Structure of the 30-kDa Sin3-associated Protein (SAP30) in Complex with the Mammalian Sin3A Corepressor and Its Role in Nucleic Acid Binding*

The ∼2-megadalton evolutionarily conserved histone deacetylase-associated Rpd3L/Sin3L complex plays critical roles in altering the histone code and repressing transcription of a broad range of genes involved in many aspects of cellular physiology. Targeting of this complex to specific regions of the genome is presumed to rely on interactions involving one or more of at least 10 distinct subunits in the complex. Here we describe the solution structure of the complex formed by the interacting domains of two constitutively associated subunits, mSin3A and SAP30. The mSin3A paired amphipathic helix 3 (PAH3) domain in the complex adopts the left-handed four-helix bundle structure characteristic of PAH domains. The SAP30 Sin3 interaction domain (SID) binds to PAH3 via a tripartite structural motif, including a C-terminal helix that targets the canonical PAH hydrophobic cleft while two other helices and an N-terminal extension target a discrete surface formed largely by the PAH3 α2, α3, and α3′ helices. The protein-protein interface is extensive (∼1400 Å2), accounting for the high affinity of the interaction and the constitutive association of the SAP30 subunit with the Rpd3L/Sin3L complex. We further show using NMR that the mSin3A PAH3-SAP30 SID complex can bind to nucleic acids, hinting at a role for a nucleolar localization sequence in the SID αA helix in targeting the Rpd3L/Sin3L complex for silencing ribosomal RNA genes.

by sap30 genetic knockouts in yeast that yield similar effects as those involving sin3 and rpd3 (20, 29 -31). SAP30 also appears to play a role in complex assembly because the protein has been proposed to interact with the ING1b/ING2 and SAP180/RBP1 subunits of the complex (26,(32)(33)(34).
At the molecular level, SAP30 harbors a poorly conserved N-terminal region of low sequence complexity, a highly conserved central region with a zinc finger motif found in organisms ranging from flies to humans, and a C-terminal Sin3 interaction domain (SID) that is well conserved from yeast to humans. Belying its small size, SAP30 appears to be targeted by a remarkably large and diverse array of cellular transcription factors and viral proteins, including CIR (CBF1-interacting factor of the Notch pathway (35)), yin yang 1 (YY1 (36)), human papillomavirus-binding factor (37), non-structural NSs protein of the Rift Valley fever virus (38), and the HTRP protein of the herpes simplex virus type 1 (HSV-1 (39)). Aside from the novel zinc finger motif that has been implicated in nonspecific interactions with nucleic acids (40,41), little is known structurally regarding how SAP30 interacted with its targets.
Here, we structurally and functionally characterize the interaction between the SAP30 SID and the mSin3A PAH3 domains. Structures of Sin3 PAH1 and PAH2 domains in complex with diverse targets have been described, providing insights into how these four-helix bundles bind to relatively short sequence motifs embedded within isolated helical segments of the SIDs (42)(43)(44)(45)(46)(47)(48). The Sin3 PAH3 domains share relatively low levels of sequence identity with the PAH1 and PAH2 domains (ϳ25 and ϳ16%, respectively). Here we show that the mSin3A PAH3-SAP30 SID complex shares several underlying themes with previously characterized PAH-SID complexes, but there are additional structural and functional elaborations on these themes that could not have been predicted solely from sequence analysis.

EXPERIMENTAL PROCEDURES
Production of SAP30 SID and mSin3A PAH3-The coding sequences of mouse SAP30 SID (residues 130 -220) and mammalian Sin3A PAH3 (residues 457-528) were amplified by PCR and inserted into the pMCSG23 and pMCSG7 expression vectors, for expression as His 6 -maltose-binding protein (MBP) and His 6 -tagged fusion proteins, respectively. The cloned gene segments were confirmed by DNA sequencing. Escherichia coli BL21(DE3) cells (Novagen) co-transformed with both plasmids were grown at 37°C in LB broth (EMD Chemicals). The growth temperature was shifted to 20°C when the A 600 nm reached about 0.6 -0.8. Protein expression was induced using 1 mM isopropyl ␤-D-thiogalactopyranoside, and the cells were harvested 16 -18 h thereafter. Cell pellets were suspended in 50 mM Tris-HCl buffer (pH 8.0) containing 0.15 M NaCl, 2 mM tris(2-carboxyethyl)phosphine hydrochloride, 1 mM phenylmethylsulfonyl (PMSF), 1 M leupeptin, 1 mM pepstatin, and 0.1% Triton X-100 and lysed via sonication followed by DNase I treatment for 15 min at room temperature and centrifugation at 12,000 rpm for 30 min. The supernatant was incubated with His-Select Ni 2ϩ -resin (Sigma-Aldrich) for 40 min, followed by extensive washing with 50 mM Tris buffer (pH 8) containing 0.15 M NaCl and 2 mM tris(2-carboxyethyl)phosphine hydro-chloride and subsequent incubation with tobacco etch virus protease at room temperature for 4 h and then at 4°C overnight. The cleaved proteins were collected and purified to homogeneity via reversed-phase HPLC using a C18 or C8 column (Grace Vydac) and a linear gradient of 0.1% trifluoroacetic acid (TFA) and 0.1% TFA in 80% acetonitrile. The mSin3A PAH3 and SAP30 SID polypeptides eluted with distinct retention times, and fractions containing each protein were pooled and lyophilized. Uniformly 15 N-and/or 13 C-labeled proteins were produced using the same procedure except that the cells were grown in M9 minimal medium containing 15 N-labeled ammonium sulfate and/or 13 C-labeled D-glucose (Cambridge Isotopes), respectively. The identity and integrity of the proteins and the extent of isotope enrichment were established by electrospray ionization mass spectrometry and SDS-PAGE.
SAP30 SID mutants V148E, F186E, F200E, V148E/F186E, V148E/F200E, F186E/F200E, and V148E/F186E/F200E for in vitro binding assays were generated using the QuikChange sitedirected mutagenesis protocol (Agilent). All mutations were confirmed by DNA sequencing. Wild-type and mutant SAP30 SID proteins for these assays were expressed as His 6 -MBP fusion proteins and purified in an analogous manner to when they were co-expressed with mSin3A PAH3, except the proteins were neither cleaved by tobacco etch virus protease nor subjected to HPLC.
Surface Plasmon Resonance-Surface plasmon resonance (SPR) experiments were performed on a BIAcore 3000 instrument (Biacore) at 25°C in HBS-P buffer (0.01 M HEPES (pH 7.4), 0.15 M NaCl, 0.005% (v/v) surfactant P20) at a flow rate of 25 l/min. The mSin3A PAH3 polypeptide was immobilized on a CM5 sensor chip, whereas the wild-type and mutant forms of the SAP30 SID were injected at different concentrations. Protein-protein association was monitored for 200 s, whereas the protein-protein dissociation was monitored for 450 and 250 s for the wild-type and the mutant proteins, respectively. The chip was regenerated using 0.01 N sodium hydroxide (NaOH) for 10 s at a flow rate of 100 l/min between runs.
SAP30 SID-mSin3A PAH3 Complex Generation and NMR Sample Preparation-The complex was generated by dissolving 15 N, 13 C-labeled SAP30 SID or mSin3A PAH3A and an equivalent amount of the unlabeled mSin3A PAH3A or SAP30 SID in 20 mM sodium phosphate buffer (pH 6.0) containing 2 mM DTT and 8 M urea, followed by dialysis against the same buffer but without any urea using a Spectrapor DispoDialyzer dialysis unit with a 1000 Da cut-off (Spectrum Laboratories). The sample was exchanged into NMR buffer containing 20 mM sodium phosphate (pH 6.5), 2 mM DTT-d 10 , and 0.2% (w/v) NaN 3 using a YM-3 Centricon unit with a 3000 Da cut-off (Millipore). Protein concentrations were determined spectrophotometrically. Samples of the protein-protein complex were lyophilized and redissolved in 99.996% D 2 O for experiments performed in D 2 O (Sigma-Aldrich).
NMR Titrations-A 25-mer single-stranded DNA oligonucleotide with the sequence 5Ј-GCGTTACGGCTTTTTGCCG-TAACGC-3Ј was purchased in lyophilized form from Integrated DNA Technologies (Coralville, IA). The DNA was used for the titrations without any additional purification. The DNA was dissolved in 20 mM sodium phosphate buffer (pH 6.5) con-taining 2 mM dithiothreitol-d 10 , 10% D 2 O, and 0.2% (w/v) NaN 3 . The DNA duplex in hairpin conformation was obtained by heating the sample to 95°C for 15 min and rapidly cooling it on ice. Hairpin formation was confirmed by one-dimensional 1 H NMR; the spectrum revealed no evidence of sample heterogeneity, implying a reasonably pure sample preparation. The complex of 15 N-labeled SAP30 SID and unlabeled mSin3A PAH3 was generated as described above; complex formation was confirmed by recording a two-dimensional 1 H-15 N HSQC spectrum. The protein-DNA complex was generated by slowly titrating the 15 N-labeled SAP30 SID-mSin3A PAH3 complex with the hairpin duplex. Titrations were performed at 0.05 mM protein concentration and terminated when five equivalents of DNA had been added.
Transcription Repression Assays-SAP30 cDNA encoding various mutations was cloned into the pCMV-BD plasmid (Agilent) to create GAL4-DBD fusions. Luciferase reporter under the control of the 14D promoter was a kind gift from D. Ayer (University of Utah). LacZ in pcDNA 3.1 plasmid (Invitrogen) was used as a transfection control in repression experiments. Repression assays were performed as described previously by Viiri et al. (49). Two independent experiments with three replicates were performed showing comparable results. Human embryonal kidney epithelial cells (HEK 293T) were cultured in DMEM (Invitrogen) supplemented with penicillin/ streptomycin antibiotics, 5% fetal bovine serum, 1 mM sodium pyruvate, and 50 g/ml uridine. Cells were plated to a density of 20,000 cells/cm 2 24 h prior to transfection. HEK293T cells were transfected with 250 ng of pCMV-BD, 200 ng of the 14D luciferase reporter, and 50 ng of pcDNA-LacZ using Fugene 6 or Fugene HD reagents (Roche Applied Science) as instructed by the manufacturer.
NMR Spectroscopy and Structure Determination-All NMR data were acquired on a Varian Inova 600-MHz spectrometer equipped with a pulsed field gradient triple resonance cold probe at 35°C. NMR data processing and analysis were performed using Felix 98.0 (Accelrys) and Sparky. Sample concentrations used for the structure determination were ϳ1 mM. Backbone and side chain 1 H, 15 N, and 13 C resonance assignments for each of the interacting proteins in the SAP30 SID-mSin3A PAH3 complex were obtained by analyzing three-dimensional HNCACB, CBCA(CO)NH, HNCA, HN(CO)CA, C(CO)NH-TOCSY, HNCO, HCACO, HCCH-COSY, and HCCH-TOCSY spectra (50,51). Aromatic resonances were assigned based on a careful analysis of two-dimensional 1 H-13 C aromatic HSQC, three-dimensional 13 C-edited NOESY, and 15 N, 13 C-double half-filtered NOESY spectra (52).
Structures were determined using ARIA 1.2 in conjunction with CNS 1.1 starting from an initial structure with extended backbone conformations (55)(56)(57). All NOEs were calibrated automatically and assigned iteratively by ARIA; the assignments were checked manually for errors after every run. Eighty conformers were calculated in the final iteration, of which the 40 conformers with the lowest restraint energies were refined in a shell of explicit water, and the 20 conformers with the lowest restraint energies, restraint violations, and r.m.s. deviations from the ideal covalent geometry were selected for further analysis. The final conformers were analyzed using CNS (57), PROCHECK (58), MONSTER (59), DALI (60), and DeepView (61). Molecular images were generated using CHIMERA (62).

RESULTS
Only the mSin3A PAH3-SAP30 SID Complex, Not the Isolated Domains, Is Amenable to Structural Studies-Previous studies localized the SAP30-mSin3A interaction to a 91-residue SID at the C terminus of SAP30 and the 74-residue PAH3 domain of mSin3A. Although SAP30 SID could be readily expressed in E. coli, the mSin3A PAH3 domain could not be stably expressed in this host unless when co-expressed with the SAP30 SID. Co-purification and subsequent dissociation of the complex was straightforward, but the individual polypeptides either formed soluble aggregates (SAP30 SID) or were not sufficiently soluble (mSin3A PAH3) at near physiological pH, precluding NMR studies. On the other hand, the mSin3A PAH3-SAP30 SID complex was readily soluble and yielded high quality NMR spectra for the individual proteins to facilitate resonance assignments using standard double and triple resonance approaches and structure determination using traditional approaches.
Both mSin3A PAH3 and SAP30 SID Exhibit Long Range Interactions Characteristic of Globular Domains in the Complex-The solution NMR structure of the mSin3A PAH3-SAP30 SID complex was determined using 4079 NOE-based distance restraints, including 617 intermolecular distance restraints, and 168 chemical shift-based torsion angle restraints ( Table 1). The final ensemble of 20 conformers exhibited reasonable precision and good covalent geometry and was in good agreement with NMR data, including no distance violations of Ͼ0.5 Å or torsion angle violations of Ͼ5°( Fig. 1A and Table 1). The vast majority of the residues of both mSin3A PAH3 and SAP30 SID adopt well defined conformations, although the four N-terminal and two C-terminal residues of mSin3A PAH3 as well as the 13 N-terminal and 11 C-terminal residues of SAP30 SID are disordered in the complex.
The core SAP30 SID comprises three helices, with the ␣A, ␣B, and ␣C helices spanning residues Val 155 -His 164 , Lys 175 -Phe 186 , and Glu 193 -Asn 206 , respectively (Fig. 1B). The ␣A and ␣B helices pack against each other and also partially against a segment that is largely in an extended conformation N-terminal to the ␣A helix forming a globular subdomain ( Fig. 2A). In contrast, the ␣C helix makes no contacts with the rest of the SID.
SAP30 SID Binds via a Tripartite Motif to Two Discrete Surfaces of the mSin3A PAH3 Domain-The mSin3A PAH3 domain features a prominent hydrophobic cleft analogous to those observed previously for the Sin3 PAH1 and PAH2 domains ( Fig. 2A (42-48)). A secondary surface, also largely hydrophobic in character, defined by residues in the ␣2, ␣3, and ␣3Ј helices as well as in the loops preceding, linking, or following these segments, is unique to the PAH3 domain. The SAP30 SID engages these two discrete surfaces of the PAH3 domain via a tripartite structural motif comprising the N-terminal extension, the globular subdomain, and the ␣C helix. The latter targets the cleft, whereas the N-terminal extension and the globular subdomain bind to the secondary surface ( Fig. 2A). The linker segment connecting the globular subdomain with the ␣C helix wraps intimately around a hydrophobic segment of the ␣2 helix. The protein-protein interface in the mSin3A PAH3-SAP30 SID complex is extensive, with the solvent-excluded surface area averaging ϳ1440 Ϯ 70 Å 2 in each of the 20 conformers of the NMR ensemble.
Non-covalent Contacts across the mSin3A PAH3-SAP30 SID Interface Are Dominated by Hydrophobic Interactions-The mSin3A PAH3-SAP30 SID complex appears to be stabilized largely by hydrophobic interactions, with Ile 143 , Thr 145 , Pro 146 , Val 148 , Leu 150 , and Leu 153 in the N-terminal segment; Thr 157 , Tyr 161 , and Phe 165 in the ␣A helix; and Val 182 and Phe 186 in the ␣B helix of SID engaging the secondary surface of PAH3 defined by Leu 475 , Ala 480 , Cys 487 , Ile 490 , Val 495 , Ile 496 , Leu 504 , Pro 507 , Phe 508 , Arg 476 , and Phe 512 , with the aliphatic portions of the side chains of Glu 479 , Glu 482 , Asn 483 , Arg 486 , Gln 493 , Glu 494 , Glu 500 , Gln 503 , and Lys 511 located in the periphery bolstering the hydrophobic character of the surface (Fig. 2, A and B) (Fig. 2, A and B).   form salt bridges with Arg 472 and Glu 464 , respectively (Fig. 2, A  and B). Furthermore, the side chains of Thr 145 , Gln 152 , Gln 154 , and Thr 157 of SID engage in hydrogen bonding interactions with those of Ser 477 , Gln 503 , Glu 500 , and Glu 500 of PAH3, respectively. Finally, at least two instances of positively charged/amino-aromatic interactions involving Gln 493 and Arg 486 of PAH3 with Phe 165 and Phe 186 of SID, respectively, were found that could lend additional stability to the complex. Residue-specific sequence conservation patterns both within and across species for Sin3 PAH3 and SAP30 SID reveal a strong correlation between conservation of a particular residue and its involvement in stabilizing either the structure of the corresponding domain or that of the PAH3-SID complex (Fig.  2B). Rather surprisingly, the lowest level of sequence conservation is seen for the N-terminal extension of SAP30 SID. However, because this segment adopts a largely extended conformation ( Fig. 2A), it is likely to be more malleable to sequence changes than if it were in a regular secondary structure. Predictably, the yeast proteins show the highest level of sequence divergence, although the plant proteins also exhibit high levels of divergence relative to the corresponding proteins from flies to humans, implying that the structures of the Sin3 PAH3-SAP30 SID complex in yeast and plants are likely to show some significant differences (Fig. 2B).
The mSin3A PAH3-SAP30 SID Interaction Is of High Affinity but Can Be Destabilized by Structure-guided Mutagenesis-To evaluate the role of various polypeptide segments in contributing to the stability of the mSin3A PAH3-SAP30 SID complex, we measured the binding affinities of SAP30 SID mutants in in vitro binding assays and compared the results with functional assays conducted in cells. SPR analysis of the mSin3A PAH3-SAP30 SID interaction using the wild-type proteins yielded an equilibrium dissociation constant of ϳ9 nM, consistent with the large size of the protein-protein interface.
Given the extensive protein-protein interface and the multitude of interactions stabilizing the complex, we mutated three key residues of SAP30 SID, including Val 148 , Phe 186 , and Phe 200 , either singly or in combination to glutamic acid because all three residues are engaged predominantly in hydrophobic interactions, and replacement with a charged residue is predicted to disrupt these interactions (mSin3A PAH3 mutants could not be studied due to difficulties with protein expression as discussed above). The choice of these three specific sites was guided by our desire to evaluate the contributions of each of the three SAP30 SID structural motifs involved in recognizing mSin3A PAH3. Val 148 is located in the N-terminal extension, whereas Phe 186 and Phe 200 are located in the globular subdomain and the ␣C helix, respectively ( Fig. 2A). Furthermore, all three residues are either invariant or conserved in orthologous proteins ( Fig. 2B; Val 148 is replaced by aromatic residues in some orthologs).
The in vitro mSin3A PAH3 binding activity of each of the SAP30 SID mutants was analyzed by SPR measurements. Association rates as well as complex dissociation rates were measured at multiple concentrations; the equilibrium dissociation constants were inferred from global fits of these rates. The V148E mutant showed a ϳ50-fold reduction in binding affinity relative to the wild-type protein, whereas the F186E and F200E mutants exhibited ϳ100and 200-fold reductions in affinity, respectively ( Table 2 and supplemental Fig. S1). Each of the three structural motifs involved in mSin3A PAH3 binding thus appears to make significant contributions to the overall affinity. Consistent with this argument, PAH3 binding activity was severely diminished for the double and triple mutants, with the upper limit for the binding affinity estimated to be in the 50 M range.

SAP30 Functional Assays in Cells Qualitatively Mimic the Trends Noted in in Vitro
Binding Assays-The ability of SAP30 to effect repression of a reporter gene when fused to a heterologous DNA-binding domain has been described previously (17). This repression was shown to be dependent on the interaction with mSin3A because deletion of the C-terminal 91 residues of SAP30 (harboring the SID) abolished both mSin3A binding and the repression activity. Therefore, we analyzed the effect of SAP30 mutations on its transrepression activity by fusing the protein with the Gal4 DNA-binding domain and using a luciferase reporter assay in HEK293 cells. Human as well as mouse full-length SAP30 proteins and a mouse SAP30 mutant lacking the C-terminal 91 residues were used as positive and negative controls, respectively. As expected, both wild-type FIGURE 2. Non-covalent interactions at the mSin3A PAH3-SAP30 SID interface and sequence conservation in the respective domains. A, three views of the protein-protein interface with the PAH3 domain rendered as a molecular surface and the SAP30 SID depicted in a worm-and-sticks representation. Only those SAP30 side chains involved in intermolecular interactions are shown, whereas mSin3A residues engaging in these interactions are colored in gold. Nitrogen, oxygen, and sulfur atoms of interacting residues in both proteins are colored blue, red, and yellow, respectively. B, ClustalW-guided multiple sequence alignments of SAP30 SID orthologs (top), the corresponding mSin3A PAH3 orthologs (middle), and the paralogous Sin3 PAH domains (bottom). Invariant and conserved residues are highlighted in green and purple, respectively. The filled circles in green and magenta denote residues engaging in intermolecular electrostatic/hydrogen bonding and hydrophobic interactions, respectively. Key PAH3 residues that lend hydrophobic character to the secondary surface, which is absent in the PAH1 and PAH2 domains, are denoted by asterisks (in orange) in the bottom panel. The presumed specificity-determining residues in the hydrophobic cleft are denoted by asterisks (in cyan). The red vertical lines represent sites of amino acid insertions in the respective PAH domains. human and mouse SAP30 potently repressed transcription of the reporter, whereas the C-terminally deleted mutant failed to effect repression (Fig. 3). All three single mutants, including V148E, F186E, and F200E, exhibited diminished repression activity, with ϳ4-fold reduction in activity compared with wildtype protein. The double and triple mutants were additionally compromised in their transrepression ability (Fig. 3), although the repression in these cases was not synergistic, which might be a consequence of the limited dynamic range of this assay. Collectively, however, the results from the repression assays implicate all three residues, including Val 148 , Phe 186 , and Phe 200 , in critical roles in Sin3 recruitment with attendant effects on transrepression activity. SAP30 SID Shares Structural Homology with the SAP Motif, a DNA-binding Motif-A DALI-based search for potential SAP30 SID structural homologues revealed that the domain shared homology with the SAP (SAF, acinus, PIAS) motif, also known as the SAF-box, which has been implicated in binding nucleic acids (63,64). A backbone C ␣ superposition of the SAP domain of E1B-55 kDa-associated protein 5 (E1B-AP5) with the SAP30 SID yields a r.m.s. deviation of 0.87 Å for the segment spanning Asp 149 -Lys 187 of SAP30, encompassing the ␣A and ␣B helices (Fig. 4). Although the SAP30 sequence by and large follows the consensus for the SAP motif (Fig. 4A), their relationship at the sequence level could not be established via Pfam and SMART data base searches, in part because of the shortness of the motif and the default settings for E-value cut-offs. We note that the character of the molecular surfaces of the SAP30 SID and E1B-AP5 domains are quite distinct (Fig. 4, B and C), consistent with the two domains sharing only 32% sequence identity and 45% sequence similarity over a 38-residue segment.
The Nucleolar Localization Sequence of SAP30 SID Can Mediate Nucleic Acid Binding-SAP30 and SAP30L (SAP30like protein) have previously been shown to localize to the nucleolus, thereby playing a vital role in targeting the Sin3L complex to this organelle. Nucleolar targeting relies on an intact nucleolar localization signal (NoLS) in both SAP30 and SAP30L that maps to the ␣A helix and a portion of the succeeding loop in the three-dimensional structure (Fig. 4A). A striking feature of this sequence is its basic character, which is propagated into the molecular surface (Fig. 4, A and C).
The molecular target of the NoLS is presently unknown. Given the function of the NoLS, the basic character of this sequence, and the structural relationship of the SAP30 SID with the nucleic acid-binding SAP motif, we asked whether SAP30 SID could interact with nucleic acids. To this end, we designed a DNA hairpin duplex as a potential mimic of the natural target of the NoLS and performed NMR titrations with 15 N-labeled SAP30 SID prebound to mSin3A PAH3 (an RNA stem-loop structure has been implicated in targeting the nucleolar remodeling complex to the nucleolus (65)). The hairpin duplex induced strong but selective perturbations characterized by shifting and broadening of resonances in the NMR spectra of SAP30 SID (Fig. 5), implicating a role for this domain in nucleic acid binding. Some of the strongest perturbations are observed for the NoLS residues, including Arg 159 , His 164 , and Lys 166 , but similar perturbations are also observed for those in the vicinity at the level of both primary and tertiary structures, including Gln 154 , Val 155 , Val 179 , Gly 183 , and the side chains of Asn 174 and Gln 177 . In contrast, perturbations are minimal for residues in other regions of the SID, including Asp 142 , Val 148 , Thr 196 , and Thr 198 . These results thus suggest that the NoLS residues may have a direct role in the recognition of a putative nucleic acid target.

DISCUSSION
Unlike the Sin3 PAH1 and PAH2 domains that are broadly available for direct interactions with sequence-specific DNAbinding transcription factors for targeting of the Rpd3L/Sin3L complex to specific gene promoters, the PAH3 domain plays an important role in Rpd3L/Sin3L complex assembly by constitutively associating with the SAP30 subunit. Notwithstanding the extensive structural studies of the Sin3 PAH1 and PAH2 domains in complex with diverse targets (42)(43)(44)(45)(46)(47)(48), the remote similarity between these domains and PAH3 precluded reliable predictions of how this domain might interact with SAP30.
The structure of the mSin3A PAH3-SAP30 SID complex reveals significant variations in the themes previously uncovered for PAH domains. For example, the ␣1 helix is shorter at the N terminus by approximately one to two turns compared with the PAH1 and PAH2 domains, whereas the ␣3 helix is kinked, formally dividing the helix into two (␣3 and ␣3Ј). The former along with the drastically bulkier amino acids in the cleft, including Glu 464 , Phe 468 , and Leu 501 , in PAH3 has the effect of dramatically altering the character of the cleft, making it considerably less cavernous than in the other PAH domains (Fig. 2,  A and B). This probably explains the contrasting sequence motif (i.e. (D/E)XXXXXX(R/K), where represents a hydrophobic residue) preferred by the PAH3 domain and also why the SID mutations that preserve an intact ␣C helix are still unable to bind PAH3 efficiently (i.e. the single helical motif that was sufficient for high affinity interactions with PAH1 and  PAH2 is necessary but not sufficient for high affinity interactions with PAH3).
Perhaps the most striking difference between the PAH3 and the PAH1/2 domains is the involvement of a secondary surface for SID binding. The kinking of the ␣3 helix and the particular amino acid composition in the vicinity has the effect of creating a narrow hydrophobic groove that is absent in the other PAH domains (Fig. 2, A and B). Additional hydrophobic residues in the ␣2 and ␣3 helices, either absent or not in the same structural context in the PAH1/2 domains (Fig. 2B), define a shallow surface that is recognized by the bipartite structural motif of the SIDs that is unique to this complex.
The mSin3A PAH3-SAP30 SID is the highest affinity PAH-SID interaction that has been structurally and functionally characterized thus far. The high affinity of the interaction, which results from the cooperative recognition of the two discrete mSin3A PAH3 surfaces by the SAP30 SID tripartite motif, provides the basis for the constitutive association of this subunit with Sin3 and by extension the Rpd3L/Sin3L complex.
Intriguingly and belying its simple fold, the globular subdomain of SAP30 SID shows striking structural homology with the SAP motif. The SAP motif was originally discovered in the scaffold attachment factor SAF-A (66), which is involved in attaching chromosomal DNA to the nuclear matrix, but has since been predicted to occur in a number of other proteins, including those involved in DNA repair, RNA splicing, and apoptotic degradation of chromatin (63,64). Structurally, the SAP motifs, like the globular subdomain of SAP30 SID, harbor two helices separated by a short loop, although isolated SAP motifs as well as those that are part of larger domains comprising four or five helices have been described (67,68). Although structures of SAP motifs bound to nucleic acids have not yet been described, the N-terminal portions of the two SAP helices have been suggested to bind nucleic acids by NMR titration experiments (67,68). More rigorous NMR analyses are required to map the precise location of the nucleic acid-binding site on the SAP30 SID when its cognate target is identified. However, the pattern of chemical shift perturbations induced by the surrogate molecule employed in this study suggests that the binding site might not be restricted only to the N termini of the ␣A and ␣B helices seen for the SAP motifs.
Given their structural and, to some extent, functional similarities, it would thus appear that the SAP30 SID and the SAP motifs might have shared a common evolutionary history. This is indeed confirmed by a recent phylogenetic analysis of the SAP30 family that also deduced the loss of one of its original functions involving association with the nuclear matrix (69). Our results would suggest that the latter function is somewhat modified to recognize nucleic acids perhaps of a different type and through a different mechanism.
The recent discovery that the SAP30 proteins are involved in the nucleolar targeting of mSin3A (and by extension the Sin3L complex) is exciting (49) because mSin3A, RbAp46, and HDAC1, all members of the mammalian Sin3L complex, have been shown biochemically to be recruited by the nucleolar remodeling complex for the repression of ribosomal gene transcription (70). Furthermore, the repression was dependent on HDAC activity because HDAC inhibitors relieved the repression. The involvement of the mammalian Sin3L complex in the repression of ribosomal gene transcription contrasts with the role of the Rpd3L/Sin3L complex in yeast, where genetic deletions of members of this complex have been implicated in enhanced ribosomal silencing (30). Interestingly, the NoLS is present only in higher organisms from flies to humans in animals (Figs. 2B and 4A), suggesting that this might have been a recently acquired function. Rather strikingly, the SAP30 proteins in the same organisms harbor a novel zinc finger motif N-terminal to the SID; the motif is notably absent in yeast and plants (40,41). The discovery that the SAP30 zinc finger can also bind to nucleic acids raises the intriguing possibility that the SID and the zinc finger motif might cooperate in the recognition of the same target that promotes their nucleolar localization (40,41). Additional studies are required to test this exciting possibility.