Report

TAF1 Activates Transcription by Phosphorylation of Serine 33 in Histone H2B

See allHide authors and affiliations

Science  14 May 2004:
Vol. 304, Issue 5673, pp. 1010-1014
DOI: 10.1126/science.1095001

This article has been retracted. Please see:

Abstract

Dynamic changes in chromatin structure, induced by posttranslational modification of histones, play a fundamental role in regulating eukaryotic transcription. Here we report that histone H2B is phosphorylated at evolutionarily conserved Ser33 (H2B-S33) by the carboxyl-terminal kinase domain (CTK) of the Drosophila TFIID subunit TAF1. Phosphorylation of H2B-S33 at the promoter of the cell cycle regulatory gene string and the segmentation gene giant coincides with transcriptional activation. Elimination of TAF1 CTK activity in Drosophila cells and embryos reduces transcriptional activation and phosphorylation of H2B-S33. These data reveal that H2B-S33 is a physiological substrate for the TAF1 CTK and that H2B-S33 phosphorylation is essential for transcriptional activation events that promote cell cycle progression and development.

Transcription initiation in eukaryotes involves dynamic changes in chromatin structure that permit assembly of the transcription machinery at a gene promoter (1, 2). The fundamental structural unit of chromatin is the nucleosome, which contains 146 base pairs of DNA wrapped around a histone octamer composed of two copies each of histones H2A, H2B, H3, and H4 (3). Distinct patterns of histone modifications (e.g., acetylation, phosphorylation, and methylation) may act as “modification cassettes” that facilitate DNA-dependent events (4, 5). For example, in vertebrates phosphorylation of H2B Ser14 is associated with apoptotic chromatin, and in all eukaryotes phosphorylation of H3 Ser10 is associated with transcriptionally active and mitotic chromatin (46). Although all histones are phosphorylated in vivo, the function of many of these modifications and the kinases that carry them out are not known (7).

With the use of an in vitro kinase assay, we found that the Drosophila general transcription factor (GTF) TFIID phosphorylates histone H2B but not H1, H2A, H3, or H4 (Fig. 1, A and B) (8). TFIID is a multiprotein complex composed of the TATA box–binding protein (TBP) and numerous TBP-associated factors (TAFs) (9). TFIID functions during transcription initiation by nucleating assembly of GTFs and RNA polymerase II at the promoter. TAF1 (formerly TAFII250) is the only TFIID subunit that possesses kinase activity, suggesting that it phosphorylates H2B (1012). In fact, recombinant TAF1 and denatured and renatured recombinant TAF1 phosphorylated H2B in vitro, demonstrating that TAF1 has intrinsic, H2B-specific kinase activity (Fig. 1, B and C) (8). Collectively, these results indicate that TAF1 alone and in the context of TFIID phosphorylates H2B.

Fig. 1.

The TAF1 CTK phosphorylates histone H2B. (A) Silver-stained gel of immunopurified TAF1 and endogenous Drosophila TFIID. The asterisk indicates the position of the antibody heavy chain. (B) Coomassie blue–stained gel (left) and corresponding autoradiogram (right) of in vitro kinase assays programmed with γ32P-ATP and recombinant histones (H1, H2A, H2B, H3, and H4) in the presence (+) or absence (–) of immunopurified TAF1 or TFIID. (C) Coomassie blue–stained gel (top) and corresponding autoradiogram (bottom) of in vitro kinase assays as in (B), but containing the indicated recombinant histones and denatured and renatured TAF1. (D) Schematic representation of TAF1 and the CTK. Positions of the NTK domain, the HAT domain, the ubiquitin activating and conjugating domain (E1/E2), the DBD (B1 and B2), and the CTK are indicated. (E) Coomassie blue–stained gel (top) and corresponding autoradiogram (bottom) of in vitro kinase assays as in (B), except that reactions were programmed with recombinant CTK or NTK and either H2B or RAP74.

TAF1 contains two kinase domains, an N-terminal (NTK, amino acids 1 to 496) and a C-terminal (CTK, amino acids 1496 to 2132) domain (10) (Fig. 1D). In vitro, the NTK and the CTK autophosphorylate and the NTK transphosphorylates the RAP74 subunit of the GTF TFIIF. To determine which domain phosphorylates H2B, we assayed NTK and CTK separately in vitro. In contrast to the NTK, the CTK did not phosphorylate RAP74 but strongly phosphorylated H2B, indicating that the CTK possesses H2B-specific kinase activity (Fig. 1E).

Protein kinases contain two essential functional motifs, an adenosine triphosphate (ATP) binding motif and an amino acid–specific kinase motif. Computational sequence comparison analyses identified a putative serine and threonine (S/T) kinase motif (amino acids 1534 to 1546) and two tandem ATP binding domains (amino acids 1747 to 1780) in the CTK (Fig. 2A and fig. S1) (8, 13). To test whether the identified motifs mediate H2B phosphorylation, we performed in vitro kinase assays with the use of CTK polypeptides lacking the S/T kinase motif (CTKΔ1600) or the ATP binding motifs (CTKΔATP). Relative to the wild-type CTK, CTKΔ1600 and CTKΔATP weakly phosphorylated H2B (Fig. 2B and fig. S2A). To confirm the role of the S/T kinase motif, we mutated a catalytically important aspartic acid to an alanine (D1538A) in the motif (Fig. 2A and fig. S1) (13). Like CTKΔ1600, CTK(D1538A) exhibited weak autophosphorylation and H2B transphosphorylation activities (Fig. 2B). Interestingly, the S/T kinase motif is located in the first bromodomain of the double bromodomain module (DBD), which binds acetylated lysines (Fig. 2A) (14). However, CTK(D1538A) retains the ability to bind acetylated H4 in vitro, indicating that the introduced mutation disrupts kinase activity rather than acetylated lysine-based substrate recognition (fig. S2B). Thus, the identified S/T kinase and ATP binding motifs of the TAF1 CTK are essential for H2B phosphorylation.

Fig. 2.

The TAF1 CTK phosphorylates Ser33 in H2B. (A) Schematic representation of TAF1 and TAF1 derivatives. The positions of enzymatic domains are indicated. The position and amino acid sequence (30) of the S/T kinase motif and the positions of the ATP-binding motifs and CTK(D1538A) mutation are indicated. (B) Autoradiograms of in vitro kinase assays containing γ32P-ATP, native nucleosomes (top) or recombinant H2B (bottom), and TFIID, TAF1, or TAF1 derivatives. (C) Coomassie blue–stained gel (top) and corresponding autoradiogram (bottom) of in vitro kinase assays programmed with γ32P-ATP, CTK, and H2B, tailless H2B core domain (H2B-core), or H2BT peptide. (D) Schematic representation of H2BT (top). Serines (S) are highlighted. Autoradiograms (A) and corresponding Coomassie blue–stained gels (C) of in vitro kinase assays containing TAF1, γ32P-ATP, and increasing amounts of H2BT or H2BT derivatives (bottom).

To identify H2B residue(s) phosphorylated by the CTK, we first examined whether the CTK phosphorylates the N-terminal tail of Drosophila H2B (amino acids 1 to 39, H2BT) or the tailless H2B core domain (amino acids 40 to 123) and found that the CTK phosphorylated H2BT but not the H2B core domain (Fig. 2C). Next, to pinpoint which residue(s) in H2BT is phosphorylated, we generated mutant H2BT peptides in which alanines replaced all or individual serines or threonines. The CTK did not phosphorylate peptides lacking all serines, suggesting that it phosphorylates either Ser5 (H2B-S5) or Ser33 (H2B-S33) (fig. S2C). To test this, we used H2BT peptides as substrates that contained alanines in place of H2B-S5, H2B-S33, or both (H2BT-S5A, H2BT-S33A, and H2BT-S5/33A, respectively). The CTK phosphorylated H2BT-S5A but not H2BT-S33A or H2BT-S5/33A, indicating that H2B-S33 is the target of the CTK (Fig. 2D).

To investigate whether H2B-S33 is phosphorylated in vivo, we raised a polyclonal antibody recognizing phosphorylated H2B-S33 (H2B-S33P) (8). On Western blots, the antibody recognized H2BT containing H2B-S33P but not recombinant, unphosphorylated H2B or an H3 peptide (amino acids 1 to 32) containing phosphorylated Ser10 and Ser28 (Fig. 3A). In addition, the H2B-S33P antibody recognized H2BT and recombinant H2B that was phosphorylated in vitro by the CTK or TFIID, indicating that the antibody specifically recognizes phosphorylated H2B-S33 (Fig. 3B). The H2B-S33 antibody also recognized a protein with a molecular weight similar to that of H2B from histone preparations from Drosophila embryos or S2 cells, providing evidence that H2B-S33 is a target for phosphorylation in vivo (Fig. 3, C and D) (15, 16). To determine whether TAF1 mediates H2B-S33 phosphorylation in vivo, we used RNA interference (RNAi) to eliminate TAF1 expression in S2 cells (8). As shown by Western blot analysis, both TAF1 expression and H2B-S33 phosphorylation were reduced in TAF1 RNAi cells compared with mock RNAi cells, suggesting that TAF1 is a major H2B-S33 kinase in vivo (Fig. 3, D and E).

Fig. 3.

H2B Ser33 is phosphorylated in Drosophila. (A) Coomassie blue–stained gel (top) and corresponding Western blot (bottom) of purified nucleosomes, H2BT, H2BT-S33P peptide (phosphorylated at Ser33), H3T peptide, H3T-S10/28P peptide (phosphorylated at Ser10 and Ser28), and recombinant H2B. Phosphorylation of H2B-S33 was monitored with the H2B-S33P antibody. (B) Western blots of in vitro kinase assays programmed with ATP, recombinant H2B, and TAF1 CTK, TBP, or TFIID and probed with the H2B-S33P antibody. (C) Coomassie blue–stained gel (top) and corresponding Western blot (bottom) of nucleosomal histones purified from embryos and recombinant H2B and probed with the H2B-S33P antibody. (D) Coomassie blue–stained gel (left) and corresponding Western blots (middle and right) of histone octamers purified from mock (+) or TAF1 (–) RNAi S2 cells. The same Western blot was probed with the H2B-S33P antibody, stripped, and reprobed with an H2B antibody as a loading control. (E) Western blots of whole cell extracts from mock(+) or TAF1 (–) RNAi S2 cells probed with antibodies to TAF1 (top) or SIN3 (bottom).

Flow cytometry analysis of TAF1 RNAi cells revealed that loss of TAF1 results in G2-M phase cell cycle arrest (fig. S3). To test the hypothesis that TAF1 controls the transcription of genes whose activities contribute to G2-M progression, we used microarray expression profiling and reverse transcription polymerase chain reaction (RT-PCR) to monitor transcription in mock and TAF1 RNAi cells. Both methods showed that transcription of string (stg), which encodes a Drosophila homolog of yeast Cdc25, was reduced (Fig. 4A) (8). The Stg protein phosphatase is predominantly expressed during G2 and activates the cell cycle by dephosphorylating Cdc2 (17). Because loss of stg from S2 cells by RNAi causes G2-M arrest, TAF1 may regulate G2-M progression by activating stg transcription (18).

Fig. 4.

Transcription activation in Drosophila coincides with TAF1-mediated phosphorylation of H2B-S33. (A) Photographs of ethidium bromide–stained agarose gels showing RT-PCR products for stg (left) and actin5C (right) transcripts in mock (+) and TAF1 (–) RNAi S2 cells. (B) Photographs of ethidium bromide–stained agarose gels showing PCR products for the stg promoter (P) or coding region (CR) in mock (+) and TAF1 (–) RNAi S2 cells. In vivo cross-linked chromatin was immunoprecipitated with the indicated antibodies or rabbit preimmune serum (control). Input represents the amount of stg promoter present in 0.1% of the chromatin used for XChIP. (C) Gt transcription in heterozygous mutant cad (top), homozygous mutant TAF1CTK (middle), or heterozygous mutant cad and homozygous mutant TAF1CTK (bottom) blastoderm-stage Drosophila embryos. Gt transcription was detected by in situ hybridization with gt anti-sense RNA. Anterior is to the left and dorsal is up. Arrows indicate the position of the posterior gt transcription domain. (D) Photographs of ethidium bromide–stained agarose gels showing PCR products for the gt promoter (P) or coding region (CR) in embryos described in (C). In vivo cross-linked chromatin was isolated from the posterior halves of embryos and immunoprecipitated with the indicated antibodies or rabbit preimmune serum (control). Input represents the amount of gt promoter present in 2% of the chromatin used for XChIP.

Chromatin immunoprecipitation (XChIP) was used to establish whether there is a direct correlation between transcriptional activation of stg and TAF1-mediated phosphorylation of H2B-S33 at the stg promoter (15). Cross-linked chromatin was isolated from mock and TAF1 RNAi S2 cells and immunoprecipitated with TAF1 or H2B-S33P antibodies. Immunoprecipitated DNA was purified and used as a template for PCR to detect the stg promoter or coding region and actin5C promoter (Fig. 4B and fig. S4). In contrast, TAF1 is not essential for actin5C transcription, and H2B-S33P antibodies do not precipitate the actin5C promoter (Fig. 4A and fig. S5A). Thus, the transcriptional dependence of a gene for TAF1 is correlated with H2B-S33 phosphorylation, not with TAF1 association.

To distinguish whether loss of H2B-S33 phosphorylation at the stg promoter is due directly to loss of TAF1 or indirectly to G2-M arrest, we performed XChIP analysis on S2 cells arrested in G2-M by RNAi of the SIN3 transcriptional corepressor. Stg transcription is repressed in SIN3 RNAi cells, yet the stg promoter remains associated with H2B-S33P and TAF1, indicating that loss of H2B-S33 phosphorylation in TAF1 RNAi cells is because of elimination of TAF1 rather than G2-M arrest (fig. S5B) (18).

In addition to kinase domains, TAF1 has a histone acetyltransferase (HAT) domain that acetylates H3 Lys14 (H3-K14) and unidentified lysines in H4 in vitro (19). XChIP analysis detected acetylated H3-K14 and H4 at the transcriptionally active stg promoter in mock RNAi cells but not at the inactive stg promoter in TAF1 RNAi cells (Fig. 4B and fig. S5C). In contrast, TAF1-independent histone modifications did not correlate with activation of stg in mock and TAF1 RNAi cells (fig. S5C). Taken together, these results indicate that TAF1-mediated phosphorylation of H2B-S33 and acetylation of H3 and H4 potentiate transcriptional activation in Drosophila cells.

To investigate the role of TAF1-mediated phophorylation of H2B-S33 during fly development, we used a recessive lethal TAF1 allele, TAF1CTK, which contains a nonsense mutation at amino acid 1728 that truncates the CTK downstream of the DBD (Fig. 2A and fig. S6A) (8). The corresponding protein (TAF1ΔCTK) is expressed in Drosophila but presumably does not have CTK activity, because it does not phosphorylate H2B in vitro (Fig. 2B and fig. S6, B and C). In situ hybridization was used to monitor transcription in embryos homozygous mutant for TAF1CTK and heterozygous mutant for the maternal activator Caudal (Cad) (20). In this genetic background, transcription of the gap gene giant (gt) was reduced (Fig. 4C). Gt is transcribed in two domains along the anterior-posterior axis of blastoderm-stage embryos. Transcription of the posterior gt domain (pgt) is Cad-dependent, whereas transcription of the anterior gt domain (agt) is Cad-independent. Relative to controls (cad/+ or TAF1CTK), pgt transcription was reduced in cad/+;TAF1CTK embryos (Fig. 4C).

XChIP analysis was used to examine whether TAF1-mediated phosphorylation of H2B-S33 contributes to pgt transcription. Cross-linked chromatin was isolated from the posterior halves of cad/+;TAF1CTK and control embryos and immunoprecipitated with antibodies to H2B-S33P, acetylated histones, or TAF1. PCR analysis detected H2B-S33P at the transcriptionally active gt promoter in control embryos, but not at the transcriptionally repressed promoter in cad/+;TAF1CTK embryos (Fig. 4D and fig. S4). To monitor TAF1 binding, we used two antibodies, TAF1-M and TAF1-C, which recognize the middle domain and the CTK of TAF1, respectively (fig. S6A). Both antibodies precipitated the gt promoter from control embryos, indicating that TAF1ΔCTK and maternally contributed, wild-type TAF1 are present at the gt promoter in the pgt (Fig. 4D and fig. S4). In contrast, although the TAF1-M antibody precipitated the gt promoter from cad/+;TAF1CTK embryos, TAF1-C did not. Because TAF1ΔCTK is present at a higher concentration in cad/+;TAF1CTK embryos than maternal TAF1, this result indicates that TAF1ΔCTK is preferentially recruited to the gt promoter in the pgt (Fig. 4D and fig. S6C). This result is supported by the presence of TAF1-mediated histone acetylation at the transcriptionally silent gt promoter. Thus, TAF1-mediated phosphorylation of H2B-S33 contributes to transcriptional activation during Drosophila embryogenesis.

Ser33 is the only evolutionarily conserved serine or threonine in the N-terminus of metazoan H2Bs (fig. S7). In the crystal structure of the Xenopus laevis nucleosome, the equivalent serine links the H2B DNA-binding N-terminal tail to the histone fold domain (3, 21). Thus, replacing the hydroxyl group on Ser33 with a bulkier, negatively charged phosphate group may drastically affect H2B tail interactions with DNA. This is important because the H2B tail regulates nucleosome mobility. Deletion of the tail bypasses the requirement for the SWI/SNF nucleosome-remodeling complex in yeast, and the tail is critical for maintaining the position of histone octamers in in vitro sliding assays (22, 23). These findings support a model in which TAF1-mediated phosphorylation of H2B-S33 disrupts DNA-histone interactions, resulting in local decondensation of chromatin. Decondensation may trigger chromatin remodeling and formation of a chromatin structure that facilitates assembly of other GTFs at a promoter, a function that is primarily attributed to TFIID (1, 2, 10).

Our data indicates that the S/T kinase motif of the CTK is located in the DBD. In the crystal structure of the DBD, the position of the S/T kinase motif does not overlap with the acetylated lysine-binding surface of the DBD, suggesting that it is an independent functional unit of the DBD (14). Members of the fsh/RING3 (BET) family of DBD proteins have kinase activity, suggesting that TAF1 is a member of a kinase family whose catalytic motif resides within the DBD (24, 25).

Phosphorylation of H2B-S33 by TAF1 is essential for transcriptional activation of stg/cdc25 and, consequently, cell cycle progression. Similarly, depletion of yeast TAF5, human TAF2, or a twofold reduction in chicken TBP results in G2-M arrest (2628). Like TAF1, TBP regulates stg/cdc25 expression, providing support for the finding that the H2B-S33 kinase activity of TAF1 occurs in the context of TFIID (28). Interestingly, depletion of yeast TAF1, which does not possess a CTK, and inactivation of TAF1 HAT activity induce G1 arrest because of reduced transcription of B- and D-type cyclins, respectively (26, 29). Thus, loss of all TAF1 activities causes G2-M arrest whereas loss of TAF1 HAT activity causes G1 arrest, suggesting gene-specific requirements for TAF1 CTK and HAT activities. In contrast, the presence of phosphorylated H2B-S33 and acetylated H3 and H4 at the stg and gt promoters implies that TAF1 CTK and HAT activities can cooperate in transcriptional activation of some genes. This proposal is supported by the finding that loss of H2B-S33P from the gt promoter results in reduced transcription, despite the presence of TAF1-mediated histone acetylation. Thus, TAF1-mediated phosphorylation of H2B-S33 may work in concert with other TAF1-mediated histone modifications, H1 ubiquitination, and H3 and H4 acetylation to contribute to the chromatin-based mechanisms underlying transcription activation of eukaryotic genes.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5673/1010/DC1

Materials and Methods

Figs. S1 to S7

References and Notes

References and Notes

View Abstract

Navigate This Article