Protein-protein interactions are known to be central to most cellular functions. These assemblies govern the dynamics of cellular processes both temporally and spatially. High throughput studies have unrevealed an unexpected extended protein networks in various bacteria that challenge our understanding in terms of functionality [1]. A typical two component system consists of a histidine protein kinase containing a conserved kinase core and a response regulator protein containing a conserved regulatory domain. Extracellular stimuli are sensed by and serve to modulate the activities of the histidine kinase. The histidine kinase transfers a phosphoryl group to the regulator protein in a reaction catalyzed by the regulator protein. Phosphotransfer to the regulator protein results in activation of a downstream effector domain that elicits the specific response [2,3]. In an elaborated way we can say that, the ATP-dependent phosphorylation of histidine is generally regulated in response to environmental signals by a family of histidineprotein kinases. The phosphorylated aspartate is present within another variety of protein termed as response regulator which undergoes a phosphorylation- induced conformational change that serves to evoke a response [4]. Thus, there are two essential families of proteins that function together the sensory kinases and their associated response regulators.

Histidine Protein Kinases (HPKs) are a large family of signaltransduction enzymes that auto phosphorylate on a conserved histidine residue. HPKs form two-component signaling systems together with their downstream target proteins, the response regulators, which have a conserved aspartate in a so-called ‘receiver domain’ that is phosphorylated by the Histidine protein kinase [5]. Response regulators of bacterial sensory transduction systems generally consist of receiver module domains covalently linked to effector domains. The effector domains include DNA binding and/or catalytic units that are regulated by sensor kinase-catalyzed aspartyl phosphorylation within their receiver modules. Most receiver modules are associated with three distinct families of DNA binding domains, but some are associated with other types of DNA binding domains with methylated chemotaxis protein demethylases or with sensor kinases [6].

Enterococci are gram-positive constituents of the normal human micro flora typically colonizing the intestinal tract and skin. However, these organisms are capable of causing disease as opportunistic pathogens, mainly in immune compromised patients. Normally, Enterococci are used as probiotics to improve the microbial balance of the intestine and to treat gastroenteritis in humans and animals. These bacteria probably now represent the greatest risk to human health of any bacterial species currently used for these purposes. General features of these organisms are hallmarks of their biology and may also contribute to their pathogenicity. Enterococci are distinguished for their ability to grow at temperatures ranging from 10 to 45°C and in 6.5% NaCl or/and to tolerate acidic and alkaline growth conditions. The above observations raise the question of how Enterococcal physiology has evolved to allow the organisms to sense environmental changes and respond to the various stimuli with adaptive behavior. Monitoring and adapting to changing environmental conditions is the key function of bacterial signal transduction, which is generally carried out by the so-called two-component systems. Studies of enterococci have identified some general stress proteins; a global view of the Enterococcal signal transduction mechanisms has not been gained [7].

For understanding the function of a particular protein, it is usually useful to identify otherproteins with which it associates. This can be done by topological identification of specific proteins. To fulfill their biological activities in the cell, most proteins function in association with protein partners or as part of large molecular assemblies. Hence, the knowledge of the interactionscontext of a protein is crucial to understand its cellular functions. A comprehensive description of the stable and transient protein–protein interactions in a cell would facilitate the functional annotation of all gene products, and provide insight into the higher-order organization of the proteome. Several methodologies have been developed to detect protein–protein interactions, and some have been adapted to chart interactions at a proteome-wide scale [8].

The two component signal transducing protein-protein interactions network analysis has been carried out for ten different strains of Enterococcus faecalis such as Enterococcus faecalis V583, Enterococcus faecalis X98, Enterococcus faecalis CH188, Enterococcus faecalis D6, Enterococcus faecalis JH1, Enterococcus faecalis HIP11704, Enterococcus faecalis TX0102, Enterococcus faecalis DAPTO 516, Enterococcus faecalis TX0635, Enterococcus faecalis 62 and Enterococcus faecalis V583 [9-11]. This V583 strain was found to lack the cytolysin gene and a surface adhesin, Esp, that contributes to urinary tract infections. Mobile genetic elements make up one quarter of the genome [12].

Protein tracing

Proteins which are participating in Two Component Signal Transduction (TCST) of specific Enterococcus faecalis strains are traces out using NCBI protein database search where we have found number of proteins in a strain participating in TCST were selected and downloaded as fasta format.

Protein interactions, sequence and network visualization

The proteins of specific Enterococcus faecalis strains which are participating in the two component signal transduction were identified from NCBI/Protein database. The protein sequences have been downloaded and group of proteins of specific strain are uploaded to the STRING database for the identification of particular protein-protein interactions. STRING does not consider any specific splicing isoforms or posttranslational modifications, but instead represents each protein-coding locus in a genome by a single protein. STRING imports protein association knowledge not only from databases of physical interactions, but also from databases of curated biological pathway knowledge. Functional partnerships between proteins are at the core of complex cellular phenotypes, and the networks formed by interacting proteins provide researchers with crucial scaffolds for modeling, data reduction and annotation. STRING is a database and web resource dedicated to protein-protein interactions, including both physical and functional interactions. It weights and integrates information from numerous sources, including experimental repositories, computational prediction methods and public text collections, thus acting as a Metadatabase that maps all interactions evidence onto a common set of genomes and proteins. The most important new developments in STRING 8 over previous releases include a URL-based programming interface, which can be used to query STRING from other resources, improved interactionsprediction via genomic neighborhood in prokaryotes, and the inclusion of protein structures. Version 8.0 of STRING covers about 2.5 million proteins from 630 organisms, providing the most comprehensive view on protein–protein interactions currently available (Figure 1) [13].

Figure 1: Summarized Flow chart of step by step methodology applied for the prediction of protein-protein interactions through STRING database.

STRING generates the protein interactions based on the Neighborhood, Gene fusion, Co-occurrence, Experiments, Databases, Text mining and Homology. Comparatively STRING gives score between 0.001 to 0.999. The STRING score is divided into Low Confidence (0.150), Medium Confidence (0.400), High Confidence (0.700) and Higher Confidence (0.900) which shows a perfect interactions network.

Analysis of host-pathogen protein-protein interactions

A pathogen causing an infectious disease generally exhibits extensive interactions with the host [14]. These complex crosstalks between a host and a pathogen may assist the pathogen in successfully invading the host organism, breaching its immune defense, as well as replicating and persisting within the organism. Systematic determination and analysis of Host-Pathogen Interactions (HPIs) is a challenging task from both experimental and computational approaches, and is critically dependent on the previously obtained knowledge about these interactions.

The molecular mechanisms of Host-Pathogen Interactions (HPIs) include interactions between proteins, nucleotide sequences, and small ligands [15-18]. “Intra-species PPI”, where two proteins from the same species interact with each other and “Inter-species PPI” where two proteins from two different species interact. Host-pathogen protein–protein interactionsplay a vital role in initiating infection are a subset of inter-species interactions. Identification and study of HPIs is critical for understanding molecular mechanisms of infection and subsequent development of drug targets (Figure 2).

Figure 2: Summarized Flow chart of step by step methodology applied for the prediction of protein-protein interactions through STRING database.

The Host-Pathogen Interactions data of 10 strains of Enterococcus faecalis was generated using HPIDB (Host Pathogen Interactions Database). The two component signal transducing proteins are given for BLASTP in HPIDB full database BLAST with parameters of Blossom 62 Algorithm Matrix and E-Value of 10.0 [19]. The obtained PPI data was loaded in to Cytoscape 2.8.2 for the identification of protein-protein interactions of specific two component signal transducing proteins of different strains of Enterococcus faecalis.

Network topology analysis

The combined BLAST data has been generated from Host Pathogen Interactions Data Base 1.0. (Figure 3) [20] which was imported into the Cytoscape 2.8.2. When the blast data was imported taking E-value of the proteins and predicted partners as source interactions, bit score was taken as a target interactions and interactions type was given as default. The Cytoscape 2.8.2 generates a grid view of protein-protein interactions which are arranged according to their individual blast scores. The topological parameters of Enterococcus faecalis strains PPI networks were analyzed using Network Analysis plugin [21] of Cytoscape.2.8.2. The edges in all PPI networks were treated as undirected. The definition of network topological measure can be found in Network Analyzer Online Help (http: //med.bioinf.mpi-inf.mpg. de/netanalyzer/help/2.6.1/index.html).

Figure 3: Flow chart representing the work flow of Host Pathogen Interactions Database.

Fitting a line

Network Analyzer provides another useful feature - fitting a line on the data points of some complex parameters. The method applied is the least squares method for linear regression [22]. Network Analyzer gives the correlation between the given data points and the corresponding points on the fitted line. In addition, the R-squared value (also known as coefficient of determination) is reported. Fitting a line can be used to identify linear dependencies between the values of the x and y coordinates in a complex parameter shows the fitted line on a neighborhood connectivity distribution. The correlation between the data points and corresponding points on the line is approximately 0.969. The R-squared value is 0.939 giving a relatively high confidence that the underlying model is indeed linear.

Fitting a power law

The degree distribution of many biological networks approximates a power law: DD (k) ~ k^α for some negative constant α. Several studies have reported similar properties of the average clustering coefficient distribution [23] and the topological co-efficient [24]. Network Analyzer can fit a power law to some topological parameters. Please note that Network Analyzer uses the least squares method [22] and only points with positive coordinate values are considered for the fit. This approach fits a line on logarithmized data and may be inappropriate for supporting certain hypotheses. Network Analyzer gives the correlation between the given data points and the corresponding points on the fitted curve. In addition, the R-squared value (also known as coefficient of determination) is reported. This coefficient gives the proportion of variability in a data set, which is explained by a fitted linear model [25,26]. Therefore, the R-squared value is computed on logarithmized data, where the power-law curve: y = βx^a is transformed into linear model: 1n y = lnβ + alnx.

We have generated protein-protein interactions networks in different strains of Enterococcus faecalis. We have observed several proteins which have not been participating in the interactions network. All the interactions are predicted under the parameters adjusted to highest confidence (0.900) and the numbers of predicted interactions are adjusted to not more than 10. The color schema of bands in the network denotes the specific parameters at which the proteins are interacting with one another.

Protein Interactions of Enterococcus faecalis 62

Enterococcus faecalis 62 contains the gelatinase gel E and serine proteinase sprE genes but displays a gelatinase-negative phenotype [27]. In this Protein-Protein Interactions data of Enterococcus faecalis 62 we have figured out three signal transducing proteins which are not participating in the interactions. In figure 3 you can observe EF_3329 DNA-binding response regulator, EF_0960 endonuclease/exonuclease/phosphatase family protein and EF_2630 transcriptional regulator which are not participating in the protein interactions network, whereas EF_2568 aminotransferase, class V and EF_2566 hypothetical protein are interacting with each other without interacting with the other network lproteins. We have identified a sub network of EF_2218 AraC family DNA-binding response regulator, EF_2219 sensor histidine kinase, EF_1822 response regulator, lytS sensor histidine kinase; member of the two-component regulatory system, agrCfs histidine kinase, putative, and lytT response regulator; Member of the two-component regulatory system lytS/lytT that probably regulates genes involved in cell wall metabolism (Figure 4).

Figure 4: Predicted Protein Interactions network generated from STRING database representing the protein networking of two component signal transducing proteins of Enterococcus faecalis 62.

Comparative network analysis of predicted protein interactions

The protein-protein interactions built for the 9 strains of Enterococcus faecalis are on the basis of protein-protein interactions of Enterococcus faecalis V583 which are available in the STRING 9.0 database. According to the protein interactions networks predicted in 10 different strains of Enterococcus faecalis, we have figured out a proteinwhich is not participating in any protein interactions network except in protein interactions network of X98 where we have identified the absence of the protein EF_3329 DNA-binding response regulator. Sub networks which are not at all having any interactions with the main network have been identified in 9/10 networks except in the Enterococcus faecalis D6where no sub networks are formed. Most of the sub networks consists of following proteins EF_2218 AraC family DNA-binding response regulator, EF_2219 sensor histidine kinase, EF_1822 response regulator, lytS sensor histidine kinase; member of the two-component regulatory system, agrCfs histidine kinase; putative, and lytT response regulator; Member of the two-component regulatory system lytS/lytT that probably regulates genes involved in cell wall metabolism which did not have any interactions with the main protein interactions network in all the 10 Enterococcus faecalis strains.

Topological analysis of Enterococcus faecalis host-pathogen PPI networks

We have calculated the topological parameters of PPI networks of Enterococcus faecalis using Network Analysis [21] plug in present in the Cytoscape.2.8.2. The obtained results consists of connected components, Network diameter, Network radius, Network Centralization, Shortest paths, Characteristic Path length, Average numbers of neighbors, Number of Nodes, Network Density and Network Heterogeneity. During the power law fit of Network analysis some data points has identified containing non-positive coordinates so, only points with positive coordinates are induced in the fit. Here we have comparatively analyzed the results of Network Analysis of ten Enterococcus faecalis strains.

Host-Pathogen protein interactions analysis of Enterococcus faecalis 62

The Network analysis data of Enterococcus faecalis 62 contains connected components 20, Network diameter is 6, Network radius is 1, Network Centralization is 0.118, Shortest paths are 1796 (10%), Characteristic Path length is 2.606, Average numbers of neighbors are 1.712, Number of Nodes are 132, Network Density is 0.013 and Network Heterogeneity is 1.311. In the graph of Node Degree Distribution (Figure 5A) the number of proteins with a given link (k) in the protein interactions network of Enterococcus faecalis 62 follows a fitted line y = a+bx (a) 23.053, (b) -1.849. Correlation is 0.377, R² = 0.142 and power law y = ax^b (a) 23.639, (b) -1.410. Correlation is 0.944, R² = 0.685 for the power law fit. The Topological Co-efficient (Figure 5B) was plotted against the number of links. The topological coefficient with a given link (k) in the Enterococcus faecalis 62protein interactionsnetwork follows a same fitted line (a) 0.528, b) -0.036. Correlation is 0.884.

Figure 5: Topological properties of Enterococcus faecalis 62 PPI Network analyzed from Cytoscape 2.8.2.
(A) Node Degree Distribution (B)Topological coefficients Distribution (C) Neighboring Connectivity Distribution (D) Stress Centrality Distribution (E) Betweenness Centrality Distribution (F) Closeness centrality Distribution.

R² = 0.781 and power law (a) 0.973, b) -0.918, Correlation = 0.987, R² = 0.929 and for the Neighboring Connectivity Distribution (Figure 5C) (a) 4.002, b) -0.252, Correlation = 0.495, R² = 0.245 for fitted line and for power law fit (a) 4.420, (b) -0.582, Correlation = 0.831, R² = 0.398. Stress Centrality Distribution (Figure 5D) (a) 42.846, (b) -0.355, Correlation = 0.325, R² = 0.106 for fitted line and for power law fit (a) 2.286, (b) 0.370, Correlation = 0.958, R² = 0.931. The stress of a node n is the number of shortest paths passing through n. The stress distribution gives the number of nodes with stress s for different values of s. The values for the stress are grouped into bins whose size grows exponentially by a factor of 10.The Betweenness centrality distribution (Figure 5E) was built as (a) 0.282, b) 0.489, Correlation = 0.346, R² = 0.236 for the power law fit, closeness centrality distribution (Figure 5F) (a) 0.486, b) 0.154, Correlation = 0.218, R2 = 0.051 for power law fit.

The correlation and R-Squared values of power law fit showed a high fit when compare to the standard (Power law perfect fit Correlation = 1.000 and R-Squared = 1.000) in Topological Co-efficient (Correlation = 0.987, R² = 0.929) and Node Degree Distribution (Correlation = 0.944, R² = 0.685). The interactions analysis also has showed shortest paths 1796 (10%). The above all parameters prove that the protein interactions network of Enterococcus faecalis 62 is perfectly fit.

Comparative network analysis of host-pathogen protein interactions

The two component signal transduction host-pathogen protein interactions network analysis of 10 different Enterococcus faecalis strains have yielded us antithetic statistical data containing Node degree distribution, Topological coefficients, Neighborhood connectivity distribution, Stress centrality distribution, Betweenness centrality and Closeness centrality. Here we have given the compared statistical data analysis.

Network parameters

The network parameters have been generated by undirected network analysis. Network parameters of 10 variant Enterococcus faecalis strains have been comparatively analyzed (Table 1) The network parameters contain Connected Components, Network Diameter, Network Radius, Network Centralization, Shortest Paths, Characteristic Path Length, Average Numbers of Neighbors, Number of Nodes, Network Density and Network Heterogeneity. Enterococcus faecalis TX0635 shows high number of connected components (Table 1)(Figure 6) compared to other strains of Enterococcus faecalis. Network diameter of Enterococcus faecalis CH188 is high (Figure 7C), Network radius of all the strains is same. Network Centralization of Enterococcus faecalis HIP11704 is higher (Figure 7A).

Table 1: Comparison of network parameters in 10 different Enterococcus faecalis strains [F-Enterococcus faecalis, C.C: Connected Components; N.Di: Network Diameter; N.R: Network Radius; N.C: Network Centralization; S.P: Shortest Paths; C.P.L: Characteristic Path Length; A.N.N: Average Numbers of Neighbors; N.N: Number of Nodes; N.De: Network Density and N.H: Network Heterogeneity].

Figure 6: Pie chart representing the connected components of 10 Enterococcus faecalis strains.

Figure 7: Graphs representing the comparative network parameter analysis of Enterococcus faecalis strains (A) Network Centralization, (B) Number of Nodes, (C) Network Diameter, (G) Average Numbers of Neighbors, (D) Network Density, (E) Characteristic Path Length and (F) Network Heterogeneity.

Shortest paths

Shortest paths are high in Enterococcus faecalis CH188, Characteristic Path Length is high in Enterococcus faecalis CH188 (Table 1)(Figure 7E), Average Numbers of Neighbors are more in Enterococcus faecalis X98 (Table 1)(Figure 7G), Number of Nodes are high in Enterococcus faecalis X98 (Table 1)(Figure 7B), Network Density is high in Enterococcus faecalis D6 and Enterococcus faecalis V583 (Table 1)(Figure 7D) and Network Heterogeneity is high in Enterococcus faecalis TX0102 (Table 1)(Figure 7F).

Node degree distribution

The correlation value of Enterococcus faecalis X86 is comparatively high (Supplementary Table 1) whereas Enterococcus faecalis TX0635 gave a peak value in the comparison of R- Squared values under the fitted line algorithm (y = a+bx). In the Fitted Power Law Y = ax^b correlation value of Enterococcus faecalis X86 is high whereas Enterococcus faecalis CH188 is high in the comparison of R- Squared values (Supplementary Table 2).

Topological coefficients

The correlation and R- Squared values of Enterococcus faecalis TX0102 (Supplementary Table 3) are high compared to the other strains of Enterococcus faecalis in the fitted line algorithm (y = a+bx). Compared to the values of all other Enterococcus faecalis strains Correlation and R-Squared values of Enterococcus faecalis HIP11704 showed a highest fit of network (Supplementary Table 4). The topological coefficient fitness is considerably high only which the protein interactions network is highly linked giving a highest fit value in proteins interacting in a network.

Neighborhood connectivity distribution

The correlation and R- Squared values of Enterococcus faecalis CH188 (Supplementary Table 5) are high in the comparison of fitted line algorithm (y = a+bx). But the values are not significant due to the low value ratio. When the fitted power law (Y = ax^b) was applied to the neighborhood connectivity distribution Enterococcus faecalis CH188 (Supplementary Table 6) showed a higher Correlation and R-Squared values compared to the other Enterococcus faecalis strains. Compared to the fitted power law (Y = ax^b) standard (Power law perfect fit Correlation = 1.000 and R-Squared = 1.000) the higher values which were observed during comparative analysis are low.

Stress centrality distribution

In the stress centrality distribution of network analysis, Correlation and R- Squared values of Enterococcus faecalis TX0635 (Supplementary Table 7) are high in the comparison of fitted line algorithm(y = a+bx). But the values are not significant due to the low value ratio. When the fitted power law (Y = ax^b) was applied to the stress centrality distribution Enterococcus faecalis DAPTO516 (Supplementary Table 8) showed a higher Correlation and R-Squared values compared to the other Enterococcus faecalis strains. Compared to the fitted power law (Y = ax^b) standard (Power law perfect fit Correlation = 1.000 and R-Squared = 1.000), the higher values which were observed during comparative analysis are high and shows a perfect fit of network. Compared to the topological coefficients fitted power law (Y = ax^b) values the stress centrality distribution values are a bit less but considerable for a high network fit.

Betweenness centrality distribution

The correlation and R- Squared value of Betweenness Centrality are very low when compared with the standard (Power law perfect fit Correlation = 1.000 and R-Squared = 1.000). Enterococcus faecalis D6 (Supplementary Table 9) showed a high correlation point where as Enterococcus faecalis 62 showed a higher R-Squared value when compare with the correlation and R-squared values of other Enterococcus faecalis strains.

Closeness centrality distribution

Closeness centrality is a measure of how rapidly information spreads from a given node to other reachable nodes in the network. The closeness centrality distribution of the Enterococcus faecalis TX0635is high (Correlation = 0.431, R-Squared = 0.181) (Supplementary Table 10) this shows that highly connected proteinsin the network have a pronounced ability to spread information in the network.The same tendency also exists in other strains,but less when compared with the Enterococcus faecalis TX0635 protein interactions network. Nodes with high closeness centrality have potential significance for responding to external perturbations and for maintaining network stabilization. This may be part of the explanation why the highly connected “hub” proteins usually play essential roles in cell processes [28].

The prevailing study presents the entire protein-protein interactions of two component signal transducing proteins where we have comparatively analyzed protein-protein interactions of ten Enterococcus faecalis strains. All interactions of proteins which are participating in the two component signal transduction of Enterococcus faecalis has been predicted using STRING database. Host-Pathogen protein interactions have been identified and the Host-Pathogen interactions networks have been analyzed. The two component signal transducing proteins formed a highly linked network. The protein interactions which we have generated are useful for the studies of drug targeting. The protein-protein interactions information was connected to the pathogenic protein secretion pathway and gene regulation. These results will serve as a unique resource for further dissection of signal transduction and infection mechanisms in Enterococcus faecalis, as well as for the development of novel drugs.

The authors thank UGC/MHRD start up grant for newly joined faculty funds and also acknowledge proteomics facility.

NCBI, HPIDB, STRING, NCBI Protein database, Cytoscape 2.8.2.

Only PPI data of E. faecalis 62 was shown in this paper, the complete PPI network high resolution images of remaining strains of Enterococcus faecalis; Host pathogen interactions graph data of remaining strains of Enterococcus faecalis; PPI enrichment analysis, Tables representing the topological analysis and measures of each PPI network; power-law distribution of node degree and topological coefficient; the distribution of protein category and PPI; conserved subnetworks; protein-protein interactions pairs of Enterococcus faecalis; are available as a supporting information.