Conserved Themes in Target Recognition by the PAH1 and PAH2 Domains of the Sin3 Transcriptional Corepressor

Summary

The recruitment of chromatin-modifying coregulator complexes by transcription factors to specific sites of the genome constitutes an important step in many eukaryotic transcriptional regulatory pathways. The histone deacetylase-associated Sin3 corepressor complex is recruited by a large and diverse array of transcription factors through direct interactions with the N-terminal PAH domains of Sin3. Here we describe the solution structures of the mSin3A PAH1 domain in the apo-form and when bound to SAP25, a component of the corepressor complex. Unlike the apo-mSin3A PAH2 domain, the apo-PAH1 domain is conformationally pure and is largely, but not completely, folded. Portions of the interacting segments of both mSin3A PAH1 and SAP25 undergo folding upon complex formation. SAP25 binds through an amphipathic helix to a predominantly hydrophobic cleft on the surface of PAH1. Remarkably, the orientation of the helix is reversed compared to that adopted by NRSF, a transcription factor unrelated to SAP25, upon binding to the mSin3B PAH1 domain. The reversal in helical orientations is correlated with a reversal in the underlying PAH1-interaction motifs, echoing a theme previously described for the mSin3A PAH2 domain. The definition of these so-called type I and type II PAH1-interaction motifs has allowed us to predict the precise location of these motifs within previously experimentally-characterized PAH1 binders. Finally, we explore the specificity determinants of protein-protein interactions involving the PAH1 and PAH2 domains. These studies reveal that even conservative replacements of PAH2 residues with equivalent PAH1 residues are sufficient to dramatically alter the affinity and specificity of these protein-protein interactions.

Introduction

The transcriptional status of a eukaryotic gene is determined, in part, by the pattern of post-translational modifications of histones in the promoter regions.¹ Chromatin-modifying activities such as histone deacetylases (HDACs) play an important role in this process erasing the marks established by acetyltransferases and setting the stage for further alterations of the histone code. In mammalian cells, the nuclear HDACs including HDAC1, HDAC2, and HDAC3 are typically found in large multi-subunit complexes with coregulator proteins that are in turn recruited by sequence-specific DNA binding transcription factors and chromatin-binding proteins to specific regions of the genome.^2-7 The Sin3 corepressor complex is one of the better-characterized HDAC-associated coregulator complexes and is found in a broad range of species from yeast to human.⁵ Besides its well-documented role in negatively regulating promoter-dependent transcription, the Sin3 complex is also implicated in heterochromatin silencing,^7-10 certain types of DNA repair,^11,12 and in the suppression of spurious intragenic transcription initiation.^13-16 Underscoring the Sin3 protein’s fundamental role in cellular physiology, sin3 knockouts are characterized by embryonic lethality or defects in development caused by deregulation of genes that control proliferation, homeostasis, and apoptosis.^17,18

The mammalian Sin3 corepressor complex consists of at least nine subunits besides Sin3 including SAP30, SAP45/Sds3, SAP130, SAP180/BCAA, RBP1, HDAC1, HDAC2, RbAp46, and RbAp48.^19-25 In mammals, Sin3 exists as two closely related polypeptides designated mSin3A and mSin3B each serving not only a scaffolding function for the assembly of the corepressor complex but also being available for direct recruitment by a surprisingly large and diverse array of transcription factors. The Sin3 proteins harbor six conserved regions, four of which are thought to be related to each other and were originally predicted to form a pair of amphipathic helices (PAH), a histone deacetylase interaction domain (HID), and a highly conserved C-terminal region (HCR). In the emerging view, the PAH1 and PAH2 domains are broadly available for recruitment by and interactions with diverse transcription factors,²⁶ whereas the segment spanning the PAH3, HID and PAH4 regions is involved in the scaffolding function of Sin3 by serving as interaction sites for other subunits of the corepressor complex.⁵ The sequence and structural requirements for recruitment of the mammalian Sin3 proteins via the PAH2 domain are well-defined.^27-31 By contrast, less is known about the mode of interaction with and recruitment of the Sin3 proteins via the PAH1 domain. Although the PAH2 domain is the closest relative of the PAH1 domain, the domains exhibit distinct patterns of sequence conservation, suggesting dissimilar modes of engaging targets (Figure 1). Like the PAH2 domain, the PAH1 domain is targeted by numerous factors including the promyelocytic leukemia-associated zinc finger (PLZF),³² the neuronal restrictive silencer factor (NRSF),³³ the PHD-containing factor 1 (Pf1),³⁴ a negative regulator of phospholipid biosynthesis Opi1,³⁵ the myelin transcription factor 1 (Myt1),³⁶ the nuclear receptor corepressor NCoR,³⁷ the host cell-proliferation factor 1 (HCF1),³⁸ and the 25 kDa Sin3 associated polypeptide (SAP25).³⁹ Although the structure of the NRSF Sin3 interaction domain (SID) in complex with mSin3B PAH1 has been reported,³³ the rules governing PAH1 interactions are not well understood, as the aforementioned PAH1-interacting proteins share little similarity with each other at the sequence level. We present below the solution structure of the mSin3A PAH1 domain in the apo-form and when bound to SAP25 SID, a recently identified component of the corepressor complex.³⁹ We identify the conformational transitions associated with the binding event, compare the interaction modes of SAP25 and NRSF SIDs with the PAH1 domains, attempt to derive the rules for PAH1 associations, and based on this knowledge, predict the precise locations of the SIDs for various PAH1 interactors along with their binding modes. Finally, we explore the determinants of specificity of protein-protein interactions mediated by the closely related mSin3 PAH1 and PAH2 domains.

Figure 1.

The Sin3 PAH1 domain is highly conserved and distinct from the other PAH domains. CLUSTAL W-guided multiple sequence alignment of Sin3 PAH1 domains from a variety of species (top panel; species abbreviations and GenBank accession codes: Mm: Mus musculus (Sin3A: Q60520; Sin3B: Q62141); Hs: Homo sapiens (Sin3A: Q96ST3; Sin3B: O75182); Xl: Xenopus laevis (AAD34644); Dm: Drosophila melanogaster (CAA07550); Sp: Saccharomyces pombe (Pst1: Q09750; Pst3: O74755); Sc: Saccharomyces cerevisiae (P22579); At: Arabidopsis thaliana (O48686); Ce: Caenorhabditis elegans (CAN86581)). Invariant and conserved residues are highlighted in yellow and blue, respectively. The cartoon on top depicts the location of various helices in the NMR structure of the SAP25-bound mSin3A PAH1. Residues forming the SAP25 SID-binding surface are indicated by solid circles. Multiple sequence alignment of the mouse Sin3A and Sin3B PAH1 and PAH2 domains (middle panel). Residues that are simultaneously invariant in the respective PAH1 and PAH2 domains are highlighted in green. Residues forming the Mad1/Mxd1 SID-binding surface of mSin3A PAH2 are indicated by solid circles below the alignment. Putative specificity determining residues that were mutated in our studies are indicated by inverted triangles. Alignment of the mSin3A and mSin3B PAH3 domain with other PAH1 and PAH2 domains (bottom panel). Identical residues across all three PAH domains are highlighted in yellow. Putative specificity determining residues of mSin3 PAH3 are highlighted in green.

Results and Discussion

Low-resolution analysis of mSin3A PAH1-SAP25 interactions

The ¹H-¹⁵N HSQC spectrum of the apo-mSin3A PAH1 domain, unlike that of the apo-mSin3A PAH2 domain,⁴⁰ is characterized by excellent chemical shift dispersion and a single set of correlations of uniform intensity indicative of a folded domain sampling a set of closely related conformations (Supplementary Figure S1(a)). A set of minor resonances could also be detected in the spectrum that is attributed to the incomplete conversion of a glutamine residue at the N-terminus to pyroglutamic acid. To test the functionality of the PAH1 construct, we titrated the protein with SAP25. Previous studies had coarsely mapped the Sin3 interaction domain (SID) of SAP25 to an ∼60 amino acid segment at the carboxy-terminus.³⁹ Titration of ¹⁵N-labeled mSin3A PAH1 with approximately one equivalent of unlabeled SAP25 SID led to substantial changes in resonance positions for a number of residues characteristic of a specific interaction (Supplementary Figure S1(b)). This titration as well as the reverse titration of ¹⁵N-labeled SAP25 SID with unlabeled mSin3A PAH1 were characterized by the disappearance of resonances of the apo-protein and appearance of resonances at new locations indicative of a complex with slow dissociation kinetics on the NMR chemical shift timescale (Supplementary Figure S1 and S2). Isothermal titration calorimetry (ITC) analysis of the interaction between mSin3A PAH1 and SAP25 SID yielded an equilibrium dissociation constant of 134 ± 13 nM, confirming a high-affinity interaction (Supplementary Figure S3(a)).

Structure and conformation of mSin3A PAH1 and SAP25 SID in the apo and complexed states

To gain insights into the mode of the SID-PAH interaction and to characterize the conformational transitions associated with the binding event, we determined the solution structures of the apo-mSin3A PAH1 domain and that of the SAP25 SID-mSin3A PAH1 complex. The backbone and side chain resonances of each polypeptide were almost completely assigned using standard triple-resonance approaches. The structures were determined using distance and torsion angle restraints derived from ¹H-¹H NOEs and chemical shifts of backbone nuclei. Structures of reasonable precision and in good agreement with input experimental data were obtained (Figure 2(a) and 2(b); Table 1). Since the ¹H-¹⁵N correlated spectrum of apo-SAP25 SID was characterized by poor amide proton chemical shift dispersion and uniformly intense correlations indicative of a natively unfolded polypeptide (Supplementary Figure S2(a)), conformational transitions accompanying PAH1 binding were inferred from an analysis of the secondary chemical shifts of backbone nuclei in the apo- and PAH1-bound states.

Figure 2.

Solution NMR structures of the apo-mSin3A PAH1 domain and the SAP25 SID-mSin3A PAH1 complex. Stereographic views of the ensemble of twenty NMR structures of (a) the apo-mSin3A PAH1 domain and (b) the SAP25 SID-mSin3A PAH1 complex following a best-fit superposition of backbone atoms in structurally ordered regions. Ribbon diagrams of the representative structures of the respective ensembles are shown in panels (c) and (d). The SAP25 SID is depicted in magenta whereas the mSin3A PAH1 domain is colored green. Residues 151-186 of SAP25, although considered for the structure calculations, are essentially unstructured and have been omitted in these views for clarity.

Table 1.

NMR Structure Determination Statistics

	mSin3A PAH1	SAP25 SID-mSin3A PAH1 complex
Restraint Statistics
NOE-based distance restraints	1506	1913
Unambiguous NOE-based restraints	1228	1420
Intraresidue	605	800
Sequential (|i-j|=1)	235	323
Medium-range (1 <|i-j|≤4)	189	175
Intramolecular long-range (|i-j|>4)	199	122
Intermolecular	-	173
Ambiguous NOE-based restraints	278	320
Intramolecular NOE-based restraints	1506 PAH1	358 SID, 1382 PAH1
Hydrogen bonding distance restraints	46	80
Torsion angle restraints	78 (36 ϕ, 36 ψ, 6 χ1)	127 (58 ϕ, 57 ψ, 12 χ1)
Structure Quality of NMR Ensemble
Restraint satisfaction
RMS differences for distance restraints	0.005 ± 0.001 Å	0.009 ± 0.001 Å
RMS differences for torsion angle restraints	0.054° ± 0.024°	0.079° ± 0.023°
Deviations from ideal covalent geometry
Bond lengths	0.001 ± 0.000 Å	0.001 ± 0.000 Å
Bond angles	0.273° ± 0.005°	0.306° ± 0.010°
Impropers	0.117° ± 0.007°	0.182° ± 0.013°
Ramachandran plot statistics
Residues in most favored regions	83.6%	71.3 (83.1%)a
Residues in allowed regions	15.8%	27.1 (16.3%)a
Residues in disallowed regions	0.6%	1.6 (0.6%)a
Average Atomic RMSDs from the Average Structure
All atoms	2.24 Å	8.30 Å
All atoms in ordered regionsb	1.15 Å	1.09 Å
Backbone atoms (N, Cα, C’)
All residues	1.50 Å	8.38 Å
Residues in ordered regionsb	0.44 Å	0.42 Å
Residues in helical regions	0.31 Å	0.31 Å

^aStatistics provided are for the entire mSin3A PAH1 domain and residues 126-150 of SAP25 SID

^bOrdered regions in the free mSin3A PAH1 domain encompass residues 127-186; in the binary complex, these include residues 122-188 of mSin3A PAH1 and 132-145 of SAP25 SID

The apo-mSin3A PAH1 domain adopts the canonical left-handed four-helix bundle fold described previously for the two N-terminal mSin3 PAH domains (Figure 2(c)).^28-31,33,41 The four helices of apo-mSin3A PAH1 span residues Tyr129-Gln136, Gln142-Lys155, Pro162-Leu171, and Asp177-Phe182. The helical spans are identical to those described previously for the NRSF SID-bound mSin3B PAH1 domain,³³ except the α1 helix of apo-mSin3A PAH1 is significantly shorter lacking six residues at the N-terminus and the α4 helix is shorter by two residues at the C-terminus. The backbone segment spanning Tyr129-Phe182 of apo-mSin3A PAH1 superimposes well with the equivalent segment of mSin3B PAH1 with an atomic root-mean-square deviation (RMSD) of 0.89 Å, consistent with the 79% sequence identity shared by these segments. In the SAP25 SID complex, the locations of the mSin3A PAH1 helices are largely unchanged, except the α1 helix is extended by five residues at the N-terminus and is initiated by Glu124 while the α4 helix is extended by two residues at the C-terminus and is terminated by Thr184. This change is accompanied by an enhancement in the precision of the NMR structures both in this region as well as in the segment immediately following the α4 helix extending up to Pro187 (Figure 2(b) and Supplementary Figure S4(a)). Collectively, these changes imply localized folding transitions within PAH1 upon SAP25 SID binding. Indeed, a best-fit superposition of backbone atoms for the segment extending from Val123 to Pro187 of the apo- and SAP25 SID-bound mSin3A PAH1 yields a somewhat high RMSD value of 1.60 Å that reflects these conformational changes. The extension of the first helix by approximately two helical turns was also inferred previously from backbone secondary chemical shifts for the mSin3A PAH2 domain upon complex formation with Mad1/Mxd1 SID.⁴⁰

The SAP25 SID in the mSin3A PAH1 complex is largely unstructured, except for the segment spanning Ala133-Met144 that adopts a helical conformation (designated the αA helix; Figure 2(b) and 2(d)). The ¹³C^α secondary chemical shifts, which are exquisitely sensitive to polypeptide backbone conformation,⁴² suggest that the helix is poorly formed in the apo-form with only seven residues at the N-terminus sampling significant amounts of helical conformations (Supplementary Figure S4(b)). Upon mSin3A PAH1 binding, the helix is stabilized and extended at the C-terminus. These results when considered in light of the folding and conformational transitions occurring in the mSin3A PAH1 domain upon binding indicate that the interacting proteins undergo mutual folding transitions. The N-terminal portion of the PAH1 α1 helix may play a key role in functioning as a gate to lock cognate ligands. Interestingly, the backbone ¹H-¹⁵N correlations belonging to residues in this region are significantly broadened (Supplementary Figure S1(b)), hinting at a dynamic process, possibly related to the folding transition, on the microsecond-millisecond timescale. Indeed, severe resonance broadening was also noted for equivalent residues in the mSin3A PAH2 domain when bound to HBP1 SID.³¹

The protein-protein interface in the SAP25 SID-mSin3A PAH1 complex extends over an area averaging 644(±41) Ų in the NMR ensemble. As in the case of previously described SID-PAH complexes,^28-31,33 the interaction involves a helix in the SID on the one hand with a hydrophobic cleft in the PAH domain on the other (Figure 3(a)). The cleft is cavernous and is dominated by hydrophobic residues and a few lysine residues at the periphery (Figure 3(b) and 3(c)). The complex appears to be stabilized by numerous intermolecular interactions that are almost exclusively hydrophobic (Figure 3(c)). In particular, the side chains of Leu138, Leu141, and Leu142 within the ‘LXXLL’ motif and that of Met144 in SAP25 SID make extensive contacts with the cleft. Consistent with these observations, the Leu138Ala mutant and the Leu141Ala,Leu142Ala double mutant exhibited diminished and complete loss of binding, respectively, to mSin3A PAH1.³⁹ In addition to the aforementioned affinity determining interactions made by SAP25 residues, Ala137 and Gly140 appear to play a role in stabilizing the complex, by eliminating or diminishing unfavorable interactions involving their side chains, and may thus serve as key specificity determinants of the interaction. Finally, in half of the structures of the ensemble, a sole intermolecular hydrogen bonding interaction is detected between the hydroxyl and carboxylate groups of mSin3A PAH1 Tyr144 and SAP30 SID Glu134, respectively.

Figure 3.

The SAP25 SID binds to a deep hydrophobic cleft in the mSin3A PAH1 domain. Molecular surface views of the cleft color-coded according to (a) residue type (hydrophobic: yellow; polar: cyan) and (b) curvature (concave: gray; planar: white; convex: green). The backbone of the SAP25 SID helix and the PAH1-interacting side chains are shown in worm and stick representations, respectively. (c) A catalogue of intermolecular interactions in the SAP25 SID-mSin3A PAH1 complex detected in ≥60% of conformers in the NMR ensemble.⁶⁴ SAP25 and mSin3A residues are presented in pale magenta and pale green backgrounds, respectively. The lines connect interacting residues. Line colors indicate the type of interaction (green: electrostatic; red: hydrogen bonding; purple: salt bridge; gray: hydrophobic) whereas font colors indicate the type of residue (green: hydrophobic; blue: polar; magenta: charged).

Elucidation of PAH1-interaction motifs

A comparison of the structure of the SAP25 SID-mSin3A PAH1 complex with that of the NRSF SID-mSin3B PAH1 complex³³ yields additional insights into mSin3 PAH1 recognition. In spite of the dissimilar nature of the SIDs at the sequence level, the polypeptide backbones of the PAH domains (corresponding to the Val123-Pro187 and Val34-Pro98 segments of mSin3A and mSin3B, respectively) in these complexes superimpose rather well with an RMSD of 1.10 Å, indicative of closely related conformations (Figure 4(a) and 4(b)). The SIDs engage the cleft through helical structures, but surprisingly, the relative orientations of the helices in the two complexes are almost exactly reversed. Furthermore, the helices engage a highly overlapping surface of the PAH1 domains that is essentially invariant in the two complexes (Figure 4(c) and 4(d)). A close examination of the non-covalent interactions in the SAP25 complex leads us to define a core PAH1-interaction motif of the form s-Φ-x-s-Φ-Φ-x-Φ (where s: residue with short side chain; Φ: bulky hydrophobic; x: any non-proline), which is exactly opposite in chain direction to that deduced for the NRSF complex, and correlates well with the opposite helical orientations. Key equivalent residues in the two PAH1-interaction motifs target equivalent binding ‘pockets’ on the surface of the PAH1 domain (Figure 4(c) and 4(d)). These observations are similar to those reported previously for the mSin3A PAH2 domain,³¹ implying that these properties are shared by the Sin3 PAH1 and PAH2 domains. Whereas the NRSF SID helix binds PAH1 with an orientation akin to that used by the Mad1/Mxd1 SID helix to bind PAH2,^28,29,33 the SAP25 SID binds PAH1 via a helical orientation similar to that used by the HBP1 SID helix to PAH2.³¹ We propose the terms type I and type II PAH binders to distinguish SIDs belonging to the former and latter categories, respectively. Although PAH binders of each type share overt similarity vis-à-vis helical orientations, the sequence motifs employed in targets recognition are quite distinct (e.g. Φ-x-Φ-Φ-s-x-Φ-s versus Φ-x-x-Φ-Φ-x-A-A used by type I PAH1- and PAH2-interactors, respectively).^28,29,33

Figure 4.

A comparative analysis of intermolecular interactions involving the Sin3 PAH1 domain with diverse targets. A side-by-side comparison of the (a) SAP25 SID-mSin3A PAH1 complex (colored in magenta and green, respectively) and (b) the NRSF SID-mSin3B PAH1 complex (colored in blue and green; PDB code: 2CZY³³). Note the contrasting chain directions of the two SID helices. The respective intermolecular interfaces in the two complexes with the backbones rendered as worms and the side chains as sticks (panels (c) and (d)). Residues in the respective SIDs deemed important for the interaction are annotated in bold and italicized. Sequence motifs deduced from structural analysis of the respective complexes (panels (e) and (f); abbreviations: s = any amino acid with a short side chain; x = any non-proline residue; Φ = any bulky hydrophobic residue). The predicted binding sites and modes of a subset of proteins previously shown to target the Sin3 PAH1 domain are also shown. With the exception of yeast Opi1, the proteins are of human or mouse origin (GenBank accession codes: SAP25: NP_001075431; NCoR: AAB17125; Myt1: AAC53456; PLZF: AAD03619; NRSF: NP_005603; Pf1: NP_001028733; Opi1: AAB65073)

The definition of two distinct PAH1-interaction motifs permits predictions to be made regarding the precise location of the SIDs in various PAH1 interacting proteins, as for most of these proteins the SIDs have been mapped only coarsely (we have refrained from using these motifs to discover novel Sin3 interactors as the information content is far too low to allow reliable predictions). In arriving at these predictions, we applied the following criteria: (i) the sequence must have the potential to form a helix and (ii) the motif-defining residues must be solvent accessible and must not engage in any domain-stabilizing tertiary interactions. Criterion i can be assessed by secondary structure prediction, whereas criterion ii requires a known three-dimensional structure. Given the limitations in the accuracy of protein secondary structure prediction and our insufficient knowledge of protein three-dimensional structures, the conclusions from such an analysis must be treated with caution. These concerns notwithstanding, some interesting insights emerge from our analysis. In the case of PLZF, two putative SIDs were previously identified experimentally, each within residues 1-120 and 121-364.³² Our analysis revealed multiple PAH1-interacting motifs in the former segment, but as the region corresponds to the BTB-POZ domain for which a structure is known and since most of the motif-defining residues were not solvent accessible, these were excluded from further consideration. However, PLZF residues 202−209 was found to harbor a type II motif and meets the aforementioned criteria, hinting that it might be a bona fide SID (Figure 4(e)). Similar considerations led us to identify type II motifs for Myt1 (residues 582-589) and NCoR (residues 1836-1843). The PAH1-interacting regions in these proteins were mapped previously using genetic and biochemical approaches to aa 502−623 and 1829−1940, respectively.^36,37 We note in passing that the physiological relevance of the interaction between NCoR and Sin3 is unclear as NCoR has been consistently detected in multi-protein complexes containing SMRT and HDAC3.⁶ The Pf1 corepressor was previously shown to harbor multiple SIDs.³⁴ Pf1 targets both the PAH1 and PAH2 domains of mSin3 via separate SIDs and whereas the PAH2-interacting SID was mapped precisely to the segment spanning aa 209-219,^28,34 the PAH1-interacting SID was found to reside within residues 300-743. ³⁴ Our analysis reveals a potential PAH1-interaction motif at aa 352-359 (Figure 4(f)). A similar analysis of the Opi1 SID, previously mapped to residues 45-106,³⁵ reveals two non-overlapping type I PAH1-interaction motifs spanning aa 53-60 and 68-75. We note that the inclusion of the yeast Opi1 protein in this analysis is justified as the S. cerevisiae Sin3 PAH1 domain shares 73% identity and 85% similarity at the sequence level with the mSin3A PAH1 domain; furthermore, residues comprising the SID binding surface are invariant in these domains (Figure 1).

Specificity of mSin3 PAH1 and PAH2 interactions

The close sequence and structural similarity of the mSin3 PAH1 and PAH2 domains (30% sequence identity, 54% sequence similarity) raises questions relating to the specificity of PAH-SID interactions. These questions assume great significance, as out of approximately twenty PAH residues forming the protein-protein interface, half of these are identical in the PAH1 and PAH2 domains, whereas a quarter of them involve conservative substitutions (Figure 1 and Figure 5(a)). However, the non-conservative changes map to the periphery of these protein-protein interfaces suggesting an important role for residues undergoing conservative substitutions in dictating the specificity of these interactions (Figure 5(a)). To test this idea, a panel of mSin3A PAH2/PAH1 chimeras was generated in which one or more PAH1 residues (up to a maximum of six) in the hydrophobic cleft were introduced in a PAH2 background and the affinities for a specific SID was measured. The residues selected included Val311 (corresponding to Leu130 in PAH1; Figure 5(a)), Leu332 (Met151 in PAH1), Phe304 (Val123), Ile308 (Leu 127), Asn312 (Asp131), and Tyr335 (Phe154). Using a fluoresceinated Mad1/Mxd1 SID peptide and employing steady-state fluorescence anisotropy assays, we evaluated the ability of the peptide to bind to single-, double- and multiple-site PAH2 mutants that we refer to as PAH2/PAH1 chimeras. As expected, the wild-type PAH2 protein bound to Mad1/Mxd1 SID with high affinity (Supplementary Figure S3(b); Table 2).^28,43 With the exception of the Val311Leu mutant, which exhibited three-fold higher affinity for Mad1/Mxd1 SID relative to wild-type, the single- and double-site mutants exhibited between two- to seven-fold reduction in binding affinities (Table 2). The Leu332→Met mutation had the most profound effect on Mad1/Mxd1 SID binding, indicating that this residue is an important affinity determinant for Mad1/Mxd1 binding. As more mutations were introduced, the binding affinity was correspondingly diminished with quadruple-site mutants exhibiting between ten- and over twenty-fold reductions in binding while the sextuple-site mutant was the most strongly affected showing a hundred-fold reduction in Mad1/Mxd1 SID binding affinity (Table 2).

Figure 5.

Specificity of protein-protein interactions involving the mSin3A PAH1 and PAH2 domains. (a) The mSin3A PAH2 (top) and PAH1 (bottom) residues involved in binding the SIDs of Mad1/Mxd1 and SAP25, respectively, are depicted. The residues are colored according to the sequence conservation patterns across the Sin3 PAH1 and PAH2 domains (invariant residues: purple; conservative substitutions: gold; non-conservative replacements: magenta). Note the essentially unperturbed side chain conformations of invariant residues. mSin3A PAH2 residues that were mutated to the equivalent PAH1 residue to yield various PAH2/PAH1 chimeras are annotated. (b) Pulldown assays conducted using His₆-tagged Mad1/Mxd1 SID, HBP1 SID and SAP25 SID immobilized on a Ni²⁺ resin with GST or GST-tagged wild-type mSin3A PAH1 and PAH2 proteins. Ten percent of the input proteins used in these assays are shown after resolving them by SDS-PAGE and staining the gels with Coomassie Blue (left panel). Proteins bound to the resin following the pulldown experiments were resolved using SDS-PAGE and visualized by Coomassie Blue staining (right panel). (c) Analogous pulldown assays conducted using the same SID polypeptides with various GST-PAH2/PAH1 chimera proteins. The amounts of the chimera proteins used in these assays are shown in the top panel.

Table 2.

Fluorescence Anisotropy Analysis of Mad1/Mxd1 SID_6-21 Binding by GST-PAH2/PAH1 Chimeras

Protein	Kd (nM)	A0	A∞
Wild-type	51 ± 4	0.034	0.210
V311L	16 ± 2	0.033	0.187
L332M	369 ± 82	0.053	0.199
V311L, L332M (Chimera 1)	104 ± 13	0.040	0.183
F304V	163 ± 22	0.038	0.171
F304V, I308L (Chimera 2)	187 ± 19	0.034	0.168
Chimera 1+2	1223 ± 72	0.037	0.163
N312D, Y335F (Chimera 3)+Chimera 1	510 ± 87	0.035	0.190
Chimera 2+3	947 ± 146	0.038	0.159
Chimera 1+2+3	5524 ± 810	0.037	0.167

If any or all six residues played a role in determining the specificity of PAH-SID interactions, not only should the PAH2 domain exhibit diminished binding as it starts to increasingly resemble PAH1, but it should also acquire the ability to bind PAH1-specific targets. To test this idea, we conducted binding assays in pulldown format with the wild-type proteins as well as the PAH2/PAH1 chimeras for three different SIDs, including Mad1/Mxd1 SID, HBP1 SID, and SAP25 SID. As expected, the Mad1/Mxd1and HBP1 SIDs bound efficiently to wild-type PAH2, but not to wild-type PAH1 (Figure 5(b)). Similarly, SAP25 SID bound well to wild-type PAH1, but not to wild-type PAH2 (the affinity of the interaction was difficult to measure reliably but from competition assays conducted using fluorescence anisotropy we estimate the K_d >> 2.5 μM; ITC measurements placed the affinity in the >100 μM range). In addition, our aforementioned findings with regard to the binding affinities of various PAH2/PAH1 chimeras and Mad1/Mxd1SID matched qualitatively with those obtained in the pulldown assays (Figure 5(c)). Unlike Mad1/Mxd1SID, HBP1 SID showed greater sensitivity towards PAH1 substitutions in the PAH2 domain. For example, the quadruple-site mutants exhibited strongly diminished binding for HBP1 than for Mad1/Mxd1(Figure 5(c)). Importantly, both Mad1/Mxd1and HBP1 exhibited severe reduction in binding for the sextuple-site mutant. On the other hand, the SAP25 SID bound to almost all PAH2/PAH1 chimeras tested with varying efficiencies. The Val311Leu mutation appears to be sufficient for SAP25 to bind efficiently. Indeed, all multiple-site mutants harboring this mutation also bound SAP25 SID efficiently (Figure 5(c)). Similarly, the Phe304Val mutation also bound SAP25 SID but comparatively with slightly lower efficiency than the Val311Leu mutant. We note that the phenylalanine residue was previously implicated as an important specificity determinant for Mad1/Mxd1interactions by another study.⁴⁴ Intriguingly, the quadruple-site mutant lacking the Val311Leu and Leu332Met mutations did not exhibit significant binding to SAP25 SID, implying that the contributions made by individual residues towards binding affinity are not necessarily additive or even approximately additive (Figure 5(c)). More detailed structural and functional studies are required to clarify the basis for these findings.

Comparisons with other transcription factor-coregulator complexes

Short amphipathic helices harboring Φ-x-x-Φ-Φ sequence motifs (where Φ and x denote bulky hydrophobic and non-proline residues, respectively) is a recurring theme in transcription factor-coregulator interactions.⁴⁵ The ubiquity of this feature in transcription factors on the one hand, and the seeming promiscuity of coregulators as reflected in their ability to interact with disparate transcription factors on the other hand, raises questions regarding how such interactions are facilitated at the molecular level while retaining some degree of specificity. The Sin3 PAH1 and PAH2 domains seem to have solved this problem by accommodating ‘reversible’ sequence and helical motifs within a common binding site. It is unclear whether this property is shared with the presently poorly characterized PAH3 domain, which is more distantly related to the PAH1 and PAH2 domains (Figure 1). Interestingly, this property is also exhibited by ubiquitin as evidenced by its interactions with the helical UIM and MIU motifs found in endocytic proteins.^46,47 The ligand-binding domains of nuclear receptors engage distinct sequence motifs of coactivators and corepressors through overlapping but somewhat altered binding surfaces.^48,49 The KIX domain of the CBP transcriptional coactivator interacts with distinct sequence motifs of the CREB and c-Myb transcription factors through an overlapping surface.^50,51 However, the domain despite its relatively small size exhibits considerable versatility in engaging the MLL transcription factor through a completely different surface.⁵²

Conclusions

In conclusion, our studies show that the mSin3A PAH1 domain shares many structural and functional features with the well-characterized mSin3A and mSin3B PAH2 domains including the formation of a four-helix bundle structure, interaction with helical targets through a comparable and an almost exclusively hydrophobic surface, and recognition of chain-reversible sequence motifs. The sequence motifs recognized by the Sin3 PAH1 domains differ significantly from those described for the Sin3 PAH2 domains. Both PAH1 and its target become more ordered and helical in the complex with the extended helix in PAH1 serving as a gate or barrier to prevent target escape. Finally, the PAH1 and PAH2 domains exhibit a high degree of specificity in their interactions with cognate targets and conservative replacements of PAH2 residues with equivalent PAH1 residues are sufficient to dramatically alter interaction affinity and specificity.

Materials and Methods

Production of the mSin3A PAH1 and SAP25 SID domains

The coding sequences for the mSin3A PAH1 (residues 119-189) and murine SAP25 SID (residues 126-186) domains were amplified by PCR and cloned into the pET-30 Xa/LIC expression vector (Novagen). Escherichia coli BL21(DE3) cells harboring these vectors were grown at 37 °C in M9 minimal medium and shifted to 20 °C 15 minutes before induction. Protein expression was induced using 1 mM isopropyl-1-thio-β-D-galactopyranoside when A_600nm was ∼0.7 and cells were harvested 7-8 hours thereafter. Cell pellets were suspended in 20 mM sodium phosphate buffer, pH 7.2, containing 0.5 M NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 μM leupeptin, 1 mM pepstatin and 0.1% Triton X-100, lysed via sonication, centrifuged and the supernatant loaded onto a HiTrap Chelating column (GE Healthcare) charged with Zn²⁺. The His₆-tagged protein was eluted using 20 mM sodium phosphate buffer, pH 7.2, containing 0.5 M NaCl and 50 mM EDTA. Protein-containing fractions were assayed by SDS-PAGE and these were further purified via reversed-phase high-pressure liquid chromatography (HPLC) using a C18 column (Vydac) and a mobile aqueous phase containing 80% acetonitrile and 0.1% trifluoroacetic acid. Fractions containing the desired polypeptide were pooled, lyophilized, dissolved in aqueous buffer, and treated with Factor Xa (Novagen) following the vendor’s recommendations. The resulting polypeptide, sans the His₆-tag, was purified using reversed-phase HPLC. The identity of the purified protein was confirmed by electrospray ionization-mass spectrometry (ESI-MS). Polypeptides uniformly labeled with ¹⁵N and/or ¹³C isotopes were produced by following the same procedure as above, except that the cells were grown in M9 minimal media containing ¹⁵N-(NH₄)₂SO₄ (Martek) and/or ¹³C₆-D-glucose (Isotec). The integrity of the purified proteins and the extent of isotope enrichment (typically >98% for ¹⁵N and >96% ¹³C) were confirmed by ESI-MS.

NMR sample generation

Lyophilized protein samples of mSin3A PAH1 and SAP25 SID were dissolved in 20 mM sodium phosphate buffer, pH 6.0, containing 2 mM DTT-d₁₁ and 0.2% NaN₃ in 90% H₂O/10% D₂O. Approximately equimolar complexes of SAP25 SID and mSin3A PAH1 were generated for NMR studies by titrating increasing amounts of unlabeled SAP25 SID with ¹⁵N/¹⁵N,¹³C-mSin3A PAH1 and separately, unlabeled mSin3A PAH1 with ¹⁵N/¹⁵N,¹³C-labeled SAP25 SID. The progress of the titration was monitored by recording two-dimensional (2D) ¹H-¹⁵N correlated spectra and the titration was terminated when no changes in the NMR spectrum were detected. NMR sample concentrations were in the 0.7-1 mM range. All protein concentrations were measured spectrophotometrically.⁵³ For experiments conducted in D₂O, the samples of the respective proteins or protein complexes were lyophilized and redissolved in 99.996% D₂O (Isotec).

NMR spectroscopy and structure determination

All NMR experiments were performed on a Varian Inova 600 MHz spectrometer equipped with a room temperature pulsed-field-gradient triple-resonance probe. NMR data were acquired either at 15 °C (free mSin3A PAH1 domain) or at 30 °C (free SAP25 SID domain and SAP25 SID-mSin3A PAH1 complex). Data processing and spectral analyses were conducted using Felix 98.0 software (Accelrys) modified in-house to facilitate accelerated resonance and NOE assignments.⁵⁴ Backbone resonance assignments were made by analyzing three-dimensional (3D) HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH, HNCO, and HCACO spectra whereas side chain resonances were assigned using 3D ¹⁵N-edited TOCSY, HCCH-COSY, HCCH-TOCSY, C(CO)NH-TOCSY spectra.^55,56 Aromatic resonances were assigned by analyzing 2D (HB)CB(CDCGCE)HE and 3D (HB)CBCG(CD)HD and (HC)C(C)CH-TOCSY spectra.^57,58

For the determining the structures of the free mSin3A PAH1 domain and the SAP25 SID-mSin3A PAH1 complex, the H^α, C^α, C^β, C’ and backbone ¹⁵N chemical shifts served as the source for backbone ϕ and ψ torsion angle restraints using TALOS.⁵⁹ These restraints were imposed only for those residues with reliability scores of 10. Side chain χ¹ restraints for isoleucine, threonine, and valine residues were derived from measured ³J_NC^γ and ³J_C’C^γ values. Inter-proton distance restraints were derived from 3D ¹⁵N-edited NOESY recorded in H₂O (τ_m = 100 ms), 3D aliphatic ¹³C-edited NOESY (τ_m = 80 ms), aromatic ¹³C-edited NOESY (τ_m = 60 ms), and 3D ¹³C-filtered, ¹³C-edited NOESY (τ_m = 120 ms) recorded in D₂O. Intermolecular NOEs were assigned manually and were calibrated indirectly by calculating a scaling factor for the intensities of well-resolved peaks in the ¹³C-edited NOESY and ¹³C-filtered, ¹³C-edited NOESY spectra. These NOEs were assigned upper bounds of 3, 4, 5, and 6 Å. All other NOEs were calibrated and assigned iteratively and automatically by ARIA (version 1.2)^60,61 and were checked manually between successive cycles of refinement. Structures were calculated from extended conformations as starting models using ARIA in conjunction with CNS.⁶² Altogether 80 structures were computed in the final iteration and the 20 structures with the lowest restraint energies, restraint violations, and RMS deviations from ideal covalent geometry were considered for structural analysis. The final structures were analyzed using PROCHECK,⁶³ MONSTER,⁶⁴ DeepView,⁶⁵ and CNS.⁶² Molecular images were generated using RIBBONS⁶⁶ and GRASP.⁶⁷

Production of mSin3A PAH2-PAH1 chimeras

Single- and multiple-site mutants of GST-mSin3A PAH2 were generated using the QuikChange site-directed mutagenesis kit (Stratagene). All mutations were confirmed by DNA sequencing. E. coli harboring the wild-type and mutant GST-PAH2 vectors were grown in Luria-Bertani medium at 37 °C. The proteins were expressed for approximately 16 h at 20 °C and prepared for purification using analogous approaches used for producing mSin3A PAH1 and SAP25 SID described above. The proteins were purified using glutathione-sepharose beads (GE Healthcare) following the manufacturer’s recommendations. To preclude adventitious protease-mediated degradation, the proteins were stored at 4 °C in 20 mM sodium phosphate buffer (pH 8.0) containing 25 mM EDTA.

Binding assays

Isothermal titration calorimetry experiments were performed on a MicroCal MCS calorimeter. Experiments were performed at 30 °C in 20 mM sodium phosphate buffer (pH 6.0). Binding isotherms were analyzed using Origin 5.0 software provided by the manufacturer.

GST-PAH binding assays were performed by immobilizing His₆-SAP25 SID, His₆-Mad1/Mxd1SID (corresponding to residues 1-35 of human Mad1) and His₆-HBP1 SID (corresponding to residues 342-398 of human HBP1) domain constructs to a Co²⁺ resin (Clonetech). The His-tagged proteins were incubated with 1 μM solutions of the various GST-tagged proteins in a binding buffer comprising 50 mM sodium phosphate (pH 7), 5 mM imidazole, and 0.1% Triton for a final reaction volume of 1 ml. Reactions were incubated for 30 min at room temperature. After washing the beads twice with 1 ml of binding buffer, the bound proteins were eluted by boiling in the SDS-PAGE sample-loading buffer containing 250 mM imidazole, resolved by gel electrophoresis, and analyzed by Coomassie Blue staining.

Equilibrium binding assays in free solution were performed using fluorescence spectroscopy. A 16-residue peptide corresponding to residues 6-21 of human Mad1/Mxd1and incorporating a fluorescein moiety at the N-terminus was synthesized using automated procedures (University of Utah DNA/Peptide Facility) and purified by reversed-phase HPLC. The peptide was dissolved in fluorescence buffer comprising 20 mM sodium phosphate buffer (pH 8) and 0.5 mM EDTA. Stock solutions of the various GST-PAH proteins were exchanged into the fluorescence buffer and were diluted to appropriate concentrations, ranging from 0.5 nM to 35 μM. The final volume of each sample was 1 ml while the concentration of the fluoresceinated peptide was maintained at 2 nM.

Steady-state fluorescence experiments were performed at room temperature on an ISS PC-1 fluorimeter configured in the L-format. The excitation wavelength was set to 492 nm while an excitation interference filter (Coherent/Ealing #35-3482; wavelength center: 490 nm, bandpass: 7.3 nm) and an emission cut-off filter (Omega optical #535AF45 multilayer) were employed to attenuate contributions from the excitation bandpass and scattering effects, respectively.⁶⁸ Since we employed these optical filters, we chose not to use the emission monochromator in order to increase the sensitivity of the instrument. Data collection and processing were performed using the program Vinci (ISS). Ten measurements were recorded for each sample, each measurement being the average of 10 readings of 100 μs-long intervals. The reported anisotropy values are the average and standard deviations from three independent samples. The average anisotropy was graphed as a function of the GST-PAH2 protein concentration and the resulting curves were fit to a dose-response function using the program Origin 5.0 (Microcal). In the fitting procedure, the EC₅₀ value and the limiting anisotropies at zero and infinite protein concentration were allowed to vary while the Hill’s coefficient was fixed at 1. The EC₅₀ values are reported as the equilibrium dissociation constants (K_d) in Table 2 and the associated errors are the fitted errors.

Abbreviations

SID

Sin3 interaction domain

PAH

paired amphipathic helix

HDAC

histone deacetylase

SAP

Sin3 associated polypeptide