Homology modelling was used to determine the 3D structure of MG_237. A BLASTP [9] search with default parameters was performed against the Brookhaven Protein Data Bank (PDB) to find suitable templates for homology modelling. PDB ID: 1TD6_A was identified as the best template based on sequence identity (63%) between query and template protein sequence. The Protein Structure Prediction Server (PS)²­v2 [10] was used for homology model construction. (PS)² is an automated homology modelling server that builds 3D models using the modelling package MODELLER. Multiple sequence alignments between query and template protein sequences made by the (PS)² server were manually curated before initiating homology modelling of MG_237.

Once the 3D model was generated, energy minimization was performed by GROMOS96 force field in a Swiss-PdbViewer [11]. Structural evaluation and stereochemical analyses were performed using ProSA-web Z-scores [12] and PROCHECK Ramachandran plots [13]. Furthermore, Root Mean Squared Deviation (RMSD), superimposition of query and template structure, and visualization of generated models was performed using UCSF Chimera 1.5.3 [14].

The predicted 3D structure of M. genitalium hypothetical protein MG_237 was submitted to the Protein Model Database (PMDB) [15] and assigned the PMDB ID: PM0077727.

M. genitalium hypothetical protein MG_237 was analysed for the presence of conserved domains based on sequence similarity search with close orthologous family members. For this purpose, three different bioinformatics tools and databases including InterProScan [16], Proteins Families Database (Pfam) [17], and NCBI Conserved Domains Database (NCBI-CDD) [18] were used. InterProScan is a tool that combines different protein signature recognition methods native to the InterPro member databases into one resource with look up of corresponding InterPro and GO annotation. Pfam is a database of protein families that includes their annotations and multiple sequence alignments generated using hidden Markov models. NCBI-CDD is a protein annotation resource that consists of a collection of well-annotated multiple sequence alignment models for ancient domains and full-length proteins.

Protein 3D structures can provide us with precise information of how proteins interact and localize in their stable conformation. Homology or comparative modelling is one of the most common structure prediction methods in structural genomics and proteomics. Numerous online servers and tools have become available for homology or comparative modeling of proteins in past years. Despite minimal modifications, one initial step that is common in all modeling tools and servers is to find the best matching template by performing a sequence homology search with BLASTP. Templates are experimentally determined 3D structures of proteins that share sequence similarity with the query sequence. The template sequence and the protein sequence whose structure is to be determined are aligned using multiple sequence alignment algorithms [23]. A well-defined alignment is very important for the prediction of a reliable 3D structure. The genome of M. genitalium consists of 94 hypothetical proteins without any known function or structure. A BLASTP search was performed for each protein sequence against the PDB to identify templates for homology modeling. MG_237 was selected for homology modeling as it showed maximum identity to its 1td6_A, which is an X-Ray diffraction model of a M. pneumoniae hypothetical protein. The query sequence and template ID was then given as input to the (PS)² server for homology modeling using MODELLER.

The predicted 3D structure of MG_237 is shown in Figure 1b. Even though there were no steric clashes in the structure generated, it was still subject to energy minimization and assessed for both geometric and energy aspects. The positioning of alpha-helices and beta-sheets was then compared using ESPript2.2. Secondary structure elements were found to be comparable to that of the template (Figure 1a). Eleven helices and two beta sheets were predicted in the 3D structure of MG_237, which implies that it is rich in helical structures (Figure 1a and 1b). Several structure assessment methods including RMSD, Z-scores, and Ramachandran plots were used to check reliability of the predicted 3D model. The RMSD value indicates the degree to which two 3D structures are similar. The lower the value, the more similar the structures. Both template and query structures were superimposed for the calculation of RMSD (Figure 1c). The RMSD value obtained from superimposition of MG_237 and 1TD6 in UCSF Chimera was found to be 0.213 Å, suggesting a reliable 3D structure. The Zscore is indicative of overall model quality and is used to check whether the input structure is within the range of scores typically found for native proteins of similar size. Z-scores of the template and query model were obtained from PROSAweb. The template Z score was ­7.97 (Figure 1e) and for the MG-237 homology model it was ­7.42 (Figure 1f), suggesting similarity between template and query structure. Finally, the Ramachandran plots were obtained for both the homology model and the template as a quality assessment. PROCHECK displayed 91% of residues in the most favored regions, with 7.6%, 0.7%, and 0.7% residues in additionally allowed, generously allowed and disallowed regions, respectively (Figure 1d). This indicated that the backbone dihedral angles, phi and psi, in the MG_237 3D model, were reasonably accurate. The Ramachandran plot for the template structure showed the amino acid residues to be 84.2%, 14.0%, 1.9% and 0.0%, in most favored, additionally allowed, generously allowed and disallowed regions respectively (Data not shown). The comparable Ramachandran plot characteristics, RMSD values, and Z-scores confirm the quality of the homology model of MG_237. The final protein structure was deposited in PMDB and is available under ID: PM0077727.