A Coil-to-Helix Transition Serves as a Binding Motif for hSNF5 and BAF155 Interaction.

Human SNF5 and BAF155 constitute the core subunit of multi-protein SWI/SNF chromatin-remodeling complexes that are required for ATP-dependent nucleosome mobility and transcriptional control. Human SNF5 (hSNF5) utilizes its repeat 1 (RPT1) domain to associate with the SWIRM domain of BAF155. Here, we employed X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, and various biophysical methods in order to investigate the detailed binding mechanism between hSNF5 and BAF155. Multi-angle light scattering data clearly indicate that hSNF5171–258 and BAF155SWIRM are both monomeric in solution and they form a heterodimer. NMR data and crystal structure of the hSNF5171–258/BAF155SWIRM complex further reveal a unique binding interface, which involves a coil-to-helix transition upon protein binding. The newly formed αN helix of hSNF5171–258 interacts with the β2–α1 loop of hSNF5 via hydrogen bonds and it also displays a hydrophobic interaction with BAF155SWIRM. Therefore, the N-terminal region of hSNF5171–258 plays an important role in tumorigenesis and our data will provide a structural clue for the pathogenesis of Rhabdoid tumors and malignant melanomas that originate from mutations in the N-terminal loop region of hSNF5.

Human SNF5 (hSNF5) regulates the cell cycle at various stages of the mitotic checkpoint and it interacts with a variety of proteins, such as HIV-1 integrase, c-Myc, p53, Epstein-Barr virus nuclear protein 2, and human papillomavirus E1 protein [19][20][21][22]. The interaction with HIV-1 integrase affects viral integration and infectivity [23,24]. hSNF5 is involved in tumor proliferation and progression in he p16-RB pathway, WNT signaling pathway, sonic hedgehog signaling pathway and Polycomb pathway [25]. Additionally, truncating mutations of hSNF5 gene lead to aggressive pediatric atypical teratoid and malignant rhabdoid tumors [25,26]. BAF155 (a human homologue of yeast SWI3) is another core component of the SWI/SNF complex that displays ATPase activity [27][28][29][30]. This protein protects the SWI/SNF complex from proteasomal degradation and it also directs the nuclear localization of the complex [29,31,32]. BAF155 is ubiquitously expressed, being similar to other members of the SWI/SNF complex [33].
Recently, it has been reported that hSNF5 and BAF155 interact via their RPT1 and SWIRM domains, respectively [32][33][34][35]. Here, we investigated the detailed interaction between an Nand C-terminal elongated hSNF5 RPT1 (hSNF5 171-258 ) domain and the BAF155 SWIRM domain while using solution nuclear magnetic resonance (NMR) spectroscopy and X-ray crystallography. Notably, we detected a discrepancy between the binding interface that was identified by NMR-titration experiments and the previous crystal structure (Protein Data Bank (PDB) code 5GJK) [34]. In addition, we found that the crystal structure of the hSNF5 171-258 /BAF155 SWIRM complex is dramatically different from that reported in a previous study [34]. Our structure reveals that hSNF5 171-258 employs its N-terminal disordered region as a binding motif for the BAF155 SWIRM domain via a coil-to-helix transition. The truncation of the BAF155-binding region from hSNF5 results in a seven-fold decrease in binding affinity, demonstrating the importance of this conformational switch for the hSNF5-BAF155 interaction. Our structure reveals the detailed binding interface between hSNF5 and BAF155 and it highlights a novel folding-upon-binding mechanism in the assembly of this chromatin-remodeling complex.
A previous report describing the crystal structure of the hSNF5 169-252 /BAF155 SWIRM complex found that these proteins form a heterotetramer [34], whereas the size exclusion chromatography (SEC) and multi-angle light scattering (MALS) data have indicated that hSNF5 184-249 is a monomer [35]. We performed SEC and MALS experiments with different constructs of hSNF5 in complex with BAF155 SWIRM to investigate this discrepancy. The calculated molecular weights of hSNF5 171-258 and BAF155 SWIRM were 10.0 kDa and 12.1 kDa, respectively. Using SEC, we found that hSNF5 171-258 elutes at a higher molecular weight than the monomer fraction, whereas BAF155 SWIRM elutes in the monomer fraction (Figure 2A,B). However, MALS data clearly indicate that both hSNF5 171-258 and BAF155 SWIRM are monomeric in solution, with absolute molar masses of 10.0 ± 0.9 kDa and 12.6 ± 1.4 kDa, respectively ( Figure 2C,D). The SEC elution profile of hSNF5 171-258 reflects the effect of a long flexible N-terminal loop, as shown in our NMR structure ( Figure 2E,F, Table 1). The N-terminal region of hSNF5 171-258 from His171 to Val185 lacks medium-and long-range nuclear Overhauser effects (NOEs) and it remains unstructured in the solution structure of free hSNF5 171-258 . Except for the N-terminal loop regions, our solution structure of hSNF5 171-258 is similar to the previously reported hSNF5 RPT1 structures covering 184-249 residues (PDB code 5L7A) [35], forming a β1β2α1α2 fold, with a Cα RMSD of 2.0 Å for the 64 atoms between Glu184 and Tyr248 ( Figure 2E,F). The hSNF5 171-258 /BAF155 SWIRM complex also eluted at an even higher molecular weight than the heterodimer from SEC ( Figure 2G), but the absolute molar mass of the hSNF5 171-258 /BAF155 SWIRM complex was found to be 22.5 ± 1.6 kDa from MALS data, which confirmed that hSNF5 171-258 and BAF155 SWIRM form a heterodimer (calculated m.w. 22,067 Da) ( Figure 2H). Therefore, we clearly defined that hSNF5 171-258 is a monomer and it binds to BAF155 SWIRM in a molar ratio of 1:1, even though their complex eluted in higher molecular weight in the SEC.    Figure 3A,B). The heteronuclear single quantum coherence (HSQC) spectra superimposed between the free and complex states of both hSNF5 171-258 and BAF155 SWIRM show that a number of backbone amide resonances are shifted upon interaction. We exploited the backbone chemical shifts of free and complexed proteins in order to obtain CSP diagrams for both hSNF5 171-258 and BAF155 SWIRM upon complex formation ( Figure 3C,D). The free form of hSNF5 171-258 has two β-strands, followed by two α-helices, and BAF155 SWIRM is composed of five α-helices, respectively. Interestingly, our NMR data show that the N-terminal region, β2 strand, and β2-α1 loop of hSNF5 171-258 exhibit significant CSPs, revealing the detailed interaction with BAF155 SWIRM ( Figure 3C). In particular, the N-terminal loop region (His171-Val185) that is near the hSNF5 RPT1 domain shows a dramatic CSP upon BAF155 SWIRM binding. In addition, line widths of the amide resonances of His171, Glu178, Asn179, Ser181, and Gln182 residues in hSNF5 171-258 are unusually broadened upon BAF155 SWIRM binding, which suggested that the N-terminal loop region of hSNF5 171-258 plays an important role in this interaction ( Figure 3C). The BAF155 SWIRM domain also exhibits CSP in the α4 and α5 helices, as well as in the α4-α5 loop, a finding that is consistent with the previous report [34] ( Figure 3D). Interestingly, our data show large CSP in E473 and Q538 in the α2 helix and the C-terminal loop of BAF155 SWIRM , which does not exhibit with the previous structure [34]. Therefore, we expect that the major binding site of hSNF5 171-258 and BAF155 SWIRM during the complex formation differs from the previously reported site. We next determined the X-ray crystal structure of the hSNF5 171-258 /BAF155 SWIRM complex to confirm our findings in the solution state for the binding interface between hSNF5 171-258 and BAF155 SWIRM .

N-Terminal Loop of hSNF5 RPT1 Reveals Conformational Change upon BAF155 SWIRM Binding
We determined the high-resolution structure of the hSNF5 171-258 /BAF155 SWIRM complex using X-ray crystallography for a better understanding of the major binding force between hSNF5 171-258 and BAF155 SWIRM . We obtained a viable crystal at conditions of neutral pH and lower ionic strength and then solved the crystal structure of the complex at 2.28 Å resolution ( Figure 4A, Table 2). In overall complex structure, hSNF5 171-258 has an antiparallel β-sheet (β1, 186-195; β2, 198-208), flanked by three α-helices (α N , 172-181; α1, 216-226; α2, 230-248), and BAF155 SWIRM has five α-helices (α1, 465-470; α2, 472-475; α3, 484-500; α4, 508-514; and, α5, 519-532) ( Figure 4A). Although our crystal structure exhibits a similar fold to the previously reported structure (PDB code 5GJK), the additional N-terminal helix (α N ) of hSNF5 171-258 is only observed in our complex structure ( Figure 4A). The electron density map of a newly formed α N helix is clearly observed in our crystal structure and it provides a detailed orientation of the bound conformation ( Figure 4B). The α N helix and β2-α1 loop in hSNF5 171-258 intramolecularly interacts via hydrogen bonds and hydrophobic interaction following the residues, such as Val175, Asn179, Lys211, and Leu212 ( Figure 4C). Strikingly, in our complex structure, their secondary structure of the N-terminal loop of hSNF5 171-258 , which shows the large CSP change during BAF155 SWIRM binding, dramatically differs from its free state structure showing the random-coil fold ( Figure 2E,F). We note that the conformational change mainly occurs in the N-terminal loop region, otherwise hSNF5 171-258 in complex maintains the same structure as free hSNF5 171-258 in solution, yielding a Cα RMSD of 2.2 Å for secondary structural regions. Therefore, we found that there is a conformational transition from the coil to the helix in the N-terminal loop of hSNF5 171-258 during the complex state with BAF155 SWIRM ( Figure 4D). 1 H-15 N heteronuclear NOE (XNOE) data for hSNF5 171-258 in solution indicates the overall structural flexibility. The compactly folded secondary structure parts, β1β2α1α2, reveals high XNOE values, whereas the N-terminal and C-terminal regions exhibited reduced XNOE values, which indicated a flexible and disordered conformation ( Figure 4E). We could not observe the XNOE values of all the participating residues binding with BAF155 SWIRM because of the line broadening. However, we find a valid increment of the XNOE values in the residues from Asp172 to His177 locating the N-terminal helix after it binds with BAF155 SWIRM . It should also be noted that the C-terminal region (residues 249-258) of hSNF5 171-258 lacks a regular secondary structure in both the free state and the complex ( Figure 4D,E). The combined data from the NMR spectroscopy and X-ray crystallography uncovered that the N-terminal region of hSNF5 171-258 undergoes a coil-to-helix transition upon its binding with BAF155 SWIRM , illustrating a coupled folding and binding mechanism.  2.4. The Interface between hSNF5 171-258 and BAF155 SWIRM Features Charge Complementarity hSNF5 171-258 is a highly acidic protein (pI = 4.1), whereas BAF155 SWIRM is basic (pI = 7.9). The inspection of surface charges at the binding interface reveals charge complementarity, such that positively charged surfaces of BAF155 SWIRM ( Figure S1A) interact with negatively charged surfaces of hSNF5 171-258 ( Figure S1B). Complex formation between hSNF5 171-258 and BAF155 SWIRM buries 1684 Å2 of solvent-accessible surface area, which is~30% larger than the buried surface area of 1308 Å2 detected in the same complex without the α N helix. The interaction surface features key salt bridges and hydrogen bonds between hSNF5 171-258 and BAF155 SWIRM (Asp202-Arg512/Arg513 and Glu210-Arg524) ( Figure 5A). In addition, the Asn207 of hSNF5 171-258 forms a hydrogen bond to Asp518 of BAF155 SWIRM , and Asp225 of hSNF5 171-258 forms a hydrogen-bonding network with Ser508/Thr509 of BAF155 SWIRM . The binding interface with hSNF5 171-258 and BAF155 SWIRM is similar to that of the previous structure [34], but the binding distance between the residues are quite different ( Figure 5A). Specifically, the salt bridge distance between Asp202 and Arg512 in our hSNF5 171-258 /BAF155 SWIRM complex structure is about 0.7Å further apart when compared with that of the previous structure [34]. On the other hand, the salt bridge distance between Glu210 and Arg524 in our hSNF5 171-258 /BAF155 SWIRM complex structure is approximately 0.3Å closer than that of the previous structure ( Figure 5A). Moreover, we also found an additional hydrogen bond network in the presence of the α N helix of hSNF5 171-258 . A hydrogen bond network formed between the Asn179 of hSNF5 171-258 and Arg524 and the Glu473 of BAF155 SWIRM . The Asn179 located in the α N helix of hSNF5 171-258 , showing a high CSP value, also interacts directly with Lys211 ( Figures 3C and 5B). Therefore, it is considered that the newly observed hydrogen bond network that is mediated by the Asn179 in the α N helix of hSNF5 171-258 is a very important binding force for the hSNF5 171-258 /BAF155 SWIRM complex as well as hydrogen and hydrophobic interactions that were observed on the charged surface ( Figure 5B). In the NMR titration result, the Glu473 of BAF155 SWIRM represents a high CSP value, which is consistent with our hSNF5 171-258 /BAF155 SWIRM complex structure (Figures 3D and 5B). In addition, hydrogen bond pairs (Ala180-Asn476, Gln182-Asn479, and Glu184-Ans479/Ser481) form in a complex formation of hSNF5 171-258 /BAF155 SWIRM , which makes hydrogen bonding networks in the N-terminal region of hSNF5 171-258 . Backbone carbonyl groups of Ala180 and Gln182 from hSNF5 171-258 form hydrogen bonds with side chains of Asn476 and Asn479, respectively, from BAF155 SWIRM . The carboxyl side chain of Glu184 from hSNF5 171-258 forms hydrogen bonds and salt bridges with the side chains of Ser481 and Lys482 of BAF155 SWIRM ( Figure 5C). The α N helix also induces some hydrophobic interactions. Three residues, Ile176, His177, and Ala180, provide the hydrophobic interfaces for Pro472 of BAF155 SWIRM .
The crystal structure in this study is fully consistent with the CSP analysis from the NMR-titration experiments. Finally, we identified that the N-terminal region of hSNF5 171-258 has a conformational change and it induces a tight binding pocket during the complex state. We further investigated how the N-terminal region of hSNF5 171-258 is important for the target binding and stability of the protein complex. Data from isothermal titration calorimetry (ITC) experiments support the importance of the α N helix for BAF155 SWIRM binding in a quantitative manner. We observed that hSNF5 171-258 and hSNF5 171-253 exhibit the same binding affinity to BAF155 SWIRM , which indicates that the five residues in the C-terminal tail do not affect protein binding. We also prepared a series of N-terminal truncation hSNF5 171-253 mutants and measured how these truncations impact BAF155 SWIRM binding (Table 3). Our data show that hSNF5 171-253 binds to BAF155 SWIRM , with a dissociation constant (K D ) of 110 ± 20 nM and a 1:1 stoichiometry ( Figure 6A).
The binding affinity reduced by 2.6-fold when the whole α N helix was removed (hSNF5 183-253 ) (KD~290 ± 80 nM). The moderate decrease in affinity suggests that the α N helix might engage with the BAF155 SWIRM via transient interactions. We note that the α N helix exhibited relatively low heteronuclear NOE (XNOE) values, even in complex with the BAF155 SWIRM , as shown in Figure 4E. Therefore, the α N helix in the hSNF5 171-258 /BAF155 SWIRM complex in solution could retain a degree of conformational mobility, providing favorable interactions for stabilizing the complex.
Progressive truncations of the α N helix show that hSNF5 174-253 maintains the same binding affinity, with a K D of~110 ± 30 nM, whereas hSNF5 179-253 and hSNF5 183-253 show decreased binding affinities, with K D values of~300 ± 50 nM and~290 ± 80 nM, respectively, similar to that observed for the whole α N helix-deletion mutant. The truncation of the N-terminal loop preceding the β1 strand results in a~7-fold reduction in binding affinity, such that hSNF5 186-253 exhibits a K D of~758 ± 80.3 nM for BAF155 SWIRM binding (Fig. 6B). This large reduction results from the loss of hydrogen bonding from Glu184 of hSNF5 171-253 , as well as the α N -helix deletion. We note that a single mutation of Glu184 reduces the binding affinity by 3.4-fold, yielding a K D of~370 ± 100 nM for the E184A mutation of hSNF5 171-253 ( Figure 6C). We further found that alanine mutations of Asn476 and Asn479 in BAF155 SWIRM , which make hydrogen bonds to Ala180 and Gln182, respectively, in hSNF5 171-253 have a negligible impact on binding. The N476A/N479A mutant of BAF155 SWIRM only shows a modest decrease in binding affinity, with a K D of~140 ± 30 nM, indicating a minor role or replaceable nature for their interactions. Finally, the N-terminal mutant of hSNF 171-253 (hSNF5 SFH1/171-253 ), followed by the human sequence (residues 185-253), exhibits a 4.4-fold lower binding affinity. The N-terminal tail region that extends 15 amino acids in the RPT1 sequence, is not conserved between the human and yeast sequences, so this reduced binding results from divergent sequences in the α N helix and its surrounding region. Lastly, we investigated whether the α N helix formation significantly contributes to the stability of the hSNF5 171-253 /BAF155 SWIRM complex by measuring the melting temperature (Tm) with circular dichroism (CD) spectroscopy. Although hSNF5 171-253 and hSNF5 186-253 both exhibit the same Tm of 49 • C in the free state, hSNF5 171-253 in complex has a higher Tm (~53 • C) than hSNF5 186-253 (Tm~48 • C), demonstrating that the α N helix is a determinant of protein stability for the hSNF5 171-253 /BAF155 SWIRM complex ( Figure 6D,E).  Table 3. Thermal stability of hSNF5 171-253 and hSNF5 186-253 (D) alone and (E) with BAF155 SWIRM . The melting temperatures (Tm) were calculated by the decrease in normalized molar ellipticity at 222 nm in the circular dichroism (CD) spectrum.

Discussion
The core complex that is composed of SNF5, BAF155, and BAF170 was found to show a remodeling activity that resembles that of the entire mammalian SWI/SNF complex [38,39]. An important step toward understanding how specific subunits function is to elucidate how they interact within the complex. To date, an atomic-resolution structure of the human SWI/SNF complex has not been obtained, although several individual subunits and domains have been described. Here, we solved the structure of the hSNF5 171-258 /BAF155 SWIRM complex at atomic resolution, providing biophysically and functionally important information. Our data indicate that the N-terminal loop preceding hSNF5 RPT1 undergoes a conformational change upon interaction with BAF155 SWIRM , which enhances the binding strength and stability of the complex state. The yeast Cryo-EM structure showed the presence of α N helix in the SWI/SNF complex [40]. Further, the hydrogen bond network and hydrophobic interactions between the α N helix and the SWIRM domain in our structure recapitulated, illustrating that the α N helix presents the bona fide interface for the SWIRM domain.
Interestingly, a number of studies report that mutations in the N-terminal residues Pro173 and Glu184 of hSNF5 lead to different cancers, such as Rhabdoid tumors and malignant melanoma [41,42].
The Glu184 residue of hSNF5 171-258 plays an important role in binding between the N-terminal region of hSNF5 171-258 and BAF155 SWIRM , and it is involved in tumor suppression [41]. We speculate that the mutation of Glu184, where the N-terminal region of hSNF5 171-258 binds to BAF155 SWIRM , interrupts the formation of the hSNF5 and BAF155 complex, making it difficult to form chromatin-remodeling complexes and chromatin, and thereby resulting in tumor genesis. When considering that mutations of Glu184 and Val185 abolish the interaction with HIV-1 integrase (IN) [43], we infer that residues on the non-structured loop following the N-terminal tail play an important role in the interaction with IN to a greater extent than with BAF155 SWIRM .
Intrinsically disordered regions can perform various biological functions via coupled folding and binding mechanisms [44][45][46][47][48]. The complex structure of hSNF5 171-258 and BAF155 SWIRM illustrates that the interaction between hSNF5 and BAF155 also involves a coupled folding and binding mechanism. The coil-to-helix transition of hSNF5 171-258 upon BAF155 SWIRM binding might help to distinguish between various targets, thereby fine-tuning individual interactions. The structural transition of hSNF5 171-258 can alter chromatin states and remodeling activity in vivo for the temporal and spatial control of biological processes, such as the recruitment of chromatin remodelers or transcriptional activation.
In summary, we have determined the crystal structure of the hSNF5 171-258 /BAF155 SWIRM complex as a heterodimer by 1:1 molar ratio, which reveals the detailed binding mode. The N-terminal disordered loop of hSNF5 171-258 undergoes a coil-to-helix transition upon binding to BAF155 SWIRM . In the N-terminal region of hSNF5 171-258 , residues between Asn179 and Glu184 were identified as core regions that bind to BAF155 SWIRM , which is critical for protein binding and it enhances the stability of the protein complex. In addition, our data will provide a structural clue to understanding the mechanism of the pathogenesis of diseases that are related to the chromatin remodeling complex and originated from mutations in the N-terminal region of hSNF5.

Cloning, Protein Expression, and Purification
The cDNA fragments encoding residues 171-258 of hSNF5 (hSNF5 171-258 ) and the various N-terminal truncation mutants and residues 449-546 of human BAF155 (BAF155 SWIRM ) and the BAF155 SWIRM mutants were amplified by PCR. The amplified cDNA fragments were inserted into the modified expression vector pET21b (Novagen, Darmstadt, Germany) for hSNF5 171-258 and the N-terminal truncation mutants and pMAL-C2X (Novagen, Darmstadt, Germany) for BAF155 SWIRM and the BAF155 SWIRM mutants as fusions with N-terminal His 6 -affinity tags and tobacco etch virus (TEV) protease cleavage sites. The plasmids were transformed into Escherichia coli Strain BL21 (DE3) (Invitrogen, Waltham, MA, USA). Cells expressing hSNF5 171-258 and BAF155 SWIRM were grown in LB or M9 minimal media with 15 NH 4 Cl and/or 13 C 6 -glucose as the sole nitrogen and carbon sources, respectively, at 37 • C to an OD 600 of~0.6. Protein expression was induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 25 • C and the cells were harvested by centrifugation after 16 h of induction and then stored frozen at −70 • C. Cell pellets were resuspended and disrupted by sonication in lysis buffer containing 25 mM sodium phosphate, pH 7.0, 300 mM NaCl, 5 mM β-mercaptoethanol, and protease inhibitor cocktail (Roche, Basel, Switzerland). The His 6 -tagged fusion proteins were purified by immobilized metal affinity chromatography on a Ni-NTA column (QIAGEN, Hilden, Germany) and then cleaved by TEV protease for 12 h. The digestion reactions were loaded onto a Ni-NTA column, and the flow-through was collected and loaded onto a Superdex 200 16/60 column (GE Healthcare, Chicago, IL, USA), equilibrated with 10 mM HEPES, pH 7.0, 100 mM NaCl, and 2 mM dithiothreitol (DTT).
SEC multi-angle light scattering (SEC-MALS) experiments were performed using the Agilent 1200 HPLC system (Agilent Technologies, Santa Clara, CA, USA), combined with a Wyatt DAWN HELEOS-II MALS instrument and a Wyatt OPtilab rEX differential refractometer (Wyatt Technology, Santa Barbara, CA, USA). For chromatographic separation, a WTC-030S5 column (Wyatt Technology, Santa Barbara, CA, USA) was used at a flow rate of 0.5 mL/min. in the same buffer as was used for SEC experiments. The same concentrations of hSNF5 171-258 , BAF155 SWIRM , and hSNF5 171-258 /BAF155 SWIRM complex that were used for SEC analysis were also used in the SEC-MALS experiments. The results were normalized and analyzed while using ASTRA software (Wyatt Technology, Santa Barbara, CA, USA).

NMR Structure Determination and Analysis
The structure calculations for BAF155 SWIRM and hSNF5 171-258 in the free state were performed using semi-automated CYANA 2.1 [53]. The initial fold was obtained by manual NOE assignment, and secondary topology was determined by peak picking. NOE cross-peak assignments were obtained using manual procedures. An initial fold of the protein was calculated on the basis of manually assigned NOEs, which is essential for secondary topology determination, with subsequent use of the program CYANA 2.1. Seven cycles of the CYANA routine were calculated; each cycle had 10,000 steps of torsion-angle dynamics, with a simulated annealing protocol. A total of 100 structures were calculated, and the 20 structures with the lowest target function values were chosen for analysis. Table 2 shows a summary of the NMR-derived restraints and structures of hSNF5 171-258 . The final structures with the lowest NOE energies were retained and validated with the PROCHECK program [54].

Crystallization and Structure Determination
Crystals of the hSNF5 171-258 /BAF155 SWIRM complex were grown under oil at 15 • C in a 1.5 µL micro-batch, containing equal volumes of protein solution and mother liquor (15% v/v Tacsimate, pH 7.0, 0.1 M HEPES, pH 7.0, 2% (w/v) polyethylene glycol 3350). Purified hSNF5 171-258 was mixed with BAF155 SWIRM at a 1:1 molar ratio. The crystals were flash-cooled in liquid nitrogen, with 25% glycerol being used as a cryoprotectant. The diffraction datasets were collected using a synchrotron radiation source at beamline 4A at the Pohang Accelerator Laboratory (Pohang, Korea) at 100 K. Collected data were integrated and scaled while using the HKL-2000 program (HKL Research Inc., Charlottesville, VA, USA). The crystal structure of the hSNF5 171-258 /BAF155 SWIRM complex was solved by the molecular replacement method with the MOLREP [56] and Phaser programs, using a high-accuracy template-based modeling method as the search model. The final structure was refined to the maximum resolution of 2.28 Å by iterative cycles of the manual building in COOT and then restrained refinement with Refmac5 and Phenix [57]. Table 1 summarizes data collection and refinement for the complex structure. The final structures were analyzed while using MolProbity [58].

Crystallization and Structure Determination
The ITC experiments were performed using a VP-ITC system (MicroCal Inc. Northampton, MA, USA) at 25 • C in a buffer containing 25 mM sodium phosphate, 100 mM NaCl, and 1 mM Tris(2-carboxyethyl)-phosphine (TCEP), pH 7.0. Typically, 160 µM hSNF5 171-253 was injected 35 times in 7 µL aliquots into the 1.4 mL sample cell, containing either the BAF155 SWIRM domain or BAF155 SWIRM mutants at a concentration of 10 µM. The data were fit with a non-linear least-squares best-fit curve, while using a single-site binding model with Origin for ITC (Microcal Inc.), by varying the stoichiometry (n), the enthalpy of the reaction (∆H), and the dissociation constant (K D ).

Circular Dichroism Spectroscopy
CD experiments to measure the melting temperatures were performed in buffer containing 25 mM sodium phosphate, 100 mM NaCl, and 1 mM Tris(2-carboxyethyl)-phosphine (TCEP), pH 7.0, as described previously [59,60]. Samples containing 10 µM hSNF5 171-253 or hSNF5 186-253 alone or with 10 µM BAF155 SWIRM were monitored at temperatures between 25 • C and 95 • C, with an interval of 5 • C. The parameters for far-ultraviolet UV CD measurements used a cell with a path length of 0.2 cm for scanning between 250-200 nm, with a 1 nm bandwidth and scan speed of 200 nm/min −1 . The values were baseline corrected by subtracting a buffer spectrum, and an average of five scans was recorded for each experiment. The apparent melting temperature at the midpoint of the transition was obtained by fitting the experimental data points (CD signal at 222 nm vs. temperature) with a Boltzmann sigmoidal function.