Prioritization of Mur family drug targets against A. baumannii and identification of their homologous proteins through molecular phylogeny, primary sequence, and structural analysis

Background The World Health Organization (WHO) report stated that Acinetobacter baumannii had been classified as one of the most important pathogenic bacteria causing nosocomial infection in hospital patients due to multi-drug resistance (MDR). It is vital to find out new bacterial drug targets and annotated their structure and function for the exploration of new anti-bacterial agents. The present study utilized a systematic route to prioritize the potential drug targets that belong to Mur family of Acinetobacter baumannii and identify their homologous proteins using a computational approach such as sequence similarity search, multiple sequence alignment, phylogenetic analysis, protein sequence, and protein structure analysis. Results From the results of protein sequence analysis of eight Mur family proteins, they divided into three main enzymatic classes namely transferases (MurG, MurA and MraY), ligases (MurC, MurD, MurE, and MurF), and oxidoreductase (MurB). Based on the results of intra-comparative protein sequence analysis and enzymatic classification, we have chosen MurB, MurE, and MurG as the prioritized drug targets from A. baumannii and subjected them for further detailed studies of inter-species comparison. This inter-species comparison help us to explore the sequential and structural properties of homologous proteins in other species and hence, opens a gateway for new target identification and using common inhibitor for different bacterial species caused by various diseases. The pairwise sequence alignment results between A. baumannii’s MurB with A. calcoaceticus’s MurB, A. baumannii’s MurE with A. seifertii’s MurE, and A. baumannii’s MurG with A. pittii’s MurG showed that every group of the proteins are highly similar with each other and they showed sequence identity of 95.7% and sequence similarity of 97.2%. Conclusion Together with the results of secondary and three-dimensional structure predictions explained that three selected proteins (MurB, MurE, and MurG) from A. baumannii and their related proteins (AcMurB, AsMurE, and ApMurG) belong to mixed αβ class and they are very similar.


Background
Acinetobacter is a genus name which encompasses species that are strictly aerophilic, and biochemically catalasepositive, indole-negative, oxidase-negative, gram-negative, and citrate-positive [1] and its utmost essential representative is A. baumannii [2]. It has virulence factors that are responsible for different infections triggered by this organism. Cell surface hydrophobicity, outer membrane proteins (OMPs), toxic slime polysaccharides, and verotoxins are the possible virulence factors. From the virulence factors, cell surface hydrophobicity is a crucial element for bacterial sticking together as well as avoiding being phagocytosed [1,3]. Extracellular enzymes named as esterases, certain amino-peptidases, acid phosphatases, and toxins which produce in the cytoplasm and other secreted substance are factors that play a substantial role in the pathogenesis, and it causes harm to host tissues mainly in the breathing system [1].
A. baumannii belongs to hospital-acquired pathogenic bacteria; therefore, the infections have been spread quickly in the hospital. The highest density of this infection occurred in the intensive care units (ICUs) in the hospitals [4]. On the other hand, the drug resistance ability of A. baumannii is also a global issue. The reports indicated that carbapenem-resistant strains are mostly responsible for epidemics. Additionally, these strains show intermediate resistance to tigecycline but vulnerability to colistin. Overall, this pathogenic bacteria have multiple mechanism for drug resistance ability to most of the drugs and they have the capacity to acquire the indicated drug resistance mechanisms ability within a short period of time [5]. According to the WHO, A. baumannii has been classified as one of the most important pathogenic bacteria causing hospital-acquired nosocomial infection due to MDR. Therefore, it is crucial to discover and prioritize the new and potential bacterial drug targets and subject them into structural and functional characterization, and facilitate to explore the new anti-microbial agents [6].
The bacterial cell is surrounded by a rigid structure called the cell wall. Cell wall in bacteria is made up of a polymeric network of peptidoglycan, which protects it from different environmental factors such as the excess amount of water in its surrounding that causes turgidity [7]. In the peptidoglycan pathway, the first step of the reaction is catalyzed using UDP-N-acetylglucosamine 1carboxyvinyltransferase (MurA); it involves the attack at the electrophilic carbon two positions of phosphoenolpyruvate (PEP) leading to the cleavage of the carbon oxygen bond [8]. Similarly, the formation of UDP-Nacetylmuramic acid (UDPMurNAc) catalyzes using UDP-N-acetylenolpyruvoylglucosaminereductase (MurB) in the peptidoglycan biosynthetic pathway [8,9]. Subsequently, the addition of a short polypeptide chain to the UDPMurNAc involves four key Mur ligase enzymes, namely MurC, MurD, MurE, and MurF. These four Mur ligases are responsible for the successive additions of Lalanine, D-glutamate, meso-diaminopimelate or L-lysine, and D-alanyl-D-alanine to UDP-N-acetylmuramic acid. All four Mur ligases are topologically similar to one another, even though they display low sequence identity [7,10,11]. The MraY is a membrane transferase enzyme that catalyzes the transfer of the phospho-N-acetylmuramoyl-pentapeptide motif onto the undecaprenyl phosphate carrier lipid [12]. The final steps of the peptidoglycan pathway, the transfer of a GlcNAc subunit on undecaprenyl-pyrophosphoryl-MurNAc-pentapeptide (lipid intermediate I) to form undecaprenyl-pyrophosphoryl-MurNAc-(pentapeptide) GlcNAc (lipid intermediate II) is catalyzed by MurG protein [13]. All of the enzymes are responsible for the synthesis of peptidoglycan in a sequential manner in bacteria, and this pathway is unique for them and it is not found in human [14]. So that, the prioritization of the enzymes involved in this pathway as a potential drug target is crucial to counterpart the severity of A. baumannii in hospital patients.
Nowadays, an identification, prioritization, and validation of suitable bacterial drug targets are crucial steps to identify or design new drug molecules against them. In connection with this, for the evaluation of gene function, essentiality, and suitability for drug development, the next generation sequencing have adequate and significant importance [15]. A few studies reported toward the identification, prioritization, and validation of drug target proteins from gram-negative pathogenic bacteria such as Klebsiella pneumonia [16], Salmonella enterica subsp. enterica serovar Poona [17], and Pseudomonas aeruginosa [18] by using different approaches. For computer-aided drug target identification, the Potential Drug Target Database (PDTD) is a very important resource and it is available at http://www.dddc.ac.cn/pdtd/. The database covers diverse information of more than 830 known or potential drug targets, including the information of protein and active sites structures in both PDB and MOL2 structural formats, related diseases, biological functions, as well as associated signaling pathways [19]. In this resource, TarFisDock (http://www. dddc.ac.cn/tarfisdock/) is used as a web server to identify the potential drug targets from the user given small molecule structure based on reverse docking approach. Apart from reverse docking approach, there are several approaches used to identify the drug targets, namely an integrative, multi-omics [16], computational subtractive genomics, molecular docking, virtual screening, [17], structural bioinformatics [15], and protein-protein interaction network [18].
The present study employed a systematic route to prioritize the pre-identified potential drug targets of Mur family from A. baumannii and identify their homologous proteins from other species using molecular phylogeny, primary sequence, secondary structure, and three-dimensional structural analysis. In this study, to our knowledge, first time, we used a combined molecular phylogeny with structural studies toward A. baumannii to prioritize the drug targets belonging to Mur family and identify other homologous drug target proteins in other species belonging to Acinetobacter genus using A. baumannii as reference organism. This methodology can also be used to prioritize and identify the drug targets in other bacteria caused by various diseases. This opens a new route to identify the effective antibacterial molecules which may act on multi-targeted proteins from various bacteria.

Primary sequence analysis of Mur family proteins
The amino acid sequence of Mur proteins involved in peptidoglycan biosynthetic pathway from A. baumannii were retrieved from UniProt database (https://www.uniprot.org/) which is a comprehensive collection of amino acid sequence and its associated information concerning sequence, structure, and function. The following set of Mur proteins were used in this study namely MurA (Ac-

Physiochemical properties
To predict the physiochemical properties, each protein sequence of Mur member was subjected into protein sequence analysis using Expert Protein Analysis System (ExPASy) Proteomics web server (www.expasy.org/tools) [20]. The physicochemical properties such as amino acid composition, molecular weight (MW), iso-electric point (pI), instability index (II), aliphatic index (AI), and extinction coefficient (EC) were computed using Prot-Param web server [20].

Prediction of functional domains, subcellular localization and antigenic sites
The functional domain, subcellular localization, and antigenic sites of Mur family protein were predicted. To identify the functional domains present in the Mur family protein, we used the Simple Modular Architecture Research Tool (SMART) [21]. The subcellular localization of the protein was also predicted using PSort-B (subcellular localization prediction for bacteria) web server (http://psort.hgc.jp/) [22]. Finally, the possible antigenic sites of Mur proteins were predicted using the European Molecular Biology Open Software Suite (EMBOSS)-Antigenic program [23].

Prediction of possible phosphorylation, glycosylation, and acetylation sites
In addition to the above-indicated parameters, the possible post-translation modifications (phosphorylation, glycosylation, and acetylation) sites of Mur family proteins were predicted by using MP Site (microbial phosphorylation site predictor) for phosphorylation [24], GLYCOPP v 1.0 for prediction of glycosylation sites in prokaryotes [25], and GPS-PAIL 2.0 software for acetylation [26].
Intra-comparative sequence alignment, phylogenetic tree constructions, and analysis The multiple protein sequence alignment (MSA) of these eight Mur family protein sequences from A. baumannii was performed by Clustal Omega server (https:// www.ebi.ac.uk/Tools/msa/clustalo/), followed by the phylogenetic tree which was constructed using four different methods, namely unweighted pair group method with arithmetic mean (UPGMA), maximum parsimony (MP), minimum evolution (ME), and neighbor joining (NJ). The reason for choosing four different methods in phylogenetic analysis is to get more reliable details in terms of evolutionary distances, group identification, group classification, and similarity statistics among eight Mur family proteins. The statistical significance of the phylogenetic tree topology was evaluated by bootstrap analysis with 1000 iterative tree constructions using the software package named as Molecular Evolutionary Genetics Analysis Package (MEGA) Version X [27]. The following sequential and evolutionary details were obtained from aforementioned analysis namely (i) number of conserved sites (C), (ii) number of variable sites (V), (iii) number of parsimony informative sites (PI), (iv) number of singleton sites (S), (v) evolutionary distances, and (vi) overall mean average. Consensus tree construction using four different methods provided a reliable and reasonable evolutionary relationship among the input dataset of A. baumannii for further analysis.
Inter-comparative sequence alignment, phylogenetic tree constructions, and analysis Of the various enzymes involved in peptidoglycan pathway, we chose three Mur enzymes (MurB, MurE, and MurG) for detailed investigations based on enzymatic as well as functional classification. The proteins with significant amino acid sequence similarity to MurB, MurE, and MurG from A. baumannii were collected using position-specific iterated BLAST (PSI-BLAST or advanced BLAST or iterative BLAST) [28] search against Non-Redundant Protein Sequence Database (NRDB) at the National Centre for Biotechnology Information (NCBI) site (https://blast.ncbi.nlm.nih.gov/Blast.cgi). This iteration was running until it did not find any homologous sequences of the given query. After the results of PSI-BLAST, we prepared three input datasets (MurB with related sequences, MurE with related sequences, and MurG with related sequences) for further detailed analysis with the exclusion of the redundant, hypothetical, putative, predicted, and uncharacterized protein sequences. The multiple protein sequence alignment (MSA) was performed for three individual groups (MurB with related sequences, MurE with related sequences, and MurG with related sequences). Once MSA alignment was completed, the phylogenetic tree was constructed for the aforementioned three groups using five different methods namely UPGMA, MP, ME, NJ, and maximum likelihood (ML).

Consensus secondary structure prediction
The secondary structural elements (Helix, Sheet, and Coil) of MurB, MurE, and MurG from A. baumannii as well as very closely related proteins were predicted using Consensus Secondary Structure Prediction (CSSP) [29]. Moreover, the secondary structural pattern of the MurB, MurE, and MurG, as well as their closely related protein sequences, was also investigated.

Three-dimensional structure prediction
The three-dimensional structure of MurB, MurE, and MurG from A. baumannii was reported previously [7,30]. Here, we used these same models from A. baumannii for comparative structural studies with their related protein models. In this study, we predicted the threedimensional structure of MurB from A. calcoaceticus, MurE from A. seifertii, and MurG from A. pittii using Swiss Model web server [31] based on following three structural templates (PDB IDs: 4JAY, 4C12, and 1F0K, respectively) using homology modeling or comparative protein modeling approach. Three template structures share significant sequence identity, sequence similarity, and query coverage with the target sequences. Once three models are generated from Swiss model [31], they will be subsequently subjected into YASARA [32] web server for refinement of the protein structures followed by validation using Verify3D [33], Procheck [34], and ERRAT [35] which are available at Structure Analysis and Verification Server (SAVES) [36] web server. The optimized models were further used for structural superposition studies with previously reported models from A. baumannii using PyMOL [37] and FATCAT programs [38], respectively. MurF. These four Mur ligases are responsible for the successive additions of L-alanine, D-glutamate, mesodiaminopimelate or L-lysine, and D-alanyl-D-alanine to UDP-N-acetylmuramic acid. All four Mur ligases are topologically similar to one another, even though they display low sequence identity. They are composed of three domains such as N-terminal Rossmann-fold domain responsible for binding the UDPMurNAc substrate, a central domain (similar to ATP-binding domains of several ATPasesand GTPases), and a C-terminal domain (similar to dihydrofolatereductase fold) that appears to be associated with binding the incoming amino acid. Based on results obtained from EMBOSS-Antigenic, the Mur enzymes have several putative antigenic sites. The reason for predicting the antigenic sites is whether the predicted binding site residues (interactions with ligand molecules) belong to antigenic site or not. Since, antigenic sites are crucial for pathogenicity and virulence of the microorganism. In this study, we found 16,22,13,14,21,19,25, and 20 putative antigenic sites for MurB, MurE, MraY, MurG, MurD, MurF, MurC, and MurA, respectively and most of the antigenic sites are found in the binding pocket of concerned protein.

Result
Together, the results obtained from aforementioned protein sequence analysis of eight Mur family proteins were divided into three main categories, namely transferases (MurG, MurA, and MraY), ligases (MurC, MurD, MurE, and MurG), and oxidoreductase enzymes (MurB) ( Table 2). The biological process of these enzymes is cell division, regulation of cell shape, cell cycle, and the cell wall organization. They are using the same pathway known as peptidoglycan biosynthesis, which is a critical role for the formation of cell wall in prokaryotic organisms. The four (MurC, MurD, MurE, and MurF) Mur ligases are responsible for the successive additions of Lalanine, D-glutamate, meso-diaminopimelate or L-lysine, and D-alanyl-D-alanine to UDP-N-acetylmuramic acid in the peptidoglycan pathways ( Table 2).

Intra-comparative protein sequence analysis of Mur family proteins
In the present study, the classification of Mur family proteins was carried out using comparative protein sequence analysis approach, according to sequence level similarity, identity, and pairwise distances; this provides information about the evolutionary relationship among Mur proteins from A. baumannii. This study includes multiple sequence alignment followed by phylogenetic analysis.

Multiple sequence alignment and phylogenetic tree construction
The results of multiple sequence alignment of Mur proteins from A. baumannii indicated that there is a significant variation among the input dataset (Fig. S16). The results of our comparative protein sequence alignment showed that there are four essential sites present in the Mur enzymes, namely conserved, variable, singleton, and parsimony-informative and their statistics are 4/520, 481/520, 275/520, and 195/520, respectively. The overall mean average of input dataset of eight Mur proteins is 2.29. The results of MSA indicated that there is a divergence relationship among the Mur proteins but all Mur ligases are topologically similar to one another, even though they display low sequence identity, similarity and pairwise distance. We observed an overall mean average of our input dataset of eight Mur proteins showing around 2.29, which indicated that dataset has diverged when we are comparing all at once. In contrast, they are nearly similar when we were comparing one with others in the input dataset of eight Mur proteins. The pairwise distance analysis indicated that the minimal distances were noticed in between MurD and MurC; on the other hand, the maximal distances were observed between MurB and MurG (Tables 4 and 5). According to the results of pairwise distance, two broad groups were found, and it indicates that MurC and MurD are highly similar with the score of 1.66 compared to other; in contrast, MurG and MurB are distant with the values of 2.86. In general, MurB is less similar to other proteins in the groups, and hence it is considered as out-group from this analysis. The MurC, MurD, MurE, and MurF belongs to the same group (group I), explained that they might have involved similar functional role concerning the synthesis of peptidoglycan. Additionally, MurG, MurA, and MraY are nearly similar to each other. Overall results conclude that selecting of MurE, MurB, and MurG are reasonable for further detailed studies for inter-species comparison.
Moreover, the inter-species comparison will help us to explore the sequential and structural properties of homologous proteins in other species and hence opens a gateway for novel antibacterial therapeutics using the common inhibitor for drug targets in different species caused for various diseases. This result also support the previous primary sequence analysis results concerning the classification of enzyme according to the function of the proteins.
The phylogenetic analysis was performed for eight Mur proteins from A. baumannii using NJ, ME, UPGMA, and MP methods. All ambiguous positions were removed for each sequence pair (pairwise deletion option). There were a total of 520 sequence sites that were observed in the final dataset. Evolutionary analyses were conducted using MEGA X [27], indicating that there are two major groups (group I contains MurC, MurD, MurE, MurF, and MurB and the group II comprises of MurG, MurA, and MraY). MurB is very far from the first group as well as the second group. Group I was divided into two subgroups which are representing the following members, subgroup 1 (SG1) and subgroup 2 (SG2). SG1 represents the following members such as MurE, MurD, MurF, and MurC; on the other hand, SG2 includes MurB. Group II also contains SG1 and SG2. The SG1 represents MurG and SG2 include MurA and MraY. From the results obtained from phylogenetic analysis ( Fig. 1), it has been observed that MurC, MurD, MurF, and MurE are very similar with each other and fall in two same groups but MurB is considered as outgroup. On the other hand, MurG, MurA, and MraY are similar to each other so that selecting one protein for further study from each group and subgroup may be representing the others concerning common structural and functional relationship.
Inter-comparative protein sequence alignment, phylogenetic tree construction, and analysis Based on the results of primary sequence analysis and phylogenetic tree construction of Mur members, we have selected MurE from ligase group, MurG from transferase group, and finally MurB from oxidoreductase group for further analysis.

MurB with closely related sequences
The MurB protein sequence was retrieved from the Uni-Prot database (https://www.uniprot.org/) and submitted into PSI-BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) against NRDB until no newly protein sequences are displayed during the cycle of PSI-BLAST. After completing the seven cycles of search of MurB against NRDB using PSI-BLAST, we identified 241 homologous sequences. Out of 241 protein sequences, 211belongs to the genus of Acinetobacter, 16 belongs to the genus of Algoriphagus, 4 belongs to the genus of Echinicola, 2 belongs to the genus of three species (Pseudomonas, Moraxellaceaeand Alkanindiges), and finally, 1 belong to the genus of four species (Serratia, Prolinoborus, Litoribacter, and Cytophagales) (Fig. 2). The amino acid sequences of most of the genus perform a similar function as MurB which are responsible for the peptidoglycan biosynthetic pathways. Since our interest is to understand the evolutionary relationship of Acinetobacter species, particularly baumannii, further screening are directed toward 211 genus alone, which are further classified based on species alone (Fig. 3). Finally, we have obtained 51 different species from the genus of Acinetobacter which are closely related to the A. baumannii. The statistical information of various species related to MurB from A. baumannii was presented in Figs. 2 and 3.

Multiple sequence alignment and phylogenetic tree construction
The results of multiple sequence alignment of MurB protein from A. baumannii with its related sequences indicated that there is a sequence variation among them. The results of our comparative protein sequence alignment showed that there are four significant sites are present in the input data set, namely conserved, variable, singleton, parsimony-informative, and their statistics were 68/399, 303/399, 17/399, and 286/399 respectively. The overall mean average of this dataset was found to be 0.45. Overall results of MSA followed by phylogenetic analysis explained that there is a significant relationship among the input dataset of MurB and its related entries. The phylogenetic analysis was performed using five different methods. For the clarity, we discussed the results obtained from the NJ method (Fig. 4) in the main-text and remaining was presented in supplementary section (Figs. S1-S5). Evolutionary analyses are conducted in MEGA X [27] which indicated that two major groups are observed in this phylogenetic tree (Figs. S1-S5). But according to our study, we are focusing on the groups and protein sequences nearest to our target, i.e., MurB from A. baumannii is presented here. This group is also further divided into two main subclasses, namely class A and class B. Since the MurB protein sequence from A. baumannii was located in class A, we considered this class for detailed investigation. In this class, we observed the protein sequences mostly from A. baumannii and interestingly one MurB from A. calcoaceticus.  From the results of phylogenetic analysis, we observed one very closely related sequence of MurB from A. calcoaceticus, which is found in the same clade where MurB of A. baumannii existed. To understand the structural and functional relationship between two proteins, we attempted detailed sequence and structural analysis, and these results were presented here (Fig. 5). The pairwise sequence alignment results between MurB protein from A. baumannii and A. calcoaceticus showed that both the proteins are highly similar with the sequence identity of 97.5% and sequence similarity of 97.2%, respectively. The physicochemical parameters of A. bau- Based on the results obtained from sequence analysis, we conclude that both the proteins from two different species are highly similar. Hence, one common molecule might bind to the active site of MurB from two different species. Together with A. baumannii, A. calcoaceticus is commonly known as A. calcoaceticus-A. baumannii complex. It can be pathogenic, and route causes opportunistic infection in patients with multiple underlying diseases. Our preliminary analysis opens the gateway to treat the infections caused by two different bacteria; the former one is a hospital-acquired nosocomial pathogen and later on, an opportunistic pathogen.

MurE with closely related sequences
The protein sequence of MurE was obtained from the UniProt database (https://www.uniprot.org/) and subsequently subjected to PSI-BLAST (https://blast.ncbi.nlm. nih.gov/Blast.cgi?PAGE=Proteins) search against NRDB until no newly protein sequences are displayed during the iterations of PSI-BLAST. After finishing the five cycles of PSI-BLAST search of MurE, we identified 179 protein sequences from various species. Out of 179 related sequences of MurE, the majority of entries belong to the genus of Acinetobacter (175 in total). However, few entries belong to the genus of Pseudomonas (2 in total), Serratia (1 in total) and Trichonephila (1 in total). Interestingly, we observed one ESKAPE pathogen from this phylogenetic analysis named as Pseudomonas, which is also showing MDR property as Acinetobacter. Statistical division of MurE from A. baumannii and its related sequence neighbors was presented in Fig. 6. Out of 175 related protein sequences of Acinetobacter, we observed diverse types of Acinetobacter species which is around 47 entries (Fig. 7).

Multiple sequence alignment and phylogenetic tree construction
The results of multiple sequence alignment of MurE proteins from A. baumannii with its related sequences indicated that there is a sequence variation among them. The statistics of conserved, variable, singleton, and parsimony-informative sites are 168/1085, 347/1085, 51/ 1085, and 296/1085, respectively. On the other hand, the overall mean average of this input dataset was found to be 0.27. Overall results of MSA indicated that there is a good relationship among the input dataset of MurE and its related entries. The phylogenetic analysis was performing using MEGA X [27], which indicated that there are two major groups in this cladogram (Figs. S6-S10). But according to our study, we are focusing on the groups and protein sequences nearest to our target, i.e., MurE from A. baumannii is presented here. This group is also divided into two main subclasses, namely class A and class B. Since the MurE protein sequence from A. baumannii was located in class A, we considered this class for detailed investigation. In this class, we observed the protein sequences mostly from A. baumannii and interestingly one from A. calcoaceticus. We already found the relationship between A. baumannii and A. calcoaceticus through phylogenetic analysis of MurB with its related sequences (discussed in the previous section). This organism is again observed in this phylogenetic tree (Fig. 8) as well, which is located in class A. To find the drug target in another Acinetobacter species, we observed A. seifertii in the neighboring clade (class B). Therefore, we used MurE from A. seifertii for a detailed investigation followed by comparison with MurE from A. baumannii. The pairwise sequence alignment results between MurE protein from A. baumannii and A. seifertii showed that both the proteins are highly similar (sequence identify: 95.8% and sequence similarity: 97.8%) (Fig. 9)  Moreover, the subcellular localization of two protein was found to be cytosol and similarly both the protein sharing common domain. Based on the results obtained from sequence analysis, we conclude that both the proteins from two different species are highly similar. Hence, one common molecule might act on the active site of MurE from two different species and thus lead to be a gateway for potent antibacterial therapeutics.

MurG with closely related sequences
The MurG protein sequence was retrieved from the Uni-Prot database (https://www.uniprot.org/) and submitted into PSI-BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) against NRDB until no newly protein sequences are displayed during the iteration cycle of PSI-BLAST. After finishing the six cycles of MurG search against NRDB database, we observed 220 homologous protein sequences. Out of 220 amino acid sequences, 215 belong to the genus of Acinetobacter, and the remaining genus is Serratia, Salmonella, Pseudomonadales, Prolinoborus, and Trichonephila (Fig. 10). Out of 215 genera of Acinetobacter, finally, we have obtained 55 different species which are strictly related to the A. baumannii. The statistical information of various species related to MurG from A. baumannii was presented in Figs. 10 and 11.

Multiple sequence alignment and phylogenetic tree construction
The results of comparative protein sequence alignment of MurG protein from A. baumannii with its homologous sequences indicated that there is a sequence variation among them. From the results of multiple protein  sequence alignment, we found that 402 sites in the final dataset of MurG and its related sequences and the statistics of conserved, variable, singleton, and parsimonyinformative sites are 141, 230, 23 and 207, respectively. The overall mean distance of this dataset was found to be 0.21. Results of MSA indicated that there is a good relationship among the input dataset of MurG and its related entries. Evolutionary analyses were carried out using MEGA X [27] indicating that there are two major groups in this cladogram (Figs. S11-S15). But according to our study, we are focusing on the groups and protein sequences nearest to our target, i.e., MurG from A. baumannii is presented here. This group also classified into two main divisions, namely class A and class B. Since the MurG protein sequence from A. baumannii was located in class B, we considered this division for detailed investigation. In this division, we observed that the protein sequence that is nearest to A. baumannii is A. pittii (Fig. 12). The pairwise sequence alignment results between MurG protein from A. baumannii and A. pittii showed that both the proteins are highly similar with the identity of 96.7% and similarity of 99.2%, respectively (Fig. 13). is the cytoplasm; in contrast, MurG protein from A. pittii is found to be cell membrane. Both these proteins shared a common domain. Based on the results obtained from sequence analysis, we conclude that both the proteins from two different species are highly similar (Fig. 13).

Consensus sequence-based secondary structure predication
The secondary structural study is significant as it provides direct imminent into the functional role of a  can lead to identifying the similarity at the sequence level alone. To find out the structural similarity of these two enzymes, we need to have the secondary and tertiary structure of both the proteins, thus will be helped to understand the proteins in a particular way. For the prediction of the secondary structure of MurB from both the species, we used Consensus Secondary Structure Prediction (CSSP), which is available at http://bioser-ver1.physics.iisc.ernet.in/cgi-bin/cssp/run_cssp.pl [29]. CSSP web server observed that both the proteins belong to mixed αβ class protein. The composition of alpha and  beta sheets for both the protein was found to be 29.74% and 11.05% for A. baumannii, whereas 27.62% and 13.66% for A. calcoaceticus (Fig. 14). These results conclude that the secondary structure composition of both the proteins is nearly similar, which corroborate with previous findings of primary sequence analysis followed by phylogenetic analysis. As MurB, the secondary structure of MurE from A. baumannii and A. seifertii was predicted through CSSP, which explained that both these proteins again belong to mixed αβ class protein.
The composition of alpha and beta sheets for both these proteins was found to be 33.8% and 15% for A. baumannii, whereas 33% and 14.6% for A. seifertii (Fig. 15). The secondary structural classes of MurG from A. baumannii and A. pittii belong to mixed αβ class. The alpha and beta sheets compositions of MurG from two different species are 47.1% and 9.9% for A. baumannii and 47.9% and 9.3% for A. pittii, respectively (Fig. 16). The alphahelical composition of MurG (from both species) is higher than MurB and MurE (from both the species). In summary, the consensus secondary structure prediction results revealed that the secondary structure of MurB, MurE, and MurG from A. baumannii is similar with the related MurB (A. calcoacetic), MurE (A. seifertii), and MurG (A. pittii) proteins.

Three-dimensional structure predication of selected three Mur targets and their related proteins
For understanding the functional relationship of three Mur proteins from A. baumannii with their related members namely A. calcoaceticus, A. seifertii, and A. pittii, we attempted three-dimensional structure prediction of selected three related proteins using a comparative protein modeling approach. The three-dimensional structure of MurB, MurE, and MurG from A. baumannii was reported previously [7,30]. Here, we used those models from A. baumannii for comparative structural studies with their related protein models. In this study, we predicted the three-dimensional structure of MurB from A. calcoaceticus, MurE from A. seifertii, and MurG from A. pittii based on following structural templates (PDB Accession No: 4JAY, 4C12 and 1F0K, respectively) using a comparative protein modeling approach using Swiss Model web server [31]. The structural templates were selected by using Protein BLAST against Protein Data Bank (www.rcsb.org). The template selection is again conformed with the results obtained from the first step of Swiss Model [28]. The sequence similarity search against PDB database of MurB amino acid sequence from A. calcoaceticus revealed that 4JAY was the best template with the query  coverage of 96%, sequence identity of 41.25%, and a statistical significance of 4e-87 compared to the other homologous sequences obtained from the Protein BLAST search. Thirty percent identity exists in between the query and template is sufficient cut-off to perform the comparative or homology modeling. In the case of MurE from A. seifertii, based on the sequence similarity search against PDB database, two templates including 4C12 (33.05%, 91%, and 2e-65) and 1E8C (35.95%, 92%, and 2e-71) were selected on the basis of sequence identity, query coverage and E (statistical significance) value, respectively. But, according to further selection criteria such as crystal structure resolution (1.80 Å for 4C12 and 2.00 Å for 1E8C) and model quality (the protein was modeled in both the templates but the model built by 4C12 had better model evaluation results), 4C12 was selected for the model building processes. The template selection for MurG from A. pittii showed that on the basis of sequence similarity search, two templates namely 3S2U (49.16%, 95%, and 2e-109) and 1F0K (44.13%, 97%, and 3e-90) were selected on the basis of sequence identity, query coverage, and E (statistical significance) value, respectively. But, according to further selection criteria such as resolution (the resolution 1.9 Å for 1F0K and 2.23 Å for 3S2U) and model quality, 1F0K was selected for the model building processes. The three-dimensional structure of these three models belongs to mixed αβ class as Mur proteins from A. baumannii.
The predicted model of three proteins geometrically optimized using YASARA [32] web server and their validation reports were given Fig. S17. The structural quality of MurB from A. calcoaceticus, MurE from A. seifertii, and MurG from A. pittii models were evaluated through stereochemical parameters (Ramachandran plot, Φ-ψ plot and G factor using PROCHECK), Verify_3D and ERRAT evaluating servers. These three servers are available in single web platform named as Structure Analysis and Verification Server (SAVES) (https://servicesn.mbi.ucla.edu/SAVES/). Based on the Ramachandran plot analysis, MurB from A. calcoaceticus model had 88.6% of all its residues in the most favorable regions and 11.4% were in the additional allowed regions. In the case of ERRAT quality factor, a significant result was obtained with a value of 97.21%. A good and reliable structure has an ERRAT score of 95% and more [35]. This model structure had a Verify_3D validation score of 94.89% and Verify_3D tool cut-off values is greater or equal to 80%. In the same way, the evaluation results of predicted model of MurE from A. seifertii revealed that 98% of the residues in the most favorable and additionally allowed region in the Ramachandran plot analysis, 91.48% ERRAT quality factor score and 91.51% Verify_ 3D validation score. Further, the model evaluation results of MurG from A. pittii indicated that around 99% residues found in the most favorable and additional allowed region in the Ramachandran Map, ERRAT quality factor of 98.25% and Verify_3D score of 86.52%, respectively (Fig. S17). Aforementioned structure validation results explained that predicted models are more reliable and reasonable and they can be used for further studies. Structure superposition studies were performed for MurB, MurE, and MurG from A. baumannii with MurB from A. calcoaceticus, MurE from A. seifertii, and MurG from A. pittii, respectively. The results obtained from superimposition ( Fig. 17) indicated that the predicted models from A. baumannii and related models from three species are significantly similar with the RMSD of 0.54 (AbMurB with AcMurB), 0.59 (AbMurE with AsMurE), and 0.50 (AbMurG with ApMurG). In  addition to this, pairwise structure alignment was also performed for all the models using FATCAT [38], and these results explained that they are significantly similar with the P value of 0.00e+00.

Discussion
The primary protein sequence analysis results (Tables 1  and 2) of Mur family proteins explained that all the proteins were hydrophobic due to the presence of polar and non-polar amino acid residues in their protein sequence (Tables 3 and 6). The hydrophobic nature of the protein indicated that the protein works well in non-polar solvents by crossing the plasma membrane for the formation of the cell wall through the peptidoglycan biosynthetic pathway. The total number of negatively and positively charged amino acid residues of these Mur proteins indicated that most of them are negatively charged, and only two proteins are positively charged, namely MurG and MraY. Both pI and the number of charged amino acids results indicated that most of the enzyme might optimally active in the acidic environment. But, MurG and MraY may be active in a basic environment. This result may be beneficial for developing buffer systems for purification of proteins in the near future by isoelectric focusing (IEF) and two-dimensional poly acrylamide gel electrophoresis (2D-PAGE) [39]. The   [40]. The conserved sequence motifs found in the four Mur enzymes also map to other members of the Mur ligase family [11]. The Mur family of enzymes have several antigenic sites, which will additionally be supported that predicted binding sites are critical and consider as putative active site region for docking and virtual screening against control and library of natural molecules in the ZINC database [41]. Moreover, the predicted antigenic sites (or binding sites) are mostly conserved in a homologous sequence of MurB, MurE, MraY, MurG, MurD, MurF, MurC, and MurA. Together, the results obtained from the protein sequence analysis of eight Mur family protein were divided into three main categories, namely transferases (MurG, MurA, and MraY), ligases (MurC, MurD, MurE, and MurG), and oxidoreductase enzymes (MurB) ( Table 2). The biological process of these enzymes is cell division, regulation of cell shape, cell cycle, and the cell wall organization. They are using the same pathway known as peptidoglycan biosynthesis, which is a critical role for the formation of cell wall in prokaryotic organisms. The four (MurC, MurD, MurE, and MurF) Mur ligases are responsible for the successive additions of L-alanine, D-glutamate, meso-diaminopimelate or L-lysine, and D-alanyl-D-alanine to UDP-N-acetylmuramic acid in the peptidoglycan pathways.
A recent study explained that post-translational modification (PTM) processes are very important for pathogenicity, virulence, and drug-resistance nature of ESKAPE microorganism [42]. Understanding these modifications and host proteins manipulated by these processes facilitate to examine host-pathogen interactions and help to design the more potent molecules against ESKAPE microorganism. Based on the previous reports, MDRAB might also undergo several PTMs, particularly phosphorylation [42], glycosylation [42], and acetylation [42]. From the post-translational modification results, the residues participated in PTM might be involved in interactions with various ligands and drug molecules, and this will additionally be supported that these residues are worth for further investigation. The results of intra comparative analysis revealed that MurC, MurD, MurE, and MurF belong to ligases; MurB belongs to oxidoreductase; and MurA, MraY, and MurG belong to transferases. Based on the overall mean average, Mur family protein are slightly divergent. The existing report indicated that the evolutionary relationship of antifreeze proteins showed that 1.589 was an overall mean value with significant conserved sites [39].
The results obtained from protein sequence analysis, inter-phylogenetic analysis, and secondary and 3D structure predictions, we identified potential drug targets from A. calcoaceticus (MurB), A. seifertii (MurE), and A. pittii (MurG) which are very similar to existing drug  targets (MurB, MurE, and MurG found in A. baumannii) as indicated in the overall workflow (Fig. 18). Active site region of the reported as well as predicted models were also very similar with each other. Moreover, these results open a new therapeutic route for treating the two or more bacterial diseases using a single potent molecule.

Conclusion
In silico-based screening of potential bacterial drug targets was identified in the present study with the aid of systematic computational workflow. Initially, the proteins that participated in the peptidoglycan pathways are MurA, MurB, MurC, MurD, MurE, MurF, MraY, and MurG which were retrieved from the UniProt database. After this, we performed primary sequence analysis, multiple sequence alignment, and phylogenetic tree construction and obtained results allowed us to classify the Mur proteins into three main enzymatic groups based on sequential properties. The sequence of each selected Mur protein was submitted into PSI-BLAST against NRDB to identify the homologous sequences. Based on the multiple sequence alignment, molecular phylogeny, and the pairwise sequence alignment results, we identified potential drug targets, namely MurB from A. calcoaceticus, MurE from A. seifertii, and MurG from A. pittii. The structural and functional similarity and identity of newly identified drug target proteins are validated using primary sequence analysis, consensus secondary structure prediction, and structural superimposition. This proposed methodology can also be used to identify and prioritize the drug targets in other bacteria which causes various diseases. This opens a new route for further computational and experimental studies to identify the common antibacterial molecules which may act on multi-targeted proteins from many bacteria.
Additional file 1: Phylogenetic analysis: Figure S1. Phylogenetic analysis of MurB Protein using UPGMA Method. Figure S2. Phylogenetic analysis of MurB using Neighbor Joining Method. Figure S3. Phylogenetic analysis of MurB using Maximum Parsimony Method. Figure S4. Phylogenetic analysis of MurB using Minimum Evolution Method. Figure S5. Phylogenetic analysis of MurB using Maximum Likelihood Method. Figure S6. Phylogenetic analysis of MurE using UPGMA Method. Figure S7. Phylogenetic analysis of MurE using Neighbor Joining Method. Figure S8. Phylogenetic analysis of MurE using Minimum Evolution. Figure S9. Phylogentic analysis of MurE using Maximum Parsimony method. Figure S10. Phylogenetic analysis of MurE using Maximum Likelihood Method. Figure S11. Phylogenetic analysis of MurG using UPGMA Method. Figure S12. Phylogenetic Analysis of MurG using Neighbor Joining Method. Figure S13. Phylogenetic analysis of MurG using Minimum Evolution Method. Figure S14. Phylogenetic analysis of MurG using Maximum Parsimony Method. Figure S15. Phylogenetic analysis of MurG using Maximum Likelihood Method. Figure S16. Multiple Sequence Alignment Eight Mur Family Proteins