto-oncogene proteins like c-Jun that regulate cell proliferation and apoptosis [Wisdom et al., 1999]. These transcription factors and H3 (on Ser 10 and 28), are phosphorylated by mitogen-activated protein kinases (MAPKs) or their effector kinases such as mitogen- and stress-activated kinases, and the phosphorylated proteins are involved in the induction of several IE genes [Deak et al., 1998; Seassone-Corsi et al., 1999; Clayton et al., 2000; Zhong et al., 2001; Duncan et al., 2006].

Both Ser 10 and 28 are preceded by Lys at −1 position, a residue not found very often in the vicinity of phosphorylated Ser [Iakoucheva et al., 2004; Qazi et al., 2006]. The position of Lys immediately before a phosphorylated Ser appears to be related with its methylation in this particular context. Interestingly, methylated Lys 9 mediates gene silencing and methylated Lys 27, gene repression [Lindroth et al., 2004]. Furthermore, an interplay between methylated and phosphorylated neighboring amino acid residues (Lys 9/Ser 10 and Lys 27/Ser 28) known as “phosphorylation/methylation switching” has been reported in H3 [Wang et al., 2004]. Clearly, the structural motifs consisting of Lys 9 and Ser 10, and Lys 27 and Ser 28 are functionally important.

Amongst the different PTMs, one of the dynamic and regulatory modifications of hydroxyl function of Ser/Thr is the O-GlcNAc modification, which influences protein folding, localization and trafficking, solubility, antigenicity, biological activity, and half-life, as well as cell–cell interactions [Love and Hanover, 2005]. Interplay between O-GlcNAc modification and phosphorylation on the same or neighboring Ser/Thr residues has been observed in several nuclear and cytoplasmic proteins [Comer and Hart, 2000; Wells et al., 2003]. The dynamic O-GlcNAc modification can regulate gene transcription by glycosylating transcription factors like Sp1 [Majumdar et al., 2003] and CREB [Lamarre-Vincent and Hsieh-Wilson, 2003].

Interplays of different PTMs on the same or neighboring residues are known to occur in proteins [Khidekel and Hsieh-Wilson, 2004], and may either facilitate or prevent other modifications, thereby regulating the function of the modified protein. Recently, it has been suggested that an interplay between O-GlcNAc modification and phosphorylation of H3 is involved in the regulation of the cell cycle in mammals [Kaleem et al., 2006], emphasizing the importance of PTMs on proteins that control gene regulation.

The specific combination of different PTMs may provide a basis for H3 to perform multiple functions, and computational methods may help evaluating H3 multifunctionality. Furthermore, these methods have an advantage of being fast, reproducible, and 70–80% accurate [Nielsen et al., 1999]. Several computational methods have been developed to predict glycosylation and phosphorylation sites in proteins. These include NetPhos 2.0 [Blom et al., 1999] and YinOYang 1.2 (unpublished). Most of these prediction methods that compute modification potential are neural network based and recognize specific sequence content through prior learning process. Amino acids involved in maintaining the 3D structure of a protein and hence its functions, have often been found to be highly conserved evolutionarily [Schueler-Furman and Baker, 2003] and interplay of phosphorylation and O-GlcNAc modification on conserved Ser/Thr residues has been proposed to act at key functional sites [Ahmad et al., 2006].

Available prediction, in silico, data for different PTMs suggest that a complex interplay or a specific combination of these PTMs may regulate repression or induction of different genes, including IE genes. When IE genes are ready for transcription, H3 is phosphorylated on Ser 10 [Thomson et al., 1999], methylated on Lys 4 [Hazzalin and Mahadevan, 2005], and acetylated on Lys 9 [Hazzalin and Mahadevan, 2005] and/or Lys 14 [Cheung et al., 2000b]. We propose that when H3 is O-GlcNAc modified on Ser 10, it may result in deacetylation of Lys 9, which consequently becomes methylated. Thus, a combination of O-GlcNAc modification of Ser 10 and methylation of Lys 9 may result in IE-gene repression.

MATERIALS AND METHODS

RESULTS

O-Linked Phosphorylation Sites in Human H3

The results of predictions of phosphorylation sites in human H3 performed by NetPhos 2.0 are given in Table I, and graphically presented in Figure 1. All of the potentially predicted Ser and Thr phosphorylation sites were conserved in vertebrate and in invertebrates as well (Fig. 2). No Tyr residues were predicted to be phosphorylated in human H3.

Table I. Predicted Phosphorylation, O-GlcNAc Modification, and Yin Yang Sites in Human H3

Residue no.	Experimental evidence		Prediction of modification potential
	Phosphorylation	O-GlcNAc modification	NetPhos 2.0	YinOYang 1.2	Yin Yang site
Ser 10	Zhang et al. 2003	By similaritya	+	+	+
Ser 28	Zhang et al. 2003	By similaritya	+	+	+
Ser 57	−	−	+	+	+/−
Ser 86	−	−	−	+	−
Thr 3	−	−	−	+	−
Thr 6	Zhang et al. 2003	By similaritya	+	+	+
Thr 11	Zhang et al. 2003	By similaritya	+	+	+
Thr 22	−	−	−	+	−
Thr 32	−	−	−	+	−
Thr 45	−	−	+	+	+
Thr 80	−	−	−	+	−
Thr 118	Zhang et al. 2003	By similaritya	+	+	+

• +, Positive prediction; −, negative prediction; +/− false/negative prediction.
• a Similarity in kinase and OGT recognition of same substrate site.

O-Linked Glycosylation Sites in Human H3

The prediction results of O-GlcNAc modification for human H3 by YinOYang 1.2 have been given in Table I and illustrated in Figure 3. All of the potentially predicted O-GlcNAc modification sites were conserved in vertebrates and in invertebrates (Fig. 2). Furthermore, human H3 showed a higher potential for O-GlcNAc modification compared to phosphorylation.

Yin Yang Sites in Human H3

Yin Yang sites in human H3 were predicted by YinOYang 1.2 and the results have been summarized in Table I and illustrated in Figure 4. All of these sites are of functional importance as these Ser/Thr residues can be modified by kinases as well as by OGT. Only one Ser at position 57 was identified as a false negative Yin Yang site (Table I, Fig. 4). All of the predicted and identified as false negative Yin Yang sites were found to be fully conserved in vertebrates and in invertebrates (Fig. 2).

It was also observed that the potential Ser phosphorylation sites in the N-terminal of H3 (Ser 10 and 28) contain the same sequence motif with Lys on −1 and Arg on −2 positions (Fig. 2).

Comparison of Human H3 With Human Histones H2A, H2B, and H4

The human histone H3 was aligned with the remainder core human histones to develop a relation between all four core histones. No appreciable sequence similarity was found in other core histones, except for H4, which showed highest sequence similarity as compared to H2A and H2B (Fig. 5).

Then it was investigated if the predicted sequence motif in human H3 (Table IV) also existed in the rest of core human histones. In human H2B, a similar sequence was found (KRS) at position Ser 33 and Ser 88 (Fig. 5). Phosphorylation and O-GlcNAc modification was predicted in H2B. As can be seen in Table VI, Ser 33 is predicted as Yin Yang site and Ser 88 showed potential for O-GlcNAc modification. When H2B was multiple aligned, it was found that Ser 33 is conserved in vertebrates, with a single substitution in Gallus gallus and Ser 88 was fully conserved in invertebrates and vertebrates (Fig. 6). Furthermore, both residues were predicted to be located in coiled regions by GOR IV (Fig. 7).

Table VI. Prediction of Potential for Phosphorylation and Glycosylation Sites in H2B

Phosphorylated residue	O-Glycosylated residue	Yin Yang sites
Ser 7, 15, 33, 37, 39, 56, 92, 113, 124	Ser 5, 7, 33, 88, 113, 124, 125	Ser 7, 33, 113, 124
Thr 89, 91, 97, 116	Thr 53, 120, 123
Tyr 38, 41, 122

The details of sequence alignment of H3 with H2A, H2B, and H4; phosphorylation, O-GlcNAc modification, and Yin Yang sites of H2B; secondary structure prediction in H2B and multiple alignment of H2B in vertebrates and invertebrates are shown in Table VI and Figures 5-7.

Comparison of the Sequence Motif of O-GlcNAc Modification Sites in Human H3 With Experimentally Known Proteins

The sequence motif of the predicted Yin Yang sites, Ser 10 and 28, utilizing YinOYang 1.2 in human H3, was compared with proteins with experimentally known O-GlcNAc modification sites. These results are given in Figure 8. These results showed a similar sequence in experimentally known glycosylated proteins compared to the sequence of human histone H3 at Ser 10 and 28.

DISCUSSION

The different PTMs of H3 result in structural and functional changes. The importance of O-GlcNAc modification in H3 functionality has been put forward and it is suggested in silico that the dynamic intracellular phosphorylation and O-GlcNAc modification of human H3 on Ser 10, together with acetylation and methylation, participate in the control of IE-gene induction.

Phosphorylation sites in bovine H3 have been identified, which include Thr 6 or 11 and 118, Ser 10 and 28 [Zhang et al., 2003]. These are also positive prediction sites for phosphorylation of human H3 (Table I). In addition, these sites have also been predicted positively for the O-GlcNAc modification, i.e., Yin Yang sites (Table I). These Ser/Thr residues of H3 are conserved in all members of vertebrates and invertebrates (Fig. 2), a finding that increases their potential to act as Yin Yang sites, where both phosphorylation and O-GlcNAc modification can occur. Though H3 is almost conserved in all diverse groups of organisms, the Ser/Thr residues, which possess higher potential for O-phosphate and O-GlcNAc modification, can be identified by the 3D structural region of that Ser/Thr.

During mitosis, H3 phosphorylation on Ser 10 is crucial for chromosome condensation and progression of the cell cycle [Prigent and Dimitrov, 2003]. However, this regulation of H3 phosphorylation is affected by other PTMs, such as acetylation and methylation, during interphase [Berger, 2001], result in activation or repression of genes [Berger, 2001]. Phosphorylation of Ser 10 enhances acetylation of Lys 14 [Lo et al., 2000; Cheung et al., 2000b]. In the c-fos promoter, phosphorylated Ser 10 and acetylated Lys 9 can coexist on the same N-terminal of H3 [Edmondson et al., 2002]. Methylation of H3 can mediate transcriptional gene silencing and repression [Bernstein et al., 2002]. Acetylation is rapidly reversible, while methylation is more persistent and can occur even after transcription ceases, providing a memory of a recent transcription. Methylation of different Lys residues of H3 produces different or even opposite gene responses [Strahl et al., 1999; Bernstein et al., 2002; Saccani and Natoli, 2002; Stewart et al., 2005]. In addition to phosphorylation of Ser 10 of H3, a combination of acetylation and methylation on Lys 4, 9, and 14 is important in the induction or repression of IE genes.

Generally, transcriptionally active or silenced genes are associated with distinct combinations of histone PTMs. O-GlcNAc modification, a dynamic modification, has been reported to play a crucial role in chromatin remodeling [Love and Hanover, 2005]. O-GlcNAc transferase (OGT), the enzyme that catalyzes the addition of an O-GlcNAc moiety to the backbone of the protein on Ser and/or Thr residues [Love and Hanover, 2005], is an ubiquitous regulator of transcription, and displays flexibility in recognizing its many substrates [Yang et al., 2002]. The O-GlcNAc modification of same protein may affect different genes differently as for transcription factor Sp1, and may result in different outcomes depending on the type of cell and cellular signaling [Comer and Hart, 1999]. The O-GlcNAc-modified Sp1 induces transcriptional activation in Hela cells, and represses transcription in vascular muscle cells [Comer and Hart, 1999]. Similarly, O-GlcNAc modification of different proteins may result in different gene regulation. For example, O-GlcNAc modification of a transcription factor PDX-1 results in increased DNA binding and hence increased insulin secretion [Gao et al., 2003], whereas, transcriptional inhibition of certain genes is associated with O-GlcNAc modification of transcriptome directly or indirectly through O-GlcNAc modification of the proteasome [Bowe et al., 2006]. This suggests that O-GlcNAc modification plays different and sometimes contrasting roles in the regulation of gene expression through an interplay with phosphorylation. The OGT is recruited to the promoter region by the mSin3A-HDAC1 complex [Yang et al., 2002], where it modifies promoter-bound proteins like histones, RNA-polymerase II, c-Fos, c-Jun, and other transcriptional activators and thus exerts its eukaryotic gene-silencing activity [Lamarre-Vincent and Hsieh-Wilson, 2003; Majumdar et al., 2003; Tai et al., 2004; Toleman et al., 2004] by adding O-GlcNAc moieties on Ser and Thr residues. In some instances, O-GlcNAc modification of proteins induces transcription like in the case of the transcription factor STAT5 [Gewinner et al., 2004]. When STAT5 is O-GlcNAc modified, it interacts with the CREB-binding protein CBP (CBP is a transcriptional co-activator with intrinsic histone acetyltransferase activity) and thereby induces transcription [Gewinner et al., 2004].

OGT [Yang et al., 2002] together with O-GlcNAcase [Toleman et al., 2004] affect gene transcription in mammals. It is well documented that OGT and kinase compete for the same substrate amino acid residue, Ser/Thr [Love and Hanover, 2005] and an interplay of phosphorylation and O-GlcNAc modification on Ser 10 and 28 is therefore most likely to occur. This interplay of O-GlcNAc modification and phosphorylation on Ser 10 and 28 may result in IE-gene regulation.

Lys 9 methylation inhibits Lys 4 methylation on H3 in heterochromatic gene silencing [Noma et al., 2001], and methylation of Lys 9 and 27 has been documented to be involved in gene silencing [Lindroth et al., 2004]. It is quite interesting that the basic amino acid, Lys, is preceded (on the left or at −1 position with reference to Ser 10/28) by both Ser 10 and 28, described as Yin Yang sites. According to earlier reports, O-glycosylation of Ser is favored by Pro or a small or neutral amino acid side chain [Christlet and Veluraja, 2001]. Similarly, Ser in close vicinity to Pro is favored for phosphorylation [Iakoucheva et al., 2004; Qazi et al., 2006]. A small fraction of phosphorylated Ser also show basic amino acid residue Lys on −1 position [Qazi et al., 2006]. Furthermore, MAPKs and its effector proteins are known to catalyze phosphorylation of Ser in close vicinity of basic amino acids [Barsyte-Lovejoy et al., 2002]. In H3, both Ser 10 and 28 are preceded by Lys and both these residues are highly conserved in all organisms (Fig. 2). We retrieved 103 phosphorylated protein sequences data from Phosphobase 3.0 [Diella et al., 2004], with 124 phosphorylated Ser and Lys at −1 position. Secondary structure prediction by GOR IV [Garnier et al., 1996; Combet et al., 2000] showed that the phosphorylated Ser residues with Lys at −1 position resides predominantly in coiled structural regions (Table II), whereas, only a fraction was found in the alpha helical region and a very small number of sites were found to be located in extended strands (Table II). Coiled structural regions may provide more space for phosphate modifications in the presence of a bulky Lys residue at −1 position. Both of Ser 10 and 28 of H3 are found in the coiled region, hence attachment of O-GlcNAc on Ser 10 and 28 by OGT can easily result in phosphorylation blockade.

Repeated sequence motifs from secondary structural data of 124 phosphorylation sites were extracted manually. It is striking that four sequence motifs were most frequent: RKS, KKS, PKS, and SKS. The motif RKS sequence is present in all selected sequences of H3 from different species for both Ser 10 and 28 (Fig. 2). Among the other 103 proteins, the most highly repeated pattern was found to be RKS and KKS (Table IV). Table III shows predicted kinases for 124 phosphorylation sites by NetPhosK 1.0 [Blom et al., 2004]. Of the 124 predicted phosphorylation sites, 46 were found in basic amino acid rich motifs (KKS and RKS). From these, 16 phosphorylation sites were predicted by NetPhosK 1.0 [Blom et al., 2004] to be catalyzed by different MAPKs, 11 were found in coiled regions (Table III). This means that 69% of MAPK-catalyzed phosphorylation of Ser residues can be expected in coiled regions. Methylation prediction by MeMo [Chen et al., 2006] showed that, among all the RKS, KKS, PKS, and SKS motifs from all known phosphorylated Ser residues of 103 proteins, only one instance of Lys was found to have potential for methylation in the PKS motifs, no Lys was observed to have potential in the SKS motif, whereas the highest number of Lys with a potential for methylation was found in the RKS and KKS motifs (Table V). Thus, basic residues at position −1 and −2 are preferred for phosphorylation and methylation of adjacent residues. Functional analysis of the 103 phosphorylated proteins on Ser accompanied with Lys at −1 position showed that most of these proteins are nucleus specific (Table V). When human histone H3 was aligned with human core histone H2A, H2B, and H4, they showed very low sequence similarity (Fig. 5), even though the core histones are highly conserved across their entire sequence. When the RKS motif was searched in all core histones, it was found in H3 of all organisms (Fig. 2) but in H2B this motif was found only in one organism Rhacophorus schlegelii (Fig. 6). A similar sequence KRS was identified in human H2B at Ser 33 and Ser 88. Both these sites showed to be conserved in mammals and exhibited a potential for O-GlcNAc modification. Furthermore, Ser 33 also showed potential for both phosphorylation and O-GlcNAc modification (Yin Yang site) (Table VI). Phosphorylation of Ser 33 is essential for transcriptional activation in eukaryotes. It is phosphorylated by the transcription factor TAF1 that is part of the protein complex TFIID in Drosophila [Maile et al., 2004]. Both residues, Ser 33 and Ser 88, were predicted to be located in coiled regions (Fig. 7) as predicted in H3 for Ser 10 and 28 (Table II). This suggests that the sequence motifs containing phosphorylated Ser in the vicinity of basic amino acids, Lys and Arg, at −1 and −2 positions, are important in gene regulation. The phosphorylated Ser 10 of human histone H3 is followed by the amino acids Thr at +1 and Gly and +2 positions (Fig. 2). In case of phosphorylated Ser 28, Ala at +1 and Pro at +2 positions are found (Fig. 2). These sequences, STG and SAP, were compared with experimentally known O-GlcNAc-modified proteins retrieved from the Swiss-Prot database [Boeckmann et al., 2003]. It was observed that several proteins contained a similar though not identical upstream sequence environment like that of Ser 10 (Fig. 8). Together, these results suggest that the sequence motif STG may provide space for OGT to add an O-GlcNAc moiety to the protein. Furthermore, these results indicate that O-GlcNAc modification is most likely to take place at Ser 10 (and Ser 28) of human histone H3.

On the basis of in silico data, we propose that a specific combination of different modifications (phosphorylation, acetylation, methylation, and O-GlcNAc modification) control the activation and repression of genes including the IE genes. It is quite obvious that methylation on Lys 9 results in IE-gene silencing, whereas phosphorylation on Ser 10, acetylation on Lys 9 and Lys 14 might regulate IE-gene induction, along with methylation on Lys 4, suggesting the following sequence of events: when phosphorylation of Ser 10 is blocked by the presence of O-GlcNAc modification, Lys 9 is methylated. Similarly, phosphorylation of Ser 10 and acetylation of Lys 9 and Lys 14 are involved in IE-gene activation. On the contrary, O-GlcNAc modification on Ser 10 and methylation on Lys 9 may lead to gene repression. Thus, a specific combination of different PTMs on Ser/Thr and Lys, involving Ser 10, regulate IE-gene expression and repression. In addition, the interplay of phosphorylation and O-GlcNAc modification emerges to be the regulator of other PTMs.