|
Dicty Cells
Dictyostelium discoideum cells are slime molds whose simple life cycle makes them an attractive species for study. Under normal conditions they reproduce by simply dividing into two cells. Under starved conditions, the Dictyostelium cells aggregate and form a slime mold. An individual cell initiates this transformation by producing cyclic adenosine monophosphate (cAMP), a small molecule that acts like a hormone. The cAMP is released into the environment and is recognized by the other cells through their cyclic adenosine monophosphate receptors (cARs ). There are four different types of cAR receptors (cAR1-cAR4). The cAR receptors belong to a family of other hormone receptors called G Protein Coupled Receptors. The functions of the different receptors in the transformation process are not fully understood. Once the neighboring cells receive the cAMP signal they produce large amounts of cAMP and release it into the environment. Subsequently, they move towards the cell that initiated the signal. This method of signaling and movement brings the cells together to form the slime mold.
Structural models of one of the receptors, cAR2, were generated to further the understanding of cAMP interactions with the massive cell, how this contact turns "on" the signal to make cAMP, and how the cells move towards the signal.
Molecular Modeling
Predicting protein structure is a difficult task because the exact structure that will be formed by a given sequence of amino acids is not known. One can base a model on homology or the degree of similarity between the amino acid sequences of the two structures. Often similar sequences will form similar three-dimensional structures. Programs are available on the Internet such as FASTA and BLAST to search for sequence homology. If the protein is not homologous to any structure there are protein structure prediction programs on the Internet. These programs include nnPredict and PHD. Transmembrane proteins contain stretches of hydrophobic or "water fearing" amino acids that are buried in the lipid bilayer of the membrane can be predicted by PHDhtm, PHDtopology, and Sosui. Based on the Internet prediction results a structure with similar characteristics can be chosen from the Brookhaven Protein Data Bank, a storage and cataloging facility for protein structures available over the Internet. Then the template and model sequences are aligned and submitted to modeling programs like WHAT IF and MODELLER. A model is created by overlaying the model sequence onto the template sequence and assigning the same amino acid positions for the model as the template.
The cAR2 Modeling Process
The cAR2 sequence exhibits no homology to a known structure so homology modeling could not be used. Transmembrane helices were predicted with the Internet programs 7TM, Sosui, TM Pred, and PHDT. Several structures with transmembrane helices were found in the Brookhaven Protein Data Bank. Bacteriorhodopsin (Brd) was chosen as the template for the model because seven transmembrane helices are present and the structure is simple compared to other structures with transmembrane helices. The transmembrane helices in Brd form a barrel that spans the membrane from the outside to the inside of the cell. The predicted seven transmembrane helices of cAR2 were aligned with the Brd helices and seven models were created with a modeling program, WHAT IF. The extra amino acids that do not make up the helices are not included in these models because the structure of bacteriorhodopsin only includes the helices. Twenty-one models were created with another modeling program, MODELLER. This program incorporates the amino acids outside of the helices into the model. However, the predictions of these regions are not as reliable because they are not based on a known structure.
Significant Findings in the Study
Once the difficulties of creating the models were surpassed, the problem of choosing the best one arose. The problem lies in identifying a method to make this determination. Model viewing programs were used to compare characteristics of the cAR2 model to the Brd template. Sybyl allows the highlighting of "bulky" amino acids that contain cyclic or long sidechains. The size of the amino acid is important to consider because bumping between amino acids could occur. The bulky amino acids in the template structure were observed and the model was checked for similar placements. In some cases the placements were similar, but how does one determine the fine line between an acceptable placement or one that is "wrong". Another protein viewing program, Rasmol was used to show differences by coloring chosen groups, components that make of the structure, and charge. The Brd structure was colored to display the helices Figure 1: ”Helices of Brd”. Also, the Brd structure was colored to show the differences in charge. High values are blue and low values are red. Figure 2: "Charge of Brd". Most of the protein is colored in red, which indicates a trend in uncharged amino acids. In comparison, one of the cAR2 models created by WHAT IF was colored to highlight the different components of the structures. Figure 3: “Structure of cAR2”. The helices are magenta, loops are purple, and all others are white. The same cAR2 model was colored to show the differences in charge. Figure 4: “Charge of cAR2” . Most of the protein is colored in green and yellow, which indicates slightly more charged amino acids than the Brd. Does this difference in charge indicate an "incorrect" model or are these differences significant? Unfortunately, these types of observations usually do not give a definitive answer to solve the dilemma of the acceptability of the model. So the problem of assessing the generated models remains a process. Programmers will continue to develop better programs of prediction and assessment and we will slowly move towards a better understanding of protein structure.
Useful WebSites
|
