Poster Session

To sign up, select the poster session checkbox on the registration page and enter your title and abstract in the text box.

Poster Exhibition Guidelines: Each exhibitor will be given a 60" wide x 40" tall space on which to mount their poster with push pins. Your poster should fit within this space and be readable from a distance of ~3 feet.

Check this space for updates. We will display new poster titles and abstracts here as we receive them.

"Solution NMR structural and dynamic analysis of a 60 kDa complex between putidaredoxin and cytochrome P450cam"

Prashansa Agrawal
The Universtiy of Tennessee, Knoxville

The 60 kDa complex formed between [2Fe-2S] ferredoxin putidaredoxin (Pdx) and cytochrome P450cam (CYP101) from the bacterium Pseudomonas putida has been investigated by high-resolution solution NMR spectroscopy. Pdx serves as both the physiological reductant and effector for CYP101 in the enzymatic reaction involving conversion of substrate camphor to 5-exo-hydroxy-camphor. In order to determine an experimental structure for the oxidized Pdx-CYP101 complex, a combined approach using orientational data on the two proteins derived from residual dipolar couplings and distance restraints from site-specific spin labeling of Pdx has been applied. Spectral changes for residues in and near the paramagnetic metal cluster region of Pdx in complex with CYP101 have also been mapped for the first time using 15N and 13C NMR spectroscopy, leading to direct identification of the residues strongly affected by CYP101 binding. The new NMR structure of the Pdx-CYP101 complex agrees well with previous mutagenesis and biophysical studies involving residues at the binding interface such as formation of salt bridge between Asp38 of Pdx and Arg112 of CYP101, while at the same time identifying key features different from earlier modeling studies. Analysis of the binding interface of the complex reveals that the side-chain of Trp106, the C-terminal residue of Pdx and critical for binding to CYP101 is located across from the heme-binding loop of CYP101 and forms non-polar contacts with several residues in the vicinity of heme group on CYP101, explaining its importance in complex formation with CYP101. Characterization of dynamic behavior of backbone amides of Pdx in the complex via T1, T2 relaxation experiments also indicates a change in mobility of residues forming the binding interface, suggesting a dynamic-modulated mode of binding between the two proteins.

"De novo high-resolution protein structure determination from sparse spin-labeling EPR data"

Nathan Alexander
Vanderbilt University

Nuclear magnetic resonance (NMR) and X-ray crystallography have resulted in over 99.99% of all protein structures currently deposited in the protein data base, but membrane proteins remain difficult targets for both techniques. Hence, membrane protein structure determination remains one key challenge in structural biology which must be addressed with alternative techniques. Accordingly, electron paramagnetic resonance (EPR) can be used to investigate membrane proteins that are not amenable to NMR and X-ray crystallography by measuring residue exposure to membrane or solvent and determining intramolecular distances of up to 50?. However, EPR data alone cannot produce a high-resolution protein structure. By coupling EPR datasets with computational protein structure prediction, atomic detail protein structure determination of previously inaccessible proteins becomes feasible.

"Binary and ternary complexes of the Y-class Sulfolobus solfataricus DNA polymerase Dpo4 with a template containing the aflatoxin B1 N7-dG-formamidopyrimidine adduct"

Surajit Banerjee
Vanderbilt University

Aflatoxin B1 (AFB1), one of the most potent known mutagens, forms an unstable N7-guanine adduct in DNA. Spontaneous hydrolysis of the deoxyguanosine N7-AFB1 adduct gives the persistent and mutagenic formamidopyrimidine (FAPY) derivative, which exists as a mixture of isomers upon ring opening. These include the alpha and beta anomers at C1' of the lesion deoxyribose. Previous study suggested that mutagenicity was attributed to the beta anomer, while the alpha anomer was a block to replication. To study the structure and activity of the Y-class DNA polymerase Dpo4 from Sulfolobus solfataricus with the template 5'-TCATTGAATCCTTCCCCC-3', where G is the site of adduction (AFB1-FAPY), we crystallized binary and ternary complexes of the polymerase in which the primer strand terminated with a cytosine located complementary to the adduct. The crystal structures solved at 2.7-3.0Å resolution with reasonable R-factors. The adduct site assumed Watson-Crick GC pairing. The AFB1-FAPY adduct existed in the beta anomer configuration.

"Solution Structure of the alpha-N7-dG Aflatoxin B1 Formamidopyrimidine (FAPY) Adduct in DNA"

Kyle Brown
Vanderbilt University

Ingestion of aflatoxin B1-contaminated food may result in chemically modified DNA. Hydrolysis of the N7-dG DNA aflatoxin B1 (AFB1) adduct yields its formamidopyrimidine (FAPY) derivative. N7-dG-AFB1-FAPY is the most stable AFB1-DNA adduct. In DNA, alpha and beta deoxyribose anomers of the AFB1 FAPY adduct exist in equilibrium. The alpha-FAPY is favored in single-strand DNA, whereas the beta-FAPY is favored in double strand DNA. Site-specific mutagenesis studies revealed that the alpha-FAPY blocks DNA replication in E.Coli. The beta-FAPY is highly mutagenic in E.coli, including G to T mutations. A solution structure of beta-FAPY in dsDNA has been published. The beta-FAPY was found to stabilize the duplex Tm by 14C. Little is known about alpha-FAPY, prompting many questions. Specifically, why is the alpha-FAPY disfavored in dsDNA and why is it such a strong block to replication?

"Expression and Characterization of Symmetric Half Barrels"

Carrie Fortenberry
Vanderbilt University

"Structure and Substrate Binding to Activated HSP 16.5 Mutant Analyzed by CryoEM, X-ray Crystallography, and Mass Spectrometry"

Jared Godar
Vanderbilt University

"Coarse Grained Models of Crystalline Free-Fatty Acids"

Kevin R. Hadley and Clare McCabe
Department of Chemical Engineering, Vanderbilt University

Free-fatty acids are an essential component of biological systems and important in the structure of certain biological membranes. As a first step towards the study of membrane lipid self-assembly coarse grained models of fatty acids have been developed. Coarse-grained models allow molecular simulations to be performed on larger spatial and temporal scales than is possible through atomistic simulation and are necessary to study the self-assembly of mixed lipid systems. The RPM method formulated by Reith, Pütz, and Müller-Plathe [1, 2] that is based on matching radial distribution functions for the coarse-grained molecules with their atomistic counterparts, has been used to parameterize the coarse grained models. Results will be presented for a number of pure and mixed lipid systems to demonstrate the suitability of the proposed force field for modeling crystalline and melt fatty acid phases as well as the transferability of the coarse-grained potentials.
References

1. Reith D, Putz M, Muller-Plathe F, "Deriving Effective Mesoscale Potentials from Atomistic Simulations." J Comp Chem, 24: 1624-1636 (2003).
2. Milano G, Goudeau, Muller-Plathe F, "Multicentered Gaussian-Based Potentials for Coarse-Grained Polymer Simulations: Linking Atomistic and Mesoscopic Scales." J Polymer Sci B, 43: 871-885 (2005).

"Residue Residue Contact Prediction using Artificial Neural Networks"

Mert Karakas
Vanderbilt University

Packing of secondary structure elements (SSEs) are shown to be an important force in stabilizing protein structures and hinting on the folding pathways they follow. Therefore, a reliable residue-residue contact predication method based on only sequencing information, would be able to reduce the conformational search space vastly in cases where experimental structural data does not exist or is incomplete. We have developed a method where data structures known as Artificial Neural Networks (ANNs) are trained with data from ~1800 proteins from a non-redundant fold database to differentiate between contacts and non-contacts. The ANN requires an input of two sequence windows spanning the potentially interacting SSEs, having the two directly contacting amino acids in the center. For each amino acid in these windows predicted secondary structure, position specific scoring matrices from PSI-BLAST, and a property profile are used as input. Five separate ANNs were trained for helix-helix, helix-sheet, sheet-helix, sheet-sheet, and strand-strand interactions. It is expected that high-resolution training of ANNs will increase accuracy of the predictions and resultant reduction of search space, when combined with the residue specific nature of the predictions would make this method candidate for to be used as a component in high-resolution side chain building, structure refinements and scoring schemes for tertiary structure prediction.

"Flexible Ligand Docking in Rosetta: Current Status and Future Directions"

Kristian Kauffman
Vanderbilt University

Rosetta was recently expanded to allow small molecule docking. Here we present the expansion of Rosetta to explicitly dock flexible small molecule through the creation of small molecule rotamers. We present the algorithm and its performance on a small set of test cases.

"Knowledge-based transfer free energies for predicting transmembrane spans"

Julia Koehler
Vanderbilt University

Here we describe a knowledge-based approach using the ProteinDataBank for transmembrane proteins (PDBTM) for the derivation of three hydrophobicity scales for the transfer between water-membrane, water interface and interface-membrane. The hydrophobicity scales are based on the free energy of the 20 amino acids for the three regions of water, interface and membrane, therefore giving additional information on an absolute scale. For the derivation of the scales 60 representative MPs from the PDBTM were used. Eight established hydrophobicity scales were compared with the derived knowledge-based scale for the ability of predicting transmembrane spans along the protein sequence. For transmembrane span prediction a triangular window was used. The agreements for different window sizes were calculated. Compared to the Wimley&White scale the knowledge-based scale performs significantly better for three regions on a per-residue basis. For two regions the knowledge-based scale performs 'only' equally well in comparison to the other scales possibly because the information of the interface region is lost. It can therefore be argued that inclusion of the transition region into transmembrane span prediction tools yields improved prediction results. Since both the prediction of secondary structure and the preference for soluble or membrane region using free energies are closely related, we aim to combine both concepts for a unified solution-membrane secondary structure prediction tool which can be used to distinguish soluble from membrane domains in proteins with enhanced secondary structure prediction.

"Cryo-EM guided de novo Protein Fold Elucidation"

Steffen Lindert
Vanderbilt University

Cryo-electron microscopy (cryo-EM) is an increasingly important technique to elucidate the structure of large macromolecular assemblies such as whole viruses. Cryo-EM generally does not provide atomic resolution information, but numerous sub-nanometer resolution structures have been reported in the last years. At 7 Å resolution helices are observed as density rods but strand and coil regions are not discernable. The goal of this project is the development of a computational protein structure prediction algorithm that incorporates the experimental cryo-EM data as restraints. The placement of helices is restricted to regions where density rods are observed in the cryo-EM density map. The Monte Carlo based search algorithm is driven by knowledge based energy functions.

This method has been benchmarked with ten highly α-helical proteins of known structure. The chosen proteins range in size from 250 to 350 residues. Starting with knowledge of the true secondary structure for these ten proteins, the method can identify the correct topology within the top scoring 10 models. With more realistic secondary structure prediction information, the correct topology is found within the top scoring 150 models for seven of the ten proteins. These results prove the effectiveness of the proposed assembly algorithm. In a next step the algorithm will be applied to one of the proteins in human adenovirus, protein IIIa. This protein, for which there is no high resolution structure, is predicted to be highly α-helical and is resolved in a 6.9 resolution cryoEM adenovirus structure as a bundle of ~13 α-helical density rods.

"Virtual High Throughput Screening for mGluR5 potentiators"

Ralf Mueller
Vanderbilt University

Metabotropic glutamate receptors play a key role in modulating synaptic activity in the Central Nervous System. As such they represent viable targets for treating diseases like schizophrenia, Parkinsons disease, and others. mGluR5 potentiators modulate the response of mGluR5 to agonists. Starting from a set of assay data for mGluR5 potentiators, an artificial neural network was trained to predict the mGluR5 potentiation (EC50) for small molecules. The input data consists of scalar molecular descriptors, autocorrelation functions, and radial distribution functions. The enrichment factor on the original screening data was calculated to be 43 (from approximately 1% compounds being active to 43%). After refining the artificial neural network, it was applied to scan a database of 450,000 commercially available drug candidates (ChemBridge) for new possible leads. A set of 800 compounds was selected for their high predicted activity to be tested at the High Throughput Screening center at Vanderbilt. The talk presents the methods developed and implemented to train this artificial neural network and discusses the results of this test.

"Three-Dimensional Models of KaiB-KaiC Protein Complexes and Functional Insights into the Cyanobacterial KaiABC Circadian Clock"

R. Pattanayek, D. R. Williams, S. Pattanayek, T. Mori, Y. Xu, C. H. Johnson, P. L. Stewart, and M. Egli
Vanderbilt University

The circadian clock of the cyanobacterium Synechococcus elongatus can be reconstituted in vitro by mixing recombinant KaiA, KaiB and KaiC proteins with ATP, producing KaiC phosphorylation and dephosphorylation cycles that have a regular rhythm with a ca. 24-h period and are temperature-compensated. KaiA enhances KaiC phosphorylation and is antagonized by KaiB. We have used X-ray crystallography, negative-stain and cryo-EM, biochemical and mutagenetic data to build 3D-models of the KaiA-KaiC and KaiB-KaiC clock protein complexes. The KaiA dimer interacts with the KaiC homo-hexamer via the flexible C-terminal region of the latter. A second, potentially more transient interaction concerns an apical loop region of KaiA and the inter-subunit ATP-binding cleft in KaiC. Two KaiB dimers bind to the C-terminal half of KaiC and the interaction results in splaying of the C-terminal domains of KaiC subunits. The model of the KaiB-KaiC complex also provides a rationalization of the earlier observation that KaiA and KaiB simultaneously bind to KaiC during the dephosphorylation phase. The two complexes and insights into clock function and regulation of phosphorylation based on them will be discussed.

Support by the National Institutes of Health (R01 GM73845 to ME, R01 GM67152 to CHJ, and F32 GM71276 to DRW) is gratefully acknowledged.

"Structural and Biochemical Characterization of the C-terminal Domain of Mcm10"

Patrick Robertson
Vanderbilt University

Eukaryotic DNA replication is tightly regulated at the initiation phase to ensure that the genome is copied only once and at the proper time during each cell cycle. During replication initiation, over twenty different proteins are recruited to origins of replication to denature the DNA duplex and assemble a functional replication fork. Mini-chromosome maintenance protein 10 (Mcm10) is a DNA binding protein that is recruited to origins in early S-phase and is required for the activation of Mcm2-7, the replicative DNA helicase. Additionally, Mcm10 is necessary for subsequent loading of downstream replication proteins, including Cdc45, replication protein A (RPA), and DNA polymerase α-primase (pol α), onto chromatin. However, the precise role of Mcm10 in DNA replication initiation still remains unclear. In order to perform structure-function studies on Mcm10, we have begun to characterize the individual domains of the full-length protein from Xenopus laevis (XMcm10), which shares a high sequence homology with human Mcm10. Our work has revealed that XMcm10 contains three structured domains: a putative oligomerization domain at the N-terminus and two independent DNA binding domains located in the interior and C-terminal regions of the protein. We have previously shown that the C-terminal domain (CTD) is stabilized by two zinc ions and binds to both single- and double-stranded DNA and to p180, the polymerase subunit of pol α. Here we present our progress towards the 3-dimensional structure of the C-terminal domain of XMcm10 using both x-ray crystallography and NMR spectroscopy. We aim to elucidate the DNA and pol α binding sites in XMcm10-CTD, with the ultimate goal of determining how the ID and CTD coordinate DNA and pol α binding.

"Structure of an Oligodeoxynucleotide Duplex Containing the Lesion 1,N2-Ethenodeoxyguanosine Opposite a One-Base Deletion"

Ganesh Shanmugam
Vanderbilt University

The structure of the 1,N2-εdG adduct, arising from the reaction of vinyl chloride with dG, was determined when placed opposite a one-base deletion (1BD) in the duplex 5'-d(CGCAT?GGAATCC)-3'· 5'-d(GGATTCATGCG)-3' at neutral pH. This duplex model a 1BD structure associated with bypass of the 1,N2-εdG lesion by a Y-family Sulfolobus solfataricus P2 DNA Polymerase Dpo4. The modified duplex showed ~8 °C and ~6 °C increases in Tm as compared to the unmodified 1BD and the fully complementary duplexes, respectively. This suggested that the stability of the oligodeoxynucleotide duplex increased when 1,N2-εdG was placed in the 1BD duplex as compared to the fully complementary duplex. 1,N2-εdG adopted the anti conformation about the glycosyl bond at neutral pH. The observed NOEs between the etheno protons and DNA protons established that the etheno moiety was positioned in the center of the helix and stacked between the 5'- and 3'-neighboring base pairs. Watson-Crick base pairing was conserved for the 5'- and 3'-neighboring base pairs. The chemical shift differences suggested that the structural perturbation was restricted to the 1,N2-εdG adduct and its neighboring base pairs. The 31P spectra revealed that the backbone was disturbed near the adduct site. The results are compared with the 1,N2-εdG adduct opposite to cytosine and with primer template complexes site-specifically modified with the 1,N2-εdG adduct in the presence of the Y-family Sulfolobus solfataricus P2 DNA Polymerase Dpo4. Supported by NIH grants ES-05355 (M.P.S. and C.J.R.) and ES-010375 (F.P.G.).

"Solution Structure of Pyk2 PAT domain and its interaction with LD motifs"

Murugendra Vanarotti
St. Jude Children's Research Hospital

Proline-rich tyrosine kinase 2 (PYK2) is a major kinase in the integrin-mediated cell adhesion. PYK2 is a non-receptor protein tyrosine kinase and is related to focal adhesion kinase (FAK). PYK2 localizes to adhesion structures through interactions between its C-terminal region and cytoskeletal proteins and plays an important role in cell adhesion and migration. The PYK2 C-terminal region is highly homologous to the focal adhesion targeting (FAT) domain of FAK. Both PYK 2 and FAT share similarity in the architecture of central kinase domain. The proline flanked N-terminal targeting domain that has shown to be the responsible domain for these two proteins in differential localization, the C-terminal targeting domain of Pyk2 (PAT) is replaced with the C-terminal domain of FAK (FAT), Pyk2 localizes to focal adhesion cites and vice versa (Zheng et al. 1998; Klingbeil et al 2001). The solution structure of the FAK targeting domain (FAT) was previously solved in our lab (Liu et al. 2002), and it folds into a right turn four helix bundle which binds to Paxillin LD2 and LD4 motifs (Betrtolucci et al. 2004). In order to reveal the structural differences between PYK2 and FAT we have solved the solution structure of C-terminal targeting domain (PAT) of PYK2 by using NMR. This structural information helps us to understand the role of these two proteins in focal adhesions and cell motility. The solution structure of PYK2 we solved reveals a four-helix bundle that resembling those found in other proteins involved in cell adhesion, FAK, alpha-catenin and vinculin. IN addition to structural studies we have also carried out the PYK2 interaction with LD2 and LD4 motifs of Paxillin. The interaction data suggests that unlike FAT it can only bind the LD2. This interaction study also helped us to exploit the information to design PYK2-specific inhibitors, therefore, serve a potential, novel therapeutic agent for bone resorption and bone metastatic disease.

"The Involvement of Thrombin Exosite I in Glycosaminoglycan-Accelerated Thrombin Inactivation by Heparin Cofactor II"

Ingrid M Verhamme
Vanderbilt University

Thrombin is covalently inactivated by the serpin heparin cofactor II (HCII), and this reaction is accelerated by long- and short chain glycosaminoglycans (GAGs). Long GAGs act as templates by bridging thrombin exosite II and HCII, whereas short GAGs may act as conformational inducers of an interaction of the N-terminus of HCII with thrombin exosite I. Hir54-65(SO3)- and an exosite I-specific DNA aptamer, with KD for thrombin binding of 29 and 25 nM, respectively, were used as exosite I probes in thrombin inactivation kinetics, and in binding studies of dead-end ternary complexes between fluorescently labeled thrombin, heparin and HCII. Displacement of fluorescently labeled Hir54-65(SO3)- from exosite I was monitored during heparin-accelerated thrombin inactivation by HCII. Both in the presence of long and short GAGs, the thrombin inactivation rate decreased hyperbolically to a limiting value as a function of Hir54-65(SO3)- and aptamer concentration, suggesting a partial competition of these probes and the HCII N-terminus for exosite I. Likewise, binding of Hir54-65(SO3)- to a fluorescent dead-end complex of thrombin, heparin and HCII resulted in a hyperbolic, partial decrease of the ternary complex fluorescence. Apparent dissociation constants for the exosite I ligands and the active and dead-end ternary complexes were considerably weaker than those for ligand binding to thrombin. The kinetics of Hir54-65(SO3)- displacement from thrombin during the inactivation process also indicated a weaker interaction. Taken together, these results are consistent with overlapping binding sites for the N-terminus of HCII, Hir54-65(SO3)- and the aptamer within thrombin exosite I.

"A structural basis for DNA binding by eukaryotic replication initiation protein Mcm10"

Eric Warren
Vanderbilt.University

Mcm10 is an essential eukaryotic DNA replication protein that is required for origin unwinding and replisome assembly. S. pombe and S. cerevisiae Mcm10 have been shown to physically associate with single-stranded (ss) DNA and the catalytic subunit of DNA polymerase α-primase (pol α) both in vitro and in vivo, suggesting that Mcm10 may recruit pol α to an assembling replisome To gain a mechanistic understanding of these interactions, we are investigating the molecular basis for DNA binding by vertebrate Mcm10. The sequence conservation between yeast and vertebrate Mcm10 is limited to a 200-amino acid region in the middle of the protein, which we have experimentally defined to be one of two DNA-binding domains within the protein. We present here crystal structures of the highly conserved internal domain of Xenopus laevis Mcm10 (Mcm10-ID) in the unliganded form and bound to ssDNA. Mcm10-ID is composed of an oligonucleotide/oligosaccharide binding (OB) fold and zinc finger in tandem--two classic ss- and dsDNA binding motifs. Together with NMR chemical shift perturbation and mutational analysis of DNA binding, these structures provide a high resolution model for how Mcm10 engages ss- and dsDNA. We speculate that Mcm10 plays a role in melting origin DNA and may preferentially associate with an emerging replication fork.

"Protein structure identification server: template prediction server for homology modeling of PDZ domains"

Linda Widjaja
St. Jude Children's Research Hospital

Sequence comparison is an important step in the template-based prediction of protein structure from existing structures in the Protein Data Bank. However, this first crucial step in homology modeling still poses many challenges. To address the difficulties of predicting the structure(s) to be used as template to model several PDZ domain targets of interest, we have developed a PDZ structure prediction Web server that identifies homologs related to the target sequence. We used the inputs selected by our Web server to build a homology protein structure model, using the program Modeller. We confirmed the quality of our PDZ server by performing trial tests using 3 human PDZ domains whose structures have been recently solved. Our PDZ template prediction server was useful in predicting the template(s) to be used for homology modeling for PDZ domains with sequence identities of 50% or more with currently available PDZ structures in Protein Data Bank. Our method can be adapted for determining templates for homology modeling of other proteins or domains too.

"Interaction mechanism of the PDZ domain of Dyl protein and the C-terminus of Frizzled protein receptor"

Xinxin Zhang
St. Jude Children's Research Hospital

Frizzled (Fz) proteins are seven-transmembrane proteins which act as Wnt protein receptors. The extra-cellular part of Frizzled protein interacts with Wnt Proteins, while the Frizzled signaling is conducted by the direct interaction of C-terminus KTXXXW motif of Frizzled with the PDZ domain of Dishevelled (Dvl) protein. There are 19 Wnt proteins, 10 Fz proteins and 3 Dvl proteins in mammals. Kinetics and evolutionary analysis indicate that combination of different types of Wnt, Fz and Dvl constitute tissue-specific signaling pathways, while the molecular mechanisms of interactions between Fz and Dvl proteins remain unclear. In order to elucidate the binding mechanisms of PDZ domain with Fz proteins, we carried out binding affinity studies by fluorescence spectroscopy and NMR spectroscopy. In particular, we conducted structural studies of Fz-PDZ complex using modern three-dimensional nuclear magnetic resonance techniques. Our preliminary results correlate wiell with previous studies on the Fz-Dvl interaction specificity. In addition, the 3-D solution structure of PDZ-peptide complex elucidates the interaction pattern of PDZ domain with Fz protein. Finally, the coordinate information of PDZ-peptide complex provides instruction on PDZ-targeted drug discovery.