CHAPTER 29
Prediction of Tertiary Structure of Protein: Sequence Homology
CS Mukhopadhyay and RK Choudhary
School of Animal Biotechnology, GADVASU, Ludhiana
29.1 INTRODUCTION
Prediction of the tertiary structure of a peptide sequence is done by either comparative modeling (when the experimentally determined structure of a homologous protein is available) or ab initio approaches (when the structure of the homologous protein is not available).
29.2 OBJECTIVE
To predict the tertiary structure of a peptide using the homology modeling approach with the online tool SWISS‐MODEL.
29.3 PROCEDURE (SWISS‐MODEL PROGRAM)
29.3.1 Query sequence
Let us say we are interested in determining (predicting) the structure of pancreatic ribonuclease of buffalo (Bubalus bubalis), which has not been experimentally determined and reported. Hence, we need to search for the reported structure from a related species (e.g., cattle), so that this can be used as a template for homology modeling.
The sequence of the 124‐residue‐long bubaline pancreatic ribonuclease (obtained from PDB) is:
>tr|Q95NE6|Q95NE6_BUBBU Pancreatic ribonuclease (Fragment) OS = Bubalusbubalis PE = 3 SV = 1
K E T A A A K F Q R Q H M D S S T S S A S S S N Y C N Q M M K S R S M T S D R C K P V N T F V H E S L A D V Q A V C S Q K N V A C K N G Q T N C Y Q S Y S T M S I T D C R E T G S S K Y P N C A Y K T T Q A N K H I I V A C E G N P Y V P V H F D A S V
29.3.2 Downloading the template structure
Download a reported structure of a related peptide (in a closely related species, with at least 70% sequence similarity) from a protein data bank (PDB). In the present example, we will download the .pdb file of the experimentally determined structure (X‐ray diffraction, resolution of 1.95 Å) of bovine pancreatic ribonuclease (PDB ID: 4RTE).
29.3.3 Pairwise sequence alignment
Determining the percentage similarity between the sequences (template vs. query sequences of pancreatic ribonucleases):
FIGURE 29.1 Pairwise sequence alignment to determine the extent of sequence identity between the query and template sequences.
The alignment result (percent identity matrix) clearly shows that there is a 95.97% match between the two sequences. Hence, the template of the target protein from bovine origin is fit to be used for homology modeling of the structure of bubaline pancreatic RNase.
29.3.4 Open SWISS‐MODEL workspace
The URL is http://swissmodel.expasy.org/workspace/index.php?func=show_workspace.
The first‐time user needs to create a new workspace and then start using the tool. A returning user needs to open the existing “workspace”, and then enter the email ID and project title.
29.3.5 Working mode
Click on “Modelling” at the top of the page, then select “Automated Mode” from the drop‐down menu.
FIGURE 29.2 Open the page to initiate homology modeling using the SWISS‐MODEL workspace
29.3.6 Input sequence
Paste the amino acid sequence (up to 1000 residues) into the sequence box, or enter the PDB ID of the peptide in the specified box. Here we will upload the template file 4RTE.pdb.
FIGURE 29.3 Window of SWISS‐MODEL workspace for providing the input parameters and starting homology modeling.
29.3.7 Job Submit
Click on “Submit Modeling Request”. You will need to wait until the work is done. The result will be emailed to the specified email address. Download the predicted structure in *.pdb format.
29.4 OUTPUT
The output page is self‐explanatory. The predicted model can be viewed on a Java‐enabled system by clicking on the image. The output page also mentions the following outputs:
FIGURE 29.4 Important sections of the SWISS‐MODEL output. One can download the complete result in PDF format, or can specifically download the sections as required.
29.4.1 Model information
This is the length of the input sequence that has been used for predicting its structure. Sequence identity (95.97%) and E‐value (4.47e‐48) are also given to indicate the degree of homology between the query and template sequences.
29.4.2 Quality information
QMEAN values (QMEAN4 and QMEAN6) are determined by the program to assess the overall quality of the model and the legitimacy of the bonds between amino acids in the predicted structure. The quality scores are also determined using methods like atomic empirical mean force potential (ANOLEA) and GROMOS. The detail on these terms can be obtained at the SWISS‐MODEL help page: http://swissmodel.expasy.org/workspace/index.php?func=special_help&=#A
29.4.3 Alignment
The pairwise alignment between the query and template sequences is displayed, along with the secondary structure information in the last row in each alignment block.
29.4.4 Ligand information
It is useful if the ligand is included in the model during prediction. Otherwise, the program predicts the possible ligand interacting sites on the predicted structure.
29.4.5 Modelling log
This is a relevant section to get information about the steps during homology modeling. If the template model is not supporting enough, the program uses an ab initio approach to model the query sequence. This log also indicates which portions have been built ab initio. In our example, the “final total energy” of the predicted model is –7759.335 kJ/mol.
29.4.6 Template log
Information about template search and use are given here.
29.5 VISUALIZING THE PREDICTED STRUCTURE
- The predicted structure can be visualized using a molecular visualization program such as Cn3D (http://www.ncbi.nlm.nih.gov/Structure/CN3D/cn3d.shtml) or RasMol (please refer to Chapter 27).
- Open the downloaded pdb file (Model_1.pdb, for instance) to visualize and assess the structural features.
29.6 INTERPRETATION OF RESULTS
The results obtained should be judged for the quality of the predicted structure using the following main parameters:
- Pairwise alignment: If the degree of homology (indicated by percent identity and E‐value) between the query and template sequence is high, the likelihood of accurate prediction of the structure is also high.
- Global model quality assessment: Check parameters such as QMEAN4 or QMEAN6. The packing quality of the model, minimum energy represented as favorable energy environment for each of the amino acid (green colored graph), is shown in the graph of ANOLEA. Lower energy indicates stability of the predicted structure.
- Parameters given under template selection log: Very useful for judging the cut‐off parameters set for BLAST and HHSearch during modeling.
Finally, after obtaining the model, the predicted structure must be validated using suitable tools (see Chapter 32) such as ProCheck, WhatCheck, What If, and so on. The correctness of the dihedral angle of rotation (peptide torsion angles Phi(φ) and Psi(ψ)) of the amino acid residues can be checked using Ramachandran’s Plot (RaptorX or any suitable tool).
29.7 QUESTIONS
- 1. What are the differences between the homology modeling based on ab initio approaches of tertiary structure prediction methods for tertiary structure prediction of protein?
- 2. Predict the tertiary structure of the same peptide (bubaline keratin) by homology modeling using a human homolog.
- 3. Given the following amino acid sequence, can you determine its structure through homology modeling, using a suitable homologous model?
>Query
M A L K S L V L L S L L V L V L L L V R V Q P S L G K E T A A A K F E R Q H M D S S T S A A S S S N Y C N Q M M K S R N L T K D R C K P V N T F V H E S L A D V Q A V C S Q K N V A C K N G Q T N C Y Q S Y S T M S I T D C R E T G S S K Y P N C A Y K T T Q A N K H I I V A C E G N P Y V P V H F D A S V
- 4. What are the parameters for evaluating the tertiary structure obtained through homology modeling? Evaluate the tertiary structure obtained in the previous question.
- 5. Enlist the factors that determine which template is best suited for homology modeling.