Report

Induced α Helix in the VP16 Activation Domain upon Binding to a Human TAF

See allHide authors and affiliations

Science  29 Aug 1997:
Vol. 277, Issue 5330, pp. 1310-1313
DOI: 10.1126/science.277.5330.1310

Abstract

Activation domains are functional modules that enable sequence-specific DNA binding proteins to stimulate transcription. The structural basis for the function of activation domains is poorly understood. A combination of nuclear magnetic resonance (NMR) and biochemical experiments revealed that the minimal acidic activation domain of the herpes simplex virus VP16 protein undergoes an induced transition from random coil to α helix upon binding to its target protein, hTAFII31 (a human TFIID TATA boxbinding protein-associated factor). Identification of the two hydrophobic residues that make nonpolar contacts suggests a general recognition motif of acidic activation domains for hTAFII31.

Activation of transcription in eukaryotes is directed by regulatory proteins that recruit the transcriptional machinery and chromatin remodeling factors to the promoter. Typically, these regulatory proteins are modular, having distinct domains for sequence-specific binding to DNA and for transcriptional activation through interactions with other proteins (1). Activation domains are classified according to the preponderance of amino acid residues such as glutamine, proline, and those bearing acidic side chains. Of these classes, the acidic activators have been the most extensively studied (2). Notwithstanding the gains that have been made in identifying the targets of acidic activation domains (1) and elucidating the importance of particular residues for their function (3), the structural basis for the ability of activation domains to stimulate transcription remains poorly understood (2, 4). Here, we report that the activation domain of the herpes simplex virus VP16 protein undergoes an induced coil-to-helix transition upon interaction with its target protein hTAFII31, with residues along one face of the nascent helix making intermolecular contacts to hTAFII31.

The acidic activation domains of VP16 and tumor suppressor p53 directly target hTAFII31 and its Drosophila homolog, dTAFII40; the strength of this interaction correlates with the ability to activate transcription in vitro (5-7). The molecular interaction has been mapped to the NH2-terminal 181 amino acids of hTAFII31 (TAF1–181) and a COOH-terminal segment of the VP16 activation domain (VP16C, residues 452 to 490) (5). Because the stability of TAF1–181 is poor, we overexpressed a smaller fragment, TAF1–140, which comprises the region of highest homology to dTAFII40 (8) and binds VP16C as tightly as does TAF1–181 (9). TAF1–140 was soluble up to ∼300 μM in pH 6.2 buffer and thus was deemed suitable for NMR studies.

To analyze the interaction between TAF1–140 and VP16C, we performed 1H-15N heteronuclear single-quantum coherence (HSQC) NMR experiments with15Nlabeled VP16C and varying amounts of unlabeled TAF1–140 (10). The enrichment of VP16C with 15N permitted selective detection of signals from VP16C, but not from unlabeled TAF1–140. The limited dispersion of the backbone 15N and1HN chemical shifts in the HSQC spectrum of15N-labeled VP16C alone (Fig.1A) and the intermediate values (∼7 Hz) of 1HN-1HCα coupling constants (11) indicated that VP16C alone has negligible secondary structure (12). Titration of15N-labeled VP16C with unlabeled TAF1–140 resulted in progressive rather than bimodal changes of the backbone 15N and 1HN chemical shifts (Fig. 1A), thus indicating that VP16C interacts weakly with TAF1–140 and hence exchanges rapidly between the free and bound states on the NMR time scale (13). Sequential assignment of the HSQC cross-peaks (14) established that the backbone-perturbed residues are located within a region at the COOH-terminal end of VP16C that encompasses residues 475 to 484 (Fig. 1B).

Figure 1

(A) Expanded1H-15N HSQC spectra of 15N-labeled VP16C (900 μM) in the absence (black) or presence (red) of TAF1–140 (300 μM). Significantly shifted cross-peaks are indicated. Glycine cross-peaks are not shown; the cross-peak from Gly484 shifted in the presence of TAF1–140. (B and C) Histogram showing 1HN and15N (B) and β1H (C) chemical shift changes in VP16C induced by binding of TAF1–140. The positions of amino acids are indicated. Chemical shifts of the β protons were determined by 1H-15N NOESY-HSQC and 1H-15N TOCSY-HSQC experiments. When diastereotopic β-proton signals were observed, only the chemical shift change of the proton having the largest effect is indicated. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Gross changes in backbone chemical shifts can generally be ascribed to global alterations in folded structure. To assess the importance of the side chains, we analyzed the chemical shift perturbations of the β protons in VP16C upon complexation with TAF1–140 (15). Significant chemical shift changes were observed only for the β protons of Asp472, Phe479, Leu483, and Asp486 (Fig.1C). These four residues lie within and adjacent to the region of VP16C that is suggested by backbone chemical shift perturbation to undergo an induced folding transition.

To analyze independently the importance of residues in VP16C for both the interaction with TAF1–140and activation of transcription, we performed in vitro biochemical assays with a series of VP16C deletion mutant proteins (Fig. 2A). For detection of TAF-binding activity, each deletion mutant was fused to glutathione-S-transferase (GST) and analyzed for the ability to pull down TAF1–140(Fig. 2B) (16). Deletion of residues 475 to 490 from GST-VP16C abolished TAF binding, whereas deletion of residues 452 to 468 retained binding activity. Further deletion of residues 486 to 490 had no detectable effect; however, removal of residues 469 to 473 or 481 to 485 resulted in a loss of binding activity. For assessment of the ability of the deletion mutants to activate transcription, each mutant was fused to the yeast GAL4 DNA binding domain (residues 1 to 147) and analyzed for the ability to activate transcription in vitro with HeLa nuclear extracts (Fig. 2C) (17). GAL4-VP16C stimulated transcription of the reporter construct containing five GAL4 recognition sites (5). A fusion construct in which residues 452 to 468 of VP16C were deleted activated transcription as strongly as did GAL4-VP16C itself. However, deletion of residues 475 to 490 reduced activation potential; this was slightly above GAL4 activation alone. Although deletion of residues 486 to 490 had no detectable effect, further deletion of residues 469 to 473 or 481 to 485 reduced transcriptional activity. Thus, the transcriptional activity of VP16C is directly correlated with the strength of its binding to TAF1–140; this is consistent with the notion that the interaction between these two proteins is responsible for the activation signal observed in our in vitro assays. Moreover, the COOH-terminal segment of VP16C(VP16469–485, residues 469 to 485) is necessary and sufficient to bind TAF1–140 and activate transcription in vitro, and it corresponds to the region that was mapped through NMR experiments to interact with TAF1–140.

Figure 2

In vitro biochemical experiments. (A) Schematic representation of VP16 deletion mutant proteins used for in vitro protein-protein interaction and transcription assays. Residues whose backbone NH chemical shifts (light gray) and β proton chemical shifts (black) were significantly changed upon binding to TAF1–140 are indicated. (B) In vitro protein-protein interaction assays. GST and GST-VP16 beads were incubated with TAF1–140 in buffer containing 50 mM NaCl and 0.01% NP-40. After extensive washing, the bound proteins were resolved by SDS-PAGE. Lanes 1 and 2 show the protein markers and TAF1–140 (20% of the input TAF1–140), respectively. Binding of TAF1–140 to GST-VP16C(lane 3), GST-VP16469–490 (lane 5), and GST-VP16469–485 (lanes 6 and 9) is evident. TAF1–140 is not retained on GST-VP16452–474(lane 4) and GST (lanes 8 and 12) resins. TAF1–140 binds to GST-VP16474–485 (lane 7) and GST-VP16469–480 (lane 10) resins much more weakly than to GST-VP16C. Replacement of both Phe479 and Leu483 with Ala in GST-VP16469–485 (lane 11) greatly reduced binding affinity. The position of TAF1–140is indicated by an asterisk. (C) In vitro transcription assays. Transcriptional activation by GAL4-VP16 mutant proteins was assayed in HeLa nuclear extracts. The reactions contained 2 pmol of the purified proteins and 100 ng of G5BCAT template containing the adenovirus E1b promoter linked to five GAL4 recognition sites. The products of the transcription reactions were analyzed by primer extension. Lane 1 shows basal transcription without GAL4 proteins. Transcriptional activation by GAL4-VP16C (lane 3), GAL4-VP16469–490 (lane 5), and GAL4-VP16469–485 (lanes 6 and 9) is greater than that observed for GAL4 alone (lane 2). GAL4-VP16452–474 (lane 4), GAL4-VP16474–485 (lane 7), and GAL4-VP16469–480 (lane 10) activate transcription much less than does GAL4-VP16C (lane 3). Lane 8 shows activated transcription by GAL4-VP16NC, which contains both the NH2-terminal and COOH-terminal subdomains (residues 413 to 490) of the VP16 activation domain. Replacement of both Phe479 and Leu483 with Ala greatly reduced transcriptional activation (lane 11). The position of the extension products is indicated by an arrowhead. The weak activation signal generated by GAL41–147 (lane 2) has been observed previously (25).

To determine the structure of VP16469 485 bound to TAF1–140, we performed transferred nuclear Overhauser effect (TRNOE) experiments. TRNOE relies on rapid exchange between the free and bound states for a relatively small ligand in the presence of its macromolecular receptor. Under conditions of rapid exchange, negative NOEs conveying conformational information about the bound ligand are transferred to the resonances of the free ligand (18). In the ideal case, NOEs from the free ligand itself approach zero because of its low molecular weight; hence, NOEs can be detected primarily from the bound ligand in the presence of a substoichiometric amount of the receptor. As anticipated, free VP16469–485 exhibited very few NOEs; only sequential NOEs between CαH and NH protons [d α N(i,i+1) NOEs] were clearly observed, indicative of an extended random-coil structure (19). In contrast, VP16469–485 in the presence of 0.1 mole equivalents of TAF1–140 showed numerous NOE cross-peaks, including a substantial number arising fromd NN(i,i+1),d α N(i,i), and d αβ(i,i+3) NOEs. The overall pattern of the NOE connectivities (Fig.3B) is characteristic of that observed for α-helical secondary structure, especially in the region from Asp472 to Leu483. Thus, residues 472 to 483 of VP16C, and perhaps even residues flanking this region, adopt an α-helical structure when bound to TAF1–140(20).

Figure 3

(A) Amide region of a 200-ms NOESY spectrum of VP16469–485 (3 mM) in the presence of TAF1–140 (0.3 mM). The identities of residues that exhibit NOE cross-peaks in this region are indicated. The intensities of transferred NOE cross-peaks are generally weak when compared with the diagonal peaks because the diagonal peaks arise from both free and bound ligands (26). (B) Summary of the NOEs observed from the TRNOE experiments. The thickness of the lines indicates the relative intensities of the NOE cross-peaks. (C) Helical wheel presentation of residues 472 to 483 of VP16. The residues whose β-proton chemical shifts were changed upon binding to TAF1–140 are indicated by boxes. (D) Sequence comparison of regions of VP16C, human p53, and human NF-κB p65. The activation domain sequences are aligned by the FXXΦΦ moiety. Purple circles mark the residues of VP16C whose β protons are perturbed upon binding to TAF1–140. Green circles mark the residues of the p53 activation domain that directly contact MDM2 (23). Acidic residues are highlighted in red. Although the β-proton chemical shift of Asp486 was significantly perturbed upon binding to TAF1–140, the present biochemical experiments indicate that Asp486 is not essential for TAF binding and transcriptional activation. Black coils above the sequences of VP16C and p53 represent the regions over which α helix induction is observed upon binding to TAF1–140 and MDM2, respectively. Amino acid abbreviations are as in Fig. 1.

The projection of residues 472 to 483 of the VP16 activation domain onto a helical wheel is shown in Fig. 3C. Three of the four residues whose side chain β protons are perturbed upon binding to TAF1–140 lie along one face of the helix. These three residues—Asp472, Phe479, and Leu483—may directly contact chemically complementary residues of TAF1–140. Consistent with this idea, replacement of Phe479 and Leu483 with Ala reduced TAF1–140 binding affinity and transcriptional activation (Fig. 2, B and C, respectively) (21). Whereas Phe479 and Leu483 of VP16Cpresumably make hydrophobic contacts to TAF1–140, Asp472 may participate in salt-bridge or hydrogen-bonding interactions. The fourth residue that exhibits β-proton chemical shift perturbation upon binding to TAF1–140, Asp486, can be deleted without significant loss of binding or transcriptional activation. This suggests that Asp486makes little energetic contribution to the protein-protein interaction, or possibly that the new COOH-terminal carboxyl generated through truncation functionally replaces the Asp486 side chain.

The acidic activation domains of p53 and nuclear factor κB (NF-κB) p65 have also been reported to bind hTAFII31 (6,22). Sequence comparison of the relevant regions of VP16C, p53, and p65 reveals a pattern of similarity (Fig.3D): All three activation domains contain a Phe separated by a two-residue spacer from a hydrophobic doublet. Our NMR studies provide direct evidence that at least two of these three residues contact TAF1–140. Furthermore, mutation of the hydrophobic residues in this motif abrogates binding of the p53 activation domain to hTAFII31 and in vitro transcriptional activation (6); the high propensity of α helix formation by the p65 activation domain has been noted, as has the importance of Phe for function (23). Thus, this FXXΦΦ motif may represent a general recognition element of acidic activation domains for TAFII31. The recent crystal structure of the p53 activation domain bound to the attenuator protein MDM2 reveals that the segment containing the FXXΦΦ moiety folds to form an α helix, from which the three hydrophobic residues project to make nonpolar contacts with MDM2 (24). Our study provides support for the hypothesis that the MDM2-p53 interaction mimics that of TAFII31–acidic activation domains, even though MDM2 and TAFII31 appear to be structurally unrelated (22,24).

A long-standing puzzle relates to the specific role of the acidic residues in the acidic activation domain. Evidence suggests that acidic residues play an important role in the function of acidic activation domains (3); however, the results presented here and elsewhere (3) indicate that hydrophobic residues also play an important role. Although 5 of the 17 residues that make up the minimal activation peptide VP16469–485 are acidic, only one of these, Asp472, exhibits significant perturbation of its β-proton chemical shift upon binding to TAF1–140. Asp472 may make a direct, specific contact; however, this contact is apparently not conserved in the activation domains of p53 and p65. Moreover, the positions of the acidic residues in acidic activation domains generally appear to be unimportant. This seeming paradox can be resolved by a model in which the acidic residues establish long-range electrostatic interactions with hTAFII31. Such electrostatic forces would attract basic hTAFII31 over relatively long distances in solution, thereby increasing the rate at which the activation domain locates its target. Once the activation domain and hTAFII31 come into close range, the activation domain undergoes an induced structural transition to an α helix, thereby enabling the establishment of direct hydrophobic contacts with nonpolar residues of hTAFII31. Because such folding transitions are highly cooperative, the coupling of folding to targeting by activation domains provides a mechanism whereby multiple weak interactions can produce a pronounced biologic response.

REFERENCES AND NOTES

View Abstract

Navigate This Article