mapping the protein conformational landscape with adaptive probabilistic search

1
2032-Pos Board B18 Molecular Replacement for Multi-Domain Structures Using Packing Models Yan Yan, Gregory S. Chirikjian. Molecular replacement (MR) is frequently used to obtain phase information for a unit cell packed with a macromolecule of unknown structure. The goal of MR searches is to place a homologous/similar molecule in the unit cell so as to max- imize the correlation with x-ray diffraction data. MR software packages typi- cally perform rotation and translation searches separately. This works quite well for single-domain proteins. However, for multi-domain structures and complexes, computational requirements can become prohibitive and the de- sired peaks can become hidden in a noisy landscape. The main contribution of our approach is that computationally expensive MR searches in continuous configuration space are replaced by a search on a rela- tively small discrete set of representative packing arrangements of a multi- rigid-body model. These arrangements are generated by minimizing a potential function that forces the model conformations to separate from each other and not overlap within the unit cell. This is done before computing Patterson cor- relations rather than only performing collision checks when evaluating the fea- sibility of peaks. The list of feasible arrangements is short because packing constraints together with unit cell symmetry and geometry impose strong constraints. In numerical trials, we found that a candidate from the feasible set is usually similar to the arrangement of the target structure within the unit cell. To further improve the accuracy, a refined search can be performed in the neighborhood of this packing arrangement. Our approach is demonstrated with multi-domain models in silico for both 2D and 3D, with ellipsoids representing both the do- mains of the model and target structures. Our results show that an approximate phase can be found with 5 percent error and significantly improved search speed. We acknowledge NIH Grant R01GM075310 for the support of this work and Dr. E. Lattman for useful discussions. 2033-Pos Board B19 Adding Dynamical Insight to X-Ray Images by Solution and Crystal Molecular Dynamics Simulation Logan S. Ahlstrom, Osamu Miyashita. X-ray crystallography is the most robust technique for protein structure deter- mination. However, this method is still largely a trial-and-error process of changing solution conditions and relies on unpredictable protein-protein inter- actions. Consequently, crystal packing may select a conformation unrepresen- tative of the physiologically relevant form of a protein and there are many examples in which multiple solved structures exist for the same protein. Here we compare solution and crystal lattice molecular dynamics (MD) sim- ulations 1 to add dynamical insight to the interpretation of X-ray images. As a model system we consider the dimeric l Cro transcription factor whose crystal structures range from a closed DNA-free conformation to an open DNA-bound form. Free energy profiles reporting on the conformational space sampled by the dimer in solution reveal that both states are accessible but that the closed form may be slightly more stable 2 . Subsequent crystal MD simu- lations were performed to establish how mutation within the dimer could have stabilized a DNA-free open conformation in contrast to the wild-type apo closed form. Both structures were simulated with wild-type as well as mutant neighbors to study a variation of crystal environment. Applying the MMPBSA approach we calculated the relative stabilities of crystal contact re- gions in the mutant and wild-type lattices. Moreover, the packing arrange- ment in the mutant lattice may have prevented the formation of an intersubunit salt bridge that stabilizes the closed conformation. These results suggest that differences in crystal packing due to mutation affected l Cro di- mer conformation in the lattice. Our methods of performing crystal MD sim- ulations may also improve the interpretation of other protein X-ray images to aid in establishing structure-function relationships. [1] Vorontsov, I.I. and Miyashita, O. (2009) Biophys. J., 97, 2532-2540. [2] Ahlstrom, L.S. and Miyashita, O., in progress. 2034-Pos Board B20 Mapping the Protein Conformational Landscape with Adaptive Probabi- listic Search Brian Olson, Amarda Shehu. Structural characterization of the protein native state is often the key to better understanding protein function. The conformations that comprise the native state reside in the lowest-energy basin of a funnel-like energy landscape. Dis- covering and populating this basin with native conformations in silico de- mands powerful search algorithms that can navigate high-dimensional conformational spaces. Even short protein chains have many degrees of free- dom. Additionally, semi-empirical energy functions employed to guide the conformational search may introduce slight distortions to the true energy land- scape, often leading to rough landscapes rich in local minima. A successful search algorithm must balance the competing goals of sampling a diverse rep- resentation of the landscape with the need to further populate promising en- ergy minima. We present a novel algorithm which effectively balances the goals of explora- tion and exploitation through the use of projection layers. A geometric layer keeps track of the structural diversity in the explored conformational space, and an energetic layer determines the relevance of a computed conformation for the native state. The algorithm conducts a probabilistic search of a coarse-grained conformational space, maintaining a representative ensemble of computed conformations. A probabilistic weighting function over the projec- tion layers determines where to guide the search in the conformational space by balancing coverage of conformational space with population of lower-energy levels. We have compiled an extensive list of structurally-diverse proteins on which we apply our algorithm. Our results show that the algorithm efficiently yields native-like coarse-grained conformations of diverse small-to-medium size pro- teins of alpha, beta, and alpha/beta folds. The conformational ensemble main- tained by the algorithm provides a representative map of the conformational landscape of each protein. The lowest-energy conformations in this map cap- ture the native state and reproduce well the known native structures upon fur- ther refinement with all-atom energy functions. 2035-Pos Board B21 Rigidity Analysis Identifies Common Features of Allostery Andrew J. Rader, Stephen M. Brown. Allosteric proteins are remarkable in that they display a dramatic change in chemical affinity at a catalytic site in response to ligand binding at some other distal site. Thus allostery is extremely important biologically as a regulatory mechanism for molecular concentrations in many cellular processes. Structural comparisons of allosteric proteins resolved in both inactive and active states in- dicate that a variety of structural rearrangement and changes in motions may contribute to general allosteric behavior. We utilize a novel examination of al- lostery using rigidity analysis of the underlying graph representing a protein structure. Our analysis of paired allosteric proteins resolved in the tense (T) and relaxed (R) states indicate a general global change in rigidity among the two states for 60% of the cases. Typically this change indicates the R state is more rigid than the T state. Different functional classes of allosteric enzymes display departure from this trend. Most notably, signaling proteins tend not to have a change in flexibility. In general it is expected that the coupling of cat- alytic and regulatory sites is responsible for allosteric behavior. We used a set of allosteric proteins with well-defined heterotropic interactions to test the hy- pothesis that catalytic and effector sites are structurally coupled. Within this set a rigid path connecting the effector and catalytic sites was observed in 69% of the structures. Such results indicate that rigidity is a possible means by which these distal sites communicate to each other and so contribute to allosteric reg- ulation. This rigidity characterization provides a descriptor to distinguish among different classes of allostery. 2036-Pos Board B22 Binding Site Flexibility in the CD44:Hyaluronan Protein:Carbohydrate Interaction Francis W. Jamison, Theresa J. Foster, Jacob A. Barker, Ronald D. Hills, Olgun Guvench. The CD44:hyaluronan protein:carbohydrate binding interaction is an impor- tant component of inflammation and cancer cell invasiveness and metastasis. Association of a key arginine sidechain in the beta1-alpha1 loop of the bind- ing site with the carbohydrate ligand has been postulated to lead to a high- affinity state of the complex based on atomic-resolution x-ray crystal studies. NMR studies point to coupling between an order-to-disorder transition of the beta9-alpha3 region of the CD44 protein, which is distant from the binding site, and the formation of the high-affinity state; however, this region of the protein is well-ordered in all crystal structures. All-atom explicit-solvent molecular dynamics simulations of the CD44:hyaluronan complex revealed beta1-alpha1 loop backbone instability when arginine is in close association with hyaluronan, whereas the loop was stable in the absence of sidechain association. Additionally, a conformational transition in the backbone phi angle of a single residue in the beta1-alpha1 loop was demonstrated to be suf- ficient to induce reversible arginine sidechain:hyaluronan association, leading to the description of a molecular switching mechanism. The instability of the beta1-alpha1 loop in the context of a well-ordered beta9-alpha3 region sug- gests possible allosteric coupling between the binding site conformation and the order-to-disorder transition. Tuesday, March 8, 2011 377a

Upload: amarda

Post on 30-Dec-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Tuesday, March 8, 2011 377a

2032-Pos Board B18Molecular Replacement for Multi-Domain Structures Using PackingModelsYan Yan, Gregory S. Chirikjian.Molecular replacement (MR) is frequently used to obtain phase information fora unit cell packed with a macromolecule of unknown structure. The goal of MRsearches is to place a homologous/similar molecule in the unit cell so as to max-imize the correlation with x-ray diffraction data. MR software packages typi-cally perform rotation and translation searches separately. This works quitewell for single-domain proteins. However, for multi-domain structures andcomplexes, computational requirements can become prohibitive and the de-sired peaks can become hidden in a noisy landscape.The main contribution of our approach is that computationally expensive MRsearches in continuous configuration space are replaced by a search on a rela-tively small discrete set of representative packing arrangements of a multi-rigid-body model. These arrangements are generated by minimizing a potentialfunction that forces the model conformations to separate from each other andnot overlap within the unit cell. This is done before computing Patterson cor-relations rather than only performing collision checks when evaluating the fea-sibility of peaks. The list of feasible arrangements is short because packingconstraints together with unit cell symmetry and geometry impose strongconstraints.In numerical trials, we found that a candidate from the feasible set is usuallysimilar to the arrangement of the target structure within the unit cell. To furtherimprove the accuracy, a refined search can be performed in the neighborhood ofthis packing arrangement. Our approach is demonstrated with multi-domainmodels in silico for both 2D and 3D, with ellipsoids representing both the do-mains of the model and target structures. Our results show that an approximatephase can be found with 5 percent error and significantly improved searchspeed.We acknowledge NIH Grant R01GM075310 for the support of this work andDr. E. Lattman for useful discussions.

2033-Pos Board B19Adding Dynamical Insight to X-Ray Images by Solution and CrystalMolecular Dynamics SimulationLogan S. Ahlstrom, Osamu Miyashita.X-ray crystallography is the most robust technique for protein structure deter-mination. However, this method is still largely a trial-and-error process ofchanging solution conditions and relies on unpredictable protein-protein inter-actions. Consequently, crystal packing may select a conformation unrepresen-tative of the physiologically relevant form of a protein and there are manyexamples in which multiple solved structures exist for the same protein.Here we compare solution and crystal lattice molecular dynamics (MD) sim-ulations1 to add dynamical insight to the interpretation of X-ray images. Asa model system we consider the dimeric l Cro transcription factor whosecrystal structures range from a closed DNA-free conformation to an openDNA-bound form. Free energy profiles reporting on the conformational spacesampled by the dimer in solution reveal that both states are accessible but thatthe closed form may be slightly more stable2. Subsequent crystal MD simu-lations were performed to establish how mutation within the dimer couldhave stabilized a DNA-free open conformation in contrast to the wild-typeapo closed form. Both structures were simulated with wild-type as well asmutant neighbors to study a variation of crystal environment. Applying theMMPBSA approach we calculated the relative stabilities of crystal contact re-gions in the mutant and wild-type lattices. Moreover, the packing arrange-ment in the mutant lattice may have prevented the formation of anintersubunit salt bridge that stabilizes the closed conformation. These resultssuggest that differences in crystal packing due to mutation affected l Cro di-mer conformation in the lattice. Our methods of performing crystal MD sim-ulations may also improve the interpretation of other protein X-ray images toaid in establishing structure-function relationships. [1] Vorontsov, I.I. andMiyashita, O. (2009) Biophys. J., 97, 2532-2540. [2] Ahlstrom, L.S. andMiyashita, O., in progress.

2034-Pos Board B20Mapping the Protein Conformational Landscape with Adaptive Probabi-listic SearchBrian Olson, Amarda Shehu.Structural characterization of the protein native state is often the key to betterunderstanding protein function. The conformations that comprise the nativestate reside in the lowest-energy basin of a funnel-like energy landscape. Dis-covering and populating this basin with native conformations in silico de-mands powerful search algorithms that can navigate high-dimensionalconformational spaces. Even short protein chains have many degrees of free-

dom. Additionally, semi-empirical energy functions employed to guide theconformational search may introduce slight distortions to the true energy land-scape, often leading to rough landscapes rich in local minima. A successfulsearch algorithm must balance the competing goals of sampling a diverse rep-resentation of the landscape with the need to further populate promising en-ergy minima.We present a novel algorithm which effectively balances the goals of explora-tion and exploitation through the use of projection layers. A geometric layerkeeps track of the structural diversity in the explored conformational space,and an energetic layer determines the relevance of a computed conformationfor the native state. The algorithm conducts a probabilistic search ofa coarse-grained conformational space, maintaining a representative ensembleof computed conformations. A probabilistic weighting function over the projec-tion layers determines where to guide the search in the conformational space bybalancing coverage of conformational space with population of lower-energylevels.We have compiled an extensive list of structurally-diverse proteins on whichwe apply our algorithm. Our results show that the algorithm efficiently yieldsnative-like coarse-grained conformations of diverse small-to-medium size pro-teins of alpha, beta, and alpha/beta folds. The conformational ensemble main-tained by the algorithm provides a representative map of the conformationallandscape of each protein. The lowest-energy conformations in this map cap-ture the native state and reproduce well the known native structures upon fur-ther refinement with all-atom energy functions.

2035-Pos Board B21Rigidity Analysis Identifies Common Features of AllosteryAndrew J. Rader, Stephen M. Brown.Allosteric proteins are remarkable in that they display a dramatic change inchemical affinity at a catalytic site in response to ligand binding at some otherdistal site. Thus allostery is extremely important biologically as a regulatorymechanism for molecular concentrations in many cellular processes. Structuralcomparisons of allosteric proteins resolved in both inactive and active states in-dicate that a variety of structural rearrangement and changes in motions maycontribute to general allosteric behavior. We utilize a novel examination of al-lostery using rigidity analysis of the underlying graph representing a proteinstructure. Our analysis of paired allosteric proteins resolved in the tense (T)and relaxed (R) states indicate a general global change in rigidity among thetwo states for 60% of the cases. Typically this change indicates the R state ismore rigid than the T state. Different functional classes of allosteric enzymesdisplay departure from this trend. Most notably, signaling proteins tend notto have a change in flexibility. In general it is expected that the coupling of cat-alytic and regulatory sites is responsible for allosteric behavior. We used a setof allosteric proteins with well-defined heterotropic interactions to test the hy-pothesis that catalytic and effector sites are structurally coupled. Within this seta rigid path connecting the effector and catalytic sites was observed in 69% ofthe structures. Such results indicate that rigidity is a possible means by whichthese distal sites communicate to each other and so contribute to allosteric reg-ulation. This rigidity characterization provides a descriptor to distinguishamong different classes of allostery.

2036-Pos Board B22Binding Site Flexibility in the CD44:Hyaluronan Protein:CarbohydrateInteractionFrancis W. Jamison, Theresa J. Foster, Jacob A. Barker, Ronald D. Hills,Olgun Guvench.The CD44:hyaluronan protein:carbohydrate binding interaction is an impor-tant component of inflammation and cancer cell invasiveness and metastasis.Association of a key arginine sidechain in the beta1-alpha1 loop of the bind-ing site with the carbohydrate ligand has been postulated to lead to a high-affinity state of the complex based on atomic-resolution x-ray crystal studies.NMR studies point to coupling between an order-to-disorder transition of thebeta9-alpha3 region of the CD44 protein, which is distant from the bindingsite, and the formation of the high-affinity state; however, this region of theprotein is well-ordered in all crystal structures. All-atom explicit-solventmolecular dynamics simulations of the CD44:hyaluronan complex revealedbeta1-alpha1 loop backbone instability when arginine is in close associationwith hyaluronan, whereas the loop was stable in the absence of sidechainassociation. Additionally, a conformational transition in the backbone phiangle of a single residue in the beta1-alpha1 loop was demonstrated to be suf-ficient to induce reversible arginine sidechain:hyaluronan association, leadingto the description of a molecular switching mechanism. The instability of thebeta1-alpha1 loop in the context of a well-ordered beta9-alpha3 region sug-gests possible allosteric coupling between the binding site conformation andthe order-to-disorder transition.