Role of Structural and Dynamical Plasticity in Sin3: The...

doi:10.1016/j.jmb.2006.01.100 J. Mol. Biol. (2006) 358, 485–497

Role of Structural and Dynamical Plasticity in Sin3:The Free PAH2 Domain is a Folded Module in mSin3B

Hugo van Ingen, Maria A. H. Baltussen, Jan Aelen andGeerten W. Vuister*

Department of PhysicalChemistry/BiophysicalChemistry, Institute forMolecules and MaterialsRadboud University NijmegenToernooiveld 1, 6525 EDNijmegen, the Netherlands

0022-2836/$ - see front matter q 2006 E

Present address: M.A.H. BaltusseFood Safety, Wageningen, The NethAbbreviations used: CPMG, Carr

Gill; HMG, high mobility group; Nhauser effect; O-GlcNAc, O-linked NRMSD, root-mean-square deviationBank; NOE, nuclear Overhauser effspectroscopy; HSQC, heteronuclearcoherence.E-mail address of the correspond

[email protected]

The co-repressor Sin3 is the essential scaffold protein of the Sin3/HDAC co-repressor complex, which is recruited to the DNA by a diverse group oftranscriptional repressors, targeting genes involved in the regulation of thecell cycle, proliferation and differentiation. Sin3 contains four repeatscommonly denoted as paired amphipathic helix (PAH1-4) domains thatprovide the principal interaction surface for various repressors. Here, wepresent the first structure of the free state of the PAH2 domain and discussits implications for interaction with the repressors. The unboundconformation is very similar to the conformation observed when boundto either the Mad1 or HBP1 repressor, suggesting that the PAH2 domainserves as a template that guides proper folding of the unstructuredrepressor. The free PAH2 domain shows micro- to millisecond confor-mational exchange between the folded, major state and a partiallyunfolded, minor state. Upon complex formation, we observe a significantdecrease in fast time-scale flexibility of local regions of the protein,correlated with the formation of intermolecular contacts, and an overalldecrease in the slow time-scale conformational exchange. On the basis ofour data and using a multiple sequence alignment of all PAH domains, wesuggest that the PAH1, PAH2 and PAH3 domains form pre-folded bindingmodules in full-length Sin3 like beads-on-a-string, and act as foldingtemplates for the interaction domains of their targets.

q 2006 Elsevier Ltd. All rights reserved.

Keywords: protein–protein interactions; PAH domain; transcription; Sin3;NMR
*Corresponding author
Introduction

The regulation of gene expression in eukaryotesrequires the formation of specific protein–DNA andprotein–protein complexes.1 Transcriptional activa-tors and repressors recognize specific DNAsequences upstream of the promoter of targetgenes, resulting in either up-regulation or down-regulation of transcription.2,3 This is achieved

lsevier Ltd. All rights reserve

n, RIKILT, Institute oferlands.–Purcell–Meiboom–OE, nuclear Over--acetylglucosamine;

; PDB, Protein Dataect; NOESY, NOEsingle quantum

ing author:

indirectly via recruitment of co-regulators andassociated proteins that together are capable ofshaping the chromatin structure into an active orrepressive state.4–7 The association between theactivator/repressor and the co-regulator is a crucialstep, as it provides the link between target gene andchromatin modification.A diverse group of repressors recruit a co-

regulator complex containing the highly conservedco-repressor Sin3,8 and the class I histone deacety-lating proteins (HDAC1/2). The HDAC proteinsdeacetylate specific lysine residues in the tails of thehistone proteins, resulting in a compact, repressivechromatin state.9–13 The core Sin3/HDAC complexfurther consists of Sin3 associated proteins (SAP18and SAP30), that are thought to stabilize thecomplex,11,14 and histone-targeting proteins(RbAp48 and RbAp46), that stabilize contact withthe nucleosome.9,11 Additionally, the core Sin3/HDAC co-repressor complex exists of SDS3,15

SAP130 and SAP18016 and can be extended with

d.

486 The Free PAH2 Domain of Sin3

extra catalytic modules, such as the monosaccha-ride O-GlcNAc transferase OGT.17

Sin3 contains four imperfect repeats, denoted asPAH domains that were originally presumed to foldinto paired amphipathic helices,18 and a highlyconserved region that is responsible for interactionwith HDAC1/2, referred to as the HDAC inter-action domain (HID) (Figure 1(a)). The second PAHdomain, PAH2 and, to a lesser extent, the PAH1 andPAH3 domains have been identified as the regionsrequired for interaction with the repressor,8

whereas the PAH4 domain is required for associ-ation with the monosaccharide transferase OGT.17

The mammalian homolog of Sin3 exists in twoisoforms, mSin3A and mSin3B,19 that most likelyhave separable functional roles.20,21

Structurally, Sin3 has remained largely unchar-acterized. So far, the interaction between Sin3 andthe transcriptional repressor Mad1 has receivedconsiderable attention, as Mad1 is part of the Myc/Max/Mad network involved in the switch betweencell proliferation and cell differentiation.22,23 Sol-ution structures of the PAH2 domain complexed tothe Sin3 interacting domain (SID) of Mad1 haveshown that the complex is folded as a “wedgedhelical bundle”, in which the PAH2 domain adoptsa four helix bundle conformation into which thea-helix of the SID is inserted.24–26 Recently, the

Figure 1. (a) Domain organization of Sin3. Protein domainhelix domain; HID, histone deacetylase interacting domaisequence alignment of the PAH domains of mousemSin3A anin the two experimentally studied sequences are indicated byor conservatively mutated in at least the PAH1-3 domains. ThResidues that engage in intermolecular interaction with the Sin(C) are indicated in the bottom row. (c)–(e) The 15N-HSQC Nconstruct (residues 148–232), (d) a 105 residue construct, exteresidue construct in complex with the SID of Mad1.

HMG box-containing repressor HBP1, which tar-gets several cell cycle-specific and differentiation-specific genes, was shown to interact with the PAH2domain.27 Interestingly, the structure of thiscomplex showed a reversed orientation of the SIDrelative to the Mad1 complex, while maintainingthe overall fold. The reversal of this helix orien-tation is correlated with a reversal in the SIDsequence motif.27

In contrast to the well-studied bound state of thePAH2 domain, the unbound state of the domain hasnot been described. However, knowledge of bothbound and unbound states is of paramountimportance to allow an evaluation of the structuraland dynamic changes induced upon complexformation, and will result in a better understandingof the molecular mechanism of repressor–corepres-sor interaction. Here, we report the solutionstructure and dynamics on fast and slow time-scales of the unbound PAH2 domain of mammalianSin3B and discuss its implications for the PAH–SIDinteraction. While previous studies have suggestedthat the PAH2 domain might not be fully folded inits unbound form, and that PAH2 and the SIDexperience a coupled folding transition, we showthat the domain is folded and that the structuraland dynamic plasticity in PAH2 upon complexformation is limited to a fine-tuning of the

s are indicated with filled boxes (PAH, poly amphipathicn; HCR, highly conserved region). (b) Structure-baseddmSin3B. Secondary structure is shown at the top. Helicesblack outlined boxes. Residues in bold type are conservede sequence used in this study is indicated with an asterisk.3 interacting domain (SID) of Mad1 (m), HBP1 (h) or bothMR spectra of the PAH2B domain using (c) an 85 residuended at the C terminus (residues148–252) and (e) the 105

The Free PAH2 Domain of Sin3 487

interaction surface. In addition, we show that thePAH1-3 domains most likely form folded moduleswithin Sin3, as beads-on-string, and are ready toaccommodate their specific interaction partners.

Results

Optimization of the PAH2 domain construct

Our previous studies of the PAH2 domain ofmSin3B in complex with SID of Mad1 used a PAH2domain construct of 105 residues and showed thatthe folded part of the domain comprises about 75residues.25,26 Therefore, for the study of the freePAH2 domain, a construct was made consisting of85 amino acid residues (148–232 of Mm. mSin3B,denoted as PAH2B-85), comprising the domain withfive extra residues at both the N and the C terminus.To assess the conformational integrity of thisdomain, a 2D 1H–15N correlation spectrum (15N-heteronuclear single quantum coherence (HSQC))was recorded (shown in Figure 1(c)). The lack ofdispersion of backbone amide resonances and thesevere line broadening of several resonancesindicate that this polypeptide does not adopt awell-defined tertiary structure in solution. Theobserved conformational heterogeneity persistsover a wide range of temperatures and underdifferent buffer conditions.

In sharp contrast, the HSQC spectrum obtainedfrom a PAH2 construct extended with an additional20 residues at the C terminus (148–252 of Mm.mSin3B, denoted as PAH2B-105) shows a greatlyimproved spectrum (Figure 1(d)). This improve-ment is probably caused by an enhanced solubilityof the longer construct in combination with reducedaggregation (see Discussion). The increased dis-persion and the more homogeneous linewidthssuggest that the PAH2B domain is folded. In orderto assess the potential interference between thePAH2 domain and its C-terminal tail, we analyzedboth the structural and dynamic properties. Both13Ca and 13Cb chemical shifts of all residues arewithin one standard deviation from random coilvalues. Only intra-residual and sequential nuclearOverhauser effect (NOE) cross-peaks could beidentified in the NOE spectroscopy (NOESY)spectra (data not shown). Furthermore, fast time-scale dynamics indicates that the tail is extremelyflexible with local motions on a time-scale of aboutone-tenth of the global tumbling rate (vide infra).Relaxation-dispersion measurements also did notgive any indication for the existence of a minor,possibly interfering, conformation for this region(vide infra).

In addition, the tail does not impede binding ofthe SID. The spectral dispersion in the HSQCspectrum of PAH2B-105 complexed to the Mad1-SID is further improved (Figure 1(e)). The overallresonance pattern does not change drasticallybetween the unbound and bound forms ofPAH2B-105, which suggests that the fold of PAH2

is essentially preserved between unbound andpeptide-bound states. Hence, we conclude that theC-terminal tail is unstructured in our construct anddoes not interact with the binding pocket of PAH2.Given its superior stability and spectral quality, thisPAH2B-105 construct was used for further struc-tural and dynamic studies.

Structure

The backbone traces of the unbound PAH2B-105ensemble and the best representative structure areshown in Figure 2(a) and (b), respectively. In total,1007 unique distance restraints and 131 f,jdihedral angle restraints were used to calculatethe solution structure. Residues 233–252, whichconstitute the unstructured C-terminal tail, werenot included in the structure calculation. Thepairwise heavy-atom RMSD of the ordered partsof the structure is 0.68 A for the backbone and1.46 A overall, indicating a well-defined ensemble(Table 1). Analysis of the per-residue qualityparameter U shows that the structural quality ofthe ensemble of structures is well determined by theexperimental data (see Supplementary Data).27

Further statistical data for the ensemble arereported in Table 1. The domain adopts a four-helix bundle fold that was found also for thestructure of the complex of PAH2 and Mad1-SID.24–26 The four helices are formed by residues153–167 (a1), 172–189 (a2), 202–212 (a3) and 217–226 (a4). Residues 168–171 and 213–216 fold intoturns, and residues 190–201 form a large loopbetween helices a2 and a3.The most striking feature of the PAH2 structure is

its large solvent-exposed hydrophobic bindingpocket of ca 510 A2 (averaged over the ensemble)(Figure 2(c)). This exposed conformation is causedby poor packing of helix a1 to the remainder ofthe helical bundle. Whereas the helices a2–a3 anda3–a4 arrange as nearly anti-parallel, the anglebetween a1 and a2 is w1458. As a result, tightpacking of helices a1 and a2 is restricted to twohelical turns adjacent to the connecting turn. Incontrast, helices a3 and a4 pack closely along theirfull lengths (Figure 2(d)). The interface between thelatter two helices is composed of complementarybulky (F206, L220, F223) and smaller side-chains(V205, V209, A210, G224). In contrast, a large anglebetween a1 and a2 is required to accommodate thebulky side-chains in their interface (Y160, I164,K165, Y175, F178, L179). Most notably, the bulkyF223 in a4 faces the smaller V205 and V209 in a3,while the corresponding residue F178 in a2 pointstowards the also bulky residues Y160 and I164 in a1.Furthermore, Y175 in the top of the a1,a2-interfaceis more bulky than the corresponding L220 in thea3,a4-interface (Figure 2(d)). The resulting poorpacking of helices a1 and a2 creates a gap, which isfurther emphasized as these helices are longer thantheir counterparts a3 and a4.To examine possible structural changes of the

PAH2B domain upon binding of target peptides, we

Figure 2. (a) Awall-eye stereo view of the backbone traces of residues 148–232 of the ensemble of 30 structures of thefree PAH2B-105 domain. Residues 233–252 in the unstructured C terminus were excluded from the structure calculation.(b) A cartoon representation of the best representative structure of the ensemble. Helices are shown in red and arelabeled a1 to a4; the turns and coil regions are shown in grey. (c) A surface representation of unbound PAH2B, showingthe binding pocket. Hydrophobic residues are shown in yellow and those in the binding pocket are labeled. (d) Acomparison of the equivalent inter-helical angles a1,a2 (blue) and a3,a4 (red). Helix a2, helix a4 and the adjacent turnsare superimposed to show differences in packing in the helix interface (residues shown in stick representation). Residuescrucial for the inter-helical angle are labeled (a1–2/a3–4).


compared the structure of the unbound domainwith the two different complexes solved to date, thePAH2-Mad124–26 complex and the PAH2-HBP1complex.28 A structural superposition of the bestrepresentatives of each ensemble is shown inFigure 3. The helices in the free domain and in theMad1 complex superimpose well, with an RMSD of1.09 A, indicating that the global fold of PAH2 isunchanged after ligand binding. A slightly higherRMSD of 1.37 A is found for the HBP1 complex.Excluding helix a1, the RMSD drops to 0.86 A,reflecting a small but noticeable difference in theangle between helices a1 and a2 in the twoensembles of structures, which has been reportedby Swanson et al.28

To compare the side-chain orientations betweenthe three structures, only residues with sufficientexperimental restraints to result in well-definedside-chain orientations in the three ensembles ofstructures were taken into account. The resultingheavy-atom RMSD is 1.40 A (1.61 A) between thefree structure and the Mad1 complex (HBP1complex). This is roughly within the precision ofthe ensembles. More specifically, crucial residues inthe binding pocket have nearly identical side-chainpositions in both the free and the bound forms(Figure 3(b)). In general, these residues form

the “floor” of the binding pocket. Several residuesat the edges of the pocket, e.g. E183, Y160, H183,Q186, Q225 and F226, are disordered in the freestructure. Interestingly, residue F154 has a welldefined side-chain orientation in both the unboundstate and in the Mad1-SID complex, but isdisordered in the HBP1 complex. In the Mad1complex it is part of the binding pocket and isresponsible for several intermolecular contacts,while in the HBP1 complex it is pointing awayfrom the binding pocket, into the solvent.

Fast time-scale dynamics

In addition to inducing structural changes,binding events usually affect the dynamic behaviorof proteins. In earlier work, we have characterizedin detail the backbone dynamics of the Mad1-SIDcomplex occurring on a pico- to nanosecond time-scale.26 Here, we investigated these fast backbonemotions in the unbound state of PAH2B. The 15Nlongitudinal and transverse relaxation rates (R1 andR1r, respectively) and the steady-state nuclearOverhauser effect ({1H}–15N NOE) were measuredfor 69 out of 102 non-proline residues (seeSupplementary Data, Figure S1). The resonancesof the remaining 33 residues were too severely

Table 1. Structural statistics for the PAH2B domain

A. Restraint informationTotal number of distance restraints 1007Intra-residual/sequential/medium/long

408/251/228/120

TALOS derived dihedral anglerestraints f/j

65/66

B. Average RMS deviation from experimental restraintsAll distance restraints (A) 0.035G0.002All dihedral angle restraints (deg.) 0.26G0.07

C. Pairwise Cartesian RMS deviationOrdered backbone heavy atoms (A) 0.68G0.12Ordered all heavy atoms (A) 1.45G0.18Global backbone heavy atoms (A) 2.66G0.62Global all heavy atoms (A) 3.48G0.52

D. Ramachandran quality parametersResidues in most favored regions (%) 86.2Residues in allowed regions (%) 12.8Residues in additionally allowedregions (%)

0.5

Residues in disallowed regions (%) 0.5

E. Abnormalities found in structural checksAbnormally short inter-atomicdistances

3G1

Unsatisfied H-bond acceptors (buried) 0G0Unsatisfied H-bond donors (buried) 3G2

F. Average RMS deviation from current reliable structuresBond lengths (A) 0.72G0.02Bond angles (deg.) 0.73G0.02Omega angle restraints 0.62G0.05Side-chain planarity 0.72G0.09Improper dihedral distribution 0.74G0.03Inside/outside distribution 1.07G0.04G. Average deviation from current reliable structuresSecond generation packing quality K0.35G0.21Ramachandran plot appearance K2.1G0.7Chi-1/Chi-2 rotamer normality K0.62G0.55

Statistics are given for residues 148–232 of PAH2B-105, excludingthe unstructured C-terminal tail. Ordered regions are residues158–189 and 201–226.


overlapping to extract peak intensities reliably. Thestructured regions have NOE values close to 0.8,indicating that fast local motion is limited. LowNOE values with concomitant low R1r/R1 ratiosindicate high local flexibility for the N terminus, the

Figure 3. (a) Superposition of the backbones of the unbouMad1-SID24 (blue; PDB code 1PD7) and PAH2A bound to HBPsuperposition displayed in (a), showing the side-chains of rColor code as in (a).

loop between helices a2 and a3 and for theC-terminal tail. A few residues have high R1r/R1

ratios, indicating the presence of chemical exchangeon the micro- to millisecond time-scale. This iselaborated on in more detail in the slow dynamicssection.The C-terminal tail is highly flexible on a fast

time-scale, judged from its very low {1H}–15N NOEvalues (average 0.02G0.47) and R1r/R1 ratios(average 3.4(G2.3) sK1). It has been shown thatthe presence of such a flexible tail frustrates thequantitative description of dynamics using themodel-free approach, as evidenced by anomaloushigh values for the model-free parameters.29 As thereorientation of the tail is on the time-scale of theglobal reorientation, this results in a time-depen-dence of the diffusion tensor, which constitutes aviolation of the central assumption of the model-free approach that local and global motions can bedecoupled.30,31 In contrast, reduced spectral densitymapping does not make any assumption about thenature or time-scales of local and global motionsand is used to qualitatively describe the localdynamics in term of the spectral densities J(u) atthe frequencies 0, uN and 0.87uH.

32–34 In short,assuming no internal motion and a rigid isotropicrotor, J(0) equals 2

5tc, whereas smaller values for J(0)

indicate internal mobility of the N–H bond on asub-nanosecond time-scale. Fast internal motionsresult also in an increase in J(0.87uH) values, whilethe J(uN) spectral density decreases for small tomedium-sized proteins. Motions on a micro- tomillisecond time-scale result in a chemicalexchange contribution to the transverse relaxationrate. When the observed R1r is not corrected for thisadditional contribution, these motions are reflectedby elevated values for J(0), but do not affect theJ(uN) or J(0.87uH) values.Figure 4(a) presents J(0.87uH)–J(0) and J(uN)–J(0)

correlation plots and illustrates the markedlydifferent dynamics of the core and the tail of the

nd PAH2B structure (red), the PAH2B domain bound to1-SID (green; PDB code 1S5R). (b) An enlarged view of theesidues that interact with the Mad1-SID and HBP1-SID.

Figure 4. Results from the (a) and (b) fast, and (c) and (d) slow dynamics analysis. (a) The results of the reducedspectral density mapping analysis. Spectral densities J(0.87uH) (left panel) and J(uN) (right panel) are plotted as functionof the corresponding apparent J(0) value (uncorrected for chemical exchange). The broken-line curves show the expectedcorrelation for a rigid isotropic rotor. Data points cluster in three groups, labeled I (very flexible; red), II (flexible;magenta) and III (rigid; blue). Group III can be divided into a subgroup without and a subgroup with significantchemical exchange contribution, indicated as IIIa and IIIb, respectively. (b) The three groups are indicated on a structure-model including the C-terminal tail. Residues with no available data are in grey. (c) Representative 15N R2 relaxation–dispersion curves at 500 (red duplicate), 600 (green; duplicate) and 800 MHz (black; triplicate) proton frequency. Thecontinuous lines in the left panel are the best-fit curves obtained from a global fit as described in Materials andMethods;a broken straight line is shown in the right panel to guide the eye. (d) Results from the dispersion data analysis. Values ofthe exchange rate constant (kex; left) and the magnitude of the chemical shift difference (jDdbaj, right) are plotted on thestructure. The backbone nitrogen nuclei are shown as spheres for residues for which dispersion data were acquired.Residues without significant dispersion are in grey in the left panel.


protein. Three clusters can be readily identified.These clusters are schematically represented on astructure model that includes the C-terminal tail(Figure 4(b)). Firstly, the disordered C-terminal andN-terminal tails (group I; residues 148–151, 238–252) have an extremely high level of local flexibility,as indicated by the very low J(0) values andcorresponding high J(0.87uH) and lower J(uN)values. The J(0.87uH) and J(0) values and theJ(uN) and J(0) values show the expected linearcorrelation, indicating the superposition of a globaland local motion.32 Secondly, the relatively dis-ordered first two turns of helix a1, the a2–a3 loopand the transition region between the core of theprotein and the tail (group II, residues 152–156, 190–200, 234–237) also have increased flexibility on asub-nanosecond and a picosecond time-scale asjudged by their lower J(0) and higher J(0.87uH)spectral densities. Group III (residues 157–189, 201–

233) contains all residues in structured regions(except the N-terminal part of helix a1) and can bedivided into two subgroups. Group IIIa has a ratherhomogeneous J(0) close to the theoretical curve for arigid isotropic rotor, reflecting restricted internalmotions for the N–H vector. The apparent isotropictc is 9 ns for this group. The spread in J(uN) andJ(0.87uH) values reflects a small anisotropy in therotational diffusion. Group IIIb has elevatedapparent J(0) values, indicating significant chemicalexchange contributions. These residues (164, 167–169, 174, 183, 201, 215, 223, 226 and 232) wereidentified also by the relaxation-dispersion experi-ments and are discussed below.

Slow time-scale dynamics

To probe the backbone dynamics on a micro- tomillisecond time-scale, effective 15N-R2 values were

Figure 5. Correlation between the signed chemical shiftdifferences between major and minor forms of the PAH2B

domain and the chemical shift difference with the randomcoil shift. The continuous line is the linear least-squaresbest fit line through the data points (RZ0.82). Residuesfor which the relative error on the effective R2 was largerthan 10% and residues for which the deviation from therandom coil shift was less than 0.5 ppm (shown in gray)were excluded from the fit.


determined as a function of pulse spacing in Carr–Purcell–Meiboom–Gill (CPMG) relaxation dis-persion experiments. Dispersion profiles could bedetermined for 68 residues at three magnetic fieldstrengths. Residues in the N terminus and theC-terminal tail did not show any dispersion of theireffective R2 values. Most residues in the core of theprotein and in the transition region between helixa4 and the tail showed significant R2 dispersion.Representative dispersion profiles are shown inFigure 4(c). Assuming two-site exchange between amajor (a) and minor (b) state, exchange parameterswere determined by global fitting of the Carver-Richards-Jen all time-scale approximation to thedispersion profiles. All residues are in fast-to-intermediate exchange. Out of 43 residues withsignificant dispersion, 33 could be fit in a singlecluster with exchange rate, kex, equal to 1.30(G0.05)!103 sK1 and the population of the minorstate, pb, equal to 0.92(G0.05)%. The remaining tenresidues were fit using a second cluster with kexequal to 2.29(G0.13)!103 sK1 and pb equal to1.20(G0.07)% (Supplementary Data). The twoclusters are mapped onto the unbound PAH2B

structure (left panel of Figure 4(d)). The first clustercontains mostly residues in helices and a few in thea2–a3 loop. The cluster with the higher exchangerate comprises residues in the turn between helix a1and a2 and the C-terminal part of a4 withsubsequent residues. Relaxation dispersion profilesdepend also on the magnitude of the frequencydifference between the two exchanging states a andb (jDubaj). This frequency difference is expressedmore conveniently as a difference in chemical shifts,jDdbaj. The right panel of Figure 4(d) displays thejDdbaj values mapped onto the PAH2B structure. Onaverage, the chemical shift difference is w2 ppm.Residues in helix a3 experience, on average, thesmallest change in chemical shift (1.38(G0.3) ppm),while three regions form hot-spots with largeaverage values for jDdbaj: the turn between a1 anda2 (3.4(G2) ppm), residues 198–201, which form arelatively ordered part of the loop adjacent tohelix a3 (3.1(G1) ppm), and the C-terminal part ofhelix a4 with adjacent residues in the linker region(4.6(G2) ppm).

Structural characterization of the minor confor-mation is possible only if both the magnitude andthe sign of Ddba are known, allowing reconstructionof the chemical shifts of the minor state.35–38 Wedetermined the sign using a new experiment,named CEESY66 and confirmed these results usingthe comparison of peak positions in heteronuclearsingle quantum and multiple quantum correlationspectra.35 The resulting signs of the chemical shiftdifferences were combined with their magnitude, asdetermined from the relaxation-dispersion experi-ments. Comparison of the resulting signed chemicalshift differences, Ddba, with the chemical shiftdifferences between the free and bound forms ofPAH2 showed no correlation (RZ0.08). However,comparison of Ddab with the deviation from therandom coil shifts39 resulted in a reasonable

correlation (RZ0.82) (Figure 5), suggesting thatthe minor conformation is a (partially) unfoldedstate. Comparison with shifts of fully unfoldedPAH2 in 8 M guanidinium chloride resulted in asimilar correlation (data not shown). Furthermore,our CEESY experiment enabled us to determine thesign of the frequency difference for the amideproton.66 As shown in Table 2, nearly all signs ofDdba for both the amide proton and nitrogen atom inthe three hot-spots indicate a shift toward randomcoil values, in agreement with an order–disordertransition.

Discussion

Influence of the C-terminal tail

We observed significant differences in the 15N-HSQC spectra of two different PAH2B domainconstructs. While a spectrum consistent with onedominant folded conformation is observed forPAH2B-105, that of PAH2B-85 is typical for a proteinin intermediate exchange between multiple confor-mations. Given the limited chemical shift dis-persion, it is likely that the exchange process inPAH2B-85 is, similar to that in PAH2B-105, aninterconversion between the folded state and apartially unfolded molten globule state. There is noevidence that the observed stabilization of thefolded conformation in PAH2B-105 is caused by a(partial) association of the tail with the bindingpocket, or any other part of the PAH2 domain.Given the extreme flexibility of the tail, transientinteractions are also unlikely. Improvement in thesolubility by the highly hydrophilic tail could

Table 2. Structural interpretation of exchange process

Residue HN/Na Sign Ddba

b Ddroc

Randomcoild

F168 HN n.d. –N C C7.11 Yes

L169 HN C C1.03 YesN – K3.65 Yes

D170 HN – K0.12 YesN C C5.97 Yes

H171 HN n.d. –N – C0.92 No

R198 HN n.d. –N – K2.97 Yes

G199 HN n.d. –N – K2.34 Yes

M200 HN n.d. –N – K0.80 Yes

S201 HN – K0.52 YesN – K4.98 Yes

F223 HN C C0.46 YesN – K0.74 Yes

K231 HN C C0.39 YesN n.d. –

R232 HN C C0.81 YesN C C1.50 Yes

a Sign information on exchange process for amide proton (HN)or nitrogen (N). Only residues in the hot-spots are listed.

b Sign of chemical shift difference between (b) minor and (a)major conformational state (n.d., under threshold, sign notdetermined).

c Difference with sign between the random coil (r) and theobserved chemical shift (o).

d Yes if the minor state is more random coil-like (Ddba and Ddrohave the same sign); No if the minor state is less random coil-like(Ddba and Ddro have opposite signs).


contribute to the observed stabilization, althoughthe short construct did not show severe problemswith precipitation. Alternatively, it is plausible thatthe tail reduces the formation of soluble micro-aggregates that could be formed by partiallyunfolded states of the protein domain. In addition,charged residues in the tail could be involved instabilizing long-range electrostatic interactions.Together, this could result in a stabilization of thefolded conformation, resembling the way in whichthe GB1 domain can be used as a solubilizationenhancement tag.40

Sequence analysis shows that the sequence of thisC-terminal region is conserved in the B-isoforms ofthe PAH2 domain of the vertebrates human, mouseand the African clawed frog, which could point to ageneral requirement for this region for the stabili-zation of the PAH2 domain of mSin3B (data notshown). After submission of the manuscript, thesolution structure of the PAH1 domain of mSin3Bbound to the SID of the NRSF/REST repressor waspublished.41 Interestingly, this structure shows thata conserved region after the C terminus of the PAH1domain is folded as a short helix, and is packedagainst helices a3 and a4 of the PAH1 domain.41

Structural plasticity of complex formation

The global folds of the unbound, Mad1-boundand HBP1-bound PAH2 domain are virtuallyidentical. This is rather surprising, as the hydro-

phobic interaction surface of the PAH2 domain isexposed to the solvent in the unbound form(Figure 2(c)). This open conformation is a conse-quence of the poor complementarity of side-chainsin the a1,a2-interface, forcing helix a1 out of thecore formed by helices a2, a3 and a4 (Figure 2(d)).The limited structural rearrangement of the mainchain is paralleled by only minor changes in side-chain conformation upon binding. Specifically,most side-chains that interact directly with the SIDhave practically identical conformations in theunbound and both the Mad1-SID and HBP1-SIDbound states.

In contrast, larger structural changes are thoughtto occur within the SID. Both the SID of Mad1 andHBP1 are most likely unstructured in their freestate, but fold to form an amphipathic a-helix whenbound to PAH2.28,42 The folded SID interacts withthe binding pocket of PAH2 via a precise arrange-ment of bulky and small side-chains, like knobs-in-holes. Interestingly, HBP1 and Mad1 bind withopposite helical orientations to complement the pre-defined knobs and holes of PAH2, as the crucial SIDresidues are reversed in the sequence.28 A recentmutation study stressed the importance of thecomplementarity of bulky and small residues inthe center of the interaction surface as a basis forspecificity and affinity of the SID–PAH2 inter-action.43 In addition to these general interactions,Mad1 and HBP1 each have a number of uniqueinteractions with PAH2, involving residues at theedge of the binding pocket. The unique stackinginteraction between F154 of PAH2 and Y18 of Mad1was identified as the critical determinant ofspecificity of Mad1 for the PAH2 versus the PAH1domain.44 Notably, F154 becomes disordered in theHBP1 complex, as the HBP1-SID does not providean aromatic residue at the corresponding positionof Mad1-Y18. Conversely, V152, E153, P228 and afew residues at the C-terminal end of helix a2interact uniquely with HBP1. In addition, residuesat the edges of the binding pocket of PAH2 areinvolved in specific intermolecular interactionswith residues outside the minimal Mad1-SID.26

The view emerging from the current dataindicates that complex formation involves a foldingtransition of the SID, while the PAH2 domainexperiences only limited structural rearrangements.Recent computational analyses of protein–proteininteractions have shown that the conformation ofcrucial residues in the unbound and in the boundstate can be very similar for at least one of theinteracting proteins.45,46 The structured, unboundPAH2 domain might then serve as a foldingtemplate, defining the proper fold of the SID. Thissuggests that the short construct, PAH2B-85, whichis only partially folded, will have a lower bindingaffinity for the SID. A reasonable mechanism forcomplex formation is a combination of induced fitand conformational selection.46 In this model, a fewresidues of the SID serve as “anchors”, whose side-chains are inserted into the well-defined pocket inthe core of the interaction surface of PAH2,


producing a native-like encounter complex. Wesuggest that the two central bulky hydrophobicresidues (F) in the FXXFFXAA minimal inter-action motif, i.e. L12 and L13 in Mad1, act asanchor-residues. Binding affinity is subsequentlyoptimized by an induced-fit process in the peri-phery of the binding interface,46,47 including theformation of SID-dependent, unique interactions.

Dynamic plasticity of complex formation

The PAH2B domain shows a rich and complexdynamics, both on the fast and slow time-scales.Reduced spectral density mapping showed that thehelices, the loop and the tail form three regions withincreasing fast local flexibility. On a micro- tomillisecond time-scale, the PAH2B domain isinterconverting with a minor form that is populatedfor roughly 1%. The turn between helices a1 and a2,a relatively ordered part of the loop and theC-terminal part of helix a4 with the adjacent linkerregion form three hot-spots that experience largechanges in chemical environment. The minor statemost likely constitutes a partially unfolded state, asindicated by a shift towards random coil values forthe chemical shifts of both the amide 1H and 15Nnucleus. Interestingly, residues in the turn betweena1 and a2 show exchange on a significantly fastertime-scale than the helices, which could be theresult of a modulation of the angle between a1 anda2, altering the accessibility of the binding pocket.

Figure 6 shows the ten most significant changesin the fast time-scale dynamics between unboundPAH2B and the Mad1–SID complex.26 Complexformation results in a reduction in fast time-scaledynamics for the first two turns of a1, as shown by

Figure 6. Comparison of the dynamics of unbound andMad1-SID bound PAH2. The ten most significant changesin the J(0), J(0.87uH) correlation are shown. Squaresrepresent data points of the unbound PAH2 domain,circles are shown for the Mad1-SID complex. Arrows areshown to connect residues in the free and bound states.Data points are labeled with their residue number. Filledsymbols are used for residues in the first turns of helix a1,open symbols are used for residues in the rigid part of thestructure.

the drastic reduction in the J(0.87uH) density with aconcomitant increase in J(0). This correlates with theformation of intermolecular interactions that “lock”this helix. Other parts of the interaction surface donot experience such large changes, indicating thatthe impact of binding on the fast time-scaledynamics seems to be restricted to the part of theinteraction surface that is more flexible in theunbound state. On the other hand, binding reducesslow time-scale motions for all residues thatexperience chemical exchange in the unboundstate, as demonstrated by a decrease in theirapparent J(0) values. This finding was confirmedalso by the absence of significant dispersion of 15Ntransverse relaxation rates, as determined withrelaxation dispersion experiments on the PAH2–Mad1 complex (data not shown). Interaction withthe SID thus stabilizes the PAH2 domain withrespect to the partially unfolded state, which maybe correlated with the protection of the hydro-phobic binding pocket from the solvent. Finally, thechange in dynamics of residue Y175 seems tosignify a more isotropic global rotational diffusionfor the complex compared to the free domain(Figure 6).

PAH2 in Sin3A versus Sin3B

The vertebrate Sin3 protein exists in two iso-forms, known as mSin3A and mSin3B,19 as a resultof gene duplication early in evolution.8 Interest-ingly, NMR spectra of the unbound PAH2 domainof mouse Sin3A, PAH2A, showed that the domainexists in two forms in a roughly one-to-one ratiothat interconvert slowly on the NMR time-scale,24,28

as opposed to the fast-to-intermediate interconver-sion with a 1% populated minor state observed forPAH2B in this study. We suggest that both exchangeprocesses are interconversions between a partiallyunfolded and a fully folded state. The observeddifferences in the kinetics and thermodynamics ofthis interconversion have to originate from differ-ences in sequence, as the experimental conditionsfor the PAH2A and PAH2B studies are comparable.Indeed, the hot spots of exchange are also theregions with the highest level of sequence varia-bility (Figure 1(b)). Different thermodynamic stabi-lity of the PAH domain could be related tofunctional differences between mSin3A andmSin3B by altering the free energy of interactionwith binding partners. Also, different life-times forthe folded form, caused by the different exchangetime-scales, can affect binding of interacting pro-teins.

The PAH domains in full-length Sin3

To assess the conservation of crucial residuesbetween the different PAH domains, we performeda multiple sequence alignment of 35 Sin3 homo-logues and subsequently aligned their PAHdomains using ClustalW48 (Supplementary Data,Figure S3). The region identified originally as the

† http://www.cgl.ucsf.edu/home/sparky/


PAH4 domain has, on average, only 9% identitywith the other three PAH domains (SupplementaryData, Table S2). This region is not identified as aPAH domain by domain detection programs suchas NCBI’s conservative domain search,49 anddifferent domain boundaries have beenreported.19,50 The alignment of the PAH4 domainwith the other PAH domain used here wasoptimized manually to match the structurallyimportant residues of the PAH2 domain. However,several buried, hydrophobic residues, which arestructurally important and highly conserved inPAH1-3, are replaced by hydrophilic and/or non-conserved residues in PAH4 (e.g. residues V209,L219 and L220 in PAH2; Supplementary Data, TableS3). Alternative alignments result in mismatches atother positions. This indicates that the PAH4 regionmost likely does not fold as a four-helix bundle, butinstead adapts a distinct fold. Contrastingly, thePAH1, PAH2 and PAH3 domains are highlyconserved and have significant homology (Sup-plementary Data, Table S2). Residues in the coreand in the interaction surface are conserved orconservatively mutated throughout these PAHdomains (Supplementary Data, Table S3), indicat-ing that the general fold and mode of interactionwith the SID is conserved. Residues at the edges ofthe binding pocket (V152-I158, Q186 and E202) areless conserved, indicating that these residues play arole in defining the specificity of eachmember of thePAH domain family (Supplementary Data, Table S3and Figure S2). Notably, A157 of PAH2B that is partof the interaction surface is conserved in the PAH1domain but is replaced by a conserved glutamicacid residue in PAH3, suggesting a role forelectrostatic interaction with the SID. Furthermore,the bulky and buried aromatic side-chains in thehelix a1,a2 interface (F178 and Y160) and the buriedside-chains in the a3,a4 interface (V205, V209, andF223) are conserved between the different PAH2sequences, and conserved or mutated conserva-tively throughout the PAH1 and PAH3 domains(Supplementary Data, Table S3). This suggestsstrongly that the “open” conformation is anintrinsic feature of the PAH domain. The solutionstructure of the PAH1 domain of mSin3B bound tothe SID of the NRSF/REST repressor, publishedafter submission of our manuscript, shows that thecomplex is very similar to the structures of thePAH2 domain complexes: the PAH1 domain isfolded as a four-helix bundle and the SID is wedgedbetween helices a1 and a2.41 NMR spectra of theunbound PAH1 domain both in that study and inour own experiments (data not shown) are of aquality similar to that of the extended PAH2B

construct, indicating that the unbound PAH1domain is probably folded as well. Thus, on thebasis of our structure of the PAH2B domain and themultiple sequence alignment of the PAH domains,we suggest that the PAH1, PAH2 and PAH3domains can form folded modules in full-lengthSin3, like beads-on-a-string, and that the PAH4domain probably has a distinct fold.

Conclusion

The solution structure of the unbound PAH2domain shows that the domain is folded and that itsconformation does not change appreciably uponbinding of its interaction partners, either the Mad1-SID or HBP1-SID. Notably, the binding pocket isaccessible in the unbound structure as a result ofconserved bulky residues in the helix a1,a2 inter-face that prevent close packing of a1 with the core.In addition, the orientation of crucial interactingside-chains is highly similar in the bound andunbound states, suggesting that they guide foldingof the unstructured SID by defining a foldingtemplate. A more pronounced change is seen inthe dynamics of unbound PAH2. The N-terminalpart of helix a1 is rigidified by intermolecularinteractions with the SID. Furthermore, confor-mational exchange with a minor, partially unfoldedstate is quenched when the SID is bound, indicatinga stabilization of the folded state of the domain. Onthe basis of our data and multiple sequencealignment of the PAH domains, we suggest thatthe PAH1, PAH2 and PAH3 domains form foldedbinding modules in full-length Sin3 like beads-on-a-string and act as a folding template for the SID incomplex formation.

Materials and Methods

Sample preparation

All NMR studies were performed using a proteinconstruct of 85 or 105 residues (denoted as PAH2B)corresponding to residues 148–232 (PAH2B-85) and 148–252 (PAH2B-105), respectively, of the long variant of Mm.mSin3B (SpTrEMBL accession number Q62141). Cloning,expression and purification of the PAH2B domain wereperformed as described.51 Uniformly 15N/13C-labeledPAH2B was prepared using 15NNH4Cl and [13C6]glucoseas sole nitrogen and carbon sources. NMR samplescontained 0.5 to 0.9 mM 13C/15N double-labeled proteinin a buffer of 50 mM KH2PO4/K2HPO4 (pH 6.3), 100 mMKCl, 95%H2O/5% 2H2O using protease inhibitor Pefablocand a trace of NaN3 as preservative.

NMR spectroscopy

All NMR experiments were carried out on Varian UnityInova 500, 600 and 800 MHz spectrometers. The data wereprocessed using the NMRPipe suite,52 and analyzed usingXEASY53 and Sparky†. Assignments were obtained using3DHNCA,HNCACB,CBCACONH,HNCAHA, (H)CCH-total correlated spectroscopy (TOCSY), CBCDHD, 15NNOESY-HSQC and aliphatic and aromatic 13C NOESY-HSQCspectra. Stereo-specific assignments of the pro-R andpro-Smethyl groups ofVal andLeu-residuesweremade onthe basis of a CT-13C-HSQC of 10% 13C-labeled PAH2B

sample.54Distance restraints for structure calculationswereobtained from the 3D 15N and 13C-separated NOESYspectra. The NOE mixing time in all NOESY experimentswas set to 100 ms. Backbone N, HN, Ca, Cb, C0, and Ha

http://www.cgl.ucsf.edu/home/sparky/


atoms of unfolded PAH2were assigned using a HNCACB,HNCO and HNCAHA experiments.The 15N-T1,

15N-T1r and {1H}–15N NOE experimentswere recorded at 20.0 8C at 14.1 T. All experiments wererecorded in an interleaved manner. Relaxationdelays were: 16 ms, 256 ms (2!), 384 ms, 512 ms,768 ms (2!) and 1024 ms for the T1-experiment and16 ms, 32 ms (2!), 48 ms, 64 ms, 96 ms (2!) and 128 msfor the T1r-experiment. For both experiments, the recycledelay was 1.55 s. The {1H}–15N NOE values were derivedfrom experiments using 3 s either on or off-resonanceirradiation in a total recycle delay of 5 s. Additionally, a12 s delay was executed after acquisition of each complexpoint. Relaxation dispersion experiments were recordedat 20 8C for 15N-SQ coherence at 11.7 T (duplicateexperiments), 14.1 (duplicate experiments) and 18.8 T(triplicate experiments) using a standard constant timerelaxation compensated CPMG scheme.55,56 The constant-time value was set to 40 ms and the effective CPMG fieldstrength was set to 50 Hz, 100 Hz, 200 Hz, 300 Hz, 400 Hz500 Hz, 700 Hz, 1000 Hz and 1400 Hz using a 15N 908pulse of w52 ms. The recycle delay was 2 s.

Structure calculations and refinement

The NOE peaks were assigned and converted intodistance restraints by CYANA v1.05.57,58 Dihedral anglerestraints for the j–f angles were derived from N, HN,C 0, Ca, Cb and Ha chemical shifts using TALOS.59 Aninitial set of 100 NMR structures was calculated withCYANA and the 70 best structures were subsequentlyrefined in water using a restrained molecular dynamicsprotocol60,61 to improve local geometry and electrostatics.Of the resulting structures, the 30 lowest energystructures with no distance restraint violations O0.5 Aor dihedral angle violationsO58were selected to form thefinal ensemble. The best representative structure (no. 16of the ensemble) is defined as the structure closest to themean structure. Structures were analyzed using theprograms PROCHECK-NMR62 and WHAT-CHECK.63

Fast time-scale dynamics analysis

All spectra were processed and zero filled to a final sizeof 8K (1H)!4K (15N) points. The longitudinal andtransverse relaxation rates were calculated by fittingpeak intensities to a mono-exponential decay using themodelXY module of NMRPipe.52 Errors in peak heightswere estimated by the RMS of the noise-level. Errors (1sconfidence limits) were estimated from 21 Monte Carlosimulations. In a conservative approach, all errors for R1,R1r and NOE were estimated to be at least 3% to preventfalsely attributed contributions of internal motions orexchange. Subsequently, the R1, R1r and NOE values wereanalyzed using reduced spectral density mapping todescribe the motion of the N–H vector in terms of thespectral density at 0, uN and 0.87uH frequencies.35–37 Thechemical shift anisotropy of the 15N nucleus was assumeduniform with a value of K170 ppm and the N–H bondlength was set to 1.02 A.

Slow time-scale dynamics analysis

All spectra were processed and zero filled to a final sizeof 8K (1H) !4K (15N) points. The effective R2 wascalculated on the basis of peak intensities.55 Uncertaintiesin peak intensities were based on the RMS of the noise-level. The dispersion of R2 values was defined as

significant when the RMSD of a fit to a straight line waslarger than twice the error based on noise. The programGLOVE (by J. C. Lansing and P. E. Wright, The ScrippsResearch Institute, La Jolla, CA, USA) was used to fit theeffective R2 values for all data sets simultaneous to theCarver–Richards–Jen all time-scale approximation.64,65

First, all residues were fit individually and the reducedc2 value of residue i, c2

i;individual, was calculated. Then,residues in the same secondary structure element werefitted simultaneous in a global fit and the reduced chi-squared value of residue i in the global fit, c2

i;global, wascalculated. In the global fit the exchange rate constant kexand the populations pa and pb were optimized globally,while the chemical shift difference Du and the exchangefree R2 were optimized for each fitted residue. Ifnecessary, these elements were redefined until the globalfit was accepted, defined as c2

i;global!1:5c2i;individual.

Subsequently, these elements were iteratively combinedin global fits as long as these fits were accepted.

Databank accession numbers

The coordinates of the structures are deposited in theRCSB Protein Data Bank under PDB accession number2F05. The chemical shifts and relaxation data of the freePAH2B domain are deposited in the BioMagResBankunder BMRB accession number 6899.

Acknowledgements

We gratefully thank Dr John Lansing andProfessor Dr P. E. Wright for the opportunity touse GLOVE for the analysis of the dynamics dataand helpful discussions. We thank Dr Mark Hilgefor help with the structure calculations, Dr SanderNabuurs for calculation of the U-factors andProfessor Dr C. W. Hilbers for his comments onthe manuscript. This research was funded by aJonge Chemici grant of the Dutch organization forscientific research, NWO-CW (JC 99-03) to G.W.V.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2006.01.100The Supplementary Data includes Figures plot-

ting the per-residue structural quality U-factors, all15N-relaxation rates, Tables containing the results ofthe dispersion data analysis, sequence conservationbetween the PAH domains and sequence alignmentof the PAH domain of 35 proteins.

References

1. Roeder, R. G. (2003). The eukaryotic transcriptionalmachinery: complexities and mechanisms unfore-seen. Nature Med. 9, 1239–1244.

2. Ptashne, M. (2005). Regulation of transcription: fromlambda to eukaryotes. Trends Biochem. Sci. 30, 275–278.

3. Ptashne, M. & Gann, A. (1997). Transcriptionalactivation by recruitment. Nature, 386, 569–577.

http://dx.doi.org/doi:10.1016/j.jmb.2006.01.100

http://dx.doi.org/doi:10.1016/j.jmb.2006.01.100


4. Berger, S. L. (2002). Histone modifications in tran-scriptional regulation. Curr. Opin. Genet. Dev. 12,142–148.

5. Cosma, M. P. (2002). Ordered recruitment: gene-specific mechanism of transcription activation. Mol.Cell, 10, 227–238.

6. Burke, L. J. & Baniahmad, A. (2000). Co-repressors2000. FASEB J. 14, 1876–1888.

7. Struhl, K. (1998). Histone acetylation and transcrip-tional regulatory mechanisms. Genes Dev. 12, 599–606.

8. Silverstein, R. A. & Ekwall, K. (2005). Sin3: a flexibleregulator of global gene expression and genomestability. Curr. Genet. 47, 1–17.

9. Hassig, C. A., Fleischer, T. C., Billin, A. N., Schreiber,S. L. & Ayer, D. E. (1997). Histone deacetylase activityis required for full transcriptional repression bymSin3A. Cell, 89, 341–347.

10. Laherty, C. D., Yang, W. M., Sun, J. M., Davie, J. R.,Seto, E. & Eisenman, R. N. (1997). Histone deacety-lases associated with the Sin3 corepressor mediateMad transcriptional repression. Cell, 89, 349–356.

11. Zhang, Y., Iratni, R., Erdjument-Bromage, H., Tempst,P. & Reinberg, D. (1997). Histone deacetylases andSAP18, a novel polypeptide, are components of ahuman Sin3 complex. Cell, 89, 357–364.

12. Vermeulen, M., Carozza, M. J., Lasonder, E.,Workman, J. L., Logie, C. & Stunnenberg, H. G.(2004). In vitro targeting reveals intrinsic histone tailspecificity of the Sin3/histone deacetylase andN-CoR/SMRT corepressor complexes. Mol. Cell. Biol.24, 2364–2372.

13. Eberharter, A. & Becker, P. B. (2002). Histoneacetylation: a switch between repressive and permiss-ive chromatin. Second in review series on chromatindynamics. EMBO Rep. 3, 224–229.

14. Zhang, Y., Sun, Z. W., Iratni, R., Erdjument-Bromage,H., Tempst, P., Hampsey, M. & Reinberg, D. (1998).SAP30, a novel protein conserved between humanand yeast, is a component of a histone deacetylasecomplex. Mol. Cell, 1, 1021–1031.

15. Alland, L., David, G., Hong, S. L., Potes, J., Muhle, R.,Lee, H. C. et al. (2002). Identification of mammalianSds3 as an integral component of the Sin3/histonedeacetylase corepressor complex. Mol. Cell. Biol. 22,2743–2750.

16. Fleischer, T. C., Yun, U. J. & Ayer, D. E. (2003).Identification and characterization of three newcomponents of the mSin3A corepressor complex.Mol. Cell. Biol. 23, 3456–3467.

17. Yang, X., Zhang, F. & Kudlow, J. E. (2002). Recruit-ment of O-GlcNAc transferase to promoters bycorepressor mSin3A: coupling protein O-GlcNAcety-lation to transcriptional repression. Cell, 110, 69–80.

18. Wang, H., Clark, I., Nicholson, P. R., Herskowitz, I. &Stillman, D. J. (1990). The Sacchoromyces cerevisaeSIN3 gene, a negative regulator of HO, contains fouramphipathic helix motifs. Mol. Cell. Biol. 10,5927–5936.

19. Ayer, D. E., Lawrence, Q. A. & Eisenman, R. N. (1995).Mad-Max transcriptional repression is mediatedby ternary complex formation with mammalianhomologs of yeast repressor Sin3. Cell, 80, 767–776.

20. Dannenberg, J.-H., David, G., Zhong, S., Torre, J.,Wong, W. H. & DePinho, R. A. (2005). mSin3Acorepressor regulates diverse transcriptional net-works governing normal and neoplastic growth andsurvival. Genes Dev. 19, 1581–1592.

21. Cowley, S. M., Iritani, B. M., Mendrysa, S. M., Xu, T.,Cheng, P. F., Yada, J. et al. (2005). The mSin3A

chromatin-modifying complex is essential for embry-ogenesis and T-cell development. Mol. Cell. Biol. 25,6990–7004.

22. Adhikary, S. & Eilers, M. (2005). Transcriptionalregulation and transformation by MYC proteins.Nature Rev. Mol. Cell Biol. 6, 635–645.

23. Grandori, C., Cowley, S. M., James, L. P. & Eisenman,R. N. (2000). The Myc/Max/Mad network of thetranscriptional control of cell behavior. Annu. Rev.Cell. Dev. Biol. 16, 653–699.

24. Brubaker, K., Cowley, S. M., Huang, K., Loo, V.,Yochum, G. S., Ayer, D. E. et al. (2000). Solutionstructure of the interacting domains of the Mad-Sin3complex: implications for recruitment of a chromatin-modifying complex. Cell, 103, 655–665.

25. Spronk, C. A. E. M., Tessari, M., Kaan, A. M., Jansen,J. F. A., Vermeulen, M., Stunnenberg, H. G. & Vuister,G. W. (2000). The Mad1-Sin3B interaction involves anovel helical fold. Nature Struct. Biol. 7, 1100–1104.

26. van Ingen, H., Lasonder, E., Jansen, J. F. A., Kaan,A. M., Spronk, C. A. E. M., Stunnenberg, H. G. M. &Vuister, G. W. (2004). Extension of the binding motif ofthe Sin3 interacting domain of the Mad familyproteins. Biochemistry, 43, 46–54.

27. Nabuurs, S. B., Krieger, E., Spronk, C. A. E. M.,Nederveen, A. J., Vriend, G. & Vuister, G. W. (2005).Definition of a new information absed per residuequality parameter. J. Biomol. NMR, 33, 123–134.

28. Swanson, K. A., Knoepfler, P. S., Huang, K., Kang,R. S., Cowley, S. M., Laherty, C. D. et al. (2004). HBP1and Mad1 repressors bind the Sin3 corepressor PAH2domain with opposite helical orientations. NatureStruct. Mol. Biol. 11, 738–746.

29. Viles, J. H., Donne, D., Kroon, G., Prusiner, S. B.,Cohen, F. E., Dyson, H. J. & Wright, P. E. (2001). Localstructural plasticity of the prion protein. Analysis ofNMR relaxation dynamics. Biochemistry, 40,2743–2753.

30. Lipari, G. & Szabo, A. (1982). Model-free approach tothe interpretation of nuclear magnetic resonance inmacromolecules. 1. Theory and range of validity.J. Am. Chem. Soc. 104, 4546–4559.

31. Lipari, G. & Szabo, A. (1982). Model-free approach tothe interpretation of nuclear magnetic resonance inmacromolecules. 2. Analysis of experimental results.J. Am. Chem. Soc. 104, 4559–4570.

32. Lefevre, J. F., Dayie, K. T., Peng, J. W. & Wagner, G.(1996). Internal mobility in the partially folded DNAbinding and dimerization domains of GAL4: NMRanalysis of the N–H spectral density functions.Biochemistry, 35, 2674–2686.

33. Ishima, R. & Nagayama, K. (1996). Quasi-spectral-density function analysis for nitrogen-15 nuclei inproteins. J. Magn. Res. ser. B, 108, 73–76.

34. Farrow, N. A., Zhang, O., Szabo, A., Torchia, D. A. &Kay, L. E. (1995). Spectral density function mappingusing 15N relaxation data exclusively. J. Biomol. NMR,6, 153–162.

35. Skrynnikov, N. R., Dahlquist, F. W. & Kay, L. E. (2002).Reconstructing NMR spectra of “invisible” excitedprotein states using HSQC and HMQC experiments.J. Am. Chem. Soc. 124, 12352–12360.

36. Grey, M. J., Wang, C. & Palmer, A. G. (2003). Disulfidebond isomerization in basic pancreatic trypsin inhibi-tor: multisite chemical exchange quantified by CPMGrelaxation dispersion and chemical shift modeling.J. Am. Chem. Soc. 125, 14324–14335.

37. McElheny, D., Schnell, J. R., Lansing, J. C., Dyson, H. J.& Wright, P. E. (2005). Defining the role of active-site


loop fluctuations in dihydrofolate reductase catalysis.Proc. Natl Acad. Sci. USA, 102, 5032–5037.

38. Beach, H., Cole, R., Gill, M. L. & Loria, J. P. (2005).Conservation of mus-ms enzyme motions in the apo-and substrate-mimicked state. J. Am. Chem. Soc. 127,9167–9176.

39. Schwarzinger, S., Kroon, G. J. A., Foss, T. R., Wright,P. E. & Dyson, H. J. (2000). Random coil chemicalshifts in acidic 8 M urea: implementation of randomcoil shift data in NMRView. J. Biomol. NMR, 18, 43–48.

40. Zhou, P., Lugovskoy, A. A. & Wagner, G. (2001). Asolubility enhancement tag (SET) for NMR studies ofpoorly behaving proteins. J. Biomol. NMR, 20, 11–14.

41. Nomura, M., Uda-Tochio, H., Muurai, K., Mori, N. &Nishimura, Y. (2005). The neural repressor NRSF/REST binds the PAH1 domain of the Sin3 corepressorby using its distinct short hydrophobic helix. J. Mol.Biol. 354, 903–915.

42. Eilers, A. L., Billin, A. N., Liu, J. & Ayers, D. E. (1999).A 13-amino acid amphipathic a-helix is required forthe functional interaction between the transcriptionalrepressor Mad1 and mSin3A. J. Biol. Chem. 274,32750–32756.

43. Cowley, S. M., Kang, R. S., Frangioni, J. V., Yada, J. J.,DeGrand, A. M., Radhakrishnan, I. & Eisenman, R. N.(2004). Functional analysis of the Mad1-mSin3Arepressor-corepressor interaction reveals determi-nants of specificity, affinity, and transcriptionalresponse. Mol. Cell. Biol. 24, 2698–2709.

44. Le Guezennec, X., Vriend, G. & Stunnenberg, H. G.(2004). Molecular determinants of the interaction ofMad with the PAH2 domain of mSin3. J. Biol. Chem.279, 25823–25829.

45. Li, X., Keskin, O.,Ma, B., Nussinov, R.&Liang, J. (2004).Protein–protein interactions: hot spots and structurallyconserved residues often locate in complementedpockets that pre-organized in the unbound states:implications for docking. J. Mol. Biol. 344, 781–795.

46. Ramajani, D., Thiel, S., Vajda, S. & Camacho, C. J.(2004). Anchor residues in protein–protein inter-actions. Proc. Natl Acad. Sci. USA, 101, 11287–11292.

47. Smith, G., Sternberg, M. J. E. & Bates, P. A. (2005). Therelationship between the flexibility of proteins andtheir conformational states on forming protein–protein complexes with an application to protein–protein docking. J. Mol. Biol. 347, 1077–1101.

48. Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994).CLUSTALW: improving the sensitivity of progressivemultiple sequence alignment through sequenceweighting, positions-specific gap penalties andweight matrix choice. Nucl. Acids Res. 22, 4673–4680.

49. Marchler-Bauer, A. & Bryant, S. H. (2004). CD-search:protein domain annotations on the fly.Nucl. Acids Res.32, W327–W331.

50. Halleck, M. S., Pownall, S., Harder, K. W., Duncan,A. M. V., Jirik, F. R. & Schlegel, R. A. (1995). A widelydistributed putative mammalian transcriptional reg-ulator containing multiple paired amphipathic helices,with similarity to yeast SIN3. Genomics, 26, 403–406.

51. Spronk, C. A. E. M., Jansen, J. F. A., Tessari, M., Kaan,A. M., Aelen, J. & Lasonder, E. (2001). Letter to theeditor: sequence-specific assignment of the PAH2domain of Sin3B free and bound to Mad1. J. Biomol.NMR, 19, 377–378.

52. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G.,Pfeiffer, J. & Bax, A. (1995). NMRPipe—a multi-dimensional processing system based on UNIX pipes.J. Biomol. NMR, 4, 277–293.

53. Bartels, C. H., Xia, T. H., Billeter, M., Guntert, P. &Wuthrich, K. (1995). The program XEASY forcomputer-supported NMR spectral analysis of bio-logical macromolecules. J. Biomol. NMR, 6, 1–10.

54. Neri, D., Szyperski, T., Otting, G., Senn, H. &Wuthrich, K. (1989). Stereospecific nuclear magneticresonance assignments of the methyl groups of valineand leucine in the DNA-binding domain of the 434repressor by biosynthetically directed fractional 13Clabeling. Biochemistry, 28, 7510–7516.

55. Mulder, F. A., Skrynnikov, N. R., Hon, B., Dahlquist,F. W. & Kay, L. E. (2001). Measurement of slow (micros-ms) time scale dynamics in protein side-chainsby (15)N relaxation dispersion NMR spectroscopy:application to Asn and Gln residues in a cavity mutantof T4 lysozyme. J. Am. Chem. Soc. 123, 967–975.

56. Loria, J. P., Rance, M. & Palmer, A. G. (1999). Arelaxation-compensated Carr-Purcell-Meiboom-Gillsequence for characterizing chemical exchange byNMR spectroscopy. J. Am. Chem. Soc. 121, 2331–2332.

57. Guntert, P., Mumenthaler, C. & Wuthrich, K. (1997).Torsion angle dynamics for NMR structure calcu-lation with the new program DYANA. J. Mol. Biol.273, 283–298.

58. Herrmann, T., Guntert, P. & Wuthrich, K. (2002).Protein NMR structure determination with auto-mated NOE assignment using the new softwareCANDID and the torsion angle dynamics algorithmDYANA. J. Mol. Biol. 319, 209–227.

59. Cornilescu, G., Delaglio, F. & Bax, A. (1999). Proteinbackbone angle restraints from searching a databasefor chemical shift and sequence homology. J. Biomol.NMR, 13, 289–302.

60. Spronk, C. A. E. M., Linge, J. P., Hilbers, C. W. &Vuister, G. W. (2002). Improving the quality of proteinstructures determined by NMR spectroscopy.J. Biomol. NMR, 22, 281–289.

61. Linge, J. P., Williams, M. A., Spronk, C. A. E. M.,Bonvin, A. M. J. J. & Nilges, M. (2003). Refinement ofprotein structures in explicit solvent. Proteins: Struct.Funct. Genet. 50, 496–506.

62. Laskowski, R. A., Rullmann, J. A., MacArthur, M. W.,Kaptein, R. & Thornton, J. M. (1996). AQUA andPROCHECK-NMR: programs for checking the qual-ity of protein structures solved by NMR. J. Biomol.NMR, 8, 477–486.

63. Hooft, R. W. W., Vriend, G., Sander, C. & Abola, E. E.(1996). Errors in protein structures. Nature, 381, 272.

64. Carver, J. P. & Richards, R. E. (1972). A general two-site solution for the chemical exchange produceddependence of T2 upon the Carr-Purcell pulseseparation. J. Magn Reson. 6, 89–105.

65. Jen, J. (1978). Chemical exchange and NMR T2

relaxation—the multisite case. J. Magn. Reson. 30,111–128.

66. van Ingen, H., Vuister, G. W., Wijmenga, S. &Tessari, M. (2006). Characterizing the conformationof unobservable protein states. J. Am. Chem. Soc. In thepress.

Edited by A. G. Palmer III

(Received 16 November 2005; received in revised form 25 January 2006; accepted 31 January 2006)Available online 13 February 2006

S1

Supplementary materials

Role of Structural and Dynamical Plasticity in Sin3: the Free PAH2 Domain is a Folded Module in mSin3B

Hugo van Ingen, Maria A.H. Baltussen†, Jan Aelen & Geerten W. Vuister*

Department of Physical Chemistry/Biophysical Chemistry, Institute for Molecules and Materials, Radboud University Nijmegen, Toernooiveld 1, 6525 ED Nijmegen, the Netherlands.

† present address: RIKILT, Institute of Food Safety, Wageningen, The Netherlands.

* Corresponding author: phone: +31-24-365-2321; fax: +31-24-365-2112, e-mail: [email protected]

Figure S1. Per-residue U-factor and Ramachandran quality Z-scores. Secondary structure is

indicated by filled boxes (α-helix) at the top of the figure.

S2

Figure S2. 15N relaxation parameters of the free PAH2B domain recorded at 600 MHz proton

frequency and 293K. R1 (top left panel), R1ρ (top right panel), NOE (bottom left panel) and

the R1ρ/R1 ratio (bottom right panel).

S3

Table S1. Results from 15N-R2 relaxation dispersion data fittinga.

res. χ2 b Δω (.101 rad/s)c

error Δω (.101 rad/s)

kex (.101s-1)

error kex (.101s-1)

pb d

(%) error pb (%)

152 0.83 51 2 130 5 0.92 0.05 153 0.66 62 3 130 5 0.92 0.05 157 0.75 61 3 130 5 0.92 0.05 160 0.60 62 4 130 5 0.92 0.05 162 0.51 56 3 130 5 0.92 0.05 163 0.91 56 5 130 5 0.92 0.05 167 0.95 83 5 130 5 0.92 0.05 168 0.70 116 7 229 13 1.20 0.07 169 0.39 184 13 229 13 1.20 0.07 170 1.15 99 5 229 13 1.20 0.07 171 0.85 33 2 229 13 1.20 0.07 173 0.53 86 8 130 5 0.92 0.05 174 0.60 82 6 130 5 0.92 0.05 176 0.93 40 3 130 5 0.92 0.05 178 0.82 56 3 130 5 0.92 0.05 181 0.76 46 3 130 5 0.92 0.05 183 0.53 62 7 130 5 0.92 0.05 185 0.60 40 4 130 5 0.92 0.05 187 0.68 45 3 130 5 0.92 0.05 188 0.67 47 3 130 5 0.92 0.05 189 0.53 66 4 130 5 0.92 0.05 190 0.66 59 3 130 5 0.92 0.05 191 1.10 73 5 130 5 0.92 0.05 193 1.13 61 3 130 5 0.92 0.05 198 0.59 116 8 130 5 0.92 0.05 200 0.57 57 4 130 5 0.92 0.05 201 0.75 130 10 130 5 0.92 0.05 202 0.56 32 3 130 5 0.92 0.05 203 0.79 42 3 130 5 0.92 0.05 206 0.52 55 3 130 5 0.92 0.05 208 0.89 45 3 130 5 0.92 0.05 213 0.44 50 4 130 5 0.92 0.05 214 0.39 36 3 130 5 0.92 0.05 215 0.42 140 23 130 5 0.92 0.05 216 1.01 33 2 130 5 0.92 0.05 218 0.70 37 2 130 5 0.92 0.05 219 1.14 84 4 130 5 0.92 0.05 223 0.56 210 17 229 13 1.20 0.07 226 0.35 173 28 229 13 1.20 0.07 231 0.68 124 6 229 13 1.20 0.07 232 0.31 196 22 229 13 1.20 0.07 234 0.95 73 4 229 13 1.20 0.07 236 1.31 85 4 229 13 1.20 0.07 a Global fits of the 15N R2-relaxation dispersion profiles to the Carver-Richards-Jen all-

time scale approximation were performed as discussed in Materials and Methods. b The reduced χ2 for the global fit.

S4

c The magnitude of the chemical shift difference, expressed in rad/s at 500 MHz proton

frequency. d population of the minor conformation.

Table S2. Sequence identity (%) between the PAH domainsa,b.

PAH1 PAH2 PAH3 PAH4 PAH1 85 ± 9 /

52 ± 23 32 ± 1 22 ± 2 13 ± 2

PAH2 28 ± 8 78 ± 16 / 50 ± 22 19 ± 2 10 ± 2

PAH3 20 ± 6 20 ± 6 78 ± 16 / 34 ± 16 14 ± 2

PAH4 9 ± 3 9 ± 4 8 ± 3 67 ± 19 19 ± 16 /

a Top half compares the PAH domain in the vertebrates Homo Sapiens, Mus Musculus and

Xenopus Laevis, bottom half compares all PAH domains (accession numbers in Figure S2).

b Domain boundaries in Mm. mSin3B: 34-98 (PAH1), 152-228 (PAH2), 293-358 (PAH3),

705-789 (PAH4).

S5

Table S3. Conservation of important residues in the PAH domainsa.

residue roleb PAH1 PAH2 PAH3 PAH4

F154 BP v,t,c f t,l,y,(x) v,i,a,(x) A157 i12, BP A A e,d l,e,n I158 i12, BP L I x f,e Y160 i12/14, BP Y Y,F F x (a,e,c) V161 i12, BP L V F(L,C) n, l I164 i12/i14 V,L I,V V(A) I,W(T,N,M) K165 i12, BP K K K,R y,g F168 core F F L,I F,E,L I174 i23 V(K,I,T) I(T,V,K) V,A,T,I,D L,V Y175 i12,BP Y Y Y,F Y,C F178 i23/i12, BP F F F L(N) L179 i12,BP L L L L(X) I181 i23 I,V I C,L I,V(K,E) L182 i12,BP M L L,I k,y Y185 core, i23, BP F Y F,Y k,a,e,n Q186 BP K Q,R s,n(v) x E202 i34, BP t i,e r,k(g,t) d,x V205 i23/34, BP V V L L,A,M(E,D) F206 i34 I Y,F V,L,Q I,X V209 i23/34 V V V,A r,q,x A210 i34 S(K) T,A,S S(X) l,x L212 i23 L L F,Y,L d,e,(x) F213 core F F L,I M,L(X) L219 i14 L L L a,s,x L220 i34 I,L L F,M,L(W) e,s,x F223 core, i34, BP F F F Y(E) F226 core, BP F F F,L F,L,V(I,T) L227 i34, BP L L L,V,M L,V(Y,K,E) a Most frequent amino acid types are listed that occur at a certain position in the PAHi

sequences (i=1,2,3,4) (80% cut-off). Residues are listed in order of decreasing occurrence.

When some amino acids occur in the majority of sequences, the ones occurring in the

minority are placed between round brackets. X denotes any amino acid. Similarities with

the Mm. PAH2B sequence are shown in uppercase, dissimilarities in italic lowercase.

Strong similarities are shown in bold-type, while strong dissimilarities are shown with

underline. b BP = part of binding pocket; i12, i23, i34, i14 = part of interface between helix α1 and

α2, α2 and α3, etc.; core = part of protein core.

S9

Figure S3. Multiple sequence alignment generated by ClustalW of all PAH domains from 35

proteins. Homologues were identified using the Pfam database of protein families (Bateman

et al., 2004). Redundant sequences were removed. The alignment was made in three steps: i)

excluding the most divergent sequences, a multiple alignment was made for every PAH

family (PAH1, PAH2, etc.); ii) the four family alignments were added together stepwise by a

profile alignment; iii) divergent sequences were added to the resulting alignment by a

sequence alignment which was further optimized manually. Sequences are labeled by their

SwissProt accession number, an identifier for the PAH domain and the organism

abbreviation. Top row displays the sequence used in the present study in which the heliccal

residues are shown in bold type, the structural role of residue (c=buried residues in the core of

the protein, i=residues in the interface between two helices) and the involvement in the

binding pocket (h = only intermolecular contacts in the HBP1-complex, m = only

intermolecular contacts in the Mad1-complex, + = intermolecular contacts in both

complexes).

Supplementary References

Bateman, A., Coin, L., Durbin, R., Finn, R.D., Hollich, V., Griffiths-Jones, S., Khanna, A.,

Marshall, M., Moxon, S., Sonnhammer, E.L., Studholme, D.J., Yeats, C. and Eddy, S.R.

(2004). The Pfam protein families database. Nucl. Acids Res. , 32, D138-41.

Role of Structural and Dynamical Plasticity in Sin3: The...

Documents

Transcript of Role of Structural and Dynamical Plasticity in Sin3: The...