1. Learning Objectives

a. This unit project is designed to reinforce the underlying basis of the concept that protein structure determines protein function. The focus is on the physical and chemical properties of a protein, ways to discern patterns of secondary structure in a collection of related protein domains, search and retrieval of specific items from the protein structure databases, and the process of modeling a 3-D protein structure from aligned sequences and a template structure.

b. Bioinformatics tools to be evaluated include query software for the protein sequence and structure databases, viewers (such as Swiss-PdbViewer) for PDB format structural data files, sequence alignment packages, motif and profile methods, the SWISS-MODEL server, and the Verify 3D Crystal Structure Verification server.

2. Project Objectives

a. Develop a plan for modeling the 3-D structure of the death domain from the human 75 kDa neurotrophin receptor. The proposed model should be sufficient to address the following question: Would the deletion of a three residue loop, in rodent NTR death domains between helix I and helix II, interfere with the functioning of these rodent death domains?

b. Prepare an appropriate multiple sequence alignment and estimate the secondary structure of the query death domain.

c. Identify an appropriate template or templates and use it or them to develop a model. Validate your model, then propose an answer to the question above based on your model.

1. Death Domains

a. Death domains come in two flavors One type is the interaction domain of adaptor proteins that form complexes with receptor death domains, eg., death domains in proteins called TRADD, FADD, etc. The other type is the collection of death domains in the cytoplasmic tails of members of the Tumor Necrosis Factor Receptor superfamily ("receptor death domains"). Some members of this superfamily are Fas (CD95), DR3, DR4, DR5, DR6, p75 NTR (NGFR, LANR, and other names), TNFR-I, etc. Many members of this family do not have death domains, for example CD40, LTbetaR, CD30, various "decoy" receptors, etc. You can ignore these death-domainless members.

b. There are two structures in the PDB for receptor death domains: the one for the rat p75 NTR (by Carlos Ibanez' group) and the one for human Fas (by Stephen Fesik's group).

2. Death domains are difficult to align

a. Making a good alignment of divergent sequences is challenging. In this project, helix I will be the most divergent. Warning: Plug and chug with CLUSTAL W will not yield a useful result. The following paper describes a method that is very helpful for collecting protein fragments and building material for a multiple alignment:

Gapped BLAST and PSI-BLAST:
a new generation of protein database search programs

Stephen F. Altschul, Thomas L. Madden, Alejandro A. Schaffer,
Jinghui Zhang, Zheng Zhang, Webb Miller and David J. Lipman
Nucleic Acids Research, 1997 25(17) 3389-3402

b. The best source of curated protein sequences is the SwissProtein Database. The NCBI server has a good PSI-BLAST utility.

1. The steps in building a homology-based model

a. Listed on page 283 of Gibas and Jambeck, the first steps depend on searching for relatives of the query sequence, building a structurally-informative multiple sequence alignment, and locating an appropriate template structure in the protein structure database.

b. The next step is to achieve the best possible global (full length) alignment of the query sequence with the chosen template sequence or sequences, lining up every secondary structure element with its appropriate counterpart.

c. The next three steps are the actual modeling of the structure. Here the backbone trace is assembled (composed of secondary structure elements without sidechains (ie, amide nitrogen, alpha carbon, carbonyl carbon for each residue in the chain). Note that reconstructive surgery may (will) be required to work around gaps and insertions in the alignment.

d. Once the backbone has been constructed, add the sidechains of the query sequence. At this point, most models blow up. Don't panic, it's normal. Adjust the side chain torsion angles to eliminate major collisions. If there is a problem in one (or more) of the loops grafted on during the plastic surgery, change it and readjust the torsion angles.

e. Optimize the structure using energy minimization. This procedure is based on the idea that strained bonds are high-energy bonds, that repulsive electrostatic forces increase potential energy, and that Van der Waals attractions become energetic repulsions when atoms get too close. Relaxing the model should bring down the potential energy stored in all these unhappy situations. A low energy model is a happy model.

2. Software and services for the project

a. For background on Principles of Protein Structure, Comparative Protein Modelling and Visualisation see Protein modelling by Nicolas Guex and Manuel C. Peitsch, Part II Chapter 6: Comparative protein modelling.

b. For relatively easy to use free modeling software check out SwissModel Optimise Mode . SwissModel project files can be produced with Swiss-PDB Viewer (spdbv). These projects contain the sequence to be modelled and the superimposed template structures. This is the default file format, in which SwissModel will return the results. SPDBV allows you to manually adjust the sequence alignment that will be used to guide the model building. One major reason for failure of "First Aproach" attempts is badly placed gaps or insertions. If you have taken care of this problem already, the modeling should go efficiently.

c. SPDBV project files are the only way to prepare and submit "Optimize Mode" modelling requests to the SwissModel server. Swiss-PDB Viewer can be downloaded for free from Swiss-PdbViewer.

1. Background on model verification

a. For a tutorial on model validation see Model quality from Principles of Protein Structure, Comparative Protein Modelling and Visualisation by Nicolas Guex and Manuel C. Peitsch: Part III: Chapter 8: How to evaluate the quality of a model.

b. "The correctness of a model is essentially dictated by the quality of the sequence alignment used to guide the modelling process. If the sequence alignment is wrong in some regions, then the spatial arrangement of the residues in this portion of the model will be incorrect."

2. Server for access to validation software

a. You can run tests on your model at Verify 3D Crystal Structure Verification Server provided by the UCLA Bioinformatics Servers and Databases.

b. Good luck.

1. Prepare a presentation [15 minutes maximum] of your findings, including a brief summary of methods, results, and conclusions. Include recommendations for further extensions of the research.

a. For visual aids, you may include overheads of results or use a computer presentation, such as PowerPoint.

b. Be prepared to field questions from other groups and from the presenter.

2. Submit the following as a typed report. As a guideline, a finished report on the search should be about three pages of text [12 pt standard font] and no more than four pages of appended graphs, tables, images, etc. [Final page length is to be determined by what the group identifies as appropriate*]:

a. Name of the project and names of the members in your group.

b. A brief report in scientific format, including abstract [100 word limit], introduction, methods, results, discussion/conclusions, and citations. Include figures as appropriate.

3. Grading will be on content, organization, spelling, & grammar.