Keywords

2.1 Introduction

Multidrug resistance (MDR) in pathogenic bacteria is a serious problem and becomes a public health threat worldwide. MDR in bacteria occurs mainly by the accumulation of antimicrobial-resistant (AMR) genes on resistance plasmid or transposons. In general, AMR genes code for resistance to a specific agent, but when the genes accumulate in bacterial plasmid, the bacteria show resistance to multiple drugs and make the treatment more critical [1]. Penicillin-binding proteins (PBPs) are one among such a class of proteins which shows reduced susceptibility to penicillin and other beta lactams. Penicillin-binding proteins (PBPs) are the enzymes that assemble the peptidoglycans, which are the main constituents of the cell wall on the outer side of the plasma membrane and help the bacteria to resist the intercellular pressure caused by any external factors such as antimicrobial drugs. Because of the important role in cell wall maintenance, PBP proteins are considered as the major molecular targets for β-lactam antibiotics. The β-lactams such as penicillin inhibit the transpeptidase activity and thus inhibit the peptidoglycan cross-links in the bacterial cell wall [2, 3]. The pathogens acquires resistance to β-lactam drugs by developing more strategies and makes the treatment difficult and cause life threatening disease. Several studies have reported that the alterations in the structure of PBP proteins are the reason for the resistance to penicillin [4, 5]. The β-lactam resistance in Gram-positive bacteria is via two main mechanisms: One mechanism is by enzymatic degradation through the production of β-lactamases, and other mechanism is by the decreasing the affinity of the antibiotics for its target [2], whereas in the Gram-negative bacteria, the β-lactamase-mediated resistance is due to either acquisition of new genes or the mutations affecting the expression of its chromosomal β-lactamases [6].

In our present study, we have analyzed the gene interaction network of the extended spectrum beta lactamase (ESBL), penA gene. penA gene belongs to the class A of beta lactamases and codes for the protein, penicillin-binding protein 2 (PBP2). Various studies have confirmed that penicillin resistance due to penA alleles has been arisen from either recruitment of sequence blocks from natural resistance species or by point mutations such as codon insertion or substitution [3]. The gene interaction network (GIN) approaches are becoming famous and drawing the scientific community attention and are considered to be the more reliable approach to study the antimicrobial-resistant mechanisms in pathogenic bacteria. The GIN analysis in AMR genes provides new insights to various drug-resistant mechanisms by analyzing the AMR genes along with their functional partners [7,8,9,10,11]. We have also collected the penA protein sequences from various bacterial strains and constructed the phylogenetic tree to understand the phylogenetic relationship between the bacterial penA gene among the bacteria. We have collected functional interactions of penA gene from both Gram-positive and Gram-negative bacteria and constructed gene interaction network to understand the molecular-level interactions of penA gene with the functional partners. We further analyzed the functional enrichment of genes in the network to understand the role of penA gene along with its functional partners. We have also constructed a phylogenetic tree of penA gene from different bacteria to understand the evolutionary relationship between various bacterial species. Our results will be helpful in better understanding the role of penA gene in drug-resistant mechanism in various pathogenic bacteria. The genes which play an important role in various biological pathways were reported, and they can be useful as potent drug targets in new drug discovery.

2.2 Materials and Methods

2.2.1 Gene Interaction Data Curation from STRING Database

We have collected the penA gene interactions from STRING database from different Gram-negative and Gram-positive bacteria. STRING databases consist of protein–protein interactions curated from various sources such as high-throughput experimental data, data mining, literature survey, and co-expression analysis studies. All the interactions are broadly classified as direct (physical interactions curated from laboratory techniques) and indirect (functional associations extracted from computational prediction). Each interaction is provided with a confidence score that lies between 0 and 1. Each interaction further classified based on the confidence scores as highest (≥0.90–1.0), high (≥0.70–0.89), medium (≥0.40–0.69), and low (≥0.15–0.39) [12].

2.2.2 Interaction Network Construction

We have used Cytoscape 3.6.1 for the gene interaction network construction and analysis. Cytoscape is an open-source visualization tool used for constructing molecular interaction networks and biological pathways. Cytoscape tool comes with the core distribution with basic features for data integration, visualization, and analysis. The additional features such as network and molecular profiling analysis, new layouts, additional file format support, and cross-reference with databases can be available as Cytoscape plugins or apps [13].

2.2.3 Clustering Analysis

Clustering analysis of interacting network was carried out by using Cytoscape plugin MCODE. MCODE is based on molecular complex detection (MCODE) algorithm. The algorithm operates mainly in three different stages such as vertex weighing, complex prediction, and optionally postprocessing. In an interaction network, every vertex is a molecule and edge is an interaction between the molecules. Based on the type of data used, the graphs are divided into direct (known cell signaling and known pathways) and indirect graph [14].

2.2.4 Shortest Path Length and Closeness Centrality Analysis

The interaction network of penA gene was analyzed by using Cytoscape app NetworkAnalyzer. It is used to compute and display topological parameters such as a number of nodes, connecting edges, the network diameter, density, radius, centralization, heterogeneity, clustering coefficient, the characteristic path length, the distribution of node degrees, neighborhood connectivity, average clustering coefficients, and the shortest path lengths. It helps in analyzing both the types of networks (directed and undirected) and also allows to construct the intersection or union of two networks [15].

2.2.5 Phylogenetic Tree Construction

For the phylogenetic tree construction, we have used MEGA7. The phylogenetic tree construction involves in multiple sequence alignment (MSA) using MEGA inbuilt tools ClustalW or MUSCLE and followed by the constriction of phylogenetic tree from the aligned sequences. The gaps and the missing data from all the positions were removed from phylogenetic analysis. For phylogenetic tree construction, there are various methods used such as neighbor-joining method, maximum likelihood method, evolutionary distance method, and maximum parsimony method [16].

2.3 Results

2.3.1 penA Gene Interaction Network Analysis

Network analysis of penA gene along with the functional partners was done using STRING database. A total of 144 penA gene interactors with 1039 functional interactions from 28 bacterial strains were curated with medium (>0.4) confidence scores. Out of 28 bacterial species used in this study, 20 belong to Gram-negative (Achromobacter piechaudii, Ahrensia sp. R2A130, Bordetella petrii, Burkholderia cenocepacia, Burkholderia mallei, Burkholderia pseudomallei, Campylobacter coli, Campylobacter jejuni 414, C. jejuni 81176, C. jejuni NCTC11168, Collimonas fungivorans, Kingella kingae, Legionella pneumophila Paris, Neisseria lactamica, Neisseria meningitidis, N. C102, Neisseria sp. F0314, Oligotropha carboxidovorans, Oxalobacter formigenes OXCC13, Rhodopseudomonas palustris CGA009) and 8 belong to Gram-positive (Brevibacillus laterosporus, Lactobacillus antri, Saccharopolyspora erythraea, Streptococcus oralis ATCC35037, Streptococcus pneumoniae D39, S. pneumoniae R6, S. pneumoniae TIGR4). A list of interactions from each bacterial species is provided in Table 2.1. There are no significant interactions with medium and above confidence scores available in STRING for the bacterial species A. piechaudii. To obtain the maximum number of penA gene interactors from all the possible bacteria, we have curated the low confidence score interactions in the case of A. piechaudii (Fig. 2.1). Out of 1039 functional interactions, there are 522, 269, 229, and 19 belong to highest, high, medium, and lowest confidence scores, respectively. Out of 144 functional interactors, 130 directly interact with penA gene and are highlighted in red color in the network (Fig. 2.2).

Table 2.1 List of bacteria along with the number of interactions extracted from STRING database
Full size table
Fig. 2.1
figure 1

List of penA gene interactions collected from STRING database for different organisms. The interactions are given scores as highest (≥0.90–1.0), high (≥0.70–0.89), medium (≥0.40–0.69), and low (≥0.15–0.39) confidence scores. Out of 28 bacterial species, 27 have interactions with medium and above medium confidence scores. In A. piechaudii, we have collected interactions with low confidence scores as there are not many interactions above the medium confidence scores

Full size image
Fig. 2.2
figure 2

penA gene interaction network along with the functional interactors. Genes are clustered into 12 densely interconnected clusters (C1–C12) using Cytoscape MCODE. For easy identification, penA gene is given in blue color, red-colored nodes are the direct interactors of the penA gene, and those nodes which are not direct interactors have given green color

Full size image

2.3.2 Network Analysis Using NetworkAnalyzer

NetworkAnalyzer is used for network analysis of penA gene. For each node, individual centrality scores, along with the average shortest path length and degree, were given (Supplementary File S1). Among the 145 genes used in the study the top 20 genes with direct interactions (degree), average shortest path length score, closeness centrality, and between centrality scores have been provided in Table 2.2. All the 145 nodes are arranged in 3 layers based on the number of individual direct interactions. There are 37 nodes with more than 10 interactions, 53 nodes with interactions in between 05–10, and 45 nodes with less than 05 direct interactions (Fig. 2.3).

Table 2.2 Network analysis using NetworkAnalyzer tool
Full size table
Fig. 2.3
figure 3

penA gene interactions: All the 145 nodes are arranged in three layers based on the number of individual direct interactions. penA gene is denoted with octagonal red color node followed by 37 nodes with more than 10 interactions in hexagonal blue color node followed by 53 nodes with interaction in between 05 and 10 in rounded edge rectangular green color nodes and 45 nodes with less than 5 direct interactions

Full size image

2.3.3 Clustering Analysis Using MCODE

MCODE has been resulted in 12 highly interconnected clusters. Out of 145 genes used in the interaction network, 90 genes were included in 12 clusters and the clusters are named as C1–C12. We have used default MCODE parameters for filtration of nodes and edges. Among the clusters, there are only C1, C3, and C5 clusters having more than 15 nodes. The penA gene was included in cluster C1, and it consists of 19 nodes and 171 edges with 13.44 MCODE score (Table 2.3).

Table 2.3 Clustering analysis of penA gene interactions
Full size table

2.3.4 Functional Enrichment Analysis Using STRING Database

All the genes from 12 clusters are analyzed for the functional enrichment of genes using STRING database. Various GO terms such as biological processes (BP), molecular functions (MF), and cellular components are enriched along with the KEGG pathway-related genes, PFAM, and InterPro domain-related genes. Out of 145 genes, 49 genes are enriched in BP, MF, CC, KEGG, PFAM, and InterPro domains (Fig. 2.4). The clusters C2, C6, and C12 have been observed with no significant enrichment results (Supplementary File S2). Cluster C1 consists of genes related to mur ligase family-related genes and peptidoglycan synthase-related genes. The enriched GO terms include the BP such as regulation of cell shape (GO.0008360), peptidoglycan biosynthetic process (GO.0009252), cell wall organization (GO.0071555), cellular component biogenesis (GO.0044085), nitrogen compound metabolic process (GO.0006807), MF acid-amino acid ligase activity (GO.0016881) and various CC’s related to cell (GO.0005623), cell part GO.0044464), intracellular (GO.0005622), cytoplasm (GO.0005737). PFAM and InterPro domains such as mur ligase family, glutamate ligase domain cell cycle protein related are enriched in cluster C1. The genes mraY, mraZ, murC, murD, murG, and rsmH are found in multiple GO terms and protein domain-related enriched entities.

Fig. 2.4
figure 4

penA gene interaction network functional enrichment analysis: Out of 145 genes in the network, 49 genes are functionally enriched with biological processes, molecular functions, cellular compartments, KEGG, PFAM, and InterPro domain-related genes. All the 49 enriched genes are highlighted in red color

Full size image

In cluster C3, there are no significant GO terms enriched, but the genes related to PFAM, InterPro domains related to transglycosylase (PF00912), penicillin-binding protein transpeptidase domain (PF00905), d-alanyl-d-alanine carboxypeptidase, penicillin-binding protein 5-C terminal domain (PF07943), beta lactamase/transpeptidase-like (IPR012338), glycosyl transferase family 51 (IPR001264), and penicillin-binding protein transpeptidase (IPR001460) were enriched. The cluster C4 genes are enriched with the KEGG pathway peptidoglycan biosynthesis (KEGG ID: 550) along with PFAM and InterPro domains related to penicillin-binding protein transpeptidase domain and transglycosylase related domains. In cluster C5, the BPs related to regulation of cell shape (GO.008360) and peptidoglycan biosynthetic process (GO.009252) and KEGG pathways such as peptidoglycan biosynthesis (KEGG ID: 550) and beta lactam resistance (KEGG ID: 312) were enriched. Along with protein domains related to penicillin-binding protein dimerization, penicillin-binding protein transpeptidase was also enriched. Whereas in cluster C7, cell cycle protein-related PFAM and InterPro domains are enriched.

In cluster C8, KEGG pathways beta lactam resistance (KEGG ID: 312), peptidoglycan biosynthesis (KEGG ID: 550), and protein domains related to penicillin-binding protein dimerization and penicillin-binding protein transpeptidase domains were enriched. In cluster C9, the KEGG pathways penicillin and cephalosporin biosynthesis (KEGG ID: 311) and biosynthesis of secondary metabolites were enriched. The protein domains penicillin amidase, penicillin acylase, and nucleophile aminohydrolases related PFAM and InterPro domains were enriched. In cluster C10 there is only one KEGG pathway glutathione metabolism (KEGG ID: 480) was enriched. In cluster C12, the KEGG pathways penicillin and cephalosporin biosynthesis (KEGG ID: 311), beta lactam resistance (KEGG ID: 312), and glutathione metabolism (KEGG ID: 480) were enriched along with cytosol aminopeptidase family domains and beta lactam transpeptidase-like domains.

2.3.5 Phylogenetic Tree Construction and Analysis

We have collected the protein sequences for penA gene for 28 bacterial strains. Out of 28 bacterial strains, we have collected available 23 protein sequences from NCBI GenBank database. By using MEGA-muscle MSA tool, we have aligned the 23 amino acid sequences. We have used the aligned sequences for constructing phylogenetic tree (Fig. 2.5). We have used neighbor-joining method with the bootstrap consensus tree inferred from 1000 replicates. The evolutionary distances were computed using the poisson correction method and are in the units of the number of amino acid substitutions per site. All positions containing gaps and missing data were eliminated. There were a total of 207 positions in the final dataset.

Fig. 2.5
figure 5

Phylogenetic analysis of penA related proteins: Out of 28 bacteria, we have curated 23 available protein sequences and multiple sequence alignment is carried out by using MEGA7-muscle MSA program. The evolutionary history was inferred using the neighbor-joining method. The optimal tree with the sum of branch length = 9.64624886. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches

Full size image

2.4 Discussion

Multidrug resistance exerted by the pathogenic bacteria has become a global threat for treatment. PBP proteins assemble at the cell wall of the bacteria and help bacteria to resistant the intracellular pressure caused by external agents such as antimicrobials. In our present study, we have used PBP protein-coding gene penA to understand the molecular-level interactions of PBP in various bacteria. The mutation in penA gene is associated with the resistance to tetracycline/azithromycin, spectinomycin, and ceftriaxone. In N. gonorrhoeae, the mutation (insertion of aspartate at 345 position) in penA gene along with the ponA, gyrA, and parC genes determines the susceptibility to penicillins, tetracyclines, and fluoroquinolones [17, 18]. In our results, we have observed that the penA gene with 142 total interactions constitutes 13.66% of total interactions in the whole interaction network followed by murG (75 interactions; 7.22%), ftsW (68 interactions; 6.54%), murC (52 interactions; 5%), ftsA (48 interactions; 4.62%), and ftsQ (36 interactions (3.46%). Along with the penA gene, these five genes constitute 40% of total interactions from the whole interaction network. With the maximum number of direct interactions, these nodes may be considered to have a high impact on the network; thus, these can be considered as functional hub nodes and further used as potent drug targets.

The functional enrichment analysis of genes in the interaction network has been shown that the penA gene along with the functional interactors is mainly responsible for peptidoglycan biosynthesis, mur ligase synthesis, β-lactam resistance, and glutathione metabolism. It is well known that Gram-positive and Gram-positive bacterial peptidoglycan composition is similar, while Gram-positive bacteria consist of more thick and cross-linked peptidoglycan layer. Peptidoglycan is the main component of the bacterial cell wall, and it plays a key role in cell shape maintenance, facilitates the attachment for surface-exposed virulence factors, and avoids the modification in internal osmotic pressure. Thus, peptidoglycan biosynthesis mechanism-related genes are the preferred targets in the discovery of new antibiotics for many decades. The antibiotic-resistant mechanisms developed by the pathogenic bacteria against the antibiotics that targets the precursors of bacterial cell wall and biosynthesis machinery-related genes draw the attention of research community in overcoming the resistance strategies exhibited by the pathogens. There is an immediate necessity to study the resistance strategies and identify the new potent drug targets [19]. In our present study, we have observed the genes OCAR_5243, dacC, BPSL3031, OCAR_6617, ddlB, BPSS1219, ftsI, BPSS1240, mtgA, and pbpA are related to peptidoglycan biosynthesis, and these genes are found to have more dense interactions within the network. There are many clinically used antibiotics, especially various β-lactams, glycopeptides, fosfomycin, and cycloserine, which targets the genes involved in peptidoglycan biosynthesis pathway. Although the amide ligases (murC, murD, murE, and murF) play an important role in peptidoglycan biosynthesis by catalyzing the non-ribosomal peptide bond formations for the addition of peptide moiety on the peptidoglycan blocks, there are no antibacterial agent targets these amide ligases [20, 21]. In our results, the amide ligase-coded genes murC, murD, murE, and murF are enriched with Pfam and InterPro domains. It is also noticed that the MUR family genes murG, murC, and murE genes were in the top 10 genes with more number of direct interactions, which show the importance of amide genes in resistance caused by the penA gene along with its interactors. The other notable result is the enrichment of KEGG pathway glutathione metabolism (KEGG ID: 480). The genes pepA, pepB, RPA3685, and SACE_1030 were involved in glutathione metabolism. Various studies have shown that the antibiotics used against the pathogenic Gram-positive and Gram-negative bacteria results in the formation of reactive oxygen species (ROS). When the cell is exposed to an antibiotic, glutathione plays a key role in maintaining the cellular redox poise by the detoxification the xenobiotics [22]. Another interesting observation in our results is the enrichment of KEGG pathway, β-lactam resistance (KEGG ID: 312), which includes the genes pbp1B, BPSL3031, SACE_1373, pbp2A, BPSS1219, SACE_3877, pbpX, and BPSS1240. The gene pbp1B is a peptidoglycan glycosyltransferase or murein synthase, and it plays an essential role in synthesizing peptidoglycan in the absence of a pre-existing template [23]. Whereas the gene pbp2A is found in staphylococcal cell wall biosynthesis along with the known PBP family proteins PBP2 and PBP4, PBP2A which is an acquired transpeptidase plays a crucial role in susceptibility to antimicrobial agents [24]. pbpX gene is a homologous to the pneumococcal PBP2x gene, and it is isolated from penicillin-sensitive S. oralis strain [25]. The genes SACE_1373 and SACE_3877 belong to S. erythraea strain and coded for beta lactamase. The genes BPSL3031, BPSS1219, and BPSS1240 belong to B. pseudomallei strain. BPSL3031 gene codes for peptidoglycan synthase, and BPSS1219 and BPSS1240 code for penicillin-binding protein.

Out of 28 bacterial penA-related proteins, there are only 23 sequences available in NCBI GenBank database. We have collected the 23 penA-related protein sequences and used for the construction of phylogenetic tree. Among the bacterial species used for the study, few are non-pathogenic, and most of them are pathogenic. The phylogenetic analysis of penA genes from various bacterial species has confirmed the lineage of different bacteria with respect to penA gene. We have used neighbor-joining method for the phylogenetic tree construction with 1000 bootstraps. In the phylogenetic tree, the bootstrap values more than 70% show more confidence and less than that show poor confidence of the phylogeny with respect to the entry. In our results, there are only two bootstrap values less than 70%. The constructed phylogenetic tree shows that the bacteria C. jejuni showed boot strap values 44% with the other Gram-negative bacteria such as Neisseria genus and K. kingae. The bacteria belonging to Neisseria genus are closely related (bootstrap values 96%) to K. kingae. The bacteria B. pseudomallei and O. formigenes OXCC13 have bootstrap value 96%. The Gram-positive bacteria used in the study have shown the bootstrap values more than 90%, which shows the confidence of the penA gene lineage among the Gram-positive bacteria. All the streptococcus-related penA proteins have shown bootstrap values 100%, and they are highly similar to other Gram-positive bacterial penA protein sequences. We have also observed that the bacteria belonging to the same genus showed less than 100% bootstrap values. The bacteria B. mallei and B. cenocepacia show the bootstrap values 80%; N. meningitidis and Neisseria mucosa C102 show 83%; and with respect to N. lactamica shows, the bootstrap value is 70%. Our phylogenetic studies have provided a detailed lineage of penA gene in various pathogenic and non-pathogenic bacteria.

2.5 Conclusion

The gene interaction analysis of PBP2 coding penA gene provides a comprehensive evidence on penA gene and their functional partners in antibiotic resistance in various bacteria. In our present study, we have identified functional interactors of the penA gene from different bacterial species, and by using functional enrichment analysis, we have analyzed the role of these genes in peptidoglycan biosynthesis, mur ligase synthesis, β-lactam resistance, and glutathione metabolism. The constructed penA gene network helps in understanding the functional relationship of these interactors in biological pathways. Our results give critical information on various biological processes such as gene functions and complex cellular mechanisms. The phylogenetic tree of penA related amino acids from various bacteria gives a glance of lineage of penA gene in various Gram-positive and Gram-negative bacteria. To conclude, from our study, we have observed the penA gene along with the interactors plays a major role in peptidoglycan biosynthesis, amide ligase biosynthesis, β-lactam resistance, and glutathione metabolism. Our results will help in better understanding the functional role of β-lactamase penA gene in β-lactam induction. The molecular interactions of penA along with the functional partners will be useful for researchers exploring the β-lactam-mediated antibiotic resistance in pathogenic bacteria, and the identified resistance genes play major roles in various biological processes, and these can be considered as potent drug targets for developing new drugs.