Fri Jul 17 14:24:01 PDT 1998 Kevin Karplus Although this is a "synthetic sequence", wu-blast finds a fairly high similarity to 1cos[ABC] "Coiled SER" and 1coi " Designed Trimeric Coiled Coil". These are both synthetic triple-stranded colied-coils, and so don't appear in the fssp database. The next hit (much weaker) is 1abz, another synthetic peptide, supposedly a helix-turn-helix. The first real protein is 1aq5[ABC] cartilage matrix protein fragment. Double-blast finds nothing, mainly because nothing in NRP matches close enough to get E=0.00005. Perhaps I need to loosen the threshold to 0.01 for this protein? Even that isn't enough! Try 0.4, which should pick up the 1aq5[ABC] match. This finally picks up the same matches as wu-blast. The problem was in the intermediate E value, which gets a min of 0.015 for 1cosA---I forgot to scale it for the increased size of NRP. The short length an paucity of homologs may require changing thresholds for the t98 method also, though the t73.remote* alignments may be ok. Some homologs actually come in on t73.t98_5, and several are present by t73.t98_6, indicating that the synthetic sequence is not completely devoid of natural homologs. The top-scoring sequence with t73.t98_6 is 1coi (not 1cos[ABC]), perhaps because 1coi is more similar to the naturally occurring sequences? The best template model is 1aq5A, since it is the closest natural protein (we didn't build template models for the synthetics). Even with summing, 1aq5A does not do as well as 1coi. I THINK the designers were trying to get a turn in the helix (like in 1aq5A), but I THINK they've just got a longer coiled-coil. Putting in cystines like in 1aq5A might get a turn, but then the helices shouldn't look so much like coiled-coils. So, do we predict an up-up-down interaction like 1cosA, or all parallel like 1coi? Since the proline will interrup the helix somewhat (but not enough, I think to force a turn), I predict a trimeric coiled-coil with parallel arrangement, using 1coi-first-hand.a2m for the first half and t73-1coi-joint.pw for the second half. This way all the prolines can line up in the structure, making a single coordinated interruption. 21 July 1998 t73 is a tough one. It is obviously a synthetic coiled-coil, but no one will miss that. The hard part is guessing whether it is dimeric, trimeric, or tetrameric, and whether the arrangment is parallel or anti-parallel. There are no natural coiled-coils in PDB that are very close. The two closest matches in PDB are both synthetic trimeric coiled-coils, one parallel and one 2-up, 1-down (though that seems to be at least partly a question of the pH at which the crystal was made, according to the papers). Unless Christian finds something suggesting a different solution in the papers, I'm going to guess a trimeric, parallel coiled-coil, though the kink in the middle may be throwing me off. Perhaps I should look for the kink sequence in PDB, and see what sort of helix it can be expected to form. Perhaps the best match to the middle is from the end of 1etc: t73 LQALKKEGFSPEELAAL 1etc CDLQSLL GYTPEELHAMLDVKPDAD >1etc mol:protein length:110 Solution Structure: Ets Domain From Murine__ GSGPIQLWQFLLELLTDKSCQSFISWTGDGWEFKLSDPDEVARRWGKRKNKPKMNYEKLSRGLRYYYDKNIIHKT AGKRYVYRFVCDLQSLLGYTPEELHAMLDVKPDAD The aldehyse reductases (1ae4,2alr,1cwn) have TFSPEEM (probably before a helix). The myrosinases (1myr,2myr) have TFSPEET (possibly before a helix). Retinol-binding protein (1aqb) has HGFSPEV, (1erb, 1fel, 1fem, 1fen, 1hbp, 1hbq) has SGFSPEV t73 LQALKKEGFSPEELAAL Antithrombin has LVDLFSPEKSKL 1antI, 1antL, 2antI, 2antL, 1athA, 1athB, 1ttA, 1attB Wed Jul 22 14:15:06 PDT 1998 Breaking t73 into heptads yields: S LAALKSE LQALKKE GFSPEE LAALESE LQALEKK LAALKSK LQALKG Note the shorter "heptad" in the middle. For the Leucine cores to line up, this will have to make 2 turns in 6 residues instead of 7---that is make a 3-10 helix. With 7 heptads, we can't get an easy super-helical crystal, but the 3-10 helix doesn't require any twist to be taken up (it is exactly 3 residues per turn, not 3.6 twisted to reduce to 3.5), so maybe this is the equivalent of a 6-heptad repeat, which would be appropriate for a dimer or tetramer. Question: is GFSPEE likely to be a 3-10 helix? 23 July 1998 Here is my draft of comments for the results section. T0073 is obviously a synthetic coiled-coil, with leucines in the a and d positions of the heptads, with the third heptad replaced by GFSPEE. Since we know that the peptide does crystallize, the question is mainly one of the form for the crystal---is it a dimer, trimer, tetramer? Do the pieces stack so that the superhelical axes line up? Are the helices parallel or anti-parallel? What happens to the GFSPEE section? We believe that T0074 makes a 180-degree turn around the superhelical axis, forming a dimeric, parallel, coiled-coil with the superhelical axes lined up. (A tetramer is possible, but less likely.) The GFSPEE forms a 3-10 helix to allow the leucine cores on either side to line up. We did not find a coiled coil with exactly the structure we are predicting, and so have provided two alignments to 1coi, which is a parallel trimer. We base our conjecture on Ogihara, Weiss, et al's theory that coiled-coils will crystalize with aligned superhelices if there are 2, 4, or 6 heptads, and each heptad will add 30 degrees of turn around the superhelical axis. Since a 3-10 helix adds no extra twist to the helix (it is an almost exact 3 residues per turn), T0074 should have 180 degrees of superhelical twist to relieve the twist of 6 heptads. We prefer a parallel orientation so that the slightly larger rise of the 3-10 helices occurs in the same place in the superhelix. A more detailed prediction would try to predict which salt bridges and hydrogen bonds are formed. Other than the standard guesses (serine forming a hydrogen bond up one turn, and E and K in the e and g positions forming salt bridges to K and E in the adjacent position g or e position in the other helix), we have not attempted a detailed prediction of the bonding. Fri Jul 24 09:22:04 PDT 1998 I sent a message to Glenn Millhauser, asking for his comments on our prediction, as suggested by Leslie Grate. I also went to the PDB web site and picked up 23 sequences with length >40 that had "coiled-coil" in the text. From glennm@hydrogen.ucsc.edu Fri Jul 24 09:53:56 1998 Return-Path: glennm@hydrogen.ucsc.edu Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 24 Jul 1998 09:51:28 +0100 To: Kevin Karplus From: glennm@hydrogen.UCSC.EDU (Glenn Millhauser) Subject: Re: coiled-coil prediction for CASP I'll look at it and get back to you today. G >One of the prediction targets for the CASP3 contest is a synthetic >coiled-coil, and Leslie Grate suggested that I ask you for comments on >our prediction. If you have time today to look at this, and provide >any info, I'd greatly appreciate it. If you want to suggest a >different solution, we could add you as a collaborator on this prediction. > >Here is the sequence we were given (I've split it into heptads for you): > >S >LAALKSE >LQALKKE >GFSPEE >LAALESE >LQALEKK >LAALKSK >LQALKG > >Here is what I said about it: > > >T0073 is obviously a synthetic coiled-coil, with leucines in the a and >d positions of the heptads, with the third heptad replaced by GFSPEE. >Since we know that the peptide does crystallize, the question is >mainly one of the form for the crystal---is it a dimer, trimer, tetramer? >Do the pieces stack so that the superhelical axes line up? Are the >helices parallel or anti-parallel? What happens to the GFSPEE section? > >We believe that T0074 makes a 180-degree turn around the superhelical >axis, forming a dimeric, parallel, coiled-coil with the superhelical >axes lined up. (A tetramer is possible, but less likely.) The GFSPEE >forms a 3-10 helix to allow the leucine cores on either side to line >up. > >We did not find a coiled coil with exactly the structure we are >predicting, and so have provided two alignments to 1coi, which is a >parallel trimer. > >We base our conjecture on Ogihara, Weiss, et al's theory that >coiled-coils will crystalize with aligned superhelices if there are >2, 4, or 6 heptads, and each heptad will add 30 degrees of turn >around the superhelical axis. Since a 3-10 helix adds no extra twist >to the helix (it is an almost exact 3 residues per turn), T0074 should >have 180 degrees of superhelical twist to relieve the twist of 6 >heptads. We prefer a parallel orientation so that the slightly larger >rise of the 3-10 helices occurs in the same place in the superhelix. > >A more detailed prediction would try to predict which salt bridges and >hydrogen bonds are formed. Other than the standard guesses (serine >forming a hydrogen bond up one turn, and E and K in the e and g >positions forming salt bridges to K and E in the adjacent position g >or e position in the other helix), we have not attempted a detailed >prediction of the bonding. ------------------------------ | Glenn L. Millhauser | | glennm@hydrogen.ucsc.edu | | glennm@chemistry.ucsc.edu | | voice: (408) 459-2176 | | fax: (408) 459-2935 | ------------------------------ Glenn recommended reading the first paper below: From MELVYL@UCCMVSA.UCOP.EDU Fri Jul 24 10:20:58 1998 Return-Path: MELVYL@UCCMVSA.UCOP.EDU Date: Fri, 24 Jul 98 10:19:57 PDT From: Melvyl System To: karplus@cse.ucsc.edu Subject: (id: MKJ64508) MELVYL system mail result Search request: F PA ALBER, T# Search result: 11 citations in the Medline database Display: ABSTR 1. Gonzalez L Jr; Brown RA; Richardson D; Alber T. Crystal structures of a single coiled-coil peptide in two oligomeric states reveal the basis for structural polymorphism. Nature Structural Biology, 1996 Dec, 3(12):1002-9. (UI: 97102426) Abstract: Each protein sequence generally adopts a single native fold, but the sequence features that confer structural uniqueness are not well understood. To define the basis for structural heterogeneity, we determined the high resolution X-ray crystal structures of a single GCN4 leucine-zipper mutant (Asn 16 to aminobutyric acid) in both dimeric and trimeric coiled-coil conformations. The mutant sequence is accommodated in two distinct structures by forming similarly-shaped packing surfaces with different sets of atoms. The trimer structure, in comparison to a previously-characterized trimeric mutant with substitutions in eight core residues, shows that the twist of individual helices and the helix-helix crossing angles can vary significantly to produce the most favoured packing arrangement. 2. Gonzalez L Jr; Woolfson DN; Alber T. Buried polar residues and structural specificity in the GCN4 leucine zipper. Nature Structural Biology, 1996 Dec, 3(12):1011-8. (UI: 97102427) Abstract: A conserved asparagine (Asn 16) buried in the interface of the GCN4 leucine zipper selectively favours the parallel, dimeric, coiled-coil structure. To test if other polar residues confer oligomerization specificity, the structural effects of Gln and Lys substitutions for Asn 16 were characterized. Like the wild-type peptide, the Asn 16Lys mutant formed exclusively dimers. In contrast, Gln 16, despite its chemical similarity to Asn, allowed the peptide to form both dimers and trimers. The Gln 16 side chain was accommodated by qualitatively different interactions in the dimer and trimer crystal structures. These findings demonstrate that the structural selectivity of polar residues results not only from the burial of polar atoms, but also depends on the complementarity of the side-chain stereochemistry with the surrounding structural environment. 3. Cronk JD; Endrizzi JA; Alber T. High-resolution structures of the bifunctional enzyme and transcriptional coactivator DCoH and its complex with a product analogue. Protein Science, 1996 Oct, 5(10):1963-72. (UI: 97052967) Abstract: DCoH, the dimerization cofactor of hepatocyte nuclear factor 1 (HNF-1), functions as both a transcriptional coactivator and a pterin dehydratase. To probe the relationship between these two functions, the X-ray crystal structures of the free enzyme and its complex with the product analogue 7,8-dihydrobiopterin were refined at 2.3 A resolution. The ligand binds at four sites per tetrameric enzyme, with little apparent conformational change in the protein. Each active-site cleft is located in a subunit interface, adjacent to a prominent saddle motif that has structural similarities to the TATA binding protein. The pterin binds within an arch of aromatic residues that extends across one dimer interface. The bound ligand makes contacts to three conserved histidines, and this arrangement restricts proposals for the enzymatic mechanism of dehydration. The dihedral symmetry of DCoH suggests that binding to the dimerization domain of HNF-1 likely involves the superposition of two-fold rotation axes of the two proteins. 4. Gonzalez L Jr; Plecs JJ; Alber T. An engineered allosteric switch in leucine-zipper oligomerization. Nature Structural Biology, 1996 Jun, 3(6):510-5. (UI: 96227970) Abstract: Controversy remains about the role of core side-chain packing in specifying protein structure. To investigate the influence of core packing on the oligomeric structure of a coiled coil, we engineered a GCN4 leucine zipper mutant that switches from two to three strands upon binding the hydrophobic ligands cyclohexane and benzene. In solution these ligands increased the apparent thermal stability and the oligomerization order of the mutant leucine zipper. The crystal structure of the peptide-benzene complex shows a single benzene molecule bound at the engineered site in the core of the trimer. These results indicate that coiled coils are well-suited to function as molecular switches and emphasize that core packing is an important determinant of oligomerization specificity. 5. Petersen JM; Skalicky JJ; Donaldson LW; McIntosh LP; Alber T; Graves BJ. Modulation of transcription factor Ets-1 DNA binding: DNA-induced unfolding of an alpha helix. Science, 1995 Sep 29, 269(5232):1866-9. (UI: 96032787) Abstract: Conformational changes, including local protein folding, play important roles in protein-DNA interactions. Here, studies of the transcription factor Ets-1 provided evidence that local protein unfolding also can accompany DNA binding. Circular dichroism and partial proteolysis showed that the secondary structure of the Ets-1 DNA-binding domain is unchanged in the presence of DNA. In contrast, DNA allosterically induced the unfolding of an alpha helix that lies within a flanking region involved in the negative regulation of DNA binding. These findings suggest a structural basis for the intramolecular inhibition of DNA binding and a mechanism for the cooperative partnerships that are common features of many eukaryotic transcription factors. 6. Nautiyal S; Woolfson DN; King DS; Alber T. A designed heterotrimeric coiled coil. Biochemistry, 1995 Sep 19, 34(37):11645-51. (UI: 96018851) Abstract: Principles that guide folding of coiled coils were tested by designing three peptides that preferentially associate with each other to form a heterotrimeric coiled coil. The core positions of the designed helices contained residues that promote formation of trimeric coiled coils. Ionic interactions were employed to mediate heterospecificity, and negative design was used to favor formation of the heterotrimer over alternative arrangements. A program was written to select sequences that maximized the number of attractive interhelical interactions in a parallel heterotrimer and the number of repulsive electrostatic interactions in alternative species. Solution studies indicate that an equimolar mixture of the three peptides forms a helical trimer with high specificity and stability. These results validate the principles used to guide the design and suggest that the heterotrimer may serve as a useful, autonomous trimerization domain. 7. Alber TR; Scheidt KA; Fajman WA. Diffuse abdominal uptake of technetium-99m-HDP after colectomy in Gardner's syndrome. Journal of Nuclear Medicine, 1995 Sep, 36(9):1611-4. (UI: 95387136) Abstract: A 37-yr-old man presented with increasing abdominal girth and multiple palpable intra-abdominal masses 3 yr after colectomy for polyposis coli. Whole-body skeletal scintigraphy performed prior to laparotomy demonstrated diffuse abdominal uptake of 99mTc-HDP consistent with mesenteric fibromatosis confirmed at surgery. When diffuse abdominal uptake of skeletal imaging agents occurs in patients with prior colectomy for polyposis coli, mesenteric fibromatosis as a manifestation of Gardner's syndrome should be suspected. This case illustrates another cause of diffuse abdominal uptake of skeletal imaging agents. 8. Woolfson DN; Alber T. Predicting oligomerization states of coiled coils. Protein Science, 1995 Aug, 4(8):1596-607. (UI: 96060091) Abstract: An algorithm based on the profile method was developed that faithfully distinguishes between the amino acid sequences of dimeric and trimeric coiled coils. Normalized sequence profiles derived from nonhomologous, two- and three-stranded, coiled-coil sequences with unambiguous registers were used to assign dimer and trimer propensities to test sequences. The difference between the dimer and trimer profile scores accurately reflected the preferred oligomerization state. The method relied on two strategies that may be generally applicable to profile calculations--profile values of solvent-exposed residues and of amino acids that were underrepresented in the data-base were given zero weight. Differences between the dimer and trimer profiles revealed sequence patterns that match and extend experimental studies of oligomer specification. 9. Endrizzi JA; Cronk JD; Wang W; Crabtree GR; Alber T. Crystal structure of DCoH, a bifunctional, protein-binding transcriptional coactivator [published erratum appears in Science 1995 Jun 9;268(5216):1421]. Science, 1995 Apr 28, 268(5210):556-9. (UI: 95242099) Abstract: DCoH, the dimerization cofactor of hepatocyte nuclear factor-1, stimulates gene expression by associating with specific DNA binding proteins and also catalyzes the dehydration of the biopterin cofactor of phenylalanine hydroxylase. The x-ray crystal structure determined at 3 angstrom resolution reveals that DCoH forms a tetramer containing two saddle-shaped grooves that comprise likely macromolecule binding sites. Two equivalent enzyme active sites flank each saddle, suggesting that there is a spatial connection between the catalytic and binding activities. Structural similarities between the DCoH fold and nucleic acid-binding proteins argue that the saddle motif has evolved to bind diverse ligands or that DCoH unexpectedly may bind nucleic acids. 10. Harbury PB; Kim PS; Alber T. Crystal structure of an isoleucine-zipper trimer. Nature, 1994 Sep 1, 371(6492):80-3. (UI: 94352397) Abstract: Subunit oligomerization in many proteins is mediated by short coiled-coil motifs. These motifs share a characteristic seven-amino-acid repeat containing hydrophobic residues at the first (a) and fourth (d) positions. Despite this common pattern, different sequences form two-, three- and four-stranded helical ropes. We have investigated the basis for oligomer choice by characterizing variants of the GCN4 leucine-zipper dimerization domain that adopt trimeric or tetrameric structures in response to mutations at the a and d positions. We now report the high-resolution X-ray crystal structure of an isoleucine-containing mutant that folds into a parallel three-stranded, alpha-helical coiled coil. In contrast to the dimer and tetramer structures, the interior packing of the trimer can accommodate beta-branched residues in the most preferred rotamer at both hydrophobic positions. Compatibility of the shape of the core amino acids with the distinct packing spaces in the two-, three- and four-stranded conformations appears to determine the oligomerization state of the GCN4 leucine-zipper variants. 11. Zhang T; Bertelsen E; Alber T. Entropic effects of disulphide bonds on protein stability. Nature Structural Biology, 1994 Jul, 1(7):434-8. (UI: 95393176) Abstract: To measure the thermodynamic consequences of the reduction in the number of polypeptide-chain conformations that accompanies protein folding, we developed a method called loop permutation analysis. In this approach, the stabilizing contributions of three engineered disulphide bonds were compared in extended and circularly permutated mutants of phage T4 lysozyme. The observed differences in disulphide contributions, although qualitatively consistent with theoretical estimates, were not solely proportional to the differences in loop length. These findings suggest that in addition to the length of the chain, the polypeptide sequence may influence the energetic consequences of conformational restrictions. From karplus@cse.ucsc.edu Fri Jul 24 12:17:06 1998 Return-Path: karplus@cse.ucsc.edu Date: Fri, 24 Jul 1998 12:17:05 -0700 From: Kevin Karplus To: cbarrett@cse.ucsc.edu Cc: karplus@cse.ucsc.edu, glennm@hydrogen.UCSC.EDU, les@cse.ucsc.edu Subject: t73 CHANGE PREDICTION After talking with Glenn Millhauser (as suggested by Leslie), I want to change the prediction for T73. We need to add another segment to the prediction to describe the newly proposed turn, and here is the new text: T0073 obviously has 2 heptads of helix, a hexad GFSPEE, and 4 more heptads of helix. The helices are classic coiled-coil formers, with leucines in both the a and d positions of the heptad. The questions are 1) what happens with the GFSPEE? 2) is this a dimer, trimer, tetramer, ...? 3) if this is a single coild-coil, how do the peptides stack? Our first guess was that the GFSPEE would form a 3-10 helix, with the whole sequence forming a single dimeric, parallel coiled-coil with 180 degrees of twist around the superhelical axis, so that the peptides could stack along that axis. However, we were uncomfortable with the proline in the middle of a helix in solution, and looked for other possibilities. After consulting with Glenn Millhauser, whose first reaction was that the GFSPEE would form a turn, we found a similar turn in 1ETC (GYPTEEL), in which the ring of the tyrosine packs nicely against the leucine that starts a helix coming out of the turn. The length of the turn is about right for going from the exposed side of one helix to the buried side of the other in an anti-parallel coiled-coil. Glenn pointed out that substituting F for Y would make it more hydrophobic, and subsituting S for T would make that exposed residue more hydrophilic, further stabilizing the turn. This lead us to the conjecture that the structure is a long and a short helix packed as a coiled-coil with a turn between them. The peptides could dimerize with the long helices of one extending the short helices of the other (like sticky tails in DNA double helices). This arrangement nicely balances the dipole moments, and buries all the leucines in coiled-coils. Fri Jul 24 14:04:06 PDT 1998 I looked over the paper Woolfson DN; Alber T. Predicting oligomerization states of coiled coils. Protein Science, 1995 Aug, 4(8):1596-607. and tried applying their method to T73. It strongly supports a dimeric prediction, Using just the 31 scores of page 1601 (taken from Table 5 on page 1602): LAALKSE 3.95 LQALKKE 3.95 LAALESE 5.41 LQALEKK 2.94 LAALKSK 2.15 LQALKG 2.15 ---- log(1142.66) = 3.06, well into the dimer region.