Portal de Periódicos da CAPES

Structural Organization of a Sex-comb-on-midleg/Polyhomeotic Copolymer

2005; Elsevier BV; Volume: 280; Issue: 30 Linguagem: Inglês

10.1074/jbc.m503055200

1083-351X

Chongwoo A. Kim, M.R. Sawaya, Duilio Cascio, Woojae Kim, James U. Bowie,

RNA Interference and Gene Delivery

The polycomb group proteins are required for the stable maintenance of gene repression patterns established during development. They function as part of large multiprotein complexes created via a multitude of protein-protein interaction domains. Here we examine the interaction between the SAM domains of the polycomb group proteins polyhomeotic (Ph) and Sex-comb-on-midleg (Scm). Previously we showed that Ph-SAM polymerizes as a helical structure. We find that Scm-SAM also polymerizes, and a crystal structure reveals an architecture similar to the Ph-SAM polymer. These results suggest that Ph-SAM and Scm-SAM form a copolymer. Binding affinity measurements between Scm-SAM and Ph-SAM subunits in different orientations indicate a preference for the formation of a single junction copolymer. To provide a model of the copolymer, we determined the structure of the Ph-SAM/Scm-SAM junction. Similar binding modes are observed in both homo- and heterocomplex formation with minimal change in helix axis direction at the polymer joint. The copolymer model suggests that polymeric Scm complexes could extend beyond the local domains of polymeric Ph complexes on chromatin, possibly playing a role in long range repression. The polycomb group proteins are required for the stable maintenance of gene repression patterns established during development. They function as part of large multiprotein complexes created via a multitude of protein-protein interaction domains. Here we examine the interaction between the SAM domains of the polycomb group proteins polyhomeotic (Ph) and Sex-comb-on-midleg (Scm). Previously we showed that Ph-SAM polymerizes as a helical structure. We find that Scm-SAM also polymerizes, and a crystal structure reveals an architecture similar to the Ph-SAM polymer. These results suggest that Ph-SAM and Scm-SAM form a copolymer. Binding affinity measurements between Scm-SAM and Ph-SAM subunits in different orientations indicate a preference for the formation of a single junction copolymer. To provide a model of the copolymer, we determined the structure of the Ph-SAM/Scm-SAM junction. Similar binding modes are observed in both homo- and heterocomplex formation with minimal change in helix axis direction at the polymer joint. The copolymer model suggests that polymeric Scm complexes could extend beyond the local domains of polymeric Ph complexes on chromatin, possibly playing a role in long range repression. The PcG 1The abbreviations used are: PcG, polycomb group; Ph, polyhomeotic; Scm, sex-comb-on-midleg; ML, mid-loop; EH, end-helix; GST, glutathione S-transferase; ADA, [(carbamoylmethyl)imino]diacetic acid; bistris, 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol; SUMO, small ubiquitin-like modifier; βME, β-mercaptoethanol. proteins form large multiprotein complexes that repress transcription over long distances and maintain repressed states in a stable and heritable manner during embryogenesis, development, and into adulthood (1Francis N.J. Kingston R.E. Nat. Rev. Mol. Cell. Biol. 2001; 2: 409-421Crossref PubMed Scopus (310) Google Scholar, 2Orlando V. Cell. 2003; 112: 599-606Abstract Full Text Full Text PDF PubMed Scopus (295) Google Scholar). PcG mutations result in both developmental defects and cancer (3Jacobs J.J. van Lohuizen M. Biochim. Biophys. Acta. 2002; 1602: 151-161PubMed Google Scholar, 4Valk-Lingbeek M.E. Bruggeman S.W. van Lohuizen M. Cell. 2004; 118: 409-418Abstract Full Text Full Text PDF PubMed Scopus (644) Google Scholar, 5van Lohuizen M. Curr. Opin. Genet. Dev. 1999; 9: 355-361Crossref PubMed Scopus (73) Google Scholar). The formation of such large complex structures requires the precise organization of many proteins of the PcG. Indeed, protein-protein interaction domains are the most common functional motifs that have been identified in the PcG family. Nevertheless, few of these domains have been characterized in detail, and the structural architecture of the resultant protein complexes is largely unknown. One protein-protein interaction domain present in the PcG family is the SAM (sterile alpha motif) domain (often called SPM in PcG literature) found in polyhomeotic (Ph) and Sex-comb-on-midleg (Scm). SAM domains are small, helical protein modules found in proteins with diverse functional roles ranging from receptor tyrosine kinases to transcription factors (6Ponting C.P. Protein Sci. 1995; 4: 1928-1930Crossref PubMed Scopus (143) Google Scholar). Unlike many other protein-protein interaction modules that have a common, well defined function such as SH2 domains, SAM domains can play diverse roles. To date, SAM domains are known to form homo- and heterooligomers that can be polymeric (7Kim C.A. Phillips M.L. Kim W. Gingery M. Tran H.H. Robinson M.A. Faham S. Bowie J.U. EMBO J. 2001; 20: 4173-4182Crossref PubMed Scopus (206) Google Scholar, 8Kim C.A. Gingery M. Pilpa R.M. Bowie J.U. Nat. Struct. Biol. 2002; 9: 453-457PubMed Google Scholar, 9Qiao F. Song H. Kim C.A. Sawaya M.R. Hunter J.B. Gingery M. Rebay I. Courey A.J. Bowie J.U. Cell. 2004; 118: 163-173Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar) or have a discrete oligomeric state (10Ramachander R. Kim C.A. Phillips M.L. Mackereth C.D. Thanos C.D. McIntosh L.P. Bowie J.U. J. Biol. Chem. 2002; 277: 39585-39593Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar), they can bind to other proteins (11Kasten M. Giordano A. Oncogene. 2001; 20: 1832-1838Crossref PubMed Scopus (95) Google Scholar, 12Zhang H. Xu Q. Krajewski S. Krajewska M. Xie Z. Fuess S. Kitada S. Pawlowski K. Godzik A. Reed J.C. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 2597-2602Crossref PubMed Scopus (165) Google Scholar), and they can even bind RNA (13Aviv T. Lin Z. Lau S. Rendl L.M. Sicheri F. Smibert C.A. Nat. Struct. Biol. 2003; 10: 614-621Crossref PubMed Scopus (165) Google Scholar, 14Green J.B. Gardner C.D. Wharton R.P. Aggarwal A.K. Mol. Cell. 2003; 11: 1537-1548Abstract Full Text Full Text PDF PubMed Scopus (89) Google Scholar). Ph and Scm proteins co-localize on polytene chromosomes (15Peterson A.J. Kyba M. Bornemann D. Morgan K. Brock H.W. Simon J. Mol. Cell. Biol. 1997; 17: 6683-6692Crossref PubMed Scopus (127) Google Scholar) and can interact via their SAM domains (15Peterson A.J. Kyba M. Bornemann D. Morgan K. Brock H.W. Simon J. Mol. Cell. Biol. 1997; 17: 6683-6692Crossref PubMed Scopus (127) Google Scholar, 16Kyba M. Brock H.W. Dev. Genet. 1998; 22: 74-84Crossref PubMed Scopus (64) Google Scholar, 17Peterson A.J. Mallin D.R. Francis N.J. Ketel C.S. Stamm J. Voeller R.K. Kingston R.E. Simon J.A. Genetics. 2004; 167: 1225-1239Crossref PubMed Scopus (55) Google Scholar). The interaction between Ph and Scm appears to be complex. In addition to binding to each other, the SAM domains of Ph and Scm can also self-associate and thus exist in a higher oligomeric state of their own (8Kim C.A. Gingery M. Pilpa R.M. Bowie J.U. Nat. Struct. Biol. 2002; 9: 453-457PubMed Google Scholar, 15Peterson A.J. Kyba M. Bornemann D. Morgan K. Brock H.W. Simon J. Mol. Cell. Biol. 1997; 17: 6683-6692Crossref PubMed Scopus (127) Google Scholar, 16Kyba M. Brock H.W. Dev. Genet. 1998; 22: 74-84Crossref PubMed Scopus (64) Google Scholar). Although they appear to work at the same sites, isolation of the soluble form of one PcG protein complex, PRC1, showed that Scm is present in substoichiometric amounts compared with other proteins in the complex, including Ph, Posterior Sex Combs (Psc), Polycomb (Pc) and dRING1 (18Saurin A.J. Shao Z. Erdjument-Bromage H. Tempst P. Kingston R.E. Nature. 2001; 412: 655-660Crossref PubMed Scopus (324) Google Scholar). Ph-SAM forms a helical, head-to-tail polymer structure similar to that formed by the SAM domain of another unrelated transcriptional repressor, TEL (7Kim C.A. Phillips M.L. Kim W. Gingery M. Tran H.H. Robinson M.A. Faham S. Bowie J.U. EMBO J. 2001; 20: 4173-4182Crossref PubMed Scopus (206) Google Scholar, 8Kim C.A. Gingery M. Pilpa R.M. Bowie J.U. Nat. Struct. Biol. 2002; 9: 453-457PubMed Google Scholar). In both structures, the individual SAM domain subunits bind using two surfaces named the mid-loop (ML) and the end-helix (EH) binding surfaces. SAM domain polymerization could provide a mechanism for long range repression by spreading repression complexes along the chromatin. Scm can indeed mediate long range repression. When Scm is tethered to a DNA binding site by fusion to a DNA binding domain, it is able to repress transcription over long distances (19Roseman R.R. Morgan K. Mallin D.R. Roberson R. Parnell T.J. Bornemann D.J. Simon J.A. Geyer P.K. Genetics. 2001; 158: 291-307Crossref PubMed Google Scholar). Long range repression by Scm is dependent on both an intact SAM domain as well as Ph function. In addition, overexpression of an isolated Scm-SAM domain results in a dominant loss of Scm function, suggesting that protein-protein interactions via the SAM module can poison full-length Scm proteins (17Peterson A.J. Mallin D.R. Francis N.J. Ketel C.S. Stamm J. Voeller R.K. Kingston R.E. Simon J.A. Genetics. 2004; 167: 1225-1239Crossref PubMed Scopus (55) Google Scholar). Here, we show that Scm-SAM forms a polymeric structure similar to Ph-SAM and investigate how the two polymeric domains of Ph and Scm can co-polymerize. Our results suggest that independent blocks of the two SAM polymers would preferentially join in one orientation, thereby forming a single junction copolymer. SAM Domain Constructs—The numbering and the procedures for the Ph-SAM mutant and Se-Met preparations have been described previously (7Kim C.A. Phillips M.L. Kim W. Gingery M. Tran H.H. Robinson M.A. Faham S. Bowie J.U. EMBO J. 2001; 20: 4173-4182Crossref PubMed Scopus (206) Google Scholar). Briefly, a fragment of Drosophila melanogaster Ph, comprising amino acids 1502–1577, was cloned into a modified pET-3c vector where Val-1502 of Ph corresponds to Val-6 in our numbering scheme. Scm-SAM mutants were cloned into the same modified pET-3c vector. The final Scm polypeptide encoded by the construct incorporates an N-terminal MEKTR leader sequence, Scm amino acids 795–870 then a C-terminal sequence of DHHHHHH. Thus Ala-795 of SCM corresponds to Ala-6 in our numbering scheme. All mutants were prepared using the Stratagene QuikChange mutagenesis kit. All Scm-SAM proteins were expressed in BL21(DE3) pLysS cells. Typically, cells from a 1-liter culture were resuspended in 10 ml of 50 mm Tris, pH 8.0, 200 mm NaCl, 10 mm βME, 10 mm imidazole, pH 7.5, and 6 m urea and lysed by sonication. The denatured soluble extract was applied to a 1.0-ml column of nickel-nitrilotriacetic acid-agarose (Qiagen) and washed extensively in the same buffer. Refolding of the peptide was performed on the resin by washing the column with 10 ml of the same buffer but with 4 m urea followed by successive washes with buffer containing 3, 2, and 1 m urea and then finally with a two 10-ml washes with buffer in the absence of urea. The protein was eluted with 10 ml of 50 mm NaCl, 10 mm βME and 300 mm imidazole, pH 7.0. The eluted solution was loaded directly onto a 5-ml HiTrap SP column (Pharmacia) equilibrated in 25 mm Tris, pH 7.0, 50 mm NaCl, 10 mm βME and eluted with a linear gradient to 25 mm Tris, pH 8.5, 1 m NaCl, 10 mm βME. The purified protein was dialyzed into 10 mm bis-tris propane, pH 6.4, 50 mm NaCl, 2 mm βME and concentrated using an Amicon ultrafiltration device. The GST-Scm-SAM fusion proteins were purified using a C-terminal His6 tag as described above except that βME in the lysis/wash buffer was replaced with 5 mm dithiothreitol. X-ray Crystallography—Crystals of the Scm-SAM/Ph-SAM complex were grown using the hanging drop vapor diffusion method. 2–3 μl of well buffer was mixed with an equal volume of an 18–25 mg ml-1 solution of a 1:1 molar ratio of the native Scm-SAM L52R/M57R double mutant and a fully incorporated Se-Met Ph-SAM L69R mutant. The best crystals were grown using a well buffer consisting of 50 mm ADA, pH 6.0, 20% polyethylene glycol 1000 w/v, 50 mm NH4 acetate. Crystals grew over 8 weeks at 20 °C. Four data sets, collected at different wavelengths for multiwavelength anomalous diffraction phasing, were obtained at the National Synchrotron Light Source at the Brookhaven National Laboratory on Beamline X8-C and a 1.8-Å data set was collected at the Advanced Light Source on Beamline 5.0.2 (see Table I for more details). All data were processed with DENZO/SCALEPACK (20Otwinowski Z. Minor W. Methods Enzymol. 1997; 276: 307-326Crossref PubMed Scopus (38782) Google Scholar). Heavy atom positions were determined using ShelxD (21Uson I. Sheldrick G.M. Curr. Opin. Struct. Biol. 1999; 9: 643-648Crossref PubMed Scopus (265) Google Scholar), which were then used to calculate phases with MLPHARE (22Number Collaborative Computational Project Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19879) Google Scholar). Solvent flattening was performed using DM (22Number Collaborative Computational Project Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19879) Google Scholar). The starting model was built using ONO with partial helices built by MAID (23Levitt D.G. Acta Crystallogr. Sect. D Biol. Crystallogr. 2001; 57: 1013-1019Crossref PubMed Scopus (118) Google Scholar) as a guide. Refinement was carried out against the high resolution data set collected at ALS, setting aside 5% of the reflections for calculation of a cross validation R-factor (Rfree). CNS (24Brunger A.T. Adams P.D. Clore G.M. DeLano W.L. Gros P. Grosse-Kunstleve R.W. Jiang J.S. Kuszewski J. Nilges M. Pannu N.S. Read R.J. Rice L.M. Simonson T. Warren G.L. Acta Crystallogr. Sect. D Biol. Crystallogr. 1998; 54: 905-921Crossref PubMed Scopus (17025) Google Scholar) was used for the refinement with the maximum likelihood target using amplitudes function. The PDB accession code for the hetero-SAM domain complex structure is 1PK1.Table ICrystallographic data tables Rsym = Σ | I - 〈I〉|/Σ 〈I〉, where I is the observed intensity, and 〈I〉 is the average intensity from observations of symmetry-related reflections. Rcryst = Σ | Fobs - Fcalc |/Σ Fobs, where Fobs and Fcalc are the observed and calculated structure factor amplitudes, respectively. Rfree is calculated for a set of reflections (5%) that were not included in atomic refinement. Rcryst and Rfree were calculated after bulk solvent correction and with no reflection intensity cutoff. Open table in a new tab For Scm-SAM alone, the best crystals were grown of the single mutant L52R. The crystals were grown by vapor diffusion, mixing equal amounts of 50 mm ADA, pH 6.4, 1.5 m ammonium acetate well buffer, with 16 mg/ml protein. A 1.85-Å data set was collected using a Rigaku FR-D generator and the images were collected on a RAXIS IV++ image plate detector. The structure was solved by molecular replacement using EPMR (25Kissinger C.R. Gehlhaar D.K. Fogel D.B. Acta Crystallogr. Sect. D Biol. Crystallogr. 1999; 55: 484-491Crossref PubMed Scopus (691) Google Scholar). A single Scm-SAM molecule from the complex with Ph-SAM was initially used as the search model. Despite having an accurate search model, EPMR was unable to find a solution for the three molecules in the asymmetric unit using a single molecule of Scm-SAM. The correct solution was found using three Scm-SAMs aligned over three Ph-SAM molecules of the Ph-SAM polymer structure. The structure was refined using CNS with the maximum likelihood target using amplitudes function. The PDB accession code for Scm-SAM is 1PK3. Surface Plasmon Resonance—The surface plasmon resonance experiments were performed at 20 °C in 25 mm bis-tris propane, pH 7.0, 200 mm NaCl and 0.005% Surfactant P20. For Scm-SAM self-association (Fig. 1e), Scm-SAM L66R was placed on a Biacore Pioneer CM5 sensor chip. The EH surface mutant ScmSAM L66R was immobilized, and various concentrations ranging from 3.8 × 10-8m to 1.9 × 10-7m of the ML surface mutant Scm-SAM L52R/Y61R were introduced in the mobile phase. For the interactions with Ph-SAM, wild-type Ph-SAM was immobilized on a Biacore Pioneer CM5 sensor chip and various concentrations of the Scm-SAM L66R (Fig. 2b, lower panel) and Y61R (Fig. 2b, upper panel) mutants injected onto the chip. The resulting binding data were analyzed with the BIAevaluation 3.0 software.Fig. 2Ph- and Scm-SAM interactions.a, GST pull-down assay. The top bands corresponds to the GST-fused SAM domains, which are identified on the top panel above each lane. The non-fused SAM domains are indicated by the lower panel. The mutations are listed by their numbered locations where the residue was mutated to Arg. For example, Scm5261 would be the double mutant of L52R and Y61R. Ph5156 and Scm5261 are ML surface mutations and Ph65 and Scm66 are EH surface mutations. Possible binding interactions for the various combinations are illustrated for each lane indicated by the arrows. Scm- and Ph-SAM are illustrated as gray and white objects, respectively, and mutations are indicated by the circled X. b, binding affinity of hetero-SAM domain interactions. The surface plasmon resonance data are shown for the indicated hetero-SAM domain binding orientations that confer a positive binding signal in a (lanes 2 and 3). The fitted curves for the binding reaction are overlaid on the actual data. The observed rate constants and the calculated dissociation constant derived from the rate constants are shown. c, the binding affinities suggest preference for the formation of a single junction copolymer. A nucleated hetero-SAM dimeric complex (Kd = 54 nm, Fig. 2b) is shown at the left. The two possible ends where either the Ph-SAM or Scm-SAM could bind are indicated with the corresponding dissociation constants for each. Scm-SAM would preferentially bind on the left (to a prior Scm-SAM) and Ph-SAM would preferentially bind on the right (to a prior Ph-SAM). The result of the favorable binding reactions is shown in the middle along with the possible binding modes for additional SAM domains. The right panel shows the expected result of further co-polymerization.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Scm-SAM Forms a Helical Polymer—To determine the structural organization of the SAM domains of Ph and Scm, it is essential to first know the individual structures of both. Scm-SAM is largely insoluble, and electron micrographs of the protein showed the formation of long fibers (not shown). We therefore suspected that Scm-SAM forms polymers, much like Ph-SAM. As formation of a heterogeneous, insoluble polymer precludes high resolution structural studies, we used an approach that has been successful for the structure determination of both Ph-SAM and TEL-SAM polymers. For these proteins, we introduced mutations that disrupted the polymer interface sufficiently so that the proteins became largely monomeric and soluble, but the interface retained adequate affinity so that the polymer reformed upon crystallization. Based on the solved Ph-SAM structure, we introduced a series of mutations into the possible Scm-SAM interface and obtained soluble mutants that were subjected to crystal trials. An L52R mutant provided the best crystals, and its structure was solved by molecular replacement and refined to an Rfree of 0.231 at 1.85 Å. As expected from the electron micrographs, subunits in the Scm-SAM crystal are arranged as a helical polymer similar to the polymers of Ph-SAM and TEL-SAM (Fig. 1a). The 52-Å repeat distance along the approximate 65 screw axis is essentially identical to the 52–53 Å observed for the TEL-SAM polymer, but somewhat longer than the 45-Å repeat distance for Ph-SAM. The apolar residues buried in the interfaces of all three SAM polymers are highlighted in red and green for the EH and ML binding surfaces, respectively, in Fig. 1b. As can be seen from the alignment, residue positions involved in Scm-SAM self-association are nearly identical to the ones observed for Ph-SAM. Like the Ph-SAM binding interface, the apolar and polar interactions are in separate clusters with the hydrophobic residues near the center axis of the helical polymer, whereas the salt bridges and hydrogen bond interactions are on the outside (Fig. 1, c and d). The apolar residues Ala-49, Met-57, and Tyr-61, which make up the EH binding surface, along with the L52R mutant side chain, are in the same positions as the residues observed for Ph-SAM. Met-59, Leu-66, Gly-67 (at the center of the interface), Leu-70, and Asn-74 constitute the ML binding surface. We have included Asn-74 as part of the apolar domain because it plays a similar role to the Ala of Ph-SAM. The Cβ of Asn in Scm-SAM replaces the Cβ of Ala in Ph-SAM, and the amide moiety escapes to solvent. Another minor difference between the apolar regions of the binding interface is the addition of the longer Met-59 in the Scm-SAM interface, whereas the equivalent, shorter Val side chain of Ph-SAM remains in the core and does not participate in the binding interface. Although the three molecules in the asymmetric unit of the crystal make slightly different polar contacts, some interactions are common to all of them (Figs. 1c and 3c). The carbonyl oxygen of His-43 on the ML surface, hydrogen bonds to the backbone nitrogen of the neighboring EH surface residue, Gly-67. Asp-46 on the ML surface forms a salt bridge with Lys-71 on the EH surface and simultaneously hydrogen bonds to the backbone nitrogen of Ala-49 in the same polypeptide chain. Lys-71 is positioned by hydrogen bonds to the γ oxygen of Ser-32, which in turn interacts with Asp-30, all within the same polypeptide chain as Lys-71. To measure the affinity between subunits in the wild-type polymer, we created a dimeric Scm-SAM complex by combining a ML surface double mutant (L52R/Y61R) with an EH surface mutant (L66R). In this complex, the wild-type surfaces can bind, but polymer extension is prevented by the mutations. Surface plasmon resonance experiments, shown in Fig. 1e, revealed a dissociation constant of 47 ± 4nm. Thus, the natural interface is quite stable. A Ph-SAM and Scm-SAM Copolymer—Given the similarity of the Scm-SAM and Ph-SAM polymers, it seems likely that a copolymer could form when they are mixed. If the binding interfaces in the two SAM domains are completely interchangeable, a random copolymer would form. If Scm-SAM can only bind to one end of the Ph-SAM polymer, however, a single joint copolymer would form, containing a block of Scm-SAM polymer at one end and a block of Ph-SAM polymer at the other. Alternatively, Scm-SAM and Ph-SAM may bind to each other using completely different binding surfaces than in the homopolymeric structures, forming a more complex, intertwined polymer. To address how the Scm-SAM and Ph-SAM polymers join, we tested the affinity of all possible binding orientations using ML and EH surface mutants to force particular binding modes. The GST pull-down experiments shown in Fig. 2a, show that binding is not observed between Ph-SAM and Scm-SAM when both bear mutations in the ML surfaces or both bear mutations in the EH surfaces. Thus "head-to-head" or "tail-to-tail" binding modes are not productive (Fig. 2a, lanes 1 and 4). We did observe strong binding between the EH surface of Scm-SAM and the ML surface of Ph-SAM (lane 3) and much weaker binding in the reverse orientation (lane 2). These results indicated that the hetero-SAM interaction utilizes the same surfaces as the homopolymers and that binding in one of the two "head-to-tail" orientations is strongly preferred. Surface plasmon resonance experiments confirm the qualitative impression provided by the GST pull-down experiments. Wild-type Ph-SAM was immobilized on the chip, and various concentrations of Scm-SAM mutants were introduced in the mobile phase. The dissociation constant between Ph-SAM and the Scm-SAM ML surface mutant, Y61R, was found to be a very stable 54 ± 2 nm. For the Scm-SAM EH surface mutant, L66R, binding to Ph-SAM was much weaker with a Kd of nearly 1 μm. The binding affinities for all the possible interactions are summarized in Fig. 2c. The affinities observed for the various possible homo- and hetero-SAM interactions suggest that the single joint copolymer (Fig. 2c) would be favored with the isolated SAM domains under equilibrium conditions. Interactions between Scm-SAM domains and Ph-SAM domains have dissociation constants of 47 ± 2 and 190 ± 23 nm, respectively. The joint between blocks of Scm-SAM and Ph-SAM is also quite strong, with a Kd of 54 ± 2nm, in the Scm-SAM EH surface/Ph-SAM ML surface binding orientation. These binding constants create a strong preference to extend the Scm-SAM block with Scm-SAM subunits and the Ph-SAM block with Ph-SAM subunits (Fig. 2c). Thus, the differential affinities favor a single joint copolymer model. Crystal Structure of the Ph-SAM/Scm-SAM Complex—To develop a detailed structural model of the copolymer, we determined a crystal structure of the Ph-SAM and Scm-SAM complex in the high affinity binding orientation. Crystals of the complex were grown using the EH surface Ph-SAM mutant, L69R, and the ML surface Scm-SAM double mutant, L52R/M57R. The structure was solved using multiwavelength anomalous diffraction phasing using the fully incorporated Se-Met residues of the Ph-SAM L69R protein and refined to and Rfree of 0.249 at 1.80 Å resolution. There are two complexes in the asymmetric unit that are joined by a disulfide bond between the Ph-SAM domains (Fig. 3a). To form the disulfide bond, the loop region housing the cysteine in one Ph-SAM reaches out and engages the same cysteine in the other Ph-SAM, which undergoes minimal conformational changes. This disulfide is unlikely to be biologically relevant given the strongly reducing environment in the cell. The interaction between Ph-SAM and Scm-SAM is very similar to the polymeric interfaces of the individual polymer structures, involving the ML and EH surfaces on the proteins. Like both the Ph-SAM and Scm-SAM polymers, the interaction can be divided into apolar and polar domains (see Fig. 3b). A detailed view of the hydrogen bonding and salt bridge interactions in the Ph-SAM/Scm-SAM complex is shown in Fig. 3b. A schematic illustration of the polar interactions for all the SAM interfaces is shown in Fig. 3c. In the hetero- and homo-SAM interfaces, the identical apolar and polar residues are used, although the specific interactions are somewhat different. With the structural information available for the individual Ph-SAM and Scm-SAM polymers along with that of their complex, a three-dimensional model of the copolymer structure can be easily generated by overlaying the individual polymer structures over the corresponding SAM domains in the Ph-SAM/Scm-SAM complex structure (Fig. 4). As a result of the similarities between binding surfaces of Ph-SAM and Scm-SAM, the model shows minimal deviation at the copolymer joint from the polymer axes of both individual polymers. Scm Polymer Structure Explains Biological Effects—Our results clearly demonstrate that Scm-SAM can form a polymer in vitro, and there is considerable evidence that the same polymer is an important aspect of the biological function of Scm. First, the high affinity of the intersubunit interaction is a strong indication that polymerization is a normal function of Scm-SAM in vivo. Second, it is hard to see how polymerization could be an in vitro artifact, because similar polymer architectures have now been seen for SAM domains from three divergent transcriptional repressors (TEL, Ph, and Scm) (7Kim C.A. Phillips M.L. Kim W. Gingery M. Tran H.H. Robinson M.A. Faham S. Bowie J.U. EMBO J. 2001; 20: 4173-4182Crossref PubMed Scopus (206) Google Scholar, 8Kim C.A. Gingery M. Pilpa R.M. Bowie J.U. Nat. Struct. Biol. 2002; 9: 453-457PubMed Google Scholar). Moreover, polymer blocking mutations in TEL render the protein unable to repress transcription (26Wood L.D. Irvin B.J. Nucifora G. Luce K.S. Hiebert S.W. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 3257-3262Crossref PubMed Scopus (101) Google Scholar), suggesting that polymerization is required for repressive function in TEL. Third overexpression of an isolated Scm-SAM generates an Scm defect in vivo (17Peterson A.J. Mallin D.R. Francis N.J. Ketel C.S. Stamm J. Voeller R.K. Kingston R.E. Simon J.A. Genetics. 2004; 167: 1225-1239Crossref PubMed Scopus (55) Google Scholar). It is easy to envision that an overabundance of the isolated SAM domain could infiltrate endogenous Scm polymers. Finally, Peterson et al. (17Peterson A.J. Mallin D.R. Francis N.J. Ketel C.S. Stamm J. Voeller R.K. Kingston R.E. Simon J.A. Genetics. 2004; 167: 1225-1239Crossref PubMed Scopus (55) Google Scholar) identified a set of mutations in Scm-SAM that failed to self-associate. When these mutants were introduced into the full-length Scm protein, they failed to complement Scm mutants in Drosophila. The SAM domain mutants that fail to self-associate and cannot rescue Scm mutant flies are readily rationalized by our polymer structure. Peterson et al. (17Peterson A.J. Mallin D.R. Francis N.J. Ketel C.S. Stamm J. Voeller R.K. Kingston R.E. Simon J.A. Genetics. 2004; 167: 1225-1239Crossref PubMed Scopus (55) Google Scholar) characterized five mutants that were found to be defective both in vitro and in vivo: I45T, G47D, M59Δ, M62R, and K71E (our numbering). As can be seen in Fig. 5, the sites of the mutations are localized to the interface seen in our Scm polymer structure. Ile-45 and Gly-47 are found in the ML binding surface. Both residues are highly buried in the monomer structure (Ile-45 is 91% buried, and Gly-47 is 100% buried). The I45T and G47D mutations are therefore likely to distort the structure of this critical binding surface. Met-59 and Met-62 are both located on helix 4, which contributes to both the ML and EH binding surfaces. Thus, helix 4 is a particularly important region for polymer formation. Moreover, Met-59 is an important hydrophobic residue in the interface. Deletion of Met-59 would therefore remove an important contribution to the interface and would necessarily distort the local structure. Met-62 is 98% buried in the monomer, making it difficult to accommodate the M62R substitution without some sort of structural distortion. Finally, Lys-71 is located on the EH interface and makes a salt bridge across the polymer interface to Asp-46. Thus a K71E mutation would eliminate this interaction and introduce unfavorable electrostatic repulsion. Overall, the results argue that the polymer structure we observed is biologically relevant. Regulating Polymerization—SAM domain polymerization must be regulated in some fashion to facilitate complex assembly and disassembly. Yan, a member of the Ets family of transcription factors, contains a SAM domain that polymerizes in the same fashion as Ph- and Scm-SAM as well as the closely related ortholog of Yan, TEL-SAM (9Qiao F. Song H. Kim C.A. Sawaya M.R. Hunter J.B. Gingery M. Rebay I. Courey A.J. Bowie J.U. Cell. 2004; 118: 163-173Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). Yan-SAM can be depolymerized via its interaction with the SAM domain of its regulator, Mae. Mae-SAM binds to a polymerization interface of Yan-SAM with 1000-fold greater binding energy than Yan-SAM has with itself thereby effectively competing away Yan-SAM self-association and ultimately leading to the down-regulation of Yan activity (9Qiao F. Song H. Kim C.A. Sawaya M.R. Hunter J.B. Gingery M. Rebay I. Courey A.J. Bowie J.U. Cell. 2004; 118: 163-173Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). Regulation of either Ph- or Scm-SAM polymer by each other in the same fashion as Yan/Mae appears unlikely because Ph/Scm-SAM lack the large disparity in binding affinities. It is possible that polymerization is regulated by some still unidentified Mae-like protein that can cap Scm and Ph polymers, or that SAM polymerization is regulated internally by another domain with Scm and Ph. Work on TEL raises the intriguing possibility that polymerization is regulated by covalent modification with small ubiquitin-like modifier (SUMO) (26Wood L.D. Irvin B.J. Nucifora G. Luce K.S. Hiebert S.W. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 3257-3262Crossref PubMed Scopus (101) Google Scholar, 27Chakrabarti S.R. Sood R. Nandi S. Nucifora G. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 13281-13285Crossref PubMed Scopus (102) Google Scholar). TEL is sumoylated at a lysine residue at the edge of the polymeric binding interface. Examination of the site of modification in the context of the TEL polymer structure strongly suggests that polymer formation and sumoylation are mutually incompatible. Thus, we would expect SUMO to disrupt TEL-SAM polymers. In this light, it is interesting to note that the polycomb group protein, Pc2, is a SUMO-ligating enzyme, indicating that sumoylation plays an important role in PcG function (28Kagey M.H. Melhuish T.A. Wotton D. Cell. 2003; 113: 127-137Abstract Full Text Full Text PDF PubMed Scopus (462) Google Scholar). Moreover, sumoylation of the Caenorhabditis elegans PcG protein SOP-2 is essential for Hox gene repression (29Zhang H. Smolen G.A. Palmer R. Christoforou A. van den Heuvel S. Haber D.A. Nat. Genet. 2004; 36: 507-511Crossref PubMed Scopus (66) Google Scholar). Both Drosophila Ph- and Scm-SAM possess potential sumoylation sites, but there is still no evidence for sumoylation of these proteins. A Ph/Scm Copolymer—Ph and Scm are known to bind to each other (15Peterson A.J. Kyba M. Bornemann D. Morgan K. Brock H.W. Simon J. Mol. Cell. Biol. 1997; 17: 6683-6692Crossref PubMed Scopus (127) Google Scholar) and cooperate in their repressive functions (19Roseman R.R. Morgan K. Mallin D.R. Roberson R. Parnell T.J. Bornemann D.J. Simon J.A. Geyer P.K. Genetics. 2001; 158: 291-307Crossref PubMed Google Scholar). Our findings that Scm-SAM and Ph-SAM both form polymers argues that they must interact in the form of a copolymer. Measurements of binding affinities in different orientations demonstrate that one of the possible joints between the two polymers is strongly preferred. This suggests that PcG complexes involving Ph and Scm would tend to form separate domains on chromatin, by means of a single joint copolymer depicted in Fig. 4 (although some intermixing of Ph and Scm is possible). If so, we would envision a domain of Ph complexes, which are known to localize within a few kilobases around a polycomb response element, extended by Scm complexes. This hypothesis is consistent with previously reported observations. First, the repressive function of Scm artificially tethered to a DNA binding site depended on the presence of the SAM domain and was enhanced by the presence of Ph (19Roseman R.R. Morgan K. Mallin D.R. Roberson R. Parnell T.J. Bornemann D.J. Simon J.A. Geyer P.K. Genetics. 2001; 158: 291-307Crossref PubMed Google Scholar). Secondly, in a Drosophila PRC1 complex, Scm co-purified with the other members including Ph and was thus originally identified as a member of the complex (30Shao Z. Raible F. Mollaaghababa R. Guyon J.R. Wu C.T. Bender W. Kingston R.E. Cell. 1999; 98: 37-46Abstract Full Text Full Text PDF PubMed Scopus (675) Google Scholar). Subsequent experiments, however, showed smaller amounts of Scm compared with the other members, Pc, Psc, dRING1, and Ph (18Saurin A.J. Shao Z. Erdjument-Bromage H. Tempst P. Kingston R.E. Nature. 2001; 412: 655-660Crossref PubMed Scopus (324) Google Scholar). From our copolymer model, we would expect variable amounts of Scm associated with a core domain of Ph complexes. Third, an analysis of regulatory DNA elements suggested the site of action of Scm is adjacent to the site of action of Psc, a member of the PRC1 complex along with Ph (31Tillib S. Petruk S. Sedkov Y. Kuzin A. Fujioka M. Goto T. Mazo A. Mol. Cell. Biol. 1999; 19: 5189-5202Crossref PubMed Google Scholar). Scm function extending over an adjacent site is exactly what is expected from our single junction copolymer model. Our results strongly argue that polymerization plays an important role in Ph and Scm function. The biological implications of these findings require further investigation, but it is reasonable to suggest that polymerization facilitates the spreading of PcG complexes along the chromosome. Although Ph is found localized around polycomb response elements, the location of Scm in repressed genes is not known. Scm is capable of long range repression (19Roseman R.R. Morgan K. Mallin D.R. Roberson R. Parnell T.J. Bornemann D.J. Simon J.A. Geyer P.K. Genetics. 2001; 158: 291-307Crossref PubMed Google Scholar), and our results suggest that Scm could be utilized for the extension of repression outside the immediate region of the polycomb response element. We thank Frank Laski for providing the Drosophila cDNA library and Aaron Chamberlain, Hoang Tran, Feng Qiao, and Sanguk Kim for helpful comments on the manuscript.