Keratitis is an inflammation of the cornea, the outermost part of the eye. It usually is caused by a plethora of bacteria, fungi and virus. It is a serious condition in which blood vessels grow into the cornea. Bacterial keratitis is a sight-threatening process. A particular feature of bacterial keratitis is rapid progression and corneal destruction, which may be complete in 24-48 hrs with some of the more virulent bacteria. Corneal ulceration, stromal abscess formation, surrounding corneal edema, and anterior segment inflammation are characteristics of this disease.

Serratia marcescens is a Gram negative, bacillus shaped bacteria that belongs to the family Enterobacteriaceae [1]. It is differentiated from Gram negative bacteria, as it is able to perform casein hydrolysis. Serratia marcescens is a potential cause of infectious keratitis that appears to be associated with abnormal corneal surface, topical medications and contact lens wear. It can also cause refractory keratitis resulting in corneal perforation and blindness. Serratia marcescens produces a cysteine protease, in addition to metalloproteases. These bacterial proteases have a number of biological activities, such as degradation of tissue constituents and host defense-oriented proteins, as well as activation of zymogens through limited proteolysis.

The aim of the present work was to isolate the genomic DNA from Serratia marcescens, and to amplify the 56 kDa cysteine protease gene. The cysteine protease gene from Serratia marcescens was for the first time amplified and sequenced in our lab, and as it was a preliminary study we were able to perform only a partial amino acid sequence. Since this is the first data regarding the sequence of cyteine protease of Serratia marcescens, we thought it necessary to study its structure. The structure of cysteine protease for S. marcescens has not yet determined experimentally (X-ray of NMR), therefore models are built following homology modeling protocol through computational methods. The structure prediction by homology modeling can help in understanding the 3D structure of the given protein and this in turn, will help to unravel its function. Therefore, in this paper we report for the first time, the in silico analysis, and homology modeling studies of cysteine protease was performed, as the 3D structures for these proteins are not available.

Corneal scrapings were collected after instillation of 4% lignocaine, without preservative under aseptic conditions, from each ulcer by an ophthalmologist using a sterile Bard Parker blade. Scrapings were performed under magnification of slit-lamp or operating microscope. Leading edge and base of each ulcer were scraped initially and the material obtained, were directly inoculated onto the surface of solid media such as sheep blood agar, chocolate agar and Sabouraud Dextrose Agar (SDA) and cultured. The battery of samples was processed with traditional isolation processes and was subjected to molecular identification by extracting genomic DNA. Isolation of DNA from S. marcescens was performed by CTAB/NaCl method [2-6]. Identification of these isolates was performed by amplification and sequencing of the 16S rDNA gene, followed by comparison of the sequence with the sequences in the GenBank DNA sequence database, using BLAST and submitted to NCBI GenBank (Accession number FJ588709).

Amplification of 56k Cysteine Protease

The amplification of 56k cysteine protease was performed using specific primers (Fwd seq: aagcgctcataaacgattgg, and Rev seq: acgtcatggcgatgatacaa). The gene was sequenced and the sequence was deposited to NCBI GenBank (Accession No: FJ810078). The same sequence was used for the further studies in silico homology modeling analysis studies.

Sequence Alignment and Structure Prediction

The FASTA sequence of the query protein was retrieved from NCBI Entrez sequence search (http://www.ncbi.nlm.nih.gov). Following BLASTp run (http://www.blast.ncbi.nlm.nih.gov).

Primary Structure Prediction

For physio-chemical characterization, theoretical isoelectric point (pI), molecular weight, total number of positive and negative residues, extinction coefficient, instability index, aliphatic index and grand average of hydropathy (GRAVY) were computed using the Expasy ProtParam server [7-9].

Secondary Structure Prediction

Secondary structure of this protein was predicted using the FASTA sequences of 56 k protease, and predicted using GOR IV and SOPMA [10,11].

Homology Modeling

The protein sequence was subjected for comparative homology modeling via Swiss model [12] and ESyPred3D (via The MODELLER 9v7 to generate putative 3D model [13]. The Swiss model performs the sequence alignments and searches for the putative template protein for generating the 3D model.

Sequence Subjected for Modeling

>target

MYQLQFTNLVYDTTKLTHLEQTIINLFIGNWSNHQLQKSICIRHGDDTSHNRY HILLIDTAHQRIKFSSIDNEGIIYILDYDDAQHILMQPSSKQGIGTSRPIVYERLVS

Template Selection

The template selection was done by performing BLAST against PDB database, to find the structural homologs. The template was selected based on the following criteria PROCHECK and PROSA analysis.

Validation

The validation of the modeled structure was carried out using PROCHECK [14], which calculates the main-chain torsion angles, i.e. the Ramachandran Plot [15]. The models were further checked with WHAT check [16]. Structural analysis was performed and figures representations were generated with Swiss PDB Viewer [17]. The rootmean- square deviation (RMSD) values were calculated using modeler, by fitting the carbon backbone of the predicted. Finally, the all-atom models were subjected to a validation using Ramachandran plot by NIH MBI server.

Cysteine protease has been found in viruses and prokaryotes, as well as in higher organisms such as plants and mammals, including humans [18]. The cysteine proteases produced by pathogenic bacteria are considered important virulence factors and have their role in the development of many diseases [19]. The precise control of proteolytic processes is essential for appropriate functioning of cells and whole organisms. This is achieved at many levels, from regulation of protease expression, secretion and maturation through specific degradation of mature enzymes, to blockage of their activity by inhibition. A search for similar sequences using BLAST showed that the Staphostatis B and Cysteine protease from S aureus showed 95.9% similarity while that from S warneri showed only 56.6% similarity (Table 1).

Table 1: BLAST results of cysteine protease.

Primary Structure Prediction

In this study, primary structure of 56k protease of Serratia marcescens were predicted using Expasy’s ProtParam server (http://expasy.org/cgi-bin/protparam), using the gene sequence retrieved from GenBank (Accession no: FJ810078) and the results are shown in Table 2. Results showed that 56k protease had 110 amino acid residues and the estimated molecular weight to be 12846.5. The calculated isoelectric point (pI) is useful, at pI, the solubility is least and the mobility in an electric field is zero. Isoelectric point (pI) is the pH at which the surface of protein is covered with charge but net charge of protein is zero. The calculated isoelectric point (pI) was computed to be 6.35. The computed value, 7 indicates that the protein is acidic. The very high aliphatic index (101.91) indicates that this protein is stable for a wide range of temperature range. While the instability index (37.3) provides the estimate of the stability of protein in a test tube, The Grand Average of Hydropathy (GRAVY) value which is as low as -0.345 indicates better interaction of the protein with water.

Table 2: Parameters computed using Expasy’s ProtParam tool.

Secondary Structure Prediction

The secondary structure is composed of alpha helix and beta sheets and the secondary structure is predicted using GOR IV and SOPMA. Table 3 presents the comparative analysis of GOR IV and SOPMA, from which it is clear that random coil is predominantly present, when the structure was predicted both by SOPMA and GOR, followed by extended strand and alpha helix.

Table 3: Secondary structure of cysteine protease by SOPMA and GOR IV.

Tertiary Structure Prediction

Swiss Model was used to study the 3D modeling of the 56 k protease. The assessment of the predicted models generated by Swiss- MODEL is shown in Figure 1. However, earlier studies have shown that the majority of cysteine protease profragment share similar fold and consist of two parts. In this study, we have modeled cysteine protease using SWISS-Modeller 9v7 (Figure 2) and the secondary structure as predicted by GOR IV and SOPMA are compared in Table 2. This structure has been further validated using Ramachandran plot by NIH MBI server (http://nihserver.mbi.ucla.edu/SAVES/ ) (Figure 3). Structural analysis was performed and representations were generated using Swiss PDB Viewer.

Figure 1: Structure of cysteine protease of Serratia marcences modeled by modbase.

Figure 2: 3D structure of the modeled protein.

Figure 3: Ramachandran plot obtained from NIH MBI server.

Protein Structure Validation

Ramachandran plot

The Ramachandran plot for the predicted structure performed using the NIH MBI server, shows that the protein structure holds totally 109 amino acids, in which 95 amino acids are in favored region, 4 amino acids in additionally allowed region, 0 amino acids in generously allowed region and 0 amino acids in disallowed region.

PROCHECK program

All proteases share in common, the general mechanism of a nucleophilic attack on the carbonyl-carbon of an amide bond. This results in a general acid-base hydrolytic process that disrupts the covalent bond. Different proteases utilize different strategies to generate the nucleophile and to juxtapose the nucleophile with the targeted bond. These distinctions serve as a useful classification scheme, and on this basis proteases can be grouped into four major classes: serine, cysteine, aspartate, and metalloproteins. Serine and cysteine proteases utilize their HO- and HS- side chains, respectively, directly as nucleophiles. Although not identical, the catalytic mechanisms of serine and cysteine proteases are remarkably similar. The latter two groups of enzymes utilize aspartate residues and heavy metals, respectively, to immobilize and polarize a water molecule so that the oxygen atom in water becomes the nucleophile. By the validation and 3D Errat done by using http://nihserver.mbi.ucla.edu server, 69.37% of the residues have score of 0.2 while errat shows 95.098 as a quality factor (Table 4).

Table 4: Analysis of Ramachandran plot statistics of predicted model.

In this study, 56k protease (Cysteine protease) of Serratia marcescens was amplified and sequenced, and the sequence was submitted to GenBank. In silico homology modeling of this protein was carried out. The primary structure prediction and physiochemical characterization were performed using ProtParam. The secondary and tertiary structures were modeled and then validated using PROCHECK and PROSA. This structure will provide a good foundation to further analyse this protein, for protein-protein interaction and functional analysis. Still, there is promise that the continued elucidation of specific physiological functions for cysteine proteases, will presage new therapeutic tools.