SUMO-1 Modification of Human Transcription Factor (TF) IID Complex Subunits

The TFIID complex is composed of the TATA-binding protein (TBP) and TBP-associated factors (TAFs) and is the only component of the general RNA polymerase II (RNAP II) transcription machinery with intrinsic sequence-specific DNA-binding activity. Binding of transcription factor (TF) IID to the core promoter region of protein-coding genes is a key event in RNAP II transcription activation and is the first and rate-limiting step of transcription initiation complex assembly. Intense research efforts in the past have established that TFIID promoter-binding activity as well as the function of TFIID-promoter complexes is tightly regulated through dynamic TFIID interactions with positive- and negative-acting transcription regulatory proteins. However, very little is known about the role of post-translational modifications in the regulation of TFIID. Here we show that the human TFIID subunits hsTAF5 and hsTAF12 are modified by the small ubiquitin-related modifier SUMO-1 in vitro and in human cells. We identify Lys-14 in hsTAF5 and Lys-19 in hsTAF12 as the primary SUMO-1 acceptor sites and show that SUMO conjugation has no detectable effect on nuclear import or intranuclear distribution of hsTAF5 and hsTAF12. Finally, we demonstrate that purified human TFIID complex can be SUMO-1-modified in vitro at both hsTAF5 and hsTAF12. We find that SUMO-1 conjugation at hsTAF5 interferes with binding of TFIID to promoter DNA, whereas modification of hsTAF12 has no detectable effect on TFIID promoter-binding activity. Our observations suggest that reversible SUMO modification at hsTAF5 contributes to the dynamic regulation of TFIID promoter-binding activity in human cells.

The TFIID complex is composed of the TATA-binding protein (TBP) and TBP-associated factors (TAFs) and is the only component of the general RNA polymerase II (RNAP II) transcription machinery with intrinsic sequence-specific DNA-binding activity. Binding of transcription factor (TF) IID to the core promoter region of protein-coding genes is a key event in RNAP II transcription activation and is the first and rate-limiting step of transcription initiation complex assembly. Intense research efforts in the past have established that TFIID promoter-binding activity as well as the function of TFIID-promoter complexes is tightly regulated through dynamic TFIID interactions with positive-and negative-acting transcription regulatory proteins. However, very little is known about the role of post-translational modifications in the regulation of TFIID. Here we show that the human TFIID subunits hsTAF5 and hsTAF12 are modified by the small ubiquitin-related modifier SUMO-1 in vitro and in human cells. We identify Lys-14 in hsTAF5 and Lys-19 in hsTAF12 as the primary SUMO-1 acceptor sites and show that SUMO conjugation has no detectable effect on nuclear import or intranuclear distribution of hsTAF5 and hsTAF12. Finally, we demonstrate that purified human TFIID complex can be SUMO-1-modified in vitro at both hsTAF5 and hsTAF12. We find that SUMO-1 conjugation at hsTAF5 interferes with binding of TFIID to promoter DNA, whereas modification of hsTAF12 has no detectable effect on TFIID promoter-binding activity. Our observations suggest that reversible SUMO modification at hsTAF5 contributes to the dynamic regulation of TFIID promoter-binding activity in human cells.
To initiate mRNA synthesis, RNA polymerase II (RNAP II) 1 must assemble in an ordered fashion with a set of general transcription factors, TFIIA, -IIB, -IID, -IIE, -IIF, and -IIH, to form a so-called preinitiation complex (PIC) at the core promoter region of protein-coding genes (1)(2)(3). The initial, and rate-limiting, step for PIC assembly is binding of TFIID, a multiprotein complex composed of the TATA-binding protein TBP and at least 13 TBP-associated factors (TAFs), to core promoter sequence elements (1)(2)(3)(4)(5). In addition to sequencespecific DNA binding, the TFIID complex provides several enzymatic activities that all reside in its largest subunit TAF1 (6). These include histone acetyltransferase activity (7), protein kinase activity (8), and ubiquitin-activating/-conjugating activity (9). How exactly TFIID enzymatic activities contribute to RNAP II transcription is still unknown.
In accordance with its crucial role in PIC assembly, TFIID is considered one of the key targets of transcription regulation pathways. Biochemical and genetic studies have shown that TFIID binding to promoters as well as the stability and functionality of TFIID-promoter complexes are subject to regulation by gene-specific activators and repressors (10,11) and by an array of ubiquitous regulators of TFIID (TBP) activity, including NC2 (Dr1/DRAP1), Mot1/BTAF1, and the NOT complex (11)(12)(13)(14). In addition, TFIID subunits are subject to post-transcriptional modifications. Earliest studies demonstrated that several subunits of TFIID are phosphorylated during mitosis and that TFIID isolated from mitotic cells fails to respond to transcription activators in vitro (15). Results of more recent studies demonstrated gene-specific effects upon hsTAF10 methylation by the SET9 protein methyltransferase (16) and that the C-terminal domain of TAF1 is a substrate for the protein kinase CKII (17). Exactly how and to what extent these TAF modifications modulate TFIID functions is not known.
In recent years, regulation of transcription factor activity through modification with SUMO proteins has attracted considerable attention (18 -21). SUMO proteins are small ubiquitinrelated modifiers that are conjugated to target proteins through an enzymatic pathway, which is similar to ubiquitylation, but which involves a distinct set of enzymes (22,23). Like ubiquitin, SUMO proteins are expressed as inactive precursors that have to be processed by SUMO-specific proteases to expose a C-terminal double-glycine motif required for conjugation. The processed form of SUMO is activated in an ATP-dependent reaction in which a thioester bond is formed between its C-terminal glycine residue and a cysteine residue in the activating E1 enzyme (SAE1/2). Following transfer to a conjugating E2 enzyme (Ubc9), SUMO is covalently attached to a target protein lysine residue via isopeptide bond formation. Whereas SUMO E1 and E2 enzymes are sufficient to modify target proteins in vitro (23), several SUMO E3 ligases have been described that enhance transfer of SUMO from the E2 enzyme to specific substrates (22). Importantly, SUMO modification is a reversible and dynamic process, and SUMO conjugates can be removed from substrate proteins by SUMO-specific proteases (22). SUMO modification occurs in many cases, but not exclusively, within a consensus motif ⌿KXE, where ⌿ is a large hydrophobic amino acid residue and X is any amino acid residue (22,24). Here we investigated SUMO-1 modification of several TFIID subunits containing the ⌿KXE motif. We show that hsTAF5 and hsTAF12 are SUMO-1-modified in human cells. We further show that recombinant hsTAF1, hsTAF5, hsTAF12, and hsTBP can be SUMO-1-modified using a minimal SUMO conjugation system composed of recombinant E1 and E2 enzymes and SUMO-1. TAF5 and TAF12 can also be efficiently sumoylated in purified TFIID complex, whereas SUMO acceptor sites in TAF1 and TBP appear to become inaccessible upon assembly into TFIID. We identified the principal SUMO-1 acceptor sites in hsTAF5 and hsTAF12 and show that SUMO modification does not affect nuclear import or the global nuclear distribution of hsTAF5 and hsTAF12 in human cells. Finally, we present results of in vitro DNA binding experiments showing that SUMO modification of purified human TFIID at hsTAF5 inhibits TFIID DNA-binding activity.

EXPERIMENTAL PROCEDURES
Cell Culture-HeLa cells were grown in Dulbecco's modified Eagle's medium, supplemented with 10% fetal bovine serum. HeLa-6His:Myc: SUMO-1 cells (25) were kindly provided by Peter O'Hare (Marie Curie Research Institute, Oxted, UK) and were grown in the presence of 2 g/ml puromycin to select for 6His:Myc:SUMO-1 expression.
Antibodies-Rabbit polyclonal antibodies for hsTAF5, hsTAF6, hsTAF9, hsTAF12, and hsTBP were a kind gift of R. G. Roeder (The Rockefeller University, New York, NY). Rabbit polyclonal antibody raised against the N-terminal part of VP22 (AGV31) has been described previously (26). Mouse monoclonal antibodies for the HA:epitope tag (F7), hsTAF1 and hsSUMO-1, as well as rabbit polyclonal antibody for the 6His:epitope tag were purchased from Santa Cruz Biotechnology.
Infection of HeLa and HeLa-6His:Myc:SUMO-1 Cells with HSV-1-HeLa or HeLa-6His:Myc:SUMO-1 cells were seeded in 6-well plates at 3 ϫ 10 5 cells/well and infected the following day with wild-type HSV-1 strain 17syn ϩ in Dulbecco's modified Eagle's medium containing 2% fetal bovine serum at a multiplicity of infection of 100 plaque-forming units per cell. The virus inoculum was allowed to adsorb to cells for 1 h at 37°C, then replaced with fresh medium, and the infected cells were further incubated at 37°C. Cells were lysed in SDS loading buffer containing 25 mM N-ethylmaleimide (NEM) 17 h post infection and analyzed by SDS-PAGE and immunoblotting.
Cold Shock Treatment of Hela-6His:Myc:SUMO-1 Cells-6His:Myc: SUMO-1 cells were seeded in 6-well plates at 3 ϫ 10 5 cells/well. The following day, cells were subjected to 4°C for 10 min. After further incubation at 37°C for 5 h cells were lysed in SDS loading buffer containing 25 mM NEM and analyzed by SDS-PAGE and immunoblotting.
Transient Transfection-Transient transfections were performed using GeneJuice transfection reagent (Novagen) according to the manufacturer's protocol. Typically, transfections were performed in 10-cm tissue culture plates with 2 ϫ 10 6 cells and 5 g of plasmid DNA.
SDS-PAGE and Immunoblotting-Proteins were separated by SDS-PAGE using pre-cast NuPAGE SDS gels (Invitrogen) and transferred to nitrocellulose membranes (Schleicher and Schuell) using an XCell II TM blot module (Invitrogen). Membranes were washed in H 2 O for 10 min, stained with Aurodye Forte (Amersham Biosciences) according to the manufacturer's instructions, blocked with 5% dried nonfat milk in TBS-T20 (10 mM Tris⅐Cl, pH 8.0, 25 mM NaCl, 0.1% Tween 20), and incubated with primary antibody in blocking buffer. Membranes were washed three times for 10 min in TBS-T20 and incubated for 1 h with the appropriate horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences) diluted in blocking buffer. Membranes were washed again three times for 10 min in TBS-T20 and detected with enhanced chemiluminescence reagent (ECL, Amersham Biosciences).
Immunocytochemistry-HeLa cells were grown on coverslips and fixed using 4% paraformaldehyde in PBS (0.43 mM Na 2 HPO 4 , 0.14 mM KH 2 PO 4 , 13.7 mM NaCl, 0.27 mM KCl) for 10 min, washed three times for 5 min in PBS, permeabilized with 0.5% Triton X-100 in PBS for 15 min, washed again three times for 5 min in PBS, and blocked with 10% fetal bovine serum in PBS for 1 h. Coverslips were then incubated for 1 h with primary antibody (1:1000 dilution in blocking buffer), washed three times for 5 min with PBS, and further incubated for 1 h with fluorescein-conjugated secondary antibody (Vector Laboratories, 1:1000 dilution in blocking buffer). Coverslips were washed again three times with PBS and mounted on a slide using Vectashield mounting media (Vector Laboratories). Cells were visualized using an Axiovert S100TV confocal microscope (Zeiss). Images were processed using Metamorph version 4.1.4 (Universal Imaging Corp.) and Photoshop version 7 (Adobe) software.
Preparation of Immobilized Promoter DNA Templates-762-bp linear biotinylated G5TdT(Ϫ41TATA/ϩ33) promoter DNA fragments were amplified by PCR from plasmid pTOG5TdT(Ϫ41TATA/ϩ33) using a 5Ј-biotinylated reverse primer. PCR products were purified from agarose gels and immobilized on M-280 streptavidin-coated magnetic beads (Dynal) according to the manufacturer's instructions.
TFIID DNA Binding Assay-TFIID DNA binding was performed at 30°C for 1 h in transcription buffer (20 mM HEPES⅐KOH, pH 8.4, 5 mM MgCl 2 , 0.05% Igepal CA-630 (Sigma), 5 mM dithiothreitol) containing 0.1 mg/ml FLAG peptide (Sigma). 50-l binding reactions contained 10 ng of highly purified recombinant TBP or 200 ng of immunoaffinitypurified FLAG:epitope-tagged TFIID (f:TFIID), and 500 fmol of immobilized promoter template. f:TFIID-promoter complexes were separated from unbound proteins in a magnetic particle separator, washed with 200 l of transcription buffer without FLAG peptide, and eluted from immobilized template with SDS loading buffer. Proteins present in the combined unbound and wash fractions were quantitatively recovered by binding to 16 l of StrataClean suspension (Stratagene) for 30 min at 4°C and elution with SDS loading buffer. In vitro SUMO-1 conjugation of f:TFIID prior to or after DNA binding was carried out in 50-l reactions as described under "In Vitro SUMO-1 Conjugation Assay."
Detection of SUMO-1 conjugates is often difficult, because only a small percentage of cellular SUMO-1 target proteins is modified at any given time (22). For this reason, SUMO conjugation is generally investigated in cells that overexpress SUMO proteins, and, as a consequence, contain elevated levels of SUMO-1 conjugates. Detection of SUMO targets is further facilitated by expression of epitope-tagged versions of SUMO proteins, allowing for enrichment of SUMO conjugates from cell lysates by affinity chromatography methods.
To investigate SUMO-1 modification of human TFIID subunits, we made use of a human cell line that constitutively expresses elevated levels of 6His:Myc:epitope-tagged human SUMO-1 protein (25). Cells were lysed in the presence of SDS and N-ethylmaleimide (NEM) to inhibit SUMO-specific proteases, and cell lysates were passed over nickel-charged agarose resin (Ni-NTA-agarose, Qiagen) to enrich proteins conjugated to 6His:Myc:SUMO-1. Analysis of Ni-NTA-bound fractions by immunoblotting using TAF-specific antibodies revealed immunoreactive bands with an apparent molecular weight expected for mono-sumoylated forms of hsTAF5 and hsTAF12 present in HeLa-6His:Myc:SUMO-1 cell lysates, but not in lysates of HeLa control cells ( Fig. 2A, compare lanes 2  and 3). SUMO-1-modified hsTAF5, but not SUMO-1-modified hsTAF12, could also be detected in whole cell lysates of HeLa-6His:Myc:SUMO-1 cells (Fig. 2B and data not shown). Unfortunately, we were unable to obtain evidence for SUMO-1 Previous studies reported that stress conditions such as heat shock, serum starvation, and infection by viruses affect SUMO-1 conjugation in human cells (25,35,36). To investigate whether sumoylation of TFIID subunits in HeLa-6His:Myc: SUMO-1 is regulated, we examined the sumoylation status of endogenous hsTAF5 protein in response to different stress conditions, including heat shock, cold shock, serum starvation, and virus infection.
Heat treatment (1 h at 42°C) and serum starvation (24 h at 37°C) had no detectable effect on hsTAF5 sumoylation levels (data not shown). In contrast, cold shock (10 min at 4°C) and infection with herpes simplex virus 1 (HSV-1) resulted in a dramatic loss of intracellular SUMO-1 conjugates and a corresponding loss of SUMO-1-modified hsTAF5 (Fig. 2B). Importantly, neither cold shock nor HSV-1 infection led to a detectable decrease in the absolute levels of cellular hsTAF5 protein.
Thus, the observed loss of SUMO-1-modified hsTAF5 in response to stress can be attributed to either down-regulation of intracellular SUMO-1-activating/-conjugating activity and/or up-regulation of intracellular SUMO isopeptidase activity. In summary, these observations demonstrate that hsTAF5 and hsTAF12 are modified by SUMO-1 in human cells and that TAF SUMO-1 conjugation observed in HeLa-6His:Myc: SUMO-1 cells is a regulated event.
Recombinant hsTBP, hsTAF1, hsTAF5, and hsTAF12 Are Substrates for SUMO-1 Modification in Vitro-Next, we analyzed SUMO-1 modification of single recombinant TFIID subunits in vitro. Highly purified bacterially expressed recombinant hsTAF12 or hsTBP, or recombinant hsTAF1, hsTAF4, or hsTAF5 expressed by coupled in vitro transcription/translation, were incubated with a minimal SUMO-1 conjugation system composed of purified recombinant human E1 (SAE1/2) and E2 (Ubc9) enzymes, and purified mature SUMO-1 (aa 1-97) (32,33). Immunoblot analysis of reaction products revealed SUMO-1 modification of hsTBP, hsTAF1, hsTAF5, and the 20-kDa variant of hsTAF12 (Fig. 3, A-D). SUMO-1 modification was not detected with hsTAF4 (Fig. 3D), which contains a SUMO consensus target site (Fig. 1), and with the 15-kDa variant of hsTAF12 (Fig. 3A), which lacks the N terminus of full-length TAF12 (28). hsTAF1 contains two SUMO-1 consensus target sequences (Fig. 1), one located close to a region harboring histone acetyltransferase and ubiquitin-activating/-conjugating (E1/E2) activities (7,9), and another located in close proximity to the hsTAF1 double-bromodomain region (37). To test whether SUMO modification of recombinant hsTAF1 occurs at a single site or at multiple sites, we divided hsTAF1 into three parts and expressed the corresponding protein fragments by coupled transcription/translation in vitro: an N-terminal fragment A spanning amino acid residues 1-500, which lacks consensus SUMO target sites, an internal fragment B (aa 501-1150) harboring HAT and ubiquitin-activating/-conjugating activi- ties, and a C-terminal fragment C (aa 1151-1872) containing the double bromodomain and C-terminal kinase domain (Fig.  3E). As shown in Fig. 3E, hsTAF1 fragments B and C, which contain consensus SUMO target sites, could be SUMO-1-modified, whereas fragment A, which lacks a consensus SUMO target site, could not be SUMO-1-modified. These results demonstrate that hsTAF1 contains at least two SUMO target sites and are consistent with the idea that hsTAF1 sumoylation occurs at the identified SUMO consensus target sequences.

Purified TFIID Complex Can Be Efficiently in Vitro SUMO-1-modified at hsTAF5 and hsTAF12, but Not at TAF1 or TBP-
The results shown in Fig. 3 demonstrated that recombinant hsTAF1, hsTAF5, hsTAF12, and hsTBP are SUMO substrates in vitro. It was important to test if these TFIID subunits could also be SUMO-modified when assembled into the TFIID complex. To answer this question we performed in vitro sumoylation experiments with immunoaffinity-purified FLAG:epitopetagged TFIID complex (f:TFIID) (34). Immunoblot analysis with antibodies specific for hsTBP and hsTAF subunits revealed mono-sumoylated forms of hsTAF5 and hsTAF12 (Fig.  4). In contrast, SUMO modification of hsTAF1, hsTAF6, hsTAF9, and hsTBP could not be detected.
These results demonstrate that TAF5 and TAF12 SUMO targets sites are exposed on the surface of the human TFIID complex. However, SUMO target sites within hsTAF1 and hsTBP appeared to be inaccessible under the in vitro conditions.
Identification of SUMO-1 Acceptor Sites in hsTAF12 and hsTAF5-Further investigations focused on hsTAF5 and hsTAF12, because these were so far the only subunits for which we could demonstrate SUMO modification in human cells and in vitro in the context of the TFIID complex.
To test whether SUMO-1 modification of hsTAF5 and hsTAF12 occurred at the SUMO-1 target sites identified by amino acid sequence analysis (Fig. 1), we generated Lys -Ͼ Arg point mutants in which the putative SUMO-1 acceptor lysine was substituted by arginine. Recombinant hsTAF5K14R and hsTAF12K19R were expressed by coupled in vitro transcription/translation, subjected to in vitro sumoylation, and analyzed by immunoblotting. As shown in Fig. 5A, in vitro SUMO modification of hsTAF5K14R and hsTAF12K19R mutant proteins was significantly reduced compared with the wild-type proteins. These data demonstrate that Lys-14 in hsTAF5 and Lys-19 in hsTAF12 are the primary acceptor sites for SUMO-1 conjugation in vitro.
To confirm that hsTAF5 Lys-14 and hsTAF12 Lys-19 are also sites for SUMO conjugation in human cells, we transiently expressed wild-type and mutant hsTAF5 and hsTAF12 proteins in HeLa-6His:Myc:SUMO-1 cells (25). Wild-type and point mutants were expressed at comparable levels, suggesting that the Lys -Ͼ Arg substitutions do not affect protein stability in human cells (data not shown). Furthermore, SUMO-1 modification of transiently expressed hsTAF5K14R and hsTAF12K19R mutant proteins was undetectable (Fig.  5B), consistent with the results of in vitro sumoylation experiments (Fig. 5A). We conclude that SUMO-1 modification of hsTAF5 and hsTAF12 occurs within the sumoylation consensus sequence ⌿KXE and that Lys-14 in hsTAF5 and Lys-19 in hsTAF12 are sites of SUMO-1 conjugation in human cells.

SUMO-1 Modification Does Not Affect hsTAF12 and hsTAF5
Intracellular Localization-Previous studies had correlated SUMO modification with the recruitment of target proteins to specialized subnuclear domains, such as PML bodies (18 -20, 38). To address this issue, we transiently expressed triple HA:epitope-tagged wild-type hsTAF5 and hsTAF12 and hsTAF5K14R and hsTAF12K19R SUMO acceptor site mutants in HeLa-6His:Myc:SUMO-1 cells and examined their intracellular localization by fluorescence microscopy.
Immunostaining of untransfected HeLa cells with anti-hsTAF5 antibody revealed that endogenous hsTAF5 is localized predominantly in the nucleus, with a diffuse nuclear distribution excluding the nucleoli (Fig. 6A, panel 4). Likewise, low levels of exogenous (HA) 3 :hsTAF5 expressed shortly after transfection (Ͻ16 h) could only be detected in the nucleus (data not shown). However, with increasing expression, exogenous (HA) 3  transfection) (Fig. 6A, upper panels 1-3), suggesting that nuclear import of hsTAF5 might depend on a limiting cellular factor. Indeed, we found that even high levels of exogenously expressed (HA) 3 :hsTAF5 were efficiently imported into the nucleus when co-expressed with hsTBP (Fig. 6A, lower panels 1). (HA) 3 :hsTAF12 was found exclusively in the nucleus, even when expressed at high levels (Fig. 6B).
Transiently expressed SUMO acceptor site mutants (HA) 3 : hsTAF5K14R and (HA) 3 :hsTAF12K19R showed diffuse nuclear localization similar to their wild-type counterparts (Fig.  6), suggesting that nuclear import of hsTAF5 and hsTAF12 is independent of SUMO modification. However, because only a small percentage of expressed proteins were modified (Figs. 2 and 5), it was possible that differences in the nuclear distribution of wild-type proteins and SUMO acceptor mutants could not be detected. We therefore wished to express hsTAF5 and hsTAF12 variants with SUMO-1 stably attached to their respective SUMO-1 lysine acceptor sites. Because the SUMO acceptor lysine residues in hsTAF5 and hsTAF12 are located very close to their N terminus (Fig. 1), we reasoned that constitutive SUMO-1 conjugation could be closely mimicked by fusing the mature form of SUMO-1 (aa 1-97) to the N terminus of hsTAF5 lacking the first 13 amino acids (hsTAF5(⌬1-13)) and to hsTAF12 lacking the first 18 amino acids (hsTAF12(⌬1-18)), respectively. However, upon expression in HeLa cells these fusion proteins were rapidly processed into free SUMO-1 (aa 1-97) and hsTAF5(⌬1-13) or hsTAF12(⌬1-18), respectively (data not shown). To prevent cleavage by SUMO-specific proteases, we substituted the C-terminal glycine in the SUMO-1 moiety of the fusion proteins (Gly-97) with alanine (G97A). The resulting fusion proteins, SUMO-1-(1-95)-GA-hsTAF5(⌬1-13) and SUMO-1-(1-95)-GA-hsTAF12(⌬1-18), are stable when expressed in HeLa and HeLa-6His:Myc:SUMO-1 cells (data not shown) and show diffuse nuclear localization with no apparent differences to wild-type proteins (Fig. 6). Collectively, these data suggest that SUMO-1 modification does not affect nuclear import or the nuclear distribution of hsTAF5 and hsTAF12.

SUMO-1 Modification at TAF5
Inhibits DNA-binding Activity of Human TFIID-We used an immobilized promoter DNA template assay to investigate whether SUMO conjugation affects the DNA-binding activity of purified TFIID complex. We first asked if human TFIID could be SUMO-1-modified when bound to DNA. To this end, linear promoter DNA templates were immobilized on magnetic beads via a biotin-streptavidin link and incubated with highly purified f:TFIID complex. TFIID-promoter complexes were separated from unbound protein using a magnetic particle separator and incubated with our in vitro SUMO-1 conjugation system. After the in vitro sumoylation reaction, DNA-bound and unbound proteins were again separated and analyzed in parallel by immunoblotting. A small fraction (10 -20%) of TFIID-promoter complexes dissociated during the incubation in sumoylation buffer (Fig. 7A). However, this was independent of ongoing sumoylation or the presence of individual components of the reconstituted human SUMO conjugation system (Fig. 7A, lanes 1-2 and 5-6 and data not shown). Furthermore, SUMO-1-modified forms of hsTAF5 and hsTAF12 could clearly be detected in the DNAbound f:TFIID fraction (Fig. 7A, lanes 4 and 8). These results demonstrate that SUMO-1 can be conjugated to hsTAF5 and hsTAF12 in promoter-bound TFIID without affecting TFIIDpromoter complex stability ( Fig. 7A and data not shown).
Next, we tested whether sumoylation affects binding of purified TFIID complex to promoter DNA. f:TFIID was first subjected to in vitro SUMO-1 modification, followed by addition of immobilized promoter DNA template and further incubation for 60 min to assemble TFIID-promoter complexes. Free TFIID and TFIID-DNA complexes were separated and analyzed in parallel by immunoblotting. We found that in vitro SUMO conjugation was strongly inhibited in the presence of 15 mM ATP␥S nucleotide analogue ( Fig. 7B and data not shown). To prevent SUMO-1 conjugation during the DNA binding reaction, ATP␥S was added along with the immobilized promoter DNA template (Fig. 7B, lanes 5-6 and 11-12).
When the DNA-binding reaction was carried out in the pres- ence of ATP␥S, SUMO-1-modified TAF5 could only be detected in the unbound TFIID fraction. In contrast, unbound TFIID and promoter-bound TFIID fractions contained a similar percentage of SUMO-1-modified TAF12 (Fig. 7, compare lanes 5  and 6 and lanes 11 and 12). When the DNA-binding step was carried out in the absence of added ATP␥S, low levels of SUMO-1-modified TAF5 could be detected in TFIID-promoter DNA complexes. According to our data shown in Fig. 7A, these can be attributed to SUMO-1 conjugation after stable TFIID-promoter complex formation. However, even under these conditions the majority of TFIID containing SUMO-1-modified TAF5 was found to be present in the unbound fraction.
In summary, the results of the DNA binding assays suggest that SUMO-1 modification at TAF5 prevents TFIID binding to promoter DNA, whereas SUMO-1 modification at TAF12 has no detectable effect on TFIID DNA-binding activity. In contrast, pre-formed TFIID-promoter complexes can be SUMO-1modified at both TAF5 and TAF12, without compromising TFIID nucleoprotein complex stability. DISCUSSION In this study we investigated SUMO-1 modification of several human TFIID subunits containing the ⌿KXE SUMO consensus target motif (22,24). We demonstrate that the human TFIID subunits hsTAF5 and hsTAF12 are SUMO-modified in human cells and in vitro, both in isolation and in the context of purified human TFIID. We further observed in vitro sumoylation of recombinant hsTBP and hsTAF1. Interestingly, hsTAF1 contains at least two SUMO acceptor sites, one located close to the hsTAF1 HAT domain, and another in close proximity to the hsTAF1 double bromodomain (Fig. 1). However, so far we were unable to demonstrate in vitro SUMO conjugation to hsTAF1 or to hsTBP in the context of purified human TFIID complex.
Whether TFIID sumoylation at hsTAF1 and/or hsTBP can occur under more physiological conditions, for example in the presence of specific E3 ligase activity, remains a possibility that has to be investigated in the future.
Investigations into the functional consequences of SUMO modification within multiprotein complexes such as TFIID present a formidable technical challenge. In the simplest scenario, conjugation of SUMO to a specific target site may either directly or indirectly modulate subunit-specific TFIID functions, such as TAF interactions with regulatory proteins, TBP DNA-binding activity (see below), or one of the enzymatic activities residing in the TAF1 subunit. However, it is equally conceivable that a single SUMO modification can affect the entire TFIID complex, for example by altering its stability or by mediating its localization to specialized nuclear domains. On the other hand, SUMO modification at a single target site may have only very subtle effects, and multiple sumoylation events may be required to modulate TFIID functions. Finally, we note that global changes in cellular TFIID activity could be brought about by transient SUMO modification of isolated TFIID subunits, which may in turn affect TFIID complex assembly or the induction of dynamic changes in TFIID subunit composition seen during developmental processes (39,40).
So far, we have no indication that SUMO modification affects the nuclear localization of TFIID. In human cultured cells, transiently expressed hsTAF5 and hsTAF12 point mutants lacking the SUMO acceptor lysine and SUMO-1-hsTAF5 and -hsTAF12 fusion proteins showed a nuclear localization pattern similar to their wild-type counterparts. However, the data obtained so far do not allow us to draw conclusions on the question whether changes in the nuclear distribution of TFIID can be brought about by simultaneous SUMO modification of FIG. 7. A, promoter-bound TFIID can be SUMO-1-modified at TAF5 and TAF12. Immunoaffinity-purified human f:TFIID complex was pre-bound to immobilized promoter DNA templates. f:TFIID-promoter complexes were isolated and subjected to in vitro SUMO-1 conjugation. After sumoylation, DNA-bound and unbound proteins were separated, quantitatively loaded onto SDS-PAGE gels, and analyzed in parallel by immunoblotting with the antibodies indicated. B, SUMO-1 conjugation at TAF5 inhibits f:TFIID promoter-binding activity. Immunoaffinity-purified human f:TFIID complex was in vitro SUMO-1-modified prior to the addition of immobilized promoter DNA template and further incubation to assemble TFIID-promoter complexes. 15 mM of the ATP analogue ATP␥S was added to prevent any SUMO-1 conjugation during the DNAbinding step. f:TFIID-promoter complexes were separated from unbound protein, and DNA-bound and unbound fractions were analyzed by immunoblotting with the antibodies indicated. As a negative control, sumoylation reactions were carried out in the absence of E1 enzyme (ϪE1). multiple subunits.
TFIID is unique in that it is the only component of the general RNAP II transcription machinery capable of sequencespecific DNA binding. TFIID interactions with core promoter DNA elements are crucial for RNAP II transcription initiation complex assembly and are therefore tightly controlled by transcription regulatory pathways.
We investigated the effect of sumoylation on the DNA-binding activity of purified human TFIID complex using immobilized promoter DNA templates. Our data suggest that SUMO-1 conjugation to hsTAF5 interferes with TFIID binding to promoter DNA. In contrast, SUMO-1 conjugation to hsTAF12 subunit appears to have no detectable effect on TFIID DNA-binding activity. These observations suggest that reversible SUMO modification at the hsTAF5 subunit can contribute to the dynamic regulation of TFIID promoter-binding activity in human cells.
Surprisingly, prebound TFIID-promoter complexes can be SUMO-modified at both hsTAF5 and hsTAF12 without compromising their stability. This observation suggests that hsTAF5 sumoylation interferes with TFIID DNA interactions through a subunit other than hsTAF5 itself. Earlier studies suggested that the in vitro promoter-binding activity of purified human TFIID complex is largely dependent on sequence-specific TBP interactions with TATA box sequences (27,41,42). Thus SUMO modification of TFIID at hsTAF5 may compromise TBP-DNA interactions. Support for this model comes from structural analyses of purified human and yeast TFIID complexes by electron microscopy and digital image processing (43)(44)(45)(46). These studies revealed a remarkable similarity in the overall architecture of human and yeast TFIID complexes, which resemble a molecular clamp formed of three major lobes connected by thin linking regions (45,46). The location of TAFs within the yeast TFIID complex has been determined by immunolabeling experiments (43,44). Importantly, yeast TFIID contains two copies of TAF5 (47), which form the bridging regions between the three lobes of the TFIID structure (Fig. 8) (44). The N termini of the two copies of TAF5 are both located within lobe C in close proximity of each other, whereas the two TAF5 C termini, containing WD40 protein-protein interaction domains (48), extend into two different lobes A and B (Fig. 8) (43, 44). TAF12 forms a heterodimeric complex with TAF4 through specific histone fold domain interactions (4,49). Yeast TFIID complexes can contain two TAF4/TAF12 submodules, which localize to lobes B and C of the TFIID structure (Fig. 8) (43). Fig. 8 shows a hypothetical model based on the yeast TFIID structure, which indicates the potential location of hsTAF5 and hsTAF12 SUMO target sites. The model predicts that SUMO conjugation to the N terminus of hsTAF5 occurs in lobe C in relative close proximity to TBP, consistent with the idea that hsTAF5 interferes with TBP DNA binding. Sumoylation of hsTAF12 may affect two distinct locations, one in lobe C close to the TAF5 N termini and one in lobe B, a good distance away from TBP. These observations suggest an additional level of complexity that must be considered when the functional relevance of SUMO modifications at individual TFIID subunits is investigated: SUMO modification of a particular TFIID subunit might potentially affect two distinct regions within the same TFIID structure.
Previous studies had shown that reversible SUMO conjugation modulates the activity of many promoter-specific transcription regulatory proteins in human cells (18,19,21). Our data extend these observations and provide the first evidence that SUMO modifications can directly affect the activity of the human TFIID complex, a key component of the general RNAP II transcription machinery. Our observations are in broad agreement with recently published results of proteomic approaches in yeast, which identified several RNAP II subunits as well as subunits of the yeast general transcription factors TFIIA, TFIID, TFIIE, and TFIIF as potential SUMO targets (50 -52). Thus SUMO modification of general RNAP II transcription factors appears to be a regulatory mechanism that is conserved throughout evolution. The three-dimensional structure of human TFIID has been determined by electron microscopy and digital image processing (45,46). The relative positions of TBP and TAF subunits within the threelobed TFIID structure are based on immunomapping of the yeast TFIID complex (43,44).