Histone Folds Mediate Selective Heterodimerization of Yeast TAF_II25 with TFIID Components yTAF_II47 and yTAF_II65 and with SAGA Component ySPT7

Abstract

We show that the yeast TFIID (yTFIID) component yTAF_II47 contains a histone fold domain (HFD) with homology to that previously described for hTAF_II135. Complementation in vivo indicates that the yTAF_II47 HFD is necessary and sufficient for vegetative growth. Mutation of highly conserved residues in the α1 helix of the yTAF_II47 HFD results in a temperature-sensitive phenotype which can be suppressed by overexpression of yTAF_II25, as well as by yTAF_II40, yTAF_II19, and yTAF_II60. In yeast two-hybrid and bacterial coexpression assays, the yTAF_II47 HFD selectively heterodimerizes with yTAF_II25, which we show contains an HFD with homology to the hTAF_II28 family We additionally demonstrate that yTAF_II65 contains a functional HFD which also selectively heterodimerizes with yTAF_II25. These results reveal the existence of two novel histone-like pairs in yTFIID. The physical and genetic interactions described here show that the histone-like yTAF_IIs are organized in at least two substructures within TFIID rather than in a single octamer-like structure as previously suggested. Furthermore, our results indicate that ySPT7 has an HFD homologous to that of yTAF_II47 which selectively heterodimerizes with yTAF_II25, defining a novel histone-like pair in the SAGA complex.

Transcription factor TFIID, one of the general factors required for accurate and regulated initiation by RNA polymerase II, comprises the TATA binding protein and TATA binding protein-associated factors (TAF_IIs) (4, 15). The cDNAs encoding many human TAF_IIs (hTAF_IIs) have been isolated, revealing a striking sequence conservation with yeast TAF_IIs (yTAF_IIs) and Drosophila TAF_IIs (dTAF_IIs). A subset of TAF_IIs are present not only in TFIID but also in the SAGA, PCAF, STAGA, and TFTC complexes (7, 13, 23, 27, 36).

Genetic studies with yeast have shown that TAF_IIs play an important role in transcriptional regulation of many genes (14). Temperature-sensitive mutations in yTAF_II145 and yTAF_II90 result in cell cycle arrest and lethality, but the expression of only a small number of genes is affected (3, 35). In contrast, tight temperature-sensitive mutations in yTAF_II17, yTAF_II25, yTAF_II60, and yTAF_II61/68, which are present in the TFIID and SAGA complexes, or in the TFIID-specific yTAF_II40 have a more dramatic effect, the transcription of the majority of yeast genes being affected (2, 21, 24–26, 29).

Initial sequence alignments indicated that hTAF_II80 (corresponding to dTAF_II60 and yTAF_II60), hTAF_II31 (dTAF_II40 and yTAF_II17), and hTAF_II20 (dTAF_II30α and yTAF_II61/68) presented obvious sequence similarity to histones H4, H3, and H2B, respectively (17, 20). Structural studies show that dTAF_II60 and dTAF_II40 interact via a histone fold and form an H3-H4-like heterotetramer (37). These findings, together with biochemical experiments and genetic interaction data obtained with yeasts, led to the proposal that TFIID and the other TAF_II-containing complexes contain a histone octamer-like substructure composed of an hTAF_II80-hTAF_II31 heterotetramer and two hTAF_II20 homodimers (8, 16).

Subsequent data show this model to be an oversimplification. hTAF_II28 and hTAF_II18 are also histone-like, since they interact via a histone fold domain (HFD) to form a heterodimer (5). The SAGA, PCAF, TFTC, and STAGA component SPT3 shows extensive sequence homology to the HFDs of both hTAF_II18 and hTAF_II28 in its N- and C-terminal regions, respectively, and could potentially form a histone-like pair by intramolecular interactions. Contrary to what was first suggested, hTAF_II20 does not homodimerize but rather heterodimerizes with hTAF_II135 (10). In yeasts, the hTAF_II20 homologue yTAF_II68 heterodimerizes with the SAGA component yADA1, and it has been suggested that yTAF_II68 may also heterodimerize with yTAF_II48, a potential homologue of hTAF_II135, in yTFIID (28, 30). These results indicate that there are many more histone-like pairs in TFIID and SAGA than originally suspected. Recent electron microscopy studies show that TFIID comprises three or four lobes arranged in a horseshoe fashion around a central groove (1, 6). Within the present limits of resolution it appears that no single lobe of TFIID would be big enough to harbor all the known histone fold TAF_IIs, suggesting that they are shared among two or more of the lobes.

We previously reported that hTAF_II135 contained an HFD with significant sequence homology to the SAGA component yADA1 (10). Now we show that the hTAF_II135 HFD also shares significant sequence homology with an HFD in yTAF_II47 (34). In complementation experiments, the yTAF_II47 HFD is necessary and sufficient for vegetative yeast growth. The temperature-sensitive phenotype of a mutation in the yTAF_II47 HFD can be rescued by overexpression of yTAF_II25 and, at less restrictive temperatures, by yTAF_II60, yTAF_II40, and yTAF_II19. In yeast two-hybrid and bacterial coexpression experiments, the yTAF_II47 HFD mediates selective heterodimerization with the conserved core domain of yTAF_II25. There are therefore both genetic and physical interactions between these two yTAF_IIs, suggesting that they form an additional histone-like pair in yTFIID. Consistent with this idea, we show the conserved core domain of yTAF_II25 to be an HFD which shares homology with the HFD of the hTAF_II28 family.

Recently, yTAF_II65 was identified as a novel component of yTFIID (30). Now we demonstrate that yTAF_II65 contains an HFD with homology to that of yTAF_II17, dTAF_II40, and hTAF_II31. In contrast to the HFD of yTAF_II47, the yTAF_II65 HFD is not sufficient for growth. Nevertheless, deletion or mutation of this domain results in temperature sensitivity, showing that it is important for yTAF_II65 function. Surprisingly, the yTAF_II65 HFD also selectively heterodimerizes with yTAF_II25, which thus has two heterodimerization partners in TFIID.

Finally, as yTAF_II47 and yTAF_II65 are not present in SAGA, we sought a heterodimerization partner for yTAF_II25 in this complex. Our results indicate that ySPT7 contains an HFD with homology to that of yTAF_II47. This domain mediates selective heterodimerization with yTAF_II25. Together our results reveal the existence of novel histone-like pairs in the TFIID and SAGA complexes. They highlight the important functional and structural role played by this motif in these complexes and provide evidence that the histone-like yTAF_IIs assemble into at least two distinct substructures within TFIID.

MATERIALS AND METHODS

Yeast strains.

The yeast strains used in this study are YSLS67 (Mata ura3-1 leu2-3,112 trp1-1 his3-11,15 ade2-11 can1-100 taf47Δhisg-hisg [pRS416-TAF47]), YSLS67/47 (MATa ura3-1 leu2-3,112 trp1-1 his3-11,15 ade2-11 can1-100 taf47Δhisg-hisg [pAS3-TAF47)], YSLS67/47HFD (MATa ura3-1 leu2-3,112 trp1-1 his3-11,15 ade2-11 can1-100 taf47Δhisg-hisg [pAS3-TAF47(1–81)]), YSLS67/47m1 (MATa ura3-1 leu2-3,112 trp1-1 his3-11,15 ade2-11 can1-100 taf47Δhisg-hisg [pAS3-TAF47(R13D, I14E)]), YSLS67/VP16AD47HFD (MATa ura3-1 leu2-3,112 trp1-1 his3-11,15 ade2-11 can1-100 taf47Δhisg-hisg [pASV3-TAF47(1–81)]), YSLS58 (MATa ura3Δ0 leu2Δ0 his3Δ1 lys2Δ0 MET15 KAN Δtaf65 [pRS416-TAF65]), YSLS58/65 (MATa ura3Δ0 leu2Δ0 his3Δ1 lys2Δ0 MET15 KAN Δtaf65 [pAS3-TAF65]), YSLS58/65ΔHFD (MATa ura3Δ0 leu2Δ0 his3Δ1 lys2Δ0 MET15 KAN Δtaf65 [pAS3-TAF65(103–510)]), YSLS58/65m1 (MATa ura3Δ0 leu2Δ0 his3Δ1 lys2Δ0 MET15 KAN Δtaf65 [pAS3-TAF65(L64P, L67P)]), YSLS58/VP16AD65 (MATa ura3Δ0 leu2Δ0 his3Δ1 lys2Δ0 MET15 KAN Δtaf65 [pASV3-TAF65]), and L40 [MATa trp1-901 leu2-3,112 his3-Δ200 ade2 LYS2::(LexAop)₄-HIS3 URA3::(LexAop)₈-Lac].

Construction of recombinant plasmids.

All yeast and bacterial expression vectors were constructed by PCR using primers with the appropriate restriction sites, and constructs were verified by automated DNA sequencing. Details of constructions are available on request. LexA fusions were constructed in the multicopy vector pBTM116 containing the TRP1 marker, and the VP16 fusions were constructed in the multicopy vector pASV3 containing the LEU2 marker (10). For complementation, wild-type or mutated yTAF_IIs were cloned in the multicopy pAS3 plasmid with a LEU⁻ marker.

Two-hybrid, complementation, and high-copy-number temperature-sensitive suppression assays.

All yeast strains were transformed by the lithium acetate technique. For two-hybrid assays, transformants were selected on Trp⁻ Leu⁻ plates. Quantitative β-galactosidase assays on individual L40 transformants were determined as previously described (10). Reproducible results were obtained in several independent experiments, and the results of a typical experiment are shown in the figures. Yeast strains YSLS67 and YSLS58, used for plasmid shuffling of TAF47 and TAF65, were derived from YJR10 (34) and YSLS41 (30) by sporulation and tetrad dissection. For complementation assays, the rescue plasmids indicated in the relevant figures were transformed and the wild-type TAF/URA3 plasmid was shuffled out by two passes on media containing 5-fluoroorotic acid. For suppression of the yTAF_II47(R13D, I14E) mutant strain, cells were transformed with high-copy-number plasmids with a URA3 marker expressing the indicated yTAF_IIs and serial dilutions of the transformants were spotted at 30, 34, and 36°C. Plates at the restrictive temperatures were photographed after 3 days of growth. In all experiments cultures were grown in yeast extract-peptone-dextrose unless selection was necessary, in which case all cultures were grown in the appropriate selective synthetic dextrose (SD) medium.

Coexpression in Escherichia coli.

Coexpression in E. coli was performed as previously described (9a, 10). All plasmids were constructed by PCR, and details are available on request. The yTAF_II47, yTAF_II65, and SPT7 histone fold regions were expressed as glutathione S-transferase (GST) fusion proteins in pGEX2T. Native untagged, yTAF_II25, hTAF_II30, and yTAF_II68 HFDs were expressed from a modified version of the vector pACYC184 (New England Biolabs). Plasmids pairs were introduced into E. coli strain BL21(DE3), and double transformants were selected on plates containing ampicillin and chloramphenicol. Bacteria were amplified to an optical density at 600 nm of 0.45 and induced for 4 h at 25°C with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Cell were lysed by sonication in buffer (25 mM Tris-HCl [pH 6.0] and 0.4 M NaCl), and the soluble fraction was collected after centrifugation at 14,000 rpm for 20 min at 4°C in an Eppendorf centrifuge. Aliquots of the soluble fraction from a 10-ml bacterial culture were then incubated with glutathione-Sepharose (Pharmacia). Binding and washing were done essentially as described previously (10). Bound proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and staining with Coomassie brilliant blue.

RESULTS

yTAF_II47 contains an HFD which suffices for vegetative growth.

Previously we reported that hTAF_II135 amino acids 870 to 944 showed significant sequence homology to H2A, NC2α and yADA1 (10) (Fig. 1). The hTAF_II135 HFD also shows significant homology to yTAF_II48, the hTAF_II135 homologue in yTFIID (28, 30) (Fig. 1). Database searches also revealed a significant similarity between the HFD of the TAF_II135 family and the N-terminal region of yTAF_II47 (Fig. 1). In the yTAF_II47 HFD, the conserved RI(V, M) residues are found in the α1 helix, followed by an amphipathic α2 helix with numerous conserved hydrophobic residues. The α3 helix is characterized by the presence of D(V, I, L) residues. This pattern of sequence conservation is analogous to that seen amongst other histone fold proteins (5, 10, 32). Several metazoan sequences encoding potential proteins with homology to the yTAF_II47 HFD were also detected in these searches (Fig. 1) (32). This observation indicates that yTAF_II47 is a histone-like protein belonging to an evolutionarily conserved family. In addition to the α1, α2, and α3 helices, which comprise the minimal HFD, there is a potential additional αC helix in this family.

FIG. 1.

Alignment of the HFD sequences of the hTAF_II135 family with those of ADA1 and the yTAF_II47 family. h, human; y, Saccharomyces cerevisiae; d, Drosophila melanogaster; S.pombe, Schizosaccharomyces pombe; Anopheles, Anopheles gambiae; mouse, Mus musculus; Candida, Candida albicans; zebra fish, Danio rerio. The positions of the predicted α helices and loops are indicated above the sequence based on homology with H2A (10, 22). Positions with conserved, mainly hydrophobic, amino acids are in white on a black background. Other residues conserved within the yTAF_II47 family are boxed in gray. Amino acids were classified as follows: small residues, P, A, G, S, and T; hydrophobic residues, L, I, V, A, F, M, C, Y, and W; polar and/or acidic residues, D, E, Q, and N; basic residues, R, K, and H. Threonine residues are occasionally present in otherwise hydrophobic positions. The amino acids sequences shown without numbers are predicted from genomic, expressed sequence tag, or sequence tagged site sequences. The accession numbers for the indicated sequences are as follows: S. pombe, SPT:CAB90151; Candida, 396380B03; Anopheles, GB CN501GI9 AL143170; zebra fish, GB AW343321fi76b06.y1; mouse, GB AA692266ur52c07.

To determine whether the yTAF_II47 HFD is important for function, we performed complementation experiments by plasmid shuffle in a yTAF_II47 null strain. Expression vectors for wild-type or mutated yTAF_II47 proteins were constructed (Fig. 2A), and their ability to rescue growth of the null strain was evaluated. As previously described, yTAF_II47 is essential for vegetative yeast growth (34). Expression of full-length yTAF_II47 efficiently restored the growth of the null strain at 30°C, whereas no growth was seen with the expression vector alone (1 and 3 in Fig. 2B). Growth was also rescued by yTAF_II47(1–81) containing only the minimal HFD (construct 4 in Fig. 2B). These two strains showed comparable growth rates in liquid culture (data not shown). This 1–81 domain rescued growth both when expressed as a native protein and when expressed as a fusion with the VP16 activating domain from a two-hybrid expression vector (construct 5 in Fig. 2B). Mutation of the highly conserved amino acids R13 and I14 in the α1 helix (construct m1 in Fig. 2A) abolished the ability of yTAF_II47(1–81) to rescue growth (construct 6 in Fig. 2B). In contrast, this mutation did not abolish yTAF_II47 function in the context of the full-length protein (construct 2 in Fig. 2B). The (1–353)m1 mutant did, however, show a temperature-sensitive phenotype, as it did not rescue growth at 37°C while both the wild-type protein and the 1–81 deletion rescued growth at this temperature (compare constructs 1 and 3 in Fig. 2C). The VP16-TAF_II47(1–81) fusion also showed a temperature-sensitive phenotype (construct 4 in Fig. 2C). These results indicate that the yTAF_II47 HFD is an essential functional domain necessary and sufficient for vegetative yeast growth.

FIG. 2.

The yTAF_II47 HFD is sufficient for growth. (A) The sequence of the yTAF_II47 HFD is shown along with that of mutant m1. The locations of the potential α helices and loops are indicated above the sequence. The ability of each mutant to complement the null strain is indicated on the right. TS, temperature sensitive. (B and C) Growth of yeasts plated at the indicated temperatures. 5-FOA, 5-fluoroorotic acid.

Genetic interaction between yTAF_II47 and other histone-like yTAF_IIs.

Possible genetic interactions between yTAF_II47 and other yTAF_IIs were examined. To look for such interactions, we tested the ability of other yTAF_IIs to rescue the temperature-sensitive phenotype of the yTAF_II47(1–353)m1 allele when overexpressed at the nonpermissive temperature.

The yTAF_II47(1–353)m1 strain did not grow at 34°C (Fig. 3). At 34°C, the temperature-sensitive phenotype was efficiently suppressed by overexpression of yTAF_II25 and partially suppressed by overexpression of yTAF_II60, yTAF_II40, and yTAF_II19 (Fig. 3). At 36°C, however, only yTAF_II25 could suppress the temperature-sensitive phenotype (Fig. 3). No significant growth was seen when any of the other yTAF_IIs were overexpressed at either temperature (Fig. 3), while all strains showed equivalent growth at 30°C (data not shown). Analogous results were obtained with the yTAF_II47(1–81)-VP16 allele (data not shown). These results show a genetic interaction between yTAF_II47 and three other known histone-like yTAF_IIs, yet the strongest interaction was observed with yTAF_II25, which has not previously been described as histone-like.

FIG. 3.

Genetic interactions among histone-like yTAF_IIs. The growth of serial dilutions of strains with the yTAF_II47(1–353)m1 allele at 34 and 36°C is shown. The overexpressed yTAF_IIs used to rescue the growth at the nonpermissive temperatures are shown on the left.

The HFD of yTAF_II47 mediates heterodimerization with yTAF_II25.

The strong genetic interaction between yTAF_II47 and yTAF_II25 prompted us to determine whether they also interact physically. The yTAF_II47 HFD was fused to the VP16 activation domain or the LexA DNA binding domain and tested for interactions with LexA or VP16 fusions of the core domain of yTAF_II25 and the HFDs of other yTAF_IIs in a series of yeast two-hybrid experiments. Interactions were assessed by measuring β-galactosidase activity in strain L40, which harbors a LexA-responsive LacZ gene (10, 33).

In two-hybrid assays, a strong interaction between yTAF_II47 and yTAF_II25(74–206) was observed (Fig. 4A, column 1). This interaction required only the HFD of yTAF_II47(1–81) and was abolished by mutation m1 in the α1 helix (Fig. 4A, columns 2 and 3). In the context of the full-length yTAF_II47, the m1 mutation reduced but did not abolish the interaction (Fig. 4A, column 4). In contrast, we detected no interactions between the HFD of yTAF_II47 and the HFDs of yTAF_II40 or yTAF_II19 (which themselves strongly interact in two-hybrid assays [data not shown]), yTAF_II60, yTAF_II68, yTAF_II48, and yTAF_II65 (summarized in Table 1). Moreover, we did not observe homodimerization of yTAF_II47. Similarly, with the exception of yTAF_II65 (see below), yTAF_II25 did not interact with the HFDs of the other yTAF_IIs tested (Fig. 4A, columns 5 to 7, and Table 1), although a possible homodimerization was seen (Table 1; also, see Discussion). These results indicate that yTAF_II47 selectively heterodimerizes with yTAF_II25.

FIG. 4.

Selective heterodimerization between yTAF_II47 and yTAF_II25 in two-hybrid assays. (A) Quantification of β-galactosidase activity in two-hybrid assays. The VP16-yTAF_II47, VP16-yTAF_II40, VP16-yTAF_II19, and yTAF_II48 fusions shown below each column were assayed in a LexA-yTAF_II25(74–206) background as indicated above the graph. β-gal, β -galactosidase. (B) Alignment of yTAF_II25 and hTAF_II30 with members of the hTAF_II28 family from yeast and D. melanogaster. Conserved positions are in white against a black background. The positions of the α helices and loops of hTAF_II28 are indicated. Alignment of the α3 helix of yTAF_II25 and hTAF_II30 with that of the other indicated histone fold proteins is also shown. (C) The sequences of the conserved region of yTAF_II25 and hTAF_II30 are shown. Conserved amino acids are white on a black background. The end points of the deletions tested in two-hybrid assays are indicated by the arrows below the sequence. Δ, internal deletion. (D) Mapping of the yTAF_II25 region required for interaction with yTAF_II47. The LexA-yTAF_II25 (LexA-hTAF_II30) deletions indicated below the graph were assayed in the VP16-yTAF_II47(1–81) strain.

TABLE 1.

Interactions between TFIID components and various HFDs

Complex	Histone fold	Interaction with HFDa
		TFIID					TFIID + SAGA
		yTAFII47	yTAFII48	yTAFII65	yTAFII40	yTAFII19	yTAFII25	yTAFII68	yTAFII60
TFIID	yTAFII47	−	−	−	−	−	+++	−	−
	yTAFII65	−	−	−	−	−	+++	−	−
TFIID + SAGA	yTAFII25	+++	−	+++	−	−	+?	−	ND

^a−, no interaction; +?, possible homodimerization; +++, strong interaction; ND, not determined. 

Sequence alignments have shown that hTAF_II30, yTAF_II25, and their homologues from other species have a bipartite structure with a highly conserved C-terminal domain and an unconserved N-terminal region (12) (Fig. 4C). In yTAF_II25, it is the conserved C-terminal region (amino acids 74 to 206) which mediates interaction with the yTAF_II47 HFD (Fig. 4A and D, column 2). We therefore compared this region of yTAF_II25 to the other known histone-like TAF_IIs. In doing this, we noted a significant similarity between the sequences of yTAF_II25/hTAF_II30 and the HFD of the hTAF_II28 family of proteins (Fig. 4B). This similarity predicted the existence of potential αN, α1, and α2 helices within yTAF_II25 and hTAF_II30, the nonconserved sequence corresponding to an insertion in the L2 loop of yTAF_II25 (Fig. 4B). In contrast, the proposed α3 helix of yTAF_II25 contains the conserved D(V, I, L) pair and shows better homology to several other known histone-like proteins than to hTAF_II28.

We next tested the effect of deleting various regions of the yTAF_II25 HFD on heterodimerization with yTAF_II47 (Fig. 4C). Deletion of the αN and a large part of the extended L2 loop [yTAF_II25(84–206)Δ] had no effect on interaction (Fig. 4D, column 3). However, deletion of the LN led to a loss of interaction despite the fact that the proposed α1 remained in this construct [yTAF_II25(90–206)] (Fig. 4D, column 4). At the C terminus, interaction with yTAF_II47 was not affected by deletion up to amino acid 199, leaving intact the proposed α3 helix [yTAF_II25(74–199)] (Fig. 4D, column 5), whereas interaction was abolished when the α3 was truncated (74–186) (column 6). The conserved C-terminal domain of hTAF_II30 interacted with yTAF_II47 as efficiently as yTAF_II25 [hTAF_II30(116–218) and yTAF_II25(74–206)] (Fig. 4D, columns 1 and 2). Together, these results indicate that the conserved C-terminal domain of the yTAF_II25 family contains an HFD which heterodimerizes with yTAF_II47.

To observe direct heterodimerization of yTAF_II47 and yTAF_II25, these proteins were coexpressed in E. coli. A native version of yTAF_II25(74–206) or hTAF_II30(116–218) was coexpressed with a GST fusion of the yTAF_II47 HFD. When expressed alone, the GST-yTAF_II47 fusion is largely insoluble and little soluble protein is recovered on the glutathione-Sepharose beads (Fig. 5A, lane 1). In contrast, coexpression with the yTAF_II25 or hTAF_II30 HFD solubilizes GST-yTAF_II47, which is retained in the form of a complex with each of these proteins on the beads (Fig. 5A, lanes 2 and 3). At first, the GST-yTAF_II47 chimera appears to be more abundant than the untagged yTAF_II25 or hTAF_II30 protein. However, the GST moiety of the chimera stains strongly with Coomassie brilliant blue. When this disproportionate staining is taken into account, the yTAF_II47-yTAF_II25 ratio would be closer to the 1:1 ratio expected for a heterodimeric complex. This is a selective heterodimerization, since neither solubilization nor complex formation was seen when GST-yTAF_II47(1–81) was coexpressed with the HFD of yTAF_II68 (lane 4), previously shown to heterodimerize with yADA1 in this assay (10), yTAF_II40, or yTAF_II19 (data not shown). In control experiments, the native yTAF_II25(74–206) and hTAF_II30(116–218) proteins were insoluble when expressed alone and were not retained on glutathione beads (Fig. 5D, lanes 4 and 5). These results confirm the direct heterodimerization of the TFIID components yTAF_II47 and yTAF_II25, indicating that they form a novel histone-like pair.

FIG. 5.

Coexpression of yTAF_II47 and yTAF_II65 with yTAF_II25 in E. coli. (A) Bacteria were transformed to express the proteins shown above each lane. Following extract preparation, the soluble protein retained on glutathione-Sepharose beads was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and staining with Coomassie brilliant blue. The locations of the GST-yTAF_II fusions and the retained yTAF_II25 and hTAF_II30 proteins are indicated. (B) yTAF_II25-ySPT7 heterodimerization in E. coli. The locations of the GST-ySPT7 fusion and yTAF_II25 are shown. (C) Heterodimerization with yTAF_II25 deletion mutants. The GST-yTAF_II47, GST-yTAF_II65, and GST-SPT7 proteins were coexpressed with the untagged yTAF_II25 deletion mutants as indicated. The soluble proteins retained on glutathione-Sepharose beads are shown. (D) Lack of evidence for yTAF_II25 and hTAF_II30 homodimerization. The GST fusions of yTAF_II25 and hTAF_II30 were expressed alone or in combination with the native HFDs indicated above each lane as in panel A.

Solubilization and complex formation were also observed when GST-yTAF_II47(1–81) was expressed with yTAF_II25(74–206)Δ in which the extended L2 loop has been deleted (Fig. 5C, lane 1). A similar result was seen with yTAF_II25(84–199)Δ, whereas almost no soluble complex was observed with yTAF_II25(90–206)Δ (Fig. 5C, lanes 2 and 3). Thus, as observed in the two-hybrid experiments, yTAF_II47 and yTAF_II25 heterodimerization does not require the extended L2 loop but does require the LN region.

yTAF_II65 contains a functional HFD which mediates selective heterodimerization with yTAF_II25.

The above results reveal the presence of considerably more histone fold proteins in TFIID than originally suspected. This prompted us to examined the sequences of other yTAF_IIs for the presence of potential HFDs. Analysis of the novel yTFIID subunit yTAF_II65 (30) indicates the presence of a potential HFD with similarity to yTAF_II17/dTAF_II40/hTAF_II31 between amino acids 37 and 103 at the N terminus of the protein (Fig. 6A). To determine whether this is a functional domain of the protein, we tested the ability of yTAF_II65 mutants to complement the growth of the yTAF_II65 null strain.

FIG. 6.

yTAF_II65 contains a histone fold motif. The sequence of yTAF_II65 is aligned with the sequences of members of the yTAF_II17 family. The conserved positions are white against a black background, and the positions of the α helices and loops are indicated. The sequences of the m1 and m2 mutants are indicated below the wild-type sequence. (B) The structures of the yTAF_II65 mutants used in complementation experiments are schematized. The HFD is depicted as a black box. TS, temperature sensitive. (C and D) Growth of yeasts plated at the indicated temperatures. 5-FOA, 5-fluoroorotic acid.

Deletions or mutations in the yTAF_II65 HFD were generated (Fig. 6A and B) and used to complement the yTAF_II65 null strain. At 30°C, growth was seen with the full-length protein, whereas no growth was seen when the null strain was complemented with yTAF_II65(1–140) containing only the HFD (Fig. 6C, constructs 2 and 3). Surprisingly, growth at 30°C was also seen using the deletion 103–510, in which the HFD is deleted, and with mutant (1–510)m1, which contains a double amino acid substitution in the α2 helix (Fig. 6C, constructs 4 and 5). These two strains were, however, temperature sensitive, since they did not grow at 37°C, while growth was seen with the wild-type protein (Fig. 6D, constructs 1 to 4). Therefore, in contrast to yTAF_II47, the yTAF_II65 HFD is not sufficient for vegetative growth. The HFD is nevertheless an important functional domain at 37°C, since its deletion or mutation generates a temperature-sensitive phenotype.

We next tested the ability of yTAF_II65(1–140) to heterodimerize with the HFDs of other yTAF_IIs in the two-hybrid assay. Surprisingly, a selective interaction was seen only with yTAF_II25 (Table 1 and Fig. 7B, column 2). Interaction with yTAF_II25 was seen with full-length yTAF_II65 (Fig. 7B, column 1), and this interaction was abolished by mutation m1, which also generated a temperature-sensitive phenotype (Fig. 7A, column 4). Interaction with the yTAF_II65 HFD alone was reduced compared to yTAF_II65(1–510) or (1–140) [yTAF_II65(37–107) (Fig. 7A, column 3). Nevertheless, interaction with yTAF_II25 was totally abolished when two amino acid changes were introduced into the α1 and α2 helices of the yTAF_II65 HFD [yTAF_II65(37–103)m2 in Fig. 6A and 7B, column 5]. Interaction with yTAF_II65 required the conserved C-terminal domain of yTAF_II25 that was required for interaction with yTAF_II47 (data not shown). However, in contrast to yTAF_II47, yTAF_II65 does not interact with hTAF_II30 (Fig. 7B, column 6).

FIG. 7.

The histone fold of yTAF_II65 is required for heterodimerization with yTAF_II25. (A) Structures of the yTAF_II65 deletion mutants used in the two-hybrid assays. (B) The VP16-yTAF_II65 deletions indicated below the graph were assayed in the LexA-yTAF_II25(74–206) strain. For column 6, yTAF_II65 was assayed in the LexA-hTAF_II30(116–218) strain. β-gal, β-galactosidase.

As described above for yTAF_II47, direct yTAF_II65-yTAF_II25 heterodimerization was verified by coexpression in E. coli. The yTAF_II65(36–107) HFD was fused to GST and expressed either alone or in combination with the HFD of yTAF_II25, hTAF_II30, or yTAF_II68. As observed for yTAF_II47, the GST-yTAF_II65 protein was essentially insoluble when expressed alone, while solubilization and complex formation were observed when GST-yTAF_II65 was coexpressed with yTAF_II25(74–206) (Fig. 5A, lanes 5 and 6). In agreement with the two-hybrid assay data, heterodimerization was also observed with yTAF_II25(74–206)Δ and yTAF_II25(84–199)Δ but was strongly reduced with (90–206)Δ (Fig. 5C, lanes 4 to 6). Similarly, no complex was formed with the hTAF_II30 core domain (Fig. 5A, lane 7), and in an additional control no heterodimerization was seen with the yTAF_II68 HFD (Fig. 5A, lane 8). Therefore, while yTAF_II47 can heterodimerize with yTAF_II25 or hTAF_II30, heterodimerization with yTAF_II65 is selective for yTAF_II25 and is not seen with the closely related hTAF_II30. These results indicate that the TFIID component yTAF_II65 selectively and directly heterodimerizes with yTAF_II25 to form an additional histone-like pair.

yTAF_II25 heterodimerizes with the SAGA component ySPT7.

The above results indicate that yTAF_II25 heterodimerizes with yTAF_II47 and yTAF_II65. yTAF_II25 is present in TFIID and SAGA, yet both yTAF_II47 and yTAF_II65 are TFIID specific and are not present in SAGA (29, 30). We therefore sought a potential heterodimerization partner for yTAF_II25 in the SAGA complex. Database searches with the yTAF_II47 HFD showed that ySPT7 contains a potential HFD between amino acids 975 and 1051 with similarity to that of yTAF_II47 (Fig. 1) (32). This observation prompted us to test the ability of the ySPT7 HFD to interact with that of yTAF_II25. In two-hybrid assays, a strong interaction between ySPT7(964–1051) and yTAF_II25(74–206) was observed (Fig. 8A, column 6, and B, column 1). The yTAF_II25-ySPT7 two-hybrid interaction was comparable to that seen with both yTAF_II47 and yTAF_II65 (Fig. 8A, columns 1, 4, 5, and 6).

FIG. 8.

Heterodimerization of the SAGA components yTAF_II25 and ySPT7. (A) The VP16 fusions shown below each column were transformed into the LexA-yTAF_II25(74–206) strain. (B) The LexA fusions shown below each column were transformed into the VP16-SPT7(964–1051) strain. β-gal, β-galactosidase.

Interestingly, the ySPT7 HFD interacted not only with yTAF_II25(74–206) but also with yTAF_II25(90–206), which did not interact with either yTAF_II47 or yTAF_II65 (Fig. 8B, columns 1 and 2). This indicates that the determinants of yTAF_II25 required for interaction with ySPT7 are not exactly the same as those required for interaction with yTAF_II47 and yTAF_II65. In contrast, ySPT7 did not heterodimerize with hTAF_II30(116–218) or with the other SAGA components yTAF_II68 and ADA1, showing that heterodimerization with yTAF_II25 is highly specific (Fig. 8B, columns 3 to 5).

Heterodimerization between yTAF_II25 and ySPT7 was verified directly by coexpression in E. coli. The GST-ySPT7(964–1051) fusion protein was insoluble when expressed alone, whereas in the presence of yTAF_II25(74–206) a soluble heterodimeric complex was formed (Fig. 5B, lanes 1 and 2). Therefore, the SAGA components yTAF_II25 and ySPT7 directly heterodimerize to form a histone-like pair. Consistent with the two-hybrid assay results, no complex was formed with hTAF_II30(116–218) or with yTAF_II68 (lanes 3 and 4). In contrast to what was observed with yTAF_II47 and yTAF_II65, complex formation with yTAF_II25(90–206)Δ was as efficient as that with yTAF_II25(74–206)Δ and yTAF_II25(84–199)Δ (Fig. 5C, lanes 7 to 9). This is in good agreement with the two-hybrid assay data and confirms that there is a differential requirement for the LN loop in heterodimerization with ySPT7 compared with yTAF_II47 and yTAF_II65.

Potential homodimerization of yTAF_II25.

It has previously been reported that both yTAF_II25 and hTAF_II30 can interact with themselves (18, 19). Indeed, in our two-hybrid experiments, we observed an interaction between LexA and VP16 fusions of the yTAF_II25 HFD (Fig. 8A, column 7, and Table 1). We therefore addressed the potential yTAF_II25 homodimerization in the coexpression assay. In contrast to GST-yTAF_II47, GST-yTAF_II65, and GST-ySPT7, both full-length yTAF_II25 and the yTAF_II25 (hTAF_II30) HFD GST fusions were soluble when expressed alone (Fig. 5D, lanes 1 to 3). However, when these GST fusions are coexpressed with the corresponding native HFDs, the untagged HFDs are not retained on the GST-Sepharose beads (Fig. 5D, lanes 6 to 8) as they are when expressed with GST-yTAF_II47, GST-yTAF_II65, or GST-ySPT7. Thus, while we readily observe heterodimerization using this assay, we fail to detect homodimerization of yTAF_II25 and hTAF_II30 via their HFDs.

DISCUSSION

Novel histone-like components in TFIID and SAGA.

We previously reported that hTAF_II135 and yADA1 contained HFDs (10). Here we show that yTAF_II47, yTAF_II25, and yTAF_II65 have significant sequence similarities with other TAF_IIs shown experimentally to contain histone folds. These similarities are comparable to those described for other histone-like proteins (9). Although definitive proof that these are bone fide HFDs will require that their structures be determined, the sequence similarities and the results of our coexpression studies suggest that yTAF_II47, yTAF_II65, yTAF_II25, and ySPT7 are potential histone-like proteins which heterodimerize to form novel pairs in the TFIID and SAGA complexes. Thus, in total there are nine histone-like yTAF_IIs, and the known genetic and physical interactions suggest that they are organized in at least two substructures within TFIID.

Our database searches using the hTAF_II135 HFD showed the presence of a potential HFD in yTAF_II47. Further support for this comes from Sullivan et al., who also identified yTAF_II47 as a histone-like protein using an alternative algorithm (32). The positions of the α helices, notably the α1 helix, proposed by these authors differ significantly from that shown in Fig. 1. This is due mainly to the absence of gaps in the loops in their alignments. Previous alignments of the hTAF_II135 sequence with that of H2A, whose structure has been determined, favor the alignment shown in Fig. 1A. Furthermore, mutation of the conserved RI pair in yTAF_II47 abolishes interaction with yTAF_II25 and the same mutation in hTAF_II135 abolishes interaction with hTAF_II20 (our unpublished data). Both observations point to these amino acids being located in the α1 helix, as indicated in our alignments. A definitive assignment of the precise α1 helix boundaries will require that the structure of this molecule be determined. The presence of the highly conserved D(I, V, L) pair allows the position of the α3 helix in each of the proteins described here to be determined based on the alignment with histone fold proteins of known structure.

Interestingly, these database searches reveal the existence of potential metazoan homologues of yTAF_II47. As the metazoan homologues of yTAF_II25 (hTAF_II30 and dTAF_II24/16) (12, 18) are known TFIID components, it is only to be expected that metazoan TFIID will contain a homologue of their heterodimerization partner, yTAF_II47. It will be interesting to determine whether the human and mouse proteins revealed in these database searches are TFIID components. These observations reinforce the idea that the structure and function of TFIID have been strongly conserved throughout evolution. Hence, it is likely that metazoan TFIID also contains a homologue of yTAF_II65 and, as yTAF_II25 and hTAF_II30 are components of SAGA, TFTC, and PCAF, we would expect these complexes to contain a homologue of ySPT7.

yTAF_II47-yTAF_II25, a novel histone-like pair in yTFIID.

Our results show that the yTAF_II47 HFD is necessary and sufficient for the function of this protein in vivo. The growth of the yTAF_II47 null strain can be complemented by expression of the HFD alone, and its function is abolished by mutation of the well-conserved RI pair in the α1 helix. Interestingly, this same mutation is not lethal in the context of the full-length protein but rather results in a temperature-sensitive phenotype. One interpretation of this result is that other regions of yTAF_II47 missing in the 1–81 mutant act to stabilize the interactions of the mutant protein at permissive temperatures. This is also suggested by the observation that heterodimerization with yTAF_II25 is abolished by mutation m1 in yTAF_II47 in the context of the minimal HFD, yet in the context of the full-length yTAF_II47 mutation m1 does not abolish heterodimerization. It is therefore likely that heterodimerization is necessary for function in vivo. In addition to the three α helices which constitute the minimal HFD, αC and αN helices can often be found. In the yTAF_II47 family, a short region downstream of the α3 helix is conserved, and computer algorithms predict that this conserved sequence may form an α helix. Therefore, this αC helix or another, as-yet-undefined domain in yTAF_II47 may act to stabilize interactions with yTAF_II25 or with other components of yTFIID, but their function becomes evident only when the HFD is mutated.

Our results show that the temperature-sensitive phenotype of the yTAF_II47(1–353)m1 allele can be suppressed by overexpression of several other histone-like yTAF_IIs. At 34°C, suppression is most efficient with the direct heterodimerization partner yTAF_II25. A weaker suppression is seen with yTAF_II40 and yTAF_II19, suggesting that this pair makes close contact with the yTAF_II47-yTAF_II25 pair in native yTFIID. Overexpression of yTAF_II60, but not its heterodimeric partner yTAF_II17, can suppress the temperature-sensitive phenotype of this allele, suggesting that this pair may also in some way contact the yTAF_II47-yTAF_II25 pair. At higher temperatures, however, only the direct heterodimerization partner yTAF_II25 can suppress the temperature-sensitive phenotype of the yTAF_II47(1–353)m1 allele.

We also tested the ability of overexpressed yTAF_II65 to suppress the TAF_II47 temperature-sensitive mutation. Interestingly, we found that even at permissive temperatures, yTAF_II65 overexpression was toxic in the yTAF_II47 temperature-sensitive background but not in the yTAF_II47 wild-type or other backgrounds. One interpretation of this result is that at 30°C, the TAF_II47-TAF_II25 interaction is already sufficiently weakened that when yTAF_II65 is overexpressed it competes with yTAF_II47(1–353)m1 and titrates yTAF_II25. Consequently, there is no longer enough of the TAF_II47-TAF_II25 heterodimer for the yeast to survive, providing evidence that formation of the yTAF_II47-yTAF_II25 heterodimer is essential for viability.

In addition to a genetic interaction, our results show that the yTAF_II47 HFD interacts physically with the conserved C-terminal domain of yTAF_II25 (and hTAF_II30) both in yeast two-hybrid assays and by coexpression in E. coli. The minimum yTAF_II25 region necessary for interaction with yTAF_II47 is located between amino acids 84 and 199. Comparison with the sequences of other families of histone-like TAF_IIs revealed a similarity with the hTAF_II28 family in the α1-L1-α2 region, but the α3 helix of yTAF_II25 shows higher homology to those of other known histone fold proteins. Like yTAF_II40 and ySPT3, yTAF_II25 contains a large insertion in the L2 loop.

This alignment predicts the existence of a possible αN helix in yTAF_II25 and hTAF_II30. The presence of an α helix at this position is also predicted by secondary structure computer algorithms (our unpublished data). Deletion of this helix, however, does not lead to a loss of interaction with yTAF_II47 or yTAF_II65. Interestingly, the putative LN loop is highly conserved in the yTAF_II25/hTAF_II30 family (12), and its deletion results in a loss of interaction with yTAF_II47 and yTAF_II65, but not with ySPT7. This strongly suggests that the LN region plays a critical role in heterodimerization by making direct contacts with yTAF_II47 and yTAF_II65 as is seen in the hTAF_II28-hTAF_II18 pair (5), while in the yTAF_II25-ySPT7 heterodimer this interaction either does not take place or is not essential for complex formation.

Introduction of stop codons truncating the α2 helix abolishes yTAF_II25 function. Moreover, in a screen for yTAF_II25 temperature-sensitive mutants, several alleles with substitutions at G101 were found (29). This position corresponds to the G residue in the L1 loop, highly conserved in both the yTAF_II25 and hTAF_II28 families. A more detailed functional analysis of yTAF_II25 reveals a tight correlation between its ability to interact with its heterodimerization partners and its function (D. Kirschner and L. Tora, unpublished data).

It has previously been suggested that yTAF_II25 and hTAF_II30 may homodimerize (18, 19). Klebanow et al. reported two-hybrid interactions using full-length yTAF_II25 fusions. Our two-hybrid experiments show that this potential homodimerization requires only the HFD of yTAF_II25. As it is possible that the two-hybrid interaction is indirect, we wished to visualize homodimerization directly. However, while the GST-yTAF_II25(74–206) fusion was relatively soluble in E. coli when expressed alone, no homodimerization was seen when it was coexpressed with native yTAF_II25(74–206). Therefore, under the conditions used to observe heterodimerization via the HFD, we detected no yTAF_II25 homodimerization. Consequently, it is unlikely that the observed yTAF_II25 oligomerization represents the formation of a histone-like homodimer. This does not, however, exclude the possibility that oligomerization of yTAF_II25 may occur in some other way, involving for example loop-loop interactions or via the exposed hydrophilic faces of the α helices.

yTAF_II65-yTAF_II25, a novel histone-like pair in yTFIID.

Here we have identified an HFD in the N terminus of yTAF_II65. This HFD shows high homology to the yTAF_II17/hTAF_II31 family. Both two-hybrid analysis and coexpression in E. coli indicate that the yTAF_II65 HFD mediates a selective heterodimerization with the same yTAF_II25 domain that is required for interaction with yTAF_II47. The yTAF_II65 HFD alone is sufficient to allow interaction with yTAF_II25, and interaction is abolished by mutations within this domain. These results strongly suggest that yTAF_II25-yTAF_II65 form an additional histone-like pair in TFIID. Unlike yTAF_II47, yTAF_II65 does not interact with the hTAF_II30 HFD despite their high homology. We have previously reported that hTAF_II20, but not yTAF_II68, interacts with hTAF_II135. In this case, an important determinant for specificity was mapped to the L2-α3 region of the hTAF_II20/yTAF_II68 HFD (10). This observation together with those reported here indicate that there is a strict specificity code which determines the choice of a heterodimerization partner.

The yTAF_II65 HFD is not sufficient for growth and in fact is not essential at 30°C. Nevertheless, the yTAF_II65 HFD contributes to function, since its deletion or mutation results in a temperature-sensitive phenotype. It has previously been shown that introduction of proline residues in the α2 helix of yTAF_II60 and yTAF_II17 results in a temperature-sensitive phenotype (24). This is also true for yTAF_II65, since the m1 mutation which introduces two prolines generates a temperature-sensitive phenotype. This same mutant abolishes interaction with yTAF_II25. In yTAF_II47 and yTAF_II68 the HFD is necessary and sufficient for growth. In contrast, however, yTAF_II65 must contain another essential domain(s) located between amino acids 103 and 510. Further complementation analysis reveals that this essential domain(s) is located between amino acids 161 and 406 (our unpublished data). In this context, it is interesting that, as with yTAF_II47, we attempted to suppress the temperature-sensitive phenotype of this mutation by overexpression of other yTAF_IIs. In this case, however, the only genetic interaction detected was with yTAF_II68, whose overexpression suppressed the temperature-sensitive phenotype not only of yTAF_II65(1–510)m1 but also of yTAF_II65(103–510), in which the HFD is deleted (our unpublished data). There is therefore a genetic interaction between yTAF_II68 and yTAF_II65, but this involves a functional domain(s) of yTAF_II65 other than the HFD.

yTAF_II25-ySPT7 a novel histone-like pair in ySAGA.

We have shown here that yTAF_II25 can heterodimerize with yTAF_II47 and yTAF_II65. Both of these heterodimerization partners are TFIID specific and are not present in SAGA. yTAF_II25 is, however, a SAGA subunit, and therefore, an additional heterodimerization partner for yTAF_II25 must exist in this complex. Our results, and those of Sullivan et al. (32), identified an HFD in ySPT7, and we show that this HFD heterodimerizes with the yTAF_II25 HFD in both the two-hybrid and bacterial coexpression assays. This HFD is found in the C-terminal half of the protein, which has been reported to partially rescue the phenotype of the SPT7 null strain (11). Genetic studies have also shown that mutations in SPT7 have the same severe phenotype as mutations in ADA1 and SPT20. Mutation of each of these proteins completely disrupts the SAGA complex, showing that they are critical for its structural integrity (31). We have previously reported a direct interaction between the TAF_II and ADA families of proteins through the heterodimerization of yTAF_II68 with yADA1 (10). Here we show that the histone fold is also the interface between the TAF_II and the SPT families in SAGA. Together with the genetic studies, this suggests that the yTAF_II68-yADA1 and yTAF_II25-SPT7 pairs are both key structural elements of SAGA.

Implications for TFIID structure.

In summary, our present results show yTFIID to comprise at least nine histone-like yTAF_IIs rather than the three originally described. These yTAF_IIs assemble into five heterodimeric pairs (yTAF_II60-yTAF_II17, yTAF_II40-yTAF_II19, yTAF_II68-yTAF_II48, yTAF_II25-yTAF_II47, and yTAF_II25-yTAF_II65). Amongst these, yTAF_II25 appears to be a key player which can form two distinct heterodimers in TFIID. Previous results have indicated that yTAF_II47 can be coimmunoprecipitated with yTAF_II65 in extracts from yeast strains harboring an epitope-tagged yTAF_II65 (30). This excludes the possibility that the yTAF_II25-yTAF_II47 and yTAF_II25-yTAF_II65 pairs are present in distinct populations of yTFIID complexes; instead, it indicates that the two pairs coexist in the same population of yTFIID.

It has previously been shown that a temperature-sensitive allele of yTAF_II17 can be suppressed by overexpression of yTAF_II60 or yTAF_II68 and that a temperature-sensitive allele of yTAF_II60 can be suppressed by overexpressed yTAF_II17 and yTAF_II68 (24). Furthermore, yTAF_II48 overexpression can suppress a temperature-sensitive mutant of yTAF_II68 (28). These genetic interactions were interpreted as providing evidence of an octamer-like substructure in TFIID (24). Our present results showing genetic interactions between yTAF_II47 and yTAF_II25, yTAF_II40, yTAF_II19, and yTAF_II60 suggest a more complex picture which is difficult to interpret in the context of a single octamer-like substructure. Altogether, the existing data suggest that there may be at least two potential substructures, one composed of yTAF_II68, yTAF_II60, yTAF_II48, and yTAF_II17 as described by Michel et al. and Reese et al. (24, 28) and the other composed minimally of yTAF_II47, yTAF_II40, yTAF_II25, and yTAF_II19 as reported here. The suppression of the yTAF_II47 temperature-sensitive allele by overexpressed yTAF_II60 further implies interplay between the substructures via yTAF_II60. The stoichiometry of the yTAF_IIs present within each substructure and whether they interact in a way analogous to the core histones in the nucleosome octamer remain to be determined. Similarly, it is not clear whether the yTAF_II25-yTAF_II65 pair associates with the yTAF_II25-yTAF_II47 substructure via the yTAF_II25-yTAF_II25 interactions discussed above or whether it is located elsewhere within TFIID.